Preface – Water-Quality Engineering K Hanaki, University of Tokyo, Tokyo, Japan & 2011 Elsevier B.V. All rights reserved...
34 downloads
1435 Views
60MB Size
Report
This content was uploaded by our users and we assume good faith they have the permission to share this book. If you own the copyright to this book and it is wrongfully on our website, we offer a simple DMCA procedure to remove your content from our site. Start by pressing the button below!
Report copyright / DMCA form
Preface – Water-Quality Engineering K Hanaki, University of Tokyo, Tokyo, Japan & 2011 Elsevier B.V. All rights reserved.
Water technology has been ever growing. It is an essential set of technologies for sustainable human society. Traditional technology, or better called just skill, to obtain, purify, and supply water was developed in the ancient era in various regions of the world. Great efforts have been made to obtain safe and adequate water as an essential resource to human life. However, still, billions of people in the world have no access to safe water. Moreover, large numbers of people have no chance to use a proper sanitation system, and this eventually deteriorates water quality and decreases the available safe water resources. Water resources are renewable theoretically. Used water does not disappear but is renewed to freshwater through evaporation by the power of solar energy. Solar energy is a natural distillation system to remove impurities present in water. However, the help of water technology is needed to maintain this renewing function in the modern world in which human activity overwhelms the natural purifying function. Conventional water technology was used as a black box through which water was purified without knowing the mechanisms, which control the physical, chemical, and biological reactions used in purification. However, such empirical use of technology cannot further improve or develop the technology. Many researchers and practitioners have developed theory-based technology, rather than mere empirical skill, for purifying water. The function of each unit process was studied and the mechanisms of separation, role of microorganisms, and process characteristics were clarified. A significant amount of knowledge has been accumulated. This knowledge improves process performance and reliability. Human beings also developed tools to examine the micro- or nanoscale reaction. Modern technology needs to be based on a deep and broad understanding of theory. Water technology is not isolated from other technologies. Many innovations to upgrade water-technology performance have been tried by applying new technologies from other fields. Membrane technology that originated in a field such as medical science or chemical engineering is an example. Nowadays, water treatment is one of the largest application areas of membrane technology. The purpose of water technology has been expanded from purification of water to water generation, energy and resource recovery. This is a practical and important area to which new
technology can be applied. Water availability is limiting human settlements. The supply of water produced from seawater or even moisture can break through this limitation. The requirements for water technology differ very much from one place to the other. The key factors are target compounds to be removed, resource and energy consideration, capacity of operating human resources, as well as economic resources. For example, a safe water-supply system in leastdeveloped areas needs technology, which can be used without frequent and sophisticated maintenance. However, such technology does not mean cheap and old technology. Newly developed innovative technology has a higher chance of implementation than old technology. Water management needs policy and system technology rather than simple connection of unit technologies. A distributed wastewater treatment system needs reliable and economically and technologically reasonable treatment technologies. A nutrient removal policy for eutrophication can be realized by introducing a technologically reasonable combination of secondary and advanced treatments. The water technology is a system technology. Resource and energy limitation has become a key factor for sustainability. Substantial amount of material use threatens the world’s resources, and energy use provokes the climate change problem. Saving resource and energy is now an indispensable aspect of water technology. The necessity of energy and resource saving further changes water technology. The current global situation regarding climate change and resource limitation enhances the recovery of resource and energy. Wastewater contains organic matter, which is biomass; therefore, obtaining carbon-neutral energy is possible. Water technology is now forming an important part of business worldwide. Every country needs safe water and environmental protection from wastewater. Technology development, implementation, and maintenance provide substantial opportunities for business. This volume includes theory, practice, and recent development of these wide range of water technologies, although all such innovative technologies cannot be included. There is no single answer to any of the particular cases. Among many options, one should choose a technology system considering the local social, economic, and engineering aspects. This volume would help such a technology choice.
1
4.01 Water and Wastewater Management Technologies in the Ancient Greek and Roman Civilizations G De Feo, University of Salerno, Fisciano (SA), Italy LW Mays, Arizona State University, Tempe, AZ, USA AN Angelakis, Institute of Iraklion, Iraklion, Crete, Greece & 2011 Elsevier B.V. All rights reserved.
4.01.1 4.01.2 4.01.3 4.01.4 4.01.5 4.01.6 4.01.7 4.01.8 References
Aqueducts Minoan and Greek Aqueducts Roman Aqueducts Cisterns and Reservoirs Water Distribution Systems Fountains Drainage and Sewerage Systems and Toilets Discussion and Conclusions
Prolegomena The past is the key for the future ‘Hydor (Water) is the beginning of everything’ Thales from Miletus (c. 636–546 BC).
Humans have spent most of their existence as hunting and food-gathering beings. Only in the last c. 9000–10 000 years, they discovered how to grow agricultural crops and tame animals. Such revolution probably first took place in the hills to the north of Mesopotamia. From there the agricultural revolution spread to the Nile and Indus Valleys. During this agricultural revolution, permanent villages replaced a wandering existence. About 6000–7000 years ago, farming villages of the Near East and Middle East became cities. Hydraulic technology began during antiquity long before the great works of such investigators such as Leonardo da Vinci (1452–1519) and Isaac Newton (1642–1727), and even long before Archimedes (287–212 BC) (Mays, 2008). During the Neolithic age (c. 5700–3200 BC), the first successful efforts to control the water flow were driven (such as dams and irrigation systems) due to the food needs and were implemented in Mesopotamia and Egypt (Mays et al., 2007). Urban water-supply and sanitation systems are dated at a later stage, in the Bronze Age (c. 3200–1100 BC). Regarding the technological principles related to water and wastewater, today it is well documented that many are not achievements of present day, but date back to 3000–4000 years ago. These achievements include both water and wastewater constructions (such as dams, wells, cisterns, aqueducts, sewerage and drainage systems, toilets, and even recreational structures). These hydraulic works also reflect advanced scientific knowledge, which allowed the construction of tunnels from two openings and the transportation of water both by gravity flow in open channels and by pressurized flow in closed conduits. Certainly, technological developments were driven by the necessities to make efficient use of natural resources, to make civilizations more resistant to destructive natural elements, and to improve the standards of life. With respect to the latter, the Greek (including Minoan) and
4 4 5 8 11 14 15 19 21
Roman civilizations developed an advanced, comfortable, and hygienic lifestyle, as manifested from public and private bathrooms and flushing toilets, which can only be compared to the modern one, re-established in Europe and North America in the beginning of the last century. Minoan technological developments in water and wastewater management principles and practices are not as well known as other achievements of the Minoan civilization, such as poetry, philosophy, sciences, politics, and visual arts. However, archaeological and other evidence indicate that, during the Bronze Age in Crete, advanced water management and sanitary techniques were practiced in several palaces and settlements. This period was called by the excavator of the palace at Knossos, Sir Arthur Evans, as Minoan after the legendary King Minos. Thus, Crete became the cradle of one of the most important civilizations of mankind and the first major civilization in Europe. One of the major achievements of the Minoans was the advanced water and wastewater management techniques practiced in Crete during that time. The advanced water distribution and sewerage systems in various Minoan palaces and settlements are remarkable. These techniques include the construction and use of aqueducts, cisterns, wells, and fountains, the water-supply systems, the construction and use of bathrooms and other sanitary and purgatory facilities, as well as wastewater and stormwater sewerage systems. The hydraulic and architectural function of the water-supply and sewer systems in palaces and cities are regarded as one of the salient characteristics of the Minoan civilization. These systems were so advanced that they can be compared with the modern systems, which were established only in the second half of the nineteenth century in European and American cities (Angelakis et al., 2010). Water and wastewater technologies developed during the Minoan, Greek, and Roman civilizations are considered in this chapter. Emphasis is given to the water resources development such as aqueducts, cisterns, wells, distribution systems, wastewater and stormwater sewerage systems construction, operation, and management beginning since Minoan times (second millennium BC). The achievements to support the
3
4
Water and Wastewater Management Technologies in the Ancient Greek and Roman Civilizations
hygienic and the functional requirements of palaces and cities during this time were so advanced that could be paralleled only to modern urban water systems that were developed in Europe and North America only in the second half of the nineteenth century (Angelakis and Spyridakis, 1996). It should be noted that hydraulic technologies developed during the Greek and Roman periods are not limited to urban water and wastewater systems. The progress in urban water supply was even more admirable, as witnessed by several aqueducts, cisterns, wells, and other water facilities discovered (Koutsoyiannis et al., 2008). These advanced Minoan technologies were expanded to the Greek mainland in later periods of the Greek civilization, that is, in Mycenaean, Archaic, Classical, Hellenistic, and Roman periods. In this chapter, a rather synoptic description of the main concepts of water and wastewater management during the Minoan, Greek, and Roman civilization is attempted. The main principles and challenges are also discussed.
4.01.1 Aqueducts Aqueducts were used to transport water from a source to the locations where the water was needed, either for irrigation or for urban water supplies, and have been used since the Bronze Age. Aqueduct bridges are man-made conduits for transporting water across rivers, streams, and valleys. As a matter of fact, a systematic evolution of water management in ancient Greece began in Crete during the early Bronze Age, that is, the Early Minoan period (c. 3500–2150 BC) (Myers et al., 1992; Mays, 2007). Starting the Early Minoan period II (c. 2990–2300 BC), a variety of technologies such as wells, cisterns, and aqueducts were used (Mays, 2007).
4.01.2 Minoan and Greek Aqueducts The water distribution system at Knossos, as well as the mountainous terrain and available springs made possible
the existence of an aqueduct (Mays, 2007; Mays et al., 2007). The Minoan inhabitants of Knossos depended partially on wells, and mostly on water provided by the Kairatos River to the east of the low hill of the palace, and on springs. Indications suggest that the water-supply system of the Knossos palace initially relied on the spring of Mavrokolybos (called so by Evans), a limestone spring located 450 m southwest of the palace (Angelakis et al., 2007; Evans, 1921–1935; Mays et al., 2007). In later periods with the increase of population, other springs at further longer distances were utilized. Thus, an aqueduct made of terracotta pipe could have crossed a bridge on a small stream south of the palace which carried water from a perennial spring on the Gypsadhes hill (Graham, 1987; Mays, 2007). A second example of an aqueduct was found in Tylissos (see Figure 1(a)). Parts of the stone aqueduct, with the main conduit at the entrance of the complex of houses, and other secondary systems led the water to a cistern dated at c. 1425–1390 BC (Mays et al., 2007). Other aqueducts were in Gournia, Malia, and Mochlos. These technologies were further developed during the Hellenistic and Roman periods in Crete, and were transferred to continental Greece as well as other Mediterranean locations (Angelakis et al., 2007; Angelakis and Spyridakis, 2010). In the Archaic and the Classical periods of the Greek civilization, aqueducts were built similar to the ones built by the Minoans and Mycenaeans. One of the most renowned watersupply systems is the tunnel of Eupalinos on Samos Island. In fact, it is the first deep tunnel in history that was dug from two openings with the two lines of construction meeting at about the central point of the distance. The construction of this tunnel was made possible by the progress in geometry and geodesy that was necessary to implement two independent lines of construction that would meet (Koutsoyiannis et al., 2008; Mays et al., 2007). The Samos aqueduct system includes the 1036-mlong tunnel and two additional parts for a total length greater than 2800 m. Its construction started in 530 BC, during the tyranny of Polycrates and lasted 10 years. It was in operation until the fifth century AD (Koutsoyiannis et al., 2008).
Figure 1 Ancient Minoan and Greek aqueducts: (a) aqueduct entering Tylissos showing the stone cover and (b) Peisistratean aqueduct consisting of terracotta pipe segments and elliptical pipe openings in each pipe. Copyright permission with LW Mays.
Water and Wastewater Management Technologies in the Ancient Greek and Roman Civilizations
Obviously, there are several other acknowledged aqueducts in Greek cities since water supply was regarded a crucial and indispensable infrastructure of every city (Tassios, 2007). Aqueducts (either tunnels or trenches) were always subterranean due to safety and security reasons. Usually, at the entrance of the city, aqueducts would branch out in the city to feed cisterns and public fountains in central locations. The aqueducts were pipes (usually terracotta) laying in the bottom of trenches or tunnels allowing for protection. One or more pipes in parallel were used depending upon the flow to be conveyed. The terracotta pipes (20–25 cm in diameter) fit into each other and allow access for cleaning and maintenance by elliptic openings that were covered by terracotta covers (Mays, 2007; Mays et al., 2007). Water conveyed by aqueducts typically originated from karstic springs. As the history teaches us, the presence of natural springs was a prerequisite for the selection of an area to settle. As a matter of fact, the Acropolis at Athens had an aquifer and a spring named Clepsydra. With the intensified urban development as well as the increase of population, the natural springs were not able to cover the water demand. Thus, the increasing water scarcity was remedied by transferring water from distant springs by aqueducts, digging wells, and constructing cisterns for rainwater storage. In Athens all these alternatives coexisted: the Peisistratean aqueduct (see Figure 1(b)) constructed by the end of the sixth century BC was accompanied with numerous wells and cisterns. Legislative and institutional tools were developed in Athens in order to wisely and effectively manage a water-supply system with public and private elements (Mays et al., 2007; Koutsoyiannis et al., 2008). Subsequently, the technologies developed in ancient Greece were transferred to the Greek colonies both to the east in Ionia (Asia Minor, nowadays Turkey) and to the west in the Italian peninsula, Sicily, and other Mediterranean sites, most of which were founded during the archaic period. A brilliant example of this was the founding of Syracuse (on Sicily) as a colony of Corinth in 734 BC (Mays et al., 2007). Later, during the Hellenistic period, further developments were accomplished by the Greeks in the construction and operation of aqueducts due to the progress in science which led a new technical expertise. Hellenistic aqueducts usually used pipes as well as they continued to be subterranean for safety reasons (war, earthquakes, etc.). The scientific progress
1
2
3
4
5
6
5
in hydraulic (especially due to Archimedes, Hero of Alexandria) allowed the construction of inverted siphons at large scales to convey water across valleys (lengths of kilometers, hydraulic heads of hundreds of meters) (Koutsoyiannis et al., 2007, 2008; Mays, 2007; Mays et al., 2007).
4.01.3 Roman Aqueducts Springs, by far, were the most common sources of water for aqueducts even with the Romans. Water sources for the Greeks and Roman systems included not only springs, percolation wells, and weirs on streams, but also lakes that were developed by building dams. At ancient Augusta Emerita, at present-day Merida, Spain, the Roman water system included two reservoirs created by the construction of the Cornalvo and the Proserpina dams. The Proserpina dam is an earthen dam, approximately 427 m long and 12 m high. The Cornalvo dam is an earthen dam, approximately 194 m long and 20 m high with an 8 m dam crest width. Both of these dams are still used in the present day, obviously with modifications over the years. Dams were built in many regions of the Roman Empire. Aqueducts consisted of many components, including open channels and pipes. The main types of conduits used by the Romans are: (1) open channels (rivi per canales structiles), (2) lead pipes (fistuli plumbei), (3) earthenware (terracotta) pipes (tubili fictiles), and (4) wood pipes. Open channels were built using masonry or were cut in the rock and flows were driven by gravity, while the lead pipes were used for pressurized conduits including inverted siphons. A scheme representing the general path of a whole aqueduct with the basic elements is presented in Figure 2. Obviously, there are many system configurations that were built by the Romans and Greeks; however, the drawing presents the major components, including the siphon (inverted siphon) which was used in some systems. Various types of pipes constructed by the Romans included terracotta, lead, wood, and stone. One of the most impressive Roman aqueducts in Roman Greece is that in the Aegean island Lesvos (Figure 3). It is probably a work of late second or early third century AD. It was mainly used for water supply of Mytilene town, the capital of the island, and for water supply and irrigation of the southeastern area of the island, by transporting water from the lake of Megali Limni (big lake), at the Olympus mountain,
7
8
9
10
11
Figure 2 Flow sheet and components of a Roman aqueduct: (1) source – caput aquae; (2) steep chutes (dropshafts); (3) settling tank; (4) tunnel and shafts; (5) covered trench; (6) aqueduct bridge; (7) inverted siphon; (8) substruction; (9) arcade; (10) distribution basin/castellum aquae divisorium; (11) water distribution system. From De Feo G and Napoli RMA (2007) Historical development of the Augustan aqueduct in Southern Italy: Twenty centuries of works from Serino to Naples. Water Science and Technology: Water Supply 7(1): 131–138.
6
Water and Wastewater Management Technologies in the Ancient Greek and Roman Civilizations
Figure 3 Part of the impressive Roman aqueduct rises 600 m west Moria, a Lesvian village at 6 km from Mytilene town: (a) general view of the remains and (b) the base of columns. Copyright permission with AN Angelakis.
where the construction begins. The aqueduct was also fed by other secondary springs, such as the springs at the Agiassou area (i.e., Karini). It was passed through a very anomalous landscape relief; thus, it includes parts on the soil surface, tunnels, and bridges. The total length of the Lesvos aqueduct is 26 km, with a uniform slope of 0.0096 m m1. Its depth ranges from 0.65 to 1.10 m and its width from 0.35 to 0.64 m (Karakostantinou, 2006). Its maximum capacity is estimated to be of 25 000 m3 d1 a along the distance of 26 km, a route that was entirely supported by gravity. Today, the maximum water supply of the town (15 000 m3 d1) is pumping from springs of Ydata located in a lower level of that of Karini (Mytilene Municipal Enterprise for Water Supply and Sewerage, 2009, personal communication. Mytileni, Greece). Its remains at the village of Moria are 170 m long and 27 m in height and consist of 17 arches, also called Kamares laying on their column (Figure 3(a)). Each opening is divided in three successive arches based on columns. The masonry is constructed with the use of emplekton system (Karakostantinou, 2006). The columns and arches were constructed from large blocks of gray marble taken from the island; these materials were very strong and resistant to decay (Figure 3(b)). The distribution of the arches along the openings consists of three at a time – up and down – for every opening. The openings are delimited by columns, and each column has an abacus. Siphons (Figure 2(g)) were built by the Romans also, in fact many of the siphons may very well have been started by the Greeks and completed by the Romans. The siphons included a header tank for transitioning the open channel flow of the aqueduct into one or more pipes, the bends called geniculus, the venter bridge to support the pipes in the valley, and the transition of pipe flow to open channel flow using a receiving tank. Locations of siphons included Ephesus, Methymna, Magnesia, Philadelphia, both Antiochias, Blaundros, Patara, Smyrna, Prymnessos, Tralleis, Trapezopolis, Apameia, Akmonia, Laodikeia, and Pergamon (Mays et al., 2007; Tassios, 2007). These siphons were initially built with terracotta pipes or stone pipes (square stone blocks to which a hole was
carved) such as the inverted siphon at Patara (Turkey), shown in Figure 4 (Haberey, 1972). As shown in the figure this siphon was constructed from carved stone segments. Nevertheless, the need for higher pressures naturally led to the use of metal pipes, specifically from lead. One of the largest siphons was the Beaunant siphon of the aqueduct of the Gier River which supplied the Roman city of Lugdunum (Lyon, France). This siphon had nine lead pipes with a total length of 2.6 km. This siphon was 2600 m long and 123 m deep with an estimated (Hodge, 2002) discharge of 25 000 m3 d1. Pergamon was a city in western Turkey at the present-day city of Bergama. The Helenistic aqueducts constructed were the Attalos, the Demophon, the Madradag, the Nikephorium, and the Asklepieion. The Roman aqueducts constructed were the Madradag channel, the Kaikos, and the Aksu. The Madradag aqueduct which had a triple pipeline (terracotta pipe) of more than 50 km long included an inverted siphon (made of lead) longer than 3.5 km with a maximum pressure head of about 190 m (Mays et al., 2007; Tassios, 2007). It took another 2000 years later before another pipeline was constructed that could bear a higher pressure (Fahlbusch, 2006). In particular, the Attalos aqueduct was the first pipeline (buried of fired clay, and 13 cm inner diameter) in Pergamon, and it was probably constructed in the middle or second half of the third century BC, bringing water from a spring in the mountains north of Pergamon (Fahlbusch, 2006; Mays, 2007; Oziz, 1987, 1996). The Romans built mega water-supply systems including many magnificent structures. As a matter of fact, Roman aqueducts became very famous all over the world, with Rome’s water-supply system being considered one of the marvels of the ancient world (Hodge, 2002; De Feo and Napoli, 2007; De Feo et al., 2009b; Mays, 2007; Mays et al., 2007). In fact, the Romans were urban people and consumed enormous amount of drinking water in order to supply baths, public and decorative fountains, residences, garden irrigation, flour mills, aquatic shows, and swimming pools (Hodge, 2002; Tolle-Kastenbein, 2005; De Feo and Napoli, 2007; De Feo et al., 2009b; Mavromati and Chryssaidis, 2007). However, the
Water and Wastewater Management Technologies in the Ancient Greek and Roman Civilizations
7
Figure 4 Inverted siphons. (a) Inverted siphon at Patara (Turkey) made of stone pipes. (b) Reconstruction of siphon of the aqueduct of Gier, near Beaunant, France that supplied water to Ancient Lugdunum, showing ramp of siphon with header tank on the top and the nine lead pipes of the siphon. (a) From Mays LW (ed.) (2010) Ancient Water Technologies. Dordrecht: Springer and (b) From Haberey W (1972) Die ro¨mischen Wasserleitungen nach Ko¨ln. Bonn: Rheinland-Verlag.
Roman aqueducts were not built with the primary purpose of providing drinking water, nor to promote hygiene, but rather to supply the thermae and baths or for military purposes (Hodge, 2002; De Feo and Napoli, 2007; De Feo et al., 2009b). The description of the ancient Roman water-supply system is contained in some recommendations of the Latin writers: Vitruvius Pollio (De Architectura, book VIII), Plinio the Elder (Naturalis Historia, book XXXVI), and Frontinus (De Aquaeductu Urbis Romae). Roman hydraulic engineering borrowed from the experiences and techniques of the Greeks and Etruscans. However, the size of the works as well as the technical-organizational features of distribution started with them. The common Greek practice was based on underground conduits, following courses determined by terrain features (Martini and Drusiani, 2009). The Etruscan civilization flourished in central Italy from the VIII century BC onward. The Etruscan talent for water and land management is highlighted by the existence of an imposing number of works (tunnels and channels) spread over their territories of Latium and, to a lesser amount, of the other Etruscan areas (Bersani et al., 2010). The construction of an ancient Roman aqueduct was not different from the modern practice, with several modern technologies coming from Roman engineering. The building of an aqueduct started with the search for a spring. Water was collected after permeating through vaults and walls of
draining channels and settled. From the spring, water flowed into an open channel flow and air was present over the water surface (Monteleone et al., 2007). The water in the aqueducts descended gently through concrete channels. During the route, there were multitiered viaducts, inverted siphons, and tunnels to exceed valleys or steep points. At the end of its course, the channel entered into a so-called piscina limaria, a sedimentation tank to settle particulate impurities. Then, the channel flowed into a partitioning tank called castellum divisorium where there were some walls and weirs to regulate the water flowing into the urban pressure pipes (De Feo and Napoli, 2007; Monteleone et al., 2007). Rome originally used water directly from the river Tiber as well as wells and many small springs existed inside its town area, such as Acque Lautole, Acque Tulliane, Fonte Giuturna, and Fonte Lupercale. However, since the fourth century BC, Rome gradually built aqueducts (Bono and Boni, 1996). Aqua Appia was the first aqueduct built in Rome in 312 BC. It was entirely underground for a total length of around 16.561 km, equivalent to 11 190 passus (1 passus ¼ 1.48 m) and an average flow rate of 73 000 m3 d1, corresponding to 1825 quinariae (1 quinaria B40 m3 d1) (Table 1; Panimolle, 1984). It is important to specify that a quinaria has not been scientifically defined. As a matter of fact, a quinaria was a pipe of 2.3125 cm diameter and there is no unanimity on how much water is a quinaria (Rodgers, 2004). During the subsequent
8
Water and Wastewater Management Technologies in the Ancient Greek and Roman Civilizations
Table 1
Characteristics of the 11 Imperial Age Roman aqueducts
Location
Dating
Length (km)
Aqua Appia Anio Vetus Aqua Marcia Aqua Tepula Aqua Julia Aqua Virgo Aqua Alsietina Aqua Claudia Anio Novus Aqua Traiana Aqua Alexandrina Average Total
312 BC 273 BC 144 BC 127 BC 33 BC 19 BC 2 BC 52 AD 52 AD 109 AD 226 AD
16.561 63.640 91.331 17.800 22.830 20.875 32.882 68.977 86.876 58.000 22.000 45.616 501.772
Underground length (km (%)) 16.472 (99.5%) 63.312 (99.5%) 80.286 (87.9%) 12.470 19.040 32.814 53.620 72.964
(54.6%) (91.2%) (99.8%) (77.7%) (84.0%)
43.872 (86.8%) 350.978
Average slope (m km1)
Flowrate (m3 d1)
0.6 3.6 2.7 5 12.4 0.2 6 3.8 3.8 3.8 1 3.9
73 000 175 920 187 600 17 800 48 240 100 160 15 680 184 280 189 520 113 100 21 025 102 393 1 126 325
From Panimolle G (1984) Gli Acquedotti di Roma Antica (The Aqueducts of Ancient Rome). Rome: Edizioni Abete; Adam JP (1988) L’Arte di Costruire presso i Romani. Materiali e Tecniche (Roman Building: Materials and Techniques). Milan: Longanesi; Bono P and Boni C (1996) Water supply of Rome in antiquity and today. Environmental Geology 27: 126–134; Hodge AT (2002) Roman Aqueducts & Water Supply, 2nd edn. London: Gerald Duckworth; Rodgers RH (2004) Sextus Iulius Frontinus. On the Water-Management of the City of Rome. De Aquaeductu Urbis Romae. Cambridge: Cambridge University Press.
500 years, 10 more aqueducts were constructed. The last great aqueduct built in Rome in ancient times was the 22-km-long Aqua Alexandrina. On the whole, the 11 Imperial Age Roman aqueducts had a total flow rate of 1.13 106 m3 d1 and a total length of more than 500 km. Since the population of Rome at the end of the first century AD was about 500 000 inhabitants (Bono and Boni, 1996), a mean specific discharge of B2000 l inhabitant1 d1 was produced. This value is extraordinary if compared with the current specific water use of B200–300 l inhabitant1 d1. Nowadays, the popular but inaccurate image is that Roman aqueducts were elevated throughout their entire length on lines of arches, called arcades. Roman engineers, as their Greek predecessors, were very practical and therefore whenever possible the aqueduct followed a steady downhill course at or below ground level (Hansen, 2006). As a matter of fact, Table 1 shows that on average 87% of the length of the Rome’s aqueduct system was underground. The longest aqueduct in the Roman world was constructed in the Campania Region, in Southern Italy. It is the Augustan Aqueduct Serino-Naples-Miseno, which is not well known due to there being no remains of spectacular bridges, but it was a masterpiece of engineering. The Serino aqueduct was constructed during the Augustus period of the Roman Empire, probably between 33 and 12 BC when Marcus Vipsanius Agrippa was curator aquarum in Rome, principally in order to refurnish the Roman fleet of Misenum and secondarily to supply water for the increasing demand of the important commercial harbor of Puteoli as well as drinking water for big cities such as Cumae and Neapolis. The main channel of the Serino aqueduct was approximately 96 km long, and had seven main branches to towns such as Nola, Pompeii, Acerra, Herculaneum, Atella, Pausillipon, Nisida, Puteoli, Cumae, and Baiae (De Feo and Napoli, 2007; De Feo et al., 2010). In summary the Romans made great contributions to the advancement of the engineering of aqueducts. Fahlbusch
(2006) points out the following from examination of many aqueducts: 1. size of the aqueduct channel was chosen according to the estimated discharge and the size varied along the course of the aqueduct; 2. the cross section was large enough for people to walk through the channel for repair and maintenance, particularly to remove calcareous deposits; and 3. the cross section was kept constant allowing manifold uses for encasings, especially the soffit scaffoldings for the vaults in a kind of industrialized construction.
4.01.4 Cisterns and Reservoirs In general, cisterns were usually constructed in order to store rainwater for domestic use (private houses), with a volume in the order of dozens of cubic meters, while reservoirs were realized in order to store flowing water with a volume in the order of thousands of cubic meters (Tolle-Kastenbein, 2005; De Feo et al., 2010). The Minoan and Mycenaean settlements used cisterns a 1000 years before the classical and Hellenistic-Greek cities. Cisterns were used to supply (store runoff from roof tops and court yards) water for the households through the dry summers of the Mediterranean. In ancient Crete, in particular, the technology of surface and rainwater storage in cisterns for water supply was highly developed and has continued to be used in modern times. One of the earliest Minoan cisterns was found in the center of a pre-palatial house complex at Chamaizi dating back to the turn of the second millennium BC. It is located on the summit of a hill and its rooms were situated around a small open court with a deep circular rock-cut cistern, 3.5 m in deep and with a diameter of 1.5 m, lined with brickwork in its upper part (Davaras, 1976; Mays et al., 2007; Angelakis and
Water and Wastewater Management Technologies in the Ancient Greek and Roman Civilizations
Spyridakis, 2010). Four of the earliest Minoan structures which may be considered to be large cisterns were built in the first half of the second millennium BC at Pyrgos-Myrtos (Ierapetra), Archanes, Tylissos, and Zakros (Cadogan, 2007; Mays et al., 2007; Angelakis and Spyridakis, 2010). While, at Phaistos, water supplied to cisterns depended on precipitation collected from rooftops and courts, a supplementary system was needed to satisfy the needs of water supply, especially in this particular area where agriculture was widely practiced. Thus, water was probably taken from wells in a location southwest of the palace which was rich in groundwater and surface water, and from the river Ieropotamos located to the north, at the foot of the Phaistos hill (Gorokhovich, 2005; Mays et al., 2007; Angelakis and Spyridakis, 2010). There were also cisterns on the high grounds above the Minoan palace in Malia, in a site lying in a narrow plain between the mountains and the sea. At the famous Phaistos palace, cisterns depended on precipitation collected from rooftops and yards. A supplementary system of water supply was needed to satisfy the needs of water supply, especially in those areas where agriculture was intensive. The cisterns were connected to small channels collecting spring water and/or rainfall runoff from catchment areas. The use of cisterns preceded channels or aqueducts in supplying the palace and the surrounding community with water (Mays et al., 2007; Angelakis and Spyridakis, 2010). Most Greek houses had a cistern supplied by rainwater for purposes of bathing, cleaning, houseplants, domestic animals, and even for drinking during shortages of water. Most likely, the water was of a quality that would be subpotable using today’s standards. Aristotle in his Politics (vii, 1330 b) written around 320 BC asserted that ‘‘cities need cisterns for safety in war.’’ During this time a severe 25-year drought required the collection and saving of rainwater. Also about this time cisterns were built in the Athenian Agora for the first time in centuries (Crouch, 1993; Mays, 2007). In particular, in the ancient Greek city of Dreros on Crete, there is a rectangularshaped cistern with dimensions of approximately 13.0 5.5 6.0 m3 (Antoniou et al., 2006; Mays, 2007). In ancient Crete, the technology of surface and rainwater storage in cisterns is continued to be used even today. Four of the earliest Minoan structures which may be considered to be large cisterns were built in the first half of the second millennium BC (the time of the first Minoan palaces) at PyrgosMyrtos (Ierapetra), Archanes, Tylissos, and Zakros (Angelakis et al., 2010). The Tylissos cistern is shown in Figure 5(a). This technology has been further improved during the Hellenistic and Roman periods. An impressive pillar of two interconnected cisterns, 40 m deep cut in the rock, has been discovered in ancient city Eleutherna (Figure 5(b)). The dimensions of the two cisterns are 40 25 m2 and the depth 4.5 m. The city flourished in the early Christian times and the water was transported from a spring through an aqueduct of about 3 km long to the cisterns. The water supply of the city including the thermes was transported through a 150-m-long channel with dimensions of 1.5 2.0 m2. The advanced water-supply technologies developed in Minoan Crete were expanded and improved during the Roman domination of the Greek world. Two such examples with a relatively small but impressive cistern in Minoan city and one of the two huge cisterns
9
(of about 3000 m3 each) in Aptera city in the western Crete are shown in Figures 5(c) and 5(d), respectively. During the classical age (the period between the Archaic and Roman epoch), the political situation was characterized in the Greek world (mainly Greece and Asia Minor) by wars among the various cities. In this period, no springs or deep wells existed, so cisterns were constructed to collect rainfall during the winter season. These cisterns were dug into the rock and were mostly pear-shaped with at least one layer of hydraulic plaster that prevented water loss. The cisterns varied in size from 10 m3 to thousands of cubic meters and possibly supplied more than 10 000-people baths and thermes. To prevent contamination of water the mouth of the cistern was covered to keep out dust and debris, and to prevent light from entering, avoiding the growth of bacteria and algae. Reservoirs constructed by the ancient Romans were set low in the ground, or actually underground, and roofed over, by means of concrete vaulting. The roofing vaults were supported by rows of columns, piers, or wall pierced with doors to allow the water to circulate. In some cases, the floor was slightly concave with a drain in the middle, to permit cleaning (Hodge, 2002; De Feo et al., 2010). In general, in the Roman world the reservoirs had two functions: a reservoir could be a reserve for use when the aqueduct ran low or by adding in a little from the tank everyday to supplement supplies until the aqueduct discharge picked up again. When the daily consumption exceeded what the aqueduct could bring in, at least in the hours of daylight, the reservoir was topped up every night to meet the next day’s demands (Hodge, 2002; De Feo et al., 2010). An example of a Roman reservoir is the Bordj Djedid at Carthage in Tunisia, into which the Carthage aqueduct emptied after a run of no less than 90.43 km from its source. This great reservoir was oblong, 39.0 154.6 m2, the size of an entire city block, and subdivided into 18 transverse compartments. Its capacity was 25 000–30 000 m3, representing about a day and a half’s discharge for the aqueduct (Hodge, 2002; De Feo et al., 2010). Remaining in Tunisia, in the center of the city of Dougga/Thugga, there are two very large reservoirs. The first one is the Ain El Hamman reservoir with five aisles, while the second one is the Ain Mizeb reservoir with seven aisles. The two reservoirs have a combined storage volume of 15 000 m3 (Tolle-Kastenbein, 2005; De Feo et al., 2010). Large reservoirs were constructed not only in Northern Africa but also in Europe, especially in Italy and in Turkey. Since a Roman thermae required an enormous quantity of water for its functioning, a huge reservoir had to be constructed. As a matter of fact, the reservoir of the Baths of Caracalla (located in an area of over 100 000 m2) could contain over 80 000 m3 in the numerous cells, situated into two parallel aisles and onto two floors. The oldest baths of Traiano received water supply from a reservoir of around 10 000 m3 (Tolle-Kastenbein, 2005; De Feo et al., 2010). The greatest baths of Diocletian occupied about the same area as those of Caracalla (a rectangle of about 356 316 m2) and closely resembled them in the plans. The reservoir by which the baths were supplied was fed by the aqua Marcia, the volume of which was increased by Diocletian. It was trapezoidal in shape, 91 m in length, with an average width of 16 m. This reservoir, called Botte di Termini (Barrel of Termini), was
10
Water and Wastewater Management Technologies in the Ancient Greek and Roman Civilizations
Figure 5 Minoan, Hellenistic, and Roman water collection and storage cisterns: (a) Minoan at the ancient town of Tylissos; (b) Hellenistic at the city of Eleutherna; (c) Roman at the Minoa town; and (d) Roman at town of Aptera. Copyright permission with AN Angelakis.
destroyed during 1876 in order to build the Termini railway station, whose name derives from that of the baths (De Feo et al., 2010). In the three centuries of the Roman imperial age, the reservoirs were designed in almost all the architectural forms and in almost all the techniques of masonry known: arcs (especially transversal arcs), turned (especially barrel vault), carrying pillars or groups of pillars, walls of stones and bricks, opus caementicium; while columns were still not used. In fact, the columns were introduced by architects famous for their works of hydraulic engineering in the present-day Istanbul. They created a host of columns hidden in the heart of the capital of the Roman Empire (Tolle-Kastenbein, 2005; De Feo et al., 2010). As a matter of fact, the name of the first reservoir means ‘with a 1001 pillars’. It is the Binbirdirek reservoir which was built under the order of Philoksenos, a Senate member in the Constantinus I period of the fourth century. During the Roman period, Istanbul’s water requirements were met by water brought from distant parts of Thrace. For this reason, the Byzantines built large reservoirs in order to be able to withstand long sieges (De Feo et al., 2010). The Binbirdirek reservoir covered an area of 3640 m2 and had a capacity of around 32 500 m3 of water. It measured 66 56 m2 and was carried by 224 columns consisting of
16 rows, each one having 14 columns, all of which are equal in length, and every column carries the signature of its master (‘1001’ was used to emphasize the great number of columns). There is a thick overlapping astragal running round the columns carrying the vaults and arches and they are in the form of a truncated pyramid and are without decoration. The relief cross on one of the columns is good proof that the reservoir was built in the fourth century, after the Byzantines accepted Christianity. In order to construct ceilings 14–15 m2 high, a second layer of columns was fixed over the marble rings on the first layer of columns. When the palace was destroyed in the sixth century, the cistern was restored. After the Ottoman conquest of Istanbul in 1453, new reservoirs were built and the Binbirdirek was no longer used (De Feo et al., 2010). One of the magnificent historical constructions of Istanbul is the Yerebatan Saray (or Basilica Cistern), located near the southwest of Ayasofya (Hagia Sophia). This huge reservoir was rebuilt by the emperor Justinian (527–565) after the Nika revolt (532). It is a large, vaulted space; the roof rests on 12 rows of 28 marble columns, which are about 9 m high. As the total surface is 65 138 m2, the maximum capacity is almost 85 000 m3, which was brought to this cistern from a well B20 km away with a new aqueduct, also built by
Water and Wastewater Management Technologies in the Ancient Greek and Roman Civilizations
Justinian. It was used to provide water to the imperial palace (hence the name, imperial cistern). The 336 columns (246 are still visible) were brought to the Basilica Cistern from older buildings. Again, it is narrated that 7000 slaves worked in the construction of the cistern. In fact, the cistern borrowed its name from Ilius Basilica in the vicinity (Lendering, 2008; Ku¨ltu¨r, 2008; De Feo et al., 2010). Another huge Roman reservoir in ancient Constantinopolis (today’s Istanbul) is the Sultan’s Cistern. We do not have any verifiable scientific evidence for its construction date; at the earliest, it could be late fourth century AD, judging by the presence of crosses carved into the upper parts of the column heads. It has a rectangular plan and the whole is divided into five equal rectangular parts by the use of 28 columns, with 7 in granite and 21 in marble, placed equidistant from each other, also supporting the roof with vaulted arches (De Feo et al., 2010). The last Roman underground hydraulic marvel is the spectacular Piscina Mirabilis in Misenum, in the Southern Italy. The Piscina Mirabilis is located in the present-day Municipality of Bacoli, in Miseno (the ancient Misenum), up the hill facing the sea in the bay of Naples. It was constructed during the Augustan Age in order to supply water to the Classis Praetoria Misenensis (Adam, 1988; Hodge, 2002; De Feo and Napoli, 2007; De Feo et al., 2010). The Piscina Mirabilis is a gigantic reservoir 72 m long and 27 m large, with a volumetric capacity of 12 600 m3 of water (Figure 6). It is dug in a tufa hill and has two step entrances in the northwest, the Ancient Roman entrance and southeast corners, the latter closed. Forty-eight pillars, arranged on four rows serving as a support to the barrel vault, divide it into five principal aisles on the long sides (Figure 7(a)) and 13 secondary aisles on the short sides (Figure 7(b)), giving it the majestic look of a cathedral. The long walls were built in opus reticolatum (reticular work) with brick bonding courses and by the technique of the tufa stone pillars, both covered with a thick waterproof layer of opus signinum (pounded terracotta). There is a basin of 1.10 m, probably a polishing pool, which is a waste bath for the maintenance of the reservoir, in the floor of the nave. It was used as a Piscina limaria for the periodical cleaning of the reservoir (Figure 7(c)). The water was lifted through a series of openings (doors) in the vault along the central nave, hydraulically to the covering terrace of the reservoir, and from there, flowed in channels to the urban area. These doors appear casually opened in the roof (Figure 7(d)), with an irregular realization being noted (Adam, 1988; Hodge, 2002; De Feo and Napoli, 2007; De Feo et al., 2010). Russo and Russo (2007) estimated a total daily demand of 12 000 m3 of water for Misenum, including 4000 m3 for the fleet and 8000 m3 for daily demands and for the thermal baths and gardens (based upon daily individual requirements of 100 liters per capita and equal requirements for thermal baths and gardens). The estimated total daily demand is similar to the capacity of the Piscina Mirabilis. Close to the Piscina Mirabilis are two other large cisterns, probably belonging to large villas, the Grotta Dragonaria and Cento Camerelle (Nerone’s jail). In Pozzuoli, the aqueduct served several cisterns, notably the Piscina Cardito (55 16 m2) from the second century, and the Piscina Lusciano (35 20 m2) from the first century AD (De Feo and Napoli, 2007; De Feo et al., 2010).
11
4.01.5 Water Distribution Systems Water distribution systems are aimed at distributing water from reservoirs or aqueducts to the end users. The modern systems are based on the use of pipes. Regarding this aspect, the Minoan society was surprisingly modern. As a matter of fact, in the Knossos palace, the water supply was furnished by means of a network of terracotta pipe conduits (60–75 cm flanged to fit into one another and cemented at the joints) beneath the floors at depths that vary from a few cm up to 3 m (Koutsoyiannis et al., 2008; Angelakis and Spyridakis, 2010). Possibly, the piping system was pressurized (Mays, 2007). Similar terracotta pipes were discovered in some other Minoan sites. In particular, Tylissos was one of the important cities in Ancient Crete during the Minoan era, flourishing (2000–1100 BC) as a peripheral center dependent on Knossos. From the aqueduct, secondary conduits were used to convey water to a sedimentation tank (Figure 8; Mays, 2010) constructed of stone before its storage to the cistern shown in Figure 5(a). Terracotta pipes have also been found at Vathypetro, as well as in the Caravanserai (Guest House), south of the Knossos palace with some also having been found scattered in the countryside (Angelakis and Spyridakis, 2010). The study of the ruins of Pompeii gives a clearer understanding of a Roman urban water distribution system. But this statement does not mean that all Roman cities are identical to Pompeii. The ending point of a Roman aqueduct was the castellum divisorium which had the double function of serving as a disconnection between the aqueduct and the urban distribution network as well as dividing the water flow to various uses and/or geographical areas of the city (Figure 9). In the beginning, Pompeii was not supplied by the Serino aqueduct. As there were no springs in Pompeii, wells were dug to supply water. It is also very likely that Pompeii received water via an aqueduct from the mountains due northeast of Avella. The town must have had a long-distance water supply, quite some time before the Augustan Age, probably around 80 BC. When the Serino aqueduct was built under Augustus, it crossed the course of the older Avella aqueduct between the Apennines and Mount Vesuvius, and both aqueducts were united into a single system (De Feo and Napoli, 2007). The castellum divisorium of Pompeii was housed inside a large brick building near the Vesuvian gate (Figure 10(a)). The supply channel entering the building is 30 25 cm (Figure 10(b)). The flow in this distribution structure was allowed to expand into a wide, shallow tank, separated into three equal compartments (masonry structures) (Figure 10(c)). Flow from each compartment entered a lead pipe. Some feel that the three pipes were connected separately to public fountains, the second to the thermal baths and the third to private users (Hodge, 2002; Russo and Russo, 2007). From the exits the water flowed into lead pipes. There is also the distinct possibility that the three pipes were directed to different geographical areas of Pompeii. Assuming that the pipes did convey water separately to the three major uses as presented by Hodge (2002), the central pipe was directed to the public fountains and had a 30 cm external diameter, whereas the two side ones were 25 cm in diameter. The three gates were of different heights. Thus, the highest gate, which was that serving private houses, cut off their supplies until and unless the water level in
12
Water and Wastewater Management Technologies in the Ancient Greek and Roman Civilizations 1.2
4.9
1.2
4.0
1.2
4.0
1.2
4.0
1.2
4.9
1.2
11.4
27.0
1.2
4.3 4.3 1.2 4.3 4.3
( Measures in meters )
1 2 3 4 5
2.0
1.2
4
Legend
N
11.4
9.4
A
Inlet water Ancient Roman entrance - 1 Piscina Limaria Outlet washing water Ancient Roman entrance - 2
1
B
1.2
4.9
1.2
4.3
2
1.2
4.3
Longitudinal section A-A
1.2
4.3
1.2
4.3
1.2
72.0
1.2
A
4.3 1.2
3
A
Plan of the Roman Piscina Mirabiliis
1.2
3.0
5
1.2
4.3
1.2
4.3
1.2
4.9
1.2
B
10.4
Trasversal section B-B Figure 6 Plan and sections of the Piscina Mirabilis. Modified from De Feo G, De Gisi S, Malvano C, and De Biase O (2010) The greatest water reservoirs in the ancient Roman world and the ‘‘Piscina Mirabilis’’ in Misenum. Water, Science and Technology: Water Supply 10(4) (in press).
Water and Wastewater Management Technologies in the Ancient Greek and Roman Civilizations
13
Figure 7 Piscina Mirabilis: (a) a cross aisle; (b) a longitudinal aisle; (c) internal piscina limaria; and (d) a hole in the barrel vaulted roof.
the main body of the castellum rose high enough to spill over it and start flowing down the channel; on the contrary, the lowest gate (that in the center) governed access to the public fountains, which, if the water level sank, were thus the least to dry up. The private users had no minimum water entitlement until the needs of the public fountains and thermal baths had been satisfied (Hodge, 2002). From the castellum divisorium, the three pipes lead the water to different parts of the city filling water towers: the castellum secondarium or castellum privatum (Figure 10(d)). The water
towers were lead tanks positioned on top of brick masonry pillars, 6 m tall, located at crossroads and connecting small numbers of customers. They also supplied public fountains. The single user had to pay to obtain water for his premises. The water was metered by means of bronze orifices, the calices connecting the customers’ pipes (usually quinariae pipes) to the castellum privatum lead tank. In Pompeii, case calices were placed at the bottom of the lead tanks, and pipes fit into cavities left in the brick pillars (Hodge, 2002; Monteleone et al., 2007).
14
Water and Wastewater Management Technologies in the Ancient Greek and Roman Civilizations
Figure 8 Water system at Tylissos, Crete, Greece with sedimentation tank in foreground with stone channel connecting to cistern in background. (Mays, 2010, Copyright permission with LW Mays).
Aqueduct
Castellum divisorium
Head 18 m
Castellum secondarium Head 6m
Figure 9 Flow sheet of a Roman urban water distribution systems based on Pompeii. Modified from Hodge AT (2002) Roman Aqueducts & Water Supply, 2nd edn. London: Gerald Duckworth.
The lead tank on the water tower acted as a disconnection between the system at high pressure upstream and the customers’ pipes downstream. Connecting water derivation pipes elsewhere in the castellum privatum was against the regulations. The only connection available had to be arranged with the water office discussing the quantities for consumption. This water-supply system clearly shows that water towers could break from the pressure built up in the mains descending from the initial castellum divisorium at the top point of the city, with excess water overflowing into streets drains. As shown in Figure 9, the maximum height of water over the tap was about 6 m, without accounting for the pressure losses in the delivering pipes (Hodge, 2002; Monteleone et al., 2007). Lead pipes (Figure 11) in Pompeii are of the same construction and appearance as found in other Roman cities. The water taps found in Pompeii were also similar to those found in other Roman cities. Only a small number of houses had
a water pipe that supplied a private bath or basins in the kitchen, in the toilet, or in the garden.
4.01.6 Fountains The Minoan civilization gave an extraordinary contribution to the development of water management practices also in terms of fountains. The main examples of Minoan fountains are subterranean structures supplied with water directly or from springs via ducts. The construction of steps or alternatively the shallow basins indicates that water was taken out with the use of a container. This recalls the type of fountain of the later Classical and Hellenistic period called arykrene. The most typical of them is that of the Zakro palace. Another fountain similar to the Tykte was found at the Guest House (Caravanserai) of Knossos in the Spring Chamber. A ritual function of
Water and Wastewater Management Technologies in the Ancient Greek and Roman Civilizations
15
Figure 10 Pompeii: (a) brick building near the Vesuvian gate housing the castellum divisorium; (b) inside castellum divisorium; (c) supply channel; and (d) a castellum secondarium.
the particular fountains is also argued, as artifacts of ritual content have also been unearthed. Another type known in later periods as rookrene, which constantly provided freshwater, was also found in Zakro with two zoomorphic waterspouts. Finally, a remarkable fragment from a fresco composition depicting a fountain of a supposedly Minoan garden was found in the House of Frescoes in Knossos (Angelakis and Spyridakis, 2010). During the Roman period, public fountains were usually located in the street. For example, in Pompeii the fountains were located at fairly evenly spaced intervals of about 100 m, and it was rare for anyone to carry their water for more than 50 m (Hodge, 2002). The simplest form of street fountain was normally equipped with an oblong stone basin, typically about 1.5 1.8 m2 and 0.8 m high, into which the spout discharged, and which presumably was normally full. The fountains were deliberately designed to overflow in order to clean the street (Hodge, 2002; De Feo et al., 2010). Not far from the city of Pompeii, in the District of Salerno, there is a Roman gallery in rock in the village of Sant’Egidio del Monte Albino in the Sarno River basin. The gallery was constructed in order to supply a public fountain which stands on the structure of an ancient Roman villae (the Helvius
villae). The Helvius fountain was a public fountain, but it was quite different from the public fountains in nearby Pompeii (Figure 12(a)). As a matter of fact, the Helvius fountain was constructed neither by means of matched slabs nor in limestone nor in Vesuvian stone. It was built as a single block of white marble. Moreover, there is another particular aspect which differentiates the Helvius fountain from the Pompeian fountains (Figure 12(b)). The Helvius fountain has a sculptural decoration on the three available sides representing the river Sarno along its path from the spring toward the sea (De Feo et al., 2010). Figure 13 shows two additional Roman fountains that are quite different from those previously mentioned. Figure 13(a) shows a fountain in Chersonesos (Crete) and Figure 13(b) the Fountain of Trajan in Ephesus (Turkey), dedicated by Aristion, AD 102/114.
4.01.7 Drainage and Sewerage Systems and Toilets Drainage systems were used for the disposal of surplus water, and were found both in cities (to carry rainfall, overflow from fountains and bathrooms) and in the country (to prevent
16
Water and Wastewater Management Technologies in the Ancient Greek and Roman Civilizations
Figure 11 Components of lead pipe system found in Pompeii: (a) lead pipe and joint found along the street; (b) junction box; and (c) manifold. Copyright permission with LW Mays.
flooding in the fields). Sewerage systems were used for the conveyance of domestic wastewater, and were only found in cities, where they were necessary due to a high population density (Hodge, 2002). However, in most cases, combined systems of flow rates composed mainly of rainfall runoff and wastewater were applied. The Minoan civilization also gave an extraordinary contribution to the development of water management practices in terms of drainage and sewerage systems. As a matter of fact, Minoan palaces were equipped with elaborate storm drainage and sewer systems (MacDonald and Driessen, 1988). Open terracotta and stone conduits were used to convey and remove stormwater and limited quantities of wastewater.
Pipes, however, were scarcely used for this purpose. Larger sewers, sometimes large enough for a man to enter and clean, were used in Minoan palaces at Knossos, Phaistos, and Zakro. These large sewers may have led to the conception of the idea of the labyrinth, the subterranean structure in the form of a maze that hosted the Minotaur, a hybrid monster. The end section of the main part of the sewerage system of the Knossos palace is shown in Figure 14(a). The outlet of the Phaistos palace system appears to be similar (Figure 9(b)). Note that Evans (1921–35) and Darcque and Treuil (1990) considered that the main part of the system had been planned and constructed originally in Middle Minoan time. The main disposal sites at the Knossos and Zakros palaces were directed
Water and Wastewater Management Technologies in the Ancient Greek and Roman Civilizations
17
Figure 12 Public fountains: (a) in Pompeii (matched slabs) and (b) in the basin of the Sarno river (single block of white marble).
Figure 13 Roman fountains: (a) fountain in Hersonissos (Crete) and (b) remains of the fountain of Trajan in Ephesus (Turkey), dedicated by Aristion, AD, 102/114. Copyright permission with LW Mays.
to the Kairatos River and to the sea, respectively. However, there are indications that in the palace of Phaistos and in the villa of Agia Triadha, cisterns were also used as disposal sites of surface water, along with appropriate landforms. Particularly in the palace of Phaistos, agricultural land located in the south site of the palace was used as disposal site of the both the wastewater and stormwater instead of the river Ieropotamos crossing the northern site of the Phaistos hill. In all cases of palaces and cities, there is an increased slope of the central sewers toward of their outlets; thus, anaerobic conditions have been maintained and the odors have been avoided. In addition to the very effective drainage and sewerage systems, some palaces had toilets with flushing systems operated by pouring water in a conduit. However, the best example of such an installation was found on the island of Thera (Santorini) in the Cyclades, Greece. This is the most eloquent and best-preserved example belonging to the early late-Minoan period (c. 1550 BC) in the Bronze Age settlement of Akrotiri, which shares the same cultural context of Crete (Angelakis and Spyridakis, 2010). At the beginning, for some centuries, the collection and discharge of rainwater runoff was managed by separate sewers.
As a matter of fact, rainwater was carried in simple channels carved into the rock in cities with bedrock (i.e., the Acropolis of Athens). Otherwise, the channels were covered with rocks. A system for the simultaneous discharge of both rainwater and domestic sewage was invented during the Greek period (Tolle-Kastenbein, 2005). Ancient drainage and sewerage systems were usually developed on four levels. The initial channels coming from buildings (first order) ended in street channels of second order, which prosecuted in principal channels with an increasing size (third order) and ended in a final huge collection channel (fourth order), usually present only in big cities. The great drain of Athens was first designed as a rainwater drainage system. However, in the first quarter of the fifth century BC, it received domestic sewage and ended in a huge collection channel (fourth order) similar to the Roman Cloaca Maxima (Tolle-Kastenbein, 2005). The Cloaca Maxima is the best-known ancient urban drain. Tradition ascribes its construction to Tarquinius Priscus, king of Rome 616–578 BC. The Cloaca Maxima (4.2 m high, 3.2 m wide) was covered by stone vaulting, while its bottom was paved with basalt pavers. It combined the three functions of
18
Water and Wastewater Management Technologies in the Ancient Greek and Roman Civilizations
Figure 14 Outlet of the central Minoan sewerage and drainage systems: (a) palace of Knossos and (b) palace of Phaistos. Copyright permission with AN Angelakis.
wastewater and rainwater removal and swamp drainage. As it is well known, the exit from the Cloaca Maxima drain into the river Tiber still exists in Rome, but now partly hidden by the modern Lungotevere Embankment (Hodge, 2002). The street drains of Pompeii are very famous. At the time of the famous Vesuvius eruption, the drains existed only in the area around the forum. The streets were a sort of open channel conveying water coming from public fountains, rainwater, and segregate sewage. Therefore, as shown in Figure 15, streets had raised sidewalks (50–60 cm high) with stepping stones (pondera) at the street corners to enable pedestrians to cross from one side to the other without stepping down (Hodge, 2002). Toilets have a long history. The first evidence of the purposeful construction of bathrooms and toilets in Europe comes from Bronze Age Minoan (and Mycenaean) Crete in the second millennium BC (Vuorinen et al., 2007). In the palace of Knossos, rainwater was probably used to flush the toilet near the Queen’s Hall (Figure 16; Angelakis et al., 2005). The Hellenistic period is considered more progressive for the sanitary and purgatory engineering during the antiquity, although the considerable spreading of these systems occurred during the Roman era. The Romans applied the earlier techniques in larger constructions, using the advantages of their
building methods with concrete walls and vaulted roofing. Moreover, due to their improved aqueduct technologies, they could provide natural water flow in most public latrines. It is also evident that such structures and installations have survived until the end of the ancient world and have been implemented during the beginning of the Byzantine period. The customs of the new religion, Christianity, modified some of the structures in terms of privacy in bathing facilities (Antoniou and Angelakis, 2009). During the Hellenistic era lavatories improved significantly, followed by their spread throughout the Roman Empire. The features of the typical ancient lavatory are the bench-type seats with keyhole-shaped defecation openings and an underneath ditch. The ditch was both a water-supply conduit for flushing and a sewer. Figure 17 shows remains of a public toilet in Ephesus (Turkey) illustrating the bench seats, the defection openings, and the small channel on the floor for cleaning the sponghia. The lavatory was usually situated in the area of the building most convenient for water supply and/or sewerage. In many cases, the water for the flushing was reused either after other domestic or communal activities. Despite privacy, lavatories were used in antiquity by many people simultaneously, from two to three people in the small domestic latrines and up to 60 people in the larger public latrines
Water and Wastewater Management Technologies in the Ancient Greek and Roman Civilizations
19
Figure 15 Stepping stones (pondera) in Pompeii.
Wooden seat
Door jamb
Gypsum floor
likely lacked running water and they were commonly located near the kitchens. All this created an excellent opportunity for the spreading of intestinal pathogens (Vuorinen et al., 2007). Hygienic conditions in both types of toilets must have been very poor, and consequently intestinal diseases were diffused. Dysentery, typhoid fever, and different kinds of diarrheas are likely candidates for diagnoses. Unfortunately, descriptions of the intestinal diseases in the ancient texts are so unspecific that the identification of the causative agent is a very problematic venture. Studies of ancient microbial DNA might offer some new evidence for the identification of microbes spread by contaminated water (Vuorinen, 2010).
Sewer
Seat
Hood
Sewer
Flushing conduit 1m Doors
Figure 16 Section and plan of ground-floor toilet in the residential quarter of palace of Minos. From Angelakis AN, Koutsoyiannis D, and Tchobanoglous G (2005) Urban wastewater and stormwater technologies in ancient Greece. Water Research 39: 210–220.
(Antoniou, 2010). Lavatories were used throughout the Roman Empire, with a more or less monumental appearance. The reader is referred to Antoniou (2010) for a detailed discussion of ancient Greek lavatories. Toilets during the Roman era can be divided into two groups: public and private. A public toilet was frequently built near to or inside a bath so that it was easily entered from both inside and outside of the bath. The abundance of water that was conducted to the bath could also be used to flush the toilet. Piped water for flushing private toilets seems to have been a rarity. The Romans, however, lacked something similar to our toilet paper. They probably used sponges or moss or something similar. In public toilets, the facilities were common to all. They were cramped, without any privacy, and had no decent way to wash one’s hands. The private toilets most
4.01.8 Discussion and Conclusions In the Minoan, Greek, and Roman cities, and other settlements, water supply varied according to local conditions, determined by climate (mainly rainfall), surface and ground water, and terrain. In these periods, various water-supply and wastewater systems and techniques were developed and applied, such as collection and storage facilities, wells and groundwater abstraction aqueducts, water distribution and use, construction and use of fountains, sewers, bathrooms, and other sanitary facilities and even recreational uses of water. These advanced technologies, which have been used in prehistoric Crete since about 4500 years ago, were subsequently expanded during the Mycenaean and then the Archaic, Classical, and Roman periods. In light of these historical and archaeological evidences, it turns out that the progress of present-day urban water and wastewater technologies as well as comfortable and hygienic living is not as significant as we tend to believe (Angelakis and Koutsoyiannis, 2003). However, a burst of achievements in water and
20
Water and Wastewater Management Technologies in the Ancient Greek and Roman Civilizations
Figure 17 Public toilet in Ephesus (Turkey): (a) the bench-shaped seats were constructed of stone slabs with another vertical stone slab that covered the opening from the void between the floor and the seat and (b) the small channel (half-pipe-shaped cross-section) on the floor in front of the seat had a continuous flow of water for cleaning the sponghia (the toilet paper of the time). Copyright permission with LW Mays.
wastewater technology was accomplished throughout the centuries of the ancient Greek and Roman civilization. With a few exceptions, the basis for present-day progress in water transfer is clearly not a recent development, but an extension and refinement of the past. In fact, the surprising features are the similarity of ancient water methodologies with those of the present and the advanced level of water and wastewater management used by the ancients. Greek and Roman technological developments in water and wastewater management principles and practices as well as other achievements of those civilizations, such as poetry, philosophy, sciences, politics, and visual arts, are not known. To put in perspective the ancient water and wastewater achievements discussed in this chapter, it is important to examine their relevance to modern times and to harvest some lessons. The relevance of ancient hydraulic works should be examined in terms of the evolution of technology, the technological advances, homeland security, and management principles. The Romans, whose empire replaced the Greek rule in most part of this area, inherited the technologies and developed them further by changing their application scale from small to large and implementing them to almost every large city. The Greek and Roman water technologies are not only a cultural heritage but also the underpinning of modern achievements in water and wastewater engineering and management practices. Apparent characteristics of technologies and management practices in many ancient civilizations are durability and sustainability. Also, there have been integrated management practices, combining both large-scale and small-scale constructions and measures that have allowed cities to sustain for millennia. Currently, engineers use return period for the design of hydraulic structures as dictated by design standards and economic considerations. Sustainability, as a design principle, has
entered the engineering lexicon within the last decade. Naturally, it is difficult to estimate the design principles of ancient engineers but it is notable that several ancient works have operated for very long periods, some until recent times. Thus, wastewater and stormwater drainage systems were functioning in Bronze Age settlements and continued during the Greek and Roman periods. These include the construction and use of bathrooms and other sanitary and purgatory facilities, as well as wastewater and storm sewer systems. In fact, the hydraulic and architectural function of sewer systems in palaces and cities are regarded as one of the salient characteristics of Minoan civilization. They were so advanced that they can be justly compared with their modern counterparts. The durability of some of the constructions that operated up to present times, as well as the support of the technologies and their scientific background by written documents, enabled these technologies to pass to present societies despite regressions that have occurred through the centuries (i.e., in the Dark Ages). The development of science and engineering is not linear but often characterized by discontinuities and regressions. Bridges from the past to the future are always present, albeit oftentimes they are invisible to those who cross them! Thus, in addition to many ancient constructions that have been continuously or intermittently in operation to date, substantial information from ancient Greek and Roman written sources has also been preserved (Angelakis and Koutsoyiannis, 2003). Thus, the major achievements were accomplished during the Greek and Roman civilizations. As a result, they represent the state-of-the-art structures that were technically feasible at that time. For example, the aqueduct of ancient Samos, called ‘&mj´istomon’ or ‘bi-mouthed’ (thus pointing out that it was constructed from two openings), is an important hydraulic monument, indicating that it was possible in the ancient world to design and construct technologically advanced water transportation projects on a large scale.
Water and Wastewater Management Technologies in the Ancient Greek and Roman Civilizations
From the preceding synoptic discussion, certain conclusions might be suggested for further reflection and systematic investigation: 1. The water and wastewater hydraulics works in Minoan, Greek, and Roman civilizations are sometimes not too different from the modern practice, since present technologies descend directly from that time’s engineering. 2. Minoan, Greek, and Roman water and wastewater public works are characterized by simplicity, robustness of operation, and the absence of complex controls. 3. The meaning of sustainability in modern times should be reevaluated in light of Minoan, Greek, and Roman hydraulic works and water and wastewater management practices. 4. Technological developments based on sound engineering principles can have extended useful lives. 5. In areas of water shortage, development of a cost-effective and environmental friendly water resources management practice, based on Minoan, Greek, and Roman civilizations principles, is essential.
References Adam JP (1988) L’Arte di Costruire presso i Romani. Materiali e Tecniche (Roman Building: Materials and Techniques). Milan: Longanesi. Angelakis AN and Koutsoyiannis D (2003) Urban water resources management in ancient Greek times. In: Stewart BA and Howell T (eds.) Encyclopedia of Water Science, pp. 999--1007. New York: Dekker. Angelakis AN, Koutsoyiannis D, and Tchobanoglous G (2005) Urban wastewater and stormwater technologies in ancient Greece. Water Research 39: 210--220. Angelakis AN, Lyrintzis AG, and Spyridakis SV (2010) Urban water management in Minoan Crete, Greece. E-Water (in press). Angelakis AN, Savvakis YM, and Charalampakis G (2007) Aqueducts during the Minoan era. Water Science and Technology: Water Supply 7(1): 95--101. Angelakis AN and Spyridakis SV (1996) The status of water resources in Minoan times – a preliminary study. In: Angelakis A and Issar A (eds.) Diachronic Climatic Impacts on Water Resources with Emphasis on Mediterranean Region, pp. 161–191. Heidelberg: Springer. Angelakis AN and Spyridakis DS (2010). Water supply and wastewater management aspects in ancient Greece. Water Science and Technology: Water Supply 10(4) (in press). Antoniou G, Xarchakou R, and Angelakis AN (2006) Water cistern systems in Greece from Minoan to Hellenistic period. In: Angelakis AD and Koutsoyiannis D (eds.) Proceedings of 1st IWA International Symposium Water and Wastewater Technologies in Ancient Civilizations, pp. 457–462. National Agricultural Research Foundation, Iraklio, Greece, 28–30 October 2006. Antoniou GP (2010) Ancient Greek lavatories: Operation with reused water. In: Mays LW (ed.) Ancient Water Technology. Dordrecht: Springer. Antoniou GP and Angelakis AN (2009) Historical development bathrooms (toilets) and other sanitary and purgatory structures in Greece. In: Proceedings of 2nd IWA International Symposium on Water and Wastewater Technologies in Ancient Technologies. Bari, Italy, 28–29 May 2009. Bersani P, Canalini A, and Dragoni W (2010) First results of a study of the Etruscan tunnel and other hydraulic works on the ‘‘Ponte Coperto’’ stream (Cerveteri, Rome, Italy). Water Science and Technology: Water Supply 10(4) (in press). Bono P and Boni C (1996) Water supply of Rome in antiquity and today. Environmental Geology 27: 126--134. Cadogan G (2007) Water management in Minoan Crete, Greece: The two cisterns of one Middle Bronze Age settlement. Water, Science and Technology: Water Supply 7(1): 103--112. Crouch DP (1993) Water Management in Ancient Greek Cities. New York: Oxford University Press. Darcque P and Treuil R (eds.) (1990) The storm drains of the east wing at Knossos. Special Issue: L’habitat e´ge´en pre´historique. Bulletin de Correspondance Helle´nique, Supple´ment 19: 141–146. Davaras K (1976) Guide to Cretan Antiquities. Park Ridge, NJ: Noyes Press.
21
De Feo G, De Gisi S, Malvano C, and De Biase O (2010) The greatest water reservoirs in the ancient Roman world and the ‘‘Piscina Mirabilis’’ in Misenum. Water, Science and Technology: Water Supply 10(4) (in press). De Feo G, De Gisi S, Malvano C, et al. (2010) The Roman aqueduct and the Helvius’ Fountain in Sant’Egidio del Monte Albino, in Southern Italy: A historical and morphological approach. In: Proceedings of 2nd IWA International Symposium on Water and Wastewater Technologies in Ancient Technologies. Bari, Italy, 28–29 May 2009. De Feo G, Malvano C, De Gisi S, and De Biase O (2009b) The ancient aqueduct from Serino to Beneventum in Southern Italy: A technical and historical approach. In: Proceedings of 2nd IWA International Symposium on Water and Wastewater Technologies in Ancient Technologies. Bari, Italy, 28–29 May 2009. De Feo G and Napoli RMA (2007) Historical development of the Augustan aqueduct in Southern Italy: Twenty centuries of works from Serino to Naples. Water Science and Technology: Water Supply 7(1): 131--138. Evans SA (1921–1935) The Palace of Minos at Knossos: A Comparative Account of the Successive Stages of the Early Cretan Civilization as Illustrated by the Discoveries, vols. I–IV, London: Macmillan (reprinted by Biblo and Tannen, New York, USA, 1964). Fahlbusch H (2006) Water management in the classic civilization. In: Proceedings of La Ingenieria Y La Gestion Del Agua a Traves de Los Tiempos. Universidad de Alicante, Spain, with the Universidad Politechnica de Valencia, Alicante, Spain, 30 May–01 June 2006. Gorokhovich Y (2005) Abandonment of Minoan palaces on Crete in relation to the earthquake induced changes in groundwater supply. Journal of Archaeological Science 32: 217--222. Graham JW (1987) The Palaces of Crete. Princeton, NJ: Princeton University Press. Haberey W (1972) Die ro¨mischen Wasserleitungen nach Ko¨ln. Bonn: RheinlandVerlag. Hansen RD (2006) Water and wastewater systems in imperial Rome. http:// www.waterhistory.org (accessed February 2010). Hodge AT (2002) Roman Aqueducts & Water Supply, 2nd edn. London: Gerald Duckworth. Karakostantinou A (2006) The Roman Aqueduct of Moria, Lesvos. Volos, Greece: Department of Elementary Education, University of Thessaly (in Greek). Koutsoyiannis D, Mamassi N, and Tegos A (2007) Logical and illogical exegeses of hydrometeorological phenomena in ancient Greece. Water Science and Technology: Water Supply 7(1): 13--22. Koutsoyiannis D, Zarkadoulas N, Angelakis AN, and Tchobanoglous G (2008) Urban water management in ancient Greece: Legacies and lessons. ASCE, Journal of Water Resources Planning and Management 134(1): 45--54. Ku¨ltu¨r AS¸ (2008) The History of the Basilica Cistern. Istanbul, Turkey. http:// www.yerebatan.com/english/itarihce.html (accessed July 2010). Lendering J (2008) Constantinople (Istanbul): Basilica Cistern. Istanbul, Turkey. http://www.livius.org (accessed July 2010). MacDonald CF and Driessen JM (1988) The drainage system of the domestic quarter in the Palace at Knossos. British School of Athens 83: 235--358. Martini P and Drusiani R (2009) History of the water supply of Rome as a paradigm of water services development in Italic peninsula. In: Proceedings of 2nd IWA International Symposium on Water and Wastewater Technologies in Ancient Technologies. Bari, Italy, 28–29 May 2009. Mavromati E and Chryssaidis L (2007) Aqueducts in the Hellenic area during the Roman Period. Water Science and Technology: Water Supply 7(1): 139--145. Mays LW (2007) Ancient urban water supply systems in arid and semi-arid regions. In: Proceedings of International Symposium on New Directions in Urban Water Management. UNESCO, Paris, France, 12–14 September 2007. Korea Water Resources Association, http://www.kwra.or.kr (accessed February 2010). Mays LW (2008) A very brief history of hydraulic technology during antiquity. Environmental Fluid Mechanics 8(5): 471--484. Mays LW (ed.) (2010) Ancient Water Technologies. Dordrecht: Springer. Mays LW, Koutsoyiannis D, and Angelakis AN (2007) A brief history of urban water supply in antiquity. Water, Science and Technology: Water Supply 7(1): 1--12. Monteleone MC, Yeung H, and Smith R (2007) A review of ancient Roman water supply exploring techniques of pressure reduction. Water Science and Technology: Water Supply 7(1): 113--120. Myers JW, Myers EE, and Cadogan G (1992) The Aerial Atlas of Ancient Crete. Berkeley, CA: University of California Press. Oziz U (1987) Ancient water works in Anatolia. Water Resources Development 3(1): 55--62. Oziz U (1996) Historical water schemes in Turkey. Water Resources Development 12(3): 347--383. Panimolle G (1984) Gli Acquedotti di Roma Antica (The Aqueducts of Ancient Rome). Rome: Edizioni Abete.
22
Water and Wastewater Management Technologies in the Ancient Greek and Roman Civilizations
Rodgers RH (2004) Sextus Iulius Frontinus. On the Water-Management of the City of Rome. De Aquaeductu Urbis Romae. Cambridge: Cambridge University Press. Russo F and Russo F (2007) Pompei. La Tecnologia Dimenticata (Pompeii. The Forgotten Technology). Naples: ESA – Edizioni Scientifiche e Artistiche. Tassios TP (2007) Water supply of ancient Greek cities. Water Science and Technology: Water Supply 7(1): 165--191.
Tolle-Kastenbein R (2005) Archeologia dell’Acqua (Water Archaeology). Milan: Longanesi. Vuorinen HS (2010) Water, toilets and public health in the Roman era. Water Science and Technology: Water Supply 10(4) (in press). Vuorinen HS, Juuti PS, and Katko TS (2007) History of water and health from ancient civilizations to modern times. Water Science and Technology: Water Supply 7(1): 49--57.
4.02 Membrane Filtration in Water and Wastewater Treatment Y Watanabe and K Kimura, Hokkaido University, Sapporo, Japan & 2011 Elsevier B.V. All rights reserved.
Membrane Application to Water Purification Current Status Membrane Fouling Main foulant Affinity of main foulant for membranes Membrane Filtration Systems for Controlling Fouling Channel flocculation in monolith ceramic membrane Pre-coagulation/sedimentation in hollow-fiber UF/MF membrane Hybrid submerged MF membrane system PVDF Membrane filtration with pre-ozonation Membrane Application to Wastewater Treatment Current Status of MBRs Mechanism of Membrane Fouling Effect of membrane permeate flux on fouling Effect of membrane material on fouling Fouling potential of carbohydrate assessed by lectin affinity chromatography
4.02.1 Membrane Application to Water Purification 4.02.1.1 Current Status The mainstay of water purification technology in the twentieth century was sand filtration, but since the late 1980s, membrane filtration technology using RO/NF/UF/MF membranes has been applied to the water and wastewater treatment, desalination, and water reuse (RO, reverse osmosis; NF, nanofiltration; UF, ultrafiltration; MF, microfiltration).
3500 Start of RO research in USA (1953)
3000
Water / wastewater treatment (UF/MF)
President J.F.Kenedy approved RO desalination as a national project (1961)
2500 Cryptosporidium infection in Milwaukee (1993)
2000 Enhanced regulations of surface water in USA (1998)
1500 Enhanced water works law in Japan (2001)
1000
Brackish water desalination / wastewater reuse (NF / RO)
2005
2000
1995
1990
1980
1975
1970
1965
1960
0
1955
500
1950
Global accumulative amount of permeate (×104 m3 d−1)
23 23 23 24 30 36 36 40 43 45 47 47 48 49 54 57 60
Figure 1 shows the historical development of membrane technology in the water and wastewater treatment. Membrane filtration has small foot print, extremely high solid–liquid separation ability, and its maintenance is easy. Water purification plants in the United States, the Netherlands, France, Australia, and Japan have introduced the membrane filtration process. Figure 2 shows the recent increase in the amount of water produced by the membrane filtration, which includes water purification, desalination, and wastewater treatment.
1985
4.02.1 4.02.1.1 4.02.1.2 4.02.1.2.1 4.02.1.2.2 4.02.1.3 4.02.1.3.1 4.02.1.3.2 4.02.1.3.3 4.02.1.3.4 4.02.2 4.02.2.1 4.02.2.2 4.02.2.2.1 4.02.2.2.2 4.02.2.2.3 References
Sea water desalination (RO)
“If we could produce fresh water from salt water at a low cost that would indeed be a great service to humanity, and would dwarf any other scientific accomplishment” John F. Kennedy Figure 1 Development of membrane filtration. MF, microfiltration; NF, nanofiltration; RO, reverese osmosis; UF, ultrafiltration.
23
24
Membrane Filtration in Water and Wastewater Treatment Global amount of water produced by membrane processes 35 000 000 32 000 000 m3 d–1, 2006 SWRO
Amount of water (m3 d–1)
30 000 000
Increase by 25% each year
NF+BWRO 25 000 000
LP+MF+UF
20 000 000 15 000 000 10 000 000
2006
2005
2004
2003
2002
2001
2000
1999
1998
1997
1996
1995
1994
1993
1992
1991
0
1990
5 000 000
Figure 2 Increase in purified water by membrane filtration. BWRO, brackish water reverese osmosis; LP, low pressure; MF, microfiltration; NF, nanofiltration; SWRO, seawater reverese osmosis; UF, ultrafiltration.
Table 1
Large-scale water purification plants in world wide
Country
Place (plant name)
Capacity (103m3d1)
Construction year
Membrane
Water source
USA Canada Singapore USA USA USA Canada UK Germany USA
Minneapolis (Fridley Plant) Mississanga, Ontario Chestnut Minneapolis (Columbia Heights) Racine, Wisconsin Thornton, Colorado Kamloops, British Columbia Clay Lane Roetgen/Aachen San Joaquin, California
360 302 273 265 189 187.5 160 160 144 136
2011 (to be built) 2006 2003 2005 2005 2005 2005 2001 2005 2005
UF UF UF UF UF UF UF UF UF UF
Surface Lake Surface Surface Surface Surface Surface Ground Reservior Surface
Source: Japan Water Research Center, Hot News in water works, No. 56.
Table 1 shows the large-scale water purification plants using membrane filtration. All plants in the table use the UF membrane but a plant using monolith ceramic MF membrane with the capacity of 173 000 m3 d1 is under construction in Japan. There has been a significant progress in the development of new robust MF membranes with new polymers such as PVDE and FTFE for water and wastewater treatment. Combining robust MF membranes and the other processes such as coagulation, ozonation, biological/chemical oxidation, and powdered activated carbon adsorption and chemically enhanced physical cleaning makes very efficient water purification system. They are very effective in the application to the large-scale water purification plant. The trend toward membrane filtration is expected to spread worldwide during this century. However, there are several limiting factors applying the UF membrane and MF membrane to the water purification. Among them, fouling in membrane is a major obstacle to widespread use of this technology. The authors have been studying the mechanism and control of membrane fouling in water treatment. This chapter
summarizes the authors’ research on membrane application to the water purification.
4.02.1.2 Membrane Fouling Several physical membrane cleaning methods such as hydraulic backwashing and air scrubbing have been developed and used routinely in many existing membrane plants to minimize membrane fouling. Despite routine physical membrane cleaning, membrane filtration resistance gradually increases over a long period of operation, indicating that membrane fouling cannot be completely controlled by physical cleaning. Fouling that cannot be controlled by physical cleaning is defined here as physically irreversible fouling. Control of physically irreversible fouling is important for the reduction of operation cost in a membrane process because this type of fouling develops even when a very efficient physical cleaning is carried out. Physically irreversible membrane fouling can only be canceled by chemical cleaning. However, chemical cleaning of the membrane should be limited to a minimum frequency because repeated chemical
Membrane Filtration in Water and Wastewater Treatment
cleaning may shorten the membrane lifetime and disposal of spent chemical reagents poses another problem. Membrane fouling strongly depends upon the structure of membrane (average size, size distribution, and density of pores). Surface morphology and roughness are surely involved in it. However, this chapter describes the effect of only nominal pore size and materials of membrane on the membrane fouling.
4.02.1.2.1 Main foulant In a number of previous studies on fouling of membranes used for water treatment, natural organic matter (NOM), composed of a variety of nonbiodegradable organic compounds including humic substances, has been shown to be the major constituent causing membrane fouling. However, it is still not clear which fraction of NOM causes membrane fouling. In early works, hydrophobic fractions of NOM, such as humic substances, were considered to be the major foulants. Hydrophobic interaction and electrostatic interaction were the explanations for the binding between hydrophobic NOM and membranes. More recently, hydrophilic NOM with features of carbohydrate or protein has been reported by several researchers to be the major foulant. As explanations for the binding between hydrophilic NOM and membranes, van der Waals attraction and hydrophobic interaction between membranes and hydrophobic domains in hydrophilic NOM have been suggested. In addition to NOM, metals and metal– NOM complexes have been reported as the constituents affecting membrane fouling (Yamamura et al., 2007a, 2007b). Physically reversible fouling and physically irreversible fouling have not been distinguished in many previous studies. In addition, many previous studies were based on short-term experiments, which are not sufficient for observing physically irreversible fouling. As a result, knowledge of physically irreversible fouling occurring in membrane filtration in drinking water treatment is very limited; therefore, further studies need to be carried out with special emphasis on physically irreversible fouling for more efficient use of membranes. In particular, investigation of the characteristics of components that cause physically irreversible fouling would be useful for the establishment of a new protocol of fouling control. In this study, three MF/UF membranes that had been fouled in long-term filtration of surface water used as a drinking water source were investigated in terms of the recovery of water permeability by chemical cleaning and the characteristics of the foulant causing physically irreversible fouling. Based on the results obtained from various analyses, a hypothesis regarding the evolution of physically irreversible fouling is proposed. Three different hollow-fiber membranes were used in this study. Two of them were MF membranes and the other was a UF membrane. The two MF membranes had the same nominal pore size of 0.1 mm but were made from different polymers such as polyethylene (PE; Mitsubishi Rayon, Tokyo, Japan) and polyvinylidene fluoride (PVDF; Asahikasei Chemicals, Tokyo, Japan). The UF membrane had a molecular weight cut-off of 100 000 Da and was made from polyacrylonitrile (PAN; Toray Industries, Tokyo, Japan). Using these three different membranes, pilot-scale membrane
25
filtration tests were carried out in parallel using the Chitose River surface water. This river flows through peat area and its surface water contains many humic substances. The concentration range of total iron and aluminum was 0.7–1.7 and 0.05 and 0.7 mg l1. About 75% of them were larger than 0.45 mm. The PVDF and the PE membranes were submerged in separate tanks and were operated under vacuum. The PAN membrane was housed in a vessel and was operated under pressure. All membranes were operated in the outside-in flow mode. The three membranes were operated with identical run cycles (filtration: 30 min; air scrubbing: 30 s; hydraulic backwashing: 60 s) at the same constant flux of 0.65 m3 m2 d1. Hydraulic backwashing was not accompanied by the addition of chlorine. When membrane fouling became significant in the submerged MF membranes despite the implementation of periodical backwashing, membrane modules were taken out from the tanks and were cleaned by spraying pressurized water on the membrane surface. The average quality of the feed water and that of membrane permeates are shown in Table 2. In the feed water, large portions of aluminum (78%) and iron (75%) were present as suspended solids (40.45 mm), while manganese, calcium, and organic matter were mainly present in dissolved forms. Aluminum and iron were effectively removed by the tested membranes due to the strict solid–liquid separation. On the other hand, removal of manganese, calcium, and organic matter was not significant in any of the membranes. This implies that the sizes of manganese, calcium, and dissolved organic carbon (DOC) were smaller than the pore sizes of the tested membranes. The UF membrane showed slightly higher rates of removal of DOC and UV absorbance than those of the two MF membranes, reflecting the difference between membrane pore sizes of the MF and UF membranes. However, the concentration of aluminum in the PAN membrane was slightly higher than the concentrations in the MF membranes. No reasonable explanation for this is available at present. Figure 3 shows the changes in transmembrane pressure (TMP) in the three membranes. The rates of increase in TMP in the three membranes were considerably different. As expected, the tightest membrane (PAN) showed the highest rate of increase in TMP. The rates of increase in the two MF membranes were different despite the fact that they had the same nominal pore size. This clearly indicates that the materials of the membrane have a substantial influence on the
Table 2
Average raw water quality during experiment
Temperature (1C) pH Turbidity (NTU) UV absorbance at 220 nm (cm1) UV absorbance at 260 nm (cm1) TOC (mg 11) DOC (mg 11) THMFP (mg 11) Manganese (mg 11) Soluble manganese (mg 11) Ammonia Nitrogen (mg 11)
11.5 7.11 16.54 0.411 0.099 2.43 2.29 0.086 0.100 0.074 0.22
DOC, dissolved organic matter; THMFP, trihalomethane formation potential; TOC, total organic carbon.
26
Membrane Filtration in Water and Wastewater Treatment 200
PAN
Time of additional physical cleaning
PVDF PE
TMP (kPa)
160
120
80
40
0
0
10
20
30
40
50
Operation time (days) Figure 3 Time course changes in transmembrane pressure (TMP) difference adjusted to 20 1C equivalent value considering the change in water viscosity. PAN, polyacrylonitrile; PE, polyethylene; PVDF, polyvinylidene fluoride; TMP, transmembrane pressure.
evolution of membrane fouling. Interestingly, the results obtained in this study showing that the PE membrane was less fouled than the PVDF membrane are opposite to the results of a previous study focusing on membrane fouling in membrane bioreactors (MBRs) used for municipal wastewater treatment. This implies that characteristics of foulants in the case of drinking water treatment were different from those in the case of wastewater treatment. Further investigation is needed to understand the influence of membrane material on the rate of fouling. In all of the tested membranes, increase in TMP was not constant and rapid increases in TMP were seen several times. After the rapid increases in TMP, however, the value of TMP gradually declined due to the periodical backwashing except for the case of the PVDF membrane. On days 31 and 41, an additional physical cleaning (spraying pressurized water on the membrane surface) was needed to maintain the permeability of the PVDF membrane. This additional physical cleaning worked well and substantial reduction in TMP in the PVDF membrane was seen after cleaning. Chemical cleaning was not carried out at that time. Based on the observations mentioned above, it is assumed that the rapid increases in TMP shown in Figure 3 were caused by the accumulation of cake on the surfaces of the membranes. The three dashed lines shown in the figure are assumed to represent the evolution of physically irreversible fouling in the three membranes, which accumulated and remained despite of the implementation of periodical backwashing and additional physical cleaning. As seen in Figure 3, the rates of occurrence of physically irreversible fouling in the three membranes were different. To investigate the features of constituents that were responsible for physically irreversible fouling, the foulants were desorbed from the fouled membranes at the termination of the operation and then their chemical characteristics were analyzed. When the pilot operations were terminated, fouled membranes were taken out from the filtration units. The
membrane fibers were immediately brought to the laboratory in a container filled with distilled water. First, each membrane fiber was manually wiped with a sponge and thoroughly rinsed with distilled water, which was carried out to minimize the influence of the accumulated cake causing physically reversible fouling in subsequent tests. By visual inspection, no accumulated cake was found on the membrane after wiping with a sponge. Using the wiped membranes, tiny membrane modules of 40 cm2 in membrane area were assembled and pure water permeability of the fouled membrane was measured by applying 30 kPa of pressure difference. Filtration was continued until a constant permeate flow rate was achieved (typically in 15 min). After measuring the pure water permeability, tiny membrane modules were soaked in various chemical solutions at 20 1C for 24 h. The chemical solutions used for cleaning were Milli-Q water, NaClO (700 ppm as free available chlorine), NaCl (0.1 M), NaOH (pH 12), HCl (pH 2), ethylenediaminetetraacetic acid(EDTA) (20 mM), and oxalic acid (0.5%). Recoveries in pure water permeability by the chemical cleaning were evaluated and the chemical solutions containing the foulants desorbed from the membranes were analyzed. Membrane specimens that were not used for assembling the tiny membrane modules were divided into two portions and were soaked in a solution of sodium hydroxide at pH 12 or hydrochloric acid at pH 2. Because a large amount of membrane specimens was available in this study, this process enabled extraction of a sufficient amount of organic matter for advanced analysis (e.g., Fourier transform infrared (FTIR) and nuclear magnetic resonance (NMR) spectra). Figure 4 shows the degree of restoration of the fouled membranes in terms of pure water flux by chemical cleaning with various reagents. In this figure, the ratio of pure water flux after chemical cleaning (J1) to the flux before chemical cleaning (J0) is used to express the degree of flux restoration. As described earlier, chemical cleaning was carried out after
Membrane Filtration in Water and Wastewater Treatment PVDF
PE 7.0
NaCIO Oxalic
PAN
NaCIO
NaCIO
Oxalic
Oxalic
HCl
HCl
HCl
EDTA
EDTA
EDTA
NaOH
NaOH
NaOH
NaCl
NaCl
NaCl
MQ
MQ
MQ
1
2 J1/ J0
3
1
27
2 J1/ J0
3
1
2 J1/ J0
3
Figure 4 Effect of chemical membrane cleaning (J0: pure water flux before chemical cleaning, J1: pure water flux after chemical cleaning). EDTA, ethylenediaminetetraacetic acid; MQ, milli-Q water; PAN, polyacrylonitrile; PE, polyethylene; PVDF, polyvinylidene fluoride.
manually removing reversible cake that had accumulated on the membrane. Therefore, it can be considered that the restoration shown in Figure 4 represents removal of the foulants causing physically irreversible membrane fouling. Actually, manual sponge cleaning carried out prior to chemical cleaning had little effect on the permeability of the fouled membranes, indicating that fouling seen at the termination of the longterm operation could be attributed mainly to physically irreversible fouling. As seen in the figure, in the case of the PVDF and PAN membranes, NaCl (0.1 M) and EDTA (20 mM) were not effective in mitigation of physically irreversible fouling in this study. Figure 4 also shows that alkaline solution (NaOH) was more efficient than acid solutions (oxalic acid and HCl) for recovery of permeability of the PVDF and PAN membranes. The oxidizing agent (NaClO) exhibited the best cleaning performance in recovery of permeability of the PVDF and PAN membranes. This implies that organic matter was mainly responsible for the evolution of physically irreversible membrane fouling in the PVDF and PAN membranes. In contrast, in the case of the PE membrane, which exhibited the least membrane fouling in the continuous run (Figure 3), the degree of recovery of water permeability following cleaning with acid, alkaline, and oxidizing reagents were comparable. This suggests that the contribution of metals to the physically irreversible fouling in the PE membrane was significant. Desorption of membrane foulants was carried out at the termination of the pilot operation. As stated above, to ensure that physically reversible cake was removed from the membrane surface, each membrane fiber was carefully wiped with a sponge prior to desorption tests. Although both aluminum and iron in the raw water were effectively removed by the membranes tested, only iron was desorbed from the fouled membranes at a significant amount. This suggests that aluminum in the feed water was rejected or deposited on the membrane surface and subsequently removed by the periodical backwashing. In contrast, iron was likely to cause the physically irreversible fouling to some extent. In the cleaning with HCl solution, not only metals but also organic matter were desorbed from the fouled membranes, particularly from the PVDF membrane. Figure 5 shows the FTIR spectra of the foulants desorbed from the fouled membranes by HCl solution. Interestingly, there were significant similarities among the three spectra. All of
1080
HCI-PVDF HCI-PE HCI-PAN 2000
1800
1600
1400
1200
1000
800
Wave number (cm–1) Figure 5 FTIR spectra of membrane foulant desorbed with HCl (pH 2) solution. PAN, polyacrylonitrile; PE, polyethylene; PVDF, polyvinylidene fluoride.
the spectra had a dominant peak near 1080 cm1, which is an indication of their carbohydrate character. Therefore, the carbohydrate-like organic matter was thought to be the main constituent in the foulants desorbed with HCl solution regardless of membrane type. In a study by Kabsch-Korbutowicz et al., it was shown that a large portion of organic matter desorbed from the fouled membrane by acid or chelating agents formed complexes with metals. Similarly, in the present study, the carbohydrate-like organic matter and metals (mainly iron) desorbed with HCl solution were assumed to form complexes and cause physically irreversible fouling. It has been reported that carbohydrate can form a complex with iron. As previously mentioned, NaOH solution restored the membrane permeability to a larger extent and desorbed a larger amount of organic matter from the fouled membranes than did HCl solution. Therefore, analysis of the foulants desorbed from the membrane with NaOH solution would be more useful in understanding the fouling, compared to the case of HCl solution. The value of specific ultraviolet absorbance (SUVA) is considered to be a surrogate measurement of aromacity of organic matter, and a high SUVA value corresponds to organic matter consisting of a large amount of double-bond or aromatic structures. The values of SUVA determined for the foulants desorbed by NaOH solution were much lower than those for the feed water on average. This
28
Membrane Filtration in Water and Wastewater Treatment 1080 1660
1390 1540 PVDF
NaOH-PVDF NaOH-PE
PE
NaOH-PAN 2000
1800
1600
1400
1200
1000
800
Wave number (cm–1)
PAN
Figure 6 Infrared spectra of membrane foulant desorbed with NaOH (pH 12) solution. PAN, polyacrylonitrile; PE, polyethylene; PVDF, polyvinylidene fluoride.
200
implies that a relatively hydrophilic fraction of the organic matter in the feed water was responsible for the physically irreversible fouling. Interestingly, the value of SUVA determined for the foulants was similar among the foulants desorbed from the three membranes. This indicates that the characteristics of the foulants desorbed from the three membranes might be similar, but this turned out to be false as discussed later. FTIR spectra of the foulants desorbed with NaOH solution from the three membranes are presented in Figure 6. There were significant similarities in the spectra obtained for the three membranes. In these spectra, peaks near 1660 and 1540 cm1 were significant. They are assigned to amido-I and II bands, respectively. In all spectra, a broad peak near 1080 cm1 was seen. This peak is an indicator of carbohydrate character. FTIR spectra shown in Figure 6 are not similar to those of humic substances. This suggests that humic substances were relatively minor components in the foulant responsible for the physically irreversible fouling. CPMAS 13C NMR spectra of the foulants desorbed with NaOH solution from the membranes are presented in Figure 7. A general similarity among the foulants desorbed from the three membranes was found in NMR analysis as well. Although a proteinaceous nature of the foulants in the membranes can be seen by peaks near 175 and 55 ppm, carbohydrate (peak at 75 ppm) was dominant in the foulant regardless of the membrane type. The aromatic carbon signal (110–165 ppm) was minor in the spectra for the two MF membranes (PVDF and PE) but was pronounced in the spectrum for the PAN membrane. This indicates that the contribution of the humic fraction of NOM to the evolution of physically irreversible fouling was more significant in the PAN membrane than in the two MF membranes. The humic fraction would be smaller than carbohydrate, as shown later. Thus, it is reasonable to assume that the contribution of the small humic fraction would become more significant in a UF membrane (PAN in this case) than in MF membrane (PVDF and PE in this case). The amount of calcium desorbed with NaOH solution was significant in the case of the PAN membrane. This calcium might have formed a complex with humic substance as suggested by several researchers. Nevertheless,
160
120 80 Chemical shift (ppm)
40
0
Figure 7 CPMAS 13C NMR spectra of membrane foulants desorbed with NaOH (pH 12) solution. PAN, polyacrylonitrile; PE, polyethylene; PVDF, polyvinylidene fluoride.
carbohydrate was dominant in the foulant desorbed form the PAN membrane as well, as shown in Figure 5. As shown above, both FTIR and NMR analyses demonstrated that carbohydrate was a dominant component causing physically irreversible fouling regardless of the type of membrane. Carbohydrate has, however, a hydrophilic nature, and hydrophobic interaction between the membranes and carbohydrate is therefore not a reasonable explanation for the participation of carbohydrate in physically irreversible fouling. To elucidate the fouling mechanisms involved in the continuous operation, changes in rejection rate of both humic acid and carbohydrate in the operation were investigated using HPLC-SEC with UV/DOC detectors. Figure 8 shows the representative molecular weight distribution of organic matter contained in the feed water used in this study. As seen in the figure, organic matter contained in the feed water could be roughly divided into two fractions: large molecules with a hydrophilic nature (little UV absorbance) and small molecules with a hydrophobic nature (high UV absorbance). A similar molecular weight distribution of organic matter was found in previous studies. It is thought that large molecules mainly consisted of carbohydrate, while small molecules mainly consisted of humic acid. Figure 9 shows changes in the removal of the large and small molecules by the three membranes determined by HPLC-SEC with UV/DOC detectors. In the case of the PVDF membrane, about 15% of the fraction of smaller organic molecules mainly composed of humic substances was initially removed. As the operation period became longer, however, the rate of removal of the small organic molecules declined and eventually no removal of small molecules was achieved by the PVDF membrane. The size of the small molecules should be considerably smaller than the nominal pore size of the PVDF membrane (0.1 mm); therefore, the sieving effect was discounted as an explanation for the initial removal of small organic molecules by the PVDF membrane. Rather, the initial
Membrane Filtration in Water and Wastewater Treatment
29
Humic acid DOC
900 600
Colloidal carbohydrate or protein
300 0
0 UV
0.01
0.02 107 106 105 104 103 102 Molecular weight (Da)
UV absorbance (cm–1)
DOC (ppb)
1200
10
Figure 8 Molecular size distribution of dissolved organic matter in the Chitose river surface water. DOC, dissolved organic carbon.
removal of the small molecules can be attributed to adsorption on/in the PVDF membrane. In contrast to the small molecules, the rate of removal of the large organic molecules by the PVDF membrane gradually increased during the operation. When the removal of the small molecules declined to a negligible level, the removal of large organic molecules increased by almost 100%. A similar trend was also seen for the other two membranes. Based on these observations, the following hypothesis regarding the evolution of physically irreversible fouling is presented. First, small molecules mainly composed of humic substances are adsorbed on/in membranes by hydrophobic interaction. As a result of adsorption of the small molecules, the sizes of membrane pores decrease and it becomes possible for large molecules mainly composed of carbohydrates to plug the pores and cause physically irreversible fouling. Also, adsorbed humic substances could work as glue for carbohydrates and facilitate the capture of carbohydrates on/in membranes. The examined PVDF was assumed to be more hydrophobic than the PE membrane because hydrophilic modification was provided for the PE membrane by the manufacture. It is likely that the hydrophobic PVDF membrane adsorbed humic substances more rapidly than did the hydrophilic PE membrane. As a result, the PVDF membrane should achieve complete rejection of carbohydrates earlier than the PE membrane (Figure 9). In discussion made above, it is assumed that foulant causing physically irreversible fouling originated from the feed water. Another possible origin of the foulant might be biofilms that cannot be removed by backwashing. It was reported that both carbohydrate and humics were excreted by microorganisms. Although the possibility that excretion from biofilms was the main source of the foulant which cannot be completely eliminated, it would be discounted by the following reasons: (1) evolution of reversible fouling (indication of biofilm formation) did not always dominate in the operation of the membranes as shown in Figure 3; (2) occasional increases in physically reversible fouling shown in Figure 3
could be explained by increases in turbidity in the feed (data not shown); and (3) water temperature was low (i.e., 5–10 1C) in the operation. To deal with the issues discussed above more precisely, establishment of the methods that can distinguish the origin of organic matter is indispensable. The following points were derived from the measurement of the zeta potentials of membranes before and after the longterm operation. The decrease in rejection of small molecules during the operation might be attributable to a decrease in favorable electrostatic interaction (repulsion) since the zeta potential of the tested membranes became slightly less negative after operation as a consequence of carbohydrate deposition. In this study, it was assumed that the decrease in favorable electrostatic interaction was not the main reason for the decrease in rejection of small molecules both because of the initial zeta potential that was close to neutral and because of the small changes in the zeta potentials after use. However, further investigation is needed to determine the influence of surface conditions of membranes on binding of NOM to membranes. To confirm the experimental results showing that carbohydrate-like substances are main substances causing the physically irreversible fouling, the authors carried out the bench-scale study where the surface water samples taken from four different sources such as Toyohira River (central Hokkaido), Kusiro River (eastern Hokkaido), Inba Lake (Chiba prefecture), and Yodo River (Osaka prefecture). Toyohira River water (total organic carbon, TOC ¼ 0.8 mg l1) is relatively clean. Kushiro river water (TOC ¼ 0.9 mg l1) is rich in humic substances. Inba Lake water (TOC ¼ 5.7 mg l1) is polluted and eutrophicated by the domestic wastewater. Yodo River water (TOC ¼ 1.8 mg l1) contained a lot of treated wastewater. Tiny membrane module with the surface area of 1.44 103 m2 was prepared with hollow-fiber membranes made of PVDF. The pore size of membranes was 0.1 mm. Membrane filtration was carried out by a peristaltic pump, and the constant-flow-rate mode of operation was applied. Permeate flux was fixed at 1.5 m d1 for all filtration experiments.
30
Membrane Filtration in Water and Wastewater Treatment PVDF 30
80 60
Large molecules
25
Small molecules
20 15
40
10
20
5
0
Removal rate of small molecules (%)
Removal rate of large molecules (%)
100
0 0
10
30 20 40 Operation time (days)
50
PE Removal rate of large molecules (%)
25
Large molecules
80
20
Small molecules
60
15 40
10
20
5
0
Removal rate of small molecules (%)
30
100
0 0
10
20 30 40 Operation time (days)
50
PAN 30
80 60
Large molecules
25
Small molecules
20 15
40
10
20 0
5 0
10
20 30 40 Operation time (days)
50
Removal rate of small molecules (%)
Removal rate of large molecules (%)
100
0
Figure 9 Changes in removal rate of large molecules (carbohydrate) and small molecules (humic acid). PAN, polyacrylonitrile; PE, polyethylene; PVDF, polyvinylidene fluoride.
Hydraulic backwashing was performed every 15 min. The duration and pressure of backwashing were 30 sec and 50 kPa, respectively. The organic matter in the four water sources was concentrated using RO (Nanomax 95, Millipore), and its recovery, defined as (DOC mass after concentration by RO)/(DOC mass before concentration), was 0.95, 0.81, 0.80, and 0.91 for Toyohira River, Inba Lake, Kushiro River, and Yodo River, respectively. Fractionation of organic matter contained in the isolates was carried out using the procedure described by Croue et al. They used the DAX-8 and XAD-4 resins. The portion that passed through both the DAX-8 and XAD-4 column was denoted the hydrophilic (HPI) fraction. The portion that retained on DAX-8 resin was denoted the hydrophobic (HPO) fraction. The portion that retained on XAD-84 was denoted the transphilic (TPI) fractions. The HPO and TPI fractions were eluted by backwashing with 2 l of 0.1 N NaOH at 100 ml min1. Each of the three fractions was desalinated by the electric dialysis until its electric conductivity became less
than 0.5 mS O1. The HPI and HPO fractions were diluted to a concentration of 2.0 mg TOC l1 with Milli-Q water and used as the feed water for the bench scale experiment. Figure 10 shows the FTIR spectra of the organic matter in the hydrophobic and hydrophilic fractions of the water from each of the four sources. FTIR analysis is a powerful tool for identifying the functional groups in organic matter and, together with the SUVA, provides useful information about the characteristics of organic matter in the feed waters. As seen in the spectra of the HPO fractions, the organic matter in the hydrophobic fraction was highly aromatic. For all the spectra of HPO fraction, a general similarly was seen in two broad peaks around 1400 and 1620–1660 cm1. These peaks are an indication of their aromatic character. The HPO fractions also seemed to contain alkyl aromatic sulfonates, as evidenced by the peaks of an aromatic sulfonic acid group (1035 and 1009 cm1) and the alkyl group (2930 cm1). In the spectra of the HPI fractions, on the other hand, a high peak at 1080 cm1 is seen for all the sources. This peak is assigned to C–O stretching of polysaccharide or aliphatic alcohol, which represent the carbohydrate-rich nature of HPI organic matters. The spectra of the HPI fractions of Inba Lake water and Yodo River water not only show the signature of carbohydrate-like substances but also have sharp peaks at 1620 and 1660 cm1 corresponding to carboxylic acid. These peaks, in combination with the peak at 1080 cm1, might indicate the presence of alginate-like substances in the feed water. The changes in TMP during filtration through the MF membrane made of PVDF differed between the HPO fraction and the HPI fraction are shown in Figure 11. Regardless of the NOM source, the TMP for the HPO fraction increased by less than 7 kPa and the TMP for the HPI fraction increased by more than 30 kPa. This clearly indicates that the HPI fraction of NOM is a major component affecting the development of physically irreversible fouling. The major differences in the characteristics of organic matter between the HPO fraction and the HPI fraction are in aromaticity and size. The organic matter in the HPO fraction consisted mainly of aromatic humic substances less than 6000 Da in size, while the HPI fraction was rich in carbohydrate-like substances having sizes between 100 000 and 1000 000 Da. These findings indicate that the development of physically irreversible fouling was caused not by aromatic humic substances but by carbohydrate-like substances. In authors’ study investigating the affinity between NOM and membrane surfaces, it was concluded that the physico-chemical interaction with the surface of membrane was more significant for carbohydrate-like substances (with hydroxyl groups) than for humic-like substances (with carboxyl groups). As a consequence, it can be hypothesized that large carbohydrate-like substances can accumulate on the membrane surface, interact with it strongly, and thereby cause physically irreversible fouling. Although some researchers suggested that physically reversible fouling is largely due to the HPO fraction of NOM, the development of physically reversible fouling was not obvious for the HPO feed waters, probably because the organic particles in the HPO fraction are smaller than the membrane pores in this study. Rather, some of the HPI fractions were found to contribute to the physically reversible fouling as well.
Membrane Filtration in Water and Wastewater Treatment Hydrophobic
31
Hydrophilic
(a)
(e)
(b)
(f)
(c)
(g)
(d)
(h)
4000 3500 3000 2500 2000 1500 1000
4000 3500 3000 2500 2000 1500 1000
Wave number (cm–1)
Wave number (cm–1)
Figure 10 Fourier transform infrared (FTIR) spectra of the natural organic matter (NOM) in hydrophobic (HPO) and hydrophilic (HPI) fractions of raw water from different sources: HPO fraction of water from (a) Toyohira river, (b) Lake Inbanuma, (c) Kushiro river, and (d) Yodo river; HPI fraction of water from (e) Toyohira river, (f) Lake Inbanuma, (g) Kushiro river, and (h) Yodo river.
Fraction contributing to memberane fouling (TOC = 2 mg l–1)
Toyohira RW (HPI)
Transmembrane pressure (kPa)
60 Ca: 0 mg l
Kushiro RW (HPI)
Inba Lake W (HPI)
–1
Yodo RW (HPI)
60
60
50
50
50
40
40
40
40
30
30
30
30
20
20
20
20
10
10
10
10
0
0
50
0
200
400
600
800
0
200
400
600
0 0
800
Toyohira RW (HPO)
Inba Lake W (HPO)
60
400
600
800
0
Operation time (min)
Operation time (min)
Operation time (min)
200
200
400
600
800
Operation time (min)
Kushiro RW (HPO)
Yodo RW (HPO)
60
60
60
Transmembrane pressure (kPa)
–1
Ca: 0 mg l
50
50
50
40
40
40
40
30
30
30
30
20
20
20
20
10
10
10
10
0
0
0
50
0
200
400
600
Operation time (min)
800
0
200
400
600
Operation time (min)
800
0 0
200
400
600
800
Operation time (min)
HPI fraction of NOM contributed to the physically irreversible fouling regardless of the types of NOM sources Figure 11 Changes in transmembrane pressure (TMP) during filtration.
0
200
400
600
Operation time (min)
800
32
Membrane Filtration in Water and Wastewater Treatment
In particular, the filtration of the HPI fractions of Lake Inbanuma water and Yodo River water induced the evolution of the physically reversible fouling to a large extent. These HPI fractions were found to contain a large amount of macromolecular polysaccharides with a negative charge at neutral pH, in which the electrostatic repulsion between the negatively charged polysaccharide and accumulated polysaccharides or membrane surface would occur. Such an electrostatic repulsion would help to weaken the binding of organic molecules to each other and thereby enable the accumulated organic matter to be easily removed by physical cleaning.
4.02.1.2.2 Affinity of main foulant for membranes In our previous study on pilot-scale filtration using hydrophilic and hydrophobic membranes, NMR analysis of the foulant demonstrated significant contribution of carbohydrate-like substances to the evolution of fouling. It was also shown that the nature of membrane materials affected the rate of accumulation of carbohydrate-like substances. However, the reason for the preferential binding of carbohydrate-like substances to membranes remains unclear. Elucidation of the physicochemical interactions between membranes and carbohydratelike substances is needed for understanding the mechanism of fouling involving carbohydrate. Several research groups have already demonstrated the usefulness of atomic force microscopy (AFM) force measurement for the quantification of the affinity between a carboxylmodified microsphere and the surfaces of NF/RO membranes. Carboxyl-modified microspheres were used as a surrogate of humic substances in their studies. Taking into account the hydroxyl-rich characteristics of carbohydrate, AFM force measurement using hydroxyl-modified microspheres and membranes would provide useful information about the affinity of carbohydrate-like substances to membranes, which has been reported in recent studies on fouling as reviewed above. Two MF membranes with the same nominal pore size of 0.1 mm were used in this study. One membrane was made of PE (Mitsubishi Rayon Engineering, Tokyo, Japan) and the other was made of PVDF (Asahi Kasei Chemicals, Tokyo, Japan). These two membranes were chosen because they are now used in many full-scale plants. Prior to the AFM force measurement, new membranes were filtered with Milli-Q water for 6 h so as to wash out impurities remaining on the membrane surface. Because of hydroxyl-rich nature of carbohydrate, Polybeads-hydroxylate microspheres (Cosmo Bio, Tokyo, Japan) were used as surrogates for carbohydrate-like substances. For comparison, Polybeads-carboxylate microspheres (Cosmo Bio, Tokyo, Japan) were also used in the AFM force measurement. In previous studies, carboxyl-modified microspheres were used as surrogates of humic substances. Both microspheres used in this study were made of polystyrene (3 mm). The characteristics of these microspheres are shown in Yamamura et al. (2008). The colloidal probes used in the AFM force measurement were prepared by attaching the microspheres to the top of a silicon nitride tip (NP-S: Veeco Instruments Inc., New York, USA) as previously described (Figure 12). Attachment of the microspheres to the cantilever tip was carried out with a micromanipulator with the aid of a
Figure 12 Scanning electron microscope image of a polystyrene bead (3 mm) glued to the top of a cantilever tip.
scanning electron microscope (TINY SEM, Technex Lab, Tokyo, Japan). After preparation, the colloidal probes were stored in a refrigerator (4 1C) prior to use. The spring constants of carboxyl- and hydroxyl-colloidal probes determined by thermal fluctuation method were 84 and 92 pN nm1, respectively. These values were used for converting cantilever deflections to loading forces. An atomic force microscope (MFP-3D, Asylum Research, Santa Barbara, CA) was used for the force measurements. Measurements were carried out in buffered water (1.0 mM NaHCO3, pH 6.8) with a trigger point of 50 nm. Divalent cations such as calcium or magnesium were not added to the buffered solution so as to prevent the formation of a bridging between polystyrene of microspheres and the membrane surface. Taking the heterogeneities of local membrane surfaces into account, measurements of force curves were made at three different locations. At each location, more than five force curves were obtained. All force curves obtained by the AFM force measurement were originally expressed as a function of force determined on the basis of the scanner position in the AFM instrument. The scanner position was converted to the separation distance by determining the onset of constant compliance between the scanner position and cantilever deflection (i.e., where cantilever deflection becomes a linear function of piezo-scanner position) and subtracting this value from all other scanner position values. In AFM force curves, the separation distance at which the interaction became either repulsive or attractive was identified as the point where the measured force is either positive or negative, respectively. At separation distances greater than this value, no force was considered to be acting on the colloidal probe and the zero force region of the plot was determined. An AFM force measurement gave two force curves: an approaching force curve and a retraction force curve. The affinity of the colloidal probe to the surface of the membrane was expressed by the adhesion force, Fad, which is defined as the force needed to separate the two from contact. Fad is determined on the basis of the maximum value of cantilever deflection in a retraction force curve (dmax) as shown in
Membrane Filtration in Water and Wastewater Treatment
33
0
Adhesion force (nN)
Deflection (nm)
12 –100 Maximium deflection –200 dmax
10 8 6 4 2
–300
0 PE (a)
0.0
0.2 0.4 0.6 0.8 Separation distance (μm)
1.0
PE
PVDF
PVDF
(b)
Figure 13 A representative retraction force curve.
Figure 14 Adhesion forces of (a) carboxyl-modified and (b) hydroxylmodified microspheres to polyethylene (PE) and polyvinylidene fluoride (PVDF) membrane surfaces in buffered solution (pH 6.8).
Figure 13. In contrast, interaction between the colloidal probe and the membrane surface when the probe was approaching the membrane surface (similar to the situation in which carbohydrate-like substances approach membranes by convection flow) was also assessed by the effective distance of the forces shown in an approaching force curve. The affinity between a carbohydrate-like substance and membrane surface would change as a result of fouling. Therefore, AFM force measurement was also carried out with membranes previously fouled in a pilot operation to investigate the change in affinity. Because of the difficulty in regular sampling of membrane specimens from the PVDF membrane module used in the pilot study, the investigation of change in affinity of the carbohydrate-like substance was carried out only with the PE membrane. Pilot-scale membrane filtration was carried out at the Kamiebetsu water purification plant (Ebetsu, Japan) using Chitose River surface water as raw water. Characteristics of the raw water used for the pilot operation are described elsewhere. In authors’ previous study using the same water, it was found that carbohydrate-like substance was dominant in the foulant causing physically irreversible fouling. After passage of the grit chamber, the raw water was delivered to the membrane units without any pretreatment. The PE membrane, which had the same properties as those described before, was assembled (3 m2) and horizontally immersed in a 300 l submersion tank. The operation was conducted using a vacuum. The filtration flux was set at a constant value of 1.0 m3 m2 d1. During the operation that continued for 49 days, periodic physical cleaning was carried out by filtration for 30 min, air scrubbing for 30 s, and hydraulic backwashing for 60 s, as recommended by the manufacturer. When the membrane was rapidly fouled or the value of TMP became excessive, the submerged membrane module was taken out from the submersion tank and was cleaned by spraying pressurized water on the membrane surface. During the pilot-scale operation, membrane fibers were sampled from the center of the membrane module six times: on days 1, 3, 5, 16, 23, and 39. After cutting the fibers, corresponding channels were closed with epoxy glue to prevent leakage, and the permeate flow rate was adjusted to maintain a constant flux of 1.0 m3 m2 d1. To check for membrane breakage, turbidity of the permeate was monitored. After
membrane fibers had been cut, they were immediately brought to the laboratory in a container filled with distilled water (resulting pH of 6.570.5), and the surface of the membrane specimen was manually wiped with a sponge and rinsed with distilled water thoroughly. This step was carried out to ensure the removal of the accumulated cake (i.e., effect of physically reversible fouling) and to specifically focus on physically irreversible fouling in this study. It was found that manual sponge cleaning had little effect on permeability of the fouled membranes at the termination of the operation, indicating that physically irreversible fouling was dominant in the pilot operation. A portion of membrane fibers was examined in a zeta potential meter (ELS-8000, Otsuka Electronics, Osaka, Japan) at pH 7.0 and 5 mM KCl. The other membrane fibers were stored in Milli-Q water until use for AFM force curve measurements. Figure 14 shows the adhesion forces (Fad) of (a) carboxylmodified and (b) hydroxyl-modified microspheres to clean PVDF or PE membranes, which were determined from the maximum values in the retraction force curves (Figure 13). From Figure 14, it is obvious that the adhesion force of the hydroxyl group was much greater than that of the carboxyl group regardless of membrane. The difference in values shown in Figure 14 is explained by differences in a balance of three relevant forces: (1) electrostatic interaction, (2) hydrogen bond (or electron transfer interaction), and (3) van der Waals interaction as seen in Figure 15. The hydrogen bond and the van der Waals interaction work as attraction forces, while the electrostatic interaction is considered to be repelling force because of the negatively charged nature of both microspheres and membrane surfaces. The electrostatic repulsion is governed by Coulomb’s force, which is proportional to the product of the two different charges to be considered. The charges of the two functionally modified microspheres were comparable. Thus, similar levels of Coulomb’s force would be exerted on carboxyl- and hydroxyl-modified microspheres with the membranes. On the other hand, the van der Waals interaction between a microsphere and a flat surface is known to be proportional to the radius of the microsphere The two types of microspheres used in the present study had the same radius of 3 mm, and therefore the levels of van der Waals attraction were also considered
34
Membrane Filtration in Water and Wastewater Treatment
Carboxyl group (COOH)
–130 mV
Hydroxyl group (OH)
Electrostatic
–126 mV
Van der Waals interaction
–4 mV
–4 mV Hydrogen bond
Figure 15 Three relevant forces: (1) electrostatic interaction, (2) van der Waals interaction, and (3) hydrogen bond.
to be similar. Considering the balance of the three relevant forces, it is reasonable to conclude that the difference in the adhesion force shown in Figure 14 may be attributed to the hydrogen bond. A hydrogen bond is generated by electron transfer reaction between electronegative atoms (e.g., O, N, F, and Cl) and H atoms that are covalently bound to similar electronegative atoms. The two functional groups examined in the present study (i.e., carboxyl and hydroxyl groups) have the possibility of forming hydrogen bonds due to their high polar nature, but the bounding power largely depends on their pKa values. If pKa value is larger than pH of the solution, the functional group is protonated, contributing to the formation of a strong hydrogen bond. In contrast, in the condition of pKa being less than pH of the solution, the functional group dissociates, resulting in an insignificant hydrogen bond. To make sure the dissociation condition of two functional groups, an investigation of the adhesion force as a function of pH is considered to be appropriate. However, because of low resistance of available AFM cells to extreme pH condition, the authors could not figure out the dissociation condition of two functional groups. In previous studies in which the pKa values of hydroxyl- or carboxyl-modified microspheres were investigated, it was estimated that carboxyl groups have pKa values between 3 and 6 and hydroxyl groups have pKa values between 9 and 13. Assuming that the pKa values obtained in those previous studies could be applied to the present study, the carboxyl groups were dissociated whereas the hydroxyl groups were not dissociated in the adhesion force measurements carried out at pH 6.8 (Figure 14). In the present study, the difference stated above presumably caused the remarkable difference in adhesion force of the two types of microspheres. An additional remark that should be made for Figure 14 is that the adhesion forces of hydroxyl-modified microspheres to the PVDF membrane and the PE membrane were quite different. As shown in Figure 14, the binding power of the hydroxyl group was much greater for the PVDF membrane than for the PE membrane. According to Ducker et al., the adhesion value possibly varies depending on surface roughness. The difference between the roughness of the PVDF membrane and that of the PE membrane was insignificant, suggesting a limited effect of roughness on the difference in adhesion force. Rather, difference in polymer materials seemed to affect the binding force of hydroxyl-modified microspheres: binding power of the hydrogen bond largely depends on hydrogen
bonded pairs. It is known that PVDF has two fluoride atoms that are arranged symmetrically with a center carbon atom, while PE has only hydrogen atoms along with carbon chain. Generally, the higher the electronegativity of the bounded atom, the greater the binding energy of the hydrogen bond becomes. Because of the high electronegative nature of fluoride atoms, a strong hydrogen bond would be formed between the surface of the PVDF membrane and hydroxyl-modified microspheres. Based on the fact that carbohydrate has many hydroxyl groups in its structure, the hydrogen bond seems to play an important role in the accumulation of carbohydrate-like substances on membranes, as indicated by previous studies on fouling. The hydrogen bond is considered as a semi-irreversible reaction, and the value of binding energy is between 10 and 40 kJ mol1, which is stronger than that of typical van der Waals attraction (B1 kJ mol1). Because of such a strong and semi-irreversible binding ability of the hydrogen bond, it is probably very difficult to remove carbohydrate-like substances from membranes by physical cleaning (e.g., backwashing) once they have adhered to the membranes by hydrogen bonds. The data shown in Figure 14 suggest that more carbohydrate would accumulate on a membrane made from polymers containing atoms with high electronegativity. For the prevention of accumulation of carbohydrate on membranes used for water treatment, it would be desirable to choose membranes that are fabricated with polymers that do not contain atoms with high electronegativity in their structure. Figure 16 shows the approaching force curves repeatedly measured with (a) carboxyl- and (b) hydroxyl-modified microspheres for new PE and PVDF membranes. As shown in the figure, features of approaching force curves were completely different depending on the type of microspheres. As the carboxyl- modified microspheres approached the membrane surface (Figure 16(a)), they encountered repulsive interaction due to repulsive electrostatic interactions between the negatively charged microspheres and the negatively charged membrane surface. It is shown in Figure 16(a) that the interaction became apparent within a distance of about 20 nm for both membranes, demonstrating that the two membranes exerted similar electrostatic repulsion against the carboxyl-modified microspheres. This is consistent with the results of measurement of zeta potentials of the membranes: the two membranes exhibited similar negative charges.
PE 0.5 0.25
0.5 0.25
0.0
0.0 0
25
50
75
0
Distance (nm)
(a)
50
25
75
Distance (nm)
0.5
0.5 PE
PVDF
0.25
0.25
0.0
0.0
–0.25
–0.25
Force (nN)
Force (nN)
35
PVDF
Force (nN)
Force (nN)
Membrane Filtration in Water and Wastewater Treatment
–0.5
–0.5
–1.0
–1.0
–1.5
–1.5
–2.0
–2.0 25
0
50
75
0
Distance (nm)
(b)
25
50
75
Distance (nm)
Figure 16 Approaching force curves of (a) carboxyl-modified microspheres and (b) hydroxyl-modified microspheres to the PE membrane (left panels) and the PVDF membranes (right panels) in buffered solution (pH 6.8).
6 Adhesion force (nN)
Adhesion force (nN)
6 5 4 3 2 1
4 3 2 1 0
0 0 (a)
5
5
10 15 20 25 30 35 40 Operation time (days)
0 (b)
5
10 15 20 25 30 35 40 Operation time (days)
Figure 17 Changes in adhesion force of carboxyl-modified microspheres (a) and hydroxyl-modified microspheres (b) to PE membranes that were sampled during the pilot-filtration test.
In contrast, as the hydroxyl-modified microspheres approached the membrane surface (Figure 16(b)), rapid decrease in the bending stresses of the cantilever or jump-in attraction forces appeared after gradual increase in repulsion force. The increase in attractive force was probably due to hydrogen bonds between the hydroxyl groups of microspheres and the membrane surface. The effective distances of hydrogen bonds were around 15 and 5 nm in the case of the PVDF and the PE membranes, respectively. This was in accordance with the strong adhesion force of the hydroxyl-modified microspheres to the PVDF membrane discussed above. The results shown in Figure 16 suggest that hydrogen bonds between foulants and membranes can be significant only when they are transported to the region where the
membrane surface is very close. Before entering the region where hydrogen bonds can be significant, foulants need to overcome repulsive forces if they bear negative charges. Otherwise, they do not adhere to the membrane surface and subsequently cause membrane fouling. Strongly negativecharged particles/molecules (e.g., humic acid) are less likely to reach the membrane surface: in contrast, it is expected that carbohydrate-like substances relatively easily access to the membrane surface because of their electrostatically neutral nature. This is an additional explanation why carbohydratelike substances have recently been reported to be major foulants. Figure 17 shows the changes in adhesion forces (Fad) of (a) carboxyl- or (b) hydroxyl-modified microspheres to the PE
36
Membrane Filtration in Water and Wastewater Treatment
membranes, which were sampled during the pilot-scale filtration on days 0, 1, 3, 5, 16, 23, and 39. Adhesion force shown in the figure was determined by the same procedure as that used for obtaining the data shown in Figure 14. As clearly shown in Figure 14, adhesion forces of both hydroxyl-modified and carboxyl-modified microspheres changed to a large extent as a result of fouling. In the case of carboxyl-modified microspheres, the adhesion force decreased rapidly to a value of 0.06 nN within 1 day and remained at a low level until the end of operation. One possible reason for the reduction in binding force of carboxyl-modified microspheres was the increase in electrostatic repulsion. The charge of the membrane surface changed from 11 to 28 mV during the pilot operation (Yamamura et al., 2008), which resulted in greater electrostatic repulsion between negatively charged microspheres and the membrane surface. In authors’ previous fouling study using the PE membrane carried out at the same plant, it was shown that negatively charged substances (e.g., humic substances) also accumulated on/in the membrane during the long-term filtration. Accumulation of such negatively charged substances presumably decreased the charge of the membrane surface. As shown in Figure 17, adhesion force of the hydroxylmodified microspheres also declined rapidly, but the values of Fad for the hydroxyl-modified microspheres were much larger than those for the carboxyl-modified microspheres except for on day 39. This result partially explains why hydrophilic NOM dominated over humic substances and was shown to be a major foulant in previous studies on fouling: hydrophilic NOM actually has a great binding power to the membrane due to hydrogen bonding.
The exponential reduction of adhesion force seen with hydroxyl-modified microspheres could presumably be explained by the decrease in binding sites available on the membrane surface due to membrane fouling and/or by the increase in repulsive forces between negatively charged microspheres and the negatively charged membrane surface.
4.02.1.3 Membrane Filtration Systems for Controlling Fouling In order to reduce the membrane fouling, we need to produce the membrane resistant to fouling and to construct hybrid membrane systems which include the existing treatment processes such as coagulation, activated carbon adsorption, and biological/chemical oxidation. Figure 18 describes such a concept, considering the size, concentration, and chemical properties of the substances to be removed.
4.02.1.3.1 Channel flocculation in monolith ceramic membrane Coagulation–flocculation process has been widely used to form aggregates (flocs), which include many fine particles contained in the raw water, for the efficient solid–liquid separation in the sedimentation basin and sand filter. Tambo and Watanabe published several papers describing the floc density and flocculation kinetics for the better understanding of flocculation process. They presented the floc density function and GC0T value. The floc density function describes the quantitative relationship between the size and effective (buoyant) density of flocs. The exponent Kr in the function is related to the fractal dimension (D) for the aggregates formed
Impurities mm Suspended matters
Organic–inorganic soil (clay, microorganisms, highmolecular-weight humics, etc.) Silts MF–UF Algae filtration Protozoa (Cryptosporidium, Giardia, etc.) Bacteria
µm Impurity size
Protein Colloidal matters
Coagulation + MF–UF
Coagulation / sedimentation + MF–UF filtration
Oxidized substances (SiO2, Fe2O3, Al2O3, MnO2, etc.) Humic acids Virus
nm
Dissolved matters
Å
Adsorption, ion exchange + MF–UF
Saccharoid
Ozonation, activated carbon adsorption, biological oxidation + MF–UF filtration
Taste and odor producing inorganic ions (Fe2+, Mn2+, etc) Fulvic acids NF filtration Synthetic organic compounds (DDT, BHC, PCB,) Inorganic compounds (arsenic, antimony, seleninum, etc.)
Concentration Figure 18 Design matrix of hybrid membrane filtration systems. DDT, dichlorodiphenyltrichloro ethane; BHC, benzene hexachloride; MF, microfiltration; NF, nanofiltration; PCB, polychlorinated biphenyl; UF, ultrafiltration.
Membrane Filtration in Water and Wastewater Treatment
in cluster–cluster aggregation (CCA) as D ¼ 3 Kr. Kr is a function of the aluminum to turbidity (ALT) ratio, which is defined as Al dosage(mg/l)/suspended solid concentration (mg l1) in raw water, and has the value of 1.00 and 1.25 for the ALT ratio of around 1/100 and 1/20, respectively. These values coincide with the fractal dimension D determined for the reaction and diffusion limited case (2.05 and 1.75), respectively. Tambo and Watanabe have proposed that the GC0T value is more useful than GT value proposed by Camp as the criterion of flocculation. These research results have been included in the membrane filtration process to improve the filterability of the membrane (Yonekawa et al., 2004). In Japan, membrane filtration plant has increased its treatment capacity since the mid-1990s. Tokyo Metropolitan Water Works Authority constructed a plant with the total capacity of 80 000 m3 d1 in April 2007 using hollow-fiber MF membranes made of PVDF. It is currently the largest plant in Japan. There has also been innovation in the membrane material and membrane module. The monolith ceramic membrane was developed in 1988 and its advances have been remarkable as seen in Figure 19. Figure 20 describes the detail of the monolith ceramic membrane. By the end of 2008, 81 plants with monolith ceramic membrane have been under operation in Japan and the maximum capacity of the plants is about 40 000 m3 d1. The pre-coagulation has been provided to all of these plants to strengthen filterability for stable filtration performance for a
Configuration
Unit
Length
mm
Diameter
mm
Channel number
wide range of raw water turbidity and enhancement of the removal of viruses and dissolved organic substances. The authors have clarified the characteristics unique to monolith ceramic membrane with pre-coagulation by referring to the behavior of microparticles. The region exists in the monolith channel with the optimum G and GC0T value for good flocculation. The flocculation of microparticles offers the reduction in the membrane fouling. The laminar flow model within dead-end hollow-fiber membranes has been presented in many studies. For example, Fujita and Takizawa developed Equation (1) from the energy equation and the material balance in the course of filtration:
dp v 8m ¼ 1 dv g rdkðp p0 Þ
ð1Þ
where p is the static pressure (m), v the axial velocity within hollow fiber (m s1), g the gravitational acceleration (m s2), m the viscosity (kg m1 s1), r the water density (kg m3), d the internal diameter of hollow fiber (m), k the membrane filterability (s1) and p0 the external pressure of membrane (m). Considering the characteristic values (d ¼ 4 104 m, k ¼ 6 106 s1) of the typical hollow fiber, the first term in Equation (1) is much smaller than the second term. Neglecting the first term, an appropriate equation to calculate an expanded approximate axial velocity in a fibre can be derived. In the case of monolith ceramic membrane (d ¼ 2.5 103 m,
Tube 1985
Stage
Monolith 1988
1990
1994
2001
2006
1000 10
1500
30
1
19
37
180 61
2000
Channel diameter
mm
7
4
3
Membrane area
m2
0.02
0.24
0.35
0.48
15
24
Packing density
m2 l–1
0.25
0.34
0.50
0.63
0.6
0.63
1.5m3 8.9m2 1000 Module capacity (m3 d–1 module)
2.5
Industrial use
Application
100
10
Figure 19 Advance in monolith ceramic membrane.
37
13m2
Water purification m–2
d–1
1.8m3 73m2
m–2
d–1
2.5m3 m–2 d–1
5m3 m–2 d–1
150m2
240m2 module–1
38
Membrane Filtration in Water and Wastewater Treatment
Figure 20 Detail of monolith ceramic membrane (META water product).
k ¼ 5 105 sec1), however, the first term in Equation (1) cannot be neglected to derive an appropriate equation for calculating axial velocity in a monolith channel. Without neglecting the first term in Equation (1), the authors have developed Equation (2) to calculate an expanded approximate axial velocity in a monolith channel:
v2 v ¼ vf coshðaxÞ b pf pe þ f sinhðaxÞ 2g rgdk 4dk 2 ; a¼ 2 b ¼ 8m d b
ð2Þ
where pf and vf are the pressure (m) and velocity at inlet of monolith channel (m s1), respectively. On the other hand, the membrane filterability k in the monolith membrane has a certain distribution. To facilitate analysis of the flow pattern on the basis of this distribution, a five-channel model with three levels of filterability was created, as described in Figure 21. Solving Equation (3) under the material balance and appropriate boundary conditions, the equation for axial velocity in the five channel model has been derived as
vi ¼ vfi coshðai xÞ bi ðpfi pe Þsinhðai xÞ ði ¼ 0; 1; 2Þ
ð3Þ
where pfi is the total pressure at channel inlet (m) and pe the external static pressure of membrane (m). The calculated flow pattern in the monolith ceramic membrane module is shown in Figure 22. A concentrate flowing out through outlets of channels 1 and 2 with lower filterability is drawn into channel 0 with higher filterability. It was also confirmed that the dead-end point is located at the position with an axial velocity vi ¼ 0 in channel 0.
In the channel of 1 m long, axial velocities calculated by Equation (9) are shown in Figure 21 for the membrane flux of 2 m3 m2 d1. The G value in the channels 0–2 was calculated at about 40 s1, which is in the range of optimum values proposed by Camp. On the other hand, the mean hydraulic residence time in the channels 0–2 was about 50 s. Therefore, the GT value in the channel is only about 2000, which is too small compared with the Camp’s proposed values. However, good flocculation was observed in the channel, because the GC0T value in the part of channel is high enough for good flocculation, explained as below. Using the data shown in Figure 23, the distribution of the local G values within the channel 2 under the membrane flux of 2 m3 m2 d1 is described as seen in Figure 23. Considering the velocity distribution in the channel and high concentration of coagulated microparticles reflected by membrane filtration, the GC0T value may be high enough for a good flocculation in the region with the local G value of 40–100 s1. In this context, C0 is defined as the coagulated microparticle concentration near the entrance of such a region. Figure 24 shows the experimental setup (large and small monolith membrane module) and sampling points. The top and bottom portion of the both modules were made of transparent material to enable a visual observation of flocs using video camera. Raw water was taken from the Kiso River near Nagoya city. The dosage of coagulant (polyaluminum chloride, PACl) was fixed at 1 mg Al l1. For rapid mixing condition, G value was fixed at 150 s1 and hydraulic detention time at 300 s. The filtration mode was dead end and membrane flux was fixed at 2 m3 m2 d1. The specifications and operation conditions of the two membrane modules are described in this chapter.
Membrane Filtration in Water and Wastewater Treatment
39
Eq. (3) Velocity equation for a 5-channels model
5 channels
Average velocity in channels (m s−1)
Channel no.0 Permeability k = 5.80 × 10–5[s–1]
Channel no.1 Permeability k = 4.65 × 10–5[s–1]
Channel no.2 Permeability k = 4.07 × 10–5[s–1]
Channel no.1 Permeability k = 4.65 × 10–5[s–1]
Channel no.0 Permeability k = 5.80 × 10–5[s–1]
i = fi cos(i x ) – i (p f,i – pe) sinh(i x ) (i = 0, 1, 2)
0.04
Channel no. 0 Channel no. 1 Channel no. 2
0.03 0.02
0.01
2 m3 m–2 d–1
0 0
0.2
0.4
0.6
0.8
1.0
Channel axial coordinate (m) Figure 21 Five-channel model and filterability k.
Module casing
Dead-end point
90%
Membrane
96.5%
Feed
Figure 22 Flow pattern in monolith module.
With the laser diffraction scattering-type particle-size distribution cell holder (Horiba LY-073), the particle-size distribution was measured to verify the predicted flocculation phenomena and its effect on the filterability of the monolith ceramic membrane. The behavior of microparticles with the size of 0.5–15 mm in the channel with lower filterability was also measured to identify the critical particle size. Polystyrenetype latex particles (JSR Stadex/Dynospheres: 0.5, 3, 5, 10, 15 mm, specific density of 1.05) were used as model particles.
The authors also investigated the correlation between microparticle concentration and TMP using the effluent from a conventional rapid sand filtration process, as shown in Figure 25. There exists a clear relationship between them. It would suggest a significant effect of the flocculation on the filterability in the monolith channel, because the microparticles, larger than 1 mm in the shear field, are subject to a lift force such as the lateral migration and shear-induced diffusion which are proportional to square and cubic power of the equivalent particle diameter, respectively, as described in Figure 26. There were no visual flocs in the bottom portion of the module where coagulated microparticles entered. Visual flocs, however, blew out at the maximum velocity of 3–8 mm s1 from the lower filterability channels in the upper portion of the module. From the analytical result with five channel model, the average outflow velocity at the membrane top was estimated to be 2–4 mm s1. The maximum flow velocity in laminar flow is twice the average velocity. Therefore, the analytical result has been confirmed by the visual experiment. The authors measured the concentration of polystyrenetype latex particles with the size range of 0.5–15 mm in the influent and effluent of the membrane. There were almost no particles in the effluent. It demonstrated that the latex particles of smaller than 15 mm are deposited onto the membrane surface in the course of membrane filtration. This result can explain the correlation of the variation of microparticle number in raw water and TMP as seen in Figure 25. The experiment on the flocculation in the monolith channel was carried out to prove that good flocculation occurs in the channel and will improve the filterability of the membrane. From the theoretical analysis, the average flow velocity in the channel with lower filterability is about 0.5 mm s1 in
40
Membrane Filtration in Water and Wastewater Treatment Recovery 90%
Channel diameter 2.5 mm
Let’s consider “flocculation condition in channel” Channel length 1000 mm
G value 20 sec–1
Especially, near the membrane surface G value 20−100 s –1 : desirable value for flocculation Contact Time
40
60
80
Enough : laminar velocity is very low 100 Concentration Highly concentrated : accompanied by filtration Flux 2 m3 m–2 d–1 Figure 23 Profile of G values in monolith channel with lowest k.
Frequency (volume based ) (%)
20
SP3
16
SP4
12 SP2 SP1
8
Sp4
4 Filtrate
0 1
10
100 Particle size (µm) Small membrane
Coagulant (PACI) M
M
P SP1
SP2
SP3
Figure 24 Experimental setup and sampling points (SPs). PACl, polyaluminum chloride.
the region of 1–200 mm from the surface, so the detention time is between a few tens of minutes and few hours. The G value in the zone is between 20 and 100 s1. The floc size distribution in each sampling point is seen in Figure 24. Flocs are lifted up by laminar flow and carried away from the outside of the channel. Therefore, the space near the membrane surface might be considered to be a high efficient field for coagulating the charge neutralized microparticles.
Figure 27 shows a schematic image of phenomenon occurred in the channel when the pre-coagulation is prepared. In order to confirm the flocculation effect on the improvement of ceramic membrane filterability, the authors carried out an additional experiment using the small module with the Nishitappu River water. It is a very clean water with annual average turbidity and DOC of 1.3 TU and 0.6 mg l1, respectively.
Membrane Filtration in Water and Wastewater Treatment
41
CSF treated water
Run 6
Pore size
Flux
Interval
Pressure
Recovery
1.0 mm
20 m3 m–2 d–1
15 min
300 kPa
93.3%
TMP (kPa)
50 10
40 5 TMP Microparticle 30 29 Oct.
30 Oct.
Microparticle count (103 ml–1) (0.5–1.0 µm)
15
0
31 Oct. Date
Figure 25 Correlation between transmembrane pressure (TMP) and microparticle concentration. CSF, coagulation/sand filtration.
0 Monolith ceramic membrane
Back transport –2 log cm s–1
F1 ux Membrane
Log transport velocity (cm s–1)
2 Minimum size of particle that will not deposit on membrane
0
DpL = 56 µm
d–1
–4 0.8 µm
–6
DpL = 87 µm
–8 –10
Ultrafiltration flux
–4
–8 –4
m–2
Microfiltration flux
–2
–6
2
m3
Brownian R = 0.03 cm u = 133 cm s–1 T = 20 °C –3
–3
Shear Calculation conditions
–2 –1 0 1 Particle diameter: log Dp (µm)
2
3
Channel diameter = 2.5 mm Water temperature = 20 °C Channel entrance
Lateral migration
–2 –1 0 1 Log particle diameter (µm)
Middle point of channel 2
Back-transport velocity and critical flux Figure 26 Particle size and lift force.
Figure 28 shows the experimental result and confirms the effect of flocculation on the fouling reduction. Further improvement is possible using the chemically enhanced backwashing (CEB) with acidic solution. Coagulant addition of 1 mg Al l1 to the monolith ceramic MF membrane system also improved the virus log removal efficiency up to 7. Figure 29 shows the experimental verification of the effect of the CEB on the membrane filterability. The reason behind the
improvement may be the removal of microflocs attached to the membrane surface by the ECB.
4.02.1.3.2 Pre-coagulation/sedimentation in hollow-fiber UF/MF membrane The surface water from Chitose River and Nisitappu River was used as the raw water in the experiment. Table 2 summarizes
42
Membrane Filtration in Water and Wastewater Treatment
L = uo2 dp3/(32 ro2)
Lateral migration Shear-induced diffusion
S = 0.05 uo dp2/(4 ro2)
Ceramic membrane surface
Disaggregated floc particles
Lift force
Ceramic membrane surface u (membrane flux) (b) Aggregation
(a) Carrying near the membrane accompanied with filtration
(c) Lifting from membrane
Figure 27 Schematic image of effect of channel flocculation.
80
4 TMP Membrane flux
TMP (kPa at 25 °C)
3 Back washing interval: 4h
2h
40
2
20
1
0 1/4
Membrane flux (m d–1)
Precoagulation
60
0 1/19
2/3
2/18
3/5
3/20
4/4
Figure 28 Effect of channel flocculation on transmembrane pressure (TMP) change.
CEB (acid)
Experimental flow Coagulation
TMP (kPa)
Mn oxidization
Ceramic membrane
40
10 m3 m–2 d–1
30
8 m3 m–2 d–1
20
6 m3 m–2 d–1
10 0 04 Jan.
4 14 Jan.
m3
m–2
d–1,
24 Jan.
with CEB
without CEB 03 Feb. Date
13 Feb.
Figure 29 Effect of CEB, chemically enhanced backwashing on TMP change under high flux operation.
23 Feb.
Membrane Filtration in Water and Wastewater Treatment
the average raw water quality of Chitose river during the experiment (Jang et al., 2004). With Chitose River water, the pilot plant consists of a rapid mixing tank, a jet mixed separator (JMS) with inclined tube settlers, and three hollow-fiber UF or MF membrane filters as described in Figure 30. The JMS is a simple but effective solid/liquid separator with several vertical porous plates in a channel; microflocs are flocculated under the turbulent flow produced by the water jets and larger parts of grown flocs settle between the plates; subsequently, residual small flocs are removed in the inclined tube settlers. The effective volume of JMS with inclined tube settlers is 7.0 m3 and flow rate to the JMS was 120 m3 d1, corresponding to the hydraulic detention time of 84 min. The operating conditions of this pilot plant are summarized (refer to Jang et al., 2004). Four processes of the pilot experiment were carried out. In processs 1 and 2, the aluminum sulfate (AS) with activated silicate and PACl was used as coagulant. The water was fed from outside to inside of hollow-fiber UF membrane, which is made of specially polymerized PAN with nominal average pore size of 0.01 mm, at a constant permeate flow rate of 0.9 m d1. The physical cleaning with back washing and air scrubbing was carried out to prevent fouling in a time interval of 30–60 min. In processes 3 and 4, polysilicato-iron (PSI) which is inorganic polymeric iron coagulant was used as coagulant PSI has a molecular ratio of Fe to Si of 1:1–1:5, but we used the molecular ratio with 1:1 in this pilot plant experiment. Coagulant dosage and coagulation pH were 0.21 mmol Fe l1 and 6.2, respectively. Results of TMP trends, for Chitose River water, with increasing UF filtration time in process 1 are shown in Figure 31. Figure 32 shows the comparison of the TMP among processes 1, 2, and 3 using the same UF membrane and AS, PACl, and PSI as coagulant, respectively.
Transmembrane pressure at 25 °C (kPa)
Considering the data shown in Figure 33 and Table 3, it may be concluded that the higher DOC removal in the precoagulation/sedimentation gives better performance of UF membrane filtration. Even though the TMP used by PACl and PSI was almost the same at about 3300 h of filtration time (the actual TMP reached about 100 kPa, which is the recommended TMF for chemical cleaning), TMF used PSI has always been lower than that by AS and PACl. Figure 33 shows the comparison of removal efficiency of DOC among the three coagulants. PSI gave the best removal efficiency resulting in the best filtration performance. With Nishitapu River water, the TMP increased in each operating condition as seen in Figure 34. When the Nishitapu River water was filtered at constant flow rate of 1.1 m d1 directly by using UF membrane, the filtration time to reach 100 kPa of TMP was only 300 h in spite of low organic content and low inorganic content. However, the filtration time for coagulated water was 4 times longer than that. In addition, hypochlorite solution was added
150
Raw-UF Coa.-UF Sed.-UF
120 90 60 30
Flux: 0.9 m d–1
0 0
500 1000 1500 2000 Membrane filtration time (h)
Jet mixed separator (JMS)
P
Rapid mixing tank Permeate
Permeate
Permeate
Drain P
Compressor
Figure 30 Schematic description of pilot plant.
2500
Figure 31 Effect of pre-coagulation/sedimentation on performance of ultrafiltration (UF) membrane system.
Coagulant
Chitose river water
43
P
P
44
Membrane Filtration in Water and Wastewater Treatment
during backwashing term; the UF membrane filterability was significantly improved. These results were also obtained in the case of using MF membrane as seen in Figure 35.
4.02.1.3.3 Hybrid submerged MF membrane system
Transmembrane pressure at 25 °C (kPa)
The hybrid MF membrane system is a combination of submerged membrane and the other processes such as the powdered activated carbon adsorption and chemical/biological oxidation. The membrane system has been developed to purify raw waters with low quality containing a lot of soluble matter such as biodegradable organics, humic substances, manganese, and ammonia nitrogen. In the hybrid system, soluble less-biodegradable organics are adsorbed to the
120
Sed.-UF (PSI: 0.21 mmol-Fe l–1)
100
Sed.-UF (PACI: 0.19 mmol-Al l–1) Sed.-UF (AS: 0.37 mmol-Al l–1)
80
powdered activated carbon, and suspended particles including powdered activated carbon are separated by the membrane filtration. The soluble biodegradable organics, manganese and ammonia nitrogen, are biologically or chemically oxidized. In the case of chemical oxidation (with prechlorination), soluble manganese is oxidized with chlorine and the catalytic reaction of powdered activated carbon, and the oxidized manganese is removed by membrane separation. Ammonia nitrogen is also oxidized by chlorine in a pre-chlorination tank. In the case of biological oxidation (without prechlorination), the iron oxidizing bacteria and ammonia oxidizing bacteria, which are concentrated in submerged membrane tank, oxidize the soluble manganese and ammonia, respectively. A schematic diagram of the pilot plant is shown in Figure 36 (Suzuki et al.,1998). The volume of membrane submerged tank and the surface area of submerged membrane were 4 m3 and 86–120 m2, respectively. Detention time in the mixing tank was 10–15 min. The raw water was fed into the mixing tanks. Four types of polytetrafluoroethylene (PTFE) membranes were used. When the first, second, and third type of membranes were used, the
60 Table 3
Physically irreversible resistance
40 Run 1(125 days)
20
Run 2(73 days)
Module Module Module Module A B A B
0 0
500
1000 1500 2000 2500 3000 3500 Membrane filtration time (h)
Figure 32 Effect of various coagulants on performance of ultrafiltration (UF) membrane system.
Membrane flux (m3m2d1) Physically irreversible filtration resistance (1011 m1)
0.2 0.29
0.6 1.52
0.4 0.69
0.8 2.48
100 Removal efficieny of DOC (%)
Removal of UF
Removal of coagulants
Removal of sedimentation
80 PSI
60 PACI
40
AS
20
Run 1
Run 2
RW -U F C oa -U JM F SU F
RW -U F C oa -U JM F SU F
RW -U F C oa -U JM F SU F
RW -U F C oa -U JM F SU F
0
Run 3
Run 4
• Direct UF; around 15%, precoagulation / sedimentation; 35–60% • The highest DOC removal efficiency was obtained in run 4 Figure 33 Removal efficiency of dissolved organic carbon (DOC) with various coagulants. AS, aluminium sulfate; PACl, polyaluminum chloride; PSI, polysilicato-iron.
Transmembrane pressure at 25 °C (kPa)
Membrane Filtration in Water and Wastewater Treatment
45
Raw-UF (hypochlorite solution was not added, flux: 1.1 m d–1) Coa.-UF (hypochlorite solution was not added, flux: 1.1 m d–1) Coa.-UF (hypochlorite solution was added, flux: 1.1 m d–1) Coa.-UF (hypochlorite solution was added, flux: 1.7 m d–1)
140 120 100 80 60 40 20 0 0
500
1000
1500
2000
Membrane filtration time (h)
Transmembrane pressure at 25 °C (kPa)
Figure 34 Effect of operation condition on performance of ultrafiltration (UF) membrane system.
100
Raw-MF (hypochlorite solution was added) Coa.-MF (hypochlorite solution was added)
80 60 Flux: 1.4 m d –1
40 20 0 0
100 200 300 400 Membrane filtration time (h)
500
Figure 35 Effect of precoagulation on microfiltration (MF) membrane system.
raw water to the pilot plant was taken from the existing water purification plant, which had already contained the powdered activated carbon in the concentration of 5–30 mg l1. In the first tank, hypochlorite was added when the chemical oxidation was applied, and the sludge containing biomass and activated carbon were returned from the membrane submerged tank and mixed with the raw water in the second tank. The same powdered activated carbon (average diameter of 10 mm) was dosed into the second tank at the constant concentration of 13 mg l1 when the fourth PTFE membrane was used. PACl was added to coagulate the small suspended particles in the third tank. The pretreated water was fed into the submerged tank where the hollow-fiber PTFE membranes were submerged and intermittent aeration was performed to supply the oxygen to the microorganisms. Air was supplied for 1 min. with the intensity of 0.64 N m3 min1. every 4 min. The raw water in the pilot plant study was surface water from Chitose River. It contained many humic substances, soluble manganese, and ammonia nitrogen. The average raw water quality is given in Table 2. The dosage of hypochlorite in the mixing tank was 4–6 mg Cl2 l1. Coagulant dosage was 2–3 mg Al l1.
In the pilot plant experiments, symmetric or composite PTFE membrane with nominal pore size of 0.1 mm was used. The thickness of skin layer in the composite membrane was changed at 60, 30, and 15 mm. The pore density was about 80%. The skin layer thickness and pore density of the newest composite PTFE membrane are about 15 mm and 80%, respectively. The hybrid membrane system is able to efficiently remove the soluble matter such as organics, manganese, and ammonia nitrogen. The soluble manganese and ammonia nitrogen were oxidized biologically or chemically and small humic substances were adsorbed to the powdered activated carbon. The removal efficiency of TOC, E260, and trihalomethane formation potential (THMFP) is shown in Figure 37, and the comparison in the removal efficiency of the soluble manganese and ammonia nitrogen between chemical oxidation with prechlorination and biological oxidation without prechlorination was made in Figure 38 when the composite PTFE membrane with the skin layer thickness of 30 mm was used. Dosage of powdered activated carbon was fixed at 13 mg l1. As previously reported by the authors, chemical oxidation is necessary to oxidize soluble manganese when the raw water temperature become less than 10 1C. The authors also reported that improved filtrate quality can contribute to keep a higher flux. Figure 39 shows the change of the permeate flux, TMP, and raw water temperature during the experiment with the symmetrical membrane. In this experiment, the hybrid MF membrane system was operated without prechlorination. The average flux was relatively low at less than 0.3 m d1 and the TMP increased to 70 kPa after 5 months of operation. To improve the permeability of the PTFE membrane, the structure of membrane was changed from symmetrical to composite. Permeability of the composite membranes with different skin layer thickness was compared in the pilot plant experiment. The thinner the skin layer thickness, the better the permeability. Figure 40 shows the change of the membrane flux, TMP, and raw water temperature with increasing operation time when the newest composite PTFE membrane was used. It demonstrates that TMP was very stable under a high flux of 1.2 m d1. It is about 4 times higher than that in the symmetrical membrane.
46
Membrane Filtration in Water and Wastewater Treatment Hybrid submerged PTFE MF membrane system including coagulation, carbon adsorption and biological oxidation Submerged MF membranes PAC Cl2 PACl Raw water
C
P Suction pump
Compressor
P
Storage tank of permeate
P Circulation pump Hydraulic retention time = 1.5 h Figure 36 Schematic description of pilot plant. PAC, powdered activated carbon; PACl, polyaluminum chloride.
4 3.5
Raw water Raw water (soluble) Membrane filtrate
0.12 0.1
2003/9/1~2004/8/31 0.099 0.086
mg l–1
2.5
2.43
1/cm mg / l
3 2.29
2 1.5
1.17
0.08 0.06 0.04
0.031
1 0.02
0.5 0
0.017
0 TOC
E260
THMFP
Figure 37 Removal efficiency of total organic carbon (TOC), E260, and THMFP in hybrid system.
4.02.1.3.4 PVDF Membrane filtration with pre-ozonation Combination of ozonation with membrane filtration is effective for the prevention of membrane fouling. Japanese membrane manufacturing companies have developed the ozone-resisting membrane module made of PVDF with potting material having a high resistance to ozone. In the developed membrane module, water containing residual ozone can be directly filtered. It is reported that this system can provide consistently high permeate flux for various raw waters, especially high turbidity water and secondary treated municipal wastewater. We studied the effect of residual ozone on fouling reduction using the ozone resisting PVDF membrane.
Figure 41 shows the schematic diagram of the experimental system. The same raw water was used as the hybrid membrane system. The average water quality is shown in Table 2. In experimental runs 1-1 and 1-2, ozone dosage was 2.0 and 4.2 mg O3 l1, in which the residual ozone concentration was 0.73 and 1.13 mg O3 l1, respectively. The ozone contact time was of 7.8 min in all experimental runs. In runs 2-1 and 2-2, ozone dosage was 1.4 and 1.9 mg O3 l1, in which the residual ozone concentration was 0.41 and 0.61 mg O3 l1, respectively. Figure 42 shows the TMP change with increasing operation time in run 1 where the constant permeate flux mode operation under 3.5 m d1 with physical cleaning of backwashing
Membrane Filtration in Water and Wastewater Treatment 0.3
mg l–1
0.15
0.2
0.090
0
0.1
0.05
Manganese
0.01
0.016
0.05
0.15
0.002
0.1
2003/11/13–2004/8/31 with prechlorination (chemical oxidation)
0.100
0.15 0.105
mg l–1
0.25
2003/6/25–11/13 without prechlorination (biological oxidation)
0
Ammonia nitrogen
Manganese
0.01
0.25
Raw water Raw water (soluble) Membrane filtrate 0.23
Raw water Raw water (solouble) Membrane filtrate
0.079
0.3
0.2
47
Ammonia nitrogen
30
Flux Water temperature
0.6
25 20 15
0.4
10
0.2
5 0
30 Nov
31 Oct
1 Oct
1 Sep
3 Jul
100 90 80 70 60 50 40 30 20 10 0
2 Aug
0
Water temperature (°C)
0.8
3 Jun
TMP (kPa)
Membrane flux (m d–1)
Figure 38 Removal efficiency of Mn and NH4–N in hybrid system.
2002
0
30
60
90 Operating days
120
150
Figure 39 Transmembrane pressure (TMP) changes and permeates flux (symmetrical PTFE membrane with pore size of 0.1 mm).
and air scrubbing was carried out. It clearly demonstrates that the residual ozone reduced the membrane fouling (Lee et al., 2004). After the continuous operation for about 1800 h, chemical cleaning of fouled membrane was conducted. The following three chemical solutions were used: NaOH solution of 1%, NaClO solution of 5 mg l1, and oxalic and nitric acid of 2%
and 5%. Figure 43 shows the extracted TOC in each chemical solution. As seen in Figure 41, preozonation with residual ozone significantly decreased the attached organic substances to the membrane causing the physically irreversible fouling. It may come from the following two ozone-induced reactions: degradation of organic substances and destabilization of particles on the membrane surface. The ozone-induced particle
Membrane Filtration in Water and Wastewater Treatment 50
1.8 1.6 1.4 1.2 1 0.8 0.6 0.4 0.2 0
Average membrane Flux: 1.2 m d –1 40
:
30 Temperature
Flux
Flux
Water temperature
20 10
Water temperature (°C)
Membrane flux (m d–1)
48
0
100 90
TMP (kPa)
80 70 60
Chemical washing
50 40 30 20
0
30
60
90
120 150 Operating days
180
30 Aug
31 Jul
1 Jul
1 Jun
2 May
2 Apr
3 Mar
2 Feb
3 Jan
0
4 Dec
10
210
240
270
Figure 40 Transmembrane pressure (TMP) changes and permeates flux (new composite PTFE membrane with a skin layer of 15 mm and porosity of about 80%).
destabilization reaction has been reported by many researchers. In the other experiment, we measured the size distribution of fouling particles in the backwash water and found that the average size of the particles was about 20 and 50 mm without and with ozonation, respectively. Increasing particle size increased the rate of back transport of organic particles, leading to the decrease in the accumulation of organic particles on the membrane surface. In this experiment, permeate TOC (i.e., DOC) concentration in the membrane filtration system without and with preozonation was the same as 2.4 mg l1 but E260 and E260/DOC were 0.062 and 0.034 cm1, and 0.026 and 0.013 cm1 mg1 l1, respectively. These results demonstrate that the biodegradability of organic particles increased due to the oxidation by O3.
4.02.2 Membrane Application to Wastewater Treatment 4.02.2.1 Current Status of MBRs Necessity of recycling use of water has been recognized to resolve the shortage of water resources. Municipal wastewaters seem to be an important water resource for recycling use. MBR is a key technology for creating the reclaimed water resource. MBR has been applied to the municipal wastewater treatment since the 1980s. The first-generation MBR combines a crossflow-type membrane with outside bioreactor and mixed liquor
is recirculated into membrane module. The operation pressure is high and recirculation pump is needed. In addition, it is reported that microorganism activity decreases due to the recirculation of the mixed liquor. The second-generation MBR submerges membrane module directly in the bioreactor. As a result, circulation pump is not needed and the operating pressure is low. Submerged MBRs have been preferred due to their lower energy consumption and smaller footprint compared with recirculated MBRs. However, it is reported that accumulated dissolved organic matter in the bioreactor decreases the membrane permeability in the submerged MBR more seriously compared with the first-generation MBR. In 2005 the European Commission decided to boost the development and application of MBR processes for municipal wastewater treatment through financing a 3-year research project within the scope of the 6th framework program: AMEDEUS (accelerate membrane development for urban sewage purification). Within AMEDEUS an analysis of the potential for MBR standardization was carried out. Based on an extensive survey of the MBR industry, the White Paper was published to provide a comprehensive overview of the market interest/expectation and technical potential of going through a standardization process of MBR technology in Europe. Due to the predominance of submerged MBR system in municipal applications, representing 99% of the installed membrane surface in Europe in the period 2002–05, the study focuses only on this configuration.
Membrane Filtration in Water and Wastewater Treatment
49
Preozonation-PVDF MF membrane system O3
Permeate
Chitose River water Ozonationmembrane Residual O3
PVDF membrane Back wash P
Run-4.1, 4.2 Run-5.1, 5.2
Air
Ozonation tank Retention tank O3 O2 Ozonationmembrane No residual O3
Back wash P
Run-4.3 Run-5.3
Air
O3 removal tank Membrane Run-4.4 Run-5.4
Pressurized membrane ⇒ Pore size: 0.1 μm (MF) ⇒ PVDF (polyvinylidenefluoride)
Back wash P
Air Figure 41 Experimental system of preozonation and membrane filtration.
300 Run 1 Run 2 Run 3 Run 4
TMP at 25 °C (kPa)
250 200 150 100 50 0
0
300
600
900
1200
1500
1800
Operation time (h) Figure 42 Effect of preozonation and residual ozone on membrane filtration. TMP, transmembrane pressure.
Figure 44 shows the number of municipal and industrial MBRs in Europe. In Japan MBR technology has been applied to the water recycling for some large business, commercial and residential complex buildings such as Roppongi Hills and Tokyo Mid Town. Membrane fouling deteriorates the membrane permeability and consequently increases energy consumption in MBR. A seriously fouled membrane must be cleaned with chemical reagents, which are costly. In addition, disposal of chemical reagents after membrane cleaning is an issue of concern, and the frequency of chemical membrane cleaning
should therefore be minimized. Thus, there is a need for efficient control of membrane fouling in MBR. In order to develop methods for efficient MBR operation, a better understanding of the mechanism of membrane fouling in MBR is needed.
4.02.2.2 Mechanism of Membrane Fouling Membrane fouling is a major obstacle for wider application of MBRs. Membrane fouling results in reduced performance, severe flux decline, high-energy consumption, and frequent
50
Membrane Filtration in Water and Wastewater Treatment 14 NaOH NaClO (Oxalic + nitric) acid
Detached organic substances per unit permeate volume (mg-TOC m–3)
12 10
Without O3
8 6 4 Residual O3 2 0 Run 1-1
Run 1-2
Run 1-4
Figure 43 Effect of residual ozone on amount of detached organic carbon.
600
Number of installations
With standardization?
Industrial (> 20 m3 d–1) Municipal (> 500 p.e.)
500 400
> 50 per year 300 200 > 20 per year 100 0 7
99
Rsm Am20 = 0.33 d−1 Kn20 = 1.0 mgN I−1
40
Influent ammonia concs
30
Rsm = For different influent ammonia concentrations
20
10
KnT ðbAT þ 1=Rs Þ Na ¼ Nae ¼ mAmT ðbAT þ 1=Rs Þ
1
ðmgN l Þ
ð128Þ 0
From Equation (128), the ammonia concentration (Na) in the reactor and effluent (Nae) are independent of the specific yield coefficient (YA) and the influent ammonia concentration (Nai). Using mAm20 ¼ 0.33 d1 and Kn20 ¼1.0 mgN l1 at 20 1C, and taking bAT ¼ 0.04 d1 (Table 10), a plot of Equation (128) with Nae versus sludge age Rs is given in Figure 24. At long sludge ages Nae is very low and remains so until the sludge age is lowered to about 4 days. Below 4 days, Nae increases rapidly and in terms of Equation (128) can exceed the influent FSA concentration, Nai. This clearly is not possible so the limit of validity of Equation (128) is Na ¼ Nai. Substituting Nai for Na in Equation (128) and solving for Rs give the minimum sludge age for nitrification, Rsm below which theoretically, nitrification cannot be achieved, that is,
Rsm ¼
1 ½mAmT =ð1 þ ðKnT =Nai ÞÞ bAT
ð129Þ
This minimum sludge age varies slightly with the magnitude of Nai (Figure 24) – higher Nai gives a slightly lower Rsm. The effect of Nai on RSm is very small because the magnitude of KnT is very small relative to Nai (o5%). So for Nai 420 mgN l1 (rarely will it be lower than this), and noting that Kn20B1 mgN l1, then KnT/Nai is negligibly small with respect to 1 (o5%). So substituting zero for KnT/Nai in Equation (129) yields
Rsm ¼
1 mAmT bAT
ðdaysÞ
ð130Þ
For all practical purposes, taking into account the uncertainty in mAm, Equation (130) adequately defines the minimum sludge age for nitrification. Conceptually, Equation (130)
0
2
4 6 Sludge age (days)
8
10
Figure 24 Effluent ammonia concentration vs. sludge age for the steady-state nitrification model.
states that if the net nitrifier multiplication rate (inverse of the net maximum specific growth rate, mAm bA) is slower than the harvesting rate of the nitrifiers via the sludge waste flow rate, then the nitrifiers cannot be sustained in the system and nitrification cannot take place. At sludge ages lower than the minimum for nitrification, nitrifiers are washed out of the system and so are called washout sludge ages. This concept of washout can be applied to any group of organisms in a bioreactor, and defines the sludge age below which the bioprocess will not take place because the organisms mediating this process are not sustained in the system. The virtually constant value for Rsm insofar as the influent FSA concentration is concerned (for the fixed values of mAmT and bAT) and the rapid decrease in effluent FSA concentration at sludge ages slightly longer than RSm is due to the very low Monod half saturation concentration for the nitrifiers (Kn20). This feature causes that in a particular plant, as the sludge age is increased, once Rs4Rsm, a high efficiency of nitrification will be observed, provided the FSA is the growth limiting nutrient for the ANOs, that is, all other requirements such as oxygen are met. Consequently, under steady-state conditions with increasing sludge age, kinetically, one would expect an AS system either not to nitrify at all, or, if it nitrifies, to nitrify virtually completely depending on whether the sludge age is
466
Biological Nutrient Removal
shorter or longer than the minimum (Rsm), respectively. Conversely, as sludge age decreases, one would expect an AS system to nitrify virtually completely and then quite suddenly cease to nitrify depending on whether the sludge age is shorter or longer than the minimum (Rsm), respectively. This behavior sometimes occurs in full-scale AS systems, where for many years the system has nitrified virtually completely, and suddenly one winter it stops nitrifying and produces very high effluent FSA concentrations. Provided the oxygen supply is not limiting, what happens in these situations is that over the years, the organic (COD) load on the system has increased and in order to maintain the reactor VSS concentration at the required level, the sludge wastage rate (Qw) has been increased, which reduced the sludge age. Then, coupled with a low winter temperature, the system sludge age falls below the minimum and nitrification ceases. This cannot happen with hydraulic control of sludge age, where a fixed proportion of the reactor volume is wasted daily to establish a constant sludge age. However, the secondary settling tank may become overloaded as the reactor TSS concentration increases with time, depending on the settleability with the AS (see Section 4.14.14). An operator therefore can choose the way an AS system fails with increasing organic loading – it does not have to be with nitrification, and so also with N removal.
4.14.20 Factors Influencing Nitrification From the discussion above, it can be seen that there are a number of factors that affect the nitrification process, the minimum sludge age required to achieve it, and the effluent FSA concentration from the AS system: 1. the magnitude of the kinetic constant mAm20 because this rate can vary considerably in different wastewaters; 2. temperature because it decreases the mAm20 rate and Kn20 coefficient; 3. unaerated zones in the reactor because ANOs are obligate aerobes and can grow only under aerobic conditions; 4. DO concentration because Monod kinetics presumes that FSA is the growth-limiting nutrient implying that the oxygen supply must be adequate; 5. cyclic flow and load conditions because FSA is dissolved and therefore the reactor (and effluent) FSA concentration is affected by the instantaneous actual hydraulic retention time; most FSA not nitrified during the actual hydraulic retention time escapes with the effluent; and 6. pH in the reactor because the mAm20 is strongly suppressed by pH outside the 7–8 range. These six factors are discussed further below.
4.14.20.1 Influent Source The maximum specific growth rate constant mAmT has been observed to be quite specific for the wastewater and also to vary between different batches of the same wastewater source. This specificity is so marked that mnmT should not be classified as a kinetic constant but rather as a wastewater characteristic. The effect appears to be of an inhibitory nature due to some substance(s) in the influent wastewater. It does not appear to
be a toxicity problem because a high efficiency of nitrification can be achieved even with a low mAmT value if the sludge age is increased sufficiently. These inhibitory substances are more likely to be present in municipal wastewater flows having some industrial contribution. In general, the higher the industrial contribution, the lower mAmT tends to be, but the specific chemical compounds that cause the reduction of mAmT have not been clearly defined. A standard temperature of 20 1C has been adopted for reporting mAm rates to take into account the effect of temperature. A range mAm20 values have been reported in the range of 0.30–0.75 d1 for municipal wastewaters. These two limits will have a significant effect on the magnitude of the minimum sludge age for nitrification. Two systems, having these respective mAm20 values, will have Rsm values differing by 250%. Clearly due to the link between the sludge age and mAmT, the latter’s value should always be estimated experimentally for optimal design. In the absence of such a measurement, a low value for mAmT necessarily will need to be selected to ensure that nitrification takes place. If the actual mAm is higher, the sludge age of the system will be longer and the reactor volume larger than necessary. However, the investment in the large reactor is not lost because in the future the plant will be able to treat a higher organic load at a shorter sludge age. Experimental procedures to determine mAm20 are given in the literature (e.g., WRC, 1984). The bn20 rate is taken as constant for all municipal wastewater flows at bn20 ¼ 0.04 d1. Its effect is small so that there is no need to enquire closely into all the factors affecting it. Little information on effects of inhibitory agents on KnT is available; very likely KnT will increase with inhibition.
4.14.20.2 Temperature The mAmT, KnT, and bAT constants are sensitive to temperature with a high-temperature sensitivity for the first two, while the endogenous rate is accepted to have the same low-temperature sensitivity as that for OHOs, viz.,
mAmT ¼ mAm20 ðyn ÞðT20Þ ðd1 Þ
ð131aÞ
KnT ¼ Kn20 ðyn ÞðT20Þ ðmgN l1 Þ
ð131bÞ
bAT ¼ bA20 ðyb ÞðT20Þ ðd1 Þ
ð131cÞ
where yn is the temperature sensitivity for nitrification ( ¼ 1.123) and yb the temperature sensitivity for endogenous respiration for ANOs ¼ 1.029. The effect of temperature on mAmT is particularly strong. For every 6 1C drop in temperature, the mAmT value halves which means that the minimum sludge age for nitrification doubles. Design of systems for nitrification, therefore, should be based on the minimum expected system temperature. The temperature sensitivity of KnT is also strong, doubling for every 6 1C increase in temperature. This does not affect the minimum sludge age for nitrification, but it does affect the effluent FSA concentration – the higher the Kn value, the higher the effluent FSA at Rs b Rsm. However, the faster mAmT rate at the higher temperature compensates for the higher KnT value so that the effluent FSA decreases with increase in temperature.
Biological Nutrient Removal
The effect of unaerated zones on nitrification can be formulated based on the following assumptions: 1. Nitrifiers, being obligate aerobes, grow only in the aerobic zones of a system. 2. Endogenous mass loss of the nitrifiers occurs under both aerobic and unaerated conditions. 3. The proportion of the ANOs in the VSS in the unaerated and aerated zones is the same so that the sludge mass fractions of the different zones of the system reflect the distribution of the nitrifier mass as well. From 1–3 above, it can be shown that if a fraction fxt of the total sludge mass is unaerated (i.e., (1 fxt) is aerated), the effluent ammonia is given by
KnT ðbAT þ 1=Rs Þ Nae ¼ mAmT ð1 f xt Þ ðbAT þ 1=Rs Þ
ð132Þ
Equation (132) is identical in structure to Equation (128), if one views the effect of the unaerated mass (fxt) as reducing the value of mAmT to mAmT(1 fxt), which conforms with (1) to (3) above. This sludge mass fraction approach is compatible with the nitrification kinetics in the AS kinetic models such as ASM1 and ASM2 (Henze et al., 1987, 1995) and UCTOLD and UCTPHO (Dold et al., 1991; Wentzel et al., 1992). In these models, nitrifier growth takes place only in the aerobic zone and endogenous respiration in all the zones. This sludge mass fraction approach is not compatible with the aerobic sludge age approach, which is used in some ND AS system design procedures (WEF, 1998; Metcalf and Eddy, 1991). In the aerobic sludge age approach, it is assumed that the growth and endogenous processes of the nitrifiers are active only in the aerobic zone, with neither processes active in the unaerated zone(s). This aerobic sludge age approach is not compatible with kinetic models and so significantly different predictions can be expected for the nitrification behavior from the aerobic sludge age-based design procedures and kinetic models. Following the same reasoning as that preceding Equation (132), it can be shown that the minimum sludge age for nitrification Rsm in an ND system having an unaerated mass fraction, fxt, is
Rsm ¼
1 mAmT ð1 f xt Þ bAT
ð133Þ
Alternatively, if Rs is specified, then the minimum aerobic sludge mass fraction (1 fxm) that must be present for nitrification to take place is found by substituting Rs for Rsm and fxm for fxt in Equation (133) and solving for (1 fxm), that is,
ð1 f xm Þ ¼ ðbAT þ 1=Rs Þ=mAmT
ð134Þ
or equivalently, from Equation (134), the maximum allowable unaerated sludge mass fraction at a sludge age of Rs is
f xm ¼ 1 ðbAT þ 1=Rs Þ=mAmT
ð135Þ
For a fixed sludge age, Rs, the design value for the minimum aerobic sludge mass fraction (1 fxm) should always be significantly higher than that given by Equation (134), because
nitrification becomes unstable and the effluent ammonia concentration increases when the aerated sludge mass fraction decreases to near the minimum value as given by Equation (134) in the same way as when the sludge age (Rs) approaches the minimum for nitrification (Rsm). This situation is exacerbated by cyclic flow and ammonia load conditions (see below). Consequently to ensure low effluent ammonia concentrations, the maximum specific growth rate of nitrifiers must be decreased by a factor of safety, Sf, to give the minimum design aerobic sludge mass fraction; from Equation (134),
ð1 f xm Þ ¼ ðbAT þ 1=Rs Þ=ðmAmT =Sf Þ
ð136aÞ
The corresponding maximum design unaerated sludge mass fraction, from Equation (136a), is
f xm ¼ 1 Sf ðbAT þ 1=Rs Þ=mAmT
ð136bÞ
With the aid of the temperature dependency equations for nitrification (Equation (131)), the maximum unaerated sludge mass fraction (fxm) from Equation (136b) is shown in Figure 25 for Sf ¼ 1.25 and mAm20 rates from 0.25 to 0.50 at 14 1C. This shows that fxm is very sensitive to mAmT. Unless a sufficiently large aerobic sludge mass fraction (1 fxm) is provided, nitrification will not take place and consequently nitrogen removal by denitrification is not possible. In fact, the selection of the maximum unaerated sludge mass fraction to achieve near complete nitrification and a required degree of N removal is the single most important decision that is made in the design of the BNR AS system because it defines the system sludge age and, for a selected reactor MLSS concentration, also the reactor volume. From Equations (132) and (136), it can be shown that at fxm for constant flow and ammonia load (i.e., steady-state conditions)
Nae ¼ KnT =ðSf 1Þ ðmgN l1 Þ
ð137Þ
From Equation (137), if Sf is selected at say 1.25 or greater at the minimum wastewater temperature, the effluent ammonia
0.80 Maximum unaerated sludge mass fraction
4.14.20.3 Unaerated Zones
467
Temperature = 14 °C
Factor of safety = 1.25
Recommended maximum 0.60 50
0.
40 36 0.
0.
0.40
30
0.
25
0.
0.20
Am20
0.00 0
10
20 30 Sludge age (days)
40
Figure 25 Maximum unaerated sludge mass fraction required to ensure nitrification vs. sludge age for maximum specific growth rates of nitrifiers mAm20 of 0.25–0.50 d1 at 14 1C for Sf ¼ 1.25.
468
Biological Nutrient Removal
concentration (Nae) will be lower than 2 mgFSA-N l1 at 14 1C for Kn20 ¼1.0 mgN l1. Although Kn is higher at higher temperature, Nae will decrease with increase in temperature because at constant sludge age, Sf increases with increase in mAmT. Consequently, for design the lower expected temperature should be selected to determine the sludge age and the aerobic mass fraction. If this is done, using say Sf ¼ 1.25, then it can be accepted from Equation (137) that the effluent ammonia concentration is below 2 mgN l1 at the lowest temperature and around 1 mgN l1 at 20 1C. In this way, explicitly calculating Nae with Equation (132) is not necessary because provision for near complete nitrification has been made by the selection of Sf. Clearly, selection of the mAm20 and Sf values has major consequences on the effluent FSA concentration and economics (size) of the ND AS system.
4.14.20.3.1 Maximum allowable unaerated mass fraction The above equations allow the two most important decisions in the design of an NDAS system to be made, the maximum unaerated sludge mass fraction and sludge age to ensure near complete nitrification. Evidently from Figure 25, for mAm20 4 0.50 the unaerated mass fraction at 14 1C can be as high as 0.7 at a sludge age of 40 days. Such a high unaerated mass fraction is apparently also acceptable at RsZ10 days at 20 1C. However, there are additional considerations that constrain the unaerated mass fraction – sludge age selection. 1. Experience with laboratory-scale ND (and NDBEPR) systems has shown that at unaerated mass fractions greater than 0.40, the filamentous bulking can become a problem, in particular at low temperatures (o16 1C). Systems with low unaerated mass fractions of o0.30 show greater tendency for good settling sludges (Musvoto et al., 1994; Ekama and Wentzel, 1999a; Tsai et al., 2003). 2. For design of BNR plants for high N and P removal, the unaerated sludge mass fraction fxm usually needs to be high (440%). If the mAm20 value is low (o 0.40 d1, which will be the usual case in designs where insufficient information on the mAm20 is available), the necessary high fxm magnitudes will be obtained only at long sludge ages (Figure 25). For example, if mAm20 ¼ 0.35 d1, then with Sf ¼ 1.3 at Tmin ¼ 14 1C, an fxm ¼ 0.45 (Equation (136b)) gives a sludge age of 25 days and for fxm ¼ 0.55 a sludge age of 37 days. Long sludge ages require large reactor volumes – increasing Rs from 25 to 37 days increases the reactor volume by 40%, whereas fxm increased only 22%. Also, for the same P content in the sludge mass, the P removal is reduced as the sludge age increases because the mass of sludge wasted daily decreases as the sludge age increases. Consequently, for low mAm20 values, the increase in N and P removal that can be obtained by increasing the unaerated sludge mass fraction above 0.50–0.60 might not be economical due to the large reactor volumes this will require, and might even be counterproductive insofar as it affects P removal. A sludge age of 30 days probably is near the limit of economic practicality which, for low mnm14 ¼ 0.16 values, will limit the unaerated mass fraction to about 0.5. At higher mnm14 values, the sludge ages allowing 50% unaerated mass fractions decrease significantly again indicating the advantages of determining
experimentally the value of mAm20 to check whether a higher value is acceptable. 3. An upper limit to the unaerated mass fraction is evident also from experimental and theoretical modeling of the BNR system. Experimentally at 20 1C with Rs ¼ 20 days, if fxm 40.70, the mass of sludge generated is found to increase sharply. Theoretically, this happens for fxm 4 0.60 at T ¼ 14 1C and Rs ¼ 20 days. The reason is that for such a high fxm, the exposure of the sludge to aerobic conditions becomes insufficient to utilize the adsorbed and enmeshed BPOs. This leads to a decrease in active mass and oxygen demand and a buildup of enmeshed nondegraded organics. When this happens, the system still functions in that the COD is removed from the wastewater, but the degradation of the COD is reduced; the system begins to behave as a contact reactor of a contact-stabilization system, that is, a bio-flocculation with minimal degradation. This critical state occurs at lower fxm as the temperature is decreased and the sludge age is reduced. From the above discussion, it would appear that the unaerated mass fraction should not be increased above an upper limit of about 60%, as indicated in Figure 25, unless there is a specific reason for this (Tsai et al., 2003).
4.14.20.4 DO Concentration High DO concentrations, up to 33 mg l1, do not appear to affect nitrification rates significantly. However, low oxygen concentrations reduce the nitrification rate. Stenstrom and Poduska (1980) have suggested formulating this effect as follows:
mAmO ¼ mAm
O ðd1 Þ KO þ O
ð138Þ
where O is oxygen concentration in liquid (mgO l1), KO the half-saturation constant (mgO l1), mAmo the maximum specific growth rate (d1), and mAO the specific growth rate at DO of O mg l1. The value of KO ranges from 0.3 to 2 mgO l1, that is, at DO values below KO the growth rate will decline to less than half the rate where oxygen is present in adequate concentrations. The wide range of KO probably has arisen because the concentration of DO in the bulk liquid is not necessarily the same as inside the biological floc where the oxygen consumption takes place. Consequently, the value will depend on the floc size, mixing intensity, and oxygen diffusion rate into the floc. Furthermore, in a full-scale reactor the DO will vary over the reactor volume due to the discrete points of oxygen input (with mechanical aeration) and the impossibility of achieving instantaneous and complete mixing. For these reasons, it is not really possible to establish a generally applicable minimum oxygen value – each reactor will have a value specific to the conditions prevailing in it. In nitrifying reactors with bubble aeration a popular DO lower limit, to ensure unimpeded nitrification, is 2 mgO l1 at the surface of the mixed liquor. Under cyclic flow and load conditions the difficulties of ensuring an oxygen supply matching the oxygen demand and a lower limit for the DO concentration are difficult.
