Thermal methods in petroleum analysis

Thermal Methods in Petroleum Analysis by. Heinz Kopsch CopyrightoVCH Verlagsgesellschaft mbH, 1995 Heinz Kopsch Therm...

Author: Heinz Kopsch

251 downloads 1854 Views 18MB 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

DOWNLOAD PDF

Thermal Methods in Petroleum Analysis by. Heinz Kopsch CopyrightoVCH Verlagsgesellschaft mbH, 1995

Heinz Kopsch

Thermal Methods in Petroleum Analysis

0VCH VerlagsgesellschaftmbH, D-69451 Weinheim, Federal Republic of Germany, 1995 Distribution: VCH, PO. Box 10 1161, D-69451 Weinheim,Federal Republic of Germany Switzerland:VCH, PO. Box, CH-4020 Basel, Switzerland United Kingdom and Ireland: VCH, 8 Wellington Court, Cambridge CB1 lHZ, United Kingdom USA and Canada: VCH, 220 East 23rd Street, New York, NY 10010-4606,USA Japan: VCH, E k o w Building, 10-9 Hongo 1-chome,Bunkyo-ku,Tokyo 113,Japan ISBN 3-527-28740-X

Heinz Kopsch

Thermal Methods in Petroleum Analysis

4b

VCH

Weinheim . New York . Base1 - Cambridge - Tokyo

Dr. rer. nat. Heinz Kopsch Institut fur Technische Chemie T.U. Clausthal ErzstraBe 18 D-38678 Clausthal-Zellerfeld Germany

This book was carefully produced. Nevertheless, the author and publisher do not warrant the information contained therein to be free of errors. Readers are advised to keep in mind that statements, data, illustrations,procedural details or other items may inadvertently be inaccurate.

Published jointly by VCH Verlagsgesellschaft mbH, Weinheim (Federal Republic of Germany) VCH Publishers, Inc., New York, NY (USA)

Editorial Director: Dr. Barbara Bijck Production Manager: Peter J. Biel The cover photo shows a view of part of the BASF steamcracker in Antwerp. (Courtesy of BASF Aktiengesellschaft Ludwigshafen,Germany)

Library of Congress Card No. applied for British Library Cataloguing-in-PublicationData: A catalogue record for this book is available from the British Library Die Deutsche Bibliothek - CIP-Einheitsaufnahme Kopsch, He& Thermal methods in petroleum analysis / Heinz Kopsch. Weinheim ; New York ; Base1 ; Cambridge ;Tokyo : VCH, 1995 ISBN 3-527-28740-X

0VCH Verlagsgesellschaft mbH, D-69451 Weinheim, Federal Republic of Germany, 1995 Printed on acid-free and low-chlorine paper All rights reserved (including those of translation into other languages). No part of this book may be reproduced in any form - by photoprinting, microfilm, or any other means - nor transmitted or translated into a machine language without written permission from the publishers. Registered names, trademarks, etc. used in this book, even when not specifically marked as such, are not to be considered unprotected by law. Composition:Graph. Werkstatten Lehne GmbH, D-41516 Grevenbroich Printing and Bookbinding: Paderborner Druck Centrum. D-33100 Paderborn Printed in the Federal Republic of Germany

Preface

The monograph Thermal Methods in Petroleum Analysis is based mainly on results of more than twelve years research work on the application of thermoanalytical methods to petroleum and its products during the activities of the author at the German Institute for Petroleum Research. It was very interesting to research the application of well defined physical methods, such as thermogravimetry and differential scanning calorimetry, to the multicomponent systems of petroleum and its products, and to understand the limits of those methods on the one hand and the excellent transferability of the results to technical processes on the other. The diversity of possible applications of thermoanalytical methods to various problems in the petroleum laboratory can only be indicated in this monograph. Many people supported my work, either by active or by indirect help. Thanks are expressed to Mrs. Elvira Falkenhagen, who has been a skilful and reliable assistant for many years, as well as to Dr.-Ing. Maria Nagel, Dr.-Ing. Ulrike Tietz, Mrs. Liliane Varoscic, Mrs. Regina Bosse, Mrs. Gerda Sopalla, and the late Mrs. Heidi Gottschalck. An acknowledgement should be made to the directors of the German Institute for Petroleum Research: Professor Dr. H. H. Oelert, Professor Dr. H.-J. Neumann, and Professor Dr. D. Kessel who granted me maximum independent research capacity. Some parts of the research work were carried out with financial support from the German Association for Research CD (Deutsche Forschungsgemeinschaft). For several years successful and pleasant cooperation was established with colleagues of the University of Belgrade, especially with Professor Dr. D. Skala, Professor Dr. M. Sokic, and Professor Dr. J. A. Jovanovic. Thanks are also expressed to those whose names do not appear in this list. All the companies which supplied me with information as well as with illustrations are likewise acknowledged; their names may be found in the appendix. I hope that this monograph will be of some help to colleagues in both academic and industrial research establishments and will encourage them towards further attempts in the application of thermal methods of analysis, even to chemically non-defined multicomponent systems. The examples presented might represent a stimulation for further experimental work. Heinz Kopsch Oktober 1995

Contents

1

1

Introduction

2

Methods and instrumentation 3

3 3.1 3.1.1 3.1.2 3.1.3 3.1.4 3.1.5 3.2 3.2.1 3.2.2 3.3 3.3.1 3.3.1.1 3.3.1.2 3.3.1.3 3.3.1.4 3.3.2 3.3.2.1

3.3.2.2 3.3.3 3.3.4 3.3.5 3.3.6

Thermal analysis on model substances 15 Thermogravimetry (TGA) 15 Thermogravimetry in an inert atmosphere 15 Simulated distillation 28 Thermogravimetry in an oxidizing atmosphere 38 Isothermal thermogravimetry 45 Experiments using the simultaneous thermal analyzer 47 Differential scanning calorimetry on model substances 54 DSC in an inert atmosphere 54 DSC in an oxidizing atmosphere 63 Reaction kinetics 68 Theoretical basis 68 Method according to ASTM E 698-79 69 Method according to Borchardt and Daniels 70 Method according to Flynn and Wall 72 Method according to McCarthy and Green 74 Kinetic investigations on model substances 75 DSC experiments according to ASTM E 698-79 heat of vaporization of n-alkanes 75 Pyrolysis kinetics according to ASTM E 698-79 82 DSC oxidation kinetics according to ASTM E 698-79 84 Kinetics according to Borchardt and Daniels 89 TGA kinetics according to Flynn and Wall 90 TGA kinetics according to McCarty and Green 94

4 4.1 4.2 4.2.1 4.2.2 4.2.2.1

Thermoanalytical investigations on petroleum und petroleumproducts 97 Crude oils (degasified crudes) 99 Refinery residues 111 Description and characterization of the samples 112 Implementation and evaluation of tests 118 Deviations in thermogravimetry 119

VIII

Contents

Thermogravimetry in an inert atmosphere 123 Directly measured index numbers 131 Derived index numbers 136 Simulated distillation 137 Directly measured index numbers in comparison with the simulated distillation 143 4.2.3.5 Derived index numbers for pracital application 144 4.2.4 Thermogravimetry in air 147 4.2.4.1 Directly measured index numbers 155 4.2.5 Correlations of analytical data with index numbers from thermogravimetry 160 4.2.6 Simulated thermal cracking by TGA 162 4.2.6.1 Index numbers from simulated cracking 164 4.2.6.2 Correlation of index numbers from simulated cracking with analytical data 165 4.2.7 Start temperature of the cracking process in an inert atmosphere 166 4.2.8 Differental scanning calorimetry (DSC) 167 4.2.8.1 Experiments in argon at atmospheric pressure 167 4.2.8.2 Experiments in methane at 10 bar pressure 171 4.2.8.2.1 Reaction enthalpy from tests at 10 bar pressure 174 4.2.8.3 Start temperatures of the cracking process at different pressures 175 4.2.8.4 Correlation of kinetic parameters with analytical data 176 4.2.9 Conclusions from experiments on refinery residues 181 4.2.9.1 Thermogravimetry 181 4.2.9.2 Reaction kinetics 184 4.2.9.3 Correlation of data from thermoanalysis with analytical data 186 4.3 Investigations on bitumen 187 4.3.1 Description and characterization of the samples 189 4.3.2 Thermoanalytical investigations 195 4.3.2.1 Thermogravirnetry in inert gas 195 4.3.2.1.1 Correlation of index number from thermogravimetry with consistency data 202 4.3.2.1.2 Correlation index numbers with analysis data 21 1 4.3.2.2 Thermogravimetry in air 217 4.3.3 Isothermal aging tests by thermogravimetry 225 4.3.4 Differential scanning calorimetry (DSC) 233 4.3.4.1 Test in argon at atmospheric pressure 233 4.3.4.2 Tests in methane at 10 bar pressure 237 4.3.4.3 Tempratures of the cracking process 243 4.3.4.4 Oxidation in air 247 4.3.5 Low temperature behavior of bitumen 258 4.3.6 Conclusions from experiments on bitumen 258 4.3.6.1 Thermogravimetry 258 4.3.6.2 Reaction kinetics 261 4.2.3 4.2.3.1 4.2.3.2 4.2.3.3 4.2.3.4

Contents

IX

Investigations on polymer modified bitumens (PMB) 265 Description and characterization of the samples 265 Thermogravimetry 269 Dynamic (temperature-programed) thermogravimetry 269 Isothermal gravimetq 275 Reaction kinetics using DSC 283 Low temperature behaviour of PMB using DSC 285 Aging properties of polymers for the modification of bitumen 287 Investigation on the hydrocracking reaction of heavy residues 296 Investigation on a vacuum residue from Kirkuk 297 Investigation on residues of different origins 304 Oil shale and shale oil 321 Investigation using TGA and DSC 322 Modelling and simulation of oil shale pyrolysis 331 Fingerprinting of oil shale by oxidation 345 Lubricants 348 Evaporation behavior of lubrication oils 349 Oxidation behavior of lubrication oils 358 Comparison of the oxidation stability of virgin oils, reclaimed oils, and synthetic lubrication oils 365 4.8 Silicone oils 376 4.9 Relation of the kinetics of pyrolysis and oxidation reactions to the system pressure: Investigations on tertiary oil recovery by in situ combustion 400 4.9.1 Pyrolysis tests 405 4.9.2 Oxidation tests 410 4.9.2.1 Range of low temperature oxidation (LTO) 412 4.9.2.2 Range of fuel deposition 415 4.9.2.3 Range of fuel combustion 421 4.9.3 Discussion 424 4.10 Comparison of commercial computer programs for investigation of kinetics of pyrolysis and oxidation reactions of heavy petroleum products 427 4.10.1 Pyrolysis reaction 429 4.10.1.1 Kinetics according to ASTM E 698-79 429 4.10.1.1.1 DSC (DTA) experiments 429 4.10.1.1.2 Kinetics according to ASTM E 698-79 from simultaneous TGA/DTA experiments 439 4.10.1.2 Kinetics according to Borchardt and Daniels 440 4.10.1.3 Kinetics according to Flynn and Wall 442 4.10.1.4 Kinetics according to McCarty and Green 453 4.10.2 Oxidation reaction 458 4.10.2.1 Kinetics according to ASTM E 698-79 460 4.10.2.1.1 DSC (DTA) experiments 460 4.10.2.1.2 Experiments using Simultaneous Thermal Analyzer 467 4.10.2.2 Kinetics according to Borchardt and Daniels 468

4.4 4.4.1 4.4.2 4.4.2.1 4.4.2.2 4.4.3 4.4.4 4.4.5 4.5 4.5.1 4.5.2 4.6 4.6.1 4.6.2 4.6.3 4.7 4.7.1 4.7.2 4.7.3

X

Contents

4.10.2.3 Kinetics according to Flynn and Wall 469 4.10.2.4 Kinetics according to McCarty and Green 473 4.10.3 Conclusions 477

5 5.1 5.2

Final consideration 485 Other applications 485 Summary on progress of instrumentation (hard and software) and advice 487

6

Appendix: Manufacturers of thermoanalytical instrumention 495

References 499 References Chapter 1 References Chapter 2 References Chapter 3 References Chapter 4 References Chapter 5

499 500 500 502 506

List of Symbols

A AR ASTM

Frequency or Pre-exponential Factor (min ') Atmospheric residue American Society for Testing and Materials

BP

Boiling Point ("C)

CCR CR

Conradson Carbon Residue (%) Crackable part of the sample (%)

DDK DIN

Dynamic Difference Calorimetry (see DSC) Deutsches Institut k r Normung e. V. (German Institute for Standardization) Differential Scanning Calorimetry Differential Thermal Analysis Differential Thermogravimetry (First differential quotient of weight loss with respect to time) (% min ')

DSC DTA DTG

Activation Energy (J . Mol-l) Base of natural logarithm Exponent with base e Residual weight at the point of inflexion of the TGA curve (%) Energy flow (pW) Enthalpy of pyrolysis (J . g-1) Heat of fusion (J . g-1) Heat of vaporization (J . g-1) Infrared Spectroscopy Reaction (Rate) Constant (min-') Natural logarithm Decimal logarithm Mean relative particle mass (Mean molecular weight) Melting Point ("C) Non-distillable part of the sample (%) Nuclear Magnetic Resonance Spectroscopy Reaction order (dimensionless)

XI1

List of Symbols

P PCR Pen PMB

Pressure (bar) Practical thermal crackable part of the sample (%) Needle penetration at 25 "C (0.1 mmj Polymer modified Bitumen

Q

Quotient of weight loss in air divided by weight loss in inert gas (Isothermal Gravimetry) Universal Gas Constant (J Mol-' K-') Residue (%) at 600 "C experimental temperature Residue (%) at 800 "C experimental temperature Coefficient of correlation (dimensionless)

R R600 R800 r

SAR Simulated atmospheric residue (%) S.P.R&B Softening Point Ring and Ball ("C) Simultaneous Thermal Analysis (or Analyzer) STA (TGA+DTA or TGA+DSC) Simulated vacuum residue (%) SVR Standard deviation S

512

Temperature (Generally OC, except kinetics with absolute temperature K) Thermal Analysis Start temperature of the thermal crack reaction ("C) Thermogravimetry Temperature of peak maximum ("C) Onset temperature ("C) Temperature of the point of inflexion of the TGA curve ("C) Temperature ("C) at 1 % weight loss Temperature ("C) at 5 % weight loss Time (min) Half life time (minj

U

Conversion (%)

V VR VVR

Coefficient of variation (%j Vacuum residue Visbreaker residue

x

Arithmetic mean

01

Fractional conversion (dimensionless) Heating rate (K min-') Solubility parameter according to Hildebrandt ( ~ M m-3) J Difference Weight loss up to 100 "C (%) Weight loss up to 200 "C (%) Weight loss up to 300 "C (%) Weight loss up to 400 "C (%)

T TA Tcrack TGA Tm Tonset Tw T1 % T5 % t

P

6

A AGlOO AG200 AG300 AG400


1 Introduction

Analytical methods describing the thermal behavior of substances during programmed temperature changes, like thermogravimetry, differential thermoanalysis, or differential scanning calorimetry are old methods, which were applied at first to problems of inorganic chemistry, mainly to minerals. The analysis of petroleum and petroleum products has been mentioned relatively late. In the literature survey by Weselowski [ 1-11 the first citation dates from 1958. Also, the oldest citation in the research report by Kettrup and Ohrbach [1-21 dates from 1965. Petroleum, especially heavy crudes, is recovered sometimes by the use of thermal processes like steam flooding or by in situ combustion. The processing of the recovered crudes in the refineries is usually done by thermal methods at very different temperatures. A review of the temperatures applied in refinery operations is given in Table 1-1. These thermal processes are performed partly by sequential heating until the desired products are obtained. The operating parameters for the different processes have been obtained to a large extent by empirical experience or partly by simulation of the processes in laboratory installations or in pilot plants. For that reason thermoanalytical methods are considered to be very useful in obtaining data concerning the thermal behavior i. e. data describing the Table 1-1: Temperature Ranges in Petroleum Processing Process Atmospheric Distillation Vacuum Distillation Thermal Cracking Catalytic Cracking Steam Cracking High Temperature Pyrolysis Hydrocracking (Gas Phase) Hydrocracking (Liquid Phase) Visbreaking Reforming (Thermal Treating) Reforming (Catalytic Treating) Isomerization Alkylation (Catalytic) Polymerization Hydrotreating Steam Reforming Bitumen Blowing

Temperature Range ("C)

350 . . . 380 350 . . . 380 400 . . . 650 450 . . . 540 650 . . . 1000 1000 340 . . . 430 340 . . . 470 460 . . . 480 510 . . . 580 500 . . . 550 60 . . . 200 0 . . . 200 170 . . . 215 250 . . . 430 700 . . . 800 230 . . . 300

2

1 Introduction

thermal and oxidation stability of petroleum and its products; data predicting the manner and quantity of products gained in the processes; and data concerning reaction kinetics which can be used to optimize the refinery processes. Thermogravimetry (TGA), differential thermoanalysis (DTA), and differential scanning calorimetry (DSC) are the main methods which can be used in the analysis of petroleum and its products. DSC is preferred to DTA, because DSC supplies values of energies directly, whereas the DTA supplies only temperature differences. These thermal methods of analysis have been described in several basic books [l-3 to 1-17]. The application to polymers is described likewise [l-18, 1-19]. So far no compilation on the application to petroleum and its products exists. The situation in the field of standards is similar. The NormenausschuB Materialpriifung im Deutschen Institut fur Normung (Committee for Testing and Materials of the German Institute for Standardization e. V., DIN) has approved only two standards (one of them contains terms of thermal analysis [ 1-20], the other is the standard for thermogravimetry [l-211). Furthermore there are three proposals (principles of differential thermal analysis [1-22], determination of melting temperatures of crystalline material by DTA [l-231, and testing of plastics and elastomers by DSC [ 1-24]). The American Society for Testing and Materials (ASTM) has to date approved forty standards for the application of thermal methods of analysis. Among them, seven standards are concerned with the testing of petroleum and its products [l-251 to [l-321, six standards are general methods [l-321 to [l-381, and four standards concerning the testings of polymers are applicable to petroleum and its products too [l-391 to 11-42].


2 Methods and instrumentation

Using thermogravimetry (TGA), the dependence of the change in sample weight (mass) on the temperature during programmed temperature changes in a chosen gas atmosphere can be measured. The first derivative of the weight (mass) signal with respect to time is called derivative thermogravimetry (DTG) and is a criterion for the reaction rate. It is usual to record both the slope of the weight (mass) versus the time or temperature (TGA), and the differentiatoed curve versus the time or temperature (DTG). The heating rate dictates the actual position of the TGA and DTG graphs; it is therefore advisable always to use the same heating rate ( p ) so that different tests may be compared. For small sample weights (masses), up to approximately 10 rng, a standard heating rate of 10 K/min is practicable. This heating rate is slow enough to avoid any temperature gradient inside the sample while permitting a reasonable utilization of the available workmg time. The shift to higher temperatures of the TGA and DTG curves as a consequence of faster heating rates permits calculation of the Arrhenius kinetic parameters and hence investigation of the reaction kinetics (see chapter 3.3). Furthermore, the position of the TGA and DTG curves will be influenced by the shape of the sample pan, especially by the ratio of surface to volume of the sample, and lastly by the quantity of gas flowing through the oven (gas flow rate). Therefore it is important that variations in sample quantity are minimized and that the gas flow rate is maintained as constant as possible. However, the gas flow rate must not fall below a certain minimum value in order to avoid condensation of evaporated sample fractions on the hangdown of the sample holder or in the gas outlet tubes. The minimum gas flow rate depends on the geometric shape of the oven and the position of the gas inlet and outlet tubes and therefore differs for different instruments. If the gas flow rate is sufficient, the evaporated portions of the sample will be discharged immediately and therefore no equilibrium between liquid and vapor will be attained. As a consequence the boiling (evaporation)temperature of the sample will decrease adequately.That can be used to perform a simulated distillation (see chapter 3.1.2). However, the application of thermoanalytic methods is limited to substances having a start temperature of evaporation at atmospheric pressure not far below 200 "C.Otherwise there is the risk that evaporation in the gas flow will begin at room temperature and thus the correct start temperature of evaporation (zero point of the TGA curve) cannot be ascertained. In principle all except very corrosive gases can be passed through a thermobalance; in practice the inert gases nitrogen, helium, and argon and the reactive gases air, oxygen, and hydrogen will be used. The weight calibration of thermobalances is done using standard weights. The temperature calibration is more difficult. The method using the Curie point temperature, as

4

2 Methods and Instrumentation

described in ASTM E 914-83, does not work if a magnetic field from outside the oven is prevented from reciprocal action with the standard inside the oven, by the construction or the material of the oven. Calibration using calcium oxalate monohydrate for standard is very common, since it has exhibited three clearly-defined steps of weight loss during heating (Fig. 2-1 to 2-3).: Reaction

CaC,O,.H,O CaC,04 CaCO,

-+ CaC,04 + H,O + CaCO, + C o t

t

CaO+CO,T

Temperature Range at p = 10 K/min ("C) 135.. . 175 4 6 3 . . ,502 660. . .740

Residue

DTG Maximum Temperature

(%I

("C) 163 49 1 722

87.7 68.5 38.4

As can be seen from the figures, the DTG maximum is found at conversions which are smaller than the maximum conversion of the reaction step concerned. The onset temperatures as well as the DTG maximum temperatures can be reproduced with coefficients of variation < 2 % of the corresponding mean value. The thermogravimetric experiments are run using open platinum sample pans. Pans made from aluminium, platinum, quartz, glass, stainless steel etc. were also available. The 5

110

0

I00

-5 90

-10

U

:m: 0

no

>

U

L

-15

PI

a

r(

x

l l N T TABLE TGA

70

-20 60

-25 50

-30 TGA

-35

40 __.

100

too

300

I

500

4bO Deg

Fig. 2-1: Thermogravimetry of CaC,O,. H,O Plot of STA 780: TGA and DTA Atmosphere: Argon 30 + 20 cm3/min Heating Rate p: 10 K/min

c

I

600

I

700

I

800

0

k

L

m

a 70

-

60

-

50

-

.3-

4

0

.2-

4l

m .,

.i0.0

-

-.l

40

L.--y.r 0

100

200

-

--

----.

300

400

600

500

800

700

-.2

-

9 900

Deg C

Fig. 2-2: Thermogravimetry of CaC,O,. H,O Plot of STA 780: TGA and DTG Atmosphere: Argon 30 + 20 cm3 Heating Rate p: 10 K/min/min 110

-

100

-

90

-

80

-

70

-

60

-

50

-

40

-

U

;

136.13

c

171.08 C

463.21 C

0

L

m

a

0

601.36

100

200

I

300

1

400

I

500

C

661.26 C

600

I

700

800

I0

6


catalytic effect of the pan material on the pyrolysis reaction could not be ascertained when comparing the reaction in platinum and quartz pans, however, it could not be completely excluded. All thermogravimetric experiments carried out by the author were run in platinum pans. Argon was used as the inert atmosphere. Oxidation experiments were run in air because the reactions are too fast in oxygen. The first stage of experiments was carried out using a Stanton-Redcroft TG 750 thermobalance connected to a three-pen recorder, recording weight (mass) loss (TGA), derivative thermogravimetry (DTG), and temperature (q. For documentation the graphs of weight (mass) versus temperature were drawn manually. Later on, the experiments were performed using a simultaneous thermal analyzer Stanton-Redcroft STA 780 (STA 1 000), which is equipped with a personal computer for control, data sampling, and data evaluation (Table 2-1). Using this device the curves of TGA, DTG, and DTA (differential thermal analysis) versus temperature can be plotted. Furthermore, the PC is equipped with extensive software to evaluate the results under varying conditions. Table 2-1: Thermobalances Instrument: System: Pressure Range: Heating Rates: Recording: Evaluation:

TG 750 Stanton-Redcroft TGA + DTG up to 1000°C normal pressure and vacuum 0.5 . . . 100 K/min 3 pen recorder manually TGA empirical index numbers evaporation pyrolysis oxidation simulated distillation DTG empirical index numbers kinetics according to ASTM E 698-79

Instrument: System: Pressure Range Heating Rates: Recording: Evaluation:

STA 780 Stanton-Redcroft (STA 1000) TGA + DTG + DTA simultaneous up to 1000°C normal pressure and vacuum 0.5 . 50 K/min PC PC TGA empirical index numbers evaporation pyrolysis oxidation simulated distillation kinetics according to Flynn & Wall kinetics according to McCarty & Green DTG empirical index numbers kinetics according to ASTM E 698-79 DTA specific heat conversion temperatures kinetics according to Borchardt & Daniels kinetics according to ASTM E 698-79


7

6‘

-5

i

1-

i

i

2a-

I t4

2-

1

312-

GA S ..tN

WALE! IN

B-‘

-7

-13 9-

WATER OUT

10-

\ FURNACE LIFTING SYSTEM’

11

Fig. 2-4: Diagram of the Thermobalance Stanton-Redcroft TG 750 1 Balance glass housing 8 Cooling water flow meter 9 Furnace 2 Glass protection tube 2a Brackets 10 Furnace lifting system 3 Glass protection tube 11 Spirit level 4 Counter weight glass housing 12 Support for glass protection tube 5 Gas inlet 13 Lower cover 6 Protection lid (Figure by Stanton-Redcroft Ltd.) 7 Gas flow meter

8


A schematic digaram of the TG 750 is shown in Fig. 2-4, of the STA 780 in Fig. 2-5. The recorder script of an experiment with a hydrocarbon using the TG 750 is depicted schematically in Fig. 2-6. Curve I represents the weight (mass) signal (TGA), curve I1 that of the first derivative (DTG), and curve I11 the temperature (T> of the thermocouple directly below the sample pan. Point A marks the start of the weight (mass) loss 1 % and the corresponding temperature T1 %; point B is the weight (mass) loss 5 % and the corresponding temperature T5 %. Point C corresponds to the weight (mass) loss at 400°C (AG400). This is the temperature limit of the thermal stability of most non-aromatic hydrocarbons and of the heterocompounds. Point D marks the weight (mass) of the coked residue at 600 "C (R600) or at 800 "C (R800). Point E represents the maximum of the DTG curve

Fig. 2-5: Cross-Section of Water-cooled Furnance for STA 1 000 (STA780) A Water cooled cold finger B Ceranuc baffles C Ceramic tube D Micro-enviromental cup E Ceramic stem gas inlet F Furnace winding G STA hangdown assembly (Figure by Rheometnc Scientific, Polymer Laboratories GmbH)


E

9

D

Fig. 2-6: Schematic Diagram of Recorder Diagram of a Test in Protecting Gas by means of TG 750. I TGA signal II DTG signal I11 Temperature signal A Start of weight loss (T1 %) B Start of weight loss (T5 %) C Weight loss up to 400 "C (AG400) Residue at 600 "C (R600) or at 800 "C (R800) D E Maximum of DTG curve (T-)

with the corresponding temperature T-. The amplitude of the DTG curve corresponds to the reaction rate. The temperature of the DTG maximum shows whether the reaction remains in the evaporation (distillation) range (Tmx< 400 "C) or if a pyrolysis (cracking) reaction has occurred (T- > 400 "C). An example of rescaling the plot of weight versus time to weight versus temperature is shown in Fig. 2-7. Here, the point of intersection of the tangents (offset point) represents the weight (mass) Gw of generated coke at the temperature Tw at the point of inflexion of the TGA curve. This happens only during experiments in inert gas. Using ash-free substances in experiments in air, a TGA curve passing through zero weight is obtained, while ash-containing substances give a constant residual weight. The DTG graph of the experiment in air always shows more than one maximum, the first of which can represent vaporization as well as oxidation. In this case the TGA graph in protecting gas must be consulted for comparison. Figs. 2-1 to 2-3 demonstrate possible evaluations using the STA 780 in an experiment with calcium oxalate monohydrate. In Fig. 2-1 the TGA curve is evaluated with respect to the weight (mass) losses of 1 %, 5 %, 10 %, and further in 10 % steps, whereas in the DTA curve the peak maximum temperature and the corresponding residual weight (mass) are plotted. Fig. 2-2 again shows the TGA and DTG curves with peak maximum temperatures and corresponding residual weights. Fig. 2-3 demonstrates the onset and offset temperatu-

10

2 Methods and Instrumentation Residue ( % I

100..

80

60.

LO

- - - - - - -9___

(c-

z

20.

I

I I Tw I *

100

300

1 500

7M

800

-

res of the three reaction steps. Theoretically all three evaluations could be drawn in one plot, but that would be very difficult to interpret. With the help of differential scanning calorimetry (DSC), events can be observed which are created by energy transfer (take up or delivery) during programmed heating or cooling of a sample, i.e. melting, crystallization, second order transitions, evaporation, pyrolysis, oxidation etc. The energetic effect in the sample is compared to a thermally inert reference substance which undergoes the same temperature programme. The differences between sample and reference in uptake or delivery of energy will be recorded as energy flows versus temperature or time. Using DSC it is likweise possible to differentiatethe resulting data with respect to time (dimensions W/s) or to temperature (dimensionsW/O C). Neither differential quotient has any meaning in the physical sense. They serve only to elucidate effects of the graph of energy flows versus temperature or time. Therefore it is not surprising, that only a very few literature references exist where differentiated curves are described. Using DSC, the position of the energy flows versus temperature curve as well as the rate of an event were influeced by the heating rate, too. Therefore the DSC tests were run likewise, using a standard heating rate p= 10 K/min with the exception of the investiga-


11

tion of reaction kinetics. There the shift of the maxima of the energy flow curve to higher temperatures as a result of increasing heating rates permits the calculation of the Arrhenius lanetic parameters (see Chapter 3.3). All the influences, such as oven geometry, shape of the sample pan, position of gas inlet and gas outlet, on the results of DSC are the same as in TGA. The gas purge with a minimum flow rate is also necessary in DSC to avoid condensations,when petroleum and its products or generally volatile substances are tested. As a consequence, the boiling (evaporation) temperature will decrease in a similar way to that in TGA. The gases used in TGA can be used also in DSC. Some additional experiments have been carried out in methane to study the influence of a hydrocarbon atmosphere. For calibration, the melting point of indium were measured, which has a temperature of fusion (MP) = 156.4"C and a heat of fusion Hf= 28.46 J/g (Fig. 2-8). Because reactions at higher temperatures occur in experiments with petroleum refinery residues, additional calibration runs were performed using pewter (MP = 231.84"C, H,= 59.61 J/g) and lead (MP = 327.40°C, Hf= 26.47 J/g). If a calibration at higher temperatures is necessary, potassium perrhenate KReO, (MP = 550°C, H,= 294.8 J/g) can be used. All DSC tests carried out by the author were run using open aluminium pans. For reference an empty pan were used. Comparative tests with platinum pans gave no indica-

0

Temperature CoC)

Fig. 2-8: Indium Calibration Curve of DSC MP : 156.4 "C Hf : 28.46 J/g

12


tion that the pan material had any influence, neither for pyrolysis nor for oxidation reactions. The first experiments were carried out with the help of a DuPont 990 Thermoanalysis System connected to a 910 DSC.This system used a two pen x-y recorder; the resultant graphs were evaluated manually. Later, a DuPont 9900 Thermoanalysis System was used, which is equipped with a PC for control, data sampling, and data evaluation (Table 2-2). A cross-section of the DSC cell is shown in Fig. 2-9.

Table 2-2: Differential Scanning Calorimetry Instrument: System: Heating Rates: Cooling Rates: Recording: Evaluation:

Fig. 2-9: DSC Cell Cross-Section 1 Gas purge inlet 2 Lid 3 Reference pan 4 Silver ring 5 Furnace winding 6 Furnace block 7 Radiation shield 8 Sample platform

DuPont 9900 Thermal Analysis System DuPont 910 DSC Pressure DSC Cell DSC -75 "C . . . +250 "C, normal pressure DSC RT . . . +650 "C vacuum till bar pressure up to 70 bar 0.5 . . . 50K/min 0.5 . . . 5K/min PC PC specific heat conversion temperatures reaction enthalpy heat of conversions kinetics according to ASTM E 698-79

9 Chromel disk 10 Chromel wire 11 Alumel wire 12 Thermocouple junction 13 Thermoelectric disk (Constantan) 14 Samplepan 15 Lid (Figure by TA Instruments Inc.)


13

Petroleum and its products are multicomponent systems of varying chemical composition. They are predominantly a mixture of hydrocarbons, usually accompanied by a small quantity of heterocompounds which contain in addition to carbon (C) and hydrogen (H) other atoms such as sulfur (S), nitrogen (N), and/or oxygen (0).Metals are present in very small concentrations, such as vanadium and nickel in organically bound forms. The average elementary composition of petroleum in weight-% lies between the following limits [2-11: C 8 3 . . . 87% H11 . . . 14% S 0.01.. . 8 %

0 0 . . .2 % N 0.01 . . . 1.7 % Metals 0 . . . 0.1 %

Petroleum contains four groups of hydrocarbons:

- alkanes (unbranched n- and branched i-alkanes) - cycloalkanes (naphthenes, unsubstituted and substituted) - aromatics (unsubstituted and substituted)

- complex hydrocarbons (naphthenoaromatics) Alkenes (olefins) and alkynes (acetylenes) are not found in petroleum (crude oils). However, they were formed during the processing of petroleum at high temperatures. With regard to the boiling behavior, the full range of substances occur, from those which evaporate early during the recovery, as a result of pressure decrease, through to substances which cannot evaporate without decomposition. It is possible to separate individual cherncally-defined substances from the low boiling fractions. From medium and high boiling fractions and from the non-distillable residues only multicomponent systems can be obtained, which can be separated into groups characterized by a similar chemical and physical behavior. Separation into individual compounds is almost impossible. Under these circumstances, it seems reasonable to study the thermal reactions such as boiling, pyrolysis, and the oxidation behavior of defined model substances first, in order to understand the behavior of petroleum and its main products and to draw some analogous conclusions.


3 Thermal analysis on model substances

3.1 Thermogravimetry (TGA) 3.1.1 Thermogravimetry in an inert atmosphere The series of n-alkanes from n-decane up to n-hexacontane (Table 3-1) and some chemicals mentioned in ASTM D 2887-84 (Table 3-2) were used for modelling purposes. Table 3-1: Boiling Points of n-Alkanes DIN 51435 [3-51 Carbon Number

Boiling Point ("C)

Carbon Number

Boiling point ("C)

10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35

174 196 216 235 253 28 1 287 302 317 331 344 356 369 380 39 1 402 412 422 432 44 1 450 459 468 476 483 49 1

36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

499 506 513 520 527 533 540 546 552 558 564 570 575 581 586 591 597 602 607 612 616 62 1 626 630 635

16

3 Thermal Analysis on Model Substances

Table 3-2: Substaiices according to ASTM D 2887-84 Table X 1.1 [3-61 Number

Substances

1 2 3 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 27 29 30 31

Benzene Toluene p-Xylene Cumene (i-Propylbenzene) 1-Decene sec-Butylbenzene n-Buty lbenzene trans-Decalin &-Decalin 1-Dodecene Naphthalene 2,3-Dihydroindole

1,3,5-Tn-i-propylbenzene 2-Methylnaphthalene 1-Methylnaphthalene Indole Acenaphthene 1-0ctadecene Phenanthrene Anthracene Acndine Triphenylmethane Pyrene 1,3,5-Triphenylbenzene Di-p-tolylsulfone

1,1,2,2-Tetraphenylethylene Palmitic acid methyl ester

BP ("C) 80.2 110.8 138.4 152.5 168 - 170 172 - 173 183.1 187 - 188 193 213.4 218.05 225 - 221 223 - 226 241 - 242 244.6 253 - 254 277.9 314.8 322 342 345 - 346 360 394.8 460 405 415 - 420 415

These chemicals were of the highest purity commercially available. All substances were tested by thennogravimetry at a heating rate p= 10 K/min in an inert gas atmosphere (argon). To avoid condensation the gas flow rate through the TG 750 thermobalance should be 25 cm3/min. Using the STA 780, a gas purge of 30 cm3/min through the oven, and an additional gas flow of 20cm3/min through the side tube of the hangdown protecting cylinder were needed. In the STA 780, the gas flow causes complete vaporization of all the n-alkanes. The STA plot of n-hexacontane is shown in Fig. 3-1 as an example. The chemicals mentioned in Table 3-2 were also completely vaporized, with the exception of 1, 3,5-triphenylbenzene, di-p-tolylsulfone, and palmitic acid methyl ester. The latter start to decompose when about 70-80 % of the sample has been evaporated. The smaller gas flow rate in the TG 750 (25 cm3/min) causes a reduction in the mass loss rate. In consequence, a pyrolysis reaction occurs, sooner or later depending on the purity of the individual substance. Whereas pyrolysis of n-hexacontane starts above 400 "C as expected (Fig. 3-2), the start of pyrolysis of n-hexatriacontane is as low as 330 'C,an unusually low temperature for n-alkanes (Fig. 3-3). The substances which pyrolyze at relatively low temperatures in the STA 780, do the same in the TG 750: i. e. 1, 3, 5-triphenylbenzene

17

3.1 Thermogravimetry (TGA)

2.0

110 1.81

d'

+,

1.e

-

1.6-

c

m

; i.4-

-a m

>

m

a

40 -

-

1.0

-

m

.E-

50

20

1.2

+I

m

n

POINT TABLE TGA

.6

-

.414

-

.2 0.0

0 -

RESIDUE .E X

0

100

1

I

I

200

300

400

Deg

I

500

I

600

,

-.2j

700

C

Fig. 3-1: Thermogravimetry of n-Hexacontane Plot of STA 780: TGA and DTG Atmosphere: Argon, 30 + 20 cm3/min Heating Rate fi: 10 K/min

(start temperature approx. 325 "C), di-p-tolylsulfone (start temperature approx. 275 "C), and palmitic acid methyl ester (start temperature approx. 280 "C). The thermogravimetry of hydrocarbons in inert gas atmosphere can be described generally as follows: At the beginning there is an induction period, the length of which is determined by the evaporation start temperature of the sample. The induction period is followed by the evaporation. The temperature TI % of a weight (mass) loss of 1 % or of 5 % (T5 %) characterizes the initial boiling temperature. As expected, the coefficient of variation k V of the mean X is smaller for T5 % than for T1 %, if a statistical evaluation of at least ten experiments of the same sample had been carried out. The coefficients of variation T1 % and T5 % will also decrease with increasing boiling points (BP) of the samples. The ratio T5 %/T1 % also decreases with increasing boiling point of the sample and attains a constant final value of T5 %/T1 % = 1.143 for boling points above 500 "C (Fig. 3-4). However, a boiling temperature above 500 "C is an extrapolated value because no organic substance can be distilled at such temperatures without undergoing decomposition. For example BP = 500 "C corresponds to the boiling point of n-hexatriacontane at atmospheric pressure. With increasing temperature the slope of the TGA curve becomes steeper and passes through the final weight value zero, unless another chemical reaction occurs in the sample sooner. Samples which disproportionate forming different volatile fragments (pyrolysis), exhibit a point of inflexion in the curve of weight versus temperature. From this

18


Residue (%) 100

Fig. 3-2: Thermogravimetry of n-Hexacontane TG 750 Atmosphere: Argon, 25 cm3/min Heating Rate p: 10 K/min

1


I

10

I I+

100

200

300

Fig. 3-3: Thermogravimetry of n-Hexatricontane TG 750 Atmosphere: Argon, 25 cm3/min Heating Rate p : 10 K/min

19

20


I

Fig. 3-4: Quotient of the Temperatures of 5 % Weight Loss Divided by 1 % Weight Loss versus Boiling Point Temperatures (BP) of n-Alkanes

point the slope of the graph becomes less steep. The temperature at the point of inflexion T , characterizes the pyrolysis behavior of the sample. The slope of the TGA curve from this point of inflexion becomes less step and meets the zeroweight value at relatively high temperatures only if the pyrolysis results in the formation of products of different boiling points. If the pyrolysis produces a coke residue as well as low boiling fragments, then the part of the TGA curve beyond the point of inflexion runs straight, nearly parallel to the temperature axis, demonstrating a constant residual weight (mass). The model substances tested mostly exhibit only one maximum in the curve of weight (mass) loss rate versus temperature (DTG). The maximum temperature will be determined by the boiling behavior of the sample. The same is valid for the DTA curves measured by the Simultaneous Thermal Analyzer (STA 780). Each of the temperatures, which describe the thermal behavior in an inert atmosphere, can be repeated within relatively narrow limits as shown in Table 3-3 by the example of four a-alkanes, and in Table 3-4 for five aroma-


21

Table 3-3: Temperature Statistics ("C) for n-Alkanes -

X

n-Undecane BP = 196 "C T1% 26.90 42.10 T5 % 81.90 T50 % 99.00 DTGrnax DTArnax n-Eicosane BP = 344 "C T1% 118.40 151.27 T5 % 206.55 T50 % 209.47 DTGrnax "A ' max n-Hexatriacontane BP = 499 "C T1% 207.78 239.40 T5 % 305.40 T50 % DTGmaX 318.73 A ''max n-Pentacontane BP = 586°C T1% 23 1.64 255.27 T5 % 347.10 T50 % 359.64 DTGrnax DTArnax

TG 750

-

STA 780

f V (%)

X

f V (%)

8.26 7.92 5.41 4.44

47.89 77.00 124.30 142.00 144.40

7.49 4.90 2.70 3.10 2.02

2.09 1.74 1.76 2.90

174.88 200.20 254.40 262.00 260.30

2.32 2.02 1.54 1.74 1.23

354.70 384.00 446.00 442.90 442.50

3.21 1.31 0.58 1.84 0.58

-

-

5.28 3.21 4.65 3.68 -

1.40 1.28 2.07 1.73 -

tics. The comparison of the boiling points (BP) to the correspondmg evaporation start temperatures T1 % and T5 % in the thermobalance, (Table 3-5) demonstrates that evaporation in the thermobalance takes place at considerably lower temperatures. This is a result of the immediate discharge of the vaporized portions of the sample by the gas flow with the consequence that an equilibrium between evaporation and condensation cannot be established. However, the sample weight (mass) of approximately 5 mg provides a very large surface compared to the volume, so that the experimental conditions resemble those of a thin-layer vacuum distillation. Nevertheless there are differences in the temperatures of equal evaporated portions, and in the DTG maximum temperatures between the two thermobalances TG 750 and STA 780 as a consequence of the different shape of the sample pans:

Diameter (mm) Wall Height (mm) Ratio Surface: Volume

TG 750 Pan

STA 780 Pan

5.0 2.0 0.5

5.3 3.9 0.26

22


Table 3-4: Temperature Statistics ("C) for Aromatics TG 750

X

+V (%)

131 159 209 234

3.15 2.36 1.95 1.86

191 228 310 318

2.98 1.86 1.82 2.02

132 157 217 235

3.28 2.86 2.77 3.12

T1% T5 % T50 % DTGmm

136 179 242 26 1

2.98 3.02 2.54 2.23

T1% T5 % T50 % DTGmax

142 169 237 257

2.78 2.75 3.15 2.26

Tiiphenylmethane BP = 360 "C T1% T5 % T50 % DTGmaA 1,3,5-TriphenylbenzeneBP = 460 "C TI % T5 % T5O % DTGm, Fluoranthene BP = 384 "C TI % T5 % T50 % m 'a'''

3-Methylpyrene BP = ?

4-Methylpyrene BP = 410 "C

Table 3-5 Boiling Point ("C)BP Temperature of I % Weight LOSST1 % Temperature of 5 % Weight - Loss T5 %

I

for TG 750 and ST* 780 (Q

TG 750 Substance n-Undecane n-Eicosane n-Triacontane n-Tetracontane n-Pentacontane n-Hexacontane n-Butylbenzene Naphthalene Anthracene Indole 1,1,2,2-Tetraphenylethylene

BP

TI %

T5 %

195.9 343.6 449.9 526.3 585.7 634.2 183.1 218.1 342.0 253.5 417.5

26.9 118.4 189.4 230.6 261.5 286.3 22.8 43.2 115.7 64.0 158.4

42.1 151.3 216.9 263.5 299.0 327.8 38.9 60.8 138.4 83.0 184.2

STA 780 T1% T5 % 47.9 174.9 237.5 278.6 327.5 340.9

77.0 200.2 256.1 309.8 346.9 358.4


23

Even if the ratio surface: volume seems to be nearly equal for equal sample sizes, the higher walls of the STA pan cause a difference in the removal of the evaporated portions. Nor did the use of TG 750 pans in the STA 780 yield identical results, because the geometric conditions of the two ovens are different. Mixtures of model substances give TGA graphs with steps or some levelling, which prevent the exact identification of the components of the mixture, even when their boiling points were considerably separated from each other. For instance Fig. 3-5 demonstrates this on a mixture of equal parts of n-undecane (BP = 196 "C), n-pentacosane (BP = 402 "C), and n-hexatriacontane (BP = 499 "C). The TGA curve of TG 750 exhibits an evaporation start temperature T1 % = 46.8 "C. The first step is terminated at an offset

100

Mo

300

400

500

600

700

Fig. 3-5: Thermogravimetry of Test Mixture of Equal Parts of n-Undecane, n-Pentacosane, and n-Hexatricontane TG 750 Atmosphere: Argon, 25 cm3/min Heating Rate p : 10 K/min

24


temperatue of 115 "C. The second step has an onset temperature of 225 "C and an offset temperature of 305 "C. The same experiment carried out using STA 780 gives the following temperatures: T1 % = 56.5 "C, offset1 = 157 "C, onset2 = 290 'C, and offset2 = 390 "C. The DTG curve from TG 750 shows two distinct peaks, at 99 "C and at 256 OC, and an additional shoulder at 327 "C. From STA 780, a DTG curve exhibits three distinct peaks, at 129 "C,296 OC, and 375 "C. In the DTA curve two melting peaks were found, between 36 and 38 "C and between 69 and 71 "C. The first one is almost 10 degreees below the fusion temperature of n-pentacosane, but the second one corresponds nearly to the fusion temperature of n-hexatriacontane (MP = 71 "C). Moreover three distinct maxima are present in the DTA curve, at 129 O C , 296 OC, and at 376 "C. The statistical evaluation of five tests demonstrates that the data described previously can be repeated with small deviations as proved by the coefficients of variation (Table 3-6). Linear polyethylenes containing minimal chain branching (ratio of CH, groups to 1 000 chain C atoms < 1) can be regarded as very long chain n-alkanes. Thermogravimetry proves that they contain minimal portions of vaporizable substances (depending on the mean molecular weight Mw)and depolymerize nearly quantitatively at temperatures above 400 "C. A small coked residue amounting to only 3-5 % was found in the TGA curve after Table 3-6: Temperature Statistics ("C) Mixture of Equal Parts of n-Undecane, npentacosane, n-Hexatriacontane -

T1% T5 % T50 % DTG 1 DTG2 DTG3 DTAl DTA2 DTA3

TG 750

STA 780

-

X

+V (%)

X

f V (%)

46.80 70.80 246.60 99.2 256.0 326.6

2.89 0.96 1.20 3.29 1.24 1.75

56.60 94.00 294.40 131.8 301.8 374.8 128.0 296.0 375.4

6.68 2.26 1.38 2.11 1.03 0.69 2.50 0.79 0.84

Table 3-7: Thermogravimetry of Polyethylene TG 750 Atmosphere: Argon 25 cm3/min Heating rate p = 10 K/min

Mw

5 400 5 900 15 000 60 000 147 000 388 000

Average of Chain Lenght C-Atoms

TI % ("C)

T5 % ("C)

T50 % ("C)

TDTG

385 42 1 1 070 4 825 10 500 27 714

203 248 378 400 398 388

323 349 452 439 427 430

495 483 500 488 489 492

509 496 5 10 492 492 493

("C)

3.1 Therrnogravirnetvy (TGA)

25

a point of inflexion in the vicinity of 500 "C (Fig. 3-6). The DTG curve demonstrates only one maximum around 500 "C. A dependence of the evaporation start temperature on the mean molecular weight M , can only be recognized on polymers having small molecular weights. The temperature of 50 % weight (mass) loss (T50 %) and the temperature of the DTG maximum do not exhibit any dependence on the molecular weight (Table 3-7). This indicates that depolymerization always follows the same reaction mechanism regardless of the C chain length (molecular weight). The aromatics tested (Tables 3-4 and 3-5) normally exhibit a levelling of the TGA curve at weight (mass) losses more than 85-95 %. There is doubt whether this is a real pyrolysis because the point of inflexion T, is in the range from 230 "C up to 300 "C. Nevertheless the length of an aliphatic side chain influences the thermal behavior of substituted pyrenes

- \

147ooc

MW 5'

1 I

d

I I

I:

AI I

d

I I

Fig. 3-6: Thermogravimetry of Polyethylenes of Different Molecular Weights M , TG 750 Atmosphere: Argon, 25 cm3/min Heating Rate p : 10 K/min

26


Table 3-8: Thermogravimetry of Pyrenes and Alkylpyrenes (TG 750) Atmosphere: Argon 25 cm3/min Heating Rate p= 10 K/min Substance

T1% ("C) 133 136 142 169 182

M

Pyrene 3-Methylpyrene 4-Methylpyrene Pentylpyrene n-Hex ylpyrene

202 216 216 272 286

TS% ("C) 162 179 169 206 214

T50% ("C) 215 242 237 272 282

-

262 255 305 305

TpTG ( C) 235 26 1 257 302 302

(Table 3-8). The increase of the side-chain by five CH, groups from methyl to hexyl pyrene has the consequence of increasing the temperatures characterizing evaporation by 3540 "C. Regarding pure n-alkanes of nearly the same boiling range, i. e. n-C,,H,, and n-C,,H6,, a difference in the corresponding temperatures of 32 "C to 42 "C is found. The data representing the boiling behavior of the homologous series of alkyl substituted pyrenes can be displayed easily as a function of the C number of the aliphatic side chain (Fig. 3-7). This is analogous to the behavior of the homologous series of the mono-, di-,

1

2

3

4

5

Fig. 3-7: Thermogravimetry of Pyrene and Alkylpyrenes TG 750 Temperature of 1 % Weight Loss (T1 %) Temperature of 5 % Weight Loss (T5 %) Temperature of 50 % Weight Loss (T50 %) Temperature of the DTG Maximum (TDTG)

6

3.1 Thevmogravirnetry (TGA)

10 0

150

200

250

27

3 00

Fig. 3-8: Boiling Point Temperatures (BP) of Aromatics versus Molecular Weights (M)

and trisubstituted alkylbenzenes, where the boiling point versus the chain length (C number) of the alkyl substituent also gives a steady function. Even the boiling points of the series of condensed aromatics versus the molecular weight results in a steady curve (Fig. 38). The behavior of the substances specified in ASTM D 2887-84, Table X 1-1, (this Chapter, Table 3-2) is different because these substances are not members of an homologous series but represent olefins, cycloalkanes, aromatics, and heterocompounds. The condensed aromatics and the polar compounds containing heteroatoms such as oxygen, sulfur, or nitrogen diverge distinctly from the boiling behavior of the non-aromatics of equal molecular weight (Fig. 3-9). Therefore it is understood that the evaporation behavior in the thermobalance ist also different.

28


450

400

350

300

250

200

150

100 I

I

100

150

I

200

250

J

300

Fig. 3-9: Boiling Point Temperatures (BP) of Substances according to ASTM 2887-84, Table X1.l (Table 3-2) versus Molecular Weights ( M )

3.1.2 Simulated distillation The fact that volatile substances will vaporize at temperatures far below their atmospheric boiling point, due to the gas flow through the oven of the thermobalance, can be used to perform a simulated distillation [3-1 to 3-41. In contrast to the simulated distillation by gas chromatography GC [3-5, 3-61 no partition or rectification effect can be assumed during evaporation in a thermobalance. Therefore calibration can only take place using individual substances of known boiling points. It seems reasonable to use the substances which are named for calibration in the standards of simulated distillation by GC [3-5, 3-61.


29

For calibration, the temperatures of the weight losses of 1 %, 5 %, 10 % and up to 100 % or up to evident cracking symptoms were assembled from the TGA graphs of individual substances, and drawn versus the evaporated portion, similar to the Engler curves in DIN 51 572 [3-71. Fig. 3-10 demonstrates this in the diagram for n-hexatriacontane. It can be seen that cracking of the sample starts at a temperature above 320 "C when 70 % has already evaporated. In the TG 750 the lower n-alkanes, up to a carbon number of approximately 40 to 44, do not crack at all. The higher ones start cracking when about 70 or 80 % of the sample has evaporated. In the STA 780 every n-alkane can be vaporized without decomposition except n-pentacontane and n-hexacontane. Fig. 3-1 1 shows the Engler curves of the n-alkanes measured using TG 750 (extrapolated to 100 % weight loss). Fig. 3-12 shows the Engler curves of the substances presented in Table 3-2, also acquired by TG 750. The repeatability is worse at small weight losses than at higher ones. However, the repeatability is much better for the homologous series of n-alkanes than for the group of chemicals from ASTM D 2887-84 (Fig. 3-13).

(TG 750)

n-C36 600 .

500

:1

T ("C)

i

400

300

200

0

20

40

60

80

100

Fig. 3-10: Thermogravimetry of n-Hexatriacontane Diagram analogous to DIN 5 1 75 1 TG 750

T ("C)

-4

n-CSO

43u

400

350

300

250

n-ci9

200

n-C17

150 -

Loss ( w t - % )

Fig. 3-11: Thermogravimetry of n-Alkanes Diagram analogous to DIN 5 1 751 TG 750

3.1 Thermogravirnetry (TGA) 550

31

-

450

400

350

300

250

200

150

100

50

Fig. 3-12: Thermogravimetry of Substances according to ASTM 2887-84 Table X1.l (Table 3-2) Diagram analogous to DIN 51 751 TG 750

32

3 Thermal Analysis on Model Substances 30

-77 -25 ba

Y

w +I

$20 4J .C

2

u

*

'\

-

---\

i '' : Ic

i2 5 \5 :

-

0

I

0

S

20

A

u "

1

--

3 l

"

'

"

'

~

60

40

L 0

l

s s

'

"

80

~

100

(wt-%)

Fig. 3-13: Coefficient of Variation (kv) of the Temperatures versus Weight Losses Curve 1: n-Alkanes (TG 750) Curve 2: Substances'according ASTM (TG 750) Curve 3: n-Alkanes (STA 780) Curve 4: Substances according ASTM (STA 780)

By plotting the temperature of the actual adequate weigJ loss TTGversus the carbon number C of the n-alkane one obtains curves of a logarithmic function TTG=a, ln(C + aJ + where and

Eq. 3-1

TTG= Temperature in the thermobalance C = Carbon number of the individual n-alkane.

This is demonstrated in Fig. 3-14, which shows the TG temperatures versus the carbon numbers of the n-alkanes for a weight loss of 50 %. The curves representing other percentages of weight loss are identical. After a corresponding temperature shift, every one is congruent to the curve of the extrapolated boiling points of the n-alkanes (Fig. 3-15) [3-81. The correlation is very good (coefficient of correlation r = 0.990 at weight loss of 1 %; starting from weight loss 5 % the correlation coefficient rises to r > 0.995). The dependence of the maximum temperature of the reaction rate TDTG obeys a second order function

TDTG = a4 C?

+ a5 C + as

Eq. 3-2

3.1 Thermogravimetry (TGA) TG 7 5 0

500

33

T 50 %

T ("C)

400

-

. .

C-Number

0

-7-T-TTT7-P I

20

10

0

30

40

50

-

9

I

70

60

Fig. 3-14: Temperatures of 50 % Weight Loss versus C-Numbers of n-Alkanes (TG 750)

800

600

400

200

C-Number

0

~

1- -----.--v

0

20

40

60

80

100

Fig. 3-15: Boiling Points (BP) of n-Alkanes versus C-Numbers (According to 0. Glinzer) [3-81

34


demonstrating a correlation coefficient Y = 0.995 (Fig. 3-16). The graph of the boiling point (BP) of the n-alkanes versus the corresponding temperature in the thermobalance TTG (Fig. 3-17), is that of a linear function for each percentage of weight loss B P = a TTG+ b

Eq. 3-3

400 -

*

300

-

*

200 -

100 -

C-Number J

Fig. 3-16: Temperatures of the Maxima of Weight Loss Rate versus C-Numbers of n-Alkanes TG 750

The correlation coefficient for 1 % weight loss is Y = 0.986, for 5 % Y = 0.994, and for weight losses exceeding 10 % the correlation coefficient increases to Y > 0.997. The coefficients of equation 3-3 are different for each weight loss and decrease at increasing weight losses (Fig. 3-18 and 3-19). The ascertained functions are always of the same kind regardless to the type of thermobalance or the series of calibration substances, but the coefficients are, naturally, different. Nevertheless there is a linear relation between the data


35

600 500

1

400

1

300

7

200

i

1 0

200

100

300

400

500

Fig. 3-1 7: Boiling Points (BP) versus Temperatures in the Thermobalance at 50 % Weight Loss (TsTA) of n-Alkanes BP = a . TsTA+ b (Equation 3-3)

x10134

132

130 m 4J .r

128

u

.r

+ 01

126

u 124

,

122

0

120 0

20

40

60

W e i g h t Loss %

80

100

Fig. 3-18: Coefficient a of Equation 3-3 versus Weight Loss AG of a-Alkanes STA 780 a=x.AG+y x=-1.18682 10.’ y = 1.3355 r = 0.9818

36


140

t

120

100

20 "O

I

0 0

20

40

60

60

100

Weight Loss %

Fig. 3-19: Coefficient b of Equation 3-3 versus Weight Loss AG of n-Alkanes STA 780 b = x . l n ( A G ) + y x = - 21.9653 y = 153.538 r= 0.99265

achieved by TG 750 and the data achieved by STA 780 regardless of the differences in oven geometry and sample pans. With the help of this linear function the temperatures of equal weight losses can be converted from one thermobalance to the other. The coefficients a and b of equation 3-3 permit the calculation of the distillation curve at atmospheric pressure of an unknown substance from its TGA curve. This is shown in Fig. 3-20, which depicts the distillation curve of a mixture of equal portions of n-dodecane (BP = 216 "C), n-pentacosane (PB = 402 "C), and n-hexatriacontane (BP = 499 "C). The step between the two substances which have a difference in the boiling point of approximately 200 "C (C12and C2J is easy to recognize. Less obvious is the transition from n-C,, to n-C,,, the signal of the DTG demonstrates two distinct peak maxima too, which permit the calculation of two boiling temperatures (213 "C and 400 "C). Moreover there is a shoulder in the curve which corresponds to a boiling temperature of 489 "C (measurements by TG 750). The result of the simulated distillation of a test mixture of substances from ASTM D 2887-84 is shown in Fig. 3-21. The diagram demonstrates the distillation curve of a mixture of equal portions of cumene (BP = 152.5 "C), 1-octadecene (BP = 315 "C), and triphenylmethane (BP = 360 "C). The application of the simulated distillation on real systems (technical substances) will be described in section 4.2.3.3.


37

Fig. 3-20: Simulated Distillation Curve of a Test Mixture of Equal Parts of n-Undecane, n-Pentacosane, and n-Hexatriacontane (TG 750)

38


Fig. 3-21: Simulated Distillation Curve of a Test Mixture of Equal Parts of Cumene, I-Octadecene, and Triphenylene methane (TG 750)

A simulated distillation by thermogravimetry may be carried out in principle by substituting the flow of inert gas by the application of vacuum, but the adjustment and the maintenance of a defined vacuum in,very narrow limits is considerably more difficult than the adjustment and maintenance of a defined gas flow rate.

3.1.3 Thermogravimetry in an oxidizing atmosphere The thermogravimetric oxidation tests were performed using air as oxidizing agent. Information trials using oxygen had disclosed very fast oxidation reactions, sometimes so fast that effects are overtaken and concealed which are clearly visible in air.With regard to


39

the transfer of the technique to practical or industrial conditions, thermogravimetry using air is preferable. Air is also preferred in tests on model substances on the grounds of comparability. However, this may have the disadvantage that sometimes substances will evaporate prior to the start of the oxidizing reaction due to their low boiling point. For example n-dodecane (BP = 216 "C) evaporates completely before any oxidation can start. The TGA curves both in argon and in air are identical (Fig. 3-22 curve No. 1). The corresponding DTG curve exhibits only one peak maximum at 150 "C in argon as well as in air, certainly caused by evaporation. Even the major portion of n-hexatriacontane (PB = 499 "C) evaporates or cracks before any deviation of the TGA curve in air from the TGA curve in argon can be perceived (Fig. 3-22 curves No. 2 and 3). The DTG curve

Fig. 3-22: Thermogravimetry (TG 750)

Heating Rate p : 10 K/min Gas Flow Rate: 25 cm3/min Curve 1: n-Dodecane in Argon and in Air Curve 2: n-Hexatriacontane in Argon Curve 3: n-Hexatriacontane in Air

40


exhibits a twin peak in argon at maximum temperatures of 309 "C and 3 18 OC: whereas the DTG curve in air displays only one sharp peak at 293 "C. The test run on n-hexacontane demonstrates a distinct difference of the TGA curves in argon and in air right from the beginning (Fig. 3-23). In argon the DTG curve exhibits only one maximum at 418 "C, which indicates a crack reaction. On the other hand there are six maxima of the DTG curve in air at temperatures of 296,357,375,388,462, and 530 "C. Those maxima correspond, at least partly, to the three theoretical steps of the oxidation reaction [3-91. In the first reaction step, addition of oxygen takes place in a temperature range between 200 "C and 300 "C, when non-volatile oxygen-containing products are formed. This range is called low temperature oxidation (LTO). The TGA curves shown before do not give any indication of

1

1 300 0

PO0

1

0

\

I

m

-yi

600

7

Fig. 3-23: Thermogravimetry of n-Hexacontane (TG 750) Heating Rate p: 10 K/min Gas Flow Rate: 25 cm3/min Curve 1: Argon Curve 2: Air


41

oxygen uptake. In a temperature range from approximately 300°C to 400 or 45OoC, simultaneous combustion and pyrolysis reactions occur accompanied by the formation of gaseous products, more or less volatile hydrocarbons, coke residue, and carbon black. This range is called fuel deposition because of the formation of non-volatile combustible residues. In the last range above 450 "C the deposited residues will be burned. Therefore this range is called fuel combustion. In tests on n-alkanes and nearly all the other substances there is no increase in sample weight in the temperature range corresponding to low temperature oxidation. Only substances with high boiling points and containing a high concentration of carbon double bonds, such as polybutadiene, have an increase in weight in the LTO range (Fig. 3-24). This increase of only + 1.4 % is very small compared to the theoretical.increase of + 29 %. As a consequence of the small increase in weight there is no

Fig. 3-24: Thermogravimetry of Polybutadiene (TG 750) Heating Rate p : 10 K/min Gas Flow Rate: 25 cm3/min

42


evaluable peak in the DTG curve. During experiments on non-evaporable n-alkanes such as high density polyethylene (M, = 15 000) no rise in weight was found, but a flattening of the TG curve in the temperature range between 220 "C and 380 "C (Fig. 3-25). Above 380 "C the weight loss becomes nearly linear with respect to the temperature. Two points of inflexion are found in the TGA curve at 400 "C and at 440 "C. The flattening between 220 "Cand 380 "C indicates that there the weight loss caused by evaporation (cracking) is compensated partly by the uptake of oxygen. The temperature range from 380 "C to 440 "C can be regarded as fuel deposition. Then the fuel combustion follows at temperatures above 440 "C. The DTG curve in argon only displays one maximum at a temperature of 510 "C, which is certainly caused by pyrolysis. On the other hand, the DTG curve of the reaction in air presents three sharp peaks with maximum temperatures of 375 'C, 410 'C, and 427 "C and a weak peak at 522 "C. When repeating the experiment using the simulta-

Argon

Air

I

__

\

I

I I

1 I

I I I \

\

.

500

- Tf0C) 3

7w

Fig. 3-25: Thermogravimetry of HD-Polyethylene (TG 750) Heating Rate p : 10 K/min Gas Flow Rate: 25 cm3/min

3.1 Therrnogravimetry (TGA)

43

neous thermal analyzer, the same DTG peaks and additional exothermal DTA peaks were found at identical temperatures. In no case were any endothermal DTA peaks found. The statistical evaluation of repeated experiments does not show any differences between the mean values of the temperatures of the DTA and DTG peaks within the confidence level. Most of the aromatics from Table 3-2, up to and including anthracene, which were tested, evaporate or sublimate below the temperature of 225 "C in the flow of inert gas through the thermobalance. Most of the aromatics also evaporate or sublimate in the flow of air, due to their high oxidation stability. Fig. 3-26 demonstrates that in experiments with n-hexylpyrene, a deviation of the oxidation from the evaporation curve first occurs above 300 "C. In the oxidation experiment a small portion of coke is deposited, which will bum at temperatures above 450 "C.

Fig. 3-26: Thermogravimetry of n-Hexylpyrene (TG 750) Heating Rate p : 10 K/min Gas Flow Rate: 25 cm3/min

44


Even relatively bulky molecules such as adamantane, evaporate or sublimate completely and thus present identical TGA and DTG curves both in argon and in air. The DTG curve exhibits one peak only at a temperature of 140 "C. The start of the oxidation reaction of an individual substance or of a mixture of several substances can be ascertained by comparing the TGA curves of experiments in argon and in air,provided that the experiments are carried out using the same heating and gas flow rates. However, the measurement in air by means of DSC or DTA gives more exact results (see chapter 3.2.2). Generally the oxidation starts at relatively high temperatures. Using a heating rate p= 10 K/min, the start temperature of oxidation for substances containing reactive carbon double bonds is in the range from 130 "C to 150 'C, for aliphatic hydrocarbons in the range from 200 "C to 220 'C, and for aromatics above 300 "C. Experiments in air using different heating rates have demonstrated not only a parallel shift of the TGA curves to higher temperatures at faster heating rates, but also a change of the appearance of the curve, as shown in Fig. 3-27 for n-hexacontane at heating rates from 5 up to 100 K/min. Whereas the curve at p= 10 K/min is more or less a parallel displacement of the curve at p= 5 K/min, the curve at p= 20 K h i n shows quite another form. The date for the TGA and DTG temperatures in Table 3-9 demonstrate, especially in the case of the DTG maximum temperatures, that at fast heating rates effects are taken over, and obscured, which are still present at slower heating rates. At heating rates exceeding 50 K/min only one sharp peak appears at 344 "C (50 K/min) or 383 "C (100 K/min)

100

75

50

25

Fig. 3-27: Therrnogravimetry of n-Hexacontane in Air (TG 750) Heating Rate p: 10 K/min Gas Flow Rate: 25 cm3/min Curve 1: 5K/min Curve2: 10K/min Curve 3: 20 K/min Curve4: 50K/min Curve 5: 100 K/min

3.1 Thermogravimetiy (TGA)

45

Table 3-9: n-Hexacontane in Air, Dependence of the TGA and DTG Data on the Heating Rate Heating Rate

p

T1 % ("C) T5 % ("C) T50 % ("C) DTG,n, ("C)

5 K/min

10 K/min

20 K/min

50 K/min

100 K/min

210 228 295 277 350 368 430 510

215 239 307 296 375 390 462 539

232 257 320 3 10 370 446 450 546

240 279 344 344

25 1 29 1 383 383

-

-

-

-

-

-

570

548

a

maximum temperature, whereas the last maximum at approximately 550 "C becomes very weak. The phenomenon of the change of the shape of the curve should be considered in the evaluation of the reaction kinetics according to Flynn and Wall (see section 3.3.5). As a matter of fact, experiments at constant temperature prove that at a very fast heating rate (p= 100 K/min) oxidation no longer occurs (see section 3.1.4).

3.1.4 Isothermal thermogravimetry The reactions which take place at elevated temperatures exhibit a dependence on the reaction temperature as well as on the reaction (residence) time. The dependence on the temperature is described by the Arrhenius kinetic equation (see section 3.3.1). The dependence on the reaction time at a constant temperature can be ascertained by isothermal thermogravimetry. This generally applies to evaporation, pyrolysis (cracking), and oxidation processes. The aimed test temperature should be attained as quickly as possible. Therefore heating rates were chosen, which are the maximum rates for each instrument (TG 750: p= 100 K/min; STA 780: p= 50 K/min). Using a heating rate of p= 100 K/min, a final constant temperature of 250 "C will be attained within 2.5 minutes. Fig. 3-28 shows the graph of the weight loss of high density polyethylene (M,= 15 000) versus the residence time at a reaction temperature of 250 "C in air and in argon (flow rate: 25 cm3/ min). The diagram shows that the oxidation process is still not complete after 150 minutes, but it has slowed down compared to the reaction rate during the starting phase. It is well known that the oxidation reaction has an induction period, but this induction period can be concealed at extremely fast heating rates as Fig. 3-29 demonstrates in the example of n-hexacontane. In all experiments a heating rate p= 100 K/min was used to attain different constant final temperatures. The loss of weight starts after almost three minutes regardless of the target final temperature. It can be seen that in the experiments with the highest final temperatures of 455 "C and 465 "C a loss of the total sample weight occurs within one minute both in argon and in air. This means that the evaporation (or cracking) is faster than the oxidation and so no oxidation reaction may start. In the temperature range of approximate 300 "C the behavior of n-hexacontane both in argon and

46


100

75

50

25

30

60

90

120

150

Fig. 3-28: Isothermal Thermogravimetry at 250 "C (TG 750) HD-Polyethylene M, = 15 000 Gas Flow Rate: 25 cm3/min

in air resembles the behavior of polyethylene (Fig. 3-28) i. e. small weight loss in argon and a weight loss increased by a factor of four in air after 15 minutes reaction time. The phenomenon that depolymerization or pyrolysis reactions are faster than oxidation reactions was unexpected and is shown in Fig. 3-30 in the example of SEBS copolymer (styrene - ethylene - butene - styrene) which represents an alkyl substituted aromatic. Total loss of the sample weight occurs at a temperature of about 450 "C within identical reaction times in both argon and air. Figs. 3-29 and 3-30 also demonstrate that in the medium temperature range, 300 "(2-350 'C, only the first two reaction steps of the oxidation reaction take place within the residence time of 15 minutes, i. e. low temperature oxidation and fuel deposition. The deposited coke residue does not oxidize during this time. This is proved by the asymptotic rise of the curve of weight loss against time. Within ten minutes at a constant temperature of approximately 350 "C the curves of both substances arrive at a final weight loss of 80 % or 85 % representing 20 % (15 %) coke residue. The termination of the experiment after 15 minutes residence time may result in misleading conclusions. Fig. 3-31 shows the isothermal oxidation of coronene at a constant temperature of 350 "C. Up to approximately 30 minutes residence time the curve seems to exhibit an asymptotic rise, but after 35 minutes the characteristicof the curve changes to an exponential rise, which is typical for autocatalytic ractions. After a residence time of 75

3.I Thermogravimetry (TGA)

47

lo(

90

80 70 60

so 10 30 20 10

5

10

15

Fig. 3-29: Isothermal Thermogravimetry at Different Final Temperatures (TG 750) n-Hexacontane Gas Flow Rate: 25 cm3/min

minutes, corresponding to a weight loss of 90 %, the slope of the curve flattens and attains the final value of 100 % weight loss after nearly 100 minutes. All experiments were carried out using TG 750, at a heating ratep=100 K/min and a gas flow rate of 25 cm3/min.

3.1.5 Experiments using the simultaneous thermal analyzer The experiments described in this section were performed using a Simultaneous Thermal Analyzer STA 780 (Stanton-Redcroft Ltd.). This instrument supplies the measuring signals temperature (T), weight (TGA), first derivative of weight with respect to time

48


t Mass Loss

Fig. 3-30: Isothermal Thermogravimetry at Different Final Temperatures (TG 750) SEBS Copolymerisate Gas Flow Rate: 25 cm3/min

(DTG), and differential thermal analysis (DTA). In the first stages of experimental work the signals were recorded by means of a four-pen line recorder and evaluated manually. Later on the instrument was equipped with a personal computer (PC) for control, data sampling, and data evaluation. At the time of acquisition, DSC systems with a maximum temperature limit of only 600 "C were available, so a DTA system possessing an upper temperature limit of 1 000 "C was purchased. In the meantime heat-flux DSC systems with upper temperature li&ts of 1 000 "C or 1 500 "C have been developed. In any case DSC is preferred to DTA systems, because the DSC supplies energy values directly, whereas DTA only provides differences between sample and reference temperatures, which then have to be converted to energy data. Simultaneous instrumentation allows changes in weight and in energy of the same sample to be measured during the same experiment. Each test carried out by the author was performed using an empty pan for reference.


49

100

75

50

25

Fig. 3-31: Isothermal Therrnogravimetry at 350 "C in air (TG 750) Coronene Gas Flow Rate: 25 cm3/min

In the experiments using n-alkanes for model substances the DTA signal supplies the temperatures and the areas of the fusion (melting) and the boiling peaks in addition to the TGA and DTG curves. Integration of the peak area gives a value possessing the unusual dimension pV . s . mg-' (Fig. 3-32). The conversion to energy (enthalpy) data, dimension J . g-l, is possible in principle but raises some difficulties. Evaluation of the maximum temperatures of the fusion (melting) peaks (MP) demonstrates good conformity of the measured data to the data from literature references [3-10, 3-11].