Biological Nutrient Removal
4
2
3
2
1
0.0 0
0 0 (a)
Max. effl. FSA/steady-state FSA ratio
Amplitude of influent flow and FSA conc. 1.00 0.75 0.50 0.25
6
Steady-state FSA conc. (mgN I−1)
Max. effl. FSA/steady-state FSA ratio
4
8
1
2
3
4
5
T = 22 °C: Raw sewage
10
5
8
0.8
0.0 Amplitude of influent flow and FSA
6
0.6
1.00 4
0.4
0.75 0.50
2
0.2
0.25
0
0.0 2
6
R s /R sm ratio
1.0
Steady-state effluent FSA
(b)
Steady-state FSA conc. (mgN I−1)
T = 14 °C: Raw sewage Steady-state effluent FSA
10
469
4
6
8
10
12
14
16
R s /R sm ratio
Figure 26 (a) Maximum to steady-state effluent FSA concentration ratio vs. sludge age to minimum sludge age for nitrification ratio for influent flow and ammonia concentration amplitude (in phase) of 0.0 (steady state) 0.25, 0.50, 0.75, and 1.0 at 14 1C. (b) Maximum to steady-state effluent FSA concentration ratio vs. sludge age to minimum sludge age for nitrification ratio for influent flow and ammonia concentration amplitude (in phase) of 0.0 (steady-state) 0.25, 0.50, 0.75, and 1.0 at 22 1C.
Where storm flows are not of long duration, flow equalization is a practical way to facilitate control of the DO concentration in the reactor. In fact, most of the diurnal variations in reactor dissolved concentrations are a direct consequence of diurnal flow variation – negligibly little is due to the kinetic rates of the biological processes, especially at long sludge ages. In the absence of flow equalization, amelioration of the adverse effects of low DO concentration during peak oxygen demand periods occurs by increasing the sludge age to significantly longer than the minimum necessary for nitrification, that is, by effectively increasing Sf.
4.14.20.5 Cyclic Flow and Load It is well known both experimentally and theoretically with simulation models that under cyclic flow and load conditions the nitrification efficiency of the AS system decreases compared with that under steady-state conditions. From simulation studies, during the high flow and/or load period, even though the nitrifiers are operating at their maximum rate, it is not possible to oxidize all the ammonia available, and an increased ammonia concentration is discharged in the effluent. This in turn reduces the mass of nitrifiers formed in the system. Equivalently, the effect of diurnal variation in flow and load is to reduce the system sludge age. The average effluent ammonia concentration from a system under cyclic flow and load conditions is therefore higher than that from the same system under constant flow and load (steady-state conditions). The adverse effect of the diurnal flow variation becomes more marked as the fractional amplitude of the flow and load variation increase and is ameliorated as the safety factor Sf increases. Simulation studies of the diurnal flow effect show a relatively consistent trend between the maximum or average effluent FSA concentrations under diurnal conditions and the steady-state effluent FSA concentration versus the ratio of system sludge age and the minimum sludge age for nitrification (Rs/Rsm). For mAm20 ¼ 0.45 d1 (other constants in
Table 10), Figures 26(a) (for 14 1C) and 26(b) (for 22 1C) show the maximum (average not shown) effluent FSA concentration as a ratio of the steady-state effluent FSA concentration versus the system sludge age as a ratio of the minimum sludge age for nitrification (Rs/Rsm) for a single reactor fully aerobic system receiving cyclic influent flow and FSA load as in-phase sinusoidally varying flow and ammonia concentration, both with amplitudes of 0.25, 0.50, 0.75, 1.00, and 0.0 (steady state). For example, at 14 1C (Figure 26(a)) if the system sludge age is 2 times the minimum for nitrification, the maximum effluent FSA concentration is 8 times the steadystate value. From Figure 26(a), the latter is 0.8 mgN l1 so the maximum is 8 0.8 ¼ 6.4 mgN l1. From Figures 26(a) and 26(b), clearly the greater the diurnal flow variation and the lower the temperature, the higher the maximum (and average) effluent ammonia concentrations. This can be compensated for by increasing Sf, which has the effect of increasing the sludge age or decreasing the unaerated mass fraction of the system. This obviously has an impact on the effluent quality and/or economics of the system. The importance of the selection of mAm cannot be overemphasized. If the value of mAm is selected higher than the actual value, even with a safety factor Sf of 1.25–1.35, the plant is likely to produce a fluctuating effluent ammonia concentration, with reduced mean efficiency in nitrification. Hence, conservative estimates of mAm (low) and Sf (high) are essential for ensuring nitrification and low effluent ammonia concentration.
4.14.20.6 pH and Alkalinity The mAm rate is very sensitive to the pH of the mixed liquor outside the 7–8 range. It seems that the free ammonia (NH3) and nitrous acid (HNO2) act inhibitorily when their respective concentrations increase too high. This happens when the pH increases above 8.5 (increasing (NH3)) or decreases below 7
470
Biological Nutrient Removal
(increasing (HNO2)); optimal nitrification rates are expected for 7opHo8.5 with sharp declines outside this range. From the overall stoichiometric equations for nitrification (Equation (118a)), nitrification releases hydrogen ions which in turn decreases H2CO3* Alkalinity of the mixed liquor. For every 1 mgFSA that is nitrified 2 50/14 ¼ 7.14 mg Alkalinity (as CaCO3) is consumed. Based on equilibrium chemistry of the carbonate system (Loewenthal and Marais, 1977), equations linking the pH with H2CO3* Alkalinity for any dissolved carbon dioxide concentration can be developed. These relationships are plotted in Figure 27. When the H2CO3* Alkalinity falls below about 50 mg l1 as CaCO3 then, irrespective of the carbon dioxide concentration, the pH becomes unstable and decreases to low values. Generally, if nitrification causes the H2CO3*Alkalinity to drop below about 50 mg l1 (as CaCO3), problems associated with low pH will arise at a plant, such as poor nitrification efficiency, effluents aggressive to concrete, and the possibility of development of bulking (poor settling) sludges (Jenkins et al., 1993). For any particular wastewater, the effect of nitrification on pH can be readily assessed, as follows: for example if a wastewater has a H2CO3*Alkalinity of 200 mg l1 as CaCO3 and the expected production of nitrate is 24 mgN l1, then the expected H2CO3*Alkalinity in the effluent will be (200 7.14 24) ¼ 29 mg l1 as CaCO3. From Figure 27, such an effluent will have a pH o7.0. Wastewaters having low Alkalinity (capital A denotes H2CO3* Alkalinity) are often encountered where the municipal supply is drawn from areas underlain with sandstone. A practical approach to treating such wastewaters is to (1) dose lime or better (2) create an anoxic zone(s) to denitrify some or all of the nitrate generated. In contrast to nitrification, denitrification takes up hydrogen ions which is equivalent to generating Alkalinity (see Section 4.14.24.2). By considering nitrate as electron acceptor, it can be shown that for every milligram of nitrate denitrified, there is an increase of 1 50/ 14 ¼ 3.57 mg Alkalinity as CaCO3. Hence, incorporating denitrification in a nitrification system causes the net loss of
10 0.5 1.0 2.0 5.0 10.0
Mixed liquor pH value
8 Carbon dioxide concentration (mg I−1 as CaCO3)
6
Saturation ~ 0.5 mg I−1 as CaCO3
4
Alkalinity to be reduced usually sufficiently to maintain the Alkalinity above 50 mg l1 as CaCO3 and consequently the pH above 7. In the example above, where the Alkalinity in the system is expected to decline to 29 mg l1 as CaCO3, if 50% of the nitrate were denitrified, the gain in Alkalinity would be (0.5 24 3.57) ¼ 43 mg l1 as CaCO3 and will result in an Alkalinity of (29 þ 43) ¼ 72 mg l1 as CaCO3 in the system. In this event the pH will remain above 7. For low Alkalinity wastewaters, it is imperative, therefore, that denitrification be built into nitrifying plants, even if N removal is not required. Incorporation of unaerated zones in the system influences the sludge age of the system at which nitrification takes place so that cognizance must be taken of the effect of an anoxic or unaerated zone in establishing the sludge age of a nitrifying– denitrifying plant (see Section 4.14.20.3). In the AS systems treating reasonably well buffered wastewaters, quantifying the effect of pH on nitrification is not critical because pH reduction can be limited or completely obviated by including anoxic zones, thereby ensuring Alkalinity recovery via denitrification. However, in poorly buffered wastewaters, or wastewaters with high influent N (such as AD liquors), the interaction between the biological processes, pH, and nitrification is the single most important one for the N removal AS system. Hence, it is essential to include the effect of pH on the nitrification rate for such wastewaters to quantify this important interaction. From Equation (121), the specific growth rate of the ANOs (mA) is a function of both mAm and Kn. It was shown above that the minimum sludge age is dominated by the magnitude of mAmT; it is only very weakly influenced by KnT. At RscRsm, the effluent ammonia concentration (Nae), although low, is, relatively speaking, significantly higher for larger KnT values: for example, if KnT increases by a factor of 2, the effluent ammonia concentration will increase correspondingly by the same factor (Equation (132)). Consequently, the value of KnT is significant insofar as it governs the effluent ammonia concentration once nitrification takes place at RscRsm. Several investigations have been made to understand the effect of pH on mAmT. These investigations generally have not separated out the effect of pH on mAmT and KnT so that most data are in effect lumped parameter estimates of mAmT. Almost no information is available on the effect of pH on KnT by itself. Quantitative modeling of the effect of pH on mAm has been hampered by the difficulty of accurately measuring the effects of pH on nitrification. Studies have shown that mAm can be expressed as a percentage of the highest value at optimum pH. Accepting this approach and that mAm is highest and remains approximately constant in the pH range for 7.2opHo8.0 but decreases as the pH decreases below 7.2 (Downing et al., 1964; Loveless and Painter, 1968), So¨temann et al. (2005a) modeled the mA pH dependency as For 5opHo7.2,
2
mAmpH ¼ mAm7:2 yns ðpH7:2Þ 0 −100
0
100
200 −1
Alkalinity (mg I
300
as CaCO3)
Figure 27 Mixed liquor pH vs. H2CO3* alkalinity for different concentrations of carbon dioxide.
400
ð139aÞ
where yns is the pH sensitivity coefficient (E2.35). Declining mAm values at pH48.0 have been observed and it has been noted that nitrification effectively ceases at a pH of about 9.5 (Malan and Gouws, 1966; Wild et al., 1971; Antoniou et al., 1990). Accordingly, for pH47.2, So¨temann et al.
Biological Nutrient Removal
(2005a) proposed Equation (139b) to model the decline in the mAm from pH 47.2 to 9.5 as a function of mAm7.2 using inhibition kinetics as follows:
mAmpH ¼ mAm7:2 KI
Kmax pH Kmax þ KII pH
ð139bÞ
where KI ¼ 1.13, Kmax ¼ 9.5, KIIE0.3. The overall effect of pH on mAm is modeled by combining Equations (139a) and (139b), which is given by Equation (139c) and shown in Figure 28. It can be seen that in the range pH ¼ 7.2–8.3, the change in mAmpH is small, with mAmpH/mAm7.2 40.9:
mAmpH ¼ mAm7:2 2:35ðpH7:2Þ KI
Kmax pH Kmax þ KII pH
ð139cÞ
where 2.35(pH7.2) is set ¼ 1 for pH47.2,
KI
Kmax pH ¼1 Kmax þ KII pH
for pH o7.2 and mAmpH ¼ 0 for pH49.5. Experimental data from the literature are also shown in Figure 28 to provide some quantitative support for Equation (139c). At low pH (o7.2), data from Wild et al. (1971) and Antoniou et al. (1990) fit the equation reasonably well. Very few data are available for pH48.5, but the few points from Antoniou et al. (1990) show reasonable agreement with Equation (139c). Accordingly, Equation (139c) was accepted to calculate mAmpH in the pH range 5.5–9.5. From Equation (139c), the minimum sludge age for nitrification (Rsm) at different pH and temperature (T) and unaerated mass fraction (fxm) is given by
Rsm ¼ 1=½mApHT ð1 f xm Þ bnT
ðdaysÞ
ð140Þ
The problem with nitrification in low alkalinity wastewater is that the pH obtained is not known, because it is interactively 1.2
Fraction Unm/Umm7.2
1 Eq (139b) 0.8 0.6 0.4 Eq (139a) Eq (139b) 0.2 0 4
5
6
7
8
9
10
471
established between the degree of nitrification, loss of alkalinity, pH, and mApHT. To investigate this interaction, the biological kinetic ASM1 model for carbon (C) and nitrogen (N) removal was integrated by So¨temann et al. (2005a) with a two-phase (aqueous-gas) mixed weak acid/base chemistry kinetic model to extend application of ASM1 to situations where an estimate for pH in the biological reactor is important. This integration, which included CO2 (and N2) gas generation by the biological processes and their stripping by aeration, made a number of additions to ASM1, inter alia the above effect of pH on the autotrophic nitrifiers (ANOs). From simulation of a long sludge age ND AS system with incrementally decreasing influent H2CO3* Alkalinity, when the effluent H2CO3* alkalinity fell below about 50 mg l1 as CaCO3, the aerobic reactor pH dropped below 6.3, which severely retarded nitrification and caused the minimum sludge age for nitrification (Rsm) to increase up to the operating sludge age of the system. The simulation confirmed the earlier conclusion that when treating low H2CO3* alkalinity wastewater (1) the minimum sludge age for nitrification (Rsm) varies with temperature and reactor pH and (2) for low effluent H2CO3* alkalinity (o50 mg l1 as CaCO3), nitrification becomes unstable and sensitive to dynamic loading conditions resulting in increases in effluent ammonia concentration, reduced nitrification efficiency, and as a result lower N removal. For effluent H2CO3* alkalinity o50 mg l1, lime should be dosed to the influent to raise the aerobic reactor pH and stabilize nitrification and N removal.
4.14.21 Nutrient Requirements for Sludge Production All live biological material and some unbiodegradable organic compounds contain nitrogen (N) and phosphorus (P). The organic sludge mass (VSS) that accumulates in the biological reactor comprises active organisms (XBH), endogenous residue (XEH), and UPOs (XI), each of which contains N and P. From TKN and VSS tests conducted on AS, it has been found that the N content (as N with respect to VSS, fn, mgN/mgVSS) ranges between 0.09 and 0.12 with an average of about 0.10 mgN/ mgVSS. Similarly, from total P and VSS tests, the P content (as P with respect to VSS, fp, mgP/mgVSS) of AS in purely aerobic and anoxic aerobic systems ranges between 0.01 and 0.03 with an average of about 0.025 mgP/mgVSS. From the steady-state model, the relative proportions of the active organisms (XBH), endogenous residue (XEH), and UPOs (XI) change with sludge age. Yet, it has been found that the fn value of the VSS is relatively constant at 0.10 mgN/ mgVSS. This indicates that the N content of the active organisms (XBH), endogenous residue (XEH), and UPOs (XI) is closely the same; if they were significantly different, it would be observed that fn changes in a consistent manner with sludge age. Likewise, for fully aerobic systems, the P content of the three constituents of AS is approximately similar at 0.025 mgP/mgVSS.
pH Figure 28 Maximum specific growth rate of nitrifiers, as a fraction of the rate at pH 7.2, vs. pH of the mixed liquor. (F), Model; (), Malan and Gouws (1966); ( ), Downing et al. (1964); ( ), Wild et al (1971); and (m), Antoniou et al. (1990).
4.14.21.1 Nitrogen Requirements The mass of N (or P) incorporated into the sludge mass is calculated from a N balance over the completely mixed AS
472
Biological Nutrient Removal
system (Figure 2) under steady-state daily conditions, viz., TKN flux out ¼ TKN flux in TKN flux in ¼ Qi Nti (mgN d1) TKN flux out ¼ TKN flux in Qe and Qw
Noting that Qw þ Qe ¼ Qi and Qw ¼ Vp/Rs yields
Qi Nte ¼ Qi Nti f n Xv Vp =Rs from which
Nte ¼ Nti f n MXv =ðRs Qi Þ ðmgN l1 Þ
ð141Þ
where Nte is the effluent TKN concentration (mgN l1). The term fnMXv/(RsQi) is denoted Ns and is the concentration of influent TKN in mgN l1 that is incorporated into sludge mass and removed from the system bound in the particulate sludge mass in the waste flow (Qw): 1
Ns ¼ f n MXv =ðRs Qi Þ ðmgN l
influentÞ
ðmgN l1 Þ
ð143Þ
From Equation (141), under daily average conditions, the concentration of N per liter influent required for incorporation into sludge mass is equal to the N content of the mass of sludge (VSS) wasted per day divided by the influent flow. Substituting Equation (106) relating the mass of sludge (VSS) in the reactor (MXv) to the daily average organic load on the reactor (FSti), cancelling Qi and dividing by Sti yields the concentration of N required per liter influent for sludge production per mgCOD/l organic load on the reactor, viz.,
ð1 f S0 us f S0 up ÞYH f S0 up Ns ¼ fn ð1 þ f EH bH Rs Þ þ Sti ð1 þ bH Rs Þ f cv ðmgN=mgCODÞ
Nae ¼ Nai þ Nobsi þ Nobpi ðNs Noupi Þ ðmgN l1 Þ
ð142Þ
From the N mass balance, this Ns concentration does not include the N in dissolved form in the waste flow. The soluble TKN concentration in the waste flow is the same as the effluent TKN concentration, Nte, which is soluble N in the form of ammonia (Nae) and unbiodegradable soluble organic N (Nouse). Therefore, from Equation (141), provided nitrifiers are not supported in the AS reactor so that nitrification of ammonia to nitrate does not take place, the effluent TKN concentration Nte is given by
Nte ¼ Nti Ns
organics (Nobsi and Nobpi) is released as FSA when these organics are broken down. This FSA adds to the FSA in the reactor from the influent. Some of the FSA in the reactor is taken up by the OHOs to form new OHO biomass. Some of the OHO biomass in the reactor is lost via the endogenous respiration process. The N associated with the biodegradable part of the OHO biomass is released back to the FSA pool in the reactor but the N in the unbiodegradable endogenous residue part remains as organic N bound in the endogenous residue VSS. Due to these interactions, it is possible that the effluent FSA concentration from a non-nitrifying AS system is higher than the influent FSA concentration – this occurs when the influent TKN comprises a high biodegradable organic N fraction. If the conditions are favorable for nitrification, the net FSA concentration in the reactor is available for the ANOs for growth with the associated generation of nitrate. Unless taken up for OHO growth or nitrified, the FSA remains as such and exits the system with the effluent. So in the absence of nitrification, the effluent ammonia concentration Nae is given by
ð144Þ
The influent TKN comprises ammonia and N bound in organic compounds of a soluble and particulate and biodegradable and unbiodegradable nature. The unbiodegradable organics, some of which contain organic N, are not degraded in the AS system. The influent unbiodegradable soluble organic N (Nousi) exits the system with the effluent (and waste flow) streams. The UPOs are enmeshed with the sludge mass in the reactor and so the organic N associated with these organics exits the system via the daily waste sludge (VSS) harvested from the system. The N bound in the biodegradable
ð145Þ and the effluent TKN (Nte) concentration by
Nte ¼ Nouse þ Nae
ðmgN l1 Þ
ð146Þ
The same approach is applied for the phosphorus (P) requirement for sludge production. Accepting that the P content of the AS in the fully aerobic system without BEPR is 0.025 mgP/mgVSS, the effluent total P (TP) concentration Pte is given by
Pte ¼ Pti Ps
ðmgP l1 Þ
ð147Þ
where
Ps MXv f p Ns ¼ fp ¼ Sti Rs Qi f n Sti
ðmgP l1 influentÞ
ð148Þ
4.14.21.2 N (and P) Removal by Sludge Production A plot of Equations (144) and (148) versus sludge age is given in Figure 29 for fn ¼ 0.10 mgN/mgVSS, fp ¼ 0.025 mgP/mgVSS for the example raw and settled wastewaters. It is evident that higher concentrations of TKN and TP are required for sludge production for raw than for settled wastewaters. This is because greater quantities of sludge are produced per mgCOD organic load on the reactor at the same sludge age when treating raw wastewaters (see Section 4.14.13). Also, the N and P requirements decrease as the sludge age increases because net sludge production decreases as sludge age increases. Generally, for sludge ages greater than about 10 days, the N removal from the wastewater attributable to net sludge production is less than 0.025 mgN/mgCOD load on the reactor. As influent TKN/COD ratios for domestic wastewater are in the approximate range 0.07–0.13 (Figure 29), it is clear that only a minor fraction of the influent TKN (A in Figure 29) is removed by incorporation into sludge mass. Additional N removal (B in Figure 29) is obtained by transferring the N from the dissolved form in the liquid phase to the gas phase
Biological Nutrient Removal Nutrient requirements 0.035 Approximate range of influent TKN/COD and P/COD ratios of municipal wastewaters
0.12 0.10
0.030 0.025
0.08
0.020
0.06
0.015 B
0.04
0.010 Raw
0.02
0.005 A
P requirement (mgP/mgCOD)
N requirement (mgN/mgCOD)
0.14
Settled
0.00
0.000 0
5
10 15 20 Sludge age (days)
25
30
Figure 29 Approximate minimum nutrient N and P requirements as mgN l1 influent TKN and mgP l1 influent total P per mgCOD l1 organic load on the activated sludge reactor vs. sludge age for the example raw and settled wastewaters at 20 1C together with influent TKN and TP to COD concentration ratio ranges for municipal wastewater.
by autotrophic nitrification and heterotrophic denitrification, which transforms the nitrate to nitrogen gas in anoxic (nonaerated) reactor(s). The details of heterotrophic denitrification are presented below. From Figure 29, normal P removal by incorporation into biological sludge mass is limited at about 0.006 and 0.004 mgP/mgCOD for raw and settled wastewaters respectively, effecting a TP removal of about 20–25% from average municipal wastewaters. As transformation of dissolved orthoP to a gaseous form is not possible, to increase the P removal from the liquid phase, additional ortho-P needs to be incorporated into the sludge mass. This can be achieved in two ways: (1) chemically and/or (2) biologically. With chemical P removal, iron or aluminum chlorides or sulfates are dosed to the influent (pre-precipitation), to the AS reactor (simultaneous precipitation) or to the final effluent (post-precipitation). The disadvantage of chemical P removal is that it significantly increases (1) the salinity of treated wastewater, (2) the sludge production due to the inorganic solids formed, and (3) the complexity and cost of the WWTP. With biological P removal, the environmental conditions in the biological reactor are designed in such a way that a specific group of heterotrophic organisms (called PAOs) grow in the AS reactor. With the accumulated polyPs, these organisms have a much higher P content than the OHOs, as high as 0.38 mgP/ mgPAOVSS (Wentzel et al., 1990). The more PAOs that grow in the reactor, the higher will be the mean P content of the VSS sludge mass in the reactor and therefore the higher the P removal via the wasted sludge. With a significant mass of PAOs present, the mean P content of the VSS sludge mass can increase from 0.025 mgP/mgVSS in aerobic systems to 0.10– 0.15 mgP/mgVSS in biological N and P removal systems. The advantage of biological P removal over chemical P removal is that (1) the salinity of the treated wastewater is not increased, (2) sludge production is increased only between 10% and 15%, and (3) the system is less complex and costly to operate.
473
A disadvantage of biological P removal is that, being biological, it is less dependable and more variable than chemical P removal. The biological processes which mediate biological N and P removal in AS systems and the different reactor configurations in which these take place are described in Section 4.14.28.
4.14.22 Nitrification Design Considerations The kinetic equations describing the interactions between the FSA and the organic N are complex and have been developed in terms of the growth–death–regeneration approach in AS simulation models such as ASM1 and ASM2. However, for steady-state conditions assuming (1) all the biodegradable organics are utilized in the reactor and (2) a TKN mass balance over the AS system, a simple steady-state nitrification model can be developed from the nitrification kinetics and the N requirements for sludge production considered above. This model is adequate for steady-state design and from it some general response graphs are developed below for the example raw and settled wastewaters. Dynamic system responses can be determined with the simulation models once (1) the AS system has been designed and sludge age, zone and reactor volumes and recycle flows are known and (2) the steady-state concentrations have been calculated to serve as initial conditions for the simulation. In the nitrifying AS system design, the (1) effluent FSA, TKN, and nitrate concentrations and (2) the nitrification oxygen demand need to be calculated.
4.14.22.1 Effluent TKN The filtered effluent TKN (Nte) comprises the FSA (Nae) and the unbiodegradable soluble organic N (Nouse). Once mAm20, fxt, Rs, and Sf have been selected, the equations for these concentrations are: 1. Effluent FSA (Nae). Nae is given by Equation (132), which applies only if Rs4Rsm, which will be the case for Sf41.0. 2. Effluent soluble biodegradable organic nitrogen concentration (Nobse). The biodegradable organics (both soluble and particulate) are broken down by the OHOs releasing the organically bond N as FSA. In the steady-state model, it is assumed that all the biodegradable organics are utilized. Hence, the effluent soluble biodegradable organic N concentration (Nobse) is zero. 3. Effluent soluble unbiodegradable organic nitrogen concentration (Nouse). Being unbiodegradable, this concentration of organic N flows though the AS system with the result that the effluent concentration (Nouse) is equal to the influent concentration (Nousi), that is,
Nouse ¼ Nousi
ð149Þ
where Nousi is the influent soluble unbiodegradable organic nitrogen, mgOrgN-N l1 ¼ fN0 ous Nti, where fN0 ous is the soluble unbiodegradable organic N fraction of the influent TKN (Nti). The two nonzero effluent TKN concentrations (FSA, Nae and OrgN, Nouse) are soluble and so exit with the effluent (and
474
Biological Nutrient Removal
waste flow). The soluble (filtered) TKN in the effluent (Nte) is given by their sum, that is,
Nte ¼ Nae þ Nousi
ðfiltered TKNÞ
ð150Þ
If the effluent sample is not filtered, the effluent TKN will be higher by the concentration of TKN in the effluent VSS, that is,
Nte ¼ Nae þ Nouse þ f n Xve
ðunfiltered TKNÞ
nitrogen required for sludge production per mgCOD applied (from Equation (144)). The nitrification capacity to influent COD concentration ratio (Nc/Sti) of a system can be estimated approximately by evaluating each of the terms in Equation (153) as follows:
•
ð151Þ
where Xve is the effluent VSS concentration (mgVSS l1) and fn the N content of VSS (B0.1 mgOrgN-N/mgVSS).
•
4.14.22.2 Nitrification Capacity From a TKN mass balance over the AS system and Rs 4 Rsm, the concentration of nitrate generated in the system (Nne) with respect to the influent flow is given by the influent TKN (Nti) minus the soluble effluent TKN (Nte) and the concentration of influent TKN incorporated in the sludge wasted daily from the AS system (Ns), that is,
Nne ¼ Nc ¼ Nti Nte Ns
ð152Þ
The Ns concentration is determined from the mass of N incorporated in the VSS mass harvested from the reactor per day (Equation (142)). The mass of VSS in the reactor (MXv) does not have to include the VSS mass of nitrifiers because this mass, as mentioned earlier, is negligible (o2–4%). In Equation (152), Nc defines the ‘nitrification capacity’ of the AS system. The nitrification capacity (Nc) is the mass of nitrate produced by nitrification per unit average influent flow, that is, mgNO3-N l1. In Equation (150), the effluent TKN concentration (Nte) depends on the efficiency of nitrification. In the calculation for the maximum unaerated sludge mass fraction (fxm) at a selected sludge age, if the factor of safety (Sf) was selected 41.25 to 1.35 at the lowest expected temperature (Tmin), the efficiency of nitrification be high (495%) and Nae generally will be less than 1–2 mgN l1. Also, with Sf 41.25 at Tmin, Nae will be virtually independent of both the system configuration and the subdivision of the sludge mass into aerated and unaerated mass fractions. Consequently, for design, with Sf41.25, Nte will be around 3–4 mgN l1 provided that there is reasonable assurance that the actual mAm20 value will not be less than the value accepted for design and that there is sufficient aeration capacity so that nitrification is not inhibited by an insufficient oxygen supply. Accepting the calculated fxm and selected sludge age (Rs) at the lower temperature, then at higher temperatures the nitrification efficiency and the factor of safety (Sf) both will increase so that at summer temperatures (Tmax), Nte will be lower, approximately 2–3 mgN l1. Dividing Equation (152) by the total influent COD concentration (Sti) yields the nitrification capacity per mgCOD applied to the biological reactor, Nc/Sti, viz.,
Nc =Sti ¼ Nti =Sti Nte =Sti Ns =Sti
ð153Þ
where Nc/Sti is the nitrification capacity per mgCOD applied to the AS system (mgN/mgCOD), Nti/Sti the influent TKN/ COD concentration ratio of the wastewater, and Ns/Sti the
•
Nti/Sti: This ratio is a wastewater characteristic and obtained from the measured influent TKN and COD concentrations – it can range from 0.07 to 0.10 for raw municipal wastewater and 0.10 to 0.14 for settled wastewater. Nte/Sti: Provided the constraint for efficient nitrification is satisfied at the lowest temperature (Tmin), the effluent TKN at Tmin (Nte) will be low at B2–3 mgN l1, that is, for influent COD concentrations (Sti) ranging from 1000 to 500, Nte/Sti will range from 0.005 to 0.010. At Tmax, NteE1– 2 mgN l1 making the Nte/Sti ratio lower. Ns/Sti: Given by Equation (144).
A graphical representation of the relative importance of these three ratios to the nitrification capacity, Nc/Sti, is shown in Figure 30(a) (for 14 1C) and 30(b) (for 22 1C) and were generated by plotting Nc/Sti versus sludge age for selected influent TKN/COD (Nti/Sti) ratios of 0.07, 0.08, and 0.09 for the example raw wastewater and settled wastewater for 40% COD and 15% TKN removal in primary settling, viz., 0.113, 0.127, and 0.141. Also shown are the minimum sludge ages for nitrification at unaerated sludge mass fractions of 0.0, 0.2, 0.4, and 0.6 for the example mAm20 value of 0.45 d1. For a particular unaerated sludge mass fraction, the plotted values of Nc/Sti are valid only at sludge ages longer than the corresponding minimum sludge age. These figures show the relative magnitudes of the three terms that affect the nitrification capacity versus sludge age and temperature. 1. Temperature. To obtain complete nitrification at 14 1C (for a selected fxm), the sludge age required is more than double that at 22 1C. The corresponding nitrification capacities per influent COD at 14 1C show a marginal reduction to those at 22 1C, because sludge production at 14 1C is slightly higher than at 22 1C due to the reduction in endogenous respiration rate of the OHOs. 2. Sludge age. For a selected influent TKN/COD ratio (Nti/Sti), the nitrification capacity (Nc/Sti) increases as the sludge age increases because the N required for sludge production decreases with sludge age, making more FSA available for nitrification. However, the increase is marginal for Rs410 days. 3. Influent TKN/COD ratio (Nti/Sti). Clearly, for both raw and settled wastewater, at any selected sludge age, the nitrification capacity (Nc/Sti) is very sensitive to the influent TKN/COD ratio (Nti/Sti). An increase of 0.01 in Nti/Sti causes equal increase of 0.01 in Nc/Sti. For the same Nti/Sti ratio for raw or settled wastewater, the nitrification capacity (Nc/Sti) for raw wastewater is lower than for settled wastewater because more sludge (VSS) is produced per unit COD load from raw wastewater than from settled wastewater because the unbiodegradable particulate COD fraction (fS’up) in raw water is higher than in settled wastewater. Apart from this difference, an increase in influent TKN/COD ratio will result in an equal increase in nitrate concentration (nitrification capacity) per influent
Biological Nutrient Removal
WW Char fS’us Raw 0.07 Settled 0.12
0.10
0.05 Raw wastewater
0.15
0.0 0.2 0.4 0.6 −1 Unaerated mass fraction 14 °C; UA20 = 45 d bA20 = 0.04 d−1; Sf = 1.25
0.00 0 (a)
fS’up Settled wastewater 0.15 0.141 0.04 0.127 0.113 TKN/COD ratio 0.10 0.09 0.08
Nitrification capacity
Nitrification capacity
0.15
5
20 10 15 Sludge age (days)
25
fs’us 0.07 0.12
fs’up 0.15 0.04
Settled wastewater 0.141 0.127
0.10
0.113 TKN/COD ratio 0.10 0.09 0.08
0.05 Raw wastewater
0.00
30
WW Char Raw Settled
475
0.2 0.6 22 °C; UA20 = 0.45 d−1 0.0 0.4 −1 b Unaerated mass fraction A20 = 0.04 d ; Sf = 1.25 0
5
(b)
10 20 15 Sludge age (days)
25
30
Figure 30 Nitrification capacity per mgCOD applied to the biological reactor vs. sludge age for different influent TKN/COD concentration ratios in the example raw and settled wastewaters at 14 1C (a) and 22 1C (b). Also shown as vertical lines are the minimum sludge ages required to achieve nitrification for Sf ¼ 1.25 for unaerated sludge mass fractions of 0.0, 0.2, 0.4, and 0.6.
COD. This decreases the likelihood, or makes it impossible, to obtain complete denitrification using the wastewater organics as electron donor. This will become clear when denitrification is considered below. Because primary settling increases the influent TKN/COD ratio, N removal via nitrification denitrification is always lower with settled wastewater than with raw wastewater. However, this lower N removal comes with the advantage of a smaller biological reactor and lower oxygen demand resulting significant savings in reactor and oxygenation costs.
4.14.22.3 Mass of Nitrifiers (MXA) and Nitrification Oxygen Demand (FOn) Once nitrification takes place because the sludge age of the system is longer than the minimum required for nitrification, the mass of nitrifiers (MXA, mgVSS) in the reactor is calculated from the flux of nitrate generated (FNne) in the same way as the mass of OHOs (MXBH) was calculated from the flux of biodegradable organics (Equation (86)), viz.,
MXBA ¼ FNne YA Rs =ð1 þ bAT Rs Þ ðmgVSSÞ
ð154Þ
where FNne is the flux of nitrate generated ¼ (Qe þ Qw)Nne ¼ Qi Nne (mgN d1) and Nne is given by Equation (152). The oxygen demand for nitrification is simply 4.57 mgO/ mgN times the flux of nitrate produced, that is,
FOn ¼ 4:57 FNne ¼ OURn Vp
ðmgO d1 Þ
ð155Þ
Table 11 Raw and settled wastewater characteristics required for calculating effluent N concentrations from nitrification AS systems Influent WW characteristic
Sym
Raw
Influent TKN (mgN l1) Influent TKN/COD ratio Influent FSA fraction Unbio sol orgN fraction Unbio partic VSS N content
Nti fns f N0 a fN0 ous fn
60 0.08 0.75 0.03 0.1
Influent pH Influent Alk mg l1 as CaCO3 ANO max spec growth rate Influent flow rate (M l d1)
Alk mAm20 Qi
7.5 200 0.45 15
Seta 51 0.113 0.88 0.034 0.1 7.5 200 0.45 15
a
Settled wastewater (WW) characteristics must be selected/calculated to be consistent with the raw wastewater ones and mass balances over the primary settling tanks, e.g., soluble concentrations must be the same in settled wastewater as in raw wastewater.
organics (COD) removal (see Section 4.14.9.5). The wastewater characteristics for the raw and settled wastewaters for COD removal are listed in Table 7 and the additional characteristics required for nitrification are listed in Table 11. The nitrifier kinetic constants in Table 10, adjusted for wastewater temperatures 14 and 22 1C, were applied. No adjustment to mAm20 for pH was made, that is, an effluent Alkalinity 450 mg l1 as CaCO3 was assumed. Also, it is accepted that all the biodegradable organics are degraded and their N content released as ammonia so the effluent soluble biodegradable organic N concentration (Nobse) is zero.
4.14.23.2 Nitrification Process Behavior
4.14.23 Nitrification Design Example 4.14.23.1 Wastewater Characteristics Design of a nitrification AS system without denitrification is considered below. For the purpose of comparison, the nitrifying AS system is designed for the same wastewater flow and characteristics accepted for the design of the AS system for
From Equation (20a), the unbiodegradable soluble organic nitrogen in the effluent is Nouse ¼ Nousi ¼ 1.8 mgN l1 for raw and settled wastewater The ammonia concentration available for nitrification (Nan) is the influent TKN concentration (Nti) minus the N concentration required for sludge production (Ns) (Equation (142)) and the USO N concentration in the effluent
476
Biological Nutrient Removal
(Nouse), viz.,
Nan ¼ Nti Ns Nouse
ðmgN l1 Þ
ð156Þ
If the sludge age of the system is shorter than the minimum required for nitrification (RsoRsm), no nitrification takes place and the effluent nitrate concentration (Nne) is zero. The effluent ammonia concentration (Nae) is equal to the nitrogen available for nitrification (Nan, Equation (156)). If Rs4Rsm for Sf ¼ 1.0, most of the FSA available for nitrification is nitrified to nitrate and the effluent nitrate concentration (Nne) is the difference between Nan (Equation (156)) and the effluent FSA concentration given by Equation (132). For both RsoRsm and Rs4Rsm, the effluent TKN concentration (Nte) is the sum of effluent ammonia and unbiodegradable soluble organic nitrogen concentrations (Nte ¼ Nae þ Nouse). For RsoRsm, no nitrification takes place so the effluent nitrate concentration (Nne) is zero and the effluent ammonia concentration (Nae) is given by Nan (Equation (156)). The nitrifier sludge mass (MXA) and the nitrification oxygen demand (FOn) are both zero because Nne is zero. With increasing sludge age starting from Rs ¼ 0, Nae from Equation (132) is first negative (which is of course impossible) and then 4Nan (which is also not possible). For a sludge age slightly longer than Rsm, the Nae falls below Nan. From this sludge age, nitrification takes place and for further (even small) increases in sludge age, the Nae rapidly decreases to low values (o4 mgN l1). Hence for Rs4Rsm, the effluent ammonia concentration (Nae) is given by Equation (132), the effluent TKN concentration by Nte ¼ Nae þ Nouse, and the effluent nitrate concentration (Nne) by
Nne ¼ Nan Nae ¼ Nti Ns Nte
ðmgN l1 Þ
ð157Þ
With nitrification, the nitrifier biomass and nitrification oxygen demand are given by Equations (154) and (155). Substituting the influent N concentrations for raw and settled wastewaters and the values of the kinetic constants at 14 1C into the above equations, the results at different sludge ages were calculated. In Figure 31(a), the different effluent N concentrations from the system versus sludge age for raw and settled wastewater at 14 1C are shown. In Figure 31(c) are shown the nitrifier sludge mass (as a % of the reactor VSS mass) and nitrification oxygen demand for raw and settled wastewater at 14 1C. Also shown in Figure 31(c) are the carbonaceous and total oxygen demands for raw and settled wastewater at 14 1C. The calculations were repeated for 22 1C and shown in Figures 31(b) and 31(d). Figures 31(a) and 31(b) show that once the sludge age is approximately 25% longer than the minimum required for nitrification, nitrification is virtually complete (for steady-state conditions) and comparing the results for raw and settled wastewater, there is little difference between the nitrification oxygen demand and the concentrations of ammonia, nitrate, and TKN in the effluent. The reasons for this similar behavior are: (1) the primary settling tank removes only a small fraction of the influent TKN and (2) settled wastewater results in lower sludge production, so that the FSA available for nitrification in raw and settled wastewater is nearly the same. Once
nitrification takes place, temperature has relatively little effect on the different effluent N concentrations. However, a change in temperature causes a significant change in the minimum sludge age for nitrification. Considering Figures 31(a) and 31(b), for RsoRsm, the effluent ammonia concentration (Nae) and hence the effluent TKN concentration (Nte) increase with increasing sludge age up to Rsm because Ns decreases for increases in Rs. For Rs4Rsm, Nae decreases rapidly to o2 mgN l1 so that for Rs41.3Rsm, the effluent TKN concentration is o4 mgN l1. The increase in nitrate concentration (Nne) with an increase in sludge age for Rs41.3Rsm is mainly due to the reduction in N required for sludge production (Ns). This is important for BNR systems – increasing the sludge age increases the nitrification capacity (see Figure 30) so more nitrate has to be denitrified to achieve the same N removal. Figures 31(c) and 31(d) show that the nitrification oxygen demand increases rapidly once Rs4Rsm but for Rs41.3Rsm, further increases are marginal irrespective of the temperature or wastewater type. This nitrification oxygen demand represents an increase of 42% and 65% above the COD for the raw and settled wastewater. However, the total oxygen demand for treating settled wastewater is only 75% of that for treating raw wastewater. In order that nitrification can proceed without inhibition by oxygen limitation, it is important that the aeration equipment is adequately designed to supply the total oxygen demand; generally, heterotrophic organism growth takes precedence over nitrifier growth when oxygen supply (or ammonia) becomes insufficient. This is because heterotrophic organisms can grow adequately with DO concentrations of 0.5–1.0 mgO l1, whereas nitrifiers tend to require higher DO concentrations. Just as the effluent FSA concentration rapidly decreases for Rs4Rsm, the nitrifier sludge mass rapidly increases once Rs4Rsm, is slightly higher at 14 1C than at 22 1C due to the lower endogenous respiration rate. Also, because the concentrations of nitrate produced with raw and settled wastewater are closely similar (B40 mgN l1), the nitrifier sludge mass is approximately the same at the same sludge age (B430 kgVSS at 10d sludge age and B900 kgVSS at 30 day sludge age). Because with raw wastewater so much more sludge mass is produced than with settled wastewater, the nitrifier sludge mass is a much smaller proportion of the VSS mass with raw waste water (B1.4% at 10 day sludge age) than with settled wastewater (B3.3% at 10 day sludge age). Comparing the nitrifier sludge mass to the heterotrophic sludge mass, as in Figures 31(c) and 31(d), the nitrifier sludge mass comprises o4% of VSS mass even at high TKN/COD ratios for settled wastewater and so is ignored in the determination of the VSS concentration in the AS reactor treating domestic wastewater. It is worth repeating that primary sedimentation removes only a minor fraction of the TKN but a significant fraction of COD (15% and 40% in this example). Even though the settled wastewater has a lower TKN concentration than the raw wastewater, the effluent nitrate concentration does not reflect this difference. This is because the N removal for sludge production is lower for settled than for raw wastewater. Consequently, the nitrate concentration for settled wastewater is nearly the same as for raw wastewater – for different
(a)
Nitrogen-FSA, TKN, NO3 0.0
0.2
0.4
0.6
0.8
1.0
1.2
0
14 °C
5
0
FOc
5
Rsm
Ns
%Nit
25
25
30
0.0
0.5
1.0
1.5
2.0
2.5
3.0
3.5
30
Raw WW Settled WW
Raw WW Settled WW
FOn FOn
FOc
FOc
FOt
FOt
15 20 10 Sludge age (days)
%Nit
20 10 15 Sludge age (days)
Nte
10
Nouse = 1.8 mg N l−1 Nte = Nae + Nouse
Nne
Influent TKN
Ns
0
Nte
Rsm
20
30
40
50
60
14 °C
% Nitrifier VSS
(b)
(d)
0.0
0.2
0.4
0.6
0.8
1.0
1.2
0 0
0
5
22 °C Rsm
5
FOn FOn
%Nit
FOt
FOt
25
25
30
0.0
0.5
1.0
1.5
2.0
2.5
3.0
3.5
30
Raw WW Settled WW
22 °C
Raw WW Settled WW
FOc
FOc
10 20 15 Sludge age (days)
%Nit
15 20 10 Sludge age (days)
Ns
10
Nouse = 1.8 mg N l−1 Nte = Nae + Nouse
Nne
Influent TKN
Ns Nte
Rsm
20
30
40
50
60
70
Figure 31 Effluent ammonia (Nae), TKN (Nte), and nitrate (Nne) concentrations and N required for sludge production (Ns) vs. sludge age at 14 1C (a) and 22 1C (b) and nitrification (FOn), carbonaceous (FOc) and total (FOt) oxygen demand in kgO/kgCOD load and % nitrifier VSS mass vs. sludge age at 14 1C (c) and 22 1C (d) for the example raw and settled wastewater.
(c)
Oxygen demand (kgO/kgCOD)
Nitrogen-FSA, TKN, NO3 Oxygen demand (kgO/kgCOD)
70
% Nitrifier VSS
478
Biological Nutrient Removal
wastewater characteristics, it can be higher than raw wastewater. In contrast, the maximum N removal by denitrification using the wastewater organics as electron donor, called the denitrification potential, mainly depends on the influent COD concentration and this concentration is significantly reduced by primary sedimentation. This may result in a situation where it may be possible to obtain near-complete nitrate removal when treating raw wastewater but not when treating settled wastewater. The difference in COD and TKN removal in PSTs therefore has a significant effect on the design of BNR systems.
4.14.24 Biological Denitrification
ages can be in the usual fully aerobic short sludge age range of 3–6 days. Unaerated zones should still be incorporated to derive the benefits of denitrification in the event nitrification does take place. When it does not, the unaerated zone will be anaerobic (no input of DO or nitrate) instead of anoxic, and some BEPR may take place. Because BEPR is not required and therefore not exploited to the full, whether or not it takes place is not important because it does not affect the system behavior very much. With some BEPR, the sludge production will be slightly higher (o5%) per COD load, the VSS/TSS ratio and oxygen demand both somewhat lower (by about 5%). However, BEPR may result in mineral precipitation problems in the sludge treatment facilities if the WAS is anaerobically digested.
4.14.24.1 Interaction between Nitrification and N Removal Nitrification is a prerequisite for denitrification – without it biological N removal is not possible. Once nitrification takes place, N removal by denitrification becomes possible and should be included even when N removal is not required (see Section 4.14.14) by incorporating zones in the reactor that are intentionally unaerated. Because the nitrifiers are obligate aerobes, nitrification does not take place in the unaerated zone(s), so to compensate for this, the system sludge age needs to be increased for situations where nitrification is required. For fully aerobic systems and a wastewater temperature 14 1C, a sludge age of 5–7 days may be sufficient for complete nitrification, taking due consideration of the requirement that the effluent FSA concentration should be low even under cyclic flow and load conditions (Sf41.3). For anoxic – aerobic systems, a sludge age of 15–20 days may be required when a 50% unaerated mass fraction is added (Figure 25). Therefore, for plants where N removal is required, invariably the sludge ages are long due to (1) the uncertainty in the mAm20 value, (2) the need for unaerated zones, and (3) the guarantee of nitrification at the minimum average winter temperature (Tmin). For plants where nitrification is a possibility and not obligatory, uncertainty in the mAm20 value is not important and unaerated zones can be smaller, with the result that sludge Table 12
4.14.24.2 Benefits of Denitrification In the design of fully aerobic systems discussed above, it was suggested that when nitrification is not obligatory but a possibility, unaerated zones should still be incorporated in the system to derive the benefits of denitrification. These benefits include (1) reduction in nitrate concentration which ameliorates the problem of rising sludge from denitrification in the secondary settling tank (Section 4.14.14), (2) recovery of alkalinity (Section 4.14.20.6), and (3) reduction in oxygen demand. With regard to (3), under anoxic conditions, nitrate serves as electron acceptor instead of DO in the degradation of organics (COD) by facultative heterotrophic organisms. The oxygen equivalent of nitrate is 2.86 mgO/mgNO3-N which means that 1 mg NO3-N denitrified to N2 gas has the same electron-accepting capacity as 2.86 mg of oxygen. In nitrification to nitrate, the FSA donates eight electrons (e)/mol, the N changing from an e state of 3 to þ 5. In denitrification to N2, the nitrate accepts 5 e/mol, the N changing from an e state of þ 5 to 0. Because 4.57 mgO/mgFSA-N are required for nitrification, the oxygen equivalent of nitrate in denitrification to N2 is 5/8 4.57 ¼ 2.86 mgO/mgNO3-N (Table 12). Therefore, for every 1 mg NO3-N denitrified to N2 gas in the anoxic zone, during which about 2.86/(1 fcvYH) ¼ 8.6 mgCOD
Comparison of nitrification and denitrification processes in single sludge activated sludge systems Nitrification
Denitrification
Form Function Half-reaction Organisms Environment
Ammonia (NHþ 4) Electron donor Oxidation Autotrophs Aerobic
Nitrate (NO 3) Electron acceptor Reduction Heterotrophs Anoxic
Compound Oxid. no.
NH4 þ 3 Nitrification (oxidation)
N2 0
NO2 þ3
8 e/atom N ¼ 4.57 mgO/mgN Net loss Denitrification (reduction) 5e/atom N ¼ 2.86 mgO/mgN Nitrification: 4.57 mgO/mgNH4-N nitrified to NO3-N Denitrification: 2.86 mgO recovered/mg NO3-N denitrified to N2 gas Therefore, denitrification allows at best 62.5% (5/8 or 2.86/4.57) recovery of the nitrification oxygen demand.
NO3 þ5
Biological Nutrient Removal Effluent TKN liquid phase
Raw wastewater TKN/COD = 0.08 1.0
Oxygen demand (kgO/kgCOD on reactor)
14 °C 22 °C
479
Total incl. nitrification
~5%
N in gas phase ~75%
0.8
N in sludge solid phase ~20%
0.6 Total incl. nitrif. and denit. 0.4
N nitrified (transformed in liquid phase) and possibly denitrified (transferred to gas phase)
Carbonaceous
Possibility of nitrification 0.2
Possibility of denitrification
Figure 33 Exit routes for nitrogen in single sludge nitrification denitrification activated sludge systems.
0.0 0
5
10
15
20
25
30
Sludge age (days) Figure 32 Carbonaceous, total including nitrification and total including nitrification and denitrification oxygen demand per unit COD load on the biological reactor vs. sludge age for the example* raw wastewater. *Note: All the figures in this part which show the behavior of the various activated sludge system configurations were generated from the example raw and settled wastewater characteristics.
is utilized, 2.86 mg less oxygen needs to be supplied to the aerobic zone. Because the oxygen requirement to form the nitrate from ammonia is 4.57 mgO/mgNO3-N, and 2.86 mgO/mgNO3-N is recovered in denitrification to N2 gas, a maximum of 2.86/4.57 (or 5/8ths) ¼ 0.63 of the nitrification oxygen demand can be recovered. A comparison of the nitrification and denitrification reactions is given in Table 12. Under operating conditions, it is not always possible to denitrify all the nitrate formed with the result that the nitrification oxygen recovery by denitrification is about 50% (see Figure 32). Therefore, once the possibility of nitrification exists, it is always worthwhile to consider including intentional denitrification because of the recovery of alkalinity and oxygen. With regard to oxygen, if the oxygen supply is insufficient to meet the combined carbonaceous and nitrification requirement, areas in the aerobic reactor will become anoxic. Under oxygen limited conditions, the aerobic mass fraction in the aerobic reactor will vary depending on the COD and TKN load on the plant over the day. At minimum load, oxygen supply may be adequate so that nitrification may be complete whereas, at peak load, oxygen supply may be insufficient so that nitrification may cease (partially or completely) and denitrification will take place on the accumulated nitrate. This behavior is exploited in the single-reactor ND configurations such as the ditch or Carousel-type systems.
4.14.24.3 N Removal by Denitrification In biological N removal systems, the N is removed by transfer from the liquid phase to the solid and gas phases. About 20% of the influent N is incorporated in the sludge mass (Figure 33) but the bulk of the N (i.e., about 75% when complete denitrification is possible) is removed by transfer to the gas phase via nitrification and denitrification (Figure 33). In the nitrification step, the N remains in the liquid phase because it is transformed from ammonia to nitrate. In the denitrification step, it is transferred from the liquid to the gas phase and escapes to the atmosphere. When complete denitrification is achieved, a relatively small fraction of the influent TKN (B5%) remains in the liquid phase and escapes as total N (TKN þ nitrate) with the effluent. For aerobic conditions, the problem of the designers is to calculate the mass of oxygen electron acceptor required by the OHOs (and ANOs) for the utilization of the known mass of organic electron donors (organics and ammonia) available. For anoxic conditions, the problem is the opposite. Here, the problem is to calculate the mass of electron donors (COD) that are required to denitrify a known mass of electron acceptors nitrate. If sufficient electron donors (COD) are not available then complete denitrification cannot be achieved. The calculation of the nitrogen removal is essentially a reconciliation of electron acceptors (nitrate) and donors (COD) taking due account of (1) the biological kinetics of denitrification and (2) the system operating parameters (such as recycle ratios and anoxic reactor sizes) under which the denitrification is constrained to take place. The electron donors (or COD or energy) for denitrification can come from two sources: (1) internal or (2) external to the AS system. The former are those present in the system itself, that is, those in the incoming wastewater or generated within the biological reactor by the AS itself; the latter are organics imported to the AS system and specifically dosed into the anoxic zone(s) to promote denitrification, (e.g., methanol,
480
Biological Nutrient Removal
acetate, and molasses; Monteith et al., 1980). Here, the focus is on internal COD sources for denitrification, but the principles and procedures are sufficiently general to be adaptable to include external COD (energy) sources also.
4.14.24.4 Denitrification Kinetics There are three internal organics sources, two from the wastewater and one from the AS sludge mass itself. The two in the wastewater are the two main forms of organics (i.e., RBSO) and slowly biodegradable organics (BPO)). The third is slowly biodegradable organics generated by the biomass itself through death and lysis of organism mass (also known as endogenous mass loss/ respiration). This self-generated BPO is utilized in the same way as the wastewater BPO, but is recognized separately because of its different source and rate of supply to that of the influent. The RBSO and BPO (influent or self-generated) are degraded via different mechanisms by the OHOs. The different RBSO and BPO degradation mechanisms lead to different COD utilization rates. The RBSO comprises small simple dissolved organic compounds that can pass directly through the cell wall into the organism, for example, sugars and short-chain fatty acids. Accordingly, the RBSO can be used at a high rate which does not change significantly whether nitrate or oxygen serves as terminal electron acceptor (Ekama et al., 1996a). Simulation models use the Monod equation to model the utilization of RBSO by OHOs under both aerobic and anoxic conditions. The BPO comprises large particulate or colloidal organic compounds, too large to pass into the organism directly. These organics must be broken down (hydrolyzed) in the slime layer surrounding the organism to smaller components, which then can be transferred into the organism and utilized. The extracellular BPO hydrolysis rate is slow and forms the limiting rate in the utilization of BPO (Section 4.14.5.1.3). This hydrolysis rate is much slower under anoxic conditions than under aerobic conditions – only about one-third (Stern and Marais, 1974, van Haandel et al., 1981). This introduces a reduction factor Z in the BPO hydrolysis rate equation for anoxic conditions (Equation (159) below). Research has indicated that the utilization of RBSO is simultaneous with the hydrolysis of BPO. Also the rate of RBSO utilization is considerably faster (7–10 times) than the rate of BPO hydrolysis so the denitrification rate with influent RBSO is much faster than with BPO. Therefore, the influent RBSO is the preferred organic for denitrification and the higher this concentration in the influent with respect to the total COD, the greater the N removal.
4.14.24.5 Denitrification Systems As a result of the different degradation mechanisms and rates of RBSO and BPO utilization, the position of the anoxic zone in the biological reactor significantly affects the denitrification that can be achieved. There are many different configurations of single sludge ND systems but from the point of view of the source of the organics (electron donors), these can be simplified to two basic types of denitrification or combinations of these. The two basic types utilizing internal organics are: (1) post-denitrification, which utilizes self-generated
endogenous organics and (2) pre-denitrification, which utilizes influent wastewater organics. With post-denitrification (Figure 34(a)), the first reactor is aerobic and the second is unaerated. The influent is discharged to the aerobic reactor where aerobic growth of both the heterotrophic and nitrifying organisms takes place. Provided the sludge age is sufficiently long and the aerobic fraction of the system is adequately large, nitrification will be complete in the first reactor. The mixed liquor from the aerobic reactor passes to the anoxic reactor, also called the secondary anoxic reactor, where it is mixed with stirring. The outflow from the anoxic reactor passes through an SST and the underflow is recycled back to the aerobic reactor. The BPO released by the sludge mass via the death of organisms provides the energy source for denitrification in the anoxic reactor. However, the rate of release of energy is low, so that the rate of denitrification is also low. To obtain a meaningful reduction of nitrate, the anoxic mass fraction of the
Anoxic reactor
Aerobic reactor
Waste flow Settler
Influent
Effluent
Sludge recycle
(a)
s
Anoxic Aerobic reactor reactor Mixed liquor Recycle
Waste flow
a
Settler
Influent
Effluent
I
Sludge recycle
(b)
Primary Aerobic anoxic reactor reactor Mixed liquor Recycle a
s
Secondary anoxic reactor Reaeration reactor Waste flow Settler Effluent
Influent
(c)
Sludge recycle
s
Figure 34 (a) The post-denitrification single sludge biological nitrogen removal system. (b) The modified Ludzack–Ettinger single sludge biological nitrogen removal system proposed by Barnard (1973), including the primary anoxic reactor only. (c) The four-stage Bardenpho single sludge biological nitrogen removal system, including primary and secondary anoxic reactors.
Biological Nutrient Removal
system (i.e., the fraction of the mass of sludge in the system that is in the anoxic reactor) must be large and this may cause, depending on the sludge age, cessation of nitrification. Thus, although theoretically the system has the potential to remove all the nitrate, from a practical point this is not possible because the anoxic mass fraction will need to be so large that the conditions for nitrification cannot be satisfied particularly if the temperatures are low (o15 1C). Furthermore, in the anoxic reactor, ammonia is released through organism death and lysis, some of which passes out with the effluent thereby reducing the total nitrogen removal of the system. To minimize the ammonia content of the effluent, a flash or re-aeration reactor sometimes is placed between the anoxic reactor and the SST. In this reactor, N2 gas is stripped from the mixed liquor to avoid possible sludge buoyancy problems in the SST and the ammonia is nitrified to nitrate to assist with compliance of ammonia standards but it reduces the overall efficiency of the nitrate reduction of the system. For these reasons, post-denitrification has not been widely applied in practice.
4.14.24.5.1 The Ludzack–Ettinger system Ludzack and Ettinger (1962) were the first to propose a single sludge ND system utilizing the biodegradable organics in the influent as organics for denitrification. It consisted of two reactors in series, partially separated from each other. The influent was discharged to the first, or primary anoxic reactor which was maintained in an anoxic state by mixing without aeration. The second reactor was aerated and nitrification took place in it. The outflow from the aerobic reactor passed to the SST and the SST underflow was returned to the aerobic (second) reactor. Due to the mixing action in both reactors, an interchange of the nitrified and anoxic liquors was induced. The nitrate which entered the primary anoxic reactor was denitrified to nitrogen gas. Ludzack and Ettinger reported that their system gave variable denitrification results, probably due to the lack of control of the interchange of the contents between the two reactors. In 1973, Barnard proposed an improvement to the Ludzack– Ettinger system by completely separating the anoxic and aerobic reactors, recycling the underflow from the SST to the primary (first) anoxic reactor and providing a mixed liquor recycle from the aerobic to the primary anoxic reactor (Figure 34(b)). These modifications allowed a significant improvement in control over the system N removal performance of the system with the mixed liquor recycle flow. The RBSO and BPO from the influent stimulated high rates of denitrification in the primary anoxic reactor and much higher reductions of nitrate could be achieved than with post-denitrification, even when the pre-denitrification reactor of this system was substantially smaller than the post-denitrification reactor. In this system, called the Modified Ludzack–Ettinger (MLE) system, complete nitrate removal cannot be achieved because a part of the total flow from the aerobic reactor is not recycled to the anoxic reactor but exits the system with the effluent. To reduce the possibility of flotation of sludge in the SST due to denitrification of residual nitrate, the sludge accumulation in the SST needed to be kept to a minimum. This was achieved by having a high underflow recycle ratio from the SST, equal to the mean influent flow (1:1).
481
4.14.24.5.2 The four-stage Bardenpho system In order to overcome the deficiency of incomplete nitrate removal in the MLE system, Barnard (1973) proposed including a secondary anoxic reactor in the system and called it the fourstage Bardenpho system (Figure 34(c)). Barnard considered that the low concentration of nitrate discharged from the aerobic reactor to the secondary anoxic reactor will be denitrified to produce a relatively nitrate-free effluent. He included a flash or re-aeration reactor to strip the nitrogen gas and to nitrify the ammonia released during the denitrification. Although in concept the Bardenpho system has the potential for complete removal of nitrate, in practice this is not possible except when the influent TKN/COD concentration ratio is quite low, o0.09 mgN/mgCOD for normal municipal wastewater at 14 1C. The low denitrification rate and ammonia release (about 20% of the nitrate denitrified) results is an inefficient use of the secondary anoxic sludge mass fraction. As a result of the competition between the aerated and unaerated sludge mass fractions from the requirement to nitrify, (Section 4.14.20.3) usually it is better to exclude the secondary anoxic (and re-aeration) reactor and enlarge the primary anoxic reactor and increase the mixed liquor recycle ratio.
4.14.25 Denitrification Kinetics 4.14.25.1 Denitrification Rates The denitrification behavior in the primary and secondary anoxic zones is best explained by considering these reactors as plug-flow reactors. However, the explanation is equally valid for completely mixed reactors because the denitrification kinetics are essentially zero order with respect to nitrate concentration (van Haandel et al., 1981; Ekama and Wentzel, 1999b). Owing to the two different kinds of biodegradable organics (RBSO and BPO) in the influent wastewater, the denitrification in the primary anoxic reactor follows two phases (Figure 35(a)) – an initial rapid phase where the rate is defined by the simultaneous utilization of RBSO and BPO (K1 þ K2) and a second slower phase where the specific denitrification rate (K2) is defined by the utilization of only BPO originating from the influent and self-generated by the sludge through organism death and lysis. In the secondary anoxic reactor, only a single slow phase of denitrification takes place (Figure 35(b)), the specific rate (K3) being about two-thirds of the slow rate (K2) in the primary anoxic reactor (Stern and Marais, 1974; van Haandel et al., 1981). In the preceding aerobic reactor all the RBSO and most of the BPO of the influent has been utilized with the result that in the secondary anoxic reactor the only biodegradable COD available is BPO from organism death and lysis; the slow rate of supply of this BPO governs the K3 rate and causes this rate to be slower than the K2 rate. The values of the K rates are given in Table 13. A further specific K rate (K4) has been defined for denitrification in intermittently aerated anoxic aerobic digestion of WAS (Warner et al., 1986). This rate is only two-thirds of the K3 rate in the secondary anoxic reactor (Table 13), but sufficiently high to denitrify all the nitrate generated in aerobic digestion of WAS if the 6 h aeration cycle is 50% anoxic and 50% aerobic. Denitrification in anoxic–aerobic digestion adds
482
Biological Nutrient Removal
NO3−N concentration
NO3−N concentration
K1XBH
K2XBH
1st
(a)
K3XBH
Single phase
Second phase
(b)
Time
Time
Figure 35 Nitrate concentration of vs. time profiles in primary anoxic (a) and secondary anoxic (b) plugflow reactors, showing the three phases of denitrification associated with the K1, K2, and K3 rates. In the primary anoxic the initial rapid rate K1 is attributable to the utilization of the influent RBSO and the second slower rate K2 to the utilization of BPO from the influent wastewater and self-generated by organism death and lysis. In the secondary anoxic reactor, the rate K3 is attributable to the utilization of the self-generated BPO only. Table 13
K denitrification rates and their temperature sensitivity
Symbol
20 1 C
y
14 1 C
22 1 C
Equation
K120a K220a K320a K420a
0.72 0.1 0.1 0
1.2 1.08 1.029 1.029
0.241 0.06 0.06 0.04
1.036 0.118 0.08 0.05
158 159 160 161
a
Units – mgNO3-N/(mgOHOVSS d).
the benefits of denitrification to this system, that is, zero alkalinity consumption, oxygen recovery, improved pH control, reduced chemical dosing (Dold et al., 1985), and additionally a nitrogen free dewatering liquor. This last advantage is significant considering the high N content of WAS compared with primary sludge. The constancy of K1, K2, K3 (and K4) specific denitrification rates under constant flow and load conditions can be explained in terms of the kinetics of RBSO and BPO organics utilization included in the AS simulation models such as ASM1 developed later. The utilization of RBSO organics is modeled with the Monod equation and expressing the K1 rate in terms of this yields
K1 ¼
ð1 f cv YH Þf cv mHm Ss 2:86YH Ks þ Ss
where
Ss E 1 ðmgNO3 -N=ðmgOHOVSS dÞÞ Ks þ Ss
ð158Þ
In the plugflow and completely mixed primary anoxic reactor, the Monod term SS/(KS þ SS) remains close to 1 down to low RBSO concentrations because the half-saturation concentration (KS) is low. Accepting YH ¼ 0.45 mgVSS and
fcv ¼ 1.48 mgCOD/mgVSS yields K1 ¼0.26 mH mgNO3-N/ (mgOHOVSS d). So for the measured K1 ¼0.72 mgNO3-N/ (mgOHOVSS d) (Table 13), the mHm must have been about 2.8 d1. This mHm rate is in the range of mHm rates measured in AS systems. In investigating the kinetics of RBSO utilization in aerobic and anoxic selectors, Still et al. (1996) and Ekama et al. (1996a, b) found mHm values ranged between 1.0 d1 in completely mixed reactor systems and 4.5 d1 selector reactor systems, which yields K1 denitrification rates around 0.26 mgNO3-N/(mgOHOVSS d) for completely mixed type systems and 1.17 mgNO3-N/(mgOHOVSS d) for systems in which a selector effect (high mH) has been stimulated in the OHO biomass. The utilization of BPO is expressed in terms of the activesite surface hydrolysis kinetic formulation, which has the form of a Monod equation, except the variable is the adsorbed BPO to active OHO ratio (Xs/XBH), not the bulk liquid BPO concentration. Hence, the K2, K3 (and K4) rates are given by
K2 ¼ K3 ¼ K4 ¼
ð1 f cv YH Þ ZKh ðXs =XBH Þ 2:86f cv YH ½Kx þ ðXs =XBH Þ mgNO3 -N=ðmgOHOVSS dÞ
ð159Þ
where XS/XBH is progressively lower in primary (K2) secondary (K3) and anoxic–aerobic digestion (K4).
Biological Nutrient Removal
In the constant flow and load primary and secondary anoxic plugflow reactors, the (Xs/XBH) ratio changes very little due to the reduced anoxic hydrolysis rate including the Z. The reason for the K2 being higher than K3 arises from different concentrations of adsorbed BPO relative to the active OHO concentration (Xs/XBH) (Figure 36). In the primary anoxic reactor, the ratio is high because adsorbed BPO originates from the influent and OHO death. In the secondary anoxic, the ratio is lower because BPO originates only from OHO death. For the K2 and K3 denitrification rates, there is no simple relationship between the K rates and the ZKh because the adsorbed BPO to OHO ratio (Xs/XBH) is different in the primary and secondary anoxic reactors (and aerobic digester) and varies somewhat with sludge age and unaerated sludge mass fraction. It was concluded that the K1, K2, K3, and K4 denitrification constants have no direct kinetic significance; their constancy is the result of a combination of kinetic reactions which show little variation with sludge age in the range 10–30 days. Temperature does affect the K rates but once these have been adjusted for temperature, again the K rates show little variation at different sludge ages (van Haandel et al., 1981). It can be concluded both from experimental observation and theoretical kinetic points of view that acceptance of constant K2 and K3 rates is acceptable for steady-state design. This is in fact done to estimate the denitrification potential (Dp) of an anoxic reactor under constant flow and load conditions. With regard to K1, this rate can change significantly because the RBSO utilization rate can change appreciably depending on the mixing regime in the anoxic (or aerobic) reactor (Ekama et al., 1986, 1996a, b and Still et al., 1996). However, its variation does not affect ND design significantly because normally primary anoxic reactors are sufficiently large to allow complete utilization of RBSO even when the utilization rate (mHm) is low. In fact, the denitrification design procedure requires that all the RBSO is utilized in the primary anoxic reactor which introduces a minimum primary anoxic sludge mass fraction (fx1 min) and a minimum a-recycle ratio (amin) to
0.12
Specific denit rates (K )
K2 0.10 0.08 0.06
K3 K4
0.04 0.02 0.00 0.0
0.1
0.2
0.3
0.4
XS /XBH ratio (mgCOD/mgCOD) Figure 36 Specific denitrification rate (K) vs. adsorbed SB organics to OHO biomass concentration ratio (XS/XBH), showing the primary anoxic (K2), secondary anoxic (K3), and anoxic–aerobic digestion (K4) specific denitrification rates.
483
ensure this. These concepts can also be used for anoxic selector reactor design (Ekama et al.,1996a). The simulation model was applied also to anoxic–aerobic digestion of WAS. It was found that the model predicted accurately both aerobic and anoxic–aerobic digester behavior under constant and cyclic flow and load conditions and validated the K4 specific denitrification rate (Warner et al., 1986); no significant adjustment to values of the kinetic constants was necessary.
4.14.25.2 Denitrification Potential The concentration of nitrate (per liter influent flow Qi) that an anoxic reactor can denitrify biologically is called that reactor’s denitrification potential. It is called a potential because whether or not it is achieved depends on the nitrate load on the anoxic reactor(s). If too little nitrate is recycled to the anoxic reactor, all the recycled nitrate will be denitrified and the actual removal of nitrate, that is, denitrification performance, will be lower than the potential. In this case the denitrification is system (or recycle) limited. An increase in the system recycle ratios will increase nitrate load on the anoxic reactor and hence also the denitrification. Once the recycle rates are such that the nitrate load on the anoxic reactor(s) equals the denitrification potential of the reactor, then the system denitrification performance is optimal and the recycle ratios are at their optimum values. At this point the anoxic and aerobic reactor nitrate concentrations are just zero and the lowest possible, respectively. Increasing the recycle rates above the optimum increases the nitrate concentration in the anoxic reactor outflow above zero but this does not improve the denitrification performance because the system has now become biological or kinetics limited. The denitrification potential of the anoxic reactor(s) has been achieved and more nitrate cannot be denitrified by the particular anoxic reactors and wastewater. Indeed, increases in the recycle ratios above the optimum values are uneconomical due to unnecessary pumping costs and introduce unnecessary additional DO into the anoxic reactors which causes an undesirable reduction in denitrification performance and increase in effluent nitrate concentration. The principle of denitrification design therefore hinges around (1) calculating the denitrification potential of the anoxic reactor(s); (2) setting the nitrate load imposed on the anoxic reactor(s) equal to the denitrification potential; and (3) calculating the recycle ratios associated with this condition. The recycle ratios so calculated are the optimum values. From the above discussion, it is clear that critical in the design for denitrification is calculation of the nitrate load and denitrification potential. The nitrate load is calculated from the nitrification capacity, which is the concentration of nitrate per liter influent flow (Qi) generated by nitrification (Section 4.14.22.2, Equation (152)). The nitrification capacity (Nc, mg N l1 influent) was shown above to be approximately proportional to the influent TKN concentration (Nti). The denitrification potential is calculated separately for the utilization of the RBSO and BPO. The RBSO gives rise to a rapid denitrification rate so that it can be assumed that it is all utilized in the primary anoxic reactor. This is in fact an objective in the design. Accordingly, the contribution of the RBSO to the denitrification potential is simply the catabolic component of its
484
Biological Nutrient Removal
electron-donating capacity in terms of nitrate as N. Therefore, in the complete utilization of the influent RBSO, a fixed proportion (1 fcvYH) of the RBSO electrons (catabolic component) will be donated to NO3 reducing it to N2. Thus, knowing the influent RBSO concentration and assuming it is all utilized, the denitrification potential of this RBSO is given by
Dp1RBSO ¼ f Sb0 s Sbi ð1 f cv YH Þ=2:86 1
ðmgNO3 -N l
influentÞ
components of the RBSO and BPO yields the total denitrification potential of primary and secondary anoxic reactors, that is,
Dp1 ¼ Dp1RBSO þ Dp1BPO ¼ f Sb0 s Sbi ð1 f cv YH Þ=2:86 þ Sbi K2 f x1 YH Rs =ð1 þ bH Rs Þ ¼ Sbi ff Sb0 s ð1 f cv YH Þ=2:86 þ K2 f x1 YH Rs =ð1 þ bH Rs Þg ðmgN l1 influentÞ
ð163Þ
ð160Þ Dp3 ¼ Dp3RBSO þ Dp3BPO
where Dp1 RBSO is the denitrification potential of the influent RBSO in primary anoxic reactor, Sbi the influent biodeg. COD (mgCOD l1), fSb0 s the RBSO fraction of Sbi, YH the OHO yield coefficient (0.45 mgVSS/mgCOD), and 2.86 the oxygen equivalent of nitrate. For the BPO, this substrate contributes to denitrification in the primary anoxic reactor and the secondary anoxic reactor. The denitrification potentials for the BPO are formulated on the basis of the K2 and K3 specific denitrification rates, respectively. These K rates are a simplification of the kinetic equations describing the utilization of BPO from the influent and/or from organism death and lysis and have a basis in the fundamental biological kinetics incorporated in the AS simulation models such as ASM1 (Henze et al., 1987). The K rates define the denitrification rate as mgNO3 -N denitrified per day per mgOHOVSS mass in the anoxic reactor. To determine the denitrification potential contributed by the BPO, the mass of OHOVSS produced per liter influent flow and the proportion of this mass in the primary and/or secondary anoxic reactors needs to be calculated and multiplied by the K2 or K3 rates. From the steady-state AS model for organics removal (Section 4.14.9.3), the OHO mass in the system (MXBH) is calculated from the biodegradable COD load (Equation (103)). Of this MXBH mass, a fraction fx1 and/or fx3 is continuously present in the primary and/or secondary anoxic reactors, respectively, that is, fx1 and fx3 are the primary and secondary anoxic sludge mass fractions, respectively. The OHOVSS mass in the primary and/or secondary anoxic reactors per liter influent flow is therefore given by
f x1 MXBH =Qi ¼ f x1 Sbi ðYH ÞRs =ð1 þ bH Rs Þ ðmgOHOVSS l1 influent in primary anoxicÞ
f x3 MXBH =Qi ¼ f x3 Sbi ðYH ÞRs =ð1 þ bH Rs Þ ðmgOHOVSS l1 influent in secondary anoxicÞ
Multiplying these masses by the respective K rates gives the primary and secondary anoxic reactor denitrification potentials attributable to BPO (Dp1BPO, Dp3BPO), viz.,
Dp1BPO ¼ K2 f x1 MXBH =Qi ¼ K2 f x1 Sbi YH Rs =ð1 þ bH Rs Þ ð161Þ Dp3BPO ¼ K3 f x3 Sbi YH Rs =ð1 þ bH Rs Þ
ð162Þ
This approach is valid because the K2 and K3 rates are continuous for the entire sludge residence time in the anoxic reactor(s), provided the nitrate concentration does not decrease to zero (Figure 35). Combining the denitrification potential
¼ 0 þ Sbi K3 f x3 YH Rs =ð1 þ bH Rs Þ ðmgN l1 influentÞ ð164Þ In Equations (163) and (164), the K2, K3, and bH rates are temperature sensitive, decreasing as the temperature decreases. The temperature sensitivity of these rates has been measured and is defined in Tables 13 and 6. From Equations (163) and (164), it can be seen that the denitrification potentials are directly proportional to the biodegradable COD concentration of the wastewater (Sbi). This is expected because in the same way that the oxygen demand is directly related to the COD load, so also is the nitrate demand (which is called the denitrification potential) because both oxygen and nitrate act as electron acceptor for the same organic degradation reactions. For the same size anoxic reactor, the primary anoxic has a much larger denitrification potential (by about 2–3 times) than the secondary anoxic because (1) K2 is larger than K3 and (2) more importantly, the RBSO makes a significant contribution to the denitrification potential in the primary anoxic reactor. For this reason the RBSO needs to be accurately specified to ensure reliable estimates of the N removal that can be achieved. For a normal municipal wastewater with an RBSO fraction (fSb0 s) of about 25% (with respect to biodegradable COD), the RBSO contributes about one-third to half of Dp1 depending on the size of the primary anoxic reactor and temperature. In a system where a high degree of N removal is required, between one-fourth and one-third of the carbonaceous oxygen demand is met by denitrification, which reduces the carbonaceous oxygen demand in the aerobic reactor by the same amount. As mentioned earlier, this reduction represents about half of the oxygen that was required to produce the nitrate by nitrification (see Figure 32). From Equation (164), the influent RBSO contribution to the denitrification potential of the secondary anoxic reactor is zero. This is because all the RBSO is utilized either in the preceding primary anoxic and/or in aerobic reactors. However, the Dp3 RBSO term has been included in Equation (164) in the event an external carbon source such as methanol, acetic acid, or high-strength organic wastewater is dosed into the secondary anoxic reactor to improve the denitrification. The sludge mass fraction approach above is valid because the fraction of the VSS (MXv) or TSS (MXt) masses that is OHO mass (MXBH) is constant for specified wastewater characteristics and sludge age and equal to the active fraction (fatOHO or favOHO – Equations (114) and (115)) and very closely the same in the anoxic and aerobic reactors of the system. Therefore, the anoxic and aerobic sludge mass fractions are the same whether calculated from the VSS, TSS, or OHO masses; for example, in an MLE system with anoxic and aerobic reactor volumes
Biological Nutrient Removal of 3000 and 6000 m3, respectively, one notes that nearly one-third of the OHO, VSS, and TSS masses in the system are in the anoxic reactor, and hence the anoxic sludge mass fraction is 0.33.
4.14.25.3 Minimum Primary Anoxic Sludge Mass Fraction In Equation (163), it is assumed that the initial rapid rate of denitrification is always complete, that is, the actual retention time in the primary anoxic reactor is always longer than the time required to utilize all the influent RBSO. This is because in Equation (163), the denitrification attributable to the influent RBSO is stoichiometrically expressed, not kinetically – it gives the concentration of nitrate the K1 rate removes when allowed sufficient time to reach completion. By considering the actual retention time (say t1) required to complete the first phase of denitrification (Figure 35(a)), and noting that t1(a þ s þ 1) is the minimum nominal hydraulic retention time to achieve this, it can be shown that the minimum primary anoxic sludge mass fraction fx1min to remove all the influent RBSO at a rate of K1 mgNO3-N/(mgOHOVSS d) is
f x1min ¼
f Sb0 s ð1 f cv YH Þð1 þ bHT Rs Þ 2:86K1T YH Rs
ð165Þ
Substituting the values of the kinetic constants into Equation (165), yields fx1mino0.08 for Rs 410 days at 14 1C. This value is much lower than most practical primary anoxic reactors so that Equation (163) will be valid in most cases. Equation (165) also applies to sizing anoxic selectors provided K1 (or mH) is appropriately selected (see Section 4.14.25.1, Equation (158); Ekama et al., 1996a).
4.14.25.4 Denitrification – Influence on Reactor Volume and Oxygen Demand From the design approach to nitrification (Equation (136)) and denitrification (Equations (163) and (164)), it can be seen that the design for N removal is done entirely using sludge mass fractions and does not require the volume of the reactor to be known. The volume of the reactor is obtained in the identical fashion as for the fully aerobic system and follows from the choice of the TSS concentration (Xt) for the reactor (Section 4.14.11). The volume of the reactor so obtained is then subdivided in proportion to the calculated primary and/or secondary anoxic and aerobic sludge mass fractions. Consequently, N removal design is grafted directly into the aerobic system design and for the same design reactor TSS concentration and sludge age, a fully aerobic system and an anoxic–aerobic system for N removal will have the same reactor volume. Research has indicated that there are many factors that influence the mass of sludge generated for a given sludge age and daily average COD load, and alternating anoxic–aerobic conditions is one of them. However, relative to the uncertainty in organic (COD) load and unbiodegradable particulate COD fraction and their daily and seasonal variation, these influences are not large enough from a design point of view to be given specific consideration in the design procedure. From a design point of view, the only significant difference between aerobic and anoxic–aerobic conditions is the oxygen demand
485
and this difference needs to be taken into account for economical design (Figure 32).
4.14.26 Development and Demonstration of Design Procedure It was concluded above that the influent wastewater characteristics that need to be accurately known are the influent TKN/COD ratio and RBSO fraction. These have a major influence on the nitrification capacity and denitrification potential, respectively, and hence on the N removal performance and minimum effluent nitrate concentration that can be achieved by biological denitrification. The effect of these two wastewater characteristics on design will be demonstrated below with numerical examples generated from the example raw and settled wastewaters with different influent TKN concentrations and RBSO fractions. The design of biological N removal is developed and demonstrated below by continuing the calculations with the example raw and settled wastewater characteristics listed in Tables 7 and 11. The only additional characteristic required for the denitrification design is the influent RBSO fraction (fSb0 s), which is 0.25 and 0.385 of the biodegradable COD for the raw and settled wastewaters, respectively. The results obtained so far for the COD removal and nitrification calculations for sludge ages 3–30 days are shown in Figures 14, 15, and 31.
4.14.26.1 Review of Calculations For the raw wastewater characteristics (i.e., fS0 up ¼ 0.15 mgCOD/mgCOD, fS0 us ¼ 0.07 mgCOD/mgCOD, Tmin ¼ 14 1C, Sti ¼ 750 mgCOD l1 – see Table 7) and 20 days sludge age, and accepting the nitrogen content of the volatile solids (fn) to be 0.10 mgN/mgVSS, the nitrogen required for sludge production Ns ¼ 17.0 mgN l1 (Equation (144)). From Section 4.14.23.2, the effluent biodegradable and unbiodegradable soluble organic nitrogen concentrations (Nobse and Nouse) are 0.0 and 1.80 mgN l1, respectively. From Equation (132) the effluent ammonia concentration Nae is 2.0 mgN l1. The effluent TKN concentration (Nte) is the sum of Nouse and Nae (Equation (150)) and hence Nte ¼ 3.8 mgN l1. The nitrification capacity (Nc) is found from Equation (152) and for the example raw wastewater (Nti ¼ 60.0 mgN l1; TKN/COD ¼ 0.08 mgN/mgCOD) at 14 1C is
Nc ¼ 60:0 17:0 3:8 ¼ 39:2 mgN l1 The nitrification oxygen demand, FOn is found from Equation (155), that is,
FOn ¼ 4:57Nc Qi ¼ 4:57 39:2 15 106 mgOd1 ¼ 2687 kgOd1 and the mass of nitrifier VSS in the reactor is given by Equation (154), that is,
MXBA ¼ 0:1 20=ð1 þ 0:034 20Þ 39:2 15 ¼ 702 kgVSS
486
Biological Nutrient Removal
In the design, because it is intended to reduce the nitrate concentration as much as possible, the alkalinity change in the wastewater will be minimized; assuming that 80% of the nitrate formed is denitrified, the H2CO3* alk change ¼ 7.14Nc 3.57 (nitrate denitrified) ¼ 7.14 39.2 þ 3.57 0.80 39.2 ¼ 168 mg l1 as CaCO3. With an influent H2CO3* alk of 250 mg l1 as CaCO3 the effluent H2CO3* alk ¼ 250–168 ¼ 82 mg l1 as CaCO3, which, from Figure 27, will maintain a pH above 7 (see Section 4.14.20.6).
4.14.26.2 Allocation of Unaerated Sludge Mass Fraction In nitrogen removal systems, the maximum anoxic sludge mass fraction available for denitrification, fxdm, can be set equal to the maximum unaerated sludge mass fraction fxm at the minimum temperature, that is,
f xdm ¼ f xm
ð166Þ
where fxm is given by Equation (136) for selected Rs, mnmT, and Tmin. This is because for N removal systems, unaerated sludge mass need not be set aside for the anaerobic reactor. In N and P removal systems, some of the unaerated sludge mass (0.12–0.25) needs to be set aside for the anaerobic reactor to stimulate BEPR. This sludge mass fraction, called the anaerobic sludge mass fraction and denoted fxa, cannot be used for denitrification. For BEPR to be as high as possible, no nitrate should be recycled to the anaerobic reactor so that zero denitrification takes place in this reactor. So, for the purposes of this development and demonstration of denitrification behavior, it will be accepted that the maximum unaerated sludge mass fraction available at 20 days sludge age (fxm) is all allocated to anoxic conditions, that is, fxdm ¼ fxm ¼ 0.534.
4.14.26.3 Denitrification Performance of the MLE System 4.14.26.3.1 Optimum recycle ratio a In the MLE system, the anoxic sludge mass fraction is all in the form of a primary anoxic reactor, that is, fx1 ¼ fxdm ¼ fxm. The denitrification potential of the primary anoxic reactor Dp1 is found from Equation (163), that is, for the example raw wastewater at 14 1C and fxm ¼ fxdm ¼ fx1 ¼0.534, Dp1 ¼ 52.5 mgN l1. The only additional wastewater characteristic required to calculate Dp1 is the influent RBSO (Sbsi) concentration or fraction (fSb’s), which for the example raw and settled wastewaters are given in Table 14, that is, 0.25 and 0.385 with respect to the biodegradable COD (Sbi), respectively. In the MLE system, if the nitrate concentration in the outflow of the anoxic reactor is zero, then the nitrate concentration in the aerobic reactor (Nnar) is equal to Nc/ (a þ s þ 1), that is, the nitrification capacity in mgN l1 influent flow diluted by the total (no nitrate containing) flow entering the aerobic reactor which is (a þ s þ 1) times the influent flow, where a and s are the mixed liquor and underflow recycle ratios (with respect to the influent average dry weather flow Qi), respectively. Accepting that there is no denitrification in the secondary settling tank (which needs to be minimized anyway due to the problem of rising sludges), the aerobic reactor and system effluent nitrate concentrations (Nnar and
Table 14 Additional wastewater characteristics required for denitrification (and BEPR) design Wastewater Readily biodegradable soluble organics (RBSO) as.y (1)y.fraction of biodegradable organics (BO, Sbi) COD (fSb0 s) (2)y.fraction of total organics (Sti, COD) (fS0 bs) VFA fraction of biodegradable soluble organics (RBSO), (fSbs0 a)
Raw
Settled
0.25
0.385
0.194
0.324
0.10
0.10
Nne, respectively) are equal and given by
Nne ¼ Nnar ¼ Nc =ða þ s þ 1Þ
ð167Þ
Knowing Nne and Nnar and taking into account DO concentrations in the a and s recycles, that is, Oa and Os mgO l1 respectively, the equivalent nitrate load on the primary anoxic reactor (Nnlp) by the a and s recycles is
Oa Os a þ Nne þ s Nnlp ¼ Nnar 2:86 2:86 The optimum denitrification (i.e., lowest effluent nitrate concentration) is obtained when the equivalent nitrate load on the anoxic reactor is equal to the denitrification potential of the anoxic reactor (i.e., Dp1 ¼ Nnlp), viz.,
Dp1 ¼
Nc Oa Nc Os þ þ aþ s ð168Þ ða þ s þ 1Þ 2:86 ða þ s þ 1Þ 2:86
Solving Equation (168) for a yields the a recycle ratio which exactly loads the primary anoxic reactor to its denitrification potential with nitrate and DO. This a value is the optimum because it results in the lowest Nne, that is,
aopt ¼ ½B þ
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi B 2 þ 4AC=ð2AÞ
ð169Þ
where
A ¼ Oa =2:86 B ¼ Nc Dp1 þ fðs þ 1ÞOa þ sOs g=2:86 C ¼ ðs þ 1ÞðDpp sOs =2:86Þ sNc and
Nnemin ¼ Nneaopt ¼ Nc =ðaopt þ s þ 1Þ ðmgN l1 Þ
ð170Þ
For a ¼ aopt, Equation (170) for Nne is valid and will give the minimum Nne attainable. When a raopt Equation (170) is also valid because the assumption on which Equation (169) is based is valid, that is, NnlprDp1 or equivalently, zero nitrate concentration in the outflow of the anoxic reactor. For a 4aopt this assumption is no longer valid and Nne increases as the a recycle ratio increases due to increasing DO flux entering the anoxic reactor. For a4aopt, Nne is given by the difference between the equivalent nitrate load on the anoxic reactor (which is the sum of the nitrification capacity Nc and the nitrate
Biological Nutrient Removal MLE system (settled) Effluent nitrate and a recycle ratio
MLE system (raw) Effluent nitrate vs. a recycle
20
487
20
14 °C s = 0.5 s = 1.0 s = 2.0 22 °C s = 1.0
15
10
Effluent nitrate (mgN I−1)
Effluent nitrate (mgN I−1)
a-opt (14 °C)
a-opt (14 °C) a-opt (22 °C)
5
14 °C s = 0.5 s = 1.0 s = 2.0
15
22 °C s = 1.0
10
N ne min (14 °C)
5
N ne min (14 °C)
0 0
0 5
10 a Recycle ratio
15
0
5
10 a Recycle ratio
15
Figure 37 Effluent nitrate concentration vs. mixed liquor a recycle ratio for the example raw (a) and settled (b) wastewaters for underflow (s) recycle ratio of 1:1 at 14 1C (bold line) and 22 1C (thin line) and for s ¼ 0.5:1 and 2.0:1 at 14 1C (dashed lines).
equivalent of the oxygen concentration with respect to the influent flow) and the denitrification potential Dp1, viz.,
Nne ¼ Nc þ
aOa sOs þ Dp1 2:86 2:86
ðmgN l1 Þ
ð171Þ
As Nc, Dp1, Os, and Oa are constants, the increase in Nne with increasing a above aopt is linear with slope Oa/2.86 mgN l1. At a ¼ aopt, Equations (170) and (171) give the same Nne concentrations. Accepting the design sludge age of 20 days, which allows a maximum unaerated sludge mass fraction fxm of 0.534, the denitrification behavior of the MLE system is demonstrated below for the example raw and settled wastewaters at 14 and 22 1C. In the calculations the DO concentrations in the a and s recycles, Oa and Os are 2 and 1 mgO l1, respectively, and the underflow recycle ratio s is 1:1. This s recycle ratio is usually fixed at a value such that satisfactory settling tank operation is obtained. Details of secondary settling tank theory, design, modeling, and operation are discussed by Ekama et al. (1997) and Ekama and Marais (2004). Substituting the values for the nitrification capacity Nc and denitrification potential Dp1 into Equations (169) and (170), the optimum mixed liquor recycle ratio aopt and minimum effluent nitrate concentration Nneaopt are obtained, for example, for the settled wastewater at 14 1C
A ¼ 2=2:86 ¼ 0:70 B ¼ 39:6 40:1 þ fð1 þ 1Þ2 þ 1 1g=2:86 ¼ þ1:52 C ¼ ð1 þ 1Þð40:1 1 1=2:86Þ 1 39:6 ¼ þ39:61 Hence, aopt ¼ 6.5 and Nnemin ¼ 4.7 mgN l1. The calculations for the example raw and settled wastewater at 14 and 22 1C show that for all four cases aopt exceeds 5. Although the calculations include the discharge of DO to the anoxic reactor, a recycle ratios above 5 to 6 are not cost effective. The small decreases in Nne which are obtained for even large increases in a recycle ratio above about 5:1 do not warrant the additional pumping costs.
This is illustrated in Figure 37 which shows Nne versus a recycle ratio for the example raw (Figure 37(a)) and settled (Figure 37(b)) wastewater at 14 and 22 1C plotted from Equations (170) and (171). For the settled wastewater (Figure 37(b)) at 14 1C and s ¼ 1:1, for aoaopt, the anoxic reactor is underloaded with nitrate and DO and as the a recycle increases up to aopt, the equivalent nitrate load increases toward the anoxic reactor’s denitrification potential. Initially, Nne decreases sharply for increases in a, but as a increases the decrease in Nne becomes smaller. At 14 1C with a ¼ aopt ¼ 6.5, the anoxic reactor is loaded to its denitrification potential by the a and s recycles and a Nnemin ¼ Nneaopt ¼ 4.7 mgN l1 is achieved. At a ¼ aopt ¼ 6.5, the greatest proportion of the anoxic reactor’s denitrification potential is used for denitrification and therefore yields the minimum effluent nitrate concentration (Nneaopt). This is shown in Figures 38(a) and 38(b) for the raw and settled wastewaters at 14 1C. For the settled wastewater at 14 1C (Figure 38(b)) at a ¼ aopt ¼ 6.5, 88% of the equivalent nitrate load (i.e. (a þ s) Nnemin ¼ 35.2 mgN l1 out of a Dp1 ¼ 40.1 mgN l1) is nitrate and therefore 88% of the denitrification potential of the anoxic reactor is utilized for denitrification and 12% for DO removal. The higher the a recycle ratio, the greater the proportion of the denitrification potential is utilized for DO removal. At 14 1C, for a4aopt, the equivalent nitrate load exceeds the denitrification potential and as the a recycle increases so Nne increases due to the increased DO mass flow to the anoxic reactor. From Equation (171), at a ¼ 15, Nne ¼ 10.6 mgN l1 and 27% of the denitrification potential is required to remove DO, leaving only 73% for denitrification (Figures 37(b) and 38(b)). For 14 1C, the plots of Nne versus a at underflow s recycle ratios of 0.5:1 and 2.0:1 are also given in Figure 37 and show that aopt is not significantly different at different s recycle ratios. Also, at low a recycle ratios, changes in s have a significant influence on Nne, but at high a recycle ratios, even significant changes in s do not significantly change Nne. This is because at high a, most of the nitrate is recycled to the anoxic reactor by the a recycle, so that changes in s do not
488
Biological Nutrient Removal MLE system (raw) use of denitrification potential 100
Denitrification potential used for nitrate removal
20
% Denit. potential
% Denit. potential
Denitrification potential used for nitrate removal
0
Denitrification potential used for nitrate removal
Denitrification potential used for nitrate removal
40
a-opt (14 °C) a-prac
0 0
(a)
60
20
a-opt (14 °C)
a-prac
Used DO removal
Unused denitrification potential
80
60 40
100
Used DO removal
Unused denitrification potential
80
MLE system (settled) use of denitrification potential
5
10 a Recycle ratio
15
0 (b)
5
10 a Recycle ratio
15
Figure 38 % Denitrification potential unused, used by dissolved oxygen in the recycles and for denitrification vs. a recycle ratio for the example raw (a) and settled (b) wastewaters for underflow (s) recycle ratio of 1:1 at 14 1C.
significantly change the nitrate load on the anoxic reactor. Hence, for the MLE system, decreases in s can be compensated for by increases in a – it makes little difference which recycle brings the nitrate to the anoxic reactor as long as the anoxic reactor is loaded as closely as practically possible to its denitrification potential in order to minimize Nne. For the settled wastewater at 22 1C and s ¼ 1:1 (Figure 37(b)), Nne versus a is similar to that at 14 1C up to a ¼ 6.5. This is because Nc values at 14 and 22 1C for the example raw and settled wastewater are almost the same (i.e., 39.9 and 41.6 mgN l1 at 14 and 22 1C, respectively). However, at 22 1C, the denitrification potential is significantly higher than at 14 1C (40.1 mgN l1 at 14 1C and 52.4 mgN l1 at 22 1C) so that a higher aopt is required (e.g., 17.9) at 22 1C to load the anoxic reactor to its denitrification potential than at 14 1C. Therefore, at 22 1C, as the a recycle increases above 6.5, Nne continues to decrease until aopt ¼ 17.9 is reached. The increase in a from 6.5 to 17.9 reduces Nne from 4.9 to 2.1, that is, only 2.8 mgN l1. This small decrease in Nne is not worth the large increase in pumping costs from 6.5:1 to 17.9:1 required to produce it. Consequently, for economical reasons, the a recycle ratio is limited at a practical maximum (aprac) of say 5:1, which fixes the lowest practical effluent nitrate concentration (Nneprac) from the MLE system between 5 and 10 mgN l1 depending on the influent TKN/ COD ratio. From the design procedure demonstrated so far, it is clear that the procedure hinges around balancing the equivalent nitrate load with the denitrification potential by appropriate choice of the a recycle ratio: for selected system design parameters (sludge age, anoxic mass fraction, underflow recycle ratio, etc.) and wastewater characteristics (temperature, readily biodegradable COD fraction, TKN/COD ratio, etc.), the denitrification potential of the MLE system is fixed. With all the above fixed, the system denitrification performance is controlled by the a recycle ratio, and this performance is optimum when the a recycle ratio is set at the optimum aopt. For aoaopt, the performance will be below optimum because the equivalent nitrate load is less than the denitrification potential (Figure 38); for a ¼ aopt, the performance is optimal because the equivalent
nitrate load equals the denitrification potential; and for a4aopt, the performance is again suboptimal because now the equivalent nitrate load is greater than the denitrification potential and more than necessary DO is recycled to the anoxic reactor which reduces the denitrification (see Figures 37 and 38). If a practical limit on a is set at say aprac ¼ 5:1 and aopt is significantly higher, then a significant proportion of the anoxic reactor’s denitrification potential is not used (Figure 38). There are two options to deal with this unused denitrification potential: (1) change the design, that is, decrease the sludge age (Rs) and/or unaerated sludge mass fraction (fxm) or (2) leave the system as designed (i.e., Rs ¼ 20 days and fxm ¼ 0.534) and keep the unused denitrification potential in reserve as a factor of safety against changes in wastewater characteristics, such as (1) increased organic load, which will require a reduction in sludge age, (2) increased TKN/COD ratio, which will load the anoxic reactor with nitrate at lower a recycle ratios, or (3) decreased RBSO fraction, which decreases the anoxic reactors denitrification potential.
4.14.26.3.2 The balanced MLE system With option (1) the anoxic sludge mass fraction fx1 is decreased to eliminate the unused denitrification potential. The decrease in fx1 increases the aerobic mass fraction and therefore the factor of safety (Sf) on nitrification. To maintain the same Sf, the sludge age of the system can be reduced to that value at which the lower fx1 is equal to the maximum unaerated sludge mass fraction fxm allowed (i.e., fx1 ¼ fxm) for the selected mAm20 and Tmin. An MLE system with a sludge age (Rs) and influent TKN concentration (Nti) such that fx1 ¼ fxm and aopt ¼ aprac (say 5:1), so that this aprac loads the anoxic reactor exactly to its denitrification potential, is called a balanced MLE system. This approach to design of the MLE system was proposed by van Haandel et al. (1982) and gives the most economical AS reactor design, that is, the lowest sludge age, and therefore the smallest reactor volume, and the highest denitrification with the a recycle ratio fixed at some maximum practical limit. The influent TKN/COD ratio, fxm ¼ fx1, fx1 min, Nne, and %N removal (%Nrem) versus sludge age for balanced
Biological Nutrient Removal MLE system (settled, 14 °C) Design at fixed a -opt = 5:1
MLE system (raw, 14 °C) Design at fixed a -opt = 5:1 Balanced sludge age
0.8
0.12
f x1 = f xm
0.6 0.4
TKN/COD Raw WW
0.08
0.2
Influent TKN/COD ratio
0.14
1.0
0.16 Balanced sludge age
0.14 0.12
0.6
Settled WW f x1 = f xm
0.10
0.4
0.08
0.0 0
5
(a)
10 20 15 Sludge age (days)
25
0.06
30
0.0 0
100
20
25
30
100
75
Balanced sludge age
10
50
Effluent nitrate
5
25
0
0 0
5
10 15 20 Sludge age (days)
25
15
75
Balanced sludge age
50
10 Effluent nitrate
5
25
0
0
30
0 (d)
% Nitrogen removal
%N removal
15
Effluent nitrate (mgN I−1)
%N removal % Nitrogen removal
Effluent nitrate (mgN I−1)
20 10 15 Sludge age (days)
MLE system (settled, 14 °C) Design at fixed a -opt = 5:1
MLE system (raw, 14 °C) Design at fixed a -opt = 5:1
(c)
5
(b)
20
0.2
f x1min
f x1min 0.06
0.8
TKN/COD
Anoxic mass fraction
1.0 Anoxic mass fraction
Influent TKN/COD ratio
0.16
0.10
489
5
10 15 20 Sludge age (days)
25
30
Figure 39 Influent TKN/COD ratio (TKN/COD), maximum unaerated (fxm), primary anoxic (fx1), and minimum primary anoxic (fx1 min) sludge mass fractions (a,b) and effluent nitrate concentration and %N removal (c,d) for balanced MLE systems with a 5:1 practical upper limit to the a recycle ratio for the example raw (a,c) and settled (b,d) wastewaters at 14 1C.
MLE systems for the example raw and settled wastewaters at 14 and 22 1C are shown in Figures 39 and 40, respectively. The sludge age which balances the MLE system for given wastewater characteristics and aprac cannot be calculated directly. It is easiest to calculate the influent TKN concentration for a range of sludge ages and choose the sludge age which matches the wastewater TKN concentration (Nti). The procedure for calculating Nti for a balanced MLE system is as follows: from the design mAm20, Tmin, Sf, and a selected sludge age, fxm is calculated from Equation (136). Provided fxm4fx1 min (Equation (165)), fx1 is set equal to fxm. Knowing fx1 and the wastewater characteristics, Dp1 is calculated from Equation (163). This Dp1 and a selected value for aprac are then substituted into Equation (168), which sets the equivalent nitrate load on the anoxic reactor equal to the denitrification potential and hence aopt equals the selected aprac. With Dp1 and a known, Nc is calculated from Equation (168). Once Nc is known, Nti is calculated from Nti ¼ Nte þ Ns þ Nc (Equation (152)), where Nte ¼ Nouse þ Nae (Equation (150)) and Nae is given by Equation (132) because with Sf fixed the Rs fxm relationship is fixed. With Nc and Nti known, the effluent nitrate concentration Nne and % nitrogen removal (%Nrem) are
found from Equation (170) and %Nrem ¼ 100[Nti (Nne þ Nte)]/Nti, respectively. This calculation is repeated for different sludge ages. The shortest sludge age allowed is the one which gives fx1 ¼ fxm ¼ fx1min. In Figure 39, for 14 1C, for the raw wastewater (Figures 39(a) and 39(c)), it can be seen that fx1( ¼ fxm) increases from about 0.09 at 8 days sludge age, at which fxm is just greater than fx1 min, to 0.60 at 26 days sludge age, at which fxm is equal to the upper limit set for it. As fx1 increases so the influent TKN/COD ratio increases from 0.061 at 8 days sludge age to 0.115 at 26 days sludge age. With the increase in TKN/COD ratio, the nitrification capacity Nc increases and hence Nne increases from about 3.2 mgN l1 at 8 days sludge age to 9.3 mgN l1 at 26 days sludge age because the a and s recycle ratios remain at 5:1 and 1:1, respectively (see Equation (170)). The %N removal, which includes the N removed via sludge wastage Ns, decreases marginally from 85% to 82% as the influent TKN/COD ratio and sludge age increase for the balanced MLE system. For the settled wastewater at 14 1C (Figures 39(b) and 39(d)), the influent TKN/COD ratio, fx1 and fx1min results are similar to those for the raw wastewater, that is, for the same
Biological Nutrient Removal MLE system (raw, 22 °C) Design at fixed a-opt = 5:1 1.0
0.6
f x1 = f xm
0.10
0.4 Raw WW
0.2
0.14
Balanced sludge age
0.12
0.0
0.10
0.4
0.08
0.2
100
75 Effluent nitrate
50
25
Balanced sludge age
% Nitrogen removal
15
5
0 0
5
10 15 20 Sludge age (days)
5
25
20 10 15 Sludge age (days)
25
30
MLE system (settled, 22 °C) Design at fixed a-opt = 5:1
20
%N removal
10
0.0 0
(b)
0 (c)
0.06
30
MLE system (raw, 22 °C) Design at fixed a-opt = 5:1
20 Effluent nitrate (mgN I−1)
25
Effluent nitrate (mgN I−1)
(a)
10 15 20 Sludge age (days)
0.6
f x1mm
0.06 5
f x1 = f xm
Settled WW
f x1mm 0
0.8
Anoxic mass fraction
0.8
0.08
1.0 TKN/COD
0.14 0.12
0.16
TKN/COD Influent TKN/COD ratio
Balanced sludge age
Anoxic mass fraction
Influent TKN/COD ratio
0.16
MLE system (settled, 22 °C) Design at fixed a-opt = 5:1
100
%N removal 15
75
10
50 Effluent nitrate
5
25
Balanced sludge age
0
30
% Nitrogen removal
490
0 0
(d)
5
10 15 20 Sludge age (days)
25
30
Figure 40 Influent TKN/COD ratio (TKN/COD), maximum unaerated (fxm), primary anoxic (fx1), and minimum primary anoxic (fx1 min) sludge mass fractions (a,b) and effluent nitrate concentration and %N removal (c,d) for balanced MLE systems with a 5:1 practical upper limit to the a recycle ratio for the example raw (a,c) and settled (b,d) wastewaters at 22 1C.
sludge age approximately the same TKN/COD ratio is found for the balanced MLE system. For the settled wastewater, the Nne is slightly lower, increasing from about 3.2 to 6.7 mgN l1 from 8 to 26 days sludge age; also the %N removal is somewhat lower, around 78% mainly due to the lower N removal via sludge wastage Ns. However, it must be remembered that the TKN/COD ratio and RBSO fraction of a settled wastewater are higher than those of the raw wastewater from which it is produced, viz. TKN/COD ratio 0.113 and 0.080 mgN/mgCOD and RBSO fraction (fSb’s) 0.25 and 0.385 for the example settled and raw wastewaters, respectively. Therefore, at 14 1C, while the raw wastewater can be treated in a balanced MLE system at about 11 days sludge age (Figure 39(a)), the sludge age for the settled wastewater balanced MLE system is about 17 days (Figure 39(b)). A comparison of the balanced MLE systems for the example raw and settled wastewaters is given in Table 15. From Table 15 it can be seen that Nne is less than 1 mgN l1 higher for the settled wastewater but the reactor volume and total oxygen demand significantly lower compared with the
Table 15 Comparison of balanced MLE systems treating the example raw and settled wastewaters at 14 1C Parameter
Raw
Settled
Influent TKN/COD ratio Unaerated mass fraction(fxm) Anoxic mass fraction (fx1) Minimum anoxic fraction a Recycle ratio (aprac ¼ aopt) Sludge age (days) Effluent nitrate (Nne, mgN l1) Effluent TKN (Nte, mgN l1) Reactor vol. at 4.5 gTSS l1 (m3) Carb O2 demand (FOc, kgO d1) Nit O2 demand (FOn, kgO d1) O2 recovered (FOd, kgO d1) Tot. O2 demand (FOtd, kgO d1) %N removal Mass TSS wasted (FXt, kg d1) Active fraction wrt TSS (fatOHO)
0.08 0.306 0.306 0.08 5:1 11 5.1 4.3 9484 6156 2492 1327 7321 84.3 3880 0.316
0.113 0.485 0.485 0.108 5:1 17 5.7 4.1 5264 4251 2685 1437 5499 80.9 1394 0.414
Biological Nutrient Removal
raw wastewater. Therefore, from an AS system point of view, treating settled wastewater would be more economical than treating raw wastewater for a comparable effluent quality. Also, both systems require sludge treatment; for the raw wastewater because 11 days sludge age waste sludge is not stable (high active fraction, favOHO) and for the settled wastewater, the primary sludge needs to be stabilized. The 11 days sludge age waste sludge can be stabilized with anoxic aerobic digestion which allows the N released in digestion to be nitrified and denitrified (Warner et al. 1986; Mebrahtu et al., 2010) and primary sludge can be anaerobically digested to benefit from gas generation. The choice of treating raw or settled wastewater therefore does not depend so much on the effluent quality or the economics of the AS system itself, but on the economics of the whole WWTP, including sludge treatment. Because the minimum wastewater temperature (Tmin) governs the AS system (and sludge treatment) design, the balanced MLE system results for 22 1C are not particularly relevant to the temperate climate regions. However, in equatorial and tropical regions, where wastewater treatment is becoming a matter of increasing concern, high wastewater temperatures are encountered. For this reason and for illustrative purposes also, the balanced MLE results for the raw and settled wastewaters are shown in Figure 40. Compared with 14 1C, the upper limit to fxm ¼ 0.60 is reached already at 7 days sludge age and significantly higher influent TKN/COD ratios can be treated at equal sludge ages. These higher TKN/COD ratios result in higher Nne, which for the raw wastewater increases from 3 to 13 mgN l1 and for the settled wastewater from 3 to 9 mgN l1 for increases in sludge age from 4 to 30 days. If Tmin were 22 1C, the example raw and settled wastewaters could be treated at 3 and 4 days sludge age, respectively, yielding Nne of 5 and 6.5 mgN l1, respectively. This reinforces the conclusion in Section 4.14.24.1 that in equatorial and tropical climates it is highly likely that AS plants will nitrify even at very short sludge ages (1–2 days) and therefore to design for denitrification for operational reasons if not for effluent quality reasons.
4.14.26.3.3 Effect of influent TKN/COD ratio When the unused denitrification potential in the anoxic reactor is kept in reserve as a safety factor (option 2), the sludge age and unaerated (anoxic) mass fraction are not changed. For this situation, it is useful to have a sensitivity analysis to see the influence of changing influent TKN/COD ratio and RBSO fraction on the a recycle ratio and effluent nitrate concentration. Continuing with the design for the example raw and settled wastewaters for fixed sludge age at 20 days and unaerated (anoxic) mass fraction at 0.534, a plot of the optimum a recycle ratio aopt and minimum effluent nitrate concentration Nneaopt for underflow recycle ratios s of 0.5, 1.0, and 2.0 versus influent TKN/COD ratio from 0.06 to 0.16 is given in Figure 41 for the raw (a), (c) and settled (b), (d) wastewaters at 14 1C (a), (b) and 22 1C (c), (d). From Figure 41, it can be seen that as the influent TKN/ COD increases, aopt decreases and Nneaopt increases. The aopt–Nneaopt lines in Figure 41 give the system denitrification performance when the denitrification potential of the anoxic reactor is fully used, that is, the system denitrification
491
performance is equal to its denitrification potential and the nitrate concentration is the lowest possible. Also, large increases in the underflow recycle ratio s (i.e., from 0.50:1 to 1.0:1 or 1.0:1 to 2.0:1) decrease aopt but do not change Nneaopt because the DO in the a and s recycles does not differ much in their influence on the anoxic reactor. Therefore, it matters little which recycle flow brings the nitrate load to the anoxic reactor. As long as the anoxic reactor is closely loaded to its denitrification potential, the same minimum effluent nitrate concentration (Nneaopt) will be obtained at aopt. The aopt–Nneaopt lines therefore give the system denitrification performance when the denitrification potential of the anoxic reactor is fully used (Figure 38(b)), that is, the systems denitrification performance is equal to its potential. A better denitrification performance is not possible – the denitrification is kinetics limited and the biomass (and so also the system) does the best it can (for the given K2 denitrification rate). From Equation (170), the system denitrification performance with increasing influent TKN/COD ratio at a fixed practical operating a recycle ratio (aprac) of 5:1 is also shown in Figure 41 as the aprac and Nneaprac lines. It can be seen that Nneaprac increases linearly with increase in influent TKN/COD ratio. For low influent TKN/COD ratios, aprac is considerably lower than aopt and the system denitrification performance is lower than its denitrification potential. This is evident from Nneaprac being greater than Nneaopt. As the TKN/COD ratio increases, aopt decreases until aopt ¼ aprac ¼ 5.0:1. For the raw wastewater at 14 1C (Figure 41(a)), this happens at an influent TKN/COD ratio of 0.104. This is the influent TKN/COD ratio which balances the MLE system for the selected design conditions, namely, 20 days sludge age, fxm ¼ 0.534 and aprac ¼ 5:1 for the example raw wastewater at 14 1C. For influent TKN/ COD ratios 40.104, the a recycle ratio should be set at aopt, which fully uses the anoxic reactor’s denitrification potential and is now lower than aprac ¼ 5:1. Therefore for aprac set at 5:1, only when the influent TKN/COD ratio is 40.104, is the denitrification potential of the anoxic reactor fully used. This same conclusion can be made from Figure 39(a) at 20 days sludge age, that is, fxm ¼ 0.534 and TKN/COD ratio ¼ 0.104. Therefore for influent TKN/COD ratioso0.104, while aprac oaopt, the system denitrification performance is lower than its denitrification potential because not all the denitrification potential of the anoxic reactor is used. Once the TKN/COD ratio increases above that value which balances the MLE system, aoptoaprac and a should be set at aopt to achieve the lowest effluent nitrate concentration (Nneaopt). For these influent TKN/COD ratios, the denitrification potential of the anoxic reactor is fully used and the system denitrification performance is defined by the aopt–Nneaopt lines. Figure 41 is useful because it combines the system denitrification performance (aprac–Nneaprac lines) and the denitrification potential (aopt–Nneaopt lines) in the same diagram as influent TKN/COD ratio increases for a particular wastewater and system design (Rs ¼ 20 days and fxm ¼ 0.534). The intersection point of the straight Nneaprac line and the curved Nneaopt line (i.e., at aopt ¼ aprac ¼ 5:1) gives the influent TKN/ COD ratio for the balanced MLE system for the selected aprac ¼ 5:1. From Figure 41(a), for the raw wastewater at 14 1C, the MLE system (at 20 days sludge age and fxm ¼ 0.534) with a
Biological Nutrient Removal
10
20 15
a-prac
5
0.06
0.08
0.10
15
a Recycle ratio
Example Raw WW
a-opt
25 20 15 s=0.5 s =1.0 s =2.0
a-prac Nne-a-prac Nne-a-opt
a-opt
0 0.06
10 5
0.10
0.12
0.14
s =0.5 s =1.0 s=2.0
a-opt Nne-a-opt
0 0.06
0.08
0.10
0.12
10 5 0
0.16
0.14
Influent TKN/COD ratio MLE system (settled, 22 °C) a Recycle ratio and effluent nitrate
15
30 a-opt 25
5
20
a-prac
s =0.5 s =1.0 s =2.0
a-opt
15 10
Nne-a-prac
5 Nne-a-opt
0
0.16
Influent TKN/COD ratio
Example Settled WW
10
0 0.08
a-prac Nne-a-prac
30
10
5
5
(b)
MLE system (raw, 22 °C) a Recycle ratio and effluent nitrate
20 15
0.16
0.14
25
10
0 0.12
Influent TKN/COD ratio
(a)
(c)
5
a-opt
Nne-a-opt
0
10
s=0.5 s=1.0 s=2.0
Nne-a-prac
a Recycle ratio
a-opt
30
Example Settled WW
a-opt
a Recycle ratio
a Recycle ratio
25
15
Effluent nitrate (mgN I−1)
30
Example Raw WW
Effluent nitrate (mgN I−1)
15
MLE system (settled, 14 °C) a Recycle ratio and effluent nitrate
Effluent nitrate (mgN I−1)
MLE system (raw, 14 °C) a Recycle ratio and effluent nitrate
Effluent nitrate (mgN I−1)
492
0.06 (d)
0 0.10 0.12 0.08 0.14 Influent TKN/COD ratio
0.16
Figure 41 Optimum (aopt) and practical upper limit (aprac ¼ 5:1) a recycle ratios (bold lines) and effluent nitrate concentration at aopt (Nneaopt, bold line) and aprac (Nneaprac, dashed line) vs. influent TKN/COD ratio at underflow (s) recycle ratio of 1:1 for the example raw (a,c) and settled (b,d) wastewaters at 14 1C (a,b) and 22 1C (c,d). The optimum a recycle ratio (aopt) values at underflow recycle ratios of 0.5:1 and 2:1 are also shown (thin lines).
recycle ratio 45:1 can maintain effluent nitrate concentrations below 8.1 (total N 12.4) mgN l1 for influent TKN/COD ratios below 0.104 (78.0 mgN l1). With settled wastewater at 14 1C (Figure 41(b)), the MLE system with a 45:1 can maintain effluent nitrate concentrations below 11.3 (total N 14.9) mgN l1 for influent TKN/COD ratios up to 0.132 (59.4 mgN l1). Similarly, from Figures 41(c) and (d), with raw and settled wastewater at 22 1C, the MLE system with a 45:1 can maintain effluent nitrate concentrations below 6.0 and 8.1 mgN l1 (total N 9.9 and 11.1 mgN l1) for influent TKN/COD ratios up to 0.119 (89.3 mgN l1) and 0.148 (66.6 mgN l1). These results show that the MLE system treating settled wastewater delivers lower Nne (by 2–3 mgN l1) than when treating raw wastewater and at influent TKN/ COD ratios significantly higher. However, it should be noted that (1) the influent TKN concentrations (given above) for the raw wastewater are considerably higher than those for the settled wastewater and (2) a settled wastewater with a TKN/ COD ratio of 0.119 (14 1C) or 0.148 (22 1C) would be produced from a raw wastewater with considerably lower influent TKN/COD ratio than 0.104 (14 1C) and 0.132 (22 1C).
4.14.26.3.4 MLE sensivity diagram In Figure 41, the system denitrification performance at a selected aprac ¼ 5 is combined with the system denitrification potential at a ¼ aopt for varying influent TKN/COD ratio and a single influent RBSO fraction value. This influent TKN/COD ratio sensitivity diagram can be extended by adding the Nneaopt lines for other influent RBSO fractions. A sensitivity analysis of the system at the design stage is useful for evaluating the denitrification performance under varying influent TKN/COD ratio and RBSO fractions. These two wastewater characteristics can vary considerably during the life of the plant and have a major impact on the N removal performance of the system. The denitrification potential and system performance are combined for varying influent TKN/COD ratio and RBSO fraction in Figure 42. For the fixed system design parameters (i.e., Rs ¼ 20 days, fxdm ¼ fxm ¼ 0.534, s ¼ 1.0), the curved (bold) lines give Nneaopt when the anoxic reactor is loaded to its denitrification potential, that is, Nne for a ¼ aopt for varying TKN/COD ratio from 0.06 to 0.16 and RBSO fractions from 0.10 to 0.35 for the example raw and settled wastewaters at
Biological Nutrient Removal MLE system (settled, 14 °C) Effluent nitrate vs. TKN/COD ratio
MLE system (raw, 14 °C) Effluent nitrate vs. TKN/COD ratio 40
20 0.5 1
30 RBCOD fraction 20
2 3
0.10 0.15
4
0.20 0.25 0.30 0.35
10
5
6
Effluent nitrate (mgN I−1)
Effluent nitrate (mgN I−1)
0
15
a recycle ratio 0 0.5 0.10
10
1 2
0.35
4
5 6 8 10
10 a recycle ratio
0.30
0.25
3
RBCOD fraction
0 0.06
0.08
0.10
0.12
0.14
0.16
Influent TKN/COD ratio
(a)
0.06
40
0.20
0 0.5
20
1
0.25 0.30 0.35
2 10
3
Effluent nitrate (mgN I−1)
0.15
15
0.15
0.5
0.20
1 10
0.25 2 3 4
5
5
4
5 6 8 10
8 RBCOD fraction
0
0.16
0.10
a recycle ratio 0
0.10 a recycle ratio
0.10 0.12 0.14 Influent TKN/COD ratio
MLE system (settled, 22 °C) Effluent nitrate vs. TKN/COD ratio
20
30
0.08
(b)
MLE system (raw, 22 °C) Effluent nitrate vs. TKN/COD ratio
Effluent nitrate (mgN I−1)
0.20 0.15
5
8
0
0.30 0.35
6 10
RBCOD fraction
0 0.06
(c)
493
0.08
0.10 0.12 0.14 Influent TKN/COD ratio
0.16
0.06 (d)
0.08
0.10 0.12 0.14 Influent TKN/COD ratio
0.16
Figure 42 Effluent nitrate concentration vs. influent TKN/COD ratio for influent readily biodegradable (RBSO) fractions (fSb0 s) of 0.10, 0.15, 0.20, 0.25, 0.30, and 0.35 and mixed liquor a recycle ratio from 0 to 10 for the example raw (a,c) and settled (b,d) wastewaters at 14 1C (a,b) and 22 1C (c,d).
14 1C (Figures 42(a) and 42(b)) and 22 1C (Figures 42(c) and 42(d)). The same Nneaopt lines are given in Figures 41(a) and 41(c) for the raw wastewater RBSO fraction (fSb0 s) ¼ 0.25. These Nneaopt lines are calculated from Equations (169) and (170). The straight lines in Figure 42 give Nneaprac for fixed a recycle ratios at indicated values ranging from 0.0:1 to 10:1. These straight Nneaprac lines give the system performance for some selected a recycle ratio and are calculated with the aid of Equation (170) from the nitrification capacity value at the given TKN/COD ratio, fixed s recycle ratio at 1.0:1, and the selected a recycle ratio. The Nneaprac lines for a ¼ aprac ¼ 5:1 are
the same as the dotted lines in Figure 41. At the intersection points of the straight Nneaprac and curved Nneaopt lines, the system performance equals the denitrification potential and represents balanced MLE designs, that is, aopt ¼ aprac. For example, for the raw wastewater at 14 1C, at a ¼ 5.0:1 and fSb0 s ¼ 0.25, the TKN/COD ratio needs to be 0.104 to give an optimal design, that is, aopt ¼ 5:1 and at this TKN/COD ratio, Nne ¼ 8.1 mgN l1. This is the TKN/COD ratio that balances the MLE system at Rs ¼ 20 days and fxm ¼ 0.534 (see Figure 39). For TKN/COD ratioso0.104, aopt increases above 5:1, but if a is maintained at 5:1 (i.e., a ¼ aprac ¼ 5:1), then Nne
494
Biological Nutrient Removal
versus TKN/COD ratio is given by the a ¼ 5:1 straight Nneaprac line. For TKN/COD ratios 40.104, aopt decreases below 5:1, and Nne versus TKN/COD ratio is given by the curved Nneaopt (bold) line. The aopt value at a particular TKN/COD ratio is given by the a recycle ratio value of the intersection point between the vertical influent TKN/COD ratio line and the curved Nneaopt line, for example, for the example raw wastewater (fSb0 s ¼ 0.25) at 14 1C (Figure 42(a)) at a TKN/COD ratio of 0.12, aopt ¼ 2:1, and Nne is 16.0 mgN l1. The usefulness of Figure 42 is that it gives a performance evaluation of an MLE system at a specified sludge age and anoxic mass fraction for varying influent TKN/COD ratio and RBSO fraction taking due account of an upper a recycle ratio limit of aprac. For the example raw wastewater at 22 1C with a RBSO fraction (fSb0 s) of 0.10 (Figure 42(c)), the influent TKN/ COD ratio needs to be greater than 0.113 for a to beo6.0:1. If a is fixed at aprac ¼ 6.0:1 and the TKN/COD is o0.113, then the anoxic reactor is underloaded with nitrate and the denitrification potential is not achieved. The system performance for influent TKN/CODo0.113 is given by the straight Nne line for a ¼ 6:1. At influent TKN/COD ¼ 0.113, the straight Nne line for a ¼ 6:1 cuts the curved Nneaopt line, a ¼ aopt ¼ 6:1 and the system performance equals the denitrification potential. If a is maintained at 6:1 for TKN/COD 40.113, then the anoxic reactor is overloaded with nitrate and optimal denitrification is not achieved due to the unnecessarily high DO load on the anoxic reactor (similar to that shown in Figure 37(b) for a 46.7). The a recycle ratio therefore should be reduced to aopt for influent TKN/COD ratios 40.113, where aopt is given by the a value along the curved Nne line, which represents system performance equal to denitrification potential. For example, if the TKN/COD ratio ¼ 0.120, a ¼ aopt ¼ 4:1 and this a recycle ratio loads the anoxic reactor to its denitrification potential giving Nne of 12.0 mgN l1. Therefore, for TKN/COD ratio 40.113, the system performance and Nne is given by the curved Nne line provided the a recycle ratio is set to aopt, which is given by a recycle ratio line which passes through the intersection point of the vertical TKN/COD ratio line and the curved Nne line. From the above, it can be seen that only on the curved Nne line for the particular RBSO fraction is the system performance equal to the denitrification potential; also the aopt that produces this is given by the a recycle ratio line that passes through the intersection point of the vertical TKN/COD ratio line and the curved Nne line. This curved Nne line (for which a ¼ aopt) marks the boundary between underloaded and overloaded conditions in the anoxic reactor. In the domain above the curved Nne line, the anoxic reactor is underloaded (left of aopt in Figures 37 and 38) and the system performance (Nne) for a particular TKN/COD ratio is given by the intersection point of the vertical TKN/COD ratio line and the straight a recycle ratio line. In the domain below, curved Nne line, the anoxic reactor is overloaded (right of aopt in Figures 37 and 38). The Nne values obtained from this domain are not valid, but if the a recycle ratio is reduced to aopt (i.e., the a value of the intersection point of the vertical TKN/ COD ratio line and the curved Nne line), then the Nne value again is valid. Valid Nne system performance values are therefore given in Figure 42 only on or above the curved Nne boundary line.
From Figure 42, it can be seen that for MLE system at the design Rs ¼ 20 days and fxm ¼ 0.534 and a recycle ratio limited at say 5.0:1 for economical reasons, then the system is best suited to treating high TKN/COD ratios, depending on the RBSO fraction: 40.091 for fSb0 s ¼ 0.10 and 40.117 for fSb0 s ¼ 0.35. This is because with only a primary anoxic reactor, the MLE system cannot produce a low effluent nitrate concentration (o4–6 mgN l1) at a recycle ratio limited at 5.0:1. If obtaining low effluent nitrate concentrations is not required at low TKN/COD ratios, then a balanced MLE design can be selected by reducing the sludge age as demonstrated in Figures 39 and 40. If obtaining low effluent nitrate concentrations is important at low (o0.10) TKN/COD ratios, then this can be achieved at high a recyle ratios (aopt4aprac) in MLE systems or at low a recycle ratios by including a secondary anoxic reactor. Incorporation of a secondary anoxic reactor (and a re-aeration reactor for practical reasons – see Section 4.14.24.5) produces the four-stage Bardenpho system (Figure 34(c)). However, because the K3 denitrification rate is so low and needs to be reduced by at least 20% to account for the ammonia released during endogenous denitrification (which is re-nitrified in the re-aeration reactor), the net additional nitrate removal achieved in a secondary anoxic reactor is very low, too low for secondary anoxic reactors to be included in N removal systems, unless the influent TKN/COD ratio is unusually low.
4.14.27 System Volume and Oxygen Demand 4.14.27.1 System Volume Having determined the subdivision of the sludge mass into anoxic and aerobic fractions to achieve the required N removal, the actual sludge mass in the system needs to be calculated to determine the volumes of the different reactors. The mass of sludge, total (MLSS) or volatile (MLVSS), in the system for selected sludge age and wastewater characteristics for N removal system is the same as for fully aerobic (COD removal) systems. The equations given in Section 4.14.9 therefore apply to N removal systems also. For the example raw and settled wastewaters, the design parameters for the MLE system are listed in Table 16. The MLSS mass values in the system at 20 days sludge age and 14 1C are 68168 and 26 422 kgTSS, respectively. Selecting an MLSS concentration of 4500 mg l1 (4 kg m3) (see Section 4.14.11) means that the volume of the system treating raw wastewater is 15148 m3 and that treating settled wastewater is 5871 m3. Because the sludge mass in the N removal systems usually is uniformly distributed in the system, that is, each reactor of the system has the same MLSS concentration, the volume fraction of each reactor is equal to its sludge mass fraction. For the example raw and settled wastewaters at 14 1C, the volume of the anoxic reactors are 0.534 15148 ¼ 8089 m3 and 0.534 5871 ¼ 3135 m3, respectively. The nominal and actual hydraulic retention times of the anoxic and aerobic reactors are calculated from the reactor volumes divided by the nominal (influent) and total flows passing through them (Equation (59) and Table 16). Note that the reactor nominal retention time is a consequence of the mass of sludge generated from the influent COD flux, the selected MLSS concentration, and the sludge mass fraction – the retention time per se has no significance in
Biological Nutrient Removal Table 16 Design details of MLE systems treating the example raw and settled wastewaters at 14 1C at 20 days sludge and 0.534 unaerated sludge mass fraction Parameter
Raw
Settled
Influent TKN/COD ratio Influent RBCOD fraction (fSb0 s) Unaerated mass fraction(fxm) Anoxic mass fraction (fx1) Minimum anoxic fraction a Recycle ratio (apracoaopt) Sludge age (days) Effluent nitrate (Nne, mgN l1) Effluent TKN (Nte, mgN l1) Effluent total N (Nne þ Nte) System vol at 4.5 gTSS l1 (m3) Anoxic volume (m3) System ret time – nom (h) Aerobic ret time – nom (h) Aerobic ret time – actual (h) Anoxic ret time – nom (h) Anoxic ret time – actual (h) Carb. O2 demand (FOc, kgO d1) Nit O2 demand (FOn, kgO d1) O2 recovered (FOd, kgO d1) Tot. O2 demand (FOtd, kgO d1) %N removal Mass TSS wasted (FXt, kg d1) Active fraction wrt TSS (fatOHO)
0.08 0.25 0.534 0.534 0.07 5:1 20 5.6 3.8 9.4 15148 8089 24.2 11.2 1.6 12.9 1.85 6679 2685 1440 7924 84.4 3408 0.23
0.113 0.385 0.534 0.534 0.105 5:1 20 5.7 3.8 9.5 5871 3135 9.4 4.4 0.63 5 0.72 4311 2719 1458 5572 81.4 1321 0.383
kinetics of and design for nitrification and denitrification (see Section 4.14.9.3).
4.14.27.2 Daily Average Total Oxygen Demand The total oxygen demand in a nitrogen removal system is the sum of that required for organic material (COD) degradation and nitrification, less than recovered by denitrification. The daily average oxygen demand for (1) organic material removal (FOc) is given by Equations (111) and (2) nitrification is given by Equation (155). These oxygen demands in the MLE system at 20 days sludge age for the example raw and settled wastewaters at 14 and 22 1C are 9364 and 7030 kgO d1 (Table 16). The oxygen recovered by denitrification (FOd) is given by 2.86 times the nitrate flux denitrified (Section 4.14.24.2) where nitrate flux denitrified is the product of the daily average influent flow Qi and the nitrate concentration denitrified. The nitrate concentration denitrified is given by the difference in the nitrification capacity Nc and the effluent nitrate concentration. Hence,
FOd ¼ 2:86ðNc Nne ÞQi
ðmgO d1 Þ
495
demand by incorporating ND is only 20% of that required for COD removal only, and (4) the effect of temperature on the total oxygen demand is marginal – less than 3% (see also Figure 32). For the settled wastewater, Table 16 shows that (1) the nitrification oxygen demand is about 63% of that required for COD removal; (2) about 54% of the nitrification oxygen demand can be recovered by denitrification; (3) the additional oxygen demand by incorporating nitrification and denitrification is about 30% of that required for COD removal only, and (4) the effect of temperature on the total oxygen demand is marginal – less than 3% more at the lower temperature. Comparing the total oxygen demand (FOtd) for the raw and settled wastewaters, the total oxygen demand for the latter is about 30% less than that of the former. This saving is possible because primary sedimentation removes 35–45% of the raw wastewater COD. Furthermore, for the settled wastewater, the nitrification oxygen demand is a greater proportion of the total; also, less of the nitrification oxygen demand can be recovered by denitrification compared to the raw wastewater. These effects are due to the higher TKN/COD ratio of the settled wastewater. Knowing the average daily total oxygen demand, (FOtd) the peak total oxygen demand can be roughly estimated by means of a simple design rule (Musvoto et al., 2002). From a large number of simulations with AS model no. 1 (ASM1), it was found that, provided the factor of safety on nitrification (Sf) is greater than 1.25–1.35, the relative amplitude (i.e., (peak average)/average) of the total oxygen demand variation is a fraction 0.33 of the relative amplitude of the TOD of the influent COD and TKN load (i.e., Qi(Sti þ 4.57Nti)). For example, with the raw wastewater case, if the peak influent TOD flux is obtained at a time of day when the influent flow rate, COD and TKN concentrations are 25 M l d1, 1250 mgCOD l1 and 90 mgN l1, respectively – that is, 25(1250 þ 4.57 90) ¼ 41 532 kgTOD d1, and the average influent TOD flux is 15(750 þ 4.57 60) ¼ 15 363 kgTOD d1, the amplitude of the total influent TOD flux is (41 532 15 363)/15 363 ¼1.70; hence, the amplitude of the total oxygen demand is approximately 0.33 1.70 ¼ 0.56; from Table 16 the average daily total oxygen demand (FOtd) is 7924 kgO d1 and hence the peak oxygen demand is (1 þ0.56) 7924 ¼12 378 kgO d1. As with all simplified design rules, the above rule should be used with discretion and caution, and where possible, the peak total oxygen demand is best estimated by means of the AS simulations models.
4.14.28 Biological Excess Phosphorus Removal ð172Þ
From the denitrification performance of the MLE system in Table 16, the oxygen recovered by denitrification for the example raw and settled wastewaters at 14 1C are 1440 and 1458 kgO d1. For the raw wastewater, Table 16 shows that (1) the nitrification oxygen demand (FOn) is about 40% that required for COD removal (FOc), (2) about 55% of FOn can be recovered by incorporating denitrification, (3) the additional oxygen
4.14.28.1 Introduction Phosphorus is the key element in aquatic environments that limits the growth of aquatic plants and algae controls eutrophication. Unlike nitrogen that can be fixed from the atmosphere which contains about 80% nitrogen gas, phosphorus can only come from upstream of aquatic systems (neglecting atmospheric deposition). Diffuse sources of phosphorus, for example, from agricultural fields, are best controlled by proper fertilization plans, while point sources of
Biological Nutrient Removal
phosphorus, for example, from WWTPs, can be removed by chemical or biological processes. Considering the benefit to aquatic environments, strict regulations are being applied for phosphorus removal from wastewaters. Considering the potential benefits of removing phosphorus biologically rather than chemically, along with organic matter and nitrogen from wastewater, BEPR has stimulated much interest in the study of the biochemical mechanisms, the microbiology of the systems, the process engineering and optimization of plants, and in mathematical modeling. Reviews of the development of BEPR have been regularly published over the years (Marais et al., 1983; Arvin, 1985; Wentzel et al., 1991; Jenkins and Tandoi, 1991; van Loosdrecht et al., 1997; Mino et al., 1998; Blackall et al., 2002; Seviour et al., 2003; Oehmen et al., 2007). This section briefly reviews the mechanisms of BEPR, outlines the practical systems to achieve it, summarizes some of the experimental research that led to the development of BEPR models (both steady state and dynamic kinetic), discusses the impact of anoxic zones for denitrification on BEPR, and sets out guidelines for design of NDBEPR systems. In order not to unduly complicate this, the concepts are presented for strictly aerobic phosphorus accumulating organisms (aerobic PAOs) which can use only oxygen as the electron acceptor for energy production. Considering that some denitrifying PAOs (DPAOs) exist and may have a significant impact on the performance of the process, their influence is discussed where appropriate, but is not included in the models described.
4.14.28.2 Principles of BEPR BEPR is the biological uptake and removal of P by AS systems in excess of the amount that is removed by normal completely aerobic AS systems. This is in excess of the normal P requirements for growth of AS. In the completely aerobic AS system, the amount of P typically incorporated in the sludge mass is about 0.02 mgP/mgVSS (0.015 mgP/mgTSS). By the daily wastage of surplus sludge phosphorus is thus effectively removed. This can give a P removal of about 15–25% of the P in many municipal wastewaters. In an BEPR AS system, the amount of P incorporated in the sludge mass is increased from the normal value of 0.02 mgP/mgVSS to values around 0.06–0.15 mgP/ mgVSS (0.05–0.10 mgP/mgTSS). This is achieved by system design or operational modifications that stimulate, in addition to the OHOs present in AS, the growth of organisms that can take up large quantities of P and store them internally in long chains called polyphosphates (polyPs); generically, these organisms are called phosphate accumulating organisms (PAOs). PAOs can incorporate up to 0.38 mgP/mgVSS (0.17 mgP/ mgTSS). In the biological P removal system both the OHOs, which do not remove P in excess, and the PAOs coexist. The larger the proportion of PAOs that can be stimulated to grow in the system, the greater the P content of the AS and, accordingly, the larger the amount of P that can be removed from the influent. Thus, the challenge in design is to increase the amount of the PAOs relative to the OHOs present in the AS as this will increase the capacity for P-accumulation and thereby high phosphorus removal efficiency. The relative proportion of the two organism groups depends, to a large degree, on the fraction
15 Example settled WW % P of VSS (mgP/mgVSS as %)
496
Settled WW 10 Example Raw WW Raw WW 5
P removal =
%P × VSS mass Sludge age × Q i
0 0
10
20
30
40
% Bio COD obtained by PAOs Figure 43 Percentage P (mgP/mgVSS 100) in VSS mass vs. the proportion of biodegradable COD mass (as %) obtained by PAOs.
of the influent wastewater biodegradable COD that each organism group obtains. The greater the fraction of PAOs in the mixed liquor, the greater the %P content of the AS and the greater the BEPR (Figure 43). Design and operational procedures are oriented toward maximizing the growth of PAOs. In an appropriately designed BEPR system, the PAOs can make up about 40% of the active organisms present (or 15% of VSS; 11% of TSS), and this system can usually remove about 10–12 mgP per 500 mg influent COD l1. From the first publications reporting enhanced P removal in some AS systems, there has been some controversy as to whether the mechanism is a precipitation of inorganic compounds, albeit perhaps biologically mediated, or biological through formation and accumulation of P compounds in the organisms. The objective here is not to discuss the evidence that supports the biological nature of enhanced P removal, but to briefly describe the theory of biological P removal and to demonstrate how this theory can be used as an aid for the design of biological P removal AS systems. This does not imply that precipitation of P due to chemical changes resulting from biological action (e.g. alkalinity and pH) does not take place. Although inorganic precipitation of P can certainly take place, it would appear that in the treatment of municipal wastewaters by an appropriately designed AS system, within the normal ranges of pH, alkalinity and calcium concentrations in the influent, enhanced P removal is principally mediated by a biological mechanism (Maurer et al., 1999; de Haas et al., 2000). These mechanisms are described below.
4.14.28.3 Mechanism of BEPR 4.14.28.3.1 Background Historically, several research groups have made a number of important contributions toward elucidating the mechanisms
Biological Nutrient Removal
4.14.28.3.2 Biological P removal microorganisms The basic requirement for BEPR is the presence in the AS system of microorganisms which can accumulate P in excess of normal metabolic requirements, in the form of polyP stored in granules called volutins. In the BEPR models, all organisms in the AS system accumulating polyP in this fashion and exhibiting the classical observed BEPR behavior – anaerobic P release, aerobic P uptake, and associated processes – are lumped together and represented by the generic PAO group. PolyPs can be accumulated by a wide range of bacteria. In general, they are accumulated as a phosphate reserve in relatively low amounts. Only very few types of bacteria seem to be able to harvest the energy that is stored in polyPs to take up VFAs and store them as PHAs under anaerobic conditions (in the absence of an external electron acceptor such as oxygen or nitrate). In the original research on BEPR microbiology conducted with cultivation studies, it was incorrectly considered that PAOs were of the genus Acinetobacter (Fuhs and Chen, 1975; Buchan, 1983; Wentzel et al., 1986), Microlunatus phosphovorus (Nakamura et al., 1995), Lampropedia (Stante et al., 1997), and Tetrasphaera (Maszenan et al., 2000). More recently, cultureindependent methods have shown that Accumulibacter phosphatis, a member of the genus Rhodocyclus (a beta proteobacterium), is a PAO which can be grown in enriched cultures (at up to 90% purity, as shown by fluorescence in situ hybridization (FISH) molecular probes) but not yet in axenic cultures (Wagner et al., 1994; Hesselmann et al., 1999; Crocetti et al., 2000; Martin et al., 2006; Meyer et al., 2006; Oehmen et al., 2007). From a modeling and design perspective, however, the identification of the exact organisms responsible for BEPR is of minor importance, although this may provide information that can be used to refine the models and design procedures; these are not based on the behavior of specific organisms, but rather on the observed behavior of groups of organisms identified by their function, in this case the PAOs.
4.14.28.3.3 Prerequisites To achieve BEPR in AS systems, the growth of organisms that accumulate polyP (PAOs) has to be stimulated. To accomplish this, two conditions are essential: (1) an anaerobic and aerobic (or anoxic) sequence of reactors/conditions and (2) the addition or formation of VFAs in the anaerobic reactor/period.
Glycogen
Concentration
of BEPR, including Shapiro et al. (1967), Fuhs and Chen (1975), Nicholls and Osborn (1979), Rensink et al. (1981), Marais et al. (1983), Lotter (1985), Comeau et al. (1986), Wentzel et al. (1986, 1991), Mino et al. (1987, 1994, 1998), Kuba et al (1993), Smolders et al. (1994a, 1994b, 1995), van Loosdrecht et al. (1997), Maurer et al. (1997), Seviour et al. (2003), Martin et al. (2006), and Oehmen et al. (2007). In this section, an explanation of the basic concepts underlying the more sophisticated mechanistic models for the biological P removal phenomenon is presented. For detailed description of the mechanisms, the reader is referred to the references mentioned earlier in this paragraph.
497
Poly P
PHA VFA Anaerobic
PO4 Aerobic
Figure 44 Schematic diagram showing the changes as a function of time in concentrations of volatile fatty acids (VFAs), P (PO4), polyphosphates (polyPs), polyhydroxyalkanoate (PHA), and glycogen through the anaerobic–aerobic sequence of reactors in a BEPR system.
4.14.28.3.4 Observations With the prerequisites for BEPR present, the following observations have been made at full, pilot, and laboratory scale (Figure 44). Under anaerobic conditions, bulk solution VFAs and intracellular polyP and glycogen decrease and soluble phosphate, Mg2þ, Kþ, and intracellular poly-b-hydroxyalcanoates (PHAs) increase (Randall et al., 1970; Rensink et al., 1981; Hart and Melmed, 1982; Fukase et al., 1982; Watanabe et al., 1984; Arvin, 1985; Hascoe¨t et al., 1985a; Wentzel et al., 1985; Comeau et al., 1986, 1987; Murphy and Lo¨tter, 1986; Gerber et al., 1987; Wentzel et al., 1988; Satoh et al., 1992; Smolders et al., 1994a; Maurer et al., 1997). Under aerobic conditions; intracellular polyP and glycogen increase; soluble phosphate, Mg2þ, Kþ, and intracellular PHA decrease (Fukase et al., 1982; Arvin, 1985; Hascoe¨t et al., 1985a; Comeau et al., 1986; Murphy and Lo¨tter, 1986; Gerber et al., 1987; Wentzel et al., 1988; Satoh et al., 1992; Smolders et al., 1994b; Maurer et al., 1997).
4.14.28.3.5 Biological P removal mechanism In describing the mechanisms of BEPR, a clear distinction is made between the PAOs and OHOs. In the anaerobic/aerobic sequence of reactors, it is considered that VFAs are present in the influent waste stream entering the anaerobic reactor or produced in the anaerobic reactor by fermenting organisms (accepted to be the OHOs in models). In the anaerobic reactor (zero nitrate and oxygen in or entering reactor), the OHOs cannot utilize the VFAs due to the absence of an external electron acceptor, oxygen or nitrate. The PAOs, however, can take up the VFAs from the bulk liquid and store them internally by linking the VFAs together to form complex long-chain carbon molecules of poly-b-hydroxyalkanoates (PHAs). The two common PHAs are poly-b-hydroxybutyrate (PHB: four-carbon compound synthesized from two acetate molecules) and polyhydroxyvalerate (PHV: five-carbon compound from one acetate and one propionate molecules) (Figure 45(a)). Forming PHAs from the VFAs requires energy for three functions: active transport of VFAs across the cell membrane, energization of VFAs into coenzyme A compounds (e.g.,
498
Biological Nutrient Removal Liquid Cell
PHA e−
Glycogen
VFA-CoA Energy
VFA
(PO4)n
VFA
(PO4)n−1 Pi
PAO
Pi
(a)
Liquid CO2
CO2
Catabolism PHA
Cell
e− Glycogen ETC
Anabolism
Energy H2O
New cells
O2 (PO4)n Pi
(b)
(PO4)n −1 PAO
Pi
Figure 45 (a) Simplified biochemical model for PAOs under anaerobic conditions. Anaerobic uptake of volatile fatty acids (VFAs), originating from the influent or from fermentation in the anaerobic reactor, and storage of polyhydroxyalkanoates (PHAs) by the PAOs with associated P release. (b) Simplified biochemical model for PAOs under aerobic conditions. Aerobic utilization of PHAs and growth of PAOs, with P uptake by existing and new PAOs.
acetyl-CoA) and reducing power (NADH) for PHA formation. PolyP degradation is associated with the formation of ADP from AMP, with the phosphokinase enzyme 2 ADP are converted to adenosine triphosphate (ATP) and adenosine monophosphate (AMP) (van Groenestijn et al., 1987). When ATP is used, orthophosphates are released and accumulate in the cell interior together with the counterions of polyP (potassium and magnesium). The efflux of these compounds might be related to building a proton motive force, which either can help in the uptake of acetate or in the generation of a small amount of extra ATP. It is observed (Smolders et al., 1994a) that the energy requirements for acetate uptake increase with increasing pH. This can be associated with the fact that the energy needed for acetate transport increases with pH. ATP is used, notably, for the energization of acetate and propionate into acetyl-CoA and propionyl-CoA. Glycogen degradation also results in ATP formation, NADH production, and intermediates that are
transformed into acetyl-CoA (or propionyl-CoA). Finally, acetyl-CoA and propionyl-CoA are stored as PHA (Comeau et al., 1986; Wentzel et al., 1986; Mino et al., 1998; Smolders et al., 1994b; Martin et al., 2006; Oehmen et al., 2007; Saunders et al., 2007). Thus, the PAOs in the anaerobic reactor have taken up for their exclusive use the VFAs under anaerobic conditions where the OHOs are unable to use these organics. To accomplish this, some of the stored polyP has been consumed and P released to the bulk solution. To stabilize the negative charges on the polyP, the cations Mg2þ, Kþ, and sometimes Ca2þare complexed, which add to the inorganic settleable solids (TSS) in the system (Ekama and Wentzel, 2004). When polyPs are consumed and P is released, mainly Mg2þ and Kþ cations are released in the approximate molar ratio P:Mg2þ:Kþ of 1:0.33:0.33 (Comeau et al., 1987; Brdjanovic et al., 1996; Pattarkine and Randall, 1999). In the subsequent aerobic reactor (presence of DO). In the presence of dissolved oxygen (or of nitrate under anoxic conditions) as an external electron acceptor, the PAOs utilize the stored PHA as a carbon and energy source for energy generation and growth of new cells as well as for regenerating the glycogen consumed in the anaerobic period. The stored PHA is also used as an energy source to take up P from the bulk solution to regenerate the polyP used in the anaerobic reactor, and to synthesize polyP in the new cells that are generated – P uptake (Figure 45(b)). The uptake of P to synthesize polyP in the new cells generated means that more P is taken up than is released in the anaerobic reactor, giving a net removal of P from the liquid phase in the AS system. Accompanying the P uptake, the cations Mg2þ and Kþ also are taken as countercharge for the negatively charged polyP polymer, in the approximate molar ratio P:Mg2þ:Kþ of 1:0.33:0.33. The PAOs, with stored polyP, are removed from the aerobic reactor of the system (where the internally stored polyP concentration in the PAOs is the highest in the system) via the waste sludge stream (wastage from the underflow recycle stream is possible, but not desirable for hydraulic control of sludge age; see Section 4.14.14). At steady state the mass of PAOs wasted per day (with stored polyP) equals the mass of new PAOs generated per day (with stored polyP). Thus, for a fixed sludge age, loading, and system operation, the mass of PAOs in the biological reactors remains constant, so that in the AS system at steady state there is neither a buildup nor a loss of PAOs, and the P/VSS ratio stays approximately constant. The mass of new PAOs formed depends on the mass of stored substrate (PHA) available to the PAOs. Accordingly, the enhanced P removal attained will depend on the mass of PHA stored in the anaerobic reactor.
4.14.28.3.6 Fermentable COD and slowly biodegradable COD As indicated above, under anaerobic conditions, PAOs can take up and store VFAs. However, some wastewaters contained very little VFAs, yet exhibited significant BEPR. This was ascribed to the influent RBOs, (Sbsi) which comprises both VFAs (Sbsai) and fermentable RBO (FBSO, Sbsfi) (Siebritz et al., 1983; Wentzel et al., 1985, 1990; Nicholls et al., 1985; Pitman et al., 1988; Randall et al., 1994). This influent FBSO is
Biological Nutrient Removal Liquid Cell F-RBCOD
F-RBCOD
Energy
VFA OHO
VFA
VFA
PAO
Figure 46 Simplified biochemical model for fermentation of RBSO to VFA by OHOs under anaerobic conditions – VFAs released by OHOs are taken up by PAOs.
fermented to VFAs by the OHOs in the anaerobic reactor, the VFAs becoming available for uptake and storage by the PAOs because the OHOs cannot utilize them due to the absence of an electron acceptor (NO3 or O) (Figure 46). Slowly biodegradable organics (SBO, XS), even though these can be hydrolyzed into RBO under anaerobic conditions, has been shown not to be linked to anaerobic phosphate release. This aspect is of crucial importance as it will influence both the design and operation of BNR systems, such as sizing and determining the number of anaerobic reactors, inclusion of primary sedimentation and maximum BEPR achievable. For the purpose of the BEPR models, the experimental evidence linking BEPR to the RBO is accepted, but a conversion of SBO to RBO is considered to be small enough to be negligible. Accordingly, where VFA production does occur, this will essentially be from the RBO. One exception to this consideration is when primary sludge is fermented in a separate fermentation reactor upstream of the anaerobic reactor – in these dedicated fermenters, some hydrolysis of SBO to RBO and VFAs takes place to augment the influent VFA and RBO concentrations (Lilley et al., 1992).
4.14.28.3.7 Functions of the anaerobic zone From the description of the mechanisms above, with normal domestic wastewater as influent, the anaerobic zone/reactor serves two functions: (1) it stimulates conversion of fermentable organics to VFAs by OHOs, that is, facultative acidogenic fermentation and (2) because it is not possible for the OHOs to metabolize the VFAs (no external electron acceptor), the PAOs take up the released VFAs and store them as PHA. Thereafter, the PAOs do not have to compete for substrate when an external electron acceptor becomes available in the aerobic (or anoxic) zone. Of the above two processes, the former is the slower and determines the size of the anaerobic reactor (Wentzel et al., 1985, 1990). Should primary sludge fermentation be implemented at the treatment plant, the first process would not be needed as much and the size of the anaerobic reactor could be decreased.
499
4.14.28.3.8 Influence of recycling oxygen and nitrate to the anaerobic reactor Numerous investigators (e.g., Barnard, 1976; Venter et al., 1978; Siebritz et al., 1980; Hascoe¨t and Florentz, 1985b) noted that the recycling of oxygen and/or nitrate to the anaerobic reactor causes a corresponding decrease in BEPR. In terms of the mechanisms described above, if oxygen and/or nitrate is recycled to the anaerobic reactor, the OHOs are able to utilize the fermentable COD for energy and growth using the oxygen or nitrate as external electron acceptor. For every 1 mgO recycled to the anaerobic reactor 3 mgCOD of fermentable RBO are consumed and for every 1 mgN of nitrate recycled 8.6 mgCOD of fermentable RBO are consumed by the OHOs. The ratio of 3 mgCOD/ mgO consumed comes from the catabolic oxygen requirement in organics utilization (i.e., 1/(1 fcvYH)E3) (Equation (46)). Similarly, considering that 1 mgNO3-N is equivalent to 2.86 mgO (Section 4.14.24.2), a ratio of 2.86/ (1 fcvYH)E8.6 mgCOD consumed by mgNO3-N reduced is obtained. The fermentable RBOs metabolized by the OHOs are not released to the bulk liquid as VFAs. Therefore, the amount of VFAs generated and released to the bulk liquid is reduced by the amount of RBO consumed by the OHOs. Consequently, the mass of VFAs available to the PAOs for storage is reduced, and correspondingly so is the P release, P uptake, and the net P removal. Should the influent RBO already consist of VFAs and oxygen and/or nitrate be recycled, the PAOs and OHOs will compete for the VFAs, the PAOs to take up the VFAs, and the OHOs to metabolize it. Accordingly, even in this situation recycling of oxygen and/or nitrate will reduce the BEPR. Thus, preventing the recycling of oxygen and nitrate to the anaerobic reactor is one of the primary considerations in the design and operation strategy for BEPR systems.
4.14.28.3.9 Denitrification by PAOs The extent of denitrification with associated anoxic P uptake by the PAOs appears to be highly variable (Ekama and Wentzel, 1999b), from near-zero anoxic P uptake (e.g., Wentzel et al., 1989a, Clayton et al., 1989, 1991) to anoxic P uptake dominant over aerobic P uptake (e.g. Sorm et al., 1996; Hu et al., 2000). Experimental evidence tends to suggest that magnitude of anoxic P uptake is influenced by the anoxic mass fraction and the mass of nitrate loaded on the anoxic reactor relative to its denitrification potential (Hu et al., 2002). For the purpose of design it will be considered that anoxic P uptake is not significant. Anoxic P uptake decreases the magnitude of P removal in the system (Ekama and Wentzel, 1999a, 1999b; Hu et al., 2002), and from a design point of view in which maximizing P removal is a priority, anoxic P uptake should be avoided in the system. Hence, in this chapter, anoxic P uptake will not be considered. It must be emphasized, however, that due to the anaerobic conversion of RBO to VFA which are taken up by PAOs, the kinetics of denitrification in the subsequent anoxic reactor change compared with that in the primary anoxic reactor of an MLE system.
500
Biological Nutrient Removal
4.14.29 Principles of Maximizing BEPR The principles of maximizing BEPR can be grouped into seven categories. A number of configurations or systems that are based on these principles are identified by specific names (Figure 47). 1. Oxygen entrainment in the anaerobic reactor should be minimized. For this purpose, mixing vortexes, upstream cascades, and screw pumps or air lift pumps should be avoided.
2. Nitrate (and nitrite) entering in the anaerobic reactor should be minimized. A number of named configurations were developed precisely for this purpose (Section 4.14.34). Based on observations a number of laboratory-, pilot- and full-scale systems (Barnard, 1974, 1975a, 1975b; Nicholls, 1975b), to achieve BEPR in the simplest configuration, Barnard (1976) proposed the Phoredox system (Figure 47(a) also known as the A/O process). This system comprises only an anaerobic and aerobic reactor and is intended not to nitrify to avoid nitrate entering the
Primary anoxic reactor
Anaerobic reactor Anaerobic reactor
Aerobic reactor
Mixed liquor recycle a Waste flow
Waste flow Settler Effluent
Influent
Settler Effl.
Infl.
Sludge recycle s
Sludge recycle s
(a)
(b)
Mixed liquor recycle a
Primary anoxic reactor
Anaerobic reactor
Primary Secondary Aerobic Reaeration anoxic anoxic reactor reactor reactor reactor
Anaerobic reactor
Aerobic reactor
r Recycle a Recycle Waste flow
Waste flow
Settler Effl.
Infl.
Settler Effl.
Infl.
Sludge recycle s
Sludge recycle s
(c)
(d) Primary anoxic reactor
Anaerobic reactor Anaerobic reactor
1st primary 2nd primary anoxic anoxic reactor reactor
r Recycle
Waste flow
a Recycle
Effl.
Sludge recycle s (e)
(f)
r Recycle
Settler Effl.
Infl.
Settler
Infl.
1st prim. Anaerobic anoxic reactor reactor
Aerobic reactor
Mixed liquor recycle a
Aerobic reactor Waste flow
2nd prim. anoxic reactor
Anoxic or aerobic Aerobic reactor reactor
a Recycle
PST
Aerobic reactor
SST Effluent
1° Solids
Waste
Phostrip II
Settler
Infl.
Sludge recycle s
Underflow anoxic reactor
Influent
Waste flow Effl.
Supernatant
Sludge recycle s (g)
Aerobic reactor
Mix Stripper tank Lime
Biospherics, USA
(h)
Figure 47 System configurations for BEPR: (a) phoredox or A/O; (b) three-stage Bardenpho or (A2/O); (c) five-stage Bardenpho; (d) University of Cape Town (UCT) or Virginia Initiative Process (VIP); (e) modified UCT; (f) Johannesburg (JHB); (g) biological chemical flexible system (BCFS); and (h) phostrip.
Biological Nutrient Removal
anaerobic reactor. Because nitrification can take place even at short sludge ages, particularly in warm climates, one or more anoxic reactors for denitrification are included in the two-reactor anaerobic–aerobic system to protect the anaerobic reactor from nitrate entering it. The position of the anoxic reactor(s) has led to a number of different configurations: (1) one between the anaerobic and aerobic reactors with the return sludge discharged to the anaerobic reactor (three-stage Bardenpho or A2/O systems, Figure 47(b)), (2) anoxic reactors before and after the aerobic reactor with the return sludge discharged to the anaerobic reactor (five-stage Bardenpho, Figure 47(c)), (3) one or two anoxic reactors between the anaerobic and aerobic reactors with the sludge return discharged to the first or only anoxic reactor (UCT, Siebritz et al., 1980 or VIP, Daigger et al., 1987; Figure 47(d) and modified UCT systems; Figure 47(e)), and (4) an anoxic reactor between the anaerobic and aerobic reactors and another in the sludge return flow (JHB system; Figure 47(f)). 3. VFA uptake by PAOs in the anaerobic reactor should be maximized. Primary sludge fermentation is an efficient way to increase the VFA content of the influent even though it also contributes to an increased loading in organic matter and ammonia to the AS system. Sodium acetate or fermentable industrial wastes can be added directly to the anaerobic reactor or industries that produce fermentable organics (e.g., breweries or food processing factories) should not be penalized for discharging their high RBO containing wastewater to the sewer. The sludge mass fraction of the anaerobic reactor can be increased to favor in situ fermentation of the influent or added fermentable organic matter. 4. Effluent particulate phosphorus should be minimized by removing TSSs efficiently. The particulate phosphorus content can reach as high as 18% gP/gTSS for enriched cultures. With a more typically 5–10% P content for municipal wastewater (Figure 43), every 10 mgTSS l1 in the effluent will contribute 0.5 to 1 mgP l1. Thus, efficient secondary clarification, avoiding floating sludge from denitrification in the settling tank, sand filtration, or even ultrafiltration (in a membrane bioreactor) are means of reducing the effluent TSS concentration. 5. Effluent soluble phosphorus should be minimized. Besides optimizing the BEPR process, chemical coagulants such as iron (e.g., FeCl3), aluminum (e.g., alum), or calcium (e.g., lime) salts can be added in the mainstream for pre-, co-, or post-precipitation (in the primary settling tank, in the AS process, downstream of the secondary settling tank, respectively, de Haas et al., 2001). Extracting the supernatant from the anaerobic tank or taking some sludge from the return AS and coagulating them can also lead to lower effluent soluble phosphorus (Sehayek and Marais, 1981; van Loosdrecht et al., 1998; e.g., BCFS process; Figure 47(g)). Sidestream lime precipitation of phosphate released anaerobically from the return sludge can also be done. More efficient phosphate release can be achieved in this sidestream tank by diverting some influent containing readily biodegradable COD (e.g., PhoStrip process, Figure 47(h)). These systems support the biological
501
process by phosphate stripping and potential P recovery in the main line, stabilizing the sludge settling properties and optimizing the control of nitrogen removal. In the BCFS system, a third recycle is added from the aerated reactor to the first anoxic reactor in order to maximize denitrification and to be able to aerate the second anoxic reactor during peak flows or cold temperatures. In this way both ammonium and nitrate can be better controlled to low effluent values (ammonium typical below 0.5 gN l1 and nitrate around 5–8 mg N l1). The recycle flows are controlled by a simple redox electrode-based controller (van Loosdrecht et al., 1998). Compartmentalizing the reactors and low effluent ammonia concentration contributes to a stable low SVI – around 120 ml g1 (Kruit et al., 2002; Tsai et al., 2003). Biological phosphorus removal can be supplemented by addition of precipitants to the anaerobic tank. Since phosphate concentrations are high in this tank, the precipitants are used effectively. Dosing chemicals, however, should be done carefully. Too much precipitation will make the phosphate unavailable for PAOs and deteriorate the BEPR efficiency (de Haas et al., 2001). A complicating factor is that the WWTP will respond rapidly to changes in addition of chemicals whereas the biological phosphorus removal process might have a response time of several days if not weeks. In the BCFS process, a small baffle is placed at the end of a plugflow anaerobic tank. The sludge will locally settle back into the anaerobic tank and a clear supernatant can be withdrawn for phosphate precipitation. The phosphorus can then be recovered (Barat and van Loosdrecht, 2006) or the chemical sludge produced can be prevented from accumulating in the AS which would limit the overall capacity of the plant by reducing the sludge age. Should anaerobic or aerobic digestion be performed with the wasted secondary sludge, essentially all of the polyPs will be hydrolyzed to ortho-P and the phosphate released in solution (Jardin and Po¨pel, 1994; Harding et al., 2009; Mebrahtu et al., 2010). Phosphorus recovery in the form of struvite (MgNH4PO4) or hydroxyapatite (Ca10(PO4)6OH2), which can be used as fertilizers, are also means of reducing the loading of soluble phosphate back to the AS process and, eventually, to the effluent. 6. Phosphorus uptake for cell synthesis should be maximized. Although more limited than the other maximization principles in its potential efficiency, maintaining the sludge age as short as possible will result in an increase in phosphorus removal by sludge production (cell synthesis). Although the endogenous respiration rate of the PAOs is low (0.04 d1), another small benefit of reducing the sludge age is that the PAOs degrade to a lower extent their polyP reserves for cell maintenance. 7. Because anoxic P uptake BEPR reduces the P content of the PAOs (Ekama and Wentzel, 1999a, b; Hu et al., 2002), growth of denitrifying PAOs should be avoided to maximize aerobic P uptake BEPR to maximize PAO P content – up to 0.38 mgP/mgPAOVSS (Wentzel et al., 1989b, 1990). For a review of how these developments took place, the reader is referred to Henze et al. (2008). In order to efficiently construct all the tanks in these complex BNR systems, it is possible
502
Biological Nutrient Removal
Figure 48 Zandvliet nitrification–denitrification (ND)BEPR WWTP Cape Town, South Africa. By arranging interconnecting recycle flows between the anaerobic (centre), anoxic (inner ring), and aerobic (outer ring) reactors, the system has the flexibility to be operated as a UCT, threestage Bardenpho or JHB system. Photo: GA Ekama.
to shift the construction from rectangular tanks to one round tank with a sloping outside wall divided in rings for the different aerobic/anoxic/anaerobic zones. In this way, the amount of concrete needed is minimized as all the walls require much less strength (Figure 48).
4.14.30 Model Development for BEPR 4.14.30.1 Early Developments When the first mainstream NDBEPR system was proposed (the five-stage Bardenpho, Figure 47(a); Barnard, 1976), initial conceptualization of the phenomenon extended little beyond recognition of (1) the necessity of an anaerobic/aerobic sequence of reactors and (2) the adverse influence of nitrate recycled to the anaerobic zone. With the inclusion of the secondary anoxic reactor, it was believed that nearly complete denitrification of nitrate would be achieved, thereby discharging very low nitrate concentrations to the anaerobic reactor. Design procedures were based on empirically based estimates for sizing denitrification and anaerobic reactors in terms of nominal hydraulic retention time, and sizing of the anaerobic reactor appeared to be linked to depression of the redox potential below some critical value. No rational method for predicting N and P removal was available and for design, removals were estimated largely from experience gained in operating experimental systems similar to the proposed systems (McLaren and Wood, 1976; Simpkins and McLaren, 1978; Osborn and Nicholls, 1978).
4.14.30.2 RBO and Anaerobic Mass Fraction In seeking an explanation for the different P release and enhanced P removal behavioral patterns in lab-scale modified UCT (Figure 47(e)) and MLE (Figure 34(b)) systems, Siebritz et al. (1980, 1983) applied the concept of RBO developed in denitrification and aerobic studies (Dold et al., 1980; van Haandel et al., 1981) to BEPR systems. They noted that the only evident difference between the modified UCT and MLE
systems lay in the concentration of RBO surrounding the organisms in the anaerobic reactor. (They also observed that the UCT and MLE systems with the same anoxic mass fractions yielded approximately the same effluent nitrate concentrations and the ND kinetic models (such as ASM1, Henze et al., 1987 or UCTOLD, Dold et al., 1991) predicted the NDBEPR system response reasonably well even at full scale (Nicholls, 1982). This implied that the anaerobic reactor did not appear to have a detrimental effect on the denitrification (the questions this raises regarding denitrification in NDBEPR systems are discussed in Section 4.14.34.) In the modified UCT system the RBO concentration in the anaerobic reactor is the maximum possible as no nitrate is recycled to the anaerobic reactor; in contrast, in the MLE system sufficient nitrate is recycled to the anoxic reactor to utilize all the RBO. Therefore, the different behavioral patterns of the systems would be consistently described if it is assumed that the concentration of RBO from the influent in the anaerobic reactor surrounding the organisms is a key parameter determining whether or not P release and BEPR take place. (Later it became clear that the parameter influent RBO concentration in the anaerobic reactor surrounding the organisms represented the influent and produced VFAs taken up by the PAOs in the anaerobic reactor.) The validity of this RBO hypothesis was established by Siebritz et al. (1983) at laboratory scale and Nicholls et al. (1985) at full scale, who found that the magnitude of the P release was proportional to the influent RBO concentration. This opened the way for enquiry into other factors affecting the P release and the BEPR and quantifying BEPR. It was concluded that the BEPR depended on two main parameters, viz. (1) influent RBO concentration and (2) the anaerobic sludge mass fraction. Testing the concepts of the parametric model did, in general, demonstrate the utility of the model. At laboratory scale, the concepts were tested in the modified UCT system at different sludge ages, temperatures, anaerobic mass fractions, and influent COD concentrations in which the RBSO fraction of the influent (unsettled municipal sewage) was augmented by the addition of glucose or acetate. Based on the influent RBO concentration and anaerobic mass fraction parameters, the predicted P removal compared quite consistently with the measured P removal. At full scale, evaluation of the Goudkoppies and Northern Works WWTPs with the parametric model provided a consistent explanation when good or poor P removal was obtained (Nicholls et al., 1985; 1986; 1987). Thus, the parametric model allowed some quantitative approach to design of N and P removal plants and provided a basis for evaluating the performance of existing plants (Ekama et al., 1983). This parametric BEPR model, as well the organics removal, ND models presented earlier in this chapter, were published in the NDBEPR system design guide (WRC, 1984). At the time of its publication (1984), the NDBEPR system design approach was criticized and rightly so, primarily because the influent RBO was used twice, once by the PAOs for P removal (uptake in the anaerobic reactor) and again by the OHOs for denitrification in the primary anoxic reactor. This would be possible only if in NDBEPR systems the PAOs utilize all the RBO in the primary anoxic reactor with nitrate as electron acceptor for growth and polyP accumulation in the same fashion as the RBO is completely utilized by the OHOs
Biological Nutrient Removal
in the primary anoxic reactor of the ND system. In this event the major portion of the P uptake and polyP storage by the PAOs should take place in the primary anoxic reactor of the NDBEPR systems. However, P uptake was observed taking place principally in the aerobic zone. This indicated that the denitrification behavior in NDBEPR systems is not the same as that observed ND systems so that the good predictions that had been obtained by the ND models for the NDBEPR systems were fortuitous. Denitrification behavior in NDBEPR systems is discussed in Section 4.14.34 after presenting the BEPR model based on PAO behavior. Essentially up to this time, models of NDBEPR system behavior did not recognize the presence of any specific organism mediating BEPR, only the OHOs for COD removal, denitrification, and RBO fermentation, and the ANOs for nitrification (Table 1). The parametric model in fact considered the active biomass as one group (OHOs) to represent a BEPR sludge with a propensity for P removal; variation in BEPR between different systems was modeled as changes in the propensity for P removal of OHO biomass caused by changes in influent RBSO concentration, anaerobic mass fraction, and/ or nitrate discharge to the anaerobic reactor. However, parallel research in the natural sciences had identified specific organism groups that have the propensity to store large quantities of P in the form of polyP (e.g., Buchan, 1983). This led to a shift in the approach to modeling BEPR in NDBEPR systems, from a representative OHO biomass to a specific organism group mediating BEPR, like the ANOs, the specific organism groups that mediate nitrification. The BEPR organism group became generically termed polyP organisms (Wentzel et al., 1986), bio-P organisms (Comeau et al., 1986), or PAOs (ASM2, Henze et al., 1995).
4.14.30.2.1 NDBEPR system kinetics Wentzel et al. (1988) set out to develop a general model that describes NDBEPR system behavior. They assumed that in an NDBEPR system treating municipal wastewaters, a mixed culture would develop which could be categorized into three groups of organisms: (1) heterotrophic organisms able to accumulate polyP, termed PAO; (2) heterotrophic organisms unable to accumulate polyP, termed OHOs; and (3) autotrophic organisms mediating nitrification, termed ANOs (Table 1). With regard to OHOs and ANOs, they accepted the ND models described in this chapter, viz., the steady-state (WRC, 1984) and general kinetic model (Dold et al., 1980, 1991; van Haandel et al., 1981). These models were extended to incorporate PAO behavior. To achieve this, the kinetic and stoichiometric characteristics of the PAOs in the AS environment needed to be established. From attempts to obtain information on the characteristics of the PAOs using mixed liquor from NDBEPR systems treating municipal wastewaters, Wentzel et al. (1988) noted that the OHO behavior masked the PAO behavior except in its P release, P uptake, and P removal. Accordingly, to isolate the PAO biomass characteristics, they developed enhanced cultures of PAOs in open (nonsterile) AS systems. (Serendipitously, because the UCT laboratory did not have the equipment to develop pure cultures, this was never attempted – in hindsight, this would have been the wrong approach because even today, a pure culture of PAOs has not
503
yet been established.) By enhanced culture is meant a culture in which (1) the growth of PAOs is selected into the system to the extent that they become the principal organism group so that their behavior dominates the system in all the measured parameters (OUR, VSS) and (2) growth of competing organisms is selected out but not positively excluded; neither are predation or other interaction effects.
4.14.30.3 Enhanced PAO Cultures 4.14.30.3.1 Enhanced culture development From the biochemical models, Wentzel et al. (1988) were able to identify conditions to be imposed in an NDBEPR AS system to produce an enhanced PAO culture – anaerobic/aerobic sequence with adequate anaerobic mass fraction; influent fed to the anaerobic reactor with acetate as substrate and with adequate macro- and micronutrients, in particular Mg2þ, Kþ, and to a lesser degree Ca2þ, and pH control in the aerobic reactor. Using the UCT and three-stage modified Bardenpho systems, with system sludge ages ranging from 7.5 to 20 days, they developed enhanced cultures of PAOs with greater than 90% of the organisms cultured aerobically being identified as Acinetobacter spp. using the analytical profile index (API) 20NE procedure. (The API 20NE procedure has subsequently been shown to overestimate Acinetobacter spp. numbers due to the testing technique (Lotter et al. 1986; Venter et al. 1989) and selection in culturing (e.g., Wagner et al. 1994). However, for the development of the design and simulation models exact identification of the PAOs in the enhanced cultures has been of minor consequence as the models are based on quantitative experimental observations.) The response of the enhanced culture systems indicated that significant concentrations of PAOs developed. For example, the UCT system (anaerobic mass fraction 15%, sludge age 10 days, and influent of acetate at 500 mgCOD l1) gave phosphate release of 253 mgP l1, phosphate uptake of 314 mgP l1, and phosphate removal of 61 mg l1, all as mgP l1 influent flow. This BEPR behavior was much higher than observed in a mixed culture NDBEPR systems with municipal wastewater influent of 500 mgCOD l1 giving a phosphate release of 45 mg l1, phosphate uptake of 57 mg l1, and phosphate removal of 12 mgP l1. In fact, the behavior of the enhanced culture systems corresponded closely to that of the mixed culture system in terms of the influent RBO/VFA fed – at 100% and 20% influent RBO/VFA respectively for 500 mgCOD l1 feed, the enhanced culture system removed 5 times more P (61 mgP l1) than the mixed culture system. The enhanced culture mixed liquor in the aerobic zone contained 0.25–0.20 mgP/mgVSS and had a VSS/ TSS ratio of 0.46–0.48 as sludge age increased from 7.5 to 20 days, much higher than for mixed culture systems at a P/VSS ratio of 0.1 and a VSS/TSS fraction of 0.78. The low VSS/TSS ratio for the enhanced culture systems is due to the high concentration of polyP with associated counterions in the PAOs, a phenomenon later included in the model by Ekama and Wentzel (2004).
4.14.30.3.2 Enhanced culture kinetic model From experimental observations on the enhanced culture steady-state systems and on a variety of batch tests (anaerobic, anoxic, and aerobic) on mixed liquor harvested from the
Biological Nutrient Removal
4.14.30.3.3 Simplified enhanced culture steady-state model Wentzel et al. (1990) simplified the enhanced culture kinetic model, to develop a steady-state model for the enhanced culture systems under constant flow and load conditions. From an examination of the kinetics of the processes under steady-state conditions, many of the processes were virtually complete so these kinetic relationships no longer serve an important function under steady-state conditions and could be replaced by stoichiometric relationships. The three examples are given as follows: (1) The anaerobic mass fractions provided in the enhanced culture systems were sufficient to ensure that all the acetate substrate was sequestered in the anaerobic zone, that is, the kinetics of acetate storage need not be incorporated. (2) Virtually, all the substrate taken up by the PAOs in the anaerobic zone was utilized in the subsequent aerobic zone, that is, the kinetics of PHA substrate utilization (and polyP storage) did not need to be incorporated. This implied that for the
10
OUR (mgO I−1 h−1)
9
Simulated Experimental
8 7 6 5 4 3 2 1 0
0
1
2
3
(a)
7
8
9
10
200
150
100
50
0
Simulated Experimental 0
1
2
3
4
5
6
7
8
9
10
Time (days)
(b)
20
100 90
PO4
80
15
70 60
NO3
50
10
40 30 20
Simulated
10
Experimental
0 0
(c)
4 5 6 Time (days)
1
2
3
4
5
6
7
8
9
5
Nitrate concentration (mgN I−1)
Soluble COD concentration (mgCOD I−1)
steady-state enhanced PAO systems, Wentzel et al. (1989a) elucidated the characteristics and kinetic response of the PAO biomass. Two characteristics of the PAOs in these enhanced cultures were of particular interest: (1) very little propensity to denitrify so that no provision for this process needed to be made in modeling PAO behavior – this has important implications in modeling denitrification in mixed culture NDBEPR systems (see Section 4.14.34) and (2) an extremely low endogenous mass loss rate, 0.04 mgPAOVSS/(mgPAOVSS d) which is much lower than that of OHOs in aerobic AS system at 0.24 mgOHOVSS/(mgOHOVSS d) (Marais and Ekama, 1976). A similar observation had been made by Wentzel et al. (1985) in studies on mixed culture NDBEPR systems treating municipal wastewaters; they noted from plots of phosphate uptake versus phosphate release for various sludge ages that, for a given phosphate release, the phosphate uptake was relatively insensitive to sludge age. In modeling PAO endogenous mass loss, Wentzel et al. (1989a) used the classical endogenous respiration approach (Equation (53)), as distinct from the death-regeneration approach used for the OHOs (Section 4.14.5.4.2), except that provision was made for the situations where no external electron acceptor is available. Taking note of the above, Wentzel et al. (1989a) developed a conceptual model for PAO behavior in the enhanced cultures incorporating the characteristics, processes, and compounds identified as important from the experimental investigation. Using the conceptual model as a basis, Wentzel et al. (1989b) formulated mathematically the process rates and their stoichiometric interactions with the compounds, to develop a kinetic model for the enhanced cultures of PAO. The kinetic and stoichiometric constants of the PAOs in the enhanced cultures were quantified by a variety of experimental procedures (Wentzel et al., 1989b). With these constants, application of the kinetic model to the various batch test responses observed with the enhanced cultures gave good correlation between observations and simulations (Figures 49 and 51). The model was then applied to simulate the steadystate behavior of the enhanced culture UCT and three-stage modified Bardenpho systems, for which good correlation was also obtained. (Wentzel et al., 1989b).
Soluble P concentration (mgP I−1)
504
0 10
Time (days)
Figure 49 Experimentally observed and simulated (a) oxygen utilization rate (OUR), (b) total soluble phosphorus (PO4) and nitrate (NO3) concentrations and (c) filtered COD concentrations with time in a batch aerobic digestion test of mixed liquor from an enhanced PAO culture system. Modified from Wentzel MC, Dold PL, Ekama GA, & Marais GR (1989b) Enhanced polyphosphate organism cultures in activatedsludge systems 3. Kinetic model. Water SA 15(2): 89–102.
PAOs, like for the OHOs, the growth process could be accepted as complete so that at steady state, for a given sludge age, a constant relationship exists between the flux of acetate fed to the system and the mass of PAOs formed with stored polyP. (3) P release for anaerobic maintenance energy requirements was small compared with P release for VFA uptake energy requirements, that is, the kinetics of phosphate release for anaerobic
60
60
50
50
PO4
40
40
30
30
20
20
10
10
Acetate
0 0
1
2
3
4
0
200
Simulated Experimental
150
150
PO4 100
100
50
50 Acetate
0
0
1
2 3 Time (h)
4
5
0
Figure 50 Experimentally observed and simulated total soluble phosphorus (PO4) and acetate concentration–time profiles with anaerobic addition of (a) 0.11 mgCODacetate/mgVSS (low) and (b) 0.265 mgCOD/mgVSS (moderate) to a mixed liquor drawn from a threestage Bardenpho enhanced PAO culture system. Modified from Wentzel MC, Dold PL, Ekama GA, and Marais GR (1989b) Enhanced polyphosphate organism cultures in activatedsludge systems 3. Kinetic model. Water SA 15(2): 89–102.
maintenance energy did not need to be incorporated. However, because the endogenous respiration process is never complete, it had to be retained in the steady-state model and, as for the OHOs, was accepted to take place in all the reactors of the system. Applying these simplifications and assumptions in the steady-state PAO model indicated that the P content of the PAOs was constant with sludge age at 0.38 gP/gPAOVSS, of which 0.03 was biomass P content and 0.35 was polyP content, to account for the observed P removal. What did vary was the relative proportion of PAOs (with stored polyP) in the VSS which accounted for the difference in P removal with sludge age. The resulting steady-state PAO model was identical to the OHO model (Section 4.14.31.1.5), including the value for the PAO yield coefficient (YG ¼ 0.45 mgPAOVSS/mgCOD), but the values for the PAO unbiodegradable residue fraction (fEG) and endogenous respiration rate (bG) were different to those of the OHOs (i.e., 0.25 and 0.04 d1, respectively). The PAO steady-state model provided the means for quantifying the PAO VSS mass and its endogenous residue in mixed culture NDBEPR systems receiving municipal wastewaters as influent.
OUR 100
0
5
PO4
0
5
10 15 Time (h)
20
500
0 25
20
450
Simulated Experimental
400 350
15
300 250
10
OUR
200 150
PO4
5
100 50 0
0
5
(b)
(c)
15
10
(a)
Soluble P concentration (mgP I−1)
200
Simulated Experimental
300
Time (h) 200
20
10 15 Time (h)
20
0 25
20
200 Simulated Experimental
15
150
100
5
50
0
10
OUR
PO4
0
5
10 15 Time (h)
20
0 25
Carbon oxygen utilization rate (mgO I−1h−1) Carbon oxygen utilization rate (mgO I−1h−1)
70
Soluble P concentration (mgP I−1)
Soluble P concentration (mgp I−1)
80
70
400
505
Carbon oxygen utilization rate (mgO I−1h−1)
80
(a)
(b)
90
Simulated Experimental
Soluble P concentration (mgP I−1)
90
Acetate concentration (mgHAc I−1)
100
100
Acetate concentration (mgHAc I−1)
Soluble P concentration (mgP I−1)
Biological Nutrient Removal
Figure 51 Experimentally observed and simulated total soluble phosphorus (PO4) concentration and carbonaceous oxygen utilization rate (OUR)–time profiles on aeration following anaerobic acetate addition of (a) 0.207 mgCOD/mgVSS (low),(b) 0.363 mgCOD/mgVSS (moderate), and (c) 0.220 mgCOD/mgVSS (high) to mixed liquor drawn from a three-stage Bardenpho enhanced PAO culture system. Modified from Wentzel MC, Dold PL, Ekama GA, and Marais GR (1989a) Enhanced polyphosphate organism cultures in activated sludge systems 2. Experimental behaviour. Water SA 15(2): 71–88.
4.14.30.3.4 Steady-state mixed culture NDBEPR systems Mixed culture steady-state model. Having developed the steadystate model for enhanced culture systems, Wentzel et al. (1990) extended this model to incorporate mixed cultures of PAOs and OHOs present in NDBEPR systems receiving
506
Biological Nutrient Removal
domestic wastewater as influent, to give a steady-state mixed culture model. This extension proved to be possible because (1) enhanced cultures rather than pure cultures were used to establish the kinetic and stoichiometric characteristics of the PAOs. In the enhanced cultures, PAOs present in mixed culture AS were enriched and no single species was artificially selected (as in pure cultures); (2) competing organisms and predators were not artificially excluded (as in pure cultures) so that the PAOs were subjected to the same selective pressures in enhanced as in mixed cultures; (3) the PAOs were also subjected to the same conditions present in mixed culture AS systems (e.g., anaerobic/aerobic sequencing, long SRT 45 days, etc.); and (4) per influent RBO/VFA, the PAOs exhibited the same behavioral patterns in the enhanced cultures as they did in mixed culture AS systems (i.e., P release/uptake, PHA/ polyP accumulation, etc.) – in fact, the similar, though magnified behavior of the PAO enhanced culture compared to the mixed culture systems was one criterion used to establish that the correct enhanced cultures had been established. In extending the model one aspect that emerged was the difference in the endogenous mass loss rate between PAO enhanced culture sludges and the normal aerobic OHO AS. As noted earlier, the high specific endogenous mass loss rate with OHO systems had been attributed to a high rate of predation and regrowth, formulated as death regeneration in the ND kinetic model by Dold et al. (1980). The low specific endogenous mass loss rate with PAOs in the enhanced cultures systems led Wentzel et al. (1989a) to conclude that the PAOs were not predated to the same degree as OHOs, and to adopt an endogenous respiration approach in modeling PAO endogenous mass loss. (From subsequent simulations with the steady-state mixed culture model, it was found that if the PAOs were subjected to a high predation rate, then significant BEPR in the mixed culture NDBEPR system would not be possible – the rate of death of the PAOs would be so high that no significant mass of these organisms could accumulate in the system, and BEPR would be near zero.) The low predation rate on the PAOs, and the fact that the PAOs and OHOs essentially do not compete for the same substrate, implied that PAO and OHO populations act virtually independently of each other in normal mixed culture NDBEPR systems. This allowed modeling the two population groups as essentially separate, except for the fermentation F-RBO to VFA conversion process in the anaerobic reactor, which could be used to quantify the proportion of the biodegradable organics (BO) obtained by the PAOs. This rate of conversion is much slower than the rate of VFA uptake, so that the rate of conversion controls the rate of VFA uptake. Hence, the flux of VFAs that becomes available in the anaerobic reactor to the PAOs is governed by the kinetics of conversion mediated by the OHOs. The work of Me´ganck et al. (1985) and Brodisch (1985) supported this conversion approach, which is also included in the NDBEPR kinetic models (UCTPHO, Wentzel et al., 1992; ASM2, Henze et al., 1995) – they showed that anaerobic/aerobic systems developed organisms which
convert sugars and similar compounds into VFAs in the anaerobic reactor. If nitrate (or oxygen) is recycled to the anaerobic reactor, RBO is utilized preferentially by the OHOs with nitrate (or oxygen) as external electron acceptor, thereby reducing the flux of VFAs available for uptake by the PAOs. A schematic diagram showing the proportion of the influent RBO obtained by the PAOs is shown in Figure 52. OHOs obtain BO that is not obtained by PAOs. From the above, the RBSO is subdivided into two fractions, VFAs (e.g., acetate) and fermentable RBSO (FBSO, e.g., glucose). Both these fractions are measured as RBO in the conventional bioassay (e.g., Ekama et al., 1986; Wentzel et al., 1995, 1999, 2000) and filtration (e.g., Dold et al. 1986; Mamais et al., 1993; Mbewe et al., 1994) tests (see Section 4.14.4.2.2). The rate of VFA uptake by PAOs is so rapid that all influent VFAs will be taken by the PAOs even in very small anaerobic reactors (Figure 50). The F-RBO is converted to VFAs by the OHOs in the anaerobic reactor and the resultant VFAs is available for uptake by the PAOs (Figure 46). The model for this conversion is given by Wentzel et al. (1985) and will be described below. The above model provided Wentzel et al. (1990) with the means for calculating the flux of BO (influent VFA and
Unbiodegradable soluble (effluent)
Influent wastewater COD Biodegradable COD
Unbiodegradable particulate COD
RBCOD F-RBCOD
SBCOD External acid fermentation
Internal acidification
Inert VSS accumulation
Volatile fatty acids (VFAs) P accumulating organisms (PAOs)
Ordinary heterotrophic organisms (OHOs)
Enhanced culture steady-state equations PAO activemass 0.03 mgP/ mg-VSS
O2,NO3−
Usual activated sludge steady-state equations
Usual OHO activemass 0.03 mgP/mg-VSS
PAD endogenous mass 0.03 mgP/mg-VSS
OHO Endogenous mass 0.03 mgP/mg-VSS
Inert mass 0.03 mgP/ mg-VSS
Mixed VSS in system has variable P content (mass P/mass VSS %) Depending on proportion of biodegradable COD obtained by PAOs
Figure 52 Schematic diagram showing the fate of various influent COD fractions in relation to the various OHO and PAO active, endogenous, and inert masses of the sludge.
Biological Nutrient Removal
Predicted P release (mgP I−1 influent)
100 90 80 70 60 50
R S (days) 3 4 5 6 8 10 15 20 21 25 28
40 30 20 10 0
0 (a)
10 20 30 40 50 60 70 80 90 100 Measured P release (mgP I−1 influent)
30
Predicted P removal (mgP I−1)
converted of F-RBO) taken up by the PAOs in the anaerobic reactor. The remainder of the BO flux is obtained by the OHOs. In effect, the conversion model splits the influent BO (COD) into two fractions, one eventually utilized by the PAOs and the other to be utilized by the OHOs. Because of the independent action of these two organism groups, the masses of PAOs (MXBG) and their endogenous residue (MXEG) in the system could be calculated from the enhanced PAO culture steady-state model and the masses of OHOs (MXBH) and their endogenous residue (MXEH) could be calculated from the steady-state OHO model. The mass UPO in the reactor from the influent (XI) could be calculated from the unbiodegradable particulate COD fraction (fS’up) as before (Section 4.14.9.3.2). The five VSS components, each with their P content – 0.38 mgP/mgPAOVSS for the PAOs and 0.025 mgP/mgVSS for the other four components – give the average P content of the VSS. The P removal achieved by the NDBEPR system is the P in sludge mass wasted per day from the system. Wentzel et al. (1990) evaluated the predictive power of the steady-state mixed culture BEPR model against observations made on 30 laboratory-scale NDBEPR systems over a 6-year period. The system configurations were Phoredox, three-stage modified Bardenpho, UCT, MUCT, and JHB with system sludge ages ranging from 3 to 28 days. For the evaluation, the measured nitrate in the recycle to the anaerobic zone was used to estimate the fermentable COD removal in the anaerobic zone by the OHOs with nitrate as external electron acceptor. The fermentable COD remaining was available for conversion in the anaerobic reactor to VFAs and uptake and storage as PHA by the PAOs. Plots of the predicted versus measured P release, P removal, and VSS concentration (Figures 53(a)– 53(c)) show good correlation.
507
25 20 R S (days)
15
3 4 5 6 8 10 15 20 21 25 28
10 5 0 0
(b)
4.14.31 Mixed Culture Steady-State Model 4.14.31.1 Division of Biodegradable Organics between PAOs and OHOs
Sbsi ¼ Sbsai þ Sbsfi
ð173Þ
The VFA in the influent (Sbsai) is directly available to the PAOs for uptake in the anaerobic reactor.
30
4000
Predicted VSS (mg VSS I−1)
4.14.31.1.1 Subdivision of influent RBO From the mechanism for BEPR, only VFAs can be taken up directly by the PAOs in the anaerobic reactor. Accordingly, the influent RBO (Sbsi) is subdivided into two fractions: (1) VFA (Sbsai) and (2) fermentable RBO (FBSO, Sbsfi). Hence,
5 10 15 20 25 Measured P removal (mgP I−1)
3000
R S (days)
2000
3 4 5 6 8 10 15 20 21 25 28
1000
0
4.14.31.1.2 Conversion of FBSO Wentzel et al. (1985) show that the FBSO component (Sbsfi) is converted to VFA in the anaerobic reactor by the OHOs, thereby making additional VFA available to the PAOs for uptake. The rate of conversion is much slower than the rate of VFA uptake, so that the rate of conversion controls the rate of uptake of generated VFA. Wentzel et al. (1985) proposed a
0 (c)
1000
2000
3000
4000
Measured VSS (mgVSS I−1)
Figure 53 Predicted vs. measured P release (a), P removal (b) and VSS concentration (c) in a variety of BEPR systems with various configurations. From Wentzel et al. (1990).
508
Biological Nutrient Removal
first-order conversion rate, viz.,
dSbsf ¼ KCT XBHn Sbsfn dt ðmgCOD l1 h1 Þ
equations for the conversion of FBSO to VFA can be developed. This yields equations for the concentration of FBSO in exiting the nth anaerobic compartment and the mass of OHOs in the entire NDBEPR reactor, MXBH viz.,
ð174Þ
where KCT is the first-order rate constant at temperature T ¼ 0.06 l/(mgOHOVSS d) at 20 1C, and XBHn and Sbsfn the concentrations of OHOs (mgOHOVSS l1) and FRBO (mgCOD l1) exiting the nth anaerobic compartment of the anaerobic reactor.
4.14.31.1.3 Effect of recycling nitrate or oxygen When nitrate or oxygen enter the anaerobic reactor via recycle and influent flows, the OHOs utilize FBSO with these electron acceptors. Hence, the OHOs do not release the VFA generated but completely metabolize the FBSO until the oxygen or nitrate is depleted. In the conversion model this is accommodated by reducing the concentration of FBSO available for conversion, that is,
S0bsfi
¼ Sbsfi 2:86=ð1 f cv YH ÞðrNnr þ Nni Þ 1=ð1 f cv YH ÞðrOr þ Oi Þ
ð175Þ
where S0bsfi is the FBSO available for conversion to VFA (mgCOD l1 influent), Sbsfi the influent FBSO concentration (mgCOD l1), r the recycle ratio to anaerobic reactor relative to the influent flow, Nnr, Or the nitrate and oxygen concentration in the recycle to anaerobic reactor (mgNO3-N l1 and mgO l1, respectively), Nni, Oi the nitrate and oxygen concentrations in the influent to anaerobic reactor (mgNO3-N l1 and mgO l1, respectively), 2.86/(1 fcvYH) ¼ 8.6 the mass of COD utilized per unit nitrate denitrified (mgCOD/mgNO3N), and 1/(1 fcvYH) ¼ 3.0 the mass of COD utilized per unit oxygen utilized (mgCOD/mgO). Kinetics of conversion of FBSO to VFA. The conversion model proposed by Wentzel et al. (1985) assumes that: 1. Only FBSO can be converted to a form suitable for uptake by the PAOs (i.e., VFA); within the timescale of the mixed liquor in the anaerobic reactor, conversion of SBO to VFA is assumed to be negligible. 2. The conversion is mediated by the OHOs in the absence of oxygen and nitrate only. 3. All VFA generated by conversion is immediately taken up by the PAOs. 4. All FBSO not converted to VFA in the anaerobic reactor is utilized subsequently by OHOs. 5. The rate of conversion of FBSO is first order with respect to the FBSO and OHO concentrations in the anaerobic reactor and given by Equation (174). 6. All VFA present in the influent to the anaerobic reactor is immediately taken up by the PAOs.
4.14.31.1.4 Steady-state FBSO conversion equation Applying Equations (174) and (175) within mass balances over the nth anaerobic compartment in a series of N equal volume anaerobic compartments in the anaerobic reactor receiving in a continuous flow NDBEPR system, the steady-state
S0bsfi =ð1 þ rÞ n Sbsfn ¼ f xa MXBH 1 1 þ KCT ð1 þ RÞ N Qi ðmgCOD l1 Þ
ð176Þ
where fxa is the anaerobic mass fraction of the NDBEPR system, N the total number of compartments of equal volume in the anaerobic reactor, n the nth compartment of the series, n ¼ 1,2,yy,N, Sbsfn the concentration of FBSO exiting the nth compartment, MXBH the mass of OHOs in the system (mgOHOVSS), and Qi the influent flow rate (l d1). Equation (176) provides the means to calculate the flux of FBSO converted to VFA in a series of N anaerobic compartments, that is,
FSbCON ¼ Qi ½S0bsfi ð1 þ rÞSbsfN
ðmgCOD d1 Þ
ð177Þ
However, to calculate SbsfN, MXBH/Qi needs to be known. This is calculated from the flux of BO not obtained by the PAOs. All the VFA generated by conversion and all the VFA in the influent are taken up by the PAOs, so the flux of COD taken up by the PAOs, FSbPAO, is given by
FSbPAO ¼ FSbCON þ Qi Sbsai ¼ Qi ½S0bsfi ð1 þ rÞSbsfN þ Qi Sbsai
ðmgCOD d1 Þ
ð178Þ
and the flux of biodegradable COD taken up by the OHOs is given by
FSbOHO ¼ Qi Sbi FSbPAO
ðmgCOD d1 Þ
ð179Þ
Hence, from Equation (103), the mass of OHOs in the NDBEPR system is given by
MXBH ¼
FSbOHO YH Rs ð1 þ bHT Rs Þ
ðmgOHOVSSÞ
ð180Þ
Substituting Equations (179) and (178) into Equation (180) and dividing by Qi yields the MXBH/Qi required in Equation (176), viz.,
MXBH ðSbi ½S0bsfi ð1 þ rÞSbsN þ Sbsai ÞYH Rs ¼ Qi ð1 þ bHT Rs Þ ðmgOHOVSSÞ
ð181Þ
Equations (176) and (181) need to be solved simultaneously to calculate the concentration of FBSO (SbsfN) exiting the last anaerobic compartment (N); the following procedure converges in three to four iterations: (1) Assume SbsfN ¼ 0 mgCOD l1, (2) calculate MXBH/Qi with Equation (181), (3) with MXBH/Qi known, calculate SbsfN with Equation (176), (4) recalculate MXBH/Qi using the new value for SbsfN, (5) repeat steps (3)–(5) until SbsfN and MXBH/Qi are constant. This procedure splits the influent BO between the OHOs and PAOs. Because the growth processes of two organism
Biological Nutrient Removal
groups after the anaerobic reactor are noncompetitive and VFA uptake process and the growth processes on the available organics are complete for both groups, the stoichiometric equations relating the flux of COD utilized and the biomass produced derived earlier (Equation (103)) can be applied to calculate the PAO and OHO masses and their endogenous residue masses.
509
(Supi, XIi) (Equation (67)), viz.,
MXI ¼ FSti f S0 up =f cv Rs
ðmgIVSSÞ
Total VSS in the NDBEPR system is the sum of the five VSS components:
MXv ¼ MXBH þ MXBG þ MXEH þ MXEG þ MXI ðmgVSSÞ
4.14.31.1.5 Mass of VSS in the NDBEPR system
ð188Þ
PAO mass
MXBG ¼ FSbPAO
YG Rs 1 þ bGT Rs
ðmgPAOVSSÞ
ð182Þ
where, YG is the PAO yield coefficient (mgPAOVSS/mgCOD utilized), FSbPAO the flux BO taken up by PAOs in the anaerobic reactor (mgCODd1), and bGT the PAO specific endogenous mass loss rate constant at temperature T (d1).
4.14.31.1.6 PAO P release From the mechanisms of BEPR (Wentzel et al., 1985, 1990), for every mole of VFA taken up, 1 mol of P is released to provide energy to synthesize and store the VFA as PHA. Accordingly, the P release in the anaerobic reactor is given by
FPrel ¼ f prel FSbPAO
ðmgP d1 Þ
ð189aÞ
PAO endogenous mass or
MXEG ¼ f EG bGT MXBG Rs
ðmgVSSÞ
ð183Þ Prel ¼ f prel SbPAO
where fEG is the fraction of PAOs that is unbiodegradable particulate endogenous residue. PAO oxygen demand
FOGc ¼ FOGs ðsynthesisÞ þ FOGe ðendogenous respirationÞ ¼ ð1 f cv YG ÞFSbPAO þ f cv ð1 f EG ÞbGT MXBG YG Rs ¼ FSbPAO ð1 f cv YG Þ þ f cv ð1 f EG ÞbGT 1 þ bGT Rs ðmgO d1 Þ
ð184Þ
OHO mass
MXBH
ðmgP l1 influentÞ
where fprel is the ratio P release/VFA uptake E1.0 molP/mol COD E0.5 mgP/mgCOD and SbPAO the concentration COD taken up by the PAOs per liter influent ¼ FSbPAO/Qi.
4.14.31.1.7 P removal The P removal via the waste sludge is calculated from the individual P content of the five VSS components, viz: By PAOs
MXBG MXEG 1 DPG ¼ f XBGP þ f XEGP Rs Rs Qi ðmgP l1 influentÞ
YH Rs ¼ FSbOHO 1 þ bHT Rs
ðmg OHOVSSÞ
ð185Þ
where YH is the OHO yield coefficient (mgOHOVSS/mgCOD utilized), FSbOHO the flux BO taken up by OHOs in the anaerobic reactor (mgCOD d1), and bHT the OHO specific endogenous mass loss rate constant at temperature T (d1).
ðmgVSSÞ
By OHOs
ð186Þ
FOHc ¼ FOHs ðsynthesisÞ þ FOHe ðendogenous respirationÞ ¼ ð1 f cv YH ÞFSbOHO þ f cv ð1 f EH ÞbGT MXBH YH Rs ¼ FSbOHO ð1 f cv YH Þ þ f cv ð1 f EH ÞbHT bHT Rs ðmgO d Þ
where PG is the P removal by the PAOs (mgP l1influent), fXBGP the P content of PAOs ¼ 0.38 mgP/mgPAOVSS, and fXEGP the P content PAO endogenous mass ¼ 0.03 mgP/ mgEVSS.
MXBH MXEH 1 DPH ¼ f XBHP þ f XEHP Rs Rs Qi
where fEH is the fraction of OHOs that is unbiodegradable particulate endogenous residue. OHO oxygen demand
1
ð190Þ
OHO endogenous mass
MXEH ¼ f EH bHT MXBH Rs
ð189bÞ
ðmgP l1 influentÞ
ð191Þ
where PH is the P removal by the OHOs (mgP l1influent), fXBHP the P content of OHOs ¼ 0.03 mgP/mgOHOVSS, and fXEHP the P content OHO endogenous mass ¼ 0.03 mgP/ mgEVSS. By inert mass
MXI 1 DPI ¼ f XIP Rs Qi
ðmgP l1 influentÞ
ð192Þ
ð187Þ
The same equations derived earlier in Section 4.14.7.1.1 apply for the UPO that accumulate in the reactor from the influent
where PI is the P removal due to inert mass (mgP l1 influent) and fXIP the P content inert VSS mass (mgP/mgIVSS) ¼ 0.025– 0.03 mgP/mgIVSS.
510
Biological Nutrient Removal
The total P removal is given by the sum of the individual P removals, i.e. Total P removal
DPT ¼ DPG þ DPH þ DPI
ðmgP l1 influentÞ
ð193Þ
The effluent P concentration is given by the difference between the influent P and the P removal, i.e. Effluent P concentration
Pte ¼ Pti PT
ðmgP l1 Þ
mgPAOVSS), and the polyP ISS, which is 3.286 mgISS/mgP times the PAO polyP content, which is its total P content (fXBGP) minus its biomass P content (Ekama and Wentzel, 2004). Hence,
ð194Þ
If the P removal is greater than the influent P concentration, then the expectation is that the effluent P concentration will be below 0.5 mgP l1. How far below 0.5 mgP l1 is uncertain because currently this appears to be plant specific. Research is being conducted to investigate what the limits of BEPR technology are and what conditions in the NDBEPR system cause them (Neethling et al., 2009). Revised PAO P content (fXBGP). If the P removal is greater than the influent P concentration, then there is insufficient P in the influent for the PAOs to take up P up to their maximum P content of 0.38 mgP/mgPAOVSS. Their P content (fXBGP) will therefore be limited by the available P. Under these conditions, the PAO P content needs to be revised to match the available P. If this is not done, the reactor ISS concentration, which is strongly influenced by the PAO P content, will be overestimated. In the calculation for the revised PAO P content, it is assumed that the effluent P concentration (Pte) is equal to the P concentration of the unbiodegradable soluble organics (USO; see Section 4.14.4.4.3) and that the P content of the non-PAO VSS components remains unchanged. Unless data are available to indicate a nonzero USO P concentration (Pousi40), it is reasonable to accept it as zero. Clearly, if the wastewater contains USO P, then this will impact achieving the very low effluent P standards that are being set for NDBEPR systems these days. However, it would appear that USO P in municipal wastewaters is effectively zero, or at least masked by the scatter of the difference between membrane filtered effluent TP and OP concentrations. The revised PAO P content (fXBGP) is found by making fXBGP the subject of Equation (193) and the P removal equal to the difference between the influent P and USO P concentrations (Equation (39)), viz.,
f XBGP ¼ ½ðPti Pousi ÞQi Rs f XEGP MXEG f XBHP MXBH f XEHP MXEH f XIP MXI =MXBG ðmgP=mgPAOVSSÞ ð195Þ
4.14.31.2 VSS and TSS Sludge Masses in the Reactor (System) The VSS mass in the NDBEPR reactor is the sum of the five VSS component masses (Equation (188)). The ISS concentration is the sum of the ISS that accumulates in the reactor from the influent (Equation (97)), the OHO ISS, and the PAO ISS. The OHO ISS is 15% of its VSS mass, that is, fiOHO ¼ 0.15 mgISS/ mgPHOVSS (Equation (98)). The PAO ISS is the sum of its biomass ISS, which is the same at the OHO ISS (0.15 mgISS/
XIO ¼ FXIOi Rs þ MXBH þ 3:286ðf XBGP f XBGPBM ÞMXBG ðmgISSÞ ð196Þ where fXBGPBM is the PAO biomass P content ¼ OHO biomass P content ¼ 0.025–0.03 mgP/mgPAOVSS. The TSS mass in the NDBEPR system is the sum of the VSS and ISS masses, that is,
MXt ¼ MXv þ MXIO
ðmgTSSÞ
ð197Þ
This TSS mass is distributed in the various reactors of the NDBEPR system, not necessary at the same TSS concentration in each reactor. The reactor configuration (Figure 47) influences the TSS concentration in the different reactors of the system. Calculating the reactor concentrations from the various mass fraction of the reactors is discussed below.
4.14.31.3 BEPR System Design Considerations 4.14.31.3.1 Process volume requirements An approximate reactor volume, that is, a nonconfigurationspecific volume, can be estimated from a selected average reactor TSS concentration required for the system, that is,
Vp ¼ MXt =Xt
ðm3 Þ
ð198Þ
where Xt is the zone/reactor volume weighed average TSS concentration in the NDBEPR system (mgTSS l1). For all NDBEPR system configurations with SSTs, or with membranes (MBR), at steady-state and average dry weather flow (ADWF) conditions, the concentrations of TSS in the preanoxic (Figure 47(f)) and anaerobic (if present) and anoxic and aerobic zones (Xtpax, Xtana, Xtanx, Xtaer), as fractions of the average system TSS concentration Xt are equal to the ratio of the sludge mass fraction and volume fraction of the zones, that is,
Xtana f mana Xtanx f manx ¼ ; ¼ ; Xt f vana Xt f vanx Xtpax f mpax Xtaer f maer ¼ ; ¼ Xt f vaer Xt f vpax
ð199Þ
where fm, fv are the zone sludge mass and volume fractions respectively, and subscripts ana, anx, aer, and pax are the anaerobic, anoxic, aerobic, and pre-anoxic zones, respectively. For BNR systems with SSTs in which the sludge mass is uniformily distributed, that is, the TSS concentrations are the same in the anaerobic, anoxic, and aerobic zones of the reactor, the sludge mass and volume fractions are equal, such as in the three- and five-stage Bardenpho systems (Figures 47(b) and 47(c)) for N and P removal and the pre- (modified Ludzack–Ettinger, MLE) and post-(Wuhrmann) denitrification and four-stage Bardenpho systems for N removal. For example, if an MLE ND system (Figure 34(b)) requires anoxic and aerobic mass fractions (fmanx, fmaer) of 0.45 and 0.55,
Biological Nutrient Removal
respectively, or a three-stage Bardenpho (Figure 47(b)) system requires anaerobic, anoxic, and aerobic mass fractions (fmana, fmanx, fmaer) of 0.15, 0.35, and 0.50 respectively, the corresponding volume fractions of these zones (fvana, fvanx, fvaer) with respect to the reactor volume (VR) will also be 0.45 and 0.55 for the MLE system and 0.15, 0.35, and 0.50 for the three-stage Bardenpho system. This is because the influent flow dilutes the SST return sludge concentration in the first zone by the same amount as the SST concentrates it after the last zone. This equality of sludge mass and volume fractions does not apply to any multizone BNR system with membrane solid–liquid separation in the aerobic zone, because the aerobic zone concentration is in effect the equivalent of the return sludge concentration from the SST (if there were SSTs). For BNR systems with SSTs, in which the TSS concentrations are not the same in the pre-anoxic, anaerobic, anoxic, or aerobic zones (e.g., in the UCT (Figure 47(d)) or in the JHB (Figure 47(f)) systems), the volume and mass fractions are not equal. For the UCT system, the volume fractions (with respect to Vp) of the anaerobic, anoxic, and aerobic zones (fvana, fvanx, fvaer), and the anaerobic, anoxic, and aerobic TSS concentrations (Xtana, Xtanx, Xtaer) at steady-state and ADWF conditions are related to the anaerobic and aerobic mass fractions (fmana, fmaer), recycle ratio (r) from the anoxic to the anaerobic reactor and system average TSS concentration Xt , as follows:
f mana ðr þ 1Þ rB
ð200aÞ
ð1 f mana f maer Þ B
ð200bÞ
f maer B
ð200cÞ
rB ðr þ 1Þ
ð200dÞ
f vana ¼ f vanx ¼
f vaer ¼
Xtana ¼ Xt
Xtanx ¼ Xtaer ¼ Xt B
ð200eÞ
f mana 1þ r
ð200f Þ
where
B¼
For the JHB system with SSTs, assuming the influent flow to the pre-anoxic zone, which is sometimes included to increase pre-denitrification, is zero, the volume fractions (with respect to Vp) of the pre-anoxic, anaerobic, anoxic, and aerobic zones (fvpax, fvana, fvanx, fvaer), and the pre-anoxic, anaerobic, anoxic, and aerobic TSS concentrations (Xtpax, Xtana, Xtanx, Xtaer) at steady-state and ADWF conditions are related to the pre-anoxic, anaerobic, and aerobic mass fractions (fmpax, fmana, fmaer), underflow recycle ratio (s) from the SST to the pre-anoxic reactor and average TSS concentration Xt , as follows:
f mana C
ð201aÞ
ð1 f mana f maer f mpax Þ C
ð201bÞ
f vana ¼ f vanx ¼
f maer C
ð201cÞ
f mpax s Cðs þ 1Þ
ð201dÞ
f vaer ¼ f vpax ¼
Xtana ¼ Xtanx ¼ Xtaer ¼ Xt C
C¼
ð201eÞ
Cs ð1 þ sÞ
ð201f Þ
f mpax 1 1þs
ð201gÞ
Xtpax ¼ Xt where
511
In BNR systems with membrane solid–liquid separation in the aerobic zone, the sludge mass distributes itself differently in the different zones of the system compared with systems with SSTs. This is because the effluent is withdrawn via the membranes from the aerobic zone which concentrates the sludge in this zone relative to that in the other zones. However, in recycling this concentrated aerobic zone sludge to an upstream zone, it is diluted by the less concentrated incoming sludge stream from the upstream zones. The higher the recycles from downstream zones to upstream zones, the more uniformily the sludge mass is distributed around the system and the closer the sludge concentrations in the different zones. For the UCT system with membranes, the volume fractions (with respect to Vp) of the anaerobic, anoxic, and aerobic zones (fvana, fvanx, fvaer), and the anaerobic, anoxic, and aerobic TSS concentrations (Xtana, Xtanx, Xtaer) at steady-state and ADWF conditions are related to the anaerobic and aerobic mass fractions (fmana, fmaer), recycle ratio (r) from the anoxic to the anaerobic zone, recycle ratio (a) from the aerobic to the anoxic zones, and system average TSS concentration Xt , as follows:
f mana ðr þ 1Þ Dr
ð202aÞ
ð1 f mana f maer Þ D
ð202bÞ
af maer ða þ 1ÞD
ð202cÞ
rD ðr þ 1Þ
ð202dÞ
f vana ¼
f vanx ¼
f vaer ¼
Xtana ¼ Xt
Xtanx ¼ Xt D Xtaer ¼ Xt where
D¼
ða þ 1ÞD a
f mana f maer 1þ r ða þ 1Þ
ð202eÞ ð202f Þ
ð202gÞ
For the JHB system with membranes, the volume fractions (with respect to Vp) of the pre-anoxic, anaerobic, anoxic, and aerobic zones (fvpax, fvana, fvanx, fvaer), and the pre-anoxic, anaerobic, anoxic, and aerobic TSS concentrations (Xtpax, Xtana,
512
Biological Nutrient Removal
Xtanx, Xtaer) at steady-state and ADWF conditions are related to the pre-anoxic, anaerobic, and aerobic mass fractions (fmpax, fmana, fmaer), recycle ratio (s) from the aerobic to the pre-anoxic zones, recycle ratio (a) from the aerobic to the anoxic zones, and average TSS concentration Xt , as follows:
f vana ¼ f vanx ¼
f mana ð1 þ sÞ sE
ð1 f mana f maer f mpax Þða þ s þ 1Þ ða þ sÞE
ð203aÞ
ð203bÞ
f vaer ¼
f maer E
ð203cÞ
f vpax ¼
f mpax E
ð203dÞ
sE ðs þ 1Þ
ð203eÞ
Xtana ¼ Xt
Eða þ sÞ Xtanx ¼ Xt ða þ s þ 1Þ Xtaer ¼ Xtpax ¼ Xt E
ð203f Þ ð203gÞ
ð1 þ sÞ ða þ s þ 1Þ þ f manx E ¼ f mana s ða þ sÞ þf maer þ f mpax
ð203hÞ
Equation (203) applies also to the MLE ND system and the three-stage Bardenpho system with membranes. In Equation (203h) for E, for the MLE system, the anaerobic and pre-anoxic mass fractions are both set to zero, the anoxic mass fraction is 1 minus the aerobic mass fraction (i.e., fmanx ¼ 1 fmaer), and the mixed liquor recycle ratio (a) is also set to zero – only one recycle (s) is required to return nitrate and sludge to the anoxic reactor. For the three-stage Bardenpho system, only the pre-anoxic sludge mass fraction (fmpax) is set to zero. From Equations (200)–(203), the volumes of, and the TSS concentrations in, the various zones of common BNR systems with SST or membrane solid–liquid separation can be calculated for selected anaerobic, aerobic, and pre-anoxic mass fractions (fmaer, fmana, fmpax), and interzone recycle ratios (a, r, and s). In the derivation of these equations, steady-state conditions were assumed and the sludge waste flow rate was ignored – the effect of this is negligible (o2%), especially if the sludge age is long. Generally, a uniform distribution of sludge mass in BNR MBR systems will not occur, even in systems with a single recycle flow from the aerobic to the zone receiving the influent flow. For example, changing an MLE ND system, or a three-stage Bardenpho system with SSTs to membrane solid– liquid separation systems, will change these systems from uniformly distributed sludge mass systems in which the sludge mass and volume fractions are equal to nonuniformly distributed sludge mass systems in which the sludge mass and volume fractions are different, the magnitude of difference depending on the magnitude of the recycle ratios. In multizone BNR systems with membranes in the aerobic reactor and fixed volumes for the anaerobic, anoxic, and
aerobic zones (i.e., fixed volume fractions), the mass fractions can be varied (within a range) by varying the inter-reactor recycle ratios. For example, in a UCT system with anaerobic, anoxic, and aerobic zone volume fractions of 0.25, 0.35, and 0.40 and an r recycle ratio from the anoxic to the anaerobic zones of 1:1, the anaerobic, anoxic, and aerobic zone mass fractions can be varied from 0 to 0.131, 0 to 0.366, and 1 to 0.503, respectively, by changing the a recycle ratio from 0:1 to 5:1. Increasing the a recycle ratio concomitantly increases the nitrate load on the anoxic reactor, thereby increasing the denitrification and N removal as the anoxic mass fraction increases. Increasing the r recycle ratio increases the anaerobic mass fraction (at the expense of the other two zone mass fractions) and increases (not proportionally) the P removal. This zone mass fraction flexibility is a significant advantage of membrane BNR systems over conventional BNR system with SSTs because it allows changing the mass fractions to optimize biological N and P removal in conformity with influent wastewater characteristics and the effluent N and P concentrations required. If required, the performance of membrane BNR systems can be simulated with current BNR AS models such as UCTOLD (for ND, Dold et al., 1991), UCTPHO (for NDBEPR with 490 aerobic P uptake BEPR, Wentzel et al., 1992; Hu et al., 2003), and IWA ASM Nos 1, 2 (ND and BEPR, Henze et al., 1987) by returning the SST underflow into the aerobic zone from which the SST feed flow exits (Parco et al., 2009). However, such simulations require a priori information on the reactor and zone volumes and recycle flows, which would need to be determined with the steady-state procedures set out in this chapter.
4.14.31.3.2 Nitrogen requirements for sludge production The form of the equation for calculating the nitrogen requirement for sludge production (Ns, mgN l1 influent) is the same as set out in Section 4.14.21, Equation (142), that is,
Qi Ns ¼ f n MXv =Rs
ðmgN d1 Þ
However, for the BEPR system the term MXv needs to take account of the changes in VSS components, that is, it must be calculated using Equation (188). Effect of this is to increase Ns because MXv is greater in the NDBEPR system than in the same sludge age ND system receiving the same wastewater. The increase in Ns decreases the nitrification capacity (Nc) (Equation (152)), and hence also the nitrification oxygen demand (FOn, mgO d1). For the rest, the nitrification model calculations remain the same.
4.14.31.3.3 Total oxygen demand The carbonaceous oxygen demand (FOc) is the sum of oxygen demands of PAOs (Equation (184)) and OHOs (Equation (187)):
FOc ¼ FOGc þ FOHc ¼ ð1 f cv YG ÞFSbPAO þ f cv ð1 f EG ÞbGT MXBG þ ð1 f cv YH ÞFSbOHO þ f cv ð1 f EH ÞbHT MXBH ðmgO d1 Þ
ð204Þ
Biological Nutrient Removal
4.14.32 Influence of BEPR on the System 4.14.32.1 Influence on VSS, TSS, and Carbonaceous Oxygen Demand The model for BEPR systems presented above enables the VSS and TSS of the mixed liquor (Equations (188) and (197), respectively) and the carbonaceous oxygen demand (Equation (204)) to be calculated. A comparison of the masses of VSS and TSS in the reactor and the carbonaceous oxygen demand per kg COD load on the bioreactor versus sludge age with and without BEPR are shown in Figures 54(a) and 54(b) for the example raw and settled wastewaters respectively, with influent RBO fractions with respect to the biodegradable COD (fSb’s) of 0.25 and 0.38, respectively (Table 14) and an VFA fraction of 25% of the RBO, that is, influent RBO and VFA concentration of 146 and 36 mgCOD l1 for both wastewaters. The features of the BEPR system are a UCT configuration operated at 20 1C with two equal-sized in-series anaerobic compartments with a total anaerobic mass fraction (fxana) of 0.15, an anoxic to anaerobic (r) recycle of 1:1, and no nitrate recycled to the anaerobic reactor. From Figures 54(a) and 54(b), BEPR in the AS system increases the VSS slightly, by about 5–12% and 15–25% for
Raw wastewater
8
0.8 MLTSS Oxygen demand
0.6
0.4
4 MLVSS 2
0.2
0.0
0 0
5
Although there is only a small difference in VSS production between a BEPR and a non-BEPR system, the constituent
10 15 20 25 Sludge age (days)
30
Settled wastewater
10 Sludge mass − kgVSS/(kgCOD/d) reactor
With BEPR No BEPR 20 °C
6
4.14.32.2 VSS Composition
0.1
Oxygen demand − (kgO/d)/(kgCOD/d)
Sludge mass − kgVSS/(kgCOD/d) reactor
10
raw and settled wastewaters respectively depending on sludge age – the longer the sludge age the greater the difference. This increase in VSS is due to the lower endogenous mass loss/ death rate of the PAOs (0.04 d1 at 20 1C) compared with the OHOs (0.24 d1 at 20 1C). However, the TSS is increased substantially, by about 20–25% and 45–55% for raw and settled wastewaters, respectively, depending on the sludge age. This higher TSS production is due to the large quantities of stored inorganic polyP and the associated inorganic cations necessary to stabilize the polyP chains – principally Mg2þ and Kþ (Fukase et al., 1982; Arvin, 1985; Comeau et al., 1986; Wentzel et al., 1988; Ekama and Wentzel, 2004). The high inorganic content of the PAOs causes the VSS/TSS to be much lower than that of the OHOs, 0.46 mgVSS/mgTSS compared with 0.87 mgVSS/mgTSS (excluding the influent ISS). Thus, the higher the PAO fraction of the mixed liquor, the higher the BEPR, but the lower the VSS/TSS ratio of the mixed liquor. The increase in TSS with the inclusion of BEPR needs to be taken into account in the design of the bioreactor volume (Equation (198)) and daily sludge production. Also, since the inorganic cations that stabilize the polyP are derived from the influent wastewater, there must be sufficient concentrations of these cations in the influent; otherwise, the BEPR may be adversely affected (Wentzel et al., 1989a; Lindrea et al., 1994). Further, because the VSS mass generated per kg COD load is greater with BEPR than without, the oxygen demand with BEPR is correspondingly reduced, by about 5–6% and 8–9% for raw and settled wastewaters, respectively (depending on sludge age, Figures 54(a) and 54(b)).
1.0
With BEPR No BEPR 20 °C 8
0.8 Oxygen demand 0.6
6 MLTSS
0.4
4
MLVSS
2
0.2
Oxygen demand − (kgO/d)/(kgCOD/d)
The total oxygen demand (FOt, mgO d1) is the sum of the carbonaceous and nitrification oxygen demands, taking due account of the change in nitrogen requirements for sludge production (Ns) and nitrification capacity (Nc). For a nonnitrifying BEPR systems, the total oxygen demand FOt is given by FOc. Including nitrification in the BEPR system necessarily means that denitrification must also be included; the effect of nitrification and denitrification on the total oxygen demand will be considered in Section 4.14.34.
513
0.0
0 0
5
10 15 20 25 Sludge age (days)
30
Figure 54 Predicted masses of volatile solids (MXV, MLVSS) and total solids (MXt MLTSS) and daily carbonaceous oxygen demand (FOC) per kg COD load on the biological reactor in ND (thin line) and BEPR (bold line) systems treating raw (a) raw and settled (b) wastewater.
Biological Nutrient Removal
Settled wastewater with BEPR
200
I−1
750 mg COD 25% RBCOD fraction % Of VSS mass (settled WW)
% Composition of VSS mass
Inert 60 OHO endogenous 40 OHO active PAO endog 20
90 80
Additional VSS mass in system treating raw WW, i.e., raw WW produces about 100% more activated sludge VSS mass
140 120
70 60
100
50 Inert
80
40
OHO endogenous
60
30
OHO active
PAO endog
40
20 10
PAO active
0
0 0
5
10
15
20
25
5
% Of VSS mass (settled WW)
Inert mass 60 OHO endogenous mass
40
20
OHO active mass
20
25
30
100
450 mg COD I−1 38% RBCOD fraction
180
80
15
Settled wastewater no BEPR
200
−1
750 mg COD I 25% RBCOD fraction
10
Sludge age (days)
(b)
Raw wastewater no BEPR
100
0 0
30
Sludge age (days)
(a)
% Composition of VSS mass
160
20
PAO active
90 80
160 Additional VSS mass in system treating raw WW. i.e., raw WW produces about 100% more activated sludge VSS mass
140 120
70 60 50
100 Inert
40
80 OHO endogenous
60
30 20
40 OHO active
20
10 0
0
0 0 (c)
450 mg COD 38% RBCOD fraction
180
80
100
I−1
5
10
15
20
25
0
30
Sludge age (days)
(d)
% Of VSS mass (raw WW)
Raw wastewater with BEPR
100
% Of VSS mass (raw WW)
514
5
10
15
20
25
30
Sludge age (days)
Figure 55 Percentage composition of VSS mass for BEPR systems (a, c) and ND systems (No BEPR, b, d) treating raw (a, b) and settled (c, d) wastewater.
sludge fractions for the two systems differ markedly. This can be readily demonstrated by comparing the percentage composition of the VSS mass generated in systems exhibiting BEPR with the ND system that does not. To illustrate, percentage composition of the VSS mass is shown in Figures 55(a) to 55(d) for systems at 20 1C with no BEPR (Figures 55(b)– 55(d)) and with BEPR (Figures 55(a) and 55(c)) respectively treating the example raw (Figures 55(a) and 55(b)) and settled (Figures 55(c) and 55(d)) wastewaters. Note that the BEPR system has a smaller OHO active mass than the no BEPR ND system, but the BEPR system has additionally a significant concentration of PAO biomass.
4.14.32.3 P/VSS ratio A parameter often used to evaluate the BEPR performance of an AS system is the P/VSS (or P/TSS) ratio of the mixed liquor. In Figures 56(a) and 56(b), the calculated P/VSS ratio for a BEPR system with a two-compartment anaerobic reactor and the example raw and settled wastewater characteristics are plotted versus sludge age. A zero discharge of nitrate to the anaerobic reactor is assumed. From Figures 56(a) and 56(b), as the system sludge age increases, the P/VSS ratio increases up to a sludge age of about 10 days. Further increases in sludge age cause a decrease in P/VSS ratio. The initial increase in
Biological Nutrient Removal 15
20
10
0.05 fxana
5
0.0
0
15 0.10 0.05
10
fxana 5 0.0
0 5
0 (a)
10 15 20 Sludge age (days)
25
30
0
5
(b)
10 15 20 Sludge age (days)
25
30
20
15
Settled WW
10
%P in TSS (mgP/mgVSS*100)
Raw WW %P in TSS (mgP/mgVSS*100)
0.25 0.20 0.15
Settled WW
0.25 0.20 0.15 0.10
%P in VSS (mgP/mgVSS*100)
%P in VSS (mgP/mgVSS*100)
Raw WW
0.25 0.20 0.15 0.10
5 fxana
0.05 0.0
15
0.25 0.20 0.15 0.10
10
0.05 fxana
5
0.0
0
0 0 (c)
515
5
10 15 20 Sludge age (days)
25
0
30 (d)
5
10 15 20 Sludge age (days)
25
30
Figure 56 Predicted percentage phosphorus to VSS (P/VSS 100; a, b) and TSS (P/TSS 100; c, d) ratios vs. sludge age for mixed liquor in a BEPR system with various anaerobic mass fractions (fxana) treating the example raw (a, c) and settled (b, d) wastewater.
P/VSS with sludge age is due increasing OHO mass with sludge age, which increase the fermentable RBO to VFAs conversion efficiency in the anaerobic reactor and accordingly yields an increased PAO mass (with associated P content of 0.38 mgP/mgVSS). The decrease in P/VSS can be ascribed to the endogenous respiration effect on PAOs. The P/VSS ratio is therefore a consequence of the selection of the system design parameters sludge age and anaerobic mass fraction and wastewater characteristics. Accordingly, the P/VSS ratio can neither fulfill a function in BEPR plant design nor in BEPR performance assessment between different BEPR plants.
4.14.33 Factors Influencing Magnitude of BEPR The influence of the main design parameters on the magnitude of P removal is demonstrated with the mixed culture steady-state BEPR model. These main parameters are: raw settled wastewater (Sti ¼ 750 and 450 mgCOD l1 respectively, Tables 7, 11, 14), sludge age (SRT ¼ 20 days), anaerobic sludge
mass fraction (fxana ¼ 0.15), influent RBO COD fraction (fSb’s ¼ 0.25 for raw and 0.385 for settled), discharge of nitrate and oxygen to the anaerobic reactor (0 for both) and subdivision of anaerobic reactor into compartments (N ¼ 2). The numbers in brackets are the default wastewater characteristics and system design parameter values.
4.14.33.1 Sludge Age and Anaerobic Mass Fraction Using the characteristics of the example raw and settled wastewater with a total influent COD of 750 and 450 mgCOD l1 respectively, assuming (1) no nitrate and DO enters the anaerobic reactor, (2) a recycle ratio to the anaerobic (r) of 1:1 and the anaerobic reactor is subdivided into two compartments, the P removal (normalized with respect to influent COD, mgPmg1 influent COD) versus sludge age is shown in Figures 57(a) (raw) and 57(b) (settled) for anaerobic mass fractions of 0.00 (no BEPR), 0.05, 0.10, 0.15, 0.20, and 0.25. In the same figures, the actual P removal in mgP l1 is shown on the right-hand axis.
Biological Nutrient Removal
0.02
0.25 0.20 0.15 0.10
22.5
15.0
Influent P 0.05
fxana
0.01
7.5
0.0
5
10 15 20 25 Sludge age (days)
0.04 0.03
Influent P
0.02
0.25 0.20 0.15 0.10
18.0 13.5 9.0
0.05
fxana 4.5
0.01 0.0
0.00
0.0 0
22.5 Settled WW
0.0 0
30 (b)
P removal (mgP I−1)
0.03
0.00 (a)
0.05
30.0 Raw WW
P removal (mgP/mg Infl COD)
P removal (mgP/mg Infl COD)
0.04
P removal (mgP I−1)
516
5
10 15 20 25 Sludge age (days)
30
Figure 57 Predicted P removal vs. sludge age for various anaerobic mass fractions (fxana) for a two-compartment anaerobic reactor BEPR system at 20 days sludge age, treating the example raw (a) and settled (b) wastewater.
The effect of sludge age on P removal is complex. For sludge age o3 days, the P removal increases with increase in sludge age and for sludge age 43 days, P removal decreases with increase in sludge age. The reason for this is that an increase in sludge age causes an increase in the system OHO mass, which in turn causes an increase in fermentable RBO conversion and, therefore, an increase in P release, P uptake, and P removal. However, the increased sludge age also causes a decrease in PAO biomass, its associated P content, due to endogenous respiration which decreases the P removal. At sludge age o3 days, the former effect dominates the P removal, while at sludge age 43 days the latter dominates. The decrease in both PAO biomass with increase in sludge age is crucially affected by the specific endogenous mass loss rate of the PAOs – should the endogenous mass loss rate of the PAOs (0.04 d1) have been the same as that of the OHOs (0.24 d1), virtually no BEPR would have been obtained. The effect of anaerobic mass fraction (fxana) on P removal also is shown in Figures 57(a) and 57(b). For a selected sludge age, an increase in fxana always gives rise to an increase in P removal. This is due to the increased conversion of fermentable RBO with larger anaerobic mass fractions. The improvement in P removal, however, diminishes with each step increase in fxana, due to the first-order nature of the RBO conversion kinetics. From Figures 57(a) and 57(b), it can be seen that for fxana 40.15 only small additional increases in P removal are obtained, which usually are not justified due to the decrease in unaerated sludge mass fraction this causes and the consequent impact on the minimum sludge age for nitrification.
4.14.33.2 Raw and Settled Influent The effect of primary settled wastewater on P removal can be seen by comparing Figures 57(a) and 57(b), which show the P removal for the raw wastewater of original COD of 750 mgCOD l1 and the settled wastewater produced from the raw wastewater with a 450 mgCOD l1. It can be seen that although the P removal per mg influent COD is higher for the settled WW, the P removal in mgP l1 is lower. This decrease is due to the decrease in the flux of biodegradable COD entering
the AS system which causes a reduction in OHO biomass and hence in the fermentable RBO converted and in the mass of PAOs generated. The P removal per influent COD entering the biological reactor is higher for the settled because the fraction of the biodegradable organics that is RBO (Sbsi/Sbi) is higher for settled than for unsettled wastewater because primary settling removes only the settleable organics (although not strictly true, RBO loss or gain in primary settling appears to be small; it is assumed that the RBO is not changed during primary settling).
4.14.33.3 Influence of Influent RBO Fraction Assuming zero discharge of nitrate to the two-compartment anaerobic reactor of mass fraction (fxana) of 0.15, the effect of the influent RBO fraction (fSb’s ¼ Sbsi/Sbi) is illustrated in Figures 58(a) and 58(b) for raw and settled wastewaters, respectively. At any anaerobic mass fraction, the higher the influent RBO fraction, the higher the P removal. In design, one option to improve the P removal is supplementation of influent RBO by, for example, acid fermentation of primary sludge (Pitman et al., 1983; Barnard, 1984; Osborn et al., 1989) or dosing other RBO or VFA into the anaerobic reactor.
4.14.33.4 Influence of Recycling Nitrate and Oxygen to the Anaerobic Reactor The influence of nitrate recycled to the anaerobic reactor is illustrated in Figures 59(a) and 59(b) which show P removal versus nitrate concentration recycled to the anaerobic reactor in a recycle ratio 1:1. Clearly, in agreement with numerous experimental and full-scale NDBEPR systems, recycling nitrate has a markedly deleterious influence on the magnitude of P removal. As the nitrate concentration recycled to the anaerobic reactor increases, the P removal decreases. The same applies to oxygen entering the anaerobic reactor, except that its effect is 1/2.86 times that of nitrate because the oxygen equivalent of nitrate is 2.86 mgO/mgNO3-N. If oxygen and/or nitrate are recycled to the anaerobic reactor, the OHOs no longer convert fermentable RBO to VFAs but instead themselves utilize it for energy and growth with the oxygen or nitrate as external electron acceptor. For every 1
Biological Nutrient Removal
0.02
13.5
9.0
0.05
fxana
0.01
4.5
0.0
0
0.1
0.2
0.3
0.4
0.5
22.5
Influent P
0.02
15.0 0.05
fxana 0.01
0.6
7.5
0.0
0.0 0
Influent RBO fraction (fSb’s)
(a)
0.03
0.00
0.0
0.00
30.0
0.25 0.20 0.15 0.10
Raw WW
P removal (mgP I−1)
Influent P
0.03
0.04
18.0
0.25 0.20 0.15 0.10
P removal (mgP/mg Infl COD)
Settled WW
P removal (mgP I−1)
P removal (mgP/mg Infl COD)
0.04
517
0.1
0.2
0.3
0.4
Influent RBO fraction (fSb’s)
(b)
Figure 58 Predicted P removal vs. readily biodegradable COD (RBCOD, Sbsi) as a fraction of the biodegradable COD (Sbi) fSb’s ¼ Sbsi/Sbi) for various anaerobic mass fractions (fxana) for a two-compartment anaerobic reactor BEPR system at 20 days sludge age, treating the example raw (a) and settled (b) wastewater.
0.25 0.20 0.15 0.10
0.02
22.5 Influent P
15.0
0.05
fxana
0.01
7.5
0.0
0 ( a)
8 2 4 6 Nitrate in recycle (NO3−N I−1)
0.25 0.20 0.15 0.10
0.03
Influent P
13.5
9.0
0.02 0.05
fxana 4.5
0.01 0.0
0.00
0.0
0.00
18.0 Settled WW
10
0.0 0
(b)
P removal (mgP I−1)
0.03
P removal (mgP/mg Infl COD)
Raw WW P removal (mgP I−1)
P removal (mgP/mg Infl COD)
0.04
30.0
0.04
2 4 6 8 Nitrate in recycle (NO3−N I−1)
10
Figure 59 Predicted P removal vs. nitrate concentration in recycle to anaerobic (recycle 1:1) for various anaerobic mass fractions (fxana) for a twocompartment anaerobic reactor BEPR system at 20 days sludge age treating the example raw (a) and settled (b) wastewater.
mgO2 and 1 mgNO3-N recycled to the anaerobic reactor, 3.0 and 8.6 mgCOD, respectively, are utilized (Equation (175)). Consequently, allowing oxygen and/or nitrate to enter the anaerobic reactor reduces the flux of VFAs available to the PAOs for storage, and correspondingly reduces the P release, P uptake, and P removal. From Figures 59(a) and 59(b), when the nitrate concentration in the recycle exceeds about 12 mgN l1, the P removal decreases to 4 (raw) and 2.2 mgP l1 (settled) which is the same as that of an ND system (fxana ¼ 0) with zero BEPR. In this situation, all the influent RBO is denitrified by the OHOs with the result that no VFAs are released and no VFAs are available to the PAOs, and BEPR no longer takes place – the P removal obtained is due to wastage of sludge with normal metabolic P content (0.03 mgP/mgVSS). If the influent RBO concentration increases or decreases, the concentration of recycled nitrate that completely consumes the RBO will increase or decrease correspondingly below about 12 mgN l1 (provided the recycle ratio remains unchanged). Clearly, one of the principal orientations in any design for BEPR is to minimize oxygen entrainment and nitrate recycling to the anaerobic reactor. To achieve this in situations where nitrification is obligatory or unaviodable, a number of different system configurations have been developed (Figure 47).
4.14.33.5 Subdivision of the Anaerobic Reactor into Compartments The effect of subdividing the anaerobic reactor into compartments is shown in Figures 60(a) (raw) and 60(b) (settled). Increasing the anaerobic reactor from a single completely mixed one to two compartments in series significantly improves the P removal, but increasing the number of compartments to greater than 3 yields little additional increase. This increase is due to the increased fermentable RBO conversion with in-series anaerobic reactor operation as a result of the first-order nature of the conversion kinetics. For design, at least two equal-sized in-series anaerobic reactors should be used.
4.14.34 Denitrification in NDBEPR Systems 4.14.34.1 Introduction Because usually N removal is also a requirement for BNR systems, nitrification is included and hence also denitrification to benefit from its advantages (Section 4.14.24). In the steady mixed culture BEPR model, the nitrate recycled to the anaerobic reactor needs to be known considering the adverse
Biological Nutrient Removal
0.25 0.20 0.15
0.03
0.02
0.10 0.05
22.5
15.0
Influent P
fxana
0.01
7.5
0.0
0.00
0.0 1
(a)
0.04
2
4 3 No anaerobic compartments
Settled WW
18.0
0.25 0.20
0.03
13.5 0.15 0.10
0.02
Influent P
9.0
0.05
fxana
0.01
4.5
0.0
0.00
5
0.0 1
(b)
P removal (mgP I−1)
30.0 Raw WW
P removal (mgP/mg Infl COD)
P removal (mgP/mg Infl COD)
0.04
P removal (mgP I−1)
518
2
4 3 No anaerobic compartments
5
Figure 60 Predicted P removal vs. the number of compartments in the anaerobic reactor for various anaerobic mass fractions (fxana) in a BEPR system at 20 days sludge age treating the example raw (a) and seattled (b) wastewater.
influence of recycling nitrate to the anaerobic reactor on P removal. Indeed, one of the principal orientations in the design procedure for P removal is to prevent nitrate recycling. Where nitrification is not required, this can be suppressed in a simple configuration such as the Phoredox (or the A/O) system (Figure 47(a)) but this option may not viable in some countries because nitrification is either obligatory or unavoidable due to high wastewater temperature. Accordingly, reliable and accurate quantification of denitrification in NDBEPR systems is essential not only for security in P removal but also for estimating the N removal by the NDBEPR system. The early approach (1976–85) to quantify denitrification in NDBEPR systems was to use the theory and procedures for ND systems, as set out in Sections 4.14.17–4.14.27 (Nicholls, 1982; Ekama et al., 1983; WRC, 1984). Experimental data indicated that this approach appeared to predict the observed denitrification quite closely. However, from the mechanisms for BEPR, which emerged later, this approach was theoretically inconsistent. The influent RBO was apparently used twice – first in the anaerobic reactor where it is converted to VFAs which are taken up and stored as PHA by the PAOs, and again in the primary anoxic reactor for denitrification via the K1 rate (Section 4.14.25.1). This situation is possible only if the PAOs denitrified significantly using most of their internally stored PHA with nitrate as electron acceptor in the anoxic reactor. This implies that the most of the P uptake should take place in the primary anoxic reactor. However, this was not usually observed. Practically, all (490%) the P uptake took place in the aerobic reactor of the UCT NDBEPR systems operated during the 1980s (Wentzel et al., 1985, 1990).
4.14.34.2 Experimental Basis for Denitrification Kinetics in NDBEPR Systems Clayton et al. (1991) undertook an experimental investigation into the denitrification kinetic behavior in mixed culture NDBEPR systems. A laboratory-scale modified UCT system (Figure 47(e)) was set up and operated for a year and a half. For the first 6 months, the first primary anoxic reactor was a plugflow reactor, thereafter a completely mixed one. The response of the system was monitored daily and profiles on the
plugflow primary anoxic reactor measured periodically. In addition, a variety of anoxic batch tests that reproduced the conditions in primary and secondary anoxic reactors were conducted on mixed liquor harvested from the MUCT system. In the plugflow reactor and batch tests, all the important parameters were measured to delineate the behavior of the OHOs and PAOs. No differences in the concentration time profiles from the plugflow reactor and batch tests were noted. From these tests: (1) Under the steady-state conditions of the MUCT system, the general denitrification formulation for ND systems dNO3/ dt ¼ KXBH applied also to NDBEPR systems. (2) In the primary anoxic reactor, (1) the rapid rate of denitrification associated with RBO in ND systems (K0 1, Figure 35; the K0 rate here for NDBEPR systems is used to distinguish it from the K rate in ND systems) was usually absent or of very short duration, (2) the slower rate of denitrification associated with BPO (K0 2) continued over the entire duration of the plugflow retention time or batch test (as in ND systems) but its rate was approximately 2 1/2 times faster than in the primary anoxic reactor of ND systems, i.e., 0.224 mgNO3-N/(mgOHOVSS d), where the OHOVSS concentration was calculated from the experimental system data with the ND model, not the BEPR model, that is, ignoring the reduction in OHOVSS due to the presence of the PAOs (Clayton et al., 1991). Based on the BEPR model, which yields lower OHOVSS due to the presence of PAOs, the K0 2 rate would be even higher compared with ND systems (Table 17). (3) In the secondary anoxic reactor, the denitrification rate (K0 3) was approximately 1 1/2 times the rate measured in secondary anoxic reactors of ND systems (K3, Figure 35(b)), also based on the ND model (Table 17). Clayton et al. (1991) proposed three possible explanations for the increased denitrification rate constant K0 2 observed in the primary anoxic zone of NDBEPR systems: 1. PAOs can denitrify, thereby contributing to the denitrification rate by utilizing the intracellular PHB acquired in the anaerobic zone. 2. PAOs cannot denitrify and the influent BPO is modified in the anaerobic zone to a more readily hydrolyzable form, thereby inducing a faster denitrification rate by the OHO in the primary anoxic reactor.
Biological Nutrient Removal
519
Table 17 Specific denitrification rates (K) at 20 1C observed by Clayton et al. (1991) in the MUCT NDBEPR system based on the WRC (1984) and Wentzel et al. (1990) models compared with those in ND systems (WRC, 1984) System
NDBEPR systems based on ND model
NDBEPR systems based on BEPR model
ND systems based on ND model
Units
mgNO3–N/(mgVSS d)
mgNO3–N/(mgAHVSS d)
mgNO3-N/(mgAHVSS d)
Primary anoxic
K10 ¼ 0.61a K20 ¼ 0.224 K30 ¼ 0.100
K10 ¼ 0.70a K20 ¼ 0.255 K30 ¼ 0.114
K1 ¼ 0.720 K2 ¼ 0.101b K3 ¼ 0.072
Secondary anoxic a
Denitrification by this rate contributes negligibly to the N removal of the NDBEPR system. In single-reactor intermittent aeration ND systems Warburton et al. (1991) obtained K2 ¼ 0.128 at 20 1C.
b
Denitrification in N removal systems Primary anoxic reactor
Aerobic reactor
Denitrification in BioP removal systems
Secondary anoxic Reaeration reactor reactor
Mixed liquor recycle a Influent
Waste flow Settler Effluent Sludge recycle s
Influent RBCOD + SBCOD
K1 + K2 K2
Influent SBCOD only
K3
Endogenous SBCOD only
Anaerobic Anoxic reactor reactors
Aerobic reactor
Mixed liquor recycle Influent
Secondary anoxic reactor Waste flow Settler
a
r
Effluent RBCOD acidified and taken up by PAOs No initial rapid K1 rate
Sludge recycle s
K 2′
Influent SBCOD only
K 3′
Endogenous SBCOD only
Figure 61 Comparison of steady-state specific denitrification rates (K, mgNO3-N/(mgOHOVSS d)) in the primary and secondary anoxic reactors of ND (a) and NDBEPR (b) systems. The rates are compared in Table 15.
3. PAOs cannot denitrify and the BPO is not modified in the anaerobic zone but a higher rate of BPO hydrolysis/utilization is stimulated in the OHOs in NDBEPR systems by the anaerobic–anoxic–aerobic sequencing. The PHB concentrations measured in the (1) anaerobic, anoxic, and aerobic zones of the MUCT parent system; (2) anoxic batch tests on mixed liquor harvested from the MUCT system; and (3) anoxic batch tests on mixed liquor harvested from the enhanced PAO cultures of Wentzel et al. (1988, 1989a) demonstrated that PHB did not serve as a substrate source for denitrification (negligible decrease). Therefore, the PAOs did not contribute significantly to the K’2 denitrification rate in the primary anoxic reactor and so cause (1) had to be rejected. This conclusion was supported from the observation that in the mixed and enhanced culture systems and the batch tests, the P uptake was predominantly (490%) aerobic – negligible anoxic P uptake was observed. If the anaerobic reactor pretreats the influent BPO to a more readily utilizable form, then the K denitrification rates should be lower when the NDBEPR sludge is mixed with influent wastewater. Batch tests on sludge from the MUCT system fed the same wastewater as the parent system, yielded the same high K0 2 denitrification rates. Therefore, the anaerobic reactor did not modify the BPO to a more utilizable form. So cause (2) was rejected and default cause (3) had to be accepted. No experimental means was devised to test this third
cause. However, it did at least provide a consistent explanation also for the higher K0 3 in the secondary anoxic reactor – causes (1) and (2) explain only a higher K0 2 rate. A comparison between the denitrification rates in the ND and NDBEPR systems is shown in Figure 61. Because the PAOs did not significantly contribute to the denitrification, the K0 rates had to be recalculated so that the denitrification process in NDBEPR systems is correctly ascribed to the OHO group performing it. The proportion of OHOs in the VSS of NDBEPR systems is smaller than in ND systems (Figure 55) because the PAOs obtain most of the influent RBO. Ekama and Wentzel (1999a) calculated the OHO fraction (favOHO) for NDBEPR systems iteratively with the aid of the steady-state BEPR model (Section 4.14.31) using the measured value for the influent RBO fraction and varying the influent UPOs fraction (UPOCOD, fS’up) until the calculated system VSS mass, now comprising active and endogenous PAO and OHO components and unbiodegradable particulate VSS from the influent, matched that measured. When the correct fS’up had been found, the calculated P removal was matched to that measured by changing the PAO P content (fXBGP) from the enhanced PAO culture value of 0.38 mgP/mgPAOVSS. For MUCT system of Clayton et al. (1991), the recalculated average VSS fractionation results are fS’up ¼ 0.15, favOHO ¼ 0.21, and fXBGP ¼ 0.388 mgP/mgPAOVSS. Because the favOHO based on ND model (Section 4.14.9.4) was 0.24, on average the K0 rates were 0.24/0.21 ¼1.14 or 14%
520
Biological Nutrient Removal
higher (Table 17). Following this calculation procedure, Ekama and Wentzel (1999a) also calculated the fS’up, favOHO, fXBGP, and K0 2 for four other UCT investigations, viz., Musvoto et al. (1994), Pilson et al. (1995), Sneyders et al. (1997), and Mellin et al. (1997). For the same wastewater source, reasonably consistent fS’up values are expected. For fully aerobic and ND systems, this has been the case in the UCT laboratory. For the Mitchells Plain unsettled wastewater usually fed to the experimental systems, this value was found to be around 0.12 for widely differing aerobic and ND systems, e.g., 0.10870.052 for aerobic systems (Mbewe et al., 1994), and 0.13570.060 (Warburton et al., 1991) and 0.1270.04 (Ubisi et al., 1997a, b) for anoxicaerobic systems. However, for NDBEPR systems, this was not the case. Not only was fS’up higher for NDBEPR systems fed the Mitchells Plain unsettled wastewater, it also varied widely in the different NDBEPR systems, from 0.0470.055 (Sneyders et al., 1997) to 0.29370.063 (Musvoto et al., 1994). Because of the method calculating the fS’up fraction, by reconciling the calculated VSS mass with the measured VSS mass, the variation in fS’up changes the favOHO. This, in turn, affects the K0 2 and K0 3, rates, which are higher for higher fS’up and lower for lower fS’up (Ekama and Wentzel, 1999b). Clearly, there are factors that affect the sludge production per unit COD load in the NDBEPR system that the models do not recognize. Two such factors appear to be the unaerated sludge mass fraction (fxt) and sludge settleability (measured as diluted sludge volume index, DSVI). The higher the fxt, the higher the fS’up, which could be due to an accumulation of undegraded BPO in the system. If this were the only factor, then the method of calculating fS’up and favOHO would be acceptable because undegraded BPO in effect is unbiodegradable particulate organics. However, this is not the only factor because systems with the same fxt yielded different fS’up and favOHO values depending on the DSVI (Musvoto et al., 1994; Casey et al., 1994a, 1994b). As the DSVI and hence AA (low F/M) filament abundance increased, so the system VSS mass decreased and vice versa. The calculated K0 rates varied accordingly, decreasing as the system VSS mass increased and vice versa. No explanation for this variation with DSVI can be advanced. The NDBEPR models, both steady-state (e.g., Wentzel et al., 1990) and dynamic simulation (e.g., ASM2, Henze et al., 1995), are extensions of their predecessors (WRC, 1984, ASM1; Henze et al., 1987) by including the kinetics of BEPR. Relatively few interactions between the ND and BEPR processes take place in these models, the main ones being that (1) the VFAs for the PAOs are generated by the OHOs in the anaerobic zone from the influent RBO, and more importantly, (2) the reduction factor for the BPO hydrolysis/utilization rate, Z, is increased from 0.33 in ASM1 (and UCTOLD, Dold et al., 1991) to 0.60 in ASM2 (and UCTPHO, Wentzel et al., 1992) to account for the increased K0 2 and K0 3 rates observed by Clayton et al. (1991). Insofar as the BEPR kinetics in the ASM2 and UCTPHO models are concerned, P release and uptake occur exclusively in the anaerobic and aerobic reactors respectively, in conformity with the observations of Siebritz et al. (1983), Wentzel et al. (1985, 1989b, 1990), Clayton et al. (1991), and Sneyders et al. (1997). Therefore, for the last two mentioned investigations, given the correct input fS’up and Z values, the NDBEPR models will satisfactorily predict the
performance of the M/UCT systems. However, in three other investigations, viz., Musvoto et al. (1994), Pilson et al. (1995), and Mellin et al. (1997), the P release, P uptake, and P removal behavior were significantly different to that observed on which the models are based. Not only was the excess P removal depressed at about 60% of that expected from the model of Wentzel et al. (1990), but also the P release to removal ratio was decreased. With the depressed P removal, significant P uptake took place in the (second) anoxic reactors of the MUCT systems. This was confirmed in the anoxic batch tests; whereas in the tests of Clayton et al. (1991) and Sneyders et al. (1997) negligible anoxic P uptake took place, in those of Mellin et al. (1997) significant (B40%) P uptake took place. The significant decrease in BEPR with anoxic P uptake BEPR was subsequently confirmed by Vermande et al. (2002) in parallel aerobic P uptake BEPR and anoxic P uptake BEPR systems. It is possible that different species of PAOs find a niche in the systems that can accomplish anoxic P uptake, but which have lower RBCOD to P release, P release to P removal, and fXBGP ratios. Biochemical assays have indicated that some PAOs can denitrify (Lo¨tter, 1985; Lo¨tter et al., 1986) and even anaerobic–anoxic (no aerobic) BEPR systems have been operated successfully (Kuba et al., 1993). Also, in several other studies significant anoxic P uptake has been observed (e.g., Vlekke et al., 1988; Kerrn-Jespersen and Henze, 1993; Bortone et al., 1996; Kuba et al., 1996; Kuba and van Loosdrecht, 1996; Hu et al., 2000, 2007a, 2007b). Denitrification by PAOs is included in the biochemical model of Wentzel et al. (1986, 1991) but is not included in ASM2 (Henze et al., 1995) and UCTPHO (Wentzel et al., 1992) kinetic models. Anoxic P uptake behavior of PAOs has been included in ASM2d (Henze et al., 1999) but it merely allows the P uptake to commence in the anoxic reactor without changing the P uptake, that is, anoxic P uptake is modeled with the same stoichiometry and kinetics as aerobic P uptake, which clearly is not observed experimentally. Realistic anoxic P uptake behavior of PAOs therefore cannot be simulated with current suite of IWA ASM models. Proposals to include PAO denitrification have been made (e.g., Mino et al., 1995; Barker and Dold, 1997; Hu et al., 2007a, 2007b), with varying success. One of the main problems with modeling anoxic P-uptake BEPR is that the triggers that stimulate it are not well understood. Hu et al. (2002) conclude that anoxic P-uptake BEPR is undesirable in NDBEPR systems due to the reduction in P removal per influent RBO it causes. For maximum BEPR with (usually) limited influent RBO, aerobic P-uptake BEPR is required. Large aerobic mass fractions (fxto0.5) and underloaded primary anoxic reactors with nitrate appear to favor aerobic P-uptake BEPR.
4.14.34.3 Denitrification Potential in NDBEPR Systems The denitrification potential is the maximum amount of nitrate per liter influent flow that can be removed by biological means in the anoxic reactors. As the experimental investigation into denitrification kinetics in NDBEPR systems indicated that the formulation
dNO3 =dt ¼ KXBH
Biological Nutrient Removal
developed for ND systems can also be applied to NDBEPR systems, the techniques set out in Section 4.14.25.2 to develop equations for denitrification potential in ND systems can be followed for NDBEPR systems also. For development of these equations, the experimental observation that the PAOs do not denitrify is accepted. Denitrification in the primary anoxic reactor is via utilization of any RBO leaking through the anaerobic reactor, and BPO. Procedures to determine the amount of RBO leaking through the anaerobic reactor to the primary anoxic reactor were set out in Section 4.14.31.1.4, where SbsfN is the concentration of FRBO exiting the last anaerobic compartment. Hence, SbsfN (1 þ recycle ratio) is the concentration FRBO per liter influent flow exiting the anaerobic reactor and available for denitrification in the primary anoxic reactor by OHOs. Accordingly, the denitrification potential in the primary anoxic reactor (Dp1) can be expressed as
Dp1 ¼ SbsfN ð1 þ rÞð1 f cv YH Þ=2:86 þ K02 XBH Rnp
ð205Þ
Following the procedures set out in Section 4.14.26.3, Equation (205) can be modified and simplified to give
Dp1 ¼ SbsfN ð1 þ rÞð1 f cv YH Þ=2:86 þ f x1 K02T ðSbOHO ÞYH Rs =ð1 þ bHT Rs Þ ðmg NO3 -N l1 influentÞ ¼ a0 þ f x1 K02T b0
ð206Þ
where fx1 is the primary anoxic sludge mass fraction and a0 ¼ SbsfN ð1 þ rÞð1 f cv YH Þ=2:86 and b0 ¼ ðSbOHO ÞY H Rs = ð1 þ bHT Rs Þ. In Equation (206), it is assumed that the initial rapid rate of denitrification (K0 1T) on FRBO leaking through the anaerobic reactor, SbsfN(1 þ r) is always complete, that is, the actual retention time in the primary anoxic reactor is longer than the time required to utilize this usually low concentration FRBO. As with ND systems, an equation can be developed to determine the minimum primary anoxic mass fraction f’x1min to deplete this RBO. This minimum will be a very low value (o0.05) which is much smaller than the primary anoxic reactors in NDBEPR systems, so generally Equation (206) is valid. However, Equation (206) is not without complication. To calculate the primary anoxic denitrification potential (Dp1), the concentration of RBO in the outflow from the anaerobic reactor (SbsfN) is required. To calculate SbsfN, the concentration of nitrate recycled to the anaerobic reactor is required which in turn requires Dp1 to be known. This problem is overcome by assuming initially that the nitrate concentration exiting the primary anoxic reactor is zero, as was done for the primary anoxic reactor of the MLE system in Section 4.14.25.2, which for the UCT systems means zero nitrate discharge to the anaerobic reactor. For brevity, other NDBEPR configurations are not considered in this chapter. However, if required, the denitrification potential of the secondary anoxic reactor is found using the principles set out in Section 4.14.25.2, viz.,
Dp3 ¼ f x3 K03T ðSbOHO Þ YH Rs =ð1 þ bHT Rs Þ ¼ f x3 K03T b0
ðmgNO3 -N l1 influentÞ
where fx3 is the secondary anoxic sludge mass fraction.
ð207Þ
521
Equation (207) applies to secondary anoxic reactors situated both in the mainstream (e.g., five-stage Bardenpho) and in the underflow recycle (e.g., JHB system). However, in applying Equation (207) to secondary anoxic reactors situated in the underflow recycle, care must be taken in evaluating fx3, because the mixed liquor concentration is increased by a factor (1 þ s)/s in the underflow anoxic reactor compared with the mainstream reactors.
4.14.34.4 Principles of Denitrification Design Procedures for NDBEPR Systems In NDBEPR systems design is oriented to achieve in a single sludge system COD removal, N removal (ND), and P removal (BEPR). Conflict between the last two objectives may arise, for example, the proportion of the total unaerated sludge mass assigned to the anoxic reactor(s) (N removal) and the anaerobic reactor (P removal). For each design, the priorities for treatment need to be assessed and a compromise reached to optimize the system for the particular effluent quality required. Because P is the element that is the main driver for eutrophication, for most designs of NDBEPR systems the focus is on BEPR with denitrification as a secondary design priority. Accordingly, the principle in denitrification design for NDBEPR systems is to ensure that the anaerobic reactor is protected from recycling of nitrate, which causes a disproportionate decrease in the magnitude of P removal (Figure 59). This principle guides selection of the system configuration (five-stage modified Bardenpho, JHB, and M/ UCT; Figure 47) and provides a starting point for sizing the anoxic reactors. When selecting a system configuration for BEPR, it is necessary to establish whether complete denitrification can be achieved. For the wastewater characteristics (i.e., influent TKN and COD concentrations (Nti and Sti)), maximum specific growth rate of the nitrifiers at 20 1C (mAm20), and the average minimum water temperature, the maximum unaerated sludge mass fraction (fxm) and the nitrification capacity (Nc) can be calculated for a selected sludge age (Rs) (Section 4.14.20.3). This fxm needs to be divided between anaerobic (for BEPR) and anoxic (for denitrification) mass fractions. Consequently, the maximum anoxic sludge mass fraction (fxdm) is the difference between the maximum unaerated mass fraction (fxm) and the selected anaerobic sludge mass fraction (fxa), that is,
f xdm ¼ f xm f xa
ð208Þ
where fxm is given by Equation (136) for a selected Rs, mAm20, Sf, and Tmin. The fxdm then can be subdivided between primary and secondary anoxic sludge mass fractions (fx1 and fx3) and this division fixes the denitrification potential of these two reactors (Dp1 and Dp3) and hence also of the system. If the denitrification potential of the system exceeds the nitrification capacity (i.e., Dp1 þ Dp34Nc), then complete denitrification is possible and the secondary anoxic reactor can be situated in the mainstream, that is, a five-stage Bardenpho system can be selected. If complete denitrification is not possible, then depending on the magnitude of the effluent nitrate
522
Biological Nutrient Removal
concentration, the underflow (s) recycle cannot be discharged directly to the anaerobic reactor. If the nitrate concentration is low (o3 mgN l1), the secondary anoxic sludge mass fraction (fx3) can be combined with the primary anoxic sludge mass fraction to form a three-stage Bardenpho, which, with a higher a recycle ratio, may produce a lower nitrate concentration in the sludge underflow (and effluent) than the fivestage Bardenpho because the K0 2 rate is higher than the K0 3 (Table 17). If the nitrate concentration is high (43 mgN l1), the secondary anoxic reactor can be moved into underflow recycle to form the JHB system, in which event the denitrification potential of the secondary anoxic reactor (Dp3) must exceed the nitrate and oxygen loads via the underflow s recycle. If this requirement is not met, nitrate will leak through the underflow secondary anoxic reactor to the anaerobic reactor. In this event, since the denitrification potential of the primary anoxic reactor (Dp1) is greater than that of the secondary anoxic reactor (Dp3) for equal anoxic mass fractions, incorporation of a secondary anoxic reactor becomes an inefficient utilization of anoxic mass fraction, and the secondary anoxic mass fraction is added to the primary anoxic reactor, the underflow s recycle needs to be denitrified in the primary anoxic reactor to form the M/UCT system. With each change of configuration, more nitrogen removal is sacrificed (i.e., the effluent nitrate concentration increases) to protect maximum BEPR, that is, zero or very low nitrate discharge to the anaerobic reactor.
4.14.34.5 Analysis of Denitrification in NDBEPR Systems Analysis of the denitrification behavior in the NDBEPR system is essentially the same as for the ND system (Section 4.14.26.3) except: (1) The maximum anoxic mass fraction for denitrification (fxdm) for the NDBEPR system is given by Equation (208), whereas fxdm for the ND system is given by Equation (166). Hence, for the same maximum unaerated sludge mass fraction (fxm), the NDBEPR system has a lower fxdm than the ND system, by an amount equal to fxa. (2) The specific denitrification rates for ND systems (K2 and K3) are substituted with the rates for NDBEPR systems (K0 2 and K0 3, Table 17). (3) The denitrification potentials for the primary and secondary anoxic reactors are modified from Equations (163) and (164) for ND systems to Equations (206) and (207) for the NDBEPR system to take account of the uptake of COD by the PAOs in the anaerobic reactor, the zero denitrification by the PAOs, and the faster OHO denitrification rates in NDBEPR systems. Taking account of the above, denitrification equations can be developed for all the NDBEPR configurations (Figure 47). However, in the interests of brevity, only the UCT configuration will be considered.
4.14.35 Denitrification in the UCT System In the UCT system the denitrification behavior is very similar to that in the MLE system, because the a and s recycles discharge into the primary anoxic reactor, so that taking due account of the effect of incorporating the anaerobic reactor, the design equations and procedures developed for the MLE system can be applied to the UCT system. Since complete
denitrification is not possible in the UCT system (high effluent nitrate concentration), the entire anoxic mass fraction (fxdm) available is used as a primary anoxic reactor (fx1). As in the MLE system, the a and s recycle ratios determine the split of the nitrate generated in the aerobic reactor (nitrification capacity, Nc) between the primary anoxic reactor and the effluent. The a recycle ratio is selected so that the equivalent nitrate load on the primary anoxic reactor via the a and s recycles is equal to its denitrification potential (Dp1). For a selected s recycle ratio, the a recycle ratio that loads the primary anoxic reactor via to its denitrification potential is the optimum a recycle ratio (aopt). The denitrification potential of the (primary) anoxic reactor (Dp1) is found from Equation (206) with fx1 ¼ fxdm. Following the same reasoning as in Section 4.14.26.3, the optimum a recycle ratio (aopt) is given by Equation (169), with the proviso that the Nc and Dp1 are applicable to NDBEPR systems, that is, Nc is lower due to the higher sludge production (Section 4.14.31.5.2) and Dp1 is based on Equation (206) with K0 2. As for the MLE system, at a ¼ aopt, Equation (170) gives the minimum effluent nitrate concentration (Nne) achievable. Equation (170) is valid for all aoaopt because for all aoaopt the assumption on which Equation (169) is based is valid, that is, zero nitrate concentration exiting the primary anoxic reactor. If the system is operated with a4aopt, the equivalent nitrate load on the primary anoxic reactor via the a and s recycles exceeds the denitrification potential and nitrate will also be recycled via the r recycle to the anaerobic reactor, to the detriment of BEPR. Furthermore, if nitrate does leak through the primary anoxic reactor, then the nitrate concentration in the outflow from the primary anoxic reactor no longer is zero, and consequently, Equation (170) for the effluent nitrate concentration (Nne) is no longer valid. Equations for the effluent nitrate concentration for a4aopt can be derived by following the principles applied above for aoaopt, but are not considered because aoaopt is required for zero discharge of nitrate to the anaerobic reactor and maximum BEPR. If Equation (169) yields aopt ¼ 0, then the equivalent nitrate load via the s recycle is sufficient to match the denitrification potential of the primary anoxic reactor; if aopto0, the equivalent nitrate load via the s recycle exceeds Dp1 and nitrate will be recycled via the r recycle to the anaerobic reactor. The implication of this is that the Nc that gives aopt ¼ 0 represents the upper limit (equivalently the maximum influent TKN/ COD concentration ratio) that the UCT system is able to treat and still protect the anaerobic reactor against nitrate entry. All Nc (equivalently influent TKN/COD ratios) above this limit will result in nitrate recycle to the anaerobic reactor, which cannot be controlled in the UCT system (a ¼ 0) except by reducing the s recycle ratio. From the above, the minimum a recycle ratio is a ¼ 0. The maximum a recycle ratio (amax) is determined by some practical upper limit (aprac), usually in the range 5–6, beyond which the higher pumping costs outweigh the small gain in lower effluent nitrate concentration (see Section 4.14.26.3). However, for oxidation ditch type systems, or for systems with ‘‘through the wall’’ a recycles via low head high volume pumps, the a recycle ratio ( ¼ aopt) may be significantly higher than the aprac of 5–6. If aopt 4 aprac and aprac is selected, then the
Biological Nutrient Removal
primary anoxic reactor is oversized. This unused denitrification potential (Figure 38) can be kept (i.e., fx1 not decreased) as a factor of safety (for uncertainty if K0 rate) or the size of the fx1 of primary anoxic reactor reduced to match the its denitrification potential (Dp1) to the equivalent nitrate load, as was done for the balanced MLE system (Section 4.14.26.3.2), which will allow a reduction in sludge age. The procedure for the balance MLE sytem can be followed to determine the new sludge age. All the aspects discussed in Sections 4.14.11 and 4.14.14– 4.14.16 regarding reactor concentration selection, system design and control, selection of sludge age, and treatment of the primary and/or secondary sludge produced also apply to NDBEPR systems and should be referred to there.
4.14.36 Conclusion The ND and NDBEPR models such as IWA ASM1 and 2 (Henze et al., 1987, 1995), UCTOLD (Dold et al., 1991), and UCTPHO (Wentzel et al., 1992) are very helpful for biological nutrient removal process description and simulation. However, models always need to be used with great circumspection and experience of real systems. It would appear that the ND models such as ASM1 and UCTOLD give an acceptably reliable description of the ND AS systems – the model predictions compare favorably with observed results and the wastewater characteristic, stoichiometric, and kinetic constants in the models to achieve this are reasonably consistent. For these models some scientific maturity is apparent, where the default kinetic and stoichiometric constants predict the performance of an ND system with acceptable risk of deviation. For the NDBEPR models, this is not the case. The experiments described in the literature point to three important observations in real NDBEPR systems not recognized in NDBEPR models that model users need to be aware of for prudent and proper application: that is, (1) the large variation in the unbiodegradable particulate COD fraction (fS’up) and hence the OHO active fraction (favOHO) and denitrification rate (K0 2); (2) the large variation in biological P removal behavior and P content of PAOs (fXBGP) with anoxic P-uptake BEPR stimulated in some systems for reasons not well defined yet; and (3) the unaccounted loss of influent COD in NDBEPR systems, in that even in carefully controlled laboratory systems, only 75–85% of the influent COD can be recovered in a COD mass balance (Ekama et al., 1999a,b).
References Alexander WV, Ekama GA, and Marais GvR (1980) The activated suldge process Part 2 – Application of the general kinetic model to the contact stabilization process. Water Research 14: 1737--1747. Antoniou P, Hamilton J, Koopman B, et al. (1990) Effect of temperature and pH on the effective maximum specific growth rate of nitrifying bacteria. Water Research 24(1): 97--101. Arvin E (1985) Biological removal of phosphorus from wastewater. Critical Reviews in Environmental Control 15: 25--64. Barat R and van Loosdrecht MCM (2006) Potential phosphorus recovery in a WWTP with the BCFSs process: Interactions with the biological process. Water Research 40(19): 3507--3516.
523
Barker PS and Dold PL (1997) General model for biological nutrient removal activated – sludge systems: Model presentation. Water Environment Research 69(5): 969--984. Barnard JL (1973) Biological denitrification. Water Pollution Control 72(6): 705--720. Barnard JL (1974) Cut P and N without chemicals. Water and Wastes Engineering 11: 33--36 (41-44??). Barnard JL (1975a) Biological nutrient removal without the addition of chemicals. Water Research 9: 485--490. Barnard JL (1975b) Nutrient removal in biological systems. Water Pollution Control 74(2): 143--154. Barnard JL (1976) A review of biological phosphorus removal in the activated sludge process. Water SA 2: 136--144. Barnard JL (1984) Activated primary tanks for phosphate removal. Water SA 10(3): 121--126. Batstone DJ, Keller J, Angelidaki I, et al. (2002) Anaerobic Digestion Model No 1 (ADM1), Scientific and Technical Report No 9. London: International Water Association (IWA). Blackall LL, Crocetti G, Saunders AM, and Bond PL (2002) A review and update of the microbiology of enhanced biological phosphorus removal in wastewater treatment plants. Antonie Van Leeuwenhoek International Journal of General and Molecular Microbiology 81(1–4): 681--691. Bortone G, Saltarelli R, Alonso V, Sorm R, Wanner J, and Tilche A (1996) Biological anoxic phosphorus removal – the DEPHANOX process. Water Science and Technology 34(1/2): 119--128. Bratby J (1978) Aspects of sludge thickening by dissolved-air flotation. Water Pollution Control 77(3): 421--432. Bratby J, Walker S, Barnes C, and Jones G (2008) Float on and on – taking flotation thickeners to the limits of performance. Water Environment and Technology 30(7): 56--59. Brdjanovic D, Hooijmans CM, van Loosdrecht MCM, Alaerts GJ, and Heijnen JJ (1996) The dynamic effects of potassium limitation on biological phosphorus removal. Water Research 30(10): 2323--2328. Brodisch KEU (1985) Interaction of different groups of microorganisms in biological phosphate removal. Water Science and Technology 17(11/12): 89--97. Brouckaert CJ, Ikumi DS and Ekama GA (2010) A 3 phase anaerobic digestion model. Proceedings of the 12th IWA AD conference, Guadalajara, Mexico, 1–4 November. Buchan L (1983) Possible biological mechanism of phosphorus removal. Water Science and Technology 15(3/4): 87--103. Burke RA, Wentzel MC, Dold PL, Ekama GA, and Marais GvR (1989) Biological excess phosphorus removal in short sludge age activated sludge systems. In: Proceedings of the 1st Biennial WISA Conference. Cape Town, March Water Institute of Southern Africa, PO Box 6011, Halfway House, 1685, South Africa. Casey TG, Wentzel MC, Ekama GA, Loewenthal RE, and Marais GvR (1994a) An hypothesis for the cause and control of anoxic-aerobic (AA) filament bulking in nutrient removal activated sludge systems. Water Science and Technology 29(7): 203--212. Casey TG, Wentzel MC, Ekama GA, and Marais GvR (1994b) Causes and control of anoxic-aerobic (AA) (or low F/M) filament bulking in long sludge age nutrient removal activated sludge systems. WRC Report No 286/2/94. Pretoria: Water Research Commission. Chao AC and Keinath TM (1979) Influence of process loading intensity on sludge clarification and settling characteristics. Water Research 13(12): 1213--1224. Clayton JA, Ekama GA, Wentzel MC, and Marais GvR (1989) Denitrification kinetics in biological nitrogen and phosphorus removal activated-sludge systems. Research Report W63, Dept. of Civil Eng., Univ of Cape Town, Rondebosch, 7701, South Africa. Clayton JA, Ekama GA, Wentzel MC, and Marais GvR (1991) Denitrification kinetics in biological nitrogen and phosphorus removal activated-sludge systems treating municipal wastewaters. Water Science and Technology 23(4–6): 1025--1035. Comeau Y, Hall KJ, Hancock REW, and Oldham WK (1986) Biochemical model for biological enhanced phosphorus removal. Water Research 20: 1511--1521. Comeau Y, Rabinowitz B, Hall KJ, and Oldham WK (1987) Phosphorus release and uptake in enhanced biological phosphorus removal from wastewater. Journal of the Water Pollution Control Federation 59: 707--715. Crocetti GR, Hugenholtz P, Bond PL, Schuler A, Keller J, and Jenkins D (2000) Identification of polyphosphate-accumulating organisms and design of 16S rRNAdirected probes for their detection and quantitation. Applied and Environmental Microbiology 66(3): 1175--1182. Daigger GT, Randall CW, Waltrip GD, Romm ED, and Morales LM (1987) Factors affecting biological phosphorus removal for the VIP process, a high-rate University of Cape Town process. In: Proceedings of the IAWPRC Biological Phosphate
524
Biological Nutrient Removal
Removal from Wastewater Seminar, Rome. London: International Water Association (IWA) Publishing. de Haas DW, Wentzel MC, and Ekama GA (2000) The use of simultaneous chemical precipitation in modified activated sludge systems exhibiting biological excess phosphate removal Part 1: Literature review. Water SA 26(4): 439--452. de Haas DW, Wentzel MC, and Ekama GA (2001) The use of simultaneous precipitation in activated sludge systems exhibiting biological excess phosphate removal. Part 5: Experimental periods using ferrous-ferric chloride blend. Water SA 27(2): 117--134. Dick RI (1976) Folklore in the design of final settling tanks. Journal of the Water Pollution Control Federation 48(4): 633--644. Dold PL, Bagg WK, and Marais GvR (1986) Measurement of the readily biodegradable COD fraction (Sbs) in municipal wastewater by ultrafiltration. Research Report No. W57, Rondebosch: University of Cape Town. Dold PL, Buhr HO, and Marais GvR (1982) Design and control of equalization tanks. Research Report No. W 42, Rondebosch: Department of Civil Engineering, University of Cape Town. Dold PL, Buhr HO, and Marais GvR (1984) An equalization control strategy for activated sludge process control. Water Science and Technology 17(Amsterdam): 221--234. Dold PL, Ekama GA, and Marais GvR (1980) A general model for the activated sludge process. Progress in Water Technology 12(Toronto): 47--77. Dold PL, Ekama GA, and Marais GvR (1985) pH control and cost savings in aerobic digestion. In: Drake RAR (ed.) Proceedings of the IAWPRC Specialized Seminar on Wastewater Systems Control. Houston, April 1985. Advances in Water Pollution Control: Instrumentation and Control of Water and Wastewater and Transport Systems, pp. 375–382. Oxford: Pergamon. Dold PL and Marais GvR (1986) Evaluation of the general activated sludge model proposed by the IAWPRC Task Group. Presented at the IAWPRC Specialized Seminar on Modelling of Biological Wastewater Treatment, Copenhagen, August 1985. Water Sciemce Technology, Vol. 18, pp. 63–89. Dold PL, Wentzel MC, Billing AE, and Marais GvR (1991) Activated sludge system simulation programs: Nitrification and nitrification/denitrification systems, Version 1. WRC Report No. TT 52/91. Pretoria, S.A. Downing AL, Painter HA, and Knowles G (1964) Nitrification in the activated sludge process. Journal and Proceedings of the Institute of Sewage Purification 64(2): 130--153. Ekama GA (2009) Using bioprocess stoiciometry to build a steady state plant wide wastewater treatment plant model. Water Research 43(8): 2101--2120. Ekama GA, Barnard JL, Gu¨nthert FW, et al. (1997) Secondary Settling Tanks: Theory, Design, Modelling and Operation. IAWQ STR No 6, pp. 216. London: International Association on Water Quality. Ekama GA, Casey TG, Wentzel MC, and Marais GVR (1996b) Filamentous organism bulking in nutrient removal activated sludge systems. Paper 6: Review, evaluation and consolidation of results. Water SA 22(2): 147--152. Ekama GA, Dold PL, and Marais GvR (1986) Procedures for determining influent COD fractions and the maximum specific growth rate of heterotrophs in activated sludge systems. Water Science and Technology 18(6): 91--114. Ekama GA and Marais GvR (1986) Sludge settleability and secondary settling tank design. Water Pollution Control 85(2): 100--113. Ekama GA and Marais P (2004) Assessing the applicability of the 1D flux theory to full scale secondary settling tank design with a 2D hydrodynamic model. Water Research 38(3): 495--506. Ekama GA, Siebritz IP, and Marais GvR (1983) Considerations in the process design of nutrient removal activated sludge processes. Water Science and Technology 15(3/4): 283--318. Ekama GA, So¨temann SW, and Wentel MC (2006b) Mass balances based plant wide wastewater treatment plant models – Part 3: Biodegradability of activated sludge organics under anaerobic conditions. Water SA 32(3): 287--296. Ekama GA and Wentzel MC (1999a) Denitrification kinetics in biological N & P removal activated sludge systems treating municipal wastewaters. Water Science and Technology 39(6): 69--77. Ekama GA and Wentzel MC (1999b) Difficulties and developments in biological nutrient removal technology and modelling. Water Science and Technology 39(6): 1--11. Ekama GA and Wentzel MC (2004) A predictive model for the reactor inorganic suspended solids concentration in activated sludge systems. Water Research 38(19): 4093--4106. Ekama GA, Wentzel MC, and So¨temann SW (2006a) Mass balance-based plant-wide wastewater treatment plant models – Part 2: Tracking the influent inorganic suspended solids. Water SA 32(3): 277--286. Ekama GA, Wentzel MC, Casey TG, and Marais GvR (1996a) Filamentous organism bulking in nutrient removal activated sludge systems. Paper 3: Stimulation of the selector effect under anoxic conditions. Water SA 22(2): 119--126.
Fuhs GW and Chen M (1975) Microbiological basis of phosphate removal in the activated sludge process for the treatment of wastewater. Microbial Ecology 2: 119--138. Fukase T, Shibata M, and Miyaji Y (1982) Studies on the mechanism of biological removal of phosphorus. Japanese Journal of Water Pollution and Research 5: 309--317. Gabb DMD, Still DA, Ekama GA, Jenkins D, and Marais GvR (1991) The selector effect on filamentous bulking in long sludge age activated sludge systems. Water Science and Technology 23(4/6): 867--877. Garrett MT (1958) Hydraulic control of activated sludge growth rate. Sewage and Industrial Wastes 30: 253. Gerber A, Devilliers RH, Mostert ES, and Winter CT (1987) Interactions between phosphate, nitrate and organic substrate in biological nutrient removal processes. Water Science and Technology 19(1–2): 183--194. Gleisberg D (1993) Reduced P-Input from detergents. In: 9th EWPCA-ISWA Symposium, Munich, pp. 23–45. European Water Pollution Control Association, now European Water Association, Hennep, D53773, Germany. Gujer W and Jenkins D (1975a) The contact stabilization activated sludge process – oxygen utilization, sludge production and efficiency. Water Research 9(5/6): 553--560. Gujer W and Jenkins D (1975b) A nitrification model for the contact stabilization activated sludge process. Water Research 9(5/6): 561--566. Harding TH, Ikumi DS, and Ekama GA (2009) A steady state stoichiometric model describing the anaerobic digestion of biological excess phosphorus removal waste activated sludge. MSc Thesis, Research Report No. W132. Rondebosch: Department of Civil Engineering, University of Cape Town. Hart MA and Melmed LN (1982) Microbiology of nutrient removing activated sludge. Poster pres. Special in Seminar of the IAWPRC, Pretoria, SA. Hascoe¨t MC and Florentz M (1985b) Influence of nitrates on biological phosphorus removal from wastewater. Water SA 11: 1--8. Hascoe¨t MC, Florentz M, and Granger P (1985a) Biochemical aspects of enhanced biological phosphorus removal from wastewater. Water Science and Technology 17(11/12): 23--41. Henze M (1992) Characterization of wastewater for modelling of activated sludge processes. Water Science and Technology 25(6): 1--15. Henze M, Dupont R, Grau P, and De La Sola A (1993) Rising sludge in secondary settlers due to denitrification. Water Research 27(2): 231--236. Henze M, Grady CPL, Gujer W, Marais GvR, and Matsuo T (1987) Activated Sludge Model No.1. IAWPRC Scientific and Technical Reports No. 1, ISSN 1010-707X. 33pp. London: International Association on Water Pollution Research and Control, IAWPRC. Henze M, Gujer W, Mino T, Matsuo T, Wentzel MC, and Marais GvR (1995) Activated Sludge Model No. 2, IAWQ Scientific and Technical Report No. 3, ISSN 1025-0913, 32pp. London: IAWQ. Henze M, Gujer W, Mino T, et al. (1999) Activated sludge Model No. 2d, ASM2d. Water Science and Technology 39(1): 165--182. Henze M, Kristensen GH, and Strube R (1994) Rate-capacity characterization of wastewater for nurient removal precesses. Water Science and Technology 29(7): 101--107. Henze M, van Loosdrecht MCM, Ekama GA, and Brdjanovic D (2008) Biological Wastewater Treatment–Principles, Modelling and Design. London: IWA Publishing. ISBN13 9781843391883. Hesselmann RPX, Werlen C, Hahn D, van der Meer JR, and Zehnder AJB (1999) Enrichment, phylogenetic analysis and detection of a bacterium that performs enhanced biological phosphate removal in activated sludge. Systematic and Applied Microbiology 22(3): 454--465. Ho¨rler A (1969) Discussion of ‘‘Performance of (activated sludge) secondary settling tanks’’ by Pflanz P (1969) In: Jenkins SH (ed.) Proceedings of the 4th IAWPR Conference, pp. 569–593. Prague. Oxford: Pergamon. Hu Z-R, Wentzel MC, and Ekama GA (2000) External nitrification in biological nutrient removal activated sludge systems. Water SA 26(2): 225--238. Hu ZR, Wentzel MC, and Ekama GA (2001) External nitrification in biological nutrient removal activated sludge systems. Water Science and Technology 43(1): 251--260. Hu Z-R, Wentzel MC, and Ekama GA (2002) The significance of denitrifying polyphosphate accumulating organisms in biological nutrient removal activated sludge systems. Water Science and Technology 46(1/2): 129--138. Hu Z-R, Wentzel MC, and Ekama GA (2003) Modelling biological nutrient removal activated sludge systems–A review. Water Research 37(14): 3430--3444. Hu ZR, Wentzel MC, and Ekama GA (2007a) A general kinetic model for biological nutrient removal activated sludge systems: Model development. Biotechnology and Bioengineering 98(6): 1242--1258.
Biological Nutrient Removal
Hu Z, Wentzel MC, and Ekama GA (2007b) A general model for biological nutrient removal activated sludge systems–model evaluation. Biotechnology and Bioengineering 98(6): 1259--1275. Jardin N and Po¨pel HJ (1994) Phosphate release of sludge from enhanced biological P-removal during digestion. Water Science and Technology 30(6): 281--292. Jenkins D, Daigger GT, and Richard MG (1993) Manual on the Causes and Control of Activated Sludge Bulking and Foaming, 2nd ed. Chelsea, MA: Lewis Publishers. Jenkins D and Tandoi V (1991) The applied microbiology of enhanced biological phosphate removal – Accomplishments and needs. Water Research 25(12): 1471--1478. Kappeler J and Gujer W (1992) Estimation of kinetic parameters of heterotrophic biomass under aerobic conditions and characterization of wastewater for activated sludge modelling. Water Science and Technology 25(6): 125--139. Kerrn-Jespersen JP and Henze M (1993) Biological phosphorus uptake under anoxic and oxic condition. Water Research 27(4): 617--624. Korentajer L (1991) A review of the agricultural use of sewage sludge: Benefits and potential hazards. Water SA 17(3): 189--196. Kruit J, Hulsbeek J, and Visser A (2002) Bulking sludge solved? Water Science and Technology 46(1–2): 457--464. Kuba T, Murnleitner E, van Loosdrecht MCM, and Heijnen JJ (1996) A metabolic model for biological phosphorus removal by denitrifying organisms. Biotechnology and Bioengineering 52: 685--695. Kuba T, Smolders G, van Loosdrecht MCM, and Heijnen JJ (1993) Biological phosphorus removal from wastewater by anaerobic-anoxic sequencing batch reactor. Water Science and Technology 27(5–6): 241--252. Kuba T and van Loosdrecht MCM (1996) Phosphorus and nitrogen removal with minimal COD requirement by integration of denitrifying dephosphatation and nitrification in a two-sludge system. Water Research 30(7): 1702--1710. Lawrence and McCarty (1972) Unified basis for biological treatment design and operation. Journal of the Sanitary Engineering Division – ASCE 96(SA3): 757--778. Lee BJ, Wentzel MC, and Ekama GA (2006) Measurement and modelling of ordinary heterotrophic organism active biomass concentration in anoxic/aerobic activated sludge mixed liquor. Water Science and Technology 54(1): 1--10. Lilley ID, Wentzel MC, Loewenthal RE, Ekama GA and Marais GvR (1991) Acid fermentation of primary sludge at 201C. Proceedings of the 2nd Biennial Water Institute of Southern Africa (WISA) Conference, Kempton Park, May. Lindrea KC, Pigdon SP, Boyd B, and Lockwood GA (1994) Biomass characterization in a nitrification-denitrification biological enhanced phosphorus removal (NDBEPR) plant during start-up and subsequent periods of good and poor phosphorus removal. Water Science and Technology 29(7): 91--100. Loewenthal RE and Marais GvR (1977) Carbonate Chemistry of Aquatic Systems: Theory and Application. Michigan, MI: Ann Arbor Science. Lo¨tter LH (1985) The role of bacterial phosphate metabolism in enhanced phosphorus removal from the activated sludge process. Water Science and Technology 17(11/12): 127--139. Lo¨tter LH, Wentzel MC, Ekama GA, and Marais GvR (1986) A study of selected characteristics of Acinetobacter spp. isolated from activated sludge in anaerobic/ anoxic/aerobic and aerobic systems. Water SA 12(4): 203--208. Loveless JE and Painter HA (1968) The influence of metal ion concentration and pH value in the growth of Nitrosomonas strain isolated from activated sludge. Journal of General Microbiology 52: 1--14. Ludzack FJ and Ettinger MB (1962) Controlling operation to minimize activated sludge effluent nitrogen. Journal of Water Pollution Control Federation 34: 920--931. Malan WN and Gouws EP (1966) Geaktiveerde slyk vir rioolwatersuiwering op Bellville. Research Report. Council for Scientific and Industrial Research, November 1966. Mamais D and Jenkins D (1992) The effects of MCRT and temperature on enhanced biological phosphorus removal. Water Science and Technology 26(5/6): 955--965. Mamais D, Jenkins D, and Pitt P (1993) A rapid physical – chemical method for the determination of readily biodegradable soluble COD in municipal wastewater. Water Research 27(1): 195--197. Marais GvR (1974) Faecal bacterial kinetics in stabilization ponds. Journal of the Environmental Engineering Division – ASCE 100(EE1): 119--139. Marais GvR and Ekama GA (1976) The activated sludge process part 1–Steady state behaviour. Water SA 2(4): 163--200. Marais GvR, Loewenthal RE, and Siebritz IP (1983) Review: Observations supporting phosphate removal by biological excess uptake. Water Science and Technology 15(3/4): 15--41.
525
Martin HG, Ivanova N, Kunin V, et al. (2006) Metagenomic analysis of two enhanced biological phosphorus removal (EBPR) sludge communities. Nature Biotech 24(10): 1263--1269. Maszenan AM, Seviour RJ, Patel BKC, et al. (2000) Three isolates of novel polyphosphate-accumulating Gram-positive cocci, obtained from activated sludge, belong to a new genus, Tetrasphaera gen. nov., and description of two new species, Tetrasphaera japonica sp, nov and Tetrasphaera australiensis sp nov. International Journal of Systematic and Evolutionary Microbiology 50: 593--603. Maurer M, Abramovich D, Siegrist H, and Gujer W (1999) Kinetics of biologically induced phosphorus precipitation in waste-water treatment. Water Research 33(2): 484--493. Maurer M, Gujer W, Hany R, and Bachmann S (1997) Intracellular carbon flow in phosphorus accumulating organisms from activated sludge systems. Water Research 31(4): 907--917. Mbaya A, Germanis J, Wilsenach JA, and Ekama GA (2010) The impact of source separation of urine on biological nutrient removal activated sludge plants. In: Proceedings of the WISA Biennial Conference, Durban, 18–22 April. Water Institute of Southern Africa, PO Box 6011, Halfway House, 1685, South Africa. Mbewe A, Wentzel MC, and Ekama GA (1994) Characterization of municipal wastewaters. Research Report W84, Department of Civil Engineering, Rondebosch: University of Cape Town. Mbewe A, Wentzel MC, Lakay MT, and Ekama GA (1998) Characterization of the carbonaceous materials in municipal wastewaters. In: Proceedings of the WISA Biennial Conference, Cape Town, 3–5 May, 2-E7. Water Institute of Southern Africa, PO Box 6011, Halfway House, 1685, South Africa. McKinney RE and Symons JM (1964) Growth and endogenous metabolism – a discussion. Proceedings of the 1st International Conference on Water Pollution Control. London, 1962. Oxford: Pergamon. McLaren AR and Wood RJ (1976) Effective phosphorus removal from sewage by biological means. Water SA 2(1): 47--50. Mebrahtu MK, Vogts M, and Ekama GA (2010) Anoxic-aerobic digestion of waste activated sludge from biological nitrogen and phosphorus removal systems. In: Proceedings of the WISA Biennial Conference. Durban, 18–22 April. Water Institute of Southern Africa, PO Box 6011, Halfway House, 1685, South Africa. Me´ganck M, Malnou D, Le Flohic P, Faup GM, and Rovel JM (1985) The importance of the acidogenic microflora in biological phosphorus removal. Water Science and Technology 17(11/12): 199--212. Mellin HKO, Lakay MT, Wentzel MC, and Ekama GA (1997) The effect of high temperatures (301C) on biological nutrient removal performance and AA (low F/M) filament bulking in ND and NDBEPR systems. Research Report No. W91. Rondebosch: Department of Civil Engineering, University of Cape Town. Metcalf and Eddy Inc (1991) Wastewater Engineering: Collection, Treatment, Disposal And Reuse, 3rd edn. New York: McGraw-Hill. ISBN 0-07-100824-1 Meyer RL, Saunders AM, and Blackall LL (2006) Putative glycogen accumulating organisms belonging to alphaproteobacteria identified through rRNA-based stable isotope probing. Microbiology–SGM 152: 419--429. Mino T, Arun V, Tsuzuki Y, and Matsuo T (1987) Effect of phosphorus accumulation on acetate metabolism in the biological phosphorus removal process. Paper presented at the IAWPRC Biological Phosphate Removal from Wastewater, Rome. Mino T, Liu WT, Kurisu F, and Matsuo T (1995) Modeling glycogen storage and denitrification capability of microorganisms in enhanced biological phosphate removal processes. Water Science and Technology 31(2): 25--34. Mino T, Van Loosdrecht MCM, and Heijnen JJ (1998) Microbiology and biochemistry of the enhanced biological phosphate removal process. Water Research 32(11): 3193--3207. Mino T, Wen-Tso L, Kurisu F, and Matsuo T (1994) Modelling glycogen storage and denitrification capability of microorganisms in enhanced biological phosphate removal processes. Paper presented at the Specialized Conference on Modelling of Wastewater Treatment. Kollekolle, Denmark. Monod J (1950) Technique of continuous culture – Theory and application (Translation from French). Annales de l’Institut Pasteur 79: 167. Monteith HD, Bridle TR, and Sutton PM (1980) Industrial waste carbon sources for biological denitrification. Progress in Water Technology 12(Tor): 127--141. Moosbrugger RE, Wentzel MC, Ekama GA, and Marais GvR (1992). Simple titration procedures to determine H2CO3 alkalinity and short-chain fatty acids in aqueous solutions containing known concentrations of ammonium, phosphate and sulphide weak acid/bases. Water Research Commission, Gezina, RSA. Muller AW, Wentzel MC, and Ekama GA (2006) Estimation of nitrification capacity of rock media trickling filters in external nitrification BNR. Water SA 32(5): 611--618. Murphy M and Lo¨tter LH (1986) The effect of acetate on polyphosphate formation and degradation in activated sludge with particular reference to Acinetobacter calcoaceticus: A microscopic study. Water SA 12: 63--66.
526
Biological Nutrient Removal
Musvoto EV, Casey TG, Ekama GA, Wentzel MC, and Marais GvR (1994) The effect of incomplete denitrification on anoxic-aerobic (low F/M) filament bulking in nutrient removal activated sludge systems. Water Science and Technology 29(7): 295--299. Musvoto EV, Samson K, Taljard M, Fawcett K, and Alexander WV (2002) Calculation of peak oxygen demand in the design of full scale nitrogen removal activated sludge plants. Water SA, Special Edn Wisa 2002: 56–60. Nakamura K, Ishikawa S, and Kawaharasaki M (1995) Phosphate-uptake and release activity in immobilized polyphosphate-accumulating bacterium Microlunatus phosphovorus strain NM-1. Journal of Fermentation and Bioengineering 80(4): 377--382. Neethling JB, Clark D, Pramanik A, Stensel HD, Sandino J, and Tsuchihashi R (2009) WERF nutrient challenge investigates limits of nutrient removal technologies. In: Proceedings of the 2nd IWA Specialized Conference Nutrient Management in Wastewater Treatment Processes. Krakow 6–9 September, 353–362. London: International Water Association (IWA), IWA Publishing. Nicholls HA (1975a) Biological Nitrogen and Phosphorus Removal in Large Plants. Internal Thesis Report, Department of Civil Engineering, University of Cape Town, Rondebosch, RSA. Nicholls HA (1975b) Full scale experimentation on the new Johannesburg extended aeration plant. Water SA 1(3): 121--132. Nicholls HA (1982) Application of the Marais–Ekama activated sludge model to large plants. Water Science and Technology 14(Cape Town): 581--598. Nicholls HA and Osborn DW (1979) Bacterial stress: Prerequisite for biological removal of phosphorus. Journal of the Water Pollution Control Federation 51: 557--569. Nicholls HA, Pitman AR, and Osborn DW (1985) The readily biodegradable fraction of sewage: Its influence on phosphorus removal and measurement. Water Science and Technology 17(11/12): 73--87. Nicholls HA, Osborn DW, and Pitman AR (1986) Biological phosphorus removal at the Johannesburg Northern and Goudkoppies wastewater purification plants. Water SA 12(1): 13--18. Oehmen A, Lemos PC, Carvalho G, Yuan Z, Keller J, and Blackall LL (2007) Advances in enhanced biological phosphorus removal: From micro to macro scale. Water Research 41(11): 2271--2300. Osborn DW and Pitman AR (1987) Improvement to the stability of the biological phosphate removal process at the Johannesburg Northern Works. In: Biological Phosphate Removal From Wastewaters, 261–272. London: International Water Association (IWA), IWA Publishing. Osborn DW, Lo¨tter LH, Pitman AR, and Nicholls HA (1989) Two-year study on the enhancement of biological phosphate removal by altering process feed composition (plant and laboratory studies). WRC Report No. 137/2/89. Pretoria: Water Research Commission. Osborn DW and Nicholls HA (1978) Optimisation of the activated sludge process for the biological removal of phosphorus. Progress in Water Technology 10(1/2): 261--277. Pagilla K, Sattayatewa C, Urgen-Demirtas M, and Baek S (2009) Effect of influent N speciation on organic nitrogen occurrence in activated sludge process effluents. In: Proceedings of the 2nd IWA Specialized Conference Nutrient Management in Wastewater Treatment Processes. Krakow 6–9 September, pp. 221–228. London: International Water Association (IWA), IWA Publishing. Parco V, du Toit GJG and Ekama GA (2009) The impact of high mixed liquor TSS concentration (3-13 gVSS/l) on the kinetics of N and P removal in biological nutrient removal activated sludge systems. Procs. 82nd Annual Water Environment Federation Technical Exhibition and conference - WEFTEC 2009. WEF, Washington DC. Parker DS, Kaufman WJ, and Jenkins D (1971) Physical conditioning of activated sludge floc. Journal of the Water Pollution Control Federation 43(9): 1817--1833. Pattarkine VM and Randall CW (1999) The requirement of metal cations for enhanced biological phosphorus removal by activated sludge. Water Science and Technology 40(2): 159--165. Pilson RA, Ekama GA, Wentzel MC, and Casey TG (1995) The effect of temperature on denitrification kinetics and biological excess phosphorus removal in biological nutrient removal systems in temperate climates (121 C–201 C). Research Report W86, Rondebosch: Department of Civil Engineering, University of Cape Town. Pincince AB, Braley BG, Sangrey KH, and Reardon RD (1995) Minimizing costs of activated sludge systems: Application to the Deer Island treatment plant. In: Proceedings of the 68th Annual WEF Conference and Exposition, Miami, vol. 1, pp. 693–699. Water Environment Federation, Washington, DC. Pitman AR (1991) Design considerations for nutrient removal activated sludge plants. Water Science and Technology 23(4–6): 781--790. Pitman AR (1999) Management of biological nutrient removal plant sludges–change the paradigms? Water Research 33(5): 1141--1146.
Pitman AR, Vandalsen L, and Trim BC (1988) Operating experience with biological nutrient removal at the Johannesburg Bushkoppie works. Water Science and Technology 20(4–5): 51--62. Pitman AR, Venter SLV, and Nicholls HA (1983) Practical experience with biological phosphorus removal plants in Johannesburg. Water Science and Technology 15(3/4): 233--259. Poinapen J and Ekama GA (2010) Biological sulphate reduction using primary sewage sludge in a upflow anaerobic sludge bed reactor–Part 5: Development of a steady state model. Water SA 36(2): 193--202. Randall AA, Benefield LD, and Hill WE (1994) The effect of fermentation products on enhanced biological phosphorus removal, polyphosphate storage, and microbial population dynamics. Water Science and Technology 30(6): 213--219. Randall CW, Marshall DW, and King PH (1970) Phosphate release in activated sludge process. Journal of the Sanitary Engineering Division, ASCE 96(SA2): 395--408. Randall EW, Wilkinson A, and Ekama GA (1991) An instrument for the direct determination of oxygen utilization rate. Water SA 17(1): 11--18. Rensink JH, Donker HJGW, and Vries HPD (1981) Biological P-removal in domestic wastewater by the activated sludge process. In: Proceedings of the 5th Europe Sewage and Refuse Symposium, Munich. European Water Association, Hennep, D53773, Germany or International Solid Waste Association (ISWA), Vienna, A-1080, Austria. Richard MG, Jenkins D, Hao O, and Shimizu G (1982) The isolation and characterization of filamentous micro-organisms from activated sludge bulking. Progress Report No. 81-2. Berkeley: SERL, University of California. Riddell MDR, Lee JS, and Wilson TE (1983) Method for estimating the capacity of an activated sludge plant. Journal of the Water Pollution Control Federation 55(4): 360--368. Samson KA and Ekama GA (2000) An assessment of sewage sludge stability with a specific oxygen utilization rate (SOUR) test method. Water Science and Technology 42(9): 37--40. Satoh H, Mino T, and Matsuo T (1992) Uptake of organic substrates and accumulation of polyhydroxyalkanoates linked with glycolysis of intracellular carbohydrates under anaerobic conditions in the biological excess phosphate removal processes. Water Science and Technology 26(5–6): 933--942. Saunders AM, Mabbett AN, McEwan AG, and Blackall LL (2007) Proton motive force generation from stored polymers for the uptake of acetate under anaerobic conditions. FEMS Microbiology Letters 274(2): 245--251. Sehayek L and Marais GVR (1981) Supplementary phosphorus removal by side-line addition of lime in the activated sludge process. Research Report W40. Rondebosch: Department of Civil Engineering, University of Cape Town. Sen D, Mitta P, and Randall CW (1994) Performance of fixed film media integrated in activated sludge reactors to enhanced nitrogen removal. Water Science and Technology 30(11): 13--24. Setter LR, Carpenter WT, and Winslow GC (1945) Practical application of modified sewage aeration. Sewage Works Journal 17(4): 669. Seviour RJ, Mino T, and Onuki M (2003) The microbiology of biological phosphorus removal in activated sludge systems. FEMS Microbiology Reviews 27(1): 99--127. Shapiro J, Levin GV, and Humberto HZ (1967) Anoxically induced release of phosphate in wastewater treatment. Journal of the Water Pollution Control Federation 39: 1810--1818. Siebritz IP, Ekama GA, and Marais GVR (1980) Excess biological phosphorus removal in the activated sludge process at warm temperature climate. In: Proceedings of the Waste Treatment and Utilization 2, pp. 233–251, Toronto: Pergamon. Siebritz IP, Ekama GA, and Marais GVR (1983) A parametric model for biological excess phosphorus removal. Water Science and Technology 15(3/4): 127--152. Simpkins MJ and McLaren AR (1978) Consistent biological phosphate and nitrate removal in an activated sludge plant. Progress in Water Technology 10(5/6): 433--442. Smolders GJF, van der Meij J, van Loosdrecht MCM, and Heijnen JJ (1994a) Stoichiometric model of the aerobic metabolism of the biological phosphorus removal process. Biotechnology and Bioengineering 44(7): 837--848. Smolders GJF, van der Meij J, van Loosdrecht MCM, and Heijnen JJ (1994b) Model of the anaerobic metabolism of the biological phosphorus removal process: Stoichiometry and pH influence. Biotechnology and Bioengineering 43: 461--470. Smolders GJF, Meij Jvd, Loosdrecht MCMv, and Heijnen JJ (1995) A structured metabolic model for anaerobic and aerobic stoichiometry and kinetics of the biological phosphorus removal process. Biotechnology and Bioengineering 47: 277--287. Sneyders MJ, Wentzel MC, and Ekama GA (1997) The effect of unstabilized landfill leachate addition on biological nutrient removal performance in activated sludge
4.15 Biofilms in Water and Wastewater Treatment Z Lewandowski, Montana State University, Bozeman, MT, USA JP Boltz, CH2M HILL, Inc., Tampa, FL, USA & 2011 Elsevier B.V. All rights reserved.
4.15.1 4.15.2 4.15.2.1 4.15.2.2 4.15.2.2.1 4.15.2.2.2 4.15.2.2.3 4.15.2.2.4 4.15.2.2.5 4.15.2.2.6 4.15.2.2.7 4.15.2.2.8 4.15.3 4.15.3.1 4.15.3.1.1 4.15.3.1.2 4.15.3.2 4.15.3.3 4.15.3.4 4.15.3.4.1 4.15.3.4.2 4.15.3.4.3 4.15.3.4.4 4.15.3.4.5 4.15.3.5 4.15.3.5.1 4.15.3.5.2 4.15.3.5.3 4.15.3.5.4 4.15.3.5.5 4.15.3.5.6 4.15.4 4.15.4.1 4.15.4.1.1 4.15.4.1.2 4.15.4.2 4.15.4.3 4.15.4.4 References
Introduction Part I: Biofilm Fundamentals Biofilm Formation and Propagation The Concepts of Biofilms and Biofilm Processes Quantifying microbial activity, hydrodynamics, and mass transport in biofilms Biofilm heterogeneity and its effects Biofilm activity Quantifying local biofilm activity and mass transport in biofilms from microscale measurements Horizontal variability in diffusivity and microbial activity in biofilms Mechanism of mass transfer near biofilm surfaces Biofilm processes at the macroscale and at the microscale Biofilms in conduits Part II: Biofilm Reactors Application of Biofilm Reactors Techniques for evaluating biofilm reactors Graphical procedure Empirical and Semi-Empirical Models Mathematical Biofilm Models for Practice and Research Biofilm Model Features Attachment and detachment process kinetics and rate coefficients Concentration gradients external to the biofilm surface and the mass transfer boundary layer Diffusivity coefficient for the rate-limiting substrate inside the biofilm Parameters: estimation and variable coefficients Calibration protocol Biofilm Reactors in Wastewater Treatment Biofilm reactor compartments Moving bed biofilm reactors Biologically active filters Expanded and fluidized bed biofilm reactors Rotating biological contactors Trickling filters Part III. Undesirable Biofilms: Examples of Biofilm-Related Problems in the Water and Wastewater Industries Biofilms on Metal Surfaces and MIC Differential aeration cells on iron surfaces SRB corrosion Biofilms on Concrete Surfaces: Crown Corrosion of Sewers Biofilms on Filtration Membranes in Drinking Water Treatment Biofilms on Filtration Membranes in Wastewater Treatment
4.15.1 Introduction Fundamental principles describing biofilms exist as a result of focused research. The use of reactors for the treatment of municipal wastewater is a common application of biofilms. Applied research exists that provides a basis for the mechanistic understanding of biofilm reactors. The empirical information derived from such applied research has been used to develop design criteria for biofilm reactors and remains the basis for biofilm reactor design despite the emergence of
529 529 529 530 531 532 532 533 535 535 536 538 541 542 542 543 543 544 546 546 547 548 548 548 549 549 549 550 556 556 558 562 563 564 564 565 565 566 567
mathematical models as reliable tools for research and practice. Unfortunately, little information exists to bridge the gap between our current understanding of biofilm fundamentals and reactor-scale empirical information. Therefore, there is a clear dichotomy in literature: micro- (biofilm) and macro(reactor) scales. This chapter highlights the division. Part I is dedicated to basic research and communicating the state of the art with respect to understanding biofilms. Part II is practice oriented and describes the use of biofilms for the sanitation of municipal wastewater. A basis for addressing this
529
530
Biofilms in Water and Wastewater Treatment
disconnection is presented by (1) describing the fundamental biofilm principles that can be uniformly applied to biofilms in several disciplines extending from medicine to environmental biotechnology and (2) describing a fundamentalbased approach in order to understand and apply biofilms in reactors. The use of mathematical biofilm models is common in both research and practice, but only a cursory presentation of their mathematical description is presented here. Finally, Part III gives examples of undesirable biofilms in water and wastewater industries and describes the attempts to mitigate their effects. Metabolic reactions mediated by microorganisms residing in biofilms promote the biodeterioration of materials, including metals, concrete, and plastics. It is estimated that microbially influenced corrosion (MIC) alone costs the US economy billions of dollars every year.
4.15.2 Part I: Biofilm Fundamentals 4.15.2.1 Biofilm Formation and Propagation Biofilm formation is a process that consists of a sequence of steps. It begins with the adsorption of macromolecules (e.g., proteins, polysaccharides, nucleic acids, and humic acids) and smaller molecules (e.g., fatty acids, lipids, and pollutants such as polyaromatic hydrocarbons and polychlorinated biphenyls) onto surfaces. These adsorbed molecules form conditioning films which may have multiple effects, such as altering the physicochemical characteristics of the surface, acting as a concentrated nutrient source for microorganisms, suppressing or enhancing the release of toxic metal ions from the surface, detoxifying the bulk solution through the adsorption of inhibitory substances, supplying the nutrients and trace elements required for a biofilm, and triggering biofilm sloughing. Once the surface is prepared, cells begin to attach. The initial stages of biofilm formation are well documented, mostly because acquiring images of microorganisms at this stage of biofilm formation is relatively easy. The adherence of bacteria to a surface is followed by the production of slimy adhesive substances, extracellular polymeric substances (EPS). These are predominantly made of polysaccharides and proteins. Although the association of EPS with attached bacteria has been well documented in the literature, there is little evidence to suggest that EPS participates in the initial stages of adhesion. However, EPS definitely assists the formation of mature biofilms by forming a slimy substance called the biofilm matrix. Figure 1 shows the steps in the formation of mature biofilms. The existence of these three phases of biofilm development, as depicted in Figure 1, is generally acknowledged, although the terminology may vary among authors. For example, Notermans et al. (1991) called these phases: (1) adsorption, (2) consolidation, and (3) colonization. Once a mature biofilm has been established on a surface, it actively propagates and eventually covers the entire surface. The mechanisms of propagation in mature biofilms are more complex than those of initial attachment, and several of these mechanisms of biofilm propagation are depicted in Figure 2. Although biofilms can be seen with an unaided eye, imaging their structure, microbial community structure, and
Biofilm formation Attachment
Colonization
Growth
Bulk fluid
Surface Figure 1 Steps in biofilm formation. & 1995 Center for Biofilm Engineering, MSU-BOZEMAN.
distribution of EPS requires the use of several types of microscopy combined with various probes, such as fluorescent in situ hybridization (FISH) probes and fluorescent proteins (FPs) used as reporter genes. The favorite types of microscopy among biofilm researchers are those that allow the examination of living and fully hydrated biofilms. In addition, sophisticated image acquisition devices are often needed that can selectively stimulate and image various probes when more than one type of multicolored probe is used simultaneously. Using these techniques in conjunction with a suitable microscopy, biofilm researchers can detect the presence of the selected physiological groups of microorganisms in the biofilm, their position in the biofilm with respect to other microorganisms and surface, and even their physiological state – dead, injured, or alive. The in vitro FISH techniques, popular in medical diagnostics, require that DNA or RNA be isolated from the sample and separated on a gel, and that the fluorescent probes then be added to the sample. The in situ variety of the hybridization technique, which is extensively used in biofilm research, does not require isolating DNA or RNA prior to the use of the probes; instead, the probes are hybridized to the respective nucleotide sequences inside the cells (Biesterfeld et al., 2001; Delong et al., 1999; Ito et al., 2002; Jang et al., 2005; Manz et al., 1999). In situ hybridization uses fluorescence-labeled complementary DNA or RNA probes, often derived from fragments of DNA that have been isolated, purified, and amplified. In microbial ecology, ribosomal RNA in bacterial cells is targeted by fluorescencelabeled oligonucleotide probes. Figure 3 shows an image of manganese-oxidizing bacteria (MOB) Leptotrix discophora stained with a FISH probe (green) and counterstained with propidium iodide (red). Propidium iodide is a general stain which is quite popular with biofilm researchers (GrayMerod et al., 2005; McNamara et al., 2003; Nancharaiah et al., 2005). In mature biofilms, microorganisms are imbedded in the layer of EPS. Figure 4 shows an image of a mature biofilm acquired using scanning electron microscopy (SEM). It shows microbes embedded in a matrix of EPS attached to a surface, although the EPS in this image were reduced to an entangled network of dry strands because the sample had to be dehydrated before the biofilm was imaged using electron microscopy.
Biofilms in Water and Wastewater Treatment
531
Streaming Detaching
Seeding dispersal
Rippling
Rolling
Figure 2 Mechanisms of biofilm propagation (MSU-CBE, P.Dirx).
Figure 4 SEM image of a biofilm of Desulfovibrio desulfuricans G20 embedded in EPS (Beyenal et al., 2004).
Figure 3 L. discophora stained with FISH probes and counterstained with propidium iodide. Red indicates cells that were stained with propidium iodide, and green indicates cells that react positively to the fluorescent FISH probe. Yellow indicates green and red overlay. The scale bar is 20 mm (Campbell, 2003).
4.15.2.2 The Concepts of Biofilms and Biofilm Processes It is difficult to offer precise definitions of biofilms and biofilm processes that will satisfy everyone who is interested in studying biofilms and biofilm-based technologies. Several currently used definitions have roots in historical approaches to biofilm studies. These approaches initially referred to biofilms as physical objects – microbial deposits on surfaces – but later expanded the concept to consider biofilms as a mode of
microbial growth, an alternative to microbial growth in suspension. Life scientists often emphasize the definitions that refer to biofilms as a mode of microbial growth. Engineers often find that the definitions that refer to aggregates of microorganisms which are embedded in a matrix composed of microbially excreted EPS and attached to a surface are useful for their applications. Here, we will refer to biofilms as microorganisms and microbial deposits attached to surfaces. We will use the term biofilm processes in reference to all physical, chemical, and biological processes in biofilm systems that affect, or are affected by, the rate of biofilm deposition or the microbial activity in biofilms. Biofilm processes are carried out in biofilm reactors. Colloquially, the terms biofilm reactors and biofilm systems are used interchangeably. However, biofilm systems exist with or without human intervention, while biofilm reactors are produced by our actions.
532
Biofilms in Water and Wastewater Treatment
When we promote or suppress a biofilm process in a biofilm system, or even when we quantify a biofilm process in a biofilm system without affecting its rate, the biofilm system becomes a biofilm reactor. For example, wetlands can be natural or constructed. However, even natural wetlands become biofilm reactors once we start monitoring biofilm processes in them. We will use the term biofilm system to refer to a group of compartments and their components determining biofilm structure and activity. Biofilm systems are composed of four compartments:
• • • •
the the the the
surface to which the microorganisms are attached; biofilm (the microorganisms and the matrix); solution of nutrients; and gas phase (if present).
Each compartment of a biofilm system can have a number of components. The exact number of components in each compartment may vary, depending on the needs of a particular description. For example, for some analyses it may be convenient to identify two components of the biofilm: (1) the EPS (matrix) and (2) the microorganisms. In another study, it may be convenient to identify three components of the biofilm: (1) the EPS, (2) the microorganisms, and (3) the particular matter trapped in the matrix. Similarly, in some studies it may be convenient to single out two components of the surface – (1) the bulk material and (2) the biomineralized deposits – or, if MIC is studied, it may be convenient to describe the surface by identifying three components: (1) the metal substratum, (2) the corrosion products, and (3) the biomineralized deposits on the surface. The needs of the specific study or analysis dictate the number of components identified in each compartment of the biofilm system. Biofilm studies can be characterized as studies of the relations among the compartments, the properties of one or more compartments, or one or more components of a compartment. Among many factors that are used to quantify biofilm processes, biofilm activity is most often used. Biofilm reactors are often designed and operated to optimize biofilm activity, as are the biofilm reactors used for wastewater treatment discussed later in the text. Typically, biofilm activity is identified with the rate of utilization of the growth-limiting nutrient. In some instances, however, rates other than the rate of substrate utilization or biofilm accumulation are better descriptors of the system dynamics. For example, in studies of MIC, the rate of anodic dissolution of the metal affected by the process may be a more useful descriptor of biofilm activity than the rate at which the growth-limiting substrate is utilized. The choice of the process for evaluating biofilm activity is dictated by the nature of the study, and sometimes by analytical convenience. Monitoring the rate of biofilm accumulation is important in many applications, whether we want to enhance or inhibit the growth of biofilms. The methods employed include optical microscopy (Bakke and Olsson, 1986; Bakke et al., 2001), measuring light intensity reflected from microbially colonized surfaces (Bremer and Geesey, 1991; Cloete and Maluleke, 2005), collecting and analyzing images of biofilm depositions (Milferstedt et al., 2006; Pons et al., 2009), surface sensors based on piezoelectric devices (Nivens et al., 1993; Pereira
et al., 2008), and electrochemical sensors in which stainless steel electrodes change their electrochemical behavior as a result of biofilm deposition (Licina et al., 1992; Borenstein and Licina, 1994).
4.15.2.2.1 Quantifying microbial activity, hydrodynamics, and mass transport in biofilms Microbial activity (biofilm activity), hydrodynamics, and mass transport in biofilms are difficult to discuss separately as they affect each other in many ways. Biofilm activity at the microscale is quantified as the flux, from the bulk solution to the biofilm surface, of the substance selected for evaluating biofilm activity. Since fluxes at the microscale are quantified locally, rather than averaged over the entire surface area as is done when biofilm activity is evaluated at the macroscale, the concentration profiles of the selected substance must be measured with microsensors to assure adequate spatial resolution. The idealized model of hydrodynamics and mass transfer in biofilms shown in Figure 5 is a good starting point for a discussion of biofilm activity at the microscale. In this model the overall flow velocity in the main stream is considered to be the average flow velocity, Cb. This decreases toward the surface of the biofilm, as required by hydrodynamics, and reaches concentration Cs at the biofilm surface. The layer of liquid just above the biofilm surface, where the flow velocity decreases as a result of proximity to the surface, is the hydrodynamic boundary layer, and it is denoted by j. As the flow velocity decreases toward the biofilm surface, the mechanism of mass transport changes from being dominated by convection at locations away from the biofilm, where the flow velocity is high, to being dominated by diffusion at locations near the biofilm surface, where the flow velocity is low. As the microorganisms in the biofilm consume nutrients at the rate at which they are delivered and the mass transport becomes less efficient near the biofilm surface, the nutrient concentration decreases near the surface, forming a nutrient concentration profile within the hydrodynamic boundary layer. The layer of liquid above the biofilm surface where the nutrient concentration decreases is the mass transport boundary layer, and it is denoted by LL and RL is the mass transfer resistance external to the biofilm.
Substrate concentration profile
Flow velocity profile
N = k(Cb − Cs) k =
vb
DW 1 = RL LL Cb C LF
LL
ϕ Biofilm Substratum
Figure 5 Profiles of flow velocity and growth-limiting nutrient concentration near the surface of an idealized biofilm.
Biofilms in Water and Wastewater Treatment 7 6 CO2 (mg l−1)
4.15.2.2.2 Biofilm heterogeneity and its effects The term biofilm heterogeneity refers to the extent of the nonuniform distribution of any selected constituent in any of the compartments of the biofilm system, such as the distribution of the biomass, selected nutrients, selected products of microbial metabolism, or selected groups of microorganisms. Since there are many choices for the constituents selected to evaluate biofilm heterogeneity, the term biofilm heterogeneity is usually combined with an adjective referring to the selected constituent, such as structural heterogeneity, chemical heterogeneity, or physiological heterogeneity. The term biofilm heterogeneity was initially used exclusively to refer to the nonuniform distribution of the biomass in a biofilm. As time has passed, more types of heterogeneity have been described, and the term biofilm heterogeneity is not self-explanatory anymore: the specific feature of the biofilm with respect to which the heterogeneity is quantified needs to be specified. Quantifying biofilm heterogeneity is equivalent to quantifying the extent of nonuniform distributions, such as the distribution of biomass in the biofilm. Several tools from the statistical toolbox are available for evaluating the extent of nonuniform distribution; the most popular is the standard deviation. The procedure for estimating the heterogeneity of a selected constituent of a biofilm is identical with the procedure for evaluating the standard deviation of a set of experimental data with one important difference: the deviations from the average are not due to errors in measurement but reflect a feature of the biofilm – heterogeneity. One of the most profound effects of biofilm heterogeneity is that microscale measurements in biofilms deliver different results at different locations. This is an obvious concern as most models referring to microbial growth and activity have been developed for well-mixed reactors, in which the result of a measurement does not depend on the location. Figure 6 shows this effect: three very different profiles of carbon dioxide concentration were measured at three locations in a biofilm. Because of the biofilm heterogeneity, it is impossible to determine a representative location to make the local measurements of biofilm activity that are used to validate models of biofilm processes. To include the effects of biofilm heterogeneity in mathematical models of biofilm processes, the extent of these effects – the spatial variability of the features measured in biofilms – needs to be evaluated experimentally using tools that can take measurements in biofilms to a high spatial resolution. Such tools are routinely used in biofilm research in the form of microelectrodes and various types of microscopy, often enhanced with fluorescent probes. These types of measurements deliver information about selected locations in the biofilm, and their results are referred to as local properties. The most common such measurements are local biofilm activity, local mass transfer coefficient, local diffusivity, and local flow velocity. The definition of the local mass transport coefficient is derived from the measurement procedure: the coefficient of the mass transport of an electroactive species to the tip of an electrically polarized microelectrode. The local mass transport coefficient is measured using an amperometric microelectrode without a membrane operated at the limiting current condition (masstransfer-limited). Local diffusivity is computed from these
533
A B C
5 4 3 2 1 0 0
100 200 300 400 Distance from the bottom (µm)
500
Figure 6 Carbon dioxide concentration profiles measured perpendicularly to the bottom (substratum) at three locations in a biofilm microcolony.
measurements by calibrating local mass transport microelectrodes in gels of known diffusivities (Beyenal et al., 1998).
4.15.2.2.3 Biofilm activity Biofilm activity in a biofilm reactor can be evaluated from the mass balance on the growth-limiting nutrient in the reactor:
Biofilm activity ¼
ðCInfluent CEffluent Þ Q A
ð1Þ
where C is the concentration of the growth-limiting nutrient (kg m3), Q the volumetric flow rate in the reactor (m3 s1), and A the surface area covered by the biofilm (m2). Therefore, biofilm activity at the scale of the reactor is the average flux of nutrients across the biofilm surface, which corresponds to the approach delineated in Equations (12) and (13) used in graphical procedure to evaluate pilot-plant observations. Average biofilm activity in a reactor is a useful descriptor of reactor performance. However, when the underlying biofilm processes are to be studied, an image of local biofilm activity is often required. This information can be extracted from growth-limiting substrate concentration profiles measured at
534
Biofilms in Water and Wastewater Treatment 7
surface on the oxygen profile coincides with the inflection point of the nutrient concentration profile. It is not easy to determine the exact position of the surface, though. We use a simplified procedure, explained later in Figure 16, to find the approximate position of biofilm surface on concentration profiles measured with microelectrodes. One use of such data is to estimate the local biofilm activity in terms of the flux of the growth-limiting nutrient at the location where the profile was measured. The flux of the nutrient across the biofilm surface, JLF at the location of the measurement is computed as the product of the slope of the concentration profile at the biofilm bulk solution interface by the diffusivity coefficient in water of the substance whose concentration was measured:
Oxygen concentration (mg l−1)
6 5 4 3 2 1
JLF ¼ Dw
0 0
300
600
900
1200
1500
Distance from the bottom (µm) Figure 7 Oxygen concentration profile. The vertical line marks the approximate position of the biofilm surface (Rasmussen and Lewandowski, 1998).
selected locations in the biofilm, as shown in Figure 7. The results from the two scales of observation – (1) the local biofilm activity evaluated from the concentration profiles and (2) the average biofilm activity evaluated from the mass balances around the reactor – provide different types of information. The measurements at the microscale deliver information that cannot be extracted from the measurements at the macroscale. For some biofilm processes, it is important to quantify the extreme values of biofilm activity because the locations in the biofilm where these extreme values occur exhibit extreme properties. For example, in studying MIC, which causes highly localized damage to metal surfaces, it is important to evaluate the extreme values of biofilm activity because the extreme, and highly localized, microbial activity in biofilms determines the extent of microbial corrosion. The average biofilm activity estimated from measurements at the macroscale cannot deliver this information.
4.15.2.2.4 Quantifying local biofilm activity and mass transport in biofilms from microscale measurements The profiles of flow velocity and growth-limiting substrate concentration shown in the conceptual image depicted in Figure 5 can be measured experimentally. Their interpretation leads to a better understanding of the processes occurring in biofilms. Figure 7 shows an oxygen concentration profile measured in a biofilm using an oxygen microelectrode. Nutrient concentration profiles, such as the one shown in Figure 7, are composed of two parts, the part above and the part below the biofilm surface. Different factors shape these parts of the profile: the shape of the profile above the surface is dominated by bulk liquid hydrodynamics, whereas the shape of the profile below the surface is dominated by microbial respiration in the biofilm. These two parts are described by different equations but are connected at the biofilm surface by the requirement of oxygen flux continuity. The position of the
dC dx ðxxs Þ¼0
ð2Þ
where Dw is the diffusivity in water of the substance selected for the evaluation of biofilm activity, usually the growthlimiting nutrient (m2 s1). Diffusivity of this substance in the biofilm is not constant, but instead it varies with distance, as explained below. Early mathematical descriptions of biofilm activity and the shape of the concentration profile within the biofilm were based on the conceptual model of so-called uniform biofilms, depicting biomass uniformly distributed in the space occupied by the biofilm (Atkinson and Davies, 1974; Williamson and McCarty, 1976). Formally, these early mathematical models of microbial activity in biofilms imitated the models of microbial activity in suspension, with the addition of mass transport resistance. They quantified the equilibrium between the rate of utilization of the growth-limiting nutrient and the rate of mass transport in one dimension, toward the surface:
2 qC q C mmax Xf C ; ¼ Df qt f qx2 f Yx=s Ks þ C
0 r x r xs
ð3Þ
At steady state, this equation delivers
Df
d 2C mmax CXf ¼ 2 Yx=s ðKs þ CÞ dx
ð4Þ
Two boundary conditions were generally used to specify the concentrations of oxygen at the bottom and surface of the biofilm:
dC dx
¼ 0;
Cðx¼xs Þ ¼ Cs;
t0
ð5Þ
ðx¼0Þ
where Df is the averaged effective diffusivity of growth-limiting nutrient in the biofilm (m2 s1); x the distance from the bottom (m); xs the distance from the biofilm surface in the new system of coordinates (m); Xf the averaged biofilm density (kg m3); Yx/s the yield coefficient (kg microorganisms/kg nutrient); mmax the maximum specific growth rate (s1); Ks the Monod half-rate constant (kg m3); C the growth-limiting substrate concentration (kg m3); and Cs the growth-limiting substrate concentration at the biofilm surface (kg m3). These early models were subsequently refined by adding additional factors affecting biofilm processes, such as bacterial
Biofilms in Water and Wastewater Treatment
d dC d 2 C dDfl dC mmax CXfl Dfl ¼ Dfx 2 þ ¼ dx dx dx dx dx Yx=s ðKs þ CÞ
ð6Þ
where Dfl is the local effective diffusivity of the growth-limiting nutrient (m2 s1) and Xfl the local biofilm density (kg m3). Accepting that diffusivity and biofilm density are variable introduces two new variables into the equation, and functions describing changes in effective diffusivity and biofilm density need to be quantified before the equations can be solved. Experimental data show that density changes are surprisingly regular in biofilms and can be described as a linear function of biofilm depth. Relative surface-averaged effective diffusivity
0.70 0.65
D*fz = 0.001z + 0.2968
0.60 0.55 D fz
growth and decay in a steady-state biofilm (Rittmann and McCarty, 1980a, 1980b) and then the model was extended to include unsteady states and dual nutrient limitations (Rittmann and Brunner, 1984; Rittmann and Dovantzis, 1983). One of the most popular biofilm models, initially marketed as a software called BIOSIM (Wanner and Gujer, 1986), was later improved to include irregular biofilm structure and renamed AQUASIM (Wanner et al., 1995; Wanner and Reichert, 1996). The growing popularity of the conceptual model of heterogeneous biofilms coincided with the growing popularity of cellular automata (CA) (Wolfram, 1986), and it is not surprising that the heterogeneous biofilm structures were modeled using CA procedures (Wimpenny and Colasanti, 1997a, 1997b). Soon after, Picioreanu et al. (1998a, 1998b) improved this model using more realistic assumptions and used differential equations to describe mass transport with the discrete model describing the structure (Picioreanu et al., 1998a, 1998b). Since its early applications, CA remains the most popular model used to generate biofilm structure. Further improvement of the biofilm model came from Kreft et al. (2001), who developed a two-dimensional (2-D) multinutrient, multi-species model of nitrifying biofilms to predict biofilm structures, that is, surface enlargement, roughness, and diffusion distance. These authors compared the predicted structure of the biofilm with the predictions of the biomass (cells and EPS)-based model developed by Picioreanu et al. (1998a, 1998b), and concluded that the two models had similar solutions. Meanwhile, biofilm researchers urgently needed mathematical description of the biofilm processes that could be used to describe recent progress in understanding biofilm processes. The main problems that needed to be addressed were horizontal and vertical profiles in mass transport and activity in biofilms. These were experimentally verified and the assumption that the effective diffusivity and biofilm density were constant across the biofilm had become difficult to defend. Biofilm diffusivity decreases toward the bottom of the biofilm and biofilm density increases. There have been attempts to include these results in the modeling of biofilm processes but they lead to more complicated mathematical expressions in which diffusivity and biofilm density are functions of distance. To simplify these expressions it is possible to model a biofilm as a stack of layers with constant diffusivity and density, which change from layer to layer rather than continuously. At steady state, this approach delivers the mass transport and activity related to the local properties of the biofilm:
535
*
0.50 0.45 0.40 0.35 0.30 0.25
50
100
150
200
250
300
Distance from the bottom, z (µm) Figure 8 The surface-averaged relative effective diffusivity (Dfz*) is multiplied by the diffusivity of the growth-limiting nutrient in the water to calculate the surface-averaged effective diffusivity (Dfz). Since, in the example, the growth-limiting nutrient is oxygen, to calculate the effective diffusivity of oxygen at various distances from the bottom, we must multiply the relative effective diffusivity at various distances from the bottom by the diffusivity of oxygen in water (2.1 105 cm2 s1) (Beyenal and Lewandowski, 2005).
profile, reproduced from Beyenal and Lewandowski (2005), is shown in Figure 8. Assuming that biofilm density varies with depth in a linear fashion, as shown in Figure 8, the diffusivity gradient (x) is constant:
dDfx ¼z dx
ð7Þ
At steady state, this simplifies Equation (5) to the form
Dfl
d 2C dC mmax CXfl ¼ þz 2 dx dx Yx=s ðKs þ CÞ
ð8Þ
Further, it has been demonstrated that in biological aggregates, including biofilms, density is related to effective diffusivity (Fan et al., 1990):
Dfl ¼ 1
0:43X0:92 fl 11:19 þ 0:27X0:99 fl
ð9Þ
Using this equation, we can estimate biofilm density from the variation in local effective diffusivity (Figure 9).
4.15.2.2.5 Horizontal variability in diffusivity and microbial activity in biofilms Concentration profiles of growth-limiting nutrients, such as the one shown in Figure 7, are taken at a specific location in a biofilm. Based on the results, the biofilm activity at that location can be computed. However, when the next profile is taken at another location, even as close as several micrometers from the first location, the two profiles can be significantly different. This is not surprising, considering that biofilms are heterogeneous. However, it brings into question the practice of
536
Biofilms in Water and Wastewater Treatment
evaluating biofilm activity based on a single measurement at an arbitrarily selected location. For microscale measurements in stratified biofilms, the selected variable, such as local effective diffusivity or local dissolved oxygen concentration, is measured at locations on a grid (Figure 10). Grids are positioned at various distances from the bottom. The results are then presented as maps of the distributions of the selected parameter at the specified distances from the bottom, as shown in Figure 11. One of the main advantages of this approach is that it allows us to average the concentrations of oxygen at the selected distances from the bottom and arrive at a representative profile of oxygen that illustrates its distribution across the biofilm and also shows the deviations from the average due to biofilm heterogeneity. The maps of oxygen distributions shown in Figure 11 served to construct the representative profile of oxygen across this biofilm shown in Figure 12.
100 Pseudomonas aeruginosa (v = 3.2 cm s−1) Mixed culture (v = 1.6 cm s−1) Mixed culture (v = 3.2 cm s−1)
Biofilm density (g l−1)
80
60
40
20
0
0
100 200 300 400 Distance from the bottom, z (μm)
500
Figure 9 Variation in biofilm density with distance from the bottom (Beyenal et al., 1998).
4.15.2.2.6 Mechanism of mass transfer near biofilm surfaces When the local nutrient concentrations measured across a biofilm are plotted versus distance, they form a nutrient concentration profile. It would be expected that the shape of the nutrient concentration profile will follow the shape of the local mass transport coefficient profile when they are measured at the same location. It would also be expected that, at locations where the local mass transport coefficient is high, the local nutrient concentration will be high as well, at least higher than at a location where the local mass transport coefficient is low. Figure 13 shows profiles of oxygen concentration and local mass transport coefficient measured at the same location in a biofilm (Rasmussen and Lewandowski, 1998). As can be seen in Figure 13, the mass transport coefficient profile does not correlate well with the oxygen concentration profile. Approaching the biofilm surface, for example, the oxygen concentration decreases rapidly and reaches quite low levels at the biofilm surface, while the local mass transport coefficient remains quite high at that location. This observation seems difficult to explain: since there is no oxygen consumption in the bulk, the oxygen concentration profile would be expected to follow the shape of the mass transport coefficient profile much closer than it does in Figure 13. However, although these two profiles do not match, each of them is consistent with our knowledge of the system’s behavior. We expect to measure a low concentration of oxygen at the biofilm surface: this result fits the concept of a mass transfer boundary layer of high mass transport resistance above the biofilm surface. Measuring a high mass transport coefficient near the biofilm surface is also not surprising because, as we have estimated, convection is the predominant mass transport mechanism in that zone. The two features cannot coexist: high mass transport resistance and convection. To explain this apparent discrepancy, we need to examine the procedure for measuring flow velocity in biofilms. All available flow velocity measurements in biofilms report only one component of the
Figure 10 Microscale measurements in stratified biofilms. The selected variable, such as the local effective diffusivity or local dissolved oxygen concentration, is measured at the locations where the gridlines intersect. Such grids are positioned at various distances from the bottom (MSU-CBE, P.Dirx).
Biofilms in Water and Wastewater Treatment
400
100
200
500
300 m)
y (c
CL (mg l−1)
CL (mg l−1)
400
100
200
500
500
200 x ( 300 cm 400 )
300 m)
7 6 5 4 3 2 1 0 200 x ( 300 cm 400 ) 500
400 500
500
300 200 m) 100 ( y c
7 6 5 4 3 2 1 0 200 x ( 300 cm 400 )
160 µm
400 300 200 cm) 100 y(
7 6 5 4 3 2 1 0 200 x ( 300 cm 400 ) 500
500
500 100
400 300 m) 200 c ( y
500
CL (mg l−1)
7 6 5 4 3 2 1 0 200 500 400 300 300 x( 400 cm 500 100 200 y (cm) )
400 300 200 m) y (c 100
500
180 µm
CL (mg l−1)
280 µm
500
0 µm (boottom)
CL (mg l−1)
CL (mg l−1)
7 6 5 4 3 2 1 0 200 x ( 300 cm 400 ) 500
400 300 200 (cm) y 100
40 µm
7 6 5 4 3 2 1 0
y (c
320 µm
CL (mg l−1)
400 300 200 cm) 100 y(
200 µm
7 6 5 4 3 2 1 0 200 x ( 300 cm 400 ) 500
7 6 5 4 3 2 1 0 200 x ( 300 cm 400 )
7 6 5 4 3 2 1 0 200 x ( 300 cm 400 ) 500
CL (mg l−1)
360 µm
80 µm
CL (mg l−1)
7 6 5 4 3 2 1 0 200 x ( 300 cm 400 ) 500
240 µm
CL (mg l−1)
CL (mg l−1)
400 µm (top)
537
500
100
400 300 200 m) y (c
500
7 6 5 4 3 2 1 0 500 200 400 300 x( 300 400 cm 200 m) ) 500 100 y (c
Figure 11 Distribution of oxygen measured in a biofilm at the specified distances from the bottom (Veluchamy, 2006).
flow velocity vector, parallel to the bottom. Based on these results, we estimated that mass transport is controlled by convection near biofilms. However, the convective mass transport rate equals the nutrient concentration times the flow velocity component normal to the reactive surface. The component of the flow velocity parallel to the surface has nothing to do with the convective mass transport toward that surface. Consequently, the estimate of the mass
transport mechanism based on flow velocity holds only in the direction in which the flow velocity was measured. Indeed, when the flow near a surface is laminar, the laminas of liquid slide parallel to the surface, and there is little or no convection across these layers: the mass transport parallel to the surface is convective, while the mass transport perpendicular to the surface remains diffusive. This mechanism is visualized in Figure 14.
538
Biofilms in Water and Wastewater Treatment 7
Bulk
Biofilm 6
CSA (mg l−1)
5 4 3 2 1 0 0
100
200
300
400
Distance from the bottom, x (µm)
1.0
7
0.9
6
0.8 5
0.7 0.6
4
0.5 3
0.4
2
0.3 Oxygen
1
k/kmax
Dissolved oxygen concentration (mg l−1)
Figure 12 Surface averaged oxygen concentrations (CSA) and standard deviations computed for each data set in Figure 11. The average oxygen concentrations form a representative profile of oxygen concentration, characterizing the area covered with the biofilm, and the envelope of the standard deviation is a measure of the heterogeneity of the measured variable, oxygen concentration in this case (Veluchamy, 2006).
0.2 0.1
k/kmax
0 0
0.0
200 400 600 800 1000 1200 1400 Distance from substratum (µm)
Figure 13 Profiles of oxygen and local mass transfer coefficient through a thin biofilm cluster (’, dissolved oxygen; , local mass transfer coefficient). The vertical line marks the observed thickness of the biofilm. At distances of less than 30 mm, the wall effect caused the local mass transport coefficient to decrease. The biofilm thickness was 70 mm in this location. The value of k/kmax was only slightly affected by the presence of the biofilm up to a distance of less than 30 mm from the substratum (Rasmussen and Lewandowski, 1998).
4.15.2.2.7 Biofilm processes at the macroscale and at the microscale Accurate mathematical models are necessary for advances in biofilm research. Biofilm researchers use mathematical models of biofilm processes not only to predict the outcome of these processes, but also to interpret the results of biofilm studies. In the absence of suitable models, the interpretation of biofilm studies is impaired. Biofilm science and technology are relatively young, and mathematical descriptions of biofilm
processes often lag behind the rapidly expanding knowledge of biofilm processes. On the other hand, most of the experience that was accumulated in modeling biofilm processes in water and wastewater treatment was based on the operating reactors with suspended biomass. Biofilm reactors are different, and some effects common in biofilm reactors are much less usual in reactors with suspended biomass. One effect that is particularly difficult to accommodate in biofilm models is the influence of biofilm heterogeneity on biofilm processes. Biofilm models that describe biofilm processes on the scale of the entire reactor assume that the biofilm is uniformly distributed and its effects do not depend on the location in the reactor. This assumption, which is justified in the case of well-mixed reactors, may or may not be justified in biofilm reactors. With the current sophistication in exploring biofilm processes at the microscale, it is not surprising to observe that the local conditions quantified in biofilms deviate widely from the average conditions described by the biofilm models. One hopes that these deviations from the idealized models cancel each other and that overall, at the macroscale, they do not matter much. One particularly troubling problem is the definition of and the existence of a steady state in biofilm reactors. Defining a steady state in a biofilm reactor may well be the most important question facing biofilm researchers, both those who focus on experiment and those who focus on modeling. The existence of a steady state is obvious in flow reactors, where microbial growth occurs in suspension. In such reactors, the interplay among the microbial growth rates, biomass concentration, and hydraulic and biomass retention times leads to a steady state in which process variables do not change for a long time. In contrast, the reasons for the existence of a steady state in a biofilm reactor are much less clear because an important condition for a steady state is not satisfied in a biofilm reactor: the concentration of biomass in a
Biofilms in Water and Wastewater Treatment
Convection Diffusion Convection and diffusion
539
Direction of mass transport Convection Diffusion Direction of measured flow velocity
Figure 14 Alternating zones of convective and diffusive mass transport in heterogeneous biofilms. This hypothetical model of mass transport is consistent with the results in Figure 13. Mass transport in the space occupied by the biofilm is convective, but the amount of nutrient delivered to this space is limited by the diffusive mass transport just above the biofilm surface (MSU-CBE, P.Dirx).
biofilm reactor is not a simple function of retention time and growth rate. Some biofilm technologies actually take advantage of this fact and grow biofilm microorganisms using retention times at which the microorganisms would be washed out from reactors operated with suspended microorganisms. Practically, this problem corresponds to the fact that we are uncertain what function describes detachment in biofilms, and what mechanisms are involved in biofilm detachment, except perhaps for shear stress. The mechanism of biofilm sloughing remains unknown. A steady state for the biomass concentration assumes that the same amount of biomass is generated as is removed by various processes, particularly biofilm detachment. One can argue that if the biofilm reactor is large enough, the microscale biofilm processes will average out on the scale of the reactor, and that this average may be stable even if the components of the average vary over time. This argument, even if it is true, however, does not settle the issue. A question follows: how large does the reactor have to be to ensure that the variations in the microscale biofilm processes average out and the reactor reaches a steady state at the macroscale? There are also difficulties at the microscale. Experimentally measured concentration profiles and flow velocity profiles corroborate the conceptual model shown in Figure 5. However, when it comes to interpreting experimental data, the idealized image of biofilms in Figure 5 is not adequate for many reasons. One reason is shown in Figure 15: the difficulty with locating the position of the biofilm surface. The position of the biofilm surface is important: one of the boundary conditions in the equation describing biofilm activity and mass transport specifies the conditions at the biofilm surface. As can be seen in Figure 15, however, locating it is not trivial. This problem has been addressed experimentally by judiciously locating the surface on a nutrient concentration profile at the location where the profile ends its curvature near the bottom. The rule of locating the biofilm surface at that location has been developed based on the results of studies in which an oxygen electrode and an optical sensor were used to measure the oxygen concentration profile and detect the biofilm surface, where optical density changed (Figure 16). The position of the biofilm surface coincides with the location where the oxygen profile becomes linear. The biofilm surface in Figure 7 was positioned using this principle.
Figure 15 Surface of a biofilm grown at a flow velocity of 0.81 m s1 (Groenenboom, 2000).
4.15.2.2.8 Biofilms in conduits Among the many possible effects that biofilms may have in water conduits, we will discuss two effects in more detail: (1) the effect on flow characteristics – pressure drop in conduits and (2) the effect on material performance – MIC. Flow velocity near the biofilm surface. It is well known that flow velocity affects biofilm processes. Figure 5 shows an example of the effect of flow velocity on mass transport dynamics near the biofilm surface. However, biofilm also affects flow velocity: flow velocity near a wall covered with biofilm is different from that near a wall with no biofilm. Figure 17 shows this effect. The effect of biofilm on flow velocity distribution most certainly influences the dynamics of mass transfer. However, this is not the only effect that biofilm has on hydrodynamics. For example, it is well known that biofilms increase the pressure drop in conduits, but it is not clear what the mechanism of this process is or how to quantify it. To predict pressure drop in pipes the Moody diagram is used, which correlates the Reynolds number and the relative roughness to provide the friction factor, f. This friction factor is then
Biofilms in Water and Wastewater Treatment 9.0
0.30
Oxygen concentration (mg l−1)
8.0
0.27
Oxygen conc. Log (lo /l)
7.0
0.24
6.0
0.21
5.0
0.18
4.0
0.15
3.0
0.12
2.0
0.09
1.0
0.06
0.0
0.03
0.00
0.30
0.60 0.90 1.20 1.50 Distance from the bottom (mm)
1.80
2.10
Log (lo /l )
540
0.00 2.40
Figure 16 Profiles of oxygen concentration and optical density in a biofilm. A combined microsensor – an oxygen microelectrode and an optical density microprobe – permitted locating the biofilm surface at 0.60 mm from the bottom. This distance, when marked on the oxygen concentration profile, indicates that the biofilm surface is at the beginning of the linear part of the oxygen profile within the mass transfer boundary layer; I is the local light intensity, and Io is the maximum light density (Lewandowski et al., 1991).
1.2 Biofilm 1.0
0.4 Biofilm
0.6
0.3 V /Vmax
V /Vmax
0.8
0.4
0.2 0.1
0.2 0 0
1000
2000
250 Depth (µm) 3000
4000
500 5000
Depth (µm) Figure 17 The flow velocity profile near a wall covered with a biofilm is different from the flow velocity profile near the same wall without the biofilm (DeBeer et al., 1994).
plugged into the Darcy–Weisbach equation to calculate the pressure drop:
HL ¼ f
l V2 D 2g
ð10Þ
where HL is the head loss due to friction, l the pipe length, V the average fluid velocity, g the gravitation constant, D the pipe diameter, and f the friction factor provided by the Moody diagram. When the flow velocity increases, the thickness of the boundary layer decreases, and the roughness elements protrude through the boundary layer, further affecting the drag and the pressure drop.
Unfortunately, the Moody diagram is of little help in predicting the pressure drop in conduits covered with biofilms. The pressure drop in such conduits is caused by different factors than the pressure drop in conduits without biofilms because different mechanisms are responsible for the shape of the pressure drop in each of these conduits. These differences sometimes demonstrate themselves in the form of puzzling experimental results, such as decreasing pressure drop resulting from increasing flow velocity, which is a consequence of the elastic and viscoelastic properties of biofilms. Microcolonies are made of bacterial cells embedded in gelatinous EPS that can change shape under stress. At high flow velocities the hydrodynamic boundary layer separates from the
Biofilms in Water and Wastewater Treatment
3 2 1 m
2.5 m
8 4
m
2.5 m
m
2.5 m
2.0 cm
1 2.0 cm
8 4
m
2.5 m
V (cm s−1)
4
2
2.5 m
2.0 cm
8
3
m
2.0 cm
V (cm s−1)
V (cm s−1)
Without biofilm
V (cm s−1)
V (cm s−1)
With biofilm
V (cm s−1)
microcolonies, causing pressure drag downstream of the microcolony and pulling the material in this direction. The microcolonies slowly flow under the strain, forming elongated shapes that we call streamers. Streamers are often seen when biofilms grow at high flow velocities. The streamers contribute to pressure drop by moving rapidly and dissipating the kinetic energy of the flowing water. Another important consequence of a streamer’s oscillations is that they are transmitted to the underlying microcolonies, which also oscillate rhythmically. This system reacts with turbulent boundary layers much differently than the rigid surface roughness elements of clean pipes do. One way to gain experimental access to the interactions between flowing water and biofilm is to monitor flow velocity profiles. Imaging flow velocity profiles makes it possible to evaluate the effect of biofilm formation on the flow in conduits by quantifying its effect on the entry length in the conduit. The hydrodynamic entry length is defined as the distance needed to develop a steady flow, after the water has passed through the entrance to the reactor. If the presence of biofilm makes the entry length longer, then the biofilm contributes to flow instability, and vice versa. There is a simple relation between the Reynolds number and the entry length: the higher the Reynolds number, the longer the entry length. This effect was used as a base for quantifying the effects of biofilm on the flow in conduits. Flow velocity distribution was measured in a rectangular reactor when the flow velocity was increasing from one measurement to another. As the flow velocity and the Reynolds number increased, the flow stability was monitored in a rectangular conduit using nuclear magnetic resonance (NMR) imaging. The results, shown in Figure 18, demonstrate that the presence of biofilm actually made the flow more stable. The entry length was shorter and the flow reached stability closer to the entrance in the presence of biofilm than in its absence. It is difficult to interpret this result immediately because it is well known that the presence of biofilm increases pressure drop in conduits: traditionally, pressure drop in pipes is related to friction. As pressure drop is larger in biofilm-covered pipes, a natural conclusion was that biofilms must increase friction and therefore the presence of the biofilm should introduce flow instability rather than reduce it. The relation between flowing water and biofilms is determined by two facts: (1) biofilms are made of viscoelastic polymers which actively interact with the oscillations generated by the flow of water and (2) the flow of water affects the biofilm structure. Based on what we now understand, at low flow velocities biofilms can effectively smooth surfaces and stabilize the flow because the oscillating layer of elastic polymeric matrix can effectively damp the vibrations coming from the flowing water. This effect delays the onset of turbulence in conduits covered with biofilm and explains the results shown in Figure 18. However, as the flow velocity increases further, the elastic polymeric matrix must oscillate faster and faster and, eventually, the frequency of its oscillation cannot follow the frequency of the incoming eddies. At that point the biofilm oscillation is out of phase and the biofilm not only fails to damp the flow instabilities but also actively introduces instability by randomly oscillating at a different frequency than the incoming eddies. The pressure drop in the conduit
541
2.0 cm
8 4
m
2.5 m
2.0 cm
Figure 18 Flow velocity profiles in a rectangular conduit whose walls were colonized with a biofilm. The increasing flow velocity did not affect the character of the velocity profiles in the reactor with biofilm. On the other hand, the same increase in velocity had a pronounced effect on the reactor without biofilm.
increases rapidly. This effect was, in early biofilm works, mistaken for a similar effect caused by rough surface elements. For example, Picologlou et al. (1980) observed a considerable increase in frictional resistance after the film thickness reached a value approximately equal to the calculated thickness of the hydrodynamic boundary layer for a clean surface. In clean pipes covered with surface roughness elements, when flow velocity increases the boundary layer becomes thinner and at some flow velocity the boundary layer thickness is smaller than the height of the roughness elements. When this happens, the roughness elements protrude through the boundary layer and cause an additional drag, which exhibits itself in a sudden increase of the pressure drop for flow velocities exceeding this critical flow velocity. This model was commonly accepted and was used to explain the pressure drop in conduits covered with biofilms, although even at that time some authors warned that this might not be the true mechanism of the process (Characklis, 1981). Currently, there are no models that can account accurately for pressure drop in conduits covered with biofilm.
4.15.3 Part II: Biofilm Reactors Biological systems treating municipal wastewater require (1) the accumulation of active microorganisms in a bioreactor and (2) the separation of the microorganisms from treated effluent. In suspended growth reactors, such as the activated
542
Biofilms in Water and Wastewater Treatment
sludge process, microorganisms grow and bioflocculate; the resultant flocs are suspended freely in the bulk phase. Flocculated bacteria are then separated from the bulk liquid by sedimentation or membranes. Clarifier-coupled suspended growth reactors rely on return activated sludge, or underflow, from the coupled clarifier to provide the desired active biomass concentration in the bioreactor. Consequently, clarification unit processes may be limited by the hydraulic loading rate (HLR) or solids loading rate (SLR). Biofilm reactors retain bacterial cells in a biofilm that is attached to the fixed or free moving carriers. The biofilm matrix consists of water and a variety of soluble (C) and particulate (X) components that include soluble microbial products, inert material, and EPS. Without suspended biomass, the bioreactor is decoupled from the liquid–solids separation unit. Active biomass concentrations inside the biofilm are large at 10–60 g of volatile suspended solids (VSS) l1 of biofilm. This biomass range can be compared with the range of concentrations expected for suspended growth reactors, which is typically 3–8 g VSS l1 of reactor volume. The lower value in this range is associated with clarifier-coupled activated sludge processes, and the upper range with membrane bioreactors. In biofilm reactors, bacteria attached to carriers periodically detach from the biofilm matrix and exit the system in the effluent stream. Figure 19 provides a conceptual illustration of different biofilm reactor types. Biofilm reactors can be classified based on the number of phases involved – gas, liquid, solid – according to the biofilm being attached to a fixed or moving carrier within the reactor. They are also classified based on how electron donors or acceptors are applied to seven basic types as listed below (adapted from Harremo¨es and Wilderer (1993)): 1. Three-phase system – fixed biofilm-laden carrier, bulk water, and air. Water trickles over the biofilm surface and
2.
3.
4.
5.
6.
7.
air moves upward or downward in the third phase (e.g., trickling filter (TF)) (Figure 19(a)). Three-phase system – fixed (or semifixed) biofilm-laden carrier, bulk water, and air. Water flows through the biofilm reactor with gas bubbles (e.g., aerobic biologically active filter (BAF)). Gravel is a fixed media and polystyrene beads are semifixed (Figures 19(b) and 19(c)). Three-phase system – moving biofilm-laden carrier, bulk water, and air. Water flows through the biofilm reactor. Air is introduced with gas bubbles (e.g., aerobic moving bed biofilm reactor (MBBR)) (Figure 19(g)). Two-phase system – moving biofilm-laden carrier and bulk water. Water flows through the biofilm reactor with the electron donor and electron acceptor (e.g., denitrification fluidized bed biofilm reactor (FBBR)) (Figure 19(g)). Two-phase system – fixed biofilm-laden carrier material and bulk water. Water flows through the biofilm reactor with the electron donor and electron acceptor (e.g., denitrification filter) (Figures 19(b) and 19(c)). Three-phase membrane system – a microporous hollowfiber membrane with biofilm and water on one side and gas on the other, diffusing through the membrane to the biofilm (e.g., membrane biofilm reactor (MBfR)) (Figure 19(h)). Two-phase membrane system – a proton exchange membrane separating a compartmentalized biofilm-laden anode from a compartmentalized cathode with water on both sides, but with the electron donor on one side and electron acceptor on the other (e.g., biofilm-based microbial fuel cell (MFC)).
Biofilms are ubiquitous in nature and in engineered systems and can be used beneficially in municipal water and wastewater treatment. Biofilm and suspended growth reactors can meet similar treatment objectives for carbon oxidation,
Air
Air (a)
(b)
(c)
(f)
(g)
Air
Air (e)
(d)
(h)
Figure 19 Types of biofilm reactors: (a) trickling filter; (b) submerged fixed bed biofilm reactor operated as up flow or (c) down flow mode; (d) rotating biological contactor; (e) suspended biofilm reactor including airlift reactor; (f) fluidized bed reactor; (g) moving bed biofilm reactor; and (h) membrane attached biofilm reactors. From Morgenroth (2008) Modelling biofilm systems. In: Henze M, van Loosdrecht MCM, Ekama G, and Brdjanovic D (eds.) Biological Wastewater Treatment – Principles, Modelling, and Design, pp. 457–492. London: IWA Publishing.
Biofilms in Water and Wastewater Treatment
nitrification, denitrification, and desulfurization. Biofilm reactors have also been used for the treatment of a variety of oxidized contaminants including perchlorate and bromate. The same microorganisms are responsible for biochemical reactions in both activated sludge and biofilm systems, and respond in the same way to local environmental conditions (i.e., pH, temperature, electron donor, electron acceptor, and macronutrient availability) (Morgenroth, 2008). A key component to be considered by anyone who is evaluating a biofilm reactor is the effect of multiple substrates and biomass fractions and the manner in which the reactor is affected by mass-transport limitations. Substrates typically considered are: 1. soluble compounds, including electron donors (e.g., readily biodegradable chemical oxygen demand (rbCOD), NHþ 4 , NO2 , and H2), electron acceptors (e.g., O2, NO3 , 2 3 NO2 , and SO4 ), and nutrients and buffers (e.g., PO4 , NHþ 4 , and HCO3 ) and 2. particulate compounds, including electron donors (e.g., slowly biodegradable COD (sbCOD)), active biomass fractions (e.g., heterotrophic and autotrophic bacteria), inert biomass, and EPS.
4.15.3.1 Application of Biofilm Reactors This section exists to provide the reader with a general overview of biofilm reactor applications. While general biofilm reactor applicability is described here, several treatment scenarios exist that are not conveniently generalized yet warrant the use of biofilm reactor technology. Water-quality regulations exist to protect human health and the water environment. Organic matter and the nutrients such as nitrogen and phosphorus are major contributors to water-quality impairment. In municipal wastewater-treatment scenarios, biofilm reactors are generally applied for the removal of carbon-based organic matter and/or nitrogenous compounds. Specifically, these biofilm reactors may achieve carbon oxidation, combined carbon oxidation and nitrification, tertiary nitrification, or tertiary denitrification. Biofilm reactors are not commonly used for biological phosphorus removal. Biofilm reactors treating industrial wastewaters have been applied to meet treatment objectives similar to those in municipal wastewater treatment and industrial pretreatment. The objective of pretreatment is to process industrial waste streams until their characteristics are similar to raw sewage (see Metcalf and Eddy (2003) for a description). As a result the industry can then discharge their treated wastewater into municipal sewers where further processing is accomplished at a municipal wastewater-treatment plant. Biofilm reactors are common for industrial applications because the processes are reliable, robust, easy to operate, and resilient to toxic or shock loading.
4.15.3.1.1 Techniques for evaluating biofilm reactors Several approaches exist to evaluate biofilm reactors. The primary objective of a biofilm or biofilm reactor model is to predict soluble substrate flux (J) through the biofilm surface. This flux (M L2 T1) can be used to obtain an estimate of the (1) overall biofilm reactor performance, (2) required biofilm surface area, (3) electron acceptor (e.g., dissolved oxygen), (4)
543
external electron donor (e.g., methanol or hydrogen), and (5) biosolids management requirements. This section discusses the relative benefits and limitations to some general methods of evaluating biofilm reactors. The use of mathematical biofilm models is common in both research and practice, but only a cursory presentation of their mathematical description is presented. Excellent resources exist describing aspects of mathematical modeling of biofilms and biofilm reactors (for additional information, see Wanner et al. (2006) and Morgenroth (2008)). The approaches discussed here include a graphical procedure, empirical models, semiempirical models, and mechanistic mathematical models.