50


1 6 7 0 9

70

10

M

-I II

190.74 i97.S 202.94 208.77 2i4.34

70.053 60.073 10.191 40.274 30.203

-

10

-

5

-

0

- a

227.07 C

POINT Tk6LE

-

-LO

-

-1B

-

-20

-

-a

2 226.70

10

1

0

B

ib0

is0 oeg

c

2bo

250

Fig. 3-32: Thermogravimetry of n-Nonadecane Plot of STA 780: TGA and DTA Atmosphere: Argon, 30 + 20 cm3/min Heating Rate p: 10 K/min

Fig. 3-33: Melting Point Temperatures (MP) versus C-Numbers of n-Alkanes Curve 1: Data from STA 780 Curve 2: Data from References [3-10, 3-11]

30


10

20

30

40

50

51

60

Fig. 3-34: Heat of Fusion Hfof n-Alkanes Area of the Melting Peak of STA 780 versus C-Numbers

In Fig. 3-33, curve No. 1 shows the DTA temperature data and curve No. 2 the fusion temperature data from the references, both plotted against the carbon number of the n-alkanes. After refining the calibration of the instrument the curves will be brought into congruence. The integral of the peak area, which should be equivalent to the heat of fusion Hfdoes not conform so well. The graph of the peak area versus the carbon number of the n-alkanes (Fig. 3-34) shows heavy scattering of the data, although the general direction is still recognizable. This tendency is even more distinct than that in the analogeous graph of the data of the heats of fusion Hffrom the literature versus the carbon number (Fig. 335). The evaporation peak results in a graph of the DTA peak maximum temperatures versus the carbon number which exhibits very little scattering (Fig. 3-36). The shape of the curve is similar to that of the boiling points (BP) from literature references (Fig. 3-15). A linear correlation exists:

BP = a . T,,+

b

with a = 1.2238 and b = 47.4372

Eq. 3-4

52


0 0

0

a 0

C-Number -r

10

20

30

I

! I

60

Fig. 3-35: Heat of Fusion Hf of n-Alkanes

Data from References [3-10, 3-11] versus C-Numbers

The correlation is excellent with the coefficient of correlation r = 0.99723. The graph of data of the peak areas versus the carbon number shows a higher degree of scattering (Fig. 3-37) which is likewise present in the graph of the data of the heats of vaporization (H,) from literature references versus the carbon number (Fig. 3-38). Nevertheless the pattern of both curves shows the same direction. Comparison of the DTG and DTA temperatures of the peak maxima of the evaporation gives an average temperature difference of 0.88 % of the measured value (maximum 1.66 %, minimum 0.27 %). The DTG maximum temperature seems to be somewhat lower than the DTA maximum but this difference is within experimental tolerance. So it may be assumed that the temperatures of the maximum of the evaporation rate and of the maximum of energy flow are identical. The conversions (weight losses) corresponding to those temperatures are also identical within experimental error.

3.1 Thermogravimetry (TGA) 500

400

300

200

100

10

20

30

40

50

60

Fig. 3-36: Temperatures of Vaporization (DTA) versus C-Numbers of n-Alkanes (STA 780) Atomosphere: Argon, 30 + 20 cm3/min Heating Rate p: 10 K/min

500

400

300

200

100

Fig. 3-37: Heat of Vaporization versus C-Numbers of n-Alkanes Area of the Vaporization Peak (STA 780)

53

54


10

20

30

40

50

60

Fig. 3-38: Heat of Vaporization versus C-Numbers of n-Alkanes Data from References [3-10, 3-11]

3.2 Differential scanning calorimetry on model substances 3.2.1 DSC in an inert atmosphere The model substances for DSC were the same as those used for thermogravimetry (see section 3.1) such as the series of n-alkanes and some of the substances listed in ASTM D 2887-84. If an appropriate gas flow rate was used, then evaporation without residue of all the tested n-alkanes resulted, as already described for thermogravimetry. Fig. 3-39 demonstratesthis behaviour for n-hexacontane. At a temperature of 99 "C there is a sharp fusion (melting) peak exhibiting a heat of fusion H,= 243.3 J . g-l. At 439 "C the maximum of energy resorption for the evaporation process appears, which has an onset temperature of 409 "C. The low energy resorption of 136 J . g-I indicates the occurrence of an evaporation process.

3.2 Differential Scanning Calorimetry on Model Substances

55

0 7.71 OC

13.3J/g

136.3319

-5

E

" -10

u. +J

(I

I

-15

99.

-20

."-c

100

I

200

I

300

I

400

500

I

600

7 0

Temperature 0.995. The coefficients a4 and as are independent of the heating rate p but depend on the gas flow rate. The boiling points (average values) of vaporizable substances can be ascertained by DSC using the coefficients a4and as. It is not possible to establish the distillation curve since there is no information about weight or volume changes, but the offset and onset temperatures may be used to calculate the initial and the final boiling points [3-121. Application of the correlation of the peak temperatures to the heating rates for calculation of the Arrhenius kinetic parameters and of the heat of vaporization will be considered in chapter 3.3.2. 10 K/min .I

Boiling Point ("C)

0

100

.

.

.

200

.

.

.

.

.

.

300

.

.

.

.

.

400

.

.

.

. . . 500

Fig. 3-42: DSC of n-Alkanes Peak Temperatures of DSC versus Boiling Points (BP) Atomosphere: Argon, 5 cm3/min Heating Rate p : 10 K/min

.

.

-

.

.

600

.

58


Table 3-10: Melting Point MP and Boiling Point BP of Aromatics Compared to Peak Temperatures in DSC (Atmosphere 10 bar Methane, Flow Rate 5 cm3/min, Heating Rate /?= 10 K/min) Substance ~~

Anthracene 2-Methylanthracene 9-Methylanthracene 9,lO-Dimethylanthracene 9,lO-Diphenyl anthracene 9,lO-Dihydroanthracene Phenanthrene 2-Methylphenanthrene Fluoranthene Pyrene 3-Methylpyrene 4-Methylpyrene Triphenylene Coronene 1,3,5-Triphenylbenzene Triphenylmethane

178.24 192.26 192.26 206.29 330.43 180.25 178.24 192.26 202.60 202.26 216.28 216.28 228.30 300.36 306.41 255.34

218 207 81 181 244-247 108 101 55-56 110 156 -

147.5-148.5 199 438-440 174 92-93

216.5 205.5 77 178.5 243.5 104.0 99.0 53.5 108.5 149.0 65 .O 146.5 195.5 -

169.5 90.5

342 subl. 196112 subl. -

313 332 175-180,,, 384 393 -

410 425 525 460 360

291.25 275.0 284.5 296.25 383.75 243.75 281.25 276.25 303.75 337.50 330.00 343.75 343.75 455.75 375.00 275.50

Aromatics also evaporate or sublimate in the flow of inert gas at significantly lower temperatures than their boiling point (Table 3- 10). Although the experiments were performed at a pressure of 10 bar in the DSC cell in order to shift the vaporization to higher temperatures, the measured peak temperatures are up to 90 "C below the boiling points (BP) of the substances (at a heating rate p= 10 K/min). The difference between the peak temperatures and the boiling points decreases at increasing heating rates; reaching approximately only 50 "C at a heating rate p= 50 K/min. Dependence of the temperature of the maximum of energy uptake on the molecular weight can be established for condensed aromatics, both alkylsubstituted and unsubstituted (Fig. 3-43). Substances which do not belong to the condensed aromatics do not satisfy this model, even though they have a corresponding molecular weight; for example, triphenylmethane (Fig. 3-43, No. 2), 1, 3,5-triphenylbenzene (Fig. 3-43, No. 3), or 9, 10-diphenylanthracene(Fig. 3-43, No, 4). Also, substances having a bulky molecule structure such as the non-aromatic adamantane (Fig. 3-43, No. 1) show a deviation from the curve. Generally a strong functionality between data from thermoanalysis, representing evaporation behavior, and molecular weight or the carbon number of the substances, can only be ascertained for the members of homologous series. This is not confined to thermoanalysis but is also evident for other physical methods. In the homologous series of alkylsubstituted pyrenes the peak maximum temperatures of evaporation increases only a little when the length of the carbon chain of the alkylsubstituent increases:


59

Fig. 3-43; DSC of Aromatics DSC Peak Temperatures of Condensed non substituted and alkane substituted Aromatics versus Molecular Weights Atmosphere: Argon, 5 cm3/min Heating Rate p : 10 K/min Point 1: Adamantane Point 2: Triphenylene methane Point 3: 1, 3, 5-Tnphenylene benzene Point 4: 9, 10-Diphenylene anthracene

pyrene 4-methylp yrene pentylpyrene n-hexylpyrene

337 343 349 35 1

An increase of pressure gives similar effects with both aromatics and n-alkanes. The dependence of the maximum temperature of the evaporation peak on the heating rate is also analogous to the n-alkanes. From DSC tests, data on the melting point MP and the heat of fusion (melting) Hf are obtained. If the experiment is started at room temperature then data of the series of n-alkanes may be obtained, probably beginning with n-heptadecane (MP = 22 "C). If a commercial refrigerator (temperature range from -65 "C up to 250 "C) is used to cool the DSC cell, fusion and crystallization events my be recorded down to n-octane (MP= -56.8 "C).The cooling/heating rates should not exceed 5 K/min in order to measure

60


correctly any crystallization phenomena. The fusion and crystallization behavior of 12heneicosane is shown in Fig. 3-44. Whereas the melting point (MP) is only slightly affected by the heating rate, fi its influence on the crystallization is so large that an activation energy of crystallization may be calculated, using the shift of the peak temperatures caused by varying heating rates (see chapter 3.3.2). The melting and crystallization peaks may be easily integrated in order to calculate the heats concerned (enthalpies) as Fig. 3-45 shows for the fusion peak of n-eicosane. The data of the onset temperature, the peak maximum temperature, and the heat of fusion are recorded. From about n-nonadecane up to n-tetxacontane, two peaks have been recorded, for both fusion (melting) and for crystallization, with few exceptions (Fig. 3-44). Both peaks can be evaluated more or less easily. The true fusion or the true crystallization peak always appears at a higher temperature than the secondary peak. With increasing carbon numbers of the n-alkanes the values found by the secondary peaks decrease (Table 3-11). The appearance of the secondary peaks is not caused by pollution by another n-alkane of a shorter carbon chain length, as may be suspected; the proportion of the two peak areas is too large for that. Rather, some change in structure or modification, similar to the memory effects of polymers, is responsible for the secondary peaks. Sometimes the secondary peak is present during the crystalli-

1

-5

237.2J/g 70360. J/mole

56.68J/g 16810. J/mole

-

-10-

32.87T

-15

J 41.35OC I

I

Tempwoture (OC)

Fig. 3-44: DSC of n-Alkanes LOWTemperature Behavior of n-Heneicosane Sealed Pan Atomosphere: Argon, static, 1bar Heating/Cooling Rate p : 2 K/min


61

I

Table 3-11: Melting Behavior of n-Alkanes by DSC Atmosphere: 1 bar Argon, Flow Rate 5 cm3/min Heating Rate p = 2 K/min Substance

Purity 76

97 (GC) 98 (GC) 97 (GC) 97 (GC) 95 (GC) 99 (GC) 98 (GC) purum puriss. 99" 98"

* CH Analysis

Secondary Peak Mean Peak MP ("C) Hf (J.g-') MP ("C) Hf (J-g-') -

-

22.31

45.1

-

-

32.87 50.00 60.10 73.97

56.7 63.9 59.7 15.8

-

-

28.96 31.71 35.45 41.35 54.3 64.1 80.1 81.79 89.01 92.78 99.09

234.2 228.5 206.4 237.2 151.6 130.3 197.6 237.8 238.5 236.7 195.9

C H, (J.g-') 234.2 273.6 206.4 293.9 215.5 190.0 213.4 237.8 238.5 236.7 195.9

Reference Data MP ("C) H, (J.g-')

28.2 32.1 36.8 40.5 53.7 65.4 81.4 85.3

241.21 170.62 247.32 160.83 163.72 162.7 234.1 237.1

62


zation of substances which do not show one during melting. The heating rate /3 has a strong influence on the appearance of the secondary peaks. They may disappear at heating rates /3= 10 K/min (and faster) or may be reduced to a shoulder upon the lower temperature flanks of either the fusion or the crystallization peak. It is not known why no secondary peak has been observed with n-alkanes having carbon numbers below C,, or above C4,,. The graph of the heats of fusion versus the carbon number of the n-alkanes does not follow any systematic course (Fig. 3-46). This is also confirmed by the literature references. The differences between the heats of fusion and the heats of crystallization, which should be characteristic for the entropy part, scatter unsystematically, so they cannot be correlated with the carbon numbers of the n-alkanes. The values found for such differences are in the range from 2.3 to 5 J . g-' coi-responding to 890 to 2 750 J . Mole-'. The low temperature behavior of high-density polyethylenes (laboratory products), possessing a minimum of carbon chain branching ( < 1 CH, groups per 1 000 C-atoms) has been examined. The polyethylenes may be regarded as long chain nonvolatile n-alkanes. The DSC tests reveal a very small increase of the melting temperature of the crystallites due to increasing molecular weight. (Table 3-12). An increase of the molecular weight by a factor of 100, effects only a rise of the crystallite melting point temperature of less than 10 "C. The heat of fusion Hf remains more or less constant with Hf=240 J . g-l. This value indicates a crystallinity of 80 %. The heat of fusion of HD-PE possessing 100 % crystallinity is known to be approximately 300 J . g-l. The influence of chain branching on the melting temperature of the crystallites and on the heat of fusion is very strong, as shown by the HD-PE possesssing about 20 CH, groups per 1 000 C-atoms (product No. 2 in Table 3-12). The maximum energy resorption for pyrolysis (depolymerization) will be found at a

10

20

30

Fig. 3-46: Heat of Fusion of n-Alkanes Data from References [3-10, 3-11]

40

50

60


63

Table 3-12: DSC of High Density Polyethylenes Atmosphere: 1 bar Argon, Flow Rate 5 cm3/min Heating Rate p= 10 K/min Mean Molecular Weight Mw ~~

C-Number

Branching CH, per 1000 C

MP ("C)

H, (J.g-')

Pyrolysis Peak ("C)

385 420 1070 4 300 10 520 16 260 27 740

< I 20 < 1 < 1 < 1 < 1 < 1

123.2 112.2 125.8 126.5 127.5 130.0 130.6

240 130 220 242 238 248 245

479.2 479.3 481.3 478.7 478.7 480.8 479.8

~

5 400 5 900 15 000 60 270 147 300 227 650 388 400

temperature of about 480 "C. This temperature is not a function of the carbon chain length (molecular weight). The aromatics tested displayed melting temperatures somewhat lower than the values from the references (Table 3-10). The average of the deviations amounts to -2.2 %.

3.2.2 DSC in an oxidizing atmosphere Oxidation experiments in air result in the same phenomena as the analogous experiments using thermogravimetry. Low boiling substances will partially or completely evaporate in the gas flow, prior to the starting of oxidation. The oxidation reaction is characterized by the appearance of several exothermic peaks in the diagram of heat flow versus temperature (Fig. 3-47). These peaks are analogous to the DTG peaks in TGA oxidation tests and were found in similar temperature ranges. As mentioned earlier, the evaporation will shift to higher temperatures due to increasing pressure. However, the starting temperature of the oxidation will decrease at increasing air pressure because the partial pressure of oxygen is increased. This permits the analysis of the oxidation behavior, especially of the blends containing lower boiling compounds. Therefore most of the oxidation tests described in this book were performed at higher air pressure. In particular, the oxidation tests on lubricants were carried out at an air pressure of 7 bar corresponding to 100 psig so that they may be compared with English-language literature references. Fig. 3-47 and Fig. 3-48 represent the behavior of n-hexacontane during the tests at 1 bar and lobar air pressure. The endothermic fusion peak at approximately 100°C is not influenced by the increase of pressure. On the other hand, the exothermic oxidation peaks were shifted to lower temperatures. The first peak (low-temperature oxidation LTO) moves from 241 "C at 1 bar to 221 "C at 10 bar. In the range of fuel deposition, the peak at 334 "C (1 bar) disappears almost completely and may be recognized only in the shoulder of the LTO peak at 300 "C (10 bar). Also, the sharp peak present at 407 "C (1 bar) disappears,

64


426,67

'C

443.33

'C

241. 4S°C

-40

,

0

loo

200

300

I

I

500

400

600

7 0

Temperature (OC)

Fig. 3-47: DSC of n-Hexacontane Atmosphere: Air, 5 cm3/min, 1 bar Heating Rate p: 10 K/min

whereas the peak at 443 "C (1 bar) is shifted to 432 "C (10 bar). The rounded fuel combustion peak at 499 "C (1 bar) moves to 488 "C (10 bar). In tests on the higher boiling members of the homologous series of n-alkanes from n-triacontane up to n-hexacontane, two strong peaks are found which are easy to evaluate. These peaks are the LTO peak and the peak in the fuel combustion range. In the range of fuel deposition more than one peak appears. Evaluation of the numerous small peaks is problematic and rarely valuable. Evaluation is normally limited to one or two well-defined peaks. The peak temperatures of eight n-alkanes (C30H62, C3,H66,C36H74,C38H78, C4,H,,, C,,Hg0, C50H,02, C6@122)measured in experiments using the same heating rates (p= 5 , 10, 20, 50 K/min), the same air pressure and flow rates are estimated to be equal, as demonstrated by the low coefficients of variation: Peak Temperatures of n-Alkanes (measured at p= 10 K/min)

Peak 1 Low Temperature Oxidation Peak 2 Fuel Deposition Peak 3 Fuel Combustion

Mean x ("C)

Coefficient of Variation k V ( % )

223.59 292.68 482.14

1.56 1.68 0.91


65

60

-40

0

200

100

300 400 Temperature (OC)

500

600

5 0

Fig. 3-48: DSC of n-Hexacontane Atmosuhere: Air. 5 cm3/min. 10 bar Heat& Rate p : '10 K/min

The onset temperature of the first peak, characterizing the start of the oxidation reaction, may be considered to be equal within the limits of tolerance:

V = f 1.09 %

Tonset: X = 181.88 "C

The HD polyethylenes which may be regarded as long chain n-alkanes exhibit the same behavior. Their peak temperatures under analogous experimental conditions (at 7 bar) are somewhat higher, however the scattering is very small and independent of molecular weight: Peak Temperatures of HD Polyethylenes ~

~

~

Peak 1 Low Temperature Oxidation Peak 2 Fuel Deposition Peak 3 Fuel Combustion

~

~

~

Mean X ( "C)

Coefficient of Variation f V (%)

232.81 300.06 485.00

1.12 0.90 1.10

Even the presence of a double bond in a longer carbon chain does not result in a marked change of the peak temperatures. The test on 1-octadecene supplies the following peak temperatures @= 10 K/min; air 7 bar, 5 cm3/min):

66


peak 1 215.0 'C; peak 2 306.25 "C, peak 3 485.12 "C. Compounds containing more double bonds behave differently. The oxidation of polybutadiene yields temperatures for peak 1 of only 156 "C (onset temperature 127.5 "C), for the second (evaluable) peak of 348 "C and for the last peak of 439 "C. These results confirm the sensitivity to oxidation of highly unsaturated compounds. Aromatics behave with more stability against oxidation. Therefore higher onset temperatures and higher peak temperatures were found (Table 3-13). The low temperature and the fuel cornbustion ranges each exhibit one distinct peak. The fuel deposition range is characterized by more than one peak. Frequently two of them may be evaluated. 9, 10Dihydroanthracene exhibits exceptional behavior with an oxidation starting temperature (onset temperature) as low as 125 O C , and the LTO peak maximum temperature as low as 164 "C. Interruption of the aromatic system due to hydrogenation in the 9, 10-position markedly reduces the oxidation stability. The alkyl-substituted pyrenes clearly reveal lower onset temperatures too, but the peak maximum temperatures of the LTO peak are only a little lower than the corresponding temperatures of the unsubstituted pyrene. This indicates the lower oxidation stability of the alkyl substituents. However the last peak (fuel combustion) of the alkylpyrenes appears at a considerably higher temperature than the corresponding peak of the pyrene, which only exhibits two evaluable oxidation peaks. Several attempts have been made to prove that the last peak of the oxidation experiment represents the combustion of deposited carbon. For that purpose the oxidation of graphite and activated carbon (charcoal) has been analyzed. When testing graphite by means of DSC no oxidation maximum could be achieved in the temperature range available. The upper temperature limit of the DSC instrument is 625 "C or 650 "C maximum, Checking by means of STA results in DTG and DTA maximum temperatures of 832 "C (p= 10 K/min; Table 3-13: DSC Oxidation of Aromatics Atmosphere: Air, 7 bar, 5 cm3/min Heating Rate p = 10 K/min Substance onset ("C) Anthracene 9,lO-Diphenylanthracene 9,lO-Dihydroanthracene Phenanthrene Pyrene 3-Methylpyrene 4-Methylpyrene Pentylpyrene n-Hexylpyrene Triphenylene Coronene 1,3,5-Triphenylbenzene Triphenylmethane I, 1,2,2-Tetraphenylethylene

266 315 125 254 300 232 278 248 247 340 263 336 205 325

Peak 1 Maximum ("C) 285 364 164 255 308 285 307 294 294 349 328 376 248 334

Peak 2 ("C)

Peak 3 ("C) -

-

-

429 383 383 386 433 420 430 419 422 429 440 381 429

499 486

226 -

-

284 -

Peak4 ("C)

-

558 48 1 530 539 499 533 -

488 -


67

air 1 bar, 30 + 20 cm3/min). However, the test on charcoal by means of DSC @= 10 K/ min; air 1 bar, 5 cm3/min) results in a peak maximum temperature of only 567 "C. The structure of the surface of the sample influences the reactivity and thus the temperature of the peak maximum. Therefore, determination of the temperature of the last peak maximum does not give any positive evidence concerning the nature of the substance oxidized at that point. Even integration of the last peak does not offer any help. A heat of combustion will be supplied in J . g-l. But this value is related to the starting weight and not to the weight of substance reacting in the fuel deposition range and is therefore too small. An attempt to ascertain the weight of substance at corresponding temperatures with the help of thermogravimetry, in order to correct the data of the integration, results in very different data for the heats of reaction. This may be demonstrated by the example of three HD-PE (measured at p= 10 K/min and 7 bar airpressure). ~

HD-PE M,

Heat of Combustion of the last Peak (kJ . g-l)

5 400 5 900 15 000

14.68 5.82 22.31

Reference [3-131 supplies different data for the heats of combustion of carbon: Heat of Combustion (W. g-l) Graphite High Temperature Coke Charcoal

32.8 33.3 31.2 - 34.0

The heat of combustion of charcoal measured by DSC amounts to much lower values depending on the air pressure in the DSC cell (tests in air, flow rate 5 cm3/min; p= 10 K/

min): DSC Oxidation of Charcoal

1 10 20

487 436 420

567 511 484

15.2 18.3 19.4

The decrease of the temperatures and the increase of the heat of combustion are evidently caused by the increase in the partial pressure of oxygen. Oxidation kinetics will be treated in chapter 3.3.2.

68


3.3 Reaction kinetics 3.3.1 Theoretical basis The kinetics of the reactions which take place during the heating of the samples may be analyzed using thermoanalytical methods. The relation of the reaction rate to the temperature is described by the Arrhenius equation: --E

k =A e

where

R.T

k A E R T

= = = = =

Eq. 3-7

(Inin-') constant of reaction (rate) or specific rate constant frequency (pre exponential) factor activation energy universal gas constant absolute temperature

mid mid J . M01-l J . Mol-' . K-' K

Division of the natural logarithm, In 2, by the reaction constant supplies the half life time, t1,2,of the reaction at the temperature T: t,/2=

In 2 k

Eq. 3-8

The differential daldt of the conversion a with respect to the reaction time t at a constant temperature T is given in equation 3-9:

Eq. 3-9

a>" where

a = fractional conversion t = time k = specific rate constant n = reaction order

dimensionless min min-' dimensionless

For the instrumentation described in chapter 2, commercial software exists for the evaluation of reaction kinetics on the basis of four methods: - ASTM E 698-79 - Borchardt and Daniels - Flynn and Wall - McCarty and Green The methods according to ASTM E 689-79, and to Borchardt and Daniels apply data from DSC or DTA test runs, whereas the methods according to Flynn and Wall, and to McCarty and Green are based upon the use of data from thermogravimetry. For the

3.3 Reaction Kinetics

69

evaluations according to ASTM E 698-79 and Flynn and Wall, the data of at least three test runs is needed, whereas the methods according to Borchardt and Daniels and to McCarty and Green require the data from only one test run each. The principles of these methods are described below.

3.3.1.1 Method according to ASTM E 698-79 The ASTM E 698-79 Method [3-141, based upon the investigations of Kissinger, Uricheck, and Ozawa [3-15 to 3-18] measures peak maximum temperature variations with changes in the linear programmed DSC heating rate p. At least three experimental runs have to be executed at different heating rates (between 1 and 20 K/min). The resulting relation of the shift of the peak maximum to the actual heating rate is used to compute the kinetic constants. The calculation makes three basic assumptions: - The peak maximum temperatures represent a point of constant conversion for each heating rate. - The temperature dependence of the reaction rate constant obeys Arrhenius’ equation. - The reaction is of first order. Each of these assumptions can be expressed mathematically:

Eq. 3-10 where

E R

p T,

= = = =

activation energy universal gas constant heating rate peak maximum temperature

J . Mol-’ J . M01-l . K-’ K . min-l K

The solution of Eq. 3-10 is obtained by measuring the slope of log Pversus 1/T, plot (Fig. 3-49) representing the value of E/R, followed by an iterative computer calculation to refine the value for the activation energy E . The second assumption is expressed by Eq. 3-7. To fulfil the third assumption the exponent in equation 3-9 must have the value n = 1. The pre-exponential factor A is then calculated from equation 3-11:

Eq. 3-11 Where: T and pare obtained from a midrange heating rate.

70


\

1.2

>'

\\ \ \

1.1-

"\

0a

* 8 sW 5

k,

1.0-

-\,

x.

0.B-

-. '\

\

\

0.9-

\

\

b

0.7-

0.6'

.

I

8

I

-

I

-

I

.

The alternative method for calculating the activation energy according to ASTM E 69879 (part X3) is preferred for a numerical solution, because the value of E requires only one calculation step with no iterations:

Eq. 3-12 All the experiments carried out using the DuPont 990 Thermal Analyzer have been evaluated by applying equation 3-12.

3.3.1.2 Method according to Borchardt and Daniels The Borchardt and Daniels method [3-191 permits the calculation of activation energy E, frequency (pre-exponential) factor A, heat of reaction AH, and order of reaction II from a single DSC or DTA scan of a reaction isotherm. In principle it relies on a comparison between a partial area and the total area of an exothermic peak (Fig. 3-50). This will


1

I

8)

T

Y

71

\’-

Ho

TEMPERATURE ~

~~

~

~~

~~

Fig. 3-50: Reaction Kinetics according to Borchardt and Daniels r3-201

preclude the evaluation of endothermic pyrolysis peaks, however this disadvantage can be overcome by reversing the positions of the sample pan and the empty (reference) pan in the instrument and thus simulating an exothermic reaction. The method assumes that the reaction follows nth order kinetics i. e. obeys the relationship:

Eq. 3-13

d d d t = k(n (1-a)”(s-~) Where:

a

fractional conversion k(T, = specific rate constant at temperature T (reaction constant) n = reaction order =

dimensionless s-l

dimensionless

Further, the method assumes that the temperature dependence of the reaction follows Arrhenius’ equation k(n = A ’ exp (-E/R ’ r ) Where: A = frequency (pre-exponential) factor E = activation energy R = universal gas constant T = absolute temperature

Eq. 3-7 S-1

J . Mol-’ J . Mol-’ . K-’ K

Taking logarithms of equation 3-7 yields:

Eq. 3-14

72


A plot of In k,, versus 1/ T (Arrhenius plot) should be a straight line. The activation energy and the frequency factor can be obtained from the slope and the intercept respectively. Substituting equation 3-7 into equation 3- 13 yields: d d d t = A . exp(-E/R . T ) .

Eq. 3-15

Taking logarithms of equation 3-15 yields: ln(da/dt) = In A - E/R . T + n . In( 1-a) Eq. 3-16 Equation 3-16 may be solved with a multiple linear regression. The measured DSC (DTA) curve is used to derive the two basic parameters da/dt and a, required to solve equation 3-15. The reaction rate is obtained by dividing the peak height (dH/dt) at temperature T by the total heat of reaction AH,: daldt = (dH/dt)/AHo Where:

Eq. 3-17

AHo= the total peak area (theoretical enthalpy)

The fractional conversion a is obtained by measuring the ratio of the partial area AHT (from reaction start up to the temperature 7) to the total peak area AH,:

a= MT/AH@

Eq. 3-18

3.3.1.3 Method according to Flynn and Wall The Flynn and Wall method [3-21, 3-22] uses the thermogravimetric curves to calculate the activation energy E and the frequency (pre-exponential) factor A of the Arrhenius' equation. In this approach the TGA curve of the analysed material is recorded from at least three test runs at heating rates p between 1 and 20 K/min. The corresponding temperatures of selceted constant conversion levels are determined for each test run (Fig. 3-51). The measured values of temperature and heating rate are then used to compute the activation energy E using the method of successive approximations: Eq. 3-19 Where:

E = activation energy R = universal gas constant

P =

heating rate

T = temperature at selected weight loss b = empirical approximation whose value depends upon E

J . M01-l J . M01-l. K-' K . min-'

K

73


I

c

.4

-

.3

-

.2

-

.l

100 -

90 -

YI

u 80

-

.u

; > 0

U L

- 0.0

(u

n

b . , i

x

70 -

60

-

50

-

-

-.l

-

-.2

- -.3

'-

I

ion

3b0

400

-

-.4

-

-.5

--r--1-7--1500 600 700 800 300

Deg C

Fig. 3-51: Reaction Kinetics according to Flynn and Wall [3-21, 3-22] Pyrolysis of CaC,O, . H,O Atmosphere: Argon, 30 + 20 cm3/min, 1 bar Heating Rates p: 5, 10, and 20 K/min

The frequency (pre-exponential) factor A is then calculated based upon the assumption of reaction order of unity (n = 1) by the method of Zsako and Zsako [3-231:

-p

A=R

Where:

ln(1-a) C

-

= frequency (pre-exponential) factor a = fractional conversion

A

Eq. 3-20 mid dimensionless

C = empirical approximation whose value depends upon E The values of E and A are used to calculate the other kinetic parameters such as specific rate constant (reaction constant) k and half life time tl,2 with the help of equations 3-7 and 3-8: The Flynn and Wall method is well-suited to demonstrate the dependence of activation energy and frequency (pre-exponential) factor upon the conversion level.

14


3.3.1.4 Method according to McCarty and Green The McCarty and Green method [3-24, 3-25] only requires a single thermogram (TGA curve). The thermogravimetric analysis curve is assumed to be a continuous function such that, to each point on the curve there corresponds a temperature T and a level of conversion

c

c = LIL,

Eq. 3-21

Where:

L = instantaneous weight loss L, = final weight loss If the reaction follows a simple nth order rate law and the temperature dependence of the rate constant is described by Arrhenius' equation the thermogravimetric loss rate is given by the slope equation:

dC1dT = (Alp) exp(-EIR . T) (1 - C)" Where T = absolute temperature A = frequency (pre-exponential) factor P = heating rate E = activation energy R = universal gas constant n = reaction order

Eq. 3-22

K min-' K . min-' J . M01-l J . M01-l . K-' dimensionless

Integration of equation 3-22, treating the parameters A, E, and n as constants, yields:

Where: x =Ell? . T p(x)= exponential temperature integral function p(x) is usually approximated by using a series expansion containing enough terms to achieve the desired degree of accuracy. Taking logarithms of equation 3-23 yields: In F(C) = ln(AE1PR) + In P(X)

Eq. 3-24

Differentiating equation 3-24 with respect to x yields: d In F(C) d In p(x) --

dx

dx

Eq. 3-25

Recalling that dxldT"= E/R, the chain rule may be applied to equation 3-25: Eq. 3-26


15

A plot of In F(C) versus 1/T yields the required slope d In F(C)/d T'.This curve is nearly linear over the temperature intervals normally encountered in thermogravimetric decomposition analysis. Equation 3-26 is not a direct solution for the activation energy because d In p(x)/dx is a funciton of E as well as of T. However by choosing a trial value of x (e. g. 40) a first approximation of E can be calculated using equation 3-26. Better values of E are then obtained by iteration of the same equation. The frequency (pre-exponential) factor A can be derived directly from equation 3-23 at each data point once E is known.

3.3.2 Kinetic investigations on model substances 3.3.2.1 DSC experiments according to ASTM E 698-79 heat of vaporization of n-alkanes Pyrolysis experiments in an inert atmosphere on the series of n-alkanes reveal that even those with the longest carbon chains and thus the highest molecular weights evaporated more or less completely in the flow of argon (at 1 bar pressure) (see chapter 3.2.1) [3-121. Experiments performed at different heating rates demonstrate the expected shift of the peak

-0. BO 250

300

360

400

450

Temparatura (*C>

Fig. 3-52: DSC in Argon of Bitumen B 80 Shift of Peak Temperatures at Different Heating Rates p (5, 10, and 20 K/min)

500

76


maximum temperatures depending upon the heating rates (Fig. 3-52). The dependence of the peak maximum temperatures upon the carbon number of the n-alkanes at four different heating rates has already been described (Fig. 3-41). the shift of peak temperatures in relation to the heating rates permits the formal calculation of the activation energy E and the frequency (pre-exponential) factor A for each of the tested n-alkanes. This may be done using ASTM E 698-79 because, in the plot of the logarithm heating rate p verusus 1/T, the values fall in a straight line. The graph of the activation energy found thus versus the carbon number of the n-alkanes shows a logarithmic function similar to that already described in chapter 3.1.2 for the dependence of the peak temperature upon the carbon number: E = a, ln(C + a,)

+ a3

Eq. 3-1

The correlation coefficient is very high Y = 0.9997 (Fig. 3-53). Apart from the low level of the values for E, the shape of the curve contradicts the existence of an activation energy for a pyrolysis reaction. According to the references, the curve of the activation energy of a pyrolysis reaction versus the carbon number decreases asymptotically, starting from C2H6, and attains a final limit of approximately 250 kJ/Mol with n-C,,H,,. A curve of such a shape will be attained if the activation energies are divided by the corresponding molecular weights, resulting in the dimension of kJ/kg (energy per unit weight). The plot of those

160 140 120 100

BO 60

40 20 0

0

10

20

30

40

50

Fig. 3-53: DSC Kinetics (ASTM E 698-79) of n-Alkanes in Argon Activation Energy E versus C-Numbers Atmosphere: Argon 5 cm3/min, 1 bar

60

70


77

Heat o f Vaporization (kJ/kg) -01

- \ 500

400

30 0

200

0

10

20

30

40

50

60

Fig. 3-54: Heat of Vaporization H , of n-Alkanes versus C-Numbers Curve 1: Data according to R. C. Weast [3-101 [3-261 Curve 2: Data according to API Curve 3: Data from DSC-Tests in 1 bar Argon

values versus the carbon numbers gives an elongation of the curve of the heat of vaporization according to reference data, which have so far been listed only up to n-C,,H,, (Fig. 3-54). Comparison of the curve derived from DSC data with the curve derived from DTG data according to ASTM E 698-79 shows that identical values are found using both methods of measurement (Fig. 3-55). The deviation at carbon-numbers < 12 is caused by uncertainties in the determination of the DTG maximum temperatures for low-boiling substances. Increase of the pressure of the inert gas from 1 to 10 bar at constant flow rates causes a rise of the peak maximum temperatures. A strong linear correlation exists between temperature and pressure (Fig. 3-56): TIObar= 0.9188 Tlbe+74.576

Eq. 3-27

The correlation is very good: r = 0.9951. Therefore a linear correlation also exists between the values of the activation energy (heat of vaporization) at 1 bar to those at 10 bar pressure (Fig. 3-57): Elobar= 1.3105 Elbar+ 30.142 with correlation: r = 0.9998

Eq. 3-28

78


320 300 280

260 240 220 200

180 160

.

20 30 40 Fig. 3-55: Heat of Vaporization H, of n-Alkanes Curve 1: DSC-Tests Curve 2: DTG-Tests Evaluation according to ASTM E 698-79 10

50

60

70

80C

70C

600

500

400

300

200

Fig.3-56: DCS Tests of n-Alkanes 100

0 0

100

200

300

400

500

600

700

800

Peak Temperatures at 1 and 10 bar Pressure Atmosphere: Argon, 5 cm3/min Heating Rate p: 10 K/min


79

/

Elbar A r (kJ/Mol)

___t

0' 0

I 40

80

120

Fig. 3-57: DCS Kinetics of n-Alkanes (ASTM E 698-79) Activation Energy E at 10 bar Pressure (Methane) versus E at 1 bar Pressure (Argon) Gas Flow Rate: Argon, 5 cm3/min Heating Rate p : 10 K/min

The graph of activation energy E versus carbon number displays curves of logarithmic functions displaced parallel (fig. 3-58): In E , bar= 0.6677 In C + 2.220 r = 0.9975

Eq. 3-29

In Elobar= 0.5330 In C + 3.218 r = 0.9978

Eq. 3-30

80


E (kJ/Mol)

o

l

C-Number 10

0

10

20

30

40

50

60

70

Fig. 3-58: Activation Energy E versus C-Numbers of n-Alkanes DSC Kinetics Atmosphere: Argon, 5 cm3/min Curve 1: 1 bar Pressure Curve 2: 10 bar Pressure

The relation of activation energy per unit weight Hv(kJ/kg) versus C number is shown in Fig. 3-59, which also shows that the values of the coefficient and exponent of the exponential function at 10 bar pressure are twice the values at 1 bar:

HvI bar = 322.159 exp(-O.Ol 11 7 C)

Eq. 3-31

r = 0.9912 E*lObar= 679.528 exp(-0.0200 C) r = 0.9833

Eq. 3-32


81

1000

800

600

400

200

C-Number 0 0

I

t

10

20

30

40

-

I

50

60

70

Fig. 3-59: Heat of Vaporization H, versus C-Numbers of n-Alkanes Atmosphere: Argon, 5 cm3/min Curve 1: 1 bar Pressure Curve 2: 10 bar Pressure

In principle the enthalpy of vaporization may obtained by integration of the peak area in the curve of heat flow versus temperature. This mathematical calculation will not be precise, since the baseline drifts a certain amount due to the change of the specific heat during the heating. In addition the baseline does not return to the same level after the reaction due to mass loss during the reaction. Exact determination of the integration limits is therefore difficult, especially with the multicomponent systems (petroleum and its products) described below. Theoretically the enthalpy of vaporization is independent of the heating rate. In practice it is not independent, but shows a variable increase at increasing heating rates. However, the maximum temperatures of the evaporation peaks do shift depending on the heating rates. This fact permits the calculation of the activation energy of the evaporation process, which does not exhibit any temperature dependence. Theoretical considerations lead to the conclusion that this activation energy is equal to the molar enthalpy of vaporization.

82


3.3.2.2 Pyrolysis kinetics according to ASTM E 698-79 Most of the members of the series of n-alkanes evaporate in the DSC experiment at atmospheric as well as at 10 bar cell pressure due to the gas flow. Decrease of the gas flow rate from 5 to 1 or 2 cm3/min causes a true pyrolysis reaction for the higher n-alkanes, starting at about n-C,,Hs2. This may recognized by the fact, that the peak maximum temperatures of the n-alkanes tested, from n-C40Hs,up to n-C6,HIz2no longer differ at the corresponding heating rates but approach a mean value. However the variations from this mean value are very small, thus enabling the calculation by statistics of mean values for the activation energy E and the frequency factor logA (Table 3-14). A pyrolysis reaction, starting with n-C,,H,, may be achieved if the experiments are carried out using hermetically sealed aluminium pans. However, some of the pans burst during the experiment, giving a wider scattering of data as shown by the coefficient of variation in Table 3-14. The supplier guarantees a pressure stability up to only 3 bar for the hermetically sealed aluminium pans. A model calculation shows why the pans will burst in the experiment despite a surrounding cell pressure of 10 bar: The volume of a hermetically sealed pan is approx. 0.04 cm3.The input is 5 mg n-C,,H,6 corresponding to 0.029 d o l e . The evaporation produces 0.029 . 22.4 = 0.65 cm3 vapor (under normal conditions i. e. 1 bar pressure and 20 "C temperature) corresponding to a pressure rise up to 16.25 bar. Heating from 20 "C up to 500 "C (or 293 and 773 K respectively) increases the pressure by the factor 2.6 giving a pressure of 42 bar (for an ideal gas). Latest during the evaporation of the n-C,,H,, the pan will burst (BP = 216 "C). n-C,,H,, has a molecular weight large enough (282. 53) that the amount of vapor formed from 5 mg results in a pressure increase of only 10 bar. With increasing molecular weight, the volume of gas evolved from a constant weight of substance decreases asymptotically as does the pressure inside the pan. In addition from about n-C,,H,, the temperature of evaporation approaches the temperature of pyrolysis. The pyrolysis of HD polyethylenes, which do not evaporate due their molecular size, gives values for the arithmetical means of E and log A which are of the same magnitude as Table 3-14: DSC Pyrolysis of n-Alkanes and HD-Polyethylenes Statistical Evaluation -

X

n-Alkanes (C,,H,, to C60H122) 10 bar open pans n-AIkanes (C20H42to C6,H12,) 10 bar sealed pans HD-Polyethylenes 1 bar, open pans HD-Polyethylenes 10 bar, open pans HD-Polyethylenes 10 bar, sealed pans

E (W/Mol) 1 V (%)

-

X

logA (min-') +V (%)

246.2

4.64

17.6547

3.52

241.2

6.50

17.0272

5.62

248.0

2.81

17.0544

2.48

255.5

2.49

17.4947

3.14

245.3

7.53

16.2904

1.57


83

the values for n-alkanes with shorter carbon chains, and which are independent of the molecular weight (A4from 5 000 up to 388 000) (Table 3-14). Experiments using pressure resistant capsules (mini-autoclaves supplied by Bodenseewerk Perkin-Elmer GmbH) were not successful, becaue the heat capacity of the relatively Table 3-15: DSC Pyrolysis of Aromatics Substance

Anthracene 2-Methylanthracene 9-Methylanthracene 9,lO-Dimethylanthracene 9,lO-Diphenylanthracene 9,lO-Dihydroanthracene Phenanthrene 2-Methylphenanthrene Fluoranthene Pyrene 3-Methylpyrene 4-Methylpyrene Pentylpyrene n-Hexy lpyrene Coronene Triphenylene 1,2,5-Triphenylbenzene Triphenylrnethane p-Terphenyl

Open Pans 10 bar Methane E (kJ/Mol) log A (min-') 119.7 112.9 107.6 192.5 156.7 118.3 100.8 79.7 125.0 115.8 123.4 135.5 146.0 156.9 182.6 210.8 152.6 132.9

10.7958 10.4146 10.7203 17.5132 12.0777 11. 6848 9.1199 7.0810 10.9722 9.4749 10.2975 1I. 1242 12.4783 13.3390 12.7002 17.6870 11.9442 12.3585

Sealed Pans 10 bar Methane E (M/Mol) log A (min-')

207.4

12.5110

327.7 275.0

24.2467 19.7889

857.2 418.2

64.6681 26.4214

255.2 195.8"

19.9254 12.1465"

*Experiments in Mini-Autoclave Table 3-16: Gas Phase Pyrolysis of Aromatics (Literature Data) Substance Benzene Naphthalene Anthracene Phenanthrene Fluorene Toluene p-Xylene rn-Xylene o-Xylene 1,3,5-Trimethylbenzene Ethylbenzene Propylbenzene Styrene Biphenyl

Temperature Range ("C)

E (kJ/Mol)

log A (min-l)

800-870 800-875 785-825 835-920

468 446 529 452 120 3 15 322 354 315 37 1 221 311 142 384

21.517 20.579 25.026 20.436 6.437 14.868 16.066 17.231 15.954 18.153 13.224 18.429 6.971 18.149

790-890

560-660 570-635 650-720 750-850

84

3 Thermal AnaZysis on Model Substances

heavy mini-autoclaves is too large compared to the amount of heat evolved by the reaction of 5 mg substance. Aromatics evaporate (sublimate) similarly to n-alkanes in the flow of inert gas even at 10 bar cell pressure. (Table 3-15). The values of the activation energy ascertained are ordinarily too low to represent apyrolysis reaction, as shown by comparison with reference data (Table 3-16). The use of hermetically sealed pans gave valid data for only a few substances. It is dubious whether the data in Table 3-15 are true values of the activation energy and the frequency factor from pyrolysis reactions. For example, activation energy E for phenanthrene according to the literature is twice the value found by the experiment using sealed pans. Similarly, the value for triphenylmethane (E = 255 W/Mol) is markedly lower than the reference value ( E = 347 W/Mol). However the values ascertained for the two methylpyrenes are in the order of magnitude of other alkane-substituted aromatics. The decomposition temperatures of unsubstituted aromatics fall within such a range, that most will exceed the upper temperature limit of the DSC instrumentation. The aromatic ring systems are very stable to decomposition due to resonance. The free bonding enthalpy (-AGJ of the aromatic C-C bond amounts to 402 kJ/Mol and that of the aromatic C-H bond to 427 W/Mole [3-271. The bonding enthalpies for the aliphatic side chain of an aromatic ring (335 kJ/Mol) and the aliphatic C-C bond (297 kJ/Mol) are somewhat weaker. These differences are also reflected in the activation energies, which for the pyrolysis of an unsubstituted aromatic ring amount to more than 450 kJ/Mol; for an alkyl-substituted aromatic ring system E amounts to 320-370 kJ/Mol and for an aliphatic chain (C number I20) to 240-250 kJ/Mol. Experiments on aromatics were discontinued because the use of mini-autoclaves gave widely scattered data and also the upper temperature limit of the instrumentation (650 "C maximum) is too low for pyrolysis experiments on aromatics.

3.3.3 DSC oxidation kinetics according to ASTM E 698-79 Although oxidation is not a first order type of reaction, it has been found experimentally that it can be treated mathematically as a first order reaction with regard to the consumption of fuel, provided that there is an excess of air (oxygen). As already discussed the DSC experiment in air always exhibits three or four measurable peaks in the curve of energy flow versus temperature, which correspond at their peak maximum temperatures to the different reaction steps. They all shift to higher temperatures with increasing heating rates. That permits the calculation of the activation energy E and the frequency factor logA according to ASTM E 698-79 for each of the reaction steps. The first peak (Low Temperature Oxidation LTO) and the last peak (Fuel Combustion) can usually be evaluated without difficulty. In the region of fuel deposition it is often difficult to correlate the corresponding peaks to the different heating rates. Sometimes that can only be done manually by trial and error. This involves fitting as many data points as possible (at least three values), in the plot

85


of logarithm of heating rate Pversus 1 OOO/T, to a straight line. This regression line does not always give correct values of the kinetic parameters. That statement is also valid for the calculations using the data of the first or the last oxidation peaks, and generally for all methods of calculation, which use the diagram log Pversus 1 000/T for the calculation of E. The slope of the regression line log P= f(l O O O / r ) represents the value of EIR. The value of E is obtained by multiplication by the numerical value of R, (8.314 J . Mol-'. K-I). This may imply larger divergences of the value of E due to small deviations of the slope of the regression line. Instead of using the convenient computer program to calculate the regression line, it is advisable to repeat one of the experimental runs or to carry out further runs at additional heating rates, until the regression line in the plot log P= f(l OOO/T) easily fits at least three data points. The n-alkanes with shorter carbon chains evaporate prior to the start of the oxidation reaction as already discussed for the TGA oxidation experiments. Valid measurement is possible beginning with n-triacontane even at elevated air pressure inside the cell. The temperature data of the onset points of the first oxidation peak and the peak maximum temperatures of the n-alkanes from n-C,,H6, up to n-C6,H1,, show little deviation from an average value at each heating rate, as the small coefficient of variation shows (V= f 3 % maximum). Therefore the arithmetic means of the temperatures may be used to compute the activation energy E and the frequency factor log A (Table 3-17). However the restrictions for the Fuel Deposition Range mentioned above still apply. The mean of the oxidation start temperature (first onset point) is approx. 182 "C. The mean of the maximum temperatures of the first peak representing the LTO is 224 "C (both values at a heating rate P= 10 K/min and 7 bar air pressure). The experiments on HD-polyethylenes also give onset and peak maximum temperatures with a very small spread, so the kinetic parameters may be calculated using this mean values (Table 3-17). The mean of the onset point temperatures is found at 184 OC; the mean of the maximum temperature of the first peak at 232 "C (P= 10 K/min, P = 7 bar). Table 3-17: Oxidation Kinetics according to ASTM E 698-79 of n-Alkanes, HD-Polyethylenes, Naphthenes, and Olefins Atmosphere: Air; Pressure 7 bar Temperature Range

n-Alkanes Mean of 8 different samples HD-Polyethylenes Mean of 7 samples of different molecular weights trans-Decalin 1-Octadecene Polybutadiene

175 . . . 275 280 . . . 350 350 . . . 450 > 450 logA E logA logA E E E logA (kJ/Mol) (min-l) (kJ/Mol) ( m i d ) (kJ/Mol) ( m i d ) (kJ/Mol) ( m i d ) 114.8

11.942

100.6 114.2 116.4 66.8

10.085 116.6 11.787 94.2 12.220 140.6 7.659 86.1

-

-

166.5

10.370

195.9

13.178

10.142 6.812 12.389 6.686

-

-

175.7

11.753

-

-

-

-

-

-

158.6 209.4

10.450 15.158

-

86


The inclusion of only one carbon double bond in a longer carbon chain does not decrease the oxidation stability significantly, as the example of 1-octadecene demonstrates (Table 3-17). The onset point has a temperature of 185 OC, the maximum temperature of the first peak is 215 "C. The kinetic parameters of the first oxidation peak are nearly equivalent to those of the n-alkanes. For the second peak the data are higher and for the last peak lower compared to the n-alkanes. The high content of carbon double bonds in polybutadiene causes a considerable decrease in its oxidation stability. The onset point temperature is as low as 117.5 "C and the maximum temperature of the first peak as low as 175 "C. The activation energy of the LTO (first peak) decreases to 66.8 kJ/Mol (Table 3-17). Considerably higher onset point and peak maximum temperatures, as well as higher values of the kinetic parameters E and log A demonstrate the higher oxidation stability of unsubstituted aromatics (Table 3-18). Only coronene does not comply with this series. Substitution on the aromatic ring results in a decrease of the temperatures of the onset point as well as of the first peak maximum. As a result, the activation energy and the frequency factor are also lower. The length of the carbon chain of alkyl-substituted aromatics influences the oxidation behavior: an increase of the length of the alkyl chain causes a decrease in the activation energy and the frequency factor, evident in the alkyl-substituted pyrenes (Table 3-18). The last oxidation peak characterizes the combustion of carbon (Fuel Combustion), which has been formed by pyrolysis reactions in the preceding reaction range (Fuel Deposition). The very different values of the activation energy and the frequency factor for the last peak in Table 3-18 appear illogical. The means of the maximum temperatures of the last peak at a heating rate p= 10 K/min are: for n-alkanes 484 "C, for HD-polyethylenes 485 'C, for 1-octadecene 485 "C, and for polybutadiene 439 "C. For the aromatics the temperatures vary between 48 1 "C (4-methylpyrene) minimum and 558 "C (3-methylpyrene) maximum. Differences in the oxidation behavior of elemental carbon are well-known. In 1962 Levy [3-281 reported an activation energy for commercial grade graphite at 1 bar air pressure of E = 155.6 kJ/Mol and for pyrolytic graphite E = 104.5 kJ/Mol, presumably due to the considerably larger surface of the latter. From experiments on petroleum recovery by in situ combustion Fashihi [3-191 reports a range of values of E from 120 up to 140 kJ/Mol for the last reaction range (Fuel Combustion). In experiments on the oxidation of diesel soot in air [3-291 non-combusted fuel oil has been detected as an inpurity, which was oxidized at relatively low temperatures in a range from 260 "C up to 330 "C, requiring an activation energy of 32 up to 68 kJ/Mol. The soot itself had an ignition temperature of 500 "C up to 600 "C and required an activation energy of E = 154 up to 277 kJ/Mol. Comparative tests using TGA on commercial grade furnace blacks show onset temperatures between 400 "C and 450 "C. A soot with a graphite-like surface show the considerably higher onset temperature of 750 "C. In experiments made by the author on graphite oxidation did not occur because the temperature range did not suffice in the DSC instrumentation at 7 bar air pressure. TGA experiments demonstrate a start temperature of the oxidation (T1 %) for graphite of 650 "C and for activated carbon (charcoal) of 427 "C at 1 bar air pressure (Table 3-19). The

Anthracene 9,lO-Diphenylanthracene 9,lO-Dihydroanthracene Phenathrene Pyrene 3-Methylpyrene 4-Methylpyrene Pentylpyrene n-Hexylpyrene Triphenylene Coronene 1,3,5-Triphenybenzene Triphenylmethane 1,1,2,2-Tetraphenylethylene

Substance Tpet (C) 266 315 125 254 300 232 278 248 247 340 263 336 205 325 182.5 108.6 86.6 189.3 162.9 101.6 125.7 68.8 59.8 138.7 118.4 86.7 105.5 132.1

Peak 1 E (kJ/Mol) log A (min-') 17.673 8.437 10.034 18.616 14.411 10.093 11.001 5.697 4.868 11.271 9.873 6.380 10.335 11.006 -

-

-

-

4.890 8.295

-

76.9 93.3 -

-

-

-

-

-

-

13.718

115.9 -

-

-

Peak 2 log A E kJ/Mol) (min-')

Table 3-18: DSC Oxidation Kinetics according to ASTM E 698-79 of Aromatics Atmosphere: Air; Pressure 7 bar

-

12.820 5.672

-

164.7 85.3

-

7.989 8.686 10.993 16.141 7.219 8.136 8.849 9.167 6.094 12.888

-

114.8 115.2 143.0 205.0 105.3 115.1 126.2 128.8 89.6 178.2

Peak 3 E log A (W/Mol) (min-')

-

154.4

-

-

10.100

-

-

8.412 9.286 6.798 8.746 8.308 3.511

128.5 141.6 115.1 145.8 126.6 67.934

-

10.681 11.332 -

165.0 170.9

-

Peak 4 log A (kJ/Mol) (niin-')

E

cu

2

3.

2

;5

Rg.

2

L,

88


Table 3-19: Oxidation Kinetics according to ASTM E 698-79 of Elementar Carbon Thermogravimetry

Graphite 1 bar Air

Activated Carbon 1 bar Air

Activated Carbon 7 bar Air

T1 % ("C)* T5 % ("C)* DTG ("C)* E (kJ/Mol)** log A (min-l)**

65 1 724 830 153.4 6.433

427 539 630

-

-

-

-

-

476 592 170.7 9.715

434 544 154.4 9.267

DSC onset point ("C)* DSC-Maximum ("C)* E (kJ/Mol)* log A (mid)* ~

~

-

-

~

* Heating rate /3= 10 K/min ** Values from DTG Maximum Temperatures according to ASTM E 698-79 oxidation kinetics of graphite have therefore been ascertained by evaluation of DTG maximum temperatures according to ASTM E 698-79. Charcoal could be oxidized in the DSC instrumentation at air pressures of 1 bar and 7 bar. Oxidation experiments on different diesel soots in the Simultaneous Thermal Analyzer showed DTG and DTA peak maximum temperatures between 580 "C and 680 "C and onset point temperatures of about 400 and 420 "C [3-201 (p= 10 K/min). Consequently the susceptibility to oxidation of elemental carbon depends strongly upon the history of its origin and of the nature of its surface. If the oxidation test of pure hydrocarbons is interrupted after the second reaction step (Fuel Deposition) then a glossy, black, lacquer-like residue will be found in the sample pan. The oxidation of the latter results in heats of combustion in the region of pure carbon. The susceptibility to oxidation of the coke formed during Fuel Deposition will also be influenced by any residues of the original substances or their conversion products which are still present in the soot. The formation of the deposited carbon in the reaction range Fuel Deposition is a complex process. Conjugated diolefins react at temperatures of 550 "C up to 750 "C forming unsaturated cycloalkenes and aromatics requiring a weak activation energy of only 104 kJ/Mol [3-321. The primary reaction step of the pyrolysis of benzene at a temperature of 700 - 800 "C is the formation of biphenyl. The reaction continues to form pyrene ( E 230 kJ/Mol). The reaction range of Fuel Deposition in DSC lies between the temperature limits 300 450°C. But local higher temperature peaks may exist. The sample temperature often overtakes the programmed instrument temperature just in the range of Fuel Deposition. This could be an indirect proof for the existence of such temperature peaks. At the other end of the scale, polymers form easily aromatics at relatively low temperatures. The polymerization of acenaphthene starts as low as 210 "C [3-331. The biacenaphthylidene formed first, will be converted at about 360 "C to fluorocyclene, which will condense at temperatures > 480 "C to graphite-like aromatic ring systems, after passing through some intermediate stages. The atomic ratio H/C decreases during this process.

-


89

This type of reaction is not restricted to oxidation processes but also occuls during pyrolysis in inert gas. However, the temperature peaks were not found during pyrolysis. The multitude of reaction mechanisms possible during the oxidation of multicomponent systems accounts for the appearance of several peaks in the Fuel Deposition reaction range. The number of peaks in this range decreases with increasing heating rates, because some reactions have such a slow reaction rate that they may be overtaken and obscured. The same applies to the effects of increasing pressure, because the reaction rates of different components will also vary.

3.3.4 Kinetics according to Borchardt and Daniels Experiments on model substances were not carried out. A comparison of the oxidation and pyrolysis kinetics according to Borchardt and Daniels with the other three methods will be made in chapter 4.10.

220

I---------’----------

0

Fig. 3-60: Reaction Kinetics according to Flynn and Wall [3-21, 3-22] Pyrolysis of CaC,O,. H,O Atmosphere: Argon, 30 + 20 cm3/min, 1 bar Plot: Activation Energy E versus Conversion

90


3.3.5 TGA kinetics according to Flynn and Wall The reactions during the heating of calcium oxalate monohydrate CaC,O,. H,O are shown in Fig. 3-51, where the points of isoconversion at three different heating rates (p= 5, 10, and 20 K/min) are marked and connected by straight lines. According to Flynn and Wall it is possible to calculate the dependence of the activation energy upon the level of conversion (Fig. 3-60). To compare the results with those found by the ASTM E 698-79 method it is useful to evaluate the three regions of reaction separately. The activation energies and the frequency (pre-exponential) factors should be calculated at the conversion levels corresponding to those at the DTA- (DSC-, DTG-) maximum temperatures. In Fig. 3-61 the first reaction region (evaporation of the water of crystallization) is shown ranging from 100 to 88 % weight. The points of isoconversion are connected by straight lines. The plot of In heating rate p versus the temperatures of the points of isoconversion (1 000/Kelvin) shows that the data obtained at different heating rates fit very closely to a straight line at any conversion level (Fig. 3-62). The plot of the activation energy versus percent conversion (Fig. 3-63) shows a descending line. The conversion at the DTG maximum temperature amounts to 7.28 % weight loss. The table of kinetic results

102

.5 .4

loo .3 98

.2 m

U

+I

.1

96

; >

0

0

L

0.0

m

a

h .-I

I

94

-.l -.2

92

-.3 90 -.4

88

Fig. 3-61: Reaction Kinetics (Flynn and Wall) Pyrolysis of CaC,O,. H,O + CaC,O, (First Reaction Step) Conversion: 11.07 % Raw Data Plot of STA 780

-.5

+ H,O?


91

Fig. 3-62: Reaction Kinetics (Flynn and Wall) Pyrolysis of CaC,O,. H,O First Reaction Step Plot of STA 780: In Heating Rate versus 1 000/T

(Fig. 3-64) demonstrates an activation energy E = 71.22 H/Mol and a frequency (preexponential) factorA = 2.084 . 108min-' (log A = 8.3189 min-') at a relative conversion of 60 %, corresponding to 7.38 % absolute conversion. The measurement of the other reaction regions may be done in a similar way. A comparison of the values found by the Flynn and Wall method, with the values from the calculation according to ASTM E 698-79, both corresponding to the DTG maximum temperatures, is set out in Table 3-20.

92


90

a5

80

75

70

65

60

I

20

I

30

I

1

50

40

-

CONVERPIOM 80 X I RXN OROER 1 I ACTIVATION ENEREY I KJ/mol. 1: PRE-EXPONENTIAL FACTOR Il/mlnl:

I

60

I

I

70

Percent Conversion

2.46

X

C RXN ORDER 3 1 1 COHVERSION: 40 X ACTIVATIOH ENEROY I KJ/mOlm I : 77.06148 PRE-EXPONENTIAL FACTOR I l h i n l : i .7909Cl5E+OP

4.92 I

I RXN ORDER 1 1 CONVERSION: ¶O X ACTIVATION ENERBY I kJ/melo I : PRE-EXPONENTIAL FACTOR 1l/mlnl:

74.46738 S.S71974E+08

6.15 X

C RXN ORDER I 1 1 CONVEREION: 80 X ACTIVATION ENEREV I kJ/nol. I : PRE-EXPONENTIAL FACTOR Il/mlnl:

71.22i?58 2.084L34E+08

CONVERBION: 70 I I RXN ORDER 1 1 ACTIVATXON ENEREY 1 YJ/mol. I: PRE-EXPONENTIAL FACTOR Ll/min):

67.91453 S.SO9329EiO7

I RXN ORDER i 1 CONVEREION: 00 X ACTIVATION ENERDY I k J / m l . I : PRE-EXPONENTIAL FACTOR I l h l n l :

84.84087 2.30089DE+07

-

-

[ RXN ORDER i 1 CONVERSION: 90 X ACTIVATION ENEROV I KJ/moI. 1: 56.9.789 PRE-EXPONENTIAL FACTOR l l h i n l : 33.7291

F i r s t Reaction Step : CaC204. H20 -CaC Theoretical Conversion : 12.3 I OTG Maximum Conversion : 7.28 X

I

100

l.E78114€+10

3.69 I

-

I

90

08.38ma3

CONVERSION ao x RXN ORDER L 1 ACTIVATION ENEROV 1 LJ/moI. 1: 79.65492 PRE-EXPONENTIAL FACTOR I l h i n l : ¶.100S86E~09

-

80

0 + H20t 2 4

7.38

s

8.61

X

9.84

s

11.07 X

Fig. 3-64: Reaction Kinetics (Flynn and Wall) Pyrolysis of CaC,04. H,O First Reaction Step Plot of STA 780: Table of Results

CaC,O,. H,O CaC,O, CaCO,

Reaction

t

t

= CaC,04 + H,O = CaCO, + CO = CaO + CO,

7 7.28 24.92 52.72

71.2 215.0 138.2

8.3189 14.5039 6.7246

Conversion at DTG Flynn and Wall E lo@ Maximum Temperature (%) absolute (k.l/Mol) (min-')

Table 3-20: Decomposition Kinetics of CaC,O,. H,O in Inert Atmospehere

69.0 7.8140 205.1 13.5846 162.2 7.8771

ASTM E 698-79 E logA (kJ/Mol) (min-')

115.38 300.24 287.77

14.0885 20.5942 15.0917

McCarty and Green E logA (kJ/Mol) (fin-')

94


3.3.6 TGA kinetics according to McCarty and Green The McCarty and Green method is of economical interest because only one test run is required for the calculation of reaction kmetics. In the evaluation of a decomposition experiment on calcium oxalate monohydrate, the temperatures of the onset and offset points of each of the three consecutive reactions were used as integration limits. The results are shown in Fig. 3-65. These data diverge considerably from those calculated using the Flynn and Wall or the ASTM E 698-79 methods (Table 3-20). Considerable differences in the data are also reported in the literature. Freeman and Carol1 /3-34/ reported the following data found by thermogravimetry: reaction 1: E = 92.1 kJiMol reaction 2: E = 309.8 kJ/Mol reaction 3: E = 163.3 WiMol While they do not report any data from references concerning reactions 1 and 2, they cite varying values from references for reaction 3, the removal of CO, from CaCO, (kJ/Mol): 205.2; 145.5 up to 175.9; 397.8; 171.7 up to 184.2; 154.9 up to 163.3; 201.0. The attraction of this quick evaluation method led to an attempt to calculate the activation energies of evaporation for the series of n-alkanes, but, depending upon the limits of integration chosen, different values showing relatively widespread scattering

-

In Z:

34.75

i/nin

lag 2 : 15.0917

I

1

iOC

200

300

I

430

I

500 rjeg 2

I

600

Fig. 3-65: Reaction Kinetics according to McCarty and Green [3-241 Pyrolysis of CaC,O,. H,O Heating Rate p: 10 K/min

I

7CO

1

ROC

0

3.3 Reactiori Kinetics

95

were found. Dividing the activation energics by the corresponding molecular weights gives the heats of evaporation E+ (in kJ/kg) (see chapter 3.2.2). The following limits of integration were tested: upper limit

lower limit

1OU '3, weight onset point 100 c/c weight onset point

50 '5, weight offset point onset point 70 % wcipht

The point at 70 % weight corresponds to thc transition from the bend to thc linear descending part of thc TGA curve. In all cases values of E* were found which considerably exceed those found by DSC experiments and calculated according to ASTM E 698-79. This is shown in Fig. 3-66 using the cxamplc of data calculated using the integration limits onset point + offset point (crosses) and 100 % weight -+ onset point (circles).

10

20

30

40

50

60

Fig. 3-66: Kinetic of Vaporizalion of n-Alkanes according to McCarty and Green Atmosphere: Argon, 30 + 20 cm3/min, 1 bar Heating Rate p: 10 K/rnin x Integration Limits: Onset Point to Offset Point o Integration Limits: 100 % Weight to Onset Point


4 Thermoanaly tical investigations on petroleum and petroleum products

Petroleum is a multicomponent system. With regard to its thermal behavior, it comprises a multitude of compounds ranging from very light components, which even evaporate during the production phase, through to heavy substances which do not evaporate due to their lower decomposition temperature. The distribution of these substances varies between the various oil fields. It is possible to differentiate between light crudes, medium crudes, and heavy crudes, but each consists of many individual compounds. A distillation fraction such as gasoline, with a boiling range of 3.5-200 "C comprises 2.5 isoparaffins, eight n-paraffins, four naphthencs, and five aromatics, as reported by Zerbe 14-11. Thermal refining of the crudes consists of two basic operations i.e. atmospheric and subsequent vacuum distillation. The boiling range of the atmospheric distillates extends to approximately 370 "C and yields the fractions: - liquid petroleum gas (LPG) -

gasoline

- light gas oil - heavy gas oil

Vacuum distillation of the atmospheric residue (temperature approximately 400 "C and 0.1 bar pressure) yields the fractions: -

spindle oil

- machine oil - cylinder oil

The vacuum residue may be used for bitumen heavy fuel oil - bunker C oil (marine diesel fuel) -

The individual components of the light fractions can be separated and defined more or less easily using modern analytical methods; for the high and non boiling fraction it is only possible to separate compounds into groups, characterized by similar analytical behavior. Individual substances can only be isolated in exceptional cases. Investigations concerning the kinctics of the pyrolysis and oxidation behavior of fractions from petroleum refining, will therefore only provide average values from a series of

98

4 Thet-rnouticilvtic,alItiwsligutions on Petroleum mid Petroleicm Products

parallel and consecutive reactions. The practical behavior of the multicomponcnt systems in thermal refinery processes may be satisfactory described by thermal methods of analy sis. Thermogravimetry is well-suited to describe the distillative, cracking, and oxidative behavior of crudes and petroleum products. The thermal behavior may be described

Table 4-1: Index Numhcrs for I'hcrinogravinietry Temperature ("C) at I % weight loss Representative of the start of evaporation (boiling) 'lempcraturc ('C) at S %: weight loss Same meaning as TI %, but better repeatability Weight loss (wt 5%) up to 400 "C Part of' thc sample which may still be separated by means of distillative method\. (Experiments in inert gas) Residue (wt 5%) at 600 "C Coke residue (Experiments in inert gas) IZesidue (wt %) at 800 "C Coke residue (Experiments in inert gas) Maximun~of thc loss (cvaporation, crack) rate (first differential quotient of thc TGA curve with respect to time) related to the weighed portion (%/min or pg/min.mg) Indicates the kinetics of the reaction Tempcraturc ("C) at the peak maximum of the DTG curve. Indicates if the reaction is still in the distillation range Tnnx < 400°C or in the crack range Tmx < 400°C. (Experiments in inert gas). During experiments in air the first DTG maximum may indicate distillation as well as oxidation. Extrapolated temperature ("C) of the point of inflexion (off-set point) of the TGA curve. End of the crack reaction (Experiments in inert gas) Weight ( ~ 1 % of' ) the residue at the tcmperature TW. Coke residue (Experiments in inert gas)

Table 4-2: Derived Index Numbers for Thermogravimetry

ND = 100-AG400

Nondistillable pait of the sample (wt %) (Experiments in inert gas)

CK = 100-(AG400+R800)

Part of the sample which may cracked thermally (wt %) (Experiments i n inert gas)

= ND

R800 ND CR ND

'

100

Conradson coke residue in the non-distillable part of the sample

("/.I

(Experiments in inert gas)

. 100

Residue at 800 "C in the non-distillable part of the sample ('70) (Experiments i n inert gas) Crackablc part in the non-distillable part of the sample (Experiments in inert gas)

3. I Crude 0il.v (Degasifietl Crudes)

99

(Table 4-1) using a series of empirical index numbers. Further characteriLation may be achieved using othcr index numbers derived from those aforementioned (Table 4-2). Quantitative statements concerning the kinetics of the pryrolysis and the oxidation behavior may be derived using IISC.

4.1 Crude oils (degasified crudes) Thernioanalysis on crude oils is only possible in hermetically sealed sample pans because these oils contain components which will degass at relatively low temperatures under atmospheric pressure. This behavior leads to considerable experimental difficulties; for example it is very difficult to avoid losscs during the filling of the sample pans. Degasified crudes on the other hand may be handled easily. The evaporation behavior of a representative selection of dcgasified crudes from German oil fields has been investigated using thc Stanton-Redcroft TG 750 thennobalance 14-21. The data characterizing these crudes are given in Table 4-3. It has been found that the evaporation of some crudes starts at room temperature, as soon as the oven is closed (Table 4-4). Thus the start tenipcrature of the evaporation cannot be clearly defined and so the beginning of the simulated distillation curve (see chapter 3. I .2) diverges considerably from the Engler boiling curve (DIN 5 1 75 I ); this is shown in Fig. 4- 1 using the example of the degasified crude (33.5 "C API) from the Landau oil field (Southwest Germany). During distillation according to DIN 5 I 75 1, cracking starts at 380 "C, whereas the simulated distillation curve gives an atmospheric boiling temperature of 440 "C. The degasified crudes from German oil fields which have been investigated, are usually light oils containing a high proportion of compounds which evaporate at experimental temperatures below 400 "C. The DTG curve often shows more than one peak maximum, the second and the third peaks usually appearing at temperatures above 400 "C, in the cracking range. A considerable residue is present at 600 "C (R600), but at 800 "C the residue (R800) is very small, thus indicating that the coking reaction is not yet complete at 600 "C. The relation of the temperature of the TGA (TTG)to the corresponding temperature of the distillation according to DIN 51 751 (T,) for corresponding evaporated portions shows an increase of TD/TTG with increasing amounts of distillate. The plot T,)/TTGversus % distillate has a linear function: TD/TTG = 0.00495 . % ' distillate

+ 0.38035

with a coefficient of correlation r = 0.98831 (Fig. 4-2).

Welpe Landau Darchi ng Monchsrot Aitingen Assling

Haime

Bramberge Gross Lessen Scheerhorn Hockstedt Diiste Aldorf

29.7

30.6 36.7 33.5 35. I 38.1 30.3 33. I

0.8695 0.8379 0.8545 0.8460 0.8312 0.87 14 0.8577

29.0 30.5 29.7 27.0 29,s 29.1

0.8747 0.8782 0.8703 0.8746 0.8896 0.8757 0.9777

33.4 31.6 33.7

0.8552 0.8642 0.8534

I .4 13.9 -

I .4 5.3 4.7

-

283 292 356

-

288 307

-

585 321 318 284

-

16.39 10.1 20.8') 3.4 14.9 3.6

-

-

435

1.49 0.48 3.00

328 342 352

0.9 1.6 2.7

-

1.5 4. I

-

0.35 I .8 8.75 2.7 3.3 5.9

-

1.03 0.95 1.70

(70 wt)

(Yc wt)

-

-

-

-

-

-

-

-

-

2.35

-

3.70

-

2.13 2.49 2.73

(% W t )

Clonradson Coke

Dogger beta Dogger beta Dogger beta Cornbrash Dogger alpha Valanginian Valanginian Valanginian Valanginian, Wcalden Valanginian, Jurassic Jurassic Medium Kitnmcridge Jurassic Upper Kimmcridge Jurassic Tertiary Molassc, Priabonian Eocene Eocene Eocene, 'Icrtiary

Asphaltenes

Petrolcutn Resins

Wesendorf-Siid Vorhop-Knesebeck Hankensbuttel-Sud Kronsbcrg

Molecular Weight (Mean)

2

"AH

Geological Formation

Oil Field Specific Gravity at 20 "C (g/cm3)

3

g-

5 k a

2

23

P

P

2

$

2

*a

2

2

B 6'

==.

.d

-2

2

h

5

0

Table 4-3: Degasified German Crude Oils

%

2

4

4. I Crude Oils (Degnsificd Crudes)

30

Tt

-

r.c

Ic,

*

3Nr-mNCrC-

mmTt-*-Tt

10-

c1-

101

Eocene, Tertiary Eocene

Molasse Priabonian

Eocene Eocene

Assling Arlesried

Holzkirchen

Monchssot Aitingen Bramhar

-

Molasse Priabonian

Darching

Landau Velden 111

'

28 36 35 43

22 18 16

39 45

38

11

30

18

22

19 25 20

Upper Kimmeridge Jurassic Tertiary Molasse

35 46 55

44

28

Welpe

52

25

Jurassic Medium Kimmeridge Jurassic

Aldorf Hasme

T5 % ("C)

Geological Formation

Oil Field

TI % ("C)

Table 4-4: Thermogravimetry of Degasified German Crude Oils Heating Rate p = 10 K/min Atmosphere Argon, Flow rate = 25 cm3/min

97.1 88.7 70.5

91.2

92.2 96.7

89.0

90.1 85.5 85.0

81.8

83.7

(%I

AG400

2.0 6.9 7.3

4.3

3.4 0.7

0

0.5 6.4 5.3

4.0

5.9

(%I

R600

0

0.8 1.1

2.0

2.0 0

0

0 0.2 4.4

1.5

0.2

(%I

R800

4.93 1.85 3.20 2.24 1.77 2.96 2.76

4.41 2.21 0.75 2.24 3.25 0.4

3.12 2.67 2.74 1.34

2.14

2.27

DTG (%/min)

c

Plateau from 40 "C up to 460°C

35 250 171 166 86 389 435

51 218 463 107 126 445

172 263 143 435

5

2

B 0

5 "cl a

z

2a

&

n

z g

2a

a

Gl

5

2. R g. 09

3

186

n

30

3

h

h,

0

4.1 Crude Oils (Degasified Crudes) -

450

1

T

("C)

-

400

350

300 -

250 I

I

re

200

150 -

100

?