4.15.3.1.2 Graphical procedure A graphical procedure can be used to determine the total hydraulic load (THL) required to decrease a substrate concentration, and by definition the biofilm surface area required to provide a desired substrate concentration remaining in the effluent stream. These items can be determined directly. The graphical procedure can be used to determine effluent substrate concentration from any series of continuous flow stirred tank reactors (CFSTRs). A stepwise procedure must be used when a series of CFSTRs will be used. Antoine (1976) and Grady et al. (1999) developed the graphical procedure described here and the approach is valid for any biofilm-based CFSTR. If multiple stages are expected to have different characteristics, then the graphical method requires different flux curves to describe system response in each of the CFSTRs. The procedure requires a graphical representation of substrate flux (J) as a function of bulk-liquid substrate concentration (CB). This relationship between flux and bulk-liquid substrate concentration can be obtained from numerical simulations, full-scale or pilot-plant observations. In practice, this graphical procedure is typically used to extend pilot-plant observations to full-scale biofilm reactor design criteria. The process designer should recognize that the relationship between flux and bulk-liquid substrate concentration is based on the system and location. Therefore, the flux curve required to implement the graphical procedure may not be obtained from or correlate well with values reported in the literature or from different systems. As a result, the process designer should consider carefully the conditions under which the flux curve was developed before applying results. A flux curve representing mass transfer and environmental conditions characteristic of a specific system and operating mode may not be the representative of different biofilm reactor types designed to meet the same treatment objectives. A flux curve generated for the same biofilm reactor type under similar operating conditions, however, may offer some direction in the absence of system-specific numerical simulation or pilot/full-scale observations. When using the graphical procedure to evaluate pilot-plant observations, fluxes should be compared to rates in full-scale systems. Any flux that deviates significantly from those reported for biofilm reactors in published studies should be used only after careful consideration. Pilot or experimental systems may promote a greater flux than expected. The basis for the graphical procedure is a material balance on a
544
Biofilms in Water and Wastewater Treatment
biofilm-based CFSTR:
0 = Q ⋅ Cin ,i − Q ⋅ C B ,i − J LF ,i ⋅ A − rB ,i ⋅ VB mass per time input
mass per time output
biofilm transformation rate
suspended growth transformation rate
ð11Þ
where Q is the flow rate through the system (m3 d1); Cin,i the influent concentration of soluble substrate i (g m3); CB,i the effluent, or bulk-liquid, concentration of soluble substrate i (g m3); JLF,i the flux of soluble substrate i across the biofilm surface equal to the average biofilm activity in the reactor, as shown in Equation (1) (g m2 d1); A the biofilm surface area (m2); rB,i the rate of substrate i conversion because of suspended biomass (g m2 d1); and VB the bulk-liquid volume (m3). Assuming that transformation occurring in the bulk liquid is negligible, the suspended growth transformation rate (Equation (11)) can be neglected. Rearranging Equation (11) provides the rationale for the graphical procedure:
JLF;i ¼
Q Q Cin;i CB;i A A |fflfflfflffl{zfflfflfflffl} |{z} const:
ð12Þ
slope
The slope, or ( (Q/A)), is referred to as the operating line and represents the total hydraulic load on each stage. Figure 20 illustrates the graphical method. The flux curves have been created based on observations in the first and second stage of a post-denitrification biofilm reactor. The ordinate represents nitrate–nitrogen flux and the abscissa nitrate–nitrogen concentration remaining in the effluent stream. The graphical solution indicates that the
first-stage denitrification biofilm reactor effluent nitrate– nitrogen concentration is approximately 3.9 mg l1. The secondstage effluent nitrate–nitrogen concentration is approximately 1.1 mg l1 with fluxes of approximately 1.6 and 1.1 g m2 d1 in the first and second stage, respectively. The graphical procedure depends on the substrate flux curve(s). The method requires development of multiple flux curves if the performance characteristics of respective stages vary significantly. When using pilot-plant data to generate a flux curve, appropriate scale considerations must be given when designing the pilot unit and experiments.
4.15.3.2 Empirical and Semi-Empirical Models Empirical models can be implemented easily by hand or using a spreadsheet, but they have limited applicability because of their black-box consideration of system parameters. Because environmental conditions and bioreactor configuration affect biofilm reactor performance, a system can respond differently from the description provided by an empirical model. The limited descriptive capacity of empirical models typically results from parameter values and model features based on data that were obtained from few system installations or operating conditions. Therefore, the engineer or scientist should be aware of conditions under which system-specific model parameters have been defined. Significant sources of variability in values include differences in biofilm carrier type and configuration, the extent of concentration gradients external to the biofilm surface, and biofilm composition. Despite their ease of implementation, empirical models can produce results that vary 50–100% of actual system performance.
3.5 Denitrification rate (g m−2 d−1 as NO3−N)
Stage 1 operating line 3.0 Stage 2 operating line
Stage 1 flux response curve
2.5 Stage 2 flux response curve
J LF1
2.0
1.5 −Q/A J LF2
1.0
0.5 CB -stage
0.0 0
1
2
3
CB -stage 1 4
5 6 CNO −N (mg-N l−1) 3
C in 7
8
9
10
Figure 20 Graphical procedure for describing the response of a denitrification moving bed biofilm reactor to defined conditions, including (1) firstand second-stage operating lines and (2) flux curves based on observations at a pilot-scale denitrification moving bed biofilm reactor (Boltz et al., 2010b).
Biofilms in Water and Wastewater Treatment
Coefficient values, and sometimes the empirical models, are typically created to describe system response for the removal of a specific material. The models can be used as an indicator of system viability to meet treatment objectives with respect to the specific process governing transformation. Empirical models are, however, inadequate for describing complex processes such as the explicit evaluation of two-step ammonium oxidation first to nitrite by ammonia-oxidizing bacteria and then to nitrate by nitrite-oxidizing bacteria. Therefore, empirical models have limited application in defining the conditions that either promote or deter complex processes in biological systems. Historically, biofilm reactors have been designed using empirical criteria and models, but this trend is changing. One should recognize that the coefficients in empirical models describing biofilm reactors include system, and many times, location-specific mass-transfer resistances (Grady et al., 1999). For this reason, the values typically differ from apparent or intrinsic values reported in the literature. Once a flux has been determined, Equation (11) can be rearranged, neglecting bulkphase conversion processes, to calculate the material concentration remaining in the effluent:
CB;i ¼ Cin
JLF;i A Q
ð13Þ
If sufficient data exist to allow for the development of parameter values and mathematical relationships capable of describing a complete range of conditions expected when treating municipal wastewater, then empirical models can be used. The addition of model components to account for specific phenomenon encroaches on the premise of mechanistic mathematical model development. For this reason, a distinction is made between empirical and semi-empirical models. Gujer and Boller (1986) and Sen and Randall (2008) provided an example of the latter describing nitrifying TFs, and MBBRs and IFAS systems, respectively.
C
LF
LL
4.15.3.3 Mathematical Biofilm Models for Practice and Research Mathematical modeling can be used to describe certain features of a biofilm or biofilm system (such as a bioreactor) by selecting and solving mathematical expressions. Biofilm reactor research and design commonly involve the use of mathematical biofilm models. These mathematical models are tools that allow the user to efficiently evaluate a variety of complex scenarios. Empirical models fail to provide information that is a concern for biofilm researchers and environmental protection such as biofilm composition and competition among bacteria for multiple substrates and space inside the biofilm, and the influence of individual processes on the interaction between several bacterial types. Mathematical biofilm models have been used as a research tool, but only recently modern biofilm reactors have encouraged the use of biofilm models in engineering practice. Submerged and completely mixed biofilm reactors allow for the application of modern biofilm knowledge, and are conducive to simulation with existing biofilm models (Boltz and Daigger, 2010). As a result, a majority of existing wastewater-treatment plant simulators have been improved to include a biofilm reactor module(s) that is based on the mathematical description of a 1-D biofilm. A user should understand the mathematical biofilm model basis, underlying assumptions, and limitations before applying the model to research or design. A biofilm schematic is shown in Figure 21. The schematic illustrates diffusion and reaction occurring inside a 1-D biofilm. In addition, concentration gradients external to the biofilm surface are illustrated in the manner that they are modeled, namely an external mass transfer resistance represented by a mass transfer boundary layer. The partial differential equation describing molecular diffusion, substrate utilization inside a biofilm, and dynamic accumulation has been presented as Equation (3). It should be emphasized that the basis for a mathematical description of the 1-D biofilm, as described by Equation (3), is simultaneously
C
LF
LL
CB
CB
C LF
C LF
Distance from growth medium
Distance from growth medium
Z Distance from surface, X
545
Z Distance from surface, X
Figure 21 Schematic of a 1-D biofilm of thickness LF having an assumed homogeneous (a) and heterogeneous, or layered, (b) biomass distribution. Soluble substrate concentration profile is illustrated with a bulk-liquid concentration (CB) decreasing through a mass transfer boundary layer of thickness LL until reaching the liquid–biofilm interfacial concentration (CLF), and then decreasing through the biofilm.
546
Biofilms in Water and Wastewater Treatment
occurring molecular diffusion and biochemical reaction. Molecular diffusion is based on Fick’s law. Monod-type kinetics is typically applied to describe the biochemical transformation rate. Analytical solutions to Equation (3) are available only for first- and zero-order rate expressions and assuming steady state. Zero-order kinetics are valid if the bulkliquid substrate concentration is well above the half-saturation concentration (i.e., CB,i 4 Ki), and first-order kinetics is applicable for low substrate concentrations (i.e., CB,ioKi). Solving the second-order differential equations requires constants that can be derived from two boundary conditions described by Equation (5). From the concentration profile (Cf,i(x)) the flux through the biofilm surface (JLF) is calculated as Equation (2). This substrate flux, JLF, is used in biofilm reactor material balances (see Equation (11)). The concentration gradient external to the biofilm surface is not explicitly modeled. Rather, it is modeled as a mass transfer resistance:
JMTBL ¼
1 ðCB;i CLF;i Þ RL
ð14Þ
•
•
Here, JMTBL is the substrate flux in the stagnant liquid layer and RL the mass transfer resistance external to the biofilm. It is helpful to visualize RL by introducing the concept of a mass transfer boundary layer. Defining the thickness of this mass transfer boundary layer provides a more intuitive understanding compared to the mass transfer resistance. Resistance to mass transfer and the mass transfer boundary layer thickness are related according to Equation (15):
RL ¼
LL Dw
ð15Þ
Here, LL is the mass transfer boundary layer thickness and Dw the solute diffusion coefficient in the water phase. The substrate flux through the mass transfer boundary layer (Equation (15)) is linked to the substrate flux across the biofilm surface (Equation (2)). This provides an additional Equation (16) (boundary condition) that is required to calculate the additional unknown value of the substrate concentration at the liquid–biofilm interface (JLF):
JMTBL ¼ JLF
ð16Þ
One of the most difficult aspects of choosing an approach to model biofilms and biofilm reactors is to choose the appropriate level of complexity. An overview of the different model approaches is provided below (after Taka´cs et al., 2010):
•
•
0-D biofilm. One aspect of modeling biofilms is that bacteria are retained in the system and are not washed out with effluent water. The simplest approach for biofilm modeling would be to assume that all biomass in the reactor is exposed to bulk phase concentrations neglecting the effect of mass transport limitations (i.e., 0-D). In wastewater treatment biofilms are relatively thick and are usually masstransfer-limited. Thus, the 0-D modeling approach that neglects mass transfer limitations is not useful except in special cases. 1-D homogeneous biofilm (single limiting substrate). This approach takes into account mass transfer limitations into
•
the biofilm and the corresponding effects on concentration profiles and substrate flux into the biofilm. It is assumed that active bacteria are homogeneously distributed over the thickness of the biofilm. The approach is valid only if calculations are performed for the limiting substrate which has to be determined a priori by the user as described in Morgenroth (2008). The flux of the nonlimiting substrates can be calculated based on reaction stoichiometry. 1-D homogeneous biofilm (multiple substrates and multiple biomass components). One key aspect of modeling biofilms is to evaluate the competition and coexistence of different groups of bacteria and local environmental conditions. Local process conditions can be accurately determined by calculating penetration depths for different soluble substrates. Based on the fluxes the growth of individual groups of bacteria can be determined. To simplify calculations it can be assumed that all bacterial groups are homogeneously distributed over the thickness of the biofilm (Rauch et al., 1999; Boltz et al., 2009a). 1-D heterogeneous biofilm. Different groups of bacteria are competing in a biofilm not only for substrate but also for space where bacteria toward the surface are less influenced by mass transport limitations. Bacteria growing toward the base of the biofilm are often rate limited by substrate availability resulting from mass transfer limitations. On the other hand, these bacteria are better protected from detachment. These 1-D heterogeneous biofilm models must keep track of local growth and decay of the different bacterial groups and of detachment to calculate biomass distributions over the biofilm thickness. 2-D and 3-D biofilm models. Practically, biofilms are not as smooth and flat as is assumed in 1-D biofilm models. Mathematical models have been developed that predict the development of biofilms in two or three dimensions, the influence of the heterogeneous structure on fluid flow, and ultimately the combination of fluid flow and biofilm structure on substrate availability and removal inside the biofilm. For most questions related to practical biofilm reactor studies, such multi-dimensional models are not necessary. However, it is important for model users to recognize that biofilm structure influences local fluid dynamics and external mass transport, which are simultaneously affected by biofilm reactor appurtenances and mode of operation. Such interactions are not accounted for in existing 1-D biofilm models due to a rigid segregation of the bulk phase, mass transfer boundary layer, and biofilm (which is assumed to have a uniform thickness and smooth surface). Multi-dimensional biofilm models have been used to quantify the influence of biofilm structure on local fluid dynamics and external mass transport (Eberl et al., 2000).
Different scales of heterogeneity are relevant for biofilm reactors. The length scale of the biofilm thickness, which is on the order of 100–1000 mm, is taken into account in 1-D and multi-dimensional biofilm models. Substrate fluxes from these simulations can then be integrated into models describing overall reactor performance where the length scale is on the order of 1 m. However, heterogeneities can also be observed in biofilm reactors in between these scales where, in some cases, patchy biofilms are observed and where certain
Biofilms in Water and Wastewater Treatment
parts of the biofilm support medium is bare while at other areas dense biofilms develop (B1–10 cm). These heterogeneities in between the small and the large scale are typically not considered in biofilm models and it is not clear to what extent they are relevant (Taka´cs et al., 2010). No simple and general recommendations can be given on what approach is the most appropriate for describing biofilm reactors. Wanner et al. (2006) provided a detailed description of different modeling approaches and a discussion on how the modeling approaches compare for different modeling scenarios. Many commercially available wastewater-treatment plant simulators used for biofilm reactor design and evaluation takes into account multiple substrates and biomass fractions in either a heterogeneous or a homogeneous 1-D biofilm. Examples of software, and references to the biofilm model that constitutes the biofilm reactor module, that is applied to design, optimize, and evaluate, typically pilot- or full-scale biofilm reactors are summarized in Table 1.
4.15.3.4 Biofilm Model Features Excellent guides exist that describe the mathematical modeling of biofilms (see Wanner et al., 2006; Morgenroth, 2008). However, the state of biofilm modeling is subject to several uncertainties. In the context of this chapter, Boltz et al. (2010a) summarized the following items which cause uncertainty when using 1-D biofilm models to describe biofilm reactors: (1) the fate of particulate substrates, (2) biofilm distribution in the reactor and the effect biofilms have on reactor components, (3) dynamics and fate of biofilm detachment, (4) quantifying concentration gradients external to the biofilm surface, and (5) a lack of generally accepted biofilm reactor Table 1
model calibration protocol. Parameter estimation and model calibration are serious concerns for process engineers who apply biofilm models in engineering practice. Therefore, parameters that are critical components of biofilm reactor models (that use a 1-D mathematical biofilm model) are introduced, including: attachment (kat) and detachment (kdet) coefficients, the mass transfer boundary layer, rate-limiting substrate diffusivity coefficient inside the biofilm (Df,ratelimiting), and the biokinetic parameters maximum growth rate (m) and the ratelimiting substrate half-saturation coefficient (Ki,ratelimiting) (Boltz et al., 2010b).
4.15.3.4.1 Attachment and detachment process kinetics and rate coefficients An accurate mathematical description of particle attachment and detachment processes is a critical component of biofilm (reactor) models. Unfortunately, attachment/detachment process mechanics are poorly understood. Conceptually, particles suspended in the bulk liquid are hydrodynamically transported to the vicinity of the biofilm. From the bulk phase, particles are subjected to concentration gradients external to the biofilm surface. Particles enter the biofilm matrix through channels, crevasses, and other structural irregularities where they attach to the biofilm surface (see Reichert and Wanner (1997) for a description of particle transport within the biofilm matrix). Once entrapped, the particles can be hydrolyzed by extracellular polymeric enzymes resulting in soluble substrate that diffuses into the biofilm. Then, the soluble substrate is subject to well-known biochemical transformation processes that yield biomass. Alternatively, particles that have attached to the biofilm surface from the bulk phase remain unaltered and exit the system after detaching from the biofilm
Biofilm models used in practice (Boltz et al. 2010b)
Software
Company
Biofilm model type and biomass distribution
Reference
AQUASIMTM
EAWAG, Swiss Federal Institute of Aquatic Science and Technology, Du¨bendorf, Switzerland (www.eawag.ch/index_EN) Aquaregen, Mountain View, California (www.aquifas.com) EnviroSim Associates Ltd., Flamborough, Canada (www.envirosim.com)
1-D, DY, N; heterogeneous
Wanner and Reichert (1996) (modified)
1-D, DY, SE and N, heterogeneous 1-D, DY, N, heterogeneous
Sen and Randall (2008)
Hydromantis Inc., Hamilton, Canada (www.hydromantis.com) CH2M HILL Inc., Englewood, Colorado (www.ch2m.com/corporate) ifak GmbH, Magdeburg, Germany (www.ifak-system.com) WRc, Wiltshire, England (www.wateronline.com/ storefronts/wrcgroup.html) MOSTforWATER, Kortrijk, Belgium (www.mostforwater.com)
1-D, DY, N, heterogeneous
AQUIFASTM BioWinTM
GPS-XTM Pro2DTM SimbaTM STOATTM WESTTM
547
1-D, SS, N(A), homogeneous (constant Lf) 1-D, DY, N, heterogeneous 1-D, DY, N, heterogeneous 1-D, DY, N(A)a, Nb, homogeneousa, heterogeneousb
Wanner and Reichert (1996) (modified), Taka´cs et al. (2007) Hydromantis (2006) Boltz et al. (2009a; 2009b) Wanner and Reichert (1996) (modified) Wanner and Reichert (1996) (modified) Rauch et al. (1999)a, Wanner and Reichert (1996) (modified)b
a
Rauch et al. (1999) is linked with the definition ’N(A)’ and ’homogeneous’. Wanner and Reichert (1996) (modified) is linked with the definition ’N’ and ’heterogeneous’.
b
1-D, one dimensional; DY, dynamic; N, numerical; N(A), numerical solution using analytical flux expressions; SE, semi-empirical; SS, steady-state. Hydromantis, Inc. (2006) Attached growth models. In: GPS-X Technical Reference, pp. 157–185 (unpublished).
548
Biofilms in Water and Wastewater Treatment
matrix. Most of the heterogeneous 1-D biofilm models listed in Table 1 describe the rate of particle attachment (rat) as a first-order process ðrat ¼ kat XTSS;bulk Þ depending on an attachment rate coefficient (kat) and the bulk-liquid particle concentration. Boltz and La Motta (2007) presented a model describing variability in this parameter with influent particle concentrations. The researchers postulated that increasing particle concentrations ultimately reduced the biofilm surface area available for particle attachment; thereby, the particle attachment coefficient decreases until reaching a plateau. The plateau was considered commensurate with a condition in which a minimum biofilm area was consistently available as a result of continuously detaching biofilm fragments (during steady operating conditions – variable hydrodynamics can influence biofilm structure). Given the current state of the science, experimental data are required to develop/validate or evaluate existing approaches for simulating the fate of particles in biofilm reactors. Steady-state biofilm models have assumed a constant biofilm thickness in which case biofilm growth is balanced by internal loss (e.g., decay and hydrolysis, or endogenous respiration) and/or detachment. This approach has been successfully applied to simulate biofilm reactors at steady state, but their dynamic simulation requires that a detachment model is included despite rather limited mechanistic understanding. The rate (Morgenroth and Wilderer, 2000; Boltz et al., 2010a) and category (i.e., abrasion, erosion, sloughing, and predator grazing) of detachment can have a significant influence on biofilm reactor simulation and performance (Morgenroth, 2003). Kissel et al. (1984) stated that problems inherent to biofilm detachment modeling include a poor understanding of fundamental (biofilm detachment) process mechanics and the inability to predict exactly at what location inside the biofilm that detachment will occur. Detachment location is important when taking into account a heterogeneous biofilm distribution throughout the reactor either by combining multiple 1-D simulations or by 2- or 3-D modeling (Morgenroth et al., 2000). Unlike attachment, Boltz et al. (2010a) described eight different biofilm detachment rate expressions (rdet) for the heterogeneous 1-D biofilm models listed in Table 1. Detachment rate equations can be categorized based on the aspect controlling detachment: biofilm thickness (LF), shear, or growth/activity. Mixed-culture biofilms, such as those growing in a combined carbon oxidation and nitrification MBBR, are subject to competition for substrate between fast-growing heterotrophic and slow-growing autotrophic organisms (primarily for dissolved oxygen). Morgenroth and Wilderer (2000) performed a modeling study that demonstrated ammonium flux was significantly influenced by the mode of simulated detachment. Essentially, biofilm (thickness) dynamics influenced competition for substrate between heterotrophic and autotrophic organisms; high variations in biofilm thickness dynamics favored the faster growing heterotrophic organisms.
4.15.3.4.2 Concentration gradients external to the biofilm surface and the mass transfer boundary layer Biofilms growing virtually in all full-scale biofilm reactors are subject to some degree of substrate concentration gradients
external to the biofilm surface. Concentration gradients external to the biofilm surface are not explicitly simulated in 1-D biofilm models. Rather, the reduction in concentration of any substrate is modeled as a mass-transfer resistance, RL ( ¼ LL/Dw). Based on the observation that the external masstransfer resistance, RL, is more dependent on biofilm reactor bulk-liquid hydrodynamics than biofilm thickness or surface heterogeneity, the impact of RL can be accounted for by empirical correlations (Boltz et al., 2010a). However, a realistic description of hydrodynamic effects ultimately depends on an accurate estimate of the mass-transfer boundary layer thickness LL. Therefore, the mass-transfer boundary layer thickness is an important facet of biofilm-reactor models that use a 1-D biofilm model. Despite the potential significant impact the mass-transfer boundary layer thickness may have on biofilmreactor model results and process design, factors influencing the interface between the biofilm model and reactor scale is one important feature of biofilm-reactor models that is not well understood.
4.15.3.4.3 Diffusivity coefficient for the rate-limiting substrate inside the biofilm Soluble substrates are primarily transported into biofilms by a combination of advection and molecular diffusion. Generally, the most important mechanism is molecular diffusion (Zhang and Bishop, 1994). The largest component of biofilm is water, but the diffusivity of a solute inside the biofilm is generally less than that in water because of the tortuosity of the pores and minimal biofilm permeability. Consequently, an effective diffusivity must be applied. Many biofilm reactor models treat this value as 80% of the diffusivity in water (i.e., Dw ¼ Df/0.80) (Stewart, 2003). However, it has been demonstrated that the effective diffusion coefficient (Df,i) for any soluble substrate i can vary with depth inside the biofilm (Beyenal and Lewandowski, 2000). The effective diffusivity decreases with depth because of increasing density and decreasing porosity and permeability of the biofilm with depth. Flow velocity past the biofilm is a major influencing factor determining biofilm density. Varying liquid velocity in the vicinity of the biofilm surface can influence a soluble substrate effective diffusivity inside a biofilm. Consequently, the varying flow rate can affect the rate of internal mass transfer and transformation rates (Bishop, 2003). Turbulent, high-sheer stress environments result in planar and denser biofilms while quiescent, low-sheer stress environments will result in rough and less dense biofilms (van Loosdrecht et al., 1995). Picioreanu (1999) defined a growth number ðG ¼ L2f mmax Xf =ðDf CB ÞÞ that can be related to biofilm roughness. According to Picioreanu (1999), the biofilm may have a dense solid matrix and a flat surface when Go5. However, if G 4 10 the biofilm may develop complex structures such as mushroom clusters and streamers.
4.15.3.4.4 Parameters: estimation and variable coefficients A parameter is an arbitrary constant whose value characterizes a system member. Biokinetic parameter estimation is a serious concern for those who seek to use biofilm models for biofilm reactor process design and research because most parameter values cannot be measured directly in full-scale municipal
Biofilms in Water and Wastewater Treatment
wastewater-treatment plants (Brockmann et al., 2008). Parameters exist for every aspect of biofilm models, including stoichiometry, kinetics, mass transfer, and the biofilm itself. A majority of parameter values in modern process models (e.g., those described by Henze et al. (2000)) have a substantial database that serves to define a relatively narrow range of values that are applicable to a majority of municipal wastewater-treatment systems. Existing biofilm models are relatively insensitive to changes in a majority of the biokinetic parameter values, most of which are described by Henze et al. (2000), within a range of values reported in the literature except for, as an example, the autotrophic nitrifier maximum growth rate (m). However, the mathematical description of some processes includes variable, or lumped, parameters. These parameter values are often system specific and subject to significant uncertainty. The lumped parameters account for an incomplete mechanistic description of the simulated process. Lumped parameters in a majority of biofilm models, including those described in this chapter, are the oxygen affinity constant for autotrophic nitrifiers (KO2,A), endogenous respiration rate constants (bres), attachment rate coefficient (kat), detachment rate coefficient (kdet), mass-transfer boundary layer thickness (LL), ratio of diffusion in biofilm to diffusion in water (Df/Dw). Indentifying parameter subsets that require definition for biofilm model calibration has been the subject of several investigations by Smets et al. (1999), Van Hulle et al. (2004), and Brockmann et al. (2008).
549
model for another period. Similarly, Bilyk et al. (2008) reported the calibration of a denitrification filter model by adjusting assumed biofilm thickness and incorporating the assimilative denitrification reaction. Both of these biofilm reactor model calibration efforts were based on bulk-phase measurements, but only Sin et al. (2008) utilized measured characteristics of the biofilm. Such adjustments to systemspecific biofilm and biokinetic parameters in order to match observed data may not produce a properly calibrated model that is capable of describing a variety of design conditions for a wastewater-treatment plant. As previously discussed, the attachment coefficient, for example, has been experimentally demonstrated (and described mathematically) to change as a function of particle (total suspended solids) load (Boltz and La Motta, 2007). Then, it may be argued that adjusting the attachment coefficient (during calibration) to match an observed dataset would naturally render the calibrated model incapable of describing another scenario with a different particle load. Suffice it to say that a reliable and transparent description of recommended approaches for the application and calibration of biofilm models are required for the models to gain general acceptance, understanding, and become effectively used in engineering practice. Protocol defining methodology for sampling, testing, evaluating and applying data to mathematical biofilm reactor models is required. It is likely that existing biofilm reactor models will require improvement for reliable dynamic simulation in practice.
4.15.3.5 Biofilm Reactors in Wastewater Treatment 4.15.3.4.5 Calibration protocol Application of a dynamic biofilm model to describe full-scale municipal wastewater-treatment processes requires a calibration of the selected model. Ad hoc expert-based trial and error and standardized systematic approaches have been used to calibrate process models. Sin et al. (2005) presented a critical comparison of systematic calibration protocols for activated sludge models. These protocols have many similarities that are applicable to biofilm reactor models including goal definition, data collection/testing/reconciliation, and validation. The major differences between protocols reported by Sin et al. (2005) are related to the measurement campaign, experimental methods for influent wastewater characterization, and parameter subset selection and calibration. The major differences speak to areas of systematic calibration protocols for activated sludge models that will almost certainly be exasperated when creating systematic protocol for calibrating a biofilm reactor model. Certainly, additional tests will be required to characterize the physical attributes of both suspended growth and biofilm compartments, and mathematical biofilm models have more parameters than activated sludge models. Furthermore, the biofilm compartment parameters must be estimated from bulk-phase measurements in order to have a timely and costeffective approach to calibrating biofilm reactor models. Sin et al. (2008) reported the calibration of a dynamic biologically active (continuously backwashing) filter model using traditional expert-based manual trial and error. The researchers manipulated system-specific parameters related to attachment, detachment, and biofilm thickness. After calibration, Sin et al. (2008) successfully tested the calibrated
Biofilm reactors play an important role in environmental biotechnology, but many aspects of their design and operation remain poorly understood. Biofilm reactors can be traced to origination of modern water sanitation. Corbett (1903) reported the use of continuously distributed sewage flow over a fixed bed, and Stoddart (1911) reported the use of a coarse biofilm-covered medium dosed with a continuous trickling flow. These accounts are acknowledged as the creation of the TF process. Approximately 100 years following these reports significant advances in the design, academic understanding, and mathematical modeling of biofilms have led to the development of new and emerging biofilm reactors conducive to fundamentally based design approaches and the application of fundamentally based design and operation procedures for traditional biofilm reactors. Two processes – mass transfer and biochemical conversion – are characteristics of all biofilm reactors and influence biofilm structure and function. Compartments that are common to every biofilm reactor exist to optimize mass-transfer and biochemical conversion.
4.15.3.5.1 Biofilm reactor compartments Biofilm reactors have five primary compartments: (1) influent wastewater (distribution) system; (2) containment structure; (3) biofilm carrier; (4) effluent water collection system; and (5) an aeration system (for aerobic processes and scour) or mixing system (for anoxic processes that require bulk-liquid agitation and biofilm carrier distribution). Five components influence local conditions inside the biofilm: (1) biofilm carrier surface (i.e., substratum); (2) biofilm (including both particulate and liquid fractions); (3) mass-transfer boundary
550
Biofilms in Water and Wastewater Treatment
layer; (4) bulk liquid; and (5) gas phase (when significant). The components typical of biofilm reactors are described in context of some commercially available biofilm reactors. The five biofilm reactors described include the MBBR, BAF, FBBR, rotating biological contactor (RBC), and TF.
4.15.3.5.2 Moving bed biofilm reactors The MBBR is a two- (anoxic) or three- (aerobic) phase system with a buoyant free-moving plastic biofilm carrier that requires mechanical mixing or aeration to distribute carriers throughout the tank. The process includes a submerged, completely mixed biofilm reactor and liquid–solids separation unit (Ødegaard, 2006). A range of pollutant loading and bulkphase external carbon sources in denitrification MBBRs and dissolved oxygen concentrations in carbon-oxidation and/or nitrification MBBRs have been applied, and system response evaluated (Lazarova and Manem, 1994). It has been demonstrated that MBBRs are capable of processing wastewater to meet a variety of effluent water-quality standards ranging, for example, from the US Environmental Protection Agency definition of secondary treatment (30 mg TSS l1 and 30 mg BOD5/l monthly average) to more stringent enhanced nitrogen removal limits (e.g., total nitrogen less than 3–5 mg l1) under a variety of loading conditions. The MBBR process is capable of meeting similar treatment objectives as the activated sludge process for carbon oxidation, nitrification, and denitrification, but the MBBR makes use of a smaller tank volume than a clarifier-coupled activated sludge system. Biomass retention is clarifier independent; therefore, solids loading in liquid–solids separation unit are significantly reduced when compared with the activated sludge process. Because it is a continuously flowing process, the MBBR does not require a special operational cycle for biofilm thickness control (e.g., backwashing in a BAF or flushing in a TF). Hydraulic head loss and operational complexity is minimal. The MBBR offers much of the same flexibility to manipulate the process flow sheet (to meet specific treatment objectives) as the activated sludge process. Multiple reactors can be configured in series without the need for intermediate pumping or return activated sludge pumping (to accumulate mixed liquor). Liquid–solids separation may be achieved with a variety of processes including sedimentation basins, dissolved air flotation, cloth-disk and membrane filters. The MBBR is well suited for retrofit installation into existing municipal wastewater-treatment plants. An MBBR may be a single reactor or several reactors in a series. Typically, each MBBR has a length-to-width ratio (L:W) in the range of 0.5:1–1.5:1. Plans with an L:W greater than 1.5:1 can result in nonuniform distribution of the biofilm carriers. MBBRs contain a plastic biofilm carrier volume up to 67% of the liquid volume. Screens are typically installed with one MBBR wall and allow treated effluent to flow to the next treatment step while retaining the free-moving plastic biofilm carriers. Aerobic MBBRs use a diffused aeration system to evenly distribute the plastic biofilm carriers and meet process oxygen requirements. On the other hand, anoxic MBBRs use mechanical mixers to evenly distribute the plastic biofilm carriers because there is no process oxygen requirements. Each process mechanical component is submerged. Figure 22
depicts the Williams-Monaco WWTP, Commerce City, Colorado, a two-train bioreactor that consists of four MBBRs in series. The biofilm carriers are extruded or molded from either virgin or recycled high-density polyethylene (HDPE). Table 2 summarizes characteristics of several commercially available plastic biofilm carriers. The carriers are slightly buoyant and have a specific gravity between 0.94 and 0.96 g cm3. Both native and biofilm-covered plastic biofilm carriers have a propensity to float in quiescent water. Biofilms primarily develop on the protected surface inside the plastic biofilm carrier. For this reason, the specific surface areas of plastic biofilm carriers listed in the table exclude areas not inside the plastic carrier. The listed bulk-specific surface area, which is based on 100% carrier fill, is characteristic of a plastic biofilm carrier. The net specific surface area is characteristic of plastic biofilm carrier and fill percentage. For example, if a plastic biofilm carrier has a 500 m2 m3 bulk-specific surface area, then the net specific surface area at 50% carrier fill is 250 m2 m3. Similarly, the net liquid volume displacement at 50% carrier fill is 0.0725 for a plastic biofilm carrier having a characteristic 0.15-bulk-liquid volume displacement (at 100% carrier fill). Plastic biofilm carriers are retained in an MBBR by horizontally configured cylindrical screens or vertically configured flat screens as shown in Figure 23. Aerobic zones typically contain cylindrical screens; anoxic zones contain the flat wall screens. Cylindrical screens are desired. They extend horizontally into the upward-flowing air bubbles imparted by the diffuser grid which aids in scouring any accumulated debris. Energy imparted by the mechanical mixers is insufficient to dislodge debris accumulated on the flat wall screen. Therefore, scouring of flat screens is accomplished with a sparging air header in a denitrification MBBR. Removing the debris retained on a screen aids in maintaining hydraulic throughput. Hydraulically, an MBBR is commonly designed to process a maximum approach velocity (based on the tank cross-sectional area perpendicular to forward flow) in the range 30– 35 m h1. Screen area is defined by the maximum allowable head loss through the screens, which is typically in the range of 5–10 cm. The screen superficial hydraulic load is typically in the range of 50–55 m h1 for average design conditions. The screens and their supporting structural assemblies, if required, are typically constructed from stainless steel and may be from wedge-wire mesh or perforated plates. Low-pressure diffused air is applied to aerobic MBBRs. The airflow enters the reactor through a network of air piping and diffusers that are attached to the tank bottom. Airflow has the dual purpose of meeting process oxygen requirements and uniformly distributing plastic biofilm carriers. To promote uniform distribution of the plastic biofilm carriers, the diffuser grid layout and drop pipe arrangement provide a rolling water circulation pattern. Coarse-bubble diffusers are typically used in moving bed reactors (Figure 25). Coarse-bubble diffusers typically used in MBBRs are stainless steel pipes with circular orifices along the underside. These coarse-bubble diffusers are less affected by scaling and fouling because of the large dimension and turbulent airflow through the discharge orifice (Stenstrom and Rosso, 2008). As a result, coarse-bubble diffusers require less maintenance than fine-bubble diffusers. The coarse-bubble diffusers are designed with a structural end
Biofilms in Water and Wastewater Treatment
551
Aerated reactor #2
Aerated reactor #1
RECIR
Mixer Mixed bed reactor #2
Screen
Effluent overflow Effluent
Airflow distribution area
RECIR pump Effluent basin
RECIR
RECIR
Mixed bed reactor #1
Effluent
Influent Influent splitter box
Aerated reactor #3
RECIR
Aerated reactor #4
(a)
Mixed bed reactor #4
Mixed bed reactor #3
RECIR
Effluent
RECIR
(b)
Figure 22 (a) Moving bed biofilm reactor at the Williams-Monaco Wastewater Treatment Plant, Colorado, USA. (b) Schematic representation of the photographed system which illustrates the system consisting of two parallel trains each with four reactors in series.
support that enables them to withstand the weight of plastic biofilm carriers when the MBBR is out of service and drained. Denitrification MBBRs use mechanical mixers to agitate the bulk of the liquid and to distribute plastic biofilm carriers uniformly throughout the tank. The mechanical mixers are typically rail-mounted submersible (wet motor) units. Stateof-the-art submersible mechanical mixers typically have a maximum 120-rpm impeller speed and a minimum of three blades per impeller. The mixer uses a stainless steel backwardcurve propeller with a round bar welded along its leading edge to avoid damage to the plastic biofilm carriers and impeller wear. The mixer has a large diameter impeller with a fairly low rotational speed (90 rpm at 50 Hz and 105 rpm at 60 Hz). The plastic biofilm carriers float in quiescent water. As a result, the mixers need to be located near the water surface but not so close as to create an air-entraining vortex. A slight negative
inclination of mixer orientation helps maintain the rollingwater circulation pattern and uniformly distribute plastic biofilm carriers (see Figure 24). Rail-mounted units facilitate access to the mixer when maintenance is required. The mixers are typically sized to input 25 W m3 of reactor volume. Carbon-oxidizing MBBRs are classified as low-rate, normalrate, or high-rate bioreactors. Low-rate carbon-oxidizing MBBRs promote conditions for nitrification in downstream reactors. High- and normal-rate MBBRs are strictly carbon-oxidizing bioreactors. In the absence of site-specific pilot-scale observations or a calibrated mathematical model, high-rate MBBRs are typically designed to receive a filtered BOD5 load in the range of 15–20 g m2 d1 at 15 1C. This corresponds to total BOD5 loads as high as 45–60 g m2 d1 at 15 1C (Ødegaard, 2006). To reach secondary treatment effluent standards, a hydraulic residence time less than 30 min is not
552
Biofilms in Water and Wastewater Treatment
Table 2
Moving bed biofilm reactor plastic biofilm carrier characteristicsa
Manufacturer
Name
Bulk specific surface area, weight, gravity
Nominal carrier dimensions (depth; diameter)
Veolia Inc.
AnoxKaldnesTM K1
500 m2 m3 145 kg m3 0.96–0.98
7.2 mm; 9.1 mm
AnoxKaldnesTM K3
500 m2 m3 95 kg m3 0.96–0.98
10 mm; 25 mm
AnoxKaldnesTM Biofilm Chip (M)
1,200 m2 m3 234 kg m3 0.96–1.02
2.2 mm; 45 mm
AnoxKaldnesTM Biofilm Chip (P)
900 m2 m3 173 kg m3 0.96–1.02
3 mm; 45 mm
ActiveCellTM 450
450 m2 m3 134 kg m3 0.96
15 mm; 22 mm
ActiveCellTM 515
515 m2 m3 144 kg m3 0.96
15 mm; 22 mm
ABC4TM
600 m2 m3 150 kg m3 0.94–0.96
14 mm; 14 mm
ABC5TM
660 m2 m3 150 kg m3 0.94–0.96
12 mm; 12 mm
BioPortzTM
589 m2 m3
14 mm, 18 mm
Infilco Degremont Inc.
Aquise
Entex Technologies Inc.
Carrier photo
a
As reported by manufacturer. Modified from Boltz JP, Morgenroth E, deBarbadillo C, et al. (2010b) Biofilm reactor technology and design. In: Design of Municipal Wastewater Treatment Plants, WEF Manual of Practice No. 8, ASCE Manuals and Reports on Engineering Practice No. 76, 5th edn, vol. 2, ch. 13, (ISBN P/N 978-0-07-166360-1 of set 978-0-07-166358-8; MHID P/N 0-07166360-6 of set 0-07-166358-4). New York: McGraw-Hill.
recommended. Medium-rate MBBRs designed for meeting basic secondary treatment standards are typically designed for a loading of 5–10 g BOD5 m2 d1 at 10 1C, depending on the choice of liquid–solids separation process. Values in the higher range are used when coagulation occurs before the separation unit; values in the lower range are used without coagulation. Studying a pilot-scale combined carbon oxidation and nitrification MBBR receiving primary effluent and a (tertiary) nitrification MBBR receiving secondary effluent while maintaining a 4–6 g m3 bulk-liquid dissolved-oxygen concentration in both units, Hem et al. (1994) observed that a total BOD5 load of 1–2 g m2 d1 resulted in nitrification rates
from 0.7 to 1.2 g m2 d1, a total BOD5 load of 2–3 g m2 d1 resulted in nitrification rates from 0.3 to 0.8 g m2 d1, and a total BOD5 load greater than 5 g m2 d1 resulted in virtually no nitrification.
4.15.3.5.3 Biologically active filters BAFs have natural mineral, structured or random plastic media that supports biofilm growth and serves as a filtration medium. Solids accumulated from filtration and biochemical transformation are removed by backwashing. Media density influences BAF configuration and backwash regimes. BAF
Biofilms in Water and Wastewater Treatment
553
(a)
(b)
Figure 23 (a) Horizontal cylindrical screens constructed of wedge wire. Stainless steel coarse-bubble diffusers typically used in aerobic MBBRs are also pictured on the tank floor. (b) Flat wall screen constructed of wedge wire. A single air-header is pictured. Air is periodically introduced to scour debris accumulated on the screen.
A
B
30°
D (a)
(b)
Figure 24 (a) Schematic and (b) picture of mechanical mixers that are specially designed for anoxic moving bed biofilm reactors.
influent requires preliminary and primary treatment. Historically, the acronym BAF has meant biological aerated filters which have been used to refer to aerated biofilters used for secondary treatment. However, Boltz et al. (2010b) revised the acronym BAF to cover all BAFs, including those that operate under anoxic conditions for denitrification. BAFs are characterized by their media configurations and flow regime.
Downflow BAFs with media heavier than water include the Biocarbones process, which was marketed during the 1980s for secondary and tertiary treatment, and packed-bed tertiary denitrification filters such as the Tetra Denites process. These BAFs are backwashed using an intermittent counter-current flow. Upflow BAFs with media heavier than water such as the Infilco Degremont Biofors process have been used for
554
Biofilms in Water and Wastewater Treatment
secondary and tertiary treatment. The systems make use of expanded clay or another mineral media. These BAFs are backwashed using an intermittent concurrent flow. BAFs with floating media such as the Veolia Biostyrs process have also been used for secondary and tertiary treatment, and uses polystyrene, polypropylene, or polyethylene media. These BAFs operate with an intermittent backwash counter-current flow. Continuous backwashing filters operate in an upflow mode and contains media that is heavier than water. The media continuously moves counter-current to the wastewater flow (i.e., downward), and is continuously channeled to a center air lift where it is scoured, rinsed, and returned to the top of the media bed. Nonbackwashing submerged filters consist of a submerged static media bed, and have been called submerged aerated filters (SAFs). Solids are not retained in these filters. Therefore, nonbackwashing submerged filters require a dedicated liquid–solids separation process. A downflow BAF with media heavier than water, such as the Tetra Denites filter, is illustrated in Figure 25. The
Denites process has been used since the late 1970s for meeting stringent total nitrogen limits while providing a filtered effluent. Methanol or another external carbon source is added to the influent wastewater stream to promote biological denitrification. A typical installation includes 1.8 m of 2–3 mm diameter sand media over 457 mm of graded support gravel. In a downflow denitrification BAF, the backwash cycle typically consists of a brief air scour followed by an air–water backwash and water rinse cycle. Backwash water and air scour flow rates are typically 15 and 90 m3 m2 h1, respectively. Backwash water usage is typically 2–3% of the average flow being treated. Nitrogen gas accumulates in the media. A releasing mechanism is pumping backwash water up through the media bed for a short duration. The denitrification capacity between nitrogen release cycles typically ranges from 0.25 to 0.5 kg NOX-N m2. An upflow BAF with media heavier than water, such as the Infilco Degremont Biofors, is illustrated in Figure 26. The Degremont Biofors operates such that solids are trapped Proces air
Raw water
Air
Backwash water extraction
Water Biofilter media Support layer
Air scour Backwash water Treated water Figure 25 Downflow BAF with media heavier than water (e.g., Biocarbones and Tetra Denites). From ATV (1997) Biologische und weitergehende Abwasserreinigung (German), 4th edn. Berlin: Ernst and Sohn as presented by Tschui (1994).
Water
Process air
Biofilter media
Backwash water extraction
Air
Support layer
Air scour Treated water Backwash water
Raw water Figure 26 Upflow BAF with media heavier than water (e.g., Infilco Degremont Biofors). From ATV (1997) Biologische und weitergehende Abwasserreinigung (German), 4th edn. Berlin: Ernst and Sohn as presented by Tschui (1994).
Biofilms in Water and Wastewater Treatment
mostly in the lower part of the filter medium during normal operation and are removed through backwashing and applying scour air. As the backwash consists of concurrent scour air and backwash water, accumulated solids travel up through the media bed before being released at the top. Three types of media can be used in the Biofors depending on the application; the media types include expanded clay, expanded shale (both in the form of spherical grains with an effective size of 3.5 or 4.5 mm), and angular grains (with an effective size of 2.7 mm). The media form a submerged, fixed bed in the bottom of the reactor. The media bed typically has a height of 3–4 m with approximately 1-m freeboard. The grains-specific surface area is approximately 1640 m2 m3. Influent water to the bed flows through a plenum and nozzle air/water distribution system. The nozzles are installed in a false floor located approximately 1 m above the filter floor. Nozzles in the false floor are subject to clogging. Therefore, backwash water and scour air flow through the same plenum/nozzle system. Process air is introduced through separate air diffusers located in the media bed above the inlet nozzles. A key issue with the backwash of sunken media systems is the potential for boils during backwashing. The flow will short-circuit through the line of least resistance. This will result in a boil, or violent eruption of the flow through the point of least resistance. Similar short circuits and boils can also occur if the nozzles are blocked. These boils can result in excessive media loss during backwashing. Therefore, to achieve even backwashing the water must be well distributed across the BAF plan area. Therefore, the headloss across the distribution system must be greater than the headloss through the bed. An upflow BAF with floating media, such as the Veolia Biostyrs, is illustrated in Figure 27. These processes use a floating bed of media to provide area for biofilm development and filtration. Coarse-bubble aeration diffusers exist at the bottom of the media to enhance the contact of air, water, and biomass (Rogalla and Bourbigot, 1990). The Biostyrs process uses light weight expanded polystyrene (specific gravity of 0.05). Alternatively, the Biobeads process uses
555
recycled polypropylene with a specific gravity slightly lower than 1. The Biostyrs reactor is partially filled with (2–6 mm) polystyrene beads. Process objectives determine selection of the bead size; larger beads can be more heavily loaded. The beads, which are lighter than water, form a floating bed in the upper portion of the reactor, typically a height of 3–4 m with approximately 1.5 m of freeboard. The top of the bed is restrained by a slab fitted with filtration nozzles to evenly collect treated wastewater. The clean specific surface area of spherical beads is 1000–1400 m2 m3. In the bottom of the reactor, influent is distributed by troughs formed in the base of the cells. Process air is distributed through diffusers located along the bottom of the reactor or within an aeration grid in the media bed. The latter is used if an anoxic zone is required for denitrification. Backwashing consists of counter-current air scour and backwash water flow. The Biobeads BAF process is similar to Biostyrs, except that the media is larger and heavier, using polypropylene or polyethylene with a density of approximately 0.95. To prevent media attrition, a metal grid is fixed near the top of the reactor. Upflow floating media BAFs may also require a certain number of mini-backwashes (typically 4–8 and, in extreme cases, more than 10) to bump the filter, remove some solids, and lower headloss to achieve a complete filtration cycle of 24 or 48 h (which is the time between normal backwashes). The requirement for minibackwashes plus normal backwashes can generate a significant backwash water volume. During demonstration testing in San Diego, California, USA, a single-stage carbon-oxidation BAF with floating media generated a backwash water volume in the range of 10.3–13.9% of influent flow, compared to a sunken media BAF which produced a backwash water volume in the range of 7.4–7.9% (Newman et al., 2005). An upflow continuous backwash BAF, such as the Parkson Dynsands, is illustrated in Figure 28. Moving bed, continuous backwash filters operate in an upflow mode and consist of media heavier than water. The media continuously moves downward, counter-current to the wastewater flow. These filters are used widely for tertiary suspended solids and turbidity Backwash water
Air scour Process air
Air Aerobic filter zone
Treated water
Anoxic filter zone Water
Recirculation pump Raw water
Backwash water extraction Figure 27 Upflow BAF with floating media (e.g., Veolia Biostyrs). Adapted from ATV (1997) Biologische und weitergehende Abwasserreinigung (German), 4th edn. Berlin: Ernst and Sohn as presented by Tschui (1994).
556
Biofilms in Water and Wastewater Treatment Central reject compartment (H)
Feed (influent) (A)
Rejects (L) Top of airlift pump (G) Filtrate weir (J)
Reject weir (K) Sand washer (L) Effluent (E)
Downward moving sand bed (D)
Downward feed (B) Feed radials (C)
Bottom of airlift pump (F) Figure 28 Parkson Dynasands process schematic, continuous backwash BAF.
removal but have also been applied to separate stage nitrification and denitrification. Two commercially available systems using this technology are the Parkson DynaSands and Paques Astrasands filters. The filter cells are supplied as 4.65-m2 modules with center airlift assembly. The effective media depth is typically 2 m, and sand media size typically ranges from approximately 1 to 1.6 mm. Influent wastewater enters the filter bed through radials located at the bottom of the filter. The flow moves up through the downward-moving sand bed and effluent flows over a weir at the top of the filter. The media, with the accumulated solids, is drawn downward to the bottom cone of the filter. Compressed air is introduced through an airlift tube extending to the conical bottom of the filter and rises upward with a velocity exceeding 3 m s1 creating an air pump that lifts the sand at the bottom of the filter through the center column. The turbulent upward flow in the airlift provides scrubbing action that effectively separates solids from the media before discharge to a wash box. There is a constant upward flow of liquid into the wash box (backwash water) controlled by the wash box discharge weir. Moving bed filter manufacturers typically set the reject weir to provide a wash water flow rate equivalent to approximately 10% of the forward flow at an average filter loading rate of 4.9 m h1. The backwash frequency is quantified by the bed turnover rate. To maintain sufficient biomass for denitrification, the bed turnover rate must be reduced to approximately 100–250 mm h1.
Several media types are available for use in BAFs. Media selection is integral to meeting treatment objectives, flow and backwashing regimes. Typically, media can be categorized as mineral media and plastic media. In most cases, mineral media is denser than water and plastic media is buoyant. The media needs to resist breakdown from abrasion during backwashing and chemical degradation by constituents in municipal wastewater. Commercially available BAF systems and their media are listed in Table 3. Backwashing BAFs maximizes solids capture and filter run time. Proper backwashing requires filter bed expansion and rigorous scouring followed by efficient rinsing. Accumulation of solids and media (mud balling) results in wastewater short-circuiting and can result in excessive media loss. Feed characteristics and type of treatment provided by the BAF affect solids production and frequency requirements for backwashing. Biomass yield in tertiary BAF systems is typically low, so backwashing is relatively infrequent (i.e., one backwash per 36–48 h). Reactor characteristics and media type influence backwash frequency. More openly structured media capture fewer solids which reduces backwash frequency. During backwashing the media bed is typically expanded or fluidized (depending on the system) to allow for grain separation and free movement in order to remove as much accumulated solids as possible. Table 4 compares typical BAF backwashing requirements. BAFs designed for carbon oxidation and suspended solids removal in secondary treatment typically have volumetric BOD loading rates in the range of 1.5–6 kg m3 d1. Average and peak HLRs for secondary and tertiary treatment systems are typically in the range of 4–8 and 10–20 m h1, respectively. As BAFs for secondary treatment are typically placed immediately downstream of primary clarifiers, the applied volumetric mass loading rate is almost always the limiting design parameter. Combined carbon oxidation and nitrification will proceed when the organic loading at lower temperatures is limited to 2.5 kg BOD m3 d1 (Rogalla et al., 1990). Under these conditions a total Kjeldahl nitrogen removal rate of 0.4 kg N m3 d1 may be achieved. Inversely, Rogalla et al. (1990) found that nitrification decreases when soluble COD loadings approach 4 kg m3 d1. Ammonium removal of 80– 90% can be achieved for ammonium loads in the range of 2.5–2.9 kg m3 d1 (Peladan et al., 1996).
4.15.3.5.4 Expanded and fluidized bed biofilm reactors Expanded bed biofilm reactors (EBBRs) and FBBRs use small media particles that are suspended in vertically flowing wastewater, so that the media becomes fluidized and the bed expands. Individual particles become suspended once the drag force of the relatively fast flowing wastewater (30–50 m h1) overcomes gravity and they are separated. In municipal applications, fluidized beds are typically used for tertiary denitrification. Design criteria for denitrifying FBBRs are listed in Table 5. When treating groundwater or industrial wastewater, FBBRs are used for the removal of oxidized contaminants such as nitrate and perchlorate. Suspension of the media maximizes the contact surface between microorganisms and wastewater. It also increases treatment efficiency by improving mass transfer because there
Biofilms in Water and Wastewater Treatment Table 3
557
Biologically active filter systems and commercially available media
Process
Supplier
Flow regime
Media
Specific gravity
Size (mm)
Astrasands Biobeads Biocarbones Biofors Biolest Biopur
Paques/Siemens Brightwater F.L.I. OTV/Veolia Degremont Stereau Sulzer/Aker Kvaerner Kruger/Veolia Severn Trent Severn Trent Parkson FB Leopold Severn Trent
Upflowa Upflow Downflow Upflow Upflow Downflow
Sand Polyethylene Expanded shale Expanded clay Pumice/pouzzolane Polyethylene
42.5 0.95 1.6 1.5–1.6 1.2
1–1.6
Upflow Upflow Downflow Upflowa Downflow Up/down
Polystyrene Sand Sand Sand Sand Slag
0.04–0.05 2.6 2.6 2.6 2.6 2–2.5
3.3–5 2–3 2–3 1–1.6 2 28–40
Washed gravel
2.6
19–38
Biostyrs ColoxTM Denites Dynasands Eliminites Submerged activated filter
2–6 2.7, 3.5, and 4.5
Specific surface area (m2 m3)
1400–1600
Structured 1000 656 656
240
a
Moving bed. From Boltz JP, Morgenroth E, deBarbadillo C, et al. (2010b) Biofilm reactor technology and design. In: Design of Municipal Wastewater Treatment Plants, WEF Manual of Practice No. 8, ASCE Manuals and Reports on Engineering Practice No. 76, 5th edn, vol. 2, ch. 13 (ISBN P/N 978-0-07-166360-1 of set 978-0-07-166358-8; MHID P/N 0-07-166360-6 of set 0-07-166358-4). New York: McGraw-Hill.
Table 4
Summary of biologically active filter (BAF) backwashing (BW) requirements
Upflow, sunken media Normal BW Energetic BWa Upflow, floating media Normal BW Mini-BWb Downflow, sunken media Upflow, moving bedf
Backwash water rate, m h1
Air scour rate, m h1
Total duration minc
Total backwash water volume per cellc
Total backwash water volume per celld
20 (8.2) 30 (12.3)
97 (5.3) 97 (5.3)
50 25
9.2 m3 m2 9.2 m3 m2
12 m3 m2 10 m3 m2
55 (22.5) 55 (22.5) 15 (6) 0.5–0.6
12 (0.65) 12 (0.65) 90 (5) Continuous through air lift
16 5 20–25 Continuous
2.5 m3 m3 mediae 1.5 m3 m3 mediae 3.75–5 m3 m2 55–67 m3 d1
2.5 m3 m3 mediae 1.5 m3 m3 mediae 3.75–5 m3 m2 55–67 m3 d1
(0.2–0.24) a
Energetic backwash once every 1–2 months depending on trend in ‘‘clean bed’’ headloss following normal backwash. Mini-backwash applied as interim measure when pollutant load exceeds design load. c Backwash duration reflects total duration of the typical backwash cycle, which includes valve cycle time and pumping and nonpumping steps. The duration of each step is adjustable via programmable logic controller and supervisory control and data acquisition control systems. d The total backwash water volume includes drain and filter to waste steps, where applicable. e Backwash volume requirements for upflow floating media BAF typically are based on media volume rather than cell area because depths vary. f Continuous backwash filter BW is based on a standard 4.65 m2 cell and a typical weir setting for reject flow of approximately 2.3–2.8 m3 h1 cell1. From Boltz JP, Morgenroth E, deBarbadillo C, et al. (2010b) Biofilm reactor technology and design. In: Design of Municipal Wastewater Treatment Plants, WEF Manual of Practice No. 8, ASCE Manuals and Reports on Engineering Practice No. 76, 5th edn, vol. 2, ch. 13 (ISBN P/N 978-0-07-166360-1 of set 978-0-07-166358-8; MHID P/N 0-07-166360-6 of set 0-07-166358-4). New York: McGraw-Hill. b
is significant relative motion between the biofilm and flowing wastewater. Because of the balance of forces involved in particle fluidization and bed expansion, the smallest particles are found at the top and the largest at the bottom of the fluid bed. Therefore, the media particles should be graded to a relatively tight size range. The degree of bed expansion determines whether a bed is deemed expanded or fluidized. The transition lies between 50% and 100% expansion over the static bed height. This discussion assumes the upper limit: beds less than double static bed height
(o100% expanded) are considered expanded; those more than double the static bed height (4100% expanded) are fluidized. A lower degree of bed expansion is advantageous, because it requires a lower flow velocity, less energy, and increases effective biomass concentration, which reduces the reactor footprint. In aerobic processes, however, it increases volumetric oxygen demand because of increased biomass concentration. The FBBR/EBBR is illustrated in Figure 29. The system consists of a column in which the particles are fluidized and a
558
Biofilms in Water and Wastewater Treatment
Table 5
Design criteria for denitrifying fluidized bed biofilm reactors
Parameter
Value
Packing Type Effective size Sphericity Uniformity coefficient Specific gravity Initial depth Bed expansion Empty-bed upflow velocity Hydraulic loading rate Recirculation ratio NO 3 N loading: 13 1C 20 1C Empty-bed contact time C:N (methanol) Specific surface areaa Biomass concentrationa
Unit
Range
Typical
mm Unitless Unitless Unitless m % m h1 m3effluent m2 Bioreactor Unitless
Sand 0.3–0.5 0.8–0.9 1.25–1.50 2.4–2.6 1.5–2.0 75–150 36–42 400–600 2:1–5:1
Sand 0.4 0.8–0.85 r1.4 2.6 2.0 100 36 500 3:1
2.0–4.0 3.0–6.0 10–20 3.0–3.5 1000–3000 15 000–40 000
3.0 5.0 15 3.2 2000 30 000
area
kg m3 d1 kg m3 d1 min Unitless m2 m3 mg l1
d1
a
Specific surface area range based on sand particles; alternate media used in fluidized bed reactors such as carbon or glassy coke may have a different specific surface area range. From Boltz JP, Morgenroth E, deBarbadillo C, et al. (2010b) Biofilm reactor technology and design. In: Design of Municipal Wastewater Treatment Plants, WEF Manual of Practice No. 8, ASCE Manuals and Reports on Engineering Practice No. 76, 5th edn, vol. 2, ch. 13 (ISBN P/N 978-0-07-166360-1 of set 978-0-07-166358-8; MHID P/N 0-07-166360-6 of set 0-07-166358-4). New York: McGraw-Hill.
Figure 29 Fluidized bed biofilm reactor process flow diagram (Shieh and Keenan, 1986).
Process flow enters at the bottom of the reactor and flows through a distribution system to ensure even dispersion and uniform fluidization. Silica sand (0.3–0.7 mm diameter) and granular activated carbon (GAC; 0.6–1.4 mm) are typically used. Other materials, however, have been used at pilot scale, such as 0.7–1.0 mm glassy coke (McQuarrie et al., 2007). Small carrier particles (1 mm) provide a large specific surface area for biofilm growth (up to 2400 m2 m3 when expanded 50%), which is one of the key advantages of this process technology. In a study of tertiary nitrification of activated sludge-settled effluent using a pilot-scale EBBR, Dempsey et al. (2006) found that the process also removed up to 56% CBOD and 62% TSS from the influent stream. Removal of these materials was attributed to the activities of protozoa (free-living and stalked) and metazoa (rotifers, nematodes, and oligochaetes) as shown in Figure 30.
recycle line that is used to maintain a fixed, vertical hydraulic flow. In this way, bed expansion is kept constant and biofilm covered particles are retained independent of influent flow. Aeration typically is achieved during recycle, during which influent wastewater mixes with effluent recycled from the top of the bed. If aeration is conducted within the fluidized bed, then a significant volume of gas disturbs the fluidized state by causing turbulence and increased force of interparticle collisions. This can cleave biofilm from the substratum. Nevertheless, this approach has been used. The advantage of adding air to the recycle stream is that biomass is not stripped from the media by turbulence of rising gas bubbles; therefore, the treated effluent typically has a lower suspended concentration (Jeris et al., 1981).
The RBC process has been applied where average effluent water-quality standards are less than or equal to 30 mg l1 BOD5 and TSS. The RBC employs a cylindrical, synthetic media bundle that is mounted on a horizontal shaft. Figure 31 illustrates the shaft-mounted media. The bundled media is partially submerged (typically 40%) and slowly (1–1.6 rpm) rotates to expose the biofilm to substrate in the bulk of the liquid (when submerged), and to air (when not submerged). Detached biofilm fragments suspended in the RBC effluent stream are removed by liquid– solids separation units. The RBC process is typically configured with several stages operating in series. Each reactorin-series may have one or more shafts. Parallel trains are
Effluent
Excess biomass
Recycle Separator
Influent
Media Reactor Bioparticle
O2 Chemicals (optional) 1 2
1 Medium 2 Biofilm
4.15.3.5.5 Rotating biological contactors
Biofilms in Water and Wastewater Treatment
(a)
(b)
(c)
(d)
(e)
(f)
559
Figure 30 Particulate biofilms with associated protozoa and metazoan from expanded bed: (a) bioparticles in expanded bed; (b) bioparticles with surface attached; (c) closeup of rotifer attached to bioparticle; (d) stalked protozoa on surface of particulate biofilms; (e) testate amoeba grazing on biofilm; and (f) oligochaete worm grazing on bioparticles (Dempsey et al., 2006).
implemented to provide additional surface area for biofilm development. Media-supporting shafts typically are rotated by mechanical drives. Diffused air-drive systems and an array of airentraining cups that are fixed to the periphery of the media (to capture diffused air) have been used to rotate the shafts. RBCs have failed as a result of shaft, media, or media support system structural failure; poor treatment performance; accumulation of nuisance macrofauna; poor biofilm thickness control; and inadequate performance of air-drive systems for shaft rotation. Typically, the RBC tank is sized at 4.9 103 m3 m2 of media for low-density units. Disks typically have a 3.5-m diameter and are situated on a 7.5-m-long rotating shaft. The RBCs may contain low- or high-density media. Low-density media has a 118-m2 m3 biofilm active specific surface; high-density units have 180 m2 m3. Low-density media typically are used in the first stages of RBC systems which are designed for BOD5 removal to reduce potential media clogging and weight problems resulting from substantial biofilm accumulation. High-density media typically is used for nitrification. Mechanical shaft drives consist of an electric motor, speed reducer, and belt or chain drive. Typically, 3.7-kW mechanical drives have been provided for full-scale RBCs. Air-driven shafts require a remote blower for air delivery. Air headers are equipped with coarse-bubble diffusers. The air flow rate is typically in the range of 4.2–11.3 m3 min1 per shaft. Air quantity required by systems using air-driven shaft rotation, however, must be evaluated on a site-specific basis. Mechanical drive units have been designed for operation from 1.2 to 1.6 rpm. Air-drive units have been designed for 1.0–1.4 rpm. Ideally, shaft rotational speed is consistent. The development of an evenly distributed biofilm is desirable to avoid an uneven weight distribution, which may cause cyclical loadings in mechanical-drive systems and loping (uneven rotation) in air-driven shaft rotating systems. A loping condition often
accelerates rotational speed and, if not corrected, may lead to inadequate treatment and the inability to maintain shaft rotation. Air-drive systems should provide ample reserve air supply to maintain rotational speeds, restart stalled shafts, and provide short-term increased speeds (2–4 times normal operation) to control excessive or unbalanced biofilm thicknesses. Available data indicate that in excess of an 11.3-m3 min1 airflow rate per shaft may be required to maintain a 1.2-rpm shaft rotational speed during peak organic loading conditions (Brenner et al., 1984). Large-capacity air cups (150 mm diameter) typically are provided in the first stage of the process to exert a greater torque on the shaft and reduce loping. The RBC process is typically covered to avoid ultraviolet (UV) light-induced media deterioration and algae growth, to prevent excessive cooling, and to provide odor control. RBCs have been installed in buildings or under prefabricated fiberglass-reinforced plastic (FRP) covers (as pictured in Figure 31).
4.15.3.5.6 Trickling filters The TF is a three-phase biofilm reactor with fixed carriers. Wastewater enters the bioreactor through a distribution system, trickles downward over the biofilm surface, and air moves upward or downward in the third phase where it diffuses through the flowing liquid and into the biofilm. TF components generally include an influent water distribution system, containment structure, rock or plastic media, and underdrain and ventilation system. Wastewater treatment using the TF results in a net production of total suspended solids. Therefore, liquid–solids separation is required, and is typically achieved with circular or rectangular secondary clarifiers. The TF process generally includes an influent/ recirculation pump station, the TF(s), and liquid–solids separation unit(s).
560
Biofilms in Water and Wastewater Treatment
Figure 31 Photograph of the Envirexs rotating biological contactor cylindrical synthetic media bundle mounted on a horizontal shaft (a) and rotating biological contactor covers (b). Photographs courtesy Siemens Water Technologies.
Figure 32 (a) Hydraulically driven rotary distributors use variable frequency drive controlled gates that either open or close distributor orifices which adjust with varying pumped flow rates to maintain a constant preset rotational speed. (b) Electrically driven rotary distributor. Photographs courtesy WesTech, Inc.
Primary effluent or screened and degritted wastewater is either pumped or flows by gravity to the TF distribution system. Essentially, there are two types of TF distribution systems: fixed-nozzle and rotary distributors. Because their efficiency is poor, distribution with fixed nozzles should not be used (Harrison and Timpany, 1988). Rotary distributors may be hydraulically or electrically driven. A properly designed rotary distribution system allows for effective media wetting and the intermittent application of wastewater to biofilm carriers. The intermittent application of influent wastewater allows the biofilm to have periods of resting which primarily serves as a process aeration mechanism. Poor media wetting may lead to dry pockets, ineffective treatment zones, and odor. An electrically or modern hydraulically driven rotary distributor
(Figure 32) controls rotational speed independent of the influent wastewater flow rate, and may be used to maintain the desired hydraulic dosing rate. Ideal TF media provides a high specific surface area, low cost, high durability, and high enough porosity to avoid clogging and promote ventilation (Metcalf and Eddy, 2003). TF media types include rock (RO), random (RA) (synthetic), vertical flow (synthetic) (VF), and cross-flow (synthetic) (XF). Both VF and XF media are constructed with smooth and/or corrugated plastic sheets. Another commercially available synthetic media, although not commonly used, is vertically hanging plastic strips. Horizontal redwood or treated wooden slats have also been used, but are generally no longer considered viable because of high cost or limited supply. Modules
Biofilms in Water and Wastewater Treatment
of plastic sheets (i.e., self-supporting VF or XF modules) are used almost exclusively for new and improved TFs, but several TFs with rock media exist, and have proven capable of meeting treatment objectives when properly designed and operated. Table 6 compares the characteristics of some TF media. The higher specific surface area and void space in modular synthetic media allow for higher hydraulic loading, enhanced oxygen transfer, and biofilm thickness control in comparison to rock media. Rock media has, ideally, a 50-mm diameter, but may range in size. Due to structural requirements associated with the large unit weight of rock, rock-media TFs are shallow in comparison to synthetic-media TFs. Their large surface area makes them more susceptible to excessive cooling. Generally, rock media is considered to have a low specific surface area, void space, and high unit weight. Although recirculation is common, the low void ratio in rock-media TFs limits hydraulic application rates. Excessive hydraulic application can result in ponding, limited oxygen transfer, and poor bioreactor performance. Performance of existing rock-media TFs may sometimes be improved by providing mechanical ventilation, solids contact channels, and/or deepened secondary clarifiers that include energy dissipating inlets and flocculator-type feed wells. Grady
Table 6
et al. (1999) suggested that under low organic loading (i.e., o1 kg BOD5 d1 m3) rock- and synthetic-media TFs are capable of equivalent performance. However, as organic loading increases, synthetic-media TFs are less susceptible to operational problems and have reduced potential for plugging. Synthetic TF media has a higher specific surface area and void space, and lower unit weight than rock media. Modular synthetic media is generally manufactured with the following specific surface areas: 223 m2 m3 as high density, 138 m2 m3 as medium density, and 100 m2 m3 as low density. Both VF and XF media are reported to remove BOD5 and NH3–N (Harrison and Daigger, 1987), but sufficient scientific evidence exists to surmise that there is a difference in the treatment efficiency of TFs constructed with XF and VF media even when manufactured with virtually identical specific surface areas. Plastic modules with a specific surface area in the range of 89–102 m2 m3 are well suited for carbon oxidation and combined carbon oxidation and nitrification. Parker et al. (1989) recommended medium-density XF media against the use of high-density XF media in nitrifying TFs. This is supported by observations from a pilot-scale nitrifying TF application data and conclusions of Gujer and Boller (1983, 1984)
Properties of some trickling filter media Nominal size (m)
Bulk density (kg m3)
Specific surface area (m2 m3)
Void space (%)
0.024–0.076
1442
62
50
0.076–0.128
1600
46
60
0.61 0.61 1.22
24–45
100, 138, and 223
95
Vertical flow
0.61 0.61 1.22
24–45
102 and 131
95
Randomb
0.185 ø 0.051 H
27
98
95
Media type Rock River
Slag
Plastica Cross flow
a
561
Material
Manufacturers of modular plastic media: (formerly) BF Goodrich, American Surf-Pac, NSW, Munters, (currently) Brentwood Industries, Jaeger Environmental, and SPX Cooling. Manufacturers of random plastic media: (formerly) NSW Corp. and (currently) Jaeger Environmental.
b
562
Biofilms in Water and Wastewater Treatment
which show lower nitrification (flux) rates for lower-density modular synthetic media. The researchers claim that lower rates occur with high-density media due to the development of dry spots below the flow interruption points (i.e., higherdensity media has more flow interruptions and, therefore, less effective wetting). Using medium-density media also reduces plugging potential. Vertically oriented modular synthetic (VF) media is generally accepted as being ideally suited for highstrength wastewater (perhaps industrial) and high organic loadings such as with a roughing TF. In some cases, XF media has been placed in the top layer to enhance wastewater distribution and VF media comprises the remainder of the TF media. Typically, the top layer of a TF’s modular plastic media is covered with FRP or HDPE grating. The grating protects modular plastic media from deterioration by UV light and potential structural damage that may result from waterinduced load exerted during periods of high-intensity dosing. Figure 33 illustrates a typical TF column and a picture of the grating. Rock and random synthetic media are not self-supporting and require structural support to contain the media within the bioreactor. These containment structures are typically precast or panel-type concrete tanks. When self-supporting media such as plastic modules are used, other materials such as wood, fiberglass, and coated steel have been used as containment structures. The containment structure serves to avoid wastewater splashing, and to provide media support, wind protection, and flood containment. In some cases TF containment structures have been designed to allow flooding of the media, which increases operator flexibility in controlling macrofauna accumulation. The TF underdrain system is designed to meet two objectives: collect treated wastewater for conveyance to downstream unit processes and create a plenum that allows for the transfer of air throughout the TF media (Grady et al., 1999). Clay or concrete underdrain blocks are commonly used for rock-media TFs because of the required structural support. A variety of support systems including concrete piers and reinforced fiberglass plastic grating are used for other media types. The volume created between concrete and media bottom creates the underdrain.
TFs require oxygen to sustain aerobic biochemical transformation processes. The VF of air through the media can be induced mechanically or by natural draft. Natural air ventilation results from a difference in ambient air temperature outside and inside the TF. The temperature causes air to expand when warmed or contract when cooled. The net result is an air-density gradient throughout the TF, and an air front either rises or sinks depending on the differential condition. This rising or sinking action results in a continuous air flow through the bioreactor. Natural ventilation may become unreliable or inadequate in meeting process air requirements when neutral temperature gradients do not produce air movement. Currently, the provision of adequate underdrain and effluent channel sizing to permit free air flow is standard. Passive devices for ventilation include vent stacks on the TF periphery, extensions of underdrains through TFs side walls, ventilating manholes, louvers on the sidewall of the tower near the underdrain, and discharge of TF effluent to the subsequent settling basin in an open channel or partially filled pipes. Drains, channels, and pipes should be sufficiently sized to prevent submergence greater than 50% of their crosssectional area under design hydraulic loading. Ventilating access ports with open grating covers should be installed at both ends of the central collection channel. Large diameter TFs typically have branch channels (to collect the treated wastewater). These branches should also include ventilating manholes or vent stacks installed at the TF periphery. The open area of the slots in the top of the underdrain blocks should not be less than 15% of the TF area. One square meter gross area of open grating in ventilating manholes and vent stacks should be provided for each 23 m2 of TF area. Typically, 0.1 m2 of ventilating area is provided for every 3–4.6 m of TF periphery, and 1–2 m2 of ventilation area in the underdrain area per 1000 m3 of TF media. Another criterion for rockmedia TFs is the provision of a vent area at least equal to 15% of the TF cross-sectional area. Mechanical ventilation enhances and controls air flow with low-pressure fans that continuously circulate air throughout the TF. Therefore, a majority of new and improved TFs use low-pressure fans to mechanically promote air flow. The air flow resulting from natural draft will distribute itself. This will
Figure 33 Skid-resistant (polyethylene or fiberglass-reinforced plastic) grating placed on top of a typical modular plastic media trickling filter column.
Biofilms in Water and Wastewater Treatment
not occur with mechanical ventilation. Pressure loss through synthetic TF media is typically low, often less than 1-mm H2O/ m of TF depth (Grady et al., 1999). The low pressure drop typically results in low fan power requirements (B3–5 kW). The head on the fan is typically less than 1500-mm H2O. Unfortunately, the low pressure drop allows air to rise upward through the TF media without distributing itself across the bioreactor section. Therefore, fans are typically connected to distribution pipes. The air flow distribution piping has openings sized such that air flow through each is equal and air flow distribution is uniform. The pipes typically have a velocity in the range of 1100–2200 m h1 in order to further promote uniform air flow distribution. Air flow requirements are calculated based on process oxygen requirements and characteristic oxygen-transfer efficiency which is typically in the range of 2–10%. The mechanical air stream may flow upward to downward. Down-flow systems can be designed without covers. However, covers are required for systems that do not have air distribution through a network of pipes under the media. Covering TFs offers a wintertime benefit of limiting cold airflow and minimizing wastewater cooling. Mechanical ventilation and covered TFs may be used to destroy odorous compounds. A critical unit in the TF process is the pump station that lifts primary effluent (or screened raw sewage), and recirculates unsettled trickling effluent (here, referred to as underflow) to the influent stream. In general, TF underflow is recirculated to the distribution system to achieve the hydraulic load (influent þ recirculation) required for proper media wetting and biofilm thickness control, and decouple hydraulic and organic loading. TF influent is generally pumped to allow TF underflow to flow by gravity to the suspended growth reactor (or solids contact basin), secondary clarifier, or other downstream of the TF. When fit with weirs, a single pump station can be used to convey both influent and recirculation streams.
Table 7
563
TFs can be classified as roughing, carbon oxidation, carbon oxidation and nitrification, and nitrification. Table 7 summarizes characteristics of each TF. The performance ranges are associated with average design conditions. Single day or average week observations may significantly be greater.
4.15.4 Part III. Undesirable Biofilms: Examples of Biofilm-Related Problems in the Water and Wastewater Industries Biofilms are unavoidably associated with water environments, so biofilm control, a component of many industrial processes, is especially important in water and wastewater treatment. Depending on the particular setting, biofilms may cause process performance problems, material performance problems, health problems, and esthetic problems. The specific problems that biofilms cause in industrial settings are as diverse as the technological processes affected by the biofilms. In this section, we discuss four biofilm-related problems that have been reported in the water and wastewater industries: 1. biofilms on metal surfaces and MIC; 2. biofilms on concrete surfaces and crown corrosion of sewers; 3. biofilms on filtration membranes in drinking water treatment; and 4. biofilms on filtration membranes in wastewater treatment.
4.15.4.1 Biofilms on Metal Surfaces and MIC In the manufacturing of metals and metal alloys, raw materials – the ores – are chemically reduced and their internal chemical energy increases. These materials are used by microorganisms as sources of energy in a sequence of processes in which the chemical energy of the affected material decreases, bringing the energy levels of the products closer to
Trickling filter classification
Design parameter
Roughinga
Carbon oxidizing (cBOD5 removal)a
Carbon oxidation and nitrificationa
Nitrificationa
Media typically used
VF
RO, XF, or VF
RO, XF, or VF
XF
Wastewater source
Primary effluent
Primary effluent
Primary effluent
Secondary effluent
Hydraulic loading m3 d1 m2 BOD5 and NH3 N Load kg m3 d1 g m2 d1
52.8–178.2
14.7–88.0
14.7–88.0
35.2–88.0
1.6–3.52 NA
0.32–0.96 NA
0.08–0.24 0.2–1.0
NA 0.5–2.4
Conversion (%) or effluent concentration (mg l1) Macro fauna
50–75% filtered cBOD5 conversion No appreciable growth
20–30 mg l1 cBOD5 and TSSb Beneficial
0.5–3 mg l1 as NH3 Nb
Depth, m (ft)
0.91–6.10
r12.2
o10 mg l1 as cBOD5; o3 mg l1 as NH3 Nb Detrimental (nitrifying biofilm) r12.2
a
Detrimental r12.2
Applicable to shallow trickling filters. gpm ft2, gallons per minute per square foot of trickling filter plan area. Concentration remaining in the clarifier effluent stream. From Boltz JP, Morgenroth E, deBarbadillo C, et al. (2010b) Biofilm reactor technology and design. In: Design of Municipal Wastewater Treatment Plants, WEF Manual of Practice No. 8, ASCE Manuals and Reports on Engineering Practice No. 76, 5th edn, vol. 2, ch. 13, p. 238 (ISBN P/N 978-0-07-166360-1 of set 978-0-07-166358-8; MHID P/N 0-07166360-6 of set 0-07-166358-4). New York: McGraw-Hill. b
564
Biofilms in Water and Wastewater Treatment
the energy levels of the materials from which they were made. MIC can affect a variety of materials, both metallic and nonmetallic. If nonmetallic materials are affected, the term biodeterioration of materials is more often used than MIC, although this terminology is not very consistent and, for example, the term crown corrosion of sewers, which in fact refers to the biodeterioration of concrete, is quite popular among water professionals. Accelerated corrosion of metals in the presence of microorganisms stems from microbial modifications to the chemical environment near metal surfaces (Beech et al., 2005; Geiser et al., 2002; Lee and Newman, 2003; Lewandowski et al., 1997). Such modifications depend, of course, on the properties of the corroding metal and on the microbial community structure of the biofilm deposited on the metal surface (Beech and Sunner, 2004; Dickinson et al., 1996; Flemming, 1995; Olesen et al., 2000, 2001). Beech et al. (2005) described MIC as a consequence of coupled biological and abiotic electron transfer reactions, that is, redox reactions of metals enabled by microbial ecology (Beech et al., 2005). Hamilton (2003) attempted to generate a unified concept of MIC but found common features in only some of the possible mechanisms (Hamilton, 2003). It is unlikely that a unified concept of MIC can be generated at all. Rather, there are many mechanisms by which microorganisms may affect metal surfaces, and we demonstrate some of them here. These do not exhaust the possibilities, of course, but are rather used to exemplify the possible mechanisms. As we have restricted the discussion of MIC to metal surfaces only, it is convenient to define corrosion as anodic dissolution of a metal. In this way we can easily separate the corrosion reaction, the anodic dissolution of the metal, from many other anodic reactions that can occur at a metal surface covered with a biofilm. These other anodic reactions deliver electrons originating from substances metabolized near the metal surface, but only the reaction in which the metal itself is oxidized is defined as corrosion. The presence of the other anodic reactions causes confusion in MIC studies, as the current between the anode and the cathode is made up of electrons originating from many anodic reactions occurring at the surface, not only from the corrosion reaction. Microorganisms generate chemical environments that are conducive to corrosion reactions even if they do not take part in the process themselves. As in most industrial processes microorganisms are always present on metal surfaces, it is not immediately obvious whether the microorganisms attached to the surface accelerate the corrosion process or are just innocent bystanders. The only way to resolve this is by demonstrating that a specific mechanism of MIC is present because a product of microbial metabolism consistent with this mechanism can be detected. Many mechanisms of MIC have been proposed. Accelerated corrosion may result from the action of acid-producing bacteria, such as Thiobacillus thiooxidans and Clostridium aceticum; iron-oxidizing bacteria, such as Gallionella, Sphaerotilus, and Leptothrix; MOB, such as L. discophora; or hydrogenproducing bacteria. These mechanisms have been studied and the results described in numerous publications. We describe here representative examples of such mechanisms: the effects of differential aeration cells, sulfate-reducing bacteria (SRB corrosion), and MOB corrosion.
4.15.4.1.1 Differential aeration cells on iron surfaces MIC caused by differential aeration cells is an example of a nonspecific mechanism of MIC, because it depends on the presence of biofilm, and not on the type of microorganisms that reside in the biofilm. If the oxygen concentrations at two adjacent locations on an iron surface are different, then the half-cell potentials at these locations are different as well. The location where the oxygen concentration is higher will have a higher potential (more cathodic) than the location where the oxygen concentration is lower (more anodic). The difference in potential will give rise to a current flow from the anodic locations to the cathodic locations and to the establishment of a corrosion cell. This is the mechanism of differential aeration cells, and the prerequisite to this mechanism is that the concentration of oxygen varies among locations (Acuna et al., 2006; Dickinson and Lewandowski, 1996; Hossain and Das, 2005). Indeed, many measurements using oxygen microsensors have demonstrated that oxygen concentrations in biofilms can vary from one location to another (Lewandowski and Beyenal, 2007). If the anodic reaction is the oxidation of iron,
Fe-Fe 2þ þ 2e
ð17Þ
and the cathodic reaction is the reduction of oxygen,
O2 þ 2H2 O þ 4e -4OH
ð18Þ
then the overall reaction describing the process is
2Fe þ O2 þ 2H2 O-2Fe2þ þ 4OH
ð19Þ
The Nernst equation quantifying the potential for this reaction is
E ¼ Eo
0:059 ½Fe 2þ 2 ½OH 4 log 4 pðO2 Þ
ð20Þ
Figure 34 visualizes this mechanism.
4.15.4.1.2 SRB corrosion SRB causes corrosion of cast iron, carbon, and low alloy steels and stainless steels. SRB corrosion of potable water mains is a common (US EPA, 1984) and well-recognized problem (Seth and Edyvean, 2006; Tuovinen et al., 1980). MIC caused by SRB is an example of a mechanism that depends on the activity of a specific group of microorganisms in a biofilm. The corrosion of mild steel caused by SRB is the most notorious case of MIC, and it provides a direct and easy-to-understand link between microbial reactions and electrochemistry (Javaherdashti, 1999). According to the mechanism that was originally proposed by Von Wohlzogen Kuhr in 1934, SRB oxidizes cathodically generated hydrogen to reduce sulfate ions to H2S, thereby removing the product of the cathodic reaction and stimulating the progress of the reaction (Al Darbi et al., 2005). This mechanism was later found to be inadequate to explain the field observations. More involved mechanisms were implicated in this type of microbial corrosion, including the puzzling effect of oxygen, which can stimulate what is apparently an anaerobic process. It is now certain that the
Biofilms in Water and Wastewater Treatment
565
Aerated water Cathodic site; corrosion products
Biofilm
OH−
OH−
Cathode − e
Biof ilm
O2
e−
Anodic site
Biofilm
O2
O O2 Aerobic 2 O2 O2 O2 Anaerobic O2 O O2 2 O2 O2 Anaerobic O 2 O2 O2 O2 M+ M+ M+ Anode
O2 O2 OH−OH− e−
Cathode e−
1 mm
Metal (b)
(a)
Figure 34 Biofilm heterogeneity results in differential aeration cells. (a) This schematic shows pit initiation due to oxygen depletion under a biofilm (Borenstein, 1994). (b) An anodic site and pit under the biofilm and corrosion products deposited on mild steel.
possible pathways for cathodic reactions include sulfides and bisulfides as cathodic reactants (Videla, 2001; Videla and Herrera, 2005). The currently accepted mechanism of SRB corrosion is composed of a network of reactions that reflects the complexity of the environment near corroding metal surfaces covered with biofilms; the following paragraphs illustrate some of this complexity. The process starts with the microbial metabolism of SRB producing hydrogen sulfide by reducing sulfate ions. Hydrogen sulfide can serve as a cathodic reactant, thus affecting the rate of corrosion (Antony et al., 2007; Costello, 1974):
2H2 S þ 2e -H2 þ 2HS
ð21Þ
Ferrous iron generated from anodic corrosion sites precipitates with the metabolic product of microbial metabolism, hydrogen sulfide, forming iron sulfides, FeSx:
Fe 2þ þ HS ¼ FeS þ H þ
ð22Þ
This reaction may provide protons for the cathodic reaction (Crolet, 1992). The precipitated iron sulfides form a galvanic couple with the base metal. For corrosion to occur, the iron sulfides must have electrical contact with the bare steel surface. Once contact is established, the mild steel behaves as an anode and electrons are conducted from the metal through the iron sulfide to the interface between the sulfide deposits and water, where they are used in a cathodic reaction. Surprisingly, the most notorious cases of SRB corrosion often occur in the presence of oxygen. As SRB is anaerobic microorganisms, this fact has been difficult to explain. This effect of oxygen can be explained based on a mechanism in which iron sulfides (resulting from the reaction between iron ions and sulfide and bisulfide ions) are oxidized by oxygen to elemental sulfur, which is known to be a strong corrosion agent (Lee et al., 1995). Biofilm heterogeneity plays an important role in this process, because the central parts of microcolonies are anaerobic while the outside edges remain aerobic
(Lewandowski and Beyenal, 2007). This arrangement makes this mechanism of microbial corrosion possible, because the oxidation of iron sulfides produces highly corrosive elemental sulfur, as illustrated by the following reaction:
2H2 O þ 4FeS þ 3O2 -4So þ 4FeOðOHÞ
ð23Þ
Hydrogen sulfide can also react with the oxidized iron to form ferrous sulfide and elemental sulfur (Schmitt, 1991), thereby aggravating the situation by producing even more elemental sulfur, and closing the loop through production of the reactant used in the first reaction, FeS:
3H2 S þ 2FeOðOHÞ-2FeS þ So þ 4H2 O
ð24Þ
The product of these reactions – elemental sulfur – increases the corrosion rate. Schmitt (1991) has shown that the corrosion rate caused by elemental sulfur can reach several hundred mpy (Schmitt, 1991). We have demonstrated experimentally that elemental sulfur is deposited in the biofilm during SRB corrosion (Nielsen et al., 1993), thereby detecting the component vital for this mechanism to occur. It is also well known that the sulfur disproportionation reaction that produces sulfuric acid and hydrogen sulfide is carried out by sulfur-disproportionating microorganisms (Finster et al., 1998). Also, several microbial species, such as T. thiooxidans, can oxidize elemental sulfur and sulfur compounds and produce sulfuric acid:
4S o þ 4H2 O-3H2 S þ H2 SO4
ð25Þ
In summary, the SRB corrosion of mild steel in the presence of oxygen is an acid corrosion: Anodic reaction:
Fe-Fe 2þ þ 2e
ð26Þ
2H þ þ 2e-H2
ð27Þ
Cathodic reaction:
566
Biofilms in Water and Wastewater Treatment 4.15.4.3 Biofilms on Filtration Membranes in Drinking Water Treatment
O2
2−
FeO(OH)
SO4
SO42−
SO42− H2S S
0
FeS2 FeS
H+
O2
Fe2+ Metal
S0
HS− H2
O2
e
Figure 35 The SRB corrosion of mild steel in the presence of oxygen is an acid corrosion (Lewandowski et al., 1997).
The mechanism of SRB corrosion involves several loops, cycles in which reactants are consumed in one reaction and recycled in other reaction; the process is spontaneous at the expense of the energy released by the oxidation of the metal. This mechanism also demonstrates how the reactants and products of corrosion processes are included in the metabolic reactions of the microorganisms. For example, hydrogen, the product of the cathodic reaction above, is oxidized by some species of SRB to reduce sulfate and generate hydrogen sulfide, H2S (Cord-Ruwisch and Widdel, 1986), which is the reactant in the first reaction we referred to in this section. Hydrogen sulfide then dissociates to bisulfides:
H2 S ¼ Hþ þ HS
ð28Þ
which are then used in the reactions described above. Figure 35 shows the network of reactions described above.
4.15.4.2 Biofilms on Concrete Surfaces: Crown Corrosion of Sewers The mechanism of crown corrosion of sewers is very similar to the mechanism of MIC corrosion of metals caused by SRB. In sewers, SRB reduces sulfate ions to sulfides, which are oxidized by oxygen to elemental sulfur. Then the elemental sulfur is further oxidized, mainly by T. thiooxidans, but also by other Thiobacillus species, such as T. novellus/intermedius and T. neapolitanus, in a complex ecosystem on the sewer pipe (Vincke et al., 2001). As a result, sulfuric acid is produced, which dissolves the concrete and damages the sewers (Padival et al., 1995; Islander et al., 1991; Sand and Bock, 1984). The following reactions illustrate this action:
H2 SO4 þ CaCO3 -CaSO4 þ H2 CO3
ð29Þ
H2 SO4 þ CaðOHÞ2 -CaSO4 þ 2H2 O
ð30Þ
Crown corrosion of sewers depends on the presence of biofilm on the concrete surface and on the generation of sulfuric acid in immediate proximity to the concrete surface.
The common use of membranes in various technologies of water and wastewater treatment is probably the most visible mark of the changes that occurred in these applications in the last decade, and it is expected that filtration membranes will be even more popular in the future than they are now (Shannon et al., 2008). The traditional use of membranes in water treatment has been in the desalination of sea and brackish waters using the reverse osmosis (RO) process, and there is a large body of knowledge accumulated on this application. RO membrane filtration is becoming even more popular as the cost of desalination decreases because of various improvements in the technology that reduce the energy consumption and because of the use of new materials that produce less expensive and more robust membranes (Veerapaneni et al., 2007). Membrane processes have been introduced into other types of water treatment, besides desalination, such as water softening (Conlon et al., 1990). The main advantages of using membrane filtration in water treatment are that the process does not require using chemicals and that the membrane modules have a smaller footprint than the conventional treatment facilities. Membrane filtration can be used instead of other traditional processes in water treatment, such as coagulation, sand and activated carbon filtration, or ion exchange, without the necessity of adding chemicals to the water, which helps prevent the formation of disinfection byproducts, for example. Membrane filtration can be used alone in water treatment or in combination with other processes, in hybrid arrangements. For example, it can be used in combination with powdered activated carbon (PAC) to remove disinfection byproducts that exist in the raw water (Khan et al., 2009). Excessive biofouling of membranes is a problem in all membrane applications, but RO and nanofiltration (NF) processes are the most sensitive to biofouling (Vrouwenveldera et al., 2009). Much research has been done toward understanding the process of biofilm formation on these membranes and developing methods for cleaning the membranes. The removal of biofilm from RO membranes can be accomplished by mechanical or by chemical methods, or by a combination of mechanical and chemical methods. Mechanical methods include flushing with water or with water and air. Mechanical cleaning can be used alone or it can be followed by chemical cleaning. The simplest method of mechanical cleaning is the forward flush, in which the water flow rate above the membrane is increased to increase the shearing force and remove the deposits from the membrane. To increase the shearing force even further, air can be introduced into the conduit delivering the cleaning water. The air bubbles introduce additional instability into the flow field and increase the shearing force exerted on the surface. The backward flush is based on reversing the direction of filtration: cleaning water is filtered in the opposite direction and the particles trapped in membrane pores are removed. Depending on the contaminants deposited on the membranes, the surface can be cleaned chemically using various type of chemicals. If the deposits are predominantly inorganic scale, then the chemical cleaning can include agents that act mostly on scale, such as hydrochloric acid (HCl) or nitric acid (HNO3). If the
Biofilms in Water and Wastewater Treatment
biofilm is the main problem, then the cleaning substance may include antimicrobial agents to remove the biofilms. Two types of antimicrobial agents are in common use for this purpose: oxidizing and nonoxidizing biocides. The oxidizing biocides popular in membrane cleaning processes include chlorine, bromine, chloramine, chlorine dioxide, hydrogen peroxide, peroxyacetic acid, and ozone. Nonoxidizing biocides include formaldehyde, glutaraldehyde, and quaternary ammonium compounds. One recent study targeted cell–cell communications in biofilms to develop a novel approach in controlling membrane fouling (Yeon et al., 2009). Much effort has been directed toward the development of membranes with new or modified materials that can resist biofouling and toward modifying the surfaces of ultrafiltration (UF) and NF membranes by the graft polymerization of hydrophilic monomers that resist biofouling or allow more aggressive chemical treatment of the membranes (Hester et al., 2002; Wang et al., 2005; Asatekin et al., 2006, 2007). According to recent studies, in spiral-wound membrane modules, biofilm accumulation has a major impact on the spacer channel but the actual fouling of the membrane contributes to the overall pressure drop to a much smaller extent than previously assumed (Vrouwenveldera et al., 2009).
4.15.4.4 Biofilms on Filtration Membranes in Wastewater Treatment Membrane filtration is used in two types of wastewater technologies: (1) membrane bioreactors (MBRs) and (2) membrane biofilm reactors (MBfRs). This terminology is somewhat confusing: the names sound similar, and the fact that the obvious acronyms for the two technologies are the same does not help. It is therefore customary to call the MBRs and the MBfRs. From the biofouling point of view, microbial growth on membranes is undesirable (Le-Clech et al., 2006) while in MBfRs biofilm growth on the membrane is necessary for process performance. MBfRs are used to deliver dissolved gases, such as oxygen, hydrogen, and methane, to the microorganisms attached to the membrane (Brindle and Stephenson, 1996; Brindle et al., 1998; Suzuki et al., 2000; Lee and Rittmann, 2000; Pankhania et al., 1999; Modin et al., 2008). MBRs are used to replace gravity settling in the secondary sedimentation tanks used in traditional biological wastewater treatment; for example, the activated sludge process where membrane processes can be used to separate the biomass of suspended microorganisms from the effluent. The membranes used in MBRs are typically UF membranes. MBR technology is well established in wastewater treatment: it has been implemented on large scales (Melin et al., 2006), and textbooks have been published describing its application (Stephenson et al., 2000; Judd, 2006). Using membrane filtration to replace gravity settling has many advantages, and one of them is avoidance of the notorious problems with sludge bulking that plague many activated sludge treatment plants. Membranes in MBRs suffer from biofouling, which decreases the permeate flow (Howell et al., 2003; Young et al., 2006; Kimura et al., 2005) Large-scale operations suffer from this problem, particularly the irreversible fouling that cleaning does not remove (Wang et al., 2005). The most common solution to the excessive accumulation of biomass is bubbling air near the
567
membrane’s surface, which creates high shear and removes the biomass (Hong et al., 2002). Basic studies on biofilm formation (Davies et al., 1998) indicate that bacteria regulate their group behaviors, such as biofilm formation, in response to population density using small signal molecules called autoinducers, or quorumsensing molecules. It is expected that interference with microbial communication systems in biofilms may lead to novel approaches to preventing biofouling in many areas. Three strategies for interfering with autoinducer molecules have been proposed: blockage of autoinducer production, interference with signal receptors, and inactivation of autoinducer molecules (Rassmusen and Givskov, 2006). In a recent study, Yeon et al. (2009) demonstrated that inactivating the autoinducer molecules in a batch-type MBR reactor decreased the amount of EPS deposited on the membrane and that interfering with cell–cell communication in biofilms can alleviate the fouling of filtration membranes.
References Acuna N, Ortega-Morales BO, and Valadez-Gonzalez A (2006) Biofilm colonization dynamics and its influence on the corrosion resistance of austenitic UNSS31603 stainless steel exposed to Gulf of Mexico seawater. Marine Biotechnology 8: 62--70. Al Darbi MM, Agha K, and Islam MR (2005) Comprehensive modelling of the pitting biocorrosion of steel. Canadian Journal of Chemical Engineering 83: 872--881. Angathevar-Veluchamy RR (2006) Structure and Activity of Pseudomonas aeruginosa PAO1 Biofilms. Thesis Defense, Montana State University, Library, Bozeman, MT. Antoine RL (1976) Fixed Biological Surfaces – Wastewater Treatment. Cleveland, OH: CRC Press. Antony PJ, Chongdar S, Kumar P, and Raman R (2007) Corrosion of 2205 duplex stainless steel in chloride medium containing sulfate-reducing bacteria. Electrochimica Acta 52: 3985--3994. Asatekin A, Menniti A, Kang S, Elimelech M, Morgenroth E, and Mayes AM (2006) Antifouling nanofiltration membranes for membrane bioreactors from selfassembling graft copolymers. Journal of Membrane Science 285: 81--89. Atkinson B and Davies IJ (1974) The overall rate of substrate uptake (reaction) by microbial films. Part I: A biological rate equation. Transactions of the Institution of Chemical Engineers 52: 248--259. Bakke R, Kommedal R, and Kalvenes S (2001) Quantification of biofilm accumulation by an optical approach. Journal of Microbiological Methods 44: 13--26. Bakke R and Olsson PQ (1986) Biofilm thickness measurement by light microscopy. Journal of Microbiological Methods 5: 93--98. Beech IB, Sunner JA, and Hiraoka (2005) Microbe-surface interactions in biofouling and biocorrosion processes. International Microbiology 8: 157--168. Beech WB and Sunner J (2004) Biocorrosion: Towards understanding interactions between biofilms and metals. Current Opinion in Biotechnology 15: 181--186. Beyenal H, Tanyolac A, and Lewandowski Z (1998) Measurement of local effective diffusivity in heterogeneous biofilms. Water Science and Technology 38: 171--178. Beyenal H and Lewandowski Z (2005) Modeling mass transport and microbial activity in stratified biofilms. Chemical Engineering Science 60: 4337--4348. Beyenal H and Lewandowski Z (2000) Combined effect of substrate concentration and flow velocity on effective diffusivity in biofilms. Water Research 34(2): 528--538. Beyenal H, Sani RK, Peyton B, Dohnalkova AC, Amonette J, and Lewandowski Z (2004) Uranium immobilization by sulfate reducing biofilms. Environmental Science and Technology 38: 2067--2074. Biesterfeld S, Figueroa L, Hernandez M, and Russell P (2001) Quantification of nitrifying bacterial populations in a full-scale nitrifying trickling filter using fluorescent in situ hybridization. Water Environment Research 73: 329--338. Bilyk K, Taka´cs I, Rohrbacher J, Pitt P, Latimer R, and Dold P (2008) Full-scale testing advances fundamental understanding of denitrification filters. Proceedings of the 81st Annual Water Environment Federation Technical Exhibition and Conference (WEFTEC). Chicago, IL. Bishop PL (2003) The effect of biofilm heterogeneity on metabolic processes. In: Wuertz S, Wilderer PA, and Bishop PL (eds.) Biofilms in Wastewater Treatment. London: IWA Publishing.
568
Biofilms in Water and Wastewater Treatment
Boltz JP and Daigger GT (2010) Uncertainty in bulk-liquid hydrodynamics creates uncertainties in biofilm reactor design. Water Science and Technology 61(2): 307. Boltz JP, Johnson BR, Daigger GT, and Sandino J (2009a) Modeling integrated fixed film activated sludge (IFAS) and moving bed biofilm reactor (MBBR) systems I: Mathematical treatment and model development. Water Environment Research 81: 576--586. Boltz JP, Johnson BR, Daigger GT, Sandino J, and Elenter D (2009b) Modeling integrated fixed film activated sludge (IFAS) and moving bed biofilm reactor (MBBR) systems II: Evaluation. Water Environment Research 81: 555--575. Boltz JP and La Motta EJ (2007) The kinetics of particulate organic matter removal as a response to bioflocculation in aerobic biofilm reactors. Water Environment Research 79: 725--735. Boltz JP, Morgenroth E, and Sen D (2010a) Mathematical modelling of biofilms and biofilm reactors for engineering design. Water Science and Technology (in press). Boltz JP, Morgenroth E, deBarbadillo C, et al. (2010b) Biofilm reactor technology and design. In: Design of Municipal Wastewater Treatment Plants, WEF Manual of Practice No. 8, ASCE Manuals and Reports on Engineering Practice No. 76, 5th edn, vol. 2, ch. 13 (ISBN P/N 978-0-07-166360-1 of set 978-0-07-166358-8; MHID P/N 0-07-166360-6 of set 0-07-166358-4). New York: McGraw-Hill. Borenstein SB (1994) Microbiologically Influenced Corrosion Handbook. New York: Industrial Press. Borenstein W and Licina GJ (1994) An overview on monitoring techniques for the study of microbiologically influenced corrosion. In: CORROSION’94, Paper 611. Houston, TX, 1994. Houston, TX: NACE International. Bremer PJ and Geesey GG (1991) An evaluation of biofilm development utilizing nondestructive attenuated total reflectance Fourier transform infrared spectroscopy. Biofouling 3: 89--100. Brenner RC, Heidman JA, Opatken EJ, Petrasek AC (1984) Design Information on Rotating Biological Contactors, EPA–600/2–84–106. Washington, DC: US Environmental Protection Agency. Brindle K, Stephenson T, and Semmens MJ (1998) Nitrification and oxygen utilization in a membrane aeration bioreactor. Journal of Membrane Science 144: 197--209. Brockmann D, Rosenwinkel K-H, and Morgenroth E (2008) Practical identifiably of biokinetic parameters of a model describing two-step nitrification in biofilms. Biotechnology and Bioengineering 101(3): 497--514. Campbell SC (2003) Influence of the Distribution of the Manganese Oxidizing Bacterium Leptothrix discophora, on Ennoblement of 316L Stainless Steel. MS Thesis, Montana State University, Library, Bozeman, MT. Characklis WG (1981) Fouling biofilm development: A process analysis. Biotechnology and Bioengineering 23: 1923--1960. Cloete TE and Maluleke MR (2005) The use of the rotoscope as an online, real-time, non-destructive biofilm monitor. Water Science and Technology 52: 211--216. Conlon WJ, Hornburg CD, Watson BM, and Kiefer CA (1990) Membrane softening: The concept and its application to municipal water supply. Desalination 78: 157--175. Corbett J (1903) A Dozen years of sewage purification experiments on a large scale at Salford, England. Engineering News Record 49: 191--192. Cord-Ruwisch R and Widdel F (1986) Corroding iron as a hydrogen source for sulphate reduction in growing cultures of sulphate-reducing bacteria. Applied Microbiology and Biotechnology 25: 169--174. Costello JA (1974) Cathodic depolarization by sulfate-reducing bacteria. South African Journal of Science 70: 202--204. Crolet JL (1992) From biology and corrosion to biocorrosion. Oceanologica Acta 15: 87--94. Davies DG, Parsek MR, Pearson JP, Iglewski BH, Costerton JW, and Greenberg EP (1998) The involvement of cell-to-cell signals in the development of a bacterial biofilm. Science 280: 295--298. DeBeer D, Stoodley P, and Lewandowski Z (1994) Liquid flow in heterogeneous biofilms. Biotechnology and Bioengineering 44: 636--641. Delong EF, Taylor LT, Marsh TL, and Preston CM (1999) Visualization and enumeration of marine planktonic archaea and bacteria by using polyribonucleotide probes and fluorescent in situ hybridization. Applied and Environmental Microbiology 65: 5554--5563. Dempsey MJ, Porto I, Mustafa M, Rowan AK, Brown A, and Head IM (2006) The expanded bed biofilter: Combined nitrification, solids destruction, and removal of bacteria. Water Science and Technology 54(8): 37--46. Dickinson WH and Lewandowski Z (1996) Manganese biofouling and the corrosion behavior of stainless steel. Biofouling 10: 79--93. Dickinson WH, Lewandowski Z, and Geer RD (1996) Evidence for surface changes during ennoblement of type 316L stainless steel: Dissolved oxidant and capacitance measurements. Corrosion 52: 910--920.
Eberl HJ, Picioreanu C, Heijnen JJ, and van Loosdrecht MCM (2000) A threedimensional numerical study on the correlation of spatial structure, hydrodynamic conditions, and mass transfer and conversion on biofilms. Chemical Engineering Science 55: 6209--6222. Fan LS, Leyvaramos R, Wisecarver KD, and Zehner BJ (1990) Diffusion of phenol through a biofilm grown on activated carbon particles in a draft-tube 3-phase fluidized-bed bioreactor. Biotechnology and Bioengineering 35: 279–286. Finster K, Liesack W, and Thamdrup B (1998) Elemental sulfur and thiosulfate disproportionation by Desulfocapsa sulfoexigens sp. nov., a new anaerobic bacterium isolated from marine surface sediment. Applied and Environmental Microbiology 64: 119--125. Flemming HC (1995) Biofouling and biocorrosion-effects of undesired biofilms. Chemie Ingenieur Technik 67: 1425--1430. Geiser M, Avci R, and Lewandowski Z (2002) Microbially initiated pitting on 316 L stainless steel. International Biodeterioration and Biodegradation 49: 235--243. Grady LE, Daigger GT, and Lim H (1999) Biological Wastewater Treatment, 2nd edn. New York: Dekker. GrayMerod R, Hendrickx L, Mueller LN, Xavier JB, and Wuertz S (2005) Effect of nucleic acid stain Syto9 on nascent biofilm architecture of Acinetobacter sp BD413. Water Science and Technology 52: 195--202. Groenenboom M (2000) Increase of frictional resistance in closed conduit systems fouled with biofilms. MS Dissertation. Montana State University. Bozeman Gujer W and Boller M (1983) Operating experience with plastic media tertiary trickling filters for nitrification. In: de Emde V and Tench HB (eds.) Design and Operation of Large Treatment Plants. Oxford: Pergamon. Gujer W and Boller M (1984) Operating experience with plastic media tertiary trickling filters for nitrification. Water Science and Technology 16: 201--213. Gujer W and Boller M (1986) Design of a nitrifying trickling filter based on theoretical concepts. Water Research 20: 1353. Hamilton WA (2003) Microbially influenced corrosion as a model system for the study of metal microbe interactions: A unifying electron transfer hypothesis. Biofouling 19: 65--76. Harremo¨es P and Wilderer PA (1993) Fundamentals of nutrient removal in biofilters. In: Proceedings from the 9th Annual EWPCA–ISWA Symposium. Mu¨nchen, Germany, 11–13 May. Abwassertechnische Vereinigung e.V. Harrison JR and Daigger GT (1987) A comparison of trickling filter media. Journal WPCF 59: 679. Harrison JR and Timpany PL (1988) Design considerations with the trickling filter solids contact process. In: Proceedings of the Joint Canadian Society of Civil Engineers, ASCE National Conference on Environmental Engineering, pp. 753– 762. Vancouver, British Columbia. Hem L, Rusten B, and Ødegaard H (1994) Nitrification in a moving bed reactor. Water Research 28(6): 1425--1433. Henze M, Gujer W, Mino T, et al. (2000) Activated Sludge Model No. 2d. In: Henze M, Gujer W, Mino T, van Loosdrecht MCM (eds.) Activated Sludge Models ASM1, ASM2, ASM2d, and ASM3. Scientific and Technical Report No. 9. London: IWA Publishing. Hong SP, Bae T, Tak T, Hong S, and Randall A (2002) Fouling control in activated sludge submerged hollow fiber membrane bioreactors. Desalination 143: 219--228. Hossain MA and Das CR (2005) Kinetic and thermodynamic studies of microbial corrosion of mild steel specimen in marine environment. Journal of the Indian Chemical Society 82: 376--378. Howell JA, Arnot TC, and Liu W (2003) Membrane bioreactors for treating waste streams. Advanced Membrane Technology 984: 411--419. Islander RL, Devinny JS, Mansfeld F, Postyn A, and Shih H (1991) Microbial ecology of crown corrosion in sewers. Journal of Environmental Engineering 117: 751--770. Ito T, Nielsen JL, Okabe S, Watanabe Y, and Nielsen PH (2002) Phylogenetic identification and substrate uptake patterns of sulfate-reducing bacteria inhabiting an oxic-anoxic sewer biofilm determined by combining microautoradiography and fluorescent in situ hybridization. Applied and Environmental Microbiology 68: 356--364. Jang A, Okabe S, Watanabe Y, Kim IS, and Bishop PL (2005) Measurement of growth rate of ammonia oxidizing bacteria in partially submerged rotating biological contactor by fluorescent in situ hybridization (FISH). Journal of Environmental Engineering and Science 4: 413--420. Javaherdashti R (1999) A review of some characteristics of MIC caused by sulfatereducing bacteria: Past, present and future. Anti-Corrosion Methods and Materials 46: 173--180. Jeris JS, Owens RW, and Flood F (1981) Secondary treatment of municipal wastewater with fluidized bed technology. In: Cooper PF, Atkinson B, Cooper PF, and Atkinson
Biofilms in Water and Wastewater Treatment
B (eds.) Biological Fluidized Bed Treatment of Water and Wastewater, pp. 112--120. Chichester: Ellis Horwood for Water Research Laboratory, Stevenage Laboratory. Judd S (2006) Principles and Applications of Membrane Bioreactors for Water and Wastewater Treatment, 1st edn. Oxford: Elsevier Science. Khan MT, Lewandowski Z, Takizawa S, et al. (2009) Continuous and efficient removal of THMs from river water using MF membrane combined with high dose of PAC. Desalination 249: 713--720. Kimura K, Yamato N, Yamamura H, and Watanabe Y (2005) Membrane fouling in pilotscale membrane bioreactors (MBRs) treating municipal wastewater. Environmental Science and Technology 39: 6293--6299. Kissel JC, McCarty PL, and Street RL (1984) Numerical simulations of mixed-culture biofilms. Journal of Environmental Engineering 110: 393--411. Kreft JU, Picioreanu C, Wimpenny JWT, and van Loosdrecht MCM (2001) Individualbased modelling of biofilms. Microbiology – Society for General Microbiology 147: 2897--2912. Lazarova V and Manem J (1994) Advances in biofilm aerobic reactors ensuring effective biofilm activity control. Water Science and Technology 29(10–11): 319–327. Le-Clech P, Chen V, and Fane AG (2006) Fouling in membrane bioreactors used for wastewater treatment – a review. Journal of Membrane Science 284: 17--53. Lee AK and Newman DK (2003) Microbial iron respiration: Impacts on corrosion processes. Applied Microbiology and Biotechnology 62: 134--139. Lee KC and Rittmann BE (2000) A novel hollow-fiber membrane biofilm reactor for autohydrogenotrophic denitrificaion of drinking water. Water Science and Technology 41: 219--226. Lee W, Lewandowski Z, Nielsen PH, and Hamilton WA (1995) Role of sulfate-reducing bacteria in corrosion of mild-steel – a review. Biofouling 8: 165--194. Lewandowski Z and Beyenal H (2007) Fundamentals of Biofilm Research. Boca Raton, FL: CRC Press. Lewandowski Z, Dickinson W, and Lee W (1997) Electrochemical interactions of biofilms with metal surfaces. Water Science and Technology 36: 295--302. Lewandowski Z, Walser G, and Characklis WG (1991) Reaction kinetics in biofilms. Biotechnology and Bioengineering 38: 877--882. Licina GJ, Nekosku C, Howard RL (1992) An electrochemical method for on-line monitoring of biofilm activity in cooling water. In: CORROSION’92, Paper 177. Houston, TX: NACE. Manz W, Wendt-Potthoff K, Neu TR, Szewzyk U, and Lawrence JR (1999) Phylogenetic composition, spatial structure, and dynamics of lotic bacterial biofilms investigated by fluorescent in situ hybridization and confocal laser scanning microscopy. Microbial Ecology 37: 225--237. McQuarrie J, Dempsey MJ, Boltz JP, Johnson B (2007) The expanded bed biofilm reactor (EBBR) – an innovative biofilm approach for tertiary nitrification. In: Proceedings of the 80th Annual Water Environment Federation Technical Exposition and Conference (CD-ROM). San Diego, CA, 13–17 October. Alexandria, VA: Water Environment Federation. McNamara CJ, Lemke MJ, and Leff LG (2003) Underestimation of bacterial numbers in starvation-survival mode using the nucleic acid stain DAPI. Archiv fu¨r Hydrobiologie 157: 309--319. Melin T, Jefferson B, Bixio D, et al. (2006) Membrane bioreactor technology for wastewater treatment and reuse. Desalination 187: 271--282. Metcalf and Eddy (2003) Wastewater Engineering: Treatment and Reuse, 4th edn. New York: McGraw-Hill. Milferstedt K, Pons M-N, and Morgenroth E (2006) Optical method for long-term and large-scale monitoring of spatial biofilm development. Biotechnology and Bioengineering 94: 773--782. Modin O, Fukushi K, Nakajima F, and Yamamoto K (2008) A membrane biofilm reactor achieves aerobic methane oxidation coupled to denitrification (AME-D) with high efficiency. Water Science and Technology 58: 83--87. Morgenroth E (2003) Detachment-an often overlooked phenomenon in biofilm research and modelling. In: Wuertz S, Wilderer PA, and Bishop PL (eds.) Biofilms in Wastewater Treatment, pp. 264--290. London: IWA Publishing. Morgenroth E (2008) Modelling biofilm systems. In: Henze M, van Loosdrecht MCM, Ekama G, and Brdjanovic D (eds.) Biological Wastewater Treatment – Principles, Modelling, and Design, pp. 457--492. London: IWA Publishing. Morgenroth E, Eberl HJ, and van Loosdrecht MCM (2000) Evaluating 3-D and 1-D mathematical models for mass transport in heterogeneous biofilms. Water Science and Technology 41(4–5): 347--356. Morgenroth ET and Wilderer PA (2000) Influence of detachment mechanisms on competition in biofilms. Water Research 34(2): 417--426. Nancharaiah YV, Venuopalan VP, Wuertz S, Wilderer PA, and Hausner M (2005) Compatibility of the green fluorescent protein and a general nucleic acid stain for quantitative description of a Pseudomonas putida biofilm. Journal of Microbiological Methods 60: 179--187.
569
Newman J, Occiano V, Appleton R, et al. (2005) Confirming BAF performance for treatment of CEPT effluent on a space constrained site. In: Proceedings of the 78th Annual Water Environment Federation Technical Conference and Exposition. Washington, DC, 12–15 October. Alexandria, VA: Water Environment Federation. Nielsen PR, Lee WC, Lewandowski Z, Morrison M, and Characklis W (1993) Corrosion of mild steel in an alternating oxic and anoxic biofilm system. Biofouling 7: 267--284. Nivens DE, Chambers JQ, Anderson TR, and White DC (1993) Long-term, online monitoring of microbial biofilms using a quartz crystal microbalance. Analytical Chemistry 65: 65--69. Notermans S, Dormans JAMA, and Mead GC (1991) Contribution of surface attachment to the establishment of micro-organisms in food processing plants: A review. Biofouling 5: 21--36. Ødegaard H (2006) Innovations in wastewater treatment: The moving bed biofilm process. Water Science and Technology 53(9): 17--33. Olesen BH, Nielsen PH, and Lewandowski Z (2000) Effect of biomineralized manganese on the corrosion behavior of C1008 mild steel. Corrosion 56: 80--89. Olesen BH, Yurt N, and Lewandowski Z (2001) Effect of biomineralized manganese on pitting corrosion of type 304L stainless steel. Materials and Corrosion – Werkstoffe und Korrosion 52: 827--832. Padival NA, Weiss JS, and Arnold RG (1995) Control of Thiobacillus by means of microbial competition: Implications for corrosion of concrete sewers. Water Environment Research 67: 201--205. Pankhania M, Brindle K, and Stephenson T (1999) Membrane aeration bioreactors for wastewater treatment: Completely mixed and plug-flow operation. Chemical Engineering Journal 73: 131--136. Parker DS, Lutz M, Dahl R, and Berkkopf S (1989) Enhancing reaction rates in nitrifying trickling filters through biofilm control. Journal WPCF 61(5): 618--631. Peladan J-G, Lemmel H, and Pujol R (1996) High nitrification rate with upflow biofiltration. Water Science and Technology 34(1–2): 347--353. Pereira A, Mendes J, and Melo LF (2008) Using nanovibrations to monitor biofouling. Biotechnology and Bioengineering 99: 1407--1415. Picioreanu C (1999) Multidimensional Modeling of Biofilm Structure. PhD Thesis, Department of Biotechnology, Delft University of Technology, Delft, The Netherlands. Picioreanu C, van Loosdrecht MCM, and Heijnen JJ (1998a) A new combined differential-discrete cellular automaton approach for biofilm modeling: Application for growth in gel beads. Biotechnology and Bioengineering 57: 718--731. Picioreanu C, van Loosdrecht MCM, and Heijnen JJ (1998b) Mathematical modeling of biofilm structure with a hybrid differential-discrete cellular automaton approach. Biotechnology and Bioengineering 58: 101--116. Picologlou BF, Zelver N, and Characklis WG (1980) Biofilm growth and hydraulic performance. Journal of the Hydraulic Division, ASCE 106(HY5): 733--746. Pons M-N, Milferstedt K, and Morgenroth E (2009) Biofilm monitoring on rotating discs by image analysis. Biotechnology and Bioengineering 103: 105--116. Rasmussen K and Lewandowski Z (1998) Microelectrode measurements of local mass transport rates in heterogeneous biofilms. Biotechnology and Bioengineering 59: 302--309. Rassmusen TB and Givskov M (2006) Quorum sensing inhibitors: A bargain of effects. Microbiology 152: 895--904. Rauch W, Vanhooren H, and Vanrolleghem PA (1999) A simplified mixed-culture biofilm model. Water Research 33(9): 2148--2162. Reichert P and Wanner O (1997) Movement of solids in biofilms: Significance of liquid phase transport. Water Science and Technology 95(2–3): 289--299. Rittmann BE and Brunner CW (1984) The non-steady-state-biofilm process for advanced organics removal. Journal (Water Pollution Control Federation) 56: 874--880. Rittmann BE and Dovantzis K (1983) Dual limitation of biofilm kinetics. Water Research 17: 1727--1734. Rittmann BE and McCarty PL (1980a) Evaluation of steady-state-biofilm kinetics. Biotechnology and Bioengineering 22: 2359--2373. Rittmann BE and McCarty PL (1980b) Model of steady-state-biofilm kinetics. Biotechnology and Bioengineering 22: 2343--2357. Rogalla F and Bourbigot M-M (1990) New developments in complete nitrogen removal with innovative biological reactors. Water Science and Technology 22(1–2): 273--280. Rogalla F, Ravarini P, DeLarminat G, and Courtelle J (1990) Large scale biological nitrate and ammonia removal. Water and Environment Journal 4(4): 319--329. Sand W and Bock E (1984) Concrete corrosion in the Hamburg sewer system. Environmental Technology Letters 5: 517--528. Schmitt G (1991) Effect of elemental sulfur on corrosion in sour gas systems. Corrosion 47: 285--308. Sen D and Randall CW (2008) Improved computational model (AQUIFAS) for activated sludge, integrated fixed–film activated sludge, and moving-bed biofilm reactor
570
Biofilms in Water and Wastewater Treatment
systems. Part I: Semi-empirical model development. Water Environment Research 80: 439--453. Seth AD and Edyvean RGJ (2006) The function of sulfate-reducing bacteria in corrosion of potable water mains. International Biodeterioration and Biodegradation 58: 108--111. Shannon MA, Bohn PW, Elimelech M, Georgiadis JG, Marinas BJ, and Mayes AM (2008) Science and technology for water purification in the coming decades. Nature 452: 301--310. Shieh WK and Keenan JD (1986) Fluidized bed biofilm reactor for wastewater treatment. Advances in Biochemical Engineering Biotechnology 33: 133--169. Sin G, Van Hulle SWH, De Pauw DJW, van Griensven A, and Van rolleghem PA (2005) A critical comparison of systematic calibration protocols for activated sludge models: A SWOT analysis. Water Research 39: 2459--2474. Sin G, Weijma J, Spanjers H, and Nopens I (2008) Dynamic model development and validation for a nitrifying moving bed biofilter: Effect of temperature and influent load on the performance. Process Biochemistry 43: 384--397. Smets BF, Riefler RG, Lendenmann U, and Spain JC (1999) Kinetic analysis of simultaneous 2,4-dinitrotoluene (DNT) and 2,6-DNT biodegradation in an aerobic fluidized-bed biofilm reactor. Biotechnology and Bioengineering 63(6): 642--653. Stenstrom MK and Rosso D (2008) Aeration and mixing. In: Henze M, van Loosdrecht MCM, Ekama G, and Brdjanovic D (eds.) Biological Wastewater Treatment – Principles, Modelling, and Design. London: IWA Publishing. Stephenson T, Judd S, Jefferson B, and Brindle K (2000) Membrane Bioreactors for Wastewater Treatment. London: IWA Publishing (Alliance House). Stewart PS (2003) Diffusion in biofilms. Guest commentaires. Journal of Bacteriology 185(5): 1485--1491. Stoddart FW (1911) Nitrification and the absorption theory. Journal of the Society of Chemical Industry 30: 236--237. Suzuki Y, Hatano N, Ito S, and Ikeda H (2000) Performance of nitrogen removal and biofilm structure of porous gas permeable membrane reactor. Water Science and Technology 41: 211--217. Taka´cs I, Bye CM, Chapman K, Dold PL, Fairlamb PM, and Jones RM (2007) A biofilm model for engineering design. Water Science and Technology 55(8–9): 329--336. Takacs I, Dold P, Belia L., et al. (2010) Structured process models for nutrient removal. In: Johnson BR (ed.) Nutrient Removal, WEF Manual of Practice No. 34, ch. 11. New York: McGraw-Hill. Tschui M (1994) Submerse Festbettreaktoren. Skriptum der EAWAG (Du¨bendorf) fu¨r den Studiengang Umwelt- und Kulturingenieure an der ETH Zu¨rich, Vertieferblock, WS 94/95 [unvero¨ffentlicht]. Tuovinen OH, Button KS, Vuorinen A, Carlson L, Mair DM, and Yut LY (1980) Bacterial, chemical, and mineralogical characteristics of tubercles in distribution pipelines. Journal of American Water Works Association 72: 626--635. US EPA (1984) Corrosion manual for internal corrosion of water distribution systems. EPA/570/9-84/001. Washington, DC: Office of Drinking Water. van Loosdrecht MCM, Eikelboom D, Gjaltema A, Mulder A, Tijhuis L, and Heijnen JJ (1995) Biofilm structures. Water Science and Technology 32(8): 35--43.
Van Hulle SWH, Maertens J, De Pauw DJW, and Vanrolleghem P (2004) Using parameter sensitivity analysis of the CANON biofilm process: What to measure, where to measure and under which conditions? Water and Environmental Management Series Young Researchers 59--66. Veerapaneni S, Long B, Freeman S, and Bond R (2007) Reducing energy consumption for seawater desalination. Journal of the American Water Works Association 99: 95--106. Veluchamy RRA (2006) Structure and activity of Pseudomonas aeruginosa PAO1 biofilms. MS Dissertation. Montana State University. Bozeman. Videla HA (2001) Microbially induced corrosion: An updated overview. International Biodeterioration and Biodegradation 48: 176–201. Videla HA and Herrera LK (2005) Microbiologically influenced corrosion: Looking to the future. International Microbiology 8: 169--180. Vincke E, Boon N, and Verstraete W (2001) Analysis of the microbial communities on corroded concrete sewer pipes – a case study. Applied Microbiology and Biotechnology 57: 776--785. Vrouwenveldera JS, Graf von der Schulenburgc DA, Kruithofa JC, Johnsc ML, and van Loosdrechtb MCM (2009) Biofouling of spiral-wound nanofiltration and reverse osmosis membranes: A feed spacer problem. Water Research 43: 583--594. Wang Y, Wang T, Su Y, Peng F, Wu H, and Jiang Z (2005) Remarkable reduction of irreversible fouling and improvement of the permeation properties of poly (ether sulfone) ultrafiltration membranes by blending with pluronic F127. Langmuir 21: 11856--11862. Wanner O, Cunningham AB, and Lundman R (1995) Modeling biofilm accumulation and mass transport in a porous-medium under high substrate loading. Biotechnology and Bioengineering 47: 703–712. Wanner O, Eberl H, Morgenroth E, et al. (2006) Mathematical Modeling of Biofilms, Scientific and Technical Report No. 18. London: IWA Publishing. Wanner O and Gujer WA (1986) Multispecies biofilm model. Biotechnology and Bioengineering 28: 314--328. Wanner O and Reichert P (1996) Mathematical modeling of mixed-culture biofilms. Biotechnology and Bioengineering 49: 172--184. Williamson K and McCarty PL (1976) Verification studies of biofilm model for bacterial substrate utilization. Journal Water Pollution Control Federation 48: 281--296. Wimpenny JWT and Colasanti R (1997a) A more unifying hypothesis for biofilm structures – a reply. FEMS Microbiology Ecology 24: 185--186. Wimpenny JWT and Colasanti R (1997b) A unifying hypothesis for the structure of microbial biofilms based on cellular automaton models. FEMS Microbiology Ecology 22: 1--16. Wolfram S (1986) Theory and Application of Cellular Automata. Addison-Wesley, MA: World Scientific. Yeon K-M, Cheong W-S, Oh H-S, et al. (2009) Quorum sensing: A new biofouling control paradigm in a membrane bioreactor for advanced wastewater treatment. Environmental Science and Technology 43: 380--385. Zhang X and Bishop PL (1994) Evaluation of tortuosity factors and effective diffusivities in biofilm. Water Research 28: 2279--2287.
4.16 Membrane Biological Reactors FI Hai, University of Wollongong, Wollongong, NSW, Australia K Yamamoto, University of Tokyo, Tokyo, Japan & 2011 Elsevier B.V. All rights reserved.
4.16.1 4.16.2 4.16.2.1 4.16.2.2 4.16.3 4.16.3.1 4.16.3.2 4.16.3.2.1 4.16.3.2.2 4.16.3.2.3 4.16.3.2.4 4.16.3.2.5 4.16.3.2.6 4.16.3.3 4.16.3.4 4.16.3.4.1 4.16.3.4.2 4.16.3.4.3 4.16.3.4.4 4.16.3.4.5 4.16.4 4.16.4.1 4.16.4.2 4.16.4.3 4.16.4.4 4.16.4.4.1 4.16.4.4.2 4.16.4.4.3 4.16.4.4.4 4.16.4.4.5 4.16.4.4.6 4.16.4.4.7 4.16.4.5 4.16.4.6 4.16.4.7 4.16.4.7.1 4.16.4.7.2 4.16.4.7.3 4.16.4.7.4 4.16.5 4.16.5.1 4.16.5.1.1 4.16.5.1.2 4.16.5.1.3 4.16.5.1.4 4.16.5.2 4.16.5.3 4.16.5.4 4.16.5.4.1 4.16.5.4.2 4.16.5.4.3 4.16.5.5 4.16.5.5.1
Introduction Aeration and Extractive Membrane Biological Reactors Aeration Membrane Biological Reactor Extractive Membrane Biological Reactor History and Fundamentals of Biosolid Separation MBR Historical Development Process Comparison with Conventional Activated Sludge Process Treatment efficiency/removal capacity Sludge properties and composition Sludge production and treatment Space requirements Wastewater treatment cost Comparative energy usage Relative Advantages of MBR Factors Influencing Performance/Design Considerations Pretreatment Membrane selection and applied flux Sludge retention time Mixed liquor suspended solids concentration Oxygen transfer Worldwide Research and Development Challenges Importance of Water Reuse and the Role of MBR Worldwide Research Trend Modeling Studies on MBR Innovative Modifications to MBR Design Inclined plate MBR Integrated anoxic–aerobic MBR Jet-loop-type MBR Biofilm MBR Nanofiltration MBR Forward osmosis MBR Membrane distillation bioreactor Technology Benefits: Operators’ Perspective Technology Bottlenecks Membrane Fouling – the Achilles’ Heel of MBR Technology Fouling development Types of membrane fouling Parameters influencing MBR fouling Fouling mitigation Worldwide Commercial Application Installations Worldwide Location-specific drivers for MBR applications Plant size Development trend and the current status in different regions Decentralized MBR plants: Where and why? Commercialized MBR Formats Case-Specific Suitability of Different Formats MBR Providers Market share of the providers Design considerations Performance comparison of different providers Standardization of Design and Performance-Evaluation Method Standardization of MBR filtration systems
571 572 572 574 574 574 576 576 576 577 577 578 579 580 581 581 581 581 581 581 582 582 583 583 584 585 585 585 585 585 586 586 586 587 588 588 588 589 593 596 596 596 596 596 598 600 600 601 601 601 602 604 604
571
572
Membrane Biological Reactors
4.16.5.5.2 4.16.6 4.16.7 References
Standardization of MBR characterization methods Future Vision Conclusion
4.16.1 Introduction
Size
Membrane biological reactors combine the use of biological processes and membrane technology to treat wastewater. The use of biological treatment can be traced back to the late nineteenth century. It became a standard method of wastewater treatment by the 1930s (Rittmann, 1987). Both aerobic and anaerobic biological treatment methods have been extensively used to treat domestic and industrial wastewater (Visvanathan et al., 2000). After removal of the soluble biodegradable matter in the biological process, any biomass formed needs to be separated from the liquid stream to produce the required effluent quality. In the conventional process, a secondary settling tank is used for such solid/liquid Apprx. molecular weight 200 µm 0.001
20k
200k
0.01
0.1
Metal ions
605 605 605 605
separation and this clarification is often the limiting factor in effluent quality (Benefield and Randall, 1980). Membrane filtration, on the other hand, denotes the separation process in which a membrane acts as a barrier between two phases. In water treatment, the membrane consists of a finely porous medium facilitating the transport of water and solutes through it (Ho and Sirkar, 1992). The separation spectrum for membranes, illustrated in Figure 1, ranges from reverse osmosis (RO) and nanofiltration (NF) for the removal of solutes, to ultrafiltration (UF) and microfiltration (MF) for the removal of fine particulates. MF and UF membranes are predominantly used in conjunction with biological reactors (Pearce, 2007). UF can remove the finest particles found in water supply, with the removal rating dependent upon the
500k 1.0
10
100
Protein/enzymes Colloidal silica
Target of separation
Acid
Viruses Pesticide
Endotoxin
Herbicide
Yeast and fungi Bacteria Algae
Humic acids
Endocrine disruptors
Giardia cyst
Natural organic matter
Human hair
Crypt ospor idium
Starch
Membrane
Reverse osmosis Nano filtration Ultrafiltration Microfiltration • Reverse osmosis (RO): salts, pesticides, herbicides, metal ions, endocrine disruptors, disinfection by-products • Nanofiltration (NF): divalents salts, pesticides, herbicides, divalent metal ions • Ultrafiltration (UF): virus, bacteria, endotoxin • Microfiltration (MF): bacteria, (virus) Figure 1 The separation spectrum for membranes.
1000
Membrane Biological Reactors
• •
Easily meets regulatory levels Suitable for discharge to pristine environment
• • •
Membrane Bioreactor
Aeration (a) Oxygen phase Transfer of organics and nutrients
Membrane
Biofilm Wastewater side
(b)
Wastewater side (Biodegradable + inhibitory organics) Selective membrane
Meets standards for potable applications Increased value for industrial applications May be useful in obtaining development permit
Figure 2 Market drivers for membranes in wastewater. Information from Howell JA (2004) Future of membranes and membrane reactors in green technologies and for water reuse. Desalination 162: 1–11; and Pearce G (2007) Introduction to membranes: Filtration for water and wastewater treatment. Filtration and Separation 44(2): 24–27.
Membrane permeate
Suspended biomass
Advantages
Reuse
Discharge
Wastewater
Selective transfer of degradable organics
Criteria
et al., 2000); for bubble-less aeration of the bioreactor (Brindle and Stephenson, 1996); and for extraction of priority organic pollutants from hostile industrial wastewaters (Stephenson et al., 2000). There are other forms of membrane biological reactors such as enzymatic membrane bioreactor (Charcosset, 2006) for production of drugs, vitamins, etc., or membrane biological reactors for waste-gas treatment (Reij et al., 1998), a discussion about which is beyond the scope of this chapter. Solid–liquid membrane-separation bioreactors employ UF or MF modules for the retention of biomass to be recycled into the bioreactor. Gas-permeable membranes are used to provide bubble-less oxygen mass transfer to degradative bacteria
Oxygen transfer
pore size of the active layer of the membrane. The complete pore-size range for UF is approximately 0.001–0.02 mm, with a typical removal capability of UF for water and wastewater treatment of 0.01–0.02 mm. MF typically operates at a particle size that is up to an order of magnitude coarser than this. In water treatment, the modern trend is to use a relatively tight MF with a pore size of approximately 0.04–0.1 mm, whereas wastewater normally uses a slightly more open MF with a pore size of 0.1–0.4 mm (though wastewater can be treated using UF membranes, or MF membranes used for water applications). The market drivers for membranes in wastewater are illustrated in Figure 2. However, as in any separation process, in membrane technology too, the management and disposal of concentrate is a significant issue. Environment-friendly management and disposal of the resulting concentrates at an affordable cost is a significant challenge to water and wastewater utilities and industry. To eradicate the respective disadvantages of the individual technologies, the biological process can be integrated with membrane technology. Although some recent studies have demonstrated case-specific feasibility of direct UF of raw sewage (Janssen et al., 2008), membranes by themselves are seldom used to filter untreated wastewater, since fouling prevents the establishment of steady-state conditions and because water recovery is very low (Schrader et al., 2005; Fuchs et al., 2005; Judd and Jefferson, 2003). However, membrane filtration can be efficiently used in combination with a biological process. The biological process converts dissolved organic matter into suspended biomass, reducing membrane fouling and allowing increase in recovery. On the other hand, in the membrane filtration process, the membranes introduced into the bioreactors not only replace the settling unit for solid–liquid separation but also form an absolute barrier to solids and bacteria and retain them in the process tank. As our understanding of membrane technology grows, we learn that membrane technology is now being applied to a wider range of industrial applications and is used in many new forms for wastewater treatment. Combining membrane technology with biological reactors for the treatment of municipal and industrial wastewaters has led to the development of three generic membrane processes within bioreactors (Figure 3): for separation and recycle of solids (Visvanathan
573
Biofilm Nutrient biomedium (c) Figure 3 Simplified representation of membrane biological reactors: (a) biosolid separation, (b) aeration, and (c) extractive membrane biological reactors.
574
Membrane Biological Reactors
present in the bioreactor. Additionally, the membrane can act as support for biofilm development, with direct oxygen transfer through the membrane wall in one direction and nutrient diffusion from the bulk liquid phase into the biofilm in the other direction. An extractive membrane process has been devised for the transfer of degradable organic pollutants from hostile industrial wastewaters, via a nonporous silicone membrane, to a nutrient medium for subsequent biodegradation. Biosolid separation is, however, the most widely studied process and has found full-scale applications in many countries. In a comprehensive review published in 2006, Yang et al. (2006) pointed out that the vast majority of research on membrane biological reactors since 1990 focused on biosolidseparation-type applications. There was no significant increase in the number of studies on gas diffusion and extractive membrane biological reactors over time. Publications on extractive and diffusive membrane biological reactors became available during 1994–95, after which a steady output of less than five publications a year was observed. This indicates that current research is predominantly in the water and wastewater-filtration area, in parallel with the commercial success in this field. In line with the current trend of research and commercial application, this chapter focuses on the biosolidseparation membrane biological reactors, which is more commonly known as membrane bioreactor (MBR). However, a brief outline of the other two types of membrane biological reactors is furnished in Section 4.16.2. The remainder of this chapter elaborates on the history, fundamentals, research and development challenges, as well as the commercial application of the biosolid-separation membrane biological reactors, which are henceforth referred to as MBRs.
4.16.2 Aeration and Extractive Membrane Biological Reactors 4.16.2.1 Aeration Membrane Biological Reactor Wastewater-treatment processes using high-purity oxygen have a greater volumetric degradation capacity compared to the conventional air-aeration process. However, conventional oxygenation devices have high power requirements associated with the need for high mixing rate, and cannot be used in conjunction with biofilm processes. In the membraneaeration biological reactors (MABRs), the capability of biofilm to retain high concentrations of active bacteria is coupled with the high oxygen transfer rate to the biofilm. The key characteristic advantages of MABRs are summarized as follows:
• •
High oxygen transfer rate, especially suitable for highoxygen-demanding wastewaters. In conventional aerobic biological wastewater treatment, volatile organic compounds (VOCs) can escape to the atmosphere without being biodegraded as a result of air bubbles stripping out the compounds from the bulk liquid. Since no oxygen bubbles are formed in MABRs, gas stripping of VOCs and foaming due to the presence of surfactants can be prevented (Rothemund et al., 1994; Semmens 1991; Wilderer et al., 1985) to a greater extent.
•
Membrane-attached biofilms are in intimate contact with the oxygen source, with direct interfacial transfer and utilization of oxygen within the biofilm. In contrast to conventional biofilm processes, in MABR biofilms, oxygen from the membrane and pollutant substrate(s) from the bulk liquid are transferred across the biofilm in countercurrent directions (Figure 4). Biofilm stratification in MABRs results from this distribution of the maximum oxygen and pollutant-substrate concentrations at different locations within the biofilm and the relative thickness of MABR biofilms; this enables the removal of more than one pollutant type. The high oxygen concentrations coupled with the low organic carbon concentrations near the membrane/biofilm interface encourage nitrification, an aerobic heterotrophic layer above this facilitates organic carbon oxidation, and an anoxic layer near the biofilm/ liquid interface supports denitrification (Stephenson et al., 2000).
MABRs have been used to treat a variety of wastewater types at the laboratory scale (Brindle and Stephenson, 1996). However, in line with the characteristics of MABRs discussed above, most investigations show that the process is particularly suitable for the treatment of high-oxygen-demanding wastewaters, biodegradation of VOCs, combined nitrification, denitrification, and/or organic carbon oxidation in a single biofilm. Bubble-less oxygen mass transfer can be accomplished using gas-permeable dense membranes or hydrophobic microporous membranes (Cote et al., 1988). Both plate and frame and hollow-fiber membrane configurations have been used to supply the oxygen. Oxygen diffusion through dense membrane material can be achieved at high gas pressures without bubble formation. In hydrophobic microporous membranes, the pores remain gas filled; and oxygen is transported to the shell side of the membrane through the pores by gaseous diffusion or Knudsen flow-transport mechanisms. The partial pressure of the gas is kept below the bubble point to ensure the bubble-less supply of oxygen (Ahmed and Semmens, 1992; Rothemund et al., 1994; Semmens, 1991; Semmens and Gantzer, 1993). Pressurized hollow fibers have been investigated in the dead-end and flow-through modes of operation. The evacuation of carbon dioxide from the bioreactor is a benefit of flow-through operation, though no quantitative work to determine removal rates has been undertaken (Cote et al., 1997; Kniebusch et al., 1990). Deadend operation has usually been avoided, due to significantly decreased performance and condensate formation in the lumen (Cote et al., 1997). The nonbiological fouling and loss of performance of dead-end porous hollow fibers due to iron oxidation, absorption of free oils and greases into pores, surfactants, and suspended solids, and fiber tangling have been reported (Semmens and Gantzer, 1993). Chemical treatment of the dead ends of these hollow fibers may provide a means for the condensate to escape. The liquid boundary layer normally has a greater impact upon the overall oxygen mass transfer than the membrane, with mixing of the liquid a key operational parameter (Cote et al., 1997; Kniebusch et al., 1990; Wilderer et al., 1985). However, wall thickness significantly affects the transport of
Membrane Biological Reactors
Anaerobic zone
Aerobic zone Microbial activity
Biofilm/liquid interface
Membrane/biofilm interface
575
Oxygen Carbon substrate
Biofilm
Oxygen Microbial activity
Carbon substrate
Biofilm/liquid interface
Aerobic zone
biofilm interface
Nonpermeable support/
Anaerobic zone
Biofilm Figure 4 Simplified representation of the steady-state concentration profiles of oxygen, carbon substrate, and microbial activity in case of MABR biofilm and conventional biofilm.
oxygen through dense polymer membranes (Wilderer et al., 1985). Oxygen transport is also controlled by the presence of membrane-attached biofilm and its thickness; the partial pressure of oxygen and flow-velocity characteristics on the lumen side; and the wastewater flow-velocity characteristics on the shell side of the membrane (Kniebusch et al., 1990; Pankania et al., 1994). Oxygen partial pressure provides the means for controlling the depth of oxygen penetration into the wastewater, with an increase in partial pressure resulting in an increase in the metabolic activity of the membraneattached biofilm (Rothemund et al., 1994). In bioreactors, most membranes used for oxygen mass transfer operate with the biofilm attached to the membrane surface. These biofilms are in intimate contact with the oxygen source and are protected against abrasion and grazing (Kniebusch et al., 1990; Rothemund et al., 1994). Scanning electron micrographs show that some attached bacteria inhabit the membrane pores, with the location of the oxygen and wastewater interphase very close to the bacteria (Rothemund et al., 1994). Thus, oxygen-transfer resistance due to the thickness of the porous membrane and the liquid boundary layer are not necessarily decisive limiting factors (Kniebusch et al., 1990; Rothemund et al., 1994; Wilderer et al., 1985).
Excessive biofilm accumulation can result in the transport limitation of oxygen and nutrients, plugging of membrane fibers, a decline in biomass activity, metabolite accumulation deep within the biofilm, and the channeling of flow in the bioreactor such that steady-state conditions may not be maintained (Debus and Wanner, 1992; Pankania et al., 1994; Yeh and Jenkins, 1978). To operate at maximum efficiency, occasional membrane washing, air scouring, backwashes, and high recirculation rate of wastewater to achieve high shear velocities have all been employed to control biomass accumulation. In the MABR process, oxygen is transferred without forming bubbles and therefore cannot be utilized to mix the bulk liquid. In laboratory scale MABRs, liquid-phase mixing has been achieved using recirculation pumps, impellers, magnetic stirrers, nitrogen, or air sparging.
4.16.2.2 Extractive Membrane Biological Reactor The extractive membrane biological reactor (EMBR) process enables the transfer of degradable organic pollutants from hostile industrial wastewaters, via a dense silicone membrane,
576
Membrane Biological Reactors
to a nutrient medium for subsequent degradation (Brindle and Stephenson, 1996). Membranes used for the extraction of pollutants into a bioreactor have been developed using pervaporation by exchanging the vacuum phase with a nutrient biomedium phase where biodegradation mechanisms maintain the concentration gradient needed to transfer organic pollutants present in hostile industrial wastewaters (Lipski and Cote, 1990; Nguyen and Nobe, 1987; Yun et al., 1992). The inorganic composition of the nutrient medium is unaffected by the industrial wastewater within the hydrophobic hollow-fiber membrane. Hence, the conditions within the bioreactor can be optimized to ensure high biodegradation rate (Brookes and Livingston, 1993; Livingston, 1993, 1994). The extraction and biodegradation of toxic volatile organic pollutants, such as chloroethanes, chlorobenzenes, chloroanilines, and toluene from hostile industrial wastewaters, with high salinity and extremes of pH, using EMBRs have been demonstrated at the laboratory scale (Stephenson et al., 2000). Further information on these two generic types of MBRs can be derived from the review papers by Brindle and Stephenson (1996) and McAdam and Judd (2006), and the book by Stephenson et al. (2000). Yang et al. (2006) argued that extractive or aeration MBRs present a significant opportunity for researchers as niche areas of application as these novel processes remain unexplored. Hazardous waste treatment and toxic waste cleanup present two potential areas for the EMBR (Brookes and Livingston, 1994; Dossantos and Livingston, 1995; Livingston et al., 1998), whereas hydrogenotrophic denitrification of groundwater (Clapp et al., 1999; Mo et al., 2005; Modin et al., 2008; Nuhoglu et al., 2002; Rezania et al., 2005) and gas-extractionassisted fermentation (Daubert et al., 2003; Lu et al., 1999) are potential research areas for the AMBR. It is also important to recognize the fact that these three membrane processes are not mutually exclusive and, if necessary, could be coupled into one bioreactor (Brindle and Stephenson, 1996). Once the research field has gained momentum, commercial interest might correspondingly follow.
4.16.3 History and Fundamentals of Biosolid Separation MBR 4.16.3.1 Historical Development Membranes have been finding wide application in water and wastewater treatment ever since the early 1960s when Loeb and Sourirajan invented an asymmetric cellulose acetate membrane for RO (Visvanathan et al., 2000). Many combinations of membrane solid/liquid separators in biological treatment processes have been studied since. The first descriptions of the MBR technology date from the late 1960s. The trends that led to the development of today’s MBR are depicted in Figure 5. When the need for wastewater reuse first arose, the conventional approach was to use advanced treatment processes. The progress of membrane manufacturing technology and its applications could lead to the eventual replacement of tertiary treatment steps by MF or UF (Figure 5(a)). Parallel to this development, MF or UF was used for solid/liquid separation in the biological treatment
process and thereby sedimentation step could be eliminated (Figure 5(b)). The original process was introduced by DorrOlivier Inc. and combined the use of an activated sludge bioreactor with a cross-flow membrane-filtration loop (Smith et al., 1969). By pumping the mixed liquor at a high pressure into the membrane unit, the permeate passes through the membrane and the concentrate is returned to the bioreactor (Hardt et al., 1970; Arika et al., 1966; Krauth and Staab, 1988; Muller et al., 1995). The flat-sheet membranes used in this process were polymeric and featured pore size ranging from 0.003 to 0.01 mm (Enegess et al., 2003). Although the idea of replacing the settling tank of the conventional activated sludge (CAS) process was attractive, it was difficult to justify the use of such a process because of the high cost of membranes, low economic value of the product (tertiary effluent), and the potential rapid loss of performance due to fouling. Due to the poor economics of the first-generation MBRs, apart from a few examples such as installations at the basement level of skyscrapers in Tokyo, Japan, for wastewater reuse in flushing toilets, they usually found applications only in niche areas with special needs such as isolated trailer parks or ski resorts. The breakthrough for the MBRs occurred in 1989, the process involved submerging the membranes in the reactor itself and withdrawing the treated water through the membranes (Yamamoto et al., 1989; Kayawake et al., 1991; Chiemchaisri et al., 1993; Visvanathan et al., 1997). In this development, membranes were suspended in the reactor above the air diffusers (Figure 5(c)). The diffusers provided the oxygen necessary for treatment to take place and scour the surface of the membrane to remove deposited solids. There have been other parallel attempts to save energy in membrane-coupled bioreactors. In this regard, the use of jet aeration in the bioreactor was investigated (Yamagiwa et al., 1991). The main feature of this process was that the membrane module was incorporated into the liquid recirculation line for the formation of the liquid jet such that aeration and filtration could be accomplished using only a single pump. Jet aeration works on the principle that a liquid jet, after passing through a gas layer, plunges into a liquid bath entraining a considerable amount of air. Using only one pump makes it mechanically simpler and therefore useful to small communities. The limited amount of oxygen transfer possible with this technique, however, restricts this process only to such small-scale applications. The invention of air-backwashing techniques for membrane declogging led to the development of using the membrane itself as both clarifier and air diffuser (Parameshwaran and Visvanathan, 1998). In this approach, two sets of membrane modules are submerged in the aeration tank. While the permeate was extracted through one of the sets, the other set was supplied with compressed air for backwashing. The cycle was repeated alternatively, and there was a continuous airflow into the aeration tank, which was sufficient to aerate the mixed liquor. Eventually, two broad trends have emerged in recent times, namely submerged MBRs and sidestream MBRs. Submerged technologies tend to be more cost effective for largerscale lower-strength applications, and sidestream technologies are favored for smaller-scale higher-strength applications. The sidestream MBR envelope has been extended in recent years by the development of the air-lift concept, which
Membrane Biological Reactors
577
Conventional activated sludge + MF(UF)
Prescreening Sludge withdrawal (a)
Biological tank
Settling tank
Sidestream MBR Effluent
Membrane
(b)
Submerged MBR (integrated)
(c)
Submerged MBR (separated)
(d)
Figure 5 Evolution of membrane use in conjunction with bioreactor.
bridges the gap between submerged and cross-flow sidestream MBR, and may have the potential to challenge submerged systems in larger-scale applications (Pearce, 2008b). The economic viability of the current generation of MBRs depends on the achievable permeate flux, mainly controlled by effective fouling control with modest energy input (typically r1 kW h1 m3 product). More efficient fouling-mitigation methods can be implemented only when the phenomena occurring at the membrane surface are fully understood. Detailed discussion on the technology bottlenecks and the design aspects are provided in Sections 4.16.4 and 4.16.5, respectively. It is worth noting that as the oxygen supply limits maximum mixed-liquor suspended solids (MLSSs) in aerobic MBR, anaerobic MBRs (AnMBRs) were also developed. The first test of the concept of using membrane filtration with anaerobic treatment of wastewater appears to have been reported by Grethlein (1978). The first commercially available AnMBR was developed by Dorr-Oliver in the early 1980s for high-strength whey-processing wastewater treatment. The process, however, was not applied at full scale, possibly due to high membrane costs (Sutton et al., 1983). The Ministry of International Trade and Industry (MITI), Japan, launched a 6-year research and development (R&D) project named Aqua-Renaissance ’90 in 1985 with the particular objective of developing energy-saving and smaller footprint water-
treatment processes utilizing sidestream AnMBR to produce reusable water from industrial wastewater and sewage. However, a high cross-flow velocity and frequent physicochemical cleaning was required to maintain the performance of such a high-rate MBR (Yamamoto, 2009). It was difficult to reduce the energy consumption significantly by adopting the sidestream operation using a big recirculation pump. On the other hand, commencing in 1987, a system known as anaerobic digestion ultrafiltration (ADUF) was developed in South Africa for industrial wastewater treatment (Ross et al., 1992). This process is currently in operation. Further details on AnMBRs can be derived from the comprehensive review by Liao et al. (2006). This chapter, however, focuses on aerobic MBRs.
4.16.3.2 Process Comparison with Conventional Activated Sludge Process Some important basic characteristics of CAS and MBR are compared in this section.
4.16.3.2.1 Treatment efficiency/removal capacity The MBR process involves a suspended growth-activated sludge system that utilizes microporous membranes for solid/ liquid separation in lieu of secondary clarifiers. The biological treatment in MBR is performed according to the principles
578
Membrane Biological Reactors
known from activated sludge treatment. However, higher suspended solids, biological oxygen demand (BOD), and chemical oxygen demand (COD) removals in MBR have been reported throughout the literature. With CAS, the colloidal fraction (that represents about 20% of the organic content of wastewater) has a residence time (hydraulic residence time (HRT)) in the range of few hours while with MBR, due to total SS retention, the residence time of this fraction (sludge retention time (SRT)) is in the range of several days. Thus, the biodegradation for this fraction is higher in MBR than in CAS. Some soluble compounds too, after being adsorbed on SS, can be retained in MBR and can be biodegraded to a better extent. Thus, some studies have ascribed the better removal of soluble COD in MBR to the fact that the effluent is particle free (Cote et al., 1997; Engelhardt et al., 1998; De Wilde et al., 2003). MBR produces quality effluent suitable for reuse applications or as a high-quality feedwater source for RO treatment. Indicative output quality includes suspended solids o1 mg l1, turbidity o0.2 nephelometric turbidity unit (NTU), and up to 4 log removal of virus (depending on the membrane nominal pore size). In addition, it provides a barrier to certain chlorine-resistant pathogens such as Cryptosporidium and Giardia. In comparison to the CAS process, which typically achieves 95%, COD removal can be increased to 96–99% in MBRs (Stephenson et al., 2000). Nutrient removal is one of the main concerns in modern wastewater treatment especially in areas that are sensitive to eutrophication. As in the CAS, currently, the most widely applied technology for N removal from municipal wastewater is nitrification combined with denitrification. Total nitrogen removal through the inclusion of an anoxic zone is possible in MBR systems. Besides phosphorus precipitation, enhanced biological phosphorus removal (EBPR) can be implemented, which requires an additional anaerobic process step. Some characteristics of MBR technology render EBPR in combination with post-denitrification as an attractive alternative that achieves very low nutrient effluent concentrations (Drews et al., 2005b).
4.16.3.2.2 Sludge properties and composition The presence of a membrane for sludge separation has many consequences. This influences the rheological properties and composition of the sludge. Defrance et al. (2000) observed in a sidestream MBR with high cross-flow velocity that MBR sludge was less viscous than conventional sludge. The same was observed by Rosenberger et al. (2002). Furthermore, with increasing shear rate, viscosity of the sludge decreases (Rosenberger et al., 2002), although in some cases, the activated sludge behaves as a Newtonian fluid (Xing et al., 2001). Defrance and Jaffrin (1999) found out that filtering-activated sludge from an MBR resulted in fouling that could be totally, physically removed, whereas filtration of CAS led to physically irremovable fouling. It is quite difficult to generalize information about sludge composition from different installations, since each installation promotes different types of activated sludge. This has its effect on the microbial community that can be found in an activated sludge system. Nevertheless, it is obvious that the presence of the membrane in an MBR system influences the biomass composition. Since no suspended solids are washed
out with the effluent, the only sink is surplus sludge. From a secondary clarifier, lighter species are washed out, whereas in an MBR they are retained in the system by the membrane. Furthermore, changes in SRT and higher MLSS concentrations might lead to changes in the microbial community. Microbialcommunity analyses have revealed significant differences between CAS system and an MBR and a higher fraction of bacteria was found in the nongrowing state in the MBR (Witzig et al., 2002; Wagner and Rosenwinkel, 2000).
4.16.3.2.3 Sludge production and treatment Small-scale laboratory studies revealed a great advantage of MBRs, that is, lower or even zero excess sludge production, caused by low loading rates and high SRTs (Benitez et al., 1995). When longer SRTs are applied, sludge production, of course, decreases in the MBR (Wagner and Rosenwinkel, 2000). However, the amount of excess secondary sludge produced in larger MBR installations operated under the practical range of SRTs is somewhat lower than or even equal to that in conventional systems (Gu¨nder and Krauth, 2000). Table 1 provides a general comparison of the sludge-production rates from different treatment processes. It should be noted that the primary sludge production in the case of the MBR is lower. The suited pretreatment for the MBR is grids and/or sieves, and in an average, screened water was observed to contain 30% more solids than settled water (Jimenez et al., 2010). MBR sludge treatment is almost the same compared to CAS systems. The dewaterability of waste-activated sludge from the MBR seems to pose no additional problem, compared to aerobic stabilized waste sludge from CAS systems (Kraume and Bracklow, 2003).
4.16.3.2.4 Space requirements One of the advantages of the MBR is its compactness, because large sedimentation tanks are not needed. An interesting parameter in this respect is the surface-overflow rates for the two systems. The overflow rate of a secondary clarifier is defined as the ratio of its flow and footprint, that is, the volume of water that can be treated per square meter of tank. In practice, values around 22 m d1 are used. For an MBR filtration tank, an overflow rate can also be estimated from the permeate flux and the membrane-packing density within the
Table 1
Sludge production in case of different treatment processes
Treatment process
Sludge production kg (kg BOD)1
Submerged MBR Structured media biological aerated filter Trickling filter Conventional activated sludge Granular media BAF
0.0–0.3 0.15–0.25 0.3–0.5 0.6 0.63–1.06
Data from Stephenson T, Judd S, Jeferson B, and Brindle K (2000) Membrane Bioreactors for Wastewater Treatment. London: IWA. Gander MA, Jefferson B, and Judd SJ (2000) Membrane bioreactors for use in small wastewater treatment plants: Membrane materials and effluent quality. Water Science and Technology 41: 205–211, and Metcalf and Eddy, Inc. (2003) Wastewater Engineering – Treatment and Reuse, 4th edn. New York: McGraw-Hill.
Membrane Biological Reactors
tank. Following this method, Evenblij et al. (2005a) showed that with an average permeate flux of 15 l m2 h1, the overflow rates of the membrane tanks are in the range 25–62 m d1 which is up to 3 times higher than the overflow rate of a conventional secondary clarifier. Compared to an average overflow rate of 22 m d1 with a secondary clarifier, the space consumption for sludge-water separation in an MBR is 10–60% lower when flux is 15 l m2 h1 and 50–80% lower when flux is 25 l m2 h1. A further reduction in footprint is caused by the higher MLSS concentration that can be applied in an MBR. This estimate however did not take into account backflushing or relaxation periods, which reduce the overflow rate. Nevertheless, full-scale MBR plants also manifest these space-saving characteristics. For instance, Brescia WWTP, in Italy, which is the world’s largest MBR retrofit of an existing conventional plant, gives a full-scale example of a ratio of 2 when comparing area needed by CAS and MBR (Brepols et al., 2008).
4.16.3.2.5 Wastewater treatment cost
Relative cost (1994 cost equals 1)
The high cost connected with MBR is often mentioned in discussions about applicability of MBR. However, it is not easy to make a general economical comparison between MBR and CAS systems. First of all, the reference system should not simply be an activated sludge system, but a system that produces an effluent of the same quality. Moreover, an MBR is a modular system, that is, easily expandable, which is often mentioned as an advantage of the system. However, this makes the system less competitive with conventional systems, since these become relatively less expensive per population equivalent (p.e.) at larger scale. It should be noted that although the equipment and energy costs of an MBR are higher than systems used in conventional treatment, total water costs can be competitive due to the lower footprint and installation costs (Pearce, 2008b; Lesjean et al., 2004; Cote et al., 2004; De Wilde et al., 2003). MBR costs have declined sharply since the early 1990s, falling typically by a factor of 10 in 15 years. As MBR technology has become accepted, and the scale of installations has increased, there has been a steady downward trend in membrane prices (Figure 6), which is still continuing. This is particularly notable with the acceptance of the MBRs in the municipal sector. The uptake of membrane technology for municipal applications has had the affect of
1.0 Membrane cost (per unit flow rate) 0.8
579
downward pressure on price. A detailed holistic cost comparison may reveal reasonably comparable results between the cost of the MBR option versus other advanced treatment options, especially if land value is considered. Studies show that depending on the design and site-specific factors the total water cost associated with MBR may be less or higher than the CAS-UF/MF option. For example, a cost comparison by the US consultant HDR in 2007 showed that MBR was 15% more expensive on a 15 million liters a day (MLD) case study, whereas a study by Zenon in 2003 gave MBR 5% lower costs (Pearce, 2008a). The differences were due to the design fluxes assumed and the capital charge rate for the project. Neither study allocated a cost advantage from the reduced footprint, which could typically translate to a treated water cost saving of up to 5%. It is interesting to evaluate the development in cost estimates over the past several years. Davies et al. (1998) made a cost comparison for two wastewater treatment plants (WWTPs), with capacities of 2350 and 37 500 p.e. With the assumptions they made (e.g., a membrane lifetime of 7 years) they conclude that depending on the design capacity (i.e., 2 times DWF to be treated) MBR is competitive with conventional treatment up to a treatment capacity of 12 000 m3 d1 (Table 2). Engelhardt et al. (1998) after carrying out pilot experiments also made a cost calculation for an MBR with a capacity of 3000 p.e., designed for nitrification/denitrification and treatment of 2*DWF. Investment costs were estimated at h3104 000 (including pretreatment) and operational cost at h194 000 yr1. Adham et al. (2001) made a cost comparison between MBR oxidation ditch followed by membrane filtration and CAS followed by membrane filtration. They concluded that MBR is competitive with the other treatment systems (Table 3). Chang et al. (2001) report experiments with low-cost membranes. The effect of membrane cost on the investment cost is considerable, but operational problems hinder further application of low-cost membranes. A drawback of the applied membranes is its limited disinfecting capacity. Van Der Roest et al. (2002a) described a cost comparison between an MBR installation and a CAS system with tertiary sand filtration. The calculations were carried out for two new WWTPs with the aim of producing effluent with low
Table 2 Capital and operating cost ratios of MBR and conventional activated sludge (CAS) process assuming a capacity of 2*(dry weather flow) Parameter
Cost ratio (MBR:ASP)
Capital cost 2350 p.e 37 500 p.e Operating costs per year
0.63 2.00
2350 p.e 37 500 p.e
1.34 2.27
0.6 0.4 0.2 0.0 (1994)
(1995)
(1997)
(1999)
(2000)
Figure 6 Sharp cost decline of membranes for MBR (cost of Zenon membranes as an example).
Data from Davies WJ, Le MS, and Heath CR (1998) Intensified activated sludge process with submerged membrane microfiltration. Water Science and Technology 38(5): 421–428.
580
Membrane Biological Reactors
concentrations of nitrogen and phosphorus. Almost the same investment costs and 10–20% higher operating costs, depending on the capacity of the plant, for MBR were estimated (Table 4). Cost differences between an MBR and a traditional WWTP concerning manpower, chemical consumption, and sludge treatment were noted to be minimal. WERF (2001) summarized operating and water-quality data obtained over 1 year from two MBR pilot plants located at the Aqua 2000 Research Center at the City of San Diego (California) North City Plant. Preliminary cost estimates of the MBR technology were also developed. MBRs demonstrated that their effluent was suitable to be fed directly into an RO process from a particulate standpoint with silt density index (SDI) values averaging well below 3. The MBR effluent water quality was superior to the quality of a full-scale tertiary conventional WWTP. The preliminary cost estimate in this report was performed for a 1 million gallons a day (mgd) scalping facility (WWTP drawing a designated amount of flow from the sewer system; excess sewage flow is treated at another plant located at the end of the sewer line). This facility produced an effluent suitable as feedwater for an RO process. Based upon this estimate, the present value was estimated as $0.81 m3, $0.96 m3, and $1.16 m3 for the MBR process, oxidation ditch with MF, and oxidation ditch with conventional tertiary lime pre-treatment, respectively. Therefore, the MBR process was reported as the most cost-effective alternative for water reclamation where demineralization or indirect drinking water-production (RO) is required. McInnis (2005) reported a detailed comparative cost analysis of two membrane-based tertiary treatment options: (1)
Table 3 Capital and total cost ratios of MBR and tertiary MF following alternative biological processes Alternatives
Oxidation ditch-MF CAS-MF
Cost ratio (MBR:alternative) Capital
Total per year
0.91 0.85
0.89 0.9
Data from Adham S, Mirlo R, and Gagliardo P (2000) Membrane bioreactors for water reclamation – phase II. Desalination Research and Development Program Report No. 60, Project No. 98-FC-81-0031. Denver, CO: US Department of the Interior, Bureau of Reclamation, Denver Office.
MBR and, (2) CAS process followed by MF (CAS/MF). According to that study, irrespective of design flow rate, the MBR entails slightly higher unit capital costs as compared to CAS/ MF process, while, depending on the design flow rate, the operation and maintenance costs (O&M) of the former are higher than or comparable to that of the latter. Comparative O&M cost breakdown revealed that MBR entails less labor cost, considerably higher power and chemical consumptions and slightly higher membrane cost, other costs remaining virtually the same. In the CAS/MF process, labor cost induces the highest cost, while in case of the MBR process, labor and electrical power-consumption costs are almost similar. Overall, the MBR imposes slightly higher capital and operating/ maintenance cost over that of CAS/MF. Cote et al. (2004) explored two membrane-based options available to treat sewage for water reuse, tertiary filtration (TF) of the effluent from a CAS process, and an integrated MBR. These options were compared from the point of view of technical performance and cost using ZeeWeed immersed membranes. The analysis showed that an integrated MBR is less expensive than the CAS-TF option. The total life cycle costs for the treatment of sewage to a quality suitable for irrigation reuse or for feeding RO decreased from 0.40$ m3 to 0.20$ m3 as plant size increased to 75 000 m3 d1. It was also shown that the incremental life-cycle cost to treat sewage to indirect potable water-reuse standards (i.e., by UF and RO) was only 39% of the cost of seawater desalination. A recent market research report (BCC Research, 2008) estimated the capital cost of a 50 000 gallons per day (gpd) (190 m3 d1) plant at US$350 000, a 100 000 gpd plant at US$500 000, and a 500 000 gpd plant at US$2 million. For systems of 1 mgd (million gallons per day) and larger, capital costs start at US$3.5 million (Table 5). The largest percentage of new system installations, 93%, continue to fall into the 5000–500 000 gpd range (most of those, about 57% of them, have capacities of less than 25 000 gpd), 2% of installations range from 0.5–1 mgd, and 5% of them are larger than 1 mgd. Tables 2–5 list cost values reported during the period 1998–2008. Obviously, the data from the initial stage of the MBR development holds little relevance today. However, these are listed here to provide a general trend of cost-data evolution.
4.16.3.2.6 Comparative energy usage Table 4 Capital and total cost ratios of MBR and tertiary sand filtration following CAS Parameter
Cost ratio
Capital cost 10 000 p.e 50 000 p.e Operating costs per year
0.92 1.01
10 000 p.e 50 000 p.e
1.09 1.21
Data from Van Der Roest HF, Lawrence DP, and Van Bentem AG (2002a) Membrane Bioreactors for Municipal Wastewater Treatment (Water and Wastewater Practitioner Series: Stowa Report). London: IWA.
MBR provides an equivalent treatment level to CAS-UF/MF, but at the expense of higher energy cost since the efficiency of air usage in MBR is relatively low. The MBR process uses more Table 5
Capital cost of MBR depending on plant sizea
Plant size, gpd 103
Capital cost, US$ 103
50 100 500 1000
350 500 2000 3500
a
1 m3 d1 ¼ 264.17 gpd. Data from BCC Research (2008) Membrane bioreactors: Global markets. Report Code MST047B, Report Category – Membranes & Separation Technology.
Membrane Biological Reactors
air, and hence higher energy than conventional treatment. This is because aeration is required for both the biological process and the membrane cleaning, and the type, volume, and location of air required for the two processes are not matched. Biotreatment utilizes fine air bubbles, since oxygen needs to be absorbed for the biological reaction step. In contrast, fouling control is best achieved by larger bubbles, since the air is required to scour the membrane surface or shake the membrane to remove the foulant. Accordingly, although the concept of MBR was first developed to exploit the fact that the biological wastewater-treatment process and the process of membrane-fouling control can both use aeration (Pearce, 2008b), the potential for dual-purpose aeration is strictly limited. Based on a survey of conventional wastewater-treatment facilities in the US, Metcalfe and Eddy, Inc. (2003) reported that the energy usage range was 0.32–0.66 kW h1 m3. Energy usage in wastewater treatment is somewhat lower in Europe, partly due to a greater consciousness for energy efficiency, and partly due to the fact that average BOD loading/ capita in the US is 20–25% greater than that in Europe (due to the use of kitchen disposal units). Long-term monitoring of wastewater-treatment systems has shown usages as low as 0.15 kW h1 m3 for activated sludge, increasing to 0.25 kW h1 m3 if a biological aerated filter (BAF) stage is included (Pearce, 2008a). Membrane filtration after conventional treatment is estimated to add 0.1–0.2 kW h1 m3 to the energy, equivalent to a total energy use for CAS-UF/MF of 0.35–0.5 kW h1 m3 in a new facility (Lesjean et al., 2004). Experience in large-scale commercial MBRs shows an energy usage of around 1.0 kW h1 m3, although smaller-scale facilities typically operate at 1.2–1.5 kW h1 m3 or higher (Judd, 2006). However, in comparison to these values, energy consumption of around 1.9 kW h1 m3 was reported in 2003 (Zhang et al., 2003) and up to 2.5 kW h1 m3 in 1999 (Ueda and Hata, 1999). This proves that there is a gradual improvement in MBR design (Figure 7). Further improvements in air efficiency and membrane-packing density are expected
Energy consumption, kW hr−1 m−3
3.0
2.0
1.0
to improve the current values in the future. Even so, it seems likely that MBR energy costs will continue to exceed those of CAS-UF/MF by 0.4 kW h1 m3 or more (Pearce, 2008a). However, the fact that membrane filtration after conventional treatment is estimated to add only 0.1–0.2 kW h1 m3 to the energy points out that the higher energy consumption of MBR over CAS-UF/MF is due to the difference in consumption in the respective biological processes. MBRs are generally operated at quite low F/M ratios (less than 0.2), or high MLSS concentrations, and this is one of the reasons for the excellent biodegradation efficiency, and high aeration cost as well. CAS plants, on the other hand, are operated at higher F/M ratios, implying lower oxygen need for biodegradation. Table 6 lists typical energy-use rates of different biologicalbased treatment combinations. Section 4.16.5 provides further information on energy comparison of the MBR formats.
4.16.3.3 Relative Advantages of MBR There are several advantages associated with the MBR technology, which make it a valuable alternative over other treatment techniques. The combination of activated sludge with membrane separation in the MBR results in efficiencies of footprint, effluent quality, and residual production that cannot be attained when these same processes are operated in sequence. The MBR system is particularly attractive when applied in situations where long biological solid-retention times are necessary and physical retention and subsequent hydrolysis are critical to achieving biological degradation of pollutants (Chen et al., 2003). The prime advantages of MBR are the treated water quality, the small footprint of the plant, less sludge production, and flexibility of operation (Visvanathan et al., 2000). First of all, the retention of all suspended matter and most of the soluble compounds within the bioreactor leads to excellent effluent quality capable of meeting stringent discharge requirements and paving the way for direct water reuse. The possibility of retaining all bacteria and viruses results in a sterile effluent, eliminating extensive disinfection and the corresponding hazards related to disinfection by-products. As the entire process equipment can be made airtight, odor dispersion can be prevented quite successfully. Since suspended solids are not lost in the clarification step, total separation and control of the SRT and hydraulic retention time (HRT) are possible enabling optimum control of the microbial population and flexibility in operation. The absence of a clarifier, which also acts as a natural selector for settling organisms, enables sensitive, slow-growing
0.0 1999
2003 Year
2006
Figure 7 Gradual reduction in reported values of energy consumption by MBR. Data from Ueda T and Hata K (1999) Domestic wastewater treatment by a submerged membrane bioreactor with gravitational filtration. Water Research 33: 2888–2892; Zhang SY, Van Houten R, Eikelboom DH, et al. (2003) Sewage treatment by a low energy membrane bioreactor. Bioresource Technology 90: 185–192; and Judd S (ed.) (2006) The MBR Book: Principles & Applications of MBRs in Water & Wastewater Treatment. Oxford: Elsevier.
581
Table 6 Comparative typical energy consumption by different treatment options Treatment option
Energy use (kW h1 m3)
CAS CAS-BAF CAS-MF/UF MBR
0.15 0.25 0.35–0.5 0.75–1.5a
a
Power consumption range for large- to smaller-scale plants.
582
Membrane Biological Reactors
species (nitrifying bacteria, bacteria capable of degrading complex compounds) to develop and persist in the system (Cicek et al., 2001; Rosenberger et al., 2002). The membrane not only retains the entire biomass but also prevents the escape of exocellular enzymes and soluble oxidants creating a more active biological mixture capable of degrading a wider range of carbon sources (Cicek et al., 1999b). MBRs eliminate process difficulties and problems associated with settling, which is usually the most troublesome part of wastewater treatment. The potential for operating the MBR at very high SRTs without the obstacle of settling allows high biomass concentrations in the bioreactor. Consequently, higher-strength wastewater can be treated and lower biomass yields are realized (Muller et al., 1995). This also results in more compact systems than conventional processes, significantly reducing plant footprint and making it useful in waterrecycling applications (Konopka et al., 1996). The low sludge load in terms of BOD forces the bacteria to mineralize poorly degradable organic compounds. The higher biomass loading also increases shock tolerance, which is particularly important where feed is highly variable (Xing et al., 2000). The increased endogenous (autolytic) metabolism of the biomass (Liu and Tay, 2001) under long SRT allows development of predatory and grazing communities, with the accompanying trophiclevel energy losses (Ghyoot and Verstraete, 1999). These factors, in addition to resulting in lower overall sludge production, lead to higher mineralization efficiency than those of a CAS process. High molecular weight soluble compounds, which are not readily biodegradable in conventional systems, are retained in the MBR (Cicek et al., 2002). Thus, their residence time is prolonged and the possibility of oxidation is improved. The system is also able to handle fluctuations in nutrient concentrations due to extensive biological acclimation and retention of decaying biomass (Cicek et al., 1999a).
4.16.3.4 Factors Influencing Performance/Design Considerations This section sheds light on some important design considerations of MBR. More detailed information on some of these parameters is provided in Section 4.16.4.7, in relation to membrane fouling.
4.16.3.4.1 Pretreatment All MBRs require pretreatment, for example, screening and grit removal, to protect the membranes. Screening has historically been limited to 3 mm; however, hair and fiber can still pass through this size of the screen and become embedded or wrapped around the hollow fibers. The MBR providers have standardized their screen selections to a 2-mm traveling band, punched screen. Conversely, the flat-sheet membranes experience less problems with hair and fiber, and are standardized to a 3-mm screen. Further discussion regarding mechanical pretreatment is provided in Section 4.16.4.6.
4.16.3.4.2 Membrane selection and applied flux An MBR membrane needs to be mechanically robust, chemically resistant to high Cl2 concentrations used in cleaning, and nonbiodegradable (Pearce, 2008a). Clean-water permeability
is not as important in an MBR as in membrane-filtration applications, since the membrane transport properties are strongly influenced by the accumulation of foulant particles at the membrane surface. However, process flux in treating a wastewater feed is important since it directly affects capital cost, due to its effect on membrane area and footprint, and operating costs due to the effect of membrane area on chemical and air use. Most MBRs operate at an average flux rate between 12.5 and 25 l m2 h1, with Mitsubishi’s unit operating in the lower range. The key flux rates that determine the number of membranes required are associated with the peak flow rates. For plants with peaking factors of less than two, an MBR can handle the plant flow variation without having a significantly impact on the average design flux rate. Otherwise, equalization needs to be provided with either a separate tank at the head of the facility or within the aeration basin, allowing sidewater depth variations during peak flow.
4.16.3.4.3 Sludge retention time In the past, most MBR systems were designed with extremely long SRTs, of the order of 30–70 days, and very few were operated at less than about 20 days. Two reasons prompted such practice: (1) the drive to minimize sludge production or eliminate it all together and (2) the concern over the reduced flux resulting from short SRT operation, presumably due to the fouling effect of extracellular excretions from younger sludge. Currently, the selection of SRT is based more on the treatment requirements, and SRTs as low as 8–10 days can now be contemplated.
4.16.3.4.4 Mixed liquor suspended solids concentration From the point of view of bioreactor volume reduction and minimization of excess sludge, submerged MBR systems have been typically operated with MLSS concentrations of more than 12 000 mg l1, and often in the range of 20 000 mg l1. Hence, they offer greater flexibility in the selection of the design SRT. However, excessively high MLSS may render the aeration system ineffective and reduce membrane flux. A trade-off, therefore, comes into play. Current design practice is to assume the MLSS to be closer to 10 000 mg l1 to ensure adequate oxygen transfer and to allow for higher membrane flux. With larger systems, it is more cost effective to reduce the design MLSS because of the high relative cost of membranes when compared to the cost of additional tank volume.
4.16.3.4.5 Oxygen transfer At high MLSS concentrations, the demand for oxygen can be significant. In some cases, the demand can exceed the volumetric capacity of typical oxygenation systems. The oxygentransfer capacity of the aeration system must also be carefully analyzed. Submerged membranes are typically provided with shallow coarse bubble air to agitate the membranes as a means to control fouling. Such aeration provides some oxygenation, but at low efficiency. In compact systems, fine bubble aeration may be placed at greater depth below the membrane aeration; however, the combined efficiency and the bubblecoalescing effects require further consideration during design (Visvanathan et al., 2000).
Membrane Biological Reactors
The lower operating cost obtained with the submerged configuration along with the steady decrease in the membrane cost encouraged an exponential increase in MBR plant installations from the mid-1990s onward. Since then, further improvements in the MBR design and operation have been introduced and incorporated into larger plants. The key steps in the recent MBR development are summarized below:
• •
•
The acceptance of modest fluxes (25% or less of those in the first generation), and the idea of using two-phase bubbly flow to control fouling. While early MBRs were operated at SRTs as high as 100 days with MLSS up to 30 g l1, the recent trend is to apply a lower SRT (around 10–20 days), resulting in more manageable MLSS levels (10–15 g l1). Thanks to these new operating conditions, the fouling propensity in the MBR has tended to decrease and overall maintenance has been simplified, as less-frequent membrane cleaning is necessary.
Further discussion on these aspects is provided in the following sections.
4.16.4 Worldwide Research and Development Challenges 4.16.4.1 Importance of Water Reuse and the Role of MBR The need for pure water is a problem of global proportions. In the Earth’s hydrologic cycle, freshwater supplies are fixed and constant, while global water demand is growing (Howell, 2004; Bixio et al., 2006). With each passing year, the quality of the planet’s water measurably deteriorates, presenting challenges for the major users: the municipal, industrial, and environmental sectors. Increasing demand for water, and drought and water scarcity are now common issues facing many urban and rural communities around the world (Howell, 2004; Tadkaew et al., 2007; Jimenez and Asano, 2008). Water treatment has, therfore, become an area of global concern as individuals, communities, industries, countries, and their national institutions strive for ways to keep this essential resource available and suitable for use. Water recycling is a pragmatic and sustainable approach for many countries to mitigate or solve the problems of water supply. There is a growing interest in using nontraditional water resources by means of water reclamation and water recycling for long-term sustainability. It can be divided into two categories, internal domestic or industrial recycling and external recycling, where high-quality reclaimed water from a sewage treatment plant is used for aquifer recharge or irrigation. With the current focus on water-reuse projects and the role they play in the water cycle, the search for cost-competitive advanced wastewater-treatment technologies has never before been so important. Treatment technology for water recycling encompasses a vast number of options. A general paucity of legislative and socioeconomic information has led to the development of a diverse range of technical solutions (Jefferson et al., 2000). Membrane processes are regarded as key elements of advanced wastewater reclamation and reuse schemes and are included in a number of prominent schemes worldwide, for example, for artificial groundwater recharge, indirect
583
potable reuse, as well as for industrial-process water production (Melin et al., 2006; Bixio et al., 2008). Among the many treatment alternatives, MBRs, which combine membrane filtration and biological process for wastewater treatment, are seen to have an effective technology capable of transforming various types of wastewater into high-quality effluent exceeding most discharge requirements and suitable for a variety of nonpotable water-reuse applications such as flushing toilets and for irrigation (Tadkaew et al., 2007; Jimenez and Asano, 2008). In some cases, treated water can be applied to recharge groundwater to halt saltwater intrusion into coastal aquifers, abate subsidence in areas sinking due to overpumping groundwater, and support aquifer storage and recovery. Issues of water quality, water quantity, and aging/nonexistent infrastructure propel the market for MBRs. Escalating water costs due to dwindling supplies for communities and businesses also drive the growing acceptance of MBRs. Anticipated stricter environmental regulations are driving sales of MBRs to industry, municipalities, and are prompting maritime users to consider MBR technology (Jefferson et al., 2000; Jimenez and Asano, 2008). This is probably due to the effectively disinfected high-quality effluent and high performance in trace organic removal for safe and environmentally benign discharge that MBRs can offer. In practical terms, the process has many benefits, which make it suitable for the size of the systems applicable to recycling. The ability to run independently of load variation and produce no sludge are critical and highlight MBRs as possibly the most viable small-footprint, high-treatment option for water recycling (Jefferson et al., 2000; Melin et al., 2006; Tadkaew et al., 2007). Comparison with other technologies used for water recycling reveals that MBRs not only produce lower residual concentrations but do so more robustly than the alternatives (Jefferson et al., 2000; Melin et al., 2006). The favorable microbiological quality of the effluent of MBRs is a major factor in their frequent selection for water reuse, even if full disinfection cannot be expected, particularly considering the distribution and storage components of a full-scale system, which can be prone to regrowth of microorganism and contamination from various sources. However, the MBR effluent is adequate for many water-reuse applications with little residual chlorine disinfection for subsequent distribution. The MBR then does provide a dual layer of protection against pathogen breakthrough, greatly lowering the risk during operation. MBRS have the greatest efficacy toward water recycling, albeit contingent upon a loading rate constrained by the operable flux. Not only do they comply with all likely waterquality criteria for domestic recycling but they also produce a product that is visibly clear and pathogen free, both of which are likely to be key concerns in terms of public acceptability. There are some issues that still need to be addressed and these are highlighted throughout Sections 96.4.6 and 96.4.7 of this chapter.
4.16.4.2 Worldwide Research Trend Early development efforts in MBR technology were concentrated in UK, France, Japan, and South Korea, whereas extensive research in China and Germany began after 2000. Much
584
Membrane Biological Reactors
of the research in the newcomer countries is building on pioneering work from the UK, France, Japan, and South Korea. Three stages may be identified in the worldwide MBR research: 1. An entry-level stage spanning from 1966 to 1980, during which lab-scale research was mainly conducted. Membranes of that period had low flux and short life span due to undeveloped membrane-manufacturing technology. 2. A slow-to-moderate growth period from 1980 to 1995, when MBR technology was well investigated especially in Japan, Canada, and the USA. During this stage, new membrane-material development, MBR configuration design, and MBR operation were critically studied. The submerged MBR concept was put forward by Japanese researchers in 1989. 3. The rapid development stage started in 1995 and continues even now, when MBR technology underwent a rapid development prompted by deep understanding of the technology in research communities and by the installation of full-scale MBRs. Much of the published information on MBRs to date has mainly focused on bench or pilot-scale studies, performance results of treating a specific type of wastewater, and short-term operations. Regardless of the source of wastewater, whether it is municipal or industrial, very few publications involved fullscale studies for long-term operational periods. In a comprehensive review, Yang et al. (2006) grouped the available worldwide publications regarding MBR into six main research areas: (1) literature and critical reviews; (2) fundamental aspect; (3) municipal and domestic wastewater treatment; (4) industrial wastewater and landfill leachate treatment; (5) drinking-water treatment; and (6) others, which include gas removal, sludge treatment, hydrogen production, and gas diffusion. The fundamental research category was based on studies that exclusively looked at membrane fouling, operation and design parameters, sludge properties, microbiological characteristics, cost, and modeling. Studies, which focused on applied research and general reactor performance, were categorized by influent (feed) type (groups 3–6). Membrane fouling, which has been widely considered as one of the major limitations to faster commercialization of MBRs, has been investigated from various perspectives including the causes, characteristics, mechanisms of fouling, and methods to prevent or reduce membrane fouling. More than one-third of studies in the fundamental aspects group were found to deal with issues related to membrane fouling.
4.16.4.3 Modeling Studies on MBR Models that can accurately describe the MBR process are important for the design, prediction, and control of MBR systems. Due to the intrinsic complexity and uncertainty of MBR processes, basic models that can provide a holistic understanding of the technology at a fundamental level are of great necessity. Complex models that are also practical for real applications can greatly assist in capitalizing on the benefits of MBR technology. However, compared to experimental R&D, followed by commercialization of the technology, modeling studies for system-design analysis and performance prediction
are at a relatively preliminary stage. In an attempt to identify the required research initiatives in this regard, this section looks briefly into the state-of-the-art MBR modeling efforts. Effluent quality and the investment and operating costs are the primary concerns for any given wastewater treatment system. Therefore, model development should center on components for which water-quality standards have been set and parameters which are strongly correlated to cost. Ng and Kim (2007) put forward a few key model components and parameters for MBR modeling:
•
•
•
•
•
The ability to quantify individual resistance (i.e., resistance from cake formation, biofilm formation, and adsorptive fouling) as a function of the various influencing parameters is important in determining which parameters have the greatest influence on fouling and for designing and optimizing the system to achieve an economical balance between production and applied pressure. Determining the relationship between biomass concentration and other parameters can aid in identifying an optimal biomass concentration for operation, which can lead to significant economical savings. Aeration accounts for a significant portion of energy costs in the operation of MBR systems. The factors that influence oxygen requirement (wastewater and biomass concentration/growth rates) and the oxygen-transfer rate (MLSS concentration, MBR configuration, type of bubbles used, and specific airflow rate) should receive due consideration in the model to optimize aeration. Carbon and nutrient (nitrogen and phosphorous components) concentrations and their influencing factors (e.g., respective concentrations and growth rates of the various types of organisms and concentration of oxygen) should be incorporated into the models. Soluble microbial products (SMPs), which comprise a major portion of the organic matter in effluents from biological treatment processes and are potentially associated with issues such as disinfection by-product formation, biological growth in distribution systems, and membrane fouling, should be given proper consideration in models.
MBR models available in the literature can be broadly classified into three categories: biomass kinetic models, membranefouling models, and integrated models to describe the complete MBR process (Ng and Kim, 2007; Zarragoitia-Gonza´lez et al., 2008). Models describing biomass kinetics in an MBR include the activated sludge model (ASM) family (Henze et al., 2000), the SMP model (Furumai and Rittmann, 1992; Urbain et al., 1998; de Silva et al., 1998), and the ASM–SMP hybrid model (Lu et al., 2001; Jiang et al., 2008). The ASMs were developed to model the activated sludge process. The MBR process is the activated sludge process with the secondary clarification step replaced by membrane filtration; therefore, it is reasonable to use ASMs to characterize the biomass dynamics in an MBR system. However, their ability to describe the MBR process accurately has not been verified by in-depth experiments. Research suggests that SMPs are important components in describing biomass kinetics due to high SRTs in MBR systems. Accordingly, the SMP model demonstrated the capability of
Membrane Biological Reactors
characterizing the biomass with a reasonable-to-high degree of accuracy. Lu et al. proposed that the modified versions of ASM1 (Lu et al., 2001) and ASM3 (Lu et al., 2002), which incorporate SMPs, demonstrated fairly reasonable accuracy in quantifying COD and soluble nitrogen concentrations. Jiang et al. (2008) extended the existing ASM No. 2d (ASM2d) to ASM2dSMP with introduction of only four additional SMPrelated parameters. In addition to minimizing model complexity and parameter correlations, the model parameter estimation resulted in reasonable confidence intervals. Models describing membrane fouling include the empirical hydrodynamic model (Liu et al., 2003), fractal permeation model (Meng et al., 2005), sectional resistance model (Li and Wang, 2006), subcritical fouling behavior model (Saroj et al., 2008), and the resistance-in-series models that were presented as a part of the integrated models. Some of them are simply based on solid–liquid separation and simulate filtration processes (Chaize and Huyard, 1991; Gori et al., 2004). Other models consider specific physical approaches: cross-flow filtration (Cheryan, 1998; Hong et al., 2002; Beltfort et al., 1994) and mass-transport models (Beltfort et al., 1994; Bacchin et al., 2002). Nevertheless, membrane fouling is generally evaluated by employing the resistance-in-series model (Wintgens et al., 2003; Wisniewski and Grasmick, 1998) or, rarely, using empirical models (Benitez et al., 1995; De Wilde et al., 2003). The integrated models, basically, couple the kinetic models with the fouling ones (such as the resistance-in-series model) and they often consider the formation and degradation of SMPs (Ng and Kim, 2007). The models reported to date are valuable preliminary attempts, but require further improvements. For instance, the empirical hydrodynamic model is too simple to describe the membrane-fouling phenomenon, and the sectional resistance model lacks accuracy. Both the fractal permeation model and resistance-in-series model by Lee et al. (2002) provide good scientific insight, but specific experimental verification is necessary for general use of the models. The resistance-in-series model developed by Wintgens et al. (2003) shows the most promise, as it is fairly accurate, accounts for cleaning cycles, and can predict permeability changes over time. Further tests are needed to determine whether the model requires calibration or if the model parameters are applicable to other MBR systems. Recently, Zarragoitia-Gonza´lez et al. (2008) included the biological kinetics and the dynamic effect of the sludge attachment and detachment from the membrane, in relation to the filtration and a strong intermittent aeration in a hybrid model. The model was established considering SMP formation–degradation kinetic based on previous published models (Cho et al., 2003; Lu et al., 2001). A modification of Li and Wang’s model (Li and Wang, 2006) allows to calculate the increase of the transmembrane pressure (TMP), evaluating, at the same time, the influence of an intermittent aeration of bubbles synchronized with the filtration cycles on fouling control, and to analyze the effects of shear intensity on sludge cake removal. On the other hand, in order to describe the biological system behavior, a modified ASM1 model was used. The final hybrid model was developed to calculate the evolution of sludge properties, its relation to sludge cake growth, and the influence of sludge properties on membrane fouling.
585
A simple model for evaluating energy demand arising from aeration of an MBR was presented by Verrecht et al. (2008) based on a combination of empirical data for membrane aeration and biokinetic modeling for biological aeration. The model assumes that aeration of the membrane provides a portion of the dissolved oxygen needed for biotreatment. The model also assumes, based on literature information sources, a linear relationship between membrane permeability and membrane aeration up to a threshold value, beyond which permeability is unchanged with membrane aeration. An analysis reveals that significant reductions in energy demand are attained through operating at lower MLSS levels and membrane fluxes. The complete organic removal in MBR is due to all the inseries phenomena: biological degradation of biomass, biological filtration of cake layer, and final filtration of physical membrane. Di Bella et al. (2008) set up a mathematical model for the simulation of physical–biological wastewater organic removal for SMBR system. The model consists of two submodels: the first one for the simulation of the biological processes and a second one for the physical processes. In particular, regarding the biological aspects, it is based on the ASM concept. On the other hand, organic-matter removal due to filtration (the physical process) was described by simple models proposed in the literature (Kuberkar and Davis, 2000; Jang et al., 2006; Li and Wang, 2006). It is conceivable that several of the existing models, particularly the ASMs, require validation to determine their applicability for modeling the MBR process and to evaluate whether they can serve as a base for future MBR model development. Membrane fouling in MBRs is affected by the biotransformation processes in the system; therefore, a more effective integration of biomass kinetics and membrane fouling into the models is required. Moreover, examination of alternative empirical modeling approaches, such as the application of artificial neural networks, is worthwhile to establish a thorough link between inputs and outputs of MBR systems and to find phenomenological interrelationships among components and parameters (Ng and Kim, 2007).
4.16.4.4 Innovative Modifications to MBR Design Researchers have put forth different modifications to the conventional design of MBRs in order to enhance removal performance and/or mitigate membrane fouling. This section highlights some of such examples (Table 7). The commercialized MBR formats are discussed separately in Section 4.16.5.2.
4.16.4.4.1 Inclined plate MBR Theoretically, an infinite SRT provides a possibility of naturally achieving zero-excess sludge discharge from MBR under normal environment. It should, however, be noted that zeroexcess sludge production is just a theoretical concept which can only be obtained with a feed containing only solutes. In real life, sewage or industrial effluents contain nonbiodegradable suspended solids and colloids that accumulate in the reactor, continuously increasing the sludge concentration. Therefore, an immediate challenge encountered at infinite SRT is the extremely high sludge concentration
586
Membrane Biological Reactors
Table 7
Examples of innovative modifications to MBR design
Modified design
Main purpose
Selected reference
Inclined plate MBR
Omit excess sludge production and thereby realize long-term stable membrane filtration Derive the simultaneous advantages of efficient nutrient removal and mitigate membrane fouling (Chae et al., 2006 a,b;). Treatment of high-strength wastewater without encountering severe fouling Enhanced removal of recalcitrant compound and/or membrane fouling mitigation Obtain in one step indirect potable reuse standard effluent Indirect potable reuse along with energy demand reduction Obtain in one step indirect potable reuse standard effluent
Xing et al. (2006)
Integrated anoxic–aerobic MBR
Jet-loop-type MBR Biofilm MBR Nanofiltration MBR Forward osmosis MBR Membrane distillation bioreactor (MDBR)
produced in the bioreactor (Wen et al., 1999). Consequently, the method to achieve zero-excess sludge discharge translates into how to realize long-term stable membrane filtration of high-concentration sludge beyond the guideline value of 10– 20 g l1 recommended for submerged MBRs when applied to domestic wastewater treatment. In order to omit excess sludge production, Xing et al. (2006) proposed an innovative MBR design comprising an anoxic tank equipped with settlingenhancer inclined plates and a subsequent aerobic tank containing the membrane. The inclined plates together with intermittent air blowing (to blow off gaseous content generated by denitrification, etc.) proved to be quite effective in confining high MLSS sludge within the anoxic tank leading to an MLSS difference of 0.1– 13.1 g l1 between the aerobic and anoxic sludge. Consequently, the capability of MBRs in handling the extremely high MLSS challenge encountered especially at zero-excess sludge could be extended. Results indicated that at an HRT of 6 h, average removals of COD, ammonia nitrogen, and turbidity were 92.1, 93, and 99.9%, resulting in daily averages of 12.6 mg COD l1, 1.3 mg NH3–N l1, and 0.03 NTU, respectively.
4.16.4.4.2 Integrated anoxic–aerobic MBR In contrast to separate anoxic tanks for denitrification or creation of alternating anoxic/oxic conditions within the same tank by intermittent aeration, an integrated anoxic/oxic MBR, containing anoxic/oxic compartments in one reactor, was developed to derive simultaneous advantages of efficient nutrient removal (Chae et al., 2006a, 2006b) and mitigated membrane fouling (Chae et al., 2006a, 2006b; Hai, 2007; Hai et al., 2007; Hai et al., 2006b; Hai et al., 2008a). Under the optimal volume ratio of anoxic and oxic zones of 0.6 and the desirable internal recycle rate and HRT of 400% and 8 h, respectively, the average removal efficiencies of total nitrogen (T-N) and total phosphorus (T-P) were 75% and 71%, respectively (Chae et al., 2006b). Furthermore, comparison with sequential anoxic/oxic MBR under the same conditions revealed the membrane-fouling reduction potential of this specific design (Chae et al., 2006a).
Chae et al. (2006a,b), Hai et al. (2006b, 2008a)
Park et al. (2005), Yeon et al. (2005) Lee et al. (2006), Leiknes and Odegaard (2007), Ngo et al. (2008), Hai et al. (2008) Choi et al. (2002) Achilli et al. (2009), Cornelissen et al. (2008) Phattaranawik et al. (2008, 2009)
Working with a high-strength industrial wastewater, Hai et al. (2006a, 2006b, 2008a) demonstrated minimization of excess sludge growth and maintenance of less MLSS concentration in contact with the membrane at the aerobic zone by exploring a similar reactor design along with a strategy of splitting the feed through the two zones.
4.16.4.4.3 Jet-loop-type MBR The so-called high-performance compact reactor (HCR) which is a jet-loop-type reactor with a draft tube and a two-phase nozzle was coupled with a submerged membrane by Park et al. (2005). The HCR is able to deal with very high organic loading rates due to the high efficiency of oxygen transfer, mixing, and turbulence achieved. The significant amount of bubbles and turbulence present in the HCR can be beneficial in retarding fouling of the submerged membrane. The developed MBR showed much greater membrane permeability than the conventional MBR, promising very high potential for the treatment of high-strength wastewater without encountering severe fouling (Park et al., 2005; Yeon et al., 2005).
4.16.4.4.4 Biofilm MBR Membrane-coupled moving-bed biofilm reactor system, wherein the membrane is submerged within the same tank (Lee et al., 2006) or in an additional tank (Leiknes and Odegaard, 2007), has been extensively studied in association with different kinds of biocarriers. Powdered activated carbon (PAC) which also acts as an adsorbent is commonly added into the bioreactor as the biocarrier (Ng et al., 2006; Hai, 2007; Hai et al., 2008b). However, carriers made of inert materials, such as plastic (Leiknes and Odegaard, 2007) and sponge (Lee et al., 2006; Ngo et al., 2008), have also been used. Biomass granulation with shell-support media coupled with membrane separation is also worth mentioning in this context (Thanh et al., 2008). The mechanisms of enhanced removal and/or membranefouling mitigation depend on the specific design and the utilized biocarrier type. For example, in an integrated membrane-coupled moving-bed biofilm reactor using sponge as the biocarrier, frictional force exerted by the circulating
Membrane Biological Reactors
carrier on the submerged membrane reduced the formation of cake layer on the membrane surface and thus enhanced the membrane permeability (Lee et al., 2006). On the other hand, Leiknes and Odegaard (2007) demonstrated that operation under high volumetric-loading rates of 2–8 kg COD m3 d1and HRTs up to 4 h and maintenance of membrane fluxes around 50 l m2 h1 were possible by placing the moving-bed biofilm reactor prior to the submerged MBR. The specific purpose of the biofilm reactor in this case was to reduce the organic loading on MBR. Ng et al. (2006) contend that the improved membrane performance of the MBR with added PAC could be due to a number of factors including, PAC providing sink for some of the fouling components and the scouring action of PAC. Hai et al. (2008b) reported that simultaneous PAC adsorption within a fungiMBR treating dye wastewater resulted in multiple advantages including co-adsorption of dye and fungal enzyme onto activated carbon and subsequent enzymatic dye degradation.
4.16.4.4.5 Nanofiltration MBR The potential for using NF technology in wastewater treatment and water reuse is noteworthy. A new concept with the addition of RO membrane after conventional MBR has been recently developed to reclaim municipal wastewater. The new MBR-RO process demonstrated the capability of producing the same or more consistent product quality (in terms of total organic carbon (TOC), NH4, and NO3) and sustained higher flux compared to the CAS-MF-RO process in reclamation of domestic sewage (Qin et al., 2006). Choi et al. (2002, 2007), on the other hand, demonstrated the technical feasibility of a submerged NF-MBR. For the initial 130 days, the NF-MBR achieved high permeate quality (DOC concentration ¼ 0.5–2.0 mg l1) and maintained reasonable water productivity. With low electrolyte rejection, operation under a low suction pressure was possible, and electrolyte accumulation in the bioreactor, which may hinder biological activity, did not occur. The permeate quality, however, deteriorated to some extent (DOC concentration ¼ 3.0 mg l1) due to the deterioration of the cellulose membrane.
4.16.4.4.6 Forward osmosis MBR The forward osmosis (FO)–MBR is an innovative technique for the reclamation of wastewater, which combines activated sludge treatment and FO membrane separation with an RO posttreatment. FO membranes, either submerged or external, are driven by an osmotic pressure difference over the membrane. Through osmosis, water is transported from the mixed liquor across the semipermeable membrane into a draw solution (DS) with a higher osmotic pressure. To produce potable water, the diluted DS is then treated in an RO unit, and the concentrated DS is reused in the FO process. The FO-MBR is expected to have the same advantages as conventional MBRs; however, it has to deal with the most important drawback, that is, a high energy demand. In this system, FO membranes with structures comparable with NF or RO membranes are used instead of MF/UF membranes for the separation of suspended solids, multivalent ions, natural organic matter, and biodegradable materials. Since fluxes are generally lower and no internal fouling occurs, fouling of NF
587
or RO membranes, compared to that of the MF or UF membranes in conventional MBR, may be dealt with easily. The RO system after FO-MBR can be operated with higher fluxes because all the bivalent ions are removed in the FO-MBR. Recent studies have demonstrated high sustainable flux and relatively low reverse transport of solutes from the DS into the mixed liquor, along with very high removal performance (Achilli et al., 2009; Cornelissen et al., 2008).
4.16.4.4.7 Membrane distillation bioreactor A novel wastewater-treatment process known as the membrane distillation bioreactor (MDBR) incorporating membrane distillation in an SMBR operated at an elevated temperature was developed and experimentally demonstrated by Phattaranawik et al. (2008, 2009). The ability of membrane distillation (MD) to transfer only volatiles means that very high quality treated water is obtainable, with TOC levels below 1 ppm and negligible quantity of salts. A unique feature is that the MDBR allows for organic retention times to be much greater than the HRT. The TOC in the permeate was consistently lower than 0.7 mg l1 for all experiments. Stable fluxes in the range 2–5 l m2 h1 have been sustained over extended periods. The MDBR was described to have the potential to achieve in a single step, the reclamation obtained by the combined MBR þ RO process. It was also suggested that for viable operation, it would be necessary to use low-grade (waste) heat and water cooling. Several other emerging approaches are also noticeable in contemporary literature. These include hybrid MBR-CAS concept (De Wilde et al., 2009), anaerobic baffled reactor-MBR combination (Pillay et al., 2008), etc.
4.16.4.5 Technology Benefits: Operators’ Perspective The relative advantages of MBR over the CAS process were outlined in Section 4.16.3.3. This section highlights the technical benefits of MBRs cited by the operators: 1. high-quality effluent, ideal for post membrane treatments (e.g., NF and UF); 2. space savings, enabling upgrading of plants without land expansion; 3. shorter start-up time compared to conventional treatment systems; 4. low operating and maintenance manpower requirement (average of 1.7 working hours per MLD); and 5. (5) automated control.
4.16.4.6 Technology Bottlenecks MBR technology is facing some research and development challenges. The technology bottlenecks as reported in the literature include (Howell, 2002, 2004; Lesjean et al., 2004; Le-Clech et al., 2005a; Yang et al., 2006; Melin et al., 2006) 1. Membrane fouling. Further understanding the mechanisms of membrane fouling and developing more effective and easier methods to control and minimize membrane fouling. 2. Pretreatment. Effective methods to limiting membrane clogging and operational failures.
588
Membrane Biological Reactors
3. Membrane life span. Increasing membrane mechanical and chemical stability. 4. Cost. Further reduction of costs for maintenance and replacement of membranes, energy requirement, and labor requirements. 5. Plant capacity. Scaling up for large plants. 6. Exchangeability of modules. Module exchangeability between different brands (reduction of costs for replacement of membranes). Some other problems often encountered by the operators include (Leslie and Chapman, 2003; Adham et al., 2004; LeClech et al., 2005a; Yang et al., 2006)
• • • • • • • • • •
membrane fouling during permeate backpulsing, entrained air impacting suction-pump operation, bioreactor foaming, inefficient aeration due to partial clogging of aerator holes, no significant decrease of biosolid production, scale buildup on membrane and piping, corrosion of concrete, hand rails, and metallic components due to corrosive vapor produced during high temperature NaOCl cleaning, membrane delamination and breakage during cleanings, odor from screening, compaction, drying beds, and storage areas (although normally less than in CAS), and failure of control system.
Although the commercialization of MBRs has expanded substantially in the past 20 years, target markets have not been tapped to a large extent and new potential areas of applications are continually developing. The R&D challenges mentioned above, when tackled, will lead to a more competitive and mature market for MBR applications. Lesjean et al. (2004) contend that academic research is addressing only some of these issues. For instance, while many publications on fouling are being produced and some cost studies are conducted, no significant research efforts have addressed membrane life span, pretreatment, and scale-up issues. Academic researchers can expect interest from MBR companies and plant operators on these subjects, and should direct some of their research programs to address these needs. Among the challenges underscored by the experts, membrane fouling is one of the most serious problems that has retarded faster commercialization of MBR technology. The causes, characteristics, mechanisms of fouling, and methods to prevent or reduce membrane fouling are discussed elaborately in Section 4.16.4.7; Section 4.16.5.5 sheds light on the issue of exchangeability of modules. The remainder of the current section will be devoted to the issues closely related to membrane fouling and performance, that is, mechanical pretreatment and membrane integrity:
•
Pretreatment. Pretreatment is one of the most critical factors for ensuring a stable and continuous MBR operation. Due to membrane sensitivity to the presence of foreign bodies, fine prescreening of the feed (and sometimes of the mixed liquors) must occur. The type of sieve installed is very important with regard to the total screening of hair and fibers. Recent studies (Frechen et al., 2006; Schier et al., 2009) have shown sieves with smaller gap sizes and with
•
two-dimensional gap geometries to perform better. On the other hand, even intensive long-term pilot plant trials can fail to suggest the effective scale-up design of the sieve (Melin et al., 2006). If too many clogging problems occur, the original pre-screen systems are usually upgraded to finer screens. However, when both the influent and the mixed liquor are filtered with a fine prescreen, a large amount of trash is produced (up to 3.8 m3 per week for a 1.4 MLD plant) (Le-Clech et al., 2005a; Melin et al., 2006; Schier et al., 2009). It should be noted that the investment in pretreatment is of little use if the bioreactor is uncovered, in which case, different sorts of debris can easily enter the bioreactor. It is recommended to remove these items using a high-pressure water hose. However, many MBR users report that this type of manual cleaning causes membrane-fiber breakage. In order to keep the membrane effectively separated from the fibrous materials, Schier et al. (2009)proposed the following mechanical-treatment concept: conventional pretreatment including screen and grit chamber/grease trap to be placed before the biological tank, causing braid of hair and fibers formed therein to be removed by the sieve placed before the separate filtration chamber housing the membrane modules. Membrane integrity. A major problem facing MBR systems is the loss of membrane integrity, which leads to the permeate-quality deterioration and ineffective backwashing. When breakage occurs in a submerged hollow-fiber MBR system, continuous filtration may allow solids and particles to quickly clog the broken fiber. However, application of backwash would force the solids out of the fiber. Accordingly, once damaged, disinfection of the product water would be compromised and it would also cause the loss of the backwash efficiency; and the faulty membrane/module would need to be changed quickly.
Faulty installation is one obvious reason for membrane failure. Once under pressure, an incorrectly installed membrane module can be compressed. Other reasons associated with regular operation include frequent and/or extended contact between membrane and cleaning solution causing delamination of the membrane, scoring and cleaving of the membrane resulting from the presence of abrasive or sharpedged materials in the influent, and operating stress and strain occurring in the system due to fiber movement and membrane backwashing. A better understanding of the effect of membrane material, age, and fouling on membrane integrity may be gained from hollow-fiber-tensile test reported in the literature (Childress et al., 2005; Gijsbertsen-Abrahamse et al., 2006). Even flat-sheet membranes used in MBRs are not immune to occasional failure (Cornel and Krause, 2003). The construction of current flat-sheet MBR membrane panels is a labor-intensive, multistep operation. These are typically sandwich constructions with three separate layers. Two of them are pre-fabricated membrane layers, while the third one is a permeate drainage layer which is sandwiched between them. The three layers of the sandwich are held together by gluing or laminating techniques over their entire surface or just at their edges. Flat-sheet membranes have been found to be sensitive to breaking near the top
Membrane Biological Reactors
due to poor adhesion of the membrane to the support layer (Doyen et al., 2010).
4.16.4.7 Membrane Fouling – the Achilles’ Heel of MBR Technology Although MBR has become a reliable alternative to CAS processes and an option of choice for many domestic and industrial applications, membrane fouling and its consequences in terms of plant maintenance and operating costs limit the widespread application of MBRs (Le-Clech et al., 2006). Membrane fouling can be defined as the undesirable deposition and accumulation of microorganisms, colloids, solutes, and cell debris within pores or on membrane surface (Meng et al., 2009). It results from the interaction between the membrane material and the components of the activated sludge liquor, which include biological flocs formed by a large range of living microorganisms along with soluble and colloidal compounds. Thus, it is not surprising that the fouling behavior in MBRs is more complicated than that in most membrane applications. The suspended biomass has no fixed composition and varies with both feedwater composition and MBR operating conditions employed. Accordingly, although many investigations of membrane fouling have been published, the diverse range of operating conditions and feedwater matrices employed, and the limited information reported in most studies on the biomass composition in suspension or on the membrane, have made it difficult to establish any generic behavior pertaining to membrane fouling in MBRs. Three fouling phenomena need to be recognized and duly addressed:
• • •
Cake formation. This results from the balance of forces (shear stress at the membrane wall and filtration force) and is evidently linked to the biomass characteristics. Blockage of bundle of fibers. The bundle of fibers act as a deep bed filter (depending on biomass characteristics and structure of the bundle). Biofilm formation. This is not strictly dependent upon biomass characteristics as, very often, the microorganisms involved in the biofilm formation are not the dominant species in the biomass.
4.16.4.7.1 Fouling development Zhang et al. (2006a) proposed a three-stage history for membrane fouling in MBRs:
• • •
Stage 1. An initial short-term rise in TMP due to conditioning. Stage 2. Long-term rise in TMP, either linear or weakly exponential. Stage 3. A sudden rise in TMP, with a sharp increase in dTMP/dt, also known as the TMP jump.
When operating at fluxes well below the apparent critical flux of the MLSS, a slow steady rise in TMP (stage 2) is observed which eventually changes to a rapid rise in TMP (stage 3). For sustainable operation, the aim would be to limit the extent of stage 1, prolong stage 2, and avoid stage 3, since it could be difficult to restore.
589
4.16.4.7.2 Types of membrane fouling Definitions based on ease of removal and a variety of confusing terminologies have been proposed in the literature to describe fouling. For example, based on the ease of removal, some authors prefer to use the term ‘irreversible fouling’ to the fouling that can be removed by chemical cleaning but not by physical cleaning. Recently, Meng et al. (2009) proposed a somewhat changed definition and used the terms ‘removable’ and ‘irremovable’ for the fouling which is easily eliminated and which requires chemical cleaning, respectively. This chapter, however, uses the more direct terms – physically removable fouling and chemically removable fouling. The formation of a cake layer which can be described as a porous media with a complex system of interconnected interparticle voids has been reported as the major contributor to membrane fouling in MBRs (Jeison and van Lier, 2007; Ramesh et al., 2007). Such fouling is usually physically removable. Recently, a large number of scientific investigations have been performed in order to gain a better understanding of cake-layer formation and cake-layer morphology employing techniques such as confocal laser-scanning microscopy (CLSM), multiphoton microscopy, etc. (Yang et al., 2007; Hughes et al., 2006, 2007). During initial filtration, colloids, solutes, and microbial cells pass through and deposit inside the membrane pores. However, during the long-term operation of MBRs, the deposited cells multiply and yield extracellular polymeric substance (EPS), which clog the pores and form a strongly attached fouling layer. Chemical cleaning is usually required to remove such fouling. Evaluation of physically removable and chemically removable fouling propensity of MBR mixed liquor has been the focus of many studies to date (Field et al., 1995; Ognier et al., 2004; Pollice et al., 2005; Bacchin et al., 2006; Guglielmi et al., 2007; Lebegue et al., 2008; Wang et al., 2008b). Some of the definitions are based on the fouling components. The fouling in MBRs can be classified into three major categories: biofouling, organic fouling, and inorganic fouling, although, in general, all of them take place simultaneously during membrane filtration of activated sludge. Biofouling refers to the deposition, growth, and metabolism of bacteria cells or flocs on the membranes. Biofouling may start with the deposition of individual cell or cell cluster on the membrane surface, after which the cells multiply and form a biocake (Liao et al., 2004; Pang et al., 2005; Wang et al., 2005; Ramesh et al., 2007). Techniques such as scanning electron microscopy (SEM), CLSM, atomic force microscopy (AFM), and direct observation through the membrane (DOTM) have been extensively used to derive valuable information regarding floc/cell-deposition process and the microstructure or architecture of the cake layer. Certain studies have also analyzed the microbial community structures and microbial colonization on the membranes in MBRs (Chen et al., 2004; Jin et al., 2006; Jinhua et al., 2006; Zhang et al., 2006b; Miura et al., 2007; Lee et al., 2009) employing molecular techniques. Such studies reported that the microbial communities on membrane surfaces were quite different from those in the suspended biomass and initially a specific phylogenetic group of bacteria may play the key role in development of the mature biofilm. However, a temporal change
590
Membrane Biological Reactors
of microbial-community structure can take place due to the development of anoxic conditions in the cake layer. Organic fouling in MBRs refers to the deposition of biopolymers on the membranes (Meng et al., 2009). Due to the small size, the soluble biopolymers can be deposited onto the membranes more readily, but they have lower back-transport velocity in comparison to large particles (e.g., colloids and sludge flocs). Powerful analytical tools such as Fourier transform infrared (FTIR) spectroscopy, solid-state 13C-nuclear magnetic resonance (NMR) spectroscopy, and high-performance size-exclusion chromatography (HP-SEC) are usually utilized for identification of the deposited biopolymers (Kimura et al., 2005; Rosenberger et al., 2006; Zhou et al., 2007; Teychene et al., 2008) and studies have confirmed that SMP or EPS is the origin of organic fouling in MBR. Inorganic elements such as Mg, Al, Fe, Ca, Si, etc. and metals can enhance the formation of biofouling and organic fouling and can together form a recalcitrant cake layer (Lyko et al., 2007; Wang et al., 2008b). Inorganic fouling can form in two ways – due to concentration-polarization-led chemical precipitation and entrapment within biopolymer gel layer (Meng et al., 2009). Chemical cleaning agents such as ethylenediaminetetraacetic acid (EDTA) might efficiently remove inorganics on the membrane surface (Al-Amoudi and Lovitt, 2007); however, the fouling caused by inorganic scaling may not be easy to eliminate even by chemical cleaning (You et al., 2006).
Figure 8 lists the membrane-fouling parameters, while Figure 9 illustrates the interrelations and combined effect of those parameters. Some of the membrane characteristics and the parameters that influence the performance of the MBRs are discussed in the following: 1. Physical parameters.
•
Pore size and distribution. Studies revealed that the pore size alone could not predict hydraulic performances. The effects of pore size (and distribution of pore size) on membrane fouling are strongly related to the feedsolution characteristics and in particular the particlesize distribution. The complex and changing nature of
Membrane fouling
4.16.4.7.3 Parameters influencing MBR fouling All the parameters involved in the design and operation of MBR processes have an influence on membrane fouling (Le-Clech et al., 2006; Meng et al., 2009). While some of these parameters have a direct influence on MBR fouling, many others result in subsequent effects on phenomena exacerbating fouling propensity. However, three main categories of factors can be identified – membrane and module characteristics, feed and biomass parameters, and operating conditions.
Membrane characteristics • Physical parameters -Pore size and distribution -Porosity/roughness -Membrane configuration • Chemical parameters -Hydrophobicity -Materials
Mixed liquor characteristics
Feed
Biomass
Figure 9 Interrelations and combined effect of the membrane fouling parameters.
Feed–biomass characteristics
Operating conditions
• Nature of feed and concentration • Biomass fractionation • Biomass (bulk) parameters -MLSS concentration -Viscosity -Temperature -Dissolved oxygen (DO) • Floc characteristics -Floc size -Hydrophobicity/surface charge
• Aeration, cross-flow velocity • Sludge retention time (SRT) • Unsteady state operation
• Extracellular polymeric substance (EPS) • Soluble microbial products (SMP) Figure 8 Membrane fouling parameters at a glance.
Operating conditions
Membrane characteristics
Membrane Biological Reactors
•
•
the biological suspension present in MBR systems and the large pore-size distribution of the membrane generally used in MBR systems are the main reasons for the undefined general dependency of the flux propensity on pore size (Chang et al., 2002a; Le-Clech et al., 2003b). It is generally expected that smaller-pore membranes would reject a wider range of materials, and the resulting cake layer would feature a higher resistance compared to large-pore membranes. However, this type of fouling is easily removed during the maintenance cleaning than fouling due to internal pore clogging obtained in larger-pore membrane systems. The chemically removable fouling, due to the deposition of organic and inorganic materials onto and into the membrane pores, is the main cause of the poor longterm performances of larger pore-size membranes (Chang et al., 2001; He et al., 2005). However, the opposite trend is sometimes reported (Gander et al., 2000). The duration of the experiment and other operating parameters such as cross-flow velocity and constant pressure or constant flux operation have a direct influence on the determination of the optimization of the membrane pore size and are responsible for contradictory reports in the literature. Porosity/roughness. Membrane roughness and porosity along with membrane microstructure, material, and pore-size distribution were suggested as potential reasons for the different fouling behaviors observed (Kang et al., 2006; Ho and Zydney, 2006). For instance, a track-etched membrane, with its dense structure and small but uniform cylindrical pores, featured the lowest resistance due to pore fouling in contrast to the other membranes having interwoven sponge-like highly porous network (Fang and Shi, 2005). Other studies have pointed out the importance of pore-aspect ratio (mean major-axis length/mean minor-axis length) (Kim et al., 2004) or roughness (He et al., 2005) on fouling in an MBR. Membrane configuration. In submerged MBR processes, the membrane can be configured as vertical flat plates, vertical or horizontal hollow fine fibers (filtration from out to in) or, more rarely as tubes (filtration from in to out). Each of hollow-fiber and flat-sheet membrane types has specific footprint and air scouring and chemical cleaning requirement, which may favor one process over another for a given application (Judd, 2002; Hai et al., 2005). Nevertheless, hollow-fiber modules are generally more economical to manufacture, provide high specific membrane area, and can tolerate vigorous backwashing (Stephenson et al., 2000). For low-flux operation, hollow fibers are attractive due to their high packing density. A higher fiber-packing density would increase productivity; however, increasing the packing density may lead to severe interstitial blockage due to the impeded propagation of air bubbles toward the core, limiting their effect on fouling limitation (Kiat et al., 1992; Yeo and Fane, 2005; Sridang et al., 2005). However, Hai et al. (2008a) developed a spacer-filled module in order to utilize high packing density without encountering
591
severe fouling. Studies have also revealed the effects of other membrane characteristics including hollowfiber orientation, size, and flexibility ( Cui et al., 2003; Ognier et al., 2004; Chang and Fane, 2002; Lipnizki and Field, 2001; Zheng et al., 2003; Zhongwei et al., 2003). 2. Chemical parameters.
•
•
Hydrophobicity. The influence of the membrane hydrophobicity on the early stage of the fouling formation may be significant; however, this parameter is expected to play only a minor role during extended filtration periods in MBRs (Le-Clech et al., 2006). Once initially fouled, the membrane’s chemical characteristics would become secondary to those of the sludge materials covering the membrane surface. Nevertheless, because of the hydrophobic interactions occurring between solutes, microbial cells and membrane material, membrane fouling is expected to be more severe with hydrophobic rather than hydrophilic membranes (Madaeni et al., 1999; Chang et al., 1999; Yu et al., 2005a), although different results have also been reported (Fang and Shi, 2005). In many reported studies, change in membrane hydrophobicity often occurs with other membrane modifications such as pore size and morphology, which make the correlation between membrane hydrophobicity and fouling more difficult to assess. Materials. The large majority of the membranes used in MBRs are polymeric based. A direct comparison between polyethylene (PE) and polyvinylidene fluoride (PVDF) membranes clearly indicated that the latter leads to a better prevention of physically irremovable fouling and that PE membrane fouled more quickly (Yamato et al., 2006). Zhang et al. (2008b) studied the affinity between EPS and the three polymeric UF membranes, and observed that the affinity capability of the three membranes was of the order polyacrylonitrile (PAN)oPVDFopolyethersulfone (PES). Although featuring superior chemical, thermal, and hydraulic resistances, ceramic (Fan et al., 1996; Scott et al., 1998; Luonsi et al., 2002; Xu et al., 2003; Judd et al., 2004) and stainless steel (Zhang et al., 2005) membrane modules are not the preferred option for MBR applications due to their high cost (around an order of magnitude more expensive than the polymeric materials).
3. Feed–biomass characteristics.
•
Nature of feed and concentration. Fouling in the MBR is mostly affected by the interactions between the membrane and the biological suspension rather than wastewater itself (Choi et al., 2005). Nevertheless, the fouling propensity of the wastewater has to be indirectly taken into consideration during the characterization of the biomass, as the wastewater nature can significantly influence the physicochemical changes in the biological suspensions (Le-Clech, 2003b; Jefferson et al., 2004), which in turn may aggravate fouling.
592
•
Membrane Biological Reactors
Biomass fractionation. The many studies (Bae and Tak, 2005; Li et al., 2005a; Itonga et al., 2004; Lee et al., (2003); Lee et al., 2001a; Wisniewski and Grasmick, 1998; Bouhabila et al., 2001) that are available on the contribution of different fractions of the biomass to fouling usually report contradictory results. Although the relatively low fouling role played by the suspended solids (biofloc and the attached EPS) compared to those of the soluble and colloids (generally defined as soluble microbial products or SMP) is usually reported, the reported relative contribution of the SMP to overall membrane fouling ranges from 17% (Bae and Tak, 2005) to 81% (Itonga et al., 2004). These wide discrepancies may be explained by the different operating conditions and biological states of the suspension used in the reported studies (Figure 10). Although an interesting approach for studying MBR fouling, the fractionation experiments neglect any coupling or synergistic effects which may occur among the different components of the biomass.
•
•
4. Biomass (bulk) parameters.
•
MLSS concentration. Although the increase in MLSS concentration has often been reported to have a mostly negative impact on the MBR hydraulic performances (Cicek et al., 1999b; Chang and Kim, 2005), controversies exist (Defrance and Jaffrin, 1999; Hong et al., 2002; Le-Clech et al., 2003b; Lesjean et al., 2005; Brookes et al., 2006). The existence of threshold values above (Lubbecke et al., 1995) or below (Rosenberger et al., 2005) which the MLSS concentration has a negative influence was also reported. Figure 11 depicts the influence of shift in MLSS concentration on flux as reported in different studies. Nowadays, information on additional biomass characteristics (e.g., composition and concentration of EPS) is deemed necessary to furnish a comprehensive picture. On the other hand,
100
Variable: Membrane type
•
Hai et al. (2006a)showed that the extent of fouling was independent of MLSS concentration itself, and was rather more influenced by the efficiency of the foulingprevention strategies adopted. Viscosity. The importance of MLSS viscosity is that it modifies bubble size and can dampen the movement of hollow fibers in submerged bundles (Wicaksana et al., 2006). The net result of this phenomenon would be a greater rate of fouling. Increased viscosity also reduces the efficiency of mass transfer of oxygen and can therefore effect dissolved oxygen (DO) (Germain and Stephenson, 2005); fouling, as discussed later, tends to be worse at low DO. Critical MLSS concentrations have been reported in the literature (Itonga et al., 2004) above which, suspension viscosity tends to increase exponentially with the solid concentration. Temperature. Experiments conducted under moderate temperature usually report greater deposition of materials on the membrane surface at lower temperatures. Temperature may impact membrane filtration by increasing fluid viscosity, causing defloculation of biomass and higher EPS secretion, reducing biodegradation rate, etc. (Jiang et al., 2005; Rosenberger et al., 2006). Dissolved oxygen. The effects of DO on MBR fouling are multiple and may include changes in biofilm structure, SMP levels, and floc-size distribution (Lee et al., 2005). The average level of DO in the bioreactor is controlled by the aeration rate, which not only provides oxygen to the biomass but also tends to limit fouling formation on the membrane surface. Optimum aeration would result in lower specific cake resistance of the fouling layer featuring larger particle sizes and greater porosity (Kang et al., 2003; Kim et al., 2006). Therefore, in general, higher DO tends to lead to better filterability, and lower fouling rate.
Variable: Sludge type
Variable: SRT
80 60 40 20 0 (a)
(b)
Soluble
Colloids
Suspended solids
(c)
Colloid + soluble
Figure 10 Influence of different parameters (membrane type, sludge type, and SRT) on the relative contributions (in %) of the different biomass fractions to MBR fouling. Data from (a) Bae TH and Tak TM (2005) Interpretation of fouling characteristics of ultrafiltration membranes during the filtration of membrane bioreactor mixed liquor. Journal of Membrane Science 264: 151–160; (b) Meng F and Yang F (2007) Fouling mechanisms of deflocculated sludge, normal sludge, and bulking sludge in membrane bioreactor. Journal of Membrane Science 305: 48–56; and (c) Lee W, Kang S, and Shin H (2003) Sludge characteristics and their contribution to microfiltration in submerged membrane bioreactors. Journal of Membrane Science 216: 217–227.
Membrane Biological Reactors
Limiting or critical or stabilized flux, (l m−2 h−1)
120
irregular floc shape, and higher hydrophobicity (Meng et al., 2006).
(1)
100 80
(2)
(7)
60
(3) (4)
40 (5) 20
(6)
0 0
5
10
15
20
25
MLSS concentration, gl−1 Figure 11 Influence of shift in MLSS concentration on flux (fouling) as reported in different studies. Data from (1) Cicek N, Franco JP, Suidan MT, and Urbain V (1998) Using a membrane bioreactor to reclaim wastewater. Journal of American Water Works Association 90: 105–113; (2) Beaubien A, Baty M, Jeannot F, Francoeur E, and Manem J (1996) Design and operation of anaerobic membrane bioreactors: Development of a filtration testing strategy. Journal of Membrane Science 109: 173–184; (3) Madaeni SS, Fane AG, and Wiley D (1999) Factors influencing critical flux in membrane filtration of activated sludge. Journal of Chemical Technology and Biotechnology 74: 539–543; (4) Han SS, Bae TH, Jang GG, and Tak TM (2005) Influence of sludge retention time on membrane fouling and bioactivities in membrane bioreactor system. Process Biochemistry 40: 2393–2400; (5) Bouhabila EH, Ben Aim R, and Buisson H (1998) Microfiltration of activated sludge using submerged membrane with air bubbling (application to wastewater treatment). Desalination 118: 315–322; (6) Bin C, Xiaochang W, and Enrang W (2004) Effects of TMP, MLSS concentration and intermittent membrane permeation on a hybrid submerged MBR fouling. In: Proceedings of the IWA – Water Environment – Membrane Technology (WEMT) Conference. Seoul, Korea, 7–10 June; and (7) Defrance L and Jaffrin MY (1999) Reversibility of fouling formed in activated sludge filtration. Journal of Membrane Science 157: 73–84.
5. Floc characteristics.
•
•
593
Floc size. The floc-size distribution obtained with the MBR sludge is lower than the results generally obtained from CASP (Zhang et al., 1997; Wisniewski and Grasmick, 1998; Lee et al., 2003; Cabassud et al., 2004; Bae and Tak, 2005). Unlike in the CAS systems, the effective separation of suspended biomass from the treated water is not critically dependent on aggregation of the microorganisms, and the formation of large floc. However, independent of their size, biological floc play a major role in the secretion of EPS and formation of the fouling cake on the membrane surface. Hydrophobicity/surface charge. The direct effect of floc hydrophobicity on MBR fouling is difficult to assess. Conceptually, hydrophobic flocs would lead to high flocculation propensity, less secretion of EPS, and low interaction with the hydrophilic membrane (Jang et al., 2006). However, reports of highly hydrophobic flocs fouling MBR membranes can be found in the literature. For instance, the excess growth of filamentous bacteria, known to be responsible for severe MBR fouling, also resulted in higher EPS levels, lower zeta potential, more
6. Extracellular polymeric substances. The term EPS is used as a general and comprehensive concept for different classes of macromolecules such as polysaccharides, proteins, nucleic acids, (phosphor-)lipids, and other polymeric compounds which have been found at, or outside, the cell surface and in the intercellular space of microbial aggregates (Flemming and Wingender, 2001). EPS are the construction materials for microbial aggregates such as biofilms, flocs, and activated sludge liquors. The functions of EPS matrix are multiple and include aggregation of bacterial cells in flocs and biofilms, formation of a protective barrier around the bacteria, retention of water, and adhesion to surfaces (Laspidou and Rittmann, 2002). With its heterogeneous and changing nature, EPS can form a highly hydrated gel matrix in which microbial cells are embedded (Nielson and Jahn, 1999). Therefore, they can be responsible for the creation of a significant barrier to permeate flow in the membrane processes. Contemporary literature is replete with reports identifying EPS as a major fouling parameter (Chang and Lee, 1998; Cho and Fane, 2002; Nagaoka et al., 1996, 1998; Rosenberger and Kraume, 2002). On the other hand, since the EPS matrix plays a major role in the hydrophobic interactions among microbial cells and thus in the floc formation (Liu and Fang, 2003), it was proposed that a decrease in EPS levels may cause floc deterioration and may be detrimental for the MBR performances. This indicates the existence of an optimum EPS level for which floc structure is maintained without featuring high fouling propensity. Many parameters including gas sparging, substrate composition (Fawehinmi et al., 2004), and loading rate (Cha et al., 2004; Ng et al., 2005) affect EPS characteristics in the MBR, but SRT probably remains the most significant of them (Hernandez et al., 2005). A functional relationship between specific resistance, mixed liquor volatile suspended solids (MLVSS), TMP, and permeate viscosity, and EPS is believed to exist (Cho et al., 2005). 7. Soluble microbial products. SMPs are defined as soluble cellular components that are released during cell lysis, diffuse through the cell membrane, and are lost during synthesis or are excreted for some purpose (Laspidou and Rittmann, 2002; Li et al., 2005a). During filtration, SMPs adsorb on the membrane surface, block membrane pores, and/or form a gel structure on the membrane surface where they provide a possible nutrient source for biofilm formation and a hydraulic resistance to permeate flow (Rosenberger et al., 2005). Since direct relationships between the carbohydrate level in SMP (SMPc) solution with fouling rate (Lesjean et al., 2005), filtration index and capillary suction time (CST) (Greiler et al., 2005; Evenblij et al., 2005b; Tarnacki et al., 2005), critical flux tests (Le-Clech et al., 2005b), and specific flux (Rosenberger et al., 2005) have been clearly described, it reveals SMPc to be the major foulant indicator in MBR systems. However, controversy over the relative contribution of carbohydrate and protein portions of SMP to fouling exists (Evenblij and Van der Graaf, 2004; Drews et al., 2005a; Drews et al., 2006).
594
Membrane Biological Reactors
The operating conditions of MBrs are discussed as follows:
•
•
•
Aeration, cross-flow velocity. Since the origin of the SMBR, bubbling has been defined as the strategy of choice to induce flow circulation and shear stress on the membrane surface. Aeration used in MBR systems has three major roles: providing oxygen to the biomass, maintaining the activated sludge in suspension, and mitigating fouling by constant scouring of the membrane surface (Dufresne et al., 1997). However, an optimum aeration rate, beyond which a further increase has no significant effect on fouling suppression, has been observed on many occasions (Ueda et al., 1997; Le-Clech et al., 2003a, 2003b; Liu et al., 2003; Psoch and Schiewer, 2005b). It is also important to note that too intense an aeration rate may damage the floc structure reducing their size, and release EPS into the bioreactor (Park et al., 2005; Ji and Zhou, 2006), and thereby aggravate fouling. Solid retention time. SRT (and thereby the F/M ratio), which greatly controls biomass characteristics, is regarded as the most important operating parameter influencing fouling propensity in MBRs. Considering the advantages of this process over the conventional activated sludge process (CASP), the early MBRs were typically run at very long SRTs to minimize excess sludge (Liu et al., 2005; Gao et al., 2004; Nuengjamnong et al., 2005). But unlike in bench-scale studies employing simpler synthetic feed, the progressive accumulation of nonbiodegradable materials (such as hair and lint) in an MBR fed with real sewage definitely leads to clogging of the membrane module (Le-Clech et al., 2005b). Operating an MBR at higher SRT leads inevitably to increase of MLSS concentration (Zhang et al., 2006c). The increase in aeration intensity to retain high MLSS levels in suspension and maintain proper oxygenation may not be a sustainable option for the treatment process. In this scenario, the increased shear provided to control fouling could cause biofloc deterioration as well as cell lysis and enhanced EPS secretion, and lead to fatal fouling. On the other hand, at infinite SRT, most of the substrate is consumed to ensure the maintenance needs and the synthesis of storage products. The very low apparent net biomass generation observed can explain the low fouling propensity observed for high SRT operation in certain studies (Orantes et al., 2004). It is likely that there is an optimal SRT, between the high fouling tendency of very low SRT operation and the high viscosity suspension prevalent for very long SRT. Unsteady state operation. In practical applications, unsteady state conditions such as variations in operating conditions (flow input/HRT and organic load) and shifts in oxygen supply could occur regularly (Drews et al., 2005a). The start-up phase can also be considered as unsteady operation and data collected before biomass stabilization (including the period necessary to reach acclimatization) may become relevant in the design of MBRs (Cho et al., 2005). Such unsteady state conditions have also been defined as additional factors leading to changes in MBR fouling propensity. For instance, the addition of a spike of acetate in the feedwater significantly decreased the filterability of the biomass in an MBR due to the rise in SMP levels resulting from the feed spike (Evenblij et al., 2005a).
4.16.4.7.4 Fouling mitigation The complex interactions between the fouling parameters complicate the perception of MBR fouling and it is therefore crucial to have a complete understanding of the biological, chemical, and physical phenomena occurring in MBRs to assess fouling propensity and mechanisms and thereby formulate mitigation strategies. As membrane fouling increases with increasing flux in all membrane separation processes, the operating flux should be lower than the critical flux. When the operating flux is below the critical flux, particle accumulation in the region of membranes can be effectively prevented. However, due to physicochemical solute–membrane material interactions, the membrane permeability decreases over time, even when MBRs are operated in subcritical (below critical flux) conditions. Other preventative methods need to be considered to maintain stable operation of MBR systems (Figure 12). Fouling can be removed by various methods and they are as discussed herein: 1. Physical cleaning. The following methods are usually used in combination to remove membrane fouling:
•
Permeate backwashing. Membrane backwashing or backflushing refers to pumping permeate in the reverse direction through the membrane. Backwashing has been found to successfully remove most of the reversible fouling due to pore blocking, transport it back into the bioreactor, and partially dislodge loosely attached sludge cake from the membrane surface (Bouhabila et al., 2001; Psoch and Schiewer, 2005a; Psoch and Schiewer, 2006). Frequency, duration, the ratio between those two parameters, and its intensity are the key parameters in the design of backwashing and different combinations of these parameters have proved to be more efficient in different studies (Jiang et al., 2005; Schoeberl et al., 2005). Between 5% and 30% of the produced permeate is used for backwashing. This also
Removal of fouling
Limitation of fouling
• Physical cleaning --Backwashing --Air backwashing --Intermittent operation --Sonification and other energy-intensive processes
• Optimization of membrane characteristics
• Chemical cleaning --Maintenance cleaning --Intensive cleaning
• Optimization of operating conditions --Aeration --Other operating conditions --Membrane module design • Modification of biomass characteristics -Aerobic granular sludge -Coagulant/flocculent -Adsorbent/flux enhancers
Figure 12 Reported membrane fouling mitigation strategies at a glance.
Membrane Biological Reactors
•
•
•
affects operating costs as, obviously, energy is required to achieve a pressure suitable for flow reversion. Certain studies are, therefore, devoted to optimization of backwashing (Smith et al., 2005). Air backwashing. Air, instead of permeate, can also be used as the backflushing medium (Visvanathan et al., 1997; Sun et al., 2004). The invention of air backwashing techniques for membrane declogging led to the development of using the membrane itself as both clarifier and air diffuser. In this approach, two sets of membrane modules are submerged in the aeration tank. While the permeate is extracted through one of the sets, the other is supplied with compressed air for backwashing. The cycle is repeated alternatively, and there is a continuous airflow into the aeration tank, which is sufficient to aerate the mixed liquor. However, air backwashing may also present potential issues of membrane breakage and rewetting (Le-Clech et al., 2006). Intermittent operation. Intermittent operation or membrane relaxation can significantly improve membrane productivity (Yamamoto et al., 1989). During relaxation, back transport of foulants is naturally enhanced as loosely attached foulants can diffuse away from the membrane surface (Ng et al., 2005). Although some studies found it more important than backwashing for fouling removal (Schoeberl et al., 2005), recent studies tend to combine intermittent operation with frequent backwashing for optimum results (Zhang et al., 2005; Vallero et al., 2005). The economic feasibility of intermittent operation for large-scale MBRs has been the focus of certain studies (Hong et al., 2002); however, it seems rather an established operation mode nowadays. Sonification and other energy-intensive processes. Although sonification would be difficult to apply at a large scale due to the focused nature of the sonic energy, laboratory-scale studies have explored sonification for breaking down cake layers in MBRs, especially in case of ceramic membranes. Certain studies have confirmed the efficiency of application of sonification alone or in combination with backwashing for removing the cake layer (Lim and Bai, 2003; Fang and Shi, 2005). However, other studies report that fouling may even worsen due to pore blocking (Hai et al., 2006a). Attempts have also been made to control fouling or modify sludge by using ozone and electric field (Chen et al., 2007; Huang and Wu, 2008; Sui et al., 2008; Wen et al., 2008).
•
Maintenance cleaning with moderate chemical concentration (weekly) is applied to maintain design permeability and it helps to reduce the frequency of intense cleaning. This may be replaced by a more frequent
(e.g., on a daily basis) chemically enhanced backwash utilizing mild chemical concentration. Intensive (or recovery) chemical cleaning (once or twice a year) is generally carried out when further filtration is no longer sustainable because of an elevated TMP.
The MBR suppliers propose their own chemical cleaning recipes, which differ mainly in terms of concentration and methods, and often site-specific protocols are followed (Kox, 2004; Tao et al., 2005; Le-Clech et al., 2005b). Mainly, sodium hypochlorite (for organic foulants) and citric acid (for inorganics) are used as chemical agents. Some pitfalls of chemical cleaning are worth noting. The detrimental effect of cleaning chemicals on biological performance has been reported (Lim et al., 2005; Hai et al., 2007). It has also been mentioned that the level of pollutants (measured as TOC) in the permeate rises just after the chemical cleaning step (Tao et al., 2005). This raises concern especially in case of MBRs used in the reclamation process trains (i.e., e.g., upstream of RO) (Le-Clech et al., 2006). Chemical cleaning may also shorten the membrane lifetime and disposal of spent chemical agents causes environmental problems (Yamamura et al., 2007). The measures to limit fouling are discussed next. Recently, there have been a significant number of studies which focused on the ways to limit fouling. The proposed strategies include (1) improving the antifouling properties of the membrane, (2) operating the MBR under specific nonor-little-fouling conditions, and/or (3) pretreating the biomass suspension to limit its fouling propensity. They are discussed as follows: 1. Membrane modification.
•
2. Chemical cleaning. The effectiveness of physical cleaning tends to decrease with operation time as more recalcitrant fouling accumulates on the membrane surface. Therefore, in addition to physical cleaning, different types/intensities of chemical cleaning are applied in practice. A combination of the following types of cleaning is usually applied (Le-Clech et al., 2006):
•
595
•
Optimization of membrane characteristics. Many studies have shown that chemical modifications of the membrane surface can efficiently improve antifouling properties. Recent examples comprise (1) increasing membrane hydrophilicity by NH3 and CO2 plasma treatments (Yu et al., 2005a, 2005b) and ultraviolet (UV) irradiation (Yu et al., 2007), (2) TiO2 entrapped membrane (Bae and Tak, 2005), and (3) applying precoating of TiO2 (Bae et al., 2006), GAC (Hai, 2007), ferric hydroxide (Zhang et al., 2004), polyvinylidene fluoride-graft-polyoxyethylene methacrylated (PVDF-gPOEM) (Asatekin et al., 2006), polyvinyl alcohol (PVA) (Zhang et al., 2008a), etc. Improved performance in case of precoated membrane has been attributed to the adsorption of soluble organics on the precoat, limiting the direct contact between the organics and the membrane. Self-forming dynamic membrane-coupled bioreactors, utilizing coarse pore-sized substrates and allowing cake and gel layers to deposit on the surface, have been reported to obtain high flux and good removal in certain studies, although stable performance cannot be expected with such a filtration barrier (Wu et al., 2004). Membrane module design. The membrane module design by optimizing the packing density of hollow fibers or flat sheets, the location of aerators, the orientation of
596
Membrane Biological Reactors
fibers, and diameters of fibers (Chang and Fane, 2001; Chang et al., 2002b; Fane et al., 2002) remains another important parameter in the optimization of the MBR operation. In a specially designed module in which air bubbles were confined in close proximity to the hollow fiber (rather than diffusing in the reactor), higher permeability was obtained (Ghosh, 2006). Two major design approaches are adopted in case of the commercially available hollow-fiber bundles. One of these approaches relies on partitioning of bundles of fibers, which are fixed at both ends, to secure flow path of air bubbles introduced from the center of the bundle at the base, thereby leading sludge out of the module. In another approach, bundle of one-end free fibers are allowed to float freely under the scouring action of air bubbles introduced from the core of the bundle to avoid accumulation of sludge. In order to utilize high packing density without encountering severe fouling, a new approach to hollow-fiber module design was explored by Hai et al. (2008a). Spacer was introduced within usual hollow-fiber bundles with the aim of minimizing the intrusion of sludge into the module. The little amount of intruded sludge was then backwashed through the bottom end, while the sludge deposited on the surface was effectively cleaned by air scouring. In this way, efficient utilization of cleaning solution and air for backwashing and surface cleaning, respectively, were possible. Recent approaches such as novel fiber sheet (FiSh) membrane (Heijnen et al., 2009), multimodule flat-sheet concept (Kreckel et al., 2009), and vacuum rotation membrane (Alnaizy and Sarin, 2009; Komesli et al., 2007) are also noticeable.
•
3. Modification of biomass characteristics.
•
•
2. Optimization of operating conditions.
•
Aeration. As mentioned earlier, bubbling is an established strategy to induce flow circulation and shear stress on the membrane surface. The aeration intensity (air/permeate ratio, m3/m3) applied by MBR suppliers may vary between 24 and 50, depending on the membrane configuration (flat sheet vs. hollow fiber) and the MBR tank design (whether the membrane and aerobic zone combined into a single tank or not) (Tao et al., 2005; Le-Clech et al., 2006). However, recent large-scale studies revealed these original ratios to be quite conservative (Tao et al., 2005). The specific design of bubble size, airflow rate and patterns, and location of aerators have been defined as crucial parameters in fouling mitigation. As the energy involved in providing aeration to the membrane remains a significant cost factor in MBR design, efforts have been focused on optimization of aeration both from the points of view of fouling mitigation and reducing energy requirement. Recent developments in aeration design include cyclic aeration systems (Rabie et al., 2003), intermittent aeration (Yeom et al., 1999; Nagaoka and Nemoto, 2005), air pulsing (Judd et al., 2006), air sparging (Ghosh, 2006), improved aerator systems (Miyashita et al., 2000; Cote, 2002; Hai et al., 2008), etc.
Other operating conditions. The overall performance of the MBR is closely related to the choice of SRT value. Further optimizations of operating conditions through reactor design have been studied and include the addition of a spiral flocculator (Guo et al., 2004), vibrating membranes (Genkin et al., 2005), helical baffles (Ghaffour et al., 2004), suction mode (Kim et al., 2004) and high-performance compact reactor (Yeon et al., 2005), novel types of air lift (Chang and Judd, 2002), porous and flexible suspended membrane carriers (Yang et al., 2006), and the sequencing batch MBR (Zhang et al., 2006d). A reasonable flux rate without significant fouling is ideally expected. The concept of sustainable flux in MBRs was introduced from this point of view (Ng et al., 2005).
•
Aerobic granular sludge. In order to obtain higher biological aggregates in the bioreactor, aerobic granular sludge has also been used in MBR systems (Li et al., 2005b). With an average size around 1 mm, granular sludge increased the membrane permeability by 50%, but lower cleaning recoveries were observed (88% of those obtained with a conventional MBR). Such granular sludge may also not be stable under long-term operation (Hai, 2007). Coagulant/flocculant. Due to back transport and shearinduced fouling control mechanisms, large microbial flocs are expected to have a lower impact on membrane fouling. Based on this expectation, studies have explored addition of coagulants such as alum (Holbrook et al., 2004), ferric chloride, zeolite (Lee et al., 2001b), chitosan (Ji et al., 2008), etc. and have shown permeability enhancement. Pretreatment of the effluent is also possible and studies based on the pre-coagulation/ sedimentation of effluent before its introduction in the bioreactor revealed the fouling limitation offered by this technique (Itonga and Watanabe, 2004; Le-Clech et al., 2006). Adsorbent/flux enhancers. Lower fouling propensity is observed in MBR processes when biomass is mixed with adsorbents in that addition of adsorbents into biological treatment systems decreases the level of pollutants, and more particularly organic compounds (Kim and Lee, 2003; Lesage et al., 2005; Li et al., 2005c; Ng et al., 2006). In view of saturation of PAC during longterm studies, researchers have suggested periodic addition of PAC (Ng et al., 2005; Fang et al., 2006). Certain studies have proposed pre-flocculation and PAC addition (Guo et al., 2004; Cao et al., 2005).
A cationic polymer-based membrane performance enhancer (MPE 50) has been commercialized by Nalco recently. The interaction between the polymer and the soluble organics was reported as the main mechanism responsible for performance enhancement (Yoon et al., 2005). The potential impacts of coagulants or adsorbents on biomass community or biomass metabolism need to be taken into account (Iversen et al., 2009), and the discharge of some chemicals that are used as coagulants or adsorbents might be a potential environmental
Membrane Biological Reactors
risk. Such flux enhancers are probably best suited for solving occasional upsets rather than their continuous addition. Emerging fouling monitoring/control techniques such as interference of microbial intercellular communication by enzymatic degradation of signal molecules (Kjelleberg et al., 2008; Yeon et al., 2009), proteins and polysaccharides sensor for online fouling control (Mehrez et al., 2007), application of two-dimensional fluorescence for monitoring MBR performance (Galinha et al., 2009), etc., are worth noting.
4.16.5 Worldwide Commercial Application 4.16.5.1 Installations Worldwide The MBR process is an emerging advanced wastewater-treatment technology that has been successfully applied at an everincreasing number of locations around the world. MBRs were first developed 40 years ago and have been used commercially in Japan for almost 30 years. Since 1990, MBR technology has been adopted in North America and Europe, and it is now experiencing rapid growth in a wide variety of applications. In Asia, the drive in Japan was followed by an enthusiastic uptake in South Korea in the 1990s, and more recently by China. The highest growth rates are found in areas of greatest water stress for reuse applications, such as the southwestern US, China, Singapore, and Australia. The low footprint of the MBR is a significant driver for developed economies.
4.16.5.1.1 Location-specific drivers for MBR applications Howell (2004) stipulated the location-specific global drivers for MBR technology as follows: 1. Asia. MBR technology is being considered at many locations all over Asia, the main driver being water reclamation. Examples of settings vary from small-scale applications in Japan, where MBR product water is reused as toilet-flushing water in apartment blocks, medium-sized industrial applications in various countries, and large-scale municipal WWTPs in China. 2. Middle East. Clean-water shortages are the obvious driver for MBR applications in the Middle East, in treatment of both municipal as well as industrial (petrochemical) wastewater. 3. Europe. In Western Europe, water reclamation is not the main driver. In the UK, an important driver is compactness and strict discharge limits due to bathing wastewater requirements. In Germany and the Netherlands, important push factors are strict discharge requirements due to ecologically sensitive surface waters and the innovative character of the technological developments related to MBR. In Southern Europe, water reclamation can be considered as the main driver. 4. Northern America. In the US and Canada, MBR initiatives are predominantly driven by strict discharge requirements due to ecologically sensitive surface waters. At some locations, water reclamation is another important driver. In the US, where wastewater-treatment infrastructure lags behind population growth, MBRs are being increasingly implemented to make up the shortfall. Where there is
597
limited space to locate treatment plants, MBRs offer the potential to meet the needs of communities. 5. Australia. Stringent effluent-quality targets and water-reuse potential are obvious drivers for drought-stricken Australia.
4.16.5.1.2 Plant size Earlier MBR technology was favored in difficult applications or those applications where compactness was important and reuse was the target; and it usually involved smaller plants. As the demand for MBR technology grows globally, both the number of installations and the capacity of the installed plants are increasing dramatically. The most optimistic industry estimates suggest that up to 1000 new MBR plants will be built annually during the survey period. The size of the constructed plants has grown from facilities treating hundreds to thousands of gallons of wastewater per day to those treating tens of millions of gallons per day in just a few years. However, the most common capacity for current worldwide MBR installations ranges from the 50 000 gpd (200 m3 d1) to 500 000 gpd systems. The largest MBR plant in the world is set to be operational in 2010/11 in King County, Washington State. When completed, the facility will have an initial peak flow capacity of 495 000 m3 d1 (average 136 000 m3 d1), rising to a daily 645 000 m3 (average 205 000 m3) by 2040.
4.16.5.1.3 Development trend and the current status in different regions Figure 13 shows the regional share of total MBR plants as of 2003. Next, we discusss the trend of MBR growth in the three continents, Asia, Europe, and North America. 1. Asia. In the 1970s sidestream technology first entered the Japanese market. By 1993, 39 of such facilities had been reported for use in sanitary and industrial applications (Aya, 1994). The application of MBR in Japan concerned Europe 11%
N. America 16%
Asia 73% Figure 13 Regional share of total MBR plants (2003). Data from Pearce G (2008a) Introduction to membranes: An introduction to membrane.
598
Membrane Biological Reactors
small-scale installations for domestic wastewater treatment and reuse and some industrial applications, mainly in the food and beverage industries where highly concentrated flows are common. The domestic application often consists of so-called Johkaso or septic-tank treatment and inbuilding (office or domestic) wastewater-collection systems. In the early 1990s, the Japanese Government launched an ambitious 6-year research and development (R&D) project which led to a major technological and industrial breakthrough of the MBR process: the conception of submerged membrane modules, working with low negative pressure (out-to-in permeate suction), and membrane aeration to reduce fouling. This paved the way toward a significant reduction of capital and operation costs, due to the reduction and simplification of equipment and the abandonment of the energy-demanding sludge-recirculation loop. Since then, commercial MBRs proliferated in Japan, which had 66% of the world’s processes in 2000 (Stephenson et al., 2000). In Japan, although MBRs have long been used for industrial wastewater treatment or for reuse of wastewater in large buildings and so on, the introduction of municipal MBRs has lagged behind compared with other water-related fields. The first MBR for municipal wastewater treatment with an installed capacity of 2100 m3 d1 (total design capacity 12 500 m3 d1) in Japan started operation in March 2005, and this accelerated the introduction of MBRs in Japanese sewerage systems. Nine MBR plants, mostly small scale, for municipal wastewater treatment, are in operation at present (Table 8). In addition, there are several MBR plants currently in the design or planning stage. The number of MBRs for municipal wastewater is expected to increase in the near future and the technology will also play an important role in retrofitting and upgrading of existing treatment plants. The MBR technology saw an enthusiastic uptake in South Korea in the 1990s following its introduction in Japan. By 2005, the number of MBR plants rose up to more than 1300 (Namkung, 2008). The plants are mostly small, with more than 60% of the total plants having a capacity of less than 50 m3 d1. The plants were predominantly built on the submerged membrane technology (hollow fiber, 79%; flat sheet, 12%), while a meager 9% facilities utilized the tubular membranes in sidestream format. China has recently emerged as a strong MBR market. Hence, it would be interesting to cast light on the specific
Table 8
mode of development in that country. While the first paper on MBR was published in 1991, the emergence of a number of local and overseas companies developing MBR market in China accelerated with the funding of R&D projects by the Ministry of Science and Technology (MOST) in 1996 (Wang et al., 2008a). Since then, much progress has been achieved both in research and in practical applications of MBR in China. This is evident by the recent yearly publication rate of 35–40 English articles on MBR in China and the construction of a total of 254 plants for municipal (137) and industrial (117) wastewater treatment by 2008. The Chinese MBR market has the presence of a total of 33 companies or institutes, including famous overseas companies such as GE–Zenon Environmental Inc., Mitsubishi–Rayon (Japan), Toray (Japan), NOVO Environmental Technology (Singapore), and XFlow (Netherlands). Among these, only three companies provide flat-sheet MBR, and, interestingly, the worldwide renowned flat-sheet membrane provider, Kubota (Japan), was not found to be very active in the Chinese membrane market. Most of the plants in operation are medium scale or small scale in terms of treatment capacity, the number of plants with treatment capacity below 1000 m3 d1 totaling 225. The largest MBR plant with a capacity of 80 000 m3 d1 for municipal wastewater treatment and reuse is located in Beijing. Several other large MBR plants are also in the planning stage. Wang et al. (2008a) contend that the increasingly stringent discharge standards and the great need of water reclamation and reuse will further push forward the application of everlarger municipal MBR plants in China, especially in North China which has severe water shortage. 2. Europe. A market survey of the European MBR industry was performed by Lesjean and Huisjes (2008). They identified MBR plants constructed up to 2005, and about 300 references of industrial applications (420 m3 d1) and about 100 municipal WWTPs 4500 p.e. were listed. In Europe, the first full-scale MBR plant for treatment of municipal wastewater was constructed in Porlock (UK, commissioned in 1998, 3800 p.e.), soon followed by WWTPs in Bu¨chel and Ro¨dingen (Germany, 1999, 1000 and 3000 p.e., respectively), and in Perthes-en-Gaˆtinais (France, 1999, 4500 p.e.). In 2004, the largest MBR plant worldwide so far was commissioned to serve a population of 80 000 p.e. (in Kaarst, Germany). The installations thus grew from small WWTPs to very large WWTPs within a few
Municipal MBR plants in Japan
Name of plant
Total design capacity (m3d1)
Capacity at commissioning (m3d1)
Membrane format
Start of operation
Fukusaki Kobuhara Yusuhara Okutsu Daito Kaietsu Zyosai Heta Ooda
12 500 240 720 580 2000 230 1375 3200 8600
2100 240 360 580 1000 230 1375 2140 1075
Flat sheet Flat sheet Flat sheet Hollow fiber Flat sheet Hollow fiber – – –
2005 2005 2005 2006 2006 2007 2008 2008 2009
Membrane Biological Reactors
years. Nevertheless, the favored range for MBR systems still appears to be only 100–500 m3 d1 and 1000–20 000 p.e. for industrial and municipal wastewaters, respectively. The design capacity of the industrial units is more than an order of magnitude smaller than for the municipal WWTPs. Lesjean and Huisjes (2008) opined that, although the construction of very large MBR plants (4100 000 p.e.) were recently announced with much publicity, this will remain the exception in Europe, because of the lower lifecycle costs (Lesjean et al., 2004) of WWTP plants equipped with tertiary-membrane filtration (Figure 14). Although not representative of the market, the very large plants will attract much attention and thereby may contribute to the market expansion. The industrial market was the pioneer in the early 1990s, whereas the municipal market took off only in 1999. In 2002, 154 MBR units could be counted, among which 85% were for industrial applications. However, taking the installed membrane surface as an indicator of market share, for the period 2003–05, the municipal sector represented 75% of the market volume. Both municipal and industrial sectors saw a sharp increase in the following years, due to the commercial success and much lower capital and operating costs. By 2005, the market growth rate was linear with at least 50 industrial units and 20 municipal plants constructed per year. This progression rate is expected to sustain in the next years or may even further accelerate owing to the evolution and implementation of European and national regulations (Lesjean et al., 2006). The survey by Lesjean and Huisjes (2008) also demonstrated the predominance of the suppliers Kubota (Japan) and GE–Zenon. Their technologies based on submerged filtration modules have been outstandingly successful since 2002. In recent years, the European market can therefore be seen as a quasi-duopoly of two nonEuropean suppliers. In contrast, the most successful MBR technologies in the 1990s, based on sidestream configurations supplied by Wehrle, Norit X-Flow, Berghof, Rodia Orelis, etc., did not experience any significant market
599
growth over the last 3 years. This could explain the recent move of companies such as Wehrle and Norit to develop and commercialize novel low-energy airlift MBR systems. They argued that the industrial market has become mature: the MBR is considered as the best available technology by many industries. On the other hand, the municipal market is expected to witness further growth over the next decade under the combined effects of the acceleration of plant construction and the capacity increase. 3. North America. Full-scale commercial applications of MBR technology in North America for treatment of industrial wastewaters dated back to 1991 (Sutton, 2003). In the early 1990s, MBR installations were mostly constructed in external configuration. After the mid-1990s, with the development of SMBR system, MBR applications in municipal wastewater extended widely. In the past 15 years, MBR technology has been of increased interest both for municipal and industrial wastewater treatment in North America. The hesitancy on the part of North American municipalities to consider alternative treatment systems to the well-established conventional treatment options delayed the introduction of MBRs into the municipal arena. Industrial applications, particularly for high-strength, difficult-to-treat waste streams, on the other hand, allowed for the considerations of alternative technologies, such as MBRs (Yang et al., 2006). Nevertheless, currently, commercial application in treating industrial wastewaters does not constitute a high percentage of total full-scale MBR plants. Zenon occupies the majority of the MBR market in North America. In 2006, the North American installations constituted about 11% of worldwide installations. As in other places, in North America too, although plant capacities of MBR systems for municipal wastewater treatment are becoming larger, most of the plants in operation are medium scale or small scale in terms of capacity. The largest capacity MBR plant in operation is in Traverse City, MI at 26 900 m3 d1, and the two largest capacity plants under construction are in Johns Creek, GA at 60 000 m3 d1 and King County, Washington State at 136 000 m3 d1.
Capacity, p.e × 104
8
4.16.5.1.4 Decentralized MBR plants: Where and why? 6
4
2
0 1996
1998
2000
2002
2004
2006
2008
Year of commissioning Figure 14 Plot of capacity of randomly selected European MBR plants showing predominance of medium size plants (similar trend prevails worldwide). Data from Schier W, Frechen FB, and Fischer S (2009) Efficiency of mechanical pre-treatment on European MBR plants. Desalination 236: 85–93.
MBR technology can also provide decentralized small-scale wastewater treatment for remote or isolated communities, campsites, tourist hotels, or industries not connected to municipal treatment plants. In small communities, houses are spread out, the population density is low, and hence the use of an on-site system for an individual home or for a cluster of homes could be a cost-effective option. For emerging nations with vast unsewered areas, the population has practically no access to water sanitation, whereby wastewater is directly discharged into water bodies or reused for irrigation without treatment, thus spreading waterborne diseases and causing eutrophication and pollution of water resources. MBR technology could provide a decentralized, robust, and cost-effective treatment for achieving high-quality effluent in such instances. MBRs also offer excellent retrofit capability for expanding or upgrading existing conventional WWTPs.
600
Membrane Biological Reactors
Even when appropriate infrastructure for large-scale water recycling facility exists, the decentralized option may be preferable in some cases. This is because the cost of large-scale water-recycling applications remains high and often uneconomical due to the need to overhaul the existing waterdistribution systems. Large-scale water-recycling applications are, hence, currently somewhat restricted. Furthermore, there is a significant risk of cross-connection associated with the dual-reticulation network, which can seriously dampen public support. While the implementation of the large-scale water recycling is expected to take many years, decentralized water recycling can be applied much more readily. It is expected that MBRs can contribute to a significant increase in decentralized water reclamation and reuse activities. The discussion now centers on the limitations of traditional onsite treatment systems. A gradual but permanent reduction in per-capita water use through socially acceptable means is widely recognized by all stakeholders in the water industry as the strategic longterm sustainable solution to address the ongoing water shortage currently experienced by many countries (Tadkaew et al., 2007). Decentralized wastewater management is not a new concept. Tchobanoglous et al. (2003) defined it as the collection, treatment, and disposal/reuse of wastewater from individual dwellings, clusters of homes or isolated communities, industries, or institution facilities. Traditional decentralized treatment systems such as septic tanks were, in the past, widely used to treat small quantities of wastewater. Due to the likely toughening of environmental legislation in the near future, many of the currently operating wastewater treatments will no longer be acceptable and there will be a need to increase their efficiency significantly. Stricter regulations are found for especially sensitive areas, drinking-waterabstraction areas, and bathing waters. The problem of meeting existing and forecasted more-stringent new regulations affects especially small communities, hotels, and campsites in relatively remote areas without access to sophisticated WWTPs. A major obstacle of decentralized water recycling remains the lack of a suitable technology that can meet the strict and unique effluent criteria required for small-scale water treatment. Some essential requirements are high and reliable treated effluent quality, robustness, tolerance to variable contaminant loading, small footprint, and ease of operation and maintenance. We now discuss the advantages of MBRs in decentralized treatment. As discussed in Section 4.16.5.1.2, historically, the largest number of MBR applications was for a capacity of less than 100 m3 d1. This suggests that the application of MBRs for on-site decentralized system is possible and can offer the most advanced wastewater-treatment options in low-density areas at a cost lower than that of conventional large-scale pipeand-plant systems. Jefferson et al. (2000) argued that smallscale WWTPs constitute a potential growth market for the next millennium and urban sustainability through domestic water recycling is a major identified source for this development. Key advantages of MBRs for decentralized wastewater treatment and reuse are:
•
High and reliable treated effluent quality, small footprint, and high tolerance to variable contaminant loading.
•
•
Due to the robustness and modular nature of MBRs, smallscale MBRs can retain the superiority over conventional treatment methods such as septic tanks with regard to effluent quality, which has been very well documented in the literature (Fane and Fane, 2005). MBRs can be easily combined with other complementary treatment technologies such as UV disinfection and prescreening, which can further enhance the robustness of the treatment system and hence make it particularly suitable for water-recycling applications.
The MBRs for decentralized treatment are not without limitations. Besides the obstacles against widespread application of MBR, in general, the high capital cost can be seen as the key limitation of small-scale MBRs although currently there is very little information to substantiate this premise. Friedler and Hadari (2006) analyzed the economic feasibility of on-site graywater-reuse systems in buildings based on MBR systems. They found that on-site MBR systems became feasible when they were used for the treatment of wastewater incorporating several buildings together because cost was sensitive to building size. Therefore, the on-site MBR system for single building might be unfeasible. This could be a limitation of decentralized MBR systems. However, the true cost of water supply, which takes into account the externalities of resource depletion, was not used in their analysis. It is expected that as the demand for decentralized MBRs increases and membrane technology continues to develop, the use of on-site MBRs even for individual dwellings can be cost competitive in the near future. Some of the examples of worldwide decentralized MBRs are discussed next. The successful introduction of MBR systems into small-scale and decentralized applications has led to the development of packaged treatment solutions from most of the main technology suppliers. Sports stadia, shopping complexes, and office blocks are becoming typical end users, especially in areas of water stress (Stephenson et al., 2000; Melin et al., 2006; Tadkaew et al., 2007). The application of MBRs in Japan to date has predominantly concerned small-scale installations for domestic wastewater treatment. One of the earliest reported case studies is on graywater recycling facilities in the Mori building, Tokyo (Stephenson et al., 2000). The plant consists of a sidestream Pleiade MBR (Ubis) to treat the building flow of 500 m3 d1. The selection of an MBR over a traditional treatment process saved an area equivalent to 25 car-parking places. The treated graywater contained less than 5.5 mg l1 BOD and belowdetection level of suspended solids, colon bacilli, and n-hexane extract, enabling reuse of the graywater. Today, the main Japanese MBR providers such as Kubota or Mitsubishi Rayon offer commercial MBR package plants for on-site domestic water treatment. In Australia, small-scale MBR systems for graywater recycling at a single household level have been marketed by several companies such as AquaCell in New South Wales and BushWater in Queensland (Tadkaew et al., 2007). Commercially available systems in Europe include the package treatment plant Clereflo MBR (Conder Products, UK), designed to service populations up to 5000 and the ZeeMOD (Zenon Environmental Inc.) which is available for flow rates
Membrane Biological Reactors of up to 7500 m3 d1. Most of the manufacturers offer similar systems which means that effluent qualities of 5:5:5 (mg l1) (BOD: NH4-N:SS) are now routinely available to end users as standard treatment options (Melin et al., 2006). Households/ community units (4–50 p.e.) is a concept pioneered by Busse (Germany) in 2000 (Lesjean and Huisjes, 2008). This has become a very competitive market (at least eight products available in Germany). The units are mostly covered by maintenance contracts. The number of sales is expected to increase to address wastewater schemes of small and remote communities, although the revenue may remain marginal in the overall European MBR market. An example in USA is in eastern San Diego County, California, where expansion of an existing casino and development of a shopping mall required extension to the existing treatment facilities. The existing extended aeration system was converted to a ZeeWeed MBR allowing almost triple the capacity of the infrastructure (Melin et al., 2006). The scheme has been operational since July 2000 with the water quality meeting the California tertiary effluent standards for waterreclamation plants.
4.16.5.2 Commercialized MBR Formats As mentioned in Section 4.16.3.1, the first-generation MBRs in wastewater treatment used a sidestream format, in which feed was pumped from the bioreactor through an external membrane system. This approach was suitable for the early stage, small-scale applications for difficult-to-treat feeds. An alternative format was developed in the 1990s using modules submerged in the bioreactor tank, or in an adjoining compartment. This was much more cost effective for treating larger-scale flows with more easily treatable wastewater. The submerged format is available with modules either in a flat-sheet configuration or as hollow fibers or capillary membranes. Originally, the favored concept was to submerge the modules directly into the bioreactor for simplicity. However, in order to gain better control of the balance between the biological and filtration-treatment capacity, it is now more common to use the membrane in an external membrane tank (Brow, 2007). The external arrangement allows the size and design of the membrane tank to be optimized independently, with practical advantages for operation and maintenance. The sidestream approaches are also divided into two formats – the long-established traditional method of crossflow, now used only for the most difficult feeds, and the newer concept of airlift, which uses air to recirculate the feed and thereby significantly reduces energy demand. Both sidestream formats use tubular membranes.
4.16.5.3 Case-Specific Suitability of Different Formats The competing MBR formats based on submerged and sidestream configurations each have their own pros and cons for different application types and plant size. The energy cost for the aeration to control membrane fouling in the MBR is of an order similar to the microbiology aeration for an easy-to-treat feed, and increases by 2.5–3.0 times for the more difficult feed (Cornel and Krause, 2006).
601
Crossflow is more energy intensive – very high cross-flow velocities (up to 5–6 m3 h1) may be necessary to control the fouling; but for the more difficult feeds, it may be the only option that works reliably. Airlift is a more cost-effective way of improving mass transfer through the creation of slug-flow conditions in the lumen of the membrane tubes (Laborie et al., 1997), but there is a limit to how much air flow can be used while retaining slug-flow conditions. Airlift technology has a power cost similar to that of the submerged technology. In general, submerged MBR formats based on hollow fibers have been found to provide the most cost-effective solution for large-scale, easy-to-treat applications. Technology has been developed with optimized packing density and aeration bubble size to achieve stable performance at minimum energy use (Fane et al., 2005). However, this format can experience operational difficulties due to fibers becoming matted close to the potted ends, and therefore pretreatment and removal of hairs and fibers is essential. Hollow-fiber technology hence requires more instrumentation and control. The submerged MBR formats based on flat sheets have been found to be cost effective for similar types of wastewater, but due to higher air use and lower compactness, tend to be selected for small- to medium-scale duties. The flat-sheet format has operational advantages in terms of plugging and cleaning, and has been used in somewhat more difficult feeds. Flat-sheet systems have the advantage of relatively low manufacturing cost compared to hollow-fiber systems. However, packing density tends to be significantly lower than a hollow-fiber system (e.g., by a factor of 2.5–3 times). Therefore, flat-sheet systems tend to have a cost advantage for smallto medium-scale systems, whereas hollow fiber becomes more attractive at large scale due to the footprint advantage (Pearce, 2008b). The comparison is made more complicated, however, since aeration costs for hollow-fiber systems are often lower. This means that the most cost-effective solution for total treatment costs at medium scale is closely contested, and both approaches are found across the size range due to site-specific circumstances, which could favor either solution. Lesjean et al. (2004), taking into account the current knowledge, anticipated a future market share as follows: for municipal applications, it is expected that the hollow-fiber submerged configuration would be competitive for mediumto large-size plants. For small to medium sizes, flat-sheet technologies would have an advantage. However, in case of larger plants, or a plant refurbishment, the alternative membrane scheme (secondary/tertiary treatment followed by an MF/UF membrane filtration) is very likely to be cost competitive, unless high-cost land has to be purchased for the construction. This multi-barrier scheme will also be easier to control and to optimize because of the disconnection of the treatment steps. It will also be associated with the lowest risk in relation to the membrane operation, as the membranes will be operated under smooth hydrodynamic conditions in terms of particle matter, turbulence, and backwash re´gime. In a recent paper, Lesjean and Huisjes (2008) reiterated this expectation despite the present trend of large MBR plant construction. The airlift format has been developed as a low-energy alternative to the energy-intensive cross-flow sidestream format,
602
Membrane Biological Reactors
which has been used historically for the most difficult feeds. As mentioned earlier, the energy cost of crossflow prohibits it as a treatment option for any application other than small scale or where there is no other treatment option. However, the airlift has very low energy use, and may even undercut the energy requirements of the submerged options, due to the advantage of containment of the feed inside the tubular membrane (Van ‘T Oever, 2005; Futselaar et al., 2007). Since airlift eliminates operator contact and has good operational characteristics, it may as well make a major impact on the MBR market in the long run. Pearce (2008a, 2008b) argued that the airlift format may find applications throughout a broader range than the submerged formats. Figure 15 depicts the concept of airlift MBR.
4.16.5.4 MBR Providers 4.16.5.4.1 Market share of the providers The global market value of MBR is expected to rise up to US$500 million by 2013 from around US$300 million in 2008 (BCC Research, 2008). The MBR market is dominated by three companies, namely GE–Zenon, Kubota, and Mitsubishi Rayon Engineering (MRE). Only GE–Zenon and Kubota have a strong presence in Europe and North America, while MRE have until now mainly focused on sales in Asia. All these companies use submerged formats, with GE–Zenon and MRE Air release
Return to bioreactor
Permeate
Permeate backwash
Air injection
Airlift Feed supply Figure 15 The concept of airlift MBR.
using hollow-fiber membranes, and Kubota, flat-sheet membranes. Another three companies too have an international presence, namely Siemens–Memcor, Norit, and Koch-Puron, but the sales for these three companies makes up a small portion of the worldwide market. Among the latter three, Norit promotes the airlift format. Figure 16(a) shows the worldwide relative market share (in terms of installations numbers) for the three large players (Yang et al., 2006; Pearce, 2008b). The MBR market has several dozen regional or application specialists, quite a few of who use flat-sheet formats as adopted by Kubota: for example, Japan’s Toray and A3 from Germany. In addition to these international companies, there are a further 30 companies in the European Union (EU) market that have either significant regional presence, or an application focus, or a low-level international presence (Lesjean and Huisjes, 2008). Many of these companies are significant in the local markets, but individually, they have a small share of the international market. It is interesting to note that the MBR market has characteristics different from that of the UF/MF market. In UF/MF, there are 10–12 significant players with worldwide presence, with four market leaders, none of who dominate the market. Besides these companies, other regional players are relatively insignificant (Pearce, 2008a, 2008b). Zenon is long established in the market and has been one of the major companies promoting the MBR concept, and the use of PVDF membranes. The North American market is dominated by Zenon (Yang et al., 2006) as shown by the revenue share illustrated in Figure 16(b) and has many more opportunities in the municipal sector than in industry. Zenon leads the European market as well (Figure 16(c)). Kubota was one of the early pioneers of the MBR concept, encouraged by a Japanese Government initiative in the 1980s. They achieved a very large number of installations in small- to medium-scale systems, initially focusing on the residential/ commercial market in Japan and have approached export markets through exclusive partnerships. Kubota has a significantly greater number of plants than Zenon, with a slightly higher proportion of industrial plants. Many of Kubota’s installations in Japan and Korea are for small-scale municipal and domestic applications. Figure 17 shows the market characteristics of the two market leaders, Kubota and Zenon, illustrating the significantly different market strategies with regard to the size of plant targeted. Kubota is the strongest market player for industrial and small-scale municipal applications. MRE is a long-established supplier of MBR, with a very strong position in the relatively mature MBR market in Japan and Korea. There are a large number of references for this technology in Asia, but relatively few installations elsewhere. MRE also has a very large number of installations, with a higher proportion of industrial users, mostly with small flowrates. Koch Membrane Systems (KMS) is a long-established membrane manufacturer and membrane-systems company. In 2004, KMS acquired the MBR start-up company Puron, which had been founded in 2001. They introduced an approach to fiber potting different from that of the other hollow-fiber module providers.
Membrane Biological Reactors
603
15
17
68
(a) Worldwide (relative installation numbers % in 2006)
2 3
6
10
20
33
65
61
(b) North America (revenue % in 2003)
GE−Zenon
Kubota
(c) Europe (installed membrane surface % in 2005)
Mitsubishi−Rayon
Siemens−Memcor
Koch−Puron
Others (N. America: Mitsubishi, Norit; Europe: Norit, Wehrle and other EU and non-EU suppliers) Figure 16 Market share of the suppliers. Data from (a) Yang Q, Chen J, and Zhang F (2006) Membrane fouling control in a submerged membrane; (b) Pearce G (2008 b) Introduction to membranes – MBRs: Manufacturers’ comparison: Part 1. Filtration and Separation 45(3): 28–31; and (c) calculated from Lesjean B and Huisjes EH (2008) Survey of the European MBR market: Trends and perspectives. Desalination 231: 71–81.
Memjet product is characterized by high permeability and packing density, providing a competitive position for capital and operating costs. However, worldwide market share for MemJet MBR is not very significant, since the company tends to focus on selected regional markets (Yang et al., 2006; Pearce, 2008b).
100 Plant capacity
80
60
No. of plants
% 40
20
0 Kubota
GE−Zenon
Figure 17 Relative market share (number of plants and capacity) showing distinct market strategies of the two market leaders.
Memcor have extensive experience in the use of their products in wastewater polishing. Their very fine polypropylene (PP) fibers developed in the 1980s were inexpensive and flexible, but unfortunately had low chlorine tolerance (Judd et al., 2004). In the late 1990s, Memcor developed a PVDF fiber, and now use the PVDF fiber for their MBR product range. The
4.16.5.4.2 Design considerations The design of the reactor (including membrane, baffle, and aerator locations) and the mode of operation of the membrane are key parameters in the optimization of the system. The leading MBR providers propose several MBR designs. In each case, the process proposed is very specific. Not only are the membrane material and configuration used different, but the operating conditions, cleaning protocols, and reactor designs also change from one company to another. For example, the flat-sheet membrane provided by Kubota does not require backwash operation, while hollow-fiber membranes have been especially designed to hydraulically backwash the membrane on a given frequency. The MBR industry first developed in Japan with the use of chlorinated polyethylene (PE) flat-sheet membrane by Kubota, and PE fibers by MRE (Stephenson et al., 2000). The modified PE is characterized by reasonable strength, flexibility, wettability, and resistance to chlorine. Although PE is normally made as an MF membrane, it has relatively low permeability, so process fluxes of PE membranes tend to be at the
Membrane Biological Reactors
Table 9
air-usage efficiency. In addition, the companies using hollow fiber use intermittent aeration, for example, based on a timer in the case of Zenon, or in proportion to flow in the case of Koch–Puron. Memcor introduced a novel cleaning method by using a mixture of air and mixed liquor, instead of using only air bubbles, to scour the membranes. The air bubbles effectively scour the membranes and the semi-crossflow of mixed liquor along the membranes continuously delivers the refresh mixed liquor to the membrane surface, minimizing the solidconcentration polarization at the membrane surface and therefore reducing filtration resistance. These enhancements have significantly reduced air usage and therefore power cost.
4.16.5.4.3 Performance comparison of different providers Few large-scale studies based on comparison of the commercially available MBR systems have been conducted. The city of San Diego, California, and the research consultant, Montgomery Watson Harza, have been evaluating the MBR process through various projects since 1997, including feasibility of using MBRs to produce reclaimed water (Adham and Gagliardo, 1998, 2000), optimization of MBR operation, and parallel comparison and cost estimations of the four leading MBR suppliers (Adham et al., 2004). MBRs were evaluated for their ability to produce high-quality effluent and to operate with minimum fouling. In terms of hydraulic performances, it (8.5−12)
500
400 (17−24) (50−60)
(17−24)
0
Toray
(29)
Norit
Siemens−Memcor
Koch−Puron
100
(30−34)
GE−Zenon
(17−24)
(14−26)
Mitsu. (PVDF)
200
Mitsubishi (PE module)
300
Kubota
low end of the range. Consequently, PE membranes are very cost effective at small scale, but struggle to compete in largerscale systems. In the 1990s, PVDF became established in MBRs through the reinforced capillary fiber in Zenon’s ZW 500 module (Yamato et al., 2006). PVDF has impressive performance in terms of strength and flexibility, but is significantly more expensive as a polymer. Nevertheless, PVDF membranes can achieve substantially higher flux, thereby overcoming price disadvantage. Recently, MRE also developed a PVDF-based membrane system. This membrane, designated as SADF, promises to be very competitive in both capital and operating costs, and despite it having a lower packing density than the PE product, it operates at much higher flux. With several companies now offering PVDF products in both capillary and flat-sheet formats, this is the dominant membrane polymer in the MBR market (Pearce, 2008c, 2008d). The third significantly used membrane polymer in MBR is a reinforced PES, used by Koch–Puron. Although PES is an important polymer in water treatment, in wastewater applications, its lack of flexibility limits the possibility of using air scour. Reinforcing the capillary does allow air scour, but at the expense of permeability. The Puron product uses reinforced PES rather than the PVDF, favored by its rivals. However, its main distinguishing feature is that the membrane fibers are potted at only one end. This overcomes the problem of fouling below the potting interface by hairs and fibers, which is a problem for the other hollow-fiber technologies (Vilim et al., 2009). Norit is the one major MBR company that offers a system based on a sidestream format with tubular membranes rather than a submerged format. Crossflow is only used for smallscale applications, with feeds that are difficult to treat, whereas airlift is cost effective for larger-scale municipal applications (Futselaar et al., 2007). Table 9 summarizes the specifications of the membranes used by different suppliers and Figure 18 compares the packing density and applicable flux of the membranes. Each of the suppliers makes regular improvements in air usage, since this has an important impact on total water cost. For example, the flat-sheet suppliers now use 1.5-m panels, which reduce air flow by up to 30% compared to the original 1 m panel (Pearce, 2008c, 2008d). In addition, they also use double-deck stacks wherever possible, which further improves
Membrane packing density, m2 m−3
604
Figure 18 Packing density (bar chart, m2 m3) and flux (values within parentheses, l m2 h1) of membranes from different suppliers.
MBR supplier specificationsa
Company
Membrane material
Pore size, mm
Membrane format
Fiber/tube dia (id,od),mm
pH tolerance
Kubota Mitsubishi Mitsubishi GE–Zenon Koch–Puron Siemens–Memcor Noritb Toray
Cl2 PE PE PVDF PVDF PES PVDF PVDF PVDF
0.4 0.4 0.4 0.04 0.05 0.04 0.03 0.08
FS HF HF HF HF HF TUB FS
– 0.37, 0.54 11, 2.8 0.8, 1.9 –, 2.6 –, 1.3 –, 5.2 or 8.0 –
1–13 1–13 1–10 2–10.5 2–12 2–10.5 1–11 1–11
a
All the membranes have moderate hydrophilicity and high chlorine resistance. All the companies except Norit use submerged format; Norit supplies airlift sidestream MBRs. FS, flat sheet; HF, hollow fiber; TUB, tubular.
b
Membrane Biological Reactors
was shown that all four processes were able to cope with flux rates exceeding 33 l m2 h1 and HRTs as low as 2 h. A 6-year development program has also been initiated for the introduction of MBR technology in the Netherlands market. Started in 2000, a comparative study of four 750 m3 d1 MBRs carried out by DHV water has been reported (van der Roest et al., 2002b). Three MBR plants, treating a design flow of 300 m3 d1 each, have been operated in parallel during 2003 and 2004 in Singapore (Le-Clech et al., 2006). A 4-year study, started in 2001, comparing the performance of Mitsubishi, Kubota, and Zenon MBR was conducted by the Swiss Federal Institute of Aquatic Science and Technology (EAWAG) (Judd, 2006). The Zenon MBR exhibited the most stable performance in the study. Although these studies have been conducted with the MBR systems running in parallel (with the same influent water), the MBR maximum flux, operating conditions and general design applied were those recommended by the suppliers, and therefore somewhat different for each system. This makes it difficult to make a fair comparison. Therefore, it is not possible to classify the MBRs as a function of their relative hydraulic performances, which need to be considered along with the cleaning protocols applied to each system. Mansell et al. (2004) performed measurements in which MS2 coliphage were seeded to the influent of a Kubota MBR (characteristic pore size 0.4 mm) and a Zenon MBR (characteristic pore size 0.04 mm). Permeate concentrations showed a log removal range of 3.2–7.4 for the Kubota installation and 5.32–7.5 for the Zenon installation. All of the heavy metals detected in the influent were removed to levels below detection limit, as well as the VOCs that were measured.
4.16.5.5 Standardization of Design and Performance-Evaluation Method The MBR market is very fragmented and exhibits many MBR filtration products with diverse geometries, module capacities, and operational modes (De Wilde et al., 2008; Lesjean and Huisjes, 2008). Although this situation promotes a competitive market, it is detrimental for the acceptance of the technology as a state-of-the-art process, and raises concern with potential clients or end users. From the point of view of the MBR operators, the possibility of interchanging filtration modules of different companies/suppliers would facilitate the replacement of the modules at the end of their life, and would reduce the risk of a supplier withdrawing from the market or releasing a new series of the product. In addition, the stakeholders in the industry employ various methods of membrane characterization and performance evaluation. This creates confusion among the users and prohibits fair comparison. Based on an extensive survey of the MBR industry, De Wilde et al. (2008) provided an overview of the market interests/expectations and technical potential of going through a standardization process of the SMBR technology in Europe. Due to the predominance of submerged filtration systems in municipal applications, the study focused only on this configuration. Two different aspects of standardization were considered:
•
standardization of MBR filtration modules toward interchangeable modules in MBRs and
•
605
standardization of MBR acceptance and monitoring test methods toward uniform quality-assessment methods of MBR filtration systems.
4.16.5.5.1 Standardization of MBR filtration systems In relation to the market expectations, about 20 potential technological, financial, economical, or environmental benefits/opportunities and drawbacks/threats of MBR module standardization for suppliers and operators were identified and mapped. It appeared that the number of advantages and disadvantages was quite balanced for both sides of the market, the main advantage perceived by the industry being that standardization should contribute to the growth of the MBR market. Other main advantages/opportunities are avoidance of vendor lock-in, price decrease, and increased trust and acceptance. Main disadvantages/threats for the end users are overdimensioning of civil constructions and supplementary works and costs to the peripherals during replacement. Main disadvantages for the module suppliers seem to be the higher competition, lower profit margins, and a limitation for innovative module producers to enter the market. From the technical point of view, the analysis showed that a standardization process common for both flat-sheet and hollow-fiber membranes/modules would not be realistic. In order to achieve interchangeability of filtration modules, not only should the prospect of pure dimensional standards for the module be considered, but also the design and mode of operation of the peripheral components, such as the filtration tank, pumps, blowers, and pretreatment should be borne in mind. More than 30 technical factors hampering or interfering with a standardization process were identified and quantified, and their relative potential for affecting the possible outcome was evaluated. For instance, four factors were grouped as the extremely high hindering factors: module dimensions, filtration tank dimensions, specific permeate production capacity, and specific coarse-bubble aeration demand. These factors are mainly the result of a completely different geometry and design of the filtration module and discussions for the standardization of MBR filtration systems should in essence focus on these factors. For each category, more or less the same number of obstacles lies ahead. Nevertheless, the nature of some of these obstacles or points of attention can be different. Some factors are specifically important for FS modules (e.g., flushing of air-supply pipes and design of a permeatecollection tank), and others for HF modules (e.g., type of prescreening, whether gravity filtration or any other type).
4.16.5.5.2 Standardization of MBR characterization methods The survey conducted by De Wilde et al. (2008) also revealed the respondents’ consensus in general on the positive impact of harmonization of membrane-acceptance tests at module delivery and monitoring methods on municipal MBR market growth. Some important parameters, for which a common definition and measurement protocol could be helpful, are mentioned below:
•
clearly defined and harmonized parameters to monitor membrane fouling, integrity, and aging;
606
• • • • • • • • •
Membrane Biological Reactors
a common definition of membrane lifetime for the guarantee clause; determination/definition of flux (operation and nominal design); common definition for sustainable peak hydraulic load; harmonized tests to check membrane performances over a defined period and under specific conditions; characterization method for membrane acceptance at module delivery; minimum requirements and technical methods to check membrane performance at plant commissioning; monitoring methods of normalized permeability in clear water, permeability in sludge, transmembrane pressure, and fouling rate; monitoring methods of sustainable flux and maximum flux; and operating conditions (biology and filtration systems) for warranty clauses.
It is interesting to note that, most of the newcomers in the market are developing their systems so that they can easily replace the products of the two main suppliers (Zenon–GE and Kubota). A standardization process driven by the end users could accelerate this evolution and contribute to the market development (Lesjean and Huisjes, 2008). Pearce (2008a, 2008b, 2008c, 2008d) also pointed out that, although the dimensions of the relatively newer Puron products are not identical to Zenon’s ZW 500d or MRE’s SADF, the elements are similar, and cassettes made from the elements could be used interchangeably. This begins to introduce retrofit possibilities into what hasuntil now been a fragmented market with no standardization.
4.16.6 Future Vision In addition to the alleviation of the technology bottlenecks illustrated in this chapter, a radical shift from the conventional concept of advanced wastewater treatment is deemed
Urine separation is also worthwhile to be considered
4.16.7 Conclusion MBR is a physicobiological hybrid process. The membrane provides a physical barrier for hygienically safe and clean water with the help of microbial–ecological treatment that can achieve good public acceptance. It is also well recognized by the experts that the clear membrane permeate makes post treatment easy; then, a variety of hybrid systems having the MBR as the core can be considered depending on the specific quality requirements of the reclaimed water . These advantages
A large amount of diluted organic wastewater (graywater)
To co-generation system A small amount of highstrength organic waste kitchen waste disposer-wastewater and toilet flushing)
imperative. In the context of sustainable water system, the advanced treatment must couple technologies to produce water of the required quality and realize material conversion from waste as well. The required quality does not always mean high quality. The quality comes from necessity. Membrane technology has the potential to be an on-demand quality provider just by separation. The conversion mainly comes from the biological reaction in the MBR. Three aspects of a sustainable society, namely, the low carbon society, sound material cycle society, and ecological society, are notable. From the point of view of sustainable water system, the advanced wastewater-treatment processes can be classified into the categories of energy saving (or productive), material productive, and ecologically oriented. The MBR technology might match more with the first two. However, present MBR technologies are still large energy consumers. Next-generation MBRs need to be developed to reduce the significant aeration requirement (by compact module design and sludge-concentration control techniques) and recover energy (e.g., by adding other organic wastes and combining anaerobic digestion for methane recovery). In line with the proposed definition of advanced treatment, the notion needs to be changed from organic wastewater treatment to water/biomass production by developing next-generation MBRs where the membrane acts as a separator of water and biomass and biomass is utilized for energy production. The concept is illustrated in Figures 19 and 20.
Anaerobic pretreatment Pretreatment Methane production
Biomass production from liquid organic waste (*)
Aerobic MBR
(*)
(A very small amount of residue) • Renewable energy utilization • IT-based maintenance service system • User participation in monitoring
(*) N,P recovery option
Figure 19 Next-generation MBR system: anaerobic combination for on-site small-scale advanced treatment.
Safe effluent
Membrane Biological Reactors
W.W.
Solid−liquid Solid-liquid separation
Solid concentration/ concentration/ Anoxic anoxic reaction reaction
Biosorption/ membrane separation/ aerobic reaction
607
Safe reclaimed water
Energy and/or material recovery process Other than biogas production, physicochemical treatments are also candidates for energy recovery, for example, supercritical water gasification of sludge−water mixture where the biomass sludge is utilized as energy source to produce hydrogen from water molecules (coupling clean energy production). Figure 20 Next-generation MBR system: renovation of existing wastewater-treatment plants.
make MBR a good device in water reclamation and/or advanced wastewater treatment. The continued push toward stricter discharge standards, increased requirement for water reuse, and greater than before urbanization and land limitations fuel the use of MBRs. However, there is room for improvement to utilize the potential of the MBR fully. The challenges will center on energy saving, ease of operation, simplified membrane cleaning and replacement strategies, and peak-flow management. The international adventure on R&D of MBR technologies continues.
References Achilli A, Cath TY, Marchand EA, and Childress AE (2009) The forward osmosis membrane bioreactor: A low fouling alternative to MBR processes. Desalination 239: 10--21. Adham S, DeCarolis J, and Pearce W (2004) Optimization of various MBR systems for water reclamation – phase III. Agreement No. 01-FC-81-0736, Program Final Report No. 103, April 2004. Denver, CO: US Department of the Interior, Bureau of Reclamation, Denver Office. Adham S and Gagliardo P (1998) Membrane bioreactors for water repurification – phase I. Desalination Research and Development Program Report No. 34, Project No. 1425-97-FC-81-3 0006. Denver, CO: US Department of the Interior, Bureau of Reclamation, Denver Office. Adham S, Gagliardo P, Boulos L, Oppenheimer J, and Trussel R (2001) Feasibility of the membrane bioreactor process for water reclamation. Water Science and Technology 43(10): 203--209. Adham S, Mirlo R, and Gagliardo P (2000) Membrane bioreactors for water reclamation – phase II. Desalination Research and Development Program Report No. 60, Project No. 98-FC-81-0031. Denver, CO: US Department of the Interior, Bureau of Reclamation, Denver Office. Al-Amoudi A and Lovitt RW (2007) Fouling strategies and the cleaning system of NF membranes and factors affecting cleaning efficiency. Journal of Membrane Science 303: 4--28. Alnaizy R and Sarin V (2009) Performance assessment of MBR applications for metropolitan wastewater treatment in the United Arab Emirates. In: Lesjean B and Leiknes T (eds.) Final MBR-Network Workshop, pp. 85–86. Berlin, Germany, 31 March–1 April. Berlin: MBR-Network Workshop. Arika M, Kobayashi H, and Hihara H (1966) Pilot plant test of an activated sludge ultrafiltration combined process for domestic wastewater reclamation. Desalination 48: 299--319. Asatekin A, Menniti S, Kang M, Elimelech M, Morgenroth E, and Mayes AM (2006) Antifouling nanofiltration membranes for membrane bioreactors from selfassembling graft copolymers. Journal of Membrane Science 285: 81--89. Aya H (1994) Modular membranes for self-contained reuse systems. Water Quality International 4: 21--22.
Bae TH, Kim IC, and Tak TM (2006) Preparation and characterization of foulingresistant TiO2 self-assembled nanocomposite membranes. Journal of Membrane Science 275: 1--5. Bae TH and Tak TM (2005) Interpretation of fouling characteristics of ultrafiltration membranes during the filtration of membrane bioreactor mixed liquor. Journal of Membrane Science 264: 151--160. Bacchin P, Aimar P, and Field RW (2006) Critical and sustainable fluxes: Theory, experiments and applications. Journal of Membrane Science 281: 42--69. Bacchin P, Si-Hassen D, Starov V, Clifton MJ, and Aimar P (2002) A unifying model for concentration polarization, gel layer formation and particle deposition in cross-flow membrane filtration of colloidal suspensions. Chemical Engineering Science 57: 77--91. BCC Research (2008) Membrane bioreactors: Global markets. Report Code MST047 B, Report Category – Membrane & Separation Technology. Beltfort G, Davis RH, and Zydney AL (1994) The behaviour of suspension and macromolecular solutions in crossflow microfiltration. Journal of Membrane Science 96: 1--58. Benefield LD and Randall CW (1980) Biological Process Design for Wastewater Treatment. Englewood Cliffs, NJ: Prentice Hall. Benitez J, Rodriguez A, and Malavar R (1995) Stabilization and dewatering of wastewater using hollow fiber membranes. Water Research 29: 2281--2286. Bixio D, Thoeye C, De Koning J, et al. (2006) Wastewater reuse in Europe. Desalination 187: 89--101. Bixio D, Thoeye C, Wintgens T, et al. (2008) Water reclamation and reuse: Implementation and management issues. Desalination 218: 13--23. Bouhabila EH, Ben Aim R, and Buisson H (2001) Fouling characterization in membrane bioreactors. Separation and Purification Technology 22: 123--132. Brepols C, Dorgeloh E, Frechen FB, et al. (2008) Upgrading and retrofitting of municipal wastewater treatment plants by means of membrane bioreactor (retrofit aoMBR) technology. Desalination 231: 20--26. Brindle K and Stephenson T (1996) The application of membrane biological reactors for the treatment of wastewaters. Biotechnology and Bioengineering 49: 601--610. Brookes A, Jefferson B, Guglielmi G, and Judd SJ (2006) Sustainable flux fouling in a membrane bioreactor: Impact of flux and MLSS. Separation Science and Technology 41: 1279--1291. Brookes PR and Livingston AG (1993) Point source detoxification of an industrially produced 3,4-dichloroaniline-manufacture wastewater using a membrane bioreactor. Applied Microbiology and Biotechnology 39: 764--771. Brookes PR and Livingston AG (1994) Biological detoxification of a 3chloronitrobenzene manufacture waste-water in an extractive membrane bioreactor. Water Research 28: 1347--1354. Brow PJ (2007) Ballyclare MBR WwTW: Cost & challenges of compliance with stringent standards. In: Proceedings of the 4th International Water Association Conference on Membranes for Water and Wastewater Treatment, CD ISBN: 978-186194-127-5. Harrogate, UK, 15–17 May. London: IWA Publishing. Cabassud C, Masse A, Espinosa-Bouchot M, and Sperandio M (2004) Submerged membrane bioreactors: Interactions between membrane filtration and biological activity. In: Proceedings of the Water Environment-Membrane Technology Conference. Seoul, Korea, 7–10 June. London: IWA Publishing. Cao JH, Zhu BK, Lu H, and Xu YY (2005) Study on polypropylene hollow fiber based recirculated membrane bioreactor for treatment of municipal wastewater. Desalination 183: 431--438.
608
Membrane Biological Reactors
Cha GC, Jeong TY, Yoo IK, and Kim DJ (2004) Kinetics characteristics of SMP and ECP in relation to loading rate in a MBR process. In: Proceedings of the Water Environment – Membrane Technology Conference. Seoul, Korea, 7–10 June. London: IWA Publishing. Chae SR, Ahn YT, Kang ST, and Shin HS (2006a) Mitigated membrane fouling in a vertical submerged membrane bioreactor (VSMBR). Journal of Membrane Science 280: 572--581. Chae SR, Kang ST, Watanabe Y, and Shin HS (2006b) Development of an innovative vertical submerged membrane bioreactor (VSMBR) for simultaneous removal of organic matter and nutrients. Water Research 40: 2161--2167. Chang IS and Lee CH (1998) Membrane filtration characteristics in membrane coupled activated sludge system – the effect of physiological states of activated sludge on membrane fouling. Desalination 120: 221--233. Chang IS, Lee CH, and Ahn KH (1999) Membrane filtration characteristics in membrane-coupled activated sludge system: The effect of floc structure on membrane fouling. Separation Science and Technology 34: 1743--1758. Chang IS, Gander M, Jefferson B, and Judd SJ (2001) Low-cost membranes for use in a submerged MBR. Transactions IChemE 79(B): 183--188. Chang IS and Judd SJ (2002) Air sparging of a submerged MBR for municipal wastewater treatment. Process Biochemistry 37: 915--920. Chang IS and Kim SN (2005) Wastewater treatment using membrane filtration – effect of biosolids concentration on cake resistance. Process Biochemistry 40: 1307--1314. Chang IS, Le-Clech P, Jefferson B, and Judd S (2002a) Membrane fouling in membrane bioreactors for wastewater treatment. Journal of Environmental Engineering ASCE 128: 1018--1029. Chang S and Fane AG (2001) The effect of fibre diameter on filtration and flux distribution-relevance to submerged hollow fibre modules. Journal of Membrane Science 184: 221--231. Chang S and Fane AG (2002) Filtration of biomass with laboratory-scale submerged hollow fibre modules – effect of operating conditions and module configuration. Journal of Chemical Technology and Biotechnology 77: 1030--1038. Chang S, Fane AG, and Vigneswaran S (2002b) Modeling and optimizing submerged hollow fiber membrane modules. AIChE Journal 48: 2203--2212. Chaize S and Huyard A (1991) Membrane bioreactors on domestic wastewater treatment sludge production and modelling approach. Water Science and Technology 23: 1591--1600. Charcosset C (2006) Membrane processes in biotechnology: An overview. Biotechnology Advances 24: 482--492. Chen CL, Liu WT, Chong ML, et al. (2004) Community structure of microbial biofilms associated with membrane-based water purification processes as revealed using a polyphasic approach. Applied Microbiology and Biotechnology 63: 466--473. Chen JP, Yang CZ, Zhou JH, and Wang XY (2007) Study of the influence of the electric field on membrane flux of a new type of membrane bioreactor. Chemical Engineering Journal 128: 177--180. Chen TK, Chen JN, Ni CH, Lin GT, and Chang CY (2003) Application of a membrane bioreactor system for opto-electronic industrial wastewater treatment – a pilot study. Water Science and Technology 48: 195--202. Cheryan M (1998) Ultrafiltration and Microfiltration Handbook. Lancaster: Technomic Publishing. Chiemchaisri C, Yamamoto K, and Vigneswaran S (1993) Household membrane bioreactor in domestic wastewater treatment. Water Science and Technology 27(1): 171--178. Childress AE, Daugherty J, Chen C, Le-Clech P, and Leslie GL (2005) Hollow fiber membrane integrity: Mechanics and modeling. Desalination 180: 5--14. Cho BD and Fane AG (2002) Fouling transients in nominally sub-critical flux operation of a membrane bioreactor. Journal of Membrane Science 209: 391--403. Cho J, Ahn KH, Seo Y, and Lee Y (2003) Modification of ASM No.1 for a submerged membrane bioreactor system: Including the effects of soluble microbial products on membrane fouling. Water Science and Technology 47(12): 177--181. Cho JW, Song KG, Lee SH, and Ahn KH (2005) Sequencing anoxic/anaerobic membrane bioreactor (SAM) pilot plant for advanced wastewater treatment. Desalination 178: 219--225. Choi H, Zhang K, Dionysiou DD, Oerther DB, and Sorial GA (2005) Effect of permeate flux and tangential flow on membrane fouling for wastewater treatment. Separation and Purification Technology 45: 68--78. Choi JH, Dockko S, Fukushi K, and Yamamoto K (2002) A novel application of a submerged nanofiltration membrane bioreactor (NF MBR) for wastewater treatment. Desalination 146: 413--420. Choi JH, Fukushi K, and Yamamoto K (2007) A submerged nanofiltration membrane bioreactor for domestic wastewater treatment: The performance of cellulose acetate nanofiltration membranes for long-term operation. Separation and Purification Technology 52: 470--477.
Cicek N, Franco JP, Suidan MT, and Urbain V (1999a) Effect of phosphorus on operation and characteristics of MBR. Journal of Environmental Engineering 125: 738--746. Cicek N, Franco JP, Suidan MT, Urbain V, and Manem J (1999b) Characterization and comparison of a membrane bioreactor and a conventional activated sludge system in the treatment of wastewater containing high-molecular weight compounds. Water Environmental Research 71: 64--70. Cicek N, Macomber J, Davel J, Suidan MT, Audic J, and Genestet P (2001) Effect of solids retention time on the performance and biological characteristics of a membrane bioreactor. Water Science and Technology 43: 43--50. Cicek N, Suidan MT, Ginestet P, and Audic JM (2002) Impact of soluble organic compounds on permeate flux in an aerobic membrane bioreactor. Environmental Technology 24: 249--256. Clapp LW, Regan JM, Ali F, Newman JD, Park JK, and Noguera DR (1999) Activity, structure, and stratification of membrane-attached methonotrophic biofilms cometabolically degrading trichloroethylene. Water Science and Technology 39(7): 153--161. Cornel P and Krause S (2003) State of the art of MBRs in Europe. In: Proceedings of the Conference on Application and Perspective of MBRs in Wastewater Treatment and Reuse. Cremona, Italy, 28–29 April. Cornel P and Krause S (2006) Membrane bioreactors in industrial wastewater treatment – European experience, examples and trends. Water Science and Technology 53(3): 37--44. Cornelissen ER, Harmsen D, de Korte KF, et al. (2008) Membrane fouling and process performance of forward osmosis membranes on activated sludge. Journal of Membrane Science 319: 158--168. Cote P (2002) Inverted Air Box Aerator and Aeration Method for Immersed Membrane. US Pat. 6,863,823. Cote P, Bersillon JL, and Faup G (1988) Bubble free aeration using membranes: Process analysis. Journal of Water Pollution Control Federation 60: 1986--1992. Cote P, Buisson H, Pound C, and Arakaki G (1997) Immersed membrane activated sludge for the reuse of municipal wastewater. Desalination 113: 189--196. Cote P, Masini M, and Mourato D (2004) Comparison of membrane options for water reuse and reclamation. Desalination 167: 1--11. Cui ZF, Chang S, and Fane AG (2003) The use of gas bubbling to enhance membrane processes. Journal of Membrane Science 221: 1--35. Daubert I, Mercier-Bonin M, Maranges C, Goma G, Fonade C, and Lafforgue C (2003) Why and how membrane bioreactors with unsteady filtration conditions can improve the efficiency of biological processes. Advances in Membrane Technology 984: 420--435. Davies WJ, Le MS, and Heath CR (1998) Intensified activated sludge process with submerged membrane microfiltration. Water Science and Technology 38(5): 421--428. Debus O and Wanner O (1992) Degradation of xylene by a biofilm growing on a gaspermeable membrane. Water Science and Technology 26(3–4): 607--617. Defrance L and Jaffrin MY (1999) Reversibility of fouling formed in activated sludge filtration. Journal of Membrane Science 157: 73--84. Defrance L, Jaffrin MY, Gupta B, Paullier P, and Geaugey V (2000) Contribution of various constituents of activated sludge to membrane bioreactor fouling. Bioresource Technology 73: 105--112. De Silva DGV, Urbain V, Abeysinghe DH, and Rittmann B (1998) Advanced analysis of membrane-bioreactor performance with aerobic–anoxic cycling. Water Science and Technology 38(4–5): 505--512. De Wilde W, Geenens D, and Thoeye C (2003) Do we really want to build MBRs for domestic wastewater treatment? In: Proceedings MBR 4. Cranfield, April 9. De Wilde W, Richard M, Lesjean B, and Tazi-Pain A (2008) Towards standardisation of the MBR technology? Desalination 231: 156--165. De Wilde W, Moons K, Bixio D, Thoeye C, and De Gueldre G (2009) Technical feasibility and optimal control strategy of dual (hybrid) MBR-CAS concepts for plant refurbishment. In: Lesjean B and Leiknes T (eds.) Final MBR-Network Workshop, pp. 41–42. Berlin, Germany, 31 March–1 April. Di Bella G, Mannina G, and Viviani G (2008) An integrated model for physical– biological wastewater organic removal in a submerged membrane bioreactor: Model development and parameter estimation. Journal of Membrane Science 322: 1--12. Dossantos LMF and Livingston AG (1995) Novel membrane bioreactor for detoxification of voc wastewaters – biodegradation of 1,2-dichloroethane. Water Research 29: 179--194. Doyen W, Mues W, Molenberghs B, and Cobben B (2010) Spacer fabric supported flatsheet membranes: A new era of flat-sheet membrane technology. Desalination 250: 1078--1082. Drews A, Vocks M, Iversen V, Lesjean B, and Kraume M (2005a) Influence of unsteady membrane bioreactor operation on EPS formation and filtration resistance. In:
Membrane Biological Reactors
Proceedings of the International Congress on Membranes and Membrane Processes (ICOM). Seoul, Korea, 21–26 August. Drews A, Evenblij H, and Rosenberger S (2005b) Potential and drawbacks of microbiology–membrane interaction in membrane bioreactors. Environmental Progress 24: 426--433. Drews A, Lee CH, and Kraume M (2006) Membrane fouling – a review on the role of EPS. Desalination 200: 186--188. Dufresne R, Lebrun RE, and Lavallee HC (1997) Comparative study on fluxes and performances during paper mill wastewater treatment with membrane bioreactor. Canadian Journal of Chemical Engineering 75: 95--103. Enegess D, Togna AP, and Sutton PM (2003) Membrane separation applications to biosystems for wastewater treatment. Filtration and Separation 40: 14--17. Engelhardt N, Firk W, and Warnken W (1998) Integration of membrane filtration into the activated sludge process in municipal wastewater treatment. Water Science and Technology 38(4): 429--436. Evenblij H and Van der Graaf J (2004) Occurrence of EPS in activated sludge from a membrane bioreactor treating municipal wastewater. Water Science and Technology 50: 293--300. Evenblij H, Geilvoet S, Van der Graaf J, and Van der Roest HF (2005b) Filtration characterisation for assessing MBR performance: Three cases compared. Desalination 178: 115--124. Evenblij H, Verrecht B, Van der Graaf JHJM, and Van der Bruggen B (2005a) Manipulating filterability of MBR activated sludge by pulsed substrate addition. Desalination 178: 193--201. Fan XJ, Urbain V, Qian Y, and Manem J (1996) Nitrification and mass balance with a membrane bioreactor for municipal wastewater treatment. Water Science and Technology 34: 129--136. Fane AG, Chang S, and Chardon E (2002) Submerged hollow fiber membrane moduledesign options and operational considerations. Desalination 146: 231--236. Fane AG and Fane SA (2005) The role of membrane technology in sustainable decentralized wastewater systems. Water Science and Technology 51: 317--325. Fane AG, Yeo A, Law A, Parameshwaran K, Wicaksana F, and Chen V (2005) Low pressure membrane processes – doing more with less energy. Desalination 185: 1585--1591. Fang HHP and Shi X (2005) Pore fouling of microfiltration membranes by activated sludge. Journal of Membrane Science 264: 161--166. Fang HHP, Shi X, and Zhang T (2006) Effect of activated carbon on fouling of activated sludge filtration. Desalination 189: 193--199. Fawehinmi F, Lens P, Stephenson T, Rogalla F, and Jefferson B (2004) The influence of operating conditions on EPS, SMP and bio-fouling in anaerobic MBR. In: Proceedings of the Water Environment-Membrane Technology Conference. Seoul, Korea, 7–10 June. London: IWA Publishing. Flemming HC and Wingender J (2001) Relevance of microbial extracellular polymeric substances (EPSs). Part I. Structural and ecological aspects. Water Science and Technology 43: 1--8. Frechen FB, Schier W, and Wett M (2006) Pre-treatment of municipal MBR applications in Germany – current status and treatment efficiency. Water Practice and Technology 1(3). doi:10.2166/WPT.2006057 Friedler E and Hadari M (2006) Economic feasibility of on-site greywater reuse in multi-storey buildings. Desalination 190: 221--234. Field RW, Wu D, and Howell JA (1995) Critical flux concept for microfiltration fouling. Journal of Membrane Science 100: 259--272. Fuchs W, Braun R, and Theiss M (2005) Influence of various wastewater parameters on the fouling capacity during membrane filtration. In: Proceedings of the International Congress on Membranes and Membrane Processes (ICOM). Seoul, Korea, 7–10 June. Furumai H and Rittmann BE (1992) Advanced modeling of mixed populations of heterotrophs and nitrifiers considering the formation and exchange of soluble microbial products. Water Science and Technology 26(3–4): 493--502. Futselaar H, Schonewille H, De Ventec D, and Broensa L (2007) NORIT AirLift MBR: Side-stream system for municipal waste water treatment. Desalination 204: 1--7. Galinha CF, Carvalho G, Silva AF, et al. (2009). New developments in membrane bioreactors characterization and monitoring. In: Lesjean B and Leiknes T (eds.) Final MBR-Network Workshop, pp. 17–18. Berlin, Germany, 31 March–1 April. Gander MA, Jefferson B, and Judd SJ (2000) Membrane bioreactors for use in small wastewater treatment plants: Membrane materials and effluent quality. Water Science and Technology 41: 205--211. Gao M, Yang M, Li H, Wang Y, and Pan F (2004) Nitrification and sludge characteristics in a submerged membrane bioreactor on synthetic inorganic wastewater. Desalination 170: 177--185. Genkin G, Waite TD, Fane TG, and Chang S (2005) The effct of axial vibrations on the filtration performance of submerged hollow fibre membranes. In: Proceedings of
609
the International Congress on Membranes and Membrane Processes (ICOM). Seoul, Korea, 2005, 21–26 August. Germain E and Stephenson T (2005) Biomass characteristics, aeration and oxygen transfer in membrane bioreactors: Their interrelations explained by a review of aerobic biological processes. Reviews in Environmental Science and Biotechnology 4: 223--233. Ghaffour N, Jassim R, and Khir T (2004) Flux enhancement by using helical baffles in ultrafiltration of suspended solids. Desalination 167: 201--207. Ghosh R (2006) Enhancement of membrane permeability by gas-sparging in submerged hollow fibre ultrafiltration of macromolecular solutions: Role of module design. Journal of Membrane Science 274: 73--82. Ghyoot W and Verstraete W (1999) Reduced sludge production in a two stage membrane-assisted bioreactor. Water Research 34: 205--215. Gijsbertsen-Abrahamse AJ, Cornelissen ER, and Hofman JAMH (2006) Fiber failure frequency and causes of hollow fiber integrity loss. Desalination 194: 251--258. Gori R, Lubello C, and Caffaz S (2004) Sviluppo di un modello matematico per la simulazione di un reattore MBR applicato al trattamento di acque reflue industriali. In: Proceedings of ANDIS. Taormina, Italy, June 2004. Grethlein HE (1978) Anaerobic digestion and membrane separation of domestic wastewater. Journal of Water Pollution Control Federation 50: 754--763. Guglielmi G, Saroj DP, Chiarani D, and Andreottola G (2007) Subcritical fouling in a membrane bioreactor for municipal wastewater treatment: Experimental investigation and mathematical modelling. Water Research 41: 3903--3914. Gu¨nder B and Krauth KH (2000) Excess sludge production and Oxygen transfer in MBR. In: Proceedings of ATSV Conference. Netherlands, 8–9 February. Guo WS, Vigneswaran S, and Ngo HH (2004) A rational approach in controlling membrane fouling problems: Pretreatments to a submerged hollow fiber membrane system. In: Proceedings of the Water Environment – Membrane Technology Conference. Seoul, Korea, 7–10 June. London: IWA Publishing. Hai FI (2007) Enhancement of Dye Degradation and Mitigation of Membrane Fouling in a Membrane-Coupled Fungi Reactor Treating Textile Wastewater. PhD Thesis, University of Tokyo. Hai FI, Yamamoto K, and Fukushi K (2005) Different fouling modes of submerged hollow-fiber and flat-sheet membranes induced by high strength wastewater with concurrent biofouling. Desalination 180: 89--97. Hai FI, Yamamoto K, and Fukushi K (2006a) Development of a submerged membrane fungi reactor for textile wastewater treatment. Desalination 192: 315--322. Hai FI, Yamamoto K, and Fukushi K (2006b) Performance of newly developed hollow fiber module with spacer in integrated anaerobic–aerobic fungi reactor treating textile wastewater. Desalination 199: 305--307. Hai FI, Yamamoto K, Fukushi K, and Nakajima F (2008a) Fouling resistant compact hollow-fiber module with spacer for submerged membrane bioreactor treating high strength industrial wastewater. Journal of Membrane Science 317: 34--42. Hai FI, Yamamoto K, Nakajima F, and Fukushi K (2007) Textile effluent treatment by fungi MBR with sludge bed /GAC adsorption: Long-term performance of compact hollow-fiber module and overall treatment. In: Proceedings of the 4th International Water Association Conference on Membranes for Water and Wastewater Treatment, CD ISBN: 978–1-86194-127–5. Harrogate, UK, 15–17 May. London: IWA Publishing. Hai FI, Yamamoto K, Nakajima F, and Fukushi K (2008b) Removal of structurally different dyes in submerged membrane fungi reactor – biosorption/PACadsorption, membrane retention and biodegradation. Journal of Membrane Science 325: 395--403. Hardt FW, Clesceri LS, Nemerow NL, and Washington DR (1970) Solid separation by ultrafiltration for concentrated activated sludge. Journal of Water Pollution Control Federation 42(12): 2135--2148. He Y, Xu P, Li C, and Zhang B (2005) High-concentration food wastewater treatment by an anaerobic membrane bioreactor. Water Research 39: 4110--4118. Heijnen M, Winkler R, Vogg M, Roeder G, and Berg P (2009) Development of a novel fiber sheet membrane for MBR: The FiSh. In: Lesjean B and Leiknes T (eds.) Final MBR-Network Workshop, pp. 53–54. Berlin, Germany, 31 March–1 April. Henze M, Gujer W, Mino T, and van Loosdrecht M (eds.) (2000) Scientific and Technical Report Series, Vol. 9: Activated Sludge Models: ASM1, ASM2, ASM2d, and ASM3 London: IWA. Hernandez Rojas ME, Van Kaam R, Schetrite S, and Albasi C (2005) Role and variations of supernatant compounds in submerged membrane bioreactor fouling. Desalination 179: 95--107. Ho CC and Zydney AL (2006) Overview of fouling phenomena and modeling approaches for membrane bioreactors. Separation Science and Technology 41: 1231--1251. Ho WSW and Sirkar KK (eds.) (1992) Membrane Handbook. New York: Kluwer.
610
Membrane Biological Reactors
Holbrook RD, Higgins MJ, Murthy SN, et al. (2004) Effect of alum addition on the performance of submerged membranes for wastewater treatment. Water Environmental Research 76: 2699--2702. Hong SP, Bae TH, Tak TM, Hong S, and Randall A (2002) Fouling control in activated sludge submerged hollow fiber membrane bioreactors. Desalination 143: 219--228. Howell JA (2002) Future research and developments in the membrane field. Desalination 144: 127--131. Howell JA (2004) Future of membranes and membrane reactors in green technologies and for water reuse. Desalination 162: 1--11. Huang X and Wu J (2008) Improvement of membrane filterability of the mixed liquor in a membrane bioreactor by ozonation. Journal of Membrane Science 318: 210--216. Hughes D, Tirlapur UK, Field R, and Cui ZF (2006) In situ 3D characterization of membrane fouling by yeast suspensions using two-photon femtosecond near infrared non-linear optical imaging. Journal of Membrane Science 280: 124--133. Hughes DJ, Cui ZF, Field RW, and Tirlapur UK (2007) Membrane fouling by cell– protein mixtures: In situ characterisation using multi-photon microscopy. Biotechnology and Bioengineering 96: 1083--1091. Itonaga T, Kimura K, and Watanabe Y (2004) Influence of suspension viscosity and colloidal particles on permeability of membrane used in membrane bioreactor (MBR). Water Science and Technology 50: 301--309. Itonga T and Watanabe Y (2004) Performance of membrane bioreactor combined with pre-coagulation/sedimentation. Water Science and Technology: Water Supply 4(1): 143--149. Iversen V, Koseoglu H, Yigit NO, et al. (2009) Impacts of membrane flux enhancers on activated sludge respiration and nutrient removal in MBRs. Water Research 43: 822--830. Jang N, Ren X, Choi K, and Kim IS (2006) Comparison of membrane biofouling in nitrification and denitrification for the membrane bio-reactor (MBR). Water Science and Technology 53(6): 43–39. Jang N, Ren X, Cho J, and Kim IS (2006) Steady-state modelling of biofuling potentials with respect to the biological kinetics in the submerged membrane bioreactor. Journal of Membrane Science 284: 352--360. Janssen A, Van Agtmaal J, Van Den Broek WBP, et al. (2008) Monitoring of SUR to control and enhance the performance of dead-end ultrafiltration installations treating wwtp effluent. Desalination 231: 99--107. Jefferson B, Brookes A, Le-Clech P, and Judd SJ (2004) Methods for understanding organic fouling in MBRs. Water Science and Technology 49: 237--244. Jefferson B, Laine AL, Judd SJ, and Stephenson T (2000) Membrane bioreactors and their role in wastewater reuse. Water Science and Technology 41(1): 197--204. Jeison D and van Lier JB (2007) Cake formation and consolidation: Main factors governing the applicable flux in anaerobic submerged membrane bioreactors (AnSMBR) treating acidified wastewaters. Separation and Purification Technology 56: 71--78. Ji J, Qiu J, Wong FS, and Li Y (2008) Enhancement of filterability in MBR achieved by improvement of supernatant and floc characteristics via filter aids addition. Water Research 42: 3611--3622. Ji L and Zhou J (2006) Influence of aeration on microbial polymers and membrane fouling in submerged membrane bioreactors. Journal of Membrane Science 276: 168--177. Jiang T, Kennedy MD, Guinzbourg BF, Vanrolleghem PA, and Schippers JC (2005) Optimising the operation of a MBR pilot plant by quantitative analysis of the membrane fouling mechanism. Water Science and Technology 51: 19--25. Jiang T, Myngheer S, De Pauw DJW, et al. (2008) Modelling the production and degradation of soluble microbial products (SMP) in membrane bioreactors (MBR). Water Research 42: 4955--4964. Jimenez B and Asano T (eds.) (2008) Water Reuse: An International Survey of Current Practice, Issues and Needs. London: IWA. Jimenez J, Grelier P, Meinhold J, and Tazi-Pain A (2010) Biological modelling of MBR and impact of primary sedimentation. Desalination 250: 562--567. Jin YL, Lee WN, Lee CH, Chang IS, Huang X, and Swaminathan T (2006) Effect of DO concentration on biofilm structure and membrane filterability in submerged membrane bioreactor. Water Research 40: 2829--2836. Jinhua P, Fukushi K, and Yamamoto K (2006) Bacterial community structure on membrane surface and characteristics of strains isolated from membrane surface in submerged membrane bioreactor. Separation Science and Technology 41: 1527--1549. Judd S (2002) Submerged membrane bioreactors: Flat plate or hollow fibre? Filtration and Separation 39: 30--31. Judd S (ed.) (2006) The MBR Book: Principles & Applications of MBRs in Water & Wastewater Treatment. Oxford: Elsevier.
Judd S, Alvarez-Vazquez H, and Jefferson B (2006) The impact of intermittent aeration on the operation of air-lift tubular membrane bioreactors under sub-critical conditions. Separation Science and Technology 41: 1293--1302. Judd S and Jefferson B (eds.) (2003) Membranes for Industrial Wastewater Recovery and Re-Use. Oxford: Elsevier. Judd SJ, Robinson T, Holdner J, Alvarez-Vazquez H, and Jefferson B (2004) Impact of membrane material on membrane bioreactor permeability. In: Proceedings of the IWA – Water Environment – Membrane Technology (WEMT) Conference. Seoul, Korea, 7–10 June. London: IWA Publishing. Kang IJ, Lee CH, and Kim KJ (2003) Characteristics of microfiltration membranes in a membrane coupled sequencing batch reactor system. Water Research 37: 1192--1197. Kang S, Hoek EMV, Choi H, and Shin H (2006) Effect of membrane surface properties during the fast evaluation of cell attachment. Separation Science and Technology 41: 1475--1487. Kayawake E, Narukami Y, and Yamagata M (1991) Anaerobic digestion by a ceramic membrane enclosed reactor. Journal of Fermentation and Bioengineering 71: 122--125. Kiat WY, Yamamoto K, and Ohgaki S (1992) Optimal fiber spacing in externally pressurized hollow fiber module for solid liquid separation. Water Science and Technology 26(26): 1245--1254. Kim HY, Yeon KM, Lee CH, Lee S, and Swaminathan T (2006) Biofilm structure and extracellular polymeric substances in low and high dissolved oxygen membrane bioreactors. Separation Science and Technology 41: 1213--1230. Kim J, Jang M, Chio H, and Kim S (2004) Characteristics of membrane and module affecting membrane fouling. In: Proceedings of the IWA – Water Environment – Membrane Technology (WEMT) Conference. Seoul, Korea, 7–10 June. London: IWA Publishing. Kim JS and Lee CH (2003) Effect of powdered activated carbon on the performance of an aerobic membrane bioreactor: Comparison between crossflow and submerged membrane systems. Water Environment Research 75: 300--307. Kimura K, Yamato N, Yamamura H, and Watanabe Y (2005) Membrane fouling in pilotscale membrane bioreactors (MBRs) treating municipal wastewater. Environmental Science and Technology 39: 6293--6299. Kjelleberg S, Mcdougald D, Rasmussen TB, and Givskov M (2008) Quorum sensing inhibition. In: Winans SC and Bassler BL (eds.) Chemical Communication among Bacteria, pp. 393--416. Washington, DC: ASM Press. Komesli OK, Teschner K, Hegemann W, and Gokcay CF (2007) Vacuum membrane applications in domestic wastewater reuse. Desalination 215: 22--28. Konopka A, Zakharova T, Oliver L, Camp D, and Turco RF (1996) Biodegradation of organic wastes containing surfactants in a biomass recycle reactor. Applied and Environmental Microbiology 62: 3292--3297. Kniebusch MM, Wilderer PA, and Behling RD (1990) Immobilisation of cells on gas permeable membranes Physiology of Immobilised Cells, pp. 149--160. Amsterdam: Elservier. Kox LSDM (2004) Membrane bioreactor in Varsseveld, the Dutch approach. In: Proceedings of the IWA – Water Environment – Membrane Technology (WEMT) Conference. Seoul, Korea, 7–10 June. London: IWA Publishing. Kraume M and Bracklow U (2003) MBR in municipal wastewater treatment – operational experiences and design guidelines in Germany (in German). In: Proceedings of the 5th Aachen Membrane Conference, pp. 1–20. Aachen, Germany. Krauth KH and Staab KF (1988) Pressurized biomembrane rector for wastewater treatment. Hydrotop 94: 555--562. Kreckel B, Bruess U, Grelot A, and Tazi-Pain A (2009) Innovative European made flatsheet module system for MBR application. In: Lesjean B and Leiknes T (eds.) Final MBR-Network Workshop, pp. 55–56. Berlin, Germany, 31 March–1 April. Kuberkar VT and Davis RH (2000) Modeling of fouling reduction by secondary membranes. Journal of Membrane Science 168: 243--257. Laborie S, Cabassud C, Durand L, and Laine JM (1997) Flux enhancement by a continuous tangential gas flow in UF hollow fibres for drinking water production: Effects of slug flow on cake structure. Filtration and Separation 34(8): 887--891. Laspidou CS and Rittmann BE (2002) A unified theory for extracellular polymeric substances, soluble microbial products, and active and inert biomass. Water Research 36: 2711--2720. Lebegue J, Heran M, and Grasmick A (2008) Membrane bioreactor: Distribution of critical flux throughout an immersed HF bundle. Desalination 231: 245--252. Le-Clech P, Chen V, and Fane TAG (2006) Fouling in membrane bioreactors used in wastewater treatment. Journal of Membrane Science 284: 17--53. Le-Clech P, Jefferson B, Chang IS, and Judd SJ (2003a) Critical flux determination by the flux-step method in a submerged membrane bioreactor. Journal of Membrane Science 227: 81--93.
Membrane Biological Reactors Le-Clech P, Jefferson B, and Judd SJ (2003b) Impact of aeration, solids concentration and membrane characteristics on the hydraulic performance of a membrane bioreactor. Journal of Membrane Science 218: 117--129. Le-Clech P, Fane A, Leslie G, and Childress A (2005a) MBR focus: The operators’ perspective. Filtration and Separation 42(5): 20--23. Le-Clech P, Jefferson B, and Judd SJ (2005b) A comparison of submerged and sidestream tubular membrane bioreactor configurations. Desalination 173: 113--122. Lee J, Ahn WY, and Lee CH (2001a) Comparison of the filtration characteristics between attached and suspended growth microorganisms in submerged membrane bioreactor. Water Research 35: 2435--2445. Lee JC, Kim JS, Kang IJ, Cho MH, Park PK, and Lee CH (2001b) Potential and limitations of alum or zeolite addition to improve the performance of a submerged membrane bioreactor. Water Science and Technology 43: 59--66. Lee W, Jeon JH, Cho Y, Chung KY, and Min BR (2005) Behavior of TMP according to membrane pore size. In: Proceedings of the International Congress on Membranes and Membrane Processes (ICOM). Seoul, Korea, 21–26 August. Lee W, Kang S, and Shin H (2003) Sludge characteristics and their contribution to microfiltration in submerged membrane bioreactors. Journal of Membrane Science 216: 217--227. Lee WN, Cheong WS, Yeon KM, Hwang BK, and Lee CH (2009) Correlation between local TMP distribution and bio-cake porosity on the membrane in a submerged MBR. Journal of Membrane Science 332: 50--55. Lee WN, Kang IJ, and Lee CH (2006) Factors affecting filtration characteristics in membrane-coupled moving bed biofilm reactor. Water Research 40: 1827--1835. Lee Y, Cho J, Seo Y, Lee JW, and Ahn KH (2002) Modeling of submerged membrane bioreactor process for wastewater treatment. Desalination 146: 451--457. Leiknes T and Odegaard H (2007) The development of a biofilm membrane bioreactor. Desalination 202: 135--143. Lesage N, Sperandio M, and Cabassud C (2005) Performances of a hybrid adsorption/ submerged membrane biological process for toxic waste removal. Water Science and Technology 51: 173--180. Lesjean B and Huisjes EH (2008) Survey of the European MBR market: Trends and perspectives. Desalination 231: 71--81. Lesjean B, Leiknes T, Hochstrat R, Schories G, and Gonzalez A (2006) MBR: Technology gets timely EU cash boost. Filtration and Separation 43(9): 20--23. Lesjean B, Rosenberger S, Laabs C, Jekel M, Gnirss R, and Amy G (2005) Correlation between membrane fouling and soluble/colloidal organic substances in membrane bioreactors for municipal wastewater treatment. Water Science and Technology 51: 1--8. Lesjean B, Rosenberger S, Schrotter JC, and Recherche A (2004) Membrane-aided biological wastewater treatment – an overview of applied systems. Membrane Technology August 2004: 5–10. Leslie G and Chapman S (2003) Membrane bioreactors for municipal wastewater treatment – an Australian perspective. In: Proceedings of Australian Water Association Annual Conference, Perth, Australia, 6–10 April. Li H, Yang M, Zhang Y, Liu X, Gao M, and Kamagata Y (2005a) Comparison of nitrification performance and microbial community between submerged membrane bioreactor and conventional activated sludge system. Water Science and Technology 51: 193--200. Li X, Gao F, Hua Z, Du G, and Chen J (2005b) Treatment of synthetic wastewater by a novel MBR with granular sludge developed for controlling membrane fouling. Separation and Purification Technology 46: 19--25. Li YZ, He YL, Liu YH, Yang SC, and Zhang GJ (2005c) Comparison of the filtration characteristics between biological powdered activated carbon sludge and activated sludge in submerged membrane bioreactors. Desalination 174: 305--314. Li XY and Wang XM (2006) Modelling of membrane fouling in a submerged membrane bioreactor. Journal of Membrane Science 278: 151--161. Liao BQ, Bagley DM, Kraemer HE, Leppard GG, and Liss SN (2004) A review of biofouling and its control in membrane separation bioreactors. Water Environment Research 76: 425--436. Liao BQ, Kraemer JT, and Bagley D (2006) Anaerobic membrane bioreactors: Applications and research directions. Critical Reviews in Environmental Science and Technology 36(6): 489--530. Lim AL and Bai R (2003) Membrane fouling and cleaning in microfiltration of activated sludge wastewater. Journal of Membrane Science 216: 279--290. Lim BR, Ahn KH, Song KG, and Jin-Woo C (2005) Microbial community in biofilm on membrane surface of submerged MBR: Effect of in-line cleaning chemical agent. Water Science and Technology 51(6–7): 201–207. Lipnizki F and Field RW (2001) Pervaporation-based hybrid processes in treating phenolic wastewater: Technical aspects and cost engineering. Separation Science and Technology 36: 3311--3335.
611
Lipski C and Cote P (1990) The use of pervaporation for the removal of organic contaminants from water. Environmental Progress 9: 254--261. Liu R, Huang X, Sun YF, and Qian Y (2003) Hydrodynamic effect on sludge accumulation over membrane surfaces in a submerged membrane bioreactor. Process Biochemistry 39: 157--163. Liu R, Huang X, Chen L, Wen X, and Qian Y (2005) Operational performance of a submerged membrane bioreactor for reclamation of bath wastewater. Process Biochemistry 40: 125--130. Liu Y and Fang HHP (2003) Influences of extracellular polymeric substances (EPS) on flocculation, settling, and dewatering of activated sludge. Critical Reviews in Environmental Science and Technology 33: 237--273. Liu Y and Tay JH (2001) Strategy for minimization of excess sludge production from the activated sludge process. Biotechnology Advances 19: 97--107. Livingston AG (1993) A novel membrane bioreactor for detoxifying industrial wastewaters: Biodegradation of 3-chloronitrobenzene in an industrially produced wastewater. Biotechnology and Bioengineering 41: 927--936. Livingston AG (1994) Extractive membrane bioreactors: A new process technology for detoxifying chemical industry wastewaters. Journal of Chemical Technology and Biotechnology 60: 117--124. Livingston AG, Arcangeli JP, Boam AT, Zhang SF, Marangon M, and dos Santos LMF (1998) Extractive membrane bioreactors for detoxification of chemical industry wastes: Process development. Journal of Membrane Science 151: 29--44. Lu SG, Imai T, Ukita M, Sekine M, Fukagawa M, and Nakanishi H (1999) Fermentation wastewater treatment in a membrane bioreactor. Environmental Technology 20: 431--436. Lu SG, Imai T, Ukita M, Sekine M, Higuchi T, and Fukagawa M (2001) A model for membrane bioreactor process based on the concept of formation and degradation of soluble microbial products. Water Research 35: 2038--2048. Lu SG, Imai T, Ukita M, Sekine M, and Higuchi T (2002) Modeling prediction of membrane bioreactor process with the concept of soluble microbial product. Water Science and Technology 46(11–12): 63--69. Lubbecke S, Vogelpohl A, and Dewjanin W (1995) Wastewater treatment in a biological high-performance system with high biomass concentration. Water Research 29: 793--802. Luonsi A, Laitinen AN, Beyer K, Levanen E, Poussade Y, and Nystrom M (2002) Separation of CTMP mill-activated sludge with ceramic membranes. Desalination 146: 399--404. Lyko S, Al-Halbouni D, Wintgens T, et al. (2007) Polymeric compounds in activated sludge supernatant – characterisation and retention mechanisms at a full-scale municipal membrane bioreactor. Water Research 41: 3894--3902. Madaeni SS, Fane AG, and Wiley D (1999) Factors influencing critical flux in membrane filtration of activated sludge. Journal of Chemical Technology and Biotechnology 74: 539--543. Mansell B, Kuo J, Tang CC, et al. (2004) Comparison of two membrane bioreactors and an activated sludge plant with dual-media filtration: Nutrient and priority pollutants removal. In: Proceedings WEFTEC 04. New Orleans, USA, 2–6 October. McInnis AB (2005) Cost comparison of MBR and CAS/MF facilities. In: Proceedings WEFTEC 05. Washington, DC, USA, 29 October–2 November. Mehrez R, Ernst M, and Jekel M (2007) Development of a continuous protein and polysaccharide measurement method by sequential injection analysis for the application in membrane bioreactor systems. Water Science and Technology 56: 163--171. Melin T, Jefferson B, Bixio D, et al. (2006) Membrane bioreactor technology for wastewater treatment and reuse. Desalination 187: 271--282. Meng F, Chae SR, Drews A, Kraume M, Shin HS, and Yang F (2009) Recent advances in membrane bioreactors (MBRs): Membrane fouling and membrane material. Water Research 43: 1489--1512. Meng F, Zhang H, Li Y, Zhang X, and Yang F (2005) Application of fractal permeation model to investigate membrane fouling in membrane bioreactor. Journal of Membrane Science 262: 107--116. Meng F, Zhang H, Yang F, Li Y, Xiao J, and Zhang X (2006) Effect of filamentous bacteria on membrane fouling in submerged membrane bioreactor. Journal of Membrane Science 272: 161--168. Metcalf and Eddy, Inc (2003) Wastewater Engineering – Treatment and Reuse, 4th edn. New York: McGraw-Hill. Miyashita S, Honjyo K, Kato O, et al. (2000) Gas Diffuser for Aeration Vessel of Membrane Assembly. US Pat. 6,328,886. Miura Y, Watanabe Y, and Okabe S (2007) Membrane biofouling in pilot-scale membrane bioreactors (MBRs) treating municipal wastewater: Impact of biofilm formation. Environmental Science and Technology 41: 632--638. Mo H, Oleszkiewicz JA, Cicek N, and Rezania B (2005) Incorporating membrane gas diffusion into a membrane bioreactor for hydrogenotrophic denitrification of groundwater. Water Science and Technology 51(6–7): 357–364.
612
Membrane Biological Reactors
Modin O, Fukushi K, Nakajima F, and Yamamoto K (2008) Performance of a membrane biofilm reactor for denitrification with methane. Bioresource Technology 99: 8054--8060. Muller EB, Stouthamer AB, Verseveld HW, and Eikelboom EH (1995) Aerobic domestic waste water treatment in a pilot plant with complete sludge retention by cross-flow filtration. Water Research 29: 1179--1189. Nagaoka H and Nemoto H (2005) Influence of extracellular polymeric substances on nitrogen removal in an intermittently-aerated membrane bioreactor. Water Science and Technology 51: 151--158. Nagaoka H, Ueda S, and Miya A (1996) Influence of bacterial extracellular polymers on the membrane separation activated sludge process. Water Science and Technology 34: 165--172. Nagaoka H, Yamanishi S, and Miya A (1998) Modeling of biofouling by extracellular polymers in a membrane separation activated sludge system. Water Science and Technology 38: 497--504. Namkung E (2008) Current status and future of membrane process in Korea. In: Proceedings 11th RECWET Symposium Entitled ‘‘Water Environment Control by Membrane Treatment Technology’’. Tokyo, Japan, 2 December. Ng CA, Sun D, and Fane AG (2006) Operation of membrane bioreactor with powdered activated carbon addition. Separation Science and Technology 41: 1447--1466. Ng CA, Sun D, Zhang J, et al. (2005) Strategies to improve the sustainable operation of membrane bioreactors. In: Proceedings of the International Desalination Association World Congress. Singapore, 11–16 September. Ng ANL and Kim AS (2007) A mini-review of modeling studies on membrane bioreactor (MBR) treatment for municipal wastewaters. Desalination 212: 261--281. Ngo HH, Guo W, and Xing W (2008) Evaluation of a novel sponge-submerged membrane bioreactor (SSMBR) for sustainable water reclamation. Bioresource Technology 99: 2429--2435. Nguyen TQ and Nobe K (1987) Extraction of organic contaminants in aqueous solution by pervaporation. Journal of Membrane Science 30: 11--22. Nielson PH and Jahn A (1999) Extraction of EPS. In: Wingender J, Neu TR, and Flemming HCE (eds.) Microbial Extracellular Polymeric Substances, pp. 49--72. Berlin: Springer. Nuengjamnong C, Kweon JH, Cho J, Polprasert C, and Ahn KH (2005) Membrane fouling caused by extracellular polymeric substances during microfiltration processes. Desalination 179: 117--124. Nuhoglu A, Pekdemir T, Yielder E, Keskinlerand B, and Akay G (2002) Drinking water denitrification by a membrane bio-reactor. Water Research 36: 1155--1166. Ognier S, Wisniewski C, and Grasmick A (2004) Membrane bioreactor fouling in subcritical filtration conditions: A local critical flux concept. Journal of Membrane Science 229: 171--177. Orantes JC, Wisniewski C, Heran M, and Grasmick A (2004) Influence of total sludge retension on the performance of a submerge membrane bioreactor. In: Proceedings of the IWA – Water Environment Membrane Technology (WEMT) Specialty Conference. Seoul, South Korea, 7–10 June. London: IWA Publishing. Pang CM, Hong P, Guo H, and Liu WT (2005) Biofilm formation characteristics of bacterial isolates retrieved from a reverse osmosis membrane. Environmental Science and Technology 39: 7541--7550. Pankania M, Stephenson M, and Semmens MJ (1994) Hollow fiber bioreactor for wastewater treatment using bubbleless membrane aeration. Water Research 10: 2233--2236. Parameshwaran K and Visvanathan C (1998) Recent developments in membrane technology for wastewater reuse. In: Proceedings of the 6th International Conference on Advanced Wastewater Treatment, Recycling and Reuse. Milan, Italy, 14–16 September. Park S, Yeon KM, and Lee CH (2005) Hydrodynamics and microbial physiology affecting performance of a new MBR, membrane-coupled high performance compact reactor. Desalination 172: 181--188. Phattaranawik J, Fane AG, Pasquier ACS, and Bing W (2008) A novel membrane bioreactor based on membrane distillation. Desalination 223: 386--395. Phattaranawik J, Fane AG, Pasquier ACS, Bing W, and Wong FS (2009) Experimental study and design of a submerged membrane distillation bioreactor. Chemical Engineering and Technology 32: 38--44. Pearce G (2007) Introduction to membranes: Filtration for water and wastewater treatment. Filtration and Separation 44(2): 24--27. Pearce G (2008a) Introduction to membranes: An introduction to membrane bioreactors. Filtration and Separation 45(1): 32--35. Pearce G (2008b) Introduction to membranes – MBRs: Manufacturers’ comparison: Part 1. Filtration and Separation 45(3): 28--31. Pearce G (2008c) Introduction to membranes – MBRs: Manufacturers’ comparison: Part 2. Filtration and Separation 45(3): 30--32. Pearce G (2008d) Introduction to membranes – MBRs: Manufacturers’ comparison: Part 3. Filtration and Separation 45(4): 23--25.
Pillay S, Foxon KM, and Buckley CA (2008) An anaerobic baffled reactor/membrane bioreactor (ABR/MBR) for on-site sanitation in low income areas. Desalination 231: 91--98. Pollice A, Brookes A, Jefferson B, and Judd S (2005) Sub-critical flux fouling in membrane bioreactors – a review of recent literature. Desalination 174: 221--230. Psoch C and Schiewer S (2005a) Critical flux aspect of air sparging and backflushing on membrane bioreactors. Desalination 175: 61--71. Psoch C and Schiewer S (2005b) Long-term study of an intermittent air sparged MBR for synthetic wastewater treatment. Journal of Membrane Science 260: 56--65. Psoch C and Schiewer S (2006) Resistance analysis for enhanced wastewater membrane filtration. Journal of Membrane Science 280: 284--297. Qin JJ, Kekre KA, Tao G, et al. (2006) New option of MBR-RO process for production of NEWater from domestic sewage. Journal of Membrane Science 272: 70--77. Rabie HR, Cote P, Singh M, and Janson A (2003) Cyclic Aeration System for Submerged Membrane Modules. US Pat. 684,406. Ramesh A, Lee DJ, and Lai JY (2007) Membrane biofouling by extracellular polymeric substances or soluble mcirobial products from membrane bioreactor sludge. Applied Microbiology and Biotechnology 74: 699--707. Reij MW, Keurentjes JTF, and Hartmans S (1998) Membrane bioreactors for waste gas treatment. Journal of Biotechnology 59: 155--167. Rezania B, Oleszkiewicz JA, Cicek N, and Mo H (2005) Hydrogen dependent denitrification in an alternating anoxic–aerobic SBR membrane bioreactor. Water Science and Technology 51(6–7): 403–409. Rittmann BE (1987) Aerobic biological treatment. Environmental Science and Technology 21: 128--136. Rosenberger S, Evenblij H, Te Poele S, Wintgens T, and Laabs C (2005) The importance of liquid phase analyses to understand fouling in membrane assisted activated sludge processes-six case studies of different European research groups. Journal of Membrane Science 263: 113--126. Rosenberger S and Kraume M (2002) Filterability of activated sludge in membrane bioreactors. Desalination 146: 373--379. Rosenberger S, Kruger K, Witzig R, Szewzyk U, and Kraume M (2002) Performance of a bioreactor with submerged membranes for aerobic treatment of municipal waste water. Water Research 36: 413--420. Rosenberger S, Laabs C, Lesjean B, et al. (2006) Impact of colloidal and soluble organic material on membrane performance in membrane bioreactors for municipal wastewater treatment. Water Research 40: 710--720. Ross WR, Barnard JP, Strohwald NK, Grobler CJ, and Sanetra J (1992) Practical application of the ADUF process to the full-scale treatment of maize-processing effluent. Water Science and Technology 25(10): 27--39. Rothemund C, Camper A, and Wilderer PA (1994) Biofilms growing on gas permeable membranes. Water Science and Technology 29(10–11): 447--454. Saroj DP, Guglielmi G, Chiarani D, and Andreottola G (2008) Subcritical fouling behaviour modelling of membrane bioreactors for municipal wastewater treatment: The prediction of the time to reach critical operating condition. Desalination 231: 175--181. Schier W, Frechen FB, and Fischer S (2009) Efficiency of mechanical pre-treatment on European MBR plants. Desalination 236: 85--93. Schoeberl P, Brik M, Bertoni M, Braun R, and Fuchs W (2005) Optimization of operational parameters for a submerged membrane bioreactor treating dye house wastewater. Separation and Purification Technology 44: 61--68. Schrader GA, Zwijnenburg A, and Wessling M (2005) The effect of WWTP effluent zeta-potential on direct nanofiltration performance. Journal of Membrane Science 266: 80--93. Scott JA, Neilson DJ, Liu W, and Boon PN (1998) A dual function membrane bioreactor system for enhanced aerobic remediation of high-strength industrial waste. Water Science and Technology 38: 413--420. Semmens MJ (1991) Bubbleless aeration. Water Engineering and Management 138(4): 18--19. Semmens MJ and Gantzer CJ (1993) Gas transfer using hollow fiber membranes. In: Proceedings of the 66th Annual Conference and Exposition of the Water Environment Federation, pp. 365–406. Anaheim, CA, USA, 3–7 October. Smith CV, DiGregorio D, and Talcott RM (1969) The use of ultrafiltration membranes for activated sludge separation. In: Proceedings of the 24th Annual Purdue Industrial Waste Conference. Lafayette, IN, USA, 6–8 May. Smith PJ, Vigneswaran S, Ngo HH, Ben-Aim R, and Nguyen H (2005) Design of a generic control system for optimising back flush durations in a submerged membrane hybrid reactor. Journal of Membrane Science 255: 99--106. Sridang PC, Heran M, and Grasmick A (2005) Influence of module configuration and hydrodynamics in water clarification by immersed membrane systems. Water Science and Technology 51(6–7): 135--142. Stephenson T, Judd S, Jeferson B, and Brindle K (2000) Membrane Bioreactors for Wastewater Treatment. London: IWA.
Membrane Biological Reactors Sui P, Wen X, and Huang X (2008) Feasibility of employing ultrasound for on-line membrane fouling control in an anaerobic membrane bioreactor. Desalination 219: 203--213. Sun Y, Huang X, Chen F, and Wen X (2004) A dual functional filtration/aeration membrane bioreactor for domestic wastewater treatment. In: Proceedings of the IWA – Water Environment – Membrane Technology (WEMT) Conference. Seoul, Korea, 7–10 June. London: IWA Publishing. Sutton PM (2003) Membrane bioreactors for industrial wastewater treatment: The state-of-the-art based on full scale commercial applications. In: Proceedings of WEF 76th Annual Conference and Exposition. Los Angeles, CA, USA, 12–15 October. Sutton PM, Li RR, and Korchin SR (1983) Dorr-oliver’s fixed film suspended growth anaerobic systems for industrial wastewater treatment and energy recovery. In: Proceedings of the 37th Annual Purdue Industrial Waste Conference, pp. 667–675. Lafayette, IN, USA, 11–13 May. Tao G, Kekre K, Wei Z, Lee TC, Viswanath B, and Seah H (2005) Membrane bioreactors for water reclamation. Water Science and Technology 51: 431--440. Tadkaew N, Sivakumar M, and Nghiem LD (2007) Membrane bioreactor technology for decentralised wastewater treatment and reuse. International Journal of Water 3(4): 368--380. Tarnacki K, Lyko S, Wintgens T, Melin T, and Natau F (2005) Impact of extracellular polymeric substances on the filterability of activated sludge in membrane bioreactors for landfill leachate treatment. Desalination 179: 181--190. Tchobanoglous G, Burton FL, and Stensel HD (eds.) (2003) Wastewater Engineering: Treatment and Reuse. New York: McGraw-Hill Higher Education. Teychene B, Guigui C, Cabassud C, and Amy G (2008) Toward a better identification of foulant species in MBR processes. Desalination 231: 27--34. Thanh BX, Visvanathan C, Sperandio M, and Ben Aim R (2008) Fouling characterization in aerobic granulation coupled baffled membrane separation unit. Journal of Membrane Science 318: 334--339. Ueda T and Hata K (1999) Domestic wastewater treatment by a submerged membrane bioreactor with gravitational filtration. Water Research 33: 2888--2892. Ueda T, Hata K, Kikuoka Y, and Seino O (1997) Effects of aeration on suction pressure in a submerged membrane bioreactor. Water Research 31: 489--494. Urbain V, Mobarry B, de Silva V, Stahl DA, Rittmann BE, and Manem J (1998) Integration of performance, molecular biology and modeling to describe the activated sludge process. Water Science and Technology 37(4–5): 223--229. Vallero MVG, Lettinga G, and Lens PNL (2005) High rate sulfate reduction in a submerged anaerobic membrane bioreactor (SAMBaR) at high salinity. Journal of Membrane Science 253: 217--232. Van Der Roest HF, Lawrence DP, and Van Bentem AG (2002a) Membrane Bioreactors for Municipal Wastewater Treatment (Water and Wastewater Practitioner Series: Stowa Report). London: IWA. Van Der Roest HF, Van Bentem AGN, and Lawrence DP (2002b) MBR-technology in municipal wastewater treatment: Challenging the traditional treatment technologies. Water Science and Technology 46: 273--280. Van’T Oever R (2005) MBR focus: Is submerged best? Filtration and Separation 42(5): 24--27. Verrecht B, Judd S, Guglielmi G, Brepols C, and Mulder JW (2008) An aeration energy model for an immersed membrane bioreactor. Water Research 42: 4761--4770. Vilim D, Hlavinek P, Kubik J, and Hlustik P (2009) Development of containerized turnkey MBR plants. In: Lesjean B and Leiknes T (eds.) Final MBR-Network Workshop, pp. 57–58. Berlin, Germany, 31 March–1 April. Visvanathan C, Byung-Soo Y, Muttamara S, and Maythanukhraw R (1997) Application of air backflushing technique in membrane bioreactor. Water Science and Technology 36(12): 259--266. Visvanathan C, Ben Aim R, and Parameshwaran K (2000) Membrane separation bioreactors for wastewater treatment. Critical Reviews in Environmental Science and Technology 30: 1--48. Wagner J and Rosenwinkel KH (2000) Sludge production in membrane bioreactors under different conditions. Water Science and Technology 41(10–11): 251--258. Wang S, Guillen G, and Hoek EMV (2005) Direct observation of microbial adhesion to membranes. Environmental Science and Technology 39(17): 6461--6469. Wang Z, Wu Z, Maib S, et al. (2008a) Research and applications of membrane bioreactors in China: Progress and prospect. Separation and Purification Technology 62: 249--263. Wang Z, Wu Z, Yin X, and Tian L (2008b) Membrane fouling in a submerged membrane bioreactor (MBR) under sub-critical flux operation: Membrane foulant and gel layer characterization. Journal of Membrane Science 325: 238--244. Wen X, Sui P, and Huang X (2008) Exerting ultrasound to control the membrane fouling in filtration of anaerobic activated sludge – mechanism and membrane damage. Water Science and Technology 57(5): 773--779.
613
Wen X-H, Xing C-H, and Qian Y (1999) A kinetic model for the prediction of sludge formation in a membrane bioreactor. Process Biochemistry 35: 249--254. WERF (2001) Membrane bioreactors: Feasibility and use in water reclamation. Report of the Water Environment Research Foundation. Alexandria, VA: WERF. Wicaksana F, Fane AG, and Chen V (2006) Fibre movement induced by bubbling using submerged hollow fibre membranes. Journal of Membrane Science 271: 186--195. Wilderer PA, Brautigam J, and Sekoulov I (1985) Application of gas permeable membranes for auxiliary oxygenation of sequencing batch reactors. Conservation and Recycling 8(1–2): 181--192. Wisniewski C and Grasmick A (1998) Floc size distribution in a membrane bioreactor and consequences for membrane fouling. Colloids and Surfaces A: Physicochemical and Engineering Aspects 138: 403--411. Wintgens T, Melin T, Scha¨fer A, et al. (2005) The role of membrane processes in municipal wastewater reclamation and reuse. Desalination 178: 1--11. Wintgens T, Rosen J, Melin T, Brepols C, Drensla K, and Engelhardt N (2003) Modelling of a membrane bioreactor system for municipal wastewater treatment. Journal of Membrane Science 216: 55--65. Witzig R, Manz W, Rosenberger S, Kru¨ger U, Kraume M, and Szewzyk U (2002) Microbiological aspects of a bioreactor with submerged membranes for aerobic treatment of municipal wastewater. Water Research 36: 394--402. Wu Y, Huang X, Wen X, and Chen F (2004) Function of dynamic membrane in selfforming dynamic membrane coupled bioreactor. In: Proceedings of the IWA – Water Environment – Membrane Technology (WEMT) Conference. Seoul, Korea, 7–10 June. Xing CH, Qian Y, Wen XH, Wu WZ, and Sun D (2001) Physical and biological characteristics of a tangential-flow MBR for municipal wastewater treatment. Journal of Membrane Science 191: 31--42. Xing CH, Tardieu E, Qian Y, and Wen XH (2000) Ultrafiltration membrane bioreactor for urban wastewater reclamation. Journal of Membrane Science 177: 73--82. Xing CH, Yamamoto K, and Fukushi K (2006) Performance of an inclined-plate membrane bioreactor at zero excess sludge discharge. Journal of Membrane Science 275: 175--186. Xu N, Xing WH, Xu NP, and Shi J (2003) Study on ceramic membrane bioreactor with turbulence promoter. Separation and Purification Technology 32: 403--410. Yamagiwa K, Ohmae Y, Dahlan MH, and Ohkawa A (1991) Activated sludge treatment of small-scale wastewater by a plunging liquid jet bioreactor with cross-flow filtration. Bioresource Technology 37: 215--222. Yamamoto K (2009) Submerged MBR technology: An unfinished international adventure of 20 years. In: Lesjean B and Leiknes T (eds.) Final MBR-Network Workshop, pp. 83–84. Berlin, Germany, 31 March–1 April. Yamamoto K, Hiasa H, Talat M, and Matsuo T (1989) Direct solid liquid separation using hollow fiber membranes in activated sludge aeration tank. Water Science and Technology 21: 43--54. Yamamura H, Kimura K, and Watanbe Y (2007) Mechanism involved in the evolution of physically irreversible fouling in microfiltration and ultrafiltration membranes used for drinking water treatment. Environmental Science and Technology 41(19): 6789--6794. Yamato N, Kimura K, Miyoshi T, and Watanabe Y (2006) Difference in membrane fouling in membrane bioreactors (MBRs) caused by membrane polymer materials. Journal of Membrane Science 280: 911--919. Yang Q, Chen J, and Zhang F (2006) Membrane fouling control in a submerged membrane bioreactor with porous, flexible suspended carriers. Desalination 189: 292--302. Yang Z, Peng XF, Chen MY, Lee DJ, and Lai JY (2007) Intralayer flow in fouling layer on membranes. Journal of Membrane Science 287: 280--286. Yeh SJ and Jenkins CR (1978) Pure oxygen fixed film reactor. Journal of Environmental Engineering (ASCE) 14: 611--623. Yeo A and Fane AG (2005) Performance of individual fibers in a submerged hollow fiber bundle. Water Science and Technology 51(6–7): 165--172. Yeom IT, Nah YM, and Ahn KH (1999) Treatment of household wastewater using an intermittently aerated membrane bioreactor. Desalination 124: 193--203. Yeon KM, Cheong WS, Oh SH, et al. (2009) Quorum sensing: A new biofouling control paradigm in membrane bioreactor for advanced wastewater treatment. Environmental Science and Technology 43: 380--385. Yeon KM, Park JS, Lee CH, and Kim SM (2005) Membrane coupled high performance compact reactor: A new MBR system for advanced wastewater treatment. Water Research 39: 1954--1961. Yoon SH, Collins JH, Musale D, et al. (2005) Effects of flux enhancing polymer on the characteristics of sludge in membrane bioreactor process. Water Science and Technology 51: 151--157. You HS, Huang CP, Pan JR, and Chang SC (2006) Behavior of membrane scaling during crossflow filtration in the anaerobic MBR system. Separation Science and Technology 41: 1265--1278.
4.17 Anaerobic Processes DJ Batstone and PD Jensen, The University of Queensland, Brisbane, QLD, Australia & 2011 Elsevier B.V. All rights reserved.
4.17.1 4.17.1.1 4.17.1.1.1 4.17.1.1.2 4.17.1.1.3 4.17.1.1.4 4.17.1.2 4.17.1.3 4.17.1.4 4.17.1.5 4.17.2 4.17.2.1 4.17.2.1.1 4.17.2.1.2 4.17.2.1.3 4.17.2.1.4 4.17.2.1.5 4.17.2.2 4.17.2.2.1 4.17.2.2.2 4.17.3 4.17.3.1 4.17.3.2 4.17.3.2.1 4.17.3.2.2 4.17.3.3 4.17.3.3.1 4.17.3.3.2 4.17.3.4 4.17.4 4.17.4.1 4.17.4.2 4.17.4.3 References
Anaerobic Process Fundamentals Anaerobic Conversion Processes Hydrolysis Fermentation/acidogenesis Acetogenesis and methanogenesis from hydrogen Aceticlastic methanogenesis Physicochemical Processes and pH Temperature Inhibition and Toxicity Rate-Limiting Steps Selection and Design of Anaerobic Technology Anaerobic Digester Technologies High-rate anaerobic digestion Anaerobic ponds Fully mixed liquid digester Plug-flow liquid digesters Solid phase (leach bed) Digester Selection and Design for Specific Applications Domestic and industrial wastewater Sewage solids and activated sludge biosolids Interpretation and Operation of Anaerobic Systems Evaluating and Determining Controlling Mechanisms Performance and Process Indicators High-rate anaerobic reactors Sludge digesters Evaluating Substrate and Microbial Properties Activity testing Biological methane potential testing Advanced Model-Based Analysis Future Applications of Anaerobic Digestion Sewage Treatment and Nutrient Removal Nutrient Recovery Future Applications in Energy Generation and Transport
4.17.1 Anaerobic Process Fundamentals Anaerobic digestion is the biological conversion by a complex microbial ecosystem of organic and occasionally inorganic substrates in the absence of an oxygen source. During the process, organic material is converted mainly to methane, carbon dioxide, and biomass. Nitrogen released from converted organics is in the form of ammonia. Anaerobic processes for wastewater treatment have advantages over aerobic treatment in that there are no power requirements for air supply, production of sludges requiring treatment and disposal is much lower, and the methane production can be used for energy production. Aerobic processes are catabolically more favorable, yielding approximately 10 times the energy, with a correspondingly higher microbial yield (Madigan et al., 2009). For this reason, yields used for mixed heterotrophic processes are of the order of
615 615 616 618 619 620 621 623 624 625 626 626 626 626 627 627 627 627 627 628 631 631 632 632 632 633 633 634 635 636 636 636 637 637
0.63 gCODX gCODS1 (Henze et al., 2000) as compared to 0.05–0.1 gCODX gCODS1 for anaerobic processes. COD is the chemical oxygen demand and is a measure of organics. In this case, gCODX represents the biomass generated (in grams COD), while gCODS represents the substrate consumed (Batstone et al., 2002). This lower microbial yield results in decreased operating costs. The lower yield generally implies that extended solid-retention times are required to avoid washout of active biomass. This can be done either in parallel with an increased liquid retention time, or by separation of liquid and solid-retention times. Operation, design, and interpretation of engineered anaerobic processes have greatly advanced over the last 20 years. This improvement is based on a very good understanding of underlying concepts, which has allowed implementation of technology such that it will stably and reliably operate without intervention. The process itself has (1) multiple microbial
615
616
Anaerobic Processes
steps, mediated by different organisms; (2) different steps that can be rate limiting under specific conditions; (3) interaction with the physicochemical system, particularly weak acid and base inhibition of microbial processes, and (4) highly nonlinear behavior, particularly with respect to pH regulation and inhibition. Therefore, application of anaerobic technology needs careful thought, especially to achieve an optimally engineered process for a specific application. Fortunately, understanding of the underlying microbial and chemical processes is very good, both in the scientific and in engineering sectors. Good understanding of fundamentals, as outlined in this section, has allowed the use of anaerobic technologies in a wide variety of applications, as outlined in Section 4.17.2.