- I -I -I

25

50

75

Fig. 4-1 Distillation Curves of Crude Oil (Landau Field, Germany) Curve 1: Simulated Distillation by TG 750 Curve 2: Boiling Curve DIN 51 751

103

104

4 Themoanalytical Investigations on Petroleum and Petroleum Products

-1

0

5

10

15

20

25

30

35 40 45 50 55 6 0 65 70

75 00

Fig. 4-2 Quotient of the Temperatures of Distillation T, (DIN 51 751) divided by the Temperatures in the Thermobalance TTGversus Quantity of Distillate. (Mean Values from German Crudes)

It is advantageous that thermogravimetry is not affected by even considerable water content in the sample, which might cause frothing over of the sample during distillation according to DIN 51 751. Fig. 4-3 shows the plot of TGA versus temperature for a degasified crude from the Scheerhorn oil field (Northwest Germany), firstly in the original state containing about 50% formation water and secondly after dewatering. Curve No. 1 in Fig. 4-3 shows the water content by the weight loss up to 100 "C. No problems arise with thermogravimetry on heavy crudes. The boiling curves simulated by means of thermogravimetry of a heavy crude from the Morichal I1 oil field (Venezuela), 9.3" API, and of crude from the Job0 oil field (Venezuela), 8.5" API, are shown in Fig. 4-4. This figure also shows a deviation of the boiling curve simulated by means of thermogravimetry, from that simulated by means of gas chromatography. The deviation is especially large in the range above 450 "C atmospheric boiling temperature. The crack reaction starts earlier in GC than in TGA. Oxidation starts at temperatures above 300 "C as already shown in the investigations on model substances. Fig. 4-5 shows the thermograms of the heavy crude from the Job0 field in argon and in air.

4.1 Crude Oils (Degasifed Crudes)

100

Res

7; ->T \

t \

\

15

--+ I!

T

1

50 __

i

\ \

25

- 400 "C represent crack (pyrolysis) processes. Of the 13 bitumens and vacuum residues in this investigation, only two samples show maxima in the evaporation range as well as in the crack (pyrolysis) range. The other 11 samples have only one maximum, in the cracking range. The four atmospheric residues possess at least two maxima, one of them in the evaporation range between 300 "C and 320 "C, the other in the cracking range between 440 "C and 465 "C. Of the products from conversion processes, samples 18, 23, 24, and 25 exhibit only one maximum each in the distillation range, whereas samples 19, 20,21, and 22 have only one maximum, in the cracking range. Statistical evaluation of all maxima in the distillation range results in a mean of the temperature Tm X = 3 11.1 "C with a coefficient of variaton & V = 8.2. The same evaluation

4.2 Refinery Residues

133

of the maxima in the cracking (pyrolysis) range results in a mean temperature of TmxX = V = 3.1 9% regardless of the origin of the sample. 451.5 "C with a coefficient The maximum of the weight loss rate DTG is found within the following limits: DTG (%/An) Minimum Maximum Distillation Processes Cracking Processes

2.72 1.11

9.04 11.43

The lower values are normally found if the process is spread over a wider temperature range. A high value of DTG is generally connected with a narrow range of temperatures within which the reaction takes place. This is an indication that a relatively uniform substance distills or is cracked.

Table 4-17: Maximum of Weight Loss Rate Related to the Actual Residual Weight Sample No.

10 11 12 13 14 15 16 17

Cracking Range DTG (%/min)

Distillation Range DTG (%/min)

-

-

5.86 4.28 5.33 5.35

319 306 300 306

18 19 20

12.05

296

21 22 23 24 25

5.09

409

8.40 28.24 17.93

298 346 274

-

-

15.69 16.85 24.99 15.63 21.44 14.56 13.89 20.24 13.14 21.60 8.80 12.40 18.48 16.33 18.49 14.95 5.84 7.73 16.05

470 475 44 1 472 478 465 470 482 433 454 419 449 448 438 456 440 480 464 62 1

12.23 12.30 15.90 5.47 11.22

447 435 45 1 456 448

-

-

-

-

-

-

134

4 Thermoanalytical Investigations on Petroleum and Petroleum Products

The DTG data of the cracking (pyrolysis) reactions for the substances which had already experienced weight loss through distillation, exhibit an additional error, because the value is related to the start weight and not to the residual weight, after distillation losses. Table 4-17 gives the corrected values of the maximum weight loss rate, referred to the residual weight at the corresponding maximum temperatures T-. The extrapolated temperature T, of the point of inflexion of the TGA curve (offset point) is found in the range from 500-550 "C with the exception of low boiling samples 23, 24, and 25, which exhibit temperatures from 325-375 "C. Statistical evaluation of the temperatures T, gives the following results: Mean Statistical evaluation of all samples (1-22) Bitumens and Vacuum residues (Samples 1-13) Atmospheric residues (Samples 14-17) Products from conversion processes (Samples 18-22)

X ("C)

Coefficient of Variation +V(%)

505.6 508.1 500.5 503.4

4.75 3.87 2.81 8.12

Variance analysis shows that all four means are within the confidence level, and do not demonstrate any statistical difference. Samples 18-22 are very different conversion products, and show the largest scattering. The residues G, are not as uniform within the different product groups as the temperatures T,. Regarding the group of vacuum residues (bitumens, samples 1- 13)the G, data of samples 6, 7, 10, and 11 stand out distinctly. Statistical evaluation of these samples separately results in a considerable higher mean: Residual weight G, (wt%) at the point of inflexion of the TGA-curve:

Samples 1-5, 8, 9, 12, 13 Samples 6, 7, 10, 11

x (wt %)

*V(%)

20.2 31.4

9.80 3.36

Contrary to the T, values there is a statistical difference in the means of G,. Temperature T, ("C) at the point of inflexion of the TGA-curve:

Samples No. 1-5, 8, 9, 12, 13 Samples No. 6, 7, 10, 11

x ("C)

* V(%)

507.4 509.5

2.85 6.17

Samples 6, 7, 10, 11 are blown bitumens (partly oxidized) whereas the other samples represent vacuum residues (distillation bitumens). The Conradson coke residue CCR of samples 6,7, 10, and 11 is within the limits of the other samples of this group. The corresponding temperatures TccR, with values of

135


600-700 "C, exceed the mean of the other nine samples substantially (2 = 522.1 "C, 5.27 %).

fV =

The data for the residue G, at the temperature T, clearly exceed the corresponding data of R800. On the other hand the G, values are comparable to the corresponding Conradson coke residues GccR.The same is valid if the temperature TccRis ascertained from the TGA diagram using the corresponding weight of CCR (GCCR).There is a relatively good coincidence of TWITccRand G,/GccR (Table 4-18). For complete coincidence both quotients should equal one. Table 4-18: Correlation of temperature and residual weight data from TGA and CCR methods. Sample No.

1 2 3 4 5 6 7 8 9 10

II 12 13 14 15 16 17 18 19 20 21 22 23 24 25

T,

roc] 515 5 12 500 530 515 518 510 520 495 480 490 485 495 485 487 520 500 450 530 555 500 482 350 375 325

G, [wt %]

TCCR

GCCR

L

G,

[%I

[wt %]

TCCR

GCCR

16.5 22.0 18.0 20.5 19.0 31.3 30.0 22.3 20.5 31.8 32.5 21.8 21.2 9.8 14.3 16.0 23.3 14.2 24.2 35.0 35.0 39.3 34.3 10.5 11.5

512 515 550 520 512 750 715 580 520 600 660 485 505 535 620 795 630 575 545 535 620 470 682

16.4 21.3 21.3 21.7 19.0 21.5 21.6 20.2 19.7 20.1 24.9 20.8 20.2 8.6 10.4 8.9 10.6 9.9 25.7 39.4 22.9 41.4 9.9 0 2.4

1.006 0.994 0.909 1.019 1.006 0.691 0.713 0.897 0.952 0.727 0.742 1.000 0.980 0.943 0.785 0.654 0.794 0.783 0.972 1.037 0.806 1.026 0.513

1.006 1.033 0.844 0.945 1000 1.456 1.389 1.104 1.041 1.532 1.305 1.048 1.050 1.140 1.375 I .798 2.198 1.434 0.942 0.888 1.528 0.949 3.465

-

630

-

-

0.516

4.792

Data for the blown bitumens (samples 6, 7, 10 and 11) and for the low boiling samples 23-25 deviate from those of the other samples. Data for 13 samples (1-5, 8,9, 12-14, 19, 20,22) show such a small scattering that statistical evaluation of the quotients TWITccRand GW/GccRis merited: Mean TWlTCCR GWGCR

X

0.980 0.999

Coefficient of Variation L V (%)

4.47 8.35

136


Thus the equivalence of T, and TccR,as well as of G, and G,, has been proved. The statistical evaluation of TCCR results in a mean X = 585.9 "C with a coefficient of variaton +_ V = 14.34 %. It is well known, that measurement of the Conradson coke residue exhibits a large scattering. DIN 51 551 speaks of a repeatability of 10 % and a reproducibility of 15 % of the values. Three groups may be formally distinguished, which show a smaller scattering of the TCCk

x ("C) Group I (Samples 1-5, 8, 9, 12-14, 18-20) Group I1 (Samples 10, 11, 15, 17, 21, 25) Group I11 (Samples 6, 7, 16, 23)

525.6 626.7 735.5

*If(%,)

5.84 3.14 6.58

Groups I and I1 are in coincidence with the results given by Adony 14-12], where a temperature range from 475-650 "C is given for the coking according to Conradson. No correlation with analytical (chemical) data of the sample can be made.

4.2.3.2 Derived index numbers The non-distillable part ND = 100 - AG400 of vacuum residues and bitumens is within the limits from 56 to 89 wt%, and for atmospheric residues from 33 to 40 wt%. For the conversion products ND depends on the distillation severity after conversion. ND increases for heavy products up to more than 75 wt%. Statistical evaluation results in the following data: ~

~~~

~

Vacuum residues and bitumens Atmospheric residues

~~

x (wt %)

f V (%)

76.9 37.2

10.4 9.3

The part of the sample which can be cracked, CR = 100 - (AG400 + R800) amounts to 43-73 wt%. The lower value is due to the fact that sample 1 contains a relatively high distillable fraction (AG400 = 44 wt%). A statistical evaluation of samples 2 to 13 results in a mean of CR X = 61.8 wt% with a coefficient of variaton + V = 12.6 %. Due to the higher distillable fraction, the CR of the atmosphericresidues has lower values between 26 and 40 wt%. For the products from conversion processes which do not possess an R800 (samples 24 and 25),the CR equals the ND. The CR of the other products from conversion processes is in the limits from 12 wt%, for the residue from a thermal cracker having a very high distillable fraction (sample 18), up to 53 wt%, for a visbreaker residue (sample 21). Statistical evaluation of the ratio CR/ND, which represents the crackable portion of the non-distillable part of the sample, gives the following results:


Vacuum residues, bitumens, and atmospheric residues (Samples 1-17) Products from conversion processes (Samples 18-20 and 22)

137

x

*V(%)

0.80 0.63

5.1 4.8

The lower value of the ratio CR/ND for products from conversion processes indicates that considerable amounts of these products have already been cracked and distilled. The high value CR/ND = 0.88 for the visbreaker residue, sample 21, demonstrates on the other hand that in that case the crack conditions were very light. The Conradson coke residue in the non-distillable part of the samples (CCR/ND) . 100 is in the limits from 23 to 33 % for vacuum residues, bitumens, and atmospheric residues. The statistics result in a mean X = 27.3 % and a coefficient of variation +V = 9.6 % (relative). The products from conversion processes scatter from 33 to 58 %. The furfural extract (sample 24) stands out because it does not possess any coke residue. The data for the residue at 800 "C in the non-distillable part ND of the sample (R800/ ND) . 100 for vacuum residues, bitumens, and atmospheric residues are in the limits from 10 to 26 % with a mean X = 20.0 %. The very high coefficient of variation fV = 20.6 % indicates that statistical evaluation is nonsensical. The inaccuracies of the measurements are cumulative here. As shown in chapter 4.2.2.1, the coefficient of variation for ND is k 6.9 %, and for R800 it is k 7.4 % (maximum each), so that the scattering of (RSOOIND) . 100 could theoretically increase up to _+ 15.4 %. The residues from thermal crackers (samples 18 and 22) as well as a visbreaker residue (sample 19) and a cat cracker residue (sample 20) exhibit data far above 30 %. The high values of (CCR/ND) . 100 and (R800/ND) . 100 for these products show again, that considerable portions from these samples have already been cracked and distilled. However, samples 17,24, and 25 do not show any value of (R800IND) . 100 as they have no R800.

4.2.3.3 Simulated distillation As discussed in chapter 3.1.2 above, it is not possible to obtain an equilibrium of evaporation in thermogravimetry,because the gas flow rate is relatively fast. The layer of substance in the sample pan has a thickness of only 0.4 or 0.5 mm, so that the conditions in the TGA resemble those of a thin-layer vacuum distillation. The distillation curve of the sample at atmospheric pressure may be calculated step by step from the TGA curve by means of a linear function [4-131:

BP = a, TTG+ a2 where

BP = boiling point TTG= temperatures of a defined weight loss a,, a2 = constants depending upon the weight loss.

138


The constants a, and a2may be determined by calibration using pure chemicals of known boiling point. In this investigation the calibration was performed using n-alkanes from C,,H,, to C,,H,,, and chemicals mentioned in ASTM D 2887-84, Table X 1.1. Figures 4-18 to 4-25 show the simulated distillation curves of the samples. Fractions of the sample boiling below 400 "C may be separated by means of atmospheric distillation whereas the parts boiling between 400 "C and about 550 "C could be removed only by means of a vacuum distillation. The distillation curves of the vacuum residues and bitumens are shown in Figs. 4-18 to 4-20. Due to the mode of manufacture, none of the samples possesses a distillable fraction. The crackable parts amount to 60 or 70 wt% maximum. Fig. 4-21 shows the distillation curves of an atmospheric residue and the corresponding vacuum residue produced therefrom (origin Kuwait). It may clearly be seen that the vacuum distillation in the refinery has been performed at a temperature in the vicinity of the cracking region. Fig. 4-22 shows the distillation curves of the atmospheric residues (samples 14-17). Approximately 40 or 50 wt% of this samples may still be separated by means of a vacuum distillation.

Fig. 4-18 Simulated Distillation of Bitumen B45 by TG 750 Sample No. 2: Safaniya Crude Sample No. 4: Venezuela Crude


139

Fig. 4-19 Simulated Distillation of Bitumen B80 by TG 750 Sample No. 3: Agha Jari Crude Sample No. 5: Venezuela Crude Sample No. 7: Middle East Crude, Semiblown

Fig. 4-20 Simulated Distillation of Bitumen B200 by TG 750 Sample No. 9: North Sea + Arabian Light Crudes Sample No. 10: Arabian Heavy Crude

140

4 Thermoanalytical Investigations on Petroleum and Petroleum Products ii"C1

Fig. 4-21 Simulated Distillation of Residues from Kuwait Crude (TG 750) Sample No. 17: Atmospheric Residue Sample No. 8: Vacuum Residue

Fig. 4-22 Simulated Distillation of Atmospheric Residues (TG 750) Sample No. 14: Arabian Light Crude Sample No. 15: Kirkuk Crude Sample No. 16: Tuimaza + Arabian Heavy Crudes Sample No. 17: Kuwait Crude


141

In Fig. 4-23 the distillation curves of samples from conversion processes are shown. The fully distilled residues from a cat-cracker (sample 20) and from a thermal cracker (sample 22) do not contain any substances separable by distillation, but do contain approximately 40-50 wt% crackable substances. From the visbreaker residues (samples 19 and 21) about 20 wt% may still be separated by vacuum distillation. Obviously the residue of a thermal cracker (sample 18) has undergone only atmospheric distillation. Approximately 5 wt% may still be gained from that sample by atmospheric distillation, and an additional 65 wt% by vacuum distillation, Fig. 24 shows the distillation curves of a waxy distillate (sample 23) and the residue from a thermal cracker (sample 22). The residue has been distilled exhaustively. It still contains 50 wt% crackable substances. From the waxy distillate about 50 wt% may still be gained by vacuum distillation. The distillate from a cat-cracker (sample 25 in Fig. 4-25) contains approximately 35 or 40 wt% which may be separated by atmospheric distillation and an additional 45 or 50 wt%

Fig. 4-23 Simulated Distillation of Samples from Conversion Processes by TG 750 Sample No. 18: Residue from Thermal Cracker (Arabian Light) Sample No. 19: Residue from Visbreaker Sample No. 20: Residue from Katalytic Cracker (Kirkuk) Sample No. 21: Residue from Visbreaker (Venezuela) Sample No. 22: Residue from Thermal Cracker (Iranian Heavy)

142


1 T["CI

Fig. 4-24 Simulated Distillation by TG 750 Sample No. 23: Waxy Distillate (Nigeria + Libya) Sample No. 22: Residue from Thermal Cracker (Iranian Heavy)

Fig. 4-25 Simulated Distillation by TG 750 Sample No. 24: Furfural Extract Sample No. 25: Distillate from Thermal Cracker


143

separable by vacuum distillation. The furfural extract (sample 24 in Fig. 4-25) demonstrates uniform boiling behavior. In the temperature range from 435-525 "C about 85 wt% may be obtained by distillation.

4.2.3.4 Directly measured index numbers in comparison with the simulated distillation Calculation of the simulated distillation curves from thermogravimetric data demands a considerable working time, even if a suitable computer program exists. Up to 18 values have to be gathered from the TGA curve and fed into the computer. Therefore an investigation has been made, as to whether the directly measured index numbers could indicate which portions of the sample may be separated by atmospheric and vacuum distillation. As will be shown below (chapter 4.2.7) the mean of the start temperature of the crack reaction in TGA is X = 414 k 17.3 "C. Therefore the weight loss up to 400 "C experimental temperature (AG400) may be considered as the upper limit of the distillation range. Table 4-19: Weight Loss in Thermogravimetry and Corresponding Temperature of Simulated Distillation Sample No

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 arithmetic mean:

AG200 [wt %] 2.0 0.5 0 0 0 0 0 0 0.5 0 0 0 0 7.2 4.75 6.0 5.7 6.3 4.0 0.2 3.5 0 5.0 2.6 15.75

corresp. T ["CI

450 -

-

415 430 424 449 419 448 -

443 -

427 452 284 435.7

AG300 [wt %]

corresp. T ["CI

AG400 [wt %I

corresp. T ["CI

18.0 4.25 1.7 7.25 6.1 1.2 2.5 1.1 2.0 6.2 2.25 1.45 4.5 29.75 28.5 34.0 28.5 50.0 17.0 1.5 16.0 3.25 45.5 27.75 45.5

518 582 625 567 572 646 602 581 544 558 612 637 581 495 496 494 501 470 508 560 523 609 478 498 350 552.4

44.0 22.5 20.0 32.5 29.0 11.0 19.7 12.7 28.0 37.1 26.5 16.7 29.8 62.7 60.0 65.0 59.7 82.9 46.5 23.8 40.0 23.7 70.5 92.0 92.8

608 648 658 629 634 628 660 617 624 620 636 662 628 589 590 589 597 600 630 644 610 647 565 568 -,560 617.2

144


Comparison of the values of AG400 (Table 4-15) with the corresponding values of the simulated distillation gives a mean of the distillation temperature (reduced to atmospheric pressure) of X = 617.2 "C with a coefficient of variation ? V = 4.83 %. Such products can only be obtained by high vacuum distillation, which is not the general refinery standard. For the vacuum distillation commonly found in refineries, an upper temperature limit from 530-550 "C maximum (at atmospheric pressure) exists. For atmospheric distillation the upper temperature limit is 400-420 "C. This would correspond to the index numbers for the weight loss up to 300 "C (AG300) for the usual vacuum distillation. AG300 corresponds to a mean of the distillation temperature X = 552.4 "C ( ? V = 9.70 %). The portion which can be separated by atmospheric distillation is respresented by the weight loss up to 200 "C (AG200) which corresponds to a mean of the distillation temperature X = 435.7 "C ( k V = 3.25 %) (Table 4-19).

4.2.3.5 Derived index numbers for pracital application From the arguments in chapter 4.2.3.4 it may be deduced that while the derived index numbers of chapter 4.2.3.2, such as ND, CR, CCR/ND, and R800/ND may be of theoretical interest, other index numbers are relevant to the practice of refineries, and may also be obtained by thermogravimetry (see Tables 4-20 and 4-21). Definitions: SAR = 100 - AG200 SVR = 100 - AG300 PCR = 100 - (AG300 + R800) PCR/SVR

(CCR/SVR) . 100 (R800/SVR). 100 (GW/SVR) . 100

simulated atmospheric residue (wt%) simulated vacuum residue (wt%) "practical thermally crackable part" (wt%) "practical thermally crackable part" in the simulated vacuum residue Conradson coke residue in the simulated vacuum residue (%) residue at 800 "C in the simulated vacuum residue (%) residue in the extrapolated point of inflexion of the TGA curve in the simulated vacuum residue (%)

Table 4-20: Derived Index Numbers of Thermogravimetry in Inert Gas (I) Sample No.

1 2 3 4 5 6 7 8 9 10 11 12

SAR=100-AG200 [wt %I

SVR=100-AG300 [wt %I

PCR=lOO-(AG300+R800) [wt %I

98.0 99.5

82.0 95.75 98.3 92.75 93.9 98.8 97.5 98.9 98.0 93.8 97.75 98.55

69.6 79.35 85.00 16.35 79.50 76.60 78.00 84.60 90.00 82.10 79.75 82.85

-

-

99.5 -

-

PCR SVR

~

0.849 0.829 0.865 0.823 0.847 0.775 0.800 0.855 0.9 18 0.875 0.816 0.841


Sample No.

13 14 15 16 17 18 19 20 21 22 23 24 25

SAR=lOO-AG200 [wt %]

SVR=lOO-AG300 [wt %]

PCR=100-(AG300+R800) [wt %]

95.5 70.25 71.5 66.0 71.5 50.0 83.0 98.5 84.0 96.75 54.5 72.25 54.5

81.80 65.25 63.20 57.00 71.50 44.50 62.00 88.50 76.90 66.05 52.50 72.25 54.5

-

92.8 95.25 94.0 95.3 93.7 96.0 99.8 96.5 100 95.0 97.4 84.25

PCR SVR ~

0.857 0.929 0.884 0.864 1.ooo 0.890 0.747 0.898 0.915 0.683 0.963 1.000 1.000

Table 4-21: Derived Index Numbers of Thermogravimetry in Inert Gas (11) Sample No.

~. CCR

100

SVR

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25

~. R800 SVR

100

& . 100

[%I

[%I

["/.I

20.0 22.2 21.7 23.4 20.3 21.8 22.2 20.4 20.1 21.4 25.5 21.1 21.2 12.2 14.6 13.5 14.8 19.8 31.0 40.0 27.3 42.8 18.2

15.1 17.1 13.5 16.6 15.3 19.4 20.0 14.5 8.2 12.3 18.4 15.9 14.3 7.1 11.6 13.6

20.1 23.0 21.6 22.1 20.2 31.7 30.8 24.6 20.9 33.9 33.2 22.1 22.2 14.0 20.1 24.2 32.6 28.4 29.2 35.5 41.7 40.6 62.9 14.5 21.1

-

4.4

-

11.0 25.3 10.2 8.5 31.7 3.7 -

145

146


The vacuum residues and bitumens (samples 1-13) no longer contain portions, which can separated by atmospheric distillation, i. e. SAR = 100 wt%, except sample 1. The four atmospheric residues (samples 14-17) possess between 92.8 and 95.25 wt% SAR. The residues from a cat-cracker (sample 20) and from a thermal cracker (sample 22) have been exhaustively distilled as the SAR of 100 wt% demonstrates. For the other products from conversion processes the SAR amounts to 84.25 wt% (distillate from the residue of a thermal cracker), and up to 97.4 wt% (furfural extract, sample 24). The SVR of the vacuum residues and bitumens (samples 2-13) has a mean X = 96.6 wt% ( k V = 2.27 %). Again sample 1 stands out from this group with SVR = 82 wt%. The atmospheric residues demonstrate a mean SVR of X = 69.8 wt% ( fV = 3.74 %). For the substances from conversion processes the SVR again depends on the degree of distillation. There SVR spreads over a range from 54.5 wt% (waxy distillate, sample 23) up to 98.5 wt% (cat-cracker residue, sample 20). The data for the PCR (part of the residue which is, in practice, crackable) of the vacuum residues and bitumens are not as uniform as the data for the SVR, due to considerable differences in their R800. Thus the PCR spreads over the range from 69.6-90 wt%. Statistical evaluation (samples 2 to 13) gives a mean X = 81.3 wt% with a coefficient of variation k V = 4.84 %. The PCR value is nearly one third higher than CR (see chapter 3.5.2.2). Higher values of the PCR have also been calculated for the atmospheric residues. For samples without any coke residue (RSOO), PCR equals SVR (samples 17, 24, and 25). The ratio PCR/SVR for vacuum residues, bitumens, and atmospheric residues amounts to X = 0.85 ( I V = 4.62 %). The clear difference found for the quotient CR/ND (see chapter 4.2.3.2) between the distillation residues and the conversion products does not hold for the ratio PCRISVR. The Conradson coke residue in the simulated vacuum residue ((CCR/SVR) . 100) for the vacuum residues and bitumens has a mean value X 21.6 % ( k V = 7.01 % relative). For the atmospheric residues the mean amounts to X = 13.8 % ( fV = 8.7 % relative). The products from conversion processes (samples 19, 20, and 22) have extremely high values demonstrating that they have been distilled exhaustively, whereas the distillate of the residue of a cat-cracker, sample 25, exhibits the extremely low value of 4.4 %. For the residue at 800 "C in the simulated vacuum residue ((RSOOISVR) . 100) there is no real differencebetween the means of samples 1- 13 (2 = 15.4 %) and samples 14- 17 (2 = 14.6 %). But here again samples 19 and 22 show considerably higher values and sample 23 a very low value. Generally the scattering of (RSOOISVR) . 100 exceeds that of (CCR/ SVK) . 100. The relation of the residue at the inflexion point of the TGA curve, to the simulated vacuum residue (G,/SVR) ' 100 for the group of the vacuum residues and bitumens shows widely differing values for blown (partly oxidized) products (samples 6, 7, 10, and 11). Statistical treatment (G,/SVR) . 100 for samples 1-13 give a mean X = 25.1 % with the very large coefficient of variation f V = 20.86 % (relative). When the blown products are excluded there is a considerably smaller scattering of results:


Mean X

Coefficient of Variation 5 V (%)

Samples No.

x (96)

&V (% of F)

1-5, 8, 9, 12, 13 6, 7, 10, 11

21.9 32.4

6.44 4.34

147

The data of (G,/SVR) . 100 for the atmospheric residues (samples 14- 17) do not demonstrate any uniformity. The products from conversion processes generally show higher values. The extremely high value for the waxy distillate is extraordinary and obviously due to the low SVR (54.5 wt%). The high values for cracker residues, which have been distilled exhaustively (samples 20, 21, 22) are understandable.

4.2.4 Thermogravimetry in air The TGA-curves of the experiments in argon are the same as those in air at lower temperatures. Differences in the slope of the curves first become noticeable in the medium temperature range starting at about 275-400 "C. The curves in air are usually shiftet to higher temperatures (Fig. 4-26, sample 8), which indicates absorption of oxygen. Sometimes a flattening of the curve occurs, as shown in Fig. 4-27, sample 11. Fig. 4-28 demonstrates that the oxidation first starts when about 55 wt% of sample 14, (atmospheric residue from Arabian Light crude) has already evaporated. The behavior of the atmospheric residue of a blend from Tuimaza and Arabian Heavy crudes (sample 16 in Fig. 429) is similar. The difference in the evaporated quantities before the start of oxidation is shown in Fig. 4-30 (visbreakerresidue, sample 21) and Fig. 4-31 (waxy distillate, sample 23). In contrast to the experiment in argon, in air, the distinct inflexion and levelling of the TGA curves in the higher temperature range does not occur. All the curves either meet the x-axis (combustion without any residue) or show a small constant residue (ash). The vast majority of DTG curves, i. e. the reaction rate, show numerous peaks (maxima) of different intensity, which are due to the evaporation losses, and at higher temperatures to multi-step oxidation reactions of the different compounds in the multi-component systems under investigation. The reaction rates (DTG) and the peak maximum temperatures of the first and the last maximum each of the DTG curves, are recorded in Tables 4-22a and 4-22b together with the temperatures of the remaining peak maxima. All the index numbers measured directly from the experiments are given in Tables 4-22a and 4-22b.

148


Fig. 4-26 Thermogravimetry (TG 750) Sample No. 8: Vacuum Residue (Kuwait Crude) Heating Rate /3: 10 K/min Atmosphere: Air 25 cm3/min


Residue ( W t - % )

I00 90

8C 7c

6C 50

LO 30

20 10

T["C 1 100

200

300

400

500

600

700

800

c

Fig. 4-27 Thermogravimetry (TG 750) Sample No. 11: Vacuum Residue, Semiblown (Forties + Arabian Light) Heating Rate p: 10 K/min Atmosphere: Air 25 cm3/min

149

150


100

90

80 70

60 50

LO

30

za 1C

100

200

300

400

500

600

700

Fig. 4-28 Thermogravimetry (TG 750) Sample No. 14: Atmospheric Residue (Arabian Light) Heating Rate p: 10 K/min Atmosphere: Air 25 cm3/min

800


Fig. 4-29 Thermogravimetry (TG 750) Sample No. 16: Atmospheric Residue (Tuimaza + Arabian Heavy) Heating Rate p : 10 K/min Atmosphere: Air 25 cm3/min

151

152

4 Themoanalytical Investigations on Petroleum and Petroleum Products Residue (Wt-Z)

T["CI 100

200

300

400

500

600

700

Fig. 4-30 Thermogravimetry (TG 750) Sample No. 21: Vibreaker Residue (Venezuela) Heating Rate p: 10 K/min Atmosphere: Air 25 cm3/min

860-


Fig. 4-31 Thermogravimetry (TG 750) Sample No. 23: Waxy Distillate (Nigeria + Libya) Heating Rate p: 10 K/min Atmosphere: Air 25 cm3/min

153

154


Table 4-22a: Directly Acquired Index Numbers of the Thermogravimetry in Air I Sample No.

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25

T1 % [“CI

T5 % [“CI

AG400 [wt %I

180 244 226 225 235 302 268 284 27 1 240 25 1 272 236 126 137 145 172 126 153 292 144 256 150 176 96

235 308 297 283 293 365 330 339 313 283 299 35 1 289 189 200 198 218 188 215 360 213 320 202 225 149

34.3 23.0 20.2 21.5 24.5 12.3 17.3 22.6 29.2 29.5 27.5 17.9 28.8 57.0 51.3 55.3 49.7 61.0 27.5 10.8 33.0 18.0 60.2 65.5 81.7

First Maximum DTG Tmaw [%/min] [“CI

2.63 4.27 2.30 2.50 3.98 3.31 2.52 4.71 3.59 3.37 1.57 2.24 2.63 4.08 3.76 3.70 4.01 5.17 2.06 2.93 2.07 1.43 5.27 6.96 6.36

318 408 419 350 392 420 360 379 349 365 355 361 334 315 312 3 10 320 316 336 389 366 362 297 343 288


155

Table 4-2213: Directly Acquired Index Numbers of the Thermogravimetry in Air I1 Sample No.

Last Maximum DTG Tmax [%/min] roc]

TEndof experimerlt ["CI

1 2 3 4 5 6 7 8 9 10 11 12

8.66 6.59 7.23 4.04 7.23 9.76 6.54 9.00 6.22 7.51 6.58 6.88

562 580 577 554 542 626 582 592 545 542 543 551

620 628 630 850 59 1 656 649 619 593 570 588 580

13

10.79

536

551

14 15 16 17 18 19 20 21 22 23 24 25

3.25 3.60 4.10 5.07 4.07 10.20 11.13 8.27 7.29 3.61 2.69 1.78

569 560 572 533 568 575 565 517 543 542 566 588

627 640 605 595 640 648 669 539 670 600 700 647

Peaks of the DTG - Curve at T ["CI

318, 397,494, 562 408, 442, 461,477, 491, 580 419,448,496, 577 350, 370, 399, 482, 554 392,415,428,456,462,484, 542 420,492, 534, 562, 596, 626 360, 450, 503, 562, 582 379, 424, 436,470, 484, 561, 592 349,375,387,398,404,430,468,545 365, 381, 398,404,418,438,542 355, 375, 389,409, 436, 543 361, 384, 392, 402, 412, 427, 436, 447, 475, 551 334, 357, 368, 380, 396, 421, 438, 456, 536 315, 387,473, 501, 569 312, 385,447,463, 560 310, 385,448, 538, 572 320, 382,435, 444, 467, 533 316,443, 475, 568 336,411,486, 537, 557, 575 389, 460, 565 366, 378, 399, 409, 441, 448, 517 362, 391,410,428,438,466, 543 297,400, 542 343, 566 288, 588

4.2.4.1 Directly measured index numbers The temperatures of the start of an evaporation T1 % or T5 % do not differ in the experiments in air from those in argon. In any case there is a more or less pronounced evaporation before the start of oxidation. At the beginning, the TGA curves of the experiments both in air and in argon are congruent. The point where the curves separate may be regarded as the start of oxidation (Table 4-23). Losses up to 400 "C (AG400) cannot be regarded in every case as the distillable fraction of the sample. In the temperature range 300-400 "C evaporation and oxidation processes could be partly superimposed. Given that the value AG400 can be reproduced with a tolerance of t- 5 %, a real distinction of the index numbers demands a difference of at least 10 % (relative). Only the AG400 values of the following samples demonstrate a real difference between the experiments in air and in argon:

156

4 Thennoanalytical Investigations on Petroleum and Petroleum Products

Table 4-23: Comparison of the TGA-Curves in Argon and in Air Sample No.

Curves identical up to T ("C)

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25

350 405 380 340 375 435 390 335 370 360 250 420 325 380 365 365 355 280 250 425 345 380 295 340 295

sample no. 1, 4-6, 8, 10, 14-19, 21-25. The value of AG400 in air is generally smaller than in argon except for samples 6 an( This indicates the formation of oxidation products, which do not evaporate in the tempc ture range below 400 "C. This is demonstrated especially by the flattening in the Tc curves of samples 7, 11, 14, 15, and 16. The following limits of the values for AG400 h; been ascertained: AG400 (wt %) Minimum Maximum Vacuum residues and bitumens (samples 1-13) Atmospheric residues (samples 14-17) Products from conversion processes (samples 18-25)

12.3

34.3

49.7

51.3

10.8

81.7


157

Coke residue does not exist at 800 "C. All the samples were totally burnt or reduced to ash at temperatures below 800 "C. The experiments in air were terminated in the temperature range from 545 "C to 700 "C (except sample 4 at 850 "C). The arithmetical mean for the temperature of the end of the experiment is X = 619.2 "C ( V = 6.25 %). For the end of the experiment in air the following temperature limits have been found:

TEnd of ~xpenment(~~)

Vacuum residues and bitumens Atmospheric residues Products from conversion processes

Miniinurn

Maximum

550 605 545

655 640 700

The DTG curves exhibit numerous maxima of different intensity. Substances, which lose most of their weight by evaporation, exhibit only two or three oxidation peaks (samples 23, 24, 25). The other samples show between 3 and 10 peak maxima. Samples, which have a peak maximum in argon in the distillation range, also show such a maximum in air (Table 4-24):

Table 4-24: Temperature of the Distillation Maximum of the DTG Curve Sample No. 1 4 14 15 16 17 18 23 24 25

Mean X ("C) Coefficient of variation (%)

Argon ("C)

eel

Air

310 362 319 300 300 306 296 298 274 274

318 350 315 312 310 320 316 297 288 288

303.9 8.16

311.4 5.51

There is no statistical difference between the two means. The temperature of the first maximum of an oxidation peak for the samples possessing a distillation peak, also exceeds those of the other samples (Table 4-25):

158

4 Therrnoanalytical Investigations on Petroleum and Petroleum Products

Table 4-25: First Oxidation Maximum of Samples possessing a Distillation Maximum Sample No.

1 4 14 15 16 17 18 23 24 25

DTG (%/min)

Tm("C)

1.88 4.08 1.67 2.13 2.23 1.65 1.53 1.87 2.69 1.78

397 399 387 385 385 382 443 400 566 588

Samples 24 and 25 do not contain any substances which can be oxidized at low temperatures. Their first oxidation appears in the temperature range of the last oxidation maximum of the other samples. For the temperatures of the first oxidation maximum the following limits were found: Maximum Temperature of the 1st. Oxidation Peak of the DTG Curve Minimum Vacuum residues and bitumens (samples 1-13) Atmospheric residues (samples 14-17) Products from conversion processes (samples 18-25)

Maximum

("(3

("C)

334

420

382

387

362

443

Mean

X ("C)

Coefficient of variation _+ V ( % )

381.4 397.9 384.8

7.65 7.36 0.54

395.2

7.63

Statistical evaluation shows the following results: Maximum Temperature of the 1st. Oxidation Peak

Statistics over samples 1-25 Vacuum residues and bitumens (samples 1-13) Atmospheric residues (samples 14- 17) Products from conversion processes (samples 18-25)

There is no statistical difference between the four means.


159

The maxima of the weight loss rate DTG of the first oxidation peak were found in the following limits: DTG (%/min) of the 1st. Oxidation Peak Minimum

Maximum ~

Vacuum residues and bitumens (samples 1-13) Atmospheric residues (samples 14-17) Products from conversion processes (samples 18-25)

1.88

4.7 1

1.65

2.23

1.53

2.93

The last oxidation peaks were measured in the temperature range from 517-626 "C. The statistical evaluation demonstrates a scattering of the data around 560 "C for most of the samples: Maximum Temperature of the Last Oxidation Peak ~

Coefficient of variation +v(%)

Mean

X ("C) ~

Statistics over samples 1-25 Vacuum residues and bitumens (samples 1-13) Atmospheric residues (samples 14-17) Products from conversion processes (samples 18-25)

562.1

4.50

561.2

3.70

558.5

3.18

558.0

4.04

Again, there is no difference between the four means. The values of the maxima of weight loss rate of the last oxidation peak are in the range from 4.04 to 11.13 %/min. Comparison of the temperatures of the crack peak of the DTG curves from the experiments in argon (Table 4-15) with last column of table 4-22b demonstrates that nearly all those samples also show a peak at corresponding temperatures. This indicates competition between cracking and oxidation reactions. Determination of ash using a thermobalance is not advisable due to the very low contents, between 0.01 and 0.9 wt%. An additional experiment with a larger sample weight and/or extension of the measuring range is necessary.

160


4.2.5 Correlations of analytical data with index numbers from thermogr avimetry The correlation of the analytical data of the samples with index numbers from thermogravimetry may be examined. The regression line (curve), R, and a zero point line (curve), Z, as well as the coefficients of correlation r, have been calculated using the least squares method. The following tables show the relationship between some of the thermoanalytical index numbers and: the average relative particle masses (average molecular weight) (Tables 4-26 and 4-30); the average C-number ZC (Tables 4-27 and 4-30); the colloidal composition (Table 4-28); and data from the structural groups analysis (Table 4-29). Since all the samples in this investigation are multicomponent systems, the coefficients of correlation found were less than one. Index numbers not listed here did not show significant correlation with the analytical data.

Table 4-26: Correlation of thermoanalyticalindex numbers with the averaqe relative particle size % (Experiments in argon) No.

Function

1 2 3 4 5

T1 % = 0.260 x @ T5 % = 0.225 x M + 93 AG400 = 172.9 x gp(-0.00193 x R 800 = 0.01572M CR = 0.0624 x M-2.331

Type of Curve

Coefficient of Correlation

Z R R Z R

0.873 0.896 0.917 0.852 0.891

R = Regression line (curve)

Z = Zero point line (curve)

Table 4-27: Correlation of thermoanalytical index numbers with the average carbon number ZC (Experiments in argon) No.

Function

6 7 8 9

T1 % = 4.004 x ZC T5 % = 3.588 x ZC+84.16 AG400 = 184.6 x exp(-0.0315 CR = 1.055 x ZC - 6.80

x

ZC)

Type of Curve


Z R R R

0.899 0.890 0.906 0.882


161

Table 4-28: Correlation of thermoanalytical index numbers with the colloidal composition (Experiments in argon) DP= wt - % Dispersion medium Asph= wt - % Asphaltenes H= wt - % Petroleum resins Mal= wt - % Maltenes = wt- % DP + H No.

Function

10 11 12 13 14 15

AG 400 = 1.895 x DP - 106.63 R 800 = 0.625 x Asph + 2.57 R 800 = 0.563 x (Asph + H) CR = 3.723 x H + 24.91 CR = 71.62 x exp(-0.0197 x (DPIH)) PCR = 0.721 x Ma1 + 57.31

Type of curve


R R Z R R

0.921 0.751 0.739 0.774 0.827 0.751

R

Table 4-29: Correlation of thermoanalytical index numbers with values of the structural groups analysis (Experiments in argon) No.

Function

Type of curve

16 17 18 19 20 21 22 23

RSOO = 75.83 x H/C + 123.8 CR = 118.1 x CA/(CN + CP) R800 = 0.941 x CA + 17.9 G, = 5.093 x RA + 5.755 G, = - 74.593 x H/C + 129.751 G,/SVR = 3.876 x R A + 13.071 G,/SVR = 0.732 x CA -+ 2.853 G,/SVR = - 85.455 x HIC + 149.57


0.938 0.662 0.887 0.720 0.711 0.500 0.522 0.640

Table 4-30: Correlation of thermoanalytical index numbers with the average relative particle size ?@ (Experiments in air) No.

Function

24 25 26

T1 % = 0.2616 x T5 % = 0.1992 x ?@ + 114.0 AG 400 = 135.0 x exp (-0.001868

x

Type of curve


Z R R

0.878 0.879 0.893

162


Table 4-31: Correlation of thermoanalytical index numbers with the average carbon number ZC (Experiments in air) ~

No.

Function

27 28 29

T1 % = 3.963 x ZC T5 % = 2.882 x ZC + 121.5 AG 400 = 148.2 x exp (-0.0304 x ZC)

~~~

Type of curve


Z R R

0.832 0.824 0.850

4.2.6 Simulated thermal cracking by TGA In order to simulate the thermal cracking process, samples of about 5 mg weight were heated in the Stanton-Redcroft thermobalance ThISD in an argon atmosphere up to 600 "C

I

3

5

7

Fig. 4-32 Pattern of Simulated Cracking in Argon by TG 750 Heating Rate p : 100 K/min Atmosphere: Argon 25 cm3/min

Time (inin)

163


using a rapid heating rate of 100 K/min. A scheme of the recorder diagram is shown in Fig. 4-32. Curve 1 represents the signal from the thermocouple directly below the sample pan; curve 2, the sample weight AG; and curve 3, the DTG signal. Samples which still have distillable fractions, give an additional shoulder before the maximum of the DTG curve. In Table 4-33 the following index numbers are listed: Evaporation: Maximum of evaporation rate Temperature of the evaporation maximum Weight loss at T,

DTG,

(% /min)

TV

("C) (wt%)

GV

Cracking process: Maximum of cracking rate Temperature of the cracking maximum Residue at T, Final temperature Residue at final temperature

DTG, TC RC

(%/min) ("C) (wt%) ("C) (wt%)

TE

RE

Table 4-33: Index numbers of the Simulated Thermal Cracking Evaporation Sample DTG, No. (%/mi@

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 ~~~

25.74

Cracking Process

Tv ("C)

GV (wt%)

454 -

37

-

14.20

388

17

-

-

-

-

-

-

-

-

-

-

-

-

-

-

31.39 27.89 32.12 33.88 56.84 23.38

417 398 416 406 396 406

-

-

44 36 44 41 60 32

25.77

423

32

-

-

-

-

-

-

57.00 76.87 66.10

386 408 377

56 72 73

~

* Preset final temperature of the instrument

DTG, (%/min) 34.83 63.62 61.80 58.89 65.26 86.17 64.00 92.45 100.00 72.95 69.70 80.33 71.33 40.54 48.78 37.72 48.34 -

39.36 56.14 43.53 49.94

TC ("C)

RC (wt%)

547 545 559 537 524 533 561 533 530 520 526 516 555 522 517 533 505

32 30 28 43 34 36 33 29 32 34 37 40 32 20 19

509 547 507 510

33 48 36 57

-

-

20 -

-

-

-

-

-

-

-

-

TE*

RE

(C)

(wt%)

593 593 593 593 593 593 593 593 593 593 593 593 593 593 593 593 593 593 593 593 593 592 593 593 593

20 14 12 20 18 16 17 11 12 15 19 19 14 8 8 8 7 1 20 29 22 0 1 0 1

164


The repeatability of the experimental data is very good. The means and the coefficients of variation from ten experiments with sample 11 are shown in table 4-32 Table 4-32: Repeatability of the Index Numbers -

DTG, TC RC TE RE

(%/min) ("C) (wt%) ("C)

(wt%)

X

+_V(%)

69.70 528.3 36.8 529.9 19.3

6.12 1.59 2.02 0.37 8.15

4.2.6.1 Index numbers from simulated cracking The data from the simulated cracking process may also be assembled in groups. Statistical evaluation of the groups A: vacuum residues and bitumens (samples 1-13), B: atmospheric residues (samples 14-17), and C: products from conversion processes which can be cracked under the given experimental conditions (samples 19-22) is shown in table 4-34: Table 4-34: Statistical evaluation of the simulated cracking process Group

A

B C

Samples

1-13 14-17 19-22

DTG,

Tc

536.7 522.0 518.3

3.04 1.81 3.67

74.58 43.84 46.88

RC

6.72 13.10 16.29

33.5 19.7 45.8

11.32 2.94 26.3

There is a statistical difference between the mean of the maximum cracktemperature,TC of groups A and B, and between groups A and C. No such difference exists between the means of groups B and C. The comparativelyhigh coefficients of variation of the means of DTG, of groups B and C, prove that they are far too inhomogeneous for a statistical evaluation. The same goes for the mean of R, of group C. For the samples showing only one evaporation range (samples l8,23-25), the statistics give a mean of T,, X = 392.3 "C ( fV = 4.16 %). The maximum of the evaporation rate DTG, has a mean X = 64.20 %/min ( + V = 14.18 %). For the samples which demonstrate two maxima, or a shoulder and a maximum, the values found for each of the reactions are relatively low. The measured data are by definition related to the initial sample weight, and are therefore interdependent (Table 4-35):


165

Table 4-35: Simulated cracking at 600 "C, Samples with two maxima Sample No.

DTG, [%/min]

1 3 14 15 16 17 19 21 ~

X

V

DTG, [%/min]

"a

TC

DTG, + DTG, [%/mi111

547 559 522 517 533 505 505 507

60.57 76.00 71.93 76.67 69.84 82.22 62.74 68.80

25.74 14.20 3 1.39 27.89 32.12 33.88 23.30 25.27

448 388 417 398 416 406 406 423

34.83 61.80 40.54 48.78 37.72 48.34 39.36 43.53

28.52* k 13.95 %

417.75 k 4.34 %

41.87*

524.9

71.10

f 12.61 %

f 3.78 %

f 10.18 %

* excluding sample 3 The sum of DTG,+ DTG, (F= 71.10) is now of the order of magnitude of the DTG, values of bitumens and vacuum residues which have only one crack maximum.

4.2.6.2 Correlation of index numbers from simulated cracking with analytical data The following data may be correlated: - Weight loss at the maximum of evaporation AGv with the concentration of dispersion

medium, DP: AGv = 0.8506 . DP - 34.2877 Coefficient of correlation r = 0.904

Evaporation rate (maximum) DTG, with the concentration of dispersion medium DP: DTG, = 5.5750 . DP - 189.4960 Coefficient of correlation r = 0.834 - Residue at the end of experiment RE with the concentration of asphaltenes ( sph): RE = 0.5603 . Asph + 3.9124 Coefficient of correlation r = 0.793 - Residue at the end of experiment, RE with the Conradson coke residue, CCR: RE = 0.6324 . CCR + 1.3637 Coefficient of correlation r = 0.801 The coefficients of correlation are high enough to confirm the general trend. -

166


4.2.7 Start temperature of the cracking process in an inert atmosphere The slope and behavior of the TGA curve does not give clear indication whether a distillation weight loss is continuing or whether the sample has already been cracked, but the start temperature of the cracking reaction may be determined from the DTG cu-ve by the method of Z. Adony [4-141. In the diagram of log DTG versus the inverse Kelvin temperature, 1/T, a straight line characterizes the distillation range. The temperature of departure from the straight line indicates the start of the cracking reaction. Table 4-36 reproduces the results of such an evaluation of the experiments in argon with a heating rate p= 10 K / i i n . For the start temperature of the cracking reaction, TB, a mean X = 414.16 "C with a coefficient of correlation k V = 4.19 % was calculated, The value 400 "C chosen empirically for the end of the distillation (boiling) range AG400 is confirmed by this evaluation. Table 4-36:Start temperature of the cracking reaction TB ("C) Thermogravimetry in Argon p= 10 K/min Sample No

TB ["Cl

~

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 (A): Evaporation peak alone (B): Data for the second DTG peak n. e.: non evaluable

414 (B) 412 396 408.5 (B) 405 423 428.5 43 1 402.5 426 371 -

-

437.5 (B) 406 (B) n. e. 444 (B) 252 (A) 402.5 (B) 419 n. e. n. e. n. e. 322 (A) 238 (A)


167

4.2.8 Differential scanning calorimetry (DSC) The DSC experiments were carried out to evaluate the kinetics of the pyrolysis (crack) reaction of the samples. They were performed using a DuPont 990 Thermal Analyzer equipped with a pressure DSC cell and a twin pen x-y-recorder. Evaluation was carried out externally. The experiments were carried out according to ASTM E 698-79, and evaluated by the method set out in chapter X3 of the ASTM E 698-79. Sample sizesbetween 1.5 and 4 mg were used and DuPont SFI aluminium sample pans were used to avoid any creeping of the sample over the side of the pan. Three heating rates pof 10,20, and 50 K/min were applied. If one of the measured data did not fit the straight line in the diagram log pversus 1 000/T an additional experiment was carried out at a heating rate p= 5 K/min. Parallel experiments have shown that quite fast heating rates may be applied to pyrolysis reactions, provided that the sample size is small, so that there is no inteinal temperature gradient. In contrast to this behavior, effects due to melting, crystallization, or oxidation measurements may be overtaken and obscured if fast heating rates are used, and may not appear in the recorded diagram. The experiments were carried out in argon at atmospheric pressure, and in methane at 10 bar pressure, using a gas flow rate of 5 cm3/min each. For samples possessing an additional evaporation peak (T < 400 "C) before the cracking reaction (temperature range from 450-550 "C) both peaks were evaluated. The activation energy calculated for the evaporation is equal to the enthalpy of vaporization (see chapter 3.3.2). Pyrolysis reactions of mixtures of hydrocarbons result in the formation of volatile products (gas and vapor) as well as solid residues (coke). The degree of conversion U of the reaction may be determined only by weighing the sample and pan before and after the reaction: U=

initial weight - residual weight . 100 (%) initial weight

4.2.8.1 Experiments in argon at atmospheric pressure The experiments were performed in argon at 1 bar pressure and a gas flow rate of 5 cm3/min. Four of the 25 samples (samples 18, 23-25) show only one maximum in the temperature range from 275 "C to 400 OC, and this represents a distillation process. Fourteen samples (samples 2-13, 20-22) possess only one maximum each in the temperature range from 450 "C to 550 "C, which clearly represents a pyrolysis reaction. The remaining six samples (samples 1, 14-17, 19) show two maxima, one each in the distillation range and the pyrolysis range. The coefficients of the Arrhenius equation, activation energy E and frequency (pre-exponential) factor A are presented in Table 4-37:

168


Table 4-37: DSC in 1 bar Argon Sample No.

Vaporization E, [kJ/Mol] log A, [min-'1

5.971 -

3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25

Cracking reaction E, [kJ/Mol] log A, [min-'I

-

-

5.258 11.372 6.224 10.449 14.134 6.120 -

-

40.3 70.3 59.7

2.969 5.400 5.063

82.8 78.7 81.5 79.7 82.0 69.2 69.9 77.1

226.7 204.4 165.9 211.6 197.4 235.9 159.8 181.5

15.973 14.390 11.492 14.903 13.707 16.577 10.975 12.560 -

-

180.6 215.8 167.3 174.1 171.4 177.1 174.2 187.0

12.514 17.927 18.538 12.123 11.827 12.341 12.237 13.038

206.8 217.0 190.4 199.6

14.542 9.855 11.299 13.707

69.5 68.2 79.1 78.4 83.9 84.8 84.1 77.1 85.5 76.1 63.8 72.0 63.6 100 100 100

-

61.7 125.4 71.1 124.0 171.9 76.1

U [%I

-

-

-

-

-

-

Statistical evaluation of the peak maximum temperatures of the pyrolysis reaction give coefficients of variation +V = 1.18 % maximum of the corresponding mean for each heating rate. Therefore the means of the peak maximum temperatures may justifiably be used to calculate the average values of the coefficients of the Arrhenius equation: Cracking reaction: Average values Activation energy E, = 190.6 kJ/Mol Frequency factor log A , = 13.339 min-' The arithmetical means calculated from the individual data of Table 4-37 are insignificantly larger in number, without any significant statistical difference. But their coefficients of variation are considerably larger: Cracking reaction: Means and coefficients of variation Activation Energy E, = 192.2 kJ/Mol *V = 11.35 % Frequency factor log A , = 13.529 min-' _+ V = 17.18 % There is no statistical difference between the average values and the means (t < t 0.95).

169


Only one of the vacuum residues (sample 1) has a substantial evaporizable fraction, as has already been shown using thermogravimetry (T1 % = 181 'C, T5 % = 237 'C, AG400 = 44 wt%). The boiling point, calculated from the DSC peak maximum temperatures and reduced to atmospheric conditions, is 410 "C. Each of the four atmospheric residues has an evaporizable fraction with the following temperatures for the evaporation maxima: Sample

DSC Peak maximum temperature ("C)

No.

10 k/min

20 K/min

50 K/min

Corresponding mean Boiling point (" C)

14 15 16 17

275 287 275 325

317.5 304.5 297.5 337

340 320 330 363

368 360 356 408

~

The distillation residues of these samples pyrolyze in the temperature range between 450 "C and 500 "C exhibiting a relative low energy of activation E, of approximately 177 kJ/Mol. The residue of a thermal cracker (sample 18) seems to demonstrate losses only by distillation since the peak maxima were found in the temperature range 350-380 "C. Those temperatures are very low for a pyrolysis reaction, whereas the activation energy E = 172 kJ/Mole could represent a substance which is either easily crackable or tough volatile. The residue at 800 "C in thermogravimetry (R8OO = 5.5 wt%) and the conversion in DSC (U= 85.5 %) could result from either type of reaction. In this case it is a disadvantage that the instrument does not permit identification of the products formed. The residue of a visbreaker (sample 19) exhibits a distillable fraction. The corresponding boiling point is 420 "C. The residue therefrom pyrolyzes between 450 "C and 490 "C. In the thermogravimetric experiment no DTG maximum was found in the distillation range. Samples 23-25 already designated as low boiling substances, by thermogravimetry, also distill totally at temperatures far below the pyrolysis range in the DSC. The corresponding boiling points are: Sample No.

Substance

23 24 25

waxy distillate furfural extract distillate from a catalytic cracker

DSC Peak maximum temperature ("C) 10 K/min 20 K/min 50 K/min

Corresponding mean boiling point ("C)

290 335

327.5 362.5

388 403

401 443

275

317.5

340

367

DSC peak maximum temperatures (at 10 K/min heating rate) in the distillation range agree quite well with the temperatures of the corresponding DTG maximum from thermo-

170


gravimetry. Comparison between the DSC maximum temperatures in the pyrolysis (crack) range and the corresponding DTG maximum temperatures shows complete agreement: Peak maximum temperatures of the pyrolysis reaction DSC (20 samples) X = 456.3 "C +V = 0.94 % DTG (20 samples) X = 458.4 "C +V = 3.28 % The considerable differences in the coefficients of the Arrhenius equation (Table 4-37) lead to different half-life times at identical temperatures. In Fig. 4-33 the plots of half-life time as a function of temperature (1 000/T) of the samples with maximum, minimum, and average values of the kinetic coefficients are shown:

1,70

1.60

1.50

1,40

130 1.20

1.10

1,oo

Fig. 4-33 Half Life Time t1,2 of Pyrolysis in 1 bar Argon Sample No. 6: E = 235.9 kJ/Mol; logA = 16.577 min-I Sample No. 7: E = 159.8kJ/Mol; logA = 10.975 min-' Mean Values (22 Samples): E = 190.6 k.J/Mol; lo@ = 13.339 min-'


Maximum: sample 6 Minimum: sample 7 Average

E, (kJ/Mol)

log A, (min-')

235.9 159.8 190.6

16.577 10.975 13.339

171

At a pyrolysis temperature of 450 "C there are no differences in the half-life time. Towards lower temperatures ( < 400 "C) and higher temperatures ( > 500 "C) the differences become evident. The coefficients of the Arrhenius equation determined in this manner, are the basic data for the calculation of the kinetics of pyrolysis (crack) reactions and therefore also the basis for the choice of process conditions, such as pre-setting of reactor temperatures, residence times etc. in thermal conversion processes. The residue R, measured at the end of reaction corresponds closely to the amounts of coke G,, which were found using thermogravimetry, with the exception of the data for samples 14 and 15 (atmospheric residues) and 21 and 22 (conversion products). There is a strong linear relation of R, (residues from the DSC experiments in inert gas at 1 bar pressure) with G, residual weight at the point of inflexion of the TGA curve:

R, = 0.9275 G, + 1.4 Coefficient of correlation r = 0.9989 Determination of the temperatures in the TGA curve, which correspond to the residual weight R, of the DSC experiment, shows a very narrow scattering of the data. Statistical evaluation of the data from samples 1-13, 16, 17, and 22, gives X = 508.6 "C fV = 2.88 %. This temperature corresponds roughly to the temperature of the formation of the Conradson coke residue. Samples which do not comply with this are: 14 and 15 (472 "C), sample 18 (443 "C), samples 19 (545 "C), 20 (547 "C), and 21 (565 "C), where the latter three represent residues from crackers. These samples show similar behavior of their temperatures at the point of inflexion Tw of the TGA curves. If the residues after the end of the DSC experiments are compared with the aromacity of the samples (H/C), a clear trend emerges that increasing H/C ratios lead to decreasing residues R,.

4.2.8.2 Experiments in methane at 10 bar pressure In order to simulate the process parameters of thermal cracking in the refinery better experiments were carried out in a hydrocarbon atmosphere (methane) at 10 bar pressure. This corresponds to the maximum pressure of the gas supply. The experimental conditions were again: heating rates p of 5, 10, 20, and 50 K/min, sample sizes from 1.0 to 4.0 mg, and gas flow rates of 5 cm3/min. Increase in the pressure leads to an increase of the peak maximum temperatures in the distillation (evaporation) range by up to 60 "C, whereas the peak maximum temperatures of the pyrolysis reaction can decrease by 10 "C.

172


The peak maximum temperatures of the pyrolysis (cracking) reaction (450-485 "C) differ so little that for each heating rate a coefficient of variation fV = 1.08 % of the corresponding mean has been calculated. Calculation of the kinetic coefficients using ASTM E 698-79 gives the following average values: Cracking process: Average values E, = 228.6 kJ/Mol log A, = 16.307 min-' From the individual values of the kinetic coefficients in Table 4-38, an arithmetic mean with considerably larger coefficients of variation is derived by statistical evaluation: Cracking process: Means and Coefficients of Variation E, = 237.8 kJ/Mol k V = 12.61 % log A, = 16.901 min-' V = 14.19 %

*

There is no statistical difference between the averages and the means. During the vaporization process, it is striking that the "enthalpy" increases considerably, sometimes much more than the Clausius-Clapeyron law leads us to expect. Vaporization will naturally become more difficult when the pressure is increased, but there is no explanation for an increase by a factor of 5 or 6. Therefore data of samples 14,24, and 25 seem to be anomalous. Pyrolysis also becomes more difficult with increasing pressure. That is logical, because pyrolysis is a reaction resulting in an increase in volume, due to the formation of crack gases or volatile products and coke. The activation energy of sample 18 (residue from a thermal cracker) rises froin 171.9 kJ/Mol at normal pressure to 187.2 kJ/Mol at 10 bar pressure. This increase of only 10 % indicates that a pyrolysis reaction occurs even at 1 bar pressure, especially since the peak maximum temperatures hardly shift. Sample 19 (visbreaker residue) does not exhibit a vaporization peak under these conditions. The plot of log half-life time of the cracking reaction versus the inverse Kelvin temperature (1 OOOlr) once more shows straight lines (Fig. 4-34). The slope of the line is determined by the value of the activation energy. The half-life times of the samples possessing the maximum and minimum values of the kinetic coefficients, as well as the average values were plotted.

Maximum: sample no. 12 Minimum: sample no. 14 Average:

E, (kJ/Mol)

log A /(min-')

288.4 197.7 228.6

20.386 13.394 16.307

The plot demonstrates clearly that in the temperature range from approximately 450470 "C there is no difference in the half life time, whereas the difference at lower temperatures ( < 400 "C) and at higher temperatures ( > 550 "C) becomes significant.


170

1,60

1,SO

1,bO

130 1.20

1,lO

173

1,oo

Fig. 4-34 Half Life Time t,,2 of Pyrolysis in 10 bar Methane Sample No. 12: E = 288.4 kJ/Mol; logA = 20.386 min-I Sample No. 14: E = 197.7 kJ/Mol; logA = 13.894 min-' Mean Values (22 Samples): E = 228.6 kJ/Mol; log = 16.307 -1

When considering technical processes, the behavior below a temperature of 400 "C is not have as much practical importance. Generally the half-life times of the reactions in methane at 10 bar pressure are shifted towards higher values, corresponding to the higher values of the activation energy and the frequency factor. The pyrolysis reaction in methane at 10 bar pressure demonstrates average values of the activation energy and frequency factor which are approximately 20 % higher than the values from the reaction in argon at 1 bar pressure. Again the trend may be recognized that a decreasing aromacity of the sample (increasing quotient H/C) leads to a decreasing residue (coke).

174


4.2.8.2.1 Reaction enthalpy from tests at 10 bar pressure Integration of the peak area can be used to determine the reaction enthalpy. Theoretically the enthalpy should be independent of the heating rate but this is not so in practice. An increase of the heating rate usually leads to an increase of the observed reaction enthalpy. The shift of the baseline during the reaction causes uncertainties in the exact fixing of the integration limits, and renders the integration more difficult. For this reason the crack enthalpies listed in table 4-39 should be treated with suspicion. Crack enthalpies H, from the experiments in 10 bar methane with a heating rate p= 10 K/min are shown. Samples which do not exhibit a pyrolysis peak were omitted (samples 18 and 23-25). Samples which have undergone a distillative weight loss do not conform because the enthalpy is related to the inilial sample weight and not to the residue after the evaporation (samples 1 and 14-17). Supposing that distillative losses only appear up to a temperature of 400 "C and that the crack process first starts above this temperature, it is reasonable to relate the crack enthalpy to the quantity of sample still present at 400 "C. This value has to be found by thermogravimetry. The value H: calculated in this way is listed in column 3 of the Table 4-39. Table 4-38: DSC in 10 bar Methane Sample No. 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18

19 20 21 22 23 24 25

Cracking reaction Vaporization E, [kJ/Mol] log A, [min-'1 E, w/Mol] log A, [min-'1

155.2 -

12.260 -

272.5 243.0

20.010 17.075

80.9 79.3

227.0 214.5 271.9 260.3 232.4 214.2 275.2 231.8 288.4 218.1 197.7 181.3 210.4 260.3

16.066 15.104 19.048 18.396 16.419 14.977 20.040 18.360 20.386 15.411 13.894 11.286 15.018 18.393

231.8 235.8 210.3 279.1

16.915 16.637 14.863 19.719

80.3 81.1 70.5 71.1 76.2 78.4 70.3 70.4 78.0 77.2 84.3 84.2 85.0 76.5 86.0 77.8 64.9 71.0 64.2 100 100 100

-

-

-

373.8 153.3 271.6 145.6 187.2

33.077 13.369 24.326 12.378 15.477

-

-

105.0 406.2 308.7

8.691 34.537 27.513

u [%I

-

-

-

-

-

-


175

Table 4-39: Enthalpy of pyrolysis in 10 bar methane Heating rate p= 10K/min H, [mJ/mgl: R400 [%I: H*, [mJ/mgl: Sample No.

1 2 3 4 5 6 7 8

9 10 11 12 13 14 15 16 17 18 19 20 21 22

Enthalpy of pyrolysis related to initial weight Residue at 400 "C Enthalpy of pyrolysis related to R400 HC

R400

H*C

202 1264

56 78 81 68 71 89 81 88 70 64 73 82 70 37 40 35 40

361 1621

-

598 230 1244 956 356 892 450 385 447 341 438 314 235 213 -

358 369 488 392

-

52 76 60 77

-

880 323 1398 1180 404 1274 704 527 545 487 1184 785 672 532 -

689 485 813 508

4.2.8.3 Start temperatures of the cracking process at different pressures The start temperature of evaporation and of pyrolysis reaction may be defined by means of the onset temperature, which is itself defined as the temperature of the intersection of the tangents to the two branches of the curve, which have different slopes. The small drift of the curve of energy versus temperature introduces errors in the determination of the onset temperatures of evaporation events. For the experiment in methane at 10 bar pressure the onset temperatures and the peak maximum (minimum) temperatures of the evaporation are sometimes, quite close, due to the mode of determination (table 4-40):

176


Table 4-40: Onset temperatures of the evaporation ("C) Sample No.

1 bar Ar

1 14 15 16 17 20 23 24 25

10 bar CH, -

320 270 287 280 275 375 247 220 175

275 320 285 317 382

The start temperature of the pyrolysis reaction may in all cases be clearly ascertained. The onset temperatures of the experiments in 1 bar argon fall between 390-422 "C and for the experiments in 10 bar methane, from 392-420 "C. Statistical evaluation gives small coefficients of variation (Table 4-41). Table 4-41: Onset temperatures of the pyrolysis reaction

Experiments in 1 bar argon Experiments in 10 bar methane

x ("C)

kV(%)

402.7 403.8

2.37 1.87

There is no statistical difference between the two means. The empirical designation of the upper limit of the temperature for the distillation (evaporation) range, as 400 "C has been confirmed by the experimental results. Therefore the weight loss up to 400 "C (AG400) may be regarded as a distillative loss, which will be followed by the weight loss due to cracking of the sample in the temperature range above 400 "C. The temperatures in Table 4-41 show good correlation with the start temperature of the cracking reaction found by thermogravimetry TB= 412.2 "C (see chapter 4.2.7).

4.2.8.4 Correlation of kinetic parameters with analytical data An attempt to correlate the kinetic parameters of the pyrolysis reaction with analytical data was only partly successful. A fairly high coefficient of correlation r was found for the relation of the activation energy E with the concentration of paraffin bonded carbon (CP) in the samples: in 1 bar argon: E = -0.3268 . CP + 339.0 in 10 bar methane: E = -0.3185 . CP + 375.5

r = 0.8659

r = 0.8702


177

Correlation of the activation energy with the components of the colloidal system, demonstrates a distinct relation with the concentration of the dispersion medium. The plot of activation energy E versus the concentration of dispersion medium DP [E = f(DP)] shows two straight lines with different slopes for each reaction in 1 bar argon and in 10 bar methane. The experiments in argon at 1 bar pressure show, in addition, a group of six samples, which gives a mean X with a small coefficient of variation kV. This mean is found in the plot, below the line with the lesser slope. Table 4-42 shows which samples appear on which line in Fig. 4-35 (experiments in 1 bar argon) and in Fig. 4-36 (experiments in 10 bar methane) as well as which samples form the group. Four samples, which are found on line no. I in the experiments at 1 bar argon, are also there in the experiments at 10 bar methane. But only six samples were found to stay on the same line 11, both in 1 bar argon and 10 bar methane (samples 2, 4, 5 , 11, 15, 21). No systematic causal relationship could be found. Statistical evaluation of the six samples forming the group in the experiments in 1 bar argon gives: -

+V 5.05 % 3.45 %

X

activation energy E dispersion medium

171.33 kT/Mol 68.6 wt%

Table 4-42: Distribution of the samples in Fig. 4-35 and Fig. 4-36

Line No. I 1 6 19 20

Experiments in 1 bar argon Line No. I1 Group 2 4 5 11 14

Experiments in 10 bar methane Line No. I Line No. II

3 7 8 10 12 13

!5

1 6 7 10 12 14 16 17 19 20 22

16 17 18 21 22

2 4 5 8 9 11 13

15 21

The coefficients of the equation of the line E = a DP + b in the Fig. 4-35 and Fig. 4-36 are listed in Table 4-43: Table 4-43: Coefficients u and b and coefficient of correlation r 1 bar argon

10 bar methane

Line No.

I

I1

I

I1

U

-7.29 805.54 0.8109

-1.23 286.17 0.9706

-2.93 482.46 0.8415

-1.72 341.27 0.8643

b r

178

160


mean o f s i x samples

4 1 30

, ( , , , , , , , , , , , ( , , ( , , ( , , , , , , , , , ( /

40

50

70

60

80

90

/

,

(

1

%DP

Fig. 4-35 Pyrolysis of Bitumen in 1 bar Argon Activation Energy E versus Concentration of Dispersion Medium

Correlation of the activation energy with the concentration of maltenes (dispersion medium + petroleum resins) shows a similar trend, i. e. decrease of the activation energy with increase of the concentration of maltenes. Only one line is produced for each pressure range and the coefficients of correlation are considerably lower, so the numerical values are not listed. It is common knowledge that the dispersion medium contains the highest content of compounds with aliphatic (paraffinic) bonds. Correlations of the activation energy E with the concentration of dispersion medium DP as well as with the carbon in paraffinic bond CP demonstrate clearly, that pyrolysis starts with the aliphatic (paraffinic) C-C-bonds.


179

10

Fig. 4-36 Pyrolysis of Bitumen in 10 bar Methane Activation Energy E versus Concentration of Dispersion Medium

The less distinct correlation of the activation energy with the concentration of maltenes is not contradictory, because the content of petroleum resins in the maltenes includes a corresponding concentration of aromatic compounds. Those aromatics could not be pyrolyzed under the current experimental conditions, however aliphatic side chains of substituted aromatics are pyrolysed. This theory is supported by comparison of the contents of the dispersion medium or the maltenes with the conversion during the DSC experiments. The ratio of the conversion in 1 bar argon to the concentration of dispersion medium (Table 4-44) has a value well above one for the vacuum residue and bitumens, showing a mean 2 = 1.14 (+V = 7.24 %). The

180


value for the semiblown residue from Kirkuk crude (sample 6) is only 0.88. The same is true for the reaction in 10 bar methane. Here a mean X = 1.11 ( -t V = 7.48 %) was found. Again sample 6 has a low value of 0.89. Table 4-44: Means of the ratios Conversion: Concentration of Dispersion medium

Samples Vacuum residues and bitumens samples 1-5, 7-13 Atmospheric residues samples 14-17 Products from conversion processes samples 18-22

-

X

Reaction in 1 bar argon i V

Reaction in 10 bar methane

-

X

iV

1.14

7.24

1.11

7.48

0.93

1.82

0.93

1.14

0.90

7.28

0.91

6.79

All the products from conversion processes (samples 18-22) show ratios well below one. Only the waxy distillate (sample 23) has a value of 1.03. The atmospheric residues (samples 14-17) demonstrate the same mean of the quotient, X = 0.93, for the reactions both in argon and in methane. The high weight loss due to evaporation may have influenced the values of the ratio. There is no statistical difference between the means of the atmospheric residues and of the products from conversion processes. But a difference does exist between these two means and the mean of the vacuum residues and bitumens. The ratio of the conversion to the concentration of maltenes demonstrates similar results (Table 4-45): Table 4-45: Means of the quotients Conversion: Concentration of maltenes

Samples Vacuum residues and bitumens samples 1-5, 7-13 Atmospheric residues samples 14-17 Products from conversion processes samples 18-22

-

X

Reaction in 1 bar argon f V (%)

Reaction in 10 bar methane X +V (%) ~

1.02

6.59

1.oo

7.80

0.89

4.07

0.89

4.39

0.88

6.78

0.88

6.34

Here again samples 6 (0.82) and 23 (1.02) do not conform to the group. Formally a ratio > 1indicates that a higher than the total quantity of dispersion medium (maltenes) present in the sample has been converted. On the contrary a ratio < 1.O indicates that a conversion was less than the total quantity of dispersion medium (maltenes).


181

However, this does not exactly describe the real conditions. On the one hand the dispersion medium contains aromatic compounds, which are not cracked under the experimental conditions applied here; on the other hand petroleum resins as well as asphaltenes contain parafinic (aliphatic) structures, which can be cracked. Further support for the theory that pyrolysis starts preferentially with the aliphatic C-C-bond, is given by the correlation of the residues after the reaction in 1 bar argon, R, or in 10 bar methane, R,,, respectively, with the atomic ratio H/C of the samples, which present roughly the aromatic portion of the samples, for example: H/C

Hydrocarbon

Class

4.0 3.0 2.2 2.03 2.0 1.33 1.22 1.oo 0.80 0.73

Methane Ethane Decane Hexacontane Cyclohexane Propy lbenzene Ethylbenzene Benzene Naphthalene Anthracene

Alkanes Alkanes Alkanes Alkanes Naphthenes Alkyl aromatics Alkyl aromatics Aromatics Aromatics Aromatics

For the pyrolysis reaction in 1 bar argon in the DSC, a correlation of the residue R, with the ratio C/H may be calculated as follows:

R, = -59.42 H/C

+ 108.68

Y

= 0.7246

The calculation for the residue of the reaction in 10 bar methane results in a similar correlation: R,, = -57.69 H/C

+ 106.19

Y

= 0.7215

Although the coefficients of correlation r are not very high, they clearly indicate the relationship, that with an increasing H/C ratio (i.e. decreasing aromacity), the quantity of coked products decreases.