The different microbial groups mediating each step have been well characterized, and are from phylogenetically defined regions. As examples, all methanogenic organisms discovered so far are archaea, while acidogens and acetogens are largely bacteria. Aceticlastic methanogens belong to one of the two specific genera: Methanosaeta or Methanosarcina. As shown in Figure 1, under different conditions, different steps can be rate limiting. Specifically, for particulate or slowly degradable materials, hydrolysis is rate limiting. Under conditions of stress, or where the primary substrate is rapidly degradable, aceticlastic methanogenesis is normally rate limiting. The first condition normally results in decreased performance as undegraded substrate is washed out, while the second condition results in elevated, effluent organic-acid concentrations.
4.17.1.1 Anaerobic Conversion Processes 4.17.1.1.1 Hydrolysis Anaerobic digestion proceeds through a series of parallel and sequential processes by a variety of consortia as represented in Figure 1 (Batstone et al., 2002; Pavlostathis and GiraldoGomez, 1991). In contrast to aerobic digestion, where oxygen is an external electron acceptor, gaseous and dissolved products (largely methane and carbon dioxide) have the same combined carbon-oxidation state as the primary substrates. Thus, anaerobic digestion is largely constrained by the need to find appropriate internal electron acceptors. When this is impossible, hydrogen ions or bicarbonate must be used as electron acceptors via anaerobic oxidation to produce hydrogen or formate. This introduces thermodynamic constraints that bring in obligate syntrophic relationships between the electron producer and the methanogenic electron consumer (Schink, 1997). It is conceptually correct and convenient to group complex organics into carbohydrates, proteins, and lipids, and their soluble analogs of sugars, amino acids, and long-chain fatty acids (LCFAs). Any mixed organic stream can be represented by these components, while preserving full information of mass, energy density (or COD), and nitrogen content (Nopens et al., 2009). Anaerobic digestion processes consist of four main steps:
• •
•
•
Hydrolysis is an enzyme-mediated extracellular step which solubilizes particulates and substrates that cannot be directly utilized by the anaerobic organisms. Acidogenesis or fermentation is the conversion of soluble substrates such as amino acids and sugars, which can be converted largely without an external electron acceptor. The products are largely organic acids and alcohols. Syntrophic acetogenesis is the degradation of fermentation products to acetate using hydrogen ions or bicarbonate as an external electron acceptor. This process is coupled with hydrogen or formate utilizing methanogenesis, which maintains a low hydrogen or formate concentration. Acetoclastic methanogenesis is the cleavage of acetate to methane and carbon dioxide.
Processes such as homoacetogenesis (conversion of hydrogen and carbon dioxide to acetate), and its reverse, acetate oxidation to hydrogen and carbon dioxide, have not been included in Figure 1, but can be important in specific circumstances as outlined further in this chapter.
While the formal definition of hydrolysis is much stricter, as a digestion component, hydrolysis is a term that is used to refer to solubilization of complex particulate materials. The material can be regarded either as a mixture of the basic components (carbohydrates, proteins, and fats), or as a composite compound (e.g., homogeneous material such as activated sludge and yeast). Separate classification and analysis of composite material as a separate input was proposed in the International Water Association (IWA) Anaerobic Digestion Model No. 1 (Batstone et al., 2002), but this was found to be cumbersome, especially when representing both composites and primary aggregates (e.g., waste-activated sludge (WAS) and primary sludges), and the current trend is to represent all feed materials as a combination of carbohydrates, proteins, and fats (Nopens et al., 2009). There are three main pathways for enzymatic hydrolysis. 1. The organisms excrete enzymes into the bulk liquid where it adsorbs onto a particle or reacts with a soluble substrate (Jain et al., 1992). 2. The organism attaches to the particle and secretes enzymes into the vicinity of the particle. The organism benefits from the soluble substrates being released (Vavilin et al., 1996). 3. The organism has an attached enzyme which may double up as a transport receptor to the interior of the cell (Tong and McCarty, 1991). This method requires the organism to adsorb onto the surface of the particle. The actual mechanism used depends heavily on the nature of the material, reactor hydraulics, and solid concentration, but forms 1 and 2 in the list are variations on the same mechanism, and are the principal forms considered here. Steps in extracellular enzymatic hydrolysis include (Figure 2): 1.
2.
4.
Production of enzyme – production rate can decrease when there is excessive soluble substrate available (Ramsay, 1997). Steps 2, 3, and 6 are transport processes, which can be limited due to large particles, or in solid-phase systems due to inadequate carrier liquid. Adsorption processes that are limited by surface area.
Anaerobic Processes 5. 7.
•
Reaction rates that are limited by surface area and enzyme concentrations. Deactivation can be excessive when away from optimal temperature and pH.
•
While there have been complex models that include all of these functions (e.g., Humphrey, 1979), in practice, it is very difficult to properly validate these models, and the most commonly used model is the first-order one. The use of first-order models has been justified as ‘‘an empirical expression that reflects the cumulative effect of all the microscopic processes occurringy’’ (Eastman and Ferguson, 1981). First order (or slightly more complex) has also been found to be just as effective as more complex models (Vavilin et al., 1996). Hydrolysis commonly becomes rate limiting when
•
In a continuous mixed digester, without retained solids, hydraulic-loading rate becomes too high (there is not enough time to hydrolyze the solids). Mass-loading rate is generally not an issue, and higher concentrations allow higher loading rates. Normally, a minimum of 9 days of hydraulic-retention time is required for any significant degradation (see Section 4.17.2.2.2). Mixed carbohydrate feeds are among the slowest to degrade.
In a batch system, there is insufficient batch time. Batch digesters have a higher volumetric efficiency, due to kinetic considerations. In a plug-flow system, there is insufficient reactor volume. Plug-flow digesters are highly efficient on a volumetric basis. Time of contact with the active biomass can also be an issue if the system is not effectively mixed at the inlet.
Particularly for mixed systems (the most common form of digester), where hydrolysis is rate limiting, the hydrolysis rate determines the size of the digester. We now discuss the hydrolysis of various feed materials: 1. Hydrolysis of WAS. There has been a large amount of work investigating the rate and extent of WAS digestion, but only limited analysis of the actual mechanisms of cell solubilization as specific to activated sludge. It is a complex process, involving lysis of the cell, and subsequent degradation of both soluble and particulate cellular components (Aquino et al., 2008; Madigan et al., 2009). This is further complicated by the issue that microbial cells are naturally resistant to cell lysis by other cells, and that the cells are in flocs, with varying sizes. Degradability and hydrolysis rate have been extensively analyzed. As mentioned earlier, activated-sludge hydrolysis
Particulate carbohydrates, proteins, and lipids
Acidogens produce enzymes
Hydrolysis
Sugars and amino acids Fermentation acidogenesis
CO2 Alcohols and Long-chain organic acids fatty acids
NH3
Acetogenesis CO2
CO2 Hydrogen
Acetic acid
Hydrogenotrophic methanogenesis
Aceticlastic methanogenesis
Methane Figure 1 Key steps in anaerobic digestion processes.
617
Methane
CO2
May be rate limiting
618
Anaerobic Processes
4. Adsorption of enzyme onto surface
6. Transport of product to bulk
5. Reaction
2. Transport to bulk or local environment 1. Production of enzyme
7. Deactivation of enzyme 3. Diffusion from bulk to particle
Figure 2 Steps in enzymatic hydrolysis.
is an extremely complicated physical and chemical process that is, of necessity, represented as a first-order process (Eastman and Ferguson, 1981). Practical batch testing indicates that this complex material is well represented by first-order kinetics (Dwyer et al., 2008), while primary sludge (for example) has a far more complex kinetic profile, due to the presence of multiple primary substrates (Yasui et al., 2008). Extensive analysis also indicates that for untreated activated sludge, hydrolysis rates are relatively constant at approximately 0.1 d1 (Batstone et al., 2002; Eastman and Ferguson, 1981; Ge et al., 2010; Pavlostathis and Giraldo-Gomez, 1991). The degradability of activated sludge can be entirely related back to upstream sludge age, and longer sludge-age material will be less degradable (i.e., have a higher inert fraction; Ekama et al., 2007; Gossett and Belser, 1982). It is now widely accepted that material that is undegradable under aerobic conditions, is also largely undegradable under anaerobic conditions (Park et al., 2006; Speece, 2008). Therefore, material that is degradable under anaerobic conditions can be numerically calculated from the degradable fraction of the active aerobic biomass in the WAS (Ekama et al., 2007; Nopens et al., 2009). There is a wide range of pretreatment methods to increase sludgedegradability extent and rate (Aquino et al., 2008), and these are discussed further in the Section 4.17.2 of this chapter. 2. Hydrolysis of carbohydrates. Carbohydrates mainly originate directly or indirectly from plants. Generally, plant material is a mixture of cellulose (25–60%), hemicellulose (15– 30%), and lignin (15–20%) (Tong and McCarty, 1991). Straw, a commonly used feed material, consists of 70% cellulose and hemicellulose, 8% lignins, 15% mineral solids, and 7% other organic compounds (Hashimoto, 1986). The remainder is tannins, soluble sugars, and ash. The first two components are very similar and are digested anaerobically via similar mechanisms. Tong and McCarty (1991) list typical chemical compositions of lignocellulosic materials. Cellulose is made up of linear chains of D-glucose units. Hemicellulose is a branched polymer comprising several natural minor sugars. Ease of degradation depends on the nature (crystalline or amorphous) and chain length. Hemicellulose is of a shorter length (200 units), while cellulose can have a chain length of up to 10 000 units. Lignin is a dense three-dimensional polymer of aromatic molecules. It is hydrophobic and is linked by carbon as well as ether bonds. Conversion of lignin by anaerobic bacteria is unknown,
H
R
O
H
R O
H R O
N
C
C
N
C
N
H
H
C
C
C
H
Figure 3 Protein chain with amino acids linked by amide groups.
and high lignin contents (together with the presence of crystalline cellulose) generally restrict or prevent hydrolysis of the underlying cellulosic material (Yang et al., 2009). 3. Hydrolysis of proteins. Proteins are natural polymers of different amino acids joined together by peptide (amide) bonds. The backbone of a protein is a repeating sequence of one nitrogen and two carbon atoms (Figure 3). There are 20 amino acids found in nature. These are differentiated by the R group, which defines the function of the amino acid. A protein has three structural components: • Amino-acid composition and sequence (primary structure). • The three-dimensional shape as set by bond angles and hydrogen bonds forms a helical shape in complex proteins. This is the secondary structure. • The tertiary structure defines the macromolecular shape as set by bonding between di-sulfide groups and to a lesser extent, other inter-R bonding. There are two major areas of importance for hydrolysis processes. Amino-acid composition (primary structure) affects the products. The tertiary structure defines the proteins as either fibrous or globular. Fibrous proteins are structural materials such as keratin, which is protective, and collagen, which is connective. Globular proteins are often chemically functional and act as enzymes, hormones, transport proteins, or storage proteins. Hydrolysis of proteins can be rate limiting in the overall process, depending on ease of structure degradation (Pavlostathis and Giraldo-Gomez, 1991). Protein structure is one of the main factors affecting the rate of hydrolysis. Globular proteins are rapidly hydrolyzable, while fibrous proteins are difficult to hydrolyze (McInerney, 1988). In general, all proteins apart from the most rigid type of keratin (such as the outer layer of hair and fingernails) are hydrolyzable (Figure 4). There are three main groups of proteases: serine, metallo, and acid proteases which have alkaline (8–11),
Anaerobic Processes
619
The 1,3-specific lipases can only act at the outside bonds of the triglycerides, yielding 1,2-diacylglycerols and 2-monoacylglycerols. These glyceride esters are unstable and undergo acyl migration to 1,3-diacylglycerol and 1-monoacylglycerol. Subsequently, these can be degraded further by the 1,3-specific lipase to glycerol and free fatty acids. Fatty-acid-specific lipases catalyze the removal of a specific fatty acid, preferentially removing cis-D9-monounsaturated fatty acids. Other fatty acids are degraded very slowly, especially those containing an additional double bond between D1 and D9. Figure 4 Cow hair from an anaerobic reactor showing intact keratin (A) compared with degradation of interior by anaerobic organisms (B). Photograph by Dr Damien Batstone.
CH2 OH CH
OH
CH2-O-fatty acid CH-O- fatty acid
CH2 OH
CH2-O-fatty acid
Glycerol
Triglyceride
Figure 5 Glycerol and triglycerides.
neutral (6–8), and acidic (4–6) pH optimums, respectively (Ramsay, 1997). Enzyme production may be suppressed when readably biodegradable substrates such as glucose or amino acids are supplied (Patterson-Curtis and Johnson, 1989; Ramsay, 1997). 4. Hydrolysis of lipids. Lipids are glycerol bonded to LCFAs, alcohols, and other groups by an ester or ether linkage (Madigan et al., 2009). Fats and oils have all the alcohol groups esterified with fatty acids as shown in Figure 5 and these form the bulk of glyceridic material in mixed oils and fat with other glyceridic compounds, usually a result of processing. Hydrolysis is catalyzed by LCFA ester hydrolases, called lipases. These act at the lipid–water interface in enzymatic hydrolysis to degrade the insoluble reactant to soluble products. There is little work on degradation of lipids in anaerobic environments when compared with that on carbohydrate and protein substrates. Most of this has been focused on the rumen, reviewed by (McInerney, 1988). One particular characteristic of lipases is increased activity with insoluble rather than soluble lipids (Martinelle and Hult, 1994), indicating that the activity of lipases increases greatly when the concentration of triglycerides reaches saturation and forms a second phase. The lipases are adsorbed at the interface. As there is an adsorption mechanism, combined reaction and adsorption rate may be dependent on the surface area of the insoluble triglycerides. Bacterial lipases can be divided into three main types: nonspecific lipases, 1,3-specific lipases, and fatty-acid-specific lipases (Finnerty, 1988). Nonspecific lipases can hydrolyze any fatty acid triglyceride regardless of structure, acting at any of the fatty acids. These can completely hydrolyze the ester bonds acting equally at all alkyl sites.
4.17.1.1.2 Fermentation/acidogenesis Fermentation and acidogenesis refer to the same process of conversion of sugars and amino acids to simpler compounds (mostly acids and alcohols). Fermentation is commonly applied in biotechnology processes where the focus is on the product. Acidogenesis is applied in wastewater processes. In our opinion, fermentation is a more precise and preferred term. Fermentation is defined as the conversion of organics without an obligate external electron acceptor to produce both reduced and oxidized products. The two major groups of compounds subject to fermentation under anaerobic conditions are sugars and amino acids, which are discussed next. Fermentation of sugars. Anaerobic fermentation from sugars is likely the most widely applied biotechnology process worldwide. It is used to produce food products, renewable fuels, pharmaceuticals, and industrial chemicals. It is currently in focus for production of biofuels (e.g., ethanol and butanol). Historically, fermentation has been carried out by pure or specialized microbial cultures, which are constrained to produce specific products from sugars, based on their physiology and genetic capabilities. In anaerobic digestion processes, fermentation is mediated by mixed culture, and a wide range of potential products can be formed. Sugars ferment via the Embden–Meyerhof–Parnas (EMP) pathway to pyruvate, and subsequently to C3 products (propionate or lactate), or C2–C6 products via acetyl-CoA (Madigan et al., 2009; Figure 6). The most common products are shown in Figure 6, as determined in practical mixedculture fermentation tests (Ren et al., 1997; Temudo et al., 2008). Smaller amounts of additional compounds, including metabolic intermediates, are also often detected. Actual product mixes are regulated by a number of environmental conditions, including pH, gas-phase hydrogen concentration, temperature, and biomass retention time. It is reasonable to assume that hydrogen-rich reactions (e.g., production of acetate) would be enhanced at low hydrogen concentrations and production of alcohols enhanced at low pH (Ren et al., 1997). Regulation of mixed-culture fermentation is exciting, as it offers the possibility of producing fuels and industrial chemicals directly from raw feedstocks such as crop residues and straw. While a number of models have been proposed (Costello et al., 1991; Mosey, 1983; Rodrı´guez et al., 2006), none of these can effectively describe the mixture of products under dynamic conditions. The most promising current approach evaluates the thermodynamic driving forces under varying conditions (Rodrı´guez et al., 2006).
620
Anaerobic Processes 1glucose
4e−
2pyruvate
4e−
4e−
8e−
2lactate 2propionate
2CO 2
2e−
2e− 2acetyl-CoA 2e−=H2
6e−
2acetate
2CO 2 2e−
2e− 1butyrate
2ethanol
Figure 6 Major products from C6 monosaccharide fermentation. Excess electrons are removed as hydrogen as shown.
Fermentation of amino acids. There are 20 common amino acids, which can be divided based on the R group (Figure 3) into the following groups:
• • • • • • •
Alkyl R groups: glycine, alanine, valine, leucine, and isoleucine. Alcohol R groups: serine and threonine. Carboxyl R groups: Aspartic and glutamic acids. Nitrogen-containing R groups: lysine, arginine, and histidine. Sulfur-containing R groups: cysteine and methionine. Aromatic R groups: phenylalanine, tyrosine, and tryptophan. Proline, which forms an amide ring with the amide group.
Fermentation of amino acids can either be by direct oxidation, or by fermentation in pairs along a coupled pathway. The coupled pathway is termed ‘Stickland digestion’, and it has several properties: a. Amino acids are degraded as a pair. b. One of the pair of amino acids acts as an electron acceptor (i.e., it is reduced), and the other as the electron donor (i.e., it is oxidized). c. The donor amino acid is oxidized to NH3, CO2, and a carboxylic acid with a chain length one carbon atom shorter than the original donor amino acid. d. The acceptor amino acid is reduced to NH3, and a carboxylic acid with a chain length equal to the original amino acid. e. Amino acids can act as an electron acceptor, an electron donor, or as both, but there is no rule based on the R chain. f. In general, there is a 10% shortfall in electron-acceptor amino acids in commonly found proteins. Due to the properties of Stickland reactions, and because the amino-acid compositions of most commonly encountered proteins are known, it is possible to estimate the organic acids
produced from a given protein (Ramsay and Pullammanappallil, 2001). This, however, assumes that Stickland reactions are used. If the hydrogen concentration is low, uncoupled oxidation of amino acids can occur (Stams, 1994). Uncoupled degradation can also result in a higher energy yield and, as in Stickland reactions, energy is only produced from the oxidation reaction (during regeneration of carboxyl-CoA). Figure 7 shows coupled and oxidation reactions for alanine (which is always a donor acid), and glycine (which is always an acceptor). The degradation of alanine is the same in both cases, as it is oxidized during the coupled reaction. The only change is that electrons are wasted into hydrogen ions, rather than glycine.
4.17.1.1.3 Acetogenesis and methanogenesis from hydrogen Organic acids and alcohols are converted to acetate (oddchained organics to propionate also) by anaerobic oxidation. This process utilizes hydrogen ions or bicarbonate ions to produce hydrogen gas or formate, respectively. The thermodynamics of the oxidation reaction require that the electronacceptor end product (hydrogen or formate) be maintained at a very low concentration, and, hence acetogenesis is obligately linked to a hydrogen-utilizing reaction, such as methanogenesis (Batstone et al., 2006b; Boone et al., 1989). Hence, interspecies electron transfer (IET), in which hydrogen is the electron carrier, is vital to the growth of both microbes. Indeed, the only the syntrophic association is obligate, and other forms of electron carriers are possible – even direct electron transfer via microbial nanowires (Reguera et al., 2005). In anaerobic biofilms, the oxidizing organism is normally a bacteria, while the methanogen is an archaea, and can be directly observed in close relationship (Figure 8). Hydrogen (plus bicarbonate) and formate are functionally, and thermodynamically, very similar, with hydrogen having a higher diffusivity, and formate having a higher solubility. Advanced modeling has indicated that their microscopic characteristics will be similar in either of the electron carriers (Batstone et al., 2006b). In addition, the free energy of conversion between formate and hydrogen is relatively low (5.7 kJ mol1), and the two may exist in enzyme-assisted equilibrium (Thiele and Zeikus, 1988). Therefore, hydrogen can be regarded as the representative electron carrier. The thermodynamics of the reactions can be assessed by a free-energy calculation. For the reaction a A þ b B3c C þ d D (with stoichiometry a, b, c, and d), the adjusted free energy of reaction is (Madigan et al., 2009)
DG0 ¼ DG00 þ RT ln
½C c ½D d ½A a ½B b
ð1Þ
where DG0 is the adjusted free energy of reaction, DG0 is the standard free energy of reaction, and ½C c ½D d =½A a ½B b is the reaction quotient, or concentration of products divided by concentration of reactants. The adjusted free energy DG0 must be less than zero for the reaction to proceed. Based on standard concentrations in a digester (0.001 M organic acids and 0.1 M bicarbonate), the hydrogen thresholds for different acetogenic reactions can be calculated. These are shown in Table 1. These are thermodynamic thresholds, and the actual
Anaerobic Processes Coupled
Uncoupled
Oxidation C
COO− Alanine (donor)
NH 2 H3C
C
Oxidation
Reduction
H H2C
Glycine (acceptor) 2
H2C
COO−
Alanine (donor)
NH2
2e−
2e− COO−
O
Pyruvate, NH3
Pyruvate, NH3 CO2
Acetate H3C COO−
2e−
CO2
2e−
Acetyl CoA
Acetyl CoA
Acetyl phosphate
H3C Alanine
H2 2H+ H2
Energy Acetate
Acetate
Acetate + CO2 + NH3 + 4H
2H+
Acetyl phosphate
Acetate H3C COO− Energy
COO−
621
2glycine + 4H
2acetate + 2NH3
Alanine
Acetate + CO2 + NH3 + 2H2
Figure 7 Coupled and uncoupled conversion of alanine.
Figure 8 Syntrophic community of bacteria and archaea (anaerobic granule), engaged in acetogenesis and methanogenesis. Bar is 500 nm.
levels are higher. This indicates there is only a narrow region of hydrogen concentrations where these reactions may proceed. The mechanism of the reaction can be explained thus. Oxidation of butyrate and larger organic acids (C4þ) is by b-oxidation, a process in which larger organic acids are sequentially oxidized in a cyclic process. Two carbon atoms are removed as acetyl-CoA per cycle, and energy is recovered by substrate-level phosphorylation (Ratledge, 1994). The cycle continues until only acetyl, or propionyl-CoA, remains. This is converted directly to acetate or propionate. Unsaturated bonds are reduced directly with hydrogen (with the unsaturated bond as electron acceptor), in a favorable reaction (Ratledge, 1994). While common organisms oxidize a range of C4þ fatty
acids (McInerney et al., 1981; Roy et al., 1985), the kinetics, particularly of branched chain fatty acids, can vary substantially (Batstone et al., 2003). Propionate conversion is by a limited number of specialized organisms, with the carboxyl group being converted to carbonate, and the two methyl groups being randomly converted to either the methyl or the carboxyl group on the final acetate product (de Bok et al., 2004; Stams and Plugge, 1994). Ethanol was the first observed syntrophic methanogenic culture (Bryant et al., 1967), and due to favorable thermodynamics, it was found to accumulate substantial hydrogen before thermodynamic limitations set in. Degradation was found to be via acetyl-CoA. The major pathway of hydrogen or formate removal in mesophilic high-rate reactors is methanogenesis. This occurs by activation of a carbon dioxide molecule or formate molecule and successive hydrogenation of this complex. As a final step, methyl-CoM is formed and this is reduced to methane with a yield of 1 (adenosine triphosphate (ATP) mol1 methane formed. None of the methanogenic archaea can utilize energy from substrate-level phosphorylation and ATP is probably generated from a proton-motive force (Boone et al., 1993). While methanogenesis is the major sink for electrons in anaerobic systems, there are a number of alternative sinks, including nitrate reduction, sulfate reduction, iron reduction, and homoacetogenesis (formation of acetate from hydrogen). Alternative electron acceptors, such as nitrate, sulfate, and Fe3þ, are preferred substrates to hydrogen ions. Homoacetogenesis can occur whenever there is elevated hydrogen, but is commonly observed under lower temperatures, where the thermodynamics of this reaction are more favorable.
4.17.1.1.4 Aceticlastic methanogenesis This is the major methanogenic step, where acetate is cleaved to methane and carbon dioxide. Only a limited number
622
Anaerobic Processes
Table 1
Acetogenic reactions and hydrogen thresholdsa
Reactant
Reaction
H2 threshold (Pa)
Propionate
CH3 CH2 COO þ 3H2 O-CH3 COO þ 3H2 þ HCO3 þ Hþ
10
þ
Butyrate
CH3CH2CH2COO þ 2H2O-2CH3COO þ 2H2 þ H
100
Valerate
CH3CH2CH2CH2COO þ 2H2O -CH3CH2COO þ CH3COO þ 2H2 þ Hþ
100
þ
Ethanol
CH3CH2OH þ H2O-CH3COO þ 2H2 þ H
Palmitate
CH3(CH2)14COO þ 14H2O -8CH3COO þ 14H2 þ 7H
H2, HCO 3
4H2 þ HCO3 þ H -CH4 þ 3H2 O
1000
þ
10
þ
0.2
a
For acetogenic reactions, concentrations must be below the threshold levels. For the last reaction, concentration must be above the threshold.
of methanogens within the archaea have been identified that are capable of cleaving acetate:
• •
Members from the genus Methanosaeta within Methanosaetaceae – these are obligate acetate cleavers. Members from the genus Methanosarcina within Methanosarcinaceae. Members of this genus can also utilize hydrogen, CO2, and methylated C1 compounds (Ferry, 1993).
Methanosaeta is more pH, and ammonia, sensitive and dominates at below 103 M acetate (Zinder, 1993), while Methanosarcina is found outside these conditions, generally in high-ammonia conditions where there is also higher-effluent organic acids (Karakashev et al., 2005). Recent work has indicated that Methanosarcina may, instead of cleaving acetate to hydrogen and carbon dioxide, oxidize acetate to hydrogen, with subsequent reduction by a syntrophic methanogenic partner to methane (Karakashev et al., 2006) (Table 2). Therefore, under these conditions, Methanosarcina does not act as a methanogen, but simply provides electrons to another methanogen via hydrogen or an alternative electron carrier.
4.17.1.2 Physicochemical Processes and pH Physicochemical processes are those that are not biologically mediated, and hence occur spontaneously in water systems. This research field is generally referred to as aquatic chemistry (Stumm and Morgan, 1996). Some important physicochemical reactions that occur in anaerobic digesters are shown in Figure 9, and include 1. Association and dissociation of weak acids and bases such as water, organic acids, carbon dioxide, and ammonia – this is a rapid process. 2. Gas–liquid transfer of carbon dioxide, methane, hydrogen, and hydrogen sulfide – this is a medium-rate process. 3. Metal-ion precipitation to form solid precipitates – this is a medium-slow process. Unlike biochemical reactions, almost all physicochemical reactions are spontaneous and reversible. Therefore, equilibrium calculations are an important issue to assess physicochemical systems. The physicochemical state is most commonly expressed by the pH, or negative log of the hydrogen-ion concentration (–log10[Hþ]). It expresses the net balance of strong
Table 2
Acetoclastic methanogenesis
Substrate
Reactiona
Acetate (cleavage)
CH3COO þ H2O - CH4 þ HCO 3
þ Acetate (oxidation) CH3COO þ 4H2O - 2HCO 3 þ 4H2 þ H
DG0 31 þ 105
a
Reactions for coupled acids are shown. DG0 was calculated for reaction at pH 7.
and weak acids present in the system, but not their individual concentrations or strength. The physicochemical and biochemical reaction system are strongly linked in anaerobic digesters, through the following mechanisms:
• • • •
•
Biochemical reactions produce weak acids and bases, including organic acids, LCFAs, ammonia, and carbon dioxide. Biochemical reactions produce gases. Low pH inhibits biological activity through disruption of homeostasis and denaturing of enzymes, though specialized organisms can operate at extremes. The free form of many weak acids and bases, particularly ammonia, organic acids, and hydrogen sulfide, is inhibitory to organisms (Batstone et al., 2002). This means that not only does the total concentration of the parent compounds (e.g., inorganic nitrogen, sulfides, etc.) have an impact, but the pH also has an influence by determining the concentration of the inhibitory form (e.g., ammonia and hydrogen sulfide). Weak acids and bases buffer around their characteristic acidity coefficient (pKa, see further). This means that bicarbonate, in particular, resists pH changes around 6.3, since that is its pKa.
Of the three key classes of physicochemical reactions – acid– base, liquid–gas, and metal-ion precipitation – only the first two have been extensively addressed in anaerobic digestion models (Batstone et al., 2002). This is a clear limitation in anaerobic-digestion modeling, since the behavior of more concentrated systems, and particularly the behavior of solids cannot be effectively described without describing metal-ion precipitation (Batstone, 2009). Acid–base reactions are characteristically rapid, and can hence be described by the equilibrium equation. For the
Anaerobic Processes
CO2
623
H2 Gas
H2O
CH4
Composites
Biochemical
Liquid Gas Inerts
Death/decay
Proteins
Carbohydrates
MS
AA +
Lipids
NH3
NH4
VFA−, HCO3−, NH4+
HVFA, CO2, NH3 HCO3−
Ca2+ Growth
H2
HAc
Microbes
−
CO2
CO32−
Gas
CH4
H2O
CaCO3
Physicochemical Figure 9 Biochemical (vertical) and physicochemical processes (horizontal) in an anaerobic digester. AA, amino acids; MS, monosaccharides; HVFA, associated organic acids; VFA, dissociated organic acids; HAc, acetic acid; Ac, acetate. Adapted from Batstone DJ, Keller J, Angelidaki I, et al. (2002) Anaerobic Digestion Model No. 1 (ADM1), IWA Task Group for Mathematical Modelling of Anaerobic Digestion Processes. London: IWA Publishing.
reaction acid 2 base þ Hþ, the equilibrium relationship is
½Base½H þ ¼ Ka ½Acid
ð2Þ
where Ka is the acidity coefficient, and is often expressed as pKa ¼ –log10Ka, in a similar way to pH. Most analytical methods measure or report the total species concentration:
½Totalmeas ¼ ½Acid þ ½Base
ð3Þ
These two equations can be combined to give either the acid concentration, or base concentration, as a function of the measured concentration, the pH, and the acidity constant:
Ka ½Totalmeas Ka þ ½Hþ
ð4Þ
½H þ ½Totalmeas Ka þ ½Hþ
ð5Þ
½Base ¼
½Acid ¼
The equation system is complicated when there are three reactive species (e.g., the inorganic carbon system containing
CO2, HCO3 , and CO3 2 ), or more (e.g., the phosphorous system containing four reactive species). Equations (4) and (5) are commonly used to produce acid–base speciation diagrams. An example is shown for the inorganic nitrogen acid– base system in Figure 10. This demonstrates the relationship between pKa, pH, and fractionation. The reason why fractionation is practically important is that many acids and bases are mainly inhibitory in their free or uncharged form. Ammonia is free as the base (NH3), which is why ammonia inhibition increases at elevated pH levels (discussed later in the chapter). Other acid/base pairs of importance are the organic acids (most volatile fatty acids (VFAs) have pKa levels of 4.6– 4.8), CO2/HCO3 pair (pKa ¼ 6.35), H2S/HS (pKa ¼ 7.05), NH4 þ /NH3 (pKa ¼ 9.25), and HCO3 =CO3 2 , (pKa ¼ 10.3) (Batstone et al., 2002). Gas–liquid transfer is normally described by equilibriumdriven dynamic gas–liquid transfer. Hydrogen and methane are relatively insoluble, while carbon dioxide and hydrogen sulfide are relatively soluble. This means that the latter two compounds have a substantial impact on the liquid system. Metal-ion precipitation is also generally described by equilibrium-driven dynamic relationships. The actual mechanism of crystallization is complex and includes a number
624
Anaerobic Processes 1 Fraction as acid (NH4+)
0.9 0.8
Fraction as base (NH3)
Fraction
0.7 0.6 0.5 0.4 0.3 0.2 0.1
pKa = 9.3
0 8
8.5
9
9.5
10
pH Figure 10 Inorganic nitrogen acid–base speciation vs. pH. Note that the total inorganic nitrogen is equally split between ammonia and ammonium at a pH of 9.3.
of different factors, including the presence of seed, presence of confounding compounds, such as inhibitors and promoters, and solution activity. While simple first-order relationships have been used in complex systems, these are normally ineffective (Batstone, 2009). The basic solubility of a precipitant is described by equilibrium. For the reaction aAbþ þ bBa–2AaBb, the equilibrium relationship is
KSP ¼ ½Aa ½Bb
ð6Þ
where KSP is the equilibrium constant, and is normally referred to as the solubility product. For convenience, it is also often represented as pKSP ¼ log10KSP. The higher the pKSP, the less soluble the compound. Examples include CaCO3 (KSP ¼ 8.25), FeS (KSP ¼ 18), and CaOH2 (KSP ¼ 5.3). Note that it is the pH-adjusted anion concentration that is to be used in Equation (6), and, therefore, CaCO3 precipitation is driven by the concentration of CO3 2 . This means that pH has a strong impact on metal-ion precipitation. Until recently, physicochemical models within anaerobic digestion models have been relatively simple, mainly consisting of acid–base equilibrium equations, a charge balance to determine hydrogen-ion concentration, and gas–liquid transfer (Batstone et al., 2002). More complex models have represented limited nonideality, including ion activity (Musvoto et al., 2000a), and precipitation (van Langeraak and Hamelers, 1997). Simpler models work well with dilute, ideal systems without metal-ion precipitation, but have poor predictive power in concentrated systems, or where precipitationaffected compounds exist. This has led to implementation of a more robust but more complex physicochemical framework for anaerobic digester systems (Batstone, 2009).
include 1. Reaction rates increase with increased temperature according to the Arrhenius equation (Siegrist et al., 2002). As a rule, anaerobic digesters are relatively sensitive to temperature, with temperatures below 30 1C causing a substantial loss in activity. 2. A rapid decrease in activity with abrupt temperature increases above the maximum (Van Lier et al., 1996). Normally temperature rises are maintained below 2 1C d1. 3. Decrease in microbial yields, and an increase in apparent saturation concentration (KS), with increased temperature (Van Lier et al., 1996) related to an increase in cell maintenance. 4. Shifts in reaction pathways due to changes in the free energy of reaction with temperature. This is particularly relevant for oxidative reactions, and acetate oxidation becomes more competitive as compared to aceticlastic methanogenesis at higher temperatures (Zinder and Koch, 1984), while the reverse reaction (homoacetogenesis from hydrogen and carbon dioxide) is more favorable at lower temperatures (Rebac et al., 1995). 5. Pathogen deactivation increases with temperature. These impacts occur across the temperature range, but operating modes have been split based on reactor operability and dominant microbial population into the following three temperature ranges:
• • •
Psychrophilic 10–30 1C. Mesophilic 30–40 1C. Thermophilic 40–70 1C.
4.17.1.3 Temperature
Psychrophilic conditions are largely environmental, while mesophilic and thermophilic conditions are largely in engineered systems. There are also a number of physicochemical impacts:
Temperature has a number of impacts on outputs and internal processes in anaerobic digesters, including both biochemical and physicochemical impacts. Biochemical impacts
1. Increased temperature causes decreased gas solubility. 2. Volumetric gas production increases with increased temperature due to thermal expansion.
Anaerobic Processes
3. A change in temperature changes the solubility of solids. This may increase or decrease depending on solid enthalpy of precipitation. 4. Gas transfer rates increase, due to increases in diffusivity. 5. Increased temperature increases the water-vapor fraction in the gas phase. 6. The acid–base pKa values change with temperature (generally decreases). The variation in this is enormous. Organic acid pKa is relatively unaffected by temperature, while ammonia pKa changes dramatically. 7. Liquid viscosity increases with increased temperature. This changes the energy required to pump and mix reactor contents. Overall, as temperature increases from mesophilic to thermophilic conditions, the combination of all of these impacts can be observed as follows:
• • • • •
•
Rates increase due to increased activity. This can be especially important in hydraulic limited systems. Effluent organic-acid levels increase due to increased maintenance and substrate-saturation levels. Gas quality drops as the water and carbon dioxide fractions increase. Gas production increases because of increased activity and thermal expansion. pH is normally relatively stable. It drops due to increased organic-acid concentrations and lower pKa values, but rises due to decreased CO2 solubility. The net effect can be an increase or decrease depending on the feed type and reactor performance. The system is more susceptible to ammonia inhibition, due to a decrease in ammonia pKa, and hence, there is a higher concentration of free ammonia (see next section).
4.17.1.4 Inhibition and Toxicity Speece (2008) uses two definitions within the area of general restriction of biological processes: ‘‘inhibition: an impairment of bacterial function’’ (p. 432) and ‘‘toxicity: an adverse effect (not necessarily lethal) on bacterial metabolism.’’ Commonly, inhibition is reversible, while the effects of toxicants are irreversible. That is, if an inhibitor is removed, bacterial function will return to normal levels, while if a toxicant is removed, a portion of the population will have residual effects (e.g., be dead). Inhibition is measured by the IC50, or concentration at which bacterial catabolic rate is reduced by 50%, while toxicity is measured by a LD50 or median dose – dose which will kill half the population. While there are mechanisms or chemicals that particularly influence specific functional groups, methanogenic archaea are generally more vulnerable to inhibition, and toxicity than bacteria. The order of the least-to–most-impacted processes is as follows: acidogenesis-hydrolysis-acetogenesis/hydrogenotrophic methanogenesis-aceticlastic methanogenesis. While it has not been well documented in the literature, propionate is an exception, in that it responds after acetate to initial overloads, but can remain in the effluent for long periods (1–2 weeks) after the initial overload. Inhibition is the more commonly observed phenomena in anaerobic digesters. The IC50 measure is directly applicable for
625
use in noncompetitive functions for dynamic modeling (Batstone et al., 2002), while toxicity is not commonly modeled, largely because modeling has limited capacity to address the impacts of toxicants. The mechanism of toxicants is often specific, acting on a particular mechanism of cellular metabolism. LCFAs are a common toxicant, which are thought to adsorb to the cell surface and block substrate and membrane proton transfer (Hwu et al., 1996). Other examples of toxicants include detergents, aldehydes, nitro-compounds, cyanide, azides, antibiotics, and electrophiles (Batstone et al., 2002; Speece, 2008). Inhibition can follow a number of different mechanisms, most of which either decrease the energy available from catabolism, or increase the amount of energy needed for maintenance. Common forms of inhibition are pH inhibition, ionic inhibition, product inhibition, and weak acid and base inhibition. They are discussed in detail in the following. pH Inhibition. pH inhibition is a combination of weak acid or base inhibition, disruption of cellular homeostasis, and reversible and irreversible protein denaturation. Most anaerobic organisms have a relatively broad pH optimum, with activity steady through the optimum. Anaerobic digestion operates best at a pH below 8.0, with activity of most organisms dropping above that pH, due to either free-ammonia inhibition, or other mechanisms. Lower pH is a combination of free-acid inhibition and pH inhibition. Since anaerobic digestion is mostly an acid-producing process, low pH inhibition is the most relevant form. Optimal pH levels for the different anaerobic biochemical functional groups are:
• • • •
Hydrolysis. Normally optimal above pH of 6.0, feasible up to 5.0. Acidogens. Optimal between 5.5 and 8.0, feasible up to 4.0 (Batstone et al., 2002). Acetogens/hydrogen-utilizing methanogens. Optimal between 6.5 and 8.0, feasible up to 5.0 (Batstone et al., 2002; Ferry, 1993). Aceticlastic methanogens. Optimal between 7.0 and 8.0, feasible up to 6.0.
As shown above, acid-producing microbes (acidogens and acetogens) have a higher tolerance for lower pH values than acid-consuming microbes (aceticlastic methanogens). An increase in load to a methanogenic digester will generally cause a decrease in pH, due to an increase in most acids, as well as the weak acid bicarbonate – even in an ideally operated digester. Where wastewater is poorly buffered, that is, where a lack of weak acids or bases causes poor resistance to pH changes, the pH can dip below 7.0 in response to substantial load increase. This can cause aceticlastic methanogens to be inhibited, which causes a further pH decrease due to accumulation of acetic acid. The overload is therefore self-reinforcing, and causes an acid overload. This can be difficult to recover from. This is mainly an issue in high-rate systems, where there is little or no ammonia release, a lower level of bicarbonate buffering, and a lower operating pH. Most high-rate anaerobic digesters operating on carbohydrate wastewaters require active base dosing to maintain a suitable pH, and this can form a major portion of the cost in these plants, though it is possible to reduce this by effluent CO2
626
Anaerobic Processes
stripping and recirculation (Ramsay and Pullammanappallil, 2005). Ionic inhibition. The mechanism of ionic inhibition involves increasing maintenance requirements, due to an increase in basic osmotic pressure. Sodium is the most relevant ion, with IC50 values between 5 and 30 g l1 depending on the level of acclimatization, function, and antagonistic or protagonistic ions (Feijoo et al., 1995). Acclimatization is possible and common. Product inhibition. Product inhibition occurs when products build up to the point where the catabolic reaction becomes unfavorable, that is, where the adjusted free energy of reaction as shown in Equation (1) becomes positive. The most common case is inhibition of propionate acetogenesis, caused by
Active transport of H+ (requires energy) CH3COOH (acetic acid) Passive transport
H+ CH3COO−
Cell; pH = 7.3
4.17.1.5 Rate-Limiting Steps
CH3COO− (Acetate) Bulk; pH = 7.0 H+ Active transport of H+ (requires energy) NH3 (Ammonia) Passive transport
H+ NH4+
Cell; pH = 7.3 NH4+ (Ammonium) Bulk; pH = 7.8 Figure 11 Mechanism of weak acid (top), and base (inhibition) by passive diffusion of the free form of the acid or base into the cell, and disruption of homeostasis.
Table 3
accumulation of hydrogen levels above those shown in Table 1, or substantial accumulation of acetate. Weak acid and base inhibition. Weak acid and base inhibition are caused by passive transport of uncharged acids (e.g., organic acids) or bases (e.g., ammonia) into the cell. These acids or bases then dissociate or associate within the cell to disrupt homeostasis (Figure 11). This causes increased maintenance requirements. Some important compounds causing free acid or base inhibition are listed in Table 3. While adjusting pH is a normal method to address free acid or base inhibition, it is an expensive exercise, due to the inherent buffering in most anaerobic digesters. Free-ammonia inhibition is likely the most commonly encountered form of inhibition, particularly in manure digesters and where the feed is proteinaceous, as the ammonia causes a high pH and acts as an inhibitory agent as well. This not only causes poorer overall performance, but can also cause more fundamental shifts, and (Karakashev et al., 2006) found that high-ammonia systems were dominated by Methanosarcina, oxidizing instead of cleaving the acetate. As the free form of ammonia is most important, and because temperature has a strong impact on the pKa, the entire system is heavily impacted by both temperature and pH. This is demonstrated in Figure 12, which shows that in a system with 2000 mg N l1 a thermophilic system (55 1C) will have a pH threshold of approximately 7.5 before strong inhibition occurs, while a 37 1C system will have a threshold of approximately 8.0.
This chapter outlines the key processes that occur in an anaerobic digester. In most cases, for a given wastewater type or reactor design, there is a rate-limiting step that needs to be managed in order to achieve optimal design and operation. The most common controlling mechanisms are hydrolysis and methanogenesis. Hydrolysis is normally the rate-limiting step for solid digesters (41% solids), where there are no other inhibitory factors present. For most solids, a retention time of 410 days is required (see Section 4.17.2.2.2), and at lower retention times, undigested solids will go to the effluent. Performance, as assessed by solid destruction, will decrease. Aceticlastic methanogenesis is normally the rate limitingstep in high-rate anaerobic wastewater-treatment systems, or where there is a higher level of inhibitors. Aceticlastic methanogenesis generally controls treatment systems where there are biomass limitations, or where the system is heavily loaded. Occasionally, (e.g., manure digesters), both hydrolysis can limit, due to slow solid degradation, and simultaneously, methanogenesis can cause elevated organic acids, due to ammonia inhibition.
Compounds causing free acid or base inhibition
Compound
Inhibitory concentration (free acid or base)
pKa
Condition at which inhibition occurs
NH3 (ammonia) H2 S (hydrogen sulfide) HVFA (organic acids)
1–2 mM (14–30 mgN l1) 2–3 mM (32–40 mgS l1) 0.2 mM (13 mg l1)
9.25 7.05 4.8
High pH Neutral and low pH Low pH
Anaerobic Processes
627
Free ammonia at total NH3/NH4+ of 2000 mg l−1 0.04
55°C
37°C
20°C
Free ammonia (M)
0.035 0.03 0.025 Inhibition strong 0.02 0.015
Inhibition significant
0.01 0.005
Inhibition starts
0 6.5
7
7.5
8
8.5
9
pH Figure 12 Free ammonia levels and ammonia inhibition.
In rarer cases, acetogenesis/hydrogenotrophic methanogenesis can be the rate-limiting step (e.g., hydrogen overload in a highly loaded high-rate system fed with soluble sugars), but increases in the higher organic acids may also be in response to an increase in acetic acid. In the following sections, these controlling mechanisms are discussed in context with technology selection, design, and operation.
4.17.2 Selection and Design of Anaerobic Technology 4.17.2.1 Anaerobic Digester Technologies Implementation of anaerobic digestion needs to address the two key issues of (1) maintaining sufficient retention time to allow for hydrolysis of particulate substrates and (2) providing beneficial conditions for aceticlastic methanogenesis, including maintenance of pH above 7.0. Technologies are split between wastewater-treatment technologies, which need to focus on goal 2, with extended sludge-retention times, but limited liquid-retention times, and those which need to focus on goal 1, with extended solid-retention times (Figure 13). Technologies except for high-rate systems are largely hydrolysis limited. Treatment technologies are summarized in, and described further, in the following sections (Table 4).
4.17.2.1.1 High-rate anaerobic digestion High-rate anaerobic digesters normally operate with extended solid-retention time, and short hydraulic-retention times, by integrating solid retention within the main digester (Figures 14 and 15). The most common type is an upflow anaerobic sludge blanket (UASB) reactor, in which liquid percolates through a partially settled sludge blanket. This operates with a flocculant sludge blanket, but relies on formation of anaerobic granular sludge (particles 4200 mm) for higher loading systems, especially if high effluent quality is to be maintained. High-rate digesters require a low solid feed, with relatively high amounts of soluble feed material, and are most often used for industrial wastewaters as well as domestic sewage
treatment (van Lier, 2008). Hydraulic-retention times are normally short with o48 h, while solid-retention times can be very long (4200 days, years). UASB reactors have a gas– liquid–solid separation in the upper part of the digester, while variations may include packing (Figure 14) in hybrid reactors, or extended super-high-rate/low footprint systems such as expanded granular sludge bed (EGSB) and internal circulation (IC) reactors. Other alternatives for high-rate anaerobic systems include anaerobic baffled reactors (multi-compartment reactors), fluidized bed or attached-growth systems, fixed-media anaerobic filters, anaerobic membrane bioreactors, and sequencing anaerobic batch reactors. UASB type systems are currently the market leaders in high-rate systems by a large margin (van Lier, 2008).
4.17.2.1.2 Anaerobic ponds Anaerobic ponds are a low-capital cost option, but they tie up land and require desludging approximately every 10 years, which can be excessively expensive (US$150 per dry ton). Anaerobic ponds are typically operated with very limited external control (e.g., temperature) and are therefore largely impacted by the local climate. This limits the effectiveness of ponds in colder regions. Overall costs are heavily driven by solid loading. Methane capture is relatively poor, and this results in an increase in greenhouse-gas emissions, and, generally, odors from the pond. Due to the large volumes, correction under failure can be extremely expensive or impractical. Anaerobic ponds have a depth of 5 m, with surface area determined by loading rates.
4.17.2.1.3 Fully mixed liquid digester Fully mixed digesters are most often applied to sewage sludge, activated sludge, and manure digestion (Speece, 2008). They are the most commonly applied configuration for anaerobic digestion. They operate as fully mixed reactors, with either gas recirculation or mechanical/liquid mixing systems. Mixing configuration is critical, and is reviewed further in (Tchobanoglous et al., 2003), particularly with respect to sludge
628
Anaerobic Processes 100
Plug flow
Hydraulic retention time (d)
Anaerobic ponds Liquid mixed digesters
10
Solid-phase leach bed
1
High-rate AD
0.1 0.01
0.1
1
10
100
Feed solids concentration (%) Figure 13 Anaerobic treatment technologies ranked by hydraulic-retention time (vertical axis) and solid concentration (horizontal axis).
digestion. Their configurations include cylindrical (normally with recirculated gas or liquid mixing) and egg-shaped (normally with mechanical mixing) systems. Maximum loading rate is heavily dependent on achievable solid levels, and performance can often be enhanced by pre-concentrating solids. Due to viscosity and heat-exchange consideration, the maximum in-reactor solid concentration is approximately 4% (feed concentration of approximately 8%). Costs are relatively high due to their engineered nature.
4.17.2.1.4 Plug-flow liquid digesters Plug-flow liquid digesters operate as a semisolid liquid (10– 20%) in a long polyethylene tube, vaulted brick, or concreteshaped reactor. Material is loaded at the front of the digester, and passes through to product at the end. As it is not mixed, contact with biomass is poor. These reactors have high kinetic efficiency, due to the plug-flow configuration, but are susceptible to lack of inoculation and topical souring. They are most often applied to agricultural solid digestion.
material removed). The latter is considerably more expensive due to solid handling and feed requirements. An alternative to in-reactor methanogenesis is recirculated leachate leach bed reactors (Figure 15). In this configuration, leachate is continuously percolated through a loop that includes the main solid phase leach bed, as well as a high-rate system to remove organic acids produced by the leach bed. This system has the advantage that overload and souring of the leach bed is far less likely, and gas production is steadier. The main disadvantage is susceptibility of the UASB reactor to solids. This type of system has been applied to municipal solid waste, and poultry litter (Rao et al., 2008).
4.17.2.2 Digester Selection and Design for Specific Applications Common wastewater types are shown in Table 5. As demonstrated in Figure 13, wastewater technologies are classified by their solid concentration. Specific considerations for application of technologies are given in the following sections.
4.17.2.1.5 Solid phase (leach bed)
4.17.2.2.1 Domestic and industrial wastewater
Solid-phase digesters are similar to an engineered, high-rate landfill, where material is loaded in a reactor, tumbler, or baskets, and leachate liquid is circulated through the reactor. Liquid percolates through the solid matrix and liberates organic acids, which are subsequently degraded to produce methane. It can be produced either in batches (where the system is reacted until no more methane is produced), or continuously (where material is continually added, and spent
The main criteria for application of high-rate granular anaerobic treatment technology are higher strength (4500 mg COD l1), low solids (2000 mg l1), and low oil and grease (o500 mg l1). Given these constraints, it is not surprising that 75% of applications of high-rate technology are on wastewater largely containing soluble carbohydrates and organic acids (e.g., cannery, brewery, confectionery, and distillery) (van Lier, 2008). High-rate anaerobic digestion has been
Anaerobic Processes Table 4
629
Anaerobic digestion technologiesa
Technology
Principle
Advantages
Disadvantages
Loading rate (kg COD m3d1)
High-rate digester/upflow anaerobic sludge blanket
Mainly liquid wastewater flows upward through a granular bed
Low footprint, low capital cost, very stable, produces good effluent
Intolerant to solids
10 (UASB) 20 (EGSB/IC)
Anaerobic pond
Large retention time mixed vessel
Low capital cost
Very high footprint Must be desludged Methane capture poor Can produce odors
0.1
Mixed tank
Dilution to 3–6%, and continuous feed in mixed tank. Retention of 20 days. Used across many industries
Established tech Easy to control Continuous gas production
Poor volumetric loading rate Expensive tanks Need dilution liquid Liquid (not solid) residue
1–3
Liquid plug flow
Dilution to 15%, and feed through a liquid plugflow reactor
Very high loading rates Continuous gas production
Need dilution liquid. Poor contact with active biomass. Liquid residue
5
Batch solid phase
Fill and react in a solidphase reactor. Can be an engineered landfill (but must be properly sealed). System is loaded, enclosed, and leachate/inoculum circulated intermittently
Can be very cheap Very high loading rates Good gas conversion due to retention of active biomass Easy to control via leachate No milling required
Non continuous system (gas–flow changes in quality and flow over time) Can be difficult to seal (gas seals) Needs loading and unloading
6–10
Continuous dry solid phase (plug flow)
Continuous feed of solid phase through a system. Recirculation of leachate around solid phase
Continuous gas and residue production Do not need dilution liquid Very good loading rates
Extremely high capital costs, and only really practical at very large scale. Very complicated mechanical system Potential solid handling issues
10
a
Note that the high loading of later options is achieved by high solids concentrations.
traditionally regarded as being less applicable to proteinaceous wastewaters, due to poor granule development (Fang et al., 1994). However, it is more likely that this is due to the particulate nature of these wastewaters (Batstone et al., 2004), and high-rate granular systems fed with soluble proteins (e.g., gelatine, casein) can be as effective as those fed with soluble carbohydrates (Moosbrugger et al., 1990). One of the main considerations associated with carbohydrate wastewater is buffering and pH. Carbohydrate wastewaters have no inherent buffering, which means that the acidity associated with carbon dioxide production needs to be offset by addition of a base. This can be a substantial cost consideration as outlined in the physicochemical section, although substantial savings can be achieved by effluent CO2 stripping and recycling. This is not as severe for protein-type wastewaters, as the weak base ammonia is produced during acidogenesis of proteins. Excessive ammonia release can cause free-ammonia inhibition. The flexibility of high-rate anaerobic digestion is illustrated by its applicability to domestic wastewater. Domestic
wastewater would normally be a poor feed source for high-rate anaerobic digestion, being low in strength (o1500 mg COD l1), relatively high in proteins, fats, and solids (often 4500 mg SS l1), and normally at lower temperatures. However, it has been successfully applied in both pilot and full-scale for removal of organics, and for sanitization (Seghezzo et al., 1998). This is further addressed in a later section.
4.17.2.2.2 Sewage solids and activated sludge biosolids Primary sewage solids (primary sludge) and activated sludge are the two main solid streams produced from activated sludge treatment plants. Primary sludge is material that can be settled out of raw sewage, and is relatively degradable, that is, 60– 100% can be anaerobically degraded, depending on the upstream catchment. Primary sludge has a relatively large lipid component (approximately 50% by COD; Siegrist et al., 2002; Speece, 2008). Activated sludge is a combination of microbial material produced during the activated sludge process
630
Anaerobic Processes Gas
Gas
Effluent
Effluent
Gas−liquid−solid Packing Granules
Inlet
Inlet
(a)
(b)
Figure 14 Upflow anaerobic sludge blanket (UASB) (a) and hybrid (b) systems.
Bleed stream
Gas Leach bed Makeup water
Overflow
Solid feed
High rate (UASB) Figure 15 Combined leach bed and high-rate system.
Table 5
Preferred technologies for different wastewater types
Application
Solids concentration (%)
Preferred technology
Design parameter
Nominal design parameter
Domestic or industrial wastewater
o0.2% (soluble solids may be up to 5) 2–7 2–7 10–30
High rate
Mass loading
10 kg COD m3d1
Mixed liquid phase Mixed liquid phase Solid phase
Retention time Retention time Retention time
10–15 days 20 days 30–50 days (batch)
Sewage solids, activated sludge Animal manure Organic solid wastes
(partially degradable), inert particulate material derived from influent material (not degradable), and undegradable cellular product (not degradable; Nopens et al., 2009). Activated sludge is a more homogeneous material than primary sludge, with a lower lipid content, and consequently higher protein and carbohydrate content. Overall, degradability is heavily dependent on sludge age (see further). The cost of biosolid handling and disposal can be a substantial fraction (30–50%) of overall wastewater-treatment costs, with cost being determined on a per wet ton basis. The
key considerations for sludge treatment are (1) volume and mass reduction, to reduce all costs associated with handling; (2) removal of unstable organics, to improve utility and storage options; and (3) pathogen removal, to increase utility and safety of the sludge product. The driving consideration is volume/mass reduction, since this determines eventual cost. Anaerobic digestion is effective in meeting all considerations, providing cost-effective solids and organics destruction, allowing essential pathogen destruction, and improving dewaterability of the final product.
Anaerobic Processes
70 60 50 40 30 20 10 0 5
10
15
20
25
30
0.9 0.8 0.7 0.6 0.5
Primary sludge khyd = 0.5 d−1
Activated sludge khyd = 0.3 d−1
Crop residues khyd = 0.1 d−1
10
15
40
Figure 17 Anaerobic degradability waste-activated sludge (WAS) vs. upstream sludge age.
0.4 5
35
Activated sludge age (d)
khyd = 1 d−1 Methanogenesis begins to fail
% of degradable fraction destroyed
1
temperatures, the degradability of activated sludge simply becomes too low for viable anaerobic digestion, as the biogas produced is insufficient to meet mixing requirements and provide sufficient energy for heating of the digester. In these cases, without a supplementary primary sludge stream, anaerobic digestion is no longer an option for sludge stabilization. For these poorly degradable streams, there are pretreatment options to improve both apparent hydrolysis coefficient and degradability. Lower-energy options, such as sonication, temperature-phased anaerobic digestion (TPAD), and enzymatic pretreatment, appear to largely act to increase apparent hydrolysis coefficient (Ge et al., 2010), and hence move the material upward as shown in Figure 16. High-energy options, such as thermal hydrolysis, increase the amount available and the hydrolysis coefficient, and hence can result in considerably enhanced performance (Batstone et al., 2009), though at higher capital and operating cost.
Activated sludge degradability (%)
Sizing of sludge digesters is driven by sludge hydrolysis rate coefficient (Speece, 2008; Tchobanoglous et al., 2003). Primary sludge is rapidly degradable, with first-order coefficients of the order of 0.3–0.5 d1 (Gujer and Zehnder, 1983; O’Rourke, 1968; Siegrist et al., 2002). Activated sludges are more slowly degradable, with hydrolysis rates of the order of 0.1–0.3 d1 (Ge et al., 2010). The impact that this has on digester sizing and performance is shown Figure 16, which demonstrates that as hydrolysis rate decreases, a longer retention time is required to achieve the equivalent efficiency. Apart from hydrolysis rate, the other major factor determining performance is degradability (fd). This represents the amount of material, either as COD, or as organic volatile solids (VS) that can be broken down to methane or biogas, respectively. For a perfect digester, it would be equivalent to the organic solids, or VS, destruction, but in most cases, the VS destruction is 70–90% of the degradable fraction. It has been extensively shown that the availability of material in both primary and activated sludge is the same across aerobic and anaerobic systems (Ekama et al., 2007; Gossett and Belser, 1982), and hence, degradability, or inert fraction can be directly translated between activated sludge and anaerobic models (Nopens et al., 2009). Primary sludge has a degradability of 60–100%, depending on the upstream catchment. A larger proportion of industrial input normally results in a lower net degradability. Activated sludge degradability depends heavily on the inert fraction remaining from the activated sludge process, and is hence heavily dependent on upstream activated sludge age. This was evaluated by Gossett and Belser (1982), and the results are summarized in Figure 17. Therefore, a WAS with an upstream sludge age of 15 days would be expected to have a degradability of 45%. With reference to Figure 16, a digester with a retention time of 20 days would be expected to have an efficiency of 85%. Therefore, feeding this digester with this sludge, an overall VS destruction of 45 85% ¼ 38% would be expected. As can be seen, WASs are generally poorer candidates for anaerobic digestion as compared to primary sludges. At higher sludge ages and/or higher activated sludge
631
20
25
30
Digester hydraulic retention time (d) Figure 16 Digester performance (% efficiency on degradable fraction) vs. hydraulic-retention time, and hydrolysis coefficient (khyd).
632
Anaerobic Processes
4.17.3 Interpretation and Operation of Anaerobic Systems Anaerobic reactor systems have been traditionally regarded as more difficult to control than aerobic wastewater systems. In practice, this is partially true, as commonly the only control handle is feed rate, which in the wastewater industry is largely determined by upstream considerations (Steyer et al., 2006). Aerobic wastewater treatment or digestion has a wider range of control handles, including aeration intensity and internal and sludge recycles and bypasses. Monitoring of anaerobic reactors and digesters is also more problematic, as the nonlinearity of the physicochemical process (see Section 4.17.1.2) means that simple sensors, such as pH and gas flow, have limited utility (Steyer et al., 2006). Conversely, anaerobic processes operate at higher loading rates and feed concentrations than aerobic systems, meaning that reactor sizes are smaller, and they can be readily overdesigned. In addition, anaerobic systems have large time constants (change relatively slowly) in the linear region, and it is suitable to choose a long-term optimal operating condition and aim for that set point.
4.17.3.1 Evaluating and Determining Controlling Mechanisms Optimal instrumentation, interpretation, and operation of digesters depend heavily on the controlling mechanism as discussed in Section 4.17.1.5, which is in turn dependent on configuration and feed type. Systems can be divided into 1. Methanogenesis controlled systems. Aceticlastic methanogenesis is the controlling mechanism in systems which are fed predominantly soluble wastewaters, or where pH buffering is poor. These are most often high-rate anaerobic digesters. Diminished performance is indicated by elevated acetate (and other organic acid concentrations), as well as liquidand gas-phase hydrogen concentrations (Pauss and Guiot, 1993), at mild overload conditions, and substantially decreased pH (o7.0) and gas flow during process failure. The kinetics are fast, and the process is highly nonlinear at the point between mild and severe overload. 2. Hydrolysis controlled systems. Hydrolysis is normally the controlling mechanism for systems fed with predominantly solids (41% solids), and performance is mainly determined by retention time of solids in the digester. The kinetics are relatively slow, and the overload mode is diminished performance in terms of gas flow, and unstabilized solids in the effluent. Poor performance is determined by analysis of these measures. Solid digesters are commonly well buffered due to release of ammonia from protein digesters, and pH is therefore a less-useful measure of process stability. Most of these systems are relatively linear and stable in response to changes in load. 3. Inhibited-hydrolysis controlled systems. These are relatively stable hydrolysis controlled systems (normally solid digesters), but which have a methanogenic inhibitor in the feed. The most common instance is manure digesters, where gas flow is largely determined by hydrolytic processes, and hence the retention time, while the presence of
ammonia causes substantially elevated organic-acid levels. In this case, the process itself is relatively stable, due to the ammonia buffering, but long-term performance may be poor both due to effluent organic acids (hence, lost biogas), and an inappropriate retention time for hydrolysis. Thus, the controlling mechanisms can be determined both from reactor and feed types (e.g., solid digesters vs. liquid-fed high-rate systems), and from direct analysis of solids destruction levels, and organic acid levels. This then leads to ongoing analysis of the most suitable performance indicators.
4.17.3.2 Performance and Process Indicators Suitable performance offline indicators or online sensors should provide an accurate reflection of process performance, related to process goals. If necessary, they should also offer potential for process correction, and in advanced cases, online process control. Again, selection of a suitable indicator or sensor depends heavily on the application or reactor type. As examples, the process goal for solid digesters is solid destruction (and hence gas production). A suitable indicator would be VS destruction, or gas yield. The process goal for high-rate anaerobic systems is good effluent quality, and process stability. Therefore, an indicator that provides early warning (prior to process failure) is desired. The process goal for fermentation systems is the extent of fermentation (or desired mix of organic acids). A good indicator would be product mix as measured by VFA concentration. Stability-state sensors for anaerobic digesters are difficult to apply online. As stated previously, there is a balance between the rate of hydrolysis and fermentation, and the rate of methanogenesis. When the production rate of organic acids exceeds the capacity of the digester to remove organic acids, the pH can drop, and the reactor sours. This is a hysteric process that can be extremely expensive, time consuming, and difficult to recover from. Start-up is a particularly hazardous period for this, as the biomass is nonacclimatized, and may vary in quality. The most simple sensors for anaerobic digesters are pH and gas flow. Due to system nonlinearity and stability characteristics, these measures are generally unsuitable (Steyer et al., 2006), only changing after the reactor sours. The best indicators are the intermediates, including volatile fatty acids (Pind et al., 2003), and measures such as bicarbonate alkalinity which indicate resistance to overload (Steyer et al., 2006). Overall, VFA concentration is by far the most widely applicable, direct, and meaningful measure of stability. However, the measurement of VFAs is generally an offline process involving measurement by gas chromatography-flame ionization detection (GC-FID), which is relatively slow and expensive, or titration, which is slow, and can be inaccurate in the presence of other buffers. While online methods for VFA measurement have been developed (Boe et al., 2007; Pind et al., 2003; Steyer et al., 2006), these are generally expensive, and/or require extensive sample preparation (e.g., online membrane filtration or gas-phase extraction), and there is still a need for a simple, relatively low cost online sensor to indicate anaerobic process stability state.
Anaerobic Processes 4.17.3.2.1 High-rate anaerobic reactors Most high-rate anaerobic digesters operate on mainly carbohydrate-based industrial wastewaters (van Lier, 2008), including agro-food, beverage, distillery, and pulp/paper. The objectives of the process are to remove organics, in order to reduce load to downstream treatment units, and potentially produce an effluent suitable for reuse or discharge to sewer. This requires good organic-acid removal, and process stability. The key performance measure is therefore VFA concentration, which indicates the level of acid-contributing compounds, as well as the bicarbonate, or partial alkalinity (PA – titration to pH 5.8). Titration to pH 4.2 indicates the total alkalinity (TA), which includes both bicarbonate, as well as organic acids. The contribution of the organic acids is termed intermediate alkalinity (IA ¼ TA PA). The current guideline for reactor stability is IA/TA r0.3, that is, the ratio of VFA contributed versus TA should be less than 0.3 (Steyer et al., 2006). Speece (2008) critiques this in influent analysis, pointing out that a large proportion of the total alkalinity must be also allocated to neutralize the CO2 produced during the digestion process, and suggests that reserve alkalinity (after production of CO2) is a better measure. For direct digester analysis however, the IA/TA measure is a reasonable indicator, though absolute values (of PA and VFA) should also be assessed. The two terms in this measure are contributed by bicarbonates and organic acids, and the IA/TA terms can either be measured directly (by offline, or online titration), or indirectly, and calculated by alternative methods, including Fourier transform-infrared (FTIR) (Steyer et al., 2006) and GC-FID, which are the standard analytical methods in commercial laboratories. For smaller systems, offline titration is simple, low cost, and relatively informative.
4.17.3.2.2 Sludge digesters As stated in Section 4.17.2.2.2, sludge and higher solid digesters can never achieve a high-quality effluent, due to the presence of inert solids. The main cost associated with primary and activated sludge digesters is disposal of the product, and the value of gas produced is relatively low compared to this. Therefore, the primary performance-related measure is organic solid destruction (or VS destruction). This is naturally related to methane production, since any solid destroyed must be created as methane. Secondary performance measures include stability (as measured by remaining degradable solids) and pathogen levels (Speece, 2008), and both of these are normally regulated in sludges, on the basis of vector attraction, usability, and disease control grounds. Additional performance measures may include mineral and nutrient content, odor, dewaterability, and texture, which are largely related to primary and secondary measures. As an example, a wellstabilized anaerobic biosolid product will generally have good dewaterability and low odor. Given that the primary measure is solid destruction, there are three ways to calculate this (Ge et al., 2010): mass-balance VS destruction, which assesses the flow of organic solids out, compared to the flow of organic solids in; Van Kleeck VS destruction, which is a modification of the mass balance to use VS fraction only; and apparent VS destruction of gas flow, which relies on the principle that organics destroyed must be
633
converted into gas. All measures assume that the system is at steady state, or values are averaged over longer terms. If the system is not at steady state, mass balance VS destruction and gas flow VS destruction need to be adjusted for flow, and van Kleeck VS destruction cannot be used. Mass-balance VS destruction is calculated as follows:
VSdestroyed ¼ ðVSconc;in VSconc;out Þ=VSconc;in
ð7Þ
where VSconc is the concentration of organics as measured by the volatile solids method (g l1), and subscript in and out indicate concentrations in the inlet and outlet streams. Van Kleeck VS destruction is calculated as follows:
VSdestroyed ¼
VSfrac;in VSfrac;out VSfrac;in VSfrac;in VSfrac;out
ð8Þ
where VSfrac is the fraction of total solids that is volatile (VSconc/TSconc). Gas flow VS destruction is calculated as
VSdestroyed ¼
CODgas ðkg COD d21 Þ CODin ðkg COD d21 Þ
ð9Þ
where CODgas is the calculated gas flow COD in kg COD d1. It can normally be calculated as
CODgas ¼ 2:9Qgas pCH4
ð10Þ
where Qgas is the gas flow at standard temperature and pressure (N m3 d1), pCH4 is the partial pressure of methane (atm), and 2.9 is a conversion factor (kg COD N m3). CODin is the incoming COD, and can be either directly measured, and multiplied by flow for a kg COD d1, or calculated as
CODin ¼ 1:5VSconc;in Qin
ð11Þ
where 1.5 is the assumed COD:VS ratio for activated sludges (approximately 1.7–1.8 for primary sludges), VSconc,in is the influent organic solids (kg m3 or g l1), and Qin is the inflow/ reactor hydraulic flow (m3 d1). Each of these measures has specific advantages, and can be influenced by different systematic and random errors: 1. Mass balance VS destruction. It is sensitive to errors in flow measurement, and systematic sampling issues. For example, it is common to have differential settling around sample points, such that the solid concentration is not representative of the in-reactor, or the outlet concentration. 2. Van Kleeck VS destruction. It is sensitive to accumulation of minerals in the reactor (which will read as a false low destruction), or precipitation (which will read as a false high). It is not as susceptible to systematic sampling issues, as dilution of mineral and organic solids are normally consistent. It is not dependent on flow-rate measurement. 3. Gas flow VS destruction. It is the least reliable, and is dependent on correct flow measurement in both liquid and gas streams, as well as correct VS inlet measurement. It is also sensitive to the assumed COD:VS ratio, and this can vary significantly (e.g., longer sludge ages normally result
634
Anaerobic Processes
in a higher COD:VS ratio). Finally, it assumes that the COD:VS ratio is the same across the digester. While it should not be used as a primary measure of VS destruction, it is a useful comparison to other terms. It is not sensitive at all on outlet flow measurement or outlet liquid-digester analysis. While VS destruction is the key indicator of performance, digester stability and health can be assessed as discussed in the previous section as either total organic-acid concentration, or a combination of organic-acid concentration and PA. Sludge and manure digesters normally have higher levels of inherent PA due to the pH rise produced by free-ammonia release during hydrolysis of proteins.
4.17.3.3 Evaluating Substrate and Microbial Properties
Methanogenic activity tests are common due to methanogenesis as a rate-limiting step, and because methane can be readily measured. In the case of methanogenic activity tests, the substrates (acetate, H2, and CO2) are direct precursors for methane, and activity may be determined from direct measurements of methane produced. This is not true for determinations of acetogenic, acidogenic, proteolytic, and hydrolytic activities where substrates are converted through several steps; thus, measurements of methane production are not sufficient to determine activity, as the methane-production rate will only reflect the slowest step of a more complex degradation process. Measurements of substrate depletion are more valuable in this type of test. Methanogenic activity testing is particularly important as methanogenesis is the final stage in any degradation process, and the slowest, and the most sensitive step. Methanogenic activity is estimated based on the initial rate of methane production during a controlled batch test. Only the initial
4.17.3.3.1 Activity testing Anaerobic activity tests are used to evaluate the performance of anaerobic sludge communities, and may be used to select an adapted sludge as inoculum, to estimate maximum applicable loading rates of certain processes or sludges, and to evaluate batch kinetic parameters. These tests can also be used to monitor possible changes in sludge activities over time due to the build up of toxic or inhibitory material or the accumulation of inert material. Activity testing cannot determine the presence or concentration of individual microbial species; however, relative activities indicate the balance of trophic groups within the community. The microbial activity of the different trophic groups determines the rate of each of the four main steps in anaerobic digestion and allows identification of the limiting step. The rate-limiting step will give information about the maximum organic load which can be applied to the system without causing a loss in its stability. Activity as identified by testing can be used together with other reactor indicators to promote stable operation. The quality of inoculum is important for prediction of degradation characteristics of novel waste materials. For example, in applications where the waste material is a complex organic solid, the hydrolysis step will limit the material available for fermentation. In applications where the waste is soluble or readily fermentable, the production of intermediates will be more rapid and a high methanogenic activity is required to balance this. For determination of activities of different trophic groups, model substrates should be used. The concentration of model substrate used in activity testing is a critical factor in the test set up. The initial concentration should be sufficiently high such that the biomass concentration is the limiting factor in the test, but sufficiently low to prevent inhibition of the microbial community as well. Model substrates used to determine the activity of the main trophic groups in anaerobic communities and recommended concentration ranges are shown in Table 5. A synthetic medium may be used in the assays to ensure that necessary nutrients/micronutrient/vitamins are available to allow optimal performance of anaerobic microorganisms. The composition of basic anaerobic (BA) medium recommended for anaerobic activity testing is given in Tables 6 and 7.
Table 6 Model substrates for determination of specific activities of trophic groups in anaerobic communities Trophic group
Substrate
Concentration range
Hydrolytic Proteolytic Acidogenic Acetogenic
Cellulose Casein Glucose Propionic acid n-butyric acid Acetic acid
1–10 g l1 1 g l1 1–2 g l1 0.5–1 g l1 0.5–1 g l1 1–2 g l1
H2/CO2 (80:20)
1 bar total
Methanogenic – acetoclastic Methanogenic – hydrogenotrophic
Table 7
Base anaerobic (BA) medium suggested for activity testing
Stock solution
Volume per l
Components (g l 1 of stock solution)
A
10 ml
B C D
2 ml 1 ml 1 ml
E
1 ml
NH4Cl, 100; NaCl, 10; MgCl2 6H2O, 10; CaCl2 2H2O, 5 K2PO4 3H2O, 200 Resazurin, 0.5 Trace metals: FeCl2 4H2O 2; H3BO3 0.05; ZnCl2 0.05; CuCl2.2H2O 0.038; MnCl2 4H2O 0.05; (NH4)6Mo7O24 4H2O, 0.05; AlCl3 0.05; CoCl2 6H2O 0.05; NiCl2 6H2O 0.092; EDTA, 0.5; conc. HCl, 1 ml; Na2SeO3 5H2O, 0.1 Vitamins: biotin, 2; folic acid, 2; pyridoxine acid, 10; riboflavin, 5; thiamine hydrochloride, 5; cyanocobalamine, 0.1; nicotinic acid, 5; P-aminobenzoic acid, 5; lipoic acid, 5; DL-pantothenic acid, 5
NaHCO3 Na2S 9H2O
2.6 g 0.5 g
From Angelidaki I and Sanders W (2004) Assessment of the anaerobic biodegradability of macropollutants. Reviews in Environmental Science and Bio/Technology 3: 117.
Anaerobic Processes
635
0.09 0.08
y = 0.2735 × −0.0191 R 2 = 0.9967
Methane (g COD)
0.07 0.06 0.05 0.04 0.03 0.02 0.01 0.00 0.00
0.05
0.10
0.15
0.20
0.25
0.30
0.35
Time (day) Figure 18 Methane generation during specific methanogenic activity assay.
linear methane-production rate is used to reduce the influence of biomass growth and adaptations or changes of the biomass characteristics. Environmental factors including substrate and nutrient concentrations and pH also vary during the tests. The methane-production curve for a specific methanogenic activity test is shown in Figure 18. This shows the slope of the curve, which in the linear region indicates the methanogenic activity. The specific methanogenic activity (SMA) is this slope divided by the VS present in the test vial. Methanogenic activity is generally higher than 0.2 g COD CH4 g VS1 d1 for digesters and industrial high-rate sludges, and may be far higher for laboratory-grown granules.
The ultimate methane yield represents the potential to recover energy during waste treatment; however, the value of energy produced is often only a small consideration in determining the feasibility of the anaerobic project. An example output from a methane potential test is shown in Figure 19. The key parameters used to indicate degradability of a complex feed are degradation extent (fd), the fraction of the substrate that may be converted to methane, and apparent first-order hydrolysis rate coefficient (khyd), an indicator of the rate at which conversion occurs. Determination of degradability parameters is critical in feasibility analysis, system design, troubleshooting, and competitive testing of inoculums. Hydrolysis is normally represented using a first-order model as discussed in Section 4.17.1 of this chapter:
4.17.3.3.2 Biological methane potential testing The biological methane potential (BMP) test is a simple batch assay used to determine the potential methane generated from anaerobic biodegradation of a mass of test substrate. In addition to potential methane (ml) generated per gram of substrate (wet, dry, and VS basis), the BMP assay is used in determination of parameters critical in process design, troubleshooting, and competitive testing of inoculums. The BMP test requires a test substrate to be mixed with a known good inoculum (containing a strong and balanced anaerobic community) in a controlled environment. BMP testing can be done at multiple scales ranging from several grams of test material up to tons (in pilot digesters). Extensive biodegradability testing of thousands of different materials in both aerobic and anaerobic conditions has been performed over the last 50 years; however, comparison of biodegradability data between studies in the literature has been limited by a lack of a common basis. Previously, factors including type of equipment, operating conditions, method of analysis, test compound, inoculum, and nutrient medium varied among studies and influenced the outcome of the batch assays (Rozzi and Remigi, 2004). However, this is improving with the publication of practical and standardized methods (e.g., activities of the IWA Anaerobic Biodegradation Activity and Inhibition (ABAI) Taskgroup; Angelidaki et al., 2009).
dS ¼ khyd S dt
ð12Þ
where S is the degradable portion of substrate, t is the incubation time, and khyd is the first-order hydrolysis rate constant. Determination of residual substrate, S, requires that the degradable fraction of the substrate is known. A simplified approach is achieved through the separation of variables and the integration of Equation (12):
ln
Pf P ¼ khyd t Pf
ð13Þ
where residual substrate at time t, is represented as the difference between the methane yield at that time P, and the ultimate methane yield Pf. Equation (13) will produce a linear curve when the degradation kinetics are of the first order, and the hydrolysis rate constant is represented by the slope of the curve. Alternatively, the first-order hydrolysis coefficient and degradability parameters can be estimated using a dynamic firstorder (single step) model. The first-order hydrolysis rate is used to estimate process-retention time and thus digester size, while the degradability fraction can be used to calculate the expected VS destruction during the process. Replicate testing
636
Anaerobic Processes 250
Methane (ml CH4 .gVS−1)
200
150
100
50
0 0
2
4
6
8
10
12
14
16
18
Time (day) Figure 19 Example output from biological methane potential (BMP) test. Error bars indicate 95% confidence errors from triplicate batches. The line indicates the model used to return key parameters.
is essential to determine repeatability and variability in the test results. More advanced modeling methods, including nonlinear parameter estimation and parameter uncertainty evaluations, are now being used to determine a multidimensional parameter surface showing the confidence region of the parameter estimations (e.g., Figure 20; Batstone et al., 2009). Batch assays are used on the basis that they are appropriate for assessing the performance or potential of full-scale anaerobic processes. It is not possible to replicate and maintain equal environmental factors such as nutrient, buffer, pH, and gas-phase conditions between the full-scale reactors and batch tests. The mode of operation (batch or continuous, mixed or plug flow) also varies between full-scale reactors systems and the standard assay (batch, no mixing). Degradability is a characteristic of test material rather than the test conditions and, indeed, degradability estimated using BMP tests are similar to the degradability performance achieved in continuous full-scale anaerobic processes; however, the first-order hydrolysis rate is influenced by test conditions and therefore is not directly comparable (Batstone et al., 2009). However, results from batch tests represent a conservative estimate of parameters needed for system-feasibility analysis and design. Environmental conditions and substrate characteristics vary between the BMP test and reactor used as an inoculum source; as a result, the inoculum is rarely optimized for the test material and significant adaptation does not occur during the batch test. Inoculum should be collected from a reactor operating on a complex feed material to provide a diverse and balanced microbial population and ensure complete breakdown of the degradable portion of the test material. The issue of inoculum to substrate ratio has been evaluated in some detail (Fernandez et al., 2001; Neves et al., 2004; Raposo et al., 2006). The inoculum to substrate ratio must be sufficient to ensure that hydrolysis is limited by surface availability or substrate concentration, rather than microbial concentration. This would typically require that the inoculum volume is greater than 50% of the test volume.
4.17.3.4 Advanced Model-Based Analysis Dynamic modeling of anaerobic systems developed reasonably quickly from simple dynamic first-order models, largely reflecting only hydrolysis (Gossett and Belser, 1982; Pavlostathis and Giraldo-Gomez, 1991; Pavlostathis and Gossett, 1986) to more complex dynamic multi-step models that include all the steps shown in Figure 1 (Costello et al., 1991; Siegrist et al., 1993). These include complex interactions such as physicochemical models, ammonia inhibition, and the production of organic acids. The wide variety of multi-step models have been largely consolidated in the IWA Anaerobic Digestion Model No. 1 (Batstone et al., 2002), which was designed to be a broadly applicable generic model of anaerobic digestion processes. This has been adapted to a number of diverse applications (Batstone et al., 2006a), including high-rate, sulfate reducing, nitrate reducing, solid and manure digestion, fermentative, and solid-phase digestion. In particular, there has been substantial effort into including anaerobic digesters in whole-plant models that largely depend on the IWA activated-sludge models (Nopens et al., 2009). This has allowed relatively easy characterization of input streams to the anaerobic digestion model – something which has been classically challenging. Dynamic modeling has a number of very practical applications, as well as enables specific areas of research. Specific practical applications include
• • •
Scenario analysis prior to major process changes – particularly with respect to particular inhibitors (Batstone and Keller, 2003). Its use to determine degradability rate and extent properties of upstream materials in situ, rather than through BMP testing (Batstone et al., 2009). Dynamic and detailed assessment of caustic dosing requirements and optimization for alkalinity addition in comparison with static analysis.
Anaerobic Processes
637
Hydrolysis rate - khyd (d−1)
0.2
0.15
0.1 0.3
0.35
0.4
0.45
Degradability fraction (f d) Figure 20 Surface estimation of degradability parameters – for two-parameter estimates on BMP tests. The 95% two-parameter region is represented by the line, while confidence intervals represent uncorrelated, linear estimates of parameter confidence.
4.17.4 Future Applications of Anaerobic Digestion 4.17.4.1 Sewage Treatment and Nutrient Removal High-rate anaerobic digestion has now been evaluated extensively for treatment of low concentration and domestic sewage (Barber and Stuckey, 1999; Foresti et al., 2006; Seghezzo et al., 1998). It is generally suitable for removal of bulk organics, and to remove some pathogens (though not to standards). In comparison with conventional aerobic treatment, high-rate anaerobic treatment of domestic wastewater is relatively low in capital costs and distinctly lower in operating cost, does not require aeration energy (and can produce energy as methane), and is relatively low maintenance. The main disadvantages are (1) low removal of nutrients; (2) relatively poor removal of organics (60–90%), and (3) release of methane dissolved either in the liquid or directly from the digester surface. While it is a suitable alternative to no treatment, high-rate anaerobic treatment cannot produce an effluent suitable for direct discharge to inland watercourses, with minimal environmental impact. Phosphorus can be removed (and recovered by precipitation) and methane can be captured during treatment, or removed in aerobic or other post treatment, which can also be used to remove residual organics and pathogens (Barber and Stuckey, 1999; Chernicharo, 2006; Foresti et al., 2006; Seghezzo et al., 1998). However, the key issue is removal of nitrogen. Currently, the main method of nitrogen removal (nitrification–denitrification) requires carbon for denitrification, and anaerobic processes remove carbon. Partial nitration to nitrite reduces the carbon load. While there are processes that can remove ammonia, such as the biological process anammox (using nitrite as electron acceptor to remove ammonia), stripping, and adsorption (Foresti et al., 2006), most of these are applicable at higher ammonia concentration. The anammox process is probably the most promising for low-concentration ammonia removal, and ammonia concentration is possible, through both adsorption and
membrane processes. While there are challenges, anaerobic processes both at low concentration and in solids digesters offer sustainable, low-cost alternatives to conventional aerobic processes.
4.17.4.2 Nutrient Recovery Phosphorus and nitrogen are key components in many organic sources, including biosolids and manure. Phosphorus in particular is a key resource, since it is a nonrenewable resource. World reserves are substantial, and depletion of ready resources is not expected until later this century (Isherwood, 2000). However, demand is also increasing substantially, and this has led to dramatic price increases, particularly through 2008. Alternative, sustainable, and low-cost alternatives are therefore highly desirable, particularly where national reserves are low, or supply restricted. Anaerobic digestion is already used to stabilize organic biosolids and manure for agricultural applications. Anaerobically stabilized organic biosolid is an excellent fertilizer, generally with comparable impacts (per unit nitrogen) to mineral fertilizer (Warne, 2009), with the added benefits of carbon, water, and trace-compound addition. However, stabilized organic solids are bulky, with nitrogen content between 3 and 10% (dry basis), or 0.3 and 2% on a wet basis. Transport costs are normally in the same cost order of magnitude as the value of the nutrients, meaning that beneficial use is driven by disposal costs, rather than the value of the nutrients. However, there is substantial scope for recovery and concentration of both nitrogen and phosphorus (De-Bashan and Bashan, 2004). Aerobic microbes can be used to accumulate phosphorus via enhanced biological phosphorus removal (EBPR), which results in a high phosphorus WAS stream. Anaerobic digestion plays a key component in phosphorus recovery, as it can be used to re-mobilize ammonia and phosphorus, which can then be recovered as precipitated phosphorus. Struvite ((MgNH4PO4 6H2O) is probably the
638
Anaerobic Processes
best mineral, as magnesium is low cost, and struvite precipitation also allows for nitrogen recovery (Munch and Barr, 2001). At the moment, struvite precipitation is largely used as a phosphorus removal method, than as a recovery technique, but this is likely to change in the future, with anaerobic digestion the major component to mobilize and recover accumulated phosphorus and nitrogen.
4.17.4.3 Future Applications in Energy Generation and Transport Until now, anaerobic digestion has been mainly industrially applied in developed nations for large-scale organic solid stabilization and destruction. Economics are normally driven by solid destruction and stabilization rather than energy production. Energy is either utilized to produce low-quality heat, or in co-generation engines. Economies of scale and maintenance requirements mean that the optimal economic size of co-generation engines is approximately 500 kW. While newer technologies such as microturbines are making anaerobic digestion more attractive at smaller scale, for electricity production, anaerobic digestion is still a large-scale proposition. It is also widely applied at very small scales across Asia, South America, and Africa and the Middle East directly for methane generation and utilization. Methane is used directly and effectively as a natural-gas replacement. In fact, anaerobic digestion is one of the only renewable energy technologies which is fully mature, completely scalable, and generates an energy product that can be stored as produced. There is a particular application in intensive agriculture and food processing, where there is a need for water treatment and energy, and much of the organics are being emitted as methane, which is currently lost (with a consequent greenhouse-gas impact). It is very likely that anaerobic digestion will be implemented increasingly at smaller scale once the technology is standardized further as a partner to other scalable renewable options such as wind and photovoltaic solar cells. The applications of small-to-medium-scale anaerobic digestion are not only limited to methanogenesis. Fermentation can also be used to produce a number of alternative products, including organic acids, alcohols, and hydrogen. While we do not yet have the knowledge to fully direct mixed-culture fermentation to specific products (Rodrı´guez et al., 2006), this is an exciting research area that will likely challenge classical pure-culture fermentation on cost, conversion, and specificity. It has been further enhanced by the application of bioelectrochemical systems, which have the capacity to utilize electrical current to drive full conversion of low-value carbon feedstocks to valuable products such as hydrogen, organic acids, and alcohols.
References Angelidaki I, Alves M, Bolzonella D, et al. (2009) Defining the biomethane potential (BMP) of solid organic wastes and energy crops: A proposed protocol for batch assays. Water Science and Technology 59: 927--934. Aquino SF, Chernicharo CAL, Soares H, Takemoto SY, and Vazoller RF (2008) Methodologies for determining the bioavailability and biodegradability of sludges. Environmental Technology 29: 855--862.
Barber WP and Stuckey DC (1999) The use of the anaerobic baffled reactor (ABR) for wastewater treatment: A review. Water Research 33: 1559--1578. Batstone DJ (2009) Towards a generalised physicochemical modelling framework. Reviews in Environmental Science and Biotechnology 8: 113--114. Batstone DJ and Keller J (2003) Industrial applications of the IWA Anaerobic Digestion Model No. 1 (ADM1). Water Science and Technology 47: 199--206. Batstone DJ, Keller J, Angelidaki I, et al. (2002) Anaerobic Digestion Model No. 1 (ADM1), IWA Task Group for Mathematical Modelling of Anaerobic Digestion Processes. London: IWA Publishing. Batstone DJ, Keller J, and Blackall LL (2004) The influence of substrate kinetics on the microbial community structure in granular anaerobic biomass. Water Research 38: 1390--1404. Batstone DJ, Keller J, and Steyer JP (2006a) A review of ADM1 extensions, applications, and analysis: 2002–2005. Water Science and Technology 54: 1--10. Batstone DJ, Picioreanu C, and van Loosdrecht MCM (2006b) Multidimensional modelling to investigate interspecies hydrogen transfer in anaerobic biofilms. Water Research 40: 3099--3108. Batstone DJ, Pind PF, and Angelidaki I (2003) Kinetics of thermophilic, anaerobic oxidation of straight and branched chain butyrate and valerate. Biotechnology and Bioengineering 84: 195--204. Batstone DJ, Tait S, and Starrenburg D (2009) Estimation of hydrolysis parameters in full-scale anerobic digesters. Biotechnology and Bioengineering 102: 1513--1520. Boe K, Batstone DJ, and Angelidaki I (2007) An innovative online VFA monitoring system for the anerobic process, based on headspace gas chromatography. Biotechnology and Bioengineering 96: 712--721. Boone DR, Johnson RL, and Liu Y (1989) Diffusion of interspecies electron carriers H2 and formate in methanogenic ecosystems and its implications in the measurement of Km for H2 or formate uptake. Applied and Environmental Microbiology 55: 1735--1741. Boone DR, Whitman WB, and Rouviere P (1993) Diversity and taxonomy of methanogens. In: Ferry JG (ed.) Methanogenesis: Ecology, Physiology, Biochemistry and Genetics, pp. 35--80. New York: Chapman and Hall. Bryant M, Wolin E, Wolin M, and Wolfe R (1967) Methanobacillus omelianskii, a symbiotic association of two species of bacteria. Archives of Microbiology 59: 20--31. Chernicharo C (2006) Post-treatment options for the anaerobic treatment of domestic wastewater. Reviews in Environmental Science and Bio/Technology 5: 73--92. Costello DJ, Greenfield PF, and Lee PL (1991) Dynamic modelling of a single-stage high-rate anaerobic reactor – I. Model development. Water Research 25: 859--871. De-Bashan LE and Bashan Y (2004) Recent advances in removing phosphorus from wastewater and its future use as fertilizer (1997–2003). Water Research 38: 4222--4246. de Bok FAM, Plugge CM, and Stams AJM (2004) Interspecies electron transfer in methanogenic propionate degrading consortia. Water Research 38: 1368--1375. Dwyer J, Starrenburg D, Tait S, Barr K, Batstone DJ, and Lant P (2008) Decreasing activated sludge thermal hydrolysis temperature reduces product colour, without decreasing degradability. Water Research 42: 4699--4709. Eastman JA and Ferguson JF (1981) Solubilization of particulate organic carbon during the acid phase of anaerobic digestion. Journal of the Water Pollution Control Federation 53: 352--366. Ekama GA, So¨temann SW, and Wentzel MC (2007) Biodegradability of activated sludge organics under anaerobic conditions. Water Research 41: 244--252. Fang HHP, Chui HK, Li YY, and Chen T (1994) Performance and granule characteristics of UASB process treating wastewater with hydrolyzed proteins. Water Science and Technology 30: 55--63. Feijoo G, Soto M, Mendez R, and Lema JM (1995) Sodium inhibition in the anaerobicdigestion process – antagonism and adaptation phenomena. Enzyme and Microbial Technology 17: 180--188. Fernandez B, Porrier P, and Chamy R (2001) Effect of inoculum substrate ratio on the start-up of solid waste anaerobic digesters. Water Science and Technology 44: 103--108. Ferry JG (1993) Methanogenesis: Ecology, Physiology, Biochemistry and Genetics. New York: Chapman and Hall. Finnerty WR (1988) B-oxidation of fatty acids. In: Ratledge C and Wilkinson SG (eds.) Microbial Lipids, vol. 2, pp. 528--566. London: Academic Press. Foresti E, Zaiat M, and Vallero M (2006) Anaerobic processes as the core technology for sustainable domestic wastewater treatment: Consolidated applications, new trends, perspectives, and challenges. Reviews in Environmental Science and Bio/ Technology 5: 3--19. Ge H, Jensen PD, and Batstone DJ (2010) Pre-treatment mechanisms during thermophilic–mesophilic temperature phased anaerobic digestion of primary sludge. Water Research 44: 123--130.
Anaerobic Processes
Gossett JM and Belser RL (1982) Anaerobic digestion of waste activated sludge. ASCE: Journal of Environmental Engineering 108: 1101--1120. Gujer W and Zehnder AJB (1983) Conversion processes in anaerobic digestion. Water Science and Technology 15: 127--167. Hashimoto AG (1986) Pretreatment of wheat straw for fermentation to methane. Biotechnology and Bioengineering 28: 1857--1866. Henze M, Gujer W, Mino T, and van Loosdrecht M (2000) Activated Sludge Models ASM1, ASM2, ASM2d, and ASM3. London: IWA Publishing. Humphrey AE (1979) The hydrolysis of cellulosic materials to useful products. In: Brown RD and Jurasek L (eds.) Hydrolysis of Cellulose: Mechanisms of Enzymatic and Acid Catalysis, vol. 181, ch. 2, pp. 25–53. Washington, DC: American Chemical Society Hwu CS, Donlon B, and Lettinga G (1996) Comparative toxicity of long chain fatty acid to anaerobic sludges from various origins. Water Science and Technology 34: 351--358. Isherwood KF (2000) Mineral Fertilizer Use and the Environment. Paris: International Fertilizer Industry Association/United Nations Environment Programme. Jain S, Lala AK, Bhatia SK, and Kudchadker AP (1992) Modelling of hydrolysis controlled anaerobic digestion. Journal of Chemical Technology and Biotechnology 53: 337--344. Karakashev D, Batstone DJ, and Angelidaki I (2005) Influence of environmental conditions on methanogenic compositions in anaerobic biogas reactors. Applied and Environmental Microbiology 71: 331--338. Karakashev D, Batstone DJ, Trably E, and Angelidaki I (2006) Acetate oxidation is the dominant methanogenic pathway from acetate in the absence of Methanosaetaceae. Applied and Environmental Microbiology 72: 5138--5141. Madigan MT, Martinko JM, and Brock TD (2009) Brock Biology of Microorganisms. San Francisco, CA: Pearson. Martinelle M and Hult K (1994) Kinetics of triglyceride lipases. In: Woolley P and Petersen S (eds.) Lipases: Their Structure, Biochemistry and Application, 363pp. Cambridge: Cambridge University Press. McInerney MJ (1988) Anaerobic hydrolysis and fermentation of fats and proteins. In: Zehnder JBA (ed.) Biology of Anaerobic Microorganisms, vol. 1, pp. 373--416. New York: Wiley. McInerney MJ, Bryant MP, Hespell RB, and Costerton JW (1981) Syntrophomonas wolfei gen. nov. sp. nov., an anaerobic, syntrophic, fatty acid oxidising bacterium. Applied and Environmental Microbiology 41: 1029--1039. Moosbrugger RE, Loewenthal RE, and Marais GvR (1990) Pelletisation in a UASB system with protein (casein) as substrate. Water SA 16: 171--178. Mosey FE (1983) Mathematical modelling of the anaerobic digestion process: Regulatory mechanisms for the formation of short-chain volatile acids from glucose. Water Science and Technology 15: 209--232. Munch EV and Barr K (2001) Controlled struvite crystallisation for removing phosphorus from anaerobic digester sidestreams. Water Research 35: 151--159. Musvoto E, Wentzel M, Loewenthal R, and Ekama G (2000a) Integrated chemical– physical processes modelling – I. Development of a kinetic-based model for mixed weak acid/base systems. Water Research 34: 1857--1867. Neves L, Oliveira R, and Alves MM (2004) Influence of inoculum activity on the biomethanization of a kitchen waste under different waste/inoculum ratios. Process Biochemistry 39: 2019--2024. Nopens I, Batstone DJ, Copp JB, et al. (2009) An ASM/ADM model interface for dynamic plant-wide simulation. Water Research 43: 1913--1923. O’Rourke JT (1968) Kinetics of Anaerobic Treatment at Reduced Temperatures. PhD Thesis, Stanford University, Stanford, CA, USA. Park C, Abu-Orf MM, and Novak JT (2006) The digestibility of waste activated sludges. Water Environment Research 78: 59--68. Patterson-Curtis SI and Johnson EA (1989) Regulation of neurotoxin and protease formation in Clostridium botulinum Okra-B and Hall-A by arginine. Applied and Environmental Microbiology 55: 1544--1548. Pauss A and Guiot SR (1993) Hydrogen monitoring in anaerobic sludge bed reactors at various hydraulic regimes and loading rates. Water Environment Research 65: 276--280. Pavlostathis SG and Giraldo-Gomez E (1991) Kinetics of anaerobic treatment: A critical review. Critical Reviews in Environmental Control 21: 411--490. Pavlostathis SG and Gossett JM (1986) A kinetic model for anaerobic digestion of biological sludge. Biotechnology and Bioengineering 28: 1519--1530. Pind PF, Angelidaki I, and Ahring BK (2003) A new VFA sensor technique for anaerobic reactor systems. Biotechnology and Bioengineering 82: 54--61. Ramsay IR (1997) Modelling and Control of High-Rate Anaerobic Wastewater Treatment Systems. PhD Thesis, University of Queensland, Brisbane, QLD, Australia. Ramsay IR and Pullammanappallil PC (2001) Protein degradation during anaerobic wastewater treatment: Derivation of stoichiometry. Biodegradation 12: 247--257.
639
Ramsay IR and Pullammanappallil PC (2005) Full-scale application of a dynamic model for high-rate anaerobic wastewater treatment systems. Journal of Environmental Engineering 131: 1030--1036. Rao AG, Reddy TSK, Prakash SS, et al. (2008) Biomethanation of poultry litter leachate in UASB reactor coupled with ammonia stripper for enhancement of overall performance. Bioresource Technology 99: 8679--8684. Raposo F, Banks CJ, Siegert I, Heaven S, and Boria R (2006) Influence of inoculum to substrate ratio on the biochemical methane potential of maize in batch tests. Process Biochemistry 41: 1444--1450. Ratledge C (1994) Biodegradation of oils, fats and fatty acids. In: Ratledge C (ed.) Biochemistry of Microbial Degredation, vol. 1, 590pp. Dordrecht: Kluwer. Rebac S, Ruskova J, Gerbens S, vanLier JB, Stams AJM, and Lettinga G (1995) Highrate anaerobic treatment of wastewater under psychrophilic conditions. Journal of Fermentation and Bioengineering 80: 499--506. Reguera G, McCarthy KD, Mehta T, Nicoll JS, Tuominen MT, and Lovley DR (2005) Extracellular electron transfer via microbial nanowires. Nature 435: 1098--1101. Ren N, Wan B, and JuChang H (1997) Ethanol-type fermentation from carbohydrate in high rate acidogenic reactor. Biotechnology and Bioengineering 54: 428--433. Rodrı´guez J, Kleerebezem R, Lema JM, and van Loosdrecht MCM (2006) Modeling product formation in anaerobic mixed culture fermentations. Biotechnology and Bioengineering 93: 592--606. Roy F, Albagnac G, and Samain E (1985) Influence of calcium addition on growth of highly purified syntrophic cultures degrading long-chain fatty acids. Applied Environmental Microbiology 49: 702--705. Rozzi A and Remigi E (2004) Methods of assessing microbial activity and inhibition under anaerobic conditions: A literature review. Reviews in Environmental Science and Bio/Technology 3: 93--115. Schink B (1997) Energetics of syntrophic cooperation in methanogenic degradation. Microbiology and Molecular Biology Reviews 61: 262--280. Seghezzo L, Zeeman G, van Lier JB, Hamelers HVM, and Lettinga G (1998) A review: The anaerobic treatment of sewage in UASB and EGSB reactors. Bioresource Technology 65: 175--190. Siegrist H, Renggli D, and Gujer W (1993) Mathematical modelling of anaerobic mesophilic sewage sludge treatment. Water Science and Technology 27: 25--36. Siegrist H, Vogt D, Garcia-Heras J, and Gujer W (2002) Mathematical model for meso and thermophilic anaerobic sewage sludge digestion. Environmental Science and Technology 36: 1113--1123. Speece RE (2008) Anaerobic Biotechnology and Odor/Corrosion Control for Municipalities and Industries. Nashville, TN: Archae Press. Stams AJM (1994) Metabolic Interactions between anaerobic-bacteria in methanogenic environments. Antonie Van Leeuwenhoek International Journal of General and Molecular Microbiology 66: 271--294. Stams AJM and Plugge CM (1994) Occurrence and function of the acetyl-CoA cleavage pathway in a syntrophic propionate oxidising bacterium. In: Drake HL (ed.) Acetogenesis, pp. 557--630. New York: Chapman and Hall. Steyer JP, Bernard O, Batstone DJ, and Angelidaki I (2006) Lessons learnt from 15 years of ICA in anaerobic digesters. Water Science and Technology 53: 25--33. Stumm W and Morgan JJ (1996) Aquatic Chemistry: Chemical Equilibria and Rates in Natural Waters. New York: Wiley. Tchobanoglous G, Burton F, and Stensel H (2003) Metcalf and Eddy Inc. Wastewater Engineering, Treatment and Reuse. New York, NY: McGraw-Hill. Temudo MF, Muyzer G, Kleerebezem R, and van Loosdrecht MCM (2008) Diversity of microbial communities in open mixed culture fermentations: Impact of the pH and carbon source. Applied Microbiology and Biotechnology 80: 1121--1130. Thiele JH and Zeikus JG (1988) Interactions between hydrogen and formate producing bacteria and methanogens during anaerobic digestion. In: Erickson CE and DanielYee-Chak-Fung (eds.) Handbook on Anaerobic Fermentations, pp. 537--595. New York, NY: Dekker. Tong Z and McCarty P (1991) Microbial hydrolysis of lignocellulosic materials. In: Isaacson R (ed.) Methane from Community Wastes, pp. 61--100. London: Elsevier. Van Langerak E and Hamelers H (1997) Influent calcium removal by crystallization reusing anaerobic effluent alkalinity. Water Science Technology 36: 341--348. Van Lier JB (2008) High-rate anaerobic wastewater treatment: Diversifying from endof-the-pipe treatment to resource-oriented conversion techniques. Water Science and Technology 57: 1137--1148. Van Lier JB, Sanz Martin JL, and Lettinga G (1996) Effect of temperature on the anaerobic thermophilic conversion of volatile fatty acids by dispersed and granular sludge. Water Resources 30(1): 199--207. Vavilin VA, Rytov SV, and Lokshina LYa (1996) A description of hydrolysis kinetics in anaerobic degradation of particulate organic matter. Bioresource Technology 56: 229--237.
4.18 Microbial Fuel Cells B Virdis, S Freguia, RA Rozendal, K Rabaey, Z Yuan, and J Keller, The University of Queensland, Brisbane, QLD, Australia & 2011 Elsevier B.V. All rights reserved.
4.18.1 4.18.1.1 4.18.1.2 4.18.1.3 4.18.2 4.18.3 4.18.4 4.18.4.1 4.18.4.2 4.18.4.3 4.18.4.4 4.18.5 4.18.5.1 4.18.5.2 4.18.5.3 4.18.5.4 4.18.6 4.18.7 4.18.8 4.18.9 4.18.9.1 4.18.9.2 4.18.9.3 4.18.9.4 4.18.10 4.18.10.1 4.18.10.2 4.18.10.3 4.18.10.4 4.18.10.5 4.18.11 References
Resource Recovery from Wastewater Water Recovery Nutrient Recovery Energy Recovery Microbial Fuel Cells Thermodynamics of Microbial Fuel Cells Factors Determining the Decrease of Cell Voltage Losses due to Mass-Transfer Limitation Losses due to Bacterial Metabolic Kinetics Losses due to Electron Transfer to (and from) the Electrode Losses due to the Resistance of the Electrolytes (Including the Ion-Exchange Membrane) and of the Electrical Interconnection to the Charges Flow Materials and Architectures Design Compartment Separation Electrodes Cathodic Compartment Electrochemically Active Microorganisms and Extracellular Electron Transfer Oxidative Processes Reductive Processes Challenges toward Improving MFC Efficiency Minimizing Electrode-Potential Losses Respiration, Fermentation, and Methanogenesis Reducing pH Gradients Wastewater and Electrode Resistance Opportunities for Bioelectrochemical Systems Wastewater Treatment Nitrogen Removal Bioremediation H2 Production Bioelectrochemical Production of Value-Added Chemicals Outlook
4.18.1 Resource Recovery from Wastewater Fossil fuel exploitation has significantly affected the economic growth of developed countries within the past century. The world energy-consumption rate is projected to double from 13.5 TW (1 TW ¼ 1012 W) in 2001 to 27 TW by the year 2050 and to triple to 43 TW by 2100 (Lewis and Nocera, 2006). Although the rise in prices of liquid fuels (e.g., crude oil, natural gas plant liquid, biofuels, oil shale, and bitumen) and natural gas is expected to rationalize energy demand, world energy consumption is still projected to increase due to continuing rapid economic growth and expanding population, particularly in the developing countries. Fossil fuels are predicted to remain the dominant sources of primary energy, accounting for close to 83% of the overall increase in energy demand between 2004 and 2030 (Figure 1). Increasing awareness of the possible anthropogenic effects on climate change, in combination with the instability of the
641 641 642 643 644 644 645 647 647 648 649 649 649 650 650 651 651 653 654 656 657 657 657 658 658 659 660 660 660 660 661 661
fossil fuel market, is motivating political consciousness to reduce greenhouse-gas emissions and to promote renewable energy. The greatest challenge in the future lies in catering to the world’s growing energy demands while simultaneously reducing emission of greenhouse gases. This is certainly predicted to provide serious challenges for fossil-fuel-based economies (Logan, 2008). Nuclear fission alone does not represent a feasible alternative, as known uranium reserves would be depleted within a few decades, not considering the environmental damage caused by the mining and disposal of radioactive material (Lewis and Nocera, 2006). Solar energy is an attractive energy source as it is both renewable and available in large amounts (Seboldt, 2004); however, a society completely dependent on solar energy is not realistic for the short term due to technological and economical difficulties. Other renewable energy technologies must be developed in conjunction with solar energy. About 200 TW of the 170 000 TW solar-radiation flux is continuously transformed
641
642
Microbial Fuel Cells 250 History
Quadrillion kJ
200
Projection
Oil Coal Natural gas Renewable Nuclear
150
100
50
0 1990
2000
2010
2020
2030
Figure 1 World energy use detailed by fuel type, 1990–2030. Data from Energy Information Administration (EIA).
into wind power, whereas 100 TW is stored as biomass through photosynthesis, and 6 TW is transformed into hydropower through the water cycle (Niele, 2005). These indirect forms of solar energy are already exploited to some extent for electricity production through wind turbines, biomass gasification/combustion, and hydroelectric dams. Nevertheless, the extent of this exploitation must be further developed in the future in parallel to elevating societal energy demands. Securing water resources to match increasing demand is also becoming challenging. Global trends such as urbanization and migration, often combined with frequent drought periods, even in areas traditionally rich in water resources, have increased the demand for water, food, and energy, and put at risk the sustainability of current living standards. The pressure of water availability has also affected other waterconsuming sectors, such as public water supply, agriculture, industry, and, of course, power generation. In this global picture of fading resources, we can therefore no longer afford to waste any potential sources of all these three key resources. For example, domestic and industrial wastewaters are ubiquitous and represent a potential source of energy, water, and nutrients. The development of technologies capable of simultaneously recovering energy, water, and nutrients from wastewater is crucial to resource management in the future.
4.18.1.1 Water Recovery Wastewater represents a valuable recyclable water resource. Although containing compounds dangerous to public health and to the environment (e.g., pathogens, chemicals, organics, and nutrients), at least 99.9% of wastewater is in fact water and, as such, it should by no means be considered as waste. Engineered technologies for the reintroduction of treated wastewater to water-supply grids appear to be an essential
priority, given the increasingly limited water resources in both quantity and quality. Wastewater treatment plants are a crucial part of the overall water-recycle process, being an important pre-treatment step for the advanced treatment processes, which are currently almost nonexistent worldwide, but which can generate water qualities suitable for reuse even in potable water applications. Wastewater-treatment processes aim to reduce the relevant concentration of pollutants by means of separation, destruction, and disinfection (Tchobanoglous et al., 2003). The efforts into improving the quality of wastewater-treatmentplant effluents have achieved levels of pollutant elimination well beyond the standards of environmental protection. An activated sludge treatment can reduce the influent biological oxygen demand (BOD) concentration from 4300 mg l1 to o5 mg l1 when upgraded for biological nutrient removal, while reducing the influent total nitrogen (N) concentration from 460 mg l1 to o3 mg l1; and influent phosphorus (P) concentrations from 412 mg l1 to o1 mg l1. (BOD is a measure of the concentration of biodegradable material present in wastewater expressed as the amount of oxygen consumed by microorganisms in breaking down the organic matter during a certain period of time. It normally represents a fraction of the chemical oxygen demand (COD), which is the total oxygen consumption consumed during chemical breakdown of organic and inorganic matter.) Further disinfection treatment can achieve up to 99.9999% removal of bacteria where membrane filtration is used (Foley and Keller, 2008). Advanced water treatments (AWTs) can further improve the efficacy of the disinfection process, by removing recalcitrant organics that are not metabolized in the biological nutrientremoval process and by reducing the content of total dissolved solids. AWT consists of a multi-barrier system against various acute and chronic risk factors, such as micropollutants and pathogens, that remain even after regular wastewater
Microbial Fuel Cells
treatment, and prior to the addition to ground or surface water for reuse. The most common forms of AWT are microfiltration and reverse osmosis, which can be followed by advanced oxidation processes (ozone and H2O2/ultraviolet (UV)) to remove recalcitrant contaminants. If an energy-recovery process is also included, the wastewater can be, for example, initially treated through anaerobic digestion, which would produce biogas that can power gas turbines for electricity generation. The effluent would then require further aerobic polishing to remove the slower biodegradable material, while achieving drinking water standards would require AWT. The water can subsequently be collected in the environment (e.g., in a dam) where time and environmental buffers will ensure that the higher-quality standards are met, even prior to any treatment processes already in place to produce safe drinking water. In this scheme, a large fraction of the wastewater is thus recovered as clean water to be reused for domestic, agricultural, and industrial purposes.
4.18.1.2 Nutrient Recovery For many years, wastewater treatment methods have been improved to achieve environmental protection from nutrient overload in receiving water bodies. Removal of carbon, nitrogen, and phosphorus from wastewater requires large amounts of energy and produces potentially useful resources of minerals and water that are normally disposed of. The most important minerals for living organisms are considered to be nitrogen and phosphorus, although potassium and sulfur should also be included as essential. Nitrogen and phosphorus for agriculture are produced from natural resources. Phosphorus is currently entirely derived from highly geographically concentrated geological reserves (mainly in North Africa, USA, China, and Russia). Phosphate rocks are finite nonrenewable resources and are therefore limited, with an estimated 50–100 years until depletion is reached under current extraction rates (Larsen et al., 2007). Moreover, extraction and production of good-quality phosphorus require energy, and produces a waste (the production of 1 kg of phosphorus produces up to 2 kg of gypsum, inclusive of heavy metals and radioactive elements). It is thus essential that phosphate is recovered efficiently in the future. Recycling of phosphorus contained in sewage is currently very limited, even though several techniques are available to incorporate it into the excess activated sludge. Nitrogen in fertilizers is almost completely supplied by atmospheric N2, which makes the source virtually infinite. However, to be accessible to living organisms, atmospheric nitrogen has to be in the form of ammonia or nitrate. Industrial processes for the conversion of N2 gas to ammonia require a large energy investment using the Haber–Bosch process (10.3 kW h1 kg1 nitrogen produced; Maurer et al., 2003). Additional energy is invested to obtain the opposite process during wastewater-treatment processes to achieve low nitrogen levels in the treated effluent. With this in consideration, wastewater treatment for nitrogen removal can be regarded as an indirect and inefficient method for nitrogen recovery (recycling over the atmosphere) since it engineers biological nitrification and denitrification to N2 gas that is
643
returned to the atmosphere. Therefore, any direct nitrogen recovery process will have to be more energy efficient than the indirect route via the atmosphere to be environmental favorable. As pointed out by Maurer et al. (2003), direct recovery of nitrogen-rich wastewater into reusable forms can be more sustainable than indirect recovery by biological nitrification/ denitrification and ammonia production through the Haber– Bosch process, as the total net amount of energy required is lower. For instance, the energy demand for nitrification and denitrification, together with ammonia production through the Haber–Bosch process, would be 42.8 kW h1 kg1 nitrogen using a denitrification process with methanol addition; 25 kW h1 kg1 nitrogen in case of a preanoxic denitrification system; and 17.8 kW h1 kg1 N for nitrogen removal through the Sharon–Anammox process (Maurer et al., 2003). This considerable energy requirement for indirect nitrogen recycling makes some direct recovery techniques such as thermal volume reduction of urine (requiring about 8.1–9.4 kW h1 kg1 N), or even struvite production, economically and environmentally interesting (Maurer et al., 2003). (Struvite is a phosphate mineral with formula NH4MgPO4 6H2O. Its production requires 28.3 kW h1 kg1 N (Maurer et al., 2003), which is higher than the energy demand for alternative N recovery processes like thermal volume reduction of urine. However, together with nitrogen, phosphorus is also recovered (struvite contains about 2.2 kg phosphorus per each kilogram of nitrogen).)
4.18.1.3 Energy Recovery Many recovery processes can provide bioenergy or valuable chemicals from relatively concentrated biomass streams, such as from wood and agricultural by-products (Hatti-Kaul et al., 2007; Petrus and Noordermeer, 2006; Ragauskas et al., 2006; van Wyk, 2001). Yet, not many conversion processes exist for energy and chemical production from diluted aqueous streams, such as industrial, agricultural, and municipal wastewater. Wastewater contains significant amounts of renewable energy in the form of chemical bonds. For example, domestic wastewater could potentially yield energy up to 2.2 kW h1 m3. (This is considering the energy content of glucose as 4.4 kW h1 kg1 COD, and a wastewater with 500 mg COD l1.) If properly recovered, the chemical energy daily wasted with sewage can potentially cover up to 7% of the energy consumption used for residential purposes in developed countries. (This is assuming an energy recovery of 1.2 kW h1 kg1 COD and considering a total residential energy consumption of 649.8 kg of oil equivalent capita1, equal to 7556 kW h1 capita1, and a total water withdrawal of 948 m3 yr1 capita1, assuming that it all ends up in sewage. Energy-consumption data include all energy used for activities by households except for transportation. Data on energy and water withdrawal are available at the World Research Institute.) This figure is expected to be much higher in developing countries, since the energy consumption tends to be far lower. Even though technologies such as anaerobic digestion have been long known and implemented for many years to recover energy from wastewater, the activated sludge process is by far the most widely applied process for wastewater treatment. The
644
Microbial Fuel Cells
process relies upon aeration of the wastewater, which allows microorganisms to convert the available organics into carbon dioxide. The solids are then separated from the treated water through sedimentation. Although this process yields valuable purified water, the high energy requirements for aeration, typically 0.5 kW h1 m3 treated water (Rabaey and Verstraete, 2005), makes anaerobic processes far more energy efficient than aerobic treatment. Methanogenic anaerobic digestion of organics has been shown to be advantageous over aerobic processes due to its high organic removal rates, low sludge production, low energy inputs, and, ultimately, for its energy production (Angenent et al., 2004). The methane produced thus has traditionally been used as the on-site fuel source for heat/electrical applications, or used to power gas turbines, with net energy efficiencies up to 35–40%. More recently, methane has also been converted into other products by catalytic conversion to syngas, a mixture of hydrogen and carbon monoxide, or into methanol for use in production of biodiesel (Angenent et al., 2004). However, whether methane is used in a gas turbine or to produce syngas, it is necessary to purify the biogas from impurities such as hydrogen sulfide also produced during anaerobic digestion, which therefore equates to an additional treatment process. Dark fermentation represents an alternative to biological methane production, which shares with it much of the same process reactions involved, except that during dark fermentation, the hydrogen-metabolizing organisms (methanogens) are inhibited through heat treatment of the initial inoculum while retaining only spore-forming fermenting bacteria in which hydrogen-forming bacteria are included. However, due to the limitations imposed by the thermodynamics of hydrogen formation through the hydrogenase reaction, the conversion yields of the total electron equivalents present as carbohydrate in wastewater does not normally exceed B15% (Angenent et al., 2004). As such, the process appears less appealing in comparison with the more reliable and mature biological methane production. Microbial fuel cells (MFCs) have been gaining increasing attention in recent times as devices able to produce electric power while simultaneously treating industrial, agricultural, or municipal wastewater (Rozendal et al., 2008a). Compared to treatment technologies, MFCs have the advantage of being able to theoretically achieve efficiencies. The underlining rationale is that fuel cells do not use heat as an intermediate form of energy for electricity production. As such, the process efficiency is not limited by the Carnot cycle, according to which, for a reversible process, the theoretical maximum conversion efficiency of heat to work is determined by the absolute temperature Th (K) of the process and the absolute temperature Tc (K) of the cold sink (i.e., the environment):
Zideal ¼ 1
Tc Th
ð1Þ
As heat-resisting properties of construction materials are limited to a certain maximal temperature, the theorem implies that the overall yield of combustion processes is usually no higher than 35–45% (Carnot, 1824). Since fuel cells do not operate on a thermal cycle, they are not constrained to thermodynamic limitations such as the Carnot’s theorem.
Therefore, they can theoretically convert the entire free energy of the fuel oxidation into electric energy (Schroder and Harnisch, 2009).
4.18.2 Microbial Fuel Cells Although the existence of a bioelectrical phenomenon was first observed by Italian physicist Luigi Galvani in 1790 (Piccolino, 1997), the principle that microorganisms could generate voltage and current was put forth by Michael Cresse Potter, a professor of botany at the University of Durham, UK, at the beginning of the twentieth century (Potter, 1911). This occurred a few years earlier than the discovery of the activated sludge process and shortly after the invention of the Imhoff tank, an early form of anaerobic digester. In 1931, Barnett Cohen confirmed Potter’s observations reporting a stacked biological fuel cell delivering 35 V at a current of 2 mA (Cohen, 1931). However, it was not until the US National Aeronautics and Space Administration (NASA) became interested in exploiting opportunities for recycling organic wastes into electricity during long space flights that MFCs regained popularity, and by the year 1963, were already commercially available as a power supply for small electrical devices (Shukla et al., 2004). Despite these early successes, the rapid advancement of alternative technologies, such as solar photovoltaic systems, and the fact that the complexity of the underlying biochemical processes became more evident, MFCs suffered an inevitable setback. However, the growing awareness to reduce society’s dependency on fossil fuel and the emerging environmental consequences of their usage has triggered the revival of MFC research in the last 10–15 years. In an MFC, the chemical energy contained in soluble organic molecules, such as carbohydrates and volatile fatty acids (VFAs), can be directly recovered as electric energy. MFCs are galvanic cells that couple the oxidation of an electron donor at an anode with the reduction of an electron acceptor at a higher redox potential at the cathode. Power output is generated as the overall reaction is exergonic. MFCs are the most extensively described bioelectrochemical system (BES) which, more generically, refers to a device where microorganisms interact electrically with electrodes (Rabaey et al., 2007). Microbial electrolysis cells (MECs) are another category of bioelectrochemical systems where the oxidation reaction at the anode is coupled to the reduction of an electron acceptor at a lower potential at the cathode (i.e., water to produce hydrogen). Since the process is endergonic, a certain voltage needs to be applied. In its standard configuration, an MFC consists of two chambers: the anode and the cathode compartments (Figure 2). Bacteria growing at the anode catalyze the electron transfer from an organic (or inorganic) molecule to the anodic electrode. The reduction of the terminal electron acceptor takes place at the cathode, generally separated from the anode by an ion-selective membrane and electrically connected to it via an external circuit containing a resistor or power user that harvests the energy liberated by the reactions. Several electron acceptors can be used, for example, oxygen (O2), potassium hexacyanoferrate (also known as ferricyanide, K3Fe(CN)6), and nitrate, (NO3 ). The cathodic reaction can be of an
Microbial Fuel Cells
e-
A
The maximal work that can be derived from such processes can be measured by means of the Gibbs free energy of the general redox reaction nA A þ nB B- nC C þ nD D:
e-
H2O
CO2
O2
COD
Figure 2 Schematic representation of a microbial fuel cell (MFC). The substances (organics represented as chemical oxygen demand (COD)) are oxidized to CO2 by microorganisms, which transfer the gained electrons to the anode. At the cathode, the electrons are used to reduce oxygen abiotically or biotically, producing water. To maintain electroneutrality within the system, positive charges have to migrate from the anode to the cathode through an ion-permeable separator (a cation exchange membrane (CEM) in this representation).
electrochemical or bioelectrochemical nature. In the latter case, bacteria are involved as catalysts at the cathode as well, promoting electron transfer from the electrode to the final electron acceptor. If glucose is taken as an example of electron donor and oxygen as electron acceptor, Equations (2) and (3) characterize the reactions occurring at the anode and cathode, respectively:
C6 H12 O6 þ 6H2 O- 6CO2 þ 24H þ 24e
ð2Þ
6O2 þ 24Hþ þ 24eþ - 12H2 O
ð3Þ
þ
645
The aqueous media in the anodic and cathodic compartments are called the anolyte and the catholyte, respectively. The role of the ion-exchange membrane (IEM) is to allow the transport of charges between the two compartments, thus maintaining the electroneutrality of the system, and also to physically separate the two redox processes, thus preventing the electron acceptor from reacting directly with the electron donor. (IEM is a type of membrane that allows the selective diffusion of certain ions. Different types of IEMs exist, depending on the species that is transported, including cationexchange membranes (CEMs), proton-exchange membranes (PEMs), and anion-exchange membranes (AEMs).)
4.18.3 Thermodynamics of Microbial Fuel Cells Chemotrophic organisms fulfill their energy requirements by transferring electrons from a low redox potential molecule (primary electron donor) to a high redox potential molecule (primary electron acceptor). MFC electrodes virtually interpose within the electron-transfer process that would naturally occur in bacteria between the electron donor and acceptor.
DGr ¼ DGr0 þ RTln
anCC anDD anAA anBB
ð4Þ
where DGr is the Gibbs free energy of a reaction at specific conditions, measured in Joules (J), DG0r (J) is the Gibbs free energy at standard conditions (usually defined as 298.15 K, 1 bar pressure, and 1 M concentration of the species), R is the universal gas constant (8.3145 J mol1 K1), T is the absolute temperature (K), and ai is the activity of reactant i, and ni the respective stoichiometric coefficient. (The Gibbs free energy represents the maximum amount of useful work that can be obtained from a reaction.) In diluted systems, the relation can be simplified by replacing the activities with the concentrations, and Equation (4) can be rewritten as
DGr ¼ DGr0 þ RTln
nC ½C ½D nD ½A nA ½B nB
ð5Þ
In order to generate a current, the overall process in an MFC needs to be thermodynamically spontaneous. This requires the Gibbs free-energy change of the process to be negative. For a bioelectrochemical conversion, it is useful to evaluate the reaction in terms of electromotive force (Eemf), which is expressed in volts (V). The electromotive force and the Gibbs free energy are related according to
DGr ¼ QEemf ¼ nFEemf
ð6Þ
where Q is the charge transferred in the reaction in coulombs (C), which is also equal to the number n of electrons exchanged in the reaction (mol) multiplied per the Faraday’s constant F (9.64853 104 C mol1). Equation (6) can therefore be rearranged, yielding
Eemf ¼
DGr nF
ð7Þ
At standard conditions, DGr is equal to DG0r , and Equation (7) can be written as
E0emf ¼
DGr0 nF
ð8Þ
where E0emf represents the electromotive force at standard conditions. Equations (4) and (8) can be combined to calculate the total electromotive force for a given redox reaction occurring at certain conditions, yielding Equation (9), which is known as the Nernst law:
Eemf ¼ E0emf
nC RT ½C ½D nD ln ½A nA ½B nB nF
ð9Þ
Positive values for Eemf refer to spontaneous processes, whereas negative values indicate a nonspontaneous reaction. MFC technology is a galvanic process characterized by positive
646
Microbial Fuel Cells
values for the electromotive force. Microbial electrolysis is instead an electrolytic process where the electromotive force assumes negative values. The Eemf in an MFC can be evaluated by considering the generic redox reaction occurring as the sum of an oxidation and of a reduction. The Eemf is the result of the difference between the reduction potential of the reactions occurring at the cathode and at the anode (according to Equation (6)), each of them evaluated through the Nernst equation applied to the half reaction. Half-cell potentials are reported under the International Union of Pure and Applied Chemistry (IUPAC) convention as reduction potentials in comparison with the standard hydrogen electrode (which has a reduction potential conventionally set to zero at pH2 ¼1 bar, [Hþ] ¼ 1 M); therefore, the reaction is always written as an electron-consuming reaction (reduction). Table 1 lists a series of half-reaction reduction potentials important in MFCs and biological systems in general (Thauer et al., 1977). For biological purposes, the redox potentials are
Table 1 systemse
Summary of redox reactions important in biological
EAn ¼ E0An
ECat ¼
RT ½C6 H12 O6 ln 24F ½CO2 6 ½Hþ 24
E0Cat
RT 1 ln 4F pO2 ½Hþ 4
!
E0 0
6CO2 þ 24Hþ þ 24e-Glucose þ 6H2O 2Hþ þ 2e-H2 NADþ þ Hþ þ 2e-NADH 2CO2 þ 8Hþ 8e-Acetate þ 2H2O S þ 2Hþ þ 2e-H2S SO4 2 þ 10Hþ þ 8e - H2 S þ 4H2 O Pyruvate þ 2Hþ þ 2e-Lactate FADþ þ 2Hþ þ 2e-FADH2 Fumarate2 þ 2Hþ þ 2e-Succinate2 Cytochrome b (Fe3þ) þ e-Cytochrome b (Fe2þ) Ubiquinone þ 2Hþ þ 2e-Ubiquinone H2 Cytochrome c (Fe3þ) þ e-Cytochrome c (Fe2þ) NO2 þ 2Hþ þ e - NO þ H2 O FeðCNÞ6 3 þ e - FeðCNÞ6 4 Cytochrome a (Fe3þ) þ e-Cytochrome a (Fe2þ) NO3 þ 2Hþ þ 2e - NO2 þ H2 O NO2 þ 8Hþ þ 6e - NH4 þ þ 2H2 O NO3 þ 6Hþ þ 5e - 0:5N2 þ 3H2 O Fe3þ þ e-Fe2þ O2 þ 4Hþ þ 4e-2H2O NO þ Hþ þ e-0.5NO þ 0.5H2O 0.5N2O þ Hþ þ e-0.5N2 þ 0.5H2O
0.43 Va 0.42 Va 0.32 Va 0.28 Va 0.28 Va 0.22 Va 0.19 Va 0.180 Vd þ 0.03 Va þ 0.035 Va þ 0.11 Va þ 0.25 Va þ 0.350 Vb þ 0.36 Vc þ 0.39 Va þ 0.433 Vb þ 0.440 Vd þ 0.74 Va þ 0.76 (pH ¼ 2)a þ 0.82 Va þ 1.175 Vb þ 1.355 Vb
From Madigan MT, Martinko J, and Parker J (2000) Brock Biology of Microorganisms. Upper Saddle River, NJ: Prentice Hall. b From Thauer RK, Jungermann K, and Decker K (1977) Energy-conservation in chemotropic anaerobic bacteria. Bacteriological Reviews 41: 100–180. c From He Z and Angenent LT (2006) Application of bacterial biocathodes in microbial fuel cells. Electroanalysis 18: 2009–2015. d From Rabaey K and Verstraete W (2005) Microbial fuel cells: Novel biotechnology for energy generation. Trends in Biotechnology 23(6): 291–298. e The standard redox potentials are measured at pH 7 and 25 1C. Redox couples are arranged from the strongest oxidant (more positive reduction potential) at the bottom, to the strongest reductants (most negative reduction potential) at the top. Electrons naturally flow from lower to higher redox potentials. The larger the difference in reduction potential between electron donor and electron acceptor, the larger is the energy released.
ð10Þ
! ð11Þ
The difference between ECat and EAn would then give
Eemf ¼ ECat EAn
Redox reaction
a
generally referred to at pH 7 and 25 1C (in which case they are indicated with the symbol E0 0). These reactions include not only oxidations of organics and reductions of terminal electron acceptors, but also redox reactions of intermediate metabolites. Based on the values in Table 1, if glucose is the electron donor (–0.43 V) and oxygen is the electron acceptor ( þ 0.82 V), an electromotive force of 1.25 V would develop across the MFC at standard conditions. Under more general conditions, the application of the Nernst law on Equations (2) and (3) would yield the following potentials for the two half-cell reactions:
ð12Þ
4.18.4 Factors Determining the Decrease of Cell Voltage Although the electromotive force represents the upper limit for the total voltage that the MFC can generate under certain conditions, the actual voltage will always be lower under practical conditions, due to a number of losses of either purely electrochemical and/or of biological nature. An ideal MFC would deliver any amount of current while maintaining a constant voltage, as determined by thermodynamics. In practice, the actual voltage output would be lower due to irreversible losses. These potential losses increase with increasing currents and can have a dramatic effect on the performance of the MFC, as the loss of voltage would result in a lower power output, accordingly to Equation (13)
P ¼ Vi
ð13Þ
where P is the power density (W cm2), V is the voltage (V), and i is the current density (A cm2). (In order to permit the comparison between different systems, current and power are usually normalized to some characteristic of the reactor, such as the projected surface area of anode or cathode, or alternatively to the compartment total volume or liquid (net) volume.) Maintaining a high voltage under high current production is therefore critical for successful MFC operations. Polarization curves represent the cell’s voltage as a function of the current. They are regarded as a useful tool for the measurement of the MFC performance (Figure 3). They are performed by periodically modifying the applied load (external resistance) and recording the resulting voltage and current, the latter evaluated through Ohm’s law (V ¼ R i). They can be performed manually or automatically by means
Microbial Fuel Cells
647
Theoretical Eemf Voltage at open circuit (OCV)
Cell voltage (V)
P-i curve (b) Activation region
Ohmic region
Mass transport region
Vopt
Current density (A.cm−2)
iopt
Power density (w.cm−2)
Pmax
V-i curve (a)
isc
Figure 3 Typical polarization (a) and power (b) curves for an MFC. The point of maximal power (Pmax) corresponds to the optimal voltage (Vopt) and the optimal current density (iopt.). The maximal current density at short circuit (iSC) is reached when the external resistance is zero.
of a potentiostat. In this case, an appropriate scan rate (e.g., 0.1 mV s1) should be chosen (Velasquez-Orta et al., 2009). Polarization curves should be recorded from high to low external load and vice versa. While the Eemf as defined earlier represents the thermodynamic potential difference achievable in an electrochemical system, its value is not normally reached in real systems. The open circuit voltage (OCV) is the maximal voltage that can in fact be measured under conditions at which there is infinite resistance (i.e., at open circuit). There is a series of limitations imposed by the specific bacterial communities catalyzing the anodic reaction (and cathodic, in case of biocathode) that reduce the overall potential difference attainable (Logan, 2008). Three zones defining as many different operating regimes can be identified in a polarization curve (Benziger et al., 2006): 1. At open circuit there is no flow of electric current. However, when the current starts flowing, the voltage drops rapidly as a result of the activation-energy barrier of the reactions occurring at the electrodes; this zone is referred to as ‘activation polarization region’. (The voltage at open circuit measures the activity of reactants at anode and cathode electrode surfaces.) 2. At medium currents, the voltage decreases almost linearly with the current; this is referred to as ‘ohmic polarization region’, as it is dominated by ohmic losses, which arise from the resistance opposed by electrolytes and the IEM to the transport of ions as well as by electrodes and interconnection circuit to the transport of electrons. 3. At higher currents, the voltage drastically drops as a result of the insufficient mass transport of reactants or reaction products to and from the electrode, which limits the
reaction. This is known as ‘concentration polarization region’. The ratio of the cell voltage (V) and the cell voltage at open circuit (OCV) gives the potential efficiency (PE) (Lee et al., 2008). It is essentially the portion of the total potential difference between electron donor and acceptor that is captured as useful electric energy. The coulombic efficiency, or charge transfer efficiency (CE, or eC), is defined as the ratio of the charge that is transferred to the anode and the maximal charge that would be yielded if all the converted substrate generate electricity (Logan et al., 2006). It represents therefore the fraction of electrons recovered as electricity from the substrate converted. The energy conversion efficiency (ECE, or eE) is obtained by multiplying the PE and the CE (Lee et al., 2008). It represents the ratio of the power delivered from the system and the power that would be delivered in the absence of internal resistances (Benziger et al., 2006). For an MFC, the objective is to maximize the power output (represented by the peak of the P vs. i curve in Figure 3) and the ECE. Maximal power is obtained when both current and voltage are maximized, whereas maximal ECE is obtained when the potential efficiency and the coulombic efficiency are both maximized. However, the potential efficiency is negatively affected by the current density, which is in turn needed to maximize the power. This aspect is very important in engineering MFC systems (and fuel cells systems in general) as it means in other words that ECE and power output cannot be simultaneously optimized. At maximal power output, the ECE is 50% (Benziger et al., 2006). Higher efficiencies are achievable but with lower power outputs. Understanding the nature of the losses is of fundamental importance for successful operations of MFCs. Electricity generation in MFC is in fact the result of several steps that
648
Microbial Fuel Cells
A
CO2
H2O Microorganism
Link
Link
Microorganism O2
Voltage (V)
COD
OCV 1
2 3 4 3 2
Cell voltage
1
Figure 4 Potential losses during electron transfer in an MFC including a bioanode and a biocathode. 1: losses due to mass transfer limitation. 2: losses due to bacterial metabolic kinetics. 3: losses due to electron transfer to/from the electrode. 4: losses due to the resistance of the electrolytes (including the ion exchange membrane) and of the electrical interconnection to the flow of charges. These losses result in the reduction of the cell’s voltage from its value at open circuit (i.e., at infinite external resistance).
necessarily need to proceed at the same rate, as they occur in series. These steps are: (1) mass transfer of reactants and products from the bulk liquid to the electrode attached biofilm, and vice versa, (2) losses due to bacterial metabolic kinetics, (3) electron transfer from microbial cells to the electrode (and vice versa), and (4) transfer of charges through the electrodes and through the electrolyte and IEM. Some of these steps may limit the overall rate of electron transfer thus causing a larger voltage loss than others (Figure 4). The majority of the investigations carried out in the field of MFCs have been aimed at the improvement of power outputs by acting on one of the limiting steps to electricity generation.
4.18.4.1 Losses due to Mass-Transfer Limitation Electricity production in MFC relies on the flux of the reaction reactants in and out the biofilm. The flux of substrates is controlled by the diffusion in the biofilm as well as by the rates of utilization or production. If the reactions involving the substrates occur at a rate that is faster than that at which the reactants or products diffuse in or out of the biofilm, accumulation or depletion of one of the components occurs within the biofilm. As a result, the electrode potential becomes modified as depicted by the Nernst equation (Equation (9)).
If we consider, for instance, the anodic oxidation of glucose (Equation (2)), per mole of glucose that diffuses and is consumed within the biofilm, 6 mol of CO2 and 24 mol Hþ are produced and have to diffuse out of the biofilm. Protons are particularly important as their accumulation may lead the acidification of the biofilm. Torres et al. (2008) have shown that current density is largely determined by proton transport out of the biofilm. Current densities higher by more than 4 times were achieved when the phosphate buffer was increased from 12.5 to 100 mM. The authors also concluded that only in systems in which low COD concentrations are required in the effluent, substrate mass transport limitation may be more important than proton transport. However, it is important that the MFC compartments receive a proper loading rate of substrate to support the biomass attached to the electrodes. In continuous systems, organic loading rates at the anode of MFCs can vary between 0.5 and 4 kg COD m3 of anode liquid volume per day, with an optimum close to 3 kg COD m3 d1 (Rabaey et al., 2003). The hydrodynamic patterns are also important in order to provide homogeneous conditions on the biofilm/liquid interface.
4.18.4.2 Losses due to Bacterial Metabolic Kinetics Bacterial metabolic losses result from the rates of substrate uptake and utilization during the microbial metabolic activity,
Microbial Fuel Cells
which depends on both the specific microbial consortium catalyzing the reactions and the biomass density on the electrode surface, which in turn depends on the specific surface area of the electrode accessible to bacteria. For instance, the main limitation in MFC anodes is often not the specific uptake rate by the bacteria, but the bacterial density at the anode. Measurements have revealed that biomass concentrations at MFC anodes are 30 times lower compared to anaerobic digesters (Aelterman et al., 2008). Improvement of current and power outputs in MFCs requires the achievement of denser microbial colonization of the electrodes, while maintaining thin biofilms and open structures to facilitate diffusion and reduce mass-transfer limitation. While bacteria attach well to graphite electrodes, plain graphite may not be satisfactory if high power outputs are desired. Extensive research has been done on anode materials to maximize surface affinity with microorganisms (Cheng and Logan, 2007, Liu et al., 2007) and to facilitate electron transfer by the immobilization of mediators (Park and Zeikus, 2003) or conductive polymers (Schroder et al., 2003) on the anode surface. However, regardless of the material or design adopted, the anode biology does not currently constitute the main bottleneck of MFCs, unless competing populations such as fermentative bacteria outgrow the anodophilic population, driving the process to a failure (Rabaey et al., 2003). In addition, microorganisms themselves require energy for growth and maintenance purposes. Therefore, an anodic biofilm, for example, would take part of the energy available from the organic substrate and release the electrons at a slightly lower energy level, thus reducing the total voltage. As the anode is virtually the final electron acceptor, its potential would affect the total energy available for the microbes. The higher the difference is between the redox potential of the substrate and the electrode, the higher is the theoretical energy gain for bacteria growing on its surface, per electron-mole transferred. To maximize the voltage, anodic and cathodic electrode potentials should be kept as negative and as positive as possible, respectively, accordingly to the limits imposed by the redox potentials of the substrates used. Nevertheless, when the anode potential becomes very low, competitive processes such as fermentation or even acetoclastic methanogenesis may be favored, as the energy gain would be comparable in that case, as was shown in some recent studies (Aelterman et al., 2008; Finkelstein et al., 2006; Freguia et al., 2007b; Virdis et al., 2009). Furthermore, while higher cathodic potentials would maximize the voltage, the lower driving force pushing electrons from the cathode to the final electron acceptor may lead to the accumulation of intermediates as has been shown in the case of cathodic denitrification (Virdis et al., 2009).
4.18.4.3 Losses due to Electron Transfer to (and from) the Electrode Voltage losses due to electron transfer to (and from) the electrode are caused by the finite rate of electron transfer between microorganisms and the solid phase of the electrode (and vice versa). At an anode, the result is an accumulation of positive charge on the electrode and negative charge in the form of anions in the adjacent liquid layer. This double layer thus established causes the development of a potential
649
difference across it, called activation overpotential, which results in a reduction of cell voltage equal to its value. Bioelectrochemical reactions in BESs differ significantly from conventional electrocatalytic reactions by the fact that while in the latter the electron-transfer step from the electron donor to the electrode proceeds only at one particular point (e.g., the catalyst particle), the oxidation of the substrate in the former occurs throughout a more complex series of enzymatic reactions, the last of which is the electron transfer to the electrode in the case of the anodic reaction. Only this last step influences the activation polarization (Schroder and Harnisch, 2009). Activation overpotentials are extensively described in the electrochemistry literature. A detailed explanation of the origin of overpotentials can be found in Rieger (1994). Activation overpotentials are mathematically described by the Butler–Volmer equation, which dictates the logarithmic increase of the overpotential with the current density:
bFZ ð1bÞFZ i ¼ i0 e RT e RT
ð14Þ
where Z (V) is the overpotential at the electrode, R is the universal gas constant (8.3145 J mol1 K1), T is the absolute temperature (K), b is the symmetry factor (unitless), which is a constant that represents the dependence of the activation energy on the electrode potential, F is the Faraday’s constant (9.648 53 104 C mol1), i is the current density (mA m2), and i0 is exchange current density (mA m2), which depends on the activation energy of the reaction at equilibrium conditions, in such a way that higher activation energy results in lower exchange currents. At overpotentials that are sufficiently high (greater than 80–100 mV at 25 1C, according to Freguia et al. (2007c)), the second term between brackets becomes negligible and Equation (14) can be rewritten in its simplified version, best known as Tafel equation:
ln
i bFZ ¼ i0 RT
ð15Þ
Equation (15) can be used to experimentally estimate the parameters i0 and b using the so-called Tafel plots (ln(i) vs. Z), generated from polarization-curve measurements (Freguia et al., 2007c). The parameters i0 and b (and thus the activation overpotential) strongly depend on the activation energy of the reaction at the electrode. Electrodes with high specific surface area can not only support increased biomass densities but also decrease activation losses by reducing the current densities at the electrode surface (Chaudhuri and Lovley, 2003; Freguia et al., 2007c).
4.18.4.4 Losses due to the Resistance of the Electrolytes (Including the Ion-Exchange Membrane) and of the Electrical Interconnection to the Charges Flow It was described earlier (Section 4.18.2) that an equimolar amount of positive and negative charge is produced during the oxidation reaction at the anode (Equation (2)). While the electrons need to travel along the electrodes and the electrical circuitry to reach the cathode where the reduction reaction
650
Microbial Fuel Cells
takes place, ions needs to travel through the electrolyte and the IEM to ensure electroneutrality (cations in case a CEM is used; anions in the case of an AEM). The resistance of the different conductors (e.g., electrodes, current collectors, wires, IEM, and electrolyte) toward this charge flow introduces a loss of voltage, which is referred to as ohmic loss. According to Ohm’s law, the voltage loss due to charges migrating through a medium is proportional to the current and to the ohmic resistance of the medium. In turn, the ohmic resistance depends on the medium resistivity r(mO cm) (or equivalently its inverse, the conductivity s, mS cm1), the average distance traveled by the ions (L, cm), and the cross-section area over which the charges move (A, cm2, often identical to the nominal surface area of the electrodes for two-dimensional configurations), according to the following equation:
L R¼r A
ð16Þ
The intrinsic resistance of the electrode could become an important limitation to the generation of electricity as the electric resistivity of the conductors can be very high (the resistivity of graphite and carbon is 1000 times higher than that of iron; Rozendal et al., 2008a). The resistance offered by the electrolytes as well as by the IEM toward this transfer of charge is often a major limitation in MFCs.
4.18.5 Materials and Architectures Although typical MFC architecture consists of an anodic chamber and a cathodic chamber separated by an IEM, as depicted in Figure 2, different materials and reactor configurations have been implemented for lab-scale studies, depending on the scope of the study itself. Designs may vary from two-compartments to single-chambered MFCs, from tubular to stacked configurations, and with or without a membrane. Sediment MFCs have also been constructed by placing one electrode into marine sediments and the other in the overlying oxic water (Reimers et al., 2001; Tender et al., 2002).
4.18.5.1 Design As explained above (Section 4.18.4), the performance of MFC is strongly affected by a number of factors, particularly the resistivity of material used for the electrodes, the resistance offered by the electrolytes toward the charge transport, and the nonperfect selectivity of the IEM, which creates pH gradients between the compartments. It is therefore not surprising that the system performances are dictated by the design and the materials used. The H-shape two-chambered design is an inexpensive and easy-to-handle laboratory design that has been widely adopted in early MFC research. It simply consists of two bottles connected by a tube that can interpose an IEM or a salt bridge between anode and cathode (Bond et al., 2002; Park and Zeikus, 1999; Min et al., 2005). H-shape systems typically produce low current densities due to the high internal resistance, which limit their use to basic parameter research, such as
examining new materials or studies of microbial communities (Logan et al., 2006). Better performance can be obtained by the two flatchamber designs first developed by Delaney et al. (1984), which offer lower internal resistance due to the proximity at which anode and cathode can be put over a generally larger IEM. This compact configuration resembles that of traditional chemical fuel cells. This strategy was adopted by Min and Logan (2004) while designing their flat-plate MFC that comprises of two polycarbonate plates bolted together and contains a carbon-cloth cathode hot-pressed to an IEM also in contact with a carbon paper that serves as an anode, obtaining up to 7271 mW m2 of power density. The flattened design MFCs can easily be stacked together and electrically connected in series or in parallel in order to increase the overall system voltage (Aelterman et al., 2006b). Tubular shapes have also been designed (Rabaey et al., 2005b), or upflow types with anode below and cathode above (He et al., 2005), with the liquid sequentially passing through the two compartments. More complex designs have also been implemented to allow more complex measurements such as gas production and consumption, pH, and dissolved oxygen (Freguia et al., 2007b). When oxygen is used as electron acceptor, the cathode can be placed directly in contact with air, thus circumventing the need for a second chamber. In the single-chamber configuration, the cathode consists either of a catalyzed electrode open to the air, or is assembled with the anode within the same unit. Park and Zeikus (2003) used an MFC made of one compartment consisting of an anode coupled with a porous air-cathode directly exposed to air. In Liu and Logan (2004), an anode and a cathode were placed on opposite sides of a Plexiglas cylindrical chamber of length 4 cm and diameter 3 cm. The anode was made of carbon paper without wet proofing, while the cathode was manufactured by bonding the IEM directly on a carbon cloth (with platinum as catalyst). Liu et al. (2004) implemented the tubular shape within a single-chamber configuration for the treatment of wastewater. The anode, consisting of several graphite rods, surrounded the cathode made of carbon/Pt/IEM layers bolted together to a plastic support through which air was blown. Rabaey et al. (2005b) manufactured a tubular MFC with an inner cylindrical anode consisting in granular packed-bed graphite and an outer cathode.
4.18.5.2 Compartment Separation In MFCs, the function of the IEM is not only to provide a physical barrier to prevent fuel crossover between the compartments, but also to create a way for the ions to selectively diffuse to ensure electroneutrality. For example, as shown earlier, for every negative charge that is transferred to the cathode through the electrical circuitry, an equal amount of positive charge needs to flow through the electrolyte to prevent charge build-up. Finally, it also prevents the electrolytes from large pH fluctuations due to proton production at the anode and proton consumption at the cathode. Nafion (DuPont Inc., USA) and Ultrex CMI-7000 (Membranes International Inc., USA) are largely applied CEMs.
Microbial Fuel Cells
Although Nafion CEMs have been widely used in fuel-cell research, they do not perform as well under typical conditions at which MFCs work, for example, neutral pH, and in the presence of other cations in concentrations that can be 105 times higher than the proton concentration. Rozendal et al. (2006a) showed that under these conditions, Nafion membranes mainly transfer other cations rather than protons, thus lacking specific selectivity for protons. Ultrex is a more general CEM with larger mechanic strength compared to Nafion (Harnisch et al., 2008). It is considered a more cost-effective alternative to Nafion. The reader can refer to the works of Rozendal et al. (2008c) and Harnisch et al. (2008) where alternative types of membranes are compared in MECs and MFCs. Several attempts have been made by researchers toward the development of membrane-less MFC, in which the IEM is absent (e.g., sediment MFCs), or is replaced by different types of separators. Liu and Logan (2004) studied how the performances are affected by the presence or the lack of a Nafion membrane. Their results showed that increased power densities were possible without the IEM. Nevertheless, the enhanced oxygen diffusion led to a decrease of Coulombic efficiency as a higher portion of the carbon source was oxidized without electrons transferring to the anode. In an attempt to increase the oxygen-diffusion resistance, Park and Zeikus replaced the IEM with a porcelain septum (100% kaolin) and despite obtaining higher power outputs, the Coulombic efficiency was fairly low (Park and Zeikus, 2002, 2003). The addition of successive layers of polytetrafluoroethylene to the cathodic air-side of a single-chamber MFC resulted in increased coulombic efficiency and increased maximal power density (Cheng et al., 2006a). Jang et al. (2004) designed a membrane-less MFC in which anode and cathode were physically assembled within the same reactor unit and separated by glass wool and glass bead layers. Anode and cathode (made of graphite felt) were placed at the bottom and the top of the reactor and an upflow was imposed through the cylinder. Oxygen was bubbled in the cathode and its back diffusion was avoided simply by the stream flow. The protons formed during the anodic reaction were transported to the cathode by the same liquid stream. The results showed that the internal resistance was excessively high (several kO) due to the large anode to cathode distance, which resulted in low power generation. Moreover, most of the COD was removed in the cathode compartment by direct reaction with oxygen rather than by bioelectrochemical oxidation at the anode.
4.18.5.3 Electrodes MFC anodes and cathodes are typically made of graphite, which can be in the form of rods, felt, carbon paper, or cloth. Reactions occurring at the electrodes are subject to activation energies that need to be reduced by the use of appropriate catalysts. In the anode compartment, bacteria normally accomplish the role of catalysts. The electrodes therefore need to provide a suitable surface for the bacterial growth. Rough surfaces may provide several opportunities for adhesion, as well as decrease the current densities and therefore the potential losses, as is further described later (see Section 4.18.9.).
651
Increasing the surface area of the anode is also the simplest way to increase the loading rates that can be processed in an MFC as higher quantities of biomass can grow in the same reactor volume. Granular graphite is considered a convenient material for MFCs due to its cost (approximately US$ 0.5 kg1) and its high surface area and roughness. The porosity of the graphite is also an important aspect that needs to be taken into account when calculating the specific active surface of an electrode, as bacteria can grow in pores with a size larger than the bacteria themselves, but the smallest pores cannot be colonized and therefore do not contribute to the active electrode surface area. In addition, graphite granules have a high internal volume, which takes up about half the total reactor volume. Thus, carbon fiber brushes are increasingly considered as promising for future applications (Logan et al., 2007). Metal electrodes made, for example, of stainless steel can also be used (Tanisho et al., 1989), but despite being suggested as a good cathodic material, it has been shown to be less effective when used at anodes (Dumas et al., 2007). Moreover, metal electrodes do not normally offer a high specific surface area and their higher cost when compared to graphite limits their application, especially with regard to larger-scale use (Rozendal et al., 2008a). Uncoated titanium was also proposed by ter Heijne et al. (2008), although, based on DCvoltammetry and on electrochemical impedance spectroscopy (EIS), it was concluded that uncoated titanium was not suitable as an anodic material. Kargi and Eker (2007) proposed the use of copper and copper–gold electrodes, obtaining current and power production largely comparable with other studies. Increased performances have been obtained by adopting chemical–physical strategies, like incorporating Mn(IV) and using neutral red covalently linked to mediate electron transfer to the anode (Park and Zeikus, 2003). The use of materials such as polyanilines was also shown to improve current generation (Niessen et al., 2004; Schroder et al., 2003).
4.18.5.4 Cathodic Compartment Oxygen is by far the most suitable electron acceptor for MFC operations, due to its high redox potential (see Table 1), low cost, availability, and the fact that it does not produce any unwanted reaction product. However, its slow reduction kinetic on plain graphite requires in most instances the use of an appropriate catalyst. Platinum has been largely used in chemical fuel cells as an abiotic catalyst of the cathodic reaction. Nevertheless, platinum is not likely to be suitable for most of MFC applications because of its poisoning sensitivity toward some components in the substrate solution, especially to H2 S. The above-mentioned ferricyanide commonly used at MFC cathodes, but it cannot be considered a mediator for oxygen reduction, as its oxidation rate is much slower than its reduction (Pham et al., 2004). It acts therefore as an electron acceptor on its own, thus needing periodical replenishment. Yet, ferricyanide has the advantage of having a very low overpotential on plain carbon electrodes and operates at a potential close to its open circuit value. In spite of its sensitivity, it has been adopted for dissolved oxygen or open-air cathodes (Liu et al., 2004; Reimers et al.,
652
Microbial Fuel Cells
2001), with 0.5 mg cm2 Pt loading being used in several studies (Liu and Logan, 2004). Platinum content as low as 0.1 mg Pt cm2 was also shown as effective (Cheng et al., 2006b). In an attempt to reduce the cost of the catalyst, some researchers have tested alternative electrode compositions where redox mediators were bound to the surface of the electrode, avoiding at the same time the use of soluble mediators. Park and Zeikus (2003) described a technique to bind ferric sulfate to woven graphite surfaces for improved oxygen reduction in an air-cathode MFC. Fe(III) was reduced to Fe(II) by the electrons generated at the anode and Fe(II) was subsequently re-oxidized by oxygen. An inexpensive cobalt-based material, cobalt tetramethylphenylporphyrin, was tested by two research groups (Cheng et al., 2006b; Zhao et al., 2006), both concluding that the performances were comparable with that of platinum but using a material less susceptible to poisoning. Zhao et al. (2005) found that transition metals phthalocyanines and porphyrins exhibit catalytic activity comparable to platinum. Other compounds have been employed to enhance cathode catalysis on active carbon or titanium electrodes, including cobalt oxide and molybdenum/ vanadium (Habermann and Pommer, 1991). Recently, the possibility of biocatalyzing the cathodic reaction has opened up a number of new opportunities. Biocathodes have a number of advantages compared to conventional chemical catalysts, such as their low cost, self(re)generation capacity, and the fact that they are less sensitive to the components typically present in the wastewater. In most of the biocathode studies in which oxygen was the final electron acceptor, microorganisms were used to transfer electrons from a reduced form of the metal compounds to oxygen itself. Manganese and iron have been used to transfer electrons from the electrode to oxygen by means of biological processes (Bergel et al., 2005; Rabaey et al., 2008; Rhoads et al., 2005). Ter Heijne et al. (2007) developed an oxygen cathode mediated by the couple Fe3þ/Fe2þ at very low pH with biological reoxidation of ferrous ions with oxygen by a culture of Acidithiobacillus ferrooxidans. Compounds other than oxygen can be also used as terminal electron acceptors. Nitrate, sulfate, iron, manganese, uranium, selenate, arsenate, urinate, fumarate, and carbon dioxide are all possible candidates for MFC applications (He and Angenent, 2006). Examples also exist of the use of oxygenase enzymes such as the multi-copper oxygenase laccase, as catalysts for oxygen reduction (Schaetzle et al., 2009). Although, the high costs together with the limited lifetime and stability of the enzymes are important drawbacks of enzymatic electrodes that need to be addressed.
4.18.6 Electrochemically Active Microorganisms and Extracellular Electron Transfer The underlying working principle of an MFC is extracellular electron transfer (EET; It refers to a mechanism by means of which bacteria donate or accept electrons to and from an electrode; Chang et al, 2006). Microorganisms use EET in order to utilize insoluble electron acceptors (or donors) that cannot enter the cell (Rabaey et al., 2007). Bacterial interaction with an insoluble electron acceptor has been first studied for
microorganisms that respire on Fe(III) and Mn(IV) or oxidize large humic substances that cannot enter the bacterial cell (Lovley et al., 1996, 1987; Myers and Nealson, 1988). Two pathways of EET are currently assumed to be used by microorganisms (Figure 5):
• •
through electron through electron
mobile components (also referred to as mediated transfer pathway) or immobilized structures (also referred to as direct transfer pathway).
Redox mediators (or shuttles) are soluble compounds that can transfer electrons between the microbial cells and the electrode surface. Reactions involving redox mediators can in principle occur outside or inside the cells (Gralnick and Newmann, 2007). In the first MFC prototypes, soluble redox mediators were added to the media to aid EET from bacteria to an electrode. The characteristics of redox mediators are: (1) the ability to be reversibly oxidized and reduced, (2) the resistance to biological degradation, (3) fast kinetics of oxidation at an electrode, (4) ease of diffusion through bacterial membranes, and (5) nontoxicity toward microbial consortia. Substances such as neutral red, hexacyanoferrate, thionin, or quinones were used to promote EET (Kim et al., 2000; Park and Zeikus, 2000). Delaney et al. (1984) and Allen and Bennetto (1993) developed and improved MFCs using different combinations of microorganisms and mediators. They showed that the use of suitable mediators could enhance both the efficiency and the rate of electron transfer. More recently, live–dead staining and confocal microscopy analysis showed that even in systems were no exogenous mediators were added, microorganisms could anyhow contribute to electricity generation. This means that bacteria growing at a certain distance from the electrode can also demonstrate EET. It was reported by Rabaey et al. (2005a) and Hernandez et al. (2004) that redox active compounds such as pyocyanin and phenazine-1-carboxamide were self-produced by Pseudomonas species. In particular, these compounds were essential for electricity production by Pseudomonas aeruginosa. The production of endogenous mediators was thereby identified as an additional strategy enabling mediated EET. Although redox mediators were long thought to be essential to enable EET, and were therefore extensively used in MFCs, the finding that bacteria have the ability to reduce insoluble electron acceptors such as Fe(III) and Mn(IV) in oxide forms by Lovley and Phillips as early as 1988 (Lovley and Phillips, 1988) already suggested that mediated EET was not necessarily the only mechanism for EET. This discovery indeed represented a landmark in MFC research and opened the door to the development of mediator-less MFCs. If neither endogenous nor exogenous redox mediators are used, a direct contact between the outer membrane of the bacterial cell and the electrode surface must be established in order to promote electron transfer. Direct electron transfer requires that the microorganisms rely on a transport structure that enables electron crossover to the outside of the cell where they can be delivered to a solid electron acceptor (a metal oxide or, more pertinently, to an MFC anode). Unusually high content of c-type cytochrome in Shewanella putrefaciens outer membrane during anaerobic growth was
Microbial Fuel Cells
A Anode
Substrate
653
CO2
e−
Medox
B Medred
Bacterial cell C
Bulk
Figure 5 Possible microbial interactions with the anode. Microorganisms can transfer electrons through immobilized structures such as membrane-bound proteins (A) or electrically conductive pilus (B), or through mobile components (redox mediators or shuttles) that are alternatively oxidized and reduced at the electrode (C).
reported in as early as 1992 (Myers and Myers, 1992). Kim et al. (1999) were the first to measure the electrochemical activity of Shewanella putrefaciens when grown under anaerobic conditions without nitrate, as revealed by cyclic voltammetry. Cyclic voltammetry is an electrochemical technique where an electrode is immersed in a medium and its potential is changed cyclically by a potentiostat while the current is measured. The current versus potential diagram (called cyclic voltammogram) shows peaks in correspondence of the potential of each reversible redox couple in contact with the electrode. A double peak in the cyclic voltammogram revealed that a redox active compound (possibly a cytochrome) was responsible for the electrochemical activity of the cells of Shewanella putrefaciens. The study also showed that the cells lost their electrochemical activity when grown aerobically. In a further study, the electrochemical activity of Shewanella putrefaciens was demonstrated by current production in a mediatorless MFC (Kim et al., 2002). Bond et al. (2002) showed that some bacteria are capable of transferring electrons from anoxic marine sediments to an anode, connected to a cathode in the overlying aerobic zone. Community analysis of these bacteria showed that many of them belonged to the d-proteobacteria phylum. In particular, Geobacter metallireducens and Desulforomonas acetoxidans were identified. Another bacterium,
Geobacter sulfurreducens, was successfully tested by Bond and Lovley (2003) as a pure culture in a two-chamber fuel cell with acetate as substrate. Chaudhuri and Lovley (2003) demonstrated that the bacterium Rhodoferax ferriducens can perform electron transfer to an anode when fed with glucose. Recently, it has been reported that outer-membrane proteins are not always sufficient for the reduction of Fe(III) oxides. Geobacter and Shewanella species were shown to produce conductive appendages that were referred to as ‘nanowires’ (Gorby et al., 2006; Reguera et al., 2005). In Gorby et al. (2006) the conductivity of these pilus-like nanowires produced by Shewanella oneidensis was measured through conductive-scanning tunneling microscopy. A similar observation was done for nanowires produced by Geobacter sulfurreducens using conducting-probe atomic-force microscopy. Results show that while Shewanella species appear to produce rather thick bundles of conductive wires, Geobacter seems to produce more thin structures. Evidence also exists of interspecies electron transfer between bacteria in mixed communities, as suggested, for instance, by the presence of filaments connecting the propionate-fermenting Pelotomaculum thermopropionicum with the methanogen Methanothermobacter thermautotrophicus in Ishii et al. (2005). Whether interspecies electron transfer occurs
654
Microbial Fuel Cells
through wired connections or the use of mediators, it may significantly impact BES engineering as it will broaden the array of organisms able to survive on an electrode. Although the nomenclature for microorganisms that perform EET is far from being uniform (Lovley, 2008), the term electrochemically active microorganisms has been introduced to refer to microorganisms that demonstrate the ability to perform EET on an electrode without the addition of exogenous mediators (Chang et al., 2006). More specifically, the term ‘electrode-reducing microorganisms’ refers to organisms that can use an electrode as electron acceptor (thus reducing the electrode), whereas ‘electrode-oxidizing microorganisms’ refers to the organisms that use the electrode as electron donor (and thus oxidize the electrode) (Lovley, 2008). Community analyses of mixed anodic cultures enriched in electrochemically active microorganisms were undertaken by several investigators (Holmes et al., 2004; Kim et al., 2007b; Rabaey et al., 2004a). The striking conclusion from these studies was that EET capacity is widespread in nature and it is found among most phyla of bacteria. The ubiquity of these microorganisms in nature is also confirmed by the diversity of inocula that can be used to start up lab-scale MFCs. Raw sewage (Liu et al., 2004), activated sludge (Lee et al., 2003), anaerobic and methanogenic sludge (Rabaey et al., 2003), river sediments (Gregory et al., 2004), and seawater (Bond et al., 2002) were all successfully used to develop bioelectrochemical activity. Pure or enriched mixed cultures have shown the ability to use anodes to oxidize a variety of organic substrates, including acetate (Bond and Lovley, 2003), propionate (Bond and Lovley, 2005), butyrate (Liu et al., 2005c), ethanol (Kim et al., 2007b), lactate (Kim et al., 1999), glucose (Chaudhuri and Lovley, 2003), domestic wastewater (Gil et al., 2003), and beer-processing wastewater (Wang et al., 2007). Whereas respiratory flexibility in mammalian mitochondria is rather poor, it can be extremely broad in Bacteria and Archaea, as a diverse range of electron acceptors can be used, including nitrogen oxyanions and nitrogen oxides, elemental sulfur and sulfur oxyanions, halogenated compounds, transition metals such as Fe(III) and Mn(IV), as well as radionuclides such as U(VI) (Richardson, 2000). It is widely agreed upon that the first respiratory processes to evolve on Earth over 3.5 billion years ago would have used Fe(III) or S(0) as electron acceptors. The fact that the ability to electrically interact with an insoluble electron acceptor is widespread in nature is not therefore particularly surprising.
4.18.7 Oxidative Processes As we have seen previously in Section 4.18.3, the difference in reduction potentials between primary electron donor and terminal electron acceptor determines the net energy change of the reaction, and thus the energy gain for chemotrophic microorganisms. Energy is conserved by production of adenosine triphosphate (ATP) molecules. Depending on the availability of electron acceptor, two possible pathways are possible: fermentation or respiration. In the case of fermentation, the electron acceptor is a compound internally generated from the initial substrate, whereas in the case of respiration, the electron acceptor is externally provided. When
oxygen is the electron acceptor, it is referred to as oxic respiration. When the electron acceptor is oxygen linked with other compounds, it is referred to as anoxic respiration. The basic respiratory process involves the transfer of electrons from a low redox potential electron donor such as nicotinamide adenine dinucleotide (abbreviated as NADþ in its oxidized form, and as NADH in its reduced form), to the terminal electron acceptor at a high redox potential. The transfer occurs through a chain of intermediate redox complexes. The case where an insoluble electron acceptor rather than a soluble compound is oxidized is still a form of respiration, with similar mechanisms for energy generation to other forms of respiration. Anodic oxidation relies on the tricarboxylic acid (TCA) cycle, which together with glycolysis and pyruvate oxidation before the TCA cycle and electron transfer chain after it, permits the chemical conversion of the organic substrates into carbon dioxide and water, and generates energy in the form of ATP (White, 1995). Many bacterial species have been shown to produce electricity in MFCs using compounds such as acetate, lactic acid, and ethanol which enter the TCA cycle through pyruvate or acetyl-CoA (Bond and Lovley, 2003; Kim et al., 1999, 2007b), whereas more complex carbohydrates such as glucose require glycolysis before entering the TCA cycle. NADH, nicotinamide adenine dinucleotide phosphate (NADPH), and flavin adenine dinucleotide (FADH2) are generated through the TCA cycle. These are reduced molecules that represent the primary electron donor for the electron transport chain, through a series of membrane-associated electron carriers, including flavoproteins, iron–sulfur proteins, quinone pool, and a series of cytochromes. The electron transport chain has two basic functions: (1) to accept electrons from an electron donor and to transfer these electrons to the next electron acceptor and (2) to conserve the energy released during electron transfer for the synthesis of ATP. The electron carriers are arranged in the membrane in such a way that the electrons are transferred from one complex to the following at a higher potential (Figure 6). Hydrogen atoms removed from carriers such as NADH are separated from the electrons. While the electrons are transferred to the following carrier, the protons are pumped outside the cell (or to the periplasm in Gram-negative bacteria). This generates a proton motive force across the membrane, which extrudes up to approximately 10 protons for each electron pair derived from 1 NADH. The proton motive force drives ATP generation through a process called phosphorylation, which involves a large membrane complex called protontranslocating ATP-synthase, an enzyme that exploits the electrochemical potential liberated by the protons as they return to the cytoplasm. It is assumed that about three protons are required to generate one molecule of ATP, although recent research suggested that this stoichiometry may vary from 3 to 5 protons per ATP (Nakanishi-Matsui and Futai, 2008). It is generally accepted that maximally three molecules of ATP are generated by prokaryotes per NADH molecule (White, 1995). Bacteria in MFCs establish a direct contact with the electrode through cytochromes or nanowires or via soluble redox mediators (see Section 4.18.6). In any case, the electrontransfer chain as explained earlier cannot be entirely exploited, as the potentials of the electron carrier used to transfer electrons to the electrode (soluble mediator or cytochrome) are
Microbial Fuel Cells
655
Periplasm 4H+
4H+
3H+
2H+
Complex I
Complex IV cyt c
Q
ATP synthase
Complex III NADH
+
NAD
Quinone pool
Cytoplasm
O2 + 4H+
2H2O ADP+P
ATP
Figure 6 Respiratory electron transport chain of an organism such as Paracoccus denitrificans, a model for studies of respiration. Electrons are sequentially transferred through a series of membrane-associated electron carriers, which are embedded in the lipid bilayer of membranes in such a manner that most have access to both the inside and the outside of the cell. Hþ atoms removed from carriers such as NADH are separated from electrons and pumped outside the cell or in the periplasm. The reduction of O2 to H2O plus the extrusions of Hþ during electron transport generate a pH gradient and an electrochemical potential across the membrane (proton motive force, expressed in volts). This potential energy is used to drive the formation of high-energy phosphate bonds in ATP. Complex I: membrane-spanning complex comprising of flavoproteins and Fe–S proteins. Quinone: lipid electron carrier. Complex III: comprises cytochrome b, Fe–S proteins, and cytochrome c1. Cyt c: cytochrome c. Complex IV: cytochrome aa3 oxidase. From Madigan MT, Martinko J, and Parker J (2000) Brock Biology of Microorganisms. Upper Saddle River, NJ: Prentice Hall.
normally not electronegative enough to receive the electrons from the next step in the electron transport chain. Therefore, the maximal ATP yield for electroactive microorganisms is limited by the redox level at which the transfer chain is interrupted. As bacterial growth depends on the availability of intracellular ATP, the growth yield for bacteria in MFC will ultimately depend on the mechanism of electron transfer. Furthermore, as the potential of the anode determines the last step of the electron transfer to the electrode, the microbial growth will ultimately depend upon the anodic potential (Aelterman et al., 2008; Finkelstein et al., 2006; Freguia et al., 2008b; Schroder, 2007). A broad range of biodegradable materials has been shown to serve as electron donors for electricity generation in MFCs. Volatile fatty acids (e.g., acetate, formate, and butyrate), alcohols (e.g., ethanol and methanol), as well as more complex carbohydrates (e.g., glucose, sucrose, cellulose, and even starch), and even amino acids and proteins were used as organic electron donors (Freguia et al., 2007b; He et al., 2005; Heilmann and Logan, 2006; Ishii et al., 2008; Liu et al., 2005b; Logan et al., 2005; Min and Logan, 2004; Rabaey et al., 2003). Inorganic compounds such as sulfide (Rabaey et al., 2006) and synthetic acid-mine drainage (Cheng et al., 2007) have also been reported. Thus far it is still not clear which role the different types of substrates play. Acetate was reported to be the preferred substrate when compared to buyrate (Liu et al., 2005b) or to wastewater (Rabaey et al., 2005b), suggesting that MFC organisms prefer rapidly biodegradable substances to more complex compounds. Coulombic efficiency of 100% (i.e., a stoichiometric conversion of the substrate into current) was
reported in Freguia et al. (2007b) in their acetate-fed anode, whereas much lower efficiencies were obtained when glucose was used instead. Acetate has therefore been the substrate of choice in a large number of studies. Nevertheless, even if acetate is considered inert to alternative biochemical conversions such as fermentation in MFCs (Aelterman, 2009; Freguia et al., 2007b, 2008b), acetoclastic methanogenesis has recently been reported as an important anodic electron sink when the operating conditions favor the establishment of a methanogen community alongside electrochemically active microorganisms (Virdis et al., 2009).
4.18.8 Reductive Processes Bioelectrochemical oxidation of organics at the anode of MFCs has to be coupled with a reduction reaction at a counter electrode (cathode). Several electron acceptors have been used, depending on the scope of the BES. When power generation is the goal, oxygen appears to be the preferred choice due to its high availability and its high redox potential (see Table 1). As seen in Section 4.18.5, hexacyanoferrate has been extensively used in laboratory studies focused on the anodic reaction, due to its ability to provide a constant potential. More recently, the demonstration that compounds such as nitrate (Clauwaert et al., 2007a), nitrite (Virdis et al., 2008), hexavalent uranium (Gregory and Lovley, 2005), perchlorate (Thrash et al., 2007), and trichloroethene (Aulenta et al., 2009) can be reduced at the cathode, has broadened the array of applications of MFCs on nutrient removal and bioremediation as well.
656
Microbial Fuel Cells
While oxygen has been studied extensively for conventional chemical fuel cells, MFCs face specific limitations that substantially differentiate them from their hydrogen, methanol, or methane counterparts. In particular, they must operate at ambient pressure and moderate temperatures; moreover, the relatively low current outputs that are achieved in MFC do not justify the use of expensive chemicals as catalysts and, on the other hand, the nature of wastewater environments exposes most metal catalysts to irreversible poisoning. Three major bottlenecks can be identified for oxygen utilization at the cathode of MFCs: (1) the low solubility of oxygen in water limits its delivery to the electrode surface; (2) direct oxygen reduction at graphite electrodes exhibits large overpotentials; and (3) oxygen can diffuse (to some degree) through most membranes causing coulombic losses by direct oxidation of the organic electron donor. As oxygen reduction on plain carbon was found to happen at limited rates (Oh et al., 2004; Zhao et al., 2005), oxygen cathodes were developed with a platinum coating, based on the knowledge acquired from years of research in the chemical fuel cell area. It was found that the use of platinum enhanced oxygen-reduction rates when applied at a surface concentration of at least 0.1 mg Pt cm2 (Cheng et al., 2006b), with best performances at a load of 0.5 mg Pt cm2. Platinumcoated graphite electrodes have thereby set a benchmark for oxygen cathodes in MFCs, despite the fact that encouraging results had previously been obtained by immobilizing Fe(III) on graphite (Park and Zeikus, 2002, 2003). If a current of 1000 A m3 is expected to be delivered by the bioanode (Rozendal et al., 2008a), the aeration capacity that would have to be provided at the cathode can be estimated to be as high as about 0.0035 m3 O2 min1 per cubic meter of liquid, which is in the same order of magnitude as that provided by aeration systems normally applied for aeration tanks of activated sludge-treatment plants (Tchobanoglous et al., 2003). Oh et al. (2004) showed that aqueous oxygen cathodes (with Pt as catalyst) are the limiting step to electron transfer in two-chambered MFCs: a cathodic overpotential of B0.5 mV implied that about half of the total electromotive force expected was lost entirely due to oxygen reduction. The same investigators found that oxygen reduction behaves accordingly to Monod-type kinetics, with a half saturation constant of 1.74 mg O2 l1, which indicates that a further reduction of the electron-transfer rates is expected at low aeration rates. The reasons for the low performance of platinum cathodes in MFCs are not well understood. It can be speculated that the relatively mild conditions at which they operate, such as pH of around 7 in most cases and ambient temperatures, may affect the reaction rate. Conventional chemical PEM fuel cells normally sustain much higher current densities but they also typically operate at very low pH values (lower than 1). Given that protons are reactants in the cathodic reaction (Equation (3)), low pH values guarantee that protons are available in high concentration. In addition, chemical fuel cells operate at temperatures ranging from 50 to 100 1C in the case of PEM fuel cells, higher than that used for microbial fuel cells, to enhance the reaction rate. While the solubility of oxygen in water is a physical property and as such, cannot be increased, the thickness of the liquid film across which O2 has to diffuse can be minimized to reduce mass-transfer resistance. These
constraints have led to the development of open-air cathodes (Liu et al., 2004), where the cathodes have a two-dimensional structure and are open to the air, letting oxygen diffuse to the electrode surface directly from the air. This passive aeration process is more sustainable for scale-up applications as it does not entail large energy requirements for air pumping. In order to increase oxygen supply to the cathode surface, rotating cathodes have also been developed (He et al., 2007). Despite the widespread use of platinum as cathode catalyst, its high cost and energy-intensive production technology make this metal usually unsuitable as a catalyst for wastewater applications. Considering that the economic feasibility of MFCs is strongly dependent on the cathodic compartment (almost half of the capital costs of MFCs are associated with the cathodic compartment when platinum is used as catalyst; Rozendal et al., 2008a), platinum needs to be replaced by alternative catalytic materials. Three strategies have been explored thus far: (1) the use of a material with increased surface area; (ii) alternative chemical catalysts; and (iii) biocathodes. Freguia and co-workers (2007c) have shown that by using a noncatalyzed material with a high surface area it was possible to decrease the overpotential for cathodic oxygen reduction. As shown by Equation (15), the activation overpotential Z at the cathode increases with the current density i. The use of a better catalyst has the effect of reducing the overpotential as it increases the exchange current i0, which is a characteristic of the material used. The approach of the researchers instead was to reduce the current density by using a material with a higher surface area (coarse highly porous industrial-grade granular graphite) rather than modify the exchange current using a catalyst. The current generated with this configuration was able to sustain COD removals up to 1.46 kg COD m3 d1, which is similar to that of a conventional aerobic process based on activated sludge. Among other strategies for enhancing the cathodic reaction, the use of bacteria as catalyst has attracted particular interest in the MFC field. Similarly to bioanodes, biocathodes utilize the electrical interactions that can be established between the microbes and the cathodic electrode. Biological catalysis has been shown to enhance the rate of cathodic oxygen reduction at a stainless-steel cathode in seawater sediments (Bergel et al., 2005). The existence of such a consortium of bacteria was also later demonstrated for freshwater applications (Clauwaert et al., 2007b; Freguia et al., 2008a). The involvement of microorganisms in the catalysis was unequivocally established by a pure culture study (Rabaey et al., 2008). Freguia et al. (2008a) observed increased current production in an MFC system operated with recirculation of the anode effluent to the cathode, and correlated this result to the increased microbial activity at the cathode. Also, Rozendal et al. (2008b) reported of a cathodic microbial consortia catalyzing hydrogen production at a graphite cathode. The concept of using denitrifying bacteria to reduce nitrate in the cathode of an MFC was first proposed by Lewis more than 40 years ago (Lewis, 1966). However, it was only recently that the presence of reactions involving nitrogen at the cathode was confirmed. Catalytic reduction of nitrate and nitrite driven by electric current has been explored by Mellor et al. (1992). Nitrate-enriched water was pumped into the anode chamber of an electro-bioreactor and recirculated into the
Microbial Fuel Cells
cathode chamber where purified NADH:nitrate reductase, nitrite reductase, and N2O reductase enzymes were immobilized on the surface of the cathode. The applied electric current provided the reducing power needed to carry out the process. Hydrogen is an excellent electron donor and it can be easily produced by electrolysis of water. When a denitrifying biofilm is growing on the surface of a hydrogen-producing electrode, it has the advantage of having a continuous supply of an electron donor to carry out the process. Several studies attempted therefore to obtain nitrate reduction by applying a potential difference to form hydrogen at the cathode (Kuroda et al., 1997; Sakakibara et al., 1994; Sakakibara and Kuroda, 1993). Kuroda et al. (1997) extended the concept to obtain simultaneous COD removal and denitrification. However, hydrogen was still the actual electron donor for nitrate reduction. Although bacteria were considered as able to directly use the electrode as the sole electron donor, it was only very recently that this microbial capability was experimentally verified. Gregory et al. (2004) showed that a bacterial culture enriched in Geobacter species could reduce nitrate (NO3 ) to nitrite (NO2 ) using the cathode as the sole electron donor, without producing hydrogen as redox mediator. As nitrate reduction did not occur in the absence of bacteria, the researchers concluded that the process was biochemically activated by the biofilm, showing for the first time that bacteria were using the cathode as the sole electron donor. A later study confirmed their hypothesis and showed that complete denitrification to nitrogen gas could be achieved (Park et al., 2005). More recently, full denitrification with simultaneous carbon removal was reported in MFCs by Clauwaert et al. (2007a) and Virdis et al. (2008). Besides denitrification, other cathodic reactions have also been reported. Perchlorate, a compound extensively used in industry, was shown to be reduced with the help of 2,6anthraquinone disulfonate (Thrash et al., 2007). Shea et al. (2008) coupled a perchlorate-reducing biocathode with an acetate-oxidizing bioanode in an MFC configuration. Recently, an as yet unknown self-produced redox mediator appeared to be involved in the reduction of trichloroethene to more reduced compounds such as vinyl chloride and ethane (Aulenta et al., 2009). Finally, the recent demonstration of methane production from carbon dioxide reduction (Cheng et al., 2009) and biocathodic alcohol production from VFAs (Steinbusch et al., 2008, 2009), has opened up new possibilities for BES applications to biofuel production.