4.2.9 Conclusions from experiments on refinery residues 4.2.9.1 Thermogravimetry Thermogravimetry on refinery residues gives results which may be repeated (by the same analyst using the same instrument) and reproduced (by another analyst using other instruments) within acceptable tolerances as has been determined by comparative experiments. The decrease of the boiling temperatures by a considerable amount, caused by a sufficient flow rate of the inert gas through the oven, makes a simulated distillation

182


feasible, and this continues directly onto the atmospheric distillation according to DIN 51 751 (ASTM D 285-62), as shown by other authors [4-151. The application of this simulated distillation is restricted to substances which do not evaporate in the gas flow at room temperature, because under those conditions, exact determination of the start temperature is impossible. Simulated distillation is best applied to substances such as atmospheric residues and higher boiling blends. Simulated distillation using thermogravimetry competes with simulated distillation by gas chromatography. However, it has the advantage that no damage is done to the thermogravimetric instrument by deposited coke residue from crachng reactions, when the latter may definitely be identified. To determine which products may be obtained by means of distillation processes and in what quantities, it is not necessary to calculate the complete distillation curve. A series of index numbers, which may be taken directly from the thermogram supply useful corresponding values. The index number, AG200, represents the quantity obtainable by atmospheric distillation and the difference, AG300 - AG200, the quantity separable by vacuum distillation. The temperature of 400 "C marks the upper limit of the distillation range and transition to the cracking range in thermogravimetry. Therefore the index number 4G400 marks the maximum quantity of substance which may be obtained by distillative methods (high vacuum distillation). The index numbers R500 or R600 roughly characterize the Conradson coke residue, and the index number R800 represents the final state of the coking process. R800, will be considerably less than the Conradson coke residue with an average of approximately 25 %, as has been described in the literature [4-12, 4-16]. The difference of 100 % minus the sum of the distillable fraction of the sample, AG400, and the coke residue, R800 gives the percentage of the sample which can theoretically, be cracked, CR: CR = 100 - (AG400 + R800) On the other hand, the difference of 100 % minus the sum 4G300 + R800 represents the portion of the sample which may be cracked under working conditions of the cracker in the refinery (in practice) PCR: PCR = 100 - (AG300 + R800) The Conradson coke residue may be characterized quite accurately by the residue at the point of inflexion Gwof the TGA curve. Statistical evaluation shows the temperature at this point of inflexion T, from 500-510 "C. Therefore the index number R500 represents the Conradson coke residue. It is likely that there is a distinction between the distillation residues, G, I25 wt%, and blown products and cracker residues of Gw > 30 %. The absolute values of the peak maxima of the DTG curve, which represent the reaction rate are not very helpful. The peak height and peak width indicate the uniformity of the substance which reacts in this temperature range. Uniform substances show high, narrow peaks, whereas flat, broad peaks represent blends of substances with a wide range of boiling or crack temperatures, The position of the peak maximum temperature of the DTG curve, T-, shows whether the reaction is still a distillative process (Tmx < 400 "C) or a cracking (pyrolysis) reaction (Tmx > 400°C). The latter gives an arithmetical mean X = 457.5 "C independent of the origin of the samples. From the derived index num-


183

bers, the ratio CR/ND, which represents the crackable part of the sample within the nondistillable part, may be used to distinguish between vacuum residues, bitumen, and atmospheric residues when CR/ND r 0.80, and heavy products from conversion processes with CR/ND = 0.63. The lower value for the conversion products indicates that considerable portions of the sample have already been cracked and removed by distillation. Correlation of the Conradson coke residue or R800 with the non-distillable part of the samples does not give significant results, nor does correlation of the reaction rate with the average molecular weight. Considering the temperature limits in use in refinery processes; a value of 100 % simulated atmospheric residue, SAR, is found in most of the samples of investigation. By this definition, the four atmospheric residues have not been fully distilled. The simulated vacuum residue, SVR, shows how far the samples are distilled. Since the SVR = 100 AG300, exhibits higher values than the non-distillable portion of the sample ND = 100 AG400, the values of that part of the sample which is crackable in practice, PCR, exceed the more theoretical values, CR. The ratio PCR/SVR does not give a clear distinction of distillation residues from conversion products than it is possible using the ratio CR/ ND. The relation of the different index numbers characterizing the coke residue to the simulated vacuum residue, SVR, does give useful results. The relation of the Conradson coke residue CCR to SVR results in ratios between 10.0 and 23.4 for vacuum residues and bitumens; to ratios from 12.2 to 14.8 for atmospheric residues; and the ratio for residues from conversion processes gives values of over 27 up to 43. On the other hand the ratio G,/SVR does not distinguish clearly between vacuum and atmospheric residues, but semi-blown products and residues from conversion processes clearly show higher values. The relation of R8O0 to the simulated vacuum residue SVR is similar. Since the values of R800 are lower than the corresponding values of CCR and G,, the ratio R800ISVR generally has lower values than both the other index numbers (CCR/SVR and G,/ SVR). The experiments in air yield less interesting results with regard to the further possibilities of processing residues, than do the experiments in inert atmosphere. There is no difference between the index numbers T1 % and T5 % from the experiments in inert atmosphere and in air, but the weight loss up to 400 'C, AG400, is less in the experiments in air than in inert gas. This indicates the formation of non-volatile oxidation products. In the experiments in air no residues from a colung process have been found. This does not exclude the formation of such residues as intermediates during the course of the reaction. In each case the TGA curve demonstrates the value of zero (or very little ash residue) at temperatures exceeding 550 "C. In contrast to the experiments in inert gas, the DTG curves for the experiments in air exhibit up to ten peaks, of which the first peak, representing a distillative process, is at the same temperature as that in the experiments in inert gas (arithmetical means: in argon X = 304 OC;in air X = 31 1 "C). Obviously the start of the oxidation depends on the chemical nature of the sample. Even if a clear correlation cannot be computed, high temperatures of

184


the first oxidation peak Toxhave been found for samples possessing high aromacity, FA: Sample no.

FA

To, ("C)

0.461 0.515 0.402 0.676

443 400 566 588

~~

18 23 24 25

For the other samples, the first oxidation peak maximum was found in a temperature range from 380-390 "C. The reaction rate does not give significant values and will not be discussed here. The existence of a cracking reaction during the oxidation cannot be excluded because corresponding maxima of the DTG curves both in air and in argon were found at equal temperatures. In simulated cracking experiments in the thermobalance, the temperatures of the cracking maximum of the DTG curves, Tc,are shifted to higher values (505-560 "C) due to the very fast heating rate, ,6 = 100 K/min. The group of 13 vacuum residues and bitumens exhibits the highest mean of the peak maximum temperatures, whereas the group of atmospheric residues and the group of conversion products demonstrate a lower temperature, but at the same level. The means of the reaction rate, DTGc, show the same trend. The lower values of the reaction rate DTG for the residues from conversion processes (samples 19-22) and the high values of their crack residue R, show once more that the samples have previously undergone crack reactions. Fui-ther comparisons such as relating the reaction rate to the average molecular weight, do not give significant results. The start temperature of the cracking process TBwas ascertained according to Adony [4-141. A mean X = 414 "C ( fV = 4.2 %) has been found independent of the provenance of the samples. The empirical determination of 400 "C as the end of the evaporation range and the start of the pyrolysis (cracking) range, is thus justfied.

4.2.9.2 Reaction kinetics Investigation of the reaction kinetics in an inert atmosphere using DSC confirms the difference between evaporation and cracking processes. However, the extrapolated onset temperatures of the evaporation exceed the T5 % index numbers of thermogravimetry considerably. On the contrary the start temperatures (onset) of the cracking reaction in DSC were found at nearly equal temperatures, to those from thermogravimetry, TB:

x ("C)


185

Neither the pressure nor the chemical nature of the inert gas influences the onset temperatures noticeably. The activation energy determined for the evaporation process is of the same magnitude as the enthalpy of vaporization, Hv, and depends closely upon the pressure. At atmospheric pressure and an equal heating rate, the temperature of the evaporation maximum of the DSC curve is found at the same temperature as the corresponding maximum in the DTG curve. The same behavior has been found for the temperatures of the DSC crack maximum and the corresponding maximum in the DTG curve (both at equal heating rates). While the enthalpy of vaporization depends on the nature and the boiling temperature of the component evaporating, nearly equal values of the activation energy and the frequency factor of the pyrolysis reaction have been found, independent of the provenance of the samples. When the activation energy is calculated according to equation 3-12, both the lineal and square values of the absolute peak temperature are used. Therefore the individual values of the kinetic coefficients for each sample show considerable scattering. Consequently, arithmetical means with relative large coefficients of variation will be given by statistical evaluation, but if statistical evaluation of the peak maximum temperatures for each heating rate is carried out first, and the resulting means are applied to the calculation of the kinetic coefficients, then average values with coefficients of vaiation fV = 1.2 % (maximum) are found. The plot of log heating rate oversus the inverse Kelvin temperature 1 000/T shows that the average values fit a straight line and do not require any regression calculation; whereas the temperatures from the individual measurements do not fit the straight line so perfectly. The average values of the activation energy and the frequency factor computed from the means of the corresponding peak maximum temperatures are considered to be reliable. Comparison of the means of the two methods of calculation shows no difference between the average values and the arithmetical means. Increasing pressure up to 10 bar gives values of the kinetic parameters approximately 20 % higher than those of the reaction at 1 bar pressure. The pyrolysis reaction produces an increase in volume, due to the formation of crack gases and easily volatile fragments from the original long-chain liquid or solid components. This behavior is in accordance with Le Chatelier’s principle. Pyrolysis is a first order reaction so the temperature function of the reaction rate constant k and the half life time t1,2may be computed easily using the coefficients of the Arrhenius’ equation: activation energy E and frequency factor A which had already been determined (see chapter 3.3.1, equations 3-7 and 3-8). Such data are the basis for the parameters of thermal conversion processes, such as temperature of the plant installations, housing time etc. It should be stated here that there is a question over the method for the calculation of the values for the activation energy E and the frequency factor A according to ASTM E 698-79, based on investigations by Kissinger [3-15, 3-16]. Kissinger makes the supposition that the temperatures of the maximum of the reaction rate coincide with the temperatures of the maximum of the caloric effect, when both are at the same degree of conversion, and independent of the heating rate. This assumption does not generally prove true. This may also explain the variable differences between the maximum temperatures of the DTG and the DTA curves at equal heating rates, and may even be the source of diverging values

186


of the Arrhenius kinetic coefficients, if the calculation has been carried out using DSC and DTG peak maximum temperatures of experiments with different heating rates. Nevertheless control experiments at different heating rates on the substances of this investigation, using a Simultaneous Thermal Analyzer, showed that differences in the maximum temperatures of the corresponding DTG and DTA curves fell within the tolerance limits of the method, for most of the samples. Moreover the maxima were found at equal conversions independent of the heating rates applied. We should not overlook the fact that the data acquired from the experiments are themselves averages of a series of parallel and/or consecutive reactions of multicomponent systems. The experiments on model substances give values of the activation energies and frequency factors comparable to data acquired by other methods. The data presented here are certainly not absolute data but solid values adequate for comparative purposes. The value of the pyrolysis enthalpy found by using integration of the peak areas should be independent of the heating rates, but is not so in practice. An increase in the enthalpy values was observed with increasing heating rates. Because of the difficulties experienced in finding the correct integration limits, the data presented here should be regarded only as approximate values. Nevertheless they exhibit orders of magnitude comparable to published data [4-17 to 4-19], which also show very varied values. The pyrolysis enthalpy must be known exactly, so that the supply of process energy can be calculated and this should be considered when the plant is set up. The considerable differences between the pyrolysis enthalpies determined in this investigation, do not only originate in the different bonding enthalpies of different kinds of chemical bonds which had to be cracked. The values also include the enthalpies of vaporization of the different crack products formed during the reaction, which may be of considerable magnitude.

4.2.9.3 Correlation of data from thermoanalysis with analytical data An attempt to correlate thermal behavior of a multicomponent system such as refinery residues with analytical data, did not give satisfactory results in most cases. The analytical data are mean values of a multicomponent system, whereas thermal behavior is strongly influenced by individual compounds or groups of similar compounds. The group of distillation residues, and products from conversion processes which already have a considerable themial history, which was selected, comprises a too wide range of products. Thus only a few thermoanalytical data can be correlated with data from the structural group analysis. The hypothetical “average molecule” of the strucutral group analysis does not exist and cannot govern the thermal behavior. Data describing the average size of the molecule, such as the average relative particle mass and the carbon number ZC, may be correlated satisfactorily with data from thermoanalysis, which describes the evaporation behavior, such as the start temperatures of evaporation, T1 % and T5 %, and the distillable fraction of the sample, AG400. Surprisingly and without logical explanation, the residue 11800 can also be correlated.

a

4.3 Investigations on bitumen

187

Data from colloid analysis show that the concentration of the dispersion medium may be related to the distillable fraction AG400, whereas the concentration of asphaltenes, or the total of asphaltenes and petroleum resins, determines the quantity of coke residue after pyrolysis. That portion of the sample which can be cracked, CR, will usually be determined from the concentration of petroleum resins. The aliphatic side chains of the alkylaromatic system of the asphaltenes have a small influence. The coke residue can be related to the data from structural group analysis which describe the aromatic character of the samples. Thermogravimetry in air only shows the correlation of data describing the molecule size with those index numbers which are relevant to the evaporation behavior. The reaction kinetic constants: activation energy E and frequency factor A , can only be correlated with the concentration of paraffinic carbon, CP (from structural group analysis); with the concentration of dispersion medium (from colloid analysis); and with the H/C ratio (from elemental analysis). These functions show correlation coefficients of an acceptable magnitude. Examination of the correlation of the concentration of maltenes revealed a similar tendency but with very low coefficients of correlation. It is well known that the dispersion medium contains the highest concentration of chemical bonds, which can be cracked under the chosen reaction conditions [4-201. In the pyrolysis experiments from distillation residues, about 92 % of the dispersion medium was converted, whereas conversion of the petroleum resins was only 83 %, despite the fact that the kinetic coefficients are of nearly the same magnitude for the two components.

4.3 Investigations on bitumen Bitumens are residues of the vacuum distillation of suitable crude oils (distillation bitumen). Residues of less suitable crudes must be partially oxidized by blowing to achieve the desired technical properties (semi-blown bitumen). Blown bitumens for special purposes can also be produced from vacuum residues. The process is executed by blowing a stream of finely distributed air through the molten bitumen (sometimes reduced in viscosity by addition of flux oil) at temperatures of 250-290 "C. The bitumens are usually characterized by needle penetration (DIN 52 010, ASTM D 5-73). The measured value corresponds to the penetration of a needle loaded with a defined weight at 25 "C within 5 seconds. The depth is calibrated in tenths of a millimeter (0.1 mm). The commercial range comprises bitumens of the grades from B200, B80, B65, B45, B25 down to B10. Blown bitumens are characterized by two values: the softening point ring & ball test (S.P. R&B) and the needle penetration described above. According to DIN 52 01 1 (ASTM D 2398-68T) the softening point R&B is defined as the temperature when a ball passes through a bitumen-filled ring in a heating bath. The designated values consist of a combination of softening pointlpenetration, such as 85/25,85140,100125, and 100/15 for the most commonly used blown bitumens [4-211.

188


These data are subject to relative wide tolerance ranges; therefore, individual testing of the penetration and the softening point R&B is advisable if eventual correlation with other values is attempted. Bitumens are colloid systems, as are crude oils, and consist of the two colloidal components, petroleum resins and asphaltenes, dispersed in a dispersion medium. To investigate the composition of the system, a colloid precipitation according to Neumann [4-101 is carried out. The chemical nature of the bitumen and its components were determined by element analysis, where the atomic ratio H/C includes an indicator of the aromacity. Further characterization is performed by measuring the average relative particle mass (mean of the molecular weight M> by vapor pressure osmometry. The consistency data (penetration and softening point), the analytical data, and the data from thermoanalysis have been investigated to discover whether there is any correlation between them. The aim is to find some simple means of characterization of the different bitumen types using thermogravimetry. Furthermore, the possibility of replacing the rotating flask aging test according to DIN 52 017 (ASTM D 2872-85), with isothermal thermogravimetry was investigated. DSC experiments in inert gas and in air on bitumen and its separated colloid components should reveal the quality and quantity of the influence of such compounds on the thermal and oxidation stability of the colloid system. The experiments were carried out on sixteen distillation bitumens of different consistency and different origin as well as on five blown bitumens [4-201. Bitumen may only be used technically within relative narrow temperature ranges, which are largely determined by consistency data, such as the softening point R&B and the brealung point according to FraaU (DIN 52 012, IP 80 153). This range, also called the span of plasticity, is bounded towards lower temperatures by increasing brittlenes and towards higher temperatures by plastic deformation under load. Addition of polymers to bitumen may extend the span of plasticity [4-221. The influence of polymers upon the aging properties and upon the low temperature behavior may be clarified using thermoanalytical measurements.


189

4.3.1 Description and characterization of the samples The type and origin of the samples is shown in table 4-46: Table 4-46: Type and origin of the samples Sample No.

1 2 3 4 5 6 7 8 9 10

Symbol

Type

Origin

Z

B 200 B 200 B 200 B 80 B 200 B 80 B 65 B 45 B 25 B 80

Venezuela Forties + Arabian Light Venezuela Venezuela Venezuela Venezuela Venezuela Venezuela Venezuela Iranian Light + Souedi + Khafji Arabian Heavy

Y

+ + 0 0 0 0 Y

14

+

15 16 I

X

I1 111

IV V

Y

*

* *

* *

Vacuum residue (abt. B 200) Vacuum residue (abt. B 130) Vacuum residue (abt. B 80) Vacuum residue (abt. B 65) B 200 B 10 Blown sample no. 11 (Laboratory product abt. 85/25) Bitumen 100/25 Bitumen 85/40 Bitumen 85/25 Bitumen 100/25

Kuwait Khafj

+ Russia

Kirkuk Venezuela unknown West German refinery West German refinery West German refinery South German refinery

Use of the same symbols in the plots shows distillation bitumens taken from the same refinery. Only the samples marked by a square symbol were taken at the same time. Each of the blown bitumens is marked with an asterisk. Table 4-47 shows the consistency data: needle penetration (DIN 52 010, ASTM D 5-73) and softening point ring & ball (DIN 52 011, ASTM D 2398-68)

190


Table 4-47: Consistency data of the samples (arithmetic means) Sample No.

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 I

39.0 39.5 43.1 52.5 47.3 55.0 57.9 61.3 70.5 46.5 40.0 50.1 52.8 53.3 42.9 90.5 86.5 117.8 100.5 91.3 110.0

I1 III

rv V

159.0 186.0 156.2 72.7 161.5 75.4 49.3 32.2 21.9 75.0 196.0 127.0 75.8 50.3 209.0 10.2 18.0 22.8 32.3 21.7 20.9

The colloidal composition of each sample was analysed using the colloid precipitation according to Neumann [4-101 (Table 4-48): Table 4-48: Colloid composition of the samples (wt %). Sample No.

1 2 3 4 5 6 7 8 9 10 11 12 13

Type

Dispersion Medium

Petroleum Resins

Asphaltenes

Resins Asphaltenes

B 200 B 200 B 200 B 80 B 200 B 80 B 65 B 45 B 25 B 80 VR (B 200) VR (B 130) VR (B 80)

78.09 76.95 73.38 63.35 65.58 59.60 66.22 64.81 63.31 75.60 71.60 71.76 61.61

2.91 8.51 9.84 23.70 19.88 21.83 15.10 15.09 13.59 2.88 7.40 10.50 19.86

19.00 14.54 16.78 15.30 14.54 18.57 18.68 20.10 23.10 21.52 21.00 17.65 18.45

0.153 0.585 0.586 1.549 1.367 1.176 0.808 0.75 1 0.588 0.134 0.352 0.594 1.076


Sample No.

14 15 16 I I1 I11 IV V

Type

Dispersion Medium

Petroleum Resins

Asphaltenes Astaltenes

Resins

VR (B 65) B 200 B 10

66.89 76.67 70.95 56.10 66.29 67.23 59.97 53.62

14.99 7.92 6.18 8.71 3.51 2.44 6.54 11.96

18.21 15.41 22.87 35.14 30.20 30.33 33.49 34.41

0.823 0.514 0.270 0.248 0.116 0.080 0.195 0.348

-

100125 85/40 85/25 100125

191

The elemental composition and the average relative particle mass of the bitumens and their separated colloid components were determined (Tables 4-49 - 4-52)

Table 4-49: Element analysis (wt %) and average relative particle mass of the bitumen. Sample No.

Bitumen

C

H

N

S

H C

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 I

I1 111 IV V

B 200 B 200 B 200 B 80 B 200 B 80 B 65 B 45 B 25 B 80 VR (€3 200) VR (B 130) VR (B 80) VR (B 65) B 200 B 10 -

100125 85/40 85/25 100125

-

M

84.24 85.61 84.46 84.76 85.50 85.13 84.44 85.00 85.05 82.15 84.13

10.45 10.21 10.34 10.31 10.22 10.28 10.66 10.08 10.10 10.19 10.14

0.66 0.39 0.50 0.61 0.56 0.60 0.47 0.65 0.60 0.50 0.42

3.17 3.95 3.18 3.16 3.15 3.21 3.01 3.20 3.20 5.54 5.18

1.489 1.431 1.469 1.460 1.434 1.449 1.515 1.475 1.425 1.488 1.446

733 929 727 805 788 865 973 1064 1123 753 839

83.92

9.84

0.33

5.02

1.407

1045

83.28

10.28

0.62

4.92

1.481

984

83.77

9.75

0.37

5.68

1.397

95 1

85.04 83.47 84.85 86.13 86.20 85.70 84.59

10.11 9.71 10.04 10.18 10.27 10.10 10.37

0.58 0.47 0.31 0.44 0.39 0.38 0.42

3.53 4.92 5.01 2.72 2.70 2.95 3.73

1.427 1.396 1.420 1.419 1.430 1.414 1.471

674 1149 1117 68 1 738 784 948

192


Table 4-50: Element analysis (wt %) and average relative particle mass of the dispersion medium Sample No. 1 2 3 4 5 6 7 8

9 10 11 12 13 14 15 16 I I1 111 IV V

Bitumen

C

H

N

S

H C

B 200 B 200 B 200 B 80 B 200 B 80 B 65 B 45 B 25 B 80 VR (B 200) VR (B 130) VR (B 80) VR (B 65) B 200 B 10

84.00 83.63 84.80 84.10 85.10 85.13 85.62 85.56 84.36 82.66 84.19

11.01 10.43 10.93 10.68 10.87 10.73 11.05 10.84 10.61 10.70 10.77

0.46 0.28 0.38 0.45 0.45 0.43 0.42 0.40 0.37 0.41 0.28

2.88 4.68 2.77 2.84 2.86 2.79 2.87 2.87 2.76 4.65 4.54

1.573 1.497 1.547 1.524 1.553 1.513 1.549 1.520 1SO9 1.553 1.535

83.69

10.39

0.32

4.61

1.490

792

83.21

10.78

0.38

4.02

1.555

823

83.21

10.31

0.45

4.99

1.485

738

84.89 83.49 84.06 83.86 84.39 85.27 84.62

10.77 10.77 11.15 11.20 11.30 11.21 11.43

0.34 0.27 0.17 0.62 0.54 0.41 0.55

3.07 3.97 4.15 2.3 1 2.31 2.45 2.67

1.522 1.540 1.592 1.603 1.607 1.578 1.621

465 727 669 569 439 573 589

-

100/25 85/40 85/25 100/25

~

M

568 705 540 576 842 966 1001 87 1

932 63 1 776

Table 4-51: Element analyis (wt %) and average relative particle mass of the petroleum resins. Sample No.

Bitumen

C

H

N

S

H

~

-

M

C ~

1 2 3 4 5 6 7 8 9 10 11 12

B 200 B 200 B 200 B 80 B 200 B 80 B 65 B 45 B 25 B 80 VR (B 200) VR (B 130)

82.12 82.78 84.68 82.66 84.46 84.41 84.44 84.91 84.29 81.92 83.92

10.64 10.92 10.46 10.53 10.58 10.60 10.66 10.77 10.45 10.90 11.05

0.55 0.32 0.57 0.62 0.42 0.42 0.47 0.44 0.45 0.54 0.20

2.69 4.78 3.01 2.87 2.93 2.74 3.01 2.82 3.15 4.48 4.63

1.564 1.583 1.482 1.529 1.503 1SO7 1.515 1.522 1.488 1.597 1.580

935 1000 968 984 1116 1171 1293 1310 1399 860 920

82.46

10.14

0.49

4.91

1.476

1I96

193


-

Sample No.

Bitumen

C

H

N

S

-

13

VR (B 80) VR (B 65) B 200 B 10

84.35

10.49

0.44

4.16

1.492

983

83.38

10.26

0.29

5.29

1.477

1028

83.99 85.25 83.30 83.80 83.61 84.62 84.98

10.05 9.97 11.10 11.43 10.62 11.10 11.40

0.74 0.24 0.20 0.21 0.28 0.22 0.26

3.54 4.40 3.56 3.17 2.45 2.64 2.11

1.436 1.403 1.589 1.637 1.524 1.574 1.610

1130 1403 989 577 818 801 887

14 15 16 I I1 I11 IV V

-

100/25 85/40 85/25 100/25

Table 4-52: Element analysis (wt %) and average relative particle mass Sample No.

Bitumen

C

H

N

S

H C

of the asphaltenes

H C

___

I 2 3 4 5 6 7 8 9 10 11 12 13 14

15 16 I 11 111 IV V

M

-

M

~

B 200 B 200 B 200 B 80 B 200 B 80 B 65 B 45 B 25 B 80 VR (B 200) VR (B 130) VR (B 80) VR (B 65) B 200 B 10 -

100/25 85/40 85/25 100/25

84.37 81.94 84.65 84.69 83.62 84.03 84.18 85.34 83.53 81.53 83.00

8.09 7.68 8.08 7.96 7.76 7.97 8.02 8.01 8.27 8.06 7.60

1.44 0.91 1S O 1.61 1.24 1.35 1.42 1.47 1.42 1.10 0.85

4.21 7.03 4.06 4.23 4.45 4.49 4.33 4.42 4.03 7.80 7.40

1.151 1.125 1.145 1.128 1.136 1.138 1.143 1.126 1.160 1.196 1.099

3680 4480 5380 8156 6139 7328 7870 7797 5567 2489 5938

82.54

8.08

1.08

7.23

1.175

6092

81.52

7.86

1.04

6.91

1.157

2767

81.06

7.86

0.76

9.07

1.164

4700

84.48 82.41 82.91 85.21 85.82 82.95 83.17

7.92 8.40 7.53 8.42 7.86 8.11 7.82

1.11 0.74 0.47 0.78 0.86 1.22 0.78

4.94 5.84 6.43 3.75 3.60 3.69 5.66

1.125 1.223 1.090 1.186 1.099 1.173 1.128

4213 6400 5897 4075 6420 5126 8373

194

4 Therrnoanalytical Investigations on Petroleum and Petroleurn Products

The atomic ratio H:C demonstrates relatively weak aromacity for bitumens, dispersion medium and petroleum resins, whereas the asphaltenes possess a considerably higher aromacity. Despite differences in the colloid composition, differences of the H C ratios of the individual bitumens are very small, as shown by the small coefficients of variation + V in Table 4-53: Table 4-53 Means X and coefficients of variation ?z V (%) of the H:C ratio of the distillation bitumens

Bitumen Dispersion medium Petroleum resins Asphaltenes

1.447 1.532 1.514 1.146

16 16 16 16

3.11 1.92 3.58 2.85

There is no statistical significant difference between the means of the dispersion medium and the petroleum resins. The other means do exhibit differences (99.9 % statistical significance). It is notable that the dispersion medium, as well as the petroleum resins of the blown bitumen are more aliphatic in character than those components of the distillation bitumens. This may be caused by the composition of the flux oils which were added. The asphaltenes of both types of bitumen display similar aromacity (Table 4-54): Table 4-54 Means mens

X and coefficients

of variation f V (%) of the H:C ratio of the blown bitu-


X

n

+V (%)

1.431 1.600 1.587 1.135

5 5 5 5

1.62 1.01 2.67 3.79

There is also no difference only between the means of the dispersion medium and the petroleum resins. Comparison of the means from the distillation and the blown bitumens reveals a difference between the petroleum resins of each type of petroleum, whereas the means of the bitumens and the asphaltenes are the same. The colloid composition of the blown bitumens shows higher asphaltene content and lower dispersion medium content than that of the distillation bitumens. As expected the penetration decreases with increasing concentration of asphaltenes, whereas the softening point increases. A plot of the consistency data versus the concentration of dispersion medium or maltenes naturally demonstrates the inverse tendency. The relation of the consistency data to the average relative particle mass M of the bitumens also reveals an increasing softening point with increasing M, whereas the penetration decreases.

4 . j Investigations on bitumen

lY3

A fairly close correlation of the consistency data with the colloid composition or the particle mass, is only possible for bitumens from the same refinery and from crude oils of nearly the same origin.

4.3.2 Thermoanalytical investigations The following thermoanalytical investigations were performed: Thermogravimetry in inert gas Therniogravimetry in air Pyrolysis (DSC) in inert gas Oxidation test (DSC) in air The thermogravimetric tests were carried out in a thermobalance TG 750 (StantonRedcroft). The pyrolysis and the tests of oxidation stability were performed with a DuPont 990 Thermoanalyzer equipped with a 910 DSC and a pressure DSC cell.

4.3.2.1 Thermogravimetry in inert gas All the bitumens were heated in a flow of inert gas (argon, flow rate 25 cm3/min) from room temperature up to 900 'C, always applying the same heating rate = 10 K/min. A plot of the TGA curve versus the temperature of the distillation bitumen B80 (sample 6) and its colloidal components is shown in Fig. 4-37. The curves of the bitumen, the dispersion medium, and the petroleum resins differ very little from each other up to 450 "C. The differences shown by the curves in the temperature range above 500 "C represent the coke residue. The curve of the asphaltenes is shifted towards higher temperatures due to a higher particle mass and also has a considerably higher coke residue. The index numbers for thermogravimetry in inert gas, for the bitumens and their colloidal components are given in Tables 4-55 to 4-58. The only derived index number is the crackable part of the sample, CR, which is listed in Table 4-59. Evaporation of the distillation bitumens and their colloidal components first starts at temperatures 2 200 "C (T1 %) or 2 259 "C (2'5 %). Since thin-layer evaporation takes place in the thermobalance, and the evaporated parts of the sample are immediately removed by the gas flow, these temperatures are lower than the real start temperatures of an equilibrium evaporation (for example: distillation according to Engler; DIN 5 1 75 1 or ASTM D 285-62). The corresponding start temperatures for an equilibrium evaporation should be more than 400 "C. The evaporation start temperatures for bitumen, dispersion medium, and petroleum resins are lower in the case of blown bitumens, influenced by the flux oils which are added in order to facilitate the blowing process. Some of the index numbers of the thermogravimetry may be correlated with consistency data, and with analysis data (see chapters 4.3.2.1.1 and 4.3.2.1.2). Other values show only a small

196

0

4 Thermonnalytical Investigations on Petroleum and Petroleum Products

200

4 00

600

8 00

Fig. 4-37 Thermogravimetry (TG 750) of Bitumen and its Colloidal Components Heating Rate p : 10 K/min Atmosphere: Argon 25 cm3/min Curve 1: Bitumen B80 Sample 6 Curve 2: Dispersion Medium Curve 3: Petroleum Resins Curve 4: Asvhaltenes

1.9

223

188

100125

V

2.0

23I

181

85/25

IV

24.3

51.1 40.8

19.3 16.7

7.8

18.7 20.5

19.3 19.9 16.5

13.8

17.3 23.0 23.9 20.0

16.4

20.6

16.2

15.0 16.7 9.1 14.5 17.0 16.0 16.0 18.9 17.7 19.3 16.9

800 "C

18.5

R

24.2

20.0

23.5 24.0 20.0 17.4 22.2 19.5 19.1 20.8 21.7 26.3 20.7

600 "C

15.8 58.9

22.7 31.8 58.6

50.0

28.1

28.0

34.5

47.1 45.9 47.3 39.8 35.0 16.5 13.6 15.7 20.0 38.5 44.8

400 "C

21.4 29.0

4.5 4.0 27.7

-

0.1 2.3

20.8

2.1

1.8

0.2

2.0

-

15.7 4.8 15.2 7.9 4.1 2.1 1.6 1.2 1.2 6.9 7.9

300 "C

3.2

AG

-

-

0.3

-

-

0.4

-

-

0.9

200 "C

3.3

306 3 10 223

247 257 174

100 "C

217

85/40

I11

229

333

329

184

262

279

315

247 302 248 278 307 331 341 356 350 287 284

205 258 206 220 25 1 277 285 290 290 287 239

264

T5 %

T1%

169

100/25

-

16 I I1

15

14

13

B 10

B 200 B 200 B 200 B 80 B 200 B 80 B 65 B 45 B 25 B 80 VR (B 200) VR (B 130) VR (B 80) VR (B 65) B 200

1 2 3 4 5 6 7 8 9 10 11

12

Type of bitumen

Sample No.

Table 4-55: Thermogravimetry of bitumen in argon. -

2.43 7.83 7.97 7.35 2.66 3.56 2.77 1.93 2.34 3.78 1.70 4.51

9.90

8.80

9.91

4.96 7.00 6.68 8.69 8.07 10.44 10.82 11.82 10.48 6.18 6.49

DTG

305 419 433 435 295 428 29 1 440 325 434 333 416

428

432

428

Tmax 422 407 426 423 427 436 435 439 437 43 1 418

9

8B

0

0,

8

5-

2

20.0

1.9

237

178

100/25

V

33.2

3.2

213

109

85/25

IV

35.0

4.4

202

163

-

85/40

25.0 6.3 5.0 31.3

3.4 0.2 0.4 2.8

215 290 300 218

153 229 232 170

I11

3.2

1.o

327

200

100/25

3.0

0.2

319

248

15 16 I I1

14

13

4.0

0.9

3 10

214

12

1.1 1.3

-

0.9

-

15.3 5.0 18.0 13.0 6.3 4.1 3.0 2.5 2.3 9.1 12.3

1.1 0.9 2.9 1.6 0.8 0.5

248 300 227 25 1 288 310 319 336 328 274 257

194 204 194 178 213 227 258 216 267 190 187

B 200 B 200 B 200 B 80 B 200 B 80 B 65 B 45 B 25 B 80 VR (B 200) VR (B 130) VR (B 80) VR (B 65) B 200 B 10

1 2 3 4 5 6 7 8 9 10 11

AG

300 "C

100 "C

200 "C

T5 %

T1 %

Type of bitumen

Sample No.

Table 4-56: Thermogravimetry of dispersion medium in argon

11.9 10.2

52.8

7.3 67.4

80.0

5.7 9.2 6.1 8.0

65.8 39.5 40.0 70.0

7.9

30.0

10.1

10.9

35.2

17.3

10.2 7.5 11.8 15.4 18.7 10.2 12.4 9.0 7.4 7.3 8.3

600 "C

52.3 36.5 57.0 51.5 38.5 31.5 30.0 22.3 24.8 46.0 50.0

400 "C

R

6.0

1.o

3.4

4.3 6.9 5.1 4.0

9.8

5.8

6.9

7.3 6.2 6.7 6.3 6.3 6.8 10.6 6.4 6.4 6.3 4.0

800 "C

427 435 426 309 420 290 404 310 446 450

6.53 9.98 8.00 3.93 2.76 4.27 2.31 4.47 1.93 4.19

440

1 1.94

437

428 10.41

11.61

426 434 419 426 436 438 43 1 44 1 444 433 427

Tmax

7.50 10.18 6.26 5.92 10.26 11.57 11.48 13.89 13.27 7.94 7.71

DTG

9.4 7.9

14.1 11.4 13.2

46.5 51.6 27.5

18.0 23.4 11.0

1.4 2.3 2.4

239 222 245

191

177

157

85/40

85/25

100125

I11

IV

V

-

100125

14

10.9

17.0 20.0 8.4 2.6 19.9 21.2 16.8 6.2

13.2 18.1 34.0 73.7

9.9 3.8 5.0 37.0

2.1 1.o 1.0 2.8

245 320 300 215

170 208 192 175

9.50 10.68 8.93 4.47 2.60 2.44 4.26 2.15 5.28 6.74

448 448 436 290 421 317 455 290-320 436 491

447 12.08 15.1

15 16 I I1

16.8

17.1

3.0

1.2

341

444 11.18 12.1

171

13.7

25.9

4.4

12.64 16.4

1.0

13

12 309

447

9.42 9.73 9.59 9.50 10.42 10.17 13.27 15.12 12.38 9.13 11.21 17.7 21.0 18.0 12.3 13.0 11.9 15.2 13.1 16.5 18.2 14.5

200

18.3

14.9

447 44 1 443 447 443 447 448 454 447 455 447

DTG

800 "C

4.0

18.6 28.1 20.3 14.5 14.6 17.0 17.1 14.0 17.8 21.8 15.6

30.0 17.8 27.5 31.0 27.8 15.8 14.0 10.0 17.9 20.2 25.0

14.1 3.0 9.4 8.2 6.0 4.0 2.2 2.0 2.1 2.6 5.5

4.0 0.6 2.3 1.o 1.1 0.9 0.9 0.9 0.1 0.5 1.2

R

1.5

600 "C

400 "C

AG

300 "C

200 "C

323

100 "c

177

T5 9% 210 304 248 27 1 288 310 350 370 34 1 337 29 I

B 200 B 200 B 200 B 80 B 200 B 80 B 65 B 45 B 25 B 80 VR (B200) VR (B130) VR (B 80) VR (B 65) B 200 B 10

1 2 3 4 5 6 7 8 9 10 11

T1 %

146 223 145 200 193 212 226 23 1 257 236 187

Type of bitumen

Sample No.

Table 4-57: Thermogravimetry of petroleum resins in argon

5,

r W W

3

2

8

2

E.

oc

2 3

15 16 I I1 I11 IV V

14

13

100/25 85/40 85/25 100/25

B 200 B 200 B 200 B 80 B 200 B 80 B 65 B 45 B 25 B 80 VR (B200) VR (B130) VR (B 80) VR (B 65) B 200 B 10

1 2 3 4 5 6 7 8 9 10 11

12

Type of bitumen

Sample No.

-

217 202 298 212 213 216 204

256

279

284

278

213 229 234 3 14 248 349 23 1 313 235

T1 %

362 340 372 323 3 10 33 1 415

381

390

389

344 380 387 397 378 400 390 397 342 323 378

T.5 % 100°C

Table 4-58: Theimogravimetry of aphaltenes in argon

1

7 .O 6.5

1.2 1.6 2.5 3.0 1.1 4.0 2.7 3.9 1.7

0.4 0.6

0.8

9.3 14.8 9.9 9.6 11.1 10.4 4.0

7.0

0.8 2.6 0.2

-

0.3

1.2

9.O 7.0 6.9 5.6 7.3 5.0 5.2 5.7 10.0 8.9 7.5

400 "C

0.2

300 "C 2.9 1.9 1.9 0.9 2.0 0.3 2.0 0.9 4.0 4.6 1.2

AG 0.9 0.9 0.8 0.2 0.3 -

200°C

's 2a

447 448 448 447 438 447

7.03 7.00 6.42 6.23 5.94 6.65 3.24 3.13 3.28 5.01

46.9 48.8 49.8 40.8 45.2 38.0 51.0 36.0 46.2

51.0 52.1 42.8 47.2 55.2 58.2 51.5 49.9

49.1

449 448 508

x

a

"n

R

s

$

2

E. 2 8

07

g

$

~

k

g g

23

47.7 49.9

450 447 453 453 447 456 449

448 447 452

4.39 6.73 5.66 6.04 5.00 6.12 5.87 5.01 5.06 4.99 6.12

Q

43.8 49.5 50.6 51.3 51.5 51.0 51.5 53.5 48.0 45.0 49.0

Tmax

54.7 51.2 52.9 54.4 56.9 54.2 54.3 56.0 50.8 56.0 51.3

DTG

800 "C

600 "C

R

0

bJ 0


201

Table 4-59 Crackable Parts of the Sample, CR (wt %) Sample No.

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 I

I1 I11 IV V

Bitumen

Dispersion Medium

Petroleum Resins

Asphaltenes

37.7 37.4 43.6 45.7 48.0 67.5 70.4 65.4 62.3 42.2 38.8 49.3 53.5 55.5 36.2 58.0 48.3 24.9 25.2 30.2 51.4

40.4 57.7 36.3 42.2 55.2 61.7 59.4 71.3 68.8 47.7 46.0 57.9 64.2 72.9 29.9 53.6 54.9 26.0 16.6 31.6 41.2

52.3 61.2 54.5 56.7 59.2 72.3 70.8 76.9 65.6 60.0 60.5 68.7 62.0 67.8 69.0 61.9 57.6 23.5 44.1 40.5 61.6

47.2 43.5 42.5 43.1 41.2 44.0 42.3 40.8 42.0 46.1 43.5 45.3 46.1 44.7 40.4 44.4 44.9 52.4 37.9 53.6 49.8

unsystematic scattering and can be evaluated by statistical methods. No correlation can be found for example for the maximum temperature Tm of the DTG. In this case arithmetical means within relatively small limits may be calculated (Table 4-60).

Table 4-60: Peak maximum temperatures T,, mens.


("C) of the DTG curve from distillation bitu-

x ("C)

n

f V (%)

428.3 429.3 447.1 448.6

16 16 16 16

1.91 2.49 0.80 0.88

The temperature range of the peak maximum indicates that crack processes occur. Comparison of the means by t-test gives the following results:

202


Table 4-61: Comparison of the means of T,, from distillation bitumens (Difference with. . . % statistical significance)

Bitumen Dispersion Medium Petroleum Resins

Dispersion Medium

Petroleum Resins

Asphaltenes

none

99.9 99.9

99.9 99.9 none

-

none

-

The blown bitumens sometimes have two peak maxima in the DTG curve, whereas the first maximum is absent in all cases involving asphaltenes. Bitumens, dispersion medium, and petroleum resins show the first maximum in the distillation range around 300 "C. The second maximum appears in the crack (pyrolysis) range. The coefficients of variation are larger (Table 4-62) because the number of samples is small. Table 4-62: Peak maximum temperatures T,, mens.


in the crack range of the DTG curves for blown bitu-

x ("C)

n

+V(%)

430.6 429.2 447.8 465.4

5 5 5 5

2.14 4.43 6.03 5.70

The small number of samples invalidates comparison of the means by t-test. The differences from the corresponding temperatures of the distillation bitumens (Table 4-60) are within experimental limits, except for the values of the asphaltenes. The higher temperature of the peak maximum for the asphaltenes of the blown bitumen indicates that very large molecules or aggregates are formed by the condensation reaction as a consequence of the blowing process. The temperatures of the DTG peak maxima above 400 "C represent thermal cracking processes (pyrolysis). The small coefficients of variation of the individual colloidal components indicate that thermal cleavage of identical or similar chemical bonds has taken place. The resistance of the bitumens to pyrolysis is governed by the properties of the dispersion medium, and this is shown by the uniformity T,, values for bitumen and dispersion medium.

4.3.2.1.1 Correlation of index number from thermogravimetry with consistency data The beginning of evaporation is shifted towards higher temperatures with increasing values of penetration. Fig. 4-38: T1 % = f(Pen) and Fig. 4-39: T5 % = f(Pen) demonstrate


50

0

to. 1

Penetration

250

200

150

100

203

mml

Fig. 4-38 Thennogravimetry of Bitumens Vaporization Start Temperature TI % versus Needle Pene&ation 400

1

0

+

0 Y

0

Y

Y

Y

+

+z Y

0

50

I00

150

Penetration I O . i mml Fig. 4-39 Thermogravimetry of Bitumens Vaporization Start Temperature T.5 % versus Needle Penetration

200

250

204


that the data of the distillation bitumens can be fitted to two straight lines whereas the values of the blown bitumens are grouped together well away from the lines. The laboratory blown product (sample I) does conform to the line. The distillable portion of the samples at normal pressure, AG400, increases due to the increase of the penetration (Fig. 4-40) AG400 = f(Pen). The data of the samples from one refinery fit a straight line very well whereas the data from the other refinery form a curve passing through a minimum. The values of the blown bitumen once again form a separate group. No clear correlation of the residues R600 or R800 with the penetration could be found. The peak maximum temperatures of the DTG, representing the rate of the cracking reaction, decrease with increasing penetration (Fig. 4-41) DTG = f(Pen). These data for the samples of distillation bitumens fit a straight line fairy well but the values from the blown bitumens lie separately. The value of the distillation bitumen €310 (sample 16) lies on the line of the blown bitumens indicating that the product must be semiblown. The index numbers T1 % and T5 % of the distillation bitumen can be correlated with the softening point (Fig. 4-42; T5 % = f(S.P. R&B). The curves of T1 % and T5 % versus S.P. R&B have very steep slopes for the distillation bitumens. The values for the blown bitumens are scattered, away from these curves. The values for the laboratory blown product (sample I) and for the bitumen €310 (sample 16) lie between the curves of the

70 60

50

A

40

69 I

c, 3 v

30

0 0

d 20

a

0 10

0

0

50

I00

150

Penetration (0.1 mm) Fig. 4-40 Thermogravimetry of Bitumens Distillable Part AG400 versus Needle Penetration

200

250


140

, -

i 20

-

,0100

-

4

1

x

-

80

S

.I-

E

2

=O

v

40

n 20 0

205

=-0 0

+\

-

: \

Y

+ Y

; \*

.

-

y

.

I

.

z

1

’

”

’

I

’

.

’

.

l

’

.

’

.

-

.

Fig. 4-41 Thermogravimetry of Bitumens Reaction Rate DTG versus Needle Penetration

* x

*

* u

20

40

60

00

SP R&B

100

120

(“C)

Fig. 4-42 Thermogravimetry of Bitumens Vaporization Start Temperature T5 % versus Softening Point Ring & Ball

140

206


distillation and the blown bitumens, giving further indication that they may be semiblown. In the plot of weight loss up to 400 OC, AG400, versus the softening point (Fig. 4-43), the line for the distillation bitumens also has a steep slope. The values of the laboratory blown product (sample I) and of the bitumen B10 (sample 16) once more lie between the line of the distillation bitumens and the widely scattered blown bitumens. The residues R600 and R800 do not correlate with the softening point nor with the penetration, and therefore no figure is provided. The maxima of the weight loss rate also rises rapidly against the softening point DTG = f(SP R&B). The values of the samples 16 and I are again situated between the distillation and the blown bitumens (Fig. 4-44). The residue which may still be cracked, CR = 100 - (AG400 + RSOO), amounts to 25-70 wt% for the bitumens, 26-73 wt% for the dispersion medium, 24-62 wt% for the petroleum resins, and 38-54 wt% for the asphaltenes. A correlation of the CR with the consistency data may be seen for the bitumens and the dispersion medium; the CR decreases with increasing penetration and rises with increasing softening point (Fig. 4-45 to Fig. 4-48). The petroleum resins only show a correlation of the CR with the softening point, whereas no correlation is seen for the asphaltenes. In a statistical evaluation, decreasing coefficients of variation are indicative of a to more homogeneity substance (Table 4-63). The dispersion medium have the highest contents of distillable fractions (AG400) whereas asphaltenes supply the highest coke residue (R800).

70

*

60

501

h: t-

*

*

* * X

\

20

40

60

80

100

SP RP.5 (“C) Fig. 4-43 Thermogravimetry of Bitumens Distillable Part AG400 versus Softening Point Ring & Ball

120

140


207

X

*

*

20

60

40

80

100

120

("C)

SP R&B

Fig. 4-44 Thermogravimetry of Bitumens Reaction Rate DTG versus Softening Point Ring & Ball

0

* *

+ Y

*

* * 20

I

0

'

,

,

'

I

50

.

' .

,

I

.

,

100

Penetration

,

.

f

1-

I

.

150 [O. 1 mml

Fig. 4-45 Thermogravimetry of Bitumens Crackable Residue CR versus Needle Penetration

'

.

200

250

140

208


8o B i t u m e n

1 70

-

60

-

X

h

$8

I

250

*

*

v

g40

* * 60

40

80 SP R&B ( " C )

120

100

Fig. 4-46 Thermogravimetry of Bitumens Crackable Residue CR versus Softening Point Ring & Ball

* 20

z

.

~

~

.

r

~

~

~

'

r

"

'

'

!

'

~

~

'


209

80

70 60 h

T o P

-

0

Y

-

5 50 bp c, I

Y

a: 0

40

90

20

* x

.

.

I

*

i

zj!

-

*

* .

.

I

.

.

.

.

,

.

.

.

,

'

.

.

,

'

Fig. 4-48 Thermogravimetry of Bitumens Crackable Residue CR of the Dispersion Medium versus Softening Point Ring & Ball of Bitumen

0

I

1

"

'

I

~

1.2

"

I

.

--,

1.4

. 1.6

.

,

,

,

,

1.8

H/C

Fig. 4-49 Thermogravimetry of Bitumens Residue at 600 "C R600 (Mean Value) versus Atomic Ratio H/C x Distillation Bitumens o Blown Bitumens

. 2

2 10


Consequently the mean of the CR shows its highest value in the case of petroleum resins. Table 4-63 shows very high coefficients of variation, showing that the samples have very little homogeneity. The exception is the asphaltenes, which is surprising, bearing in mind that the bitumens supplying those asphaltenes, are of very different origin. As there were only a small number of samples of blown bitumen, no attempt was made to correlate index number with consistency data. Table 4-63: Mean X of the CR for distillation bitumens

X (wt%)

n

i V (%)

50.71 54.08 63.71 43.57

16 16 16 16

22.54 23.49 10.66 4.59

Bitumen Dispersion medium Petroleum resins Asphal tenes

Since the residues at 600 "C and 800 "C do not show any correlation with the consistency data, a statistical evaluation was perfornied (Tables 4-64 and 4-65): Table 4-64: Means X of the coke residue for distillation bitumens X

R600 n

(wt%) Bitumen Dispersion medium Petroleum resins Asphaltenes

21.42 9.73 17.64 52.46

16 16 16 16

+V

-

X

(%)

(wt%)

24.52 24.89 19.73 6.56

16.54 6.74 15.46 48.97

R800 n

+V

("/.I 16 16 16 16

15.69 15.02 11.02 6.64

These coefficients of variation further support the relative homogeneity of the asphaltenes, whereas the dispersion medium and the petroleum resins are heterogeneous, and this is the source of the inconsistent values of the bitumens themselves. The coefficients of variation for the blown bitumens are even larger, but it is impossible to decide whether that is due to the heterogeneous nature of the samples or to the small number of samples (Table 4-65): Table 4-65: Means X of the coke residue for blown bitumens X

R600 n

(wt%) Bitumen Dispersion medium Petroleum resins Asphaltenes

22.02 8.70 12.34 52.4

5 5 5 5

+V

-

X

(%)

(wt%)

8.94 26.77 31.98 8.30

15.74 3.90 7.88 43.28

R8OO n

i V

(%I 4 5 5 5

30.08 48.85 38.86 14.28


21 1

Nevertheless there is a trend for the coke residue to rise from the dispersion medium towards the asphaltenes, and this is related to the atomic ratio H:C. Fig. 4-49 shows the plot of the means of R600 versus the means of H:C. The means of R800 versus the means of H:G give a similar plot. The fact that the formation of coke is not complete at 600 “C in every case, has already been proved by other means. This is confirmed by comparing the data of R600 and R800, with the mean of the colloid component, but this calls in question to a certain extent the importance of the values of the Conradson coke residue.

4.3.2.1.2 Correlation of index numbers with analysis data The term ‘average molecular weight’ is not entirely accurate: it would be more correct to use the term ‘average relative particle mass’ since the vapour pressure osmometry method does not distinguish between molecules and molecular aggregates in solution. The average molecular weight @ is the deciding factor for the start temperature of the evaporation, and for the fraction of the bitumens and their dispersion medium which can still be separated by distillation (AG400 in wt%) (Fig. 4-50 to 4-53). Correlation of the start temperatures of the evaporation, T5 %, for the bitumens and their dispersion medium, with the data representing the molecular mass or size can be expressed by an exponential function. The correlation is not exact for the petroleum resins, and even less for the asphaltenes. The data for most of the blown bitumens do not fit the function as closely as the values for the distillation bitumens.

400

Bitumen

X

350

X

I

U I

ae 300 m

+ 250

y*

L

*

200

I 700

800

900

1000

1100

1200


Fig. 4-50 Thermogravimetry of Bitumens Vaporization Start Temperature T5 % versus Mean Molecular Weight of Bitumens x Distillation Bitumens * Blown Bitumens

2 12


aoo goo 1000 1100 Molecular Weight (Mean)

700

1200

Fig. 4-51 Thermogravimetry of Bitumens Distillable Part AG400 versus Mean Molecular Weight x Distillation Bitumens * Blown Bitumens Y

4

340

X X

-

320

X

x X

300

-"

-

..

0280

-

x ,260

-.

+ 240

*

200 400

. 500

600

,7-,---.---?-

700

BOO

900

1000


Fig. 4-52 Thermogravimetry of Bitumens Vaporization Start Temperature T5 % versus Mean Molecular Weight of the Dispersion Medium x Distillation Bitumens * Blown Bitumens

For the distillable fraction AG400 there is only a weak correlation with the molecular weight 2,but the correlation is good for the dispersion medium and the petroleum resins, which are the main components of the distillable fraction. The data of these components from blown bitumens also show a small scattering around the regression curve.


213

X X X

I . . . . ~ . " ' ~ " " ~ ' . . ' " ' " " " ' ~ ' ' 400

500

600

700

800

I000

900


Fig.4-53 Thermogravimetry of Bitumens Distillable Part AG400 versus Mean Molecular Weight of the Dispersion Medium x Distillation Bitumens

*

Blown Bitumens

The total colloids (petroleum resins and asphaltenes) can be correlated quite well with the coke residues R600 and R800 (Fig. 4-54 and 4-55). There is no difference between the distillation and the blown bitumens. Correlation of the maxima of the reaction rate DTG with the colloid content gives two straight lines for the distillation bitumens (independent of the origin of the samples) and a third line for the blown bitumens (Fig. 4-56). No other correlations with the colloid content were found, nor were any expected.

28

bp

Bitumen

26

I

+J

Y

24

I

=:

22

W

DI-

20

16

1 + P , --

20

25

30

35

40

45

50

Colloids ( W t - % )

Fig. 4-54 Thennogravimetry of Bitumens Residue at 600 "C R600 versus Total Colloids in Bitumens (Symbols see Table 4-46)

2 14


0

-

120

-

100

-

80

-

I 0 -4

X

. S

'E 60 be

u

a + 40

-

20

-

n

15

+ ____II_ 0

Y

*

20

25

30

Colloids

35 40 (kit-%)

45

50

Fig. 4-56 Thermogravimetry of Bitumens Reaction Rate DTG versus Total Colloids in Bitumens (Symbols see Table 4-46)

The supposition that the concentration of the dispersion medium determines both the start temperature of the evaporation (Tl % and T5 %) and the quantity of evaporation loss up to 400 "C (AG400) has been confirmed (Fig. 4-57 and 4-58). A distinct difference between distillation and blown bitumens is evident.


215

Bitumen 360

+ +

340 320

-,y

300

-

280

2 260 k-

240 220 200

50

Jj:... 60 70 D i s p e r s i o n Medium

ao

90

(Wt-%)

Fig. 4-57 Thermogravimetry of Bitumens Vaporization Start Temperature 7'5 % versus Concentration of Dispersion Medium x Distillation Bitumens * Blown Bitumens

Bitumen

60

-:

-50 -0

I

+ 3

1

.

40

0

.

-

'

$30

u

.

220

-

d

.

a l -

a

10

.

-

0 l " " I " " ~ " " ~ ' . 50 55 60

T_p__r__c( 65

D i s p e r s i o n Medium

70

75

80

(Wt-%)

Fig. 4-58 Thermogravimetry of Bitumens Distillable Part AG400 versus Concentration of Dispersion Medium x Distillation Bitumens * Blown Bitumens

The maxima of the reaction rate can also be correlated with the concentration of the petroleum resins (Fig. 4-59), and the difference between the distillation and the blown bitumens is repeated. Further correlation with the content of petroleum resins was not been found.

2 16


0

0

5

10

15

Petroleum Resins

25

20

(Wt-%)

Fig. 4-59 Thermogravimetry of Bitumens Reaction Rate DTG versus Concentration of Petroleum, Resins (Symbols see Table 4-46)

The asphaltenes are responsible for the formation of coke residue due to their high content of condensed aromatic ring systems. This is shown by the correlation of the coke residues R600 or R800 with the contents of asphaltenes (Fig. 4-60). Independent of the origin of the samples, two straight lines result for the distillation bitumens, whereas the data of the blown bitumens fit a third line with a steeper slope. Correlation of the maxima of the reaction rate DTG upon the concentration of asphaltenes presents a similar picture (Fig. 4-61).

r

25 Bitumen

* 5

--,-.7---.

. . ,

.

.

,

.

Fig. 4-60 Thermogravimetry of Bitumens Residue at 800 "C R800 versus Concentration of Asphaltenes (Symbols see Table 4-46)


10

15

20

25

30

35

217

40

Asphal tenes (Wt-%)

Fig. 4-61 Thermogravimetry of Bitumens Reaction Rate DTG versus Concentration of Asphaltenes (Symbols see Table 4-46)

The temperature which marks the beginning of thermal cracking (pyrolysis) must be known for the production and manufacture of bitumens. This temperature may be determined according to the method of Z. Adony [4-141 (see chapter 4.2.7). The results of such an evaluation are given in Table 4-66. These values, TB,do not correlate with analysis or consistency data, so statistical evaluation was used (Tables 4-67 and 4-68). Since the mean temperatures for the blown bitumens are somewhat doubtful, because the number of samples was small, they were not compared with the corresponding mean temperatures for the distillation bitumens, nor with each other. Comparison of the means for the distillation bitumens shows that the crack temperature of the bitumens is the same as that of the petroleum resins. The crack temperatures of the dispersion medium are lower, and those of the asphaltenes are higher. The supposition that the weight loss up to 400 "C (AG400) represents the upper limit for the distillable fraction at atmospheric pressure has been confirmed by these experiments.

4.3.2.2 Thermogravimetry in air Thermogravimetry in air was carried out with the same heating and gas flow rates as for the experiments in argon. Tables 4-69 to 4-72 give the results for distillation and blown bitumens as well as for some dispersion media, petroleum resins, and asphaltenes. Up to a temperature of approximately 250 "C the TGA curves of the experiments in air are the same as those in argon (Fig. 4-62). Even during the experiments in air there is a small weight loss from evaporation, which can start, for some bitumens and their colloidal components, at temperatures below 200 "C. In contrast to the experiments in argon, the

2 18


Table 4-66: Start temperature of the cracking process T, (TG 750) Sample No.

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 I

II 111

IV V

Type of Bitumen

Bitumen

B 200 B 200 B 200 B 80 B 200 B 80 B 65 B 45 B 25 B 80 VR B 200 VR B 130 VR B 80 VR B 65 B 200 B 10 laboratory product 100125 85/40 85/25 100125

400 406 415 390 387 389 374 376 385

Table 4-67: Means

Dispersion medium

Petroleum resins

Asphaltenes

-

403

3891383 384 375 383 375 379 388

393 394 394 393 391 42 1 39613921385 395 393 405 388

4291440 430 411 429 432 408 419 426 41 1 430 439

390

381

388

412

425

395

393

430

392

367

393

407

385 378 371

374 362 391 267 249 254 262

380 391 395 262 272 425 361

424 406 394 449

-

-

387 361/367

-

-

450

X of start temperature TBfor distillation bitumens

x Bitumen Dispersion medium Petroleum resins Asphaltenes

364

(OC)

392.07 379.64 393.68 420.82

n

+V(%)

15 14 17 17

3.92 2.64 2.10 3.28

Table 4-68: Means 5 of start temperatures T, for blown bitumens

X (“C) Bitumen Dispersion medium Petroleum resins Asphaltenes

371.5 258.0 -

n

* V(%) 2.99 3.12 -

449.5

The very low value for the dispersion medium of the blown bitumens indicate that a distillation process has taken place, rather than cracking. The influence of low boiling flux oils is evident.


219

Table 4-69: Thermogravimetry of bitumen in air ~

Sample NO.

Type of Bitumen

T1 %

B 200

198

B 200

T5 %

100 -c

AG

R

200°C

300°C

400 "C

600'C

800°C

DTG

TmX

246

1.o

15.3

35.4

0.4

0.4

254

305

0.1

4.2

29.4

1.6

1.6

B 200

198

248

1.1

15.2

36.1

0.3

0.3

B 80

213

269

0.7

9.5

35.0

0.1

0.1

B 200

235

283

0.2

7.7

30.0

1.0

1 .o

B 80

267

325

0.1

2.7

19.3

1

.o

0.8

B 65

257

324

0.2

2.5

25.0

1.1

0.6

B 45

281

364

1.8

22.9

0.1

0.1

2.33 2.37 2.37 2.10 2.10 3.92 4.15 4 90 8.34 3.39 3.39 3.69 3.91 1.30 5.77 6.51 6.60 2.87 5.38 2.05 3.43 2.48 3.13 4.28 4.78 8.35 1.61 10.02 6.64 3.64 7.66 10.18 2.87 1.68 2.43 2.65 8.16 5.13 8.16 0.62 3.80 1.84 3.00 2.45 4.72 0.78 8.66 4.71 7.70 5.97 10.15 0.32 5.80 5.89 6.96 3.00 6.59 6.96 7.07 6.86 10.71 0.44 7.66 3.17 3.52 5.19 5.63 7.88 9.99

310 331 350 357 379 410 428 460 517 338 350 360 365 383 411 424 430 460 534 352 363 378 402 414 445 512 307 369 385 402 443 501 307 362 371 393 47.4 441 503 325 387 412 473 492 538 305 361 380 407 433 503 308 362 370 378 403

B 25

28 1

360

0.2

- -

1.9

25.0

0.3

0.2

47.4

446 456 462 506 304 364 390 399 420 427 438 509

220


Table 4-69: Thermogravimetry of Bitumen in Air (Continuation) Sample No.

rype of Mumen

T1 %

T5 %

R

AG

100 "C

200 "C

300 "C

$00 "C

600 "C

800 "C

DTG

2.97 1.60 1-89 2.12 2.00 4.40 4.63 4.74 4.46 6.29 2.88 2.70 3.24 3.62 3.91 4.95 5.18 6.04 0.86 9.76 3.60 4.07 6.59 1.29 8.79 7.67 6.55 6.98 8.53 3.04 2.28 3.10 3.74 4.69 5.70 9.06 2.47 1.68 1.78 2.32 2.52 2.76

~

10

B 80

243

290

0.2

6.9

38.0

1.2

1.2

11

VR (B 200)

225

297

0.5

8.9

32.2

1.2

1.1

12

VR (B 130)

274

326

2.5

32.9

0.5

02

VR

269

321

2,7

29.2

1.1

1.o

270

315

3.1

21.2

0.3

0.2

0.2

0.1

13

(B 80)

14

VR (B 65)

B 200

15

186

231

1.8

20.9

37.7

7.11

16

B 10

235

300

0.2

5.0

28.9

0.4

0.2

II

100/25

197

238

0.1

18.6

43.2

0

0

m

85/40

186

224

1.6

25.2

38 5

2.0

2.0

~

__

-

3.55 9.03 3.25 3.67 2.70 2.46 3.37 4.00 4.81 5.05 8.95 1.88 17.22 10.17 10.17 5.38 2.94 3.61 3.43 3.12 3.73 6.31 5.41

331 356 370 380 395 414 428 439 455 541 326 341 362 408 414 436 446 552 320 362 40 1 449 535 321 365 416 437 439 519 339 363 405 413 417 441

540 308 349 370 384 392 396 415 438 507 349 359 384 392 409 418 438 447 529 281 360 384 428 535 279 349 353 360 400 419 503


Sample No.

IV

V

Type of

AG

221

R

bitiirncn

300 "C

400 "C

600 "C

800 "C

DTG

TIMX

85/25

13.7

35.5

0.5

0.5

1.44

287 358 375 380 390 399 433 519 300 37 I 386 419 445 530

13.3

10Ol25

27 7

0.3

0.3

11 41 3.00 2.73 5.85 9.97 3.97 5.95 1.60 5.76 3.54 3.70 16.69 4.93

Table 4-70:Therrnogravimetry of Dispersion Medium in Air Sample No.

Type of Bitumen

l00OC

I 200°C

R

AG

300°C

400 "C

600 "C

800 "C

B 200

19.4

34.0

1.2

1.1

B 80

5.9

24.0

2.0

1.8

B 80

11.1

35.8

1.2

1.2

100/25

26.4

59.9

4.0

0.5

85/40

41.8

65.0

0

0

85/25

31.9

48.8

0.2

0

~

-

DTG -

3.27 8.23 3.82 3.38 4.91 4.36 1.03 2.84 2.39 2.72 2.07 3.15 3.75 5.25 4.88 4.88 4.34 2.31 1.82 1.89 2.10 2.10 4.78 1.53 1.71 1.24 1.89 2.54 2.71 4,19 2.26 2.04 2.92 2.20 3.03 3.08 3.20

TmaX

315 363 400 420 445 555 340 383 419 477 546 331 312 414 441 552 320 400 423 430 471 556 294 344 363 377 385 450 520 303 352 373 391 413 440 450 518

222


Table 4-71: Thermogravimetry of Petroleum Resins in Air. ~

~

T1 %

Sample No.

R

AG

T5 % 100 "C

~

200 "C

-

300 "C

400 'C ~

600 "C

__

800 "C

DTG

1.09 2.17 2.40 2.57 5.43 6.69 6.63 8.40 8.17 7 77 0.43 1.08 1.60 2.07 2 59 2.20 1.68 146 1.20 1.39 2.03 2.28 3.29 5.13 5.64 4.88 5.07 5.31 2.41 2.90 2.69 4.76 2 56 5 00 3.25 3.31 5 63 3.13 2.09 5 19 2.36 3.14 5.25 2.62

Tmax _ __ ___ _

B 200

145

210

4.0

13 9

24.7

4.1

4.1

B 80

236

314

0.7

3.7

14.2

1.1

1.o

B 80

232

315

0.6

3.5

13.7

1 20

0.95

100/25

169

204

4.0

44.3

63.2

1.0

0.7

85/40

188

238

1.7

18.9

38.4

4.5

1.7

85/25

203

256

1.0

12.9

36.2

5.2

2.0

-

- __

Table 4-72: Thermogravimetry of Asphaltenes in Air

5.31 4.39 9.74 9.58

287 357 377 402 427 4.38 452 462 467 526 331 383 408 434 445 483 545 362 372 390 404 419 430 446 466 522 570 281 362 419 447 521 305 378 408 428 448 556 325 388 412 429 456 521


0

200

4 00

6 00

223

800

Fig. 4-62 Thermogravimetry of Bitumen and its Colloidal Components in Air (TG 750) Heating Rate p : 10 K/min Atmosphere: Air 25 cm3/min Curve 1: Bitumen B80 Sample 6 Curve 2: Dispersion medium Curve 3: Petroleum resins Curve 4: Asphaltenes

TGA curves in air flatten more or less distinctly at approximately 350 "C as a consequence of the formation of non-volatile oxidation products. This levelling of the curve does not occur in the case of asphaltenes. In a temperature range from 400 to 550 "C combustion takes place marked by a near lineal decrease of the TGA curve. In this range consecutive and/or simultaneous cracking and oxidation processes may take place with the formation of highly volatile and solid intermediates as well as of combustion products such as CO,, SO,, NO,. Above 550 "C only ash remains. The DTG curves have up to ten maxima, in the temperature range from 280 to 550 'C, except for the asphaltenes, which usually only have two maxima between 450 and 550 "C. The start temperature of the oxidation reaction, marked by the temperature of the first maximum of the DTG curve, and the end of the reaction (last maximum) may be elucidated using statistics (Tables 4-73 to 4-76). The peak heights, and therefore the DTG values, increase with increasing temperature. The peaks of the distillation and blown bitumens, and of the petroleum resins, occur most often at temperatures of approximately 360 OC, 380 'C, 410 OC, and 440 "C. These are due to the reaction of compounds of different susceptibility to oxidation and to pyrolysis, which are present in the bitumens and their components.

224


Table 4-73: Oxidation start (TmJ -

Substance

n

X

V

[*%I

["CI

323.4 292.0 317.2 315.2 469.4

Distillation bitumen Blown bitumen Dispersion Medium Petroleum resins Asphaltenes

16 5 6 6 6

4.87 4.90 5.39 9.61 4.66

Table 4-74: Comparison of the means of oxidation start temperatures (% statistical difference significancy)

Distillation bitumen Dispersion Medium Petroleum resins

Blown Bitumen

Dispersion Medium

Petroleum Resins

Asphaltenes

99.9

none

-

-

none none

> 99.9 > 99.9 > 99.9

-

-

~~

Table 4-75: End of oxidation (TmJ -

n

X

V

[+%I

["CI 521.6 533.2 541.2 531.8 531.2

Distillation bitumen Blown bitumen Dispersion Medium Petroleum resins Asphaltenes

16 5 6 6 6

3.19 4.22 3.24 2.82 2.94

Table 4-76: Comparison of the means of oxidation end temperatures (% statistical differences significancy

Distillation bitumen Dispersion medium Petroleum resins

Blown bitumen

Dispersion medium

Petroleum resins

Asphaltenes

90.0

99.0

-

-

none none

none none none

-

The asphaltenes do not contain any compounds which react below 450 "C. Correlation of the oxidation behavior with the chemical composition, for example the aromacity, was unsuccessful. Any investigation as to whether the temperature of the oxidation peaks might be coordinated to the chemical nature of these substances, should first be tried using pure chemicals.


225

In principle the start of the oxidation reaction ought to be recognized by comparing the TGA curves of corresponding experiments in air and in argon. It is now known that measurement of the oxidation process using DSC will supply more precise results. The residues at 600 "C and 800 "C are very small and relatively constant (Tables 4-69 to 4-72). They represent the inorganic part of the samples. The sample weight for thermogravimetry is only 5 mg and the experimental error is so great that exact determination of the ash content by thermogravimetry is not valid, especially for substances with ash content below 5 wt%.

4.3.3 Isothermal aging tests by thermogravimetry Aging behavior is usually simulated by measuring the weight loss after thermal treatment in a rotating flask according to DIN 25 017 (similar to ASTM D 2872-85). The test temperature is 165 "C, duration 150 minutes. During the test, air passes through the flask with a flow rate of 500 cm3/min. For comparison the test is often repeated using inert gas instead of air, in order to decide whether the weight loss is a consequence of evaporation or of oxidation. The present results of thermogravimetry indicate that 165 "C is still too low a temperature to initiate an oxidation process. Therefore isothermal aging tests were carried out for 120 minutes at 250 "C both in argon and in air. Samples of approximately 5 mg weight were Table 4-77:Isothermal aging of bitumen in argon (25cm3/min). T(const.) = 250 "C weight loss [%I Sample No. Type 1 2 5 6 7 8 9 10 11 12 13 14 15 I I1

III IV V

B 200 B200 B200 B 80 B 65 B 45 B 25 B 80 VR(B 200) VR(B 130) VR(B 80) VR(B 65) B200 laboratory product 100/25 85/40 85/25 100125

1

5

10

15

30

60

90

120

3.7 1.7 0.5 0.8 0.5 0.6 0.5 1.6 1.3 0.7 0.8 0.7 5.9 0.9 5.5 5.7 3.9 6.4

7.2 2.7 1.5 1.4 0.9 0.9 0.9 3.0 2.7 1.1 1.5 1.2 11.1 1.8 11.7 12.1 8.3 10.6

9.9 3.5 2.4 2.0 1.2 1.2 1.2 4.7 4.1 1.6 2.2 1.4 15.0 2.6 16.7 17.1 11.9 13.5

11.8 4.1 3.2 2.6 1.5 1.5 1.5 5.9 5.2 2.0 2.8 1.7 17.7 3.2 20.0 20.5 14.5 15.2

15.4 5.6 4.8 3.9 2.0 2.3 2.3 9.0 7.7 3.0 4.3 2.3 22.2 5.0 26.0 27.0 19.3 18.0

19.0 7.5 7.1 5.8 3.0 3.1 3.5 12.8 10.8 4.9 6.4 3.3 26.8 7.7 31.4 33.3 24.3 21.2

22.0 9.2 8.3 6.9 4.0 3.7 4.2 15.3 12.5 6.0 8.0 4.3 29.0 9.5 34.3 36.7 27.2 23.0

23.7 10.5 9.0 8.2 5.0 4.2 5.0 17.0 14.0 7.0 9.2 5.2 30.8 10.8 36.1 38.8 29.2 24.5

226


Table 4-78: Isothermal aging in air (25 cm3/min). T = 250 "C weight loss [%] Sample No.

1 2 5 6 7 8 9 10 11 12 13 14 15 I 11 I11 IV V

1

5

10

B 200 B 200 B 200 B 80 B 65 B 45 B 25 B 80 V R(B 200) V R(B 130) VR(B 80) VR(B 65) B 200 laboratory product 100125 85/40 85/25 100/25

15

30

60

90

120

15.2 5.5 6.6 4.4 3.0 2.2 4.0 6.7 7.1 3.7 5.3 2.7 19.4 5 20.7 21.9 14.7 16.8

17.9 8.1 8.3 6.8 4.7 3.1 5.0 9.3 10.2 6.7 7,7 4.7 22.8 7 22.0 22.8 16.2 19.3

18.2 10.0 8.5 7.3 5.2 3.2 5.4 9.7 10.7 7.9 8.0 5.3 23.1 7.1 22.0 23.1 16.5 19.5

18.5 11.0 8.7 7.4 5.5 3.4 5.5 9.8 11.0 8.2 8.1 5.5 23.3 7.2 22.0 23.1 16.5 19.5

[mini

Type

3.0 1.5 0.9 0.4 0.4 0.6 0.7 1.1 0.7 0.4 0.6 0.2 4.0 0.7 -

4.8

6.0 2.3 2.0 1.1 0.8 0.8 1.3 2.3 1.9 0.8 1.3 0.5 8.8 1.5 2.9 2.5 1.5 8.8

8.7 3.1 3.0 1.8 1.3 1.1 1.9 3.4 3.2 1.4 2.2 0.8 12.5 2.5 9.9 9.0 6.3 11.6

10.7 3.8 4.1 2.4 1.7 1.2 2.5 4.3 4.5 2.0 3.0 1.3 14.9 3.2 14.8 13.7 9.7 13.5

heated up to 250 "C constant temperature by applying a heating rate of 100 K/min, and maintaining this temperature for 120 minutes. The results of the tests in air and argon are shown in Tables 4-77 and 4-78. Isothermal aging of a series of distillation bitumens from B200 to B45 (samples 5-8) is given in argon (Fig. 4-63) and in air (Fig. 4-64). A direct comparison of the aging in air and argon of a distillation bitumen B200 (sample 15) is given in Fig. 4-65 and for a blown bitumen 100/25 (sample V), in Fig. 4-66. It is evident that the aging behavior of the distillation bitumens is not uniform. Some of them undergo higher weight loss in argon than in air (for example: sample 15, bitumen B200, in Fig. 4-65) whereas others exhibit higher weight loss in air (for example bitumen B65, sample 7) (Tigs 4-63 and 4-64) as does the blown bitumen (Fig. 4-66). A quotient Q = weight loss in air divided by weight loss in argon signifies: Q < 1.0 loss in argon higher than in air Q = 1.0 equal losses in air and in argon Q > 1.0 loss in air higher than in argon

Table 4-79 shows that the quotient Q generally rises with increasing duration of test, passes through a maximum and decreases thereafter. There is a relationship with the consistency, as shown in Fig. 4-67 for the distillation bitumens B200, B80, B65, and B45.


0

20

40

60

80

100

120

140

Time ( m i n )

Fig. 4-63 Isothermal Thermogravimetry at 250 "C (TG 750) Atmosphere: Argon 25 cm3/min Sample 5 Distillation Bitumen B200 Sample 6 Distillation Bitumen B80 Sample 7 Distillation Bitumen B65 Sample 8 Distillation Bitumen B45

10

,

-x

0

20

40

60

80

100

120

Time ( m i n )

Fig. 4-64 Isothermal Thermogravimetry at 250 "C (TG 750) Atmosphere: Air 25 cm3/min Sample 5 Distillation Bitumen B200 Sample 6 Distillation Bitumen B80 Sample 7 Distillation Bitumen B65 Sample 8 Distillation Bitumen B45

140

227

228

-

4 Themoanalytical Investigations on Petroleum and Petroleum Products

30

M I 42

x 20 m YI

0 4

10

0 0

20

40

60

100

80

120

140

Time ( m i n )

Fig. 4-65 Isothermal Thermogravimetry at 250 "C (TG 750) Distillation Bitumen B200 Sample 5 Gas Flow Rate: 25 cm3/min Curve 1: Argon Curve 2: Air

30

25

'+

i

/ +

' /

-J

---

?O

a

0

M

x

15

m VI

10

2

5

0

L

,

0

20

,

.

I

40

.

.

1

I

60

80

100

120

Time (rninl

Fig. 4-66 Isothermal Thermogravimetry at 250 "C (TG 750) Blown Bitumen 100/25 Sample V Gas Flow Rate: 25 cm3/min Curve 1: Argon Curve 2: Air

140

4.3 Investigations on bitumen Table 4-79:

Sample No. 1 2 5 6 7 8 9 10 11 12 13 14 15 I I1 III IV V

Quotient Q =

Loss in Air Loss in Argon

1

5

10

0.81 0.88 1.80 0.50 0.80 1.oo 1.40 0.69 0.54 0.57 0.75 0.29 0.68 0.78

0.83 0.86 1.33 0.79 0.89 0.89 1.44 0.77 0.70 0.73 0.87 0.42 0.72 0.83 0.25 0.21 0.18 0.83

0.71 0.89 1.20 0.90 1.08 0.92 1.58 0.92 0.78 0.88 1.00 0.57 0.83 0.96 0.59 0.53 0.53 0.86

Type B 200 B 200 B 200 B 80 B 65 B 45 B 25 B 80 V R(B 200) VR(B 130) VR(B 80) VR(B 65) B 200 laboratory product 100125 85/40 85/25 100125

-

0.75

15 30 [min] 0.91 0.93 1.28 0.92 1.13 0.80 1.67 0.73 0.87 1.00 1.07 0.76 0.84 1.00 0.74 0.67 0.67 0.89

2

YI w _I 0

-

1

L .r

4

.-

S

0.5

w

w

0

_I

" U

229

0

Fig. 4-67 Isothermal Thermogravimetry at 250 "C (TG 750) Quotient Q (Loss in Airboss in Argon) versus Time Sample 5 Distillation Bitumen B200 Sample 6 Distillation Bitumen B80 Sample 7 Distillation Bitumen B65 Sample 9 Distillation Bitumen B25

0.99 0.98 1.38 1.13 1.50 0.95 1.74 0.74 0.92 1.23 1.23 1.17 0.87 1.00 0.80 0.81 0.76 0.93

60

90

120

0.94 1.08 1.17 1.17 1.57 1.00 1.43 0.73 0.94 1.37 1.20 1.42 0.85 0.91 0.70 0.68 0.67 0.91

0.83 1.09 1.02 1.06 1.30 0.86 1.29 0.61 0.86 1.32 1.oo 1.23 0.80 0.75 0.64 0.63 0.61 0.85

0.78 1.05 0.97 0.90 1.10 0.81 1.10 0.58 0.58 1.17 0.90 1.06 0.76 0.67 0.61 0.60 0.57 0.77

230

4 Therrnoanalytical Investigcitions on Petroleum and Petroleum Products

The influence of the concentration of the dispersion medium or the maltenes and their average relative particle mass is seen. If we define this influence using factors:

a

P , = (concentration of dispersion medium) . p2 = (concentration of petroleum resins) . G, the plot of the maximum value of Q (Q,,) versus P , or versus ( P , + P2) shows a degree of correlation (Fig. 4-68 and 4-69). The coefficients of correlation are not very high but they do reveal a trend ( r = 0.64 and r = 0.70). Isothermal thermogravimetry has the advantage that the actual weight loss at any time is visible, so it is possible to determine whether the process is still running or is already complete. In the test according to DIN 52 017 a final value is found only after test duration of 150 minutes. Comparison of the tests in air and in argon using the quotient Q shows the passage through a maximum after a certain duration of the test, which cannot be recognized in the test according to DIN 52 017. Nor is it possible to tell from the DIN test whether the weight losses are a result of the evaporation of highly volatile parts of the bitumen or of volatile oxidation products. A practical example for the behavior of a bitumen B80 (sample 6) during manufacture is shown in Fig. 4-70 and 4-71. Fig. 4-70 shows the plot of weight loss versus time at three different isothermal temperatures, for virgin material. Fig. 4-71 shows a similar plot for the bitumen after passage through the asphalt mixing plant (extracted sample). Differences between the virgin and the extracted material are not visible at test temperatures of 200 "C and 250 "C but appear above 300 "C. The plots of weight loss versus temperature with time as parameter (Fig. 4-72 and 4-73) show the differencesmore clearly. The behavior of the virgin material may be described by an exponential function. The behavior of the extracted material is described by a linear function. The small differences between the samples at the test temperatures 200 "C and 250 "C indicate only losses of volatile portions. Even at 300 "C there is no difference between the samples until the test 2 ,

1.6 i.B

g

1.4

i

x

x

L 3

1.2

1 0.8 0.6

1

30

40

50

60

Pi

Fig. 4-68 Maxima of Quotient Q versus Product P1

70

80 ri~3


23 1

duration is 30 minutes. After 60 minutes the difference is 2.5 wt% and after 300 minutes the loss from the virgin material exceeds that of the extracted material by 6.8 wt% (or 30 % relatively). These results do not permit positive identification of the source of the differences between the samples. This may be evaporation of some volatile parts during the mixing process; losses due to the formation of volatile oxidation products; or losses during evaporation of the chloroform extraction fluid. The exponential function AG =AT) for the

x X

30

6a

50

40

Pi

+

70

80

90

"103

P2

Fig. 4-69 Maxima of Quotient Q versus Sum of Products P1 + P2

35 30

2 0

10

5 0

a

50

100

im 200 Time (rnin)

250

300

350

Fig. 4-70 Isothermal Thermogravimetry in Air (TG 750)

Distillation Bitumen B80 Sample 6, Virgin Material Weight Loss versus Time Parameter: Temperature

232


virgin material indicates an oxidation process, whereas the extracted sample may not contain compounds which can be oxidized under the conditions of the TGA experiments. This may explain the linearity of the function AG =f(T) for the extracted material.

25

20

5

0 50

0

100

150

200

250

300

350

Time (min)

Fig. 4-71 Isothermal Thermogravimetry in Air (TG 750) Distillation Bitumen B80 Sample 6, Extracted Sample after Asphalt Mixing Process Weight Loss versus Time Parameter: Temperature

35

-

30

-

25

e

.

I

2 20 3 y1

m 0

.

15

--

10

-

-

A

-

30 '

5 -

0

-

1-

180

200

220

240

260

280

300

Temperature ("C)

Fig. 4-72 Isothermal Thermogravimetry in Air (TG 750) Distillation Bitumen B80 Sample 6, Virgin Material Weight Loss versus Temperature Parameter: Time

320


I

233

300' 240

'

$80 '

120 '

60 '

30 '

1

10 '

-I

. , . . . , . . . , ~ . . , . . . , . . . , . . . , . . . , . . . 160

180

220

200

240

260

280

300

320

Temperature ( " C )

Fig. 4-73 Isothermal Thermogravimetry in Air (TG 750) Distillation Bitumen B80 Sample 6, Extracted Sample after Asphalt Mixing Process Weight Loss versus Temperature

Parameter: Time

4.3.4 Differential scanning calorimetry DSC 4.3.4.1 Test in argon at atmospheric pressure The tests were carried out in argon atmosphere at 1 bar pressure and a flow rate of 5 cm3/min. The sample weight was between 1.5 and 4.0 mg. The maxima of energy consumption in the DSC curves lie at temperatures between 448 and 496 "C.The resulting coefficients of the Arrhenius equation calculated according to ASTM E 698-79 and the conversion factor, U , are shown in Table 4-80. No correlation could be shown for the activation energy E or the frequency factorA with analytical data. Statistical evaluation of the peak maximum temperatures of the distillation bitumens results in means with coefficients of variation 2 V = 1.01 % maximum. When using these means in the calculation, the following average values of the Arrhenius coefficients are found: E = 204.5 (kJ/Mol) log A = 14.362 (min-') When individual data from Table 4-80 are used, the mean for E is insignificantly smaller, but has a significant larger coefficient of variation. There is no significant difference

234


Table 4-80: Arrhenius coefficients and conversion for DSC pyrolysis in 1 bar argon: Distillation bitumen Sample No. Type

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16

B 200 B 200 B 200 B 80 B 200 B 80 B 65 B 45 B 25 B 80 V R (B 200) V R (B 130) V R (B 80) V R (B 65) B 200 B 10

Mean X Coefficient of variation V+ %

E (kJ/Mol)

log A (min-')

u (%I

196.1 178.2 192.0 180.1 225.6 226.0 196.1 219.4 198.8 179.3 211.2 211.7 205.0 176.5 196.3 213.1

13.804 12.512 13.467 12.656 14.888 15.824 13.629 15.293 13.854 12.520 14.830 15.016 14.561 12.345 13.850 15.099

85.70 85.70 85.22 84.58 85.58 83.01 81.25 80.82 79.84 81.40 84.09 83.72 83.47 82.51 85.08 78.94

200.33 8.27

13.970 8.14

83.17 2.60

between the average value of E and the mean from Table 4-80. The mean of the conversion is 83.17 % k V = 2.60 % (relative). The plot of the resulting half life time against the temperature is shown in Fig. 4-74. Extrapolation to lower temperatures shows the considerable thermal stability of the bitumens at manufacturing temperatures, provided that air is excluded. If bitumens are regared as model substances for heavy residues, then their Arrhenius coefficients can serve as the basis for the calculation of the kinetics of pyrolysis reactions in thermal conversion processes (thermal cracking, visbreaking, hydrotreating etc.). Integration of the peak areas gives a value for the energy required for pyrolysis reactions. The peak temperatures are not so uniform for the blown bitumens, so only the individual data are given in Table 4-81.


I

250

3CO

LOO

450

ti 12 [ T i n I

-loo

'

I

235

SO0 550 600zp["c 1

\!

I I

! A l l 1 i

I\!

I I

Fig. 4-74 Pyrolysis of Distillation Bitumen in 1 bar Argon Half Life Time t1,2versus Temperature

Table 4-81: Arrhenius coefficients and argon for blown bitumen. ~

Sample No.

I

II I11 IV V

Type exp. product 100125 Peak I Peak 2 85/40 Peak 1 Peak 2 85/25 Peak 1 Peak 2 100125

DSC

conversions in ~

~

pyrolysis ~

E (W/Mol)

log A (min-')

U (%)

213.6 94.9 238.9 62.2 192.2 78.5 201.8 217.9

15.097 7.902 17.072 5.317 12.458 7.079 14.657 14.660

78.94 81.96 82.29 82.24 80.80

in

1 bar

236


The blown bitumens from the same refinery (samples 11-IV) have two peaks each. The first peak represents an evaporation and the second one a pyrolysis reaction. The dispersion medium of those samples generally shows only one peak in the evaporation range (Table 4-82) whereas the petroleum resins exhibit two peaks (Table 4-83). For the asphaltenes, normal pyrolysis behavior with no evaporation peak was observed (Table 4-84). Table 4-82: Arrhenius coefficients and conversion in the DSC pyrolysis in 1 bar argon. Dispersion medium from blown bitumens Sample No. Type

I I1 III IV V

E (kJ/Mol)

exp. product 100/25

85/40 Peak 1 Peak 2 85/25 100125

log A (min-I)

u (%I

-

-

-

81.1 99.7 138.4 93.1 204.0

6.982 8.341 12.041 7.420 14.637

93.95 96.29 96.15 97.01

Table 4-83: Arrhenius coefficients and conversion in the DSC pyrolysis in 1 bar argon Petroleum resins from blown bitumens Sample No. Type

I I1 111 N V

exp.product

100/25 Peak Peak 85/40 Peak Peak 85/25 Peak Peak 100125

1 2 1 2 1 2

E (kJ/Mol)

log A (min ')

u (%I

-

-

-

74.6 244.3 66.9 227.0 63.7 240.6 198.3

7.616 17.308 5.500 16.282 6.018 17.126 15.361

95.85 88.40 90.7 1 89.35

Table 4-84: Anhenius coefficients and conversion of DSC pyrolysis in 1 bar argon. Asphaltenes from blown bitumens. Sample No.

I 11 III IV V

Type exp.product 100/25 85/40 85/25 100/25

E (kJ/Mol)

log A (min-')

u (%I

-

-

-

263.2 241.8 263.7 255.0

18.589 16.996 19.030 18.278

47.14 48.39 50.69 53.28

The conversion decreases from the dispersion medium down to the asphaltenes (Table 4-85):


237

Table 4-85: Mean X of the conversion U (%) in DSC pyrolysis of blown bitumens.

Blown bitumen Dispersion medium Petroleum resins Asphaltenes

81.26 95.85 91.08 49.88

5 4 4 4

1.74 1.38 3.65 5.42

Some overlapping of the evaporation and pyrolysis reactions was observed, due to the content of low boiling flux oils, especially in the experiments with separated components of the colloid system. In order to avoid such overlapping as much as possible and to simulate the technical conditions of the crack processes, further investigations on bitumens and their colloid components were carried out in a hydrocarbon atmosphere (methane) at 10 bar pressure.

4.3.4.2 Tests in methane at 10 bar pressure Pyrolysis experiments were carried out in methane at 10 bar pressure and 5 cm3/min gas flow rate. All the experiments were run using at least three different heating rates (between 5 and 50 K/min). The sample weights were between 1.0 and 4.0 mg. The individual data of the Arrhenius coefficients for the bitumens and their colloid components are shown in Tables 4-86 to 4-89. Table 4-86: Arrhenius coefficients and conversion for DSC pyrolysis in 10 bar methane Bitumen Sample No. Type

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16


E (kJ/Mol)

log A (min-l)

u (%I

226.2 188.5 192.0 240.4 235.5 242.0 221.8 239.5 238.5 227.3 237.3 229.1 195.8 203.2 208.8 236.7

15.878 13.083 13.390 16,441 16.441 16.885 15.418 16.753 16.756 15.936 16.645 16.231 13.741 14.163 44.649 16.678

84.40 82.95 84.60 83.30 82.91 81.43 79.45 78.44 78.77 80.45 82.48 81.36 81.53 78.70 84.43 77.32

238

4 Thermoanalytical Investigations on Petroleum and Petroleum Products Bitumen

~~

Sample No. Type

I I1 I11 IV V

exp.product 100/25 85/40 85/25 100125

E (kJ/Mol)

log A (min-I)

u (%I

216.0 239.4 210.5 241.8 235.2

15.250 16.567 15.373 17.202 16.522

77.24 80.45 81.15 79.21 79.44

Table 4-87: Arrhenius coefficients and conversion for DSC pyrolysis in 10 bar Methane. Dispersion medium ~

Sample No. 1

2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 I

II 111 IV V

~~

~

Type

B 200 B 200 B 200 B 80 B 200 B 80 B 65 B 45 B 25 B 80 V R (B 200) V R (B 130) V R (B 80) V R (B 65) B 200 B 10 exp.product 100/25 85/40 85/25 100/25

E (kJ/Mol)

log A (min-')

254.2 219.6 204.6 219.1 212.6 224.3 269.7 244.4 25 1.7 247.7 204.1 239.3 256.0 181.8 208.9 209.0 232.6 234.3 125.7 115.0 123.4 225.5

18.030 15.333 14.322 15.320 14.751 15.364 18.592 16.928 17.542 17.425 14.374 16.877 17.963 12.644 14.613 14.462 16.388 16.360 9.405 8.394 8.654 15.879

lJ

(a)

93.46 89.92 95.33 93.47 93.57 91.45 91.67 90.62 93.13 90.15 93.77 89.24 91.31 91.61 94.30 93.20 95.37 95.46 94.26 -

93.75 94.91


Table 4-85: Arrhenius coefficients and conversion for DSC pyrolysis in 10 bar Methane. Petroleum resins Sample No. Type

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 I I1

III 1v

V

B 200 B 200 B 200 B 80 B 200 B 80 B 65 B 45 B 25 B 80 V R (B 200) V R (B 130) V R (B 80) V R (B 65) B 200 B 10 exp.product 100/25 85/40 85/25 100/25

E (kJ/Mol)

log A (min-')

213.2 210.2 211.2 208.2 262.7 297.9 268.4 240.6 233.6 240.5 221.3 223.1 257.0 244.0 224.2 188.5 264.0 235.8 219.9 228.0 240.0

14.879 14.602 14.616 14.290 18.144 20.576 18.650 16.516 16.151 16.751 15.426 15.579 17.961 17.064 15.567 13.214 18.547 17.220 15.667 16.118 16.911

("/.I 81.07 74.31 81.95 87.52 84.91 86.28 8 1.45 84.98 82.23 78.37 84.42 80.15 85.49 83.45 80.54 77.03 89.71 92.97 86.08 87.25 87.25

Table 4-89: Arrhenius coefficients and conversion for DSC pyrolysis in 10 bar Methane. Asphaltenes Sample No. Type

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16


E (H/Mol)

log A (min-')

U (%)

242.0 212.7 224.8 193.9 212.0 249.1 234.0 244.7 221.1 246.2 213.5 220.6 243.5 228.2 214.2 229.3

16.968 14.923 15.508 13.166 14.591 17.304 16.156 16.912 15.365 17.366 15.089 15.422 16.893 15.935 14.804 16.015

48.35 50.45 46.98 46.38 46.50 45.19 44.65 44.97 51.29 47.36 49.51 50.88 51.04 49.20 48-22 54.11

239

240

4 Tlzermoanalyticul Investigations on Petroleum and Petroleum Products

Asphaltenes Sample No. Type I I1 UI IV V

E (kJ/Mol)

log A (min-')

U (%)

238.0 228.1 200.0 213.5 242.2

16.655 16.221 14.296 15.088 17.312

51.23 50.48 46.86 48.97 51.28

exp.product 100/25 85/40 85/25 100/25

Statistical evaluation of the peak maximum temperatures of the distillation bitumens gives very small coefficients of variation, as shown in Table 4-90.

Table 4-90 Maxima of the coefficients of variation for the peak temperatures from the pyrolysis of distillation bitumens in 10 bar methane. +V (%) Bitumen Dispersion medium Petroleum resins Asphaltenes

0.47 0.82 0.61 0.62

The coefficients of variation of the means of the conversions show considerably more scattering (Table 4-91).

Table 4-91: Coefficients of variation for the conversions from the pyrolysis of distillation bitumens in 10 bar methane. + V (%) Bitumen Dispersion medium Petroleum resins Asphaltenes

3.06 1.67 10.74 5.45

The very low variance of the means of the peak maximum temperatures permit their use in calculating the averages of the Arrhenius coefficients (Table 4-92). Again, there are only small differencesbetween the average values of the activation energy E and the means of E from the individual data of the Tables 4-86 to 4-89, with no statistical significance.


241

Table 4-92: Average values of the Arrhenius coefficients and the conversion for the pyrolysis of distillation bitumens in 10 bar methane.


E (kJ/Mol)

log A (min-')

u (%I

222.8 225.5 232.0 226.1

15.638 15.744 16.094 15.709

81.16 92.39 82.58 48.61

There is no statistical difference in the Arrhenius coefficients between the bitumen and the dispersion medium, nor between the petroleum resins and the asphaltenes. The statisti250

300

w)O

450 Ii

500 /

5y

600Gc

l

Fig. 4-75 Pyrolysis of Distillation Bitumen and its Colloidal Components in 10 bar Methane Half Life Time t1,2versus Temperature Line 1: Distillation bitumen Line 2: Dispersion medium Line 3: Petroleum resins Line 4: Asphaltenes

242


cal significance of the difference between the Arrhenius coefficients of the dispersion medium and the asphaltenes is only 95 %. For the other differences the statistical significance is as high as 99.9 %. A comparison of the means of the corresponding peak temperatures shows that there is a significant difference (99.9 %) between the data from the experiments in 1 bar argon and in 10 bar methane, whereas the means of the conversions do not exhibit significant differences. Experimental results confirm the theory that a reaction is rendered more difficult by increasing pressure, which causes an increase of volume due to the formation of gases and vapors (LeChatelier's principle). Differences in the Arrhenius coefficients for the three components of the colloid system (the dispersion medium, petroleum resins, and asphaltenes) do exist, but prove to be smaller than was at first assumed. The straight lines for the distillation bitumens and their colloid components in the plot of half life time, tliz, versus temperature run almost parallel, because the differences in the Arrhenius coefficients are small (Fig. 4-75). Even the corresponding peak temperatures of the blown bitumens show very small variances in the tests in 10 bar methane and also permit the calculation of statistical means. The resulting coefficients of variation are f 3 . 0 % maximum. This is also true for the colloid components, except for the dispersion medium of the two bitumens: 85/40 (sample 111) and 85/25 (sample IV). Here again a weight loss caused by distillation even occurs under pressure with the consequence of low values for the activation energy and frequency factor. Only the data of the other three samples was included in the statistics. The average values of the Arrhenius coefficients calculated in this manner and the means of the conversion are shown in Table 4-93. Table 4-93: Average values of the Arrhenius coefficients and means of the conversion from the pyrolysis of blown bitumens in 10 bar methane.


E (kJ/Mol)

log A (min-l)

u (%I

239.5 205.4" 236.3 222.6

17.104 14.488" 16.818 15.753

79.50 94.74 88.65

49.76

* from the means of the temperatures of 3 samples Comparison of the means shows that there is a significant difference in the Arrhenius coefficients for distillation and blown bitumens and their colloid components. The conversions only differ for the dispersion medium whereas the bitumens, the petroleum resins, and the asphaltenes do not differ in the conversions, either in 1 bar argon or in 10 bar methane. In the plot of half life time, tli2versus temperature, the differences of the activation energies for the blown Bitumens and their colloid components cause a clear divergence of the straight lines from the parallel (Fig. 4-76).


243

Fig. 4-76 Pyrolysis of Blown Bitumen and its Colloidal Components in 10 bar Methane Half Life Time t1,2versus Temperature

Line 1: Line 2: Line 3: Line 4:

Blown bitumen Dispersion medium Petroleum resins Asphaltenes

4.3.4.3 Temperatures of the cracking process The start of the cracking process is marked in the diagram of energy flow versus temperature of the DSC by a change from a nearly linear run, which shows only a small drifting due to the changes in specific heat with increasing temperature, to one with a considerable slope. By constructing the tangents to the induction period and to the curve of the reaction peak, the point of intersection is found, which marks the onset point and its temperature.

244


Table 4-94: Start temperature of the cracking reaction (Tons,,)of bitumen in DSC (p = 10 Klmin). Sample No. Type

Tonset

[“CI in 1 bar argon 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16

I I1 IlI

Iv

V

B 200 B 200 B 200 B 80 B 200 B 80 B 65 B 45 B 25 B 80 V R (B 200) V R (B 130) V R (B 80) V R (B 65) B 200 B 10 exp.product

400 40 1 409 414 405 405 415 415 409 378 413 403 420 390 415 395 375

lOOl25 85/40 85 I25 100125

n.e. 248

n.e. 378

‘onset

t“C1 in 10 bar Methane 400 402.5 -

402 405 406 412.5 410 404 398 406 400 410 405 413 385 385

n.e. n.e. 402 408

n.e.: it was not possible to evaluate this experiment

The temperature of the corresponding maximum of energy consumption in the DSC curve can be used as a second index number for the cracking reaction. DSC experiments with a heating rate p= 10 K/min are used, so that the results may be compared with thermogravimetry. The individual data of the onset temperature for pyrolysis both in 1 bar argon and in 10 bar methane are shown in table 4-94. Table 4-95 gives the onset temperatures of the bitumens and their colloid components from the reaction in 10 bar methane. The small variations of the individual data permit the calculation of statistical means (Table 4-96). There is no correlation with analytical data. Comparison of the means shows a significant difference (99.0 %) between the distillation bitumens and their petroleum resins, and between the dispersion medium and the petroleum resins. The blown bitumen sample is too small to examine correlation. There is no significant difference between the onset temperatures of the distillation and blown bitumens.


245

Table 4-95: Start temperature of the cracking reaction (Tonset)of DSC pyrolysis in 10 bar methane (p= 10 K/min). Sample No.

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 I I1 I11 IV V

5Pe

Bitumen

Dispersion medium

Petroleum resins

Asphaltenes

B 200 B 200

400 402.5

B 200 B 80 B 200 B 80 B 65 B 45 B 25 B 80 V R (B 200) V R (B 130) V R (B 80) V R (B 65) B 200 B 10 Exp. product 100/25 85/40 85/25 100/25

402 405 406 412.5 410 404 398 406 400 410 405 413 385 385 n.e. n.e. 402 408

415 406 397.5 405 399 402.5 415 400 398 385 393 402 395 392 407.5 395 394 385 n.e. 390 408

398 410 408 412.5 410 41 1 400 407.5 417.5 405 411 410 402.5 408 409 403 402 n.e. 400 400 409

402 40 1 402 387 410 398 398 400 398 400 417.5 404 418 403 398 398 410 360 368.75 383 378

-

n.e.: it was not possible to evaluate this experiment

Table 4-96: Means of the start temperature of the cracking reaction Tonsetin the DSC (p= 10 Klmin). ~

1 Destillation bitumen 2 Distillation bitumen 3 Dispersion medium from no.2 4 Petroleum resins from no.2 5 Asphaltenes from no.2 6 Blown bitumen

~~

~

-

n

f V

Reaction conditions

(“C)

1 bar argon

403.65

16

3.20

10 bar methane 10 bar methane

402.94

16

1.93

400.09

16

1.99

10 bar methane

407.35

16

1.23


402.62

16

1.90

405.33

4

0.75

X

(%)

246


Reaction conditions 7 Dispersion medium from no.6 8 Petroleum resins from no.6 9 Asphaltenes from no.6

-

X

n

("C)

f V (%)

10 bar methane

394.25

4

2.5 1

10 bar methane

405.00

4

1.44

10 bar methane

379.95

5

4.99

Comparison of the means of the onset temperatures of the distillation bitumens in 1 bar argon with the the means of the start temperature TBfrom thermogravimetry (Tables 4-67 and 4-68) shows that the difference has only a low significance of 95 %. The maximum of heat flow during the reaction is marked by the corresponding peak temperature (Table 4-97)

Table 4-97: Means of the peak temperatures from the DSC cracking experiments (p= 10 K/min)

1 Distillation bitumen 2 Distillation bitumen 3 Dispersion medium from no.2 4 Petroleum resins from no.2 5 Asphalteues from no.2 6 Blown bitumen 7 Dispersion medium from no.6 8 Petroleum resins from no.6 9 Asphaltenes from no.6

X

Reaction conditions

("C)

1 bar argon

453.96

16

1.01


459.95

16

1.36

46 1.49

16

0.82

10 bar methane

468.36

16

0.61

10 bar methane 10 bar methane 10 bar methane

465.51

16

0.55

447.50

5

2.61

442.43

4

6.28

10 bar methane

450.76

5

2.99

10 bar methane

452.09

5

1.62

n

f V

(%I


247

Comparison of the means of Table 4-97 gives the following results: Comparison

No. 1 No.2 No. 1 No.3 No.3 No.4 N0.7 No.7 No.8 No.3 No.4 No.5

Significant difference

No.2 No.6 No.6 No.4 No.5 No.5 No.8 No.9 No.9 No.7 No.8 No.9

Comparison of the means of the DSC peak temperatures (Table 4-97) with the means of the DTG maximum temperatures Tmx(Tables 4-60 and 4-62) for the distillation bitumens shows considerably higher temperatures of the DSC peaks even at atmospheric pressure. There is no reasonable explanation for this phenomenon. The higher values of the DSC peak temperatures at increased pressures correspond to the theoretical expectations.

4.3.4.4 Oxidation in air Investigations on the oxidation behavior of bitumen using DSC give precise results [4-23 to 4-27]. The experiments are preferably carried out in air, since the reaction in oxygen runs so fast that the actual sample temperature increases faster than the pre-set heating rate of the instrument. Consequently the temperature of the sample overtakes the temperature of the instrument, and a loop will be plotted in the diagram instead of a measurable peak. The same phenomenon occurs when the sample mass is too large. Even so, the correct peak maximum temperature may be determined by plotting the heat flow versus time, marking the peak maximum and rescaling to heat flow versus temperature. Increased pressure should be applied if the evaporation and the oxidation processes compete within the same temperature range. The application of increased air pressure will increase the boiling temperatures and decrease the oxidation start temperature, due to the increased oxygen partial pressure. The international standard of pressure for DTA and DSC oxidation tests on lubricants is 100 psi (7 bar), and we used this, so that comparisons were feasible. The plot of heat flow versus temperature of the oxidation experiments shows an induction period, which lasts up to approximately 200 O C , followed by three or four distinct peaks. Above a temperature of 500 to 550 "C the oxidation process is complete. Fig. 4-77 shows a diagram of the recorder graph of the DSC oxidation test of the distillation bitumen B65 (sample 9) in 7 bar air at a heating rate p= 20 K/min. The influence of the heating rate

248


HIpWI

1

0

100

200

3 00

L 00

5 00 600 T 1°C I

-

Fig. 4-77 Generalized DSC Recorder Graph during Oxidation of Distillation Bitumen B65 Sample 7 Heating Rate p: 20 K/min Atmosphere: Air, 7 bar pressure, 5 cm3/min

on the peak maximum temperatures is the same as that in argon. The kinetics of the oxidation reactions can be calculated according to ASTM E 698-79. The onset temperature and the maximum temperature of the first peak ( T I )are conventionally used as criteria for oxidation stability, and these values are listed in Tables 4-98 to 4-101 for bitumens and their colloid components. A statistical evaluation of the individual data is given in Table 4-102 (heating rate p= 10 K/min). Comparison of the distillation and the blown bitumens reveals differences between Tenser and TI.The colloid components show differences between the means, except for the means of Tonset of the dispersion medium and the asphaltenes. The temperatures of the peak maxima of the bitumens and their colloid components are quite uniform for equal heating rates (p = 5, 10, 20 K/min). Statistical evaluation gives coefficients of variation between f0.7 % and f2.5 % maximum. Therefore, the data were not calculated for the individual samples; instead the average values of the Arrhenius coefficients were calculated using the means of the corresponding peak maximum temperatures. For all the bitumens and their colloid components the first oxidation peak appears between 275 and 325 "C depending on the heating rate. The other peaks do not appear at equal temperatures for all of the samples, but the individual data may be concentrated in groups, whlch have means with low tolerances (Tables 4-103 to 4-107).


249

Table 4-98: Onset temperature Tonset and maximum temperature of the 1st peak T I of the oxidation of bitumen in 7 bar air (p = 10 K/min) Sample No.

1 2 3 4 5 6 7 8 9 10 I1 12 13 14 15 I

I1 I11

Iv V

Bitumen

B 200 B 200 B 200 B 80 B 200 B 80 B 65 B 45 B 25 B 80 V R (B 200) V R (B 130) VR (B 80) V R (B 65) B 200 exper. product 100125 85/40 85/25 100125

["CI

Tl ["Cl

227.50 235.00 200.00 191.25 227.50 225.00 212.50 225.00 215.00 225.00 200.00 220.00 222.50 215.00 220.00 218.75 192.50 182.50 195.00 232.50

297.50 300.00 283.75 285.00 297.50 295.00 295.00 295.00 295.00 300.00 29 1.25 300.00 295.00 297.50 300.00 293.75 285.00 286.25 282.50 293.50

TOllSCt

Table 4-99: Onset temperature Tonset and maximum temperature of the 1st peak TI of the oxidation of dispersion medium in 7 bar air (p= 10 K/min). Sample No.

10 12 15 V

Dispersion medium from B 200 B 200 B 200 B 200 B 80 B 65 B 45 B 25 B 80 VR (€3 130) B 200 100125

["Cl

T, ["CI

195.00 202.50 195.OO 182.50 192.50 197.50 215.00 200.00 212.50 185.00 180.00 212.50

290.00 300.00 292.50 288.75 290.00 290.00 295.00 292.50 297.50 295.00 290.00 305.00

Tollset

250


Table 4-100: Onset temperature Tonset and maximum temperature TI of the 1st peak of the oxidation of petroleum resins in 7 bar air (p= 10 K/min). Sample No.

1 2 3 5

6 7 8 9 10 12 15 V

Petroleum resins from B 200

Tonset

["el 235

-

B 200 B 200 B 80 B 65 B 45 B 25 B 80 V R (B 130) B 200 100/25

285

-

-

205 205 205 200 207.5 220 210 210 207.5 232.5

285 285 287.5 290 287.5 287.5 295 292.5 280 295

Table 4-101:Onset temperature To,,,, and maximum temperature TI of the 1st peak of the oxidation of asphaltenes in 7 bar air (B = 10 K/min). Sample No.

1 2 3 5 6 7 8 9 10 12 15 V

Asphaltenes from B 200 B 200 B 200 B 200 B 80 B 65 B 45 B 25 B 80 V R (B 130) B 200 100/25

Tonset

["el

TI ["el

190 190 205 180 182.5 175 190 207.5 185 175 192.5 205

297.5 295 3 10 295 295 300 300 307.5 3 10 307.5 302.5 307.5

Table 4-102: Means of the onset temperature (Tonset) and the maximum temperature T I of the 1st peak of the oxidation in 7 bar air (fi = 10 K/min). TOIlW

V[*%]

T1

V[*%]

217.4 204.3

5.56 10.11

295.2 288.2

1.72 1.78

197.5

5.94

293.9

1.68

212.5

5.47

288.2

1.60

189.8

5.93

302.3

1.98

x ["C] Distillation bitumen Blown bitumen Dispersion medium from distillation bitumen Petroleum resins from distillation bitumen Asphaltenes from distillation bitumen

x ["C]

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15

SampL No.

VR VR VR VR

B 200 B 200 B 200 B 80 B 200 B 80 B 65 B 45 B 25 B 80 (B 200) (B 130) (B 80) (B 65) B 200

Temperature range

111.5

9.875

275-320 "C

1

208.2

-

-

208.2

158.9

-

-

158.9 -

14.790

-

-

-

-

-

14.790

11.222

-

-

11.222

420-460 "C

155.4

-

94.3

155.4

-

I-

}

}

-

-

10.496

-

-

148.7

148.7

-

-

148.7 148.7

-

-

-

10.496

-

-

-

-

-

148.7

9.683

-

9.683

-

log A [min-'1

465-520 "C E [k.J/Mol]

-

9.031

10.496

155.4

log A [min-'1

445-490 "C E [k.J/Mol]

Table 4-103: Arrhenius coefficients of the oxidation of distillation bitumen in 7 bar air.

2

g.

38

B 200 B 200 B 200 B 200 B 80 B 65 B 45 B 25 B 80 V R (B 130) V R (B 80) B 200 100125

1 2 3 5 6 7 8 9 10 12 13 15 V

E [kJ/Moll

log A [min-']

275-3 15 "C

172.3

-

172.3

E kJiMol1

Petroleum resins from

B 200 B 200 B 80 B 65 B 45 B 25 V R (B 130) V R (B 80) B 200 100125

1 3 6 7 8 9 12 13 15 V

Temperature range

Sample No.

98.5

8.751

log A [rnin-'1

275-305 "C E [kJIMoll

J

I

I

i

134.4

134.4

E [kJIMol]

9.634

9.634

log A [min-'1

400-445 "C

12.567

-

12.567

log A [min-11

4 15-450 "C

Table 4-105: Arrhenius coefficients of the oxidation of petroleum resins in 7 bar air.

Dispersion medium from

Sample No.

Temperature range

Table 4-104: Arrhenius coefficients of the oxidation of dispersion medium in 7 bar air.

J

log A [min-']

-

7.521

log A [min-'1

470-515 "C

E [kTIMoll

-

116.4

E [kJ/Moll

450-510 "C

as 5

9

"a

R

5

a s

s a

U

3

3

a

h)

E

B 200 B 200 B 200 B 200 B 80 B 65 B 45 B 25 V R (B 130) V R (B 80) B 200

1 2 3 5 6 7 8 9 12 13 15 V

118.8

E [kJ/Moll

9.768

log A [min-'1

280-325 "C

J

171.10 -

171.0

E [kT/Moll

IV

Sample No.

I

85/25 1OOD5

\%?:

exper. product

$1 1

Temperature range

101.4

E [kJ/Mol]

9.078

[mm ' -1 ]

log A

275-300 "C

1

I

1 i 138.3

E [kJ/Mol]

10.258

log A [min-l]

380-425 "C

1

89.2

E [kJ/Mol]

6.028

log A [min-'1

405-460 "C

12.366 -

12.366

log A [min-'1

410-450 "C

Table 4-107: Arrhenius coefficients of the oxidation of blown bitumen in 7 bar air.

100/25

Asphaltenes from

Sample No.

Temperature range

Table 4-106: Arrhenius coefficients of the oxidation of asphaltenes in 7 bar air.

-

9.073 -

-

140.4

152.2 154.0

9.875

9.875 10.660 6.152

log A [min-'1

470-520 "C

E [kJ/Mol]

140.4

9.073

9.073

140.4 -

log A [min-'1

E [kJ/Mol]

465-520 "C

2a

P

s

2

g.

2 6'

254


The appearance of up to four peaks, with three in the range 400-500 "C indicate the presence of groups of compounds with different susceptibility to oxidation and the occurrence of a series of consecutive and/or parallel reactions. In order to elucidate this behavior the peak maximum temperatures from experiments with a heating rate p= 10 K/min are listed in Table 4-108. The onset temperature, and the temperature and kinetic data of the first oxidation peak maximum are the criteria which define the practical behavior of bitumens in its applications. Calculation of the reaction rate constants and of the half-life time tin, using the Arrhenius coefficients gives values, which may be reproduced by other methods, although the oxidation does not obey the first order reaction law. The plot of the log t1,2versus the inverse Kelvin temperature (1 000/T) is shown in Fig. 4-78. Correspondinggraphs for the other peaks show the increase in the half-life times. However, they are only of theoretical interest and do not have any relevance to practical behavior in production and manufacturing. Fig. 4-78 and Table 4-102 (Tonset> show that oxidation commences at temperatures which are commonly applied in asphalt mixing plants. This may result in the formation of longlife radicals, which may serve as initiators for the oxidative aging of the binder in the asphalt layer on roads, under the influence of ultraviolet light [4-281. The tests also confirm that the oxidation stability of blown bitumens is inferior to that of distillation bitumens. The average onset temperature of the blown bitumens is about 13 "C below that of distillation bitumens. The activation energy for the first oxidation peak of blown bitumens is approximately 10 W/Mol (or approximately 10 %) below the value of distillation bitumens, with the corresponding consequences for the half-life times. For example, the half-life time of a distillation bitumen is approximately 60 % greater than the half-life time of a blown bitumen (at 200 "C).

5

4

3

2

1

Distillation bitumen Dispersion medium from no.1 Petroleum resins from no.] Asphaltenes from no. 1 Blown bitumen 288.2

302.3

288.2

293.9

295.2

5

11

11

11

15

1.78

1.98

1.60

1.68

1.72

408.0

428.6

422.0

446.25 443.25 430.9

1.18

5

8 2.65

2.75

11

9

-

0.88 0.69

2

5

X

-

428.8

-

-

476.2

468.8

5

-

-

10

8

Peak 3 n V

1.60

-

-

1.69

1.53

[*%I

(p = 10 K/min).

["CI

Table 4-108: Means of the peak maximum temperatures of the DSC oxidation in 7 bar air ~

489.5 476.2 457.5

492.3

489.3

488.6

["CI

X

3 1 1

12

9

5

Peak 4 n

0.54

1.62

1.50

0.26

V [k%]

$

wl

wl

h,

s

0

cl

0

s.

3. 0-a

256


Fig. 4-78 Low Temperature Oxidation of Bitumens (First DSC Peak) Half Life Time tl,2 versus Temperature (Mean Values) Curve 1: Distillation bitumens Curve 2: Dispersion medium from 1 Curve 3: Petroleum resins from 1 Curve 4: Asphaltenes from 1 Curve 5: Blown bitumens


-9.5

I

-60

-40

I

-20 Tempfrotura ('C>

Fig. 4-79 Low Temperature DSC of Bitumen B200 Section of the Heating Curve Heating Rate p : 5 K/min

,

0

I

257

I Genarol V 2 . 2 A DuPont 9900

20

258


4.3.5 Low temperature behavior of bitumen Decreasing temperature causes increasing brittleness of bitumes. The breaking point according to Fraalj (DIN 52 012) must not exceed -15 "C for road bitumen B200 and - 2 "C for bitumen B25. Glass transition or second order transition is found at defined temperatures and may be recognized by steps in the heating curves (heat flow versus temperature) of DSC or DTA tests. The appearance of such steps in the cooling curves is less often observed [4-31 to 4-35]. Fig. 4-79 shows an enlarged section of the heating curve of a bitumen, B80, which has a glass transition in the temperature range between -27.6 "C and -10.4 "C (average value -23.0 "C). The brealung point of this bitumen has been determined as -19 "C. Frequently the effects are so small that they may erroneously be ascribed to artefacts, or may be overlooked. In theory the paraffins, such as n-alkanes, may be traced using low temperature DSC, but the lower limit for detection of such substances is approximately 2 %, which is also the upper limit for the paraffin content in road bitumen.

4.3.6 Conclusions from experiments on bitumen 4.3.6.1 Thermogravimetry Thermogravimetry in an inert gas facilitates the determination of correlations of index numbers from TGA, with some consistency and analytical data of bitumens. Correlation of the penetration with the index numbers, T1 %, T5 %, and AG400, produces two groups of similar behavior. A regression line can be calculated for each of the two groups (Fig. 4-38 to 4-40). It is surprising that no correlation was found of the penetration with R800, since R8OO approximately represents the concentration of asphaltenes, and the penetration does correlate with the concentration of asphaltenes themselves. The index numbers mentioned above do not correlate with the penetration of blown bitumen either. A better result was found for the correlation of the penetration with the maximum weight loss rate, DTG. Definite correlations can also be found for the softening point R&B with the index numbers T1 %, T5 %, AG400, and DTG, for the distillation bitumens (Fig. 4-42 to 4-44), but not for the blown bitumens. The correlation of the content of crackable residue, CR, with the penetration is not as good as the correlation of the CR with the softening point R&B for distillation bitumens (Fig. 4-45 and 4-46). Comparable correlations for the dispersion medium of the distillation bitumens result in two regression lines each for the correlation with penetration, whereas only one regression line results for the correlation with the softening point R&B (Fig. 4-47 and 4-48). The results mentioned indicate that the scattering is too large, and therefore the coefficients of correlation are to small, to obtain a positive description of the consistency data of


259

bitumens from thermogravimetric index numbers. The sixteen distillation bitumens tested often show two regression lines running more or less parallel. Nevertheless the same bihimens were found on the same regression line in all cases. The separation of the data for blown bitumens from those for distillation bitumens is quite sharp, whereas the semiblown bitumens fall between the other two bitumens. The second objective of this investigation was an attempt to correlate the index numbers of thermogravimetry with analytical data of the bitumens. Very good results were found for the correlation of the means of the residues at 600 "C or 800 "C with the means of the H/C ratio (Fig. 4-49), where coefficients of correlation > 0.97 were found. There is a relationship between the mean molecular weight, the start temperature of the evaporation (T5 %) and the distillable fraction (AG400) for bitumen and its colloid components (Fig. 4-50 to 4-53). The dispersion medium shows the highest values for the coefficients of correlation, which decrease via the petroleum resins down to the asphaltenes. There is no difference between distillation and blown bitumens. The residue at 600 "C or 800 "C respectively may be correlated fairly with the concentration of colloids (total petroleum resins and asphaltenes). Here also the distillation and the blown bitumens fall on the same straight line (Fig. 4-54 and 4-55). Correlation of the maximum of the reaction rate DTG with the concentration of the colloids results in two regression lines for the distillation bitumens, and a third line for the blown bitumens (Fig. 4-56). The concentration of the dispersion medium determines the evaporation start temperature (T1 % or T5 %) of the distillation bitumens (Fig. 4-57). For the blown bitumens no relation was found, but for both the distillation and the blown bitumens a clear correlation of the distillable fraction AG400 with the concentration of the dispersion medium was recognized (Fig. 4-58). The Concentration of petroleum resins influences the maximum of the reaction rate DTG considerably (Fig. 4-59). This correlation exhibits higher coefficients of correlation for the regression lines of the distillation and blown bitumens than does a similar correlation for the sum of colloids. The correlation of the maximum of the reaction rate DTG with the concentration of asphaltenes again results in two lines for the distillation bitumens and a third line for the blown bitumens (Fig. 4-61). Similar behavior was found for the correlation of the residues at 600 "C and 800 "C with the asphaltene content (Fig. 4-60). These results prove a more or less good correlation of some index numbers from thermogravimetry in inert atmosphere with a series of analytical data. We did not at first expect that the coefficients of correlation would be high, since bitumen, as well as its separated colloid components, still represents a multicomponent system of varying composition. The temperature limit of the thermal stability of bitumen and its components was found to be relatively uniform independent of the origin of the samples. Mean values of this temperature with very small coefficients of variation were calculated (Tables 4-67 and 4-68). The sequence of the cracking temperatures: dispersion medium < petroleum resins < asphaltenes

260


was as expected. The lower start temperature for the dispersion medium of the blown bitumens is due to the predominating evaporation of low boiling flux oils. Thermogravimetry in air reveals the oxidation behavior of the bitumens. The DTG curve supplies the temperatures at the beginning and end of the oxidation. There was no difference in the oxidation behavior of the colloid components from distillation or from blown bitumens. The origin of the bitumens had no effect. One cannot be certain about the beginning of oxidation of the blown bitumens, their dispersion medium, or their petroleum resins, because this type of bitumen contains relative low boiling flux oils. There are only small, statistically insignificant, differences between the means of the DTG maximum temperatures of the blown bitumens, their dispersion medium and their petroleum resins in air and in argon. Thermogravimetry cannot determine whether evaporation or oxidation takes place. Use of a simultaneous thermal analyzer might prove whether the reaction is exo- or endothermic. The results of the test at constant temperatures, dependent upon the test time, are more interesting for practical application. This experiment may be carried out more easily using a thermobalance, and with better interpretation than the standard test using the rotating flask (DIN 52 017). The experiment in argon and in air provides additional information from quotient Q:

Q

loss in air = loss in argon.

Q < 1indicates predominating evaporation of compounds in the sample. Q > 1 indicates the predominance of highly volatile oxidation products, whereas Q = 1 suggests an equilibrium between evaporation of original compounds of the sample and of volatile oxidation products. However, it may also indicate evaporation of original components both in argon and in air. With reference to Fig. 4-67 and considering the value of Q after 30 minutes test time, there is a sequence:

B200 < B80 < B65 < B25. Thermogravimetry, both isothermal and temperature-programmed, provides information about the thermal and oxidative behavior of bitumens and their colloid components. The results of the temperature-programmed method are valuable in the assessment of heavy residues with regard to their suitability for conversion processes. The results are not so useful in bitumen technology, since the effects in temperature-programmed thermogravimetry do not fall within the range of the manufacturing temperatures of bitumens. Isothermal thermogravimetry is superior in these circumstances although some experiments may be performed using other (cheaper) instruments. The consistency data (penetration and S.P. R&B) of bitumens cannot be simulated by thermogravimetry except to discriminate between distillation, semi-blown, an blown bitumens.


261

4.3.6.2 Reaction kinetics The coefficients of the Arrhenius equation, i.e. activation energy E and frequency (pre-exponential) factor A , of pyrolysis reactions may be determined using DSC, as exactly as possible for multicomponent systems when very small sample masses ( < 5 mg) are used. Determination of the reaction enthalpy is more difficult since the base line of the recorded signal does not run on equal levels before and after the reaction. Therefore it is not possible to determine the exact integration limits with sufficient certainty. The equation for the numerical calculation of the activation energy E involves the peak maximum temperature in linear and square terms. This is a source of errors. For graphic determination, log P(heating time) is plotted versus the inverse Kelvin temperature 1 OOO/ T (log P = f( 1 O O O / r ) ). The slope of the regression line is equivalent to the term E/R. Calculation of E requires multiplication by a factor of 8.319. As a consequence the means of E and A for distillation bitumens, computed from the individual data of Table 4-80, have relatively greater coefficients of variation ( fV = 8.27 % maximum) than the average values of E and A, computed using the means of the corresponding peak maximum temperatures ( fV = 1.01 %). These results had not been expected, considering the different origins of the samples, their different colloid compositions, and their different average molecular weights. Element analysis proved that the H/C ratios of the distillation bitumens, and of the blown bitumens and their colloid components, show very small variations (Tables 4-53 and 4-54). The peak maximum temperatures of the blown bitumens are not so uniform and therefore the calculation of means is not worth while. Some of the blown bitumens, their dispersion medium and their petroleum resins demonstrate peak maxima both in the distillation and in the pyrolysis range because of the content of low boiling flux oils. The activation energies computed for the distillation range are equal to the enthalpies of vaporization, as shown by experiments on model substances. The activation energies of pyrolysis are in the order of magnitude of the values for n-alkanes with a carbon chain length from C5 up to C,,, or of i-alkanes with carbon chain lengths > C,, (ascertained by other methods and cited in papers and compilations). The enthalpies of vaporization correspond to the literature values for n-alkanes with carbon chain lengths between C,, and C2*.The enthalpies of condensed aromatics such as naphthalene, anthracene, and phenanthrene are also found in this region. For bitumen technology, DSC measurements provide important evidence that the thermal stability of bitumen is not jeopardized at normal manufacturing temperatures, provided that oxygen (air) is excluded. Additional experiments were performed, to reduce the evaporation and favor the pyrolysis reactions, and the better to simulate the parameters of industrial thermal cracking. That was achieved by using methane as an inert hydrocarbon atmosphere at 10 bar pressure and 5 cm3/min gas flow rate. These experiments are quite meaningless with regard to bitumen technology, but bear some importance with regard to the behavior of heavy distillation residues and heavy crude oils in pyrolysis reactions such as visbreaking, thermal cracking,

262


and coking. In order to investigate the influence of chemical composition, the bitumens and their separated colloid components were pyrolysed. The observations made during the pyrolysis of distillation bitumens in 1 bar argon have proved to be generally true: that there is no significant difference between the means from individual calculated data of E and log A , and the average values of E and log A calculated using the means of the peak maximum temperatures. Nevertheless the coefficients of variation are much greater if the means are calculated from the individual data of E and log A. The experiments in 10 bar methane show that neither the origin nor the composition affect the results for the distillation bitumens and for their colloid components. No correlation was found for the Arrhenius coefficients with analytical data. The average values of these coefficients for the colloids components showed only small differences, but the conversions show considerable differences. The conversion may be correlated with the H/C ratio in a similar way to the correlation of the H/C ratio with the residues R600 and R800 in thermogravimetry (Fig. 4-49). Decreasing HIC ratios, i.e. increasing aromacity, lead to a decrease of the conversions (Fig. 4-80). In this case the data of the blown bitumens also fit the regression line very well. The blown bitumens do not exhibit peaks in the evaporation range when the system pressure is increased to 10 bar, except for the dispersion medium of the bitumens 85/40 and 85/25, which demonstrate only an evaporation loss. The Arrhenius coefficients of the blown bitumen showed greater differences than those of the distillation bitumens. In the plot of half life time, t,,2, versus the inverse Kelvin temperature, the distillation bitumens and their colloid components follow almost parallel lines, whereas the graphs for the blown bitumens and their colloid components diverge. The plot of t1,2 versus 1 OOO/T shows the residence time required to achieve a conversion of fifty percent, at a preset reaction temperature, or which temperature is required to achieve a preset conversion at a preset residence time. This information is valuable in thermal processing, for example in selection of the crack severity of the visbreaking process. The order of magnitude of the activation energy E and the frequency factor A from the experiments in 10 bar methane corresponds to the values of n-alkanes, n-C,,H,, to ~Z-C,~H,, (at normal pressure). No data for compariso 10 bar pressure can be found in the literature. According to the literature the activati nergy E and the frequency factor A tend asymptotically towards a final value with increasing carbon chain length. We have confirmed this in experiments on low density linear polyethylenes with an average chain length of 4300 to 27 740 C atoms. No differences of E or A could be verified. According to Zdonik [4-291, this final value is already reached with n-pentane or n-hexane, but this is not borne out by other authors. The plot of the data of the activation energy E of the n-alkanes versus the carbon numbers has an asymptotic slope above about C,6. During the pyrolysis experiments using DSC in 1 bar argon, vaporization of the n-alkanes up to a chain length of C, was found, whereas in the experiments in 10 bar methane a change from vaporization to pyrolysis occurred at a chain length of about C3,. Thus any comparison with the activation energies of the pyrolysis of bitumens can only be hypothetical. The


I

1.1

1.2

1.3 1.4 H / C

Fig. 4-80 Pyrolysis of Bitumens Conversion U versus Atomic Ratio H/C (Mean Values) z)r Distillation Bitumens 0 Blown Bitumens

1.5

1.6

263

1.7

264


value of 220-240 kJ/Mol for the activation energy of the bitumen components is in the order of that of long chain n-alkanes. The aliphatic C-C bond requires the lowest quantity of energy for splitting [4-301, its average free bonding energy being 297 kJ/Mol. Therefore one may deduce that this bond is cracked in each of the three components of bitumen. This assumption is supported by the very small and statistically insignificant differences of the Arrhenius coefficients of the colloid components. The content of aliphatic bonds decreases from the dispersion medium via the petroleum resins down to the asphaltenes, as shown by decreasing conversions. Determination of the start temperature of the cracking process using DSC gives very similar data for the series of the distillation bitumens, both in 1 bar argon and in 10 bar methane. This provides further support for the theory. The tolerances are very small and permit the calculation of means. Once again, the origin of the samples has no effect. The data from DSC and from TGA show such very small differences that they may be regarded as comparable, and confirm that no cracking processes occure below a temperature of approximately 400 "C. There is a considerable difference between the peak maximum temperatures Tmxof the reaction rate DTG in thermogravimetry, and the maximum temperatures of the peaks of the energy consumption in DSC. The mean of the DSC maximum in 1 bar argon (p = 10 K/min) exceeds the mean of Tmx from TGA by approximately 25 "C or 6 %. The suggestion that in DSC a energy flow takes place, and in TGA a mass transport, does not explain this difference. Determination of the onset temperature in DSC experiments in air give a more exact determination of the start temperature of the oxidation reaction than comparing the TGA curves in argon and in air. DSC also supplies exact evidence concerning the direction of heat flow (endo- or exothermic) and thus makes a clear distinction between evaporation and oxidation possible. TGA in air shows up to ten peak maxima of the DTG curve, whereas DSC in air shows only up to four distinct peak maxima. It is not possible to relate this to analytical data. Statistical evaluation of the onset temperatures gives coefficients of variation from k 5.5 % to 10 %, which indicates a systematic influence. The corresponding peak maximum temperatures of the first and the last oxidation peaks were found in DSC at lower values than in the DTG curves, since the oxidation experiments in the DSC were carried out at 7 bar air pressure. However the peak maximum temperatures in DSC for the different ranges show good uniformity. The coefficients of variation of the corresponding means range only from f0.3 % to ? 2.75 %, and may therefore be used to calculate the Arrhenius coefficients. ravimetry in air and the DSC oxidation experiments confirm the inferior ability of blown bitumens. DSC peak maximum temperatures, the activation energy of the first oxidation peak, and the temperatures of corresponding DTG maxima for blown bitumens are inferior compared to distillation bitumens. Knowledge of the kinetics of the oxidation reaction permits extrapolation of the resulting half life time to lower temperatures, producing information about the oxidation stability at manufacturing temperatures and the application of the bitumens. This only gives a rough estimate of the behavior on the actual road surface, since there are additional

4.4 Investigations on polymer modified bitumens (PMB)

265

effects of ultraviolet radiation. For the blowing process, parameters such as material temperature and residence time for the required degree of oxidation can be preselected, using the calculated kinetics for oxidation.

4.4 Investigations on polymer modified bitumens (PMW Bitumen can only be used industrially in a limited range of temperatures. The upper limit is marked by the softening point R&B (DIN 52 01 1, ASTM D 2398-68) and the lower limit by the breaking point according to FraaB (DIN 52 013, IP 80 153). This range, also called the plasticity span, is limited towards lower temperatures by increasing brittleness and towards higher temperatures by plastic deformation under load. To extend this plasticity span, mixtures have been tested, with nearly every polymer, but only a few of these became commercial products. Nowadays commercial mixtures of bitumens with uncured synthetic elastomers are produced, e.g. ethylene-propylene-diene terpolymers (EPDM), styrene-butadiene sequence copolymers (SBS), and ethylene-acrylic ester-acrylic acid terpolymers (AECM). Mixtures with some thermoplastics are also commercial products, e.g. polyethylene (PE), ethylene-propylenecopolymers (EPM), alpha-olefinic copolymers, atactic polypropylene (aPP), and ethylene-vinyl acetate copolymers (EVA). For technical applications, PMB with a relatively low polymer content (3-5 %) is used for road construction; mixtures with a polymer content of 15-20 % are employed as roofing felts [4-22, 4-35, 4-36]. Characterizationof such mixtures using thermoanalytical methods is presented in chapter 4.4, using the examples of four commercial PMBs and their basic bitumens.

4.4.1 Description and characterization of the samples The samples in this investigation are given in Table 4-109: Table 4-109: Samples. Sample No.

Type

1 2 3 4

Bitumen B80/1 PMB/1 = B80/1 Bitumen B80/2 PMB/2 = B80/2

+ 8 % EPM

+ approx. 3 % SBS

266


Sample No.

5Pe Bitumen B200/3 PMB/3 = B200/3 + EPDM Bitumen B80/4 PMB/4 = B80/4 + SBR cured

The bitumens and the PMB are products from German refineries. PMB/l, PMB/2, and PMB /3 are commercial products, whereas PMB/4 is an experimental product. The consistency data of the bitumens and the PMB are given in Table 4-110: Table 4-110: Consistency data of bitumen and PMB. Substance B80/1 PMB/l B80/2 PMB /2 B200/3 PMB /3 B80/4 PMB 14

Pen (0.1 mm)

SP R&B ("C)

BP ("C)

P1.Sp. ("C)

63.6 54 85.5 76.6 160.2 62.6 80.4 72.7

50.6 57.4 47.4 52.9 40.4 54.3 48.4 52.3

-16 -19 -14 -14 -16 -15 -13 -16

66.6 76.4 62.4 66.9 56.4 69.3 61.4 68.3

= needle penetration = softening point ring & ball

Pen (0.1 mm) SP R&B ("C) BP (oc)o P1.Sp. ( C )

= breaking point according to FraaS = plasticity span (SP R&B minus BP)

The data of the element analysis are listed in Table 4-1 11. Nitrogen content is not declared since the concentrations are of an order of magnitude within the limits of error of the analytical method. The content of oxygen is analysed directly to avoid the accumulating errors of calculation by differences. Table 4-111: Element analysis of the virgin samples (wt%). Sample

C

H

S

0

B80/1 PMB/1 B80/2 PMB/2 B200/3 PMB /3 B80/4 PMB 14

84.81 84.90 84.81 84.70 84.98 85.52 83.30 83.22

10.46 10.79 10.25 10.23 10.50 10.31 10.09 10.05

4.46 4.25 4.47 4.18 4.40 3.28 6.60 6.69

0.60 0.78 0.83 0.82 0.60 0.91 0.60 0.50


267

The samples were aged using the rotating flask test (DIN 52 017 similar to ASTM D 2872-85) both in argon and in air. The weight control did not show any change of weight, since all the differences were within experimental error. Notwithstanding the absence of any weight changes, considerable alterations to the consistency data occured, both to the penetration and to the softening point. The alterations of the breaking points were also within the experimental error. For ease of comparability, the alterations to the consistency data, both of the penetration and of the softening point, are reduced to their relative values (related to the start value of each) in Fig. 4-81 and 4-82. Element analysis of the samples after the rotation aging in air shows only small alterations in the composition. The assumption that alterations in the consistency data are due to oxygen uptake during oxidation is not justified. Decreases in the concentrations of oxygen as well as increases were observed. The absolute alterations of oxygen concentration are certainly small but related to the initial concentrations, they are considerable and exceed the tolerances of the analytical method (Table 4-1 12):

Penetration Decrease ( $ 1 40 35 30

25 20

15 10

5 0

B€JO/l

PmB/i 3 8 0 / 2 PmB/2 3200/3 PmW3 B80/4 PmB/4

Air

Argon

Fig. 4-81 Aging in the Rotating Flask of Bitumen and PMB (DIN 52 017). Relative Change in Needle Penetration (DIN 52 010)

268


Softening Point (Ring & B a l l )

Increase ( % I 20

16 14 12 10 8 6

4

2 0

BBO/i

PmB/f

B 8 0 / 2 PmB/2 B200/3 PmB/3 B80/4 PmB/4 Air

Argon

Fig. 4-82 Aging in the Rotating Flask of Bitumen and PMB (DIN 52 017) Relative Change in Softening Point Ring & Ball (DIN 52 01 1) Table 4-112. Oxygen concentration before and after the rotation aging in air.

Sample B80/1 PMB/l B80/2 PMB /2 B200/3 PMB/3 B80/4 PMB 14

Oxygen concentration (%) unaged aged 0.6 0.78 0.83 0.82 0.60 0.91 0.60 0.50

0.54 0.68 0.80 0.84 0.64 1.14 0.55 0.59

Alteration (%) absolute relative

-0.06 -0.10 -0.03 +0.04 +0.04 +0.23 -0.05 +0.09

-10 -12 - 3.6 + 2.5 + 6 +25 - 8.3 +18


269

4.4.2 Thermogravimetry 4.4.2.1 Dynamic (temperature-programmed) thermogravimetry The Stanton-Redcroft TG 750 thermobalance was used, applying the standard parameters: heating rate p= 10 K/min and a gas flow rate of 25 cm3/min. The TGA curves already known for bitumens were obtained. The TGA curves of B80/1 and PMB/1 are congruent up to approximately 475 "C. There is only a difference in the coking behavior (Fig. 4-83). Comparison between B200/3 and PMB/3 reveals differences: the curve of PMB/3 is shifted towards lower temperatures (Fig. 4-84). This due to the

Fig. 4-83 Thermogravimetry of Bitumen and PMB (TG 750) Heating Rate p: 10 K/min Atmosphere: Argon 25 cm3/min Curve 1: Bitumen B80/1 Curve2: PMB/1

270


100

200

300

400

500

600

700

800

Fig. 4-84 Thermogravimetry of Bitumen and PMB (TG 750) Heating Rate p: 10 Klmin Atmosphere: Argon 25 cm3/min Curve 1: Bitumen B200/3 Curve2: PMBl3

presence of a furfural extract, which had been used to swell the polymer, prior to mixing with bitumen. The curves of the pairs: B80/2 and PMB/2, and B80/4 and PMB/4, are completely congruent in the experiments in argon. During the experiments in air the pair B200/3 and PMB/3 shows divergence of the curves at low temperatures (Fig. 4-85). The low values of T1 % and T5 % for PMB/3 indicate loss by evaporation, prior to the start of the oxidation at approximately 330 "C. The other three pairs also have congruent curves in the oxidation experiments. The index numbers of the experiments in argon are listed in Table 4-113, those of the experiments in air in Table 4-1 14. The residues at 600 "C and 800 "C of the experiments in air consist of inorganic material (ash). The experiments in argon demonstrate only one DTG peak maximum for the

bitumens (PMB)

Fig. 4-85 Thermogravimetry of Bitumen and PMB (TG 750) Heating Rate p : 10 Klmin Atmosphere: Air 25 cm3/min Curve 1: Bitumen B200/3 Curve2: PMB/3

27 1

B

P D

c1

-n$ r.

B 2 @€

a

.'L Yi

.'Y -i

u '3

4

10'6 1.51 L'91

1.91

I'PI L81 S'LE 0'91 L'IE

O'SI S'IZ

16P -

0'8I

Z.PI -

OP'8

O'SI

O'EE O'EI

ElP 89P

8Z'S 90'1 L6'9 8'91 L'LI

6SP 5.51

L'PI

5' I

I'Z

z.1

-

-

If0 -

-

-

P'8 €'I

-

001

9'0 6'0

002

2'8 8'6 0' I

I8'0I

O'EI

P8'9

S'OZ

OOE

19P 899

6'1 I

OOP

EP'6

009

26'6

tl

008

0 LP 180

ma

E9P -J

P / 8Md P/088 E / 8Nd 6 I ooz 8 Z I tIMd ZlO88 1 I 8Nd

SIZ

062

1zz

OSE

99z

9LZ

8 LZ

882

POZ 06Z

1/088

6PE

IS€

IOE

amqsqng

9LZ

09E

% IJ.

ZEE

% SL


273

Table 4-114: TGA in air (25 cm3/min) (p= 10 Klmin) Substancl T1 To

T5 %

AG

100

200

R

300

- -

400 ~

600

__

B80/1

272

316

-

3.2

22.3

0

PMB I 1

261

316

-

3.3

22.0

0

B8012

207

256

0.7

13.6

34.0

0

PMB 1 2

195

247

1.0

14.6

34.7

0.8

B 200 1 3

267

307

-

4.0

30.5

0

PMB 1 3

210

248

0.6

18.7

39.0

0

DTG 800

0.61 8.46 3.12 4.00 6.00 5.82 7.90 4.90 7.84 1.04 5.09 6.68 6.74 12.87 9.38 8.64 2.22 3.85 2.76 5.05 6.06 9.31 2.51 6.27 3.17 6.13 7.38 6.64 11.55 1.69 10.81 2.19 5.00 5.31 6.06 5.56 5.31 3.75 7.44 3.08 2.94 3.56 3.49 4.93 4.79 6.84

308 370 392 403 410 416 434 456 511 308 370 407 412 428 477 510 302 347 367 410 438 501 298 353 383 422 430 442 495 310 362 382 398 400 414 419 433 459 528 303 333 388 404 439 456 538

274


Table 4-114 continued: TGA in air (25 cm3/min) Substance T1 %

(p= 10 K/min) DTG

R

AG

100

200

300

400

600

800

__

B80/4

255

-

6.0

29.0

0

-

PMB / 4

243

0.2

6.2

28.9

1.6

1.2

- __ __

__

1.84 3.15 3.28 12.52 7.21 5.24 5.97 6.69 14.88 1.89 5.40 7.81 8.14 3.95 5.99 5.21 7.36

310 365 373 389 399 418 424 438 528 311 360 380 395 423 446 456 521

samples. The maximum appears above 400 "C in the craclung region every time. The experiments in air show multiple DTG peaks, of which the first may indicate losses due to evaporation, as well as losses due to oxidation. For the DTG peak maximum temperatures of the experiments in argon a mean X = 470.4 "C and a coefficient of variation V = 2.16 % can be calculated. The experiments in air give different numbers of peaks in the eight samples, but six of the peaks appear at the same temperature level for each of the samples (Table 4-115):

*

Table 4-115: Means of the DTG maximum temperatures of the experiments in air.

Peak No.

x ("C)

f V (%)

1 2 3 4 5 6

306.7 365.6 403.8 436.4 456.8 518.6

1.49 1.73 1.42 1.32 1.50 2.91

A distillable fraction Am00 from 13- 19 % corresponds to the usual behavior of distillation bitumens. The high evaporation losses of the bitumen B80/2 and its PMB/2 are extraordinary whereas the high evaporation loss of the PMB/3 may be explained by the addition of furfural extract, to swell the polymer prior to mixing. The start of the oxidation reaction can be defined either by the first DTG peak maximum or by comparing the TGA curves in argon and in air. In the latter case, lower temperature

4.4 Investigations on polymer mod$ed bitumens (PMB)

275

data will be found. The weight loss of the samples generally starts at temperatures above 250 "C both in argon and in air. The TI % and T5 % values of the experiments in air are lower than those of the experiments in argon. The samples B80/2 and PMB/2 demonstrate a superimposition of evaporation and oxidation effects in the first stage. The flattening of the TGA curves of the experiments in air at temperatures between 350 and 370 "C indicates the formation of low-volatile oxidation products. An increase of the weight as a consequence of the resorption of oxygen in this range of temperatures (low temperature oxidation LTO) has only been observed with polymers which have many double bonds in the carbon chain (see chapter 4.4.5). Thermogravimetry did not show that addition of polymers had any stabilizing effect on the oxidation of bitumens.

4.4.2.2 Isothermal gravimetry Since no weight losses were found during rotation aging according to DIN 52 017, we tried to find the start temperatures of weight loss by using isothermal gravimetry. The samples were heated, applying a fast heating rate of p = 100 K/min, up to a preselected isothermal temperature and this was maintained for 150 minutes. The experiments were carried out both in argon and in air with a gas flow rate of 25 cm3/min. The sample weight was 5 to 8 mg. Therefore the conditions were similar to those of a thin-layer vacuum distillation [3-41 as described above. The essential difference compared to rotation aging, consists in the thickness of the material layers and in the quantity of gas per unit sample weight during the test time: Rotation aging:

gas flow rate sample weight test time calculated gas flow

500 cm3/min 100 g 150 minutes 0.75 cm3/l mg sample

Thermogravimetry:

gas flow rate sample weight test time calculated gas flow

- 6 mg

25 cm3/min

150 minutes 625 cm3/l mg sample

Experiments were carried out at isothermal temperatures of 165 "C (equivalent to the test temperature according DIN 52 017), 200 "C, and 250 "C. Disregarding the thickness of the material layers, the gas flow per unit weight in the thermobalance is many times that in the rotating flask. Nevertheless relatively small losses of weight are observed at 165 "C in argon. Five samples demonstrate losses of only up to 1 % (B80/1, PMB/1, B200/3, B80/4, PMB/4), whereas three samples demonstrate weight losses between 2 % and 4 % (B80/2, PMB/2, PMB/3). These samples also show weight losses between only 2 % and 4 % in air. At 200 "C the weight losses increase to noticeable values (Fig. 4-86 and 4-87). At 250 "C a distinct differentiation of the products occurs (Fig. 4-88 and 4-89).

276


10

9 8

7 6

5 4

3 2

1

Fig. 4-86 Isothermal Thermogravimetry at 200 "C (TG 750) Atmosphere: Argon 25 cm3/min Curve 2: PMB/1 Curve 1: Bitumen B80/1 Curve 3: Bitumen B80/2 Curve 4: PMB/2 Curve 5 : Bitumen B200/3 Curve 6: PMB/3 Curve 7: Bitumen B80/4 Curve 8: PMB/4


''If

Weight Loss ( I )

-

d

-I

15

30

45

60

75

.-. B

90

i

105

i

120 -Time

Fig. 4-87 Isothermal Thennogravimetry Atmosphere: Air 25 cm3/min Curve 1: Bitumen B80/1 Curve 3: Bitumen B80/2 Curve 5: Bitumen B200/3 Curve 7: Bitumen B&0/4

at 200 "C (TG 750)

Curve 2: Curve 4: Curve 6: Curve 8:

PMB/1 PMB/2 PMB/3 PMB/4

i

135

i

150 (min)

277

278


Fig. 4-88 Isothermal Thermogravimetry at 250 "C (TG 750) Atmosphere: Argon 25 cm3/min Curve 2: PMBI1 Curve 1: Bitumen B80/1 Curve 4: PMBI2 Curve 3: Bitumen B80/2 Curve 5: Bitumen B200/3 Curve 6: PMBl3 Curve 8: PMBI4 Curve 7: Bitumen B80/4


25

279

t

jt

Weight LOSS ( 2 )

20.