4.18.9 Challenges toward Improving MFC Efficiency Currently, several bottlenecks of both microbiological and technological nature limit the efficiencies of MFCS. It implies that despite the fact that laboratory MFCs already produce current densities suitable for practical applications, full-scale implementations are not necessarily straightforward (Rozendal et al., 2008a). The presence of alternative electron acceptors in the anode compartment (e.g., due to crossover of electron acceptors from the cathodic compartment to the anode), competitive processes such as fermentations and methanogenesis, and bacterial growth, are recognized as responsible for diverting a part
657
of the total electrons provided by the electron donor from the electricity-generation process, thus reducing the coulombic efficiency. The coulombic efficiencies reported in literature range from 1% (Kim et al., 2002) to about 100% (Freguia et al., 2007b). Besides, the presence of overpotentials at the electrodes, ohmic losses due to wastewater and electrode conductivity, and pH gradients due to imperfect ion selectivity of IEMs, are responsible for reducing the ECE as they reduce the cell voltage, described previously in Section 4.18.4. Their practical implications are described in this section.
4.18.9.1 Minimizing Electrode-Potential Losses Overpotentials at the electrodes can significantly limit the performance of MFCs as they decrease the actual voltage attainable, and therefore the energy efficiency. These losses are due to the electron-transfer kinetics from the microbial cells to the electrode (and vice versa) and to the bacterial metabolic kinetics. Moreover, as all heterotrophic bacteria retain a portion of their carbonaceous substrate to produce more biomass, a coulombic (and energetic) loss from this activity has to be taken into account as well. Interestingly, the potential losses at the anode have been reported to be much lower than that at the cathode. If, for example, an MFC can theoretically produce up to 1.1 V, less than 0.1 V is typically lost at the anode and more than 0.5 V can be lost at the cathode under working conditions (Logan et al., 2006). This would leave 0.5 V for power generation, without taking into account other losses such as ohmic losses. It is therefore obvious that any strategy intended to reduce the MFC overpotentials would have to pay particular attention to the cathodic reaction. As discussed above, the reasons for the high cathodic overpotential are mainly due to the slow kinetics of oxygen reduction. The use of biocathodes may be a valuable alternative to improve this catalytic process.
4.18.9.2 Respiration, Fermentation, and Methanogenesis In a complex environment such as wastewater, a multitude of other processes may occur alongside the conversion of organic molecules to electrons, thus competing with electricity generation in MFCs. When electron acceptors other than the electrode are present in the anode chamber, the organic electron donor can be oxidized using alternative pathways, thus reducing the electron transfer efficiency. In particular, nitrate and sulfate are commonly found in wastewater and are likely to divert the substrate electrons from the anode, as their redox potential makes them often more favorable electron acceptors than the anode. The presence of oxygen in the anodic compartment depletes substrate electrons through aerobic oxidation. Oxygen may be present due to possible pre-treatment of the MFC influent or through diffusion from the cathodic compartment in the case of oxygen cathodes. The potential for oxygen crossover to the anode is considerable in membraneless configurations (Liu and Logan, 2004). Methane is the end product of most anaerobic processes and it is regarded as one of the major bottlenecks of anode operations in MFCs, because methanogens compete with electrochemically active microorganisms for the organic material in the wastewater. It was recently shown that notable
658
Microbial Fuel Cells
amounts of methane were produced with glucose as substrate, consuming between 15% and 50% of the fermentable substrate electrons (Freguia et al., 2007a, 2007b, Lee et al., 2008). Methane was also observed in ethanol-fed MFCs (Torres et al., 2007). Fermentable substrates such as ethanol and glucose yield hydrogen when fermented. Hydrogen can be used by electrochemically active microorganisms for electricity production, or it can be further converted into methane through hydrogenotrophic methanogenesis. Observations suggest that electrochemically active microorganisms cannot completely outcompete methanogens for hydrogen (Freguia, 2008). One possible explanation for this behavior is that contrary to electroactive microbes, methanogens can grow at any distance from the electrode, as they do not need it as an electron acceptor. Therefore, they can grow on the top layer of the anodic biofilm, where they can scavenge the hydrogen that is formed by fermentation before it reaches the underlying electrochemically active biofilm. In the perspective of a real application, pre-fermentation would be required whenever fermentable substrates are present in the wastewater, in order to convert fermentable substrates into nonfermentable substrates such as acetate, thus providing electrode-reducing organisms with better chances to compete with methanogens. However, it is worth noting that when a bioelectrochemical system is operated at a controlled potential, or when a low anodic potential is the result of a low-current-producing process, the energy gain for electrochemically active organisms may become too low to successfully compete with low-energyyielding biological processes (e.g., acetoclastic methanogenesis). To conclude, controlling competitive processes requires a complex synergy of operational strategies in order to avoid the conditions at which methanogens are likely to scavenge the electrons away from electrochemically active microorganisms.
4.18.9.3 Reducing pH Gradients As extensively recalled throughout the text, anodic reactions produce protons while cathodic reactions (such as oxygen reduction) consume them. IEMs are typically used to provide physical separation between the two compartments while enabling the transport of charge at the same time (Section 4.18.2). The bottleneck that typically draws from the use of IEMs in MFCs derives from their lack of selectivity. Rozendal et al. (2006a) showed that Naþ, Kþ, and NHþ 4 normally account for most of the ionic-charge transfer across Nafion CEMs, due to their typically much higher concentrations in wastewaters compared to the Hþ concentration. Further research revealed that limited proton transfer occurs with most types of membranes, including AEMs and charge mosaic membranes (CMMs) (Rozendal et al., 2008c). The inefficient proton transport causes a pH gradient across the membrane, which results in an acidic anolyte and an alkaline catholyte, accordingly to the stoichiometry described in Equations (2) and (3). The consequence of the membrane gradient is a significant decrease in performance due to the reduction of the electromotive force. From the Nernst equation (Equation (9)), an increase in proton concentration at the anode results in a higher anodic potential and, similarly, a
reduction in the cathodic potential, causing an estimated loss of B0.06 V per pH unit (Rozendal et al., 2007). Both domestic and industrial wastewaters are characterized by limited alkalinity, which during MFC treatment has to approximately match the quantity of protons produced by the anodic reaction. According to Equation (2), 24 mol of Hþ are produced per mole of glucose, which translates to 4 mol Hþ per mole of COD. The alkalinity should therefore be about 4 times the influent COD molar concentration. This is normally not a problem in lab-scale MFCs that work on highly buffered synthetic media (in the range of 60–100 meq l1). Nevertheless, it would represent an important limitation when treating real wastewater. A domestic wastewater with 500 mg COD l1 would in fact require a 62.5 mM buffer, which is already much higher than the typical alkalinity reported (50–200 mg l1 as CaCO3 (Tchobanoglous et al., 2003), equivalent to 1–4 meq l1 buffer). Membrane-less designs would partially solve the issue. Yet, as mentioned previously, the lack of physical separation between anode and cathode would lead to the crossover of electron acceptor to the anode with significant reduction of the electron recovery. Bipolar membranes (BPMs, the twolayer combinations of a cation and an anion exchange membrane) can partially solve the problem by splitting water into Hþ and OH in the liquid space between the two membranes. However, they do so at the expense of a larger membraneinternal resistance, resulting again in a reduced power output. The sequential loop operation of anode and cathode (Freguia et al., 2007a, 2008a; Virdis et al., 2008) partly alleviates the problem by enabling convective transfer of protons from anode to cathode together with the liquid stream. Proton production by the anodic process may also negatively affect the performance of MFCs at the biofilm level, when the protons do not leave fast enough and accumulate within the biofilm. The Nernstian effect discussed earlier would occur at a smaller scale with the same effect of reducing the total electromotive force attainable. Increasing the specific surface area of the electrode may provide a significant benefit when it promotes an increased biofilm/liquid contact interface area as it would increase the proton flux (Torres et al., 2008).
4.18.9.4 Wastewater and Electrode Resistance Ohmic losses derive from the resistance of materials to the transfer of charged particles (see Section 4.18.4). Ohmic losses can be considerable in real wastewaters as they typically have low conductivity (of the order of only 1–4 mS cm1). As noted by Rozendal et al. (2008a), in full-scale MFC applications delivering 10 A m2 anode surface area, the ohmic loss that is encountered would be B1 V cm1 distance between anode and cathode for a wastewater with conductivity of 1 mS cm1, which is already B90% of the theoretical maximal voltage attainable. As increasing the ionic strength of wastewater by salt addition is not economically feasible in practice, the design and the operation of a full-scale MFC can significantly affect the extent of the ohmic losses. Researchers have attempted to reduce the ohmic voltage losses by testing different types of IEMs or even by completely removing them from the system. These investigations have revealed that the internal ohmic resistance is not controlled by
Microbial Fuel Cells
the membrane but by the electrolyte. Kim et al. (2007a) tested several membranes in identical MFCs. These included cation and anion exchange membranes as well as different kinds of ultrafiltration membranes. All MFCs tested exhibited similar internal resistances, confirming that the membranes did not control the overall resistance. Liu and Logan (2004) showed that the complete removal of the CEM did increase the power output, but in that case the increase was attributed to a higher cathodic potential and not to a reduction of the internal resistance. Moreover, the coulombic efficiency dropped to 12% as no barrier for oxygen diffusion was present in the system. Liu et al. (2005a) observed that the current output increased with the ionic strength, which was set by the salt concentration. This result further confirmed that the ionic conductivity of the medium plays a crucial role in MFC performance by determining its internal resistance. Keeping the electrodes in very close proximity is crucial for reducing the ohmic losses; flat compartment systems have therefore been used in research, either singly, or as multiple stack electrically connected in series or in parallel. Flat systems can minimize the ohmic losses as the compartments can be placed very close, thereby reducing the travel distance of ions through the electrolyte between the electrodes. However, the drawback of this configuration when expanded to larger scale is that in order to keep the same electrode distance with a rather larger volume, the electrons would need to travel longer distances to reach the cathode, thus significantly increasing the electrode ohmic loss if the material that is used is not sufficiently conductive. Highly conductive current collectors such as stainless-steel meshes can be used alongside the carbon/ graphite electrodes, although they can significantly increase the overall cost of the MFC. The use of bipolar plates can partially solve this problem as the distance that the electrons need to travel is reduced compared to the single-cell design (Shin et al., 2006). Bipolar plates (usually made of graphite) connect the anode side of one cell to the cathode of the next cell. The electrons generated at the anode only need to cross the bipolar plate to the cathode. However, this creates a stackin-series arrangement of the cells and one of the common problems encountered in such systems is the cell reversal, that is the reversal of the cell polarity, which turns some cells in the stack into electrolytic cells (Aelterman et al., 2006b).
4.18.10 Opportunities for Bioelectrochemical Systems In spite of the great need for improvement, MFCs are undoubtedly a promising technology that offers a vast range of potential applications. Bioelectrochemical wastewater treatment is a novel and promising approach to the production of renewable energy and thus has been the main focus of investigations. Nevertheless, the key characteristic of BESs highlights the fact that they decouple the oxidative and the reductive process, offering the unique opportunity of having clean electrons (reducing power) derived from renewable resources that can potentially be used to drive a multitude of biotechnological processes. Important examples for bioremediation processes include the reductive dechlorination of chlorinated compounds (Aulenta et al., 2007), the reduction
659
of soluble metals like hexavalent uranium into more insoluble forms (Gregory and Lovley, 2005), and denitrification (Clauwaert et al., 2007a, Virdis et al., 2008). Other niche applications of MFCs include 1. Electricity production in remote areas. MFCs can produce power from waste biomass, which is ubiquitous and could supply those areas with small amounts of electricity while reducing the environmental impact on local waterways. 2. Bacterial batteries. Energy stored in the form of sugars or other organic substrates can produce environmental friendly power that could be used for the small appliances. 3. Online sensors. The production of electric current in the presence of biodegradable material could be exploited for the online detection and quantification of soluble organics in waterways or wastewater treatment plant (WWTP) effluents.
4.18.10.1 Wastewater Treatment The power densities generated by MFCs are much smaller than those of chlorofluorocarbons (CFCs). Therefore, MFCs cannot compete with CFCs as power producers, but they become much more attractive if the production of electricity is combined with wastewater treatment. The COD contained in wastewaters can be thoroughly removed while producing a CO2-neutral power, which could potentially cover at least the electricity requirements of the WWTP. Complex substrates have been successfully used to generate power in MFCs, including domestic wastewater (Liu et al., 2004), anaerobic digesters’ effluent (Aelterman et al., 2006a), brewery wastewater (Feng et al., 2008), and paper-recycling wastewater (Huang and Logan, 2008). Additionally, if anodic carbon oxidation is coupled with cathodic nitrogen removal, the use of MFCs opens up new perspectives for an integrated and sustainable wastewater treatment process. Wastewater treatment with MFCs would also offer the unique feature of online monitoring of the process through current and electrode-potential measurement that can rapidly advise of system failures. A drop in the current coupled with a rise of the anodic potential would indicate, for instance, a drop in the catalytic activity of the anodic biofilm or a failure in the feeding system. The removal of organics contained in wastewater is considered as energy efficient when no energy is consumed for aeration. If this is provided by means of passive aeration, for instance, it would save around 0.7–2 kW h1 kg1 COD removed (Logan et al., 2008) for conventional aeration, and energy can indeed be harvested from the substrate, with a theoretical upper limit of 4.4 kW h1 kg1 COD. Moreover, as the energy gain for bacteria growing at the MFC anode is generally lower than that for aerobic processes (Section 4.18.6), sludge production would also be reduced. The bacterial-growth yield in MFCs is in fact expected to be between that of high energy-yielding aerobic processes (0.4– 0.6 g COD biomass g1 COD substrate according to Heijnen (1999)) and that of low energy-yielding anaerobic treatment (0.01–0.14 g COD biomass g1 COD substrate, Heijnen (1999)). Rabaey et al. (2003) reported a measured yield of 0.07–0.22 g COD biomass g1 COD substrate for a glucose-fed
660
Microbial Fuel Cells
MFC. Freguia et al. (2007b) reported growth yields at anodes ranging from 0 to 0.3 g COD biomass g1 COD substrate acetate. MFCs can achieve significant increased organic removal rates compared to aerobic processes. Laboratory reactors have in fact reached current densities of the order of B10 A m2 anode surface area (Fan et al., 2007; Torres et al., 2007). Rozendal et al. (2008a) evaluated that this would correspond to a volumetric wastewater treatment capacity of B7.1 kg COD m3 reactor d1, assuming a minimal compartment thickness of 1 cm. If the same performances could be obtained on a larger scale, wastewater treatment with MFC would even outcompete traditional aerobic treatments, which are able to process B0.5–2 kg COD m3 d1 (Logan et al., 2006), with few advantages:
•
•
•
MFCs can produce power and treat wastewater in a single stage, whereas anaerobic processes require expensive power-generation facilities: an anaerobic digester for the production of methane and a power-production stage in a gas turbine/engine. MFCs can obtain good effluent quality while treating dilute influents at temperatures below 20 1C (Pham et al., 2006), while anaerobic digesters work best with high-strength wastewater at increased temperatures (30–37 1C) and require further polishing of the effluent. In a (bio)electrochemical system, it is possible to utilize streams containing sulfur (Dutta et al., 2008; Rabaey et al., 2006) with no need for gas treatment, whereas the methane produced during anaerobic digestion normally contains traces of H2S that need to be removed before using the gas in a turbine (due to corrosion concerns and environmental regulations) or fed to a chemical fuel cell (as sulfide poisons the Pt catalyst).
However, to achieve practical implementation at a reasonable scale, several challenges will have to be solved (Section 4.18.9). Moreover, the capital costs of MFCs have to be reduced drastically as the material costs are very high and there is a limited economy of scale benefit, primarily due to the close anode/cathode distance required. In conclusion, MFCs should be considered more as a complementary system rather than as in competition with anaerobic digestion. Anaerobic digestion can be applied to the treatment of high-strength waste streams. Industrial effluents represent perfect examples of these applications. MFCs may operate better at a smaller scale, when anaerobic digestion would suffer from the high costs of gas treatment and handling; and with a more dilute waste stream such as, for example, the effluent from an anaerobic process. Furthermore, the challenge of wastewater complexity is yet to be addressed for real-scale applications. More studies using real wastewater are required to improve the knowledge of the degradation pathways of complex substances. Laboratory MFCs fed with wastewater mainly convert readily biodegradable organics, whereas more complex materials generally pass straight through the system. This can be due to the generally short hydraulic-retention time (Rabaey et al., 2005b) and the slower conversion rate of the more recalcitrant material.
4.18.10.2 Nitrogen Removal Nitrogen removal represents a topic of particular interest for MFC application. Currently, nitrogen is removed from wastewater by means of two sequential processes, both promoted by microorganisms: nitrification and denitrification. Nitrification is an autotrophic process that converts ammonium into nitrate using oxygen and an inorganic carbon source. Denitrification is instead a heterotrophic anoxic process that utilizes nitrate as an electron acceptor during the oxidation of an organic carbon source. Due to the competition between aerobic and anoxic organisms for the available organics, supplementary carbon supply is often used (typically methanol) in addition to the carbon already present in the wastewater, to increase the efficiency of denitrification. Cathodic denitrification was recently demonstrated in MFCs (Clauwaert et al., 2007a). In Virdis et al. (2008), the MFC was integrated with an external aerated vessel for nitrification, and the system was able to simultaneously remove carbon and nitrogen. As in the MFC configuration, the oxidative and reductive biomasses are kept physically separate by the IEM, and the competition between organisms can be minimized to achieve highly efficient denitrification at lower C/N ratios than generally required by heterotrophic denitrification. This represents an important advantage of MFCs as denitrification is driven by electrons directly supplied by the anode with no need for the organics to be added directly into the denitrification stage.
4.18.10.3 Bioremediation The possibility of removing metals and chlorinated compounds by means of bioelectrochemical systems is of particular importance for applications of this technology in bioremediation. A study by Gregory and Lovley (2005) reported the microbial reduction of uranium using an electrode as an electron donor which caused the conversion of soluble U(VI) into the rather insoluble U(IV), which precipitated onto the electrodes. Cathodic reduction of perchlorate, an industrial by-product (i.e., from the production of pyrotechnic compounds and lubricant oils) found in the environment due to a historical lack of regulation in its manufacturing and discharge, was also recently described (Thrash et al., 2007). Hydrogen likely served as an electron shuttle for dissimilatory perchlorate-reducing bacteria, although an isolate from a perchlorate-reducing reactor could accept electrons from the cathode via an added redox mediator. Reductive dechlorination of chlorinated compounds, such as trichloroethene (TCE), is typically achieved through the oxidation of an organic electron donor. It was recently shown that TCEdechlorinating bacteria could directly use a cathode as an electron donor (electrode polarization: 450 mV vs. SHE) (Aulenta et al., 2009).
4.18.10.4 H2 Production Hydrogen gas can effectively be produced through MECs. These are electrolysis-type BESs that are capable of producing hydrogen at the cathode on applying a small voltage (40.2 V in practice), while oxidizing organic matter at the anode (Liu et al., 2005c; Rozendal et al., 2006b).
Microbial Fuel Cells
The architecture of MECs is nearly identical to that of MFCs, except for the fact that an MEC requires gas collection at the cathode. Cathodic hydrogen production on plain carbon electrode is very slow due to high overpotentials. Platinum has been the most commonly used catalyst (Rozendal et al., 2006b). However, it was recently discovered that bacteria could be effectively used as catalysts for hydrogen production (Rozendal et al., 2008b), thus overcoming the disadvantages connected to the use of platinum-based cathodes (Section 4.18.5). More recently, the use of nickel and stainless steel in the form of flat sheets or brushes was observed to outcompete platinum as cathodic catalyst (Call et al., 2009; Selembo et al., 2009). MECs are a promising technology for sustainable hydrogen production from wastewater. While MFCs recover energy from wastewater in the form of electricity, MECs recover energy in the form of hydrogen. Nevertheless, to function as wastewater treatment systems, MECs need to guarantee reasonable COD conversion rates. Based on current H2-production performances, COD-loading rates would need to be of the order of 6.5 kg COD m3 d1 (Logan et al., 2008), which is between the range of activated sludge systems and anaerobic digesters, thus making the MEC technology competitive when compared with traditional wastewater treatment. However, MECs need to be more cost effective than existing technologies and since electric energy is consumed during their operation, the higher costs must be compensated for by sufficient hydrogen production. It is estimated that full-scale MEC systems require 1 kW h1 m3 of H2 produced and can produce up to 10 m3 H2 m3 d1 (Rozendal et al., 2007), which is equivalent to an energy requirement of B1.5 kW h1 kg1 COD treated (Logan et al., 2008), and which is similar to the energy consumption for activated sludge treatment (Rozendal et al., 2008a). On the contrary, energy recovery through anaerobic digestion does not require significant energy inputs. However, compared to MECs, anaerobic digestion produces a gas (methane) that is less valuable than hydrogen. On the other hand, anaerobic digestion is a well-established technology, whereas microbial electrolysis requires great research efforts on both engineering and biochemical aspects.
4.18.10.5 Bioelectrochemical Production of Value-Added Chemicals Currently, it is expected that the capital costs for a full-scale BES will always remain several times higher than that of conventional wastewater treatment systems (Rozendal et al., 2008a). Therefore, bioelectrochemical wastewater treatment will become economically advantageous when the larger investments are compensated for by the larger value of the products obtainable. Electricity production using MFCs has the disadvantage of its low revenue, which puts electricity among the least valuable products (Rozendal et al., 2008a). As we have seen in Section 4.18.10.4 energy can be recovered in a BES not only as electricity but also as hydrogen. In addition, BESs can offer other interesting opportunities to improve their economical feasibility. For instance, the hydrogen produced in a BES can be used to create other products in situ. Several researchers have already reported methane production as a side product in membrane-less MECs, due to hydrogen
661
scavenging (Call and Logan, 2008; Clauwaert and Verstraete, 2009). When the IEM is omitted, hydrogenotrophic methanogens can use the hydrogen produced at the cathode and combine it with the carbon dioxide produced at the anode, thus producing methane. Even though methane has a lower energy content compared to hydrogen per unit of mass, removing the IEM from the MECs would significantly lower its capital costs as well as reduce the system’s ohmic losses and pH gradients. MECs could thus be used in combination with anaerobic digestion facilities at the polishing stage by treating the residual organics present in the effluent (Clauwaert and Verstraete, 2009). In addition, direct methane production without intermediate hydrogen production was also recently observed in a biocathode dominated by Methanobacterium palustre (Cheng et al., 2009), demonstrating that BESs can be used to convert electricity into a biofuel while also capturing carbon dioxide. It is expected that, in future, BES innovations will proceed on these lines. A whole range of value-added chemicals requires the reduction of power for their production. When CO2 and O2 impurities are present together with H2, the production of biopolymers such as polyhydroxyalkanoates (PHA) by hydrogen-oxidizing bacteria can be foreseen in membrane-less or loop-based MECs (Ishizaki et al., 2001). Moreover, alcohols can also be produced from VFAs using hydrogen as an electron donor (Steinbusch et al., 2008), or a mediator (methyl viologen) (Steinbusch et al., 2009). Moreover, hydrogen peroxide production has been obtained by coupling organic oxidation at the anode with oxygen reduction at the cathode and adding a small voltage (Rozendal et al., 2009).
4.18.11 Outlook Current approaches to waste management will have to change in the future since waste will have to be considered as an alternative resource rather than an inconvenient burden to dispose of. Wastewater, in particular, represents an important resource of nutrient (primarily nitrogen and phosphorus), energy (as energy contained in chemical bonds of organic matter), and water itself. In a sustainable society, wastewater treatment will no longer be regarded as a treatment per se, its sole purpose being the removal of contaminants, often requiring a great deal of nonrenewable energy (e.g., from coal extraction), which may ultimately cause more environmental damage than the direct discharge of the untreated wastewater. In the future, we will no longer refer to wastewater treatment plants but rather to bioelectrochemical-resource-recovery plants, or biorefineries. With this new picture emerging, the raw wastewater would follow several sequential treatment stages, the first stage of which would be a pre-treatment to remove the solids, for instance, through dissolved air flotation. The solid fraction can then be sent to an anaerobic digester wherein some biogas is formed and solid liquid/separation creates a sludge that can be used for composting, while the supernatant can be sent back to the main flow. A pre-fermenter would be likely added after the pre-treatment to breakdown complex organics and produce an effluent richer in VFAs that are better metabolized in a BES anode for electron extraction. After the anodic
662
Microbial Fuel Cells
passage, the effluent would be primarily rich in nitrogen as ammonium, which can be recovered through thermal volume reduction, or as struvite. Further specific treatments would depend on the final utilization of the effluent; tertiary treatments will produce water suitable for use in other processes (as cooling water, for instance), or even be able to reach drinking standards through advanced treatment processes. The electrons harvested during the anodic passage would be conveyed to the cathodic side of the BES where they can be used to drive a wide array of processes, from direct electricity production through to MFC, or perhaps to produce hydrogen through MECs, or moreover to produce other value-added chemicals such as methane, hydrogen peroxide, alcohols, biopolymers, or biofuels as seen previously.
References Aelterman P (2009) Microbial Fuel Cells for the Treatment of Waste Streams with Energy Recovery. PhD Thesis, Ghent University. Aelterman P, Freguia S, Keller J, Verstraete W, and Rabaey K (2008) The anode potential regulates bacterial activity in microbial fuel cells. Applied Microbiology and Biotechnology 78: 409--418. Aelterman P, Rabaey K, Clauwaert P, and Verstraete W (2006a) Microbial fuel cells for wastewater treatment. Water Science and Technology 54: 9--15. Aelterman P, Rabaey K, Pham HT, Boon N, and Verstraete W (2006b) Continuous electricity generation at high voltages and currents using stacked microbial fuel cells. Environmental Science and Technology 40: 3388--3394. Allen RM and Bennetto HP (1993) Microbial fuel-cells – electricity production from carbohydrates. Applied Biochemistry and Biotechnology 39: 27--40. Angenent LT, Karim K, Al-Dahhan MH, Wrenn BA, and Domiguez-Espinosa R (2004) Production of bioenergy and biochemicals from industrial and agricultural wastewater. Trends in Biotechnology 22: 477--485. Aulenta F, Canosa A, Reale P, Rossetti S, Panero S, and Majone M (2009) Microbial reductive dechlorination of trichloroethene to ethene with electrodes serving as electron donors without the external addition of redox mediators. Biotechnology and Bioengineering 103: 85--91. Aulenta F, Catervi A, Majone M, Panero S, Reale P, and Rossetti S (2007) Electron transfer from a solid-state electrode assisted by methyl viologen sustains efficient microbial reductive dechlorination of TCE. Environmental Science and Technology 41: 2554--2559. Benziger JB, Satterfield MB, Hogarth WHJ, Nehlsen JP, and Kevrekidis IG (2006) The power performance curve for engineering analysis of fuel cells. Journal of Power Sources 155: 272--285. Bergel A, Feron D, and Mollica A (2005) Catalysis of oxygen reduction in PEM fuel cell by seawater biofilm. Electrochemistry Communications 7: 900--904. Bond DR, Holmes DE, Tender LM, and Lovley DR (2002) Electrode-reducing microorganisms that harvest energy from marine sediments. Science 295: 483--485. Bond DR and Lovley DR (2003) Electricity production by Geobacter sulfurreducens attached to electrodes. Applied and Environmental Microbiology 69: 1548--1555. Bond DR and Lovley DR (2005) Evidence for involvement of an electron shuttle in electricity generation by Geothrix fermentans. Applied and Environmental Microbiology 71: 2186--2189. Call D and Logan BE (2008) Hydrogen production in a single chamber microbial electrolysis cell lacking a membrane. Environmental Science and Technology 42: 3401--3406. Call DF, Merrill MD, and Logan BE (2009) High surface area stainless steel brushes as cathodes in microbial electrolysis cells. Environmental Science and Technology 43(6): 2179--2183. Carnot S (1824) Reflexions sur la puissance motrice du feu. Paris: Gauthier-Villars. Chang IS, Moon H, Bretschger O, et al. (2006) Electrochemically active bacteria (EAB) and mediator-less microbial fuel cells. Journal of Microbiology and Biotechnology 16: 163--177. Chaudhuri SK and Lovley DR (2003) Electricity generation by direct oxidation of glucose in mediatorless microbial fuel cells. Nature Biotechnology 21: 1229--1232. Cheng S, Dempsey BA, and Logan BE (2007) Electricity generation from synthetic acid-mine drainage (AMD) water using fuel cell technologies. Environmental Science and Technology 41: 8149--8153.
Cheng S, Liu H, and Logan BE (2006a) Increased performance of single-chamber microbial fuel cells using an improved cathode structure. Electrochemistry Communications 8: 489--494. Cheng S, Liu H, and Logan BE (2006b) Power densities using different cathode catalysts (Pt and CoTMPP) and polymer binders (Nafion and PTFE) in single chamber microbial fuel cells. Environmental Science and Technology 40: 364--369. Cheng S and Logan BE (2007) Ammonia treatment of carbon cloth anodes to enhance power generation of microbial fuel cells. Electrochemistry Communications 9: 492--496. Cheng S, Xing D, Call DF, and Logan BE (2009) Direct biological conversion of electrical current into methane by electromethanogenesis. Environmental Science and Technology 43: 3953--3958. Clauwaert P, Rabaey K, Aelterman P, et al. (2007a) Biological denitrification in microbial fuel cells. Environmental Science and Technology 41: 3354-3360. Clauwaert P, Van der Ha D, Boon N, et al. (2007b) Open air biocathode enables effective electricity generation with microbial fuel cells. Environmental Science and Technology 41: 7564--7569. Clauwaert P and Verstraete W (2009) Methanogenesis in membraneless microbial electrolysis cells. Applied Microbiology and Biotechnology 82: 829--836. Cohen B (1931) The bacterial culture as an electrical half-cell. Journal of Bacteriology 21: 18--19. Delaney GM, Bennetto HP, Mason JR, Roller SD, Stirling JL, and Thurston CF (1984) Electron-transfer coupling in microbial fuel-cells. 2. Performance of fuel-cells containing selected microorganism mediator substrate combinations. Journal of Chemical Technology and Biotechnology B – Biotechnology 34: 13--27. Dumas C, Mollica A, Feron D, Basseguy R, Etcheverry L, and Bergel A (2007) Marine microbial fuel cell: Use of stainless steel electrodes as anode and cathode materials. Electrochimica Acta 53: 468--473. Dutta PK, Rabaey K, Yuan ZG, and Keller J (2008) Spontaneous electrochemical removal of aqueous sulfide. Water Research 42: 4965--4975. Fan YZ, Hu HQ, and Liu H (2007) Enhanced Coulombic efficiency and power density of air-cathode microbial fuel cells with an improved cell configuration. Journal of Power Sources 171: 348--354. Feng Y, Wang X, Logan BE, and Lee H (2008) Brewery wastewater treatment using aircathode microbial fuel cells. Applied Microbiology and Biotechnology 78: 873--880. Finkelstein DA, Tender LM, and Zeikus JG (2006) Effect of electrode potential on electrode-reducing microbiota. Environmental Science and Technology 40: 6990--6995. Foley J and Keller J (2008) Current wastewater treatment processes. In: Gardener T, Yeates C, and Shaw R (eds.) Purified Recycled Water for Drinking: The Technical Issues, pp. 57--84. Brisbane, QLD: Queensland Water Commission. Freguia S (2008) Fundamental Studies of Anodic and Cathodic Processes in Microbial Fuel Cells. PhD Thesis, The University of Queensland. Freguia S, Rabaey K, and Keller J (2007a) Competition and interaction between fermentation and electricity generation in microbial fuel cells anodes. In: Proceeding of the 11th International Water Association World Congress on Anaerobic Digestion (AD11). Brisbane, QLD, Australia, 23–27 September 2007. Freguia S, Rabaey K, Yuan Z, and Keller J (2007b) Electron and carbon balances in microbial fuel cells reveal temporary bacterial storage behavior during electricity generation. Environmental Science and Technology 41: 2915--2921. Freguia S, Rabaey K, Yuan Z, and Keller J (2007c) Non-catalyzed cathodic oxygen reduction at graphite granules in microbial fuel cells. Electrochimica Acta 53: 598--603. Freguia S, Rabaey K, Yuan Z, and Keller J (2008a) Sequential anode–cathode configuration improves cathodic oxygen reduction and effluent quality of microbial fuel cells. Water Research 42: 1387--1396. Freguia S, Rabaey K, Yuan Z, and Keller J (2008b) Syntrophic processes drive the conversion of glucose in microbial fuel cell anodes. Environmental Science and Technology 42: 7937--7943. Gil GC, Chang IS, Kim BH, et al. (2003) Operational parameters affecting the performance of a mediator-less microbial fuel cell. Biosensors and Bioelectronics 18: 327--334. Gorby YA, Yanina S, McLean JS, et al. (2006) Electrically conductive bacterial nanowires produced by Shewanella oneidensis strain MR-1 and other microorganisms. Proceedings of the National Academy of Sciences of the United States of America 103: 11358--11363. Gralnick JA and Newman DK (2007) Extracellular respiration. Molecular Microbiology 65(1): 1--11. Gregory KB, Bond DR, and Lovley DR (2004) Graphite electrodes as electron donors for anaerobic respiration. Environmental Microbiology 6: 596--604.
Microbial Fuel Cells
Gregory KB and Lovley DR (2005) Remediation and recovery of uranium from contaminated subsurface environments with electrodes. Environmental Science and Technology 39: 8943--8947. Habermann W and Pommer E-H (1991) Biological fuel cells with sulphide storage capacity. Applied Microbiology and Biotechnology 35: 128--133. Harnisch F, Schroder U, and Scholz F (2008) The suitability of monopolar and bipolar ion exchange membranes as separators for biological fuel cells. Environmental Science and Technology 42: 1740--1746. Hatti-Kaul R, Tornvall U, Gustafsson L, and Borjesson P (2007) Industrial biotechnology for the production of bio-based chemicals – a cradle-to-grave perspective. Trends in Biotechnology 25: 119--124. He Z and Angenent LT (2006) Application of bacterial biocathodes in microbial fuel cells. Electroanalysis 18: 2009--2015. He Z, Minteer SD, and Angenent LT (2005) Electricity generation from artificial wastewater using an upflow microbial fuel cell. Environmental Science and Technology 39: 5262--5267. He Z, Shao H, and Angenent LT (2007) Increased power production from a sediment microbial fuel cell with a rotating cathode. Biosensors and Bioelectronics 22: 3252--3255. Heijnen JJ (1999) Bioenergetics of microbial growth. In: Flickinger MC and Drew SW (eds.) Bioprocess Technology: Fermentation, Biocatalisys, Bioseparation, pp. 267--291. New York: Wiley. Heilmann J and Logan BE (2006) Production of electricity from proteins using a microbial fuel cell. Water Environment Research 78: 531--537. Hernandez ME, Kappler A, and Newman DK (2004) Phenazines and other redox-active antibiotics promote microbial mineral reduction. Applied and Environmental Microbiology 70: 921--928. Holmes DE, Bond DR, O’Neill RA, Reimers CE, Tender LR, and Lovley DR (2004) Microbial communities associated with electrodes harvesting electricity from a variety of aquatic sediments. Microbial Ecology 48: 178--190. Huang L and Logan BE (2008) Electricity generation and treatment of paper recycling wastewater using a microbial fuel cell. Applied Microbiology and Biotechnology 80: 349--355. Ishii S, Kosaka T, Hori K, Hotta Y, and Watanabe K (2005) Coaggregation facilitates interspecies hydrogen transfer between Pelotomaculum thermopropionicum and Methanothermobacter thermautotrophicus. Applied and Environmental Microbiology 71: 7838--7845. Ishii S, Shimoyama T, Hotta Y, and Watanabe K (2008) Characterization of a filamentous biofilm community established in a cellulose-fed microbial fuel cell. BMC Microbiology 8: doi: 10.1186/1471-2180-8-6. Ishizaki A, Tanaka K, and Taga N (2001) Microbial production of poly-D-3hydroxybutyrate from CO2. Applied Microbiology and Biotechnology 57: 6--12. Jang JK, Pham TH, Chang IS, et al. (2004) Construction and operation of a novel mediator- and membrane-less microbial fuel cell. Process Biochemistry 39: 1007--1012. Kargi F and Eker S (2007) Electricity generation with simultaneous wastewater treatment by a microbial fuel cell (MFC) with Cu and Cu–Au electrodes. Journal of Chemical Technology and Biotechnology 82: 658--662. Kim BH, Ikeda T, Park HS, et al. (1999) Electrochemical activity of an Fe(III)-reducing bacterium, Shewanella putrefaciens IR-1, in the presence of alternative electron acceptors. Biotechnology Techniques 13: 475--478. Kim HJ, Park HS, Hyun MS, Chang IS, Kim M, and Kim BH (2002) A mediator-less microbial fuel cell using a metal reducing bacterium, Shewanella putrefaciens. Enzyme and Microbial Technology 30: 145--152. Kim JR, Cheng S, Oh S-E, and Logan BE (2007a) Power generation using different cation, anion, and ultrafiltration membranes in microbial fuel cells. Environmental Science and Technology 41: 1004--1009. Kim JR, Jung SH, Regan JM, and Logan BE (2007b) Electricity generation and microbial community analysis of alcohol powered microbial fuel cells. Bioresource Technology 98: 2568--2577. Kim N, Choi Y, Jung S, and Kim S (2000) Effect of initial carbon sources on the performance of microbial fuel cells containing Proteus vulgaris. Biotechnology and Bioengineering 70: 109--114. Kuroda M, Watanabe T, and Umedu Y (1997) Simultaneous COD removal and denitrification of wastewater by bio-electro reactors. Water Science and Technology 35: 161--168. Larsen TA, Maurer M, Udert KM, and Lienert J (2007) Nutrient cycles and resource management: Implications for the choice of wastewater treatment technology. Water Science and Technology 56: 229--237. Lee HS, Parameswaran P, Kato-Marcus A, Torres CI, and Rittmann BE (2008) Evaluation of energy-conversion efficiencies in microbial fuel cells (MFCs) utilizing fermentable and non-fermentable substrates. Water Research 42: 1501--1510.
663
Lee JY, Phung NT, Chang IS, Kim BH, and Sung HC (2003) Use of acetate for enrichment of electrochemically active microorganisms and their 16S rDNA analyses. FEMS Microbiology Letters 223: 185--191. Lewis K (1966) Symposium on bioelectrochemistry of microorganisms. 4. Biochemical fuel cells. Bacteriological Reviews 30: 101--113. Lewis NS and Nocera DG (2006) Powering the planet: Chemical challenges in solar energy utilization. Proceedings of the National Academy of Sciences of the United States of America 103: 15729--15735. Liu H, Cheng SA, and Logan BE (2005a) Power generation in fed-batch microbial fuel cells as a function of ionic strength, temperature, and reactor configuration. Environmental Science and Technology 39: 5488--5493. Liu H, Cheng SA, and Logan BE (2005b) Production of electricity from acetate or butyrate using a single-chamber microbial fuel cell. Environmental Science and Technology 39: 658--662. Liu H, Grot S, and Logan BE (2005c) Electrochemically assisted microbial production of hydrogen from acetate. Environmental Science and Technology 39: 4317--4320. Liu H and Logan BE (2004) Electricity generation using an air-cathode single chamber microbial fuel cell in the presence and absence of a proton exchange membrane. Environmental Science and Technology 38: 4040--4046. Liu H, Ramnarayanan R, and Logan BE (2004) Production of electricity during wastewater treatment using a single chamber microbial fuel cell. Environmental Science and Technology 38: 2281--2285. Liu JL, Lowy DA, Baumann RG, and Tender LM (2007) Influence of anode pretreatment on its microbial colonization. Journal of Applied Microbiology 102: 177--183. Logan B (2008) Microbial Fuel Cells. Hoboken, NJ: Wiley. Logan B, Cheng S, Watson V, and Estadt G (2007) Graphite fiber brush anodes for increased power production in air-cathode microbial fuel cells. Environmental Science and Technology 41: 3341--3346. Logan BE, Call D, Cheng S, et al. (2008) Microbial electrolysis cells for high yield hydrogen gas production from organic matter. Environmental Science and Technology 42: 8630--8640. Logan BE, Hamelers B, Rozendal R, et al. (2006) Microbial fuel cells: Methodology and technology. Environmental Science and Technology 40: 5181--5192. Logan BE, Murano C, Scott K, Gray ND, and Head IM (2005) Electricity generation from cysteine in a microbial fuel cell. Water Research 39: 942--952. Logan BE and Regan JM (2006) Electricity-producing bacterial communities in microbial fuel cells. Trends in Microbiology 14: 512--518. Lovley DR (2008) The microbe electric: Conversion of organic matter to electricity. Current Opinion in Biotechnology 19: 564--571. Lovley DR, Coates JD, Blunt-Harris EL, Phillips EJP, and Woodward JC (1996) Humic substances as electron acceptors for microbial respiration. Nature 382: 445--448. Lovley DR and Phillips EJP (1988) Novel mode of microbial energy-metabolism – organic-carbon oxidation coupled to dissimilatory reduction of iron or manganese. Applied and Environmental Microbiology 54: 1472--1480. Lovley DR, Stolz JF, Nord GL, and Phillips EJP (1987) Anaerobic production of magnetite by a dissimilatory iron-reducing microorganism. Nature 330: 252--254. Madigan MT, Martinko J, and Parker J (2000) Brock Biology of Microorganisms. Upper Saddle River, NJ: Prentice Hall. Maurer M, Schwegler P, and Larsen TA (2003) Nutrients in urine: Energetic aspects of removal and recovery. Water Science and Technology 48: 37--46. Mellor RB, Ronnenberg J, Campbell WH, and Diekmann S (1992) Reduction of nitrate and nitrite in water by immobilized enzymes. Nature 355: 717--719. Min B, Kim JR, Oh SE, Regan JM, and Logan BE (2005) Electricity generation from swine wastewater using microbial fuel cells. Water Research 39: 4961--4968. Min B and Logan BE (2004) Continuous electricity generation from domestic wastewater and organic substrates in a flat plate microbial fuel cell. Environmental Science and Technology 38: 5809--5814. Myers CR and Myers JM (1992) Localization of cytochromes to the outer-membrane of anaerobically grown Shewanella putrefaciens MR-1. Journal of Bacteriology 174: 3429--3438. Myers CR and Nealson KH (1988) Bacterial manganese reduction and growth with manganese oxide as the sole electron-acceptor. Science 240: 1319--1321. Nakanishi-Matsui M and Futai M (2008) Stochastic rotational catalysis of proton pumping F-ATPase. Philosophical Transactions of the Royal Society B: Biological Sciences 363(1500): 2135--2142. Niele F (2005) Energy: Engine of Evolution. Amsterdam: Elsevier. Niessen J, Schroder U, Rosenbaum M, and Scholz F (2004) Fluorinated polyanilines as superior materials for electrocatalytic anodes in bacterial fuel cells. Electrochemistry Communications 6: 571--575. Oh S, Min B, and Logan BE (2004) Cathode performance as a factor in electricity generation in microbial fuel cells. Environmental Science and Technology 38: 4900--4904.
664
Microbial Fuel Cells
Oh SE and Logan BE (2006) Proton exchange membrane and electrode surface areas as factors that affect power generation in microbial fuel cells. Applied Microbiology and Biotechnology 70: 162--169. Park DH and Zeikus JG (1999) Utilization of electrically reduced neutral red by Actinobacillus succinogenes: Physiological function of neutral red in membranedriven fumarate reduction and energy conservation. Journal of Bacteriology 181: 2403--2410. Park DH and Zeikus JG (2000) Electricity generation in microbial fuel cells using neutral red as an electronophore. Applied and Environmental Microbiology 66: 1292--1297. Park DH and Zeikus JG (2002) Impact of electrode composition on electricity generation in a single-compartment fuel cell using Shewanella putrefaciens. Applied Microbiology and Biotechnology 59: 58--61. Park DH and Zeikus JG (2003) Improved fuel cell and electrode designs for producing electricity from microbial degradation. Biotechnology and Bioengineering 81: 348--355. Park HI, Kim DK, Choi YJ, and Pak D (2005) Nitrate reduction using an electrode as direct electron donor in a biofilm-electrode reactor. Process Biochemistry 40: 3383--3388. Petrus L and Noordermeer MA (2006) Biomass to biofuels, a chemical perspective. Green Chemistry 8: 861--867. Pham TH, Jang JK, Chang IS, and Kim BH (2004) Improvement of cathode reaction of a mediatorless microbial fuel cell. Journal of Microbiology and Biotechnology 14: 324--329. Pham TH, Rabaey K, Aelterman P, et al. (2006) Microbial fuel cells in relation to conventional anaerobic digestion technology. Engineering in Life Sciences 6: 285--292. Piccolino M (1997) Luigi Galvani and animal electricity: Two centuries after the foundation of electrophysiology. Trends in Neurosciences 20: 443--448. Potter MC (1911) Electrical effects accompanying the decomposition of organic compounds. Proceedings of the Royal Society of London Series B – Containing Papers of a Biological Character 84: 260--276. Rabaey K, Boon N, Hofte M, and Verstraete W (2005a) Microbial phenazine production enhances electron transfer in biofuel cells. Environmental Science and Technology 39: 3401--3408. Rabaey K, Boon N, Siciliano SD, Verhaege M, and Verstraete W (2004a) Biofuel cells select for microbial consortia that self-mediate electron transfer. Applied and Environmental Microbiology 70: 5373--5382. Rabaey K, Clauwaert P, Aelterman P, and Verstraete W (2005b) Tubular microbial fuel cells for efficient electricity generation. Environmental Science and Technology 39: 8077--8082. Rabaey K, Lissens G, Siciliano SD, and Verstraete W (2003) A microbial fuel cell capable of converting glucose to electricity at high rate and efficiency. Biotechnology Letters 25: 1531--1535. Rabaey K, Read ST, Clauwaert P, et al. (2008) Cathodic oxygen reduction catalyzed by bacteria in microbial fuel cells. ISME Journal 2: 519--527. Rabaey K, Rodriguez J, Blackall LL, et al. (2007) Microbial ecology meets electrochemistry: Electricity-driven and driving communities. ISME Journal 1: 9--18. Rabaey K, Van De Sompel K, Maignien L, et al. (2006) Microbial fuel cells for sulfide removal. Environmental Science and Technology 40: 5218--5224. Rabaey K and Verstraete W (2005) Microbial fuel cells: Novel biotechnology for energy generation. Trends in Biotechnology 23: 291--298. Ragauskas AJ, Williams CK, Davison BH, et al. (2006) The path forward for biofuels and biomaterials. Science 311: 484--489. Reguera G, McCarthy KD, Mehta T, Nicoll JS, Tuominen MT, and Lovley DR (2005) Extracellular electron transfer via microbial nanowires. Nature 435: 1098--1101. Reimers CE, Tender LM, Fertig S, and Wang W (2001) Harvesting energy from the marine sediment–water interface. Environmental Science and Technology 35: 192--195. Rhoads A, Beyenal H, and Lewandowski Z (2005) Microbial fuel cell using anaerobic respiration as an anodic reaction and biomineralized manganese as a cathodic reactant. Environmental Science and Technology 39: 4666--4671. Richardson DJ (2000) Bacterial respiration: A flexible process for a changing environment. Microbiology – SGM 146: 551--571. Rieger PH (1994) Electrochemistry. New York: Chapman and Hall. Rozendal RA, Hamelers HVM, and Buisman CJN (2006a) Effects of membrane cation transport on pH and microbial fuel cell performance. Environmental Science and Technology 40: 5206--5211. Rozendal RA, Hamelers HVM, Euverink GJW, Metz SJ, and Buisman CJN (2006b) Principle and perspectives of hydrogen production through biocatalyzed electrolysis. International Journal of Hydrogen Energy 31: 1632--1640.
Rozendal RA, Hamelers HVM, Molenkamp RJ, and Buisman CJN (2007) Performance of single chamber biocatalyzed electrolysis with different types of ion exchange membranes. Water Research 41: 1984--1994. Rozendal RA, Hamelers HVM, Rabaey K, Keller J, and Buisman CJN (2008a) Towards practical implementation of bioelectrochemical wastewater treatment. Trends in Biotechnology 26: 450--459. Rozendal RA, Jeremiasse AW, Hamelers HVM, and Buisman CJN (2008b) Hydrogen production with a microbial biocathode. Environmental Science and Technology 42: 629--634. Rozendal RA, Leone E, Keller J, and Rabaey K (2009) Efficient hydrogen peroxide generation from organic matter in a bioelectrochemical system. Electrochemistry Communications 11: 1752--1755. Rozendal RA, Sleutels T, Hamelers HVM, and Buisman CJN (2008c) Effect of the type of ion exchange membrane on performance, ion transport, and pH in biocatalyzed electrolysis of wastewater. Water Science and Technology 57: 1757--1762. Sakakibara Y, Araki K, Tanaka T, Watanabe T, and Kuroda M (1994) Denitrification and neutralization with an electrochemical and biological reactor. Water Science and Technology 30: 151--155. Sakakibara Y and Kuroda M (1993) Electric prompting and control of denitrification. Biotechnology and Bioengineering 42: 535--537. Schaetzle O, Barriere F, and Schroder U (2009) An improved microbial fuel cell with laccase as the oxygen reduction catalyst. Energy and Environmental Science 2: 96--99. Schroder U (2007) Anodic electron transfer mechanisms in microbial fuel cells and their energy efficiency. Physical Chemistry Chemical Physics 9(21): 2619--2629. Schroder U and Harnisch F (2009) Electrochemical losses. In: Rabaey K, Angenent LT, Schroder U, and Keller J (eds.) Bio-Electrochemical Systems: From Extracellular Electron Transfer to Biotechnological Application, pp. 119–133. IWA Integrated Environmental Technology Series. IWA. Schroder U, Niessen J, and Scholz F (2003) A generation of microbial fuel cells with current outputs boosted by more than one order of magnitude. Angewandte Chemie – International Edition 42: 2880--2883. Seboldt W (2004) Space- and earth-based solar power for the growing energy needs of future generations. Acta Astronautica 55: 389--399. Selembo PA, Merrill MD, and Logan BE (2009) The use of stainless steel and nickel alloys as low-cost cathodes in microbial electrolysis cells. Journal of Power Sources 190(2): 271--278. Shea C, Clauwaert P, Verstraete W, and Nerenberg R (2008) Adapting a denitrifying biocathode for perchlorate reduction. Water Science Technology 58: 1941--1946. Shin SH, Choi Y, Na SH, Jung S, and Kim S (2006) Development of bipolar plate stack type microbial fuel cells. Bulletin of the Korean Chemical Society 27: 281--285. Shukla AK, Suresh P, Berchmans S, and Rajendran A (2004) Biological fuel cells and their applications. Current Science 87: 455--468. Steinbusch KJ, Hamelers HV, Schaap JD, Kampman C, and Buisman CJ (2009) Bioelectrochemical ethanol production through mediated acetate reduction by mixed cultures. Environmental Science and Technology 44: 513--517. Steinbusch KJJ, Hamelers HVM, and Buisman CJN (2008) Alcohol production through volatile fatty acids reduction with hydrogen as electron donor by mixed cultures. Water Research 42: 4059--4066. Tanisho S, Kamiya N, and Wakao N (1989) Microbial fuel cell using Enterobacter aerogenes. Bioelectrochemistry and Bioenergetics 21: 25--32. Tchobanoglous G, Burton FL, and Stensel HD (2003) Wastewater Engineering, Treatment and Reuse. New York: McGraw-Hill. Tender LM, Reimers CE, Stecher HA, et al. (2002) Harnessing microbially generated power on the seafloor. Nature Biotechnology 20: 821--825. Ter Heijne A, Hamelers HVM, and Buisman CJN (2007) Microbial fuel cell operation with continuous biological ferrous iron oxidation of the catholyte. Environmental Science and Technology 41: 4130--4134. Ter Heijne A, Hamelers HVM, Saakes M, and Buisman CJN (2008) Performance of non-porous graphite and titanium-based anodes in microbial fuel cells. Electrochimica Acta 53: 5697--5703. Thauer RK, Jungermann K, and Decker K (1977) Energy-conservation in chemotropic anaerobic bacteria. Bacteriological Reviews 41: 100--180. Thrash JC, VanTrump JI, Weber KA, Miller E, Achenbach LA, and Coates JD (2007) Electrochemical stimulation of microbial perchlorate reduction. Environmental Science and Technology 41: 1740--1746. Torres CI, Kato Marcus A, and Rittmann BE (2007) Kinetics of consumption of fermentation products by anode-respiring bacteria. Applied Microbiology and Biotechnology 77: 689--697. Torres CI, Marcus AK, and Rittmann BE (2008) Proton transport inside the biofilm limits electrical current generation by anode-respiring bacteria. Biotechnology and Bioengineering 100: 872--881.
Microbial Fuel Cells
van Wyk JPH (2001) Biotechnology and the utilization of biowaste as a resource for bioproduct development. Trends in Biotechnology 19: 172--177. Velasquez-Orta SB, Curtis TP, and Logan BE (2009) Energy from algae using microbial fuel cells. Biotechnology and Bioengineering 103(6): 1068--1076. Virdis B, Rabaey K, Yuan Z, and Keller J (2008) Microbial fuel cells for simultaneous carbon and nitrogen removal. Water Research 42: 3013--3024. Virdis B, Rabaey K, Yuan Z, Rozendal R, and Keller J (2009) Electron fluxes in a microbial fuel cell performing carbon and nitrogen removal. Environmental Science and Technology 43: 5144--5149. Wang X, Feng YJ, and Lee H (2007) Electricity production from beer brewery wastewater using single chamber microbial fuel cell. In: 11th World Congress on International Water Association World Congress on Anaerobic Digestion (AD11), Brisbane, QLD, Australia, 23–27 September, 2007. White D (1995) The Physiology and Biochemistry of Prokaryotes. New York: Oxford University Press. Zhao F, Harnisch F, Schroder U, Scholz F, Bogdanoff P, and Herrmann I (2005) Application of pyrolysed iron(II) phthalocyanine and CoTMPP based oxygen
665
reduction catalysts as cathode materials in microbial fuel cells. Electrochemistry Communications 7: 1405--1410. Zhao F, Harnisch F, Schroder U, Scholz F, Bogdanoff P, and Herrmann I (2006) Challenges and constraints of using oxygen cathodes in microbial fuel cells. Environmental Science and Technology 40: 5193--5199.
Relevant Website http://www.wri.org World Research Institute (WRI).
4.19 Water in the Pulp and Paper Industry H Jung and D Pauly, Papiertechnische Stiftung, Munich, Germany & 2011 Elsevier B.V. All rights reserved.
4.19.1 4.19.2 4.19.2.1 4.19.2.2 4.19.2.3 4.19.3 4.19.3.1 4.19.3.2 4.19.3.2.1 4.19.3.2.2 4.19.3.2.3 4.19.3.2.4 4.19.3.2.5 4.19.3.2.6 4.19.3.3 4.19.3.3.1 4.19.3.3.2 4.19.4 4.19.4.1 4.19.4.2 4.19.4.2.1 4.19.4.2.2 4.19.4.2.3 4.19.4.3 4.19.4.3.1 4.19.4.3.2 4.19.4.3.3 4.19.5 4.19.5.1 4.19.5.2 4.19.5.3 4.19.6 References
Overview of Pulp and Papermaking Water in the Pulp and Paper industry Functions of Water in Papermaking Historical Evolution of Water Systems Current Water Consumption Levels in the Pulp and Paper Industry Water Use Freshwater Process Water Circuitry Primary, secondary, and tertiary water circuits Detrimental substances General principles of circuitry Closed water circuits Assessment of freshwater use and circuitry Wastewater Characterization of wastewater from the pulp and paper industry Wastewater discharging Water Treatment Freshwater Treatment Circuit Water Treatment Objectives of circuit water treatment Mechanical circuit water treatment Advanced circuit water treatment Wastewater Treatment Preliminary mechanical treatment: Mechanical processes for removal of solids Biological treatment Advanced and tertiary treatment Potentials and Limits of Water Saving Limiting Effects of System Closure Heat Balance Economic Benefits Improving Water Efficiency in Paper Manufacturing Industries – 30 Years of Success
4.19.1 Overview of Pulp and Papermaking Paper is currently a commodity product. The worldwide consumption of paper is growing steadily and it is hard to imagine the world without paper. Papermaking is based on a principle that is roughly 2000 years old. Today, in principle, the same process steps, which were used in the past, are included. The papermaking process can be divided into four main process steps (Figure 1), which can be either integrated at one site or located at several different sites. These process steps include
• • • •
pulp production (chemical or mechanical pulp and recycled fiber pulp (RCF)), stock preparation, paper machine, and coating and finishing.
Papermaking starts with the provision of the stock components such as fibers, fillers, and chemical additives. Primary
667 667 667 667 667 668 668 668 668 669 669 670 670 671 673 673 674 674 674 675 675 675 676 678 678 678 679 680 680 680 681 682 682
fibers (chemical and mechanical pulp) are obtained from wood and annual plants by chemical pulping or mechanical defibration. Secondary fibers are produced from recovered paper. All these components have to be properly prepared for optimum use in papermaking. Stock preparation is followed by the approach flow system, which links stock preparation to the paper machine. Paper or board is produced at the paper machine. In doing so, a sheet is formed from a highly diluted fiber suspension and dewatered by means of filtration, pressing, and thermal drying. Coating and calendaring improve the surface quality of the paper and board. The final steps include slitting, sheeting, and packaging of the final product for shipment.
4.19.2 Water in the Pulp and Paper industry 4.19.2.1 Functions of Water in Papermaking Water is one of the key components in pulp and papermaking. Without water, the production of paper would be unthinkable.
667
668
Water in the Pulp and Paper Industry Raw material Wood / recovered paper / annual plants
Pulp production Chemical / thermal / mechanical
Pulp
Stock preparation Suspending / screening and cleaning / refining
Machine stock
Papermaking process Approach flow / paper machine
Paper / board Coating / finishing
Final product Figure 1 Process steps in papermaking.
Water performs numerous functions in the papermaking process. It is used as medium for suspension, dwelling, and transfer processes, and it serves to separate as well as to restore the bonds between fibers. Other uses include showers in the wire and press sections, sealing of pumps, and cooling and cleaning purposes. Furthermore, water in the form of steam is used as an energy carrier.
4.19.2.3 Current Water Consumption Levels in the Pulp and Paper Industry At the onset of industrial papermaking, paper was produced with high specific water consumption. The pulp and paper industry has improved the processes in the last few decades for economical and ecological reasons and, as a result, was able to reduce water consumption significantly. This was only possible because of increasing closure of in-mill water circuits and consistent reuse of clarified process water by former freshwater consumers. A survey conducted by the Papiertechnische Stiftung (PTS) and the German Pulp and Paper Association (VDP) showed that the average specific effluent volumes of Germany’s pulp and paper industry decreased from 46 to approximately 10 m3 per metric ton of product produced between 1974 and 2007 (Figure 2). Nevertheless, the German pulp and paper industry remains one of the six biggest consumers of industrial water (Federal Statistical Office, 2008). The consumption level in the different pulp and paper mills can vary because of both general and process-related reasons such as raw materials used, paper grades produced, and plant structure. Furthermore, local boundary conditions, such as requirements on wastewater discharge, have an impact on the consumption level. High specific effluent volumes occur particularly in specialty paper grades. These mills are often faced with structural handicaps that cause increased specific effluent volumes: small and obsolete paper machines, low production rates, frequent grade changes, and often very high quality requirements on the final product. The lowest water requirements can be found in mills that produce packaging papers, such as corrugated base paper or board. Some of these mills have already managed to close their water circuits completely, resulting in a zero effluent production.
4.19.3 Water Use 4.19.3.1 Freshwater
4.19.2.2 Historical Evolution of Water Systems According to Zippel (2001), there are three phases in the historical evolution of paper-mill water systems. Phase 1 began in the 1920s. During this phase, the basics of water circuit design were established. Freshwater saving potentials were initially introduced predominantly for economic reasons. Phase 2 began in the 1960s. During this phase, the final effluent became more important for paper mills. This was caused as a means of reducing solid losses and thereby increasing the yield on the one hand and, on the other hand, as a result of the increasing ecological awareness of the general population. Subsequently, mechanical and biological wastewater treatment plants were installed. A few mills even managed to close their water circuits completely. Phase 3 (starting in the 1970s) was marked by initial attempts undertaken to deal with the consequences of system closure. Thus, the third phase was characterized by basic investigational work on the constituents of the process water and their impact on runnability of the paper machine and paper quality.
Depending on the availability and local conditions, either surface water or groundwater is used as freshwater. Drinking water is used for certain purposes, such as trim squirts. In the German pulp and paper industry, roughly 80% of the fresh water is taken from surface waters (Jung et al., 2009). In stateof-the-art mills, there are only few freshwater consumers. Typical freshwater consumers include
• • • •
high- and low-pressure showers for felt conditioning and wire cleaning, trim squirts, sealing water for liquid-ring vacuum pumps and packing glands, and additive preparation and dilution.
In view of the limited freshwater volume available, it must be used efficiently. Hence, freshwater used for cooling purposes (oil coolers and steam condensers) should be collected and reused as fresh warm water in the paper machine. Process water should be used for all other purposes, such as stock dilution, consistency control, or cleaning.
Water in the Pulp and Paper Industry
669
Specific effluent volume (m3 per metric ton of product)
50
40
30
20
10
0 1972
1976
1980
1984
1988
1992
1996
2000
2004
2008
Year Figure 2 Averaged specific effluent volume in the German pulp and paper industry (Jung et al., 2009).
Typically, approximately 40% of the entire freshwater volume is used for the high- and low-pressure showers in the wire and press sections. Depending on the paper grade and the nozzles used, different flow rates are used for showers in the wire and press sections. The entire consumption in the European paper industry for both showers in the wire section and showers in the press section averages out to approximately 1.0–2.5 m3 per metric ton of paper, depending on the degree of water circuit closure (Kappen et al., 2004). The sealing water consumption in liquid-ring vacuum pumps is highly dependent on the installed system. If a sealing water circuit is installed with a cooling tower, the freshwater can be less than 0.5 m3 per metric ton of paper. Without a sealing water circuit, the consumption typically amounts to approximately 4–5 m3 per metric ton of paper (Kappen et al., 2004). Sealing water is needed in packing glands to lubricate the sealing faces and to remove solids. According to Kappen et al. (2004), freshwater requirements amount to 0.15 m3 h1 in pumps and agitators and 0.2 m3 h1 in refiners and deflakers.
4.19.3.2 Process Water 4.19.3.2.1 Circuitry In papermaking, it is quite important to provide both adequate water quality and the required volume of water for every single consumer. Using freshwater for all purposes would consume several 100 m3 per metric ton of paper. The objectives of the water circuit system are to provide the required amount and quality of water for every consumer paying attention to economical and, at the same time also, to ecological aspects. In meeting these requirements, most of the water used in the pulp and paper industry is process water that has been recycled in different loops. Hence, the installation and proper design of water circuits are of fundamental importance for pulp and papermaking, since it contributes to enhanced product quality and reduced effluent volume.
Process water is mainly used for pulping and consistency control in the individual process steps. It is also used to a greater extent for purposes for which freshwater was formerly used, such as (low-pressure) showers, foam destruction, sealing water of liquid-ring vacuum pumps, or additive preparation. Process water is produced in the thickening and dewatering stages of the papermaking process by separating liquid phase from solid phase. In the stock preparation loops, this is done by disk filters, screw and double wire presses, and drum thickeners. Wire section, press section, and savealls provide the required process water volumes at the paper machine. If water quality achieved is still inadequate, advanced treatment technologies such as membrane or ozone treatment can be employed. The possibilities for designing water circuits greatly vary and depend on a number of parameters. One important parameter is the grade of paper being produced and the corresponding raw material being used.
4.19.3.2.2 Primary, secondary, and tertiary water circuits Based on the connection to the core process (sheet formation on the wire), it is generally possible to differentiate between three categories of water circuits: primary, secondary, and tertiary water circuits. Figure 3 illustrates the primary and secondary water circuits. The primary circuit consists of white water 1 originating from the wire section. This circuit is the largest as far as the volumetric flow rate is concerned. The circulating flow rate depends on the retention in the wire section and the consistency in the headbox. Its objective is to dilute the main stock flow after the machine chest in the approach flow system to a consistency of approximately 0.7–1.5%. The excess flow rate is part of the secondary circuit. Besides the excess flow rate of white water 1, the secondary circuit originates from the forming section and from the press section (see Figure 3). Most of this water is preferably fed to a saveall, and the recovered fibers are sent to the blend or machine chest and stock preparation, respectively. The clarified
670
Water in the Pulp and Paper Industry Mixing chest
Machine chest
Cleaner
Screen
Wire section
Press section
From stock preparation (DIP, chemical pulp, mechanical pulp, fillers, etc.
Primary circuit
White water 1
Showers, etc.
To stock preparation
White water 2
Secondary circuit Saveall
Waste water
Figure 3 Schematic illustration of principal water and stock flows in a paper machine (Hutter, 2008).
water is sent to a buffer tank and from there it is supplied to the process. The possible fields of application of clarified water are manifold: pulping, consistency control, foam destruction, and showers (mainly in the wire section). Further treatment (e.g., membrane filtration) might be necessary in the case of sensitive applications such as sealing waters or high-pressure showers. As the papermaking process is typically supplied by freshwater, there is always an excess of process water. This excess water is part of discharged wastewater. A tertiary circuit is required when, at least, a part of the treated wastewater is recirculated. In zero effluent systems, all treated wastewater is recirculated. In order to eliminate detrimental substances from the papermaking process, the recirculated wastewater should undergo full biological treatment. Possible fields of application of the recirculated wastewater are manifold and depend on the water quality attained. Besides being used as pulping or cleaning water, it may also be used as sealing water or as spraying water in showers after adequate pretreatment. Attention must be drawn to the danger of scale formation as biologically treated water often has a high calcium concentration (Demel et al., 2004a, 2004b).
• • • • • • •
a reduction in additive efficiency, a reduction in optical and strength properties, negative effects on drainage and paper drying, negative impacts on sizing, odor formation, deposits, and/or foam generation.
The main sources of detrimental substances and contraries in paper-mill process water are fibrous raw materials, additives, and freshwater (Negro and Tijero, 1998). Table 1 provides an overview of the composition and origin of detrimental substances. The content of detrimental substances is typically measured using sum parameters, such as anionic trash, cationic demand, or chemical oxygen demand (COD). Inorganic dissolved substances are measured as increased conductivity (Stetter, 2006). The COD denotes the volume of oxidizable substances in a water sample. It is considered to be balanceable and is thus a suitable optimization parameter.
4.19.3.2.4 General principles of circuitry 4.19.3.2.3 Detrimental substances Due to the increasing use of recovered paper and the reduced freshwater consumption, constituents known as detrimental substances have accumulated in water circuits, leading to growing problems in the papermaking process. Detrimental substances are substances that have a negative impact on the papermaking process and on product properties. Auhorn defined them as follows: "Detrimental substances are dissolved or colloidally soluble anionic oligomers or polymers and nonionic hydrocolloids" (Auhorn, 1984). They can result in
Figure 4 schematically illustrates a simplified water and stock system in a paper mill. Both the stock preparation loop and the paper machine loop can use freshwater. Wastewater is discharged mainly from the paper machine loop. Moreover, water is exchanged between the paper machine and stock preparation loops depending on the transfer consistency of the pulp coming from stock preparation. Based on a specific effluent volume of 10 m3 per metric ton of paper and a COD input of 10 kg per metric ton of raw material, this results in a COD concentration of 1.7 g l1 in the stock preparation loop and 1.2 g l1 in the paper machine loop.
Water in the Pulp and Paper Industry
There are two general principles used in designing water circuits that are described on the basis of this simplified model mill:
• •
loop separation and countercurrent arrangement.
Water circuits can be subdivided into separate loops by installing thickening units such as screw presses or double wire presses. In many cases, these thickening units are also necessary for downstream units such as dispergers. At the same time, soluble detrimental substances will be retained in the stock preparation water system. An increase of up to 30% in the transfer consistency between stock preparation and the Table 1 contraries
Composition and origin of detrimental substances and
Chemical compounds
Origin
Sodium silicate
Peroxide bleaching, deinking, recovered paper Filler dispersing agent Filler dispersing agent Coated broke, recovered paper Freshwater Chemical and mechanical pulp
Polyphosphate Polyacrylate Starch Humic acids Lignin derivates, lignosulfonates, hemicelluloses Fatty acids Volatile fatty acids
Chloride Calcium Sulfides Exopolymer saccharides
Mechanical pulp, deinking Anaerobic processes (high hydraulic retention times, spoiled recovered paper) Chemical additives Recovered paper, fillers Anaerobic processes, sulfate High C/N ratio
671
paper machine results in a significant reduction in the exchanged water volume. As the wastewater is still being discharged from the paper machine loop, there is no sink for the detrimental substances in the first loop (stock preparation). Detrimental substances build up in the stock preparation loop, resulting in a COD concentration of 4.9 g l1. It is not possible to relieve the paper machine loop (Figure 5). A countercurrent arrangement (Figure 6) completes the above-described principle of loop separation. The highly concentrated filtrate from the thickening unit is discharged to the wastewater treatment plant. The water deficit in the stock preparation loop is compensated by adding water from the paper machine loop. Hence, the most contaminated water is being discharged, while the better-quality water is being used in the more sensitive paper machine loop. The water flows in a direction opposite to the stock flow. This leads to significant relief of the paper machine loop, resulting in a COD concentration of 0.5 g l1 or 58% of the initial situation described above. Strict separation of the stock preparation water loops from the paper machine loop combined with a well-designed countercurrent arrangement is essential to meet high runnability and quality requirements because this strategy keeps detrimental substances out of the paper machine. Unlike the countercurrent dewatering arrangement in paper mills, a countercurrent washing arrangement is typically installed in chemical pulp mills. Substances that are dissolved during digestion, delignification, and bleaching are carried along into the next process steps together with the fibers. To accumulate and recirculate these substances to the digester, the washing liquor passes through a countercurrent washing arrangement within the different process steps (Figure 7). This ensures that most of the organic load and the digesting chemicals are recirculated to the digester, which guarantees the efficiency of the bleaching chemicals. Furthermore, it helps to
5 COD SP loop (g l−1)
Raw materials 10 kg COD t−1 Freshwater Stock preparation
4 3 2 1
20 l kg−1 0 5
Paper machine 10 l kg−1 Wastewater treatment plant
Paper Figure 4 Simplified schematic illustration of the water and stock systems of a paper mill.
COD PM loop (g l−1)
5%
4 3 2 1 0
672
Water in the Pulp and Paper Industry 5 COD SP loop (g l−1)
Raw materials 10 kg COD t−1 Freshwater Stock preparation
4 3 2 1
20 l kg−1 0 5
Paper machine 10 l kg−1 Wastewater treatment plant
COD PM loop (g l−1)
30%
Paper
4 3 2 1 0
Figure 5 Loop separation.
5 COD SP loop (g l−1)
Raw materials 10 kg COD t−1 Freshwater Stock preparation
4 3 2 1
10 l kg−1
2 l kg−1 10 l kg−1 Wastewater treatment plant
Paper
COD PM loop (g l−1)
Paper machine
0 5
8 l kg−1
30%
4 3 2 1 0
Figure 6 Countercurrent arrangement.
relieve the pulp dewatering machine from detrimental substances as efficient as possible (Borschke, 2006).
4.19.3.2.5 Closed water circuits Complete closure of water circuits implies eliminating any sort of effluent discharge. For some mills, it is the last resort to be able to continue production at that particular location. Motivating factors include costs of discharging effluents, absence of receiving waters, or the necessary discharge rights if the mill is moved to a new location. The specific effluent volume in the case of closure is 0 m3 per metric ton of paper. Freshwater is
used only to compensate for the loss of water by evaporation and in the finished product, and for the water removed together with the rejects. This volume normally amounts to approximately 1.5 m3 per metric ton of paper. The process is thus subject to massive limitations. Only few consumers can continue to be supplied with freshwater, leading to extremely high concentrations of detrimental substances which will be bled out of the system only by transferring them into the paper. Hence, when preparing water circuits for mill closure, first, all options for optimization of the water circuits must be exhausted. A subsequent installation of internal circuit water
Water in the Pulp and Paper Industry Deknotting and screening
Washing
Washing
Oxygen Washing delignification
Pulp from digester
673
Liquor Subsequent process steps
O2
Filtrate
Digester Evaporation plant
FT
FT
FT
FT
Filtrate tank
Filtrate tank
Filtrate tank
Filtrate tank
COD in the water circuit
Figure 7 Countercurrent washing in chemical pulp manufacturing (Borschke, 2006).
2. Circuit water treatment
1. Optimization of water circuits
Closure of water circuits
most delicate part of the papermaking process, which is why the white water should contain as few disturbing substances as possible. By comparing the COD levels (filtered samples) in the wastewater prior to biological treatment and in white water 1, the K1 value makes it possible to determine the utilization of freshwater (Equation (1)). K1 value significantly less than 1 indicates freshwater which is discharged to the effluent treatment plant directly without relieving the paper machine loop:
K1 ¼ Specific effluent volume
CODEffluent CODWhite water
ð1Þ 1
Figure 8 A stage-by-stage approach for water circuit closure.
treatment units makes it possible to remove dissolved substances (kidney technology). Only then subsequent closure can be met successfully (Figure 8). Possible kidney technologies include integrated biological treatment, membrane filtration, or ozone treatment (see Section 4.19.4.2).
4.19.3.2.6 Assessment of freshwater use and circuitry In order to be able to optimize a water circuit, it must first be clarified whether or not the following conditions are fulfilled:
• • •
freshwater should be used effectively and not passed directly to the wastewater treatment plant, the contaminant load at the paper machine should be as low as possible, and contaminants should be discharged wherever possible using the smallest effluent volumes.
The K-values established by Kappen (Kappen and Wilderer, 2002) are capable of quantifying the most important goals in optimizing circuit design. They work through comparisons of the COD levels at different locations in the water circuit. K1 value. Sheet formation is brought about by dewatering fiber suspensions in the paper machine followed by the subsequent formation of hydrogen bonds between fibers. It is the process stage that is decisive for mechanical and optical characteristics of the paper. Sheet formation constitutes the
K2 value. The K2 value expresses the COD concentration ratio in water loops of the stock preparation and paper machine (Equation (2)). To achieve maximum relief of the paper machines, the COD level in white water 1 ought to be substantially lower than that of the stock preparation system. K2 41 means that detrimental substances that give rise to COD are retained in the stock preparation system, that is, the sheet formation section and white water 1 are relieved:
K2 ¼
CODStock preparation CODWhite water 1
ð2Þ
K1/K2 ratio. K1/K2 indicates whether the wastewater discharged from the papermaking system is the optimum solution in terms of paper machine relief. To obtain a maximum COD relief through a minimum effluent flow, the water highest in COD loading must be discharged to the wastewater treatment plant. This maximum loading is found in the stock preparation system where detrimental substances accumulate. The quotient of COD levels in the wastewater and stock preparation system is referred to as the K1/K2 ratio:
K1 CODEffluent ¼ K2 CODStock preparation
ð3Þ
K1/K2 close to 1 indicates that the COD loadings of the effluents and stock preparation are nearly equal, that is, the
674
Water in the Pulp and Paper Industry
effluents are discharging detrimental substances mainly from the section with the highest COD loading. This, in turn, ensures maximum paper machine relief by the given effluent volume and corresponds to the countercurrent arrangement. The K1/K2 ratio helps to evaluate the circuitries of a paper mill. A countercurrent arrangement is only realized for K1/K2 values close to 1. The above definitions apply exclusively to simple systems with one stock preparation system and one paper machine. In more complex systems, characterized by more than one line or loop in stock preparation or more than one paper machine, the COD loading of the wastewater, stock preparation, and paper machine loop are calculated as the weighted average of the individual COD loadings determined. Graphic representation of K1 and K2. The relationship between these K-values may be visualized in a K1–K2 performance characteristic (Figure 9). The K1–K2 performance characteristic depicts the current situation in a paper mill (operating point). The two lines to the left and right of the diagonal define the target range of best performance achievable under practical conditions. An operating point within the target range indicates an optimized water loop. The circuitry status and the optimization potential that exists can be visualized as the K1–K2 section that has experienced a local shift out of the target range. The success of optimization measures can be documented without much difficulty using these key parameters. The K1 value, the K2 value, and the K1/K2 ratio are the principal characteristics that make assessment of the efficiency of freshwater use and circuitry possible. The characteristics K1 and K2 make it possible to quantify the primary objectives of circuit optimization, that is, effective freshwater use, maximum paper machine relief, and effective elimination of anionic trash (Kappen and Wilderer, 2002).
4 Target area with loop separation 3
K2
Operating point 2
1 Target area without loop separation 0 0
1
2 K1
Figure 9 K1–K2 performance characteristic.
3
4
4.19.3.3 Wastewater 4.19.3.3.1 Characterization of wastewater from the pulp and paper industry In general, wastewater in the pulp and paper industry is produced in the form of excess process water, which is displaced by the freshwater input. The wastewater is loaded primarily with organics that enter the production process together with raw materials and additives. Effluents from the pulp and paper industry are still not completely understood in terms of their chemical composition (Hynninen, 2000). In the majority of cases, however, wastewater of paper mills is nontoxic and easily degradable biologically. Higher concentrations of dissolved organic and inorganic compounds are observed in productions, enabling a particularly intensive utilization of water (Mo¨bius, 2002). As far as organic loads are concerned, COD, biochemical oxygen demand (BOD5), and adsorbable organic halogens (AOX) are the key parameters that characterize papermaking effluents. Today, however, the total organic carbon (TOC) parameter is becoming more important – a development which is reflected in a growing number of measuring methods such as the cuvette test or online measuring systems. Effluent concentrations vary widely depending on
• • • •
raw materials, paper grades, specific freshwater consumption, and available installations.
The assessment of the biodegradability of effluents is based on parameters such as BOD5, COD, and their ratio in wastewater. For a completely degradable compound such as glucose, which resembles the dissolved material in paper-mill effluents, the BOD5/COD quotient is typically approximately 0.6, suggesting a very good biodegradability, whereas a lower quotient is indicative of poorer degradability and a higher residual COD. The BOD5/COD quotient from different paper mills varies roughly between 0.35 and 0.5. Moreover, some important inorganic effluent parameters, such as salt loads, have to be taken into consideration. Calcium and sulfate concentrations play a special role in the operation of anaerobic treatment plants. When treating effluents with a high calcium content, poorly soluble calcium carbonate may be precipitated. In plants employing carrier material, such precipitation products may cause deposit formation. In mixed reactors, precipitation products tend to accumulate in the sludge, impeding thorough mixing of effluents and sludge and finally reducing the share of active biomass. When treated effluents are recirculated back into production, additional precipitation problems may arise in the consumers due to the pH shift of the decreased buffer capacity. The growing use of calcium carbonate as a filler and coating pigment and the ever-tighter-closed water circuits increase the calcium concentrations in circuit water and the wastewater. This applies in particular to mills that convert recycled paper. In the case of high sulfate concentrations, anoxic conditions may trigger sulfate reduction and lead to sulfide formation. This may disturb the degradation processes (methanogenesis) in anaerobic treatment plants, whereas in
Water in the Pulp and Paper Industry
aerobic biological treatment such high sulfate concentrations may foster the growth of undesirable filamentous microorganisms. As an additional drawback, the hydrogen sulfide that forms may give rise to bad odors and corrosion phenomena. Sulfate concentrations up to 600 mg l1 are to be expected in the wastewater of paper mills producing mechanical paper due to the aluminum sulfate used for resin sizing. The sulfate concentrations are substantially lower for woodfree papers. Even higher concentrations can occur in the production of recycled fiber-based papers. Sulfate originates from recovered papers and becomes increasingly concentrated as a result of tightly closed water circuits, typical for paper mills converting recovered paper. Depending on the treatment process used, other parameters such as pH, conductivity, and temperature are important for operational safety. Normally, in paper industry effluents, phosphorus and nitrogen compounds serving as nutrients for microorganisms are either absent or only available in insufficient quantities (Hamm, 2006). Therefore, it must be ensured that dosages of nutrients in treatment plants provide a sufficient supply for the microbiota. However, simultaneously, the permissible limit values in final effluents have to be met.
4.19.3.3.2 Wastewater discharging In the German pulp and paper industry, most effluents undergo full biological treatment. Ninety-five percent of the production volume is produced in mills with an integrated biological wastewater treatment plant or mills that discharge their wastewater to municipal wastewater treatment plants; 4% of the annual production volume is produced in mills with a closed water circuit; and only 1% of the production volume comes from mills that discharge their effluents without biological treatment (Jung et al., 2009).
4.19.4 Water Treatment 4.19.4.1 Freshwater Treatment As mentioned in Section 4.19.3.1, the source of freshwater in the pulp and paper industry is usually surface water. Typically, freshwater does not meet the required quality parameters of the manufacturing process and therefore has to be treated. Well water seldom needs treatment. Objectives of freshwater treatment in the pulp and paper industry include
• • • •
removal of solids, removal of color and organic substances, decrease in hardness and removal of other dissolved salts, and, in some cases, the disinfection of the water.
Water quality can be improved by a range of treatment measures. Factors influencing the choice of the treatment method and equipment include required water quality, water volume to be treated, space available for freshwater treatment plant, and, to some extent, how well the plant operation and supervision can be integrated with the other operations in the mill (Hynninen, 2000).
675
Predominantly mechanical or chemical–mechanical treatment technologies are used for freshwater treatment in the pulp and paper industry. According to a survey conducted by PTS and VDP, more than 90% of the surface water used as freshwater for the papermaking process is treated by filtering. An additional 75% of this water is conditioned by chemical coagulation, flocculation, and subsequent sedimentation. The volume of freshwater treated with biocides has increased significantly, whereas the use of chlorine has decreased in the past few years (Jung et al., 2009). Freshwater is softened and desalinated for boiler house use and for the production of some specialty papers (e.g., photographic base paper or cigarette paper; Stetter, 2006).
4.19.4.2 Circuit Water Treatment 4.19.4.2.1 Objectives of circuit water treatment At the beginning, the objective of circuit water treatment was primarily to recover fiber furnish from papermaking effluents. Under the economic and ecological necessity of reducing effluent volumes and loads, the circulation water treatment process took on ever-greater importance and function: circuit water treatment must provide clarified water with a predefined quality and has to remove interfering substances from the system. In doing so, circuit water treatment became responsible for removing not only insoluble and colloidal components but also dissolved substances. Therewith, circuit water treatment helps to stabilize production processes and ensures product quality. The objectives of circulation water treatment include
• • •
recovery of raw materials, production of mill water with a low solid concentration available, and reduction of contaminants in the circulation water.
Depending on the required water quality, the requirements on circuit water treatment vary from reducing solid losses in the case of relatively coarse treatment to preparing shower and sealing water in the high-pressure range in the case of precision treatment.
4.19.4.2.2 Mechanical circuit water treatment Sedimentation, flotation, and filtration methods in particular are employed in mechanical circuit water treatment. These techniques can also be used in combination with one another. Methods for screening and classification are mainly used in stock preparation. The market share of the individual types of savealls moves in the direction of a two-part system, as sedimentation is steadily declining, whereas filtration and flotation are both expanding due to innovations and technical improvements (Zippel, 2001). Hydraulic surface load, solid surface load, and purification performance are the key parameters for the layout of the treatment units. Other parameters that have to be considered are the concentration of suspended solids in the feed, presence of colloidal and dissolved substances, additive demand, available space, and overall energy consumption. Two of the most important factors are the investment and operational costs. Cost effectiveness is obtained by reducing raw material losses,
676
Water in the Pulp and Paper Industry
increasing process performance, and the benefits arising from stable and efficient paper production (Weise et al., 2000). The fiber recovery unit, or synonymously the saveall, is supplied with the excess from the primary circuit (wire pit overflow) and water removed from the wire and press sections by the vacuum system. In rare cases, water from the floor channels or from the wet broke, for example, is supplied to circuit water treatment. The process connected to the saveall has to be designed so that a constant feed flow is maintained. Sedimentation. Sedimentation is generally the simplest form of a saveall, and conventional sedimentation savealls have long been known to be reliable and safe to operate. Nowadays, sedimentation plays a minor role and is commonly used only in old plants for circuit water treatment. A general disadvantage of sedimentation plants is a low density of the sediment. Hydraulic retention times are very long in some cases and can also provoke anaerobic degradation accompanied by the correspondingly disadvantageous consequences (odor, microbial contamination, etc.) that affect the entire water circuit. Long hydraulic retention times also become a problem if rapidly changing production programs are to be run on the paper machine. Flotation. Flotation denotes the use of air bubbles to float undissolved substances to the surface of a suspension. Hydrophobic or hydrophobized particles adhere to the air bubbles, rise through the suspension, and are carried along to the surface and scooped off there by a suitable skimming device. Different flotation processes vary according to how bubbles are introduced into the suspension. Dissolved air flotation (DAF) has established itself in circuit water treatment. In this process, water is supersaturated with compressed air and then supplied to the flotation chamber (Figure 10). The resulting reduction in pressure causes very fine air bubbles to form that become attached to the suspended particles. Pressure saturation current can be the entire inflow, a partial flow, or recirculated clarified water (recycling process). A general problem associated with flotation units is a sharp fluctuation in inflow loadings. Fluctuations both in the volumetric flow rate and in the solid surface loading produce poor results.
Filtration. Filtration technologies are well suited for separating solid particles from suspension with assistance of a porous filter medium. Compared with other processes, good separation properties and high-quality clarified water that can be achieved are advantageous. Disadvantages are high investment and operating costs due to the considerable amount of maintenance work. In the pulp and paper industry, disk filters (Figure 11) are by far the most common type for mechanical circuit water treatment. A disk filter comprises several disks that consist of individual segments covered with a filter medium that rotate in a vat. The filtrate consistency declines during the filtration process and the filtrates are typically collected separately as cloudy filtrate and clear filtrate. In some cases, super-clear filtrate may also be produced. As an alternative to disk filters, drum filters can be used, which usually reduces the cost factor. However, for most applications, the hydraulic capacity of drum filters is too low and only one filtrate quality is produced. This normally makes them unsuitable for saveall application. Drum filters are often used as simple but reliable thickeners, for example, in the broke-handling system (Zippel, 2001; Weise et al., 2000).
4.19.4.2.3 Advanced circuit water treatment As a result of an increasing closure of water circuits, the use of freshwater for a steadily growing number of consumers in a system must be restricted. In order to replace the freshwater at these locations, the clarified water must be of high quality. In many cases, complete elimination of solids is required, especially for showers in the high-pressure range and sealing water. Large volumes of clarified water needed in a closed or virtually closed system that at the same time places high requirements on clarified water quality for only very few consumers have promoted the use of multistage treatment processes. Methods for fine cleaning of pretreated circuit water are based on filtration. The objectives include continuously improving the water quality and serving as a police filter, if any upstream treatment method fails. A wide variety of methods are employed, including drum filters with very fine filter media
Rotary contact Spiral scoop
Clarified water pipes
Inlet distribution Rotating carriage ADT distribution Floated sludge outlet Recycle suction Raw water inlet Inspection window Settled material outlet
Clarified water outlet
Dissolved air in water inlet
Figure 10 Dissolved air flotation unit for circuit water treatment. Adapted from Krofta Waters International (2010).
Water in the Pulp and Paper Industry
Cake removal
677
Cake removal
Filtrate
Suspension
outflow
inflow
Overflow
Suspension vat
Filter cake
Filter cake
Figure 11 Disk filter saveall system. Adapted from Wilichowski M (2009) Folien zur Vorlesung Mechanische Verfahrenstechnik I þ II. http:// www.mb.hs-wismar.de (accessed March 2010).
Process water
Freshwater
Paper and board production at 55−60 °C 100% recycled paper
Quality E
Quality D
Quality A
Quality C
Quality B
Biogas Acid
Buffer tank / sedimentation
pH
UASB/IC thermophil
Sedimentation
Aeration
Police filter
Solids
Solids Ozonation (opt.)
Heating
NF/ RO
Biogas Retentate Effluent
Figure 12 Kidney technology concept (Pauly, 2001).
(microfiltration), cartridge and backwash filters, sand filters, and membrane technology. The treatment units for elimination of dissolved substances accomplish important tasks in narrowing water circuits. They relieve the water circuits of detrimental substances and therewith avoid production restrictions due to limited freshwater capacities. The so-called kidney technologies, such as integrated biological treatment, softening, membrane technology, and ozonization, are promising approaches for obtaining effluent-free paper production by way of circuit closure. Combinations of treatment technologies make it possible to provide optimized solutions for different objectives. The decisions as to which concept and which treatment technology is to be used depend on specific boundary conditions. Figure 12 is a schematic view of a potential kidney concept for water circuit closure in a paper mill converting 100% recovered papers. Full circuit closure is not necessarily the solution of choice. Nevertheless, in many cases, advanced integrated treatment steps yield both economical and
ecological advantages (Pauly, 2002). Starting in 2008, the European Aquafit4Use research project focused on high waterreuse rates. The project also highlights maximum reduction in energy and chemicals, leading to more efficient use of limited resources by developing tailor-made treatment technologies and concepts (Pauly, 2008). Biokidney. Progressive system closure in a paper mill leads to increased concentrations of dissolved and colloidal compounds that in turn can give rise to increased microbial activity, slime formation, foaming, pitch disposition, corrosion, altered wet-end chemistry, and odor problems. Biological treatment of the circuit water can reduce or eliminate the buildup of troublesome compounds. Integrated biological treatment process is called the biokidney. Biological processes are state of the art in paper-mill wastewater treatment plants and are suitable for the elimination of biologically degradable substances, the reduction of sulfates, and the preliminary purification for nanofiltration and reverse osmosis.
678
Water in the Pulp and Paper Industry
Flow
Conc.
Exhaust gas p
Inflow UF
Aeration
T pH
Sedimentation
T
Fi
Retentate overflow
Storage vessel
LC
Storage tank
Permeate vessel
Red.
Pellets
Exhaust air
UASB outlet
Separator
Permeate
Compressed air
Module
T Fi
p
Exhaust air
Insulated heated room
Outflow
Figure 13 Kidney concept – thermophilic anaerobic treatment and softening step – implemented at a Belgian paper mill in the frame of EU-project Paper Kidney; further downstream options: membrane treatment and ozonation (Pauly, 2002).
Biological treatment of process water has been carried out within a very wide range of operating conditions. Both anaerobic and aerobic process designs have been shown to be successful. Thermophilic treatment of process waters (Figure 13; Pauly, 2002) has a distinct advantage of eliminating the need for process water cooling and reheating for water recycling. A biokidney improves runnability due to a better overall quality of the process water caused by a decrease in soluble organic matter. Using biological treatment for COD reduction has allowed some packaging paper mills to operate with zero effluent systems (see, e.g., Stra¨tz, 2008; Herberz and Bahn, 2006). Practical trials conducted with aerobic and anaerobic laboratory-scale biokidneys also demonstrated that biokidneys have the potential to remove odorous compounds. (Jung et al., 2007). Membrane technology. This technology has experienced difficulties in trying to make headway into the papermaking sector. There are a variety of reasons for these difficulties, including scaling and fouling caused by high concentrations of salts and other detrimental substances, not fully developed concepts, and high investment costs due to high volumetric flow rates. A constant increase in the interest expressed by the European paper industry in membrane technology confirms that this method is due to evolve into a key technology for ¨ ller, continued water savings in the future (Simstich and O 2007). Membrane technology typically improves the quality of process waters substantially, since it removes suspended solids, microorganisms, and colloidal COD. Even salts can be separated out using reverse osmosis. Ultrafiltrated water is free of suspended solids and colloids. Bacteria, latex, and other micro-stickies are removed, for example. As anionic trash is cut to approximately half of its original level, the quality of the filtrate makes trouble-free recirculation back into the process possible (Sutela et al., 2006).
Fields of application for membrane technology in the pulp and paper industry are numerous and may offer many advantages, depending on treated partial flow such as
• • • • •
reduction in volumes of freshwater and effluent, increased product quality due to reduced system loading, recovery of raw materials from the effluents (e.g., coating color pigments), enhanced possibilities for the recirculation of biologically treated wastewater, and compliance with statutory limits on effluent concentrations.
In paper machine water circuits, ultrafiltration systems are mainly employed to provide high-quality water for use in high-demand consumers such as high-pressure showers or chemical dilutions. This leads to reduced freshwater consumption and a better machine runnability and paper quality. Ozone treatment. This treatment can be used as an internal treatment technology for decoloration of process waters. The brown color of paper-mill waters is mainly caused by derivatives similar to lignin and humic acid that are characterized by C¼C double bonds. These are the preferred sites of attack by ¨ ller, 2007). ozone, which then destroys them (Bierbaum and O As ozone is one of the strongest oxidants known, it is also possible to eliminate COD by oxidizing water components. ¨ ller and Offermanns (2002) have shown that recirculation of O ozonized waters into papermaking process is possible on the mill scale without any adverse effects on the process or product quality. Numerous other ozone applications – such as COD and AOX reduction, decoloration, microbial and odor control, and improvement of biosludge characteristics – in mill water systems (freshwaters, circuit waters, or effluents) are conceivable as well.
Water in the Pulp and Paper Industry 4.19.4.3 Wastewater Treatment 4.19.4.3.1 Preliminary mechanical treatment: Mechanical processes for removal of solids Effluents from pulp and paper mills contain solids and dissolved matter. The principal methods used to remove solids from pulp and paper mills effluents include screening, settling/clarification, and flotation. The choice of method depends on the characteristics of the solid matter to be removed and the requirements placed on the purity of the treated water. The separation of solids from the effluents is accomplished with the help of screens, grid chambers, and settling tanks. Screens are units which operate according to the sieving/filtration process. The function of the screens is to remove coarse, bulky, and fibrous components from the effluents. If necessary, fractionated particle separation can be achieved by graduating the gap width (bar screen, fine screen, inlet screen, and ultrafine screen). For reasons of operating reliability of wastewater treatment plants, it is also necessary to separate the grit transported with the effluents and other mineral materials from the degradable organic material. Grit separation from effluents can prevent operational troubles such as grit sedimentation, increased wear, and clogging. The grit separating systems currently in use are subdivided into longitudinal grit traps, circular grit traps, and vortex grit traps, depending on their design and process layout. Sedimentation technology is the simplest and most economical method of separating solid substances from the liquid phase. High efficiency is achieved in subsequent effluent treatment processes when the solid substances suspended in the effluents settle in a sedimentation tank as completely as possible, and settled sludge is removed from the sedimentation tank. Sedimentation tanks must be appropriately designed and operated. Alternative sedimentation equipment, with sets of lamella-shaped passages, is employed in the paper industry, especially for effluents with high fiber concentrations. Mechanical effluent treatment alone, however, is not sufficient to keep lakes and rivers clean, since it is incapable of removing colloidal, suspended, and dissolved substances.
4.19.4.3.2 Biological treatment Biological wastewater treatment is designed to degrade pollutants dissolved in effluents by the action of microorganisms. The microorganisms utilize these substances to live and reproduce. Pollutants are used as nutrients. A prerequisite for such degradation activity, however, is that the pollutants are soluble in water and nontoxic. Degradation process can take place either in the presence of oxygen (aerobic treatment) or in the absence of oxygen (anaerobic treatment). Both these naturally occurring principles of effluent treatment give rise to fundamental differences in the technical and economic processes involved (Table 2). The paper industry uses a variety of effluent treatment systems. The preferred process combination for each individual case depends on the grade-specific quality of the effluent that is to be treated. Experience shows that multistage processes based on an aerobic–aerobic or anaerobic–aerobic processing principle enable significantly more reliable
Table 2 treatment
679
Main characteristics of anaerobic and aerobic wastewater
Anaerobic treatment l
COD41000 mg l Low amount of excess sludge Energy generation by use of biogas Low energy demand Low required space Sensitive against high sulfate and calcium concentrations No fully biological degradation
Aerobic treatment High High High Fully
amount of excess sludge energy demand required space biological degradation
operation of the plant. The same effect can be achieved through a cascade system, which allows a graduation of the loading conditions. Among the German pulp and paper mills with onsite wastewater treatment plants, 60% have only aerobic treatment (operated as one- or two-stage processes) for their effluents, whereas 40% have an additional anaerobic stage (Jung et al., 2009). Anaerobic treatment. Anaerobic processes are employed for treatment of more highly polluted effluents such as effluents from recovered paper converting mills (Hamm, 2006). Anaerobic microorganisms conduct their metabolism only in the absence of oxygen. Anaerobic processes are characterized by a small amount of excess sludge produced and low energy requirements. As biogas is produced during the degradation process, anaerobic processes produce an excess of energy. Biogas is a mixture of its principal components, methane and carbon dioxide, with traces of hydrogen sulfide, nitrogen, and oxygen. Biogas is energetically utilized mainly in internal combustion engines or boilers. In its function as a regenerative energy carrier, biogas replaces fossil fuels in the generation of process steam, heat, and electricity. The composition and quality of biogas depend on both effluent properties and process conditions such as temperature, retention time, and volume load. Before discharge into surface waters, anaerobically treated effluents have to undergo aerobic posttreatment, because – according to the current state of the art – fully biological degradation of paper-mill effluents is not feasible (Mo¨bius, 2002). When introducing anaerobic technology into the pulp and paper industry, operational problems and their possible consequences, shown in Table 3, must be taken into account: Among different types of anaerobic reactors, ICs reactors (internal circulation) have achieved a share of more than onethird of the operating reactors and are currently the most frequently used reactors in the German pulp and paper industry. The rest of the market is shared by Biobeds and UASB reactors (UASB, Upflow Anaerobic Sludge Blanket) as well as reactors operating according to the contact sludge principle. Aerobic treatment. Aerobic microorganisms require oxygen to support their metabolic activity. In effluent treatment, oxygen is supplied to the effluent in the form of air by special aeration equipment. Bacteria use dissolved oxygen to convert organic components into carbon dioxide and biomass. In addition, aerobic microorganisms convert ammonified organic nitrogen compounds and oxidize ammonium and
680
Water in the Pulp and Paper Industry
nitrite to form nitrate (nitrification). The key factors for the success of an aerobic process are an adequate amount of nutrients in relation to the amount of biomass, a certain temperature and pH regime, and the absence of toxic substances (Hynninen, 2000). Aerobic processes are characterized by high volumes of excess sludge and higher energy demands compared to anaerobic processes. Furthermore, these reactors typically have large space requirements. Aerobic treatment allows fully biological degradation of paper-mill effluents. The BOD5 efficiency achievable with welloperated activated sludge processes is typically within the range of 90–98% (Hamm, 2006). The drawbacks of aerobic treatment technology include the relatively high operating costs due to the aeration of the effluent. On the other hand, aerobically operated plants exhibit higher plant stability and are less sensitive to fluctuations in effluent and plant parameters. Among different types of aerobic treatment technologies, activated sludge processes are currently the most frequently used treatment technologies in the German pulp and paper industry and have achieved a share of three-quarters of the operating reactors. Both moving-bed bioreactors (MBBRs) and biofilters represent another 10% of the reactors used (Jung et al., 2009). Secondary clarification. Secondary clarification is intended to separate the biomass (activated sludge) formed in biological reactors and is therefore a key element in all processes employed in the final stage of a treatment plant. The quality of the separation process is just as crucial for the final effluent quality as is for the biological treatment itself. As far as activated sludge process is concerned, secondary clarification determines the bioreactor performance. Separation and thickening of the recirculated sludge are crucial for
Table 3 Operational problems and possible consequences on anaerobic treatment in the pulp and paper industry
sludge volumes in biological treatment and for the potential sludge loading as well. Correct dimensioning of secondary clarification is therefore of great importance for overall plant performance.
4.19.4.3.3 Advanced and tertiary treatment Tertiary and/or advanced wastewater treatment is used to remove specific wastewater constituents that cannot be removed by secondary treatment. Different treatment processes are necessary to remove nitrogen, phosphorus, additional suspended solids, refractory organics, or dissolved solids. Sometimes it is referred to as tertiary treatment because advanced treatment usually follows high-rate secondary treatment. However, advanced treatment processes are sometimes combined with primary or secondary treatment (e.g., chemical addition to primary clarifiers or aeration basins to remove phosphorus) or used in place of secondary treatment (e.g., overland flow treatment of primary effluent). The reasons for advanced effluent treatment include
• • •
reduction in costs (discharge fee), compliance with limit values, and increase in production.
Advanced wastewater treatment in the pulp and paper industry is mainly focused on additional biological membrane reactors, membrane filtration techniques such as micro-, ultra-, or nanofiltration, and ozone treatment. Due to the relatively limited full-scale experience, relatively high costs, and greater complexity of water treatment, there have been only few fullscale applications of tertiary treatment of mill effluents up to now. The method that is ultimately chosen depends on the treatment aim and economic efficiency of the method in a given application. Table 4 shows the treatment aims that can be achieved by the different methods.
Operational problem
Possible consequences
4.19.5 Potentials and Limits of Water Saving
High concentrations of suspended solids in the feed flow High sulfate concentrations
Displacement of biomass Loss of pellets Displacement of methane
4.19.5.1 Limiting Effects of System Closure
bacteria
Inhibiting or toxic effects of
When reducing specific effluent volume within the framework of water circuit optimization, typical limits occur that usually require considerable investment to ensure that they will not be
sulfide High calcium concentrations Additives used in production (especially biocides and detergents)
Performance losses Precipitation of CaCO3 Displacement of biomass Inhibiting/toxic influences Poorer degradation
Insufficient supply of nitrogen and phosphorus Temperature variations Fluctuating organics loads (e.g., shock loads)
performance Decomposition/washout of pellets Unstable operation Performance losses Loss of pellets Unstable operation Performance losses Excessive production of organic acids Methanation disturbed
Table 4
Treatment aims of different advanced treatment methods
Treatment method
Aim of treatment
Biofiltration
Reduction in COD and BOD concentration
Ozone treatment Membrane treatment
Filtration processes Denitrification and phosphate precipitation
Removal of suspended solids Elimination of residual COD Decoloration Elimination of residual COD Elimination of suspended solids Demineralization Decoloration Removal of suspended solids Nitrogen and phosphate elimination
Water in the Pulp and Paper Industry
exceeded (Figure 14). A limit in this sense is the freshwater volume that is taken into the system as process freshwater and is used for cooling prior to its final use (2). The second limit is water volume that accumulates together with the rejects and is discharged together with the effluents (3). The third limit is the maximum COD value that the respective product can tolerate in the white water (4). In a selected circuit, this value also corresponds to a minimum effluent volume for the respective system. The above-mentioned limits differ in every individual system. The factors that influence these limits include the existing plant technology, raw materials used, and paper grades produced. A limit encountered in narrowing water circuits that is similar to the cooling water requirements discussed above are the rejects that accumulate when discharged with the paper mill effluent. The proportion of effluents contained in the rejects compared to the total effluent volume may amount to 40–50% of the total effluent volume, especially in paper mills with an integrated deinking plant. If the effluent volume of such a plant is to be reduced drastically, water volume added to the effluents together with the rejects constitutes a lower limit. If a further reduction in the specific effluent volume is intended, then the rejects must be dewatered and part of the filtrate returned to the water circuit. Low specific effluent volumes result in growing system loads in process waters in terms of dissolved and colloidal material (Figure 15) that cause severe quality deterioration (slime spots, odor, color shifts, etc.) and a drop in productivity (machine failures due to scaling and corrosion, slime formation, web breaks, etc.). This situation is aggravated by the use of heavily loaded waste paper.
681
If specific effluent volume is to be reduced successfully, the impact of such measures on the papermaking process must also be taken into consideration. Only if we succeed in reconciling the goal of preventing effluent production with the goal of reliable production and satisfactory product quality, can the narrowing and ultimate closure of water circuits come about successfully.
4.19.5.2 Heat Balance Narrowing and closure of water circuits lead to increased temperatures in the stock and water systems of paper mills, taking a constant energy input into account. Nowadays, temperatures of 40–50 1C are achievable in the paper machine loop without additional steam heating (Zippel, 2001). Loop separation and the countercurrent arrangement enable the paper mills to reduce the transfer of detrimental substances coming from highly loaded loops (e.g., stock preparation) into the subsequent process steps, thus relieving paper machine loop. Regarding heat balance of the stock and water system, this is disadvantageous as the highly loaded loops are typically also the hottest (e.g., thermomechanical pulp plant). Heat with quite a high-temperature level is transferred to the effluent. However, at the paper machine, a higher-temperature level would be desirable to improve mechanical dewatering and in turn decrease the energy consumption for thermal drying. Besides the above-mentioned effects, there are other positive and negative impacts of higher process temperatures:
Specific effluent volume l kg−1 Cooling Cooling water water
1
In receiving waters
Fresh water (process water)
2
Waste water
3 4
Cooling water (process water)
Rejects (waste water)
5 COD 5' 5
Rising COD
4
1
To the wastewater treatment plant
} Fresh water
Production
Evaporation
Waste water
Figure 14 Limits in reducing the specific effluent volume. (1) Current situation; (2) cooling water limitation; (3) reject limitation; (4) maximum white water loading; (5) closed water circuit.
682
Water in the Pulp and Paper Industry
COD concentration in WW1 (g l−1)
30.000 25.000 20.000 15.000 10.000 5.000 0 0
2
4
6
8
10
12
Specific effluent volume (m3 per metric ton of paper) Figure 15 Chemical oxygen demand (COD) concentration in white water 1 of European paper mills producing corrugated base paper as a function of the specific effluent volume.
• • • • •
•
The solubility and activity of most functional chemicals increase with increasing water temperature. The consumption of certain additives such as wet strength agents may increase due to increased temperatures in the water circuit. Slime formation can be restricted by increasing the process temperature above a certain limit. There might be greater formation of anaerobic metabolic products such as hydrogen sulfide. High water temperatures in the papermaking process reduce the energy consumption for pulping and increase the cleaning efficiency of showers. On the other hand, high water temperatures have negative impacts on the energy efficiency of liquid-ring vacuum pumps and the hall climate. Finally, high process temperatures lead to high effluent temperatures. Without any countermeasures, this can cause problems in aerobic effluent treatment plants (poor oxygen solubility) and with the statutory temperature limits.
Integration of waste heat streams is one possibility for paper mills to reduce their energy consumption, but presents them with the conflicting challenges of ensuring both maximum waste heat utilization and safe compliance with statutory limits on effluent temperature. Heat integration measures help optimizing heat balance of paper mills and are a cost-effective way to reduce the specific energy demand of paper mills, thus achieving a productivity increase. Apparently, conflicting objectives, such as increased process and decreased effluent temperatures, may be achieved by appropriately selected measures for heat-balance optimization. Based on available heat sources and sinks and considering other boundary conditions, there are several potential scenarios for heat integration and utilization of waste heat by means of water–water heat exchange or air–water heat exchange. Pinch analysis and process simulation are useful tools for an evaluation of the individual scenarios and an optimization of heat balances. Studies have shown that the
replacement of steam used for process- or freshwater heating yields particularly profitable energy savings (Jung, 2008).
4.19.5.3 Economic Benefits There are many reasons to reduce the specific effluent volume. One important reason is the reduction in water-related costs. In the German pulp and paper industry, the costs of discharging effluents into receiving waters are high and average h0.40 m3 for direct dischargers. Discharging and treating effluents for indirect dischargers, however, are considerably more expensive. The latter involves average costs amounting to h1.12 m3. Reducing effluent volume is very attractive, especially for indirect dischargers. Additional costs arise due to a user fee for freshwater outtake and the operational costs for freshwater treatment (Jung et al., 2009). Despite the above-mentioned problems encountered in narrowing the water circuits, potentials for a reduction of the effluent volume have been discovered in many paper mills studied by PTS in the past few years (Figure 16). Besides water-related costs, another possibility is to reduce energy-related costs by reducing energy consumption due to increased process temperature. As a rule of thumb, every 101 increase in process temperature equals approximately 1% increase in dryness after mechanical dewatering in the wire and press sections. This allows energy consumption in the drying section to be reduced by up to 4%.
4.19.6 Improving Water Efficiency in Paper Manufacturing Industries – 30 Years of Success Water is one of the key components in papermaking. Using more than 1 billion m3 of water per year, the paper industry in Europe had been challenged to reduce the impact on regionally available water resources as one of the most important industrial water consumers. Legislation, stringent discharge
Water in the Pulp and Paper Industry
683
Specific effluent volume (m3 per metric ton of product)
25 Production rate proportional weighted mean 20
15
10
5
8 mills
6 mills
4 mills
4 mills
From recovered paper
Wood containing
Wood free
Specialty paper
0
Figure 16 Optimization potentials of paper-specific effluent volumes.
standards, as well as process and product demands force industry to ensure higher water quality corresponding to increasing costs. For the water-consuming industry, water is no longer regarded as a consumable or utility but as a highly valuable asset. Attention to water scarcity and pollution results in new legislative directives, forcing industries to reduce water use and pollution, and motivating them to implement innovations and carefully observe the impact of measures. The Water Framework Directive (WFD) is one of the main drivers for sustainable water use in Europe, which forced the member states to pay more attention to sustainable and efficient water use. Competent decision making at the top management and well-trained and motivated staff delivered substantial progress in reducing the water consumption in the pulp and paper industry: high competence in closing water circuits, substantially supported by process modeling and automation, and kidney technologies as internal process water treatment, lead to a significant decrease of the average specific effluent volume in the past 30 years. The European collaborative research project, AquaFit4Use, started in 2008, focuses on optimization of existing water circuits and development of new treatment concepts to support the European sustainability policy, such as reducing the use of scarce freshwater, improving the water quality (micropollutants, salts, etc.), and sharing corresponding experiences with other sectors.
References Auhorn W (1984) Das Sto¨rstoff-Problem bei der Verringerung der spezifischen Abwassermenge. Wochenblatt fu¨r Papierfabrikation 2: 37--48. Bierbaum S and O¨ller H-J (2007) Anlagenkonzepte zur Ozonbehandlung von Papierfabriksabwa¨ssern. Allgemeine Papier Rundschau 3: 38--40. Borschke D (2006) Zellstoff- und Papierfabrikation – Prozesswassersysteme im Vergleich. Wochenblatt fu¨r Papierfabrikation 17: 971--981.
Demel I, Dietz W, Bobek B, and Hamm U (2004a) Criteria for the recirculation of biologically treated water to the production. ipw – Das Papier 1: 37--40. Demel I, Dietz W, Bobek B, and Hamm U (2004b) Criteria for the recirculation of biologically treated water to the production (II). ipw – Das Papier 2: 33--35. Federal Statistical Office (2008) Statistical Yearbook 2008. Wiesbaden, Germany. http:// www.destatis.de (accessed March 2010). Hamm U (2006) Environmental aspects. In: Holik H (ed.) Handbook of Paper and Board, pp. 208--218. Weinheim: Wiley-VCH. Herberz J and Bahn W (2006) 10 Jahre Betriebserfahrungen mit einer integrierten biologischen Reinigung im geschlossenen Wasserkreislauf. In: Jung H and Simstich B (eds.) Proceedings Wasserkreisla¨ufe in der Papiererzeugung Verfahrenstechnik und Mikrobiologie, pp. 8/1–8/10. Munich, Germany, 05–06 December. Munich: PTS. Hutter A (2008) Wasserkreisla¨ufe und Wasserqualita¨t in der Papiererzeugung. In: Jung H and Simstich B (eds.) Proceedings Wasserkreisla¨ufe in der Papiererzeugung, pp. 1/1–1/20. Munich, Germany, 02–03 December. Munich: PTS. Hynninen P (ed.) (2000) Papermaking Science and Technology Book 19 Environmental Control. Helsinki, Finland: Fapet Oy. Jung H (2008) Optimisation of the heat balance of papermills. PTS News 1: 30--33. Jung H, Hentschke C, Pongratz J, and Go¨tz B (2009) Wasser- und Abwassersituation in der deutschen Papier- und Zellstoffindustrie – Ergebnisse der Wasserumfrage 2007. Wochenblatt fu¨r Papierfabrikation 6–7: 280–283. Jung H, Pauly D, Beimfohr C, et al. (2007) Odour control – eliminating odour problems in the paper industry. PTS-News 2: 25--29. Kappen J, Hutter A, Bobek B, and Hamm U (2004) Qualitative and quantitative requirements on the water supply of internal consumers. INFOR-Project No. 52R, Munich/Darmstadt. Kappen J and Wilderer PA (2002) Key parameter methodology for increased water recovery in the pulp and paper industry. In: Lens P, Hulshoff Pol L, Wilderer P, and Asano T (eds.) Water Recycling and Resource Recovery in Industries: Analysis, Technologies and Implementation, pp. 229--251. London: IWA Publishing. Mo¨bius CH (2002) Waste Water of the Pulp and Paper Industry, 3 rd edn., Revision December 2008. Augsburg, Germany. http://www.cm-consult.de (accessed March 2010). Negro C and Tijero J (1998) Water in the pulp and paper industry. In: Blanco MA, Negro C, and Tijero J (eds.) Paper Recycling: An Introduction to Problems and their Solutions, pp. 17--46. Luxembourg: European Communities. O¨ller H-J and Offermanns U (2002) Successful start-up of the world’s 1st ozone-based effluent re-circulation system in a paper mill. In: Graham NJD (ed.) Proceedings of the International Conference Advances in Ozone Science and Engineering: Environmental Processes and Technological Applications, pp. 365–372. Hong Kong, People’s Republic of China, 15–16 April. Hong Kong: The Hong Kong Polytechnic University and The International Ozone Association. Pauly D (2001) Kidney-technology opens up new opportunities of integrated white water treatment in recycling mills. In: Gopalaratnam N and Panda A (eds.)
4.20 Water in the Textile Industry J Volmajer Valh, A Majcen Le Marechal, S Vajnhandl, T Jericˇ, and E Sˇimon, University of Maribor, Maribor, Slovenia & 2011 Elsevier B.V. All rights reserved.
4.20.1 4.20.1.1 4.20.1.2 4.20.1.2.1 4.20.1.2.2 4.20.2 4.20.2.1 4.20.2.2 4.20.2.2.1 4.20.2.2.2 4.20.3 4.20.3.1 4.20.3.1.1 4.20.3.1.2 4.20.3.1.3 4.20.3.2 4.20.3.2.1 4.20.3.2.2 4.20.3.2.3 4.20.4 References
Textile Industry Textile and Clothing Industry in Europe Processes in Textile Industry Fibers Finishing processes Characteristic of Textile Water and Wastewater Supply Water Textile Wastewater Textile wastewater from different process steps General characteristics of textile wastewater Treatment and Reuse of Textile Wastewater Wastewater Treatment Technologies Physical methods Chemical processes Biological treatment processes Reuse Pollution-prevention techniques Chemicals and water reuse and recycle: Start-of-pipe approach Process-water reuse and recycle: End-of-pipe approach Conclusions
4.20.1 Textile Industry The textile industry is one of the longest and most complicated industrial chains in the manufacturing industry. It is a fragmented and heterogeneous sector dominated by smalland medium-sized enterprises (SMEs), with a demand mainly driven by three main end uses: clothing, home furnishing, and industrial use. The textile industry is composed of a wide number of subsectors, covering the entire production cycle from the production of raw materials (man-made fibers) to semiprocessed (yarn, and woven and knitted fabrics with their finishing processes), and final products (carpets, home textiles, clothing, and industrial-use textiles) (EURATEX, 2000). The textile industry is a very diverse and heterogeneous industry, with its products being used by virtually everybody – private households and businesses alike. Downstream parts of the textile industry – such as the clothing industry – consume the output of more upstream parts (such as fabrics of all types and colors). The textile industry is also intertwined with the agricultural sector when it needs inputs in the form of natural fibers (such as cotton or wool), and with the chemical industry when it comes to the wide range of man-made fibers (such as nylon or polyester). Hardly any other industrial sector could do without the so-called technical textiles, which include products which are as diverse as filters, optical fibers, packing textiles, ribbons and tapes, air bags, insulation, and roofing materials (Stengg, 2001).
685 685 686 686 686 687 687 689 689 692 695 695 696 697 699 701 702 702 702 703 703
The textile industry is a significant contributor to many national economies, encompassing both small- and large-scale operations worldwide. In terms of its output or production and employment, the textile industry is one of the largest industries in the world. The textile manufacturing process is characterized by high consumption of different resources: water, fuel, and a variety of chemicals in a lengthy process that generates a significant amount of waste. The main environmental problems associated with the textile industry are typically those associated with water pollution caused by the discharge of untreated effluents. Other environmental issues of equal importance are air emission, notably volatile organic compounds (VOCs), excessive noise or odor, as well as workspace safety (UNEP, 1994).
4.20.1.1 Textile and Clothing Industry in Europe The textile and clothing sector is an important part of the European manufacturing industry, giving employment to more than 2 million people. Its importance for social and economic cohesion is increased by the fact that it is dominated by a large number of SMEs, which are often concentrated in particular regions, thus contributing greatly to their wealth and cultural heritage (Stengg, 2001). Being one of the oldest sectors in the history of industrial development, the textile and clothing industry is often referred to as a ‘traditional industry’, as a sector belonging to the socalled ‘old economy’. These notions divert attention from the fact that the European textile and clothing industry has
685
686
Water in the Textile Industry
undergone significant restructuring and modernization efforts during the past 10–15 years, making redundant about onethird of the total work force, increasing productivity throughout the production chain, and reorienting production toward innovative, high-quality products. Like many other sectors, the textile and clothing industry has been greatly affected by the phenomenon of globalization. Europe and the United States are not only important producers of textile and clothing products, but also the most attractive outlets for the so-called exporting countries, many of which are situated in South-East Asia. It should be noted that many developing countries and, indeed, even least developed countries have become very competitive in textiles and clothing, as they combine low-wage costs with high-quality textile equipment and know-how imported from more industrialized countries (Stengg, 2001). The textile and clothing industry is one of the world’s most global industries, and constitutes an important source of income and employment for many European Union (EU) countries. It is important to be aware of how the European textile and clothing industry operates, as well as its many complex structures and processes. The textile industry is a multifaceted area requiring a deep understanding of design, management, and technology. It plays a crucial role in creating innovative and attractive products of multiple uses for various users. It accounts for 5.7% of the production value of world manufacturing output, 8.3% of the value of manufactured goods traded in the world, and over 14% of world employment (Perivoliotis, 2002). Research and innovation have been important tools for the European textile and clothing industry to assert its leading position in global markets. The importance of research and innovation for continued industrial competitiveness is on the increase. The importance of the textile (and clothing) industry in the European economy is shown in Table 1 (EURATEX, 2002). The figures in Table 1 cover only a part of the total number of manufacturing companies in 2000 (i.e., they cover only companies with more than 20 employees). This portion of the industry represents
• • •
• • •
textile finishing, industrial and other textiles (including carpets and wool scouring), and home textiles.
4.20.1.2 Processes in Textile Industry The textile chain begins with the production or harvest of raw fiber. The basic steps in this chain are schematically represented in Figure 1 (US EPA/625/R-96/004, 1996).
4.20.1.2.1 Fibers Two general categories of fibers are used in the textile industry: natural and man-made (comments made by UK to the First Draft of the BREF Textiles, UK, 2001). Man-made fibers encompass both purely synthetic materials of petrochemical origin and regenerative cellulosic materials manufactured from wood fibers. A more detailed classification of fibers is presented in Table 2.
4.20.1.2.2 Finishing processes Pretreatment. Pretreatment processes should ensure (UBA, 1994)
• • •
the removal of foreign materials from the fibers in order to improve their uniformity, hydrophilic characteristics, and affinity for dyestuffs, and finishing treatments; the improvement of the ability to absorb dyes uniformly; and the relaxation of tensions in synthetic fibers.
Pretreatment processes and techniques depend
3.4% of EU manufacturing, 3.8% of the added valued, and 6.9% of industrial employment.
•
The textile chain is composed of a wide range of industrial subsectors, using the entire range of fibers. European industry is still engaged in all production stages, ranging from raw materials (in particular, the production of man-made fibers), to semiprocessed products (in particular, spinning, weaving, knitting, Table 1
and finishing activities), to the final products (e.g., home textiles, carpets, technical textiles, and garments) (Stengg, 2001). The complexity of the sector is also reflected in the difficulty of finding a clear-cut classification system for the different activities involved. As for the scope of this chapter, it is confined to those activities in the textile industry that involve wet processes. This refers primarily to those activities falling within the following new Classification of Economic Activities in the European Community (NACE):
• •
on the kind of fiber to be treated (natural or synthetic fibers), on the form of the fiber (flock, yarn, woven, or knitted fabrics), and on the amount of material to be treated.
Pretreatment operations are often carried out in the same type of equipment used for dyeing (in batch processing, in
Share of the EU-15 textile–clothing industry in the manufacturing industry (companies with 20 employees or more)
2000
Turnover (EUR, billion)
Added value at factor costs (EUR, billion)
Employment (million)
Turnover (%)
Added value (%)
Employment (%)
Textile Clothing
100.5 61.5
31.2 18.2
0.89 0.73
2.1 1.3
2.4 1.4
3.8 3.1
Total textile and clothing
162.0
49.4
1.62
3.4
3.8
6.9
4756.8
1308.0
23.62
100.0
100.0
100.0
Total manufacturing
Water in the Textile Industry
Polymers
Fibers manufacturing
Man-made fibers
Natural fibers
Fibers preparation
Finishing processes Pretreatment Dyeing
Loose fibers /stock
Yarn manufacturing – Spinning
Printing
Yarn
Finishing Fabric production Coating and laminating Carpet back coating
– – – –
Weaving Knitting Tufting Needle felt
particular, the material is most often pretreated in the same machine in which it is subsequently dyed). Dyeing. It is a method for coloring a textile material in which a dye is applied to the substrate in a uniform manner to obtain an even shade with a performance and fastness appropriate to its final use (Bailey et al., 2000; EURATEX, 2000). From a molecular point of view, four different steps are involved: 1. The dye, previously dissolved or dispersed in the dye liquor, diffuses from the liquor to the substrate. 2. The dye accumulates on the surface of the textile material. 3. The dye diffuses/migrates into the interior of the fiber until this is uniformly dyed. 4. The dye must be anchored (fixation) to suitable places within the substrate. Textiles can be colored at any of several stages of the manufacturing process and therefore the following coloring processes are possible:
• • •
Washing Fabric Drying Manufacture of end products
687
• • •
flock or stock dyeing; top dyeing, wherein fibers are shaped in lightly twisted roving before dyeing; tow dyeing, which consists in dyeing the mono-filament material (called tow) produced during the manufacture of synthetic fibers; yarn dyeing; piece (e.g., woven, knitted, and tufted cloths) dyeing; and ready-made goods (finished garments, carpet rugs, bathroom sets, etc.).
Clothing, knitwear, carpet, etc.
Figure 1 Schematic presentation of textile production.
Table 2
Classification of fibers
Natural fibers Animal origin Raw wool Silk fiber Hair Vegetable origin Raw cotton fiber Flax Jute Chemical fibers Natural polymers fibers Viscose, cupro, lyocell Cellulose acetate Triacetate Synthetic polymer fibers Inorganic polymer Glass for fiber glass Metal for metal fiber Organic polymer Polyester Polyamide Polyacrylonitrile Polypropylene Elastane
Dyeing can be carried out in a batch or in continuous/semicontinuous mode. The choice between the two processes depends on the type of makeup, the chosen class of dye, the equipment available, and the cost involved. Both continuous and discontinuous dyeing involve the following steps:
• • • •
preparation of the dye, dyeing, fixation, and washing and drying.
Printing. This is a process for applying color to a substrate. Print color is applied only to defined areas to obtain the desired pattern. This involves different techniques and different machinery with respect to dyeing, but the physical and chemical processes that take place between the dye and the fiber are analogous to dyeing. A typical printing process involves the following steps:
• • • •
Color-paste preparation. When printing textiles, the dye or pigment is not in an aqueous liquor; instead, it is usually finely dispersed in a printing paste, in high concentration. Printing. The dye or pigment paste is applied to the substrate using different techniques: Fixation. Immediately after printing, the fabric is dried and then the prints are fixed mainly with steam or hot air. After-treatment. This final operation consists in washing and drying the fabric (it is not necessary when printing with pigments or with other particular techniques such as transfer printing).
688
Water in the Textile Industry
Finishing (functional finishing). The term finishing covers all those treatments that serve to impart to the textile the desired end-use properties. These can include properties relating to visual effect, handling, and special characteristics such as waterproofing and nonflammability. Finishing may involve mechanical/physical and chemical treatments. Washing. Washing with water is normally carried out in hot water (40–1001C) in the presence of wetting agent and detergent. The detergent emulsifies the mineral oils and disperses the undissolved pigments. Washing always involves a final rinsing step to remove the emulsified impurities. Dry cleaning is sometimes necessary, especially for delicate fabrics. In this case, the impurities are carried away by the solvent, which is usually tetrachloroethylene (perchloroethylene). In the same step, softening treatments may also be carried out. In this case, water and surfactant-based chemicals are added to the solvent. Drying. It is necessary to eliminate or reduce the water content of the fibers, yarns, and fabrics following wet processes. Drying, in particular, by water evaporation, is a highenergy-consuming step.
Synthetic
Cotton
Wool
Fiber preparation
Scouring
Spinning
Carbonizing
W W Texturing
Warping Knitting
W Yarn dyeing W
Knitting Sizing
Heat setting
Weaving
Carbonizing
Singeing W Desizing W
W Scouring / washing
Wool felting
W Bleaching Singeing W
W Dyeing
Mercerizing
W Printing
4.20.2 Characteristic of Textile Water and Wastewater
W Finishing
4.20.2.1 Supply Water Cutting / sewing
The textile industry is very water intensive. Water is used for cleaning the raw material and for many flushing steps during the whole production (Water Treatment Solutions, 2010). In Figure 2, a general flowchart for processes in textile manufacturing is shown, and the processes that need the input water (marked with rounded W) (Bisschops and Spanjers, 2003). Processes using water are desizing, scouring or kiering, bleaching, mercerizing, dyeing, washing, neutralization, and salt bath. Most of them are presented in Tables 3–5. Textile operations vary greatly in water consumption. Wool and felted fabrics processes are more water intensive than other processing subcategories such as wovens, knits, stock, and carpet. Water use can vary widely between similar operations as well (US EPA/625/R-96/004, 1996). The highest water use generally refers to natural fibers. Synthetic fibers require lower water volumes per unit of product, mainly due to the lower cleaning and scouring needs (Matioli et al., 2002). Water consumption varies greatly among unit processes. Certain dyeing processes and print after washing are among the more intensive unit processes. Within the dyeing category, certain unit processes are particularly low in water consumption (e.g., pad batch) (US EPA/625/R-96/004, 1996). An abundant supply of clean water is necessary in order to run a dyeing and finishing plant. Dye houses are usually located in areas where the natural water supply is sufficiently pure and plentiful. Rivers, lakes, and wells represent the major sources of freshwater available for use in wet processing (Tomasino, 1992). Almost all dyes, especially chemicals, and finishing additives are applied to textile substrates from water baths. In addition, most fabric-preparation steps, including desizing, scouring, bleaching, and mercerizing, use aqueous systems.
End product
Figure 2 General flowchart for processes in textile manufacturing.
Table 3 Average water supply for different textile wet processes (Correia et al., 1994) Material
Process
Water usage (l kg1)
Cotton
Desizing Scouring or kiering Bleaching Mercerizing Dyeing
3–9 26–43 3–124 232–308 8–300
Wool
Scouring Dyeing Washing Neutralization Bleaching
46–100 16–22 334–835 104–131 3–22
Nylon
Scouring Dyeing
50–67 17–33
Acrylic
Scouring Dyeing Final scour
50–67 17–33 67–83
Polyester
Scouring Dyeing Final scour
25–42 17–33 17–33
Viscose
Scouring and dyeing Salt bath
17–33 4–13
Acetate
Scouring and dyeing
33–50
Water in the Textile Industry Table 4
Water usage (l kg1) for different materials and processes (Correia et al., 1994)
Material
Process Desizing
Wool Cotton Synthetic Nonspecified
Scouring
Bleaching
Dyeing
Printing
4–77.5 2.5–43 17–67
40–150 38–143 38–143
280–520
30–50
12.5–35
20–300
Table 5 Average, minimum, and maximum water supply for different textile operations (US EPA/625/R-96/004, 1996) Subcategory
689
Table 6
Liquor ratio for various dyeing processes
Process
l kg1
Dyeing winches Hank machines Jet dyeing Package dyeing Pad batch ULLR dyeing
20–30 30 7–10 5–8 5 5
Water usage (l kg1) Minimum Average Maximum
Wool scouring 4.2 Wool finishing 110.9 Low water use processing 0.8 Woven fabric finishing Simple processing 12.5 Complex processing 10.8 Complex processing plus desizing 5.0 Knit fabric finishing Simple processing 8.3 Complex processing 20.0 Hosiery processing 5.6 Carpet finishing 8.3 Stock and yarn finishing 3.3 Nonwoven finishing 2.5 Felted fabric finishing 33.4
11.7 283.6 9.2
77.6 657.2 140.1
78.4 86.7 113.4
275.2 276.9 507.9
135.9 83.4 69.2 46.7 100.1 40.0 212.7
392.8 377.8 289.4 162.6 557.1 82.6 930.7
The amount of water used varies widely in the industry, depending on the specific processes operated at the mill, the equipment used, and the prevailing management philosophy concerning water use. Different types of processing machinery use different amounts of water, particularly in relation to the bath ratio in dyeing processes (the ratio of the mass of water in an exhaust dyebath to the mass of fabric). Washing fabric processes greater quantities of water than dyeing. Water consumption of a batch-processing machine depends on its bath ratio and also on mechanical factors, such as agitation, mixing, bath and fabric turnover rate (called contact), turbulence, and other mechanical considerations, as well as physical flow characteristics involved in washing operations. All these factors affect washing efficiency (US EPA/625/R-96/004, 1996). The influence of the equipment and process selected is presented in Table 6 (EPA Victoria, 1998). From Table 6 we can see that hank machines and dyeing winches are the biggest water consumers (20–30 l kg1). Pad batch and ultralow liquor ratio dyeing processes need only 5 l kg1. The quantity of water used for a particular process also depends on equipment modernization and development. As an example, batch dyeing machines for knitwear have gone from 30 l kg1 to only 6 l kg1 of treated material over the last four decades (Wenzel and Knudsen, 2005). In general, heating of dyebaths constitutes the major portion of energy consumed in dyeing. Therefore,
low-bath-ratio dyeing equipment not only conserves water but also saves energy, in addition to reducing steam use and air pollution from boilers. Low-bath-ratio dyeing machines conserve chemicals as well as water and also achieve higher fixation efficiency. However, the washing efficiency of some types of low-bath-ratio dyeing machines, such as jigs, is inherently poor; therefore, a correlation between bath ratio and total water use is not always exact (US EPA/625/R-96/004, 1996). Water quality for all processes should be of such quality as to avoid any process and final-product-quality problems. Mostly, fresh softened water is used for all processes, although sometimes water of lower quality can be used as well. Three types of water quality are suggested for use in textile industry (Lockerbie and Skelly, 2003; Vandevivere et al., 1998): 1. High-quality water. It can be used for all processes, such as dyebaths, print pastes, finishing baths, and final rinse bath (Table 7). Consumption of such water is 10–20% of the total water consumption. Four different sources are presented: fresh softened water, recycled effluent (proposed), mains drinking-water prescribed concentrations or values (PCVs), and Confederation of British Wool Textiles (CBWT) water specification. 2. Moderate-quality water. It is used for washing-off stages after scouring, bleaching, dyeing/printing, and finishing (Table 8). About 50–70% of total water consumption consists of such water needs. Final rinse bath in the washing processes should be always high-quality water to ensure that the material is free from traces of contamination. 3. Low-quality water. It can be used for washing-down equipment, screen washing in print works, and general washdown of print paste containers and floors (Table 9). Quantity presents only 10–20% of total water consumption, but it is wasteful to use high-quality water for such operations.
690
Water in the Textile Industry
Table 7
Water quality suitable for all processes
Colora (mg l1 Pt scale) COD (mg l1 O2) pH Total hardness (mg l1) Chloride (mg l1) Sulfate (mg l1 SO4) Fe (mg l1) Cu (mg l1) Cr (mg l1) Al (mg l1) Mn (mg l1) Zn (mg l1)
Fresh softened water
Recycled effluent
Main water PVCs
CBWT specification
None visible
None visible 20–50 6.5–7.5 90b 500
20 5.5–9.5
None visible 6.0–8.0
250 (Ca), 50 (Mg) 400 250 0.2 3 0.05 0.2 0.050 5
60–80b
6.5–7.5 50b 300 0.05 0.05
0.1 0.005 0.01 0.02
0.1 0.1
0.05 0.1
a
Suggested specification for water with no visible color absorbance in 10 mm cell: 450 nm, 0.020.04; 500 nm 0.020.05; 550 nm, 0.010.03; 600 nm, 0.010.02. Measured as ppm CaC03. COD, chemical oxygen demand; PVC, polyvinylchloride; CBWT, Confederation of British Wool Textiles.
b
Table 8 Suggested water quality suitable for washing-off processesa
Table 9 only
Parameter
Parameter
b
Color COD (mg l1) pH Total hardness (ppm CaCO3) Chloride (mg l1) Fe (mg l1) Cu (mg l1) Cr (mg l1)
Maximum recommended level None visible 200 7.0–8.0 100 500–2000 0.1 0.05 0.1
a
Final rinse bath to use high-quality water. Suggested specification for water with no visible color absorbance in 10 mm cell: 450 nm, 0.020.04; 500 nm, 0.020.05; 550 nm, 0.010.03; 600 nm, 0.010.02.
a
Suggested water quality suitable for equipment washdown
Color COD (mg l1) pH Total hardness (ppm CaCO3) Chloride (mg l1) Fe (mg l1) Cu (mg l1) Cr (mg l1)
Maximum recommended level None visible 500–2000 6.5–8.0 100 3000–4000 0.1 0.05 0.1
a
b
Suggested specification for water with no visible color absorbance in 10 mm cell: 450 nm, 0.020.04; 500 nm, 0.020.05; 550 nm, 0.010.03; 600 nm, 0.010.02.
4.20.2.2 Textile Wastewater
In Europe, 108 million tons of wastewater is produced on a yearly basis and 36 million tons of chemicals and auxiliaries have to be removed from the textile wastewater. Textile wastewater typically contains a complex mixture of organic and inorganic chemicals, due to the wide variety of the process steps.
The textile industry is one of the most polluting industries. Many different processes are used and almost all of them generate wastewater. Wastewater from textile sector is composed of cleaning water, process water, noncontact cooling water, and storm water. The amount and the composition of wastewater vary and depend on different factors, including the nature of the processed fabric, applied dye, or special finishing; the type of the process; the equipment used; and the prevailing management philosophy regarding water use. Changes in machines, used chemicals, or any characteristic of the processes also change the nature of the generated wastewater. Scouring, dyeing, printing, finishing, and washing generate the majority of the textile wastewater. Large-volume wastes include wash water from preparation and continuous dyeing, alkaline wastewater from preparation, and batch dye wastewater containing large amounts of dye, salts, acids, or alkalis, and also other toxic additives in smaller amounts. Primary sources of biological oxygen demand (BOD) include waste chemicals or batch dumps, starch-sizing agents, knitting oils, and degradable surfactants.
4.20.2.2.1 Textile wastewater from different process steps The following processes in the textile industry produce wastewater containing different pollutants: 1. Desizing. It is the process for removing the size chemicals from the textile. Wastewater from the desizing process varies according to the used sizes and recipes and contains pollutants such as different additives, surfactants, enzymes, acids or alkalis, as well as the size themselves. The generated wastewater can be the largest contributor to the overall BOD and the total suspended solids (TSSs). When the natural sizes, based on starch or proteins, are used for sizing, the wastewater after desizing is characterized by high BOD and BOD/COD ratio. If sizing is carried out using synthetic materials, such as polyvinyl alcohol or carboxymethyl cellulose, the BOD reduction can be up to
Water in the Textile Industry
90%. Possible pollutants in wastewater after desizing process are shown in Table 10 (Correia et al., 1994). 2. Scouring. It is the process for removing different impurities from both natural and synthetic materials. The intensity of the scouring process depends on the type of material. Oils, fats, waxes, minerals, and plant matter can be present in natural fibers, whereas synthetic fibers can contain spin finishing and knitting oils. These impurities can be removed either with water or with organic solvents. Water scouring is usually preferred over solvent scouring, because water is nonflammable, nontoxic, plentiful, and cheaper. For cotton scouring, hot alkaline solutions, containing detergents or soaps, are used. Sourcing effluents can also contain herbicides, insecticides, defoliants, and desiccants, which are used in the growing of cotton, as well as fungicides such as pentachlorophenols used to prevent mildew during storage and transportation of cotton. Raw-wool scouring is the most polluting process in the textile industry. The pollution load results from impurities present in raw wool such as wax, suint, urine, feces, vegetable
691
matter, mineral dirt, and, on the other hand, the soap detergent and alkali used during the scouring and washing processes. Wool grease is the major problem in treating wool, because of its nonbiodegradability. It is a mixture of cholesterol esters, long-chain fatty acids, free fatty acid, free alcohol, and hydrocarbons. Synthetic souring requires less scouring than cotton or wool. Inorganic and organic substances which could be present in wastewater after scouring, for different fibers, are shown in Table 11 (Correia et al., 1994). 3. Bleaching. It is commonly used to remove natural coloring of cotton and other fibers. In this step, the most common agents are hydrogen peroxide, sodium hypochlorite, sodium chlorite, and sulfur dioxide gas. Auxiliary chemicals such as sulfuric acid, hydrochloric acid, sodium hydroxide (caustic soda), sodium hydrogen sulfite (sodium bisulfite), surfactants, and chelating agents are also used and released into the wastewater. Bleaching wastewater usually has high solid content with low-to-moderate BOD levels. Inorganic and organic substances which could be present in wastewater after bleaching, for different fibers, are shown in Table 12 (Correia et al., 1994). 4. Mercerizing. It improves strength, luster, and dye affinity of cotton fabrics. Cotton fabrics are treated with solutions of sodium hydroxide (caustic soda) followed by neutralization and several rinses. Wastewater generated by mercerizing has low BOD and total solid levels but high pH (Table 13) (Correia et al., 1994). 5. Dyeing. The dyeing operations of textiles may take part in the process chain at different stages of production (fibers, yarn, or piece dyeing). Stock dyeing is used to dye fibers. Top dyeing is used to dye combed wool silver. Yarn dyeing and piece dyeing are used after the yarn has been constructed into the fabric.
Table 10
Possible pollutants in desizing effluents
Fibers
Inorganic substances
Organic substances
Cotton Linen Viscose
Naþ, Ca2þ, NH4 þ , SO4 2 , CI
Silk Acetates Synthetics
Naþ, NHþ 4, CO3 2 , PO4 3
Carboxymethyl cellulose, enzymes, fats, hemicellulosses, modified starches, nonionic surfactants, oils, starch, waxes Carboxymethyl cellulose, enzymes, fats, gelatine, oils, polymeric sizes, polyvinyl alcohol, starch, waxes
Table 11
Possible pollutants and characteristics of effluents from scouring
Fibers
pH
BOD (mg l1)
TSS (mg l1)
Inorganic substances
Organic substances
Cotton
10–13
50–2900
7600–17 400
Naþ, CO3 2 , PO4 3
Anionic surfactants, cotton waxes, fats, glycerol, hemicelluloses, nonionic surfactants, peptic matter, sizes, soaps, starch
Viscose
8.5
2832
3334
Na þ , CO3 2 , PO4 3
Acetates
9.3
2000
1778
Anionic detergents, fats, nonionic detergents, oils, sizes, soaps, waxes
Naþ, CO3 2 , PO4 3
Anionic surfactants, antistatic agents, fats, nonionic surfactants, oils, petroleum spirit, sizes, soaps, waxes
Naþ, NH4 þ , CO3 2 , PO4 3
Anionic detergents, glycol, mineral oils, nonionic detergents, soaps
Naþ, NH4 þ , Kþ, Ca2 þ , CO3 2 , PO4 3
Acetate, anionic surfactants, formate, nitrogenous matter, soaps, suint, wool grease, wool wax
Synthetics
Wool (yarn and fabric)
Wool (loose fiber)
9–14
3000–40 000
1129–64 448
692
Water in the Textile Industry
Textiles are dyed using a wide range of dyestuffs, techniques, and equipment. Each dyeing process requires different amounts of dye per unit of fabric to be dyed. In the textile industry, synthetic dyes, derived from coal tar and petroleum-based intermediates, are used. Dyes can be present as powders, granules, pastes, and liquid dispersions, with concentrations of active ingredients ranging typically from 20% to 80%. Dyeing can be performed by using continuous or batch processes. Auxiliary chemicals and controlled dye-bath conditions accelerate and optimize the migration of the dye molecules from the solutions to the fiber. The dye is fixed on the fiber thermally and/or chemically. Table 12
Possible pollutants in bleaching effluents
Fibers
Inorganic substances
Organic substances
Cotton Linen Viscose Jute
Naþ, NH4 þ CIO, CI, O2 2 , F, SiO3 2
Formate
Synthetics Acetates
SiO3 2 , PO4 3 , F
Wool
Naþ, O2 2
Table 13
The water consumption in dyeing processes is very high (up to 300 l kg1). Water is used not only in the dyeing process itself, but also for rinsing operations of the dyed material. Dyes and different auxiliaries such as organic acid, fixing agents, defoamers, oxidizing/reducing agents, and diluents are typical pollutants generated in the dyeing step. Quite a large amount of the unfixed dye leaves the dyeing unit. Metals and almost all of the salts and dyes present in the overall textile wastewater originate from dyeing operations. The possible pollutants and characteristics of effluents from dyeing processes for different fibers are listed in Table 14 (Correia et al., 1994). 6. Printing. For fabric printing, many different colorants and patterns, including a variety of techniques and machines, are used. The most common printing techniques used are rotary screen, and other methods such as direct, discharge, resist, flat screen, and roller printing often used commercially. Pigments are used for about 75–85% of all printing operations. Pigments do not require washing steps and generate little waste. Compared to the dyes, pigments are typically insoluble and have high affinity for the fibers. An important component in textile printing is the print paste, which consists of water, thickeners, dyes, urea, and various other chemicals such as surfactants and organic solvents. The printing method determines the wastewater characteristics. Printing wastewaters are small in volume
Oxalate
Possible pollutants and characteristics of effluents from mercerizing
Fibers
pH
BOD (mg l1)
TSS (mg l1)
Inorganic substances
Organic substances
Cotton Linen
5.5–9.5
45–65
600–1900
Naþ, NH4 þ , CO3 2 , SO4 2
Alcohol sulfates, anionic surfactants, cyclohexanol
Table 14 Fibers Cotton Linen
Possible pollutants and characteristics of effluents from dyeing pH
BOD (mg l1)
TSS (mg l1)
Polyester
3þ
2þ
Organic substances
11–1800
500–14 100
Na , Cr , Cu , Sb3þ, Kþ, NH4 þ , CI CO3 2 , CO4 2 , F, NO2 , O2 2 , S2, S2 O3 2 , SO3 2 , SO4 2
Naphtol, acetate, amides of naphtoic acid, anionic dispersing agents, anionic surfactants, cationic fixing agents, chloro amines, formaldehyde, formate, nitro amines, nonionic surfactants, residual dyes, soaps, soluble oils, sulfated oils, tannic acid, tartrate, urea
4.8–8
380–2200
3855–8315
Naþ, Cr3þ, Cu2þ, Sb3þ, Kþ, NH4 þ , Al3þ CI, CO3 2 , S2 O4 , SO3 2 , SO4 2 Naþ CI, CO3 2
Acetate, dispersing agents, formate, lactate, residual dyes, sulfated oils, tartrate
Naþ, NH4 þ , Cu2þ, SO4 2
Acetate, aromatic amines, formate, leveling agents, phenolic compounds, residual dyes, retardants, surfactants, thiourea dioxide
Naþ, NH4 þ , Cl, S4 O6 2 , CIO, SO3 2 , NO3
Acetate, anionic surfactants, antistatic agents, dispersing agents, dye carriers, EDTA, ethylene oxide condensates, formate, mineral oils, nonionic surfactants, residual dyes, soaps, solvents
Polyamide Acrylic
þ
5–10
Viscose
Wool
Inorganic substances
1.5–3.7
175–2000
480–27 000
833–1968
Acetate, formate, polyamide oligeines, residual dyes, sulfonated oils
Water in the Textile Industry
and contain urea, dyes or pigments, organic solvents, and metals. The concentration of the pollutants in printing wastewater is higher than that in dyeing wastewater. 7. Finishing. This can refer to the chemical or mechanical treatments performed on fiber, yarn, or fabric to improve appearance, texture, or performance. Mechanical finishes can involve brushing, ironing, or other physical treatments used to increase luster and feel of textiles, such as heat setting, napping, softening, optical finishing, shearing, and compacting. The application of chemical finishes to textile can impart a variety of properties ranging from decreasing static cling to increasing flame resistance. Chemical treatments are optical finishes, adsorbent and soil-release finishes, softeners and abrasion-resistant finishes, and physical stabilization and crease-resistant finishes. Wastewaters from the finishing units are extremely variable in composition and can contain resins, waxes, softeners, acetate, stearate, as well as toxic organic compounds (pentachlorophenols and ethylchlorophosphates).
4.20.2.2.2 General characteristics of textile wastewater Textile wastewater is characterized mainly by measuring BOD, chemical oxygen demand (COD), suspended solids, and dissolved solids. Typical characteristics of textile industry wastewater are presented in Table 15. Wastewaters from the textile industry are usually polluted with recalcitrant or hazardous organics, such as dyes, surfactants, metals, salts, and persistent organic pollutants (POPs) as well. They are discussed in the following: 1. Dyes. Most of the wastewater produced during the textile material processing is colored. The main sources of color in the textile effluents arise from dyes and pigments in the dyeing and printing operations. It is known that the presence of very small quantities of dyes in water (less than 1 ppm) is highly visible due to their brilliance. There are more than 10 000 commercially available dyes with a production of over 7 105 tons yr1 (Zollinger, 1987). The exact data on the quantity of dyes discharged into the environment are also not available. It is assumed that 2% of the dyes produced is discharged directly in aqueous effluent, and B10% is subsequently lost during the textile coloration process (Easton, 1995). Dyes cause a lot of problems in the environment. They can remain in the environment for an extended period of time, because of high thermal and photo stability (the half-life of hydrolyzed Reactive Blue 19 is about 46 years at pH 7 and 25 1C Table 15
Characteristics of textile wastewater
Parameters
Values
pH BOD (mg l1) COD (mg l1) TSS (mg l1) TDS Chloride (mg l1) Total Kjeldahl nitrogen (mg l1) Color (Pt–Co)
1.9–13 50–40 000 150–12 000 15–64 000 2900–3100 1000–1600 70–80 50–2500
693
(Hao et al., 2000)). Depending on dye concentration and exposure time, dye can have acute and/or chronic effects on exposed organism. The greatest environmental concern with dissolved dyes is their absorption and reflection of sunlight entering the water (rivers, lakes, etc.). Light absorption diminishes photosynthetic activity of algae and seriously influences the food chain. Many dyes and their breakdown products are carcinogenic, mutagenic, and/or toxic to life. Mathur et al. (2005) studied the influence of textile dyes (known only by their trade name) on the health of textile-dyeing workers and the environment. The dyes were used in their crude form (without previous purification), because they wanted to test the potential danger that dyes represent in actual use. The results clearly indicated that most of the used dyes are highly mutagenic. Brown and DeVito (1993) studied how it is possible to predict the toxicity of new azo dyes. The systematic backtracking of the flows of wastewater from textile-finishing companies led to the identification of textile dyes as a cause for strongly mutagenic effects. Several textile dyes used in the textile-finishing companies in the European Union were examined for mutagenicity. According to the obtained results, the dyes which were considered to present a potential toxicity have been withdrawn from the market and have been replaced with less harmful and biodegradable substances (Ja¨ger et al., 2004; Schneider et al., 2004). Degradation of dye Direct Blue 14 led to the carcinogenic aromatic amine o-tolidine (Platzek et al., 1999). Dyes can cause allergies such as contact dermatitis (Pratt and Taraska, 2000) and respiratory diseases(Estlander, 1988; Wilkinson and McGechaen, 1996; Zuskin et al., 1998), allergic reaction in eyes, skin irritation, and irritation to mucous membrane and the upper respiratory tract. As it is known, reactive dyes form covalent bonds with cellulose, woolen, and polyacrylate fibers. It is assumed that in the same manner, reactive dyes can bond with –NH2 and –SH groups of proteins in living organisms. Many investigations have been made on respiratory diseases in workers dealing with reactive dyes. Certain reactive dyes have caused respiratory sensitization of workers occupationally exposed to them (Majcen Le Marechal et al., 1996). Organic dyes contain substituted aromatic and heteroaromatic groups. The color of dyes results from conjugated chains or rings that can absorb different regions of wavelength. The chromophores of organic dyes are usually composed of double carbon–carbon bonds, double nitrogen– nitrogen bonds, double carbon–nitrogen bonds, and aromatic and heterocyclic rings containing oxygen, nitrogen, or sulfur. Azo dyes, which contain one or more azo bonds, are the most widely used synthetic dyes and are present in 60–70% of all textile dyestuffs produced (Carliell et al., 1995). Azo dyes can be used on natural fibers (cotton, silk, and wool) and synthetic fibers (polyesters, polyacrylic, rayon, etc.). Azo dyes are mostly used for yellow, orange, and red colors. Biodegradation of more than 100 azo dyes have been tested and it was found that only a very few were degraded aerobically. The degree of stability of azo dyes under aerobic conditions depends on structure of the molecule. Dye C.I. Acid Orange 7 is one of the rare dyes which is aerobically biodegradable. Under anoxic conditions, azo dyes
694
Water in the Textile Industry
are cleaved to aromatic amines, which are not further metabolized under anaerobic conditions but are readily biodegraded in an anaerobic environment (Figure 3) (Vandevivere et al., 1998). Anthraquinone dyes constitute the second most important class of textile dyes. They have a wide range of colors in almost the whole visible spectrum, but they are most commonly used for violet, blue, and green colors. With regard to method and domain of usage, dyes are classified into acid, reactive, direct, basic, disperse, metal complex, vat, mordant, and sulfur dyes. Most commonly in use today are reactive and direct dyes for cotton and viscose-rayon dyeing and disperse dyes for polyester dyeing. Reactive dyes are termed chemically as colored compounds with a functional group capable of forming a covalent bond with a suitable substrate. Reactive dyes represent 20– 30% of the total dyes in the market. Reactive dyes are characterized by low fixation rate, and around 30% of the applied reactive dyes are wasted because of dye hydrolysis in the alkaline conditions of the dyebath. As a result, dyehouse effluents typically contain 0.6–0.8 g dye dm3 (Stenken-Richter and Kermer, 1992). Generally, dyes can be classified with regard to (1) their chemical structure, (2) the method and domain of usage, and (3) chromogen (Table 16).
HO
N HO3S
Ar — NH2
N
Figure 3 Chemical structure and degradation under anoxic condition of the azo dye with C.I. Acid Orange 7.
Table 16
In the textile-dyeing process, dyes are always used in combination with other chemicals (acids, alkali, salts, fixing agents, carriers, dispersing agents, and surfactants) which are partly or almost completely discharged in the wastewater together with the numerous additives and impurities present in the commercial dye products. Public perception of water quality is greatly influenced by the color. Therefore, the removal of color from wastewater is often more important than the removal of the soluble colorless organic substances. 2. Metals. Many textile mills have metals in their effluent, but their concentration decreased in the last decade, mainly because of the reduction of the metal contents in the dye. Metals include copper, cadmium, chromium, nickel, zinc, and lead. Metals enter the textile effluents in many ways: incoming supply water, metal parts (such as pumps, pipes, and valves), oxidizing and reducing agents, electrolyte, acid and alkali, dyes and pigments, certain finishes, herbicides, and pesticides. However, the main source of heavy metals is the dyeing process. Dyes may contain metals such as zinc, cobalt, and chromium. In some dyes, metals can form an integral part of the dye molecule; metals are functional, but in most dyes metals are just impurities generated during the dye manufacture. Mercury or other metals may be used as catalysts in the synthesis of dyes and may be present as by-products. Concentrations of metals in the dyeing effluents can be in the range 1–10 mg l1. For example, after dyeing of wool with basic dyes, the concentration of cadmium in wastewater is 7.5 mg l1. The concentration of chromium in dyeing effluents after dyeing cotton with direct dyes is 12.05 mg l1. Dyeing viscose with direct dyes revealed measurements of 2.7 mg of chromium l1, 8.52 mg of copper l11, and 1.95 mg of lead l1 in the wastewater. (EURATEX, 2000).
Classification of the dyes Classification
With regard to chemical structure (C.I.) With regard to method and domain of usage (C.I.) With regard to chromogen n-p* With regard to the nature of donor– acceptor couple With regard to the nature of polyenes Acyclic and cyclic
Cyanine
Subclass
Characteristic
Azo, anthraquinone, triphenylmethane, indigo, etc. Direct, acid, basic, reactive, reductive, sulfuric, chromic, metal-complex, disperse, pigment, etc. Absorptive, fluorescent and dyes with energy transfer, etc. 1-Aminoanthraquinone, p-nitroaniline, etc.
The classification of a dye by chemical structure into a specific group is determined by the chromophore Dyes used in the same technological process of dyeing and with similar fastness are classified into the same group This classification is based on the type of excitation of electrons, which takes place during light adsorption These chromogens contain a donor of electrons (unbound electron couple), which directly bonds to the system of conjugated p electrons
Polyolefins, annulenes, carotenoids, rhodopsin, etc.
Polyene chromogen contains sp2 (or sp) hybridized atoms. The molecules enclose single and double bonds that form open chains, circles, or a combination of both Cyanine chromogens have a system of conjugated p electrons, in which the number of electrons matches the number of p-orbitals
Cyanines, amino-substituted di- and tri-arylmethane, oxonols, hydroxyarylmethanes, etc.
Water in the Textile Industry
3. Salts. The presence of salts in textile wastewater has been identified as a potential problem by several authors. Salts in textile processes are used as raw materials or produced as by-products of neutralization, or in other reactions. Salt is used mostly to assist the exhaustion of ionic dyes, particularly anionic dyes, such as direct and reactive dyes on cotton. Typical cotton batch-dyeing operations use salts in the range 20–80% weight of dyed material. The concentration of salts in such wastewater is 2000–3000 ppm (Matioli et al., 2002). Sodium chloride (common salt) and sodium sulfate (Glaubers salt) constitute the majority of total salt use. Other salts used as raw materials or formed during the textile operations include magnesium chloride (Epson salt) and potassium chloride, and others in low concentrations. 4. Persistent organics or hazardous organics. The persistent molecules present in textile wastewater belong to very diverse chemical classes, each used in relatively small amounts. The persistent organics include surfactants or their byproducts, dyeing auxiliaries such as polyacrylates, phosphates, sequestering agents (ethylenediaminetetraacetic acid (EDTA)), deflocculating agents (lignin or naphtahalenesulfonates), antistatic agents for synthetic fibers, carriers in disperse dyeing of polyester, fixing agents in direct dyeing of cotton, preservatives (substituted phenol), and a large number of finishing auxiliaries used for fireproofing, mothproofing, and water proofing. The most toxic among POPs are the commonly named dioxins and dioxin-like compounds. Dioxin (Figure 4) is the term for a group of chemical compounds with 75 polychlorinated dibenzo-p-dioxins (PCDDs) and 135 polychlorinated dibenzofurans (PCDFs). The textile industry is a potential source of PCDD/Fs. They can arise from the various processes involved in the industry (Krizˇanec and Majcen Le Marechal, 2006): Pesticide pentachlorophenol (PCP) is used as a biocide for cotton and other materials. Pesticides, such as pentachlorophenol, are known to be contaminated with PCDD/Fs. Dyestuffs are contaminated by PCDD/Fs. Textile processes may utilize chlorinated chemicals contaminated by PCDD/Fs. Washing processes in alkaline media are part of the textile finishing processes. Large volumes of effluent water are released into the environment. The main source of dioxins in the textile industry are dioxazine and antraquinone dyes and pigments, produced 9 Clx
O
1
8 7 6
O
9 2
ClY
3 4
Clx
1
8 7 6
O
2
ClY
3 4
X + Y = 1– 8 (75 congeners)
X + Y = 1– 8 (135 congeners)
Polychlorinated dibenzo-p-dioxins
Polychlorinated dibenzofurans
Figure 4 Molecular structure of the polychlorinated dibenzo-p-dioxins and dibenzofurans.
695
from chloranil as intermediate product, and chloranil itself used as a catalyst in the production of dyes and pigments. Various dyes and pigments were analyzed for the presence of PCDD/Fs. Considerable levels of PCDD/Fs were determined in some dioxazine dyes and pigments, phatalocyanine dyes, and in printing inks. Concentrations of PCDD/Fs in Direct Blue 106 dye, Direct Blue 108 dye, and Violet 23 pigment were in the mg kg1 range with octachlorodibenzodioxin (OCDD) and octachlorodibenzofuran (OCDF) as dominant homologs. The concentration of OCDD in Direct Blue 106 was 41.9 mg kg1 and the concentration of OCDF was 12.4 mg kg1 (Williams et al., 1992). Hutzinger and Filder (US Environmental Protection Agency, 2000) found mg kg1 range levels of PCDD/Fs for higher chlorinated congress in sample of Ni-phthalocyanine dye. Results of the analyses of PCDD/Fs were reported for four printing inks obtained from a supplier in Germany. In the two inks used for rotogravure printing and two used for offset printing, the content of PCDD/Fs ranged from 17.7 to 87.2 ng TEQ kg1 (TEQ, toxicity equivalent; Santl et al., 1994). A high concentration of mixed polychlorinated and polybrominated dibenzo-p-dioxins and polychlorinated and polybrominated dibenzofurans (PBCDD/Fs) was detected after flame-retardant finishing-textile processes. A flame-retardant finish on upholstery material on the basis of PVC, Sb2O3, and hexabromocyclododecane results in the final product concentrations up to 19 mg kg1 of PBCDD/Fs. PCP and other chlorophenols can be the source of PCDD/Fs in wastewaters. A generation of dioxins was reported from the direct photolysis of pentachlorophenolcontaining water. Waddell et al. (1995) investigated the formation of dioxins by the ultraviolet (UV) photolysis of pentachlorophenol with or without addition of H2O2. Their study showed high levels of PCDD, especially OCDD. The presence of halogenated organic compounds (adsorbable organic halides (AOX)) in textile wastewater may derive from hypochlorite bleaching operations or from spent liquors following shrink-proofing finishing treatment by chlorine. The effluents after bleaching with hypochlorite may contain up to 100 mg dm3 AOX including considerable amounts of chloroform. Some reactive dyes also contain AOX. In the effluent from textiledyeing operation, an average of 0.75 mg dm3 was measured (Grutner et al., 1994). 5. Toxicity of wastewater. The toxicity of textile wastewater varies considerably among different processes in textile industry. Wastewater of some processes have high aquatic toxicity, while others show little or no toxicity. It is impossible to identify all toxic compounds used in textile production, because of the huge variety of chemicals used and the lack of data about their toxicities. Textile wastewater can contain thousands of different compounds, and identifying and testing all of them are practically impossible and too expensive. In general, the overall toxicity is determined by the toxicity test of the whole effluent stream on aquatic organisms, which is a cost-effective method. Table 17 summarizes the results for about
696
Water in the Textile Industry
Table 17 textile mills
Results from aquatic toxicity testing of effluent from 75
Toxicity (%)
Number of tests
o9 10–19 20–29 30–39 40–49 50–59 60–69 70–79 80–89 90–100 4100 (no toxicity)
7 6 8 2 4 9 3 8 2 3 38
Table 18 Agent Salt Surfactants Metals Organics Biocides Toxic anions
Typical causes of aquatic toxicity Chemical example NaCI, Na2SO4 Ethoxylated phenols Copper, zinc Chlorinated solvents Pentachlorophenol Sulfide
Source Dyeing Multiple processes Dyes Scour, machine cleaning Wool fibers contaminant Sulfur dyeing
75 companies (Horning, 1977); toxicity in the table is LC50 in percent and the higher number represents the lower toxicity. The source of aquatic toxicity can be dyes, salt, surfactants, ionic metals, toxic organic chemicals, biocides, and toxic anions. Examples of compounds in each of these classes and their source are shown in Table 18 (EPA/625/R96/004, 1996).
4.20.3 Treatment and Reuse of Textile Wastewater 4.20.3.1 Wastewater Treatment Technologies Textile wastewater may be treated by physical, chemical, or biological methods. For decoloration and degradation of textile wastewaters, many treatment technologies have been developed, but every existing technology presents limitations – advantages and disadvantages. Textile wastewater is very complex, so the use of a universal wastewater treatment seems to be impossible. The wastewater-treatment technologies used will depend on the wastewater characteristic (type, dye concentration and auxiliaries, and pH). It is apparent that a single wastewater-treatment system is unable to overcome all problems by itself to provide an efficient treatment of effluents and be cost effective at the same time. In this section, an overview of treatment technologies used in textile effluents is presented. Dyes containing wastewater can be treated by chemical or physical methods of dye removal, which refer to the process called decoloration, and by means of biodegradation, which tells us more about the fate of dyes in the environment.
Physical methods include different precipitation methods (coagulation, flocculation, and sedimentation), adsorption (on a wide variety of inorganic and organic supports), filtration, reverse osmosis, ultrafiltration, and nanofiltration. Biological treatments differ according to the presence or absence of oxygen and are termed aerobic and anaerobic treatment, respectively. Since biological treatments simulate degradation processes that occur in the environment, they are also called biodegradation. Chemical treatment methods are those in which the removal or conversion of dyes and other contaminants is brought about by the addition of chemicals or by chemical reactions (reduction, oxidation, compleximetric methods, ion exchange, and neutralization). The treatment of colored wastewaters is therefore restricted not only to the reduction of ecological parameters (COD, BOD, total organic carbon (TOC), AOX, temperature, and pH), but also to reduction of dye concentrations in wastewaters.
4.20.3.1.1 Physical methods The physical methods of treating textile wastewater are as follows: Adsorption. It is the process of collecting soluble substances that are in solution on a suitable interface. Adsorption methods for decoloration are based on the high affinity of many dyes for adsorbent materials. Some physical and chemical factors have an influence on dye removal by adsorption. These factors are dye-adsorbent interactions, adsorbent surface area, particle size, temperature, pH, and contact time. The main criteria for selection of an adsorbent should be based on characteristics such as high affinity and capacity for target compounds and the possibility of adsorbent regeneration (Santos et al., 2007). Adsorption on sludge is the main abiotic mechanism of removing dyes from wastewater. The most important factors influencing the adsorption test are sludge quality, water hardness, duration of the test, and test-substance concentration. Pagga and Taeger (1994) have described the static and dynamic removal studies involving water-soluble dyes (acid and reactive) and poorly soluble dyes (disperse). Activated carbon is the most commonly used method of dye removal by adsorption. It is very effective in adsorbing cationic, mordant, and acid dyes, and to a slightly lesser extent, disperse, direct, vat, pigment, and reactive dyes (Nassar and El-Geundi, 1991; Raghavacharya, 1997). Its performance depends on the type of carbon used and the characteristic of the wastewater. It is, like many other dye-removal treatments, well suited for one particular waste system and ineffective for another. Activated carbon is relatively expensive and has to be regenerated offsite with losses of about 10% in the thermal regeneration process (Robinson et al., 2001). Biomass referring to the dead plant and animal matter is also a suitable adsorbent for wastewater treatment. The adsorption of organic material onto various types of waste biomass such as sawdust (Poots et al., 1976a), peat (Poots et al., 1976b), chitin (McKay et al., 1982), bagasse pith (Al-Duri et al., 1990), carbonized wool waste (Malmary et al., 1985), wood chips (Nigam et al., 2000), maize cob (El Geundi, 1991), banana pith (Namasivayam et al., 1993), rice husk,
Water in the Textile Industry
hair, cotton waste, and bark (McKay et al., 1987) has been studied. The capacities of these materials have been examined through their adsorption of synthetic dyes. Two mechanisms are presented on the decoloration occurring in the biomass – adsorption and ion exchange. Both of them are influenced by dye–sorbent interaction, sorbent surface area, particle size, temperature, pH, and contact time. Biomass of different origins has been used for decoloration of acid, direct, and reactive dyes. Of all the described adsorbents, only a few have characteristics necessary for commercial use. Considering the price and binding capacity, quarternized lignocellulose-based adsorbents are the most appropriate for treating wastewatercontaining acid dyes. After the adsorption processes, the adsorbent needs to be regenerated, which adds to the cost of the process, and is sometimes a very time-consuming procedure. Decoloration with alternative materials such as zeolites, polymeric resins, ion exchangers, and granulated ferric hydroxide has also been studied in order to decrease adsorbent losses during regeneration. Filtration methods. Ultrafiltration, nanofiltration, and reverse osmosis can be used in the textile industry. These methods can be used not only for both filtering and recycling pigment-rich streams, but also for mercerizing and bleaching wastewaters. The specific temperature and chemical composition of the wastewater determine the type and porosity of the filter to be applied. The main drawbacks of membrane technology are high investment costs, potential membrane fouling, and the production of a concentrated dyebath which needs to be treated (Mishra and Tripathy, 1993; Xu and Lebrun, 1999). Coagulation and flocculation processes. These are widely used in several wastewater treatments in Germany and France. Coagulant agents such as aluminum sulfate, ferrous and ferric sulfate, ferric chloride, calcium chloride, copper sulfate, as well as several copolymers such as pentaethylene, hexamine, and ethylediene dichloride are used to form flocks with the dye, which are then separated by filtration or sedimentation. Coagulation–flocculation methods were successfully applied for decoloration of sulfur and disperse dyes, whereas acid, direct, reactive, and vat dyes presented very low coagulation–flocculation capacity. Polyelectrolyte can also be dosed during the flocculation phase to improve the flock settleability (Lee, 2000; Anjaneyulu et al., 2005). The main advantage of these processes is decoloration of the waste stream due to the removal of dye molecules from the dyebath effluents, and not due to a partial decomposition of dyes, which can lead to an even more potentially harmful and toxic aromatic compound. The major disadvantage of coagulation–flocculation processes is the production of sludge.
4.20.3.1.2 Chemical processes Some of the chemical processes are described in the following: Oxidation. The simplicity of its application makes oxidation the most commonly used chemical decoloration process. With conventional oxidation treatments, it is difficult to oxidize dyes (mainly for removing color) and toxic organic compounds in textile effluents. The development of so-called advanced oxidation processes (AOPs) has overcome the chemical limitations of conventional chemical oxidation
697
techniques. The goal of AOPs is to generate free hydroxyl radicals (OHd) which may represent a rate increase of one to several orders of magnitude compared with normal oxidants in the absence of catalysts. Hydroxyl radicals oxidize the dyes and toxic organic compounds. In AOPs, oxidizing agents such as ozone and hydrogen peroxide are used with catalysts (Fe, Mn, and TiO2), either in the presence or in the absence of an irradiation source. Table 19 shows the oxidation potential of common species. Fenton’s reagent. Hydroxyl radicals are activated by Fe2þ (ferrous ions) in an acid solution (pH ¼ 3–4) (Table 20) from hydrogen peroxide. In this process, it is important to find the optimal concentration of hydrogen peroxide because excess of H2O2 acts as a scavenger of radicals, disturbs the COD measurements, and is toxic for microorganisms. This method is suitable for the oxidation of wastewaters, which inhibit biological treatment or are poisonous. Fenton’s reagent offers a cost-effective source of hydroxyl radicals and it is easy to operate and maintain. The advantages of this system are COD, color, and toxicity reduction and the disadvantage is sludge generation, through flocculation; impurities are transferred from the wastewater to the sludge, which contains the concentrated impurities and is still ecologically questionable. Conventionally, it has been incinerated to produce power, but such a disposal, according to some, is far from being environment friendly. To avoid this problem, Gnann et al. (1993) suggest the regeneration of Fe2þ from iron sludge at pHo1, with the so-called Fenton sludge recycling system (FSRS), in which Fe(III)-sludge deposition is eliminated. Fenton’s reagent as a decoloration agent has been studied by many authors and it is suitable for different dye classes: acid, reactive, direct, metal-complex, disperse, and vat dyes, as well as pigments. Low decoloration rates were observed when C.I. Vat red (50%) and C.I. Disperse Blue (0.5%) were treated (Slokar and Majcen Le Marechal, 1997). Studies on the decoloration and mineralization of commercial reactive dyes using solar Fenton and photo-Fenton reaction indicated good color removal. The use of solar light was proved to be clearly
Table 19
Oxidation potential of common oxidizing agents
Oxidizing agents
Oxidation potential (V)
Fluorine (F2) Hydroxyl radical (OHd) Atomic oxygen Ozone (O3) Hydrogen peroxide (H2O2) Potassium permanganate (KMnO4) Hypochlorous acid (HCIO) Chlorine (Cl2) Bromine (Br2) Molecular oxygen (O2)
3.06 2.80 2.42 2.07 1.78 1.67 1.49 1.36 1.09 1.23
Table 20 Degradation of hydrogen peroxide into hydroxyl radicals activated by Fe2þ Fe2þ þ H2O2-Fe3þ þ OHd þ HO
698
Water in the Textile Industry
beneficial for the removal of color, aromatic compounds, and TOC (Garcı´a-Montan˜o et al., 2006; Torrades et al., 2004). Ozone. Once dissolved in water, ozone reacts with a great number of organic compounds in two different ways, namely direct molecular and indirect free radical-type reactions. The direct reactions are often highly solute selective, slow, and are dominant in acidic solutions. They are suitable for opening aromatic rings by means of ozone cycloaddition. The indirect hydroxyl radical reactions are nonselective, fast, proceed more rapidly with increasing pH, and constitute a significant portion of ozonation at basic pH. Indirect attacks are suitable for mineralization of TOC (Zhao et al., 2004). Although the original purpose of oxidation with ozone is disinfection of potable water, it can also be used for removing many toxic chemicals from wastewater to facilitate the decomposition of detergents, chlorinated hydrocarbons, phenols, pesticides, and aromatic hydrocarbons (Science Applications International Crop., 1987). The advantages of ozonation include
• • • • • •
decoloration and degradation occur in a single step, danger to humans is minimal, no sludge remains, all residual ozone can be decomposed easily into oxygen and water, little space is required, and ozonation is easily performed (Oguz et al., 2005).
The disadvantage is its very short half-life in water – ozone decomposes in about 20 min. The time can be significantly shortened if compounds such as dyes are present (Rice et al., 1986). Ozone stability is affected by the presence of salts, pH, and temperature. If alkaline salts are present, the solubility of ozone is reduced, while neutral salts may increase its solubility (Mallevialle, 1982). Under alkaline conditions, ozone decomposes more rapidly than under acidic conditions. With increasing temperature, ozone solubility decreases (Perkins et al., 1980). Studies of decoloration presented by several authors revealed that ozone decolorizes all dyes, except nonsoluble disperse and vat dyes which react slowly and take longer time (Namboodri et al., 1994; Marmagne and Coste, 1996; Liakou et al., 1997). Color removal strongly depends on dye concentration. Ozonation alone has low TOC and COD removal. Species such as oxalic, glyoxalic, and acetic acids cannot be completely mineralized by ozone alone at least at neutral or acidic pH (Hoigne and Bader, 1983). To enhance the efficiency of ozonation, a combination of various advanced oxidation processes has been developed, such as ozon/ UV, ozon/H2O2, and catalytic ozonation. Ozone–UV. Combination of ozone with UV results in a net enhancement of organic-matter degradation due to direct and indirect production of hydroxyl radicals upon ozone decomposition and H2O2 formation (Table 21). UV radiation decomposes ozone in water and generates highly reactive hydroxyl radicals. Hydroxyl radicals oxidize organics more rapidly than ozone itself. The efficiency of ozone/UV treatment depends on operating temperature (at higher temperature the ozone solubility is lower), pH (degradation favors neutral or slightly alkaline medium), and ozone-flow rate. For comparison of both ozonation and
Table 21 Direct and indirect production of hydroxyl radicals in O3– UV process Direct O3 þ hn-O2 þ O O þ H2O-OHd þ OHd O þ H2O-H2O2 H2O2 þ hn-OHd þ OHd Indirect O3 þ H2O þ hn-O2 þ H2O2 H2O2 þ hn-OHd þ OHd
Table 22
Reactions between O3 and H2O2
Initiation HO2 þ O3 -HO2 þ O3 H þ þ O3 # HO3 -OH þ O2 H2 O2 þ O3 -H2 O þ 2O2
kr ¼ 2.2 106 l mol1 s1 kr ¼ 1.1 105 l mol1 s1 kro102 l mol1 s1
Promotion OH þ O3 -O2 þ HO2 OH þ H2 O2 -H2 O þ HO2 OH þ HO2 -H2 O þ O2
kr ¼ 1.1 108 l mol1 s1 kr ¼ 2.7 107 l mol1 s1 kr ¼ 7.5 109 l mol1 s1
ozone/UV process, the degradation of eight commercial azo dyes in water (Shu and Huang, 1995a) and a model dyehouse wastewater (Perkowski and Kos, 2003) has been studied. In both studies, the ozone/UV process did not significantly enhance the degradation rates; the dye competed with ozone for UV absorbance. However, ozone/UV treatment, in terms of COD removal, is more effective compared to that by ozone (Bes-Pia et al., 2003). Ozone/H2O2. Addition of hydrogen peroxide to ozone enhances the production of hydroxyl radicals. The aqueous reactions between ozone and hydrogen peroxide are rather complex. The mechanisms and the kinetics of the production of hydroxyl radicals from ozone and hydrogen peroxide are known. The reactions and reaction rate constants are shown in Table 22. In the initiation sequence, reactive OHd radicals are generated. During the promotion reactions, the hydroxyl radicals are converted into the peroxy radical. At acidic pH, H2O2 reacts only very slowly with ozone, whereas at pH values greater than 5, a strong acceleration of ozone decomposition by hydrogen peroxide has been observed. The ozone decomposition rate increases with increasing pH. Decoloration with O3/H2O2 process is applicable for direct, metal-complex, or blue disperse dyes. There are some problems with the decoloration of acid and red disperse dyes, though, as well as with mixtures of direct, metal-complex, disperse, and reactive-dye decoloration. The efficiency of the decoloration with O3/H2O2 for a few of the dyes is presented in Table 23. H2O2/UV. In H2O2/UV processes, hydroxyl radicals are formed when water-containing H2O2 is exposed to UV wavelengths of 200–280 nm. The most commonly used UV source is low-pressure mercury vapor lamps with a 254-nm peak emission.
Water in the Textile Industry Table 23
Decoloration of dyes with O3/H2O2
Table 25
Textile dye
Decoloration (%)
Time (min)
Red 219 Blue 186 Direct Yellow 44 Direct Yellow 50 Red 23 Red 26 Direct Red 5B Direct Blue 1 Direct Blue 25 Direct Blue 71 Disperse Yellow 3 Disperse Yellow 64 Red 13 Red 60 Red 279 Blue 60 Palanil Blue 3RT Sulfo/disperse dye Reactive Yellow 37 Reactive Yellow 125 Reactive Yellow 125 Remazol Yellow RNL Reactive Red 35 Reactive Red 195 Blue 27 Blue 221 Green 13 Reactive dyes Vat dyes Azoic dyes
100 85 100 100 100 100 99 100 100 90 95 100 100 100 99 100 90 98 93 98 100 93 99 100 94 100 98 100 80 87
5 1 0.5 0.5 0.5 0.5 45 0.5 0.5 7 1 4.5 0.7 1 98 0.7 31 30 4 2.5 7 4 4.5 6 0.9 9 4 1 30 30
Adapted from Slokar YM and Majcen Le Marechal A (1997) Methods of decoloration of textile wastewaters. Dyes and Pigments 37(4): 335–356.