I

15

30

I

I

1

45

60

75

I

1

90

105

120

I

135

I

150

Fig. 4-89 Isothermal Thermogravimetry at 250 "C Atmosphere: Air 25 cm3/min Curve 1: Bitumen B80/1 Curve 2: PMB/1 Curve 3: Bitumen B80/2 Curve 4: PMB/2 Curve 5: Bitumen B200/3 Curve 6: PMB/3 Curve 7: Bitumen B80/4 Curve 8: PMB/4

The quotient Q =

loss in air

loss in argon provides evidence as to whether the weight loss is predominantly due to the evaporation of non-oxidized portions of the sample, Q < 1.O, or is mainly a consequenceof the formation of highly volatile oxidation products, Q > 1.0. The plot of the loss quotient Q against the test time is shown in Fig. 4-90 to 4-92.

280


*

Loss i n Argon

1.0.

--

A- -L-o-co-0--3 A3-- + T-. . d; F--'-9-* -p- ~4- a--B-0- -H 2-4-

*

b),

YV-

' + "*\

"'":

*

-96

*

3-

.7

' I ' P

'9'

6

I

I

-.

*2 I

L


Loss i n Argon

-4-.

Fig. 4-91 Quotient of Loss Q of the Isothermal Thermogravimetry at 200 "C Curve 2: PMBI1 Curve 1: Bitumen B80/1 Curve 4: PMBl2 Curve 3: Bitumen B80/2 Curve 6: PMBl3 Curve 5: Bitumen B200/3 Curve 7: Bitumen B80/4 Curve 8: PMBl4

-0 6

28 1

282


Time

15

30

45

60

75

90

105

120

135

(mi2 150

Fig. 4-92 Quotient of Loss Q of the Isothermal Thermogravimetry at 250 "C Curve 2: PMB/1 Curve 1: Bitumen B80/1 Curve 4: PMB/2 Curve 3: Bitumen B80/2 Curve 5: Bitumen B200/3 Curve 6: PMB/3 Curve 8: PMBI4 Curve 7: Bitumen B80/4

The three samples which experience weight loss at 165 "C (Q < 1)show that this loss is due to evaporation of non-oxidized parts of the sample (Fig. 4-90). At 200 "C there is no such uniformity (Fig. 4-91). For samples B80/1 and PMB/2 Q is permanently > 1.0, i.e. their considerable losses are due to volatile oxidation products. The values for the PMB/4 are still higher: t (min)

Q

75 90 105 135 150

7.00 4.50 3.33 2.00 1.83


283

At 200 "C the quotient Q of PMB/l, B80/2, B200/3, and B80/4 incrcases with increasing test times, but PMB/1 only exceeds Q = 1.0 after 90 minutes test time. At 250°C only samples PMB/I and B200/3 exceed Q = 1.0, whereas all the other samples have values of Q nearly equal to one (PMB/4) or less than 1.0 (B80/1, B80/2, PMB/3 F, PMB/4, B80/4). (Fig. 4-92).

4.4.3 Reaction kinetics using DSC Arrhenius' equation gives the best description of the dependence of the reaction rate upon the temperature, for pyrolysis (cracking) and oxidation reactions. It is possible to extrapolate the reaction rate constant and the half life time to higher or lower temperatures. Therefore the coefficients of the Arrhenius equation, (activation energy E and the frequency (pre-exponential) factor A) were determined using the method according to ASTM E

Sample B80 / 1 PMB 1 1 B8012 PMB / 2 Peak 1 Peak 2 B 200 I 3 PMB 1 3 Peak 1 Peak 2 B8014 PMB 1 4

Temperature range ("C)

E (kJ1Mol)

log A (mid)

Coke residue

450-490 440-480 440-475 310-370 430-480 440-475 310-370 435-470 430-475 440-475

273.8 239.5 259.3 78.0 268.5 258.7 98.0 258.1 219.7 266.6

19.249 16.942 18.548 6.447 19.177 18.433 8.072 18.404 15.641 18.970

15.75 14.75 15.70

(%I

-

14.04 12.75 -

12.95 15.19 15.42

Table 4-117: DSC oxidation in 7 bar air Temperature range ("C)

Peak 1 275-335

Peak 2 395-455

Peak 3 4 15-480

E (kJ/Mol) B80/1 PMB I 1 B80/2 PMEil2 B 200 1 3 PMB 1 3 B8014 PMB 1 4

102.8

9.134

111.2

10.001

114.4 102.2

8.536 10.234 8.386 9.116

183.8 103.8 187.2 175.1

13.310 7.747 13.481 12.453

157.7

11.913

196.6

14.382

-

-

-

Peak 4 460-540

log A (min-l)

148.6 139.0 125.6 209.0 222.3 191.5 193.6 131.3

10.143 9.885 8.681 14.850 16.088 13.757 13.860 9.050

163.8 187.8 146.2 175.8 134.0 121.5 138.6 123.5

10.594 12.659 9.427 11.871 8.625 7.946 9.347 7.773

284


698-79. Four different heating rates @ = 5, 10, 20, and 50K/min) were applied. The experiments were carried out in argon at 1 bar pressure and in air at 7 bar with a gas flow rate of 5 cm3/min each, using a sample weight between 1.5 and 4.0 mg. The results of the pyrolysis and oxidation experiments are listed in Tables 4-1 16 and 4-1 17. The temperature range includes the values of the reaction maximum temperatures for each of the four heating rates. The coke residue of the pyrolysis reaction was determined after the reaction and after cooling down to room temperature. The results in Table 4-116 show that pyrolysis reactions occur in temperature ranges which are normal for bitumens. The activation energies and the frequency factors are also as expected. The samples PMB/2 and the PMB/3 each have an additional peak, which must be associated with a distillation event, as indicated by its temperature and its activation energy. The activation energy equals the enthalpy of vaporization. The peak of PMB/3 derives from the content of a furfural extract, which was added to swell the polymer. The same assumption may be made for PMB/2, since there is no corresponding peak in the DSC experiment on bitumen B80/2, despite a high evaporation loss during thermogravimetry in argon (AG300 = 9.8 wt%, AG400 = 33.0 wt%). Polymer modification of bitumen B80/1 results in a decrease of the activation energy of PMB/1. The activation energies of PMB/2 and PMB/4 are increased relative to the basic bitumens, whereas the activation energy of PMB/3 is equal to that of bitumen B200/3. This must be an accident, since the production of PMB/3 includes blowing of the mixture of bitumen, extract, and polymer. But for no one product activation energy of the pyrolysis reaction fa11 so low that difficulties in manufacturing of the PMB would be incurred. The half life times of PMB/l and PMB/3 are reduced slightly, and those of PMB/2 and PMBI4 increased slightly, relative to the basic bitumens. DSC oxidation in 7 bar air produce three or four very easily evaluable peaks, which may be related to the three reaction steps: low temperature oxidation LTO (peak 1); fuel deposition (peak 2 and sometimes peak 3); and fuel combustion (last peak). The first peak for bitumen and for PMB appears at temperatures of 275-335 "C depending on the heating rate. The Arrhenius coefficients of PMB/1 and PMB/2 are less than for the basic bitumens, whereas the coefficients of PMB/3 and PMB/4 are higher. The second peak is found between 395 "C and 455 "C and the third peak (if there is an additional peak in the region of fuel deposition) between 425 "C and 480 "C. We could not determine how much overlap there was between the endothermal pyrolysis reaction and the exothermal oxidation, which was due to the combustion of the products from the preceding cracking, although both peaks are of an exothermal nature. The last peak is situated between 460 "C and 540 "C and represents the combustion of the coke which was formed by pyrolysis in the preceding reaction region. The activation energy of the last reaction often corresponds to the activation energy of elemental carbon (activated charcoal E = 154 kJ/Mol). The value of the first oxidation peak is relevant to practical applications since this peak marks the start of oxidation. In the graph of log t1,2 versus 1 000/T, the lines for the bitumen and the corresponding PMB each have a point of intersection, due to the different slopes of the lines, which are defined by the activation energy. This point of intersection for the first oxidation peak appears at nearly identical temperatures for each pair, except B80/ 1-PMB 11

4.4 Invsstigations on polymer modified bitumens (PMB)

Pair B80/1-PMB/1 BS0/2-PMB/2 B200/3-PMB/3 BSOI4-PMB 14

285

Point of intersection ("C) 150 270 250 260

The half life times of bitumen B80/1 and B80/2 are higher than those of the corresponding PMB/I and PMB/2 at temperatures below the point of intersection. The pairs B200/3-PMB/3 and B80/4-PMB/4 demonstrate an inverse behavior. The software for the computation of the Arrhenius coefficients only provides values of the reaction enthalpy incidentally. This enthalpy is always dependent on the initial weight, so the results for the oxidation experiment are not valid. The data for the pyrolysis experiments are listed in table 4- 118: Table 4-118: Enthalpy of pyrolysis. Sample B80/1 PMB/1 B80/2 PMB/2 Peak 1 Peak 2 B200/3 PMB/3 Peak 1 Peak 2 B80/4 PMB 14

Enthalpy (kJ/Mol) 365.4 244.9 232.9 24.4 123.9 319.9 50.5 195.5 152.2 147.3

The low values of peak 1 of PMB/2 and PMB/3 are evident and cannot be associated with a pyrolysis reaction.

4.4.4 Low temperature behavior of PMB using DSC Changes in the crystalline phase, such as melting, crystallization, or glass transitions of polymers and bitumens may be recognized using low temperature DSC. The influences of the polymer additive on the glass transition temperature of bitumen, and the reciprocal influence of the components of bitumens on the crystalline structure of the polymers, may be investigated if the polymer concentration is sufficient. This minimum polymer concentration should be approximately 3 wt%. Fig. 4-93 shows the behavior of a low molecular high density polyethylene HD-PE (M, = 5.900) during heating and cooling. In the heating curve a slight endothermal signal is recognized at 92 'C, which unquestionably represents a

286


0

Cooling

c 5/

E " L

: LL

0

0-

4J

B

I \

-5

-

\

I

\

-------zJ;04'C I


287

memory effect from processing, since this signal is never found in a second heating run of the same sample. At 111 "C the crystalline fraction of the sample (approx. 43 %) melts. Formation of the crystallites is shifted down to 103 "C due to the heating-cooling rate p= 5 K/min, which is quite high for such effects. Fig. 4-94 shows a bitumen B200 which has a glass transition at -23 "C (heating-cooling rate p= 5 K/min). The mixture of this bitumen with 5 wt% HD-PE is shown in Fig. 4-95. The oily phase of the bitumen (the maltenes) acts as a plasticizer for HD-PE and decreases the melting temperature of its crystalline phase to 98 "C. Both the melting and the crystallization peaks became considerably flatter and broader. The glass transition temperature of the bitumen is no longer evident.

2-

2

" 3

2u.

0-

+I

B -2-

-4 -

\

Heating

\______

'\

\

9 a . 1 3 ~

\ -6

-so

I

-40

-20

D

25

.

2

4n

.

I

60

.

I

eo

.

I

ion

.

1 0

Fig. 4-95 DSC HeatingKooling Curve of a Compound Bitumen B200 (95 wt %) + HD-PE (5 wt %) Heating/Cooling Rate p : 5 K/min

4.4.5 Aging properties of polymers for the modification of bitumen Polymers which are used in the modification of bitumen must posses sufficient compatibility, and high thermal stability; they will then be able to resist the high temperatures during the mixing process in the asphalt mixing plant, and during transport, from the

288


mixing plant to the road surface or to a substrate in the production of roofing felts. High oxidation stability, at high temperatures during manufacturing, and at lower temperatures during application, is required. The assumption that an easily oxidizable polymer will protect the bitumen against oxidation is not corroborated by the results of the investigations on bitumen and PMB (chapter 4.4). Oxidation consists of chain reactions of radicals, which attack the bitumen, as well as the polymers and other organic substances [4-371. Peroxide and hydroperoxide radicals formed elsewhere act as chain propagators, and start the oxidation of the bitumen [4-381. Therefore the polymers must be sufficiently stable to oxidation either based on their chemical constitution or on corresponding stabilizer content. Both the pyrolysis and the oxidation stability of six commercial polymers were investigated [4-391. The abbreviations for these polymers and their average molecular weights (number average Mn) are listed in Table 4-119. Table 4-119: Polymers under investigation Sample

Substance

1

Low molecular weight high density polyethylene Ethylene-propylene copolymer Ethylene-prop ylene-norbornene sequence terpolymer Styrene-butadiene radial block copolymer Styrene-ethylene-butene sequence terpolymer Low molecular weight polybutadiene (liquid)

2 3 4

5

6

Abbrevation

M"

HD-PE EPM

5.000 8.000

EPDM

SBS

233.300

SEBS

54.200

BR

3.000

A preliminary review of the thermal and thermooxidative behavior is provided by temperture-programmed thermogravimetry both in argon and in air. The index numbers T1 % and T5 % represent the weight loss due to evaporation of volatile (low molecular) fractions in argon, and the weight loss due to evaporation of volatile oxidation products in air (Table 4-120): Generally the experiments in argon start with a weight loss due to the evaporation of volatile parts. Then follows a more or less rapid depolymerization or pyrolysis of the samples as shown above for polybutadiene (BR) in Fig. 3-24 and for HD-PE in Fig. 3-25. Above 500 "C the polymers under investigation volatilize quantitatively. The DTG curves each show one maximum in the temperature range from 350 "C (evaporation of BR) to 460 "C (pyrolysis). This may indicate that in the absence of air thermal degradation does not occur at customary processing temperatures. TGA in air provides different results. All the saturated hydrocarbons behave in the same way as in argon, at least at lower temperatures. However, compounds which contain reactive double bonds, such as EPDM, SBS, and BR, show an increase of the sample weight at relatively low temperatures, attaining a maximum for the oxidation of BR as


289

Table 4-120: Start temperature of weight loss in thermogravimetry ("C). Sample No.

Substance

1 2 3 4 5 6 7

HD-PE EPM EPDM SBS SEBS BR* Bitumen B80

*

T1 %

T5 %

T1 %

in argon

378 266 239 304 292 310 204-301

T5 % in air

430 352 354 343 335 373 276-360

244 245 237 302 259 333 207-272

318 280 305 345 28 1 423 256-3 16

the oxidation of BR starts at 117 "C with an increase of the sample weight due to resorption of oxygen

shown in Fig. 3-24. This increase of the sample weight is less than that theoretically predicted by the double bond content. It is due to the resorption of oxygen and the formation of non-volatile oxygen compounds, and can be measured clearly in the unsaturated polymers mentioned above. TGA curves of the other polymers, which only contain saturated carbon chains, do not give any information concerning the possible resorption of oxygen. Above the temperature range 240-330 "C all the polymers suffer a continous weight loss, until at approximately 550 "C they have undergone quantitative combustion. The DTG curves of the experiments in air exhibit up to ten satisfactory oxidation peaks. Those which represent the increase of the sample weight of the unsaturated polymers appear from 180 "C to 190 OC, indicating that a more or less severe oxidation of the polymers would take place at the temperatures customary in the asphalt mixing plants. The dependence of changes with test time at lower temperatures has been examined using isothermal gravimetry. In order to reach the test temperatures as rapidly as possible, a heating rate /3 = 100 K/min was used. The samples remained at the preselected isothermal temperature for 150 minutes. The standard gas flow rate of 25 cm3/minwas used for both argon and air. In the experiment in argon, none of the samples suffered any weight loss, at 165 "C or 200 "C test temperature, except BR and EPM. The weight loss at 165 "C was 0.2 % for EPM and 0.3 % for BR, whereas the loss at 200 "C was 1.5 % for EPM and 2.3 % for BR, each after 150 minutes test time. Isothermal aging in air presents a different picture (Table 4-121). EPM is subject to an unexpectedly high weight loss, both at 165 "C (-10.7 %) and 200 "C (-32.1 %). The plot of weight loss against test time showed that weight loss was not even complete after 150 minutes. The behavior of BR is similar; it firstly undergoes a small increase of the sample weight, but has lost 2.0 % after 150 minutes at 165 "C test temperature and 6.5 % at 200 O C , without reaching equilibrium. At 165 "C EPDM starts with an increase of the sample weight, which reaches a maximum after 15 minutes test time. After 150 minutes the weight loss amounts to only 2.3 % but equilibrium has, again, not been reached. At 200 "C weight is lost from the beginning. After 150 minutes the weight loss amounts to 10.1 % with no sign that this is the final stage. SBS, SEBS, and HD-PE were most stable in these experiments. These results

290


Table 4-121: Isothermal gravimetry in air. Alteration of weight (%) Time

HD-PE

EPM

EPDM

SBS

Temperature: 165 "C

5 10 15 30 45 60 75 90 105 120 135 150

0

0

0 -0.1 -0.1 -0.1 -0.2 -0.2

0 -0.9 -1.9 -3.2 -4.4 -5.6 -6.9 -8.1 -9.4 -10.7

I

+0.2 +0.8 +0.9 +0.8 +0.5 +o.1 -0.2 -0.8 -1.1 -1.4 -1.9 -2.3

+1.1 +1.4 +1.8 +2.2 +2.2 +2.2 +2.3 +2.4 +2.4 +2.4 +2.4 +2.4

SEBS

BR

B 80

9

+0.5

i;;

0

0 -0.1 -0.1 -0.2 -0.2 -0.3 -0.3 -0.3

-0.1 -0.3 -0.6 -0.9 -1.3 -1.5 -1.7 -2.0

0 -0.1 -0.2 -0.3 -0.4

1

+0.5

Temperature: 200 "C

5 10 15 30 45 60 75 90 105 120 135 150

0 -0.1 -0.1 -0.2 -0.3 -0.4 -0.5 -0.7

-0.2 -1.7 -3.2 -7.8 -11.9 -14.8 -18.7 -21.7 -24.5 -27.0 -29.4 -32.1

0 -0.6 -1.1 -2.3 -3.4 -4.7 -5.7 -6.7 -7.8 -8.6 -9.4 -10.1

+1.1 +1.1 +1.1 +0.9 +0.6 +0.2 0 0 -0.3 -0.7 -1.0 -1.1

0 -0.1 -0.1 -0.1 -0.2 -0.3 -0.4 -0.5 -0.7

1::: +0.7 0 -1.6 -2.3 -3.1 -3.9 -4.8 -5.6 -6.5

-0.6 -0.7 -0.7 -0.9 -1.0 -1.1 -1.3 -1.6 -1.9 -2.1 -2.6 -2.8

indicate that EPDM and BR exhibit quite insufficient stabilization, and that EPM shows a complete absence of any oxidation-stabilizer content. The kinetics of pyrolysis and oxidation reactions were investigated according to ASTM E 698-79, applying the standard experimentalconditions.The DSC plot of the pyrolysis of EPDM (heating rate p= 5 K/min) is given as an example in Fig. 4-96. For each of the polymers only one peak maximum of the endothermal heat flow in the pyrolysis experiments in argon was confirmed, and this was between 430 "C and 490 OC, depending on the heating rate. Only SBS has an additional exothermic peak maximum between 360 "C and 390 'C, which indicates that thermal crosslinking took place prior to the pyrolysis (Fig. 497). The plot of log tl,2versus 1 000/T shows that all of the polymers have thermal stability exceeding the temperatures customary for processing bitumen or polymers. For example, the extrapolated half life time at 300 "C exceeds a value of lo3 minutes for all of the polymers tested.


0-

2"

-2-

-*

U. 0

P

I

-4-

-5

-8

-

1

0

1on

zoo

I

300 Tempordturo C°C)

Fig. 4-96 DSC Pyrolysis of EPDM Heating Rate p: 5 K/min Atmosphere: Argon 5 cm3/min

400

500

6 1

291

292

4 Thermoanalytical Investigations on Petroleum and Petroleum Products 4

4

-6 0

100

200

Temperature 300 (DC)

400

500

0

Fig. 4-97 DSC Pyrolysis of SBS Heating Rate p : 10 K/min Atmosphere: Argon 5 cm3/min

DSC in air give three to five maxima of exothermal energy flow, as Fig. 4-98 shows for the example of SBS (heating rate p = 10 K/min). The first peak at approximately I88 "C represents the low temperature oxidation, LTO, where oxygen-containing, low-volatile products are produced, which are responsible for the increase of the sample weight in thermogravimetry. The first maximum of the DSC curve for this sample was found at 189 "C. The second peak at 363 "C and the third peak at 447 "C are associated with the region of fuel deposition. The coke residue which was formed in this region of reaction will subsequently be combusted at temperatures around 490 "C (last peak). The coefficients of the Arrhenius equation (activation energy E and frequency factor A ) may be computed for each step of the reaction (Table 4-122). The LTO region is of practical relevance for oxidation effects in mixing and manufacturing processes, and for long-term oxidation stability, since any oxidation reaction would start there. Taking the half life time t1,2as a criterion for oxidation stability, then the plot of log tlizversus 1 OOO/T (Fig. 4-99) may be interpreted as follows:

4.4 Investigations on polymer modified bitumens (PMB) I20

100-

ao-

60-

363.50-C

lea. 550c

40 -

20-

0-

-20 0

Fig.

100

I

200

300

8

400

I

500

600

293

294


4.4 Investigations on polymer mod& !d bitumens (PMB)

250

20

1,50

Fig. 4-99 Low Temperature Oxidation of Polymers (First DSC Peak) Half Life Time t,,2 versus Temperature Line 1: HD-PE Line 4: SBS Line 5: SEBS Line 2: EPM Line 6: BR Line 3: EPDM Line 7: Bitumen B80 Line 4: EPDM

295

296


The half life time at 200 "C for EPM, EPDM, SBS, and BR is less than one minute. For HD-PE and SEBS the half life times range from eight to twenty minutes. Thus, any of these polymers might be oxidized to some extent during the production of asphalt (bitumen/ mineral) mixtures or roofing felts. Extrapolation to lower temperatures shows that EPM and BR still have low oxidation stability, while the other polymers are sufficiently stable, but they all have lower LTO stability than bitumen. During the oxidation process long-life radicals are formed, which are mesomeiically stabilized. These radicals can induce the oxidation of bitumen binder in the road surface or roofing felt, even at low temperatures under the influence of UV radiation. Moreover temperatures up to 60 "C or 80 "C may be reached at the surfaces of roads or roofing felts as a result of the IR radiation in sunlight. However, the energetic UV rays can only penetrate the surface for a few micrometers due to the dark color of the bitumen. The oxidation stability of the polymers is inferior to that of the bitumen. Most of the polymers which are used in the plastics industries are more or less adequately stabilized against oxidative attacks by the manufacturer. Polymers with reactive double bonds, which are for the most part used in the rubber industry, should be protected against oxidation by those antioxidants which are in common use in that industry. Easily oxidizable polymers cannot protect bitumen against oxidative attacks. The radicals which form during the oxidation of the polymers, may also start chain reactions in the bitumen. Effective protection against oxidation may only be achieved using radical scavengers or suitable antioxidants, which terminate the radical chain reaction or deactivate the free radical starters.

4.5 Investigation on the hydrocracking reaction of heavy residues Hydrocraclng of heavy refinery residues comprises a series of parallel and/or consecutive reactions, which cannot be analyzed in isolation. A model of the reaction may be made by substituting analogous substances for the process intermediates. The first step, which precedes hydrogenation, is pyrolysis which may be investigated using thermoanalytical methods, such as TGA and DSC. Depending on the reaction conditions used, varying quantities of reaction products result from hydrocraclung: gases, high-volatile liquid products (e. g. naphtha with a boiling limit of 200 "C), low-volatile heavy oils, and nonboiling solid residues. The gases foimed and the high-volatile products can be analyzed by conventional methods, but the non-boiling residues and the low-volatile heavy oils cannot be separated by distillation. In this case index numbers, which can be correlated with the reaction conditions, can be found using thermoanalytical methods. Two basic investigations will be described here. In both experimental series, a batch reactor with a volume of 108 cm3 was used for the hydrocracking reaction. The first

4.5 Investigation on the hydrocrucking reaction of heavy residues

297

experimental series was carried out on one vacuum residue (VR Kirkuk), using a cold hydrogen pressure of 120 bar and applying two different reaction temperatures (435 "C and 455 "C). The aim was to test the application of thermoanalytical methods in this field of investigation and to acquire information concerning the influence of the reaction temperature and the residence time on the yields of the different products. The residence time was varied from 0 to 100 minutes [4-40,4-411. Zero residence time means that the system was quenched immediately, as soon as the pre-selected temperature was attained in the reactor. The second, more extensive, investigation on nine vacuum residues (VR) and two derived visbreaker residues (VVR) used reactions at three different temperatures (410 OC,440 OC, and 460 "C), with a constant residence time of 30 minutes, and cold hydrogen pressure of 90 bar. The aim of this experiment was to investigate the influence of the origin of the eleven residues, from five different oil regions, upon their reaction behavior [4-421. The products of the reactions, that is, the gases, high-volatile oil (naphtha), low-volatile heavy oil, and solid residues, were isolated after the reaction (after quenching) for analysis. The isolated heavy oils were investigated using thermoanalytical methods. In addition the kinetics of the pyrolysis reaction were investigated in the second series of experiments.

4.5.1 Investigations on a vacuum residue from Kirkuk The vacuum residue from Kirkuk was characterized by elemental analysis, average molecular weight (vapor pressure osmometry), viscosity, density, asphaltenes content, and the usual index numbers of thermogravimetry (table 4-123): Table 4-123: Analysis of vacuum residue Kirkuk. Element analysis: H (%I c (%I H/C* s 11.27 85.27 1.59 2.76 *: atomic ratio Average molecular weight: 1 600 g/Mole Viscosity: 31 376 mm2/s (50 "C); 2 333 mm2/s (70 "C) Asphaltenes: 1.3 % (Cyclohexane precipitation) Density: 0.96 g/cm3 Thermogravimetry: T1 % T5 % ("C) ("C) 300 360

AG300 (wt%) 1

AG400 (wt%) 12.9

N (%I 0.1

R800 (wt%) 10.4

DTG (%/min) 5.26

0 (%I 0.6

Tm ("C) 460

298


The quality of the heavy oil produced bears a direct relation to the addition of hydrogen during the hydrocracking reaction. The hydrogen addition (wt%) was calculated according to equation 4-1: (Eq.4-1) where:

H2add= addition of hydrogen (wt%) Pstat = pressure of H, prior to the reaction (cold pressure) (bar) Pend = cold pressure after the reaction (bar) = volume of hydrogen in the gas after the reaction (~01%) = residual gas volume in the reactor prior to the reaction (1) T = room temperature (K) = molecular weight of hydrogen (g1Mole) = weight of vacuum residue prior to the reaction (g)

%

The results of the element analysis and the average molecular weight of the heavy oil produced, as well as the consumption of hydrogen in relation to the selected reaction parameters are listed in Table 4-124. The results of thermogravimetry of the heavy oils produced is shown in Table 4-125. Fig. 4-100 shows the plots of T5 YO versus residence time and Fig. 4-101 the plot of AG400 and R800 versus residence time, each at two different temperatures. Depending on the reaction conditions, T5 % varies from 90 "C up to 220 "C. The evaporation start temperature is strongly influenced by the reaction temperature, as Fig. 4-100 clearly shows. Heating up to the reaction temperature influences the results: a difference in the reactor temperature of 20 "C (435 "C and 455 "C) results in a decrease of the evaporation start temperature by 65 "C. The value at zero residence time is taken on quenching after the Table 4-124: Element analysis, molecular weight and hydrogen reception of heavy oils produced by hydrocracking.

No.

Reaction Parameters PH2/ Tlt bar1"Clmin

-

M

C

H

gm01

%

%

86.1 89.5 86.1 86.9

11.47 11.05 11.12 11.24 11.26

1.60 1.54 1.55 1.56 1.56

0.17 0.26 0.48 0.60 0.64

86.7 87.0 86.9 86.8 87.6

11.32 11.0 10.83 10.20 9.70

1.57 1,51 1.50 1.41 1.33

0.22 0.50 0.86 1.o 1.30

HIC

H2 iecepted

%

~

1 2 3 4 5

120/43510 1201435I15 1201435130 120/435I60 12'01435I90

6 7 8 9 10

120145510 l20l455l15 1201455130 1201455160 1201455190

86.6

~

-

M: Average molecular weight

620 450 323 427

4.5 Investigation on the hydrocracking reaction of heavy residues

299

Table 4-125: Thermogravimetry of the products from hydrocracking experiments (Argon 25 cm3/min, p = 10 Klmin). Hydrocracking

__

No.

1

Therinogravime try

Reaction Parameters T1 %

PlTlt bar1"CImin

"C

T5 %

R

AG

DTG

100

200

300

400

600

800

"C

%

%

%

%

%

%

222 154 135

0.3 1.0 1.8

3.7 10.4 18.3 10.6 26.7 49.0 17.7 42.0 66.4

13.3 11.9 12.8 11.8 7.8 7.0 10.7

10.0

1 2 3

120143510 123 120/435/15 100 120/435/30 84

4

120/435/60

62

94

5.9

21.8

5

1201435190

73

107

3.7

34.1 58.5

67.4

13.6

7.1

6 7

1201455l0 106 1201455/15 80

157 107

0.8 3.2

9.2 22.9 28.8 47.6

44.0 65.3

12.0 13.9

11.1 11.0

8

120/455/30

67

100

1.0

40.0

63.9

79.2

7.0

6.5

9

120/455/60

66

94

6.6

38.4

57.6 73.5

7.3

6.8

10

120/455/90

83

120

2.2

23.5

42.4

7.1

7.0

-

54.6 71.8

T,

%/min

"C

10.21 5.79 2.34 3.96 1.97 1.95 3.11 4.00 2.17

456 450 249 446 215 447 157 410 452

6.51 2.50 3.33 3.19 2.00 2.95 2.70 2.74 548

459 145 44 1 147 438 140 447 167 454

I

59.2

pre-selected reactor temperature had been reached. This difference decrease with increasing residence times. The curves for 435 "C and 455 "C reaction temperatures have a point of intersection at 65 minutes residence time, which is also the minimum of the evaporation start temperature. Above 65 minutes residence time the slopes of both curves reverse. This is because of the increasing tendency of the radicals to recombine when the quantity of hydrogen is no longer sufficient to saturate the radical fragments. If we remember that T5 % for the vacuum residue itself is 360 "C, and that the real boiling temperature is about 100 "C higher, we must attribute weight losses in thermogravimetry, above an instrument temperature of 400 OC, to pyrolysis. The curve of AG400 from the experiment at 435 "C versus residence time (Fig. 4-101) is comparable to that of T5 % versus residence time. This implies that an improvement occurs in the quantity and quality of the heavy oil produced, with increasing residence time at lower temperatures, since the quantity of the distillable fraction AG400 increases whereas the evaporation start temperature decreases. On the other hand a drastic decrease of A6400 occur at reaction temperature 455 OC, when the residence time is > 30 minutes, whereas an increase of T5 % starts at 60 minutes residence time. An explanation might be that, at a certain residence time, naphtha starts to form from the heavy oils, since there is a limited quantity of compounds within the vacuum residues which can be hydrogenated. This assumption is corroborated by the constancy of

300


0

100

P2

3

10

20

30

40

50

60

70 80 90 R e s i d e n c e Time ( m i n )

Fig. 4-100 Thermogravimetry of Reaction Products from Hydrocracking. Vaporization Start Temperature T.5 % of the Heavy Oil Fraction versus Residence Time Curve 1: Reactor Temperature 435 "C Curve 2: Reactor Temperature 455 "C according to [4-411

the total oil formation i.e. the sum of naphtha and the distillable fraction AG400 [4-401. This kind of secondary hydrogenation of the heavy oil results in a decrease of AG400 but only has a minor influence on T5 % at first. Later on the recombination of radicals leads to an increase in T5 %. This might provide a plausible explanation for the time lag between the points of inflexion of the curves of T5 % and AG400 versus time (Fig. 4-100 and 4-101). Comparison of R600 and R8OO shows that, with one exception, there is only a difference of 1 or 2 %, which implies that, above 600 "C, hardly any degradable material is left. The coke residue decreases slightly with increasing residence time, but the amount of coke is small compared to that of the vacuum residue. This indicates that the high-molecularweight material does not undergo conversion under the reaction conditions of the hydrocracking process. The maximum of the weight loss rate, DTG, is a useful indicator for the reaction kinetics at the corresponding temperatures, Tm. Below 400°C Tmx represents the kinetics of substance transportation, whereas above 400 OC, it represents the reaction kinetics of pyrolysis. The DTG of 5.3 %/min with a T-. of 460 "C for the vacuum residue definitely lies in the reaction lunetic region. The products from hydrocracking experiments exhibit DTG peak maxima at very varying temperatures, and sometimes there is more than one maximum. In Table 4-126 the values of the DTG maxima are listed and associated with the

301


:{ /I' /

20

lUh-

A

= 7

L

I

I

I

-

433

0 I

4

I

I

455 "C

435 " c

Transportation kinetics

Reaction kinetics

DTG (%/min)

Tmax ("C)

t* (mid

DTG (%/min)

Tm ("C)

0 15 30 60 90

-

-

-

-

-

-

2.34 1.97 3.11

249 215 157

2.50 3.19 2.95 2.74

145 147 140 167

0 15 30 60 90

10.21 3.96 3.96 1.95 2.17

456 446 446 447 452

6.51 3.33 2.00 2.70 5.48

459 44 1 438 447 454

t*: Residence time for the hydrocracking

302

4 Theimoanalytical Investigations on Petroleum and Petroleum Products

residence times. The rather mild reaction temperature of 435 "C does not produce any data for transportation kinetics at the beginning, notwithstanding the existence of some evaporation losses (AG100-AG300). Maxima of the weight loss rate DTG were first observed at 30 minutes residence time, which may be related to the increasing quantitiy of evaporated substance (AG300). The peak maximum temperatures Tmaxin the region of transportation kinetics decrease with increasing residence time whereas the maximum of the weight loss rate, DTG, passes through a minimum value at a residence time of 60 minutes. At the higher reaction temperature of 455 "C the maxima of the weight loss rate DTG in the transportation region increase slightly with increasing residence times. The start of evaporation is earlier and evaporation clearly increases. The rise of the reaction temperature leads to decrease of Tlmxin the region of transportation kinetics, which implies the formation of lower-boiling substances. No further distinct relationship of the temperature with the residence time is evident. In the region of reaction kinetics, and for experiments at 435 O C , heating up to the reaction temperature only yields twice the weight loss rate for the heavy oil produced, compared to the vacuum residue. The maxima of the weight loss rates DTG decrease drastically with increasing residence time. Nevertheless the reaction temperature of 455 "C gives considerably less reduction of the weight loss rates DTG with reagard to the increasing residence times, indicating that the reaction starts very early. Independent of the experimental parameters, Tmx varies in a relatively small range between 440 "C and

I

T5X X

25.

A3

/-

x

Y

-

H2-Consumption ( w t X )


303

460 "C. The highest value of the peak maximum temperature has been found to be 460 OC, which is regarded as the upper limit for all the heavy oils, and which is also T,, of the vacuum residue. The maximum value of DTG depends on the absolute quantity of transportable substance at the temperature T-. For equal quantities of substance the maximum value of DTG depends on the range of the boiling or crack temperature. Therefore the size of the DTG peak and its maximum can also be an indicator for the homogeneity of the boiling or reacting substance. The quality of the heavy oil depends on the intensity of hydrogenation, therefore the index numbers from thermogravimetery are expected to depend on the consumption of hydrogen. Fig. 4-102 shows the plot of T5 % and AG400 both versus hydrogen consumption. Two curves which are nearly mirror images are produced for all temperatures and all residence times. The evaporation start temperature T5 % passes through a minimum at a hydrogen consumption of 0.8 %, whereas the distillable fraction AG400 passes through a maximum at that point. The further consumption of hydrogen is explained by the conversion of heavy oils to naphtha and gases. The average molecular weight 2 can be correlated with data from thermogravimetry as shown in Fig. 4-103. Correlation of T5 % with M is given in equation 4-2:

T5 % = 0.2 M

200

+ 33.1

Eq. 4-22

-

175 150

-

--

-- -

Rolecu ar Weigh

-4G400 -0

( M e a n r

I

350 4 0 0

500

600

700

800

900

Fig. 4-1 03 Thermogravimetry of Reaction Products from Hydrocracking. Vaporization Start Temperature 7'5 % and Distillable Part AG400 of the Heavy Oil Fraction versus Mean Molecular Weight. according to [4-411

304


The coefficient of correlation is r = 0.98. A similar correlation of the weight loss up to 400 "C, AG400, with the average molecular weight @ also exists (equation 4-3): AG400 = -8.93 ;i? + 102.5 r = 0.97

Eq. 4-3

For both correlations very high coefficients r were computed, which confirm the relationships. The linear equations permit a rough estimate of the thermal behavior if the molecular weight is known and vice versa. Combination of the equations 4-2 and 4-3 gives the formal relation of AG400 with T5 % (equation 4-4). AG400 = -44.65 T5 % + 1 580.4

Eq. 4-4

4.5.2 Investigation on residues of different origins Eleven different distillation residues from five oil regions were used for investigations on the hydrocracking reaction. Vacuum residues (VR) and a visbreaker residue (VVR) produced from each were available from a Mexican and a Libyan crude. The residues and their origins are listed in Table 4-127; their analytical data is given in Table 4-128. It is evident that the content of heteroatoms varies considerably depending on the origin of the samples, whereas the atomic H/C ratio exhibits only small differences (mean value X = 1.415, standard deviation k s = 0.022 equals a coefficient of variation k V = 1.56 % relative). All experiments were once again carried out in the batch reactor with 108 cm3volume, and cold pressure of hydrogen of 90 bar. The residence time was constant at 30 minutes, Table 4-127: Residues under investigation ~~~~

~

Sample No.

Oil region

Field

Nature of the sample

1 2 3 4 5 6 7 8 9 10 11

Mexico Mexico Libyan Libyan Middle East Middle East Middle East Venezuela Venezuela Canada Canada

Isthmus Isthmus Zueitina Zueitina Arabian Medium Arabian Heavy Kuwait Tia Juana Morichal Cold Lake Marguerite Lake

Visbreaker residue (VVR) Vacuum residue (VR) Visbreaker residue (VVR) Vacuum residue (VR) Vacuum residue (VR) Vacuum residue (VR) Vacuum residue (VR) Vacuum residue (VR) Vacuum residue (VR) Vacuum residue (VR) Vacuum residue (VR)

* Cyclohexane precipitation ** Conradson coke residue

6 7 8 9 10 11 -

-

1.0369

-

-

-

-

0.9866 0.9792 0.9864 0.9826

1 2 3 4 5

Isthmus VVR Isthmus VR Zueitina VVR Zueitina VR Arabian Medium VR Arabian Heavy VR Kuwait VR Tia Juana VR Morichal VR Cold Lake VR Marguerite Lake VR

Density [g/cm3]

Sample No.

16.2 10.7 14.0 6.3 12.7 0.75 18.60 0.92 12.3 19.7 20.8

1045 975 1107 876 837 892 872

Asphaltenes* [wt%]

785 938 994 1276

[g/Mol]

a

Table 4-128: Analytical data of the original samples (feed).

21.02

20.86 21.65 22.04 21.42 20.53 20.28

24.93 24.0 28.88 18.17

CCR** [wt%]

84.03

84.28 84.69 83.81 84.60 84.46 84.08

86.27 85.82 86.85 85.68

C [wt%]

9.80

10.10 10.26 10.09 9.91 9.89 10.07

10.03 10.25 10.11 10.43

H [wt%]

0.18

0.16 0.36 0.42 0.46 0.90 0.18

0.44 0.42 0.60 0.58

N [wt%]

5.69

4.95 4.15 5.01 3.21 4.84 5.20

2.70 2.64 2.21 2.54

S [wt%]

0

0.30

0.51 0.54 0.73 1.82 0.70 0.47

1.06 1.04 0.79 0.58

[wt%]

1.39

1.43 1.44 1.43 1.40 1.40 1.43

1.39 1.42 1.39 1.45

H/C atomic

Q

4

a

as

$

F:

$ 2

5

8

$ .1'

3

g.

0s

$,

?

L,

306


fixed from reaching the pre-set nominal temperature (410 "C, 440 "C, and 460 "C). After quenching the gases, the low-boiling oil (final boiling point 200 "C), and the solid residue were separated using published methods [4-431. The isolated product which still contains the asphaltenes is defined as heavy oil and the fraction with a final boiling point of 200 "C as light oil. The distributions of the various products as a result of the reaction temperature are listed in Tables 4-129 to 4-131. Table 4-129: Distribution of products from hydropyrolysis at 410 "C (wt %) Sample No.

Heavy Oil

Light Oil

Residue

Gases

Asphaltenes*

95.19 94.31 93.46 93.50

2.05 3.23 4.57 4.21

1.22 1.07 0.98 0.82

0.33 0.30 0.34 0.32

12.70 4.90 13.30 0.80

0.20 0.16 0.17 0.16

91.36

5.81

0.79

0.44

4.30

0.18

91.81 91.60 90.62 87.63 88.56

6.45 5.03 7.0 9.19 8.39

0.77 0.68 0.70 0.81 0.81

0.43 0.39 0.37 0.58 0.59

2.70 4.60 3.10 10.40 9.30

0.17 0.20 0.20 0.22 0.30

88.36

8.18

0.75

0.50

9.80

0.33

~~~~~~

6 7 8 9 10 11

*

Isthmus VVR Isthmus VR Zueitina VVR Zueitana VR Arabian Medium VR Arabian Heavy VR Kuwait VR Tia Juana VR Morichal VR Cold Lake VR Marguerite Lake VR

~~

H, Consumption ~

~~

Component of the heavy oil

Table 4-130: Distribution of products from hydropyrolysis at 440 "C (wt %). Sample No.

1 2 3 4 5 6 7 8 9 10 11

*


Heavy Oil

Light Oil

Residue

Gases

Asphaltenes"

H, Consumption

73.46 7 1.05 72.76 73.48

17.88 21.56 17.38 21.49

6.52 6.09 8.24 4.02

1.37 1.49 1.37 1.31

16.50 15.30 17.70 11.40

0.46 0.49 0.44 0.41

64.92

25.94

6.61

1.64

13.00

0.47

64.92 65.65 62.00 58.74 59.47

24.56 25.17 28.91 29.60 29.46

7.83 6.53 8.09 8.31 6.70

1.62 1.68 1.57 2.15 2.38

12.70 13.60 12.40 14.00 14.00

0.46 0.50 0.45 0.56 0.68

61.31

27.24

6.39

2.61

14.90

0.68



307

Table 4-131: Distribution of products from hydropyrolysis at 460 "C (wt %) Sample No.

6 7 8 9 10 11

*


Heavy Oil

Light Oil

Residue

Gases

Asphaltenes*

H, Consumption

49.48 40.77 44.85 49.20

32.11 4 1.99 34.27 38.04

15.31 14.45 18.32 10.12

2.13 3.06 2.80 3.35

14.20 10.30 13.20 9.60

0.75 0.83 0.73 0.83

43.24

38.88

12.94

3.93

10.10

0.88

43.90 44.31 42.78 44.78 43.95

35.74 37.90 36.91 35.73 37.76

14.52 13.17 14.62 12.80 10.80

4.38 3.84 3.58 4.24 4.59

10.90 9.90 9.30 11.30 11.90

0.76 0.87 0.86 0.95 1.10

43.95

38.80

11.05

4.47

12.80

1.08


The content of heavy oil decreases with increasing temperature whereas the content of light oil increases, as desired. Unfortunately the undesired by products such as gases and residues also increase due to increasing temperatures. At equal reaction temperatures the distribution of the products is related to the origin of the samples. Generally the hydrogen consumption rises, with increasing reaction temperatures and is also dependent on the origin. The original samples and the heavy oils produced therefrom were investigated using thermoanalytical methods. Thermogravimetry in inert atmosphere, of the original samples and the heavy oils produced, gives graphs which are recognizable from the investigation on other heavy residues (see chapter 4.2). Calculation of the simulated distillation therefrom results in the distillation curves, which are shown in Fig. 4-104, for the example of Arabian Heavy VR and the three heavy oils produced at the three different reaction temperatures. The index numbers of thermogravimetry in inert gas of the original samples and the corresponding heavy oils are listed in the Tables 4-132 to 4-134. For all the products from the experiments at 410 "C and 440 "C reaction temperature, the evaporation start temperatures (T1 % and T5 %) and the maxima of the weight loss rate, DTG, are reduced, whereas the distillable fractions, AG400 increase. Consequently the crackable residue (CR) also decreases with increasing reaction temperature. At the reaction temperature of 460 "C the trend is partly inverted. The evaporation start temperatures of some products do increase again and the distillable fractions decrease. We may conclude that for the samples Isthmus VR, Kuwait VR, Tia Juana VR, Cold Lake VR, and Marguerite Lake VR the optimal reaction temperature has already been exceeded, which would deliver the desired product qualities. This is analogous to the results published by P. Zhang et al. [4-411, which also describe the recovery of the rise in the evaporation start temperatures with increasing reaction temperatures.


800

1

BP

("C)

700.

600

500

Feed

400

410°C 440°C 460'C

300

Loss ( w t Z ) ,

200

10

20 3 0

40

50 60 70 80

90

.

100

Fig. 4-104 Simulated Distillation (TG 750) * Feed Arabian Heavy VR x Heavy Oil Fraction, Reactor Temperature 410 "C 0 Heavy Oil Fraction, Reactor Temperature 440 "C 0Heavy Oil Fraction, Reactor Temperature 460 "C

7 8 9 10 11

6

1 2 3 4 5

No.

Sample


~

140 107 127 144 118 111

224

124

263 254 263 271 201 229

127 134 129 143

86

113 87 95 98 95

103

94 95 100 89

106

105 89 105 90 100

100

87 104 97 78

410°C 440°C 460°C

177 192 270 283

Feed

T1 % ["C]

283

312 339 325 263 277

329

263 314 348 350

Feed

167

210 175 192 169 169

193

204 218 216 224

123

154 127 133 137 132

144

134 142 143 142

147

143 130 147 129 136

139

122 148 131 116

410°C 440°C 460°C

T5 % ["C]

Table 4-132: Index numbers from thermogravimetry of orginal samples (feed) and heavy oils. Evaporation start temperatures and distillable fraction.

30.5

20.4 15.5 20.8 32.5 32.6

18.0

24.9 17.5 15.5 15.1

Feed

49.2

35.2 40.9 41.2 51.5 49.0

34.9

31.5 27.2 27.7 28.5

62.3

55.8 55.8 62.8 61.3 59.7

52.9

52.6 50.6 48.0 54.8

56.3

55.8 51.4 52.3 60.5 61.3

50.1

52.1 54.6 49.5 55.7

410 "C 440°C 460°C

A G 400 [wt%]

0 Q

W

4

R

5

f:

2

$

5

$

s

%

8

3

5

%

%,

3

cr,

.a

1 2 3 4 5 6 7 8 9 10 11

Sample No.

Isthmus VVR Isthmus VR Zueitina VVR Zueitina VR Arabian Medium VR Arabian Heavy VR Kuwait VR Tia Jnana VR Morichal VR Cold Lake VR Marguerite Lake VR

5.88 6.77 6.80 7.97 4.08 4.58 4.76 4.62 3.78 3.86 3.29

7.25

1.86

2.57 2.81 2.86 2.82 2.99 2.53 2.94 1.91 2.20 1.71

DTG (%/min) 410°C 440°C

5.06 6.97 8.85 10.49 6.91 8.79 9.86 7.99 6.61 6.73

Feed

3.02

3.25 3.23 3.92 2.62 2.26 2.46 3.34 2.75 2.06 2.83

460°C

473

478 458 462 477 475 417 476 417 470 472

Feed

462

485 484 472 473 473 47 1 455 411 462 466

460

412 473 472 469 477 462 460 456 462 467

440 "C

Tux ("C) 410 "C

Table 4-133: Index numbers from the thermogravimetry of original samples (feed) and heavy oils. Maximum of the weight loss rate DTG and corresponding peak maximum temperature in the cracking region.

473

472 477 412 452 450 440 460 471 458 463

460°C

R

ad

9

E

h;

3

2

&

a

5

h;

3

s2

2

h 0.

5'

23.4 18.0 17.6 17.5 18.8 20.7

15.2 15.3 17.7 17.0 15.6 15.5

7 8 9 10 11

4 5 6

20.2 16.9 20.8 13.7 15.0

31.7 26.7 22.3 18.5 17.0

1 2


Feed

15.8

22.8 18,8 16.2 16.8 19.7

19.7 19.4 20.2 15.2 16.9

14.8

25.2 20.1 26.2 17.1 13.0

17.2 14.6 16.2 20.2 25.1

410 "C 440 "C 460 "C

R 600 [wt %]

No.

Sample

19.0 15.8 19.9 12.2 13.5 21.2 14.8 16.2 15.8 17.2 19.8

14.3 14.2 15.9 15.8 14.5 14.4

14.8

21.6 16.3 15.1 15.5 18.6

18.4 18.3 19.2 14.3 15.7

14.1

22.1 17.6 21.1 16.2 12.0

16.2 13.9 15.3 18.2 21.6

410°C 440°C 460°C

21.5 18.2 20.9 9.7 15.0

Feed

R 800 [wt %]

Table 4-134: Index numbers from thermogravimetry of original samples (feed) and heavy oils. Coke residues and crackable part.

43.6 44.3 42.6 32.7 33.8 31.0

52.3

49.5 57.0 52.4 59.3 51.6

21.9

22.6 27.9 22.1 23.2 21.7

29.0 31.1 32.8 30.9 31.4

29.6

22.1 31.0 26.6 23.0 26.7

31.7 31.5 35.2 26.1 28.3

410°C 440°C 460°C

65.3 70.3 63.3 51.7 50.2

53.6 64.3 63.6 75.2 67.3

Feed

CR [wt %]

n

?i

%

$.

a

$ 2

5i

ff

8.s

0s

5.

L,

3 12


The curves of simulated distillation for the products from the experiments at 410 "C already show the existence of considerably lighter products compared to the original samples. The products of the experiments at 440 "C and 460 "C temperature are still lighter, but their evaporation behavior is the same. The coking reaction is not complete at 600 "C as discussed above [4-12,4-161. That is also true for the residues at 600 "C (12600). This may clearly be demonstrated for the vacuum and visbreaker residues of the Mexican crude Isthmus, where the residue at 800 "C is only two thirds of that at 600 "C. The heavy oils of these samples exhibit considerably smaller differences between R600 and R800, which is also true for the other original samples and their reaction products. Scarcely any difference can be observed between residues R600 and R800. The quantity of residue does not bear a simple relationship to the reaction temperature; both increasing and decreasing values of R600 and R800 were observed as the reaction temperature increased. The increase of R600 and R800 for the heavy oils produced indicates that extensive hydrogenation of the components of the samples which are subject to hydrocracking, has occurred and that the coke-forming content is enriched. The visbreaker residues contain less of the crackable substances (CR) than the corresponding vacuum residues. The difference of the CR of approximately 10 wt% between the two residues shows that the visbreaking process uses relatively mild crack severity. The CR content is than decreased drastically during the hydrocracking experiments. Whereas the reaction temperature of 410 "C is still a rather mild crack severity, 440 "C represents the optimal reaction temperature for most cases. The differences of the CR from the experiments at 440 "C and 460 "C are less than those from the experiments at 410 "C and 440 "C. Some samples (Isthmus VVR, Zueitina VVR, Kuwait VR, Tia Juana VR, Cold Lake VR, and Marguerite Lake VR) gave a higher CR in the experiments at 460 "C. This might be a consequence of the recombination of preformed radicals if the supply of hydrogen is insufficient or the radical recombination may have higher reaction rate than that of the hydrogenation. The maximum of the weight loss rate DTG is an indicator for the reaction kinetics. The peak maximum temperature Tmxmay give the preliminary information as to the nature of the reaction. A Tmx < 400 "C corresponds to the transportation kinetics of evaporating substances and Tmx> 400 "C corresponds to the reaction kinetics of pyrolysis. Each of the original samples exhibits a Tmax between 460 "C and 480 OC, which is unquestionably in the region of reaction hnetics. The absolute values of DTG scatter unsystematically between 5.0 and 10.5 %/min. The DTG of the visbreaker residues are considerably smaller (from 15 to 28 % relative) than those of the corresponding vacuum residues, which also indicates less content of crackable substances. Each of the products from the experiments at 460 "C exhibit two DTG peak maxima (with only one exception), which appear in the region of transportation kinetics at varying temperatures. The absolute value of DTG is rather slow (approximately 2.0 %/min). On the other hand, only five products out of eleven, from the experiments at 440 "Ctemperature show one peak maximum in the region of transportation kinetics (Arabian Heavy VR; Tia Juana VR, Morichal VR, Cold Lake VR, and Marguerite Lake VR). The heavy oils produced from the experiments at 410 "C do not have a DTG maximum below 350 "C temperature (Table 4-135).

313


Table 4-135: Maximum of weight loss rate DTG and corresponding peak maximum temperature in the evaporation region Sample No.

1 2 3 4 5 6 7 8 9 10 11

410°C Isthmus VVR Isthmus VR Zueitina VVR Zueitina VR Arabian Medium VR Arabian Heavy VR Kuwait VR Tia Juana VR Morichal VR Cold Lake VR Marguerite Lake VR

DTG (%/min) 440°C 460°C -

410°C

T,, 03 440°C

-

305

-

-

-

295 189 352 301 247 360

2.32 2.40 2.02 2.11

-

-

2.26 1.85 2.20 1.71

1.98 1.76 2.16 1.81 2.28

336

338 29 1 154

1.86

1.88

-

313

-

460°C

-

In the region of reaction kinetics, a clear decrease in the DTG values of the heavy oils produced was observed with increasing reaction temperature, with the exception of some products from experiments at 460 "C. Also, an increase of the peak maximum temperatures T,, was discovered, mostly due to increasing reaction temperature. The Tm values range between 440 "C and 480 "C. As mentioned above, the height of the DTG peak depends on the homogeneity of the substance, which evaporates or cracks at the average temperature T-. Individual substances will have high, narrow peaks whereas multi-component systems have low, broad peaks. For example, the products from the reactions at 440 "C from the Mexican and Libyan vacuum and visbreaker residues all have very broad, flat peaks. The uptake or consumption of hydrogen represents a criterion of the intensity of the hydrogenation. It is also expected that some index numbers of thermogravimetry will correlate to the consumption of hydrogen. This is verified, for example with the evaporation start temperature T5 % (Fig. 4-105), r = 0.924 coefficient of correlation, or with the distillable fraction AG400 (Fig. 4-106), r = 0.836. Figs. 4-105 and 4-106 clearly show that a higher consumption of hydrogen (0.4 wt%) leads to higher quantities of light products. This is primarily a consequence of the height of the reaction temperature, but the origin can also have an effect, although this influence cannot be quantified. At both reaction temperatures (410 "C and 440 "C) the T5 % values of the heavy oils produced fall into a series of their origins: Mexico > Middle East + North Africa > Venezuela + Canada. At 460°C such differences no longer exist. For the amounts of distillable fractions AG400 the order of succession is reversed. The differences for the characteristics of the products formed decrease with the rise of the reaction temperatures, which can be shown by correlating the hydrogen consumption

3 14


250

T5X ("C) f

410'C

x

440'C

0

460°C

x

5 f

x

200

X

150 c

XXXX

X X rn

H2 Consumption (wtX1 __e

100 0:5

1

1,5

Fig. 4-105 Thermogravimetry of Reaction Products from Hydrocracking. Vaporization Start Temperature T.5 % of the Heavy Oil Fraction versus H,-Consumption during the Reaction

with the content of crackable substances, CR. The coefficient of correlation r = 0.831 at 410 "C decreases to r = 0.254 at 460 "C. This leads to the conclusion that at 460 "C another reaction mechanism takes place in addition to the hydrogenation. The function of the relative decrease of the crackable substance ACR ACR =

CR(origina1) - CR(product) . 100 (%) CR(original)

versus the hydrogen consumption is represented by a regression curve possessing a coefficient of correlation r = 0.858, with a maximum which corresponds to the hydrogen consumption at 440 "C (Fig. 4-107). The Correlation of thermogravimetric index numbers with analytical data are listed in Table 4-136. Out of the three index numbers which describe the evaporation behavior,


315

70 A6400 ( % I X

x X

60

-

m

i

X

X

xx

7

clL7 rn

50

40

30

*

410'C

x

440'C 460'C

-

H2Consumption ( w t X ) 20

7

I

0,5

1

1,s

Fig. 4-106 Thennogravimetry of Reaction Products from Hydrocracking. Distillable Part AG400 of the Heavy Oil Fraction versus H2-Consumption during the Reaction

AG400 may be correlated satisfactorily with the average molecular weight of the original samples and their reaction products (heavy oils). The evaporation start temperatures T1 % and T5 % only give satisfactory correlation with the average molecular weight of the original samples. The crackable residue, CR, correlates with the average molecular weight of the original samples, and of the heavy oils from the reactions at 410 "C and 440 OC, but not with the molecular weight of the products from the 460 "C experiments. The coke residues R600 and R800 do not correlate with the asphaltene content in the original samples, but do correlate with that of the products from the reaction at 410 "C and 440 "C. If the values €or the reaction products from Arabian Heavy VR are excluded, the coefficients of correlation increase considerably (Table 4-136, no. 7 and no. 8). The Conradson coke residues (DIN 51 551) were only determined for the original samples. As expected, the correlation of the Conradson coke with R800 is very good [4-161.

3 16


t GI3 713

-

50

-

ACR (%’

48 -

“I

Y )k*

Y

30

2I3 18

-

x iy

iK

x:

410’C

x

440’C

IN

460’C

Y

Y

Hi! C o n s u m p t i o n ( w t X ) O J

(3.5

L3

-

1

1 ~ 5

1

Fig. 4-107 Thermogravimetry of Reaction Products from Hydrocracking. Change in Quantity of Crackable Substance ACR of the Heavy Oil Fraction versus H2-Consumption during the Reaction.

Table 4-136: Correlations of index numbers from thermogravimetry with analytical data. Sample No.

Index number Analysis of ThermoValue gravimetry :

Feed

Coefficient of correlation r Products of the reaction at 410°C 440°C 460°C

-

1 2 3 4 5 6 7

T1% T5 % AG 400 CR R 600 R 800 R 600

8

R 800

9 10 11 12 13 14 15 16

R 800 CR T1 % T5 % AG 400 R 600 R 800 CR

M M M

M Asphaltenes Asphaltenes Asphaltenes (without Arabian Heavy) Asphaltenes (without Arabian Heavy) Conradson coke H2-Consumption H/C H/C H/C H/C H/C H/C

0.7 18 0.758 0.610 0.793 0.054 0.197

0.353 0.423 0.543 0.449 0.449 0.534 0.843

0.247 0.389 0.546 0.525 0.377 0.398 0.738

0.045 0.026 0.697 0.118 0.581 0.523

-

0.854

0.778

-

-

0.887 -

-

-

-

-

-

-

0.83 1 0.458 0.532 0.593 0.622 0.550 0.713

0.628 0.096 0.211 0.494 0.086 0.053 0.531

0.254 0.146 0.192 0.610 0.225 0.270 0.383

4.5 investigation on the hydrocracking reaction of heavy residues

3 17

The H/C ratio only correlates with the distillable fraction AG400. Neither the evaporation start temperatures T1 % and T5 %, nor the coke residues R600 and R8OO give a satisfactory coefficient of correlation with the H/C ratio. All the above-mentioned examples show linear relationship. Hydrogenation is always preceded by a pyrolysis reaction, which supplies shorter, unsaturated fragments of the original long-chain molecules. The pyrolytic behavior of the original samples, which may be investigated very well by DSC is of great interest [4-201. The DSC experiments were carried out using a DuPont 990 Thermal Analyzer equipped with a 910 DSC and pressure cell. The experiments and their evaluation were carried out according to ASTM E 698-79 [3-131 using at least three different heating rates in a hydrocarbon atmosphere (CH,) at 10 bar pressure and with 5 cm3/min gas flow rate in order to simulate the process parameters of thermal cracking. As usual the coefficients of the Arrhenius equation (Eq. 3-37) were calculated, i.e. the activation energy E (ld/Mole) and the frequency factor log A (min-l). The temperature ranges of the maxima of energy consumption are listed in Table 4-137. Four samples showed two maxima of energy consumption each, whereas the other samples showed only one maximum each. The first of the maxima represents an evaporation, as determined by previous investigations. The activation energy calculated here equals the enthalpy of vaporization (Table 4-138). The sample Marguerite Lake VR demonstrated only one DSC peak in the region of evaporation. The Arrhenius coefficients E and log A and the conversions U during the pyrolysis are listed in Table 4-138, where

U=

start weight - end weight start weight

. 100 (%).

The average conversion is approximately 83 % which is comparable with the value found for bitumen and other heavy residues.

Table 4-137: DSC of the original samples (feed) in 10 bar methane. Peak maximum temperatures ("C) Sample No.

1 2 3 4 5 6 7 8 9 10 11


Peak 1

Peak 2

-

445-470 450-475 456-47 1 441-461 438-460 454-475 450-478 438-464 442-463 438-462

364-429 368-405 355-394 350-388

-

332-388

-

3 18


Table 4-138: Coefficients of the Arrhenius equation and conversion of the pyrolysis of the original samples (feed ) in 10 bar methane.

E [kJ/Mol]

Sample No.

4 5 6 7 8 9 10

11

Isthmus VVR Peak 1 Peak 2 Isthmus VR Peak 1 Peak 2 Zueitina VVR Peak 1 Peak 2 Zueitina VR Peak 1 Peak 2 Arabian Medium VR Peak 1 Peak 2 Arabian Heavy VR Peak 1 Peak 2 Kuwait VR Peak 1 Peak 2 Tia Juana VR Peak 1 Peak 2 Morichal VR Peak 1 Peak 2 Cold Lake VR Peak 1 Peak 2 Marguerite Lake VR Peak 1 Peak 2

log A [min-'1

u [%I

-

-

233.6

16.369

79.00

86.8 237.2

6.232 16.579

8 1.66

125.3 195.9

9.471 13.678

77.64

114.0 214.9

8.724 15.276

87.36

115.9 254.8

8.982 18.211

83.55

-

-

239.9

16.769

-

-

206.0

14.271

-

-

229.6

16.325

-

-

219.4

15.501

-

-

233.2

16.576

73.4

5.428

-

-

83.51 83.33 83.28 84.21 84.17 -

83.55

The first peaks were observed in a very broad range of temperatures (for example 230-270 "C at p= 5 K/min; 390-430 "C at p= 20 K/nlin). The temperature ranges for the second peaks are much more confined (440-455 "C at p = 5 K/min; 460-480 "C at p = 20 K/min). The differences of the H/C ratio of the samples are very small despite their different origins (H/C = 1.39-1.45). The kinetic coefficients of the vacuum residues are greater than those of the corresponding visbreaker residues. This indicates that the visbreaker residues are subject to thermal damage. The activation energy of the first peaks El may be correlated with analytical data, e.g. the average molecular weight, the asphaltene nitrogen, and sulfur content, (Table 4-1 39).


319

Table 4-139: Coefficient of correlation rl for peak 1 Correlation o f the activation energy El with analysis data. Activation energy

Analysis value M Asphaltenes N S

El

El El E,

0.610 0.536 0.946 0.512

Correlation of El with thermogravimetric index numbers, i.e. TI %, T5 %, AG400, and CR is shown in Table 4-140. Table 4-140: Coefficient of correlation r, for peak 1. Correlation of the activation energy E, with index numbers of thermogravimetry. Activation energy

r1

Index number of thermogravimetry T1 % R5 % AG 400 CR

0.816 0.933 0.805 0.685

It is not surprising that the activation energy E , which represents a vaporization enthalpy is related to the average molecular weight M and the index numbers which represent the evaporation behavior (T1 %, T5 %, AG400). On the contrary, the correlation of E, with the nitrogen and sulfur contents is unusual and these results inight be suspected since they are only based on four values each. The activation energy of the pyrolysis reaction E, shows a linear relationship with analytical data such as the nitrogen and sulfur contents, whereas no correlation could be found with the average molecular weight, the colloid composition, or the H/C ratio (Table 4-141). Table 4-141: Coefficient of correlation r2 for peak 2 Correlation of the activation energy E7 with analysis data Activation energy

Analysis value

r2

N

0.605 0.821 (without Morichal, Cold Lake) 0.617 (without Kuwait)

N S

E, decreases with increasing nitrogen concentration and increases with increasing sulfur concentration. That might be a consequence of the weakness of the aliphatic C-N bond,

320


Fig. 4-1 08 Pyrolysis of Heavy Residues H2-Consumption during Hydrocracking versus natural logarithmus of the Reaction Constant from DSC Pyrolysis

"?

'1

H2Consumption (wt%) Mean of Samples Yo. 10

w k a n o f Samples Yo.

1 -

I) " rl

-

0 "(1

-

(3.4

-

If.2

-

1'1

-

11

8 - 9

-

7

w a n o f Samples Yo.

5

Mean o f Samples l o .

1- 4

T ("C) ___._c

0 -

I

4.6 Oil shale and shale oil

321

whose free bonding enthalpy (243 kJ/Mol) is considerably less than that of the aliphatic C-C bond (297 kJ/Mol) [4-441. This implies that the radical chain reaction of the pyrolysis is accelerated by the greater number of more-easily-formed radicals from the C-N bond. On the other hand it is well known that some sulfur compounds act as radical deactivation agents and thus may retard the pyrolysis reaction. We could not find a relationship of the activation energy E, with process data, e.g. hydrogen consumption or formation of gases, nor with formation of light crack products. However, the plot of hydrogen consumption against the natural logarithmus of the reaction rate constant shows three groups of data corresponding to the three reaction temperatures. A linear regression confirms the correlation by a high coefficient of correlation r = 0.887 (Fig. 4- 108). The consumption of hydrogen increases with increasing reaction rate constant, since more radicals are formed which may be hydrogenated. The plot of hydrogen consumption versus reaction temperature confirms this relationship and indicates that the origin of the samples has an effect, although this is not quantifiable (Fig. 4-109).

4.6 Oil shale and shale oil Any kind of minerals which contain organic substances in large quantities, may be characterized as oil shale regardless of their mineralogical structure. Thus most of the oil shales are not shales in the mineralogical sense, and they do not contain oil, but higher molecular hydrocarbons and heterocompounds, which are only partly soluble and are called kerogens. Besides kerogens, bitumens are also present in variable quantities and these may be extracted using organic solvents. The kerogens are classified as three types: type I is predominantly of marine origin and exhibits the highest H/C ratio (normally 1.4). Type III is predominantly of terrestrial origin and exhibits the lowest H/C ratio (- 1.0). Type I1 is regarded as a blend of type I and type I11 and its H/C ratio lies between those of type I and type III. The O/C ratios increase from type I up to type III. Table 4-142: Element analysis of kerogens [4-451.

Schandelah (Germany) Estonia Aleksinac (Yugoslavia) Green River (USA)

\

\

65.8 77.1

7.7 9.4

2.1 0.4

6.9 2.0

17.5 11.1

1.40 1.46

0.20 0.11

76.6

10.7

3.9

3.9

4.9

1.67

0.05

80.5

10.3

2.4

1.o

5.8

1.53

0.05

322


The elemental analysis of some kerogens is listed in Table 4-142. Kerogens are insoluble, in organic solvents and may be isolated using anorganic acids (HCl, HF), which dissolve the mineral matrix. It is disputed whether the material thus isolated still consists of the original kerogens. The processing of oil shale to obtain liquid substances is performed by retorting. Smouldering or low temperaturecoking analysis (in inert gas at 520 "C) according to Fischer is the method for identifying the recoverable oil products (Table 4-143). Table 4-143: Smouldering analysis of oil shales according to Fischer [4-451. Origin

Schandelah (Germany) Estonia Aleksinac (Yugoslavia) Green River (USA)

Humidity

Water from Decomposi tion

Oil

Residue

Gases t Loss

(%)

(%I

(%I

(%I

(%I

3.7 dry

1.7 1.9

5.7 22.0

84.7 70.5

4.2 5.6

5.5

2.1

12.6

15.3

4.5

dry

1.4

10.4

85.7

2.5

In view of current oil prices, inining and processing of oil shale has been suspended in most countries on economic grounds. The fact cannot be ignored that retorting of oil shale produces considerable quantities of carcenogenic polycyclic aromatics. On the other hand known oil shale reservoirs represent a large reserve of fuel for the future.

4.6.1 Investigation using TGA and DSC The largest oil shale reservoir in Turkey is situated near Goynuk representing an estimated capacity of lo9 tons in layers of 100 to 150 meters deep which are not very Table 4-144: Proximate and ultimate analysis of Goynuk oil shale (air-dried basis) [4-461. Moisture (wt %) Ash (wt %) Volatile matter (wt %) Fixed carbon (wt %) C (organic) (wt %) H (wt %) s (wt %) N (wt %) Caloric value (kcal kg-I) Kerogene H/C ratio* Kerogene O/C ratio*

* From demineralized oil shale

6.7 32.6 50.1 10.6 43.5 7.1 1.5 1.4 4840 1.52 0.19


323

homogeneous. The results of a proximate and ultimate analysis [4-461 are given in Table 4-144. The thermogravimetric graphs both in inert gas and in air, of Goyniik oil shale and of its kerogen concentrate (demineralized oil shale) are shown in Figs. 4-110 and 4-11 1. Thermogravimetry in argon shows that there are highly volatile constituents in the oil shale as well as in the kerogen concentrate (T1 %: 27-29 OC; T5 %: 166-243 "C). The distillable fraction AG400 comprises 24-30 wt%. The higher value for the original oil shale indicates that some alteration to the kerogen concentrate occurred as a consequence of the preparation. The residue at 600 "C of the oil shale is 39.2 wt% and that of the kerogen concentrate is 29.1 wt%. At 800 "C the difference in the residues from oil shale and from kerogen concentrate becomes very small (oil shale R800 = 25.0 wt%; kerogen concentrate R800 = 26.4 wt%). Such a result was unexpected, since the oil shale still contains all the mineral

L L 1

200

300

4ud

500

600

Fig. 4-110 Thermogravimetry of Oil Shale Goyniik Heating Rate p: 10 K/min Gas Flow Rate: 25 cm3/min Curve 1: Argon Curve 2: Air

324


Fig. 4-1I 1 Thermogravimetry of Kerogen Concentrate from Oil Shale Goyniik Heating Rate p : 10 K/min Gas Flow Rate: 25 cm3/min Curve 1: Argon Curve 2: Air

parts of the sample. The graph of weight loss rate, DTG, exhibits two peak maxima at 300 "C and 468 "C for the oil shale whereas the kerogen concentrate has only one rnaximum at 468 "C. TGA in air displays distinct steps in the curve for the oil shale, at temperatures between 225 "C and 270 OC, 290 "C and 400 OC, and 430 "C and 525 "C. The corresponding DTG maximum temperatures occur at 262 "C, 383 OC, and 443 "C. These steps in the TGA curve do not appear so clearly in the thermogravimetry in air of the kerogen concentrate, but the DTG peak maxima are also present at temperatures of 269 'C, 383 "C, and 455 "C. The ash at 800°C of the oil shale amounts to 14.3 wt% and that of the kerogen concentrate to 4.4 wt % indicating that complete demineralization has not taken place. It is not possible to tell whether this weight is partly lost from the minerals of the oil shale.

21

30

27

25

29

22

3.5

400

440

480

520

560

595

Kerogen concentrate at 520 "C

Pyrolysis Temoerature

~

52

35

45

42

45

5.5

35

(%I

45.3 46.5

17.8 17.4

14.0

37.5

51.8

53.3

18.7

24.2

41.5

14.8

90.4 91.9

80.0 80.5

67.5

85.0

82.0

95.0

86.9

75.5

81.6

88.0

(%I

400°C

73.7

85.7

(%I 44.0

(%)

24.0

300°C

200°C

100°C

AG

I

0

2.8

6.0 4.0

4.7

3.0

2.4

2.8

3.2

8.9

5.1

9.1

0

(%I

(%I 0.9

800°C

R

600°C

Table 4-145: Thermogravimetry of the pyrolysates from the retorting of Goyniik oil shale Heating rate p = 10 K/min, Argon 2.5 cm3/min

I

2.48 2.88

2.41 3.45 1.35 2.86 1.67 3.49 2.01 3.55 2.00 3.23 3.03 2.13

(%/min)

DTG

42 258

("CI 24 163 30 244 21 245 28 253 31 254 39 215

Tm

h

2.

; i .

R

b

8a

&

9

9

-

.a

on

326


Thermogravimetry of the products from retorting at increased temperatures (400 'C, 440 'C, 480 O C , 520 'C, 560 OC, and 595 "C) indicates that this process generates considerably lighter products (Table 4-145). The evaporation start temperature T1 % remains between 21 and 35 "C but T5 % has decreased to between 35 and 55 "C. The distillable fraction AG400 has been increased to values between 82 and 95 wt% with a maximum for the product from retorting at 480 "C. The residue at 600 "C increases with increasing retorting temperature (Fig. 4-112). Each sample exhibits two peak maxima in the DTG curve which are situated in the evaporation region (maximum temperature 258 "C). The results from the product of retorting the kerogen concentrate (at 520 "C) are not very different from the other retorting products. The curves of a simulated distillation are similar to the curve of a degassed German crude oil from a Dogger-pformation reservoir with approximate 28 API. Thermogravimetry in air of the retorting residues provides data of R800 for the oil shale residues from 37-45 wt%, which correspond to nearly twice the amount of the original samples. That implies that approximately a quarter of the organic substance originally present has been distilled by retorting. If the retorting temperature increases, the value of R800 also rises from 4.4 (original sample) to 18 wt% in the retorting residue from the

Fig. 4-112 Thermogravimetry of Pyrolysates from Oil Shale Goyniik Residue at 600 "C R600 versus Pyrolysis Temperature x Oil Shale o Kerogen Concentrate

Benzene/ Methanol (2: 1) Chloroform Toluene

Pyridine

Extraction medium

226 258 238

132 181 157

19.0 18.2

181

20.3

75

0 0.2

0.7

2.3

1.5 2.3

3.1

5.9

16.6

38.9

I 15.8

14.2

11.0 14.7

16.2 40.0 50.8

46.4

10.9 9.0

12.3

3.1 1.5

10.9

5.14 3.88 3.88

4.78

2.22 4.60

:%/min)

[dxiiizl

21.3

DTG

6

Table 4-146: Thermogravimetry of the extracts from Goyniik oil shale Heating rate p = 10 K/min, Argon 25 cm3/min

469 370 448

441

327 467

("c)

T,,

-a

P

b\

328


experiment at 520°C temperature. This implies that 14 wt% of the organic substance originally present has been pyrolysed and distilled. Only a small portion of the organic substance may be extracted from the oil shale by organic solvents. Extraction using solvents with different solubility parameters 8, according to Hildebrandt, yields extracts, which have altered characteristicsas a consequence of this solubility parameter (Table 4-146). The following solvents were used:

Pyridine Benzene/Methanol (2 : 1) Chloroform Toluene

21.3 20.3 19.0 18.2

Thennogravimetry in argon shows that the evaporation start temperatures of the extracts (T1 %, T5 %) increase with increasing solubility parameter 8, whereas the distillable fraction AG400 decreases and the residues R600 and R800 increase (Fig. 4-1 13). This is a consequence of the rising amounts of aromatic, coke-generating substances, which were extracted with solvents of increasing solubility parameter. The curves of the simulated distillation of the extracts correspond approximately to that of an atmospheric residue of a Middle East crude (for example AR Kirkuk).

15

10

5

Fig. 4-113 Thermogravimetry of Extracts from Oil Shale Goyniik Residues at 600 "C R600 versus Solubility Parameter 6 of the Extraction Solvent


329

Table 4-147: Kinetics of the pyrolysis of oil shale Goyniik. Argon 5 cm3/min Oil shale

Kerogene concentrate

Peak I Temperature range ("C) E (kJ/Mol) log A (min-')

445-480 262.1 18.551

444-478 294.0 21.014

Peak 2 Temperature range ("C) E (kJ/Mol) log A (min-')

495-550 394.4 22.227

520-550 229.9 14.338 71.4

Conversion (%)

57.9

The investigation of pyrolysis kinetics according to ASTM E 698-79 using DSC in inert gas shows that there are two distinct maxima of the energy consumption for both the oil shale and the kerogen concentrate (Table 4- 147). The activation energy and the frequency factor, which were determined for the first peak, are similar to those of heavy distillation residues. The values of the second peak are extraordinarily high, thus indicating that reactions of parts of the mineral matrix may have an effect. The residues after the reaction amount to 42.1 wt% for the oil shale and to 28.6 wt% for the kerogen concentrate. Since TGA has shown ash content of 14.3 wt% for the oil shale, the amount generated from the organic substance of the oil shale must be 42.1 %-14.3 % = 27.8 %. That means that approximately one-third of the organic substance has been converted to non-volatile products. The analogous calculation for the kerogen concentrates proves that a fraction of 24.8 % of the residue was generated from the organic substance and also confirms the Table 4-148: Kinetics of the oxidation of oil shale Goyniik. Air (7 bar, 5 cm3/min) Oil shale

Kerogen concentrate

Peak 1 Temperature range ("C) E (kJ/Mole) log A (min-')

300-355 132.8 11.772

295-330 124.0 11.057

Peak 2 Temperature range ("C) E (kJ/Mole) log A (min-')

420-460 177.6 12.987

435-475 174.4 12.485

Peak 3 Temperature range ("C) E W/Mole) log A (min-') Conversion (%)

550-570 357.8 22.907 76.9

-

95.7

330


conversion of approximately one-third of the organic substance in the kerogen concentrate to non-volatile products. Investigation of oxidation kinetics displays three distinct and easily-evaluated peak maxima for the oil shale and only two peaks for the kerogen concentrate. The activation energy and frequency factor for both the first and the second peak demonstrate similar values for both oil shale and kerogen concentrate. The third peak, which is only present in the DSC of the oil shale, reveals fairly high values of the kinetic coefficients (Table 4-148). During thermoanalytical investigations of oil shales, the reactions of the mineral matrix have to be considered. For example, any water of crystallization will be released in a temperature range between 75 and 150 "C depending upon the heating rate and the gas flow rate. The water which is bonded in the crystal lattice of clay or zeolite, requires temperatures as high as 300 "C for its release. An approximately equal temperature will be required to release the fifth water molecule from CuSO, . 5 H,O. Above temperatures of 550 OC, separation of carbon dioxide from carbonate minerals has to be taken into consideration. The reservoir matrix of Green River oil shale consists of varying amounts of nine minerals (according to [4-451) which are listed in Table 4-149. The TGA curves of a Green River oil shale and of the minerals Illite (Morris, Illinois USA) and Montmorillonite (Clay Spur, Wyoming USA) are given in Fig. 4-1 14. Illite and Montmorillonite have congruent curves in argon and in air. Naturally differences were found in the TGA curves of Green River oil shale both in argon and in air as a consequence of the oxidation of the organic substances. DSC (heating rate p = 10 K/min) in argon of Montmorillonite has a peak maximum of energy consumption at 136 "C. Above 550 "C the curve descends again but another maximum may not be recorded since the upper temperature limit of the instrument is 625 "C and it certainly lies in the temperature range above 650 "C. Illite exhibits three endothermic maxima at 102 'C, 372 OC, and 525 "C. The Green river oil shale also exhibits three endothermic maxima, at 82 "C, 455 OC, and 532 "C. The organic content of the Green River oil shale is reported to amount to 15.1 wt%. The Karl-Fischer water titration yields 6.8 wt% water in the air-dried sample. Similar titrations with Illite result in 2.8 wt% water and for Montmorillonite in 9.1 wt% water. From TGA, water contents of 3 wt% for Green River oil shale, about the same value for Illite, and of 8 wt% for Montmorillonite may be assumed. The weight loss of 5 wt% in the temperature Table 4-149: ,Mineral composition of Green River oil shale according to /4-45/ Dolomite Calcite Quartz Clays (Illite and Montmorillonite Orthoclase Plagioclase Zeolite Pyrite, Marcasite

CaMg(CO3)z CaCO, SiO, 5 0 x 3 4 0 , x 6Si0, x 2H20 Al2(0H),(Si4O,,,) x nHzO (Mg & Ca) KAlSi,O, NaAlSi,O, x CaAI$i,O, NaA1Si206x H20 FeS,

32-33 % 16-20 % 10-15 % 11-19 % 4- 6 % 10-12 % 1- 7 % 1- 3 %


331

Fig. 4-114 Thermogravimetry of Oil Shales and Minerals Heating Rate p: 10 K/min Gas Flow Rate: 25 cm3/min Curve 1: Oil Shale Green River in Argon Curve 2: Oil Shale Green River in Air Curve 3: Montmorillonite (Clay Spur, Wyoming/USA) in Argon Curve 4: Illite (Morris, Illionois/USA) in Argon

range between 400°C and 800°C for Montmorillonite may be interpreted as a water release from the aluminum hydroxide (part of the sample, theoretically 5 wt% maximum), which attains its highest reaction rate between 600 "C and 700 "C temperature. It is possible that organic components of the Green River oil shale may have reacted with the iodine in the Karl-Fischer reagent, thus simulating a higher water concentration.

4.6.2 Modelling and simulation of oil shale pyrolysis The knowledge that oil shales contain kerogens, which are insoluble in organic solvents, and varying amounts of bitumens, which are soluble in toluene, is state-of-the-art [4-47, 4-48]. Pyrolysis of the kerogens and bitumens during retorting is a complex process, which comprises more than one subsequent and/or parallel reaction step and which cannot be described by assuming that only one first order reaction takes place. Investigations firstly on Jugoslavian oil shales and later on others [4-47 to 4-51], which contain different types

332


Table 4-150: Composition of the oil shales (wt %) according to [4-47. 4-49]. Sample Aleksinac (A) Knjazevac (K) Korea (KOR) Estonia (E)

Mineral matrix

Organic substance

Kerogen

Bitumen

77.7 93.2 58.6 53.5

22.3 6.8 41.1 46.5

21.5 6.3 39.8 45.5

0.8 0.5 1.6 1.o

of kerogens and different total amounts of organic material (table 4-150), showed that thermogravimetry and DSC gave different values of the Arrhenius coefficients which appear in different temperature regions. Some pyrolysis kinetics were determined using DSC according to ASTM E 698-79. The mathematical procedures used in the determination of the TGA-derived kinetics correspond to the integral method according to Doyle [4-521 and Gorbachev [4-531. Assuming first order for the weight loss rate during TGA experiments at constant heating rate p :

p (da/dT) = A[exp (-EIRT)] . (1-a) where:

p

= heating rate

a: A E R T

= fractional weight loss = frequency factor = activation energy

= universal gas constant = Kelvin temperature

Eq. 4-5 (K min-') (dimensionless) (min-l) (kJ Mole-') (J Mole-' K-l) (K)

Integration of equation 4-5 using either approximations by series or partial integration gives: ln[-ln(l-a)/P] = ln[AR/P ( E + 2 R q ] - E/RT

Eq. 4-6

The Arrhenius coefficients, E and A , may be found by plotting In[-ln (l-a:)/P]versus 1/T. To validate of the methods, a series of original and modified samples from the oil shales listet in Table 4-150 were tested by thermoanalytical methods. The samples were designated as follows:

1. Oil shale (original) 2. Kerogen concentrate from No. 1 3. Bitumen (toluene extract from No. 1) 4. Bitumen" (toluene extract after heating up to 773 K of oil shale No. 1) 5. Extraction residue after heating up to 773 K of oil shale No. 1

Sample A (K, KOR, E) Sample A-K (K-K etc.) Sample A-B (K-B etc.) Sample A-B-773 (K-B-773 etc.) Sample A-773 (K-773 etc.)


333

Table 4-151: Regions of temperature and conversion for first order reaction in thermogravimetric pyrolysis according to [4-47, 4-49]. Sample

Heating rate (K/mnin)

Temperature region (K)

Conversion region

A

10 5 2

A-K A-B A-B-773

10 5 2 10 10

A-B-773

5

A-773 K

10 10 5 10 5 2 10 10 20 10 5 20 10 5

633-753 613-753 593-733 593-733 593-753 593-733 573-753 573-673 693-813 573-673 693-813 633-773 613-873 593-873 613-8 13 593-813 633-873 733-873 733-853 673-773 653-753 633-733 653-873 653-873 613-673 713-813 613-693 713-873 653-753 653-753 633-713 673-753 673-753 653-733 613-833 633-713 733-813 613-713 733-813 673-753 653-773 653-733

1-59 1-64 1-65 1-41 1-42 1-52 15-71 2- 8 11-68 4-14 14-69 1-66 3-61 2-76 1-26 1-27 3-36 24-92 27-74 9-SO 7-45 6-43 30-72 34-82 35-46 55-72 37-51 51-82 7-33 10-43 7-39 10-78 12-90 9-85 34-84 51-69 69-86 40-64 75-87 9-65 8-62 9-68

K-K K-B K-B-773 KOR KOR-B-773

2 KOR-773 E

E-B-773

20 10 5 20 10 5 10

5 2 E-773

20 10 5

Total conversion 573-873 K

(%I 73 77 82 57 50 61 87 75 74 81 61 76 34 36 36 92 80 60 61 60 78 82 75 82 57 59 58 96 99 99 87 90 92 83 84 81

140

-

113 140

123 29.9

-

-

-

-

-

-

-

-

-

-

20

(a) First region 573 to 693 K (b) Second region 630 to 813 K

E-773

KOR-773 E E-B-773

A-773 K K-K K-B K-B-773 KOR KOR-B-773

(b)

A-B-773 (a)

A-K A-B

A

Heating rate (K/min) Sample

137

114 154 37.9

33.3 81.4 117.4 93.6 48.0 78.9 75.1 125 31.8

125.7 117.4 86.3 30.8

-

9.5478

-

-

7.0607 9.5478

7.8470 1.0934

-

65.8 18.8

-

-

52.7 16.4

-

-

-

133 64.0 21.4 127 150 41.2 23.7 136

-

39.6

-

-

-

-

-

-

-

-

-

-

20

55.2 46.6

-

-

-

-

33.0 55.8

96.7

-

102.6

2

88.3

113.0

5 Activation energy (kJ/Mol) 10

Table 4-152: Kinetic coeficients from thermogravimetry according to [4-47, 4-49]

9.0682

7.0170 10.6513 1.5185

0.5705 4.5302 7.4843 5.2330 1.6085 4.3096 3.8426 7.8774 0.9100

18.2014 7.6656 5.1584 1.1959

8.371 1 0.4099 -0,1530 0.8176 10.2480 1.6503 0.2201 8.9903

-

-

2.1703 1.2355

-

0.4843 2.4579

-

5.0934

7.0864

10 5 log A (min-')

0.2648 -0.5258

-

-2.1635 -0.9393

-

0.2967

-

-

-

-

-

5.5276

6.1367

2

2

2

P

3 5

P

2 A 2

s 2 a

ol

8 a .g.

s

E

8 cR?

3

2

3

A

P

w

w


335

Table 4-153: Kinetic coefficients from DSC according to [4-47 , 4-49]. Sample

A ( < 773K) A ( > 773K) A-K ( < 773 K) A-K ( > 773 K) A-B-773 (623-873 K) K K-K E ( < 773K) E ( > 773K) E-B E-B-773 E-773 KOR KOR-B KOR-773

400

450

Activation energy (kJ/Mole)

log A (min-')

215 320 227 232 307 539 293 200 273 182 146 182 149 92.3 333

14.9895 20.68 12 15.7782 14.0414 21.8639 35.8621 18.6435 14.0253 17.3324 14.3304 10.4150 12.7513 9.7160 6.2833 23.0899

500

550

T

("C)

Fig. 4-115 Thermogravimetry of Oil Shale Aleksinac (STA 780) Heating Rate p: 5 K/min Atmosphere: Argon 30 + 20 cm3/min The TGA curve shows the weight change of that organic part of the sample which remains after heating up to 300 "C.

336


We evaluated regions of both temperatures and conversions from thermogravimetry where the reaction obeys first order kinetics (Table 4-151). The Arrhenius coefficients from TGA are listed in Table 4-152, those from DSC in Table 4-153. Fig. 4-1 15 shows the modification of the original TGA plot (Simultaneous Thermal Analyzer STA 780) of pyrolysis of the oil shale Aleksinac (sample A) in inert gas. The percentage weight change (ordinate) is calculated in relation to the organic material present in the oil shale after heating up to 300 "C (573 K) i. e. the maximum weight loss assumes an organic content of 22.3 wt%. The maximum weight loss occurs in the temperature range between 380 "C and 420 "C, but the peak maxima of DTG and DTA appear at considerable higher temperatures (520-530 "C). The DTA peak proves that this is an endothermal reaction. The bitumen from oil shale Aleksinac (sample A-B) has two regions of different slopes of the straight line in the plot In[-ln (l-a)]/P versus 1/T, which result in two different pairs of the Arrhenius coefficients, below and above 420 "C (693 K). The low values of E and log A for the region below 420 "C indicate that evaporation takes place. For the oil shale Knjaievac (sample K) the range of both temperature and conversion which obeys first order reaction is wider. This range is limited to 150 K for the oil shale Aleksinac (sample A) and to 100 K for the two oil shales Korea (sample KOR) and Estonia (sample E). Nevertheless 70 or 80 % of the conversion takes place in just this range of temperature. The thermally generated bitumens* also follow first order reactions in an ample range of temperature at least at higher heating rates fl Lower heating rates reveal two regions of different kinetic coefficients, the higher ones being in the higher temperature range. The oil shale from Estonia (sample E) has the highest conversion. This oil shale also supplies the highest activation energy in TGA. This holds also for the extraction residue from the sample which has been heated up to 773 K before extraction (sample E-773). The Arrhenius coefficients of the bitumens are generally lower than those of the corresponding oil shales. The values of E and log A from the DSC experiments considerably exceed those from TGA. The DSC pyrolysis of oil shale Aleksinac has two endothermal peak maxima. Depending upon the heating rates (p= 5, 10, 20 K/min) the first peak maxima appear at temperatures of 457 "C, 467 "C, and 477 OC,whereas the second peak maxima appears at 550 OC, 568 O C , and 583 "C. The kerogen concentrate sample A-K behaves in a similar €ashion, whereas the bitumen* A-B-773 exhibits only one maximum at temperatures of 448 "C, 454 OC, and 469 "C. may be interpreted as pyrolytic decomposition of the kerogen, since the foil and gases will be found in the temperature range from 350 "C up to 50 "C the heavy fractions formed during the pyrolytic decomposition of the kerogen, undergo coking. The activation energy E and the frequency factor logA of the pyrolysis of oil shale Knjaievac (sample K) have high values which cannot be explained. The oil shale Estonia :sample E) has two distinct peaks in the DSC curves, below and above 500 "C (773 K). 4fter several hours of pyrolysis at 550 "C (823 K) or above, an additional weight loss xcurs, indicating reactions of the mineral matrix of the oil shale. Therefore the activation


337

energy of 273 kJ/Mol determined for the second peak should not be completely ascribed to pyrolysis of the organic substance. The significant differences between kinetic coefficients derived from TGA and DSC result in significant differences in the basic data obtained by these two methods. The relation ship between weight loss rates registered by TGA and energy flow rates measured by DSC can be clearly understood for a simple physical or chemical process, but with the complex pyrolysis process combined with the partial evaporation of liquids produced, the relations between heat absorption and weight loss are more complex, especially in the case of multicomponent systems. Considerable energy demand during kerogen + bitumen" decomposition in the first stage of oil shale pyrolysis is accompanied by only a small weight loss. The same holds for the final stage when pyrolysis of the heaviest intermediates takes place. Therefore DSCderived kinetics would appear to be more appropriate to describe those stages of the pyrolysis process where decomposition of the organic substances mainly leads to the formation of non-volatile products i. e. solids and very heavy liquids. This could be important when the DSC-and TGA-derived kinetics, obtained here using the first order kmetic model, are used as starting points for the construction of a more complex multi-step kinetic model of oil shale pyrolysis. Mathematical modelling and simulation of pyrolysis of powdered oil shale samples allows predictions to be made of different effects which are relevant to the subsequent modelling and design of the retorting process [4-48 to 4-5 11. The primary aim is to predict the total gas and oil evolution rates during isothermal and non-isothermal pyrolysis of powdered oil shale samples, i. e. under conditions at which the heat and mass transfer do not govern the overall pyrolysis rate. Although some positive results were obtained using a single-step model (SSM) in the simulation of pyrolysis of oil shale samples, this kmd of model was inadequate because it did not account for differences in the observed pyrolysis rates of kerogen and bitumen originally present in the oil shale; it gave poor simulation of the actual weight loss and energy consumption rates; and it gave poor simulation of isothermal pyrolysis. The change from the single-step to a two-step model (TSM) for the pyrolysis process amplified the range of applicability, but neither model gives a satisfactory description of this complex process. They can in practice only give a rough prediciton of the cumulative sample weight loss during non-isothermal pyrolysis. Therefore we attempted to simulate advanced pyrolysis using a multi-step model (MSM). This model was developed using TGA- and DSC-derived kinetic coefficients, determined for chemically and thermally treated oil shale samples by modelling particular reaction steps. The MSM is based on the reaction scheme shown in Fig. 4-116 which displays a series of parallel and consecutive first order reactions. and B denote the kerogen and bitumen originally present in the oil shale; B,, B,, and R, to R, are nonvolatilized intermediates and products (solids and liquids); P, to P5are volatilized products (gases and vapors); and& tof,,, are the stoichiometric coefficients that fulfil the condition:

338

4 Thermounalytical Investigations on Petroleum and Petroleum Products

+

Fig. 4-116 Pattern of the Multistep Model for Pyrolysis of Oil Shale according to [4-491

As the intermediates and products were not physically or chemically characterized, their precise definitions are omitted to avoid introducing arbitrary terms. The reaction scheme reveals a similarity between intermediate B, and the bitumen in oil shale (B).Intermediate B, could be associated with bitumen generated by the thermal treatment of oil shale, which pyrolyses at lower temperatures in the same way as the bitumen originally present in oil shale. This is indicated by the almost identical activation energies obtained by TGA analysis. The kinetics of the reaction steps are represented by DSC-derived rate constants for oil shale samples (steps 1 and 2), and TGA-derived rate constants for bitumen samples (steps 3,4, and 5). The rate constant for the coking process (step 6) was taken from the literature. The stoichiometric coefficientsf, to&, were determined for each oil shale sample, giving cumulative effects which are in agreement with the experimentally determined conversions at particular temperature intervals. The coefficients for the coking process f9 andf,, were taken from the literature [4-561. The mathematical model of non-isothermal pyrolysis at a constant heating rate of p= dT/dt and based on the reaction scheme consists of the system of twelve equations 4-7 to 4-18 (Table 4-154). The start temperature is To= 573 K, while the start values of and B correspond to the weight fractions of kerogen and bitumen in the organic part of the oil shale sample. The initial values of all other components in the scheme are zero. The only exception is the initial P , value of 0.04 in the case of sample KOR, which corresponds to the fractional weight loss registered The process is simulated by solving equations 4-7 to 4-18 with a temperature increment of 0.5 K up to the final temperature of 873 K. The actual fractional weight loss values, necessary for the simulation of TGA curves, were calculated by equation 4-19: Eq. 4-19 where mo and m denote the initial and actual weights of the organic part of the sample.


339

Table 4-154: Reaction equations of the non-isothermal pyrolysis according to [4-491. (4-7) (4-8) (4-9) (4-10) (4-1 1) (4-12) (4-13) (4-14) (4- 15) (4-16) (4-17) (4-18)

The same mathematical model is used for the simulation of DTG and DSC curves by numerical derivation according to equation 4-20: ‘weight loss = dKmo-

m)/m,I/dT

Eq. 4-20

The actual heat consumption rates, necessary for DSC simulation, were calculated by equation 4-2 1: 5 ‘heat consumphon

=

1 ‘iQ,

Eq. 4-21

The DSC simulation was only carried out to determine the position of the peaks on the DSC curve (peak maximum temperatures), so the corresponding heats of reaction (Q,) were taken from literature. The rate constants and the stoichiometric coefficients (k, andJ respectively) used in the simulation are given in Table 4-155. It is evident that the frequency factors and the activation energies of the DSC-derived constants k, and k2 exceed all the other kinetic coefficients shown in Table 4-155. The high activation energy of 539 kJ/Mol (Table 153) indicates the negligible influence of the heating rate on the temperature at which the maximum rate of heat consumption is registered by DSC analysis. Similar activation energies have been reported for pyrolysis of medium volatile bituminous coal. These values are intermediate between the energy needed for the linear single-bond scission (368 kJ/Mol for ethane) and the double-bond scission (720 kJ/Mol for ethylene). Generally high activation energy values are followed by high values of the corresponding frequency factors as a consequence of the mode of calculation. To illustrate the combined effects of the activation energy and frequency factor on the rate of the particular pyrolysis step, the k, values were calculated for the samples K and KOR at several temperatures in the region 723-823 K (450-550 “C). Comparison of the results demonstrates that at 723 K, k, for the sample K is two orders of magnitude lower than for sample KOR. This implies relatively slow production of the intermediate B2

2.00 x 1015 exp(-24100/T) 1.06 x 1014 exp(-24100/T) 44.7 exp(-4950/T) 44.7 exp(-4950/T) 44.7 exp(-4950/T) 3.00 x lo5 exp(-11000/T) 0.84 0.16 0.10 0.90 0.12 0.88 0.12 0.88 0.80 0.20

5.00 x 1013exp(-17900/T) 5.20 x lo9 exp(-17900/T) 80 exp(-383O/T) 8.13 exp(-3830/T) 8.13 exp(-383O/T) 3.00 x lo5 exp(-11000/T) 1 .oo 0.00 0.43 0.57 1.oo 0.00 0.45 0.55 1.oo 0.00

7.28 x 1039 exp(-64800/T) 7.28 x 1035exp(-648OO/T) 6.96 x lo3 exp(-9030/T) 2.04 x lo4 exp(-9490/T) 6.96 x lo3 exp(-9030/T) 3 . 0 0 ~lo5 exp(-11000/T) 0.91 0.09 0.7 0.30 0.90 0.10 0.10 0.90 0.80 0.20

1.50 x 1017 exp(-25900/T) 9.76 x 1014exp(-25900/T) 3.00 exp(-4000/T) 15.7 exp(-37OO/T) 3.72 exp(-4000/T) 3.00 x lo5 exp(-113OO/T) 0.93 0.07 0.40 0.60 0.90 0.10 0.03 0.70 0.80 0.20

k, (min-I) k, (min-1) k,(min-') k4(min-') k,( min-') k6(min-') f'(1) f 2 (1) f 3 (1) f4(1) fs (1) f 6 (l)

fg

(1) f,o(l)

f* (1)

f 7 (1)

E

KOR

K

A

Model Parameter

Table 4-155: Model parameters for the simulation of pyrolysis according to [4-491.

W

B

aF

"cr

5

6

2a

R

5a

$

3

"cr

3

5

5'

fs

B G.

R,

%

__

a 3 a

2

2

A

P 0


341

during low temperature pyrolysis of sample K. At 773 K, & values for both samples are of the same order of magnitude, and at 823 K the k, value for sample K is one order of magnitude higher than for sample KOR. The sharp increase in the k2value for sample K, as a consequence of the higher activation energy, indicates that production of the intermediate B2 during non-isothermal pyrolysis of sample K could be expected to take place in a relatively narrow range of higher temperatures. The production of B, during pyrolysis of sample KOR could be expected to take place in a wider interval of intermediate temperatures. A similar relation holds in the case of pyrolysis step 1, i. e. in the case of k , values. However, the increased frequency factor values, as a result of the fitting procedure, indicate that pyrolysis step 1 is almost over before the second step occurs. In the case of samples KOR and E, the lunetics of reaction step 4 are represented by the same rate constant used for steps 3 and 5, i. e. by the rate constant obtained for thermally generated bitumen. This is because kinetic data were not available for the pyrolysis of bitumen originally present in these samples. It was assumed that this deviation from the standard modelling procedure would not change the simulation results significantly, because of the very similar pyrolysis kinetics of the two bitumens, from samples K and A. Good agreement is found between experimental and simulated curves in the case of sampIe E (Fig. 4-117), and only small differences between the curves for sample KOR (Fig. 4118), which validates the model for the prediction of weight loss of these oil shales during pyrolysis. The differences between experimental and simulated curves increase for sample

1

0.8 0

E -0.6 \

E I

00.4

E Y

0.2

0

600

650

700

750

800

€350

Temperature. K Fig. 4-117 Non-isothermal Pyrolysis of Estonian Oil Shale (Sample E) Heating Rate p: 5 K/min Experimental Curve Simulated Curve According to [4-491 ~

342

4 The~i?~ounulyti&ul Investigations on Petroleum and Petroleum Products 1

0.8

E"

10.6 E I

00.4

-

E

v

0.2

-

---.-.-. 0

I " " I " " I " " I " " I " " I '

600

650

700

750

Temperature.

800

850

K

Fig. 4-118 Non-isothermal Pyrolysis of Korean Oil Shale (Sample KOR) Heating Rate p : 5 K/min -Experimental Curve - . - Simulated Curve According to [4-491

0.8 0

E"

\

-0.6 E I

00.4

E

v

0.2

-

0 600

650

700

800

750

Temperature,

850

K

Fig. 4-119 Non-isothermal Pyrolysis of Jugoslavian Oil Shale Knjaievac (Sample K) Heating Rate p: 10 K/min -Experimental Curve - . - Simulated Curve According to [4-491


343

K (Fig. 4-1 19). This proves that the model is not suitable for all types of oil shales. Sample K has a relatively high bitumen content, and somewhat different behavior during pyrolysis, compared with samples E and KOR. Most of its organic substance is pyrolyzed at temperatures above 773 K (500 "C) whereas in the case of samples E and KOR and Aleksinac oil shale (sample A) most of the organic material is pyrolyzed at 673-773 K (400-500 "C). Observation of the volatile products leads to the conclusion that the first part of the pyrolysis process of sample E starts below 673 K (400 "C) and is governed by the kinetics of reaction step 1. The reaction rate constant k, determines the reaction in the temperature range from 673 to 733 K (400-460 "C), and applies partially until the final stage at 873 K (600 "C). The final period of pyrolysis (above 733 K = 460 "C) is controlled by reaction steps 3, 4, 5, and 6, as indicated by moderate increases in contributions of corresponding products. The same analysis for sample KOR gives similar results, indicating that the most important part of the pyrolysis process, occurring between 673-773 K (400-500 "C), is controlled by reaction step 2. In the case of sample K, the changes in product distribution are quite different, indicating reaction step 4 as the controlling step below 693 K (420 "C), and reaction step 1 as the controlling step from 713-733 K (440-460 "C). At higher temperatures (733-793 K = 460-520 "C) pyrolysis is controlled by reaction step 2 while the final pyrolysis period is controlled by step 5. These differences in pyrolysis behavior of the oil shales can be explained by structural differences in the corresponding kerogen types. The kerogens of oil shales Aleksinac, Estonia, and Korea are associated with type I, which is of predominantly paraffinic nature. Oil shale Knjaievac is associated with kerogen type 111, which is of predominantly aromatic nature. Thus the multi-step model appears to be suitable for simulating the pyrolysis of oil shales with kerogen type I, but cannot be properly adjusted for the other kerogen types. A rough estimation of reliability in predicting the actual weight loss rates (DTG) and heat consumption rates (DSC) was performed by comparing the peak maximum temperatures obtained experimentally with those obtained using equations 4-20 and 4-21, for the examples of oil shales E and KOR (Table 4-156). The difference of 4 % maximum between the experimental and the simulated data is acceptable. Table 4-156: Comparison between the experimental and simulated DSC and DTG curves acording to [4-491. Sample

Curve

E

DTG DSC

KOR

DTG DSC

Peak maximum temperatures (K) at different heating rates 5 K/min 10 K/min 20 K/min exp. sim. exp. sim. exp. sim. exp. sim.

788 793 794 803

713 713 712 713 721 743 785 753

72 1 725 721 725

803 813 805 823 738 763 811 778

740 743 749 743

816 833 825 843 756 788 833 803

344


The multi-step model was also tested for prediction of the effects of isothermal pyrolysis, which represents the retorting process. Isothermal TGA experiments were performed using powdered samples E and KOR. The experiments were carried out at 673,708,748, 773, and 794 K (sample E) and 723, 748, 773, and 794 K (sample KOR). Although a maximum heating rate was used to attain the pyrolysis temperature, a significant part of the sample was pyrolyzed in a non-isothermal regime, especially in the experiments performed at higher temperatures. Therfore, the simulation was performed in two steps. In the first step, non-isothermal pyrolysis was simulated at a maximum heating rate, by solving the equations 4-7 to 4-18. When the temperature of isothermal pyrolysis was reached, the simulation was continued at constant temperature by switching from temperature to time as independent variable in equations 4-7 to 4-18. This was done by replacing the constant heating rate term by dT/dt. The initial conditions for the second step of simulation were the final conditions of the first step. Comparison of the experimental and the simulated values (Fig. 4-120) for sample A shows good agreement at 708 K (435 "C) and a perfect agreement at 748 K (475 "C). The simulation of sample E does not agree well with experimental values: At both temperatures 708 K (435 "C) and 748 K (475 "C) the predicted fractional weight loss values are considerably higher than those obtained experimentally. At the end of the isothermal test time, the experimental data at 708 K are approximately 45 %, and at 748 K approximately 36 %, lower than the simulated values. For samples K and KOR there is a better fit between the simulation results and the experimental data.

2

6

a t,min

Fig.4-120 Isothermal Pyrolysis of Jugoslavian Oil Shale Aleksinac (Sample A) at 435 "C (708 K) and 475 "C (748 K) Curve l a and lb: Experimental Curves Curve 2a and 2b: Simulated Curves According to [4-491


345

Modelling and simulation do not have to be limited to the pyrolysis of oil shales. Any pyrolysis process can be treated in similar manner, breaking it down into a series of consecutive and parallel reactions. The simulation will fit the real behavior so much the better, when as much information as possible is available on the original substances and intermediates. Fractions of similar chemical behavior from the original sample should therefore be characterized thermoanalytically and their reaction kinetics determined. Low and nonvolatile intermediates from the pyrolysis process should be isolated and their thermal behavior and reaction kinetics investigated.

4.6.3 Fingerprinting of oil shale by oxidation The DTG curve of oil shale in thermogravimetry in air exhibits several peak maxima at increasing temperatures similar to those of other organic substances. At a heating rate p= 10 K/min the first DTG maximum appears at approximately 250 "C and the last one generally above 450 "C. In the temperature interval between these limits up to five additional peak maxima appear depending on the origin of the oil shale and the type of kerogen. A similar behavior is observed during DSC experiments in air. The first (exotherm) peak maximum of both the DTG and DSC curves represents low temperature oxidaton (LTO). Its temperature depends on the type of kerogen in the oil shale. From modelling experiments we know that the LTO peak maximum temperatures of alkanes or other aliphatic hydrocarbons are lower than those of aromatics. The activation energies are also different. Thermogravimetric experiments on oxidation of oil shale in oxygen with kinetic evaluation according to Doyle and Gorbatchev [4-52, 4-53] resulted in considerably varied activation energies for the LTO region (Table 4-157). Generally the activation energies obtained from DSC experiments in air are considerably higher than those from experiments in oxygen. As usual the first peak (LTO) and the last one (fuel combustion) can be evaluated without difficulty, whereas the co-ordination of corresponding peaks from experiments with different heating rates in the interval can Table 4-157: LTO activation energy in oxygen of oil shale. TGA evaluation according to [4-571. Sample

Kerogen Type

Korea Aleksinac (Yugosl.) Aleksinac (Yugosl.) Creveny (France) Knjaievac (Yugosl.) Estonia

I I blend of I and I1 I1 blend of I1 and 111 -

E (kJ/Mol) 56.2 34-39 19-24 27-31 23-29 38.5

346


Table 4-158: Activation energy of DSC oxidation in air of oil shale LTO 250-350 "C E (kJ/Mol)

Fuel Deposition 350-450 "C E (kJ/Mol)

192.6

165.1

144

144

289

134.5

163.7

265.2

45.5

138.6

423.5

Goyniik (Turky) Kerogen type I

132.8

177.6

375.8

Goyniik (Turky) Kerogen concentrate Kerogen type I

124.0

174.5

Sample

Aleksinac (Yugosl.) Sample A Kerogen type I Aleksinac (Yugosl.) Sample A 132 Kerogen types I and I1 Aleksinac (Yugosl.) Kerogene concentrate Kerogen type I Knjaievac (Yugosl.) Kerogen concentrate Kerogen types II and I11

. - . --.

Fuel Combustion > 450°C E (kJ/Mol)

-.

75

50

25

5

10

Fig. 4-121 Isothermal Thermogravimetry of Kerogen Concentrate Aleksinac (Kerogen Type I) at Different Temperatures Atmosphere: Air 25 cm3/min Mass Loss versus Residence Time

15


347

only be done by trial and error. The results of DSC lunetics are listed in Table 4-158. Unfortunately there are not enough data to support any firm statement, but it does seem possible to differentiate the kerogen types using thermoanalytical methods. Isothermal gravimetry at different temperatures did not give very useful results for this purpose. Isothermal gravimetry experiments were performed at final temperatures between 300 "C and 470 "C in air (flow rate 25 cm3/min) using a heating rate p= 100 K/min. This implies that non-isothermal gravimetry occurs for nearly five minutes in the case of the highest final temperature of 470 "C. Fig. 4-121 demonstrates the behavior of kerogen concentrate from Aleksinac oil shale (kerogen type I); Fig. 4-122 for kerogen concentrate from Knjaievac oil shale (blend of kerogen types I1 and 111). Most weight is lost during the heating up to final temperature. All curves in the plot of conversion versus time become asymptotic after, at most, 10 minutes test time, when all of the organic material has been oxidized to volatile products. This holds especially for the low temperature experiments. At around 350 "C the conversion behavior of the samples Aleksinac and Knjaievac differs. Similar oxidation behavior has already been observed for coronene (Fig. 3-31) and for asphaltenes isolated from bitumen (Fig. 4-123).

75

50

25

Fig.4-122 Isothermal Thermogravimetry of Kerogen Concentrate Knjaievac (Kerogen Type I1 and Type 111) Atmosphere: Air 25 cm3/min Mass Loss versus Residence Time

348


N9"C Air

5

10

15 Timelmirtl

Fig. 4-123 Isothermal Thermogravimetry of Asphaltenes from Bitumen B80 (Venezuela) Gas Flow Rate: 25 crn3/min Mass Loss versus Residence Time Parameter: Oven Temperature, Nature of Gas

4.7 Lubricants Lubricants, especially lubrication oils for Otto or Diesel and for jet engines, are exposed to high temperatures in inert and in oxidizing atmospheres during use. They may suffer losses by evaporation and also loss or alteration due to cracking and oxidation reactions. High duty lubricants contain additives, which inhibit early aging, and additives which alter the temperature-viscosity behavior. Papers appear quite early in the literature, which deal with the aging of lubrication oils, and also with the dependence of the low temperature behavior on oil composition and the influence of additives [4-58 to 4-80]. The literature reflects progress in lubricant development and development of thermoanalytical methods of investigation. Difficulties in the development of test methods which have practical relevance and which are able to simulate long term behavior are highlighted. Thermogravimetry, differential thermoanalysis, and differential scanning calorimetry are used as test instruments. Most of the papers describe simulation of the evaporation behavior and try to establish selected values or index numbers from the thermoanalysis as relevant to aging and oxidation stability. The aging behavior is usually described by TGA onset temperatures, and by onset and peak maximum temperatures of DTG, DTA, and DSC curves. Only one publication [4-631 deals with the attempt to ascertain the reaction kinetics of oxidation using the calculation of reaction rates and half life times from the reaction heat, dHldt. Moreover, some of the papers refer to testing of wax content, glass temperatures, and cloud points, but most of the publications describe only the evaporation behavior of lubricants.

4.7 Lubricants

349

4.7.1 Evaporation behavior of lubrication oils We shall consider the practical relevance of the Noack test of evaporation loss (DIN 51 581). An oil sample is held at 250 "C for 60 minutes in a defined flow of air, and weighed before and after this procedure. We cannot say whether the weight loss is based only on evaporation of parts of the original sample, or whether parts of the sample have been oxidized and then evaporated. The repeatability is f 5 % and the reproducibility k 10 %, each for the mean value. ASTM D 972-86 and IP 183/79 use an instrument of different geometric construction and permit a free choice of temperature between 99 "C and 150 "C. An air flow of 2 l/min is applied and the test time is 22 h. The main difference of the ASTM 2595-85 lies in the extension of the temperature range from 99 "C to 316 "C. A reproducibility of & 15 % is quoted for these methods. Blends of standard lubrication oils have been produced for a systematic investigation of the application of thermal methods of analysis. Starting from four vacuum distillates (samples FVAl-FVA4), seven base oils were blended (samples MI-MVII). The subsequent addition of 8 wt% of a performance additive package (Orogil XOA 560, Orogil Table 4-159: Viscosity of Vacuum Distillates (mm2/s) Sample

at 40 "C

at 100°C

VI

FVA 1 FVA2 FVA3 FVA4

16.26 29.52 95.98 483.55

3.54 5.07 10.69 31.15

94 94 94 94

VI = Viscosity Index

Table 4-160: Composition and Viscosity of Base Oils. Sample

FVANo.

wt %

~

MI MI1 MI11 MIV MV MVI MVII

at 40 "C

Viscosity at 100 "C

(mm2/S)

(mm2/S)

~

1 2 2 3 2 3 2 3 3 4 3 4 3 4

35 65 76 24 40 60 15 85 90 10 34 66 48 52

~

at -17.8 "C (mPa. s) ~

24.32

4.54

920

39.94

6.14

2400

59.50

7.89

4850

79.54

9.41

9000

110.64

11.66

15000

161.77

15.08

200.06

17.32

350


Company) to each, provides seven single grade oils with SAE specifications from 1OW to 50W (samples 01-07). A further addition of a methacrylate-based viscosity improver, G 545, to the single grade oil (samples 01 and 02) gave the multigrade oils of SAE specifications 1OW-30, 1OW-40, 15W-30, and 15W-50 (samples 08-01 1). The composition of the blends and their characteristic values are listed in Tables 4-159 to 4-162. The element analysis and the average relative particle weight (@ of the vacuum distillates are given in Table 4-163. Table 4-164 shows the structural group analysis according to Oelert [4-111, which shows predominantly aliphatic structures (approx. 50 %) and naphthenic as well as aromatic structures in equal parts (one quarter each). The structural composition and the viscosities of the base oils MI-MVII can be composed additively from the corresponding data of the vacuum distillates. As expected the viscosity depends on the molecular size and has linear correlations with the molecular weight, the sum of the carbon atoms, ZC, and the maximum of the carbon chain length of the paraffinic molecule part, KLMAX. Thermogravimetry in inert gas of the vacuum distillates and base oils shows that the molecule size governs the evaporation behavior. The evaporation start temperatues T1 % and T5 % are linear functions of the average molecular weight M (Fig. 4-124) and of the chain length of the paraffinic molecule part, KLMAX (Fig. 4-125). Also, the peak maximum temperature T,, of the DTG increases linearly with increasing molecular weight (Fig. 4-126) whereas the evaporation rate decreases (Fig. 4-127). There is a clear linear

a,

Table 4-161: Viscosity and SAE Specification of Single Grade Oils. Sample

(mm2/S)

Viscosity 100 "C (mm2/s)

28.25 44.15 66.15 87.98 121.64 168.49 213.52

5.09 6.77 8.65 10.22 12.63 15.72 12.28

40 "C 01 02 03 04 05 06 07

VI

SAE Specification

114 107 102 98 95 95 95

1ow 20w 20w 20W-30 30W 40W 40W

-17.8 "C (mPa. s) 980 2600 5400 8700 20000 -

Table 4-162: Composition, Viscosity and SAE Specification of Multi Grade Oils Single Grade Oil

Multi Grade Oil

01 01 02 02

08 09 010 01 1

G 545 (wt %)

40°C (mm3/s)

Viscosity 100°C (mm2/s)

-17.8"C (mPa. s)

8.0 16.0 8.0 16.0

48.17 85.39 72.86 123.02

9.45 16.15 11.96 20.06

1350 2120 3350 4650

VI

SAE Specification

184 204 161 187

1OW-30 1OW-40 15W-30 15W-50

351

4.7 Lubricants

Table 4-163: Elemental Analysis (wt %) and Average Molecular Weight of Vacuum Distillates. -

FVA 1 FVA2 FVA3 FVA4

C

H

N

S

H/C

M

86.37 86.11 86.09 85.82

13.39 13.41 13.36 13.12

0.05 0.05 0.05 0.05

0.45 0.39 0.39 1.34

1.86 1.87 1.86 1.83

345 398 526 718

Table 4-164: Structure Group Analysis of Vacuum Distillates.

No. of C-Atoms ZC No. of H-Atoms ZH Aromatic Carbon CA Naphthenic Carbon CN Paraffinic Carbon CP No. of Aromatic Rings RA No. of Naphth.Rings RN Aromacity Max. Chain Length of the Paraffinic Part KLMAX

FVA1

FVA2

FVA3

FVA4

24.81 45.83 24.13 24.76 51.11 0.44 1.12 0.107

28.53 53.03 19.63 26.78 53.58 0.41 1.39 0.085

37.70 69.72 18.62 25.09 56.28 0.53 1.72 0.084

51.36 93.45 23.69 17.07 59.24 0.99 1.59 0.1 16

5.89

7.95

10.65

16.28

- 300 u

0 d

s I n +

250

200

150

400

600

800

R Fig. 4-124 Vaporization Start Temperature T5 % of Vacuum Distillates (FVA1 - FVA4) and Base Oils (MI - MVII) versus Mean Molecular Weight.

352


5

10

15 CLMAX

Fig. 4-125 Vaporization Start Temperature T5 % of Vacuum Distillates (FVA1 - FVA4) and Base Oils (MI - MVII) versus Chain Length KLMAX of the Paraffinic Part of the Molecule.

0

I

I

Fig. 4-126 DTG Maximum Temperature T,, of Vacuum Distillates (ETA1 - FVA4) and Base Oils (MI - MVII) versus Mean Molecular Weight

4.7 Lubricants

-

353

1 5 .O

E

. .r

E

-

12.5

69

a I-

n

10.0

7 -5

200

400

600

m

Fig.4-127 Maximum of Loss Rate DTG of Vacuum Distillates (FVA1 - FVA4) and Base Oils (MI MVII) versus Mean Molecular Weight

correlation of the DTG maximum temperature Tmxwith the chain length KLMAX. For the weight loss rates, DTG, such a correlation cannot be clearly proved (Table 4-165). Thermogravimetry in air of the vacuum distillates is no different to that in argon for temperatures up to 200 "C (Table 4-166). Above 200 "C the curve of the TGA in air is shifted to higher temperatures as shown by the value of AG300. Low-volatile oxidation products are probably present. Of some interest is the question whether the Noack evaporation loss test could be replaced by isothermal gravimetry. In contrast to the Noack test, which only delivers one value after one hour test time, the weight loss versus the test time is recorded by the thermobalance. The important information is whether the evaporation has already reached its final stage. But there is a significant difference in the ratios of surface to volume S : V between the cup of the Noack tester and the pan of the thermobalance (Stanton-Redcroft TG 750): Noack Tester S : V = 0.028 (m-') Thermobalance S : V = 2.0 (mm-') There is a higher weight loss in the thermobalance experiment than in the Noack test, even at equal test times and temperatures because of this. Therefore the temperature in the thermobalance was reduced to 200 'C, and a heating rate p= 100 K/min was applied in order to attain the final isothermal temperature as rapidly as possible. The gas flow rate had the standard value of 25 cm3/min. The results of the Noack test are listed in Table 4-167, those of isothermal gravimetry in Table 4-168. Since it was difficult to establish an exact final temperature in the TG 750, the correct final temperatures are also listed in Table 4-168. Repeated tests showed that the temperature remained within k 0.57 % of the mean during the 90 minutes test time. Comparison between tables 4-167 and 4-168 shows that the losses from the light oils (samples 01, 02, 08-011) in the thermobalance after 10

274 265

0 0.2

25.0 10.0 1.7 0.5 27.5 11.0 2.1 1.o 0.6

FVAl FVA2 FVA3 FVA4

Sample

162 190 215 278

("Cj

("C)

133 157 180 243

T5 %

T1 %

87.8 85.3 30.0 11.3

(%I

(%I 27.0 8.5 3.7 0

AG 300 "C

200 "C

15.3 13.0

99.0 93.8 62.5 10.0 91.0 91.4 81.0 20.3 16.6

(%I

(%I

AG

300 "C

200 "c

Table 4-166: Thermogravimetry of Vacuum Distillates p = 10 K/min; Air 25 cm3/min

227 225

MVI MVII

153 186 238 278 160 181 194 226 26 1

("C)

("C)

131 153 190 227 121 154 160 177 215

T5 %

TI %

FVAl FVA2 FVA3 FVA4 MI MU MI11 MIV MV

Sample

Table 4-165: Thermogravimetry of Vacuum Distillates and Base Oils. p = 10 K/min; Argon 25 cm3/min

77.5 70.0

100 100 94.7 74.5 99.0 96.6 94.5 92.4 94.0

(%I

400 "C

~~

(%/minj 14.29 14.98 16.41 13.50

99.0 95.6 86.1 69.5

DTG

223 256 300 396 235 257 286 350 346 367 370 362

("C)

(%I

400 "C

13.32 11.33 14.27 9.42 12.48 11.87 9.79 9.47 12.11 13.24 8.38 6.95

(%/min)

DTG

234 265 325 334

("C)

597 735 921 0 0 3.0

453 530 596 537

("CI

TEnd

312 399 566 1000 475 527 518 828 0 0 0 0.5 0 0 0 0.5

2

0

F

a

+u

3

P

8&-

El

a a

0

A

P

w

4.7 Lubricants

355

Table 4-167: Evaporation Loss in the Noack Test (DIN 51 581) Sample

Loss (wt %)

Single Grade Oils

01 02 03 04 05 06 07

26.3 16.6 12.0 8.1 5.7 5.0 4.7

Multi Grade Oils

08 09 010 01 1 XOA 560

27.1 24.1 16.6 15.6 24.5

minutes test time, are greater than those at the end of the corresponding Noack test by a factor of 1.5-1.7. The lower volatile oils (samples 03-07) show losses after 10 minutes at 200 "C in the thermobalance, which are similar to those at the end of the Noack test. The plot of weight loss versus test time shows a final stage after only 30 minutes test time for the light oils (samples 01, 02). The other five oils (samples 02-07) did not even reach this stage after 90 minutes (Fig. 4-128). A similar plot for the multigrade oils (sample 08-01 1) shows that a decrease of the evaporation losses occurs as a consequence of the addition of the VI-improver G545 to the single grade oils (samples 01 and 02). This is not so clearly evident in the Noack test. The VI-improver G545 itself loses nearly half of its weight during the test in the thermobalance, since a light oil acts as solvent for the methacrylate polymer. All the samples were tested by isothermal gravimetry in air in order to assess whether the evaporation losses of the additive-treated lubricants could be calculated from the losses of the basic components (Table 4-169). Samples FVA1, FVA2, and FVA3 reach a final value after only 30 minutes test time; the others did not reach this stage, even after 90 minutes. Theoretical evaporation losses were calculated for the additive-treated lubrication oils using the data from the basic compounds (Table 4-170). Comparison between Table 4-168 and Table 4-170 shows that the components of a blended lubrication oil influence one another in such a manner that precise prediction is impossible, and only approximate values of their evaporation losses can be given. Both tests have practical relevance since both kinds of evaporation take place in the oil sump of an automotive engine. The circulating oil passes over plane surfaces where it undergoes thin layer evaporation similar to that in the thermobalance. On the other hand the oil accumulated in the sump has a ratio of surface to volumen similar to that in the cup of the Noack tester.

01 02 03 04 05 06 07 08 09 0 10 01 1

Sample

198.5 199.0 200.5 198.5 198.0 197.5 198.5 198.0 198.0 197.0 197.5

T ("C)

10 min

45.0 36.0 12.5 8.0 4.2 4.8 7.3 45.0 37.6 32.0 23.6

AG (wt %> 199.0 198.5 200.5 198.0 198.5 197.0 198.0 197.5 198.5 197.0 197.5

T ("C)

30 m n

85.0 67.7 30.6 17.6 8.5 8.6 11.8 75.6 66.5 59.6 48.0

AG (wt %)

Table 4-168: Evaporation Loss in the Isothermal Gravimetry in Air at 200 "C. Single and Multi Grade Oils

199.5 198.5 200.5 198.0 199.0 197.5 198.5 198.0 198.5 196.5 197.5

T ("C)

60 min

91.0 79.8 46.4 28.8 14.0 12.8 18.2 86.6 80.0 73.8 63.6

AG (wt %)

198.6 198.5 200.5 198.5 199.0 197.5 198.0 197.5 198.5 196.5 197.5

T ("C)

90 min

92.7 82.0 57.0 38.0 17.0 17.2 22.0 88.8 84.4 79.0 70

AG (wt %)

A

8E:

i l

a

%

5

aiF

z2

pa

h

2

g. 2 9

n

G'

%

3

k

n

3

82

h

3

FVA 1 FVA2 FVA3 FVA4 XOA 560 G 545 MI MII MI11 MIV MV MVI MI1

Sample

198.0 197.4 199.0 199.0 199.5 199.4 197.0 198.0 199.0 198.0 198.0 198.0 199.1

T ("C)

10 inin

56.0 38.0 4.2 1.0 31.3 24.4 50.1 26.3 16.6 4.9 3.9 3.2 3.9

AG (wt %)

198.0 197.5 198.5 199.0 200.0 199.0 197.0 198.0 199.0 198.0 198.0 197.8 201.0

T ("C)

30 min

87.2 81.9 15.3 3.4 44.5 41.5 84.0 67.7 47.4 22.0 16.2 12.1 11.8

AG (wt %)

Table 4-169: Evaporation Loss in the Isothermal Gravimetry in Air at 200 "C. Vaccum Distillates, Base Oils, and Additives.

197.5 197.5 198.5 199.0 200.0 198.0 197.0 197.0 198.5 198.0 198.0 197.8 201.0

T ("C)

60 min

89.0 85.9 31.0 5.3 51.2 51.6 87.1 77.2 62.8 39.8 31.8 23.1 19.5

AG (wt %)

197.4 197.4 198.5 199.0 200.5 198.0 197.0 197.0 198.0 198.0 198.5 197.8 201.0

T ("C)

90 min

90.1 87.1 42.0 8.0 53.4 56.7 88.2 80.9 69.7 49.9 42.0 30.8 25.3

AG (wt %)

u

-a

358


01

02

03 04 05

06

07

I i

Fig. 4-128 Isothermal Thermogravimetry of Single Grade Oils (01 - 07) at 200 "C in Air (25 cm3/min)

Table 4-170: Calculated Evaporation Loss of Additive-treated Oils (wt %). Sample No.

10 min

30 min

60 min

90 min

Single Grade Oils

01 02 03 04 05 06 07

43.3 30.0 18.8 11.0 6.1 5.4 5.0

80.6 70.5 42.1 26.8 16.5 13.9 12.4

84.1 71.0 52.8 40.2 30.3 24.6 21.3

85.4 74.5 59.5 49.1 39.8 32.3 21.9

Multi Grade Oils

08 09 010 01 1

41.8 40.2 29.6 29.1

77.5 74.4 62.4 60.1

81.5 78.9 69.4 67.9

83.1 80.8 73.0 71.6

4.7.2 Oxidation behavior of lubrication oils Testing the oxidation behavior of low boiling lubrication oils does not give satisfactory results, because losses caused by evaporation of parts of the original sample and oxidation products can occur in the same temperature region. Comparison of the TGA curves in argon and air shows that they agree up to a temperature of approximately 200 O C , and thereafter diverge. The formation of low-boiling high-volatile oxidation products shifts the TGA curve towards lower temperatures, whereas formation of low-volatile oxidation products causes a shift towards higher temperatures. Thus the temperature at which the TGA curve in argon first deviates from the curve in air does not always represent the start L

4.7 Lubricants

359

Table 4-171: Start of Oxidation from TGA Experiments.

FVAl FVA2 FVA3 FVA4

240 280 320 310

of the oxidation reaction, as sometimes postulated in the literature. The start temperature of the vacuum distillates (samples FVAl to FVA4)derived from TGA is listed in Table 4-171. But before changes in weight energy effects occur in this method, so measurement by DSC in air at 100 psi (7bar) pressure was proposed by Noel [4-601. The increase of pressure increases the evaporation temperature and decreases the start of oxidation. This method avoids overlap of the endothermal evaporation and the exothermal oxidation in most cases. The oxidation in air at 7 bar pressure of petroleum lubrication oils, polyalphaolefins, and ester lubricants gives diagrams of DSC which display only two distinct peaks, as shown schematically in Fig. 4-129.The curve in the plot of energy flow versus temperature starts with a weak descent. The oxidation starts at temperatures above 150 OC, marked by the temperature TBor by the onset temperature at point A of Fig. 4-129.A first peak maximum occurs between 250 and 320 "C (point B), then the curve descends towards a minimum

H [pwl ex0

I

endo

I

I 100

I

I

2w

I

I

I

3w

4cG

I

I Mx)

I

T["Cl

Fig. 4-129 Pattern of the DSC Recorder Graph of Lubricating Oils Heating Rate p: 10 K/min Atmosphere: Air, Pressure 7 bar, Flow rate 5 cm3/min TB: Reaction Start Temperature A : Onset Point Temperature B : Maximum Temperature of Peak 1 C : Temperature of Minimum 1 D : Maximum Temperature of Peak 2 E : Temperature of Reaction End

360


between 370 and 410 "C (point C). The first peak represents the energies of two oxidation reaction steps i. e. low temperature oxidation, LTO, and fuel deposition. These results are from the DuPont 990 Thermoanalyzer. We report later that measurement with the more sensitive DuPont 9900 Thermoanalyzer provides separation into at least three peaks. This phenomenon is proved by terminating the test at the temperature of the minimum (point C), when the sample pan no longer contains liquids, but a brownish-black lacquer: the 'deposited fuel' which is also found on the pistons of an engine. Sometimes the reaction step of fuel deposition is revealed as a more or less pronounced shoulder on the descending flank of the first peak. If the test is continued beyond the point C, then a second peak appears at approximately 480 "C (point D) and the curve descends thereafter to the base line (point E). The second peak is a typical combustion peak, which represents the heat of combustion of amorphous carbon (30-32 Hlg) when the energy release in this peak is related to the weight still present at point C. This value is quite independent of the type of oil sample, as is the temperature of the second peak maximum. Therefore, the oxidation stability of the sample, can be defined by the first peak. The literature suggests the following data as empirical values indicating the oxidation stability (Fig. 4-129): - Oxidation start temperature as marked by the temperature TB or the onset point A - Temperature of the first peak maximum T-, (point B) - The ratio FV (in %) of the area AGH of peak 1 up to an empirically fixed temperature

(for example 250 "C) FZsodivided by the total area of peak 1 Fpeakl: FV= F250 100 % Fpeakl Indicators for high oxidation stability are: - a high temperature of the onset (point - a high start temperature TB - a small value of the ratio FV

A)

Evaluation of the oxidation tests on the vacuum distillates showns an increasing oxidation start temperature T, (point A) and also an increasing maximum temperature of the first peak Tml as a consequence of the rise of the sity (Fig. 4-130,4-131). The corresponding temperatures for the performance ad XOA560 are similar to those of the high-viscous vacuum distillates, whe rresponding temperatures of the VIimprover G545 are compatible to those of the low-viscous vacuum distillates. The results of the area ratios, FV, are also similar (Fig. 4-132). Blending the four vacuum distillates to the seven base oils leads to a standardization of the oxidation start temperatures, TBto a mean X = 243.4 "C ( k V = 1.70 %). The maximum temperatures of the first oxidation peak were also increased and standardized when the performance additive XOA 560 was used, but not to the same extent as the reaction start temperature (Fig. 4-133). The influence of the performance additive may be observed more explicitly by examining the area ratio FV (Fig. 4-134). Addition of the VI-improver G545 gives unsystematic results for the oxidation stability of the multigrade oils, which sometimes rises and sometimes falls.

4.7 Lubricants

7

-

I

I

I

I

I

I

I

FVA 1

I

N A2

-

I

361

NA 3

1

NA4

I

1

XOA 560

I

G 545

Fig. 4-130 DSC Oxidation of Vacuum Distillates and Additives Heating Rate p : 10 K/min Atmosphere: Air, Pressure 7 bar, Flow Rate 5 cm3/inin Reaction Start Temperature TB

FVA 1

J

FVA 2 M A3 M A4

I I

I XOA 560

Fig. 4-131 DSC Oxidation of Vacuum Distillates and Additives Heating Rate p : 10 Kimin Atmosphere: Air, Pressure 7 bar, Flow Rate 5 cm3/min Maximum Temperature of Peak 1

362


N

[$I 3

I

FVA 1 N A2

I

FVA3

I

NA4

~~~~

~

Fig. 4-132 DSC Oxidation of Vacuum Distillates and Additives Heating Rate p: 10 K/min Atmosphere: Air, Pressure 7 bar, Flow Rate 5 cm3/min Area Ratio FV

Base and s i n g l e grade o i l s

Fig. 4-133 DSC Oxidation of Base Oils and Doped Oils Heating Rate p: 10 K/min Atmosphere: Air, Pressure 7 bar, Flow Rate 5 cm3/min ' Maximum Temperature of Peak 1

Mu1 ti grade oi 1 s

363

4.7 Lubricants

-

15 Base and s i n g l e grade o i l s

k d

Il u l ti grade o i 1 s

2

10

5

i

1 fln .J

Fig. 4-134 DSC Oxidation of Base Oils and Doped Oils Heating Rate p: 10 K/min Atmosphere: Air, Pressure 7 bar, Flow Rate 5 cm3/min Area Ratio FV

Table 4-172:DSC Oxidation in Air, Temperature of the Second Peakmaximum Tm2 Substance

XOA 560 G 545 Vacuum distillates FVA1-FVA4 and Base oils MI-MVII Additive-treated Oils Sample 01-0111

T-~("C)

Coefficient of Variation +V (%)

370 484 479.9

-

1.3

478.7

2.2

-

The maximum temperature of peak 2 is independent of the viscosity of the oils and their additive treatment, and this is shown in the statistical evaluation (Table 4-172). As expected, the temperatures which characterize the start of the oxidation (TB,Ton,,,, and Tm,.) are lower in the experiments at increased pressure than at normal pressure. This gives a more realistic simulation of conditions of pressure loading in an engine. Determination of the oxidation kinetics is a more exact method in the physical sense but it takes rather more time. Evaluation may be restricted to the first oxidation peak. Comparative experiments on the base oil MIII and the single grade oil SAE 20W (sample 03) show no significant differences between the activation energies nor between the frequency factors:

364


Sample Base oil MI11 Single grade oil Sample 03

First oxidation peak E (kJ/Mol) log A (min-') 99.43

9.538

94.14

8.457

13 ("C) 200

250

300-

Fig. 4-135 DSC Oxidation Half Life Time of the First Oxidation Peak (LTO) versus Temperature Line 1: Base Oil MI11 Line 2 Single Grade Oil SAE 20W No. 03

4.7 Lubricants

365

The half life times t1,2calculated using the coefficients above have nearly equal slopes in the diagram of log t1,2versus 1 OOO/T, but the line of the additive-treated single grade oil 03 has been shifted towards higher temperatures (Fig. 4-135). Thus the half life time at equal temperatures of the latter is increased by approximately half a power of ten. Comparison between the area ratios, FV, of the two samples also shows an increase in the oxidation stability of the additive-treated sample, as shown by the quotient FV,, : FVMIn=0.11. The other samples do not show such good agreement of half life times and area ratios. We have shown that the relationship between empirical index numbers is in quite good accordance with the relationship between exactly defined physical values.

4.7.3 Comparison of the oxidation stability of virgin oils, reclaimed oils, and synthetic lubrication oils. The test method for oxidation stability, DIN 51 780 determines the oxygen absorption at a constant elevated temperature as a function of time (it is equivalent to ASTM D 525-86 and IP 40/92, Oxidation stability of gasoline; Induction period method). It is unusual to use this method for lubricants but it is very sensitive. The test records the decrease of oxygen pressure, starting from 7 bar, at 100 "C over 7 days. The oils under this investigation did not experience any measurable drop in pressure. In compensation infrared spectroscopy of the aged samples no new carbonyl or carboxyl frequencies appeared and those already present in the original samples were not reinforced. Therefore we attempted to use thermoanalytical methods for this investigation. The samples for investigation were: two virgin lubrication oils of SAE specification 10W (samples E1.O and E2.0) and SAE 30 (sample E3.0 and E4.0); reclaimed lubrication oils of identical SAE specifications (R1.O, R2.0, R3.0, and R4.0) (Table 4-173); three synthetic lubricants: polyalphaolefin (PA0 5.0); a trimethylol propane ester of short chain Table 4-173: Viscosity and Molecular Weight of Virgin Oils and Reclaimed Oils. Sample No.

Viscosity

VI

SAE Specification

Average Molecular Weight

104 98 93 93 99 109 102 104

low low

400 378 537 529 362 392 509 488

(mm2/S)

E l .O E2.0 E3.0 E4.0 R1.O R2.0 R3.0 R4.0

at 40 "C

at 100 "C

23.59 23.80 110.44 104.06

4.52 4.48 11.71 11.28

25.45 38.20 97.60 80.25

4.67 6.19 11.32 9.98

30 30

low low 30 30

366


Table 4-174: Physical Data of Synthetic Lubricants (Suppliers data).

Density at 20 "C (g/cm3) Neutralization Number (mg KOH/g) Base Number (mg KOH/g) Pour Point ("C) Flash Point ("C) Oxide Ash (wt %) Oxidation Start Temperature ("C) Average Molecular; Weight (g/Mole)* Viscosity at 40 "C (mm/s)* at 100 ' (mm/s)* Viscosity Index*

PA0

TMP

DITDA

XOA 626

0.823 0.01

0.949 0.2 4 -53 240

0.912 0.2 4 -52 250

0.99 (15 "C)

-

-60 222 -

-

-

203 439 22.7 4.8 133

210 48 1 17.8 4.1 139

197 475 26.6 5.3 138

-

86 -

208 12.1 -

1500 70.2 103

~~

*

Measurements : German Petroleum Institute

fatty acids (TMP 6.0); and di-iso-tridecyladipate (DITDA 7.0). The characteristicphysical properties of these products are listed in Table 4-174. The experiments were performed on lubricants without any additive (second digit .O), with 5 wt% additive (second digit .1), and with 11.1 wt% additive (second digit .2). The additive used in these experiments is an additive package for diesel engine oils XOA 626 (Orogil Company). The supplier recommends a concentration of 11.1 wt% of the additive package. The physical characteristicsof XOA 626 are also listed in Table 4-174. According to the suppliers specifications XOA contains 36.2 wt % petroleum based oil and the following elements (wt%): Calcium Phosphorous Zinc Sulfur Nitrogen

3.2 0.8 0.9 3.1 0.48

Thus the main component must be a zinc dithiophosphateacting as antioxidant.Thennogravimetry of XOA (Fig. 4-136) demonstrates low evaporation start temperatures (T1 % = 110 "C, T5 % = 198 "C). Up to a temperature of 250 "C approximately 36 wt% is evaporated. Pyrolysis of the residual components takes place in a temperature region from 300 to 525 "C. The residue at 800 "C amounts to 13 wt%. The experiment in air yields 11.5 wt% ash (inorganic substances). The results of thermogravimetry in argon of the petroleum based lubrication oils without any additive are listed in Table 4-175; those in air in table 4-176. The thermogravimetry of the synthetic lubricants on samples with and without any additives both in argon and in air is shown in Tables 4-177 and 4-178. The oxidation start temperatures, determined by comparisqn of the TGA curves in argon and air, are listed in Table 4-179. The additivetreatment 'shifted the TOxtowards higher temperatures. The values for the synthetics without any additive tabulated here are higher than the values in the suppliers specifications(Table 4-174).

4.7 Lubricants

f

361

Residue ( % )

Fig. 4-136 Thermogravimetry of the Additive Package XOA 626 Heating Rate p : 10 K/min Atmosphere: Argon 25 cm3/min

Table 4-175: Thermogravimetry of Virgin Oils and Reclaimed Oils in Argon. p = 10 K/min; Argon 25 cm3/min AG

Sample E1.O E2.0 E3.0 E4.0 R1 .O R2.0 R3.0 R4.0

T1 %

T5 %

200

300

400

DTG

Tm

153 145 198 179 126 148 170 167

177 179 225 213 172 170 210 202

13.8 12.0 1.5 3.0 13.5 10.0 3.3 2.8

84.9 85.6 30.3 48.3 84.9 79.4 54.0 68.5

94.8 97.6 88.4 85.0 95.4 91.2 88.6 90.8

11.93 11.48 10.59 8.66 11.13 8.99 7.36 8.50

254 254 329 325 264 265 309 297

368


Table 4-176: Theimogravimetry of Virgin Oils and Reclaimed Oils in Air. p = 10 K/min; Air 25 cm3/min AG Sample

TI %

T5 %

200

300

400

DTG

Tm

142 140 198 174 137 127 162 167

177 169 225 215 171 169 200 204

13.0 20.0 1.1 2.8 16.3 14.7 5.0 4.2

85.2 85.3 50.0 53.1 82.5 80.4 68.8 71.0

94.4 95.4 84.8 84.0 93.0 89.5 85.8 87.0

12.31 10.60 17.38 12.34 11.99 9.07 8.82 11.14

256 244 3 10 309 240 259 288 293

E1.O E2.0 E3.0 E4.0 R1.O R2.0 R3.0 R4.0

Table 4-177: Thermogravimetry of Synthetic Lubricants in Argon. p = 10 K/min; Argon 25 cm3/min AG Sample ~~

PA0 PA0 PA0 TMP TMP TMP DITDA DITDA DITDA XOA 626

XOA 626 (wt %)

T1 %

T5 %

200

300

400

DTG

Tmx

163 165 167 179 171 163 146 140 154 110

198 200 206 212 222 207 213 213 205 184

5.7 5.0 3.9 3.0 2.5 3.8 3.5 3.1 4.2 11.0

80.0 80.0 67.0 76.1 60.0 60.0 82.0 70.0 76.0 50.0

91.8 91.7 94.4 89.4 93.0 91.2 87.0 94.2 94.0 65.0

9.74 11.30 8.48 11.85 13.94 9.27 16.19 16.37 14.58 5.15 3.35

266 264 275 283 295 303 287 298 300 239 397

~

0 5.0 11.1 0 5.0 11.1 0 5.0 11.1 100

4.7 Lubricants

369

Table 4-178: Theimogravimetry of Synthetic Lubricants in Air. p= 10 K/min; Air 25 cm3/min

Sample PA0 PA0 PA0 TMP TMP TMP DITDA DITDA DITDA XOA 626

XOA 626 (wt %)

T1%

23%

200

300

400

DTG

Tm

0 5.0 11.1 0 5.0 11.1 0 5.0 11.1 100

163 169 157 183 187 176 143 136 144 116

20 1 204 205 213 227 215 198 214 213 178

4.9 4.1 4.3 2.5 1.o 2.8 5.8 3.5 3.8 9.1

82.1 82.1 83.0 80.0 66.0 70.0 83.9 86.7 80.0 50.0

90.8 90.0 91.5 93.9 93.0 93.2 92.1 94.2 95.0 75.0

11.61 14.66 18.77 12.50 16.32 17.38 10.55 19.74 10.55 4.84 3.26

26 1 259 278 265 293 292 259 279 259 248 361

Table 4-179: Oxidation Start Temperature by Thermogravimetry Tax, p= 10 K/min, Air 1 bar and 25 cm3/min Substance

To, ("C)

Virgin Oils E1.O E2.0 E3.0 E4.0 Reclaimed Oils R1.O R2.0 R3.0 R4.0 Synthetic Lubricants P A 0 5.0 TMP 6.0 DITDA 7.0 Synthetic Lubricants + 5.0 % XOA 626 PA0 5.1 TMP 6.1 DITDA 7.1 Synthetic Lubricants + 11.1 % XOA PA0 5.2 TMP 6.2 DITDA 7.2

285 320 310

Additive Package XOA 626

305

310 265 3 10 310 305 345 340 350 265 280 270 280 310 280

370


The evaporation loss was also determined using the Noack test (DIN 51 581). Simulation by isothermal gravimetry results in considerably higher evaporation losses even at 200 "C test temperature, as shown in tables 4-180 and 4-1 8 1. Despite the specification by the supplier of an oil content of 36.2 wt% in XOA 626, the loss in isothermal gravimetry at 200 "C amounts to nearly 48 wt% after 90 minutes test time. During the first hour of the test the losses of XOA 626 are higher than those of TMP or DITDA. The evaporation losses of the synthetic lubricants decrease with increasing concentrations of additive. We cannot confirm whether this is a consequence of the content of high-boiling components in XOA 626, or whether the generation of volatile oxidation products is prevented. Comparison of the TGA curves indicates that the high-boiling components of the additive package are responsible. Generally, the evaporation losses from the synthetic lubricants are of the order of magnitude of those from petroleum based oils of comparable average molecular weight. The oxidation behavior was further investigated by DSC in air (7 bar pressure, 5 cm3/ min flow rate, heating rate p= 10 K/min). The evaluations were carried out as described in chapter 4.7.2. The petroleum based oils supplied diagrams shown schematically in fig. 4129. No significant differences were observed between virgin and reclaimed oils. In the diagrams of the synthetic lubricants, the first peak appears more slender and higher, the Table 4-180: Evaporation Loss of Virgin Oils and Reclaimed Oils (wt %) Sample E1.O El.l E1.2 E2.0 E2. I E2.2 E3.0 E3.1 E3.2 E4.0 E4. I E4.2 R1.O R1.l R1.2 R2.0 R2.1 R2.2 R3.0 R3.1 R3.2 R4.0 R4.1 R4.2

XOA 626 (wt %) -

5.0 11.1 -

5.0 11.1 -

5.0 11.1 -

10'

42.2 41.0 40.6 59.3 50.7 32.8 7.0 4.3 3.3 7.0 -

11.1

8.1

-

53.9 37.3 26.0 29.0 28.2 27.3 20.2 13.9 12.3 18.7 17.3 14.3

5.0 11.1 -

5.0 11.1 -

5.0 11.1 -

5.0 11.1

Gravimetry at 200 "C 30' 60'

DIN 51581 90 '

84.3 84.3 82.0 91.0 88.0 78.0 16.2 12.1 9.6 18.5

90.6 88.2 84.6 92.3 89.5 84.3 26.3 20.3 16.7 27.0

92.3 89.2 85.3 94.0 91.0 87.9 34.0 22.1 22.2 34.0 -

-

17.0 90.3 76.8 62.3 58.5 58.7 54. I 43.1 31.0 27.7 40.4 34.0 31.0

26.3

32.6 93.0 92.0 85.6 76.4 75.8 72.2 65.7 51.1 46.6 68.0 54.0 51.2

5.2 20.8

-

92.0 83.7 81.8 72.2 70.8 67.3 58.2 43.8 39.4 59.0 46.4 43.9

22.6 -

19.6 27.9 -

18.2 2.6 -

3.6 5.3

-

20.7 17.3 -

16.7 3.4 -

3.5 9.5 9.4

4.7 Lubricants

371

Table 4-181: Evaporation Loss of the Synthetic Lubricants (wt %) ~

Sample

PA0 5.0 PA0 5.1 PA0 5.2 TMP 6.0 TMP 6.1 TMP 6.2 DITDA 7.0 DITDA 7.1 DITDA 7.2 XOA 626

t

XOA 626 (wt %) -

5.0 11.1 -

5 .0 11.1 -

5.0 11.1 100

10' 19.1 17.0 21.0 11.0 10.8 11.0 12.0 7.5 6.3 42.0

Gravimetry at 200 "C 30' 60'

53.8 52.2 51.8 31.2 32.0 31.3 35.5 33.8 33.5 41.8

75.0 72.8 67.8 65.9 60.5 56.1 58.0 55.2 49.9 46.1

EX0

END0

Fig. 4-137 DSC Oxidaton of Pure P A 0 Heating Rate p: 10 K/min Atmosphere: Air, Pressure 7 bar, Flow Rate 5 cm3/min

DIN 51581 90' 81.0 78.0 74.4 72.8 69.2 66.5 70.0 66.3 61.8 47.9

16.4 11.1 12.6 10.7 8.0 7.0 9.2 8.0 6.9 9.8

372


second peak smaller; this is contrary to the corresponding peaks of petroleum based lubricants as shown in Fig. 4-137 (PA0 5.0 without additive). The additive treatment results in a shift of the oxidation start temperatures (TB,To,,,,) towards higher values for each of the substances tested and makes the first peak of the synthetics more slender (Fig. 4-138 and 4-139). The maximum temperatures of the second peak for petroleum based and for synthetic lubrication oils both with and without additives are in the same temperature range. XOA 626 only presents the second maximum at a considerably lower temperature. The additive treatment demonstrates an essential influence on only the first oxidation peak. According to the schematic diagram Fig. 4-129 the following data were drawn from the recorder plots (Table 4-182):

TB

Oxidation start temperature Temperature of the first deviation of the left branch of the curve from linearity

Fig. 4-138 DSC Oxidation of PA0 + 5 % (wt.) XOA 626

Heating Rate p : 10 K/min Atmosphere: Air, Pressure 7 bar, Flow Rate 5 cm3/min

4.7 Lubricants

460

373

T ["C 1

Fig. 4-139 DSC Oxidation of P A 0 + 11.1 % (wt.) XOA 626 Heating Rate p : 10 K/min Atmosphere: Air, Pressure 7 bar, Flow Rate 5 cm3/min

Tonset Onset point temperature Oxidation start temperature Maximum temperature of the first oxidation peak Tmxl Ratio of the partial area (AGH) up to 250 "C temperature upon the total area FV,,, of peak 1 (ABC) Maximum temperature of the second oxidation peak Tm, According to the hypothesis, a substance which has high temperatures TBand Tonset and a small area ratio FV,,, has a high oxidation stability. Each of the substances was tested in five identical experiments and the data were evaluated by statistics. Table 4-182 demonstrates the means X and the coefficients of variations fV for each of the index numbers mentioned above. The values for the virgin and the reclaimed oils of both specifications SAE 1OW and SAE 30 exhibit only small tolerances, so that they may be combined. The denomination A in Table 4-182 includes all four virgin oils, the denomination B all four reclaimed oils.