Table 24
The main reactions that occur during the H2O2/UV process
H2O2 þ hn-OHd þ OHd RH þ OHd-H2O þ Rd-further-oxidation
Problems such as sludge formation and regeneration, and increased pollution of wastewater caused by ozone, can be avoided by oxidation with hydrogen peroxide activated with UV light. The only chemical used in the treatment is H2O2, which, due to its final decomposition into oxygen, is not problematic. The most direct method for generation of hydroxyl radicals is through the cleavage of H2O2. Photolysis of H2O2 yields hydroxyl radicals by direct process with a yield of two radicals formed per photon absorbed at 254 nm. Hydroxyl radicals can oxidize organic compounds (RH)-producing organic radicals (Rd), which are highly reactive and can be further oxidized (Table 24) (Tuhkanen, 2004). The maximum absorbance of H2O2 is needed to generate sufficient hydroxyl radicals because of low absorption coefficient. However, high concentration of H2O2 scavenges the radicals, making the process less effective, while low concentration of hydrogen peroxide does not generate enough hydroxyl radicals to be consumed by the dye and this leads to
699
Reactions of H2O2 as a radical scavenger
OH þ H2 O2 -HO2 þ H2 O HO2 þ H2 O2 -HO2 þ H2 O þ O2 HO2 þ HO2 -H2 O2 þ O2
a slow rate of oxidation. Therefore, an optimum hydrogenperoxide dose needs to be verified experimentally (Table 25). The rate of dye removal is influenced by the intensity of UV radiation, pH, dye structure, and dyebath composition. In general, decoloration is most effective at neutral pH medium, at higher UV radiation intensity (1600 W rather than 800 W), with an optimal H2O2 concentration, which is different for different dye classes, and with a dyebath that does not contain oxidizing agents having an oxidizing potential higher than that of peroxide. According to Shu and Huang (1995b) acid dyes are the easiest to decompose, and with an increasing number of azo groups, the decoloration effectiveness decreases. Yellow and green reactive dyes need longer decoloration times, while other reactive dyes as well as direct, metal-complex, and disperse dyes are decolorized quickly. In the group of blue dyes examined, only blue vat dyes were not decolorized. For pigments, H2O2/UV treatment is not suitable, because they form a film-like coating on the UV lamp, which is difficult to remove. Several authors (Georgiou et al., 2002; Neamtu et al., 2002; Galindo and Kalt 1999; Colonna et al., 1999) reported complete decoloration of reactive and azo dyes in 30–90 min. The results indicated that H2O2/UV processes could be successfully used for the decoloration of acid, direct, basic, and reactive dyes but it proved to be inadequate for vat and disperse dyes (Yang et al., 1998). A comparative study between ozone and H2O2/UV was carried out on simulated reactive dyebath effluent containing a mixture of monochlorotriazinetype reactive dyes and various auxiliary chemicals. The H2O2/ UV process presented the decoloration rates close to those rates obtained with ozone but at a lower cost (Alaton et al., 2002). H2O2/UV systems may be set up in a batch or in a continuous column unit (Namboodri and Walsh, 1996). Decoloration of some dyes with H2O2/UV is presented in Table 26. Ultrasound. Sonolysis is a relatively innovative advanced oxidation process and was found to be a suitable method for the destruction of textile dyes. The ultrasonic irradiation of liquids generates cavitation (typically in the range 20–1000 kHz). Cavitation is a phenomenon of micro-bubble formation. Micro-bubbles grow during the compression/rarefaction cycles until they reach a critical size, and implode generating heat and highly reactive radical species. Inside the cavitation bubbles, the temperature and pressure rise to the order of 5000 K and 100 MPa, respectively. Under such conditions, water molecules degrade releasing hydroxyl radicals (OHd) and hydrogen radicals (Hd) as mentioned in Table 27. These radical species can either recombine or react with other gaseous molecules within the cavity, or in the surrounding liquid, after their migration. Pyrolitic and radical reactions inside, or near, the bubble and radical
700
Water in the Textile Industry
reactions in the solution are two major pathways for sonochemical degradation. Hydrophilic and nonvolatile compounds mainly degrade through hydroxyl-radical-mediated reactions in the solution, while hydrophobic and volatile
Table 26
Decoloration of dyes with H2O2/UV
Textile dye
Decoloration (%)
Time (min)
Reactive Yellow 37a Reactive Yellow 125a Remazol Yellow RNLa Reactive Red 35a Reactive Red 195b Reactive Black 5c Acid Yellow 17d Orange 10d Blue 21b Blue 27a Green 13a Vat Bluea Red 1d Red 14d Red 18d Blue 186e Black 1d Direct Yellow 4d Direct Blue 71a Palanil Blue 3RTa
85 96 85 100 100 100 98.2 100 95 100 93 15.5 99.9 100 99.1 80 89.9 83.2 98.5 96
8 8 8 8 60 4 40 60 150 5 8 10 30 60 40 10 60 60 3 10
a
Hoigne and Bader (1983). Liakou et al. (1997). c Ince and Go¨rnec (1997). d Bes-Pia et al. (2003). e Pittroff and Greorg (1992). b
Table 27
Radical formation and depletion during water sonolysis d
H2O-))) OH þ Hd OHd þ Hd-H2O 2OHd-H2O þ Od 2OHd-H2O2
Table 28
species degrade thermally inside or in the vicinity of the bubble. Reactive azo dyes are nonvolatile, water-soluble compounds and their passage into the gas cavity is unlikely. Hence, oxidative radical reactions in the bulk solution are expected to be the major route for their destruction. According to several studies, it is difficult to obtain the total mineralization (degradation to carbon dioxide, short-chain organic acid, oxalate, formate, and inorganic ions such as sulfate and nitrate) of the complex textile dyes with ultrasound alone. For this reason, the combination of ultrasound with other advanced oxidation processes is a more convenient approach in the remediation of such pollutants. Sonochemical degradation of textile dyes has become quite an interesting research area confirmed by several reports over the last few years (Vajnhandl and Majcen Le Marechal, 2005). In Table 28, a comparison of individual AOP is given.
4.20.3.1.3 Biological treatment processes Biological degradation or breakdown by living organisms is the most important removal process of organics, which are transferred from industry processes into solid and aquatic ecosystems. The application of microorganisms for the biodegradation of synthetic dyes is an attractive method and offers considerable advantages. The process is relatively inexpensive, the running costs are low, and the end products of complete mineralization are not toxic. An extensive review of large numbers of different species of microorganisms tested for decoloration and mineralization of different dyes has been published by Forgacs et al. (2004). The efficiency of biological-treatment systems is greatly influenced by the operational parameters. To produce the maximum rate of dye reduction, the level of aeration, temperature, pH, and redox potential of the system must be optimized. The concentration of the electron donor and the redox mediator must be balanced with the amount of biomass in the system and the quantity of the dye present in the wastewater. The compounds present (sulfur compounds and salts) in the wastewater may have an inhibitory effect on the
Technical comparison of oxidative decoloration
Oxidation process
Advantages
Disadvantages
Fenton
Effective decoloration of both soluble and insoluble dyes. Simple equipment and easy implementation. Reduction of COD (except with reactive dyes). No alternation in volume. Simple equipment and implementation. Reduction of COD (except with reactive dyes). Applied in gaseous state. No alteration of volume. No sludge production. Effective for azo dye removal. No sludge formation. No salt formation. Short reaction times. Very short reaction times for reactive dyes. No sludge formation. No salt formation. Short reaction times. Reduction of COD. Simplicity in use. Very effective in integrated system.
Sludge formation. Long reaction times. Salt formation. Hazardous waste. Prohibitively expensive.
FSR (Fenton sludge recycling system) Ozone Ozone/H2O2
H2O2/UV Ultrasound
Salt formation. Formation of gasses (H2, O2 during electrolysis). Short half-life (20 min). Not suitable for disperse dyes. Releases of aromatic amines. Not applicable for all types. Toxicity, hazard, problematic handling. No COD reduction. Additional load of water with ozone. Not applicable to all types of dyes. Requires separation of suspended solid particles. Relatively new method and awaiting full scale application.
Water in the Textile Industry
dye-reduction process. For these reasons, it is important to study the effect of these factors on decoloration before the biological system can be used to treat industrial wastewater (Pearce et al., 2003). Biodegradation processes may be anaerobic, aerobic, or involve a combination of both. Anaerobic biodegradation. Under anaerobic conditions, a low redox potential (o 50 mV) can be achieved, which is necessary for the effective decoloration of dyes. Color removal under anaerobic conditions is also referred to as dye reduction. Many bacteria under anaerobic conditions reduce the highly electrophilic azo bond in the dye molecules and produce colorless aromatic amines. The anaerobic decoloration of azo dyes was first investigated using intestinal anaerobic bacteria (Allan and Roxon, 1974; Brown, 1981; Chung et al., 1992). Later, it was found that azo dyes can also be decolorized with various other anaerobical cultures (Brown and Laboureur, 1983; Beydilli et al., 1998; Donlon et al., 1997). The efficacy of various anaerobic-treatment applications for the degradation of a wide variety of synthetic dyes has been demonstrated in several experiments. The exact mechanism of azo dye reduction is not clearly understood yet. There may be different mechanisms involved, such as enzymatic (Haug et al., 1991; Rafii et al., 1990), nonenzymatic (Gingell and Walker, 1971), mediated (Kudlich et al., 1997), intracellular (Mechsner and Wuhrmann, 1982; Wuhrmann et al., 1980), extracellular (Carliell et al., 1995), and various combinations of these mechanisms. A complete anaerobic mineralization of the azo dye azodisalicylate was observed under methanogenic conditions (Razo-Flores et al., 1997). The reduction of azo dye under anaerobic conditions strongly depends on the presence and disponibility of the cosubstrate. It acts as an electron donor for the azo dye reduction. The decoloration of reactive water-soluble azo dyes was achieved under anaerobic conditions using glucose as a co-substrate (Carliell et al., 1996). Anaerobic decoloration of reactive dyebath effluents with tapioca as a co-substrate also enhances color-removal efficiency (Chinwetkitvanich et al., 2000). The other suitable co-substrates were hydrolyzed starch, yeast extract, and a mixture of acetate, butyrate, and propionate. Much effort has been devoted to the study of the influence of various modern technologies on the decomposition rate of the dyes and the effect of the presence of the other compounds in the media. It has been recently established that the development of high rate systems, in which the hydraulic-retention times are decoupled from the solid-retention times, facilitates the removal of dyes from textile-processing wastewater (Rice et al., 1986). The effect of nitrate and sulfate salts used in textile dyeing on the microbial decoloration of a reactive azo dye has been studied. The results indicated that nitrate delays the onset of decoloration while sulfate did not influence the biodegradation process (Carliell et al., 1998). The reduction of azo dyes proceeds better under anaerobic thermophilic conditions than under mesophilic conditions, although the thermophilic process seems to be less stable compared to the mesophilic process (Willetts et al., 2000).
701
Carliell et al. (1994) studied the biodegradation of reactive dyes and they decolorized 80% of a range of tested dyes. From a detailed study of a selected dye, it was proposed that this occurred via a reduction mechanism. The results were supported by tentative chemical identification of the dyedegradation products. Hu (1994) isolated Pseudomonas luteola bacteria; after a 6-month adaptation in colored wastewater, he obtained microorganisms capable of reductive cleavage of the azo group in the dye. Decoloration with these microorganisms was complete within 4 days. Van der Zee et al. (2001) studied the decoloration of 20 selected azo dyes by granular sludge from an upward-flow anaerobic sludge-bed reactor and for all the azo dyes tested, complete reduction was achieved. Aromatic amines, due to azo dye reduction, are not commonly degraded under anaerobic conditions. Many aromatic amines were tested, but only a few were degraded. Some aromatic amines, substituted with hydroxyl or carboxyl group were degraded under methanogenic and sulfate-reducing conditions (Kalyuzhnyi et al., 2000; Kuhn and Suflita, 1989; Razo-Flores et al., 1999). Aerobic biodegradation. It is a process that often takes place in the environment, for example, in natural ecosystems such as soil or surface waters, and it is often associated with technical systems such as wastewater-treatment plants. Although for long, it was considered that azo dyes cannot readily metabolize under aerobic conditions, some specific aerobic bacterial cultures were found to be able to reduce the azo linkage via an enzymatic reaction. The aerobic conversions of sulfonated azo dyes were studied by Heiss et al. (1992) and Shaul et al. (1991), and sometimes even a complete mineralization of sulfonated azo dyes was found. In some studies, aerobic color removal of certain azo dyes was achieved, but all these stains required an additional energy and carbon source for growth. Since the supply of this additional substrate could have easily led to the formation of anaerobic microniches, the occurrence of anaerobic azo dye reduction certainly cannot be excluded (Govindaswami et al., 1993; Horitsu et al., 1977; Wong and Yuen, 1996; Zissi et al., 1997). The aerobic biodegradation of different aromatic amines (aniline (Lyons et al., 1984), carboxylated aromatic amines (Stolz et al., 1992), chlorinated aromatic amines (Loidl et al., 1990), benzidines (Baird et al., 1977), and sulfonated aromatic amines) has been extensively studied and many of these compounds were found to be degraded. Sulfonated aromatic amines are difficult to degrade. Combination of anaerobic/aerobic biodegradation. Although the anaerobic reduction of azo dyes is generally more satisfactory than aerobic degradation, carcinogenic aromatic amines, as products of anaerobic degradation, have to be degraded by an aerobic process. Diverse technologies for the successive anaerobic/aerobic treatment of textile wastewater have been developed. Anaerobic/aerobic conditions can be implemented by spatial separation of the two sludges using a sequential anaerobic/aerobic reactor system (Zitomer and Speece, 1993). These conditions can also be imposed on a single reactor in the so-called integrated anaerobic/aerobic reactor system (Field et al., 1995).
702
Water in the Textile Industry
4.20.3.2 Reuse New ecolabels for textile products and tighter restrictions on wastewater discharges are forcing textile wet processors to reuse process water and chemicals (Vandevivere et al., 1998). It is quite difficult to define a general quality standard for textilewater reuse because of the different requirements of each fiber (silk, cotton, polyester, etc.) of the textile process (e.g., scouring, desizing, dyeing, and washing) and because of the different quality required for the final fabric (Water Treatment Solutions, 2010). The actions aimed at the reduction of pollution and reuse of wastewater may normally be ranked in the following order according to their cost effectiveness: 1. prevention of pollution generation, 2. treatment of polluted streams close to the source of pollution (start-of-pipe approach), and 3. treatment of the final effluents (end-of-pipe approach). Pollution-prevention actions can normally succeed in all companies, while the start-of-pipe approach is mostly indicated in medium or big enterprises and the simplicity of the end-of-pipe approach makes it suitable in small as well as medium enterprises (Matioli et al., 2002).
4.20.3.2.1 Pollution-prevention techniques Pollution-prevention techniques have proved to be an effective means to improve process efficiency and to increase company profits, and at the same time, they minimize environmental impact. During the implementation of each of these techniques, the specific conditions must be carefully considered and every option and change must be examined, to understand how it could affect air, land, and water-pollutant releases (Matioli et al., 2002). Some of the pollution-prevention techniques that can be adopted are as follows: 1. Quality control for raw materials. Textile companies can reduce waste emissions by working with suppliers to find out less-polluting raw materials. Pre-screening raw materials is a useful practice to determine interactions among processes, substrates, and other chemicals with the aim to reduce waste production (Matioli et al., 2002). 2. Chemical substitution. Textile manufacturing is a chemically intensive process, and therefore a primary focus for pollution prevention should be on textile process chemicals. Opportunities for chemical substitution vary substantially among mills because of differences in: (US EPA/625/R-96/004, 1996) • environmental conditions, • process conditions, • product, and • raw materials. Possible actions are replacement of chemicals as desizing agents, dyes and auxiliaries with less-polluting ones, and replacement of chemical treatment in some processes with mechanical or other nonchemical treatment (Matioli et al., 2002). 3. Process modification. Optimization of the processes can be obtained by modifying some operations. Examples of possible modifications are (Matioli et al., 2002)
•
substitution of dyeing machines using low liquor ratio (equipment able to substantially reduce bath ratio and allow considerable savings of energy, water, dyes, and chemicals), • optimization of process conditions (temperature and time), and • combining operations to save energy and water (combining scouring and bleaching). 4. Equipment modification. An effective way to reduce waste is also by modifying, retrofitting, or replacing equipment and introducing automation (Matioli et al., 2002). 5. Good operating practices. A suitable way to prevent pollution without changing industrial processes is introduction of pollution-prevention procedures, including pollution-prevention objectives in research, new facility design, and ad hoc worker-training programs (Matioli et al., 2002).
4.20.3.2.2 Chemicals and water reuse and recycle: Start-of-pipe approach Recycling (reusing water and chemicals in the same process that produced the effluent) can save water, chemicals, and energy as well. An example is the reuse of exhausted hot dyebaths to dye further batches of material. In order to reuse the dyebath, it is necessary to determine the exact quantities of residual chemicals remaining in the dyebath. As a following step, to respect the characteristics demanded by the next dyeing cycle, the dyebath must be reconstituted by adding water, auxiliary chemicals, and dyestuffs (Matioli et al., 2002; EPA/310-R-97-00, 1997). Several examples of water reuse without treatment are based on the recovery of the water used in rinsing operation. Implementation of countercurrent washing (reusing the last contaminated water from the final wash for the next-to-last wash and so on) can significantly reduce the overall water consumption and is already applied in continuous textile operations. A systematic analysis of the water networks is required every time the overall use of water needs to be optimized and new options of water treatment and reuse have to be evaluated. Tools such as pinch analysis provide a formal procedure to determine near-optimal designs of energy and mass-transfer networks (Matioli et al., 2002; Majozi et al., 1998). When applied to water-use optimization, pinch analysis allows the identification of reuse, regeneration, and treatment opportunities. This approach normally generates start-of-pipe solutions implementing specific-process effluent treatment. The process-integrated wastewater treatment required by start-of-pipe solutions is based upon the possibility of efficient, reliable, cost-effective, and easy-to-operate treatment of single wastewater streams. These results can be obtained by proper applications of membrane technology (Matioli et al., 2002).
4.20.3.2.3 Process-water reuse and recycle: End-of-pipe approach In some cases, the classical end-of-pipe approach for reuse and recycling of industrial final effluents can also be efficient and cost effective. It fits very well in some typical European areas that can be defined as ‘textile districts’. A textile district refers to an area where many textile factories, mainly small and
Water in the Textile Industry
medium enterprises, are widespread and utilize the same water and wastewater facilities (Matioli et al., 2002). The end-of-pipe treatment was the first approach examined for cleaning up the total effluent flow in order to meet the standards for reuse. End-of-pipe treatment involves multistage-process combinations typically composed of biological and physicochemical techniques. Recently, the interest in membrane processes applied to textile-wastewater reuse is increasing, thanks to technological innovations that render them as reliable and feasible alternatives to other systems (Schoeberl et al., 2004). Membrane systems can successfully remove the large amount of suspended solids in wastewater (Chen et al., 2005). Centralized treatment plant for mixed industrial and municipal wastewater uses an aerobic biological stage. Some compounds are completely degraded, while others (dyes, surfactants, and their metabolites) are either absorbed on the sludge or discharged into the final effluent. Textile wastes contain poorly degradable organics (at least in aerobic conditions). Many contain toxicants, which are also often poorly biodegradable. Traditional aerobic biological process presents serious technical limitations for the purification of textile wastewaters (Matioli et al., 2002). The EU founded the Research and Technological Development (RTD) project Integrated Waste Recycling and Emission Abatement in the Textile Industry (EU, 1999) proposed several combined process modules to improve the actual wastewater-treatment plants, aimed at the reuse of final effluents: 1. A module for chemical precipitation of heavy metals and adsorption of dyes on anaerobic sludge (consisted in a pretreatment option) (Terras et al., 1999; O’Neill et al., 1999). 2. Enhancement of the biological treatment to a sensor-protected aerobic stage to remove biodegradable organics and to oxidize reduced nitrogen compounds while monitoring potential toxicity (Terras et al., 1999; Massone et al., 1998; Guwy et al., 1998). 3. The optimization of the final polishing involving various tertiary treatment lines to bring the water up to the standard required for use by the industries (Bergna et al., 1999; Bianchi et al., 1999; Rozzi et al., 1997, 2000). Posttreatment for mixed textile and domestic effluents has been successfully tested on the following unit process: ozonation, clariflocculation, multimedia filtration, granular activated carbon adsorption, ceramic crossflow and hollow-fiber microfiltration, nanofiltration, and low-pressure reverse osmosis. All the processes were investigated at medium and large pilot scale (Matioli et al., 2002). New advanced respirometric methodologies based on respirometry and titration may be used as wastewater-characterization techniques. They are particularly suited to evaluate the possible effects of a given wastewater on the final wastewater-treatment plant, due to their organic biodegradable and refractory load and inhibitory potential. The use of these characterization methods makes it possible to prevent treatment problems due to toxic discharges (Rozzi et al., 1999). The fee lever based on the treatability of the discharges can also be used to design the influent wastewater in a given
703
treatment plant, discouraging the discharge of refractory and inhibitory compounds. It can also lead to the introduction of cleaner technologies when an industry, billed with high fees for the presence of inhibitory compounds in its wastewater, is pushed toward the application of pollution-prevention techniques. The concept of waste design should not be limited to an offline procedure of characterization and of request to industries of qualitative or quantitative changes to their discharges. This concept should be extended to an online management system, based on a network of sensors, actuators, and facilities that can allow the plant manager to detect in the sewer (or before to discharge to it) the presence of excess hydraulic loading, organic or nutrient loading, or toxicants, and put in operation measures that can allow to maintain an optimal treatment result (Matioli et al., 2002; Bortone et al., 1997).
4.20.4 Conclusions Textile processing is one of the largest and oldest industries worldwide and it is responsible for substantial resource consumption and pollution. The wet processing, that is, pretreatment, dyeing, printing, and finishing, is especially polluting and resource consuming in terms of water, energy, and chemicals and like in most industries, freshwater is used in all processes with almost no exceptions. Textile industry has significant impact on the aquatic system, both by consuming a lot of water, freshwater sources, and also by discharging effluents into the environment. Water savings, reclamation, and reuse in industry are topics of increasing economic interest due to increasing water scarcity and costs. For this reason, research and development activities within this topic are increasing, methods and tools for analyzing water savings and reuse possibilities are being developed, and solutions are being implemented. The problem of water scarcity and the need for a rational water management has raised an interest in the use of recycled and reclaimed water as well as further water-loop closure. The typical textile SME today does not implement water reuse, while fresh high-quality water is used in all the production processes. Furthermore, the process effluents are mixed and discharged after onsite or centralized treatment in conventional wastewater-treatment plants. Despite the fact that during the last decades, new knowledge and technologies related to process-water production, wastewater treatment, and water-loop closure have been developed and implemented, current available technologies for textile wastewater treatment are often limited in efficiency and cost, and are not environmentally selective enough. On the other hand, it is not always clear which treatment lines are best suited to achieve the desired water quality at the lowest cost. Besides, textile companies are mainly SMEs and the small scale could represent a problem, because the water streams might have very different compositions. Over the last 30 years, drought and water scarcities have cost the European economy an estimated h100 bn. The most severe impacts of climate change that the world is facing are related to water. Climate change is intensifying the hydrological cycle. Risks from flood, drought, and coastal
704
Water in the Textile Industry
inundation, melting of glaciers and changes in the flow regimes of rivers are growing. Despite an understanding of the dangers to the economy, social stability, and the environment, not enough attention was given until recently to reduce the impacts of climate change on water and to increase adaptation efforts. In light of this, the vision of technological platforms (Water Supply and Sanitation Technology Platform (WSSTP), textile platform) and industrial associations (European Water Partnership (EWP) and European Apparel and Textile Organization (EURATEX)) is trying to follow some new ideas and approaches that would bring water to the forefront of a comprehensive strategy and promote adaptation measures across all water-related sectors.
References Alaton IA, Balcioglu IA, and Bahnemann DW (2002) Advanced oxidation of a reactive dyebath effluent: Comparison of O3, H2O2/UV-C and TiO2/UV-A processes. Water Research 36(5): 1143--1154. Al-Duri B, McKay G, El Geundi MS, and Wahab MZA (1990) Three-resistance transport model for dye adsorption onto bagasse pith. Journal of Environmental Engineering 116: 487--502. Allan R and Roxon JJ (1974) Metabolism by intestinal bacteria: The effect of bile salts on tartrazine azo reduction. Xenobiotica 4(10): 637--643. Anjaneyulu Y, Chary NS, and Raj DSS (2005) Decolourization of industrial effluents – available methods and emerging technologies – a review. Reviews in Environmental Science and Bio/Technology 4: 245--273. Bailey JE, Bohnet M, Brinker CJ, et al. (2000) Ullmann’s Encyclopedia of Industrial Chemistry, 40 volume set edition. 6th edn. Weinheim: Wiley-VCH. Baird R, Camona L, and Jenkins RL (1977) Behavior of benzidine and other aromatic amines in aerobic waste water treatment. Journal – Water Pollution Control Federation 49: 1609--1615. Bergna G, Bianchi R, and Malpei F (1999) GAC adsorption of ozonated secondary textile effluents for industrial water reuse. Water Science and Technology 40(4/5): 518--525. Bes-Pia A, Mendoza-Roca JA, Roig-Alcover L, Iborra-Clar A, Iborra-Clar MI, and AlcainaMiranda MI (2003) Comparison between nanofiltration and ozonation of biologically treated textile wastewater for its reuse in the industry. Desalination 157: 81--86. Beydilli MI, Pavlostathis SG, and Tincher WC (1998) Decolorization and toxicity screening of selected reactive azo dyes under methanogenic conditions. Water Science and Technology 38(4–5): 225--232. Bianchi R, Antonelli M, Malpei F, Mattioli D, and Rozzi A (1999) Wastewater recycling in the textile industrial distric of Como (Italy). In: Proceedings of the 18th IFATCC Congress 1999. Copenhagen, Denmark. Bisschops I and Spanjers H (2003) Literature rewiew on textile wastewater characterisation. Environmental Technology 24: 1399--1411. Bortone G, Gemelli S, Tilche A, Bianchi R, and Bergna G (1997) A new approach to the evaluation of the biological treatability of industrial wastewater for the implementation of the waste design concept. Water Science and Technology 36(2–3): 81--90. Brown D and Laboureur P (1983) The degradation of dyestuffs: Part I. Primary biodegradation under anaerobic conditions. Chemosphere 12(3): 397--404. Brown JP (1981) Reduction of polymeric azo and nitro dyes by intestinal bacteria. Applied and Environmental Microbiology 41(5): 1283--1286. Brown MA and DeVito SC (1993) Predicting azo dye toxicity. Critical Reviews in Environmental Science and Technology 23(3): 249--324. Carliell CM, Barclay SJ, and Buckley CA (1996) Treatment of exhausted reactive dyebath effluent using anaerobic digestion – laboratory and full-scale trials. Water SA 22: 225--233. Carliell CM, Barclay SJ, Nadidoo N, Buckley CA, Muholland DA, and Senior E (1994) Anaerobic decolourisation of reactive dyes in conventional sewage treatment processes. Water SA 20(4): 341--344. Carliell CM, Barclay SJ, Naidoo N, Buckley CA, Mulholland DA, and Senior E (1995) Microbial decolourisation of a reactive azo dye under anaerobic conditions. Water SA 21: 61--69. Carliell CM, Barclay SJ, Shaw C, Wheatley AD, and Buckley CA (1998) The effect of salts used in textile dyeing on microbial decolourisation of a reactive azo dye. Environmental Technology 19(11): 1133--1137.
Chen X, Shen Z, Zhu X, Fan Y, and Wang W (2005) Advanced treatment of textile wastewater for reuse using electrochemical oxidation and membrane filtration. Water SA 31(1): 127--132. Chinwetkitvanich S, Tuntoolvest M, and Pansawd T (2000) Anaerobic decolorization of reactive dyebath effluents by a two-stage UASB system with tapioca as a cosubstrate. Water Research 34(8): 2223--2232. Chung KT, Stevens SEJ, and Cerniglia CE (1992) The reduction of azo dyes by the intestinal microflora. Critical Reviews in Microbiology 18(3): 175--190. Colonna GM, Caronna T, and Marcandalli B (1999) Oxidative degradation of dyes by ultraviolet radiation in the presence of hydrogen peroxide. Dyes and Pigments 41(3): 211--220. Correia VM, Stephenson T, and Judd SJ (1994) Characterisation of textile wastewaters – a review. Environmental Technology 15: 917--929. Donlon BA, Razo-Flores E, Luijten M, Swarts H, Lettinga G, and Field JA (1997) Detoxification and partial mineralization of the azo dye mordant orange 1 in a continuous upflow anaerobic sludge-blanket reactor. Applied Microbiology and Biotechnology 47(1): 83--90. Easton J (1995) The dye marker’s view. In: Cooper P (ed.) Colour in Dyehouse Effluent, p. 11. Bradford: Society of Dyers and Colourist. El Geundi M (1991) Colour removal from textile effluents by adsorption techniques. Water Research 25: 271--273. EPA/310-R-97-00 (/310-R-97-00, 1997) Profile of the Textile Industry, Sector Notebook Project. Washington, DC: US Environmental Protection Agency. EPA Victoria (1998) Environmental Guidelines for the Textile Dyeing and Finishing Industry. Melbourne, VIC: EPA Publications. Estlander T (1988) Allergic dermatoses and respiratory diseases from reactive dye. Contact Dermatitis 18(5): 290--297. EU (1999) Integrated water recycling and emission abatement in the textile industry. Final Report for EC Contract ENV4-CT95-0064 (DG12-DTEE), Programme IV Environment and Climate, pp. 1994–1998. EURATEX (2002) Data submitted by EURATEX – Share of the EU-15 Textile and Clothing Industry Sector in the Manufacturing Industry (data 2000 source: EUROSTAT). EURATEX (European Apparel and Textile Organisation) (2000) Bulletin 2000/1. EURATEX E-D (2000) Textile Industry BREF Document (Chapter 2–6). Field JA, Stams AJM, Kato M, and Schraa G (1995) Enhanced biodegradation of aromatic pollutants in cocultures of anaerobic and aerobic bacterial consortia. Antonie van Leeuwenhoek 67: 47--77. Forgacs E, Cserhati T, and Oros G (2004) Removal of synthetic dyes from wastewaters: A review. Environment International 30: 953--971. Galindo C and Kalt A (1999) UV/H2O2 oxidation of azodyes in aqueous media: Evidence of a structure–degradability relationship. Dyes and Pigments 42(3): 199--207. GarcI´a-Montan˜o J, Torrades F, Garcı´a-Hortal JA, Dome`nech X, and Peral J (2006) Degradation of Procion Red H-E7B reactive dye by coupling a photo-Fenton system with a sequencing batch reactor. Journal of Hazardous Materials 134(1–3): 220--229. Georgiou D, Melidis P, Aivasidis A, and Gimouhopoulos K (2002) Degradation of azoreactive dyes by ultraviolet radiation in the presence of hydrogen peroxide. Dyes and Pigments 52(2): 69--78. Gingell R and Walker R (1971) Mechanisms of azo reduction by Streptococcus faecalis. II. The role of soluble flavins. Xenobiotica 1(3): 231--239. Gnann A, Gregor CH, and Schelle S (1993) Chemical Oxidative Process for Purifying Highly Contaminated Wastewater. WO Pat. to Peroxide-Chemie GmbH, Germany, 29 August 1993. Govindaswami M, Schmidt TM, White DC, and Loper JC (1993) Phylogenetic analysis of a bacterial aerobic degrader of azo dyes. Journal of Bacteriology 175(18): 6062--6066. Gruttner H, Winther-Nielsen M, Jorgensen L, Bogebbjerg P, and Sinkjear O (1994) Inhibition of the nitrification process in municipal wastewater treatment plants by industrial discharges. Water Science and Technology 29: 69--77. Guwy AJ, Buckland H, Hawkes FR, and Hawkes DL (1998) Active biomass in activated sludge: Comparison of respirometry with catalase activity measured using an online monitor. Water Research 32(12): 3705--3709. Hao OJ, Kim H, and Chiang P (2000) Decolorization of wastewater. Critical Reviews in Environmental Science and Technology 30: 449--502. Haug W, Schmidt A, Nortemann B, Hempel DC, Stolz A, and Knackmuss HJ (1991) Naphthalene 1,2-dioxygenase reactions. Applied and Environmental Microbiology 57(11): 3144--3149. Heiss GS, Gowan B, and Dabbs ER (1992) Cloning of DNA from a Rhodococcus strain conferring the sulfonated azo dyes. FEMS Microbiology Letters 99: 221--226.
Water in the Textile Industry
Hoigne J and Bader H (1983) Rate constants of reactions of ozone with organic and inorganic compounds in water–II: Dissociating organic compounds. Water Research 17(2): 185--194. Horitsu H, Takada M, Idaka E, Tomoyeda M, and Ogawa T (1977) Degradation of paminoazobenzene. European Journal of Applied Microbiology and Biotechnology 4: 217--224. Horning RH (1977) Characterization and treatment of textile dyeing wastewaters. Textile Chemist and Colourist 9: 24--27. Hu TL (1994) Decolourization of reactive azo dyes by transformation with Pseudomonas luteola. Bioresource Technology 49(1): 47--51. Ince NH and Go¨rnec DT (1997) Treatability of a textile azo dye by UV/H2O2. Environmental Technology 18: 179--185. Ja¨ger I, Hafner C, and Schneider K (2004) Mutagenicity of different textile dye products in Salmonella typhimurium and mouse lymphoma cells. Mutation Research 561: 35--44. Kalyuzhnyi S, Sklyar V, Mosoloava T, Kucherenko I, Russkova JA, and Degtyaryova N (2000) Methanogenic biodegradation of aromatic amines. Water Science and Technology 42: 363--370. Krizˇanec B and Majcen Le Marechal A (2006) Dioxins and dioxin-like persistent organic pollutants in textiles and chemicals in the textile sector. Croatica Chemica Acta 79: 177--186. Kudlich M, Keck A, Klein J, and Stolz A (1997) Localization of the enzyme system involves in anaerobic reduction of azo dyes by Sphingomonas sp. strain BN6 and effect of artificial redox mediators on the rate of azo dye reduction. Applied and Environmental Microbiology 63(9): 3691--3694. Kuhn EP and Suflita JM (1989) Anaerobic biodegradation of nitrogen substituted and sulfonated benzene aquifer contaminants. Hazardous Waste and Hazardous Materials 6(2): 121--134. Lee R (2000) Coagulation and flocculation in wastewater treatment. Water Wastewater 141: 29--32. Liakou S, Pavlou S, and Lyberatos G (1997) Ozonation of AZO dyes. Water Science and Technology 35(4): 279--286. Lockerbie M and Skelly JK (2003) Water quality requirements for treated effluent produced by recycling. Water Recycling in Textile Wet Processing, SDC 212–223. Loidl M, Hinteregger C, Ditzelmuller G, Ferschl A, and Streichsbier F (1990) Degradation of aniline and monochlorinated anilines by soil-borne Pseudomonas acidovorans strains. Archives of Microbiology 155(1): 56--61. Lyons CD, Katz S, and Bartha R (1984) Mechanisms and pathway of aniline elimination from aquatic environments. Applied and Environmental Microbiology 8(3): 491--496. Majcen Le Marechal A, Slokar YM, and Lobnik A (1996) Some aspects on dyes harmfulness and testing methods parameters. Tekstilec 39: 4--10. Majozi T, Brouckaert CJ, and Buckley CA (1998) Effluent reclamation and reuse in a group of agrochemical manufacturing plants using process integration. In: Proceedings of the 4th International Symposium on Waste Management Problems in Agro-Industries, pp. 433–438. Istambul, Turkey, 23–25 September 1998. Mallevialle J (1982) Ozonation Manual for Water and Wastewater Treatment. p. 53. New York: Wiley. Malmary G, Perineau F, Molinier J, and Gaset A (1985) Continuous process of dye removal from liquid effluents using carbonized wool waste. Journal of Chemical Technology and Biotechnology A 35: 431--437. Marmagne O and Coste C (1996) Color removal from textile plant effluents. American Dyestuff Reporter 85(4): 15--21. Massone A, Gernaey K, Rozzi A, and Verstraete W (1998) Measurement of ammonium concentration and nitrification rate by a new biosensor. WEF Research Journal 70(3): 343--350. Mathur N, Bhatnagar P, and Bakre P (2005) Assessing mutagenicity of textile dyes from Pali (Rajasthan) using Ames Bioassay. Applied Ecology and Environmental Research 4(1): 111--118. Matioli D, Malpei F, Bortone G, and Rozzi A (2002) Water minimisation and reuse in the textile industry. In: Lens P, Pol LH, Wilderer P, and Asano T (eds.) Water Recycling and Resource Recovery in Industry, Analysis, Technologies and Implementation, pp. 545--584. London: IWA Publishing. McKay G, Blair HS, and Gardner JR (1982) Adsorption of dye on chitin. I. Equilibrium studies. Journal of Applied Polymer Science 27: 3043--3057. McKay G, Ramprasad G, and Mowli P (1987) Desorption and regeneration of dye colours from low-cost materials. Water Research 21: 375--377. Mechsner K and Wuhrmann K (1982) Cell permeability as a rate limiting factor in the microbial reduction of sulfonated azo dyes. European Journal of Applied Microbiology and Biotechnology 15: 123--126. Mishra G and Tripathy M (1993) A critical review of the treatments for decolourization of textile effluent. Colourage 40: 35--38.
705
Namasivayam C, Kanachana N, and Yamuna RT (1993) Waste banana pith as adsorbent for the removal of rhodamine-B from aqueous solutions. Waste Manage 13(1): 89--95. Namboodri CG, Perkins WS, and Walsh WK (1994) Decolorizing dyes with chlorine and ozone. Part II. American Dyestuff Reporter 83(4): 17--26. Namboodri CG and Walsh WK (1996) Ultraviolet light/hydrogen peroxide for decolorizing spent reactive dyebath waste water. American Dyestuff Reporter 85(3): 15--25. Nassar NM and El-Geundi MS (1991) Comparative cost of colour removal from textile effluents using natural adsorbents. Journal of Chemical Technology and Biotechnology 50: 257--264. Neamtu M, Siminiceanu I, Yediler A, and Kettrup A (2002) Kinetics of decolorization and mineralization of reactive azo dyes in aqueous solution by the UV/H2O2 oxidation. Dyes and Pigments 53(2): 93--99. Nigam P, Armou G, Banat IM, Singh D, and Marchant R (2000) Physical removal of textile dyes from effluents and solid-state fermentation of dye-adsorbed agricultural residues. Bioresource Technology 72(3): 219--226. Oguz E, Keskinler B, and Celik Z (2005) Ozonation of aqueous Bomaplex Red CR-L dye in a semi-batch reactor. Dyes and Pigments 64(2): 101--108. O’Neill C, Hawkes FR, Esteves SR, Hawkes DL, and Wilcox SJ (1999) Anaerobic and aerobic treatment of a simulated textile effluent. Journal of Chemical Technology and Biotechnology 74(10): 993--999. Pagga U and Taeger K (1994) Development of a method for adsorption of dyestuffs on activated sludge. Water Research 28(5): 1051--1057. Pearce CI, Lloyd LR, and Guthrie JT (2003) The removal of colour from textile wastewater using whole bacterial cells: A review. Dyes and Pigments 58(3): 179--196. Perivoliotis MC (2002) Textile production within a co-operative applied to rural European women. Fibres and Textiles of Eastern Europe 10(3, 38): 8--13. Perkins WS, Judkins JF, and Perry WD (1980) Renovation of dye bath water by chlorination or ozonation. Textile Chemist and Cholorist 12(8): 27. Perkowski J and Kos L (2003) Decolouration of model dye house wastewater with advanced oxidation processes. Fibres and Textiles in Eastern Europe 11: 67--71. Pittroff M, Greorg KH (1992). Decolorization of textile wastewater by UV-radiation with hydrogen peroxide. Melliand International 6 (in English), translation of Melliand Textilberichte 73: 526. Platzek T, Lang C, Grohmann G, Gi U-S, and Baltes W (1999) Formation of a carcinogenic aromatic amine from an azo dye by human skin bacteria in vitro. Human and Experimental Toxicology 18: 552--559. Poots VJP, McKay G, and Healy JJ (1976a) The removal of acid dye from effluent using natural adsorbents – I. Peat. Water Research 10(12): 1061--1066. Poots VJP, McKay G, and Healy JJ (1976b) The removal of acid dye from effluent using natural adsorbents – II. Wood. Water Research 10: 1067--1070. Pratt M and Taraska V (2000) Disperse blue dyes 106 and 124 are common causes of textile dermatitis and should serve as screening allergens for this condition. American Journal of Contact Dermatitis 11(1): 30--41. Rafii F, Franklin W, and Cerniglia CE (1990) Azoreductase activity of anaerobic bacteria isolated from human intestinal microflora. Applied and Environmental Microbiology 56(7): 2146--2151. Raghavacharya C (1997) Colour removal from industrial effluents. Chemical Engineering World 32(7): 53--54. Razo-Flores E, Lettinga G, and Field JA (1999) Biotransformation and biodegradation of selected nitroaromatics under anaerobic conditions. Biotechnology Progress 15(3): 358--365. Razo-Flores E, Luijten M, Donlon BA, Lettinga G, and Field JA (1997) Complete biodegradation of the azo dye azodisalicylate under anaerobic conditions. Environmental Science and Technology 31(7): 2098--2103. Rice RG, Bollyky LJ, and Lacy WJ (1986) Analytical Aspects of Ozone Treatment of Water and Wastewater. p. 7. Chelsea, MI: Lewis. Robinson T, McMullan G, Marchant R, and Nigman P (2001) Remediation of dyes in textile effluent: A critical review on current treatment technologies with a proposed alternative. Bioresource Technology 77: 247--255. Rozzi A, Bianchi R, Liessens J, Lopez A, and Verstraete W (1997) Ozone, GAC and membrane treatment of secondary textile effluents for direct use. In: Colloquium: Treatment of Wastewaters from Textile Processing, pp. 25–47. Berlin, 17–18 November 1997. Rozzi A, Ficare E, Cellamare CM, and Bortone G (1999) Characterization of textile wastewater and other industrial wastewaters by respirometric and titration biosensors. Water Science and Technology 40(1): 161--168. Rozzi A, Malpei F, Bianchi R, and Mattioli D (2000) Pilot-scale membrane bioreactor and reverse osmosis studies for direct reuse of secondary textile effluents. Water Science and Technology 41(10–11): 189--195.
706
Water in the Textile Industry
Santl H, Gruber L, and Stoehrer E (1994) Some new sources of polychlorinated dibenzodioxins (PCDDs) and dibenzofurans (PCDFs) in waste papers and recycled pulps. Chemosphere 29: 1995--2003. Santos AB, Cervantes JF, and Lier JB (2007) Review paper on current technologies for decolourisation of textile wastewaters: Perspectives for anaerobic biotechnology. Biresource Technnology 98(12): 2369--2385. Schneider K, Hafner C, and Ja¨ger I (2004) Mutagenicity of textile dye products. Journal of Applied Toxicology 24(2): 83--91. Schoeberl P, Brik M, Braun R, and Fuchs W (2004) Treatment and recycling of textile wastewater – case study and development of a recycling concept. Desalination 171: 173--183. Science Applications International Crop (1987) Electrotechnologies for Waste and Water Treatment. p. 4. Palo Alto, CA: Electric Power Research Institute. Shaul GM, Holdsworth TJ, Demmpsey CR, and Dostal KA (1991) Fate of water soluble azo dyes in the activated sludge process. Chemosphere 22: 107--119. Shu HY and Huang CR (1995a) Degradation of commercial azo dyes in water using ozonation and UV enhanced ozonation process. Chemosphere 39: 3813--3825. Shu HY and Huang CR (1995b) Ultraviolet enhanced oxidation for colour removal of azo dye wastewater. American Dyestuff Reporter 84(8): 30--34. Slokar YM and Majcen Le Marechal A (1997) Methods of decoloration of textile wastewaters. Dyes and Pigments 37(4): 335--356. Stengg W (2001) The Textile and Clothing Industry in the EU a Survey. Enterprise Paper No. 2. Brussels: European Commission. Stenken-Richter I and Kermer WD (1992) Decolorising textile effluents. Journal of the Society of Dyers and Colourists 108: 182--186. Stolz A, Nortemann B, and Knackmuss HJ (1992) Bacterial metabolism of 5-aminosalicylic acid. Initial ring cleavage. Biochemical Journal 282(3): 675--680. Terras C, Vandevivere PC, and Verstraete W (1999) Optimal treatment and rational reuse of water in textile industry. Water Science and Technology 39(5): 81--88. Tomasino C (1992) Chemistry & Technology of Fabric Preparation & Finishing. Raleigh, NC: Department of Textile Engineering, Chemistry and Science College of Textiles North Carolina State University. Torrades F, Garcı´a-Montan˜o J, Garcı´a-Hortal JA, Nu´n˜ez L, Dome`nech X, and Peral J (2004) Decolorisation and mineralisation of homo- and hetero-bireactive dyes under Fenton and photo-Fenton conditions. Coloration Technology 120(4): 188--194. Tuhkanen TA (2004) UV/H2O2 processes. In: Parson S (ed.) Advanced Oxidation Processes for Water and Wastewater Treatment, pp. 86--92. London: IWA Publishing. UBA (1994) Reduction of Waste Water in the Textile Industry, Texte 3/94. UNEP (United Nations Environment Programme) (1994) The textile industry and the environment. Technical Report No. 16. Paris: UNEP/IE. US Environmental Protection Agency (2000) Exposure and Human Health Reassessment of 2,3,7,8-Tetrachlorodibenzo-p-Dioxin (TCDD) and Related Compounds, vol. 2. http://www.epa.gov/ncea/pdfs/dioxin/part1/volume2/chap8.pdf (accessed April 2010).
US EPA/625/R-96/004 (1996) Best Management Practices for Pollution Prevention in the Textile Industry, Manual. Cincinnati, OH: US Environmental Protection Agency. Vajnhandl S and Majcen Le Marechal AM (2005) Ultrasound in textile dyeing and the decolouration/mineralization of textile dyes. Dyes and Pigments 65: 89--101. Van der Zee FP, Lettinga G, and Field JA (2001) Azo dye decolourisation by anaerobic granular sludge. Chemossphere 44(5): 1169--1176. Waddell DS, Boneck-Ociesa H, Gobran T, Ho TF, and Botton JR (1995) PCDD/PCDF formation by UV photolysis of pentachlorophenol with and without the addition of hydrogen peroxide. Organohalogen Compounds 23: 407--412. Wadel DS, Bonek-Ociesa, and Gobrab T (1995) PCDD/PCDF formation by UV photolysis of pentachlorophenol with and without the addition of hydrogen peroxide. Organohalogen Compounds 23: 407--412. Water Treatment Solutions (2010) Water recycling in the textile industry. http:// www.lenntech.com/water_reuse_textile_industry.htm (accessed April 2010). Wenzel H and Knudsen HH (2005) Water savings and reuse in the textile industry. In: Omelchenko A, Pivovarov AA, and Swindall WJ (eds.) Modern Tools and Methods of Water Treatment for Improving Living Standards, vol. 48, pp. 169--189. Dordrecht: Springer. Wilkinson SM and McGechaen K (1996) Occupational allergic contact dermatitis from reactive dyes. Contact Dermatitis 35(6): 376--378. Willetts JR, Ashbolt NJ, Moosbrugger RE, and Aslam MR (2000) The use of a thermophilic anaerobic system for pretreatment of textile dye wastewater. Water Science and Technology 42(5–6): 309--316. Williams DT, LeBel GL, and Benoit FM (1992) Polychlorodibenzodioxins and polychlorodibenzofurans in dioxazine dyes and pigments. Chemosphere 24: 169--180. Wong PK and Yuen PY (1996) Decolourization and biodegradation of methyl red by Klebsiella pneumoniae RS-13. Water Research 30(7): 1736--1744. Wuhrmann K, Menscher K, and Kappeler T (1980) Investigation on rate-determining factors in the microbial reduction of azo dyes. European Journal of Applied Microbiology and Biotechnology 9: 325--338. Xu Y and Lebrun RE (1999) Treatment of textile dye plant effluent by nanofiltration membrane. Separation Science and Technology 34: 2501--2519. Yang YQ, Wyatt DT, and Bahorsky M (1998) Decolorization of dyes using UV/H2O2 photochemical oxidation. Textile Chemist and Colorist 30: 27--35. Zhao W, Shi H, and Wang D (2004) Ozonation of cationic red X-GRL in aqueous solution: Degradation and mechanism. Chemosphere 57: 1189--1199. Zissi U, Lyberatos G, and Pavlou S (1997) Biodegradation of p-aminoazobenzene by Bacillus subtilis under aerobic conditions. Journal of Industrial Microbiology and Biotechnology 19(1): 49--55. Zitomer DH and Speece RE (1993) Sequential environments for enhanced biotransformation of aqueous contaminants. Environmental Science and Technology 27(2): 227--244. Zollinger H (1987) Colour Chemistry-Synthesis, Properties of Organic Dyes and Pigments. pp. 92–100. New York: VCH Publishers. Zuskin E, Mustajbegovic J, Schachter EN, and Doko-Jelinic J (1998) Respiratory function of textile workers employed in dyeing cotton and wool fibers. American Journal of Industrial Medicine 31(3): 344--352.
4.21 Water Availability and Its Use in Agriculture D Molden, International Water Management Institute, Battaramulla, Sri Lanka M Vithanage, Institute of Fundamental Studies, Kandy, Sri Lanka C de Fraiture, International Water Management Institute, Accra, Ghana JM Faures, Food and Agriculture Organization of the United Nations (FAO), Rome, Italy L Gordon, Stockholm University, Stockholm, Sweden F Molle, Institut de Recherche pour le De´veloppement and International Water Management Institute, Colombo, Sri Lanka D Peden, International Livestock Research Institute (ILRI), Addis Ababa, Ethiopia & 2011 Elsevier B.V. All rights reserved.
4.21.1 Water Availability and Its Use in Agriculture 4.21.1.1 Sources of Water for Agriculture, Their Distribution, Use, and Possible Climate Change Effects 4.21.1.1.1 Green water 4.21.1.1.2 Agricultural water use in river basins 4.21.1.1.3 Open, closing, and closed river basins 4.21.1.1.4 Groundwater 4.21.1.1.5 Wetlands 4.21.1.1.6 Water consumption 4.21.1.1.7 Water use 4.21.1.1.8 Climate change, agriculture, and water 4.21.1.1.9 Drivers of water use 4.21.1.2 Physical and Economic Water Scarcity 4.21.1.3 Future Demands for Water 4.21.1.4 Future Scenarios for Rainfed and Irrigated Agriculture 4.21.2 Productive Use of Agricultural Water 4.21.2.1 Water Productivity in Agriculture 4.21.2.2 Rainfed Agriculture Productivity 4.21.2.3 Irrigated Agriculture and Productivity 4.21.2.4 Livestock 4.21.2.5 Aquaculture and Fisheries 4.21.3 Environmental and Health Implications of Agricultural Water Use 4.21.3.1 Impact on Rivers, Wetlands, and Biodiversity 4.21.3.1.1 Aquatic ecosystems 4.21.3.1.2 Terrestrial ecosystems 4.21.3.2 Health Impacts 4.21.3.3 Environmental and Health Mitigation 4.21.4 Water Governance 4.21.4.1 Definition 4.21.4.2 Types of Governance for River Basin Management 4.21.4.3 Basin Governance Challenges Acknowledgments References
4.21.1 Water Availability and Its Use in Agriculture With growing populations, shifting geographies, and changing dietary patterns, agriculture and food production face formidable challenges in the near future. Understanding issues of water availability and its use is fundamental for assessing and responding to these challenges. The following section examines the topics of water availability and use as they relate to agricultural production. While 3% of Earth’s total water volume is fresh (most of it is found in the form of ice in polar regions), only 1% is easily accessible for human use and is found in the physical forms of lakes, rivers, and shallow
707 707 707 708 709 709 710 711 711 711 711 712 713 714 716 716 717 718 719 720 721 721 721 723 723 724 725 725 725 727 728 728
aquifers (UN, 1997). Water for food and agricultural production is the largest use of this finite resource.
4.21.1.1 Sources of Water for Agriculture, Their Distribution, Use, and Possible Climate Change Effects Conventionally, agricultural water resources have been thought of in terms of surface water and groundwater. This approach, however, can be limiting. Besides considering surface- and groundwater, accounting for rainfall and soil moisture, as they factor into hydrological and agricultural systems,
707
708
Water Availability and Its Use in Agriculture
allows for a systematic understanding of water. The concepts of blue water and green water are useful in thinking about water availability in relation to a broader range of agricultural practices and a variety of users (CA, 2007). Blue water refers to water found in rivers, lakes, reservoirs, and aquifers. In addition to its use in agriculture, blue water is the measured and managed freshwater resource needed to meet domestic, commercial, and hydroelectric power demands while also functioning to sustain ecosystems (UN, 2006). Of total renewable blue water resources, 9% is used annually. Cities and industries extract 1 200 km3 of blue water per year but return more than 90% of it. This return is often of degraded quality and much of the flow returns to the sea, where it supports coastal ecosystems (Figure 1). Green water refers to
soil moisture available to plants generated by infiltrating rainfall. Green water is the main source of water for rainfed agriculture, whereas blue water is the main source for irrigated agriculture. Rainfed agriculture strictly depends on green water only, whereas irrigated agriculture uses blue water to supplement soil moisture. By adding blue water to crops, farmers can maintain soil moisture in dry periods and allow their crops to fulfill yield potentials. Through the process of evapotranspiration both green and blue water are ‘‘consumed’’ by vegetation and not returned to the system like in the case of other sections. The implications of green and blue water use are quite different. Increased evapotranspiration of blue water reduces stream flow and groundwater levels. Agricultural
Global water use Rainfall (thousands of cubic kilometers per year) 110 100%
Green water Bioenergy Forest products Grazing lands Biodiversity Landscape 56%
Blue water Rivers Wetlands Lakes Groundwater
Soil moisture from rain Crops Livestock Rainfed agriculture 4.5%
Water storage Aquatic biodiversity Fisheries
Crops Livestock Aquaculture Irrigated agriculture 0.6% 1.4%
Open water evaporation 1.3%
Green water
Cities and industries 0.1%
Blue water
Ocean 36%
Landscape Dam and reservoir
Landscape Irrigated agriculture
Wetlands
Rainfed agriculture
Cities
Figure 1 Global water uses. Source: Comprehensive assessment of water management in agriculture (2007), Water for Food, Water for Life (Earthscan, 2007). From Oki and Kanae (2006) Global hydrological cycles and world water resources. Science 313(5790): 1068–1072; UNESCO–UN World Water Assessment Programme (2006) Water: A Shared Responsibility, The United Nations World Water Development Report 2. New York: UNESCO and Berghahn Books.
Water Availability and Its Use in Agriculture
709
4.21.1.1.2 Agricultural water use in river basins
evapotranspiration is necessary for food production, and generally as food production increases, so does evapotranspiration. Increased evapotranspiration from green water sources is usually due to expansion of agricultural land area, a terrestrial impact, but has less impact on blue water flows. Still, any change in land use can affect river flows. In South Africa, recognition of the effects of stream flow-reducing activities has led to initiatives to control commercial forestry and to remove invasive tree species in order to reduce evapotranspiration and increase river flow (Hope, 2006).
The remaining 20% of crop evapotranspiration is from blue water drawn from surface- and groundwater sources. Blue water resources are, systematically, part of hydrological regions called river basins. River basins bounded by the area that catches water and directs it to common outlets. Basins and serve as important units of analysis because they connect various water in the basins uses. A change in use in one area often influences other uses of water. Efforts to control rivers go back many thousands of years, similarly, the practice of using these physical areas as regional units of organization for planning, developing, and managing water. More recently, in the latter half of the twentieth century, major dams were constructed which resulted in the multipurpose development and management of river basins. Hydroelectric power, flood control, water storage, and navigation became linked politically, economically, and ecologically in these river systems. Meanwhile, investment in irrigation accelerated rapidly in the 1960s and the 1970s, with irrigated area expansion in developing countries at 2.2% a year reaching 155 million hectares in 1982. During the same period, total global irrigated lands rose from 168 to 215 million hectares (Carruthers et al., 1997).
4.21.1.1.1 Green water Globally, about 80% of agricultural evapotranspiration comes directly from green water (Figure 2). This implies that the majority of the world’s agricultural production comes predominantly from rainfed lands despite major increases in large-scale irrigation infrastructure over the past half century. Some 55% of the world’s gross value of crop production is grown under rainfed agriculture on 72% of harvested land (Table 1). There are, however, large geographical differences in the percentages of rainfed and irrigated agricultural lands. For instance, over 95% of sub-Saharan Africa’s cultivated lands are strictly rainfed agriculture. Similarly, Latin America’s cultivated lands are 90% rainfed agriculture. In several countries of the Near East and North Africa, more than 40% of cultivated areas is irrigated. Meawhile investment in irrigation accelerated, in the 1960s and 1970s, from about 150 million hectares to a present total of over 270 million hectares (Faures et al., 2007). Hence, irrigated agriculture is relatively important in Asia and North Africa, while rainfed agriculture dominates in sub-Saharan Africa and Americas.
4.21.1.1.3 Open, closing, and closed river basins When a river basin can supply water to meet withdrawal demands and maintain its ecological functions, it is considerd an open basin (Seckler, 2006). A river basin is closing or closed when the volume of water use approaches or exceeds the volume of discharge. Often this is a problem of overcommitment, where water resources have been allocated beyond availability. As infrastructure develops around rivers,
More than half of production from rainfed areas
More than half of production from irrigated areas
More than 75% of production from rainfed areas
More than 75% of production from irrigated areas Global total: 7130 km3 (80% from green water, 20% from blue water) 780
220
650
235
1670 Blue water
Green water
905
1080
1480 110
Figure 2 Food crop evapotranspiration from rain and irrigation. Production refers to gross value of production. The pie charts show total crop water evapotranspiration in km3 by region. From International Water Management Institute analysis done for the comprehensive assessment for water management in agriculture using the Watersim model. Source: Comprehensive assessment of water management in agriculture (2007), Water for Food, Water for Life (Earthscan, 2007).
710
Water Availability and Its Use in Agriculture
Table 1
Global water and land statistics 3
Water (km ) Use
Land (millions of hectares) Statistics
Use
Total precipitation over continents 11 000 Vapor flow back to the Runoff to the oceans 40 000 atmosphere 70 000 Evapotranspiration Biomass consumed by grazing livestock Rainfed crops Irrigated crops Irrigation Rainfall Municipal use Industrial use Reservoirs
Total terrestrial land 13 000
Withdrawals
840 4910 2664 1570 650 53 88 208
Statistics
Grazing lands
3430
Rainfed harvested lands Irrigated cultivated lands
860 Harvested 340a
381 785
a
Of which 277 are equipped. From For water withdrawal statistics and equipped irrigation area, FAO (2006a); for evapotranspiration, International Water Management Institute analysis using the Watersim model; for harvested irrigated crop area, Chapter 3 on scenarios; for biomass consumed by grazing livestock, Stockholm Environment Institute calculations for the Comprehensive Assessment of Water Management in Agriculture; for municipal, industrial, and reservoir use, Shiklomanov (2000); for land statistics, FAO (2006b). Source: Comprehensive assessment of water management in agriculture (2007), Water for Food, Water for Life (Earthscan, 2007).
streams can become increasingly diverted, controlled, and utilized. In closed or closing basin scenarios, water users (especially those downstream) will not have full access to withdraw from the resource. Many rivers around the globe are closing or closed. Closing basins are sensitive to seasonal and inter annual variations of rainfall. Meanwhile, the ecological functions of a river will suffer, it is important for rivers to have adequate flow for fish and wildlife habitat, flushing sediments, diluting pollutants, preventing salinity intrusion, and sustaining estuarine and costal ecosystems. Increasing supply through inter-basin transfers is a common response to reopen closed basins, and desalinization to increase supply is a much discussed option (Falkenmark and Molden, 2008). Many closing basins are typically under stress for 1–6 months a year. China’s Yellow River dried up for the first time in 1972. In 1997, the dry-up lasted 226 days and reached 700 km upstream (Ren and Walker, 1998). The Colorado in the United States, the Indus flowing through India and Pakistan, the Murray-Darling in Australia, and most rivers in the Middle East and Central Asia are also severely overcommitted. Even basins in monsoon regions, such as Chao Phraya River in Thailand and the Cauvery River in India, experience months of closure, when salinity creeps inland as outflows of freshwater do not flush into the sea.
4.21.1.1.4 Groundwater The Earth’s fresh groundwater resources are estimated at approximately 10 000–12 000 km3, more than 200 times the volume of global annual rainfall. Only a tiny proportion, approximately 12 000 km3, of the total volume of groundwater reserves is recharged each year, compared to the large volume in stock (Doll and Fiedler, 2008). It is
estimated that, on average, 2091 m3 per capita are withdrawn from groundwater stores, with agriculture withdrawing the majority. About 2 billion people worldwide use groundwater, making it the single most utilized natural resource on the planet. The estimated annual use of groundwater is between 600 and 700 km3 (Struckmeier et al., 2005) and it keeps increasing. In the United States, for example, groundwater use in irrigation water has increased from 23% in 1950 to 42% in 2000 (Winter et al., 1998). This trend is reflected around the globe and particularly in Asia. There are many reasons why irrigation is a major user of groundwater. For farmers, the water is available when it is needed, is of reasonable quality, and very often can be abstracted without gaining permission or consulting with other users, a situation which is often much simpler than obtaining irrigation water from a canal system. Although agriculture is the largest user of groundwater, domestic dependence on groundwater use is increasing. Groundwater has historically supplied domestic water requirements in numerous urban and rural human settlements around the world. According to one estimate, more than half of the world’s population relies on groundwater for its drinking water supply (Coughanowr, 1994). In Spain, from 1960 to 2000, groundwater use increased from 2 to 6 km3 yr1 (Martinez-Cortina and Hernandez-Mora, 2003). In the Indian subcontinent, groundwater use soared from around 10–20 km3 yr1 before 1950 to 240–260 km3 yr1 by the year 2000 (Shah et al., 2003). In the United States, the volume of groundwater used as irrigation water increased from 23% in 1950 to 42% in 2000 (Winter et al., 1998). Chinese history records occasional cases of farmers lifting water from shallow wells by barrels to irrigate vegetables; however, North China had very little irrigation
Water Availability and Its Use in Agriculture
until 1950, and its tubewell irrigation revolution took off only after 1970. In total, then, the silent revolution in groundwater irrigation is essentially a story of the past 50 years (Llamas and Custodio, 2003). These can be considered as global pockets of intensive groundwater irrigation areas (Shah et al., 2007). Now there are pockets of intensive groundwater use, usually in food-producing areas of the world, such as the North China Plains, western and southern India, and parts of Mexico and the Ogallala aquifer of the USA.
4.21.1.1.5 Wetlands Wetlands act as sources of water for the majority of the global population. Agriculture’s impacts and dependencies upon wetlands are becoming increasingly significant. Wetlands are the key areas for managing extreme water flows after heavy rainfall and for providing water during droughts. Two recent global estimates have reported on the distribution of wetlands. Compiling national inventories, Finlayson et al. (1999) estimate global wetland area at 1280 million hectares. A more recent study by Lehner and Do¨ll (2004) used multiple geospatial data sets to estimate global wetland area at 917 million hectares. Accurate information on the distribution and extent of wetland ecosystems, both regionally and globally, is clearly an area requiring further work. Nevertheless, taking these data as the best-available estimates, a minimum of 131 million hectares of wetlands occur in Africa and 286 million hectares in Asia. The millennium ecosystem assessment (MEA, 2005a) identified agriculture as the major cause of wetland degradation and loss because it is the major economic activity in and around many wetlands, where crops such as rice, maize, and various vegetables and fruits are cultivated (Dries, 1989; Soerjani, 1992; Omari, 1993). However, agricultural development has considerably decreased the ecosystem services of wetlands (FAO, 2008). More recently, the comprehensive assessment of water management in agriculture (Falkenmark et al., 2007) concluded that pressures on wetlands would probably increase, with the prospect of serious loss of wetlands and ecosystem degradation. The needs of agriculture for flat, fertile land with a ready supply of water frequently make wetlands a valuable agricultural resource. In many arid and semi-arid regions of seasonal rainfall, where even major rivers can run dry for parts of the year, wetlands function to retain moisture. For this reason, they also make attractive resources for agriculture. Where people have to cope with both seasonal and interannual shortages of water, wetlands continue to be a vital resource for cultivators and pastoralists. In recent decades, agricultural use of wetlands has increased significantly in many developing countries, particularly in Africa, where they are perceived by some as the new frontier for agriculture (Wood and Dixon, 2009). This increase is driven partly by population growth, partly by the degradation of overexploited upland fields, and partly by market opportunities and the need to earn cash income (Wood and van Halsema, 2008). For poor rural households short of food, wetlands can offer good soils as well as water for irrigation, fisheries, and edible plants. In this way, wetlands can provide a safety net for poor households.
711
Some rural households increasingly use wetlands to supply local markets with irrigated vegetables and other products, which generate income. Seasonal wetlands also provide an important resource for livestock grazing. Sometimes these act as grazing land, but in some cases, they are used for hay production. This is prominent in many of African savannahs where the climate is semi-arid, rainfall is seasonal, and wetland grazing is widespread (FAO, 2008). For these households, wetlands represent a development opportunity that can lead them out of poverty.
4.21.1.1.6 Water consumption In an agricultural context, water consumption refers to water rendered immediately unusable by way of evaporation and transpiration from crops, soil, and open water bodies. Of total agricultural water consumption, the sources are estimated at 78% green water and 22% blue water. However, blue water withdrawal rates are greater than blue water consumption rates because not all water used for irrigation evaporates. The 22% of blue water consumed equals 1570 km3, whereas 2630 km3 are withdrawn for irrigation annually. In total, 60% of the water withdrawn for agriculture is consumed, while 40% returns to surface water or groundwater. The ratio of consumption to withdrawal is commonly referred to as the consumptive fraction or depleted fraction (Molden, 1996). Consumptive fractions tend to be low in water-abundant areas (where intensive water management is not cost effective), whereas they tend to be higher in water-scarce areas (where plants use shallow groundwater and farmers reuse drainage water). In the arid Middle East and North Africa, for example, the consumptive fraction is 77% with peak values close to 100%. In water-abundant areas, the consumptive fraction can be as low as 35%. Generally, it is not feasible or desirable to have a consumptive fraction higher than 70% at the basin scale, due to substantial infrastructure and environmental costs (Molden et al., 2000). However, in all scenarios the demand for freshwater increases to meet future food demands. Water consumption increases substantially in irrigated and rainfed areas).
4.21.1.1.7 Water use Annual global water withdrawals are estimated at 3830 km3, 70% of which is used for agriculture (i.e., 2664 km3) (FAO, 2006a). The net evapotranspiration from irrigation is 1570 km3 yr1, while the majority of total evapotranspiration is directly from rainfall. About 1000 km3 or 25–30% of the 3830 km3 of total water withdrawals originate from groundwater, and is mostly used for drinking and irrigation purposes. In the past century, industrial and municipal water demands, including those for energy generation, have grown in relative proportion to agricultural water demands. As competition between these sectors intensifies, agriculture can expect to receive decreasing shares of developed freshwater resources. Again, geographical differences are important to note. Approximately 70% of the world’s irrigated land is in Asia. Of Asia’s total cultivated land, however, only 34% is irrigated. Furthermore, China and India alone account for more than half of the irrigated land in Asia. By contrast, however, there is very little irrigation in sub-Saharan Africa.
712
Water Availability and Its Use in Agriculture
There has been relatively small investment in irrigation in Africa compared to massive investments in Asia, where irrigation fueled the green revolution. Demand for water for industrial and municipal uses, including for energy generation, is growing relative to demand for agriculture. As competition for water from these other sectors intensifies, agriculture can expect to receive a decreasing share of developed freshwater resources.
4.21.1.1.8 Climate change, agriculture, and water Predicted climate change scenarios show many potential risks to agriculture and agricultural water use:
• • • • •
increased precipitation intensity and variability could therefore the occurrence of flood and drought conditions as well as runoff patterns and the risk of crop failure; changing sea levels, water logging and after cause, could cause seawater intrusion and groundwater salinization in delta zones and other coastal areas; increasing temperatures will likely reduce crop productivity and increase water requirements in low latitude regions, thereby directly decreasing water-use efficiency; irrigation demands will likely increase; and water scarcity will increase in areas with growing populations and decreased precipitation.
At the watershed scale, changes in evaporation, precipitation, and water-storage cycles will alter the seasonal, annual, and interannual water availability for both terrestrial and aquatic agro-ecosystems (FAO, 2003). Several studies have also linked increased temperatures and evaporation, and decreased rainfall with greater needs for irrigation (Barnett et al., 2005; Bates et al., 2008; IPCC, 2001). Therefore, under these conditions, issues surrounding water demand and availability will increasingly affect agricultural activities, food security, forestry, and fisheries (Bates et al., 2008). In addition to these longterm climate issues, the severity of specific climate events will also influence agriculture around the globe. For example, more than 90% of simulations predict increased droughts in the subtropics by the end of the twenty-first century, while increased extremes in precipitation are projected in the major agricultural production areas of southern and eastern Asia, eastern Australia, and northern Europe (Bates et al., 2008).
4.21.1.1.9 Drivers of water use Population growth and changing diets are the two prominent drivers of increased food demand and, as it follows, increased water use (the following discussion is after Fraiture et al., 2007). From 6.1 billion people living on the planet in 2000, global population is projected to grow to 7.2 billion in 2015, 8.1 billion in 2030, and 8.9 billion in 2050 (UN, 2003). This growth curve is projected to level off after mid-century, except in sub-Saharan Africa where populations are projected to continue to grow. Furthermore, in regions where incomes increase, diets often change. In these scenarios, while the production of staple cereals goes up, greater numbers of people will also shift away from eating cereals as their primary food source and begin consuming greater quantities of livestock products, such as fish, and high-value crops. The world food supply increased from about 2400 kcal per person per day in
1970 to 2800 kcal per person per day in 2000, a 16.6% increase. However, geographical differences must be used in context here. In developed countries during the same period, food supply increased from 3050 to 3450 kcal per person per day, while in sub-Saharan Africa supply only increased from 2100 kcal per person per day to about 2200 kcal per person per day. The growth in per capita food consumption has been accompanied by significant changes in the commodities people choose to consume. Meat consumption has increased in all regions except sub-Saharan Africa, and industrial countries are by far the largest meat consumers, at 103 kg per person per year, a trend that is projected to continue for the next 50 years. The same patterns apply to dairy products as well. In total, wealthier populations consume more food per person and eat richer, more varied diets, while producing these foods means using more water. Increased urbanization and urban migration also drive food production and agricultural water demands. In the 1960s, two-thirds of the world’s population lived in rural areas, and 60% of the economically active population worked in agriculture. Today these ratios have changed. About half of the people alive today live in rural areas. Furthermore, a little more than 40% of the economically active population depends directly on agriculture as a means of well-being. In absolute terms, rural populations will begin declining in the next few years, and by 2050, two-thirds of the world’s people will live in cities and mega-cities. But, again, global averages do not express significant regional variations. In many poor countries in South Asia and sub-Saharan Africa, the rural population will continue to grow until about 2030, while the number of people depending on agriculture in these places will continue to rise (CA, 2007). Rapid rural-to-urban migration in developing countries also influences farming practices and water demand. In this process, more men are migrating to urban centers leaving women, older people, and children behind in rural areas. Consequently, in developing countries, women’s presence in agricultural economies is growing, rising from 39% in 1961 to 44% in 2004, whereas in developed countries these numbers are falling, dropping from 44% to 35% over the same period (FAO, 2006b). As this happens, issues of gender will be increasingly important to water management, productivity, equity, and governance. As cities expand in size, their demands on and claims to water resources increase, often at a loss to rural agricultural areas. In many cities today, poor or nonexistent urban planning and enforcement of land-use regulations compound water management problems. Urbanization also physically affects hydrological environments. Buildings, roads, and parking lots, among other human structures, create impermeable surfaces while sewage systems redirect large volumes of water. As a result, surface areas available for water infiltration are decreased, and increased runoff can become a significant problem. In river basins affected by urban footprints, peak discharge occurs quicker and reaches higher volumes. This can result in greater stream channel erosion, possible channel destruction, and habitat degradation, while it can also damage human life and property. Furthermore, these surfaces reduce groundwater recharge and can decrease long-term groundwater inflow to streams. Urban centers are cites of water
Water Availability and Its Use in Agriculture
pollution. The increased presence of sediments, nutrients, microbes, toxic metals, and organics is a major externality of urbanization. All of these factors can have significant effects on human health, downstream environments, and agricultural systems. Hydroelectric power generation is another significant driver influencing water availability and agricultural production. Dams worldwide produce 715 000 MW or 19% of the world’s electricity. The process of hydroelectric power generation requires water storage and stream flow regulation, both of which can influence water availability for agriculture and other users. Thus, significant volume of the world’s blue water resources is held in river basins where multipurpose water management is linked to energy production. Many factors relate to the nature of hydroelectric production condition the way these water resources are managed. Dams have the ability to store and release water at specific times and this means electricity can be generated on demand. In the same way, dams also have the ability to regulate water for irrigation, navigation, and recreation. As dams are not sources of CO2 emissions, provided vegetation is cleared before following up hydroelectric power may be an attractive energy option in the future. However, other issues, including habitat degradation and cost efficiency, are at stake and need to be considered in analyzing trade-offs. At a different scale, smallscale hydro or micro-hydro power has been increasingly popular, especially in remote areas where other power sources are not feasible. Small-scale hydropower systems are installed in small rivers or streams with little or no marked environmental effects. In poor areas, many remote communities have no electricity. Micro-hydro power, with a capacity of 100 kW or less, allows communities to generate electricity. Changing consumption patterns, more people moving into urban areas, and increasing demand for low-carbon energy mean that agricultural water use will see greater outside competitive pressures. Small holders and individual water users, with little political power in water governance processes, will face greater risks in these conditions. The following section examines concepts of water scarcity as a means of understanding how these risks affect people differently. Water scarcity is often viewed as a physical issue, where aggregated demand for water by all potential uses is lays than the available supply (FAO, in preparation). Such approach, however does not capture the knowledge.
4.21.1.2 Physical and Economic Water Scarcity Another tool for examining water availability is the concept of water scarcity (Seckler et al., 2000; Molden et al., 2007). Rather than analyzing water availability from a hydrological approach (using river basins as units of analysis), water scarcity focuses on social and political regions (using populations as units of analysis). Evaluating water scarcity begins at a micro-level, one can the water security of individuals. Individuals are water secure when they have consistent access to safe and affordable water to satisfy their needs for drinking, washing, food productions, and other livelihood endeavors; they are water insecure when these needs cannot be met. A region is water scarce when a large number of people
713
are water insecure (Rijsberman, 2006). In adapting such approaches, economic, financial, social political and institutional dimensions of the problem of access to water become as relevant as the physical availability of water. These multiple dimensions of the problem have been captured in the dual concept of physical and economic water scarcity. Access to water is difficult for millions of people for social, political, and economic reasons, in addition to physical resource constraints. About 2.8 billion people live in areas facing water scarcity, and more than 1.2 billion of them – onefifth of the world’s population – live in areas of physical water scarcity (Molden et al., 2007 – trends chapter). Another 1.6 billion people live in basins that face economic water scarcity, where human and institutional capacity or financial resources are likely to be insufficient to develop adequate water resources even though adequate water is available to meet human needs (Figure 3). Within these regions, poor people suffer disproportionately from the implications of scarcity. Lack of finance, lack of human capacity, poor management, and a lack of good governance all contribute to water scarcity. Physical water scarcity occurs when available water resources are insufficient to meet all demands, including minimum environmental flow requirements (Figure 3). Arid regions are most often associated with physical water scarcity; however, an alarming new trend of artificial physical water scarcity is affecting even regions where water is abundant. This problem is due to the over-allocation and over-development of water resources, leaving no scope for making water available to meet new demands except through interbasin transfers. In these scenarios, there is not enough water to meet both human demands and environmental flow needs. The implications of physical water scarcity include severe environmental degradation, such as river desiccation and pollution, declining groundwater tables, water allocation disputes, and failure to meet the needs of individuals and groups. Some 1.2 billion people live in river basins where human water use has surpassed sustainable limits. Meanwhile, another 500 million people live in river basins that are fast approaching physical water scarcity. While physical scarcity introduces complex problems, investments in good management can mitigate many of the issues. Economic water scarcity occurs when the investments needed to keep up with growing water demand are constrained by limited financial, human, or institutional capacities. Much of the scarcity felt by people is due to problems with institutions and politics, favoring one group over another, not listening to the voices of women and disadvantaged groups, for instance. Problems of economic water scarcity include: inadequate infrastructure development, where people have trouble getting enough water for agriculture and domestic purposes; high vulnerability to seasonal water fluctuations, including floods and long- and short-term droughts; and inequitable distribution of water even though infrastructure exists. Much of sub-Saharan Africa experiences economic water scarcity, and there are many areas across the globe where water resources are inequitably distributed. Further water development could ease problems of poverty and inequality.
714
Water Availability and Its Use in Agriculture
Little or no water scarcity
Approaching physical water scarcity
Physical water scarcity
Economic water scarcity
Not estimated
Figure 3 Areas of physical and economic water scarcity. Definitions and indicators: (1) Little or no water scarcity. Abundant water resources relative to use, with less than 25% of water from rivers withdrawn for human purposes. (2) Physical water scarcity (water resources development is approaching or has exceeded sustainable limits). More than 75% of river flows are withdrawn for agriculture, industry, and domestic purposes (accounting for recycling return flows). This definition – relating water availability to water demand – implies that dry areas are not necessarily water scarce. (3) Approaching physical water scarcity. More than 60% of river flows are withdrawn. These basins will experience physical water scarcity in the near future. (4) Economic water scarcity (human, institutional, and financial capital limit access to water even though water in nature is available locally to meet human demands). Water resources are abundant relative to water use, with less than 25% of water from rivers withdrawn for human purposes, but malnutrition exists. From International Water Management Institute analysis done for the comprehensive assessment for water management in agriculture using the Watersim model. Source: Comprehensive assessment of water management in agriculture (2007), Water for Food, Water for Life (Earthscan, 2007).
Table 2
Withdrawals by nonagricultural sector will increase by a factor of 2.2 by 2050
Region
Sub-Saharan Africa East Asia South Asia Central Asia and Eastern Europe Latin America Middle East and North America OECD countries World
Agriculture
Domestic
Manufacturing
Thermo cooling
Total nonagricultural
2000
2000 2050
2000
2050
2000
2050
2000
2050
68 518 1095 244 175 173 233 2630
7 48 15 40 31 14 121 278
2 21 4 68 12 3 135 245
8 159 29 236 42 10 131 617
1 32 15 48 10 7 262 376
18 75 55 52 134 22 307 664
10 101 34 156 53 24 518 902
60 419 175 377 254 82 590 1963
35 185 90 88 78 51 152 681
Annual increase (%) 2000–50
3.7 2.9 3.3 1.8 3.2 2.5 0.3 1.6
Note: Units are in km3 unless otherwise indicated. From Comprehensive Assessment for Water Management in Agriculture (2007). Source: Comprehensive assessment of water management in agriculture (2007), Water for Food, Water for Life (Earthscan, 2007).
4.21.1.3 Future Demands for Water If improvements in land and water productivity or major shifts in production patterns do not occur in the near future, crop water consumption would increase 70–90% by 2050 depending upon actual population growth, changing income levels, and water requirements for livestock and fisheries. In this scenario, crop water consumption would go from a current rate of 7130 km3 yr1 to somewhere in the range of 12 050–13 500 km3 yr1. This estimated range accounts for crop water depletion for food and feed production,
plus losses through evaporation from soil and open water sites. Nevertheless, even with improvements in water productivity, agriculture will continue to consume a large portion of the world’s developed water supply. This topic is discussed further in the following section. Industrial and domestic demand for water will continue to increase with urbanization. Withdrawals for nonagricultural sectors are expected to more than double by 2050, and, as it follows, there will be increasing competition for water between sectors (Table 2). In most countries, urban water demands receive customary or legal priority over water for
Water Availability and Its Use in Agriculture
715
10
8 Projection (high)
mt ha–1
6
OECD countries FAO
IWMI Projection (low)
4 World IWMI
2 FAO
0
sub-Saharan Africa
1961
1970
1980
1990
2000
2010
2020
2030
2040
2050
Figure 4 Global water withdrawals increase. Points marked FAO (Food and Agriculture Organization) are based on projections in Bruinsma (2003); those marked IWMI (International Water Management Institute) are based on projections in Seckler and others (2000). From FAOSTAT (2006), for 1960–2003; International Water Management Institute analysis done for the comprehensive assessment for water management in agriculture using the Watersim model, for 2000–50. Source: Comprehensive assessment of water management in agriculture (2007), Water for Food, Water for Life (Earthscan, 2007).
agriculture (Molle and Berkoff, 2006). Greater competition for water will leave less for agriculture, particularly near large cities in water-short areas. The regions of the Middle East, North Africa, Central Asia, India, Pakistan, Mexico, and northern China, among other areas, will see greater competition for water as urban centers continue to develop there. Estimates also show that while the proportion of water diverted for nonagricultural sectors increases, agriculture remains the largest water user among the productive sectors. Although major trade-offs will occur between all water-using sectors, they will be particularly pronounced between agriculture and the environment as the two largest water-demanding sectors (Figure 4) (Rijsberman and Molden, 2001). Unlike agricultural water consumption, only a small part of the water diverted for domestic and industrial purposes is consumed. In urban areas, 75–85% of water diverted flows back to rivers, lakes, and groundwater as return flow. In many urban areas, particularly in water-scarce developing countries, wastewater is used for high-value vegetable production, a livelihood activity for millions of city dwellers (Gupta and Gangopadhyay, 2006; Hussain et al., 2001, 2002; Raschid-Sally et al., 2005). The use of urban wastewater for irrigation will increase as water becomes scarcer in urbanizing areas. If by 2050 half of return flows from cities are reused, 200 km3 of wastewater could be used for irrigation. This would represent only 6–8% of future agricultural withdrawals, but the economic values generated could be substantial. Much of the wastewater would likely be used to produce highly valued vegetables, helping sustain the livelihoods of millions of small farmers (Hussain et al., 2001, 2002). As reuse of city wastewater for agriculture poses environmental and health risks, these can be minimized with proper management. Water demand for managing ecological functions has also created greater resource competition, as reflected in changing
policies for water allocation and pricing. In many countries, rising incomes are correlated with increasing demands for restoring and maintaining environmental services. The demand for environmental amenities adds pressure on scarce water resources. A first-cut estimate by Smakhtin et al. (2004) indicates that 20–45% of long-term annual flows must be preserved to maintain essential ecosystem services. UNESCO (2006) suggests that 100 km3 need to be added to estimates of future water demands to account for current overexploitation of groundwater and 30 km3 must be added to account for the mining of nonrenewable groundwater, or fossil groundwater.
4.21.1.4 Future Scenarios for Rainfed and Irrigated Agriculture As outlined above, there will be greater demands on agriculture and water by the year 2050. Considering agricultural productivity, in the context of different irrigation scenarios, is an effective way to understand how these demands can be met. The comprehensive assessment of water management in agriculture (CA, 2007) provided scenarios to allow us to explore various futures. Most importantly, if irrigation development were to remain static from now until 2050, the agricultural potential of rainfed agriculture would be sufficient for meeting the projected additional food requirements in 2050 such as cultural values. In an optimistic yield growth scenario, in which cereal yields grow by 72%, the demand for agricultural commodities is met by increasing rainfed-harvested area by 7% (this work follow Fraiture et al., 2007). The contribution of rainfed agriculture to the total gross value of food supply would increase from 52% in 2000 to 60% in 2050 (CA, 2007). In the optimistic yield scenario, sub-Saharan Africa, Asia, and Latin America can be largely self-sufficient in producing major food
716
Water Availability and Its Use in Agriculture
crops. East Asia, however, would need to import maize to meet the large increase in feed demand. In addition, the Middle East and North Africa would need to import food due to lack of suitable lands for rainfed production. Global food trade would increase from 14% to 17% of total production. The scenario analysis also demonstrates the risks inherent in a rainfed-based strategy. In the pessimistic scenario, with a low rate of adoption of water harvesting and only modest improvements in rainfed yields, the area of rainfed production must increase by 53% to meet future food demands (an additional 400 million hectares as compared with the optimistic yield scenario; see Figure 5). The Food and Agriculture Organization of the United Nations (FAO) estimates suggest ample capacity for increasing the area under cultivation, except in South Asia, the Middle East, and North Africa. In sub-Saharan Africa and Latin America, only one-fifth of the potential land area is already in use. Although there are significant amounts of land available for cultivation, more than half are now forested or protected areas (Alexandratos, 2005). Furthermore, some of these lands might be of marginal quality (Bruinsma, 2003) or not suitable for cereal crops. In the pessimistic yield scenario, countries without potential to expand rainfed areas – due to either
lack of suitable land or unreliable rainfall – must increase food importation. In this case, the Middle East and North Africa would import more than two-thirds of their agricultural needs. South and East Asia, due to land limitations, would become major importers of maize and other grains, importing 30–50% of their domestic needs. Latin America, developed countries, Central Asia, and Eastern Europe, having the potential to expand land in agriculture, would increase their exports. Globally, food trade would increase from 14% of total agricultural production to 22% in 2050. Large grain imports by East and South Asia would put upward pressure on food prices (the model results suggest an increase of 11%). There is a risk that poor countries may not be able to afford food imports, and household-level food insecurity and inequity might worsen. Climate change, which is expected to increase the variability and intensity of weather events, exacerbates the risks of rainfed production, particularly in semi-arid areas vulnerable to drought (Kurukulasuriya et al., 2006). Both the optimistic and pessimistic rainfed scenarios lead to substantial increases in soil water consumption. While the global average of rainfed cereal yield would improve by 72%, crop water productivity would improve by 35%. In the pessimistic yield scenario,
Evapotranspiration by rainfall
Evapotranspiration by irrigation
Without productivity improvement (worst case)
Difference (pessimistic–optimistic)
Irrigation withdrawals
Crop evapotranspiration and irrgation withdrawals Today Rainfed scenario Irrigation scenario Trade scenario Comprehensive assessment scenario Without productivity improvement 0
2000
4000
6000
8000
10 000
12 000
14 000
km3 Irrigation area
Rainfed area
Difference (pessimistic–optimistic)
Without productivity improvement (worst case) Harvested area Today Rainfed scenario Irrigation scenario Trade scenario Comprehensive assessment scenario Without productivity improvement 0
500
1000
1500
2000
2500
Millions of hectares Figure 5 The optimistic and pessimistic scenarios. The comprehensive assessment scenario combines elements of the other approaches. The purple segments of the bars show the difference between optimistic and pessimistic assumptions for the two rainfed and two irrigated scenarios. The brown bar shows the worst-case scenario of no improvement in productivity. From International Water Management Institute analysis done for the comprehensive assessment for water management in agriculture using the Watersim model. Source: Comprehensive assessment of water management in agriculture (2007), Water for Food, Water for Life (Earthscan, 2007).
Water Availability and Its Use in Agriculture
global rainfed cereal yields improve by 20% and water productivity by 10%, while total crop water consumption increases by 54% to 10 980 km3, an additional 3850 km3 after the year 2000. Increases agricultural evapotraspiration of that order of magnitude will have impacts on river flows and groundwater recharge, with implications for downstream water users and those relying on groundwater resources.
4.21.2 Productive Use of Agricultural Water The long-term sustainability of food production and agriculture depends on the efficient management of limited water resources. Moreover, as these resources come into greater demand, driven by a broadening range of applications and functions, the implications of agricultural water use will have increasing effect on other water users and the environment. In other words, the impact of agricultural water use is increasing. By analyzing agricultural water productivity, crop production can be assessed in terms of its social, economic, and ecological costs and benefits. The following section outlines several approaches for assessing water productivity and contextualizes the topic in relation to rainfed and irrigated agricultural systems. From here, future productivity scenarios are considered, and livestock and fisheries agriculture are discussed (this section draws from analysis presented in de Fraiture et al., 2007).
4.21.2.1 Water Productivity in Agriculture Water productivity is defined as the ratio of the net benefits from crop, forestry, fishery, livestock, and mixed agricultural systems to the amount of water required to produce these benefits (this section draws from Molden et al., 2007). In its broadest sense, water productivity reflects the objectives of producing more food, income, livelihoods, and ecological benefits at less social and environmental cost per unit of water used. Water productivity can be defined in several ways. Physical water productivity is the ratio of the mass of agricultural output to the amount of water used, while economic
water productivity is defined as the value derived per unit of water used (Figure 6). Other modes of analysis include crop water productivity, where specific crops are measured individually for comparative purposes, and livestock water productivity where the ratio of the net beneficial livestock-related products and services is calculated in relation to the volume of water depleted in production including the water to feed them (Peden et al., 2007). In areas of the world already exhibiting high physical water productivity, the scale for improvement is limited. Many rainfed, irrigated, livestock, and fisheries systems across the globe, however, do not exhibit high physical water productivity. Many farmers in developing countries could raise their water productivity by adopting better management practices. These include supplemental irrigation; soil fertility maintenance; deficit irrigation; small-scale water storage, delivery, and application; modern irrigation technologies (such as pressurized systems and drip irrigation); and soil water conservation through mulching zero or minimum tillage. Breeding technologies and biotechnology can also indirectly help agricultural systems become more efficient by reducing biomass losses through increased resistance to pests and diseases, enhancing the vigorous early growth for fast ground cover to reduce soil surface evaporation, and by reducing drought susceptibility for specific crops. However, water productivity gains are context dependent, and, in some cases, a gain for one group of people can mean a loss for the others. Water productivity, especially in physically water-scarce basins, can be properly assessed only by taking an integrated basin perspective where trade-offs between uses are considered. Employment opportunities, income generation, nutrition, and opportunities for women can all be linked to agricultural productivity; in this way, increasing values derived per unit of water is important to poverty reduction. However, carefully crafted programs are required to ensure that these gains reach the poor, especially rural women, and are not captured exclusively by wealthier or more powerful users. As described above, rising demand for livestock and fish products also leads to rising demand for water. In producing
Purely rainfed
Field conservation practices
717
Fully irrigated
Supplemental irrigation
Water harvesting
Groundwater irrigation
Surface-water irrigation
Drainage
Figure 6 Agricultural water management: a continuum of practices. Source: Comprehensive assessment of water management in agriculture (2007), Water for Food, Water for Life (Earthscan, 2007).
718
Water Availability and Its Use in Agriculture
these animal products, water productivity gains can be made by carefully considering feed sources and feeding strategies animal species produced (chicken use less water than cattle), improving the quality of produce, and integrating fisheries and livestock into farm production systems. Because freshwater fisheries are increasingly threatened by reductions in stream flows, basin water productivity analysis should consider the social and ecological values generated by fisheries before reducing river flows that support them. Several studies describe multiple uses of agricultural water and the ways in which these uses can improve productivity. Poor rural households use agricultural water in multiple ways (Laamrani et al., 2000; Moriarty et al., 2004; Jehangir et al., 2000). Agricultural water can be used for drinking, sanitation, home gardens, livestock, rural industries, and aquaculture. The increase of agricultural water productivity can result in many benefits such as:
• • •
helping to meet food demand; contributing to poverty reduction and economic growth; and helping to reduce pressures to reallocate water from agriculture and to ensure that water is available for environmental uses.
Integrated and multiple-use systems – in which water serves crops, fish, livestock, and domestic purposes – can increase the value derived per unit of water used. Gains in crop production have often come, for instance, at the expense of fisheries. Values generated by fisheries, including ecosystem sustenance values, are routinely underestimated. Recognizing these values helps us to understand where there are win–win situations and what trade-offs will have to be made. However, these values are poorly recognized today and rarely influence the decision-making processes.
4.21.2.2 Rainfed Agriculture Productivity Rainfed agriculture includes both permanent crops (such as rubber, tea, and coffee) as well as annual crops (such as wheat, maize, and rice). For example, tubers, a staple crop for subSaharan Africa, have been all but uninfluenced by the technological developments of the green revolution. Rainfed farming constitutes 80% of the world’s cropland and produces more than 60% of the world’s cereal grains, generating livelihoods in rural areas while producing food for cities. In temperate regions with relatively reliable rainfall and good soils, rainfed agriculture generates high yields. Supplemental irrigation practices boost yields even higher. With rising concerns over the high cost of expanding largescale irrigation and the environmental impacts of large dams, upgrading rainfed agriculture is gaining increased attention (Rockstro¨m et al., 2007). Many people dependent on rainfed agriculture are highly vulnerable to both short-term dry spells and long-term droughts. Exposure to these risks can contribute to a reluctance to invest in agricultural inputs that could increase crop yields. Moreover, changing precipitation patterns resulting from climate change will compound this issue for many small farmers. There are several compelling reasons to invest in agricultural water management technologies and institutions
connected to rainfed agriculture (Rockstro¨m et al., 2007). To start, there is high potential to improve productivity, especially where yields are low. A majority of the rural poor are small holders who depend on rainfed rather than irrigated agriculture. Improving productivity in rainfed areas is therefore a way of supporting the poor. Boosting the potential of existing rainfed areas reduces the need for new large-scale irrigation development, which can generate adverse environmental and social impacts. Furthermore, the cost of upgrading rainfed areas is generally lower than the cost of constructing new irrigation schemes, particularly in sub-Saharan Africa. Even with these incentives, the potential contributions of rainfed agriculture to world food production are debatable, and forecasts regarding the relative roles of irrigated and rainfed agriculture vary considerably. Relying on rainfed agriculture also involves considerable risk. Water-harvesting techniques are useful for bridging short-term dry spells. Investments in water management are thus a way to decrease risk in rainfed agriculture. However, adoption rates of waterharvesting techniques are low, and extending successful local techniques over larger areas has proved to be difficult in the past. As longer dry spells may lead to crop failure, rainfed agriculture generally entails more risk than fully irrigated agriculture. Farmers adopt risk management strategies in line with the level of risk. There is a range of ’soft’ and hard’ measures that are available to integrate the risk related to climate variability in agriculture. Better control of water, either through full-fledged irrigation or supplemented irrigation, in costs or systems of crop insurances, can also integrate risk and provide farmers with better incentives to invest in their crop (Faures, 2010).
4.21.2.3 Irrigated Agriculture and Productivity The last 50 years have seen major investments in large-scale public surface irrigation as part of a global effort to increase staple food production, ensure food self-sufficiency, and to avoid devastating famine (this section follows Faures et al., 2007). From 1961 to 2008, for example, the world’s cultivated land increased approximately 12% (i.e., from 1368 to 1526 million hectares). At the same time, irrigated land area increased by 120%. The percentage of cultivated land equipped with irrigation rose from 10% in 1961 to 20% in 2008 (i.e., from 139 to 306 million hectares). Paralleling these global trends, irrigation investments in developing countries also accelerated rapidly in the 1960s and 1970s. On average, irrigated land in these regions grew by 2.2% per year reaching 155 million hectares by 1982. Widespread use of newly developed, high yielding, and fertilizer-responsive crops partially constituted the increased demand for water during this period. To achieve the higher yields now possible from these new crops, agriculture simply needed more water. In the developing world, other factors were important to the increased use of irrigation. Private and community-based investments in these countries, particularly programs aimed at groundwater pumping, grew from the 1980s onward. These projects were propelled by cheap drilling technology, rural electrification, and the availability of inexpensive small water pumps. Approximately 70% of the world’s irrigated land is in Asia (Figure 7), where it accounts for 34% of cultivated land.
Water Availability and Its Use in Agriculture 300
719
World
250 Millions of hectares
Developing countries 200 Asia 150
100 Developed countries 50 Africa 0 1961
1965
1970
1975
1980
1985
1990
1995
2000
2003
Figure 7 The area equipped for irrigation. Source: Comprehensive assessment of water management in agriculture (2007), Water for Food, Water for Life (Earthscan, 2007).
China and India alone account for more than half of irrigated land in Asia. Over time, Asia, with its high population densities, has come to rely increasingly on irrigated agriculture to boost agricultural productivity and thus to ensure domestic food security. Sub-Saharan Africa is much different. Notwithstanding a few large commercial irrigation schemes developed during the colonial period and a relatively modest small-scale irrigation subsector, there is very little irrigation in sub-Saharan Africa where water application methods are largely surface irrigation based, and little has been done to improve water productivity. The 1990s, however, saw a substantial rise in private irrigated peri-urban agriculture in sub-Saharan Africa in response to higher demand from growing cities for fresh fruits and vegetables (FAO, 2005). Today, it is suggested that global harvested irrigated area, which includes double cropping (two crops are grown in the same year), is estimated at 340 million hectares, although new incomplete evidence suggests otherwise. According to some studies, global harvested irrigated area might actually be higher than previously calculated after adjusting for higher cropping intensity and unreported, often informal, groundwater, or private irrigation use (Thenkabail et al., 2006). By the mid-1990s, irrigation projects leveled off around the world. Before this, the rapid growth in irrigated area, along with the other technological advancements of the green revolution – such as improved crop varieties and substantial growth in fertilizer use – led to a steady increase in staple food production and a reduction of real-world food prices. Until very recently, food prices in developed countries have been kept low by agricultural subsidies (Rosegrant et al., 2002), and since the late 1970s, the annual growth rate of global irrigation development, particularly in large-scale public schemes, has decreased. Other factors also contributed to the post-green revolution slow down of irrigation development. Most areas best suited for dams and irrigation have been developed, and as a result, new dams for irrigation and the related infrastructure for water delivery will cost more to construct in less
ideal locations. As a result, these geographic and economic conditions have led to overall less economic incentives for the development of large-scale irrigation projects. Other recent factors have created disincentives for irrigation investments as well. Some research has shown that the underperformance of large-scale irrigation (Chambers, 1988) has reduced donor interest (Merrey, 1997). Attention to the negative social and environmental externalities of dams – particularly the displacement of residents in affected communities and the calls for increased in-stream flows for environmental purposes – has discouraged the lending markets for irrigation investment. More competition for water from other sectors (as mentioned above) has reduced the scope for further development of irrigation. Irrigation is particularly crucial in sustaining agriculture across the dry belt, a region that extends from North Africa, the Middle East, through Northern China to Central America and parts of the United States (Figure 8). The advent of affordable drilling and pumping technologies in India and Pakistan in the mid-1980s led to the rapid development of shallow tube wells and the combined or conjunctive use of surface water and groundwater (Shah, 1993; Palmer Jones and Mandal, 1987). These technologies enabled farmers to have direct, individual control over water resources. By harvesting water by way of groundwater pumping, drainage reuse, or direct pumping from ponds, canals, and rivers, small holders gained flexibility and reliability in water delivery. Large-scale surface distribution systems did not offer these advantages. Yet, these technologies also contributed new challenges to water management. The indirect subsidization of electricity enabled farmers to pump water at zero to little cost. As a result, water tables have fallen in many regions of the world. In 1995, 38% of cereals grown in developing countries were on irrigated land, accounting for just less than 60% of all cereal production (Ringler et al., 2003). Rainfed cereal yields averaged 1.5 Mt ha1 in the developing world in 1995, but irrigated yields were 3.3 Mt ha1 (Rosegrant et al., 2002). The
720
Water Availability and Its Use in Agriculture
Less than 5%
5%–15%
15%–40%
More than 40%
No data
Inland water bodies
Figure 8 Irrigated areas as a share of cultivated area by country, 2003. Source: Comprehensive assessment of water management in agriculture (2007), Water for Food, Water for Life (Earthscan, 2007). From FAO (2006a) AQUASTAT database. Rome.
Oil crops Fodder Cotton Vegetables and pulses Sugar crops Other cereals Maize Wheat Rice 0
10
20
30
40
50
60
70
80
90
Millions of hectares Figure 9 Global distributions of crops under irrigation, 2000. Source: Comprehensive assessment of water management in agriculture (2007), Water for Food, Water for Life (Earthscan, 2007). From FAO (2006a) AQUASTAT database. Rome.
difference in productivity between irrigated and rainfed agriculture varies widely, depending on the climate, combination of crops, and technologies used. Typically, land productivity is 2–4 times higher in irrigated agriculture. Moreover, cropping intensity is typically higher under irrigation, with up to three rice crops per year in parts of Southeast Asia and two crops per year in most of the Asian subcontinent. Figure 9 shows the distribution of crops under irrigation worldwide.
4.21.2.4 Livestock Keeping livestock is one of the most important, complex, and diverse subsectors of world agriculture and for many people it
is a primary means of escaping poverty in rural areas. Modest amounts of meat in the diets of African children appear to improve mental, physical, and behavioral development (Sigman et al., 2005; Neumann et al., 2003). This suggests that meat production and water productivity must account for social and health values as well as produced food mass. However, current literature on livestock–water interactions does not address this important topic. Moreover, past research has underestimated the contributions of livestock to rural livelihoods in part because studies were predominantly concerned with food mass productivity. Limited consideration has been given to the nonmonetized products and services associated with livestock.
Water Availability and Its Use in Agriculture
Poor and subsistence households obtain multiple benefits from the use of livestock (Shackleton et al., 1999; Landefeld and Bettinger, 2005). Therefore, assessing the water resources used to support these animals must account for values beyond meat production. Livestock contribute to the livelihoods of at least 70% of the world’s rural poor and strengthen their capacity to cope with income shocks (Ashley et al., 1999). They provide milk, blood, manure, hides, and farm power essential for the cultivation and marketing of crops. Livestock assets are often an important source of wealth security. As mentioned above, livestock water productivity examines the net beneficial livestock-related products and services in relation to water use. As a systems concept, livestock water productivity attempts to account for the complex relationships among food production, livelihood, and water demands. The implications of livestock on water use have been generally overlooked by research. Animals ‘consume’ water directly for drinking purposes, but it is the food they eat that requires large quantities of water, as discussed earlier. The type, quality and origin of the feed used for animals, together with animal management practices can have major impact on livestock water productivity. Livestock water productivity differs from water or rainuse efficiency because it examines water depleted rather than applied or inflowing water. Four basic strategies help to increase livestock water productivity: improving water supply, feed sourcing, enhancing animal productivity water conservation, and spatially optimistic distributing of watering points, animal stocking rates, and pasture productivity (CA, 2007). Providing sufficient and adequate quality drinking water also improves livestock water productivity as it keeps the animal healthy and productive. However, it does not factor directly into the livestock water productivity equation because water that has been consumed remains inside the animal and thus within the production system, although subsequent evaporative depletion may follow. A balanced, site-specific approach that considers all four strategies will help increase the benefits derived from the use of agricultural water for the production of animal products and services. Children, women, and men often receive different benefits from animal keeping and have different roles in managing livestock–water interactions These are considerations that need to be taken into account in attempts to improve livestock water productivity. Livestock water productivity does not necessarily seek to maximize the number of livestock or the production of animal products and services. Rather it seeks to reach a higher level of animal products per unit of water consumed.
4.21.2.5 Aquaculture and Fisheries Inland fisheries and aquaculture contribute about 25% to the world’s fish production and are a fast growing sector (see Dugan et al., 2007). In addition, many important estuarine and coastal fisheries are closely linked to the ecological processes that occur in freshwater systems. Fisheries and aquaculture from lakes, reservoirs, rivers, ponds, and wetlands contributed about 25% (i.e., 34 million metric tons) of the reported world fisheries production in 2003 (FAO, 2004). However, catches in rivers and wetlands are easy to
721
underestimate because the contributions of numerous fisheries on smaller tributaries and water bodies are generally overlooked (Coates, 2002). Reported harvests from river fisheries alone have been shown to account for only 30–50% of actual catch (Kolding and van Zwieten, forthcoming), and the contribution from inland fisheries is therefore believed to be underestimated. Aquaculture uses water in two ways. Blue water is needed for the fish ponds and the processes of aquaculture; blue and green water is also necessary for feed production. Water productivity in terms of aquaculture is defined as the mass or value of the aquaculture produce divided by the amount of water required for feed plus the amount of evaporation from the pond. On-farm water use in aquaculture can be as low as 500–700 l in super-intensive recirculation systems and as high as 45 000 l of water (evaporation plus seepage plus feed) per kg of produce in extensive ponds (Verdegem et al., 2006). Fish can often be integrated into water management systems with the addition of little or no water (Prein, 2002). Renwick (2001) found that the fisheries in irrigation reservoirs at Kirindi Oya, Sri Lanka, contributed income equal to 18% of the rice production in the system. Haylor (1994, 1997) assessed the potential for aquaculture in small- and large-scale irrigated farming systems in the Punjab, Pakistan. The study noted that aquaculture in the region was almost entirely focused on carp production using groundwater sourced from tube wells. It also concluded that there was economic justification for expanding such aquaculture using local shallow tube wells. The study also found that the revenue potential for cage aquaculture in irrigation canals was also attractive, but operational conflicts in the use of water for agriculture would need to be resolved. Murray et al. (2002) have pointed out that traditional power structures may undermine attempts to integrate aquaculture in irrigation systems and that changes in laws and regulations would be required from community to national levels. In coastal areas, aquaculture may severely degrade land and water quality and biodiversity, requiring special attention (Gowing et al., 2006). Fisheries in lakes, rivers, and wetlands present a special case for water productivity assessment because fish are only one of the many ecosystem services provided by aquatic ecosystems. The values and livelihood benefits of fisheries are high and often ignored or underestimated, but considering only the values of fish produced would grossly underestimate the value of water in these aquatic ecosystems. The water productivity of fisheries systems needs to be considered in terms of ecosystem services and livelihoods supported per unit of water. Thus, maintenance of wetlands and biodiversity should be considered as potential benefits for leaving water in these aquatic ecosystems.
4.21.3 Environmental and Health Implications of Agricultural Water Use Agricultural water management has both negative and positive impacts on environment and health (this section follows Falkenmark et al., 2007). On the one hand, agricultural water management can improve health status through better
722
Water Availability and Its Use in Agriculture
nutrition, increasing the availability of drinking water, and controlling diseases such as malaria. On the other hand, extensive reporting shows many cases where agricultural water management has encouraged the spread of some waterborne diseases (Ersado, 2005); impacted upstream and downstream ecological services (Gichuki, 2004); affected water logging and salinization (Khan et al., 2006); and increased agrochemical usage, pollution, and eutrophication (Hendry et al., 2006).
4.21.3.1 Impact on Rivers, Wetlands, and Biodiversity Water management affects the physical and chemical characteristics of inland and coastal aquatic ecosystems, the quality and quantity of water, as well as direct and indirect biological change (Finlayson and D’Cruz, 2005; Agardy and Alder, 2005; Vo¨ro¨smarty et al., 2005). It has also affected terrestrial ecosystems through the expansion of agricultural lands and changes in water balances (Foley et al., 2005). Regulation of the world’s rivers has altered water regimes, with substantial declines in discharges to the ocean (Meybeck and Ragu, 1997). A long-term trend analysis (i.e., more than 25 years) of 145 major world’s rivers indicates that discharge has declined in one-fifth of the basins (Walling and Fang, 2003). Worldwide, large artificial impoundments hold vast quantities of water and cause significant distortion of flow regimes (Vo¨ro¨smarty et al., 2003).
4.21.3.1.1 Aquatic ecosystems Water diversion and the construction of hydraulic infrastructure have had the following negative effects: loss of local livelihood options, fragmentation, destruction of aquatic habitats, changes in the composition of aquatic communities, species loss, and health problems resulting from stagnant water. Improved flood control – an important agricultural mechanism for reducing risk – has led to the reduction of sedimentation and the deposition of nutrients on floodplains, as well as reduced flows and nutrient deposition to parts of coastal zones (Finlayson and D’Cruz, 2005). Inter basin transfers of water, particularly large transfers between major river systems have been in consideration in India and China, for example, are expected to be particularly harmful to downstream ecosystems (Gupta and Deshpande, 2004; Alam and Kabir, 2004) and will likely exacerbate pressures from hydrological regulation (Snaddon et al., 1999). Junk (2002) has highlighted the similar adverse consequences on water regimes expected from the construction of industrial waterways (i.e., hydrovias) through large wetlands, such as the Pantanal of Mato Grosso, Brazil. Shrinking lakes and rivers. There are many instances where consumptive water use and water diversions have contributed to the severe degradation of downstream ecosystem services by shrinking lakes and drying rivers. The degradation of the Aral Sea in Central Asia is an extreme case. Similarly, Lake Chapala, the world’s largest shallow lake, situated in the Lerma-Chapala Basin in central Mexico, is an example of consumptive water use upstream affecting lake-size downstream. From 1979 to 2001, water volume in the lake dropped substantially to about 20% of volume capacity due to excessive water extraction for agricultural and municipal needs.
Stream flow depletion is a widespread phenomenon in tropical and subtropical regions in river basins with large-scale irrigation, including the Pangani (IUCN, 2003), Yellow (He et al., 2005), Aral Sea tributaries, Chao Phraya, Ganges, Incomati, Indus, Murray-Darling, Nile, and Rio Grande (Falkenmark and Lannerstad, 2005). Smakhtin et al. (2004) have suggested that environmental flow (i.e., the stream flow required for aquatic ecosystem health) has already been overappropriated in many river basins. For example, in the United States the construction of dams and water diversions for irrigation and other purposes in the Colorado Basin, together with large-scale inter-basin transfers, have greatly reduced the flow of the river to the delta. As a result, a considerable portion of the delta has been transformed into mudflats, salt flats, and exposed sand. With the loss of the delta habitats, wetlands now exist mainly in areas where agricultural drainage has occurred (Postel, 1996). The Ganges is among the major rivers of South Asia that no longer discharges year round to the sea. As a result, there is a rapid upstream advance of saline water, with consequent changes in mangrove communities, fish habitat, cropping, and human livelihoods (Postel, 1996; Mirza, 1998; Rahman et al., 2000). On the Zambezi River in Southern Africa, damming for electricity and agriculture has reduced flows to the coast and led to a decline in shrimp production that could have been worth as much as $10 million a year (Gammelsrod, 1992). The regulation of rivers has brought many benefits to people, but the adverse impacts, especially those related to reduced downstream flows, have often failed to receive adequate and transparent consideration (WCD, 2000; Revenga et al., 2000; MEA, 2005a). Effects on wetlands. Water regulation and drainage for agricultural development are the main causes of wetland habitat loss and degradation (Revenga et al., 2000; Finlayson and D’Cruz, 2005) as well as the consequent loss of ecosystem services. By 1985, drainage and conversion of wetlands, mainly for agriculture, had affected an estimated 56–65% of inland and coastal marshes in Europe and North America and 27% in Asia (OECD, 1996). Drainage of wetlands often reduces important regulating ecosystem services, with such outcomes as increased vulnerability to storms and flooding and further eutrophication of lakes and coastal waters. The loss of small wetlands (regionally referred to as potholes) on the prairies of Canada and the United States through drainage and infilling has led to the loss of habitat for large numbers of migratory water birds (North American Waterfowl Management Plan, 2004). The loss of forested riparian wetlands adjacent to the Mississippi River in the United States was seen as an important factor contributing to the severity and damage of the 1993 flood in the Mississippi Basin (Daily et al., 1997). Changes in water quality. The use of fertilizers has brought major benefits to agriculture, and has also led to widespread contamination of surface water and groundwater through runoff. Over the past four decades, excessive nutrient loading has emerged as one of the most important direct drivers of ecosystem change in inland and coastal wetlands, with reactive nitrogen entering oceans at an increased rate of nearly 80% from 1860 to 1990 (MEA, 2005b). Phosphorus applications have also increased, rising threefold since 1960, with a steady increase until 1990 followed by a leveling off at approximately the application rates of the 1980s (Bennett et al., 2001). These
Water Availability and Its Use in Agriculture
changes are mirrored by phosphorus accumulation in soils, with high levels of phosphorus runoff. Excessive nutrient loading causes algal blooms, decreased drinking water quality, eutrophication of freshwater ecosystems and coastal zones, and hypoxia in coastal waters. In Lake Chivero, Zimbabwe, agricultural runoff is responsible for algal blooms, infestations of water hyacinth, and fish declines as a result of high levels of ammonia and low oxygen levels (UNEP, 2002). In Australia, extensive algal blooms in coastal inlets and estuaries, inland lakes, and rivers have been attributed to increased nutrient runoff from agricultural fields (Lukatelich and McComb, 1986; Falconer, 2001). Diffuse runoff of nutrients from agricultural land is considered a major cause for increased eutrophication of coastal waters in the United States as well as for the periodic development, often varying from year to year, of anoxic conditions in coastal water in many parts of the world, such as the Baltic and Adriatic Seas and the Gulf of Mexico (Hall, 2002). Extensive evidence shows that up to 80% of the global incidents of nitrogen loading can be retained within wetlands (Green et al., 2004; Galloway et al., 2004). However, the ability of such ecosystems to cleanse nutrient-enriched water varies and is limited (Alexander et al., 2000; Wollheim et al., 2001). Verhoeven et al. (2006) pointed out that many wetlands in agricultural catchments receive excessively high loads, with detrimental effects on biodiversity. Bioaccumulation, as a consequence of the wide use of agrochemicals, has had dire outcomes for many species that reside in or feed predominantly in wetlands or lakes where residues from pesticides have accumulated. The declined breeding success of raptors was a turning point in developing awareness about the dangers of pesticide use (Carson, 1962). An increasing amount of analytical and eco-toxicological data has become available for aquatic communities, and more recent research has also focused on risk assessments and the development of diagnostic tests that can guide management decisions about the use of such chemicals (van den Brink et al., 2003). Taylor et al. (2002) have highlighted the high levels of pesticide use and low levels of environmental risk assessment in developing countries. Vo¨ro¨smarty et al. (2005) reported that water contamination by pesticides has increased rapidly since the 1970s despite increased regulations, especially in developed countries, of xenobiotic substances (i.e., those chemical compounds foreign to a living organism). However, bans on the use of these chemicals have generally been imposed only two to three decades after their first commercial use, as with dichlorodiphenyltrichloroethane (DDT) and the common herbicide atrazine. Many of these substances are highly persistent in the environment, but because of the generally poor monitoring of their long-term effects, global and long-term implications of their use cannot be fully assessed.
4.21.3.1.2 Terrestrial ecosystems In many parts of the world, extensive sheet wash and gully erosion, due to poor land management practices, have had significant environmental effects. Large tracts of land have been devastated resulting in reduced agricultural productivity. Erosion has also contributed to the rapid siltation of reservoirs
723
and increased sediment loads in many rivers (CA, 2007). On a regional scale, some reservoirs in Southern Africa are at risk of losing more than a quarter of their storage capacity within 20–25 years (Magadza, 1995). While many Australian and Southern African waters are naturally silty, many have experienced increased silt loads as a result of agricultural practices (Davies and Day, 1998). Zimbabwe’s more than 8000 small- to medium-size dams, for example, are threatened by sedimentation from soil erosion, while the Save River, an international river shared with Mozambique, has been reduced from a perennial to a seasonal river system in large part due to increased siltation caused by soil erosion. While it is not always easy to differentiate natural erosion from human-induced erosion. The high sediment loads carried by Asian rivers are partly a consequence of land-use practices, particularly land-clearing practices for agriculture that lead to erosion, a situation likely to continue as a consequence of the expansion of agriculture in Africa, Asia, and Latin America (Hall, 2002). Changes in the water table. Water builds up in a soil profile when the rate of input exceeds the rate of throughput (e.g., when irrigation volumes are greater than crop water consumption by way of evapotranspiration). This can cause water logging and salinization, which are extensively described for irrigated agriculture (Postel, 1998). Excessive irrigation can result in soil salinization in areas where the water table rises close to the surface and evaporation leaves salts behind in the soil profile. Salt-affected soils in irrigation schemes are often related to poor soil and water management, in addition to the unsuitability of many soils for irrigation. Clearing woody vegetation for pastures and crops can also lead to dryland salinization. Tree-covered landscapes provide an important regulating service by consuming rainfall through high evapotranspiration, limiting groundwater recharge, and keeping the groundwater low enough to prevent salt from being carried upward through the soil. Australia has had major problems with soil salinization as native woody vegetation was cleared in the 1930s for pastures and agricultural expansion (Farrington and Salama, 1996). Consumptive water use has declined there, the water table has risen, and salt has moved into the surface soils so that large tracts of land have become less suitable or unusable for agriculture (Anderies et al., 2001; Briggs and Taws, 2003). The overall trend, however, in irrigation is one of increased pumping and reduced water levels, but good salt management increases a critical issue in particular in arid regions. Moisture recycling. Increased irrigation and land clearing for agriculture have modified green water flows across the globe, reducing them by 3000 km3 through forest clearing and increasing them by 1000–2600 km3 in irrigated areas (Do¨ll and Siebert, 2002; Gordon et al., 2005). Changes in land cover affect evapotranspiration and ultimately impact the hydrologic cycle. It has been suggested that large-scale deforestation can reduce moisture recycling, affect precipitation (Savenije, 1995, 1996; Trenberth, 1999), and alter regional climates, with indications of global impacts (Kabat et al., 2004; Nemani et al., 1996; Marland et al., 2003; Savenije, 1995). Pielke et al. (1998) concluded that the evidence is convincing that land cover changes can significantly influence weather and climate and are as important as other human-induced changes for
724
Water Availability and Its Use in Agriculture
the Earth’s climate. Regional studies in West Africa (Savenije, 1996; Zheng and Eltathir, 1998), the United States (Baron et al., 1998; Pielke et al., 1998), and East Asia (Fu, 2003) have illustrated the ways in which changes in land cover affect green water flows, with impacts on local and regional climates. There are also indications that increased vapor flows through irrigation can alter local and regional climates (Pielke et al., 1997; Chase et al., 1999). The conversion of dry lands to irrigated croplands in Colorado resulted in a 120% increase in evapotranspiration, contributing to higher precipitation, lower temperatures, and an increase in thunderstorm activity (Pielke et al., 1997).
4.21.3.2 Health Impacts Among the agrochemicals that pose the greatest threats to domestic use of groundwater are nitrate and biocide residues. In addition, arsenic contamination in groundwater has emerged as a major health issue in Asia recently. Other health aspects concern malnutrition and vector-borne diseases. Many of the rural poor in Asia obtain water for drinking and household use from shallow aquifers under agricultural land. Irrigated rice fields can serve as breeding sites for mosquitoes, snails, and other intermediate hosts capable of transmitting human parasites. In particular, before transplanting and after harvest, puddles in rice fields are attractive breeding grounds for the mosquito Anopheles gambiae, Africa’s most efficient malaria vector. The conditions for mosquito breeding in rice fields have been identified and management practices, such as alternate wetting and drying of fields, exist to mitigate the problem. Moreover, countries such as Sri Lanka have made great strides in controlling epidemics through broad-based public health campaigns. Japanese B-encephalitis is highly correlated with rice irrigation in Asia, especially where pigs are also reared, as in China and Vietnam. Again, alternate wetting and drying can help reduce the breeding of disease vectors (Keiser et al., 2005a). Nitrate leaching from flooded rice fields is normally negligible because of rapid denitrification under anaerobic conditions (the following section follows from Bouman et al., 2007). In the Philippines, for example, nitrate pollution of groundwater under rice-based cropping systems exceeded the 10 mg l1 limit for safe drinking water only when highly fertilized vegetables were included in the cropping system (Bouman et al., 2002). In the Indian Punjab, however, an increase in nitrate of almost 2 mg l1 was recorded between 1982 and 1988, with a simultaneous increase in nitrogen fertilizer use from 56 to 188 kg ha1, most of it on combined rice–wheat cultivation (Bijay-Singh et al., 1991). These may lead to the blood disorder methemoglobinemia in human populations, especially in babies. Mean biocide use in irrigated rice systems varies from some 0.4 kg active ingredients per hectare in Tamil Nadu, India, to 3.8 kg ha1 in Zhejiang Province, China (Bouman et al., 2002). In the warm and humid conditions of the tropics, volatilization is the major process of biocide loss, especially when biocides are applied on water surfaces or on wet soil. Relatively high temperatures favor rapid transformation of remaining biocides by photochemical and microbial degradation, but little is known about the toxicity of the residual components.
In case studies in the Philippines, mean biocide concentrations in groundwater under irrigated rice-based cropping systems were one to two orders of magnitude below the single and multiple biocide limits for safe drinking water (i.e., 0.1 and 0.5 mg l1), although temporary peak concentrations of 1.14–4.17 mg l1 were measured (Bouman et al., 2002). Biocides and their residues may be directly transferred to open water bodies through drainage water that flows overland from rice fields. The potential for water pollution from biocides is greatly affected by field water management. Different water regimes result in different pest and weed populations and densities, which farmers may combat with different amounts and types of biocides. Agricultural use of untreated wastewater can affect human health through exposure to pathogens, parasite infections, and heavy metals. Leafy vegetables, eaten raw, can transmit contaminations from farm fields to consumers. Hookworm infections are transmitted by direct exposure to contaminated water and soils. A survey along the Musi River in India revealed the transfer of metal ions from wastewater to cow’s milk through fodder irrigated with wastewater. About 4% of grass samples showed excessive amounts of cadmium and all samples showed excessive lead levels. Milk samples were contaminated with metal ions ranging from 1.2 to 40 times permissible levels (Minhas and Samra, 2004). Farmers and their families using untreated wastewater are exposed to health risks from parasitic worms, viruses, and bacteria. Many farmers cannot afford treatment for some of the health problems caused by exposure. Generally, farmers irrigating with wastewater have higher rates of parasite infections than farmers using freshwater do, but there are exceptions (Trang et al., 2006). In addition, skin and nail problems occur more frequently among farmers using wastewater (Van der Hoek et al., 2002).
4.21.3.3 Environmental and Health Mitigation Agricultural water use is closely linked with health and environmental impacts. For health, the negative impacts of irrigation development can be mitigated through better design and operation of new and existing systems, especially through the multiple uses of irrigation water. Integrated approaches have taken many forms, including integrated river basin management, integrated land and water management, ecosystem approaches, integrated coastal zone management, and integrated natural resources management. These management strategies often seek to do the following: address the integration of a broad range of benefits and costs associated with land-use and water decisions, including effects on ecosystem services, food production, and social equity; involve key stakeholders at cross-institutional levels; and address interconnectedness across subbasin, river basin, and other biophysical scales. The MEA (2003) has provided a major advancement in understanding the links between the provision of ecosystem services and human well-being. Increased awareness is still needed on several different levels. The scientific knowledge of how ecosystem services contribute to human well-being within and between different sectors of society, and the role of water in sustaining these services, needs to be improved.
Water Availability and Its Use in Agriculture
Dissemination of information on these issues and dialog with stakeholders should be enhanced. Civil society organizations can help to ensure that appropriate consideration is given to the voices of individuals and social groups and to nonutilitarian values in decision making. Minority groups and disadvantaged groups, such as indigenous people and women, in particular, need to be heard. Women play a critical and increasing role in agriculture in many parts of the developing world (Elder and Schmidt, 2004). Mitigation measures are modifications to the design or operation of agricultural water development projects to reduce negative environmental and health impacts. However, present levels of understanding mean that very often some negative impacts are not foreseen prior to project implementation. Consequently, there should be a constant reevaluation of the need for mitigation measures throughout the life of a project. This requires monitoring so that measures can be introduced retrospectively when necessary. Monitoring enables health authorities to target resources and treatment interventions at times and locations of greatest need. A wide range of technical mitigation measures to prevent environmental damage has been developed for formal irrigation schemes. Measures that promote high water-use efficiency also tend to mitigate negative environmental and health impacts. For example, good water management and drainage are prerequisites to preventing water logging, decreasing habitats for mosquitoes, and minimizing salt accumulation in soils. Over the last 20 years, considerable progress has been made in the development of methods to determine environmental flows downstream of dams and extractions for irrigation. Increasingly, these techniques are taking a systems approach that includes holistic ecosystem assessments and works to predict the impacts of different flow regimes on the livelihoods of water users (Dyson et al., 2003). For example, downstream response to imposed flow transformation (DRIFT) is a scenario-based environmental flow assessment process designed specifically for use in negotiations over water resources. It is designed to quantify the linkages between changing river conditions, and the social and economic impacts for riparian people who rely on rivers for their livelihoods (Brown and King, 2000). Hydrological environmental flow assessment methods are being developed for use in areas where insufficient ecological data exist for conclusive analyses (Smakhtin and Shilpakar, 2005). Such approaches help define environmental targets and thus facilitate the design of mitigation measures by specifying desired environmental conditions. Concerning disease vectors, the primary approach is to design irrigation or pastoral water systems that do not provide habitats for vectors, while also conducting health education. In these cases, good design and construction of canal and drainage systems, and proper leveling of fields, can ensure that water is fast flowing and stagnant pools do not occur. Other options include direct vector control using chemicals or biological methods. However, with chemical control, care is required in application to ensure that the vectors do not develop resistance. Physical removal of habitats, for example, through manual or mechanical cutting of weeds is also possible. Good cleaning and preventative maintenance of all infrastructures, including canals, cattle troughs, hydraulic
725
structures, and drains reduce the breeding of vectors and intermediate hosts, as well as improve irrigation performance. In several sub-Saharan African countries, weed control in canals has been applied as an effective method of vector control. However, routine cleaning work can itself be a health hazard. In the Gezira scheme in Sudan, canal cleaning personnel became the group most infected with the disease schistosomiasis (Fenwick et al., 1982). In some places, attempts have been made to minimize risk exposure by adapting the time of cleaning activities to the cycle of the parasite or by providing alternative tools (Euroconsult, 1993). A recent review of 40 largely pre-DDT interventions suggests that environmental modification can be a very effective malariacontrol strategy (Keiser et al., 2005b). Adapting water management to modify the vector habitat is another approach that has often been proposed in biomedical studies as an easy and cheap measure for vector control. However, there are very few examples where this type of environmental manipulation has been applied in practice (Matsuno et al., 1999; Laamrani and Boelee, 2002; Boelee and Laamrani, 2004). This is because, in reality, it is neither simple nor cheap to change established water management patterns. Water management interacts not only with vector breeding or disease transmission, but also with the irrigation system itself. Changes to water distribution often require modifications in design, notably the sizing of canals and type of structures. For example, if continuous delivery is replaced by rotation of the water flow to disrupt breeding sites, the discharge in the canals alters from constant low flows to intermittent high flows, requiring larger canals. At the same time, the wider human environment is influenced. With water flowing in the canals continuously, farmers can irrigate their crops at any time. With rotation, the flow has to be divided over time between users, demanding a higher level of organization. Water scheduling to meet crop water requirements is complicated, especially when conflicting interests between higher water-use efficiencies and farmers demanding flexibility have to be accounted for. If disease-control measures have to be observed as well, scheduling and management become very complicated (Boelee, 1999). Adaptive water management in rice fields may result in reduced vector breeding and hence reduced transmission of Japanese encephalitis and malaria (van der Hoek et al., 2001; Keiser et al., 2005b). However, these studies mainly report from Asia. In an African context, with constraints on resources and capacity, it may be especially difficult to achieve the required water deliveries and level of water management (Mutero et al., 2000). In reality, effective health interventions require an integrated approach that simultaneously implements avoidance and mitigation measures in collaboration between the water and health authorities.
4.21.4 Water Governance Improving agricultural water productivity in the future will take careful attention to the ways in which water use is linked to economies, social well-being, and ecological systems. Implementing these changes will require more effective water governance, which means rethinking how water is managed. Beyond national-scale water legislation, water governance
726
Water Availability and Its Use in Agriculture
encompasses various bodies of formal and informal regulations and institutions and also the way decision-making and political power is exercised. In short, it refers to the social mobilization and other actions designed to promote ownership, co-investment, capacity building, incentives for participation, and willingness to pay for services at the community level (UNDP, 2004). The following section outlines various forms of water governance and points to future challenges water governance will face.
4.21.4.1 Definition Governance is the way authority is organized and executed in society, and often includes the normative notion of the necessity for good governance (Merrey et al., 2007). The Global Water Partnership defines water governance as ‘‘the range of political, social, economic, and administrative systems that are in place to develop and manage water resources, and the delivery of water services, at different level of society’’ (Rogers and Hall, 2003). Governance is therefore a broad term that includes institutions, organizations, and policies. Effective water governance builds institutional capacity from the local level upward and empowers stakeholders with knowledge and the ability to make decisions about matters that directly affect their lives. It promotes the equal participation of women and men in decision making. Water governance is critical for resource planning and allocation among riparian states (those sharing a water basin) and vital for conflict resolution to defuse upstream–downstream tensions and balance the needs of different groups sharing water resources. Good water governance determines the appropriate role for the government in service delivery (i.e., as a facilitator or as a service provider) and ensures that water and sanitation services provided by both public and private actors meet the needs of the people they serve and do not fall prey to corruption. Good water governance corrects market distortions, perverse incentives, and pricing that shuts out the poor (UNDP, 2004).
basin, essential functions are partly or completely carried out, with their sum constituting basin governance (Table 3). Much attention has been given to the ideal organizational model for river basin management, while much less emphasis has been placed on the process of developing, managing, and maintaining collaborative relationships for river basin governance. More fundamentally, the essential function in river basin management – allocating water between competing uses and users, including the environment – has not received sufficient attention, although it is at the heart of integrated water resources management. Moreover, agricultural water and land practice, such as rainfed agriculture, livestock and fisheries practices often do not feature strongly. There are two main trends in basin governance. One trend concerns watersheds, or sub-basins of a limited size (typically from 10–1000 km2), where local stakeholders and agencies attempt to solve their land- and water-related problems. A second trend concerns the management of wider river basins. This trend has three salient aspects (Svendsen and Wester, 2005). First is the consensus that integrated water resources management should be carried out at the river basin level. This, together with the desire to realize the promise of integration, has placed river basin management on the agenda of
Table 3
Functions for river basin management
Function
Description
Plan
Formulation of medium- to long-term plans for managing and developing water resources in the basin. Activities executed for the design and construction of hydraulic infrastructure. Activities executed to maintain the serviceability of the hydraulic infrastructure in the basin. Mechanisms and criteria by which water is apportioned among different use sectors, including the environment. Activities executed to ensure that allocated water reaches its point of use. Activities executed to monitor water pollution and salinity levels and ensure that they remain at or below accepted standards. Flood and drought warning, prevention of floods, and development of emergency works, drought preparedness, and coping mechanisms. Provision of space or mechanisms for negotiation and litigation. Priorities and actions to protect ecosystems, including awareness campaigns. Harmonization of policies and actions undertaken in the basin by state and nonstate actors relevant to land and water management.
Construct facilities Maintain facilities
Allocate water
Distribute water
4.21.4.2 Types of Governance for River Basin Management The growing pressure on water resources and the increasing hydrological, social, and ecological interdependencies in closing river basins have led to widespread recognition of the need for holistic approaches to water management. There is a renewed emphasis on river basins as the most appropriate spatial unit for water management. The decision to manage water on the basis of river basins is a political choice, and river basins thus become a scale of governance in which tensions arise among effectiveness, participation, and legitimacy (Barham, 2001; Schlager and Blomquist, 2000; Wester and Warner, 2002). Progress in establishing adaptive, multilevel, collaborative governance arrangements for river basin management has been weak, with undue emphasis on form (setting up river basin organizations) over process. Although there may not be a central basin manager, this does not mean that river basins are not managed (Schlager and Blomquist, 2000). This can be accomplished by identifying the roles of various actors engaged in river basin water management, asking who does what, where, to what end, and how well. In any river
Monitor and enforce water quality Preparedness against water disasters
Resolve conflicts Project ecosystems Coordinate
Note: The functions listed here subsume supporting functions such as data collection and resource mobilization, which are not ends in themselves, but rather facilitate the higher-level functions listed. From Svendsen M, Wester P, and Molle F (2005) Managing river basins: An institutional perspective. In: Svendsen M (ed.) Irrigation and River Basin Management: Options for Governance and Institutions. Wallingford: CABI Publishing.
Water Availability and Its Use in Agriculture
governments and international funding agencies and has led to many new river basin initiatives. Second, the number of public and private sector actors involved in, or concerned with basin planning and management, is increasing, from environmental agencies and civil society or interest groups to regulatory bodies and service providers for agricultural, municipal, and industrial water users. With rising standards of living, urbanization, and continuing environmental deterioration, more diverse stakeholders and worldviews need to be integrated. Third, organizations associated with basin planning and management have become more specialized and differentiated into regulators, resource managers, and service providers (Millington, 2000). Regulation and standard setting are carried out in the public interest and are necessarily functions of government, but other tasks may be fulfilled by commercial or hybrid public–private organizations. River basin organizations cover a wide gamut of organizations with quite varied roles and structures. At first this may seem a source of confusion, but it also suggests that both the nature of the problems faced (e.g., development or management) and the particular history and context of each basin reflect on each river basin organization. The following typology can be inferred from a broad-brush review of river basin organizations, keeping in mind that there are no clear-cut definitions and that there are large variations in roles and power, even within the same category. In other words, the generic terms may not correspond to particular bodies. Basin authorities are autonomous executive organizations with extensive mandates for their river basin, undertaking most water-related development and management functions. They serve as regulator, resource manager, and service provider all in one. The Damodar Valley Corporation in India, the Mahaweli Authority in Sri Lanka, the Companhia de Desenvolvimento dos Vales de Sa˜o Franciscoe do Parnaiba in Brazil, and the Confederaciones Hidrograficas in Spain are examples of such basin authorities. Authorities generally exhibit poor responsiveness to local demands and are often undermined by bureaucratic conflict because they infringe on the competence of other government agencies and line ministries. Some of these authorities receive basin-wide, multifunctional mandates covering various domains but are not endowed with the legal, political, or administrative power to achieve them. They generally end up focusing on construction projects and dam management (mostly for hydropower or flood control). Examples include the Damodar Valley Corporation in India (Saha, 1979), the River Basin Development Authorities in Nigeria (Adams, 1985), and the China River Commissions (Millington, 2000). Some authorities were designed to ensure regional infrastructure development (the early River Basin Commissions in Mexico), others endured as powerful manager/operators (Brantas basin in Indonesia, Tarim in China), while others shrank and were confined to one issue or degenerated into powerless parallel structures with narrow scope and erratic funding (A. Dourojeanni, personal communication). Basin commissions or committees focus on policy setting, basin-wide planning, water allocation, and information management, with varying degrees of stakeholder participation. They are usually endowed with authority to manage water resources (allocating permits, defining taxation,
727
negotiating water allocations, and defining effluent standards), and sometimes to plan future developments, but are not involved in operation or construction. Examples include the Delaware Commission in the United States, the Murray–Darling Commission in Australia, the British water authorities, and the French Agences de l’eau. Coordinating councils are deliberative decision-making bodies incorporating public and private stakeholders and integrating policymaking across different policy areas. They are not organizations in the strict sense, but rather bring together stakeholders from various agencies and water-use sectors. Their role is coordination, conflict resolution, and review of water resources allocation or management. Examples include the river basin councils in Mexico (Wester et al., 2005), the proposed catchment management agencies in South Africa (Waalewijn et al., 2005), the Zimbabwean catchment councils (Jaspers, 2001), the river basin committees and users commissions in Brazil (Lemos and Oliveira, 2004), and several river commissions in the United States. International river commissions are unique because coordination is achieved between countries rather than among stakeholders and because political dimensions are pervasive. They were frequently established as part of a treaty signed among riparian countries or to manage dams on shared rivers (e.g., Senegal, Volta, or Zambezi rivers) (Barrows, 1998). They mediate water conflicts through consultation and cooperation and may also manage common databases, and their work may lead to concrete agreements. From a governance perspective, institutional arrangements for river basin management may be distributed along two axes, one that distinguishes between state-driven and stakeholder-driven functioning, and the other that contrasts centralized and decentralized modes. This yields four models for basin governance: unicentric (state-driven, centralized), deconcentrated (state-driven, decentralized), coordination (stakeholder-driven, centralized), and polycentric (stakeholder-driven, decentralized). Under the unicentric model, a basin authority or line ministry manages the river basin. In the polycentric model, the actions of existing organizations, layers of government, and stakeholder initiatives are coordinated to cover an entire river basin or sub-basin.
4.21.4.3 Basin Governance Challenges River basin governance is about the emergence of the appropriate blend of government, civil society, and markets in decision making and regulation. In addition to greater control, rigor, and openness for water resource planning and allocation, as just described, integrated river basin management demands adequate governance. This brings out two main challenges: ensuring that all stakeholders, including the environment, have a voice, and coordinating uses and policies within the basin. Although frequently advocated as a key to achieving effective water management (Rogers and Hall, 2003), stakeholder participation in river basin management is not straightforward, and achieving substantive stakeholder representation have proved to be elusive in practice (Wester et al., 2003). Emphasizing participation in river basin management may draw attention away from the very real social and economic differences among people and the need for
728
Water Availability and Its Use in Agriculture
redistributing resources, entitlements, and opportunities. This is unlikely to happen without challenges, and decision makers committed to social equity need to devise mechanisms that strengthen the representation of marginal groups in river basin management and empower them. Stakeholder platforms, whether river basin councils, catchment management agencies, or watershed councils, democratize river basin management by giving voice to multiple actors. However, much depends on the institutional arrangements from which these river basin management institutions emerge, as many roles, rights, and technologies and physical infrastructure for controlling water are already in place. Stakeholders have different levels and types of education, differ in access to resources and politics, hold different beliefs about how nature and society function, and often speak different languages (Edmunds and Wollenberg, 2001). If these differences are not taken into account when creating new rules, roles, and rights, the institutional outcome can easily privilege those who are literate and have access to the legal system and eventually institutionalize inequality and power differentials instead of giving voice to marginal groups (Wester and Warner, 2002). This review of basin governance patterns identified the various types of organizations and arrangements for basin management. A strong civil engineering body capable of planning, designing, and constructing infrastructure to tap available water is useful and effective when resources are plentiful and management is not a strong requirement. In the later phases of basin closure, however, experience shows that large civil engineering organizations (and agricultural or other line agencies) are not well suited to deal with the challenges of basin governance. They have limited experience in political negotiation or interaction with key stakeholders and lack the breadth of experience in dealing with complex, broad-based issues, and multiple values. Further, they often tend to adopt stances based on vested interest in continuing infrastructure development, a position antagonistic to that of stakeholders with ecosystem concerns. Countries that have strong civil engineering organizations reluctant to cede any power will face intense negotiations and struggles before an acceptable form of river basin coordination emerges that is capable of undertaking the key tasks required. However, wherever the scope for construction is reduced and societal values have changed, the trend is likely to follow that of countries such as Australia and the United States, where engineering bodies have contracted and evolved into environmental agencies. Decision makers should not infer from the integrated water resources management message that river basin management needs a strong centralized organization. Basins facing complex problems of conflicting societal values and pressure on resources will probably not be well managed by a single body. Nested or polycentric patterns of basin governance, in which user and community organizations, layers of government, and stakeholder initiatives are coordinated at the basin level, perform better and can be especially effective in settings where participation and democratic practices are well established. Moving toward sustainable river basin management requires much more emphasis on developing, managing, and maintaining collaborative relationships for river basin governance, building on existing organizations, customary practices, and administrative structures.
Acknowledgments The material from this chapter was drawn largely from the Comprehensive Assessment of Water Management in Agriculture which drew together hundreds of researchers and practitioners to find water solutions for tomorrow.
References Adams WM (1985) River basin planning in Nigeria. Applied Geography 5(4): 297--308. Agardy T and Alder J (2005) Coastal systems. In: Millennium Ecosystem Assessment: Vol. 1, Ecosystems and Human Wellbeing: Current State and Trends. Findings of the Conditions and Trends Working Group, pp. 513--550. Washington, DC: Island Press. Alam M and Kabir W (2004) Irrigated agriculture in Bangladesh and possible impact due to inter-basin water transfer as planned by India. Paper presented at the Annual Meeting of the American Society of Agricultural and Biological Engineers, 1–4 August, Ottawa, Canada. Alexander RB, Smith RA, and Schwarz GE (2000) Effect of stream channel size on the delivery of nitrogen to the Gulf of Mexico. Nature 403(6771): 758--761. Alexandratos N (2005) Countries with rapid population growth and resource constraints: Issues of food, agriculture, and development. Population and Development Review 31(2): 237. Anderies JM, Cumming G, Janssen M, et al. (2001) A resilience centered approach for engaging stakeholders about regional sustainability: An example from the Goulburn Broken catchment in Southeastern Australia. Technical Report. Canberra: Commonwealth Scientific and Industrial Research Organisation Sustainable Ecosystems. Ashley S, Holden S, and Bazeley P (1999) Livestock in Poverty-focused Development. Crewkerne: Livestock in Development. Barham E (2001) Ecological boundaries as community boundaries: The politics of watersheds. Society and Natural Resources 14(3): 181--191. Barnett TP, Adam JC, and Lettenmaier DP (2005) Potential impacts of a warming climate on water availability in snow-dominated regions. Nature 438(17): 303--309. Baron JS, Hartman MD, Kittel TGF, Band LE, Ojima DS, and Lammers RB (1998) Effects of land cover, water redistribution, and temperature on ecosystem processes in the South Platte Basin. Ecological Applications 8(4): 1037--1051. Barrows CJ (1998) River basin development planning and management: A critical review. World Development 26(1): 171--186. Bates BC, Kundzewicz ZW, Wu S, and Palutikof JP (2008) Climate change and water. Technical Paper of the Intergovernmental Panel on Climate Change, 210pp. Geneva: IPCC Secretariat. Bennett EM, Carpenter SR, and Caraco NF (2001) Human impact on erodable phosphorus and eutrophication: A global perspective. BioScience 51(3): 227--234. Bhatnagar PR, Sikka AK, Prassad LK, Sharma BR, and Singh SR (2004) Water quality dynamics in secondary reservoir for intensive fish production. Abstract 176. Paper presented at International Association of Theoretical and Applied Limnology (SIL) Congress. Lahti, Finland, 8–14 August 2004. Bijay-Singh Sadana US and Arora BR (1991) Nitrate pollution of ground water with increasing use of nitrogen fertilizers and animal wastes in the Punjab, India. Indian Journal of Environmental Health 33: 57--67. Boelee E (1999) Irrigation Ecology of Schistosomiasis: Environmental Control Options in Morocco. PhD Thesis, Wageningen University, Wageningen, The Netherlands. Boelee E and Laamrani H (2004) Environmental control of schistosomiasis through community participation in a Moroccan oasis. Tropical Medicine and International Health 9(9): 997--1004. Bouman B, Barker R, Humphreys E, et al. (2007) Rice: Feeding the billions. In: Molden D (ed.) Water for Food, Water for Life: A Comprehensive Assessment of Water Management in Agriculture, pp. 515--549. London, UK: Earthscan; Colombo, Sri Lanka: International Water Management Institute (IWMI). Bouman BAM, Castan˜eda A, and Bhuiyan SI (2002) Nitrate and pesticide contamination of groundwater under rice-based cropping systems: Evidence from the Philippines. Agriculture, Ecosystems and Environment 92(2–3): 185--199. Bouman BAM, Hengsdijk H, Hardy B, Bindraban PS, Tuong TP, and Ladha JK (2003) Water-Wise Rice Production. Los Ban˜os: International Rice Research Institute. Briggs SV and Taws N (2003) Impacts of salinity on biodiversity – clear understanding or muddy confusion? Australian Journal of Botany 51(6): 609--617.
Water Availability and Its Use in Agriculture
Brown C and King J (2000) A Summary of the DRIFT Process. Mowbray, South Africa: Southern Waters Ecological Research and Consulting Pty. Bruinsma J (2003) World Agriculture: Towards 2015/2030. An FAO Perspective. London: FAO and Earthscan. Carruthers I, Rosegrant MW, and Seckler D (1997) Irrigation and food security in the 21st century. Irrigation and Drainage Systems 11: 83--101. Carson R (1962) Silent Spring. Boston, MA: Houghton Mifflin. Chambers R (1988) Managing Canal Irrigation: Practical Analysis from South Asia. New York: Cambridge University Press. Chase TN, Pielke RA, Kittel TGF, Baron JS, and Stohlgren TJ (1999) Potential impacts on Colorado Rocky Mountain weather due to land use changes on the adjacent Great Plains. Journal of Geophysical Research – Atmospheres 104(D14): 16673--16690. Coates D (2002) Inland Capture Fishery Statistics of Southeast Asia: Current Status and Information Needs, RAP Publication 2002/11. Bangkok: Food and Agriculture Organization, Regional Office for Asia and the Pacific, Asia-Pacific Fishery Commission. CA (Comprehensive Assessment of Water Management in Agriculture) (2007) Water for Food, Water for Life: A Comprehensive Assessment of Water Management in Agriculture. London: Earthscan. Colombo, Sri Lanka: International Water Management Institute. Coughanowr C (1994) Ground water. Water-related issues of the humid tropics and other warm humid regions. IHP Humid Tropics Programme Series 8. Paris: United Nations Educational, Scientific and Cultural Organization. Daily GC, Alexander S, Ehrlich PR, et al. (1997) Ecosystem Services: Benefits Supplied to Human Societies by Natural Ecosystems. Washington, DC: Island Press. Davies BR and Day JA (1998) Vanishing Waters. Cape Town: UCT Press. de Fraiture C, Wichelns D, Rockstrom J, et al. (2007) Looking ahead to 2050: Scenarios of alternative investment approaches. In: Molden D (ed.) Water for Food, Water for Life: A Comprehensive Assessment of Water Management in Agriculture, pp. 91--145. London, UK: Earthscan; Colombo, Sri Lanka: International Water Management Institute (IWMI). Doll P and Fiedler K (2008) Global-scale modeling of groundwater recharge. Hydrology and Earth System Sciences 12: 863--885. Do¨ll P and Siebert S (2002) Global modeling of irrigation water requirements. Water Resources Research 38(4): 1037. Dries I (1989) Development of wetlands in Sierra Leone: Farmers’ rationality opposed to government policy. In: Marchand M and Udo HA (eds.) The People’s Role in Wetland Management, Proceedings of the International Conference in Leiden, pp. 833–843. Leiden, The Netherlands, 5–8 June 1989. Leiden: Centre for Environmental Studies of Leiden University. Dugan P, Sugunan VV, Welcomme RL, et al. (2007) Inland fisheries and aquaculture. In: Molden D (ed.) Water for Food, Water for Life: A Comprehensive Assessment of Water Management in Agriculture, pp. 459--483. London, UK: Earthscan; Colombo, Sri Lanka: International Water Management Institute (IWMI). Dyson M, Bergkamp G, and Scanlon J (2003) The Essentials of Environmental Flows. Gland, Switzerland: World Conservation Union. Edmunds D and Wollenberg E (2001) A strategic approach to multistakeholder negotiations. Development and Change 32(2): 231--253. Elder S and Schmidt D (2004) Global employment trends for woman, 2004. In: Employment Strategy Papers. Geneva: International Labour Organisation. Ersado L (2005) Small-scale irrigation dams, agricultural production, and health: Theory and evidence from Ethiopia. In: Policy Research Working Paper 3494. Washington, DC: World Bank. Euroconsult (1993) Fayoum Weed Control Project Completion Report. Arnhem: FAO. Falconer IR (2001) Toxic cyanobacterial bloom problems in Australian waters: Risks and impacts on human health. Phycologia 40(3): 228--233. Falkenmark M, Finlayson M, Gordon LJ, et al. (2007) Agriculture, water, and ecosystems: Avoiding the costs of going too far. In: Molden D (ed.) Water for Food, Water for Life: A Comprehensive Assessment of Water Management in Agriculture, pp. 233--277. London, UK: Earthscan; Colombo, Sri Lanka: International Water Management Institute (IWMI). Falkenmark M and Molden D (2008) Special Issue: Closed Basins – Highlighting a Blindspot. Wake up to realities of river basin closure. International Journal of Water Resources 24(2): 201--215. Falkenmark M and Lannerstad M (2005) Consumptive water use to feed humanity – curing a blind spot. Hydrology and Earth System Sciences 9(1/2): 15--28. FAO (2003) World Agriculture Towards 2015/2030. http://www.fao.org/documents/ show_cdr.asp?url_file ¼ /docrep/004/y3557e/y3557e00.htm (accessed April 2010). FAO (2004) The State of Food Insecurity in the World 2004. Rome: FAO. FAO (2005) Irrigation in Africa in Figures. AQUASTAT Survey – 2005, FAO Water Report 29. Rome: FAO.
729
FAO (2006a) AQUASTAT database. Rome. http://www.fao.org/ag/aquastat (accessed April 2010). FAO (2006b) FAOSTAT database. Rome. http://faostat.fao.org (accessed April 2010). FAO (2008) Scoping agriculture–wetland interactions: Towards a sustainable multipleresponse strategy. FAO Water Reports 33. Rome: Food and Agriculture Organization of the United Nations. FAO and UNESCO (2003) Digital soil map of the world and derived soil properties, Rev. 1. Rome: FAO. http://www.fao.org/AG/agl/agll/dsmw.htm (accessed April 2010). Farrington P and Salama RB (1996) Controlling dryland salinity by planting trees in the best hydrogeological setting. Land Degradation and Development 7(3): 183--204. Faures JM, Svendsen M, Turral H, et al. (2007) Reinventing irrigation. In: Molden D (ed.) Water for Food, Water for Life: A Comprehensive Assessment of Water Management in Agriculture, pp. 353--394. London, UK: Earthscan; Colombo, Sri Lanka: International Water Management Institute (IWMI). Fenwick A, Cheesmond AK, Kardaman M, Amin MA, and Manjing BK (1982) Schistosomiasis among labouring communities in the Gezira irrigated area, Sudan. Journal of Tropical Medicine and Hygiene 85: 3--11. Finlayson CM and D’Cruz R (2005) Inland water systems. In: Hassan R, Scholes R, and Ash N (eds.) Ecosystems and Human Well-Being: Current State and Trends: Findings of the Condition and Trends Working Group. Washington, DC: Island Press. Finlayson CM, Davidson NC, Spiers AG, and Stephenson NJ (1999) Global wetland inventory – status and priorities. Marine and Freshwater Research 50: 717--727. Foley J, De Fries R, Asner GP, et al. (2005) Global consequences of land use. Science 309: 570--574. Fu C (2003) Potential impacts of human-induced land cover change on East Asia Monsoon. Global and Planetary Change 37(3–4): 219--229. Galloway JN, Dentener FJ, Capone DG, et al. (2004) Global and regional nitrogen cycles: Past, present and future. Biogeochemistry 70(2): 153--226. Gammelsrod T (1992) Variation in shrimp abundance on the Sofala Bank, Mozambique, and its relation to the Zambezi River runoff. Estuarine, Coastal and Shelf Science 35(1): 91--103. Gichuki F (2004) Managing the externalities of dry season river flow: A case study from the Ewaso Ngiro North River Basin, Kenya. Water Resources Research 40(8): W08S03. Gordon LJ, Steffen W, Jo¨nsson BF, Folke C, Falkenmark M, and Johannesen A˚ (2005) Human modification of global water vapor flows from the land surface. Proceedings of the National Academy of Sciences of the United States of America 102(21): 7612--7617. Gowing JW, Tuong TP, and Hoanh CT (2006) Land and water management in coastal zones: Dealing with agriculture–aquaculture–fishery conflicts. In: Hoanh CT, Tuong TP, Gowing JW, and Hardy B (eds.) Environment and Livelihoods in Tropical Coastal Zones: Managing Agriculture–Fishery–Aquaculture Conflicts. Wallingford and Colombo, Sri Lanka: CABI Publishing and International Water Management Institute. Green P, Vo¨ro¨smarty CJ, Meybeck M, Galloway J, and Peterson BJ (2004) Preindustrial and contemporary fluxes of nitrogen through rivers: A global assessment based on typology. Biogeochemistry 68(1): 71--105. Gupta MV, Sollows JD, Maxid MA, Rahman M, and Hussain MG (1999) Integration of aquaculture with rice farming in Bangladesh: Feasibility and economic viability – its adoption and impact. Technical Report 55. Penang: World Fish Center. Gupta R and Gangopadhyay SG (2006) Peri-urban agriculture and aquaculture. Economic and Political Weekly 41(18): 1757--1760. Gupta SK and Deshpande RD (2004) Water for India in 2050: First-order assessment of available options. Current Science 86(9): 1216--1224. Hall SJ (2002) The continental shelf benthic ecosystem: Current status, agents for change and future prospects. Environmental Conservation 29(3): 350--374. Haylor GS (1994) Fish production from engineered water systems in developing countries. In: Muir JF and Roberts RJ (eds.) Recent Advances in Aquaculture V. London: Blackwell Science. Haylor GS (1997) Aquaculture systems research: Participatory research projects involving fish production in agro-ecosystems in Asia. Aquaculture News 23: 18--20. He C, Cheng S-K, and Luo Y (2005) Desiccation of the Yellow River and the south water northward transfer project. Water International 30(2): 261--268. Hendry K, Sambrook H, Underwood C, Waterfall R, and Williams A (2006) Eutrophication of Tamar Lakes (1975–2003): A case study of land-use impacts, potential solutions and fundamental issues for the water framework directive. Water and Environment Journal 20(3): 159--168. Hope RA (2006) Water, workfare, and poverty: The impact of the working for water programme on rural poverty reduction. Environment, Development, and Sustainability 8(1): 139--156.
730
Water Availability and Its Use in Agriculture
Hussain I, Raschid L, Hanjra MA, Marikar F, and van der Hoek W (2001) A framework for analyzing socioeconomic, health and environmental impacts of wastewater use in agriculture in developing countries. In: IWMI Working Paper 26. Colombo, Sri Lanka: International Water Management Institute. Hussain I, Raschid L, Hanjra MA, Marikar F, and van der Hoek W (2002) Wastewater use in agriculture: Review of impacts and methodological issues in valuing impacts. In: IWMI Working Paper 37. Colombo, Sri Lanka: International Water Management Institute. IPCC (2001) Climate Change 2001. IPCC Third Assessment Report. Cambridge: Cambridge University Press. IUCN (2003) The Pangani River Basin. A Situation Analysis. Gland, Switzerland: IUCN. Jaspers FGW (2001) The new water legislation of Zimbabwe and South Africa – comparison and legal and institutional reform. International Environmental Agreements: Politics, Law, and Economics 1(3): 305--325. Jehangir W, Mudasser M, and Ali N (2000) Domestic uses of irrigation water and its impact on human and livestock health in Pakistan. In: Proceedings of the US Committee on Irrigation and Drainage. Boulder, CO, USA, June. Junk W (2002) Long-term environmental trends and the future of tropical wetlands. Environmental Conservation 29(4): 414--435. Kabat P, Claussen M, Dirmeyer PA, et al. (2004) Vegetation, Water, Humans and the Climate: A New Perspective on an Interactive System. Berlin: Springer. Keiser J, Maltese MF, Erlanger TE, et al. (2005a) Effect of irrigated rice agriculture on Japanese encephalitis, including challenges and opportunities for integrated vector management. Acta Tropica 95(1): 40--57. Keiser J, Singer BH, and Utzinger J (2005b) Reducing the burden of malaria in different eco-epidemiological settings with environmental management: A systematic review. The Lancet Infectious Diseases 5(11): 695--708. Khan S, Tariq R, Yuanlai C, and Blackwell J (2006) Can irrigation be sustainable? Agricultural Water Management 80(1–3): 87--99. Kolding J and van Zwieten PAM (2006) Improving productivity in tropical lakes and reservoirs. Challenge Program on Water and Food – Aquatic Ecosystems and Fisheries Review Series 1. Theme 3 of CPWF, C/o WorldFish Center, Cairo, Egypt, 64 pp. ISBN: 977-17-3087-8. Kurukulasuriya P, Mendelsohn R, Hassan R, et al. (2006) Will African agriculture survive climate change? World Bank Economic Review 20(3): 367--388. Laamrani H, Khallaayoune K, Laghroubi M, et al. (2000) Domestic use of irrigation water: The Metfia in Central Morocco. Water International 25(3): 410--418. Laamrani H and Boelee E (2002) Roˆle des parame`tres de conception, de gestion et de maintenance des perimeters irrigue´s dans la transmission et la lutte contre la bilharziose au Maroc central. Cahiers Agricultures 11(1): 23--29. Landefeld M and Bettinger J (2005) Water effects on livestock performance. Fact Sheet ANR-13-02. Columbus, OH: Ohio State University, Agriculture and Natural Resources. Lehner B and Do¨ll P (2004) Development and validation of a global database of lakes, reservoirs and wetlands. Journal of Hydrology 296(1–4): 1--22. Lemos MC and Oliveira JLF (2004) Can water reform survive politics? Institutional change and river basin management in Ceara´, Northeast Brazil. World Development 32(12): 2121--2137. Llamas R and Custodio E (2003) Intensive Use of Groundwater – Challenges and Opportunities. Leiden: Balkema. Lukatelich RJ and McComb AJ (1986) Nutrient levels and the development of diatom and blue-green algal blooms in a shallow Australian estuary. Journal of Plankton Research 8(4): 597--618. Magadza CHD (1995) Special problems in lakes/reservoir management in Tropical Southern Africa. Presented at the 6th International Conference on the Conservation and Management of Lakes. 23–27 October 1995. Tsukuba City: University of Tsukuba, Japan. Marland G, Pielke RA, Apps M, et al. (2003) The climatic impacts of land surface change and carbon management, and the implications for climate-change mitigation policy. Climate Policy 3(2): 149--157. Martinez-Cortina L and Hernandez-Mora N (2003) The role of groundwater in Spain’s water policy. International Water Resources Association 28(3): 313--320. Matsuno Y, Konradsen F, Tasumi M, van der Hoek W, Amerasinghe FP, and Amerasinghe PH (1999) Control of malaria mosquito breeding through irrigation water management. Water Resources Development 15(1/2): 93--105. McCartney M, Boelee E, Cofie O, Amerasinghe F, and Mutero C (2005) Agricultural Water Development in Sub-Saharan Africa: Planning and Management to Improve the Benefits and Reduce the Environmental and Health Costs. Final Report (Health). Investments in Agricultural Water Management in Sub-Saharan Africa: Diagnosis of Trends and Opportunities Project. International Water Management Institute, Colombo. MEA (2005a) Ecosystems and Human Well-Being: Health Synthesis. Washington, DC: World Resources Institute.
MEA (2005b) Ecosystems and Human Well-Being: Wetlands and Water Synthesis. Washington, DC: World Resources Institute. MEA (Millennium Ecosystem Assessment) (2005c) Ecosystems and Human Wellbeing: Synthesis. Washington, DC: Island Press. Meinzen-Dick R (1997) Valuing the multiple uses of water. In: Kay M, Franks T, and Smith L (eds.) Water: Economics, Management and Demand. London: E&FN Spon. Merrey DJ (1997) Expanding the Frontiers of Irrigation Management Research: Results of Research and Development at the International Irrigation Management Institute 1984–1995. Colombo, Sri Lanka: International Water Management Institute. Merrey DJ, Meinzen-Dick R, Mollinga PP, et al. (2007) Policy and institutional reform: The art of the possible. In: Molden D (ed.) Water for Food, Water for Life: A Comprehensive Assessment of Water Management in Agriculture, pp. 193--231. London, UK: Earthscan; Colombo, Sri Lanka: International Water Management Institute (IWMI). Meybeck M and Ragu A (1997) Presenting the GEMS-GLORI, a compendium of world river discharge to the oceans. In: Webb B (ed.) Freshwater Contamination. Proceedings of a Symposium held during the Fifth IAHS Scientific Assembly at Rabat. Morocco, April–May 1997. Publication 243. Wallingford: International Association of Hydrological Sciences. Millington P (2000) River Basin Management: Its Role in Major Water Infrastructure Projects. Thematic Review V. 3. Cape Town: World Commission on Dams. Minhas PS and Samra JS (2004) Wastewater Use in Peri-Urban Agriculture: Impacts and Opportunities. Karnal, India: Central Soil Salinity Research Institute. Mirza MMQ (1998) Diversion of the Ganges water at Farakka and its effects on salinity in Bangladesh. Environmental Management 22(5): 711--722. Molden D, Frenken K, Barker R, et al. (2007) Trends in water and agricultural development. In: Molden D (ed.) Water for Food, Water for Life: A Comprehensive Assessment of Water Management in Agriculture, pp. 57--89. London, UK: Earthscan; Colombo, Sri Lanka: International Water Management Institute (IWMI). Molden D, Oweis TY, Pasquale S, et al. (2007) Pathways for increasing agricultural water productivity. In: Molden D (ed.) Water for Food, Water for Life: A Comprehensive Assessment of Water Management in Agriculture, pp. 279--310. London, UK: Earthscan; Colombo, Sri Lanka: International Water Management Institute (IWMI). Molden DJ, Sakthivadivel R, and Habib Z (2000) Basin-level use and productivity of water: Examples from South Asia. Research Report 49. Colombo, Sri Lanka: International Water Management Institute. Molle F and Berkoff J (2006) Cities versus agriculture: Revisiting intersectoral water transfers. Potential gains and conflicts. Comprehensive Assessment of Water Management in Agriculture 10. Colombo, Sri Lanka: International Water Management Institute. Molle F, Wester P, Hirsch P, et al. (2007) River basin development and management. In: Molden D (ed.) Water for Food, Water for Life: A Comprehensive Assessment of Water Management in Agriculture, pp. 585--625. London, UK: Earthscan; Colombo, Sri Lanka: International Water Management Institute (IWMI). Moriarty P, John B, and van Koppen B (2004) Beyond Domestic: Case Studies on Poverty and Productive Uses of Water at the Household Level, IRC Technical Papers Series 41. Delft: International Water and Sanitation Centre, Natural Resources Institute, and International Water Management Institute. Murray F, Little DC, Haylor G, Felsing M, Gowing J, and Kodithuwakku SS (2002) A framework for research into the potential for integration of fish production in irrigation systems. In: Edwards P, Little DC, and Demaine H (eds.) Rural Aquaculture. Wallingford: CABI Publishing. Mutero CM, Blank H, Konradsen F, and van der Hoek W (2000) Water management for controlling the breeding of Anopheles mosquitoes in rice irrigation schemes in Kenya. Acta Tropica 76: 253--263. Nemani RR, Running SW, Pielke RA, and Chase TN (1996) Global vegetation cover changes from coarse resolution satellite data. Journal of Geophysical Research 101(D3): 7157--7162. Neumann C, Bwibo N, Murphy S, et al. (2003) Animal source foods improve dietary quality, micronutrient status, growth and cognitive function in Kenyan School Children: Background, study design and baseline findings. Journal of Nutrition 133(11): 3941S--3949S. Nguyen-Khoa S, Smith L, and Lorenzen K (2005) Impacts of irrigation on inland fisheries: Appraisals in Laos and Sri Lanka. Comprehensive Assessment of Water Management in Agriculture, Research Report 7. Colombo, Sri Lanka: Comprehensive Assessment Secretariat. North American Waterfowl Management Plan (2004) Strengthening the Biological Foundation. Ottawa, ON; Washington, DC; and Mexico City: Environment Canada; US Department of the Interior, Fish and Wildlife Services; and Secretaria de Medio Ambente Recursos Naturales. OECD (1996) Guidelines for Aid Agencies for Improved Conservation and Sustainable Use of Tropical and Subtropical Wetlands. Paris: OECD.
Water Availability and Its Use in Agriculture
Omari CK (1993) Social and cultural values of wetlands in Tanzania. In: Kamukala GL and Crafter S (eds.) Wetlands of Tanzania, pp. 95--102. Gland, Switzerland: IUCN. Oweis T and Hachum A (2003) Improving water productivity in the dry areas of West Asia and North Africa. In: Kijne JW, Barker R, and Molden D (eds.) Water Productivity in Agriculture: Limits and Opportunities for Improvement. Wallingford and Colombo, Sri Lanka: CABI Publishing and International Water Management Institute. Palmer Jones RW and Mandal MAS (1987) Irrigation groups in Bangladesh. Irrigation Management Network Paper 87/2c, pp. 4–34. London: Overseas Development Institute. Pielke RA, Avissar R, Raupach M, Dolman AJ, Zeng X, and Denning AS (1998) Interactions between the atmosphere and terrestrial ecosystems: Influence on weather and climate. Global Change Biology 4(5): 461--475. Pielke RA, Lee TJ, Copeland JH, Eastman JL, Ziegler CL, and Finley CA (1997) Use of USGS – provided data to improve weather and climate simulations. Ecological Applications 7(1): 3--21. Postel S (1996) Last Oasis: Facing Water Scarcity. New York: WW Norton. Postel S (1998) Water for food production: Will there be enough in 2025? BioScience 48(8): 629--637. Prein M (2002) Integration of aquaculture into crop–animal systems in Asia. Agriculture Systems 71(1): 127--146. Qadir M, Wichelns D, Raschid-Sally L, et al. (2007) Agricultural use of marginalquality water: Opportunities and challenges. In: Molden D (ed.) Water for Food, Water for Life: A Comprehensive Assessment of Water Management in Agriculture, pp. 425--457. London, UK: Earthscan; Colombo, Sri Lanka: International Water Management Institute (IWMI). Rahman MM, Hassan MQ, Islam M, and Shamsad SZKM (2000) Environmental impact assessment on water quality deterioration caused by the decreased Ganges outflow and saline water intrusion in South-Western Bangladesh. Environmental Geology 40(1–2): 31--40. Raschid-Sally L, Carr R, and Buechler S (2005) Managing wastewater agriculture to improve livelihoods and environmental quality in poor countries. Irrigation and Drainage 54(S1): S11--S22. Ren M and Walker HJ (1998) Environmental consequences of human activity on the Yellow River and its Delta, China. Physical Geography 19(5): 421--432. Renwick ME (2001) Valuing water in a multiple-use system – Irrigated agriculture and reservoir fisheries. Irrigation and Drainage Systems 15(2): 149--171. Revenga C, Brunner J, Henniger N, Kassem K, and Payner R (2000) Pilot Analysis of Global Ecosystems, Freshwater Systems. Washington, DC: World Resources Institute. Rijsberman FR (2006) Water scarcity: Fact or fiction? Agricultural Water Management 80(1–3): 5--22. Rijsberman FR and Molden DJ (2001) Balancing water uses: Water for food and water for nature. In: Thematic Background Paper for the International Conference on Freshwater. Bonn, Germany, 3–7 December. Ringler C, Rosegrant M, Cai X, and Cline S (2003) Auswirkungen der zunehmenden Wasserverknappung auf die globale und regionale Nahrungsmittelproduktion. Zeitschrift fu¨r angewandte Umweltforschung (ZAU) 15/16(3–5): 604--619. Rockstro¨m J, Barron J, and Fox P (2003) Water productivity in rain-fed agriculture: Challenges and opportunities for smallholder farmers in drought-prone tropical agroecosystems. In: Kijne JW, Barker R, and Molden D (eds.) Water Productivity in Agriculture: Limits and Opportunities for Improvement. Wallingford and Colombo, Sri Lanka: CABI Publishing and International Water Management Institute. Rockstro¨m J, Hatibu N, Oweis TY, et al. (2007) Managing water in rainfed agriculture. In: Molden D (ed.) Water for Food, Water for Life: A Comprehensive Assessment of Water Management in Agriculture, pp. 315--352. London, UK: Earthscan; Colombo, Sri Lanka: International Water Management Institute (IWMI). Rogers P and Hall A (2003) Effective Water Governance, GWP Technical Committee Background Paper 7. Stockholm: Global Water Partnership. Rosegrant MW, Cai X, and Cline S (2002) World Water and Food to 2025: Dealing with Scarcity. Washington, DC: International Food Policy Research Institute and International Water Management Institute. Saha SK (1979) River basin planning in the Damodar Valley of India. Geographical Review 69(3): 273--287. Savenije HHG (1995) New definitions for moisture recycling and the relation with landuse changes in the Sahel. Journal of Hydrology 167(1): 57--78. Savenije HHG (1996) Does moisture feedback affect rainfall significantly? Physical Chemistry of the Earth 20(5–6): 507--551. Schlager E and Blomquist W (2000) Local communities, policy prescriptions, and watershed management in Arizona, California, and Colorado. Paper presented at
731
The Eighth Conference of the International Association for the Study of Common Property. Bloomington, IN, USA, 31 May–4 June. Seckler D (1996) The New Era of Water Resources Management: From ‘‘Dry’’ to ‘‘Wet’’ Water Savings. Colombo, Sri Lanka: International Irrigation Management Institute (IIMI), iii, 17p. (IWMI Research Report 001/IIMI Research Report 001). Seckler D, Barker R, and Amarasinghe U (1999) Water scarcity in the twenty-first century. International Journal of Water Resources Development 15(1/2): 29--42. Seckler D, Molden D, Amarasinghe U, and de Fraiture C (2000) Water Issues for 2025: A Research Perspective. Colombo, Sri Lanka: International Water Management Institute. Shackleton C, Shackleton S, Netshiluvhi T, Mathabela F, and Phiri C (1999) The direct use value of goods and services attributed to cattle and goats in the sand river catchment, Bushbuckridge. Report ENV-P-C 99003. Pretoria: Council for Scientific and Industrial Research. Shah T (1993) Groundwater Markets and Irrigation Development: Political Economy and Practical Policy. Mumbai: Oxford University Press. Shah T, Burke J, Villholth K, et al. (2007) Groundwater: A global assessment of scale and significance. In: Molden D (ed.) Water for Food, Water for Life: A Comprehensive Assessment of Water Management in Agriculture, pp. 395--423. London, UK: Earthscan; Colombo, Sri Lanka: International Water Management Institute (IWMI). Shah T, DebRoy A, Qureshi AS, and Wang J (2003) Sustaining Asia’s groundwater boom: An overview of issues and evidence. Natural Resources Forum 27(2): 130--140. Sigman M, Whaley S, Kamore M, Bwibo N, and Neumann C (2005) Supplementation Increases Physical Activity and Selected Behaviors in Rural Kenyan Schoolchildren. CRSP Research Brief 05-04-CNP. University of California, Global Livestock Collaborative Research Support Program, Davis, CA, USA. Smakhtin V, Revenga C, and Do¨ll P (2004) A pilot global assessment of environmental water requirements and scarcity. Water International 29(3): 307--317. Smakhtin VU and Shilpakar RL (2005) Planning for environmental water allocations: An example of hydrology based assessment in the East Rapti River, Nepal. Research Report 89. Colombo, Sri Lanka: International Water Management Institute. Snaddon CD, Davies BR, and Wishart MJ (1999) A global overview of inter-basin water transfer schemes, with an appraisal of their ecological, socio-economic and sociopolitical implications, and recommendations for their management. Water Research Commission Technology Transfer Report TT 120/00. Pretoria: Water Research Commission. Soerjani M (1992) Utilisation of wetland plant resources by rice farmers in Indonesia. In: Maltby E, Dugan PJ, and Lefeuvre JC (eds.) Conservation and Development: The Sustainable Use of Wetland Resources, Proceedings of the Third International Wetlands Conference, Rennes, France, 19–23rd September 1988, pp. 21--29. Gland, Switzerland: IUCN. Struckmeier W, Rubin Y, and Jones JAA (2005) Groundwater – a Reservoir for Thirsty Planet? Leiden: Earth Science for Society Foundation. Svendsen M and Wester P (2005) Managing river basins: Lessons from experience. In: Svendsen M (ed.) Irrigation and River Basin Management: Options for Governance and Institutions. Wallingford: CABI Publishing. Taylor GJ, Baird DJ, and Soares AMVM (2002) Ecotoxicology of contaminants in tropical wetlands: The need for greater ecological relevance in risk assessment. SETAC Globe 3(1): 27--28. Thenkabail PS, Biradar CM, Turral H, et al. (2006) An Irrigated Area Map of the World (1999) Derived from Remote Sensing. Research Report # 105, pp. 74. International Water Management Institute. Trang DT, van der Hoek W, Cam PD, Vinh KT, Hoa NV, and Dalsgaard A (2006) Low risk for Helminth infection in wastewater-fed rice cultivation in Vietnam. Journal of Water and Health 4(3): 321--331. Trenberth KE (1999) Conceptual framework for changes of extremes of the hydrological cycle with climate change. Climatic Change 42(1): 327--339. UNDP (2004) Water governance for poverty reduction; Key issues and the UNDP response to Millennium Development Goals. New York: UNDP. UNEP (2002) Africa Environment Outlook: Past, Present and Future Perspectives. Hertfordshire: Earthprint. UNESCO (2006) Exploitation and Utilization of Groundwater Around the World. CD-ROM. United Nations Educational, Scientific and Cultural Organization International Hydrological Programme and BRGM, Paris and Orle´ans, France. UN (1997) Commission on sustainable development. Comprehensive assessment of the freshwater resources of the world. Report of the Secretary General, 39p. New York. UN (2003) World Population Prospects: The 2002 Revision. New York.
732
Water Availability and Its Use in Agriculture
UN (2006) Water – a Shared Responsibility: The United Nations World Water Development Report 2. Paris and New York: United Nations Educational, Scientific and Cultural Organization and Berghahn Books. Van den Brink PJ, Sureshkumar SN, Daam MA, et al. (2003) Environmental and human risks of pesticide use in Thailand and Sri Lanka: Results of a preliminary risk assessment. Alterra-rapport 789. MAMAS Report Series No. 3/2003. Green World Research. Wageningen: Alterra. Van der Hoek W, Sakthivadivel R, Renshaw M, Silver JB, Birley MH, and Konradsen F (2001) Alternate wet/dry irrigation in rice cultivation: A practical way to save water and control malaria and Japanese encephalitis? Research Report 47. Colombo, Sri Lanka: International Water Management Institute. Van der Hoek W, Ul Hassan M, Ensink JHJ, et al. (2002) Urban wastewater: A valuable resource for agriculture. A Case Study from Haroonabad, Pakistan. Research Report 63. Colombo, Sri Lanka: International Water Management Institute. Van Koppen B, Moriarty P, and Boelee E (2006) Multiple-use water services to advance the millennium development goals. Research Report 98. Colombo, Sri Lanka: International Water Management Institute. Verdegem MCJ, Bosma RH, and Verreth JAJ (2006) Reducing water use for animal production through aquaculture. Water Resources Development 22(1): 101--113. Verhoeven JTA, Arheimer B, Yin C, and Hefting MM (2006) Regional and global concerns over wetlands and water quality. Trends in Ecology and Evolution 21(2): 96--103. Vo¨ro¨smarty CJ, Le´veˆque C, and Revenga C (2005) Fresh Water. In: Hassan R, Scholes R, and Ash N (eds.) Ecosystems and Human Well-Being: Current State and Trends: Findings of the Condition and Trends Working Group. Washington, DC: Island Press. Vo¨ro¨smarty CJ, Meybeck M, Fekete B, Sharma K, Green P, and Syvitski J (2003) Anthropogenic sediment retention: Major global impact from registered river impoundments. Global and Planetary Change 39(1–2): 169--190. Waalewijn P, Wester P, and van Straaten K (2005) Transforming river basin management in South Africa: Lessons from the lower Komati River. Water International 30(2): 184--196. Walling DE and Fang D (2003) Recent trends in the suspended sediment loads of the world’s rivers. Global and Planetary Change 39(1–2): 111--112. WCD (2000) Dams and Development: A New Framework for Decision-Making. London: Earthscan.
Wester P, Merrey DJ, and de Lange M (2003) Boundaries of consent: Stakeholder representation in river basin management in Mexico and South Africa. World Development 31(5): 797--812. Wester P, Shah T, and Merrey DJ (2005) Providing irrigation services in water scarce basins: Representation and support. In: Svendsen M (ed.) Irrigation and River Basin Management: Options for Governance and Management. Wallingford: CABI Publishing. Wester P and Warner J (2002) River basin management reconsidered. In: Turton A and Henwood R (eds.) Hydro-Politics in the Developing World – a Southern African Perspective, pp. 61--71. Pretoria: African Water Issues Research Unit. Winter TC, Harvey JW, Franke OL, and Alley WM (1998) Ground Water and Surface Water: A Single Resource. U.S. Geological Survey Circular 1139. U.S. Department of the Interior, US Geological Survey. Denver, CO: US Geological Survey. Wollheim WM, Peterson BJ, Deegan LA, et al. (2001) Influence of stream size on ammonium and suspended particulate nitrogen processing. Limnology and Oceanography 46(1): 1--13. Wood A and Dixon AB (2009) Valuing wetlands for livelihoods as the basis for sustainable management: The SAB approach. Striking a Balance: Policy Briefing Note 1. Wetland Action. Wood A and van Helsema GE (2008) Scoping agriculture-wetland interactions: Towards a sustainable multiple response strategy. FAO Water Report 33. Rome: Food and Agriculture Organization of the United Nations. Zhang H (2003) Improving water productivity through deficit irrigation: Examples from Syria, the North China Plain and Oregon, USA. In: Kijne JW, Barker R, and Molden D (eds.) Water Productivity in Agriculture: Limits and Opportunities for Improvement. Wallingford and Colombo, Sri Lanka: CABI Publishing and International Water Management Institute. Zheng X and Eltathir EAB (1998) The role of vegetation in the dynamics of West African monsoons. Journal of Climate 11(8): 2078--2096.
Relevant Websites http://www.maweb.org Millennium Ecosystem Assessment.