*

Up to 250 "C no area

5.1 PA0 5.2 PA0 6.0 TMP 6.1 TMP 6.2 TMP 7.0 DITDA 7.1 DITDA 7.2 DITDA 8.0 XOA 626

A.0 Virgin Oil A.l Virgin Oil A.2 Virgin Oil B.0 Reclaimed Oil B.1 Reclaimed Oil €3.2 Reclaimed Oil 5.0 PA0

Substance

5.0 11.1 100

-

5.0 11.1

-

5.0 11.1

-

5.0 11.1

-

5 .0 11.1

-

XOA 626 (wt%)

237.0 239.4 173.5 260.0 255.6 166.0 242.5 250.0 181.3

175.6 226.5 235.6 181.0 219.4 234.4 180.0

X("C)

Table 4-182: Oxidation Stability by DSC. TB

2.98 1.78 2.85 1.36 1.23 0.85 2.92 2.83 6.82

2.70 3.67 3.71 5.96 2.35 1.02 3.93

+V(%)

237.0 244.3 191.0 260.0 271.3 173.5 242.5 258.2 196.3

192.5 243.7 254.1 210.0 238.9 253.5 189.0 2.98 0.61 2.96 1.36 0.53 2.85 2.92 0.69 4.50

2.60 1.66 0.91 1.94 1.33 0.75 2.99

?V(%)

Tonset

X("C)

254.0 273.0 271.5 281.3 285.1 262.5 282.5 282.5 322.3

285.0 250.8 261.3 281.5 247.8 260.0 255.0 0.56 2.32 2.34 0.63 0.58 6.25 1.35 1.25 1.10

9.37 1.88 3.24 5.25 1.53 0.79 2.77

kV(%)

TlTlaXl

X("C)

31.25 8'15 0.70 1.38

*

.k

12.2 0.92 22.03

''.I5 8.28 14.35

8.56

13.26 12.17 8.44

8.69 9.28 10.51

7.53 0.89 10.14 6.90

FV2m L-V(%)

21.55 11.26 0.63 16.24 15.76 0.81 50.04

X(%)

Tm2

0.31 0.43

413.8

0.44 461.3

478.5

3.22

0.74

477.5 463.7

1,69

0.4 467,0

477.0

2,96

1.09

480.3 470.0

3.77 2.71

463.9 460.8

X("C) kV(%)

x-

0

;a

a 5 ap

2 2

F7

2 3

2

&.

g. ::

$

g. *

-.

Q

2 s kg

i?

A Y

P

-1

w

4.7 Lubricants

375

All the means were compared to determine any statistical differences. Thus the following statements may be made:

- Oxidation start temperatures TB and Tonset: For each of the samples the oxidation start temperature increases on additive treatment. A statistical difference exists only for the TB of the reclaimed oils containing 5 wt% and 11.1 wt% additive. To,,,, demonstrates differences for the reclaimed oils, TMP, and DITDA.

- Tmxl:

The additive treatment of the petroleum based oils effects a decrease of Tmxl.Neither virgin oils nor reclaimed oils demonstrate any differences neither with nor without additive. For PA0 the values are equal for the samples without additive and with 5 wt% additive. A significant difference exists between the values of 5 wt% additive treatment and those of 11 wt%. For TMP and DITDA, 5 wt% of additive is sufficient to cause T-, to rise considerably, whereas difference exists between the values of 5 wt% and 11.1wt% additive treatment.

-

Area ratio FV,,,: Generally the values of FV,,, decrease as a consequence of increasing additive concentration. With TMP the additive treatment has such a pronounced effect that no partial area appears, neither with 5 wt% nor with 11.1wt% additive. On the contrary the reclaimed oils do not demonstrate any effect with an additive concentration of 5 wt %. Indeed the area ratios FV,,, of the reclaimed oils are considerably smaller than those of the virgin oils. Whether this is a consequence of the presence of residual amounts of additives in the reclaimed oils or a consequence of the lack of easily oxidable components, which are still present in the virgin oils, cannot be decided.

- Tmxi

The virtual differences are mostly within the tolerances of the method or are caused by different surface structures of the coke residues, which bum in this reaction step. The surface effects are too great for positive statements concerning Tm2.

-

XOA626: T' and Tonset of XOA 626 are of the order of magnitude for lubricants without additives. T-, exceeds the corresponding temperatures considerably whereas T-, exhibits quite low temperatures. FV,,, also demonstrates a low value of 1.38 % in the range of highly additive-treated lubricants. A critical review of this comparative investigation leads to the conclusion that thermogravimetry is suitable for investigation of the evaporation behavior. Isothermal gravimetry supplies results which definitely give more information than the Noack evaporation test. Comparison between the curves of the temperature programmed TGA both in argon and in air should supply the oxidation start temperature at atmospheric pressure. The experimental data do not confirm that the synthetic lubricants should be more resistent to oxidative attacks than petroleum based oils, even without any additive. Generally, unrealistically high oxidation start temperatures have been found by comparison of the TGA curves.

376


Disregarding the fact that, at the beginning of an oxidation, non-volatile oxygen-containing products are generated, in addition simultaneous evaporation and oxidation take place in the same temperature interval. We can only give a limited recommendation of thermogravimetry as a tool for the investigation of the oxidation stability. Increase of air pressure in DSC effects an increase of the evaporation temperature, whereas the oxidation shifts to lower temperatures due an increase of the oxygen partial pressure. This allows some differentiations to be made. Critical comparison of the oxidation start temperatures, the maximum temperature of the first oxidation peak, and the area ratio does not support the assertion that the synthetic lubricants, even without additive, are more stable to oxidation than additive-treated petroleum based oils. The reclaimed oils with no additive exhibit a somewhat higher oxidation stability than the coiresponding virgin oils. Each of the lubricants investigated reacted well to additive treatment and there were no differences between virgin and reclaimed oils. Synthetic lubricants react better to the additive treatment than do the petroleum based oils. This is most clearly demonstrated by TMP. Generally the oxidation start temperatures increase considerably as a consequence of additive treatment if the recommended concentration is used. The increase amounts to approximately 60 "C for virgin oils, to approx. 50 "C for reclaimed oils, to approx. 60 "C for PAO, to 80 "C for TMP, and to 85 "C for DITDA. Although the area ratio FV,,, does not have any physical significance, it provides a very useful method of discrimination. It is evident that a concentration of 11.1 wt% of XOA will be needed to generate sufficent oxidation stability in petroleum based oils, whereas 5 wt% of additive is enough to give valuable protection for synthetic lubrication oils. These investigations cannot tell us anything about the permanence of the oxidation stability: for that purpose we would have to use the correlation of the half life time with the temperature, as determined by the Arrhenius equation, which take much more experimental time. Nevertheless the tests presented are very useful for screening purposes and for comparisons.

4.8 Silicone oils Silicone oils are high temperature lubricants, which are becoming more frequently used because of claims about their thermal stability, and their ability to prolong the interval between oil changes. A particular applicaion is in a continuous laminating press for the production of chipboard, where the lubricant acts also as a heat conducting fluid as well as an hydraulic fluid. Contamination of the fluid by organic substances (wood chips, glues) cannot be excluded and may influence its thermoxidative behavior. A study on the hydrostatic bearings of continuous laminating presses [4-8I] prompted investigation of different silicone oils and other lubricants [4-821. Silicone oils have high thermal and oxidation stability of the base oil, which may be increased by additive treatment.

4.8 Silicone oils

377

Silicone oils or siloxanes are polymeric compounds of silicon, characterized by

groups. The free valences of four-valent silicon are bonded to organic groups such as methyl or phenyl groups. The technically most important silicone oils are polydimethylsiloxanes and the polymethylphenylsiloxanes. Oxidation stability increases with increasing substitution of methyl by phenyl groups. The aging behavior of silicone oils is quite different to that of other lubricants. The oxidation products of hydrocarbons, such as petroleum-based lubrication oils or polyalphaolefins largely cause a continuous increase in viscosity. Some highly volatile cracking and oxidation products are generated simultaneously. Contaminations may accelerate the reaction rate of such an oxidation process. The polyglycol lubricants undergo a rapid drop of viscosity after an induction period of varying length, with concomitant failure of lubricating properties. A sharp decrease of the pH value announces this oxidation effect. The oxidative decomposition of polysiloxanes starts at the organic groups and first causes a decrease in the viscosity. Generation of a sufficent quantity of hydroperoxide and peroxide radicals may induce cross-linking, which might go as far as to change the liquid to the solid phase, with formation of a kind of silicone rubber [4-831. That would have dire consequences for the oil pipe system of the machinery. Table 4-183 lists the characteristic data of the oils under investigation in the study [4-811. The quantities of methyl and phenyl groups were determined by NMR analysis and are listed in Table 4-184. Investigation of the aging behavior, i. e. the oxidation stability, of polysiloxanes involves the well-known problem of simulating long-term aging by a relevant short-term test. The temperatures in the presumed application will range from 190 "C up to 225 OC, which determines the lower temperature limit of the tests. For a conventional test, oil samples of 8 0 g each in open ceramic cups were oven aged in a heating cabinet at 220°C for a prolonged period. At defined intervals the samples were cooled down to 25 "C and their viscosity was tested using a rotation viscosimeter, applying equal shear rates each time. The influence of contamination was simulated by the addition of wood chips in a concentration of 1 wt% and 2 wt%. In an open system weight losses occur by evaporation at the test temperature of 220 OC, so the weight losses were checked hourly, over 8 hours, on different samples. At the same time, isothermal gravimetry was carried out at 200 "C in argon. Fig. 4-140 shows the results from the oven aging experiments and Fig. 4-141 those of isothermal gravimetry. The evaporation of the polymethylsiloxanes corresponds to their molecular size as defined by the viscosity. The evaporation behavior of the polymethylphenylsiloxanes is influenced additionally by the phenyl group content (Fig. 4-142 and 4-143). The different ratios of surface to volume of the sample cups is also taken into account (cup from oven aging: S:V N 0.58; thermobalance pan: S:V 2.0). In addition the high gas flow rate of the thermogravimetry effects a more rapid evaporation. Nevertheless both test methods show very similar trends. N

151 650

Medium content of phenyl groups

High content of phenyl groups

Perfluoroalkylether

25 26

29

1980

366 737 860 96 1 800 100 852

18 19 20 21 22 23 24

1012 209 209 1000 1000

776 104 190 118

2450

Low content of phenyl groups

Polymethylphenylsiloxane

Polymethylsiloxanes Polydimethylsiloxane

Polyalphaolefin

(mPa. s)

lynamic Visosity at 25 "C

13 14 15 16 17

9 10 11 12

I

Sample No.

130

1.04 1.04

-

450

Viscosity Index

1.915

1.065 1.103

1.07 1.10

1.03 1.03

Density

Table 4-183: Characteristic data of the oils under investigation (Suppliers specifications)

I

3 . lo-'

10-6 10-6

Vapor Pressure at 200 "C

21

24.5 28.5

24.2

23.1 23.1

19-23

31

(mN/m)

Surface Tension

0.24

1.46 1.50

1.56 1.56

2

!?

?

F

2a

R

9a

6

2a

a

4.8 Silicone oils

Table 4-184: NMR analysis of Polysiloxanes Sample No.

Phenyl groups

Methyl groups

(%)

(%I

9 10 11 12 17 19 21 24 26

0 0 0 0 16.3 29.6 30.3 30.2 40.4

100 100 100 100 83.7 70.4 69.7 69.8 59.6

1st I-

-I

I

1,0

2.0

3,O

.

I

.

1

.

4,0

I

.

I

.

5,0

I

,

I

.

I

6,0

.

I

.

l

7,0

.

8,0

Fig. 4-140 Vaporization Loss of Polymethylsiloxanes during Oven Aging, 8 hours at 220 "C [4-811

379

380


Evaporation Loss

0

20

LO

(%I

60

80

100

120

KO

Fig. 4-141 Isothermal Thermogravimetry of Polymethylsiloxanes Temperature: 200 "C Residence Time: 120 min Vaporization Loss [4-811

Fig. 4-142 Vaporization Loss of Polymethylphenylsiloxanes during Oven Aging, 8 hours 220 "C [4-8 11

4.8 Silicone oils

Evaporation Loss (%I 4,o

Fig.4-143 Isothermal Therrnogravimetry of Polyrnethylphenylsiloxanes Temperature: 200 "C Residence Time: 120 min Vaporization loss [4-811

100

80

60

LO

20

0

100

200

300

h 0

500

Fig. 4-144 Thermogravirnetry of Virgin Polysiloxanes Heating Rate p : 10 K/min Atmosphere: Argon 25 crn3/rnin .

600

700 T ("C)

-

800

381

239

267

212

259

240

262

220 223

217

10

11

17

19

20

22

24 26

29

246

235 263

323

311

317

287

350

326

313 388

TI

Tmax

426 553 472 539 533

484 307 512 310

5.74 539

2.87 2.49 4.22 99.1 18.75

23.5 34.0

16.0

1.59 4.21 18.9 3.13 3.43 17.8 3.45

14.0

35.6 16.20 436 6.4 3.50 464 7.63 545 15.9 3.01 497 3.73 610 10.7 5.61 606

T5 % AG400 DTG [ T I [wt%1 [%/min]

Virgin Oil

0

32.7 16.4

31.1

32.5

26.0

17.8

2.9

9.8

0 0.7

R800 [wt%]

310 310

310

305

300

333

321

357 362

370

353

350

408

387

377

-

321

-

-2)

["C]

2,

measuring not possible Gelation

12.0 12.3

8.2

10.5

12.1

4.5

7.3

8.6

-

-

-

"C]

Tm

2.56 2.86 8.57 8.63 10.12 9.72 3.82 4.81

517 550 550 504 515 530 528 457

3.38 510 4.44 626 6.26 540

4.81 628

-

DTG [wt%] (%/rnin]

TS % AG400

-1)

T1 % ["C]

Oven-aged pure oil (0 % wood chips)

Sample from a plant, 120 days at an average temperature of 220 "C

237 253

T1 % ["Cl

1 9

Sample No.

33.3 30.1

20.3

14.3

40.0

32.9

7.0

6.0

-

-

[wt%]

R800

209

309 312

332

318

250

369 365

383

378

-

-

-

388

-

2.30

2.89

-

2.36

-

-

4.72

9.4 0.82 11.2 2.45

7.1

7.2

-

6.7

-

556

481 500

511

562

-

586

-

Oven-aged contaminated oil (2 wt % wood chips)

279

Table 4-185: Index numbers of thermogravimetry in inert atmosphere of virgin and oven-aged oils.

SL

Q

s

43.8

- $

P

- 2 g5 51.1 18.7

27.6

- 2

0

2 19.2 3 3

1.3

2 8 E.

3 - ' =

0

3

2+

Y

4.8 Silicone oils

383

Temperature-programmed thermogravimetry in inert gas gives a quick survey of the evaporation behavior, as shown in Fig. 4-144 for some polysiloxanes. For comparison, the TGA curves of a polyalphaolefin (sample 1) and of a perfluoroalkylether (sample 29) are also plotted. Those substances evaporate at considerably lower temperatures than the polysiloxanes despite their substantially higher viscosities. Similar behavior was observed in the isothermal gravimetry. For comparison the oven-aged samples (120 days at 220 "C both with and without wood chips) were also tested by thermogravimetry. The index numbers from those experiments are listed in Table 4-185. Oven aging both with and without organic contamination causes an increase of the evaporation start temperature (T1 % and T5 %) and a decrease of the quantity of the

loor

'0

E

Fig. 4-145 Thermogravimetry of Polymethylsiloxane No. 11 Heating Rate p : 10 K/min Atmosphere: Argon 25 cm3/min Curve 1: Virgin Oil Curve 2: After Oven Aging 120 Days 220 "C Curve 3: After Oven Aging with Addition of 2 wt% Wood chips 120 Days 220 "C

384


distillable fraction (AG400). The TGA results do not indicate whether the samples have undergone an evaporation loss during the oven-aging or whether the molecule size is increasing due to partial cross-linking. The rise of T1 % and T5 % of the polymethylsiloxanes is less in the presence of contaminants than without them. There is an increased reduction oft AG400 for polymethylsiloxanes and for polymethylphenylsiloxanes due to the contamination. The other data, such as DTG, Tm,, and R800, do not show such regular behavior. In Fig. 4-145 to 4-147 the TGA curves of some polysiloxanes are compared. The three curves represent the behavior of the virgin oil, of the oven-aged sample without wood chips, and of the oven-aged sample with 2 wt% wood chips. The oven aging was carried out at 220 "C for 120 days. The polymethylsiloxane (sample 11) shows very small differenr Residue ( I 100 -

f

\

,"\ \

*\

\

'I+

75 -

\

50

-

25

-

-

100

;

400

5

1

600

700

8

Fig. 4-146 Thermogravimetry of Polymethylphenylsiloxane No. 19 (30 % Phenyl Groups) Heating Rate p : 10 K/min Atmosphere: Argon 25 cm3/min Curve 1: Virgin Oil Curve 2: After Oven Aging 120 Days 220 "C Curve 3: After Oven Aging with Addition of 2 wt% Wood chips 120 Days 220 "C

o

4.8 Silicone oils

385

ces between the three curves, whereas considerable differences are shown by the polymethylphenylsiloxanes with a medium phenyl group content (sample 19, Fig. 4-146) and high phenyl group content (sample 26, Fig. 4-147). The results of thermogravimetry in air of the virgin samples are tabulated in Table 4-1 86; those of oven-aged, contaminated samples in Table 4-1 87. There are considerable differences between the TGA curves for the experiments in argon and in air for polymethylsiloxane, sample 11 (Fig. 4-148). Those differences diminish substantially for polymethylphenylsiloxanes, sample 19 (Fig. 4-149) and sample 26 (Fig. 4-150). The shape of the curve is maintained in the graphs from oven-aged, contaminated samples, although they are shifted towards higher temperatures. Residue

o

(%I

T\

l

I

600

700

800

Fig. 4-147 Thermogravimetry of Polymethylphenylsiloxane No. 26 (40 % Phenyl Groups) Heating Rate p : 10 K/min Atmosphere: Argon 25 cm3/min Curve 1: Virgin Oil Curve 2: After Oven Aging 120 Days 220 "C Curve 3: After Oven Aging with Addition of 2 wt% Wood chips 120 Days 220 "C

386

4 Thermoanalytical Investigations on Petroleum and Petroleum Product5

Table 4-186: Thermograviinetry of silicone oils in air (virgin oils). p= 10 K/min, Gas flow rate 25 cm3/min Sample No

TI %

T5 %

AG

R

200

300

400

600

DTG

Tmax

14.74 2.29 1.53 2.03 3.97 4.21 3.97 9.32 5.63 2.44 8.00 6.17 4.29 4.78 2.63 5.23 3.32 3.06 2.49 4.48 2.29 1.89 6.29 4.60 2.60 2.87 2.49 4.44 18.75

335 41 1 518 476 424 476 520 37 1 397 470 370 377 509 375 419 43 1 470 500 553 412 489 506 469 500 529 484 370 512 310

800

1

247

299

0.2

6.5

78.2

0

9

265

333

0.4

1.9

17.0

39.6

39.1

10

250

320

0.1

3.7

38.8

35.7

33.1

11

257

343

0.1

2.8

30.9

35.1

34.2

12

200

273

1.o

7.5

52.1

18.7

16.8

17

220

288

0.7

6.2

15.7

32.1

27.9

19

250

310

0.1

4.3

28.1

33.9

25.7

20

242

305

0.3

4.8

20.8

23.0

19.0

24 26

220 223

285 263

0.7 0.3

6.9 13.4

23.5 34.0

35.1 17.3

32.7 16.4

29

217

246

0.3

58.5

99.1

0

0

Table 4-187: Thermogravimetry of oven-aged silicone oils (aged with 2 wt % wood chips) in air. p = 10 Klmin, Gas flow rate 25 cm3/min AG 300

11

315

350

0.2

19 20 24

301 312 298

360 368 364

0.2 0.1 0.3

318

366

0

26

1 1 1

I 400

R

600

DTG

800 41.0

0.9 0.8 1.1

14.7 12.3 12.3

33.3 40.3 50.1

23.4 28.6 35.6 28.8

1.02 2.09 1.78 1.93 1.41 0.6 1.64 0.93

353 516 430 425 42 1 652 432 645

4.8 Silicone oils

Fig. 4-148 Thermogravimetry of Polymethylsiloxane No. 11 Heating Rate p: 10 K/min Gas Flow Rate: 25 cm3/min Curve 1: Argon Curve 2: Air

381

388

100


Residue

(%I

w \

\\ \

I

75

b\

l

'\

v \

T\J 50

\

\

1 25

2

r

(oc)

__c

Fig. 4-149 Thermogravimetry of Polymethylphenylsiloxane No. 19 (30 % Phenyl Groups) Heating Rate p: 10 K/min Gas Flow Rate: 25 cm3/min Curve 1: Argon Curve2: Air

4.8 Silicone oils

100

Residue

(XI

389

I

Fig. 4-150 Thermogravimetry of Polymethylphenylsiloxane No. 26 (40 % Phenyl Groups) Heating Rate p : 10 K/min Gas Flow Rate: 25 cm3/min Curve 1: Argon

Curve 2: Air

Alteration of the viscosity can be regarded as an indicator of the aging behavior of the sample as a result of oven-aging. For example, the behavior of two polymethylphenylsiloxanes of low (sample 13) and high, (sample 24) phenyl group concentration, is shown in fig. 4-151. It is evident that the increase of viscosity takes place in two stages: the first seems to be only evaporation, and is terminated after approximately 60 days for sample 13. Beyond 100 days, the viscosity starts to rise again, now exponentially,which indicates that an autocatalytic reaction is occurring. For sample 24, the increase of viscosity caused by evaporation continues smoothly after 70 days, joining the increase due to oxidation. The point of inflexion of the curve marks the start of the oxidation reaction. The slope of the

390


12

-

10

v)

2 8 Y

%

c,

.r

2

6

VI .7

w

L

2 0

0

LO

80

120

160

200

210

280

Time (days)

Fig. 4-151 Oven Aging at 220 "C of Polylmethylphenylsiloxanes No. 13 16 % Phenyl Groups No. 24 30 % Phenyl Groups Viscosity versus Residence Time Parameter: Concentration of Wood chips Table 4-188: Oxidation start time to, (days) during oven aging at T = 220 "C Sample No.

9 10 11 12 14 15 16 17 18 19 20 21 22 24 25 26

Stabilization

without wood chips

20 400 325 45 75 100 65 30 290 65 207 68 23 8 71 > 400 > 400

with 2 wt % wood chips

35 200 165 20 145 170 65 50 290 115 113 90 205 55 > 400 > 400

4.8 Silicone oils

391

oxidation curve in the plot of viscosity versus time is described by an exponential equation similar to the Arrhenius equation:

q = a(1) .

' f

Eq. 4-22

The coefficient a(2) is responsible for the slope of the viscosity - time function. Its reciprocal value a(2)-' is a useful measure of the aging behavior. The start times of the oxidation reaction to,marked by the point of inflexion of the curve in the plot of viscosity versus time are listed in Table 4-1 88. Table 4-1 89 gives the values of the aging coefficients, a(1) and a(2), calculated by approximation, and the mean deviation of the calculated curve from the experimental value. From Table 4-188 it is evident that the oxidation start times display very different values, from 20 days up to more than 400 days. The stabilizing effect of the phenyl group becomes visible at high phenyl group content, for example with sample 25 and 26. Some of the samples under investigation had been stabilized by the manufacturers (sample 15, 17, 20, 22, and 24). A distinct prolongation of the induction period is not found in every case. Organic contamination has very different effects on the induction periods, prolonging it for samples 9, 14, 15, 17, 19, and 21, but on the contrary reducing it for the other samples. The coefficient a(2)-' has a high value when the phenyl group content is high, but there is no functional relationship betweeen a(2)-' and the concentration of phenyl groups. The coefficient a(2)-' of the oven-aged samples 14, 16, 17, 19, and 21, which are contaminated by wood chips, is higher than for the samples without contamination. Exact correlations with the index numbers of thermogravimetry have not been found. Only the evaporation behavior may properly be described by isothermal and temperature-programmed TGA. The test times for oven aging at 200 "C are too long for practical application (20 days = 2.9 .lo4 minutes up to more than 400 days = 5.8 . lo5 minutes). Attempting to reduce the experimental period by increasing the test temperature leads to increased overlap of evaporation and oxidation and does not supply any data which are useful in practice. It is more worthwhile to use DSC at moderately elevated pressure to determine the coefficients of the Arrhenius equation for the oxidation reactions, because it is then possible to calculate the reaction rate constant k and the half life time tl,2 and to extrapolate them to temperatures which are used in practical applications. DSC experiments in air at 7 bar pressure supplied completely different plots compared to the DSC oxidation of hydrocarbons (Fig. 4-152). The oxidation start temperature Tonset of the virgin samples is 370 to 400 "C (except the polymethylsiloxane sample 9) (Table 4-190). A first, weak, peak maximum was found between 350 "C (sample 9) and 450 "C (sample 26). A more or less pronounced shoulder on the ascending flank of peak 2 appears in all the samples. The distinct second peak is situated in the temperature range from 415 "C (sample 10) up to 547 "C (sample 20) (Table 4-191). The first and second peaks are shifted towards higher temperatures when the concentration of phenyl groups increases, but this phenomeneon cannot be quantified. The experiments on aged samples give similar results. Contaminationby wood chips causes shift towards higher or lower peak maximum temperatures compared to the virgin samples. In some cases, the activation energy E and

0% 2%

0% 2%

2%

0%

Wood Chips

11

12

41.07 27.62 9.71 3.42 2.52 1.84 20.04 3.95

-

-

-

-

-

40 46

-

37 42

-

50

22 112

-

-

-

Deviation (mean)

17

18

19

167 181

154 63

0.53 0.0065 0.11 0.016

20

189 227

0.78 0.0053 0.81 0.0044

21

18.19 4.82 6.16 7.09 14.41 15.29 28.14 6.02

200 200

0.705 1.43 0.17 0.60 0.027 0.025 0.005 0.006 0.37 0.79 0.198 0.51 0.024 0.0217 0.005 0.0055

16

-

3.37

-

-

15

0.0067 0.032 0.046 0.020 0,167 0.0089 -

14

-

6.88

-

-

0.169 0.248 0.0018 0.0009 -

-

-

-

-

-

10

a(21-l a(21-1

a(1) a(2) a(l) 42)

Coefficient 9

Table 4-189: Coefficients a( 1) and a(2) of the exponential equation for viscosity.

7.43 6.31

263 167

1.18 0.0038 0.69 0.006

22

26

500 500

15.25 8.24

1000 333

0.147 1.33 0.002 0.001 0.11 0.73 0.002 0.003

25

18.73 5.52 9.81 6.23

39 39

0.07 0.026 0.16 0.026

24

g-

a

2 h

R

3

R

3

E

az2

3

0

A

4.8 Silicone oils

Heat Flow

(mu)

No. 2>

Fig. 4-152 DSC Oxidation of Polysiloxanes Heating Rate p: 10 K/min Atmosphere: Air, Pressure 7 bar, Flow Rate 5 cm3/min No. 11: Polymethylsiloxane No. 26: Polymethylphenylsiloxane (40 % Phenyl Groups)

393

265 289 376 365 393 363 408 409 395 384 345

292.5 351.3 379.1 367.2 403.8 411.1 432.5 433.8 420.3 450.0 385.0

256.1 438.8 428.3 439.8 217.3 193.8 217.8 208.7 239.5 358.1 327.0

Virgin oils

23.704 36.906 34.421 36.099 16.547 14.496 15.916 15.195 14.909 25.833 26.261 69.43 63.34 67.25 53.31 46.99 43.30 40.85 44.76 31.34 55.72

0

-

2,

measuring not possible Gelation

-

-

14.235 14.184 22.303 14.499 15.879 18.961 15.080 20.722

-

185.2 180.1 287.8 192.3 218.6 258.8 202.8 278.1

-

360.0 333.8 400.0 406.3 435.0 432.5 419.4 420.0

353 323 383 365 421 411 393 375 -

-

-

-

-1) -2)

Oven-aged pure oils (0 % wood chips)

ample from a plant; 120 days at an average temperature of 220 "C

I 9 10 11 17 19 20 22 24 26 29

Sample No.

Table 4-190: Values of the DSC oxidation of virgin and oven-aged oils.

-

69.86 67.33 48.24 48.63 41.49 42.70 50.98 44.29

-

-

-

352.5 285

9

-

220.6 175.6 -

243.9

-

437.5 420.0 -

471.1

-

408

-

-

-

16.839 47.54

-

15.996 52.34 12.868 47.15

165.4 12.290 49.86 165.0 12.198 49.08

-

B

2

"a

s

$

5

"a

2

'

$. -

CI

b

9

0

01

-

-

-

-

-

-

.:

189.4 12.635 76.12

-

410.0 412.5

-

363 365

350

350

-

-

-

-

-

[rnin-l]

[kJ/Mol] ["C]

c"L 00

Ash [wt%I

logA

E TpiMXl

2 I

Oven-aged contaimed Nils (2 wt % wood chips)

4.8 Silicone oils

395

Table 4-191: DSC oxidation of silicone oils. Temperature ("C) of oxidation peak 2 Sample

Virgin oils

No. 9 10 11 17 19 20 22 24 26

427 415 442 525 541 547 562 531 520

Oven-aged without wood chips

Oven-aged with 2 wt % wood chips

-

-

409 435 511 536 537 54 1 555 535

-

405 -

546 532 -

535 550

the frequency factor A of the first peak are multiples of the values for hydrocarbons, but this could not be correlated with the phenyl group content either. Oven-aging both with and without the presence of wood chips influences the coefficients of the Arrhenius equation in a different manner. Samples 10 and 11 exhibit a considerable decrease. For samples 19 and 29 there is no difference between the virgin samples and the samples aged without contamination. Addition of wood chips effects a substantial decrease of E and A for the aged samples. That is also true for sample 26, i. e. the polymethylphenylsiloxane possessing the highest concentration of phenyl groups. This

Temperature

("C)

___---

1000 f Kelvin

Fig. 4-153 DSC Oxidation of Polysiloxanes Virgin Oils Low Temperature Oxidation (LTO, Peak 1) Half Life Time t,,2 versus Temperature

396


different reaction to organic contamination in DSC corresponds very well to the behavior of the samples in oven-aging. The effects will be further elucidated when we consider the correlation of the reaction constant k or the half life time t1,2 with the temperature (Table 4-192 and Fig. 4-153 to Temperature (‘c)

Fig. 4-154 DSC Oxidation of Polysiloxanes after Oven Aging 120 Days Low Temperature Oxidation (LTO Peak 1) Half Life Time t1,2 versus Temperature

Fig. 4-155 DSC Oxidation of Polysiloxanes after Oven Aging with Addition of 2 wt% Wood chips 120 Days 220 “C Low Temperature Oxidation (LTO Peak 1) Half Life Time t1,2versus Temperature

1 9 10 11 12 17 19 20 22 24 26 29

No.

Sample

kZ50

Virgin oils

2.7081 x 1.3636 x 2.8881 x 10-l2 1.2325 x 3.8410 x 4.3480 x 3.4810 x 1.5229 x lo-* 3.0630 x 1.0142 x 3.6407 x 7.1388 x 1.2449 x 1.3821x 7.3779 x 1.4665 x 1.4134 x lo-* 2.2528 x 2.9269 x 9.8546 x 1.9719 x 1.1851 x 1.4500 x 2.8791 x

boo 2.3189 x loo 8.1870x 6.4621 x 1.0320 x 1.2201 x 10-2 5.5731 x 6.7969 x 1.1578 x 1.4822 x 1.2021 10-4 1.5606 x 2.8832 x

k300

! 6.0919 x 1.8619 x 3.3615 x 1.8657 x 5.5940 x 2.4713 x 4.8995 x 1.04553 x

k200

-

3.6658 x low7 1.9937 x 1.1325 x 10-lo 1.3283 x 6.7622 x 10-lo 8.9376 x

-

loM7 5.4785 x 1.4784 x

k250

1.1784 x 9.4317 x 9.0780 x 2.3841 x 3.9505 x 2.3771 x

-

2.2504 x 5.4722 x

k300

Oven-aged without wood chips

Table 4-192: Reaction rate constant k (min-') of the oxidation of silicone oils (peak 1).

I

1.6347 x 1.4416 x

-

5.9220 x 5.2660 x -

9.2708 x 2.1764 x

-

4.3576 x 3.0563 x

7.7416 x 7.3633 x

-

-

2.3755 x

k300

5.3455 10-9 1.0640 x 9.5576 x

5.3234 x

k250

Oven-aged with 2 wt % wood chips k200

398


4.155). The plot of half life time tI12versus 1 OOO/Tdemonstrates a clear separation of the different species of polysiloxanes and other oils. In Fig. 4- 153 the polymethylsiloxanes, samples 9, 10, and 11 exhibit steeper slopes compared to the polymethylphenylsiloxanes, samples 17, 19,20,22, and 24. They are situated in the region between the polyalphaolefin, sample 1, which has a low thermooxidative stability, and the highly stable polymethylphenylsiloxane, sample 26, which has the highest content of phenyl groups. Oven-aging of the pure samples without contamination results in different decreases of the remaining life time, which may be recognized by flatter slopes in Fig. 4-154. The catalytic effect of the added wood chips increases this reduction of life time generally, except for sample 24 (Fig. 4-155). The half life times t,12at 200 "C and at 250 "C of a modified polymethylsiloxane, sample 11, of three polymethylphenylsiloxanes with approximately 30 % phenyl groups (sample 19, 20, and 24), and of a polymethylphenylsiloxane with approximately 40 % phenyl groups (sample 26), are tabulated in Table 4-193. According to the specifications of the suppliers, samples 20 and 24 contain stabilizer, whereby sample 20 represents the stabilized sample 19. At a temperature of 200 "C the values of tIi2do not reflect any stabilizer effect but at 250 "C such an effect may be recognized. This was also indicated by the behavior of the viscosity after oven aging. Mostly a higher reduction of the remnant stability is found after oven-aging of the samples containing wood chips. The kinetics of the first oxidation peak of polysiloxanesmay be regarded as a criterion of the oxidation stability and this is analogous to the behavior of hydrocarbons. The reaction rate constants k or better the half life times t1,2,determined by DSC, can be correlated very well upon the long-term viscosity behavior during oven-aging of the different types of polysiloxanes. The alteration of the viscosity may be described quite satisfactorily by an approximation equation similar to the Arrhenius equation. Comparison between the exponential coefficient a(2) and the reaction rate constant k proves the trend to conformity of the aging behavior of the pure samples both in DSC and in oven-aging (Fig. 4-156). Organic Table 4-193: Half life times t1,2 (min) of the oxidation of polysiloxanes (DSC peak 1). ~~

~

Sample No.

~~

11

19

20

24

26

tIl2 at 200 "C

Virgin oil Oven-aged without wood chips Oven-aged with 2 wt % wood chips

1.99 x 10l2 3.72 x lo5

5.57 x lo6 3.72 x lo6

9.39 x lo7 1.24 x lo5

2.37 x lo8 1.41 x lo7

3.52 x 1013 6.63 x lo9

1.30 x lo8

6.51 x lo5

7.25 x lo5

1.59 x lo8

2.27 x lo6

t1,2at 250 "C

Virgin oil Oven-aged without wood chips Oven-aged with 2 wt % wood chips

4.55 107 4.69 x lo3

5.02 104 3.48 x lo4

4.73 x 105 6.12 x lo5

7.03 x 105 1.03 x lo5

5.85 x 109 7.71 x lo6

1.30 x lo6

1.17 x lo4

1.32 x lo4

7.48 x lo5

3.18 x lo4

4.8 Silicone oils

399

I Fig.4-156 Correlation of the Oxidation Reaction Constant k (DSC, LTO) to the Viscosity Coefficient a(2) of the Oven Aged Polysiloxanes No. 11, 19, 20, 24, 26. k20GT

' h50-C

contaminants clearly influence the aging behavior of polysiloxanes leading to a reduction of the life time (Fig. 4-157). The calculated reaction rate constants for the temperature of 250 "C give better evidence and should be used for the determination of life time. In the logarithmic plots Fig. 4-156 and 4-157, an increase of the values of k and -a(2) represents increasing oxidation stability. Fig. 4-156 shows that the values horn the long-term ovenaging correlate well with the values of the reaction rate constant for samples 19 and 26, whereas not such a good correlation is found for sample 24. In DSC and TGA experiments, both in air, a residue of SiO, remains. It should be possible to draw conclusions concerning the composition of the polysiloxanes from the quantity of this residue. A pure polymethylsiloxanetheoretically yields 80.87 wt% SiO,, a pure polyphenylsiloxane30.38 wt%. However experiments in air yield considerably smaller residues, which are different in the DSC and TGA (Table 4-194).

- -1 24x

024 20x

19x 26X

020

- -2

019 026

6

W

--3

P N

-

- -4 \

Fig.4-157 Correlation of the Oxidation Reaction Constant k (DSC, LTO) to the Viscosity Coefficient a(2) of the PolysiloxanesNo. 11, 19,20,24,26 (Samples Oven Aged with Addition of 2 wt% Wood chips)

' &JOT

' hSGT

400


Table 4-194: Ash from the oxidation of silicone oils (wt %). Sample No.

DSC in air at 7 bar

TGA in air at 1 bar

9 10 11 12

69.43 63.34 67.25

17 19 21 24 26

53.31 46.99

39.1 33.1 34.2 16.8 27.9 25.7

-

44.76 3 1.34 Polymethylsiloxane Polyphenylsiloxane

-

23.7 13.5

Theoretical

soo,

I 1

80.87 72.59 65.84 65.48 65.53 60.35 80.87 30.38

DSC in air acts as a qualified screening test with relevance to the thermooxidative stability of silicone oils, and saves substantially on experimental time. TGA experiments give additional information on the evaporation behavior. It is impossible to draw conclusions on the composition of silicone oils from the quantity of ash.

4.9 Relation of the kinetics of pyrolysis and oxidation reactions to the system pressure: Investigations on tertiary oil recorvery by in situ combustion. The processes in oil recovery and petroleum manufacture mostly take place at more or less elevated pressures. The manufacture of petroleum includes pyrolysis processes such as delayed coking (pressures from 1 to 6 bar), visbreaking (from 15 to 20 bar), or thermal reforming (from 29 to 70 bar). During tertiary oil recovery by steam flooding, high pressures of saturated steam are required, depending on the pressure in the reservoir. The high temperatures of this steam sometimes cause spontaneous visbreaking and continuing cracking reactions [4-841. During in situ combustion, consecutive and parallel pyrolysis and oxidation reactions occur at the pressure of the reservoir. The fundamental principles employed in selecting process parameters and for predicting temporal progress and yields are derived from an understanding of the influence of pressure upon the reactions concerned. Pyrolysis of hydrocarbons is a first order type reaction (see chapter 3.3.1, Eqs. 3-7, 3-8, and 3-9); whereas oxidation does not obey first order. But it has been found experimentally that it may be treated mathematically as a first order reaction with respect to the consumption of fuel, provided there is an excess of air (oxygen). The relation of the reaction rate to the temperature is described by the Arrhenius equation (Eq. 3-7).

4.9 Relation of the kinetics of pyrolysis and oxidation reactions

401

Such an equation does not exist to describe the relation of the reaction rate to the system pressure. Several papers have been presented concerning the temperature dependece of pyrolysis and oxidation reactions, but only a few investigations have been reported on the pressur dependence. The reaction rate constant of the pyrolysis reaction of pure saturated hydrocarbons at low pressures (1 bar to 7 bar) on the one hand and at very high pressures (140 bar to 1980 bar) on the other, can be found in Reference 4-85, but the pressure dependence has not been investigated systematically. No data of activation energy and frequency factor have been reported. Crudes from Huntington Beach and from West Newport oil fields have been investigated with regard to in situ combustion [4-861in a pressure range from 0.7 bar to 20.7 bar. Visbreaking reactions were postulated with activation energies between 66.6 and 28.3 kJ/Mol and frequency factors (log A) between 9.897 and 6.853 min I. The subsequent cracking reactions give values of E between 21 1.8 and 179.4 kJ/Mol. The frequency factors log A have values between 20.746 and 17.962 min Systematicresearch on pressure dependence has not been reported in this or other publications [4-87 to 4-92]. Pyrolysis is a simple one-step reaction. However, the reaction mechanism of the oxidation of pure hydrocarbons and multi-component systems such as petroleum and its products are much mor complicated. We may generally describe in situ combustion in three consecutive reaction steps [3-91: At first low-temperature oxidation takes place in the temperature range between approximately 210 "C and 260 'C, with the formation of non-volatile oxygen-containing reaction products. The activation energy in this range is 60-70 kJ/Mol. Between 280 "C and 350 "C oxidation and pyrolysis reactions run both consecutively and in parallel. In this stage volatile products are generated, accompanied by the deposition of coke, so this reaction range is called fuel deposition or fuel formation. The average activation energy resulting from the two ractions is approximately 60-80 kJ/ Mol. This range is followed by the last reaction step, which is called fuel combustion, when combustion of the deposited coke takes place with activation energies between 120 and 140 kJ/Mol. The kinetics of LTO have been published [4-93 to 4-96], but only one of them [4-961 deals with the relation of the coefficients of the Arrhenius equation to the system pressure between 1 and 63 bar. Whilst the activation energies between 63 and 65 kJ/Mol are not related to the pressure, a clear correlation has been found for the frequency factor, with log A = 6.18 min-' at 1 bar pressure and log A = 8.08 min-I at 63 bar. The half-life times tliz at 200, 250, and 300 "C calculated using those coefficients are given in Fig. 4-158. The kinetics of the second reaction step, the fuel deposition, are also discussed in the literature [3-9,4461 where pressure dependence of visbreaking and the cracking reaction has been described in a pressure range from 0.7 up to 20.7 bar. The activation energy and the frequency factor decrease linearly with increasing pressure, whereas both coefficients pass a maximum value at about 10 bar pressure in the cracking reaction. The kinetics of fuel combustion are also the subject of publications [3-9, 4-97]. In Reference 4-97 no dependence of the kinetics on the partial pressure of oxygen was found in reactions at atmospheric pressure, whereas Reference 3-9 describes an unquantified dependence on the system pressure.

'.

402


I

I

3‘0

4’0

I

__c

1’0

20

5 0

6 0 P (bar)

Fig. 4-158 Oxidation of Athabasca Bitumen, LTO Range Half Life Time t1,2versus Pressure P [4-871

As mentioned above, knowledge of the relation of the lunetics on the system pressure, for both pyrolysis and oxidation, is of interest for processes in the production of oil and manufacture of petroleum. In tertiary recovery using in situ combustion, considerable pressures, more than 200 bar, may be encountered depending upon the depth of the deposits. Investigations on the kinetics at such high pressures require very robust instrumentation and are not so easy to perform. However, experiments may be performed quite easily using pressure DSC; the instrument used by the author was constructed for pressures up to 70 bar (1 000 psi). Previous experimental work [4-201 provided information on the influence of pressure on the course of pyrolysis and oxidation reactions. We have used a pressure DSC instrument to examine whether the lunetics of pyrolysis and oxidation reactions can be quantifiably related to the system pressure in such a manner that they may be extrapolated to higher pressures. The first investigations were carried out on two model substances (pure hydrocarbons) and three petroleum components (real systems).

*

5

3 4 -

-

-

-

85.13

85.13 84.03

92.26

-

101

C

85.42

("C)

BP 635*

98-99

MP ("C)

from the extrapolated distillatioii curve of n-alkanes

n-Hexacontane n-C60H122 n-Hexylpyrene n-C22H22 Vacuum residue Asphaltenes Dispersion medium

1

2

Substance

Sample

Table 4-195: Analysis data of the samples.

10.73

10.28 7.97

7.74

14.58

0.43

2.79

3.21 4.49

-

-

0.60 1.35

-

S

-

Elemental analysis (wt %) H N

~

865

966 7328

286.4

843.5

M

1.513

1.449 1.138

1.000

2.048

H/C atomic ratio

-R

w

P 0

2

f:g.

2

s

gg.

0

R

a

2.

3

'a

+

b K' ol

5

2

E. o

re

h

;o

404


A long-chained unbranched normal paraffin (n-hexacontane C,,H,,,) and an alkylaromatic consisting of a condensed aromatic four-ring system with a side-chain of an unbranched paraffin (n-hexylpyrene) were selected as model substances. As real substances we used the vacuum residue of a Venezuela crude and its separated components, i. e. asphaltenes and the oily phase (dispersion medium). This selection was made to avoid additional complications due to distillation effects in the experiments on the multi-component systems. The physical and analytical data, i. e. melting point, MP, boiling point, BP, average relative particle weight, M, element analysis, and atomic ratio H/C are presented in Table 4-195. The simulated distillation curves (by TGA) are shown in Fig. 4-159. The asphaltenes do not contain any volatile components but do undergo a cracking reaction above 400 "C. The distillation (sublimation) curve of n-hexylpyrene is nearly 100 "C-150 "C below the corressponding curves of the other substances. The experiments were carried out using a DuPont 9900 Thermoanalyzer equipped with a pressure DSC cell. Gas pressures of 1, 10,20, and 50 bar were applied at a gas flow rate of 5cm3/min to avoid an uncontrolled rise of pressure as a consequence of the rise of temperature, and to avoid the adjustment of any equilibrium. Argon was used as inert gas

Fig. 4-159 Simulated Distillation by Thermogravimetry Curve 1: n-Hexacontane Curve 2: n-Hexylpyrene Curve 3: Vacuum Residue Curve 5: Dispersion Medium from Vacuum Residue

405


in the pyrolysis experiments and air as oxidant in the oxidation experiments. The open solid fat index SFI pans (DuPont) were used for sample pans with empty pan as reference. Measurement and evaluation were performed according to ASTM E 698-79. Each substance was measured at every pressure, applying heating rates p= 5 , 10,20, and 50 K/min. The coefficients of the Arrhenius equation E and A, the reaction rate constant k and the half life time tl,2were calculated using the DuPont DSC Thermal Stability Kinetics - ASTM E 698 software, unless an additional regression calculation was required; this was particularly necessary for the second region of oxidation reaction i. e. fuel deposition.

4.9.1 Pyrolysis tests Pyrolysis presents only a more or less broad peak in the temperature region above 400 O C , as Fig. 4-160 shows for n-hexacontane at 1 bar argon pressure and p = 10 K/min. The peak at 100 "C is the melting peak which will not shift due to alterations in the heating rate. The peak maximum at equal heating rates will be displaced towards higher temperatures as a consequence of rising pressures from 455 "C at 1 bar up to 481 "C at 50 bar. The dependence of the peak maximum temperature upon the pressure at a constant heating rate

10

G

A

a

"

454.77'C

3 3 0

-10

I L 4J (P

L

-20

-30

I

1 OD

20G

I

300 Temperature

400

I

500

(OC)

Fig. 4-160 DSC Pyrolysis of n-Hexacontane Heating Rate p: 10 K/min Atmosphere: Argon, Pressure 1 bar, Flow Rate 5 cm3/min

SOD ~-

3

406


p=

10 K/min is demonstrated in Table 4-196. The low peak maximum temperatures of a-hexylpyrene, are unusual because they are more than 100 "C below those of the other Table 4-196: Relation of the peak maximum temperature ("C) upon the pressure in pyrolysis in argon. Heating rate fi = 10 K/min Sample No.

1 2 3 4 5

Pressure (bar) n-Hexacontane n-Hex y lpyrene Vacuum residue Asphaltenes Dispersion medium

"1

350

1

10

20

50

455 325 450 453 455

47 1 333 455 456 457

477 354 458 461 46 1

48 1 366 412 465 466

,3 5 I

2

L

P (bar)

I

Fig. 4-161 DSC Pyrolysis in Argon Activation Energy Ekersus Pressure P Curve 1: n-Hexacontane Curve 4 Asphaltenes from Vacuum Residue Curve 2: n-Hexylpyrene Curve 5: Dispersion Medium from Vacuum Residue Curve 3: Vacuum Residue


407

substances. Fig. 4-161 represents the plot of activation energy E versus pressure P , and Fig. 4-162 the plot of the frequency factor log A versus pressure P . The low values for n-hexylpyrene indicate that evaporation (sublimation) occurs. No logical explanation can be given for the decrease in the values with increasing pressure. The plot of half life time t1,2versus pressure P at constant temperature demonstrates an increase of t,,2 as a consequence of increasing pressure P for n-hexacontane and for n-hexylpyrene (Fig. 4-163). For the vacuum residue and its components the half life time tIizincreases more or less steeply with increasing pressure at 400 "C. The curve of the vacuum residue becomes flatter as temperatures increase, whereas the curves of the asphaltenes and the dispersion medium pass through a maximum value at higher temperatures (Fig. 4-164). This behavior may be understood from the mode of calculation of the reaction rate constant according to Arrhenius (Eq. 3-7). In this equation the frequency factor is a linear term whereas the activation

25

20

15

18

Fig. 4-162 DSC Pyrolysis in Argon

Frequency (Pre-Exponential) Factor log A versus Pressure P Curve 1: n-Hexacontane Curve 4: Asphaltenes from Vacuum Residue Curve 2: n-Hexylpyrene Curve 5 : Dispersion Medium from Vacuum Residue Curve 3: Vacuum Residue

408


Fig. 4-163 DSC Pyrolysis in Argon Half Life Time t1,2versus Pressure P Parameter: Temperature Curve 1: n-Hexacontane Curve 2: n-Hexylpyrene


I

I

1

I

I

I

I

I

I

I

I

'.5 I k.-

1 0

20

30

4 0

1

409

I

I

.."" Enno

5 0

-

P ( ba ' )

Fig. 4-164 DSC Pyrolysis in Argon Half Life Time t1,2 versus Pressure P Parameter: Temperature Curve 3: Vacuum Residue Curve 4: Asphaltenes Curve 5: Dispersion Medium

energy appears in the exponent only as a fraction (after division by the universal gas constant R and the Kelvin temperature K). It may be seen from Fig. 4-163 and 4-164 that the pyrolysis behavior of the vacuum residue is governed fundamentally by its oil component (dispersion medium). The steadiness of the functions E = f ( P ) and log A = f(P) permits extrapolation towards higher pressures. These plots show that neither n-hexacontane nor n-hexylpyrene are suitable as model substances for simulation of the pressure dependence of the pyrolysis behavior of heavy components o f petroleum.


4 10

4.9.2 Oxidation tests The oxidation reaction comprises three ranges of reaction, i.e. low temperature oxidation LTO, fuel deposition, and fuel combustion, which manifest discrete peaks at different temperatures. For example Fig. 4-165 presents the DSC plot of the oxidation of nhexacontane in 1 bar air at a heating rate p= 5 K/min. An increase of the heating rate shifts the peak maximum temperatures towards higher values, as expected. As a consequence additional peaks appear in the range of fuel deposition, as Fig. 4-166 shows for the example of oxidation of the dispersion medium in 1 bar air at a heating rate p= 20 K/min. An increase of the pressure causes an increase of the area of the LTO peak, whereas peaks in the range of fuel deposition disappear and display only a shouIder on the flank of the LTO peak. The peak of the fuel combustion also becomes wider and flatter (Fig. 4-167, n-hexacontane in 50 bar air, p = 20 K/min).

226.39"C

-2oJ

C

100

200

300

400

500

Temperature ( O C )

Fig. 4-165 DSC Oxidation of n-Hexacontane Heating Rate p : 5 K/min Atmosphere: Air, Pressure 1 bar, Flow Rate 5 cm3/min

60C

7

4.9 Relation

41 1

of the kinetics of pyrolysis and oxidation reactions

I

-20

I

G

'

Id0

'

zcc

3co 400 iempQr-tu.ru ('9

5G0

Fig. 4-166 DSC Oxidation of Dispersion Medium Heating Rate p: 20 K/min Atmosphere: Air, Pressure 1 bar, Flow Rate 5 cm3/min

600

i

0

4 12


Fig. 4-167 DSC Oxidation of n-Hexacontane Heating Rate p: 20 K/min Atmosphere: Air, Pressure 50 bar, Flow Rate 5 cm3/min

4.9.2.1 Range of low temperature oxidation (LTO) The increase of pressure effects a shift of the peak maximum temperatures towards lower values (Table 4-197) at equal heating rates. The dependence of the activation energy E and the frequency factor logA on the pressure are shown in figs. 4-168 and 4-169. Now Table 4-197: Relation of the peak maximum temperature ("C) of the low temperature oxidation (LTO) on pressure in oxidation in air Heating rate p = 10 K/min Sample No. (bar)

1 2 3 4 5

Pressure

n-Hexacontane n-Hex ylpyrene Vacuum residue Asphaltenes Dispersion medium

1

10

20

50

24 1 310 319 326 331

223 300 282 303 287

22 1 277 272 272 268

218 270 256 256 238


I

!i i I

i

-**-t4 I

Fig. 4-168 DSC Oxidation in Air, LTO Range Activation Energy E versus Pressure P Curve 1: n-Hexacontane Curve 4: Asphaltenes from 3 Curve 2: n-Hexylpyrene Curve 5: Dispersion Medium from 3 Curve 3: Vacuum Residue

Fig.4-169 DSC Oxidation in Air, LTO Range Frequency (Pre-Exponential) Factor log A versus Pressure P Curve 1: n-Hexacontane Curve 4: Asphaltenes from 3 Curve 2: n-Hexylpyrene Curve 5: Dispersion Medium from 3 Curve 3: Vacuum Residue

413

414


the n-hexylpyrene reacts similarly to the other substances in the LTO range. It may be observed that the rise in pressure, which implies an increase of the oxygen partial pressure, causes an asymptotic slope of the curves of E and log A versus the pressure P. This will be reflected by the half life times tIiz at the temperatures 200 'C, 250 O C , and 300 "C for example (Fig. 4-170 and 4-171). As expected the asphaltenes possess the highest oxidation stability in the LTO range, which is manifested by the high values of the activation energy, the frequency factor, and thereupon of the half life time. The low values for the kinetic coefficients E and log A for n-hexacontane do not noticably affect the half life times. In contrast to the values for Athabasca bitumen (Fig. 4- lSS), the real systems tested display higher values for their half life times, by approximate half a power of ten.

Fig.4-170 DSC Oxidation in Air, LTO Range Half Life Time tl,* versus Pressure P Parameter: Temperature Curve 1: n-Hexacontane Curve 2: n-Hexylpyrene


415

Fig. 4-1 71 DSC Oxidation in Air, LTO Range Half Life Time tllZ versus Pressure P Parameter: Temperature Curve 3: Vacuum Residue Curve 4: Asphaltenes Curve 5: Dispersion Medium

4.9.2.2 Range of fuel deposition In this range, oxidation and pyrolysis take place immediately consecutively or even parallel, so there is a choice as to whether the exothermal or the endothermal peak maxima should be evaluated. Since the peaks undoubtedly have an exothermal appearance at low heating rates and at low pressures, the temperatures of the exothermal summits were selected, even when several peaks were present in the range of fuel deposition. Sometimes it was difficult to coordinate the corresponding temperatures of the peaks at alternating heating rates, so each of the reciprocal Kelvin temperatures of adjoining peak maxima were

4 16

4 Thermoanalytical Investigations on P e t d e u m and Petroleum Products

used to calculate the regression line. In spite of that, it was not possible to calculate an exact correlation of the activation energy or the frequency factor with the pressure. There is a trend towards a decrease of the kinetic coefficients as a consequence of a rise of the pressure. At a pressure of 50 bar neither n-hexylpyrene nor the dispersion medium displayed any peaks in the the range of fuel deposition. Additional experiments at 35 bar pressure also failed in this respect. On the other hand a second peak can be evaluated for the vacuum residue in the pressure range from 1-20 bar. The dependence of peak maximum temperatures on pressure is non-uniform as Table 4-198 shows. The peak maximum temperatures decrease due to increasing pressures for n-hexacontane, n-hexylpyrene, and asphaltenes, which indicates that oxidation predominates. For the vacuum residue and the dispersion medium the peak maximum temperatures increase as consequence of increasing pressure, which indicates that pyrolysis predominates. The plots of activation energy E and frequency factor log A versus pressure P are given in Fig. 4-172 and 4-173. The half life times at equal temperatures of n-hexacontane, n-hexylpyrene, and asphaltenes decrease as a result of the increase of pressure. On the

300

250

200

150

Fig. 4-1 72 DSC Oxidation in Air, Fuel Deposition Range Activation Energy E versus Pressure P Curve 1: n-Hexacontane Curve 2: n-Hexylpyrene Curve 3: Vacuum Residue Curve 4: Asphaltenes Curve 5 : Dispersion Medium


417

contrary, the values of the vacuum residue and of the dispersion medium rise (Fig. 4-174 and 4-175). The vacuum residue alone supplies two pairs of Arrhenius coefficients at pressures from 1 bar to 20 bar in the range of fuel deposition (Fig. 4-176 and Fig. 4-177). The half life time exhibits a sharp increase caused by the pressure increase similar to the behavior of the activation energy and the frequency factor (Fig. 4-178). This behavior also indicates the predomination of the pyrolysis reaction. Table 4-198: Relation of the maximum temperature ("C) of the fuel deposition peak on the pressure in oxidation in air Heating rate p = 10 K/min Sample No. (bar)

1 2 3 4 5

Pressure

n-Hexacontane n-Hexylpyrene Vacuum residue Asphaltenes Dispersion medium

1

10

20

50

334 427 408 420 409

298 412 435 396 429

285 390 429 385 442

276

Frequency (Pre-Exponential) Factor log A versus Pressure P Curve 1: n-Hexacontane Curve 2: n-Hexylpyrene Curve 3: Vacuum Residue Curve 4: Asphaltenes Curve 5: Dispersion Medium

-

452 363 -

418


Fig. 4-174 DSC Oxidation in Air, Fuel Deposition Range Half Life Time t1,2 versus Pressure P Parameter: Temperature Curve 1: rz-Hexacontane Curve 2: n-Hexylpyrene


I

I

!

I

I

I

415

I

Fig. 4-175 DSC Oxidation in Air, Fuel Deposition Range Half Life Time t,,2 versus Pressure P Parameter: Temperature Curve 3: Vacuum Residue Curve 4: Asphaltenes Curve 5: Dispersion Medium

4

P (bar)

D

Fig. 4-176 DSC Oxidation in Air, Fuel Deposition Range Activation Energy E versus Pressure P Second Value for Vacuum Residue

420


Fig. 4-177 DSC Oxidation in Air, Fuel Deposition Range Frequency (Pre-Exponential) Factor log A versus Pressure P Second Value for Vacuum Residue

Fig. 4-178 DSC Oxidation in Air, Fuel Deposition Range Half Life Time t,,* verus Pressure P Parameter: Temperature Second Value for Vacuum Residue

4.9 Relation of the kinetics of pyrolysis and oxidation reuctions

421

4.9.2.3 Range of fuel combustion The last peak becomes wider and flatter as a consequence of the rise of pressure. This peak represents the combustion of deposited coke, generated by cracking reactions in the preceding reaction range. Moreover the increase of pressure causes a shift of the peak maxima towards lower temperatures (at equal heating rates) except for n-Hexylpyrene, as Table 4-199: Relation of the maximum temperature ("C) of the fuel combustion peak on the pressure in oxidation in air Heating rate fi = 10 K/min Sample No. (bar)

Pressure

1

10

20

50

499 510 513 519 526 592

487 540 493 491 489 537

475 550 455 444 473 504

462 552 452 404 454 482

~

n-Hexacontane n-Hexylpyrene Vacuum residue Asphaltenes Dispersion medium Charcoal

1 2 3 4 5 6

200

150

100

50

Fig. 4-179 DSC Oxidation in Air, Fuel Combustion Range Activation Energy E versus Pressure P Curve 1: n-Hexacontane Curve 2: n-Hexylpyrene Curve 3: Vacuum Residue Curve 4: Asphaltenes Curve 5 : Dispersion Medium Curve 6: Activated Carbon (Charcoal)

422


shown in Table 4-199. For comparison, the data of activated carbon (charcoal) measured under identical experimental conditions are also presented (No. 6 in Table 4-199). The plots of activation energy E and frequency factor log A versus pressure P are shown in figs. 4-179 and 4-180. The coke generated from the asphaltenes exhibits the minor slopes for the E = f(P) and log A = f(P) curves (sample No. 4). Figs. 4-181 and 4-182 show the plots of half life time tl,* versus pressure P. The slope of the curves becomes flatter as a consequence of increasing pressure. The coke from n-hexacontane (sample No. 1) reaches the asymptote at 20 bar pressure. The behavior of the coke from n-hexylpyrene (sample No. 2) is completely different: here the tl,2 = f(P) cnrves increase firstly due to increasing pressure and pass through a maximum at 20 bar pressure. The curves of the coke residues of all the other substances resemble that of activated carbon (sample No. 6 in Fig. 4-182). In Fig. 4-182, the curve of the coke from the dispersion medium (sample No. 5 ) was omitted for clarity since it is nearly the same as that of the coke from the vacuum residue.

Fig. 4-180 DSC Oxidation in Air, Fuel Combustion Range Frequency (Pre-Exponential) Factor log A versus Pressure P Curve 1: n-Hexacontane Curve 2: n-Hexylpyrene Curve 3: Vacuum Residue Curve 4: Asphaltenes Curve 5: Dispersion Medium Curve 6: Activated Carbon (Charcoal)


Fig. 4-181 DSC Oxidation in Air, Fuel Combustion Range Half Life Time t1,2versus Pressure P Parameter: Temperature Curve 1: n-Hexacontane Curve 2: n-Hexylpyrene

423

424


Fig. 4-182 DSC Oxidation in Air, Fuel Combustion Range Half Life Time t1,2versus Pressure P Parameter: Temperature Curve 3: Vacuum Residue Curve 4: Asphaltenes Curve 6: Activated Carbon (Charcoal)

4.9.3 Discussion In chapter 4.9 the results of a first experimental series of the investigation using DSC, of the dependence of pyrolysis and oxidation reactions on pressure are described. First of all the experimental difficulties were rather underestimated by the author. Regulation of the system pressure by pre-setting the pressure on the gas inlet valve sometimes leads to incorrect values since the readings of high pressure reduction valves are too insensitive. That is also true for the manometer installed in the base of the pressure DSC cell. When working with a gas flow, it is advisable to measure the gas pressures at both the gas inlet and the gas outlet of the DSC cell. The application of a gas flow is required in order to guarantee a constant pressure even during variations of temperature. On the other hand the gas flow should help to avoid any condensation of evaporated components within the instrumentation and to avoid reverse reactions as much as possible. Nevertheless there is a

4.9 Relation of the kinetics of pyrolysis and oxidation. reactions

425

danger of alterations to the gas flow rate during the test run, caused by obstruction of the outlet valve by reaction products, such as coke etc., leading to faulty results. We may wish to consider whether the gas flow rate selected in this first experimental series was the optimum already. For these reasons the measured data sometimes showed scattered distribution. The results presented in this chapter all derive from regression calculations, which exhibit coefficients of correlation of at least r 2 0.85. The results should not be regarded as representing absolute values, but they do demonstrate clear trends. During pyrolysis n-hexylpyrene seems to evaporate (sublimate) in spite of high pressures. Although values of the half life time increase due to increasing pressure in spite of a decrease of E and log A, this is a consequence of the mode of calculation as discussed already. n-Hexacontane also demonstrates a different shape of the curves for E = f(P) and log A = f(P) from the petroleum products. The choice of these two substances for modelling the pyrolysis behavior of petroleum products was not very valuable. In the oxidation test the behavior of the substances does not differ so much from that of petroleum products. Generally we have to ask whether using pure individual chemicals to model the pyrolysis and oxidation behavior of multi-component systems makes much sense. The values of the real systems, obtained from experiments at pressures up to 50 bar, may be extrapolated to still higher pressures since E = f(P) and log A = f(P) are continuous functions. The supply of oxygen in the oxidation experiments at 50 bar pressure is sufficient to ensure attainment of the asymptotic limits at least in the first reaction step (LTO). Evaluation of the second reaction step of the oxidation (fuel deposition) is more difficult because an increase of the heating rate provokes the occurrence of additional peaks, which will be flattened as a consequence of a rise of the pressure. For the consecutive and parallel oxidation and pyrolysis reactions in this step, overall values of E and log A have been found, which only give steady functions for the vacuum residue. The data of the last reaction step (fuel combustion) may be evaluated very easily. They also give steady functions for E = f(P) and log A = f(P). All substances tested behave similarly to activated carbon (charcoal). Only the coke residue of n-hexylpyrene reacts completely differently and demonstrates different curves in the plots of the reaction rate constant and the half life time versus the pressure. In this reaction step the curves did not reach the asymptote even at pressures of 50 bar, but they may be extrapolated to higher pressures. There is some uncertainty about all the results since the total conversions can only be ascertained after the termination of the three reaction steps. So the conversion levels at the individual peak maxima of the curves of energy flow versus temperature are unknown. It has been proven by simultaneous TGA/DTA experiments that an isoconversion exists at the maximum of energy flow, independent of the heating rate, which is the condition for the applicabilityof ASTM E 698-79. Howevers the conversionlevel does not influence the external calculation according to ASTM, whereas the DSC software (DSCASTMKin v1.00) comprehends the Ozawa correction [3-181 for the dependence of the activation energy upon the conversion level. Since that conversion level is not known during the DSC experiments, the total conversion after termination of the experiments was found by weighing the sample cup. This total conversion of the pyrolysis was fed into the computer

426


program. The calculation for oxidation experiments was performed with the input of 100 % conversion since there was no residue after the experiment. Evaluation of thermogravimetric experiments according to Flynn and Wall gives slightly higher values of E and log A than DSC experiments. This deviation appears in the values for pyrolysis and low temperature oxidation, as well as for fuel deposition, because of the low conversion levels. However the results from a high pressure thermobalance would fit more exactly. If the reaction rate constants computed using the Arrhenius equation, are published then the values should be given for several temperatures at considerable intervals so that

Fig. 4-183 Oxidation in Air of Activated Carbon and Graphite Half Life Time t1,2versus Temperature (Pressure 1 bar)

4.10 Comparison of commercial computer programs

427

extrapolation is possible. The statement of an activation energy alone does not suffice. The corresponding frequency factors should be also stated. For example, in the oxidation of graphite in1 bar air an activation energy E = 153.36 kJ/Mole and a frequency factor logA = 6.433 min-' have been ascertained. For activated carbon (charcoal) at 7 bar air pressure an nearly equal activation energy E = 154.36 kJ/Mole results but the frequency factor differs by nearly three powers of ten (log A = 9.267 min-I). In the diagram of log t,,2 versus 1000lT' E governs the slope of the line but the intercept is governed by the frequency factor A. Therefore the lines for the two substances are parallel but with a distance of nearly three powers of ten (Fig. 4-183). It is well known from the literature that the chemical composition of the mineral matrix in petroleum reservoirs influences the course of pyrolysis and oxidation reactions. In order to simulate the conditions in petroleum reservoirs, the experiments should be carried out with the addition of finely ground acidic or basic minerals to the organic material. It would be important to add equal quantities of mineral powder to the sample and to the reference pan.

4.10 Comparison of commercial computer programs for investigation of kinetics of pyrolysis and oxidation reactions of heavy petroleum products Four different methods are in use to investigate the dependence of the reaction kinetics on temperature in thermoanalysis, as described in chapter 3.3. The suppliers of instrumentation offer different software packages for that subject. The application of such software to specified inorganic and organic chemicals, to model blends thereof, and to polymers, proceeds without particular difficulties, and has been widely published. Very few publications deal with the application to petroleum and its products, and these are sometimes contradictory. Investigations on the pyrolysis kinetics of oil shale, shale oils, and kerogens have been described in References [4-98 to 4-1021 and [4-47 to 4-51], and investigations on native bitumens in References [4-96,4-1001. Very little has been published yet on the oxidation behavior of crude oils [4-103,4-1041, of native bitumens [4-961, or of oil shale [4-1051. There are few publications on the cracking and oxidation behavior of distillation residues [4-20, 4-46, 4-1061. There is to date no systematic comparison of the different methods of investigation of reaction kinetics. That is understandable since petroleum and its products (especially the distillation and conversion residues) are multi-component systems of an unknown number of non-defined species in variable quantities. The kinetic constants which have been found, are average or overall values of a series of parallel or consecutive reations. That average value does, for the most part, describe the behavior of the multi-component system in the

428


respective process very well. However, petroleum is manufactured by thermal processes, and thermal methods of analysis are most appropriate. A comparative investigation of the applicability of commercial kinetics software was carried out using: - DuPont de Nemours 9900 (990) Thermal Analyzer equipped with pressure DSC cell. Software: DSCASTMKin v. 1.00 (Kinetics according to ASTM E 698-79) - Stanton-Redcroft Simultaneous Thermal Analyzer STA 780 (STA 1000) for TGA, DTG, and DTA. Software: ASTM E 698-79 Borchardt and Daniels Flynn and Wall McCarty and Green Version C 4.20 The principles of the methods are described in chapter 3.3. The behavior of a vacuum residue from a Venezuelan crude was simulated by a distillation bitumen B80 (according to DIN 1995). Further, a vacuum residue of a Middle East crude (VR Kuwait) and its colloid components, i.e. dispersion medium, petroleum resins, and asphaltenes were investigated. Those substances were characterizedby element analysis and average relative particle weight (molecular weight) (Table 4-200) and by analysis of their colloid composition according to Neumann [4-101 (Table 4-201). The pyrolysis reactions were carried out in argon and the oxidation in air in a temperature range from room temperature up to 600 "C (DSC) or 800 "C (STA), applying heating rates p= 5, 10,20, and 50 K/min. The gases had constant flow rates of 5 cm3/min(DSC) or 50 cm3/min(STA). DSC experiments were performed using open aluminum pans (SFI pan from DuPont). In the STA the flat platinum pan (TG 750 crucibles from Stanton-Redcroft) was used. An empty pan was used for reference in each case. Sample weights were in the range from 1 to 3 mg. Table 4-200: Elemental Analysis and average molecular weight. -

Substance Bitumen B80 VR Kuwait Dispersion medium Petroleum resins Asphaltenes

C

H

N

S

H~CatoInar

M

85.13 83.92 83.69 82.46 82.54

10.28 9.84 10.39 10.14 8.08

0.60 0.33 0.32 0.49 1.08

3.21 5.02 4.61 4.91 7.23

1.449 1.407 1.490 1.476 1.175

865 1045 792 1196 6092

Table 4-201: Colloid composition (wt %). Substance Bitumen B80 VR Kuwait

Dispersion medium

Petroleum resins

Asphaltenes

59.60 71.76

21.83 10.50

18.57 17.65


429

Sometimes plots which are produced from the oxidation experiments cannot be evaluated as described above. That happens when the generation of heat occurs so fast that the sample temperature overruns the pre-set heating rate. Smaller sample weights and/or faster heating rates may help to avoid this effect. The DuPont software allows the diagram of heat flow versus time to be called up (to the monitor). After marking the peak maxima, the diagram can be rescaled to heat flow versus temperature and the correct peak maximum temperatures can then be defined. Unfortunately this information cannot be processed by the DSCASTMKin software, so the data are used in external computation of the reaction kinetics.

4.10.1 Pyrolysis reaction 4.10.1.1 Kinetics according to ASTM E 698-79 4.10.1.1.1 DSC (DTA) experiments The pyrolysis of petroleum and its products is characterized by the appearance of only one endothermal peak in the plot of heat flow versus temperature, above 400 "C. Sometimes additional flat peaks will be found below 400 OC, and these represent evaporation processes. The software DSCASTMKm v l .OO (DuPont) supplies the following information, which is shown as plots of the pyrolysis of bitumen B80: Heat Flow versus Temperature log (Heat Rate) versus 1 00O/T Data File Summary (Table) Kinetic Parameters at Different Conversion Levels (Table) Rate Constant and Half Life Time versus Temperature (Table) Half Life Time versus Temperature Specific Rate Constant k versus Temperature % Conversion versus Time (Isothermal) Conversion Time versus Temperature (Isoconversion)

(Fig. 4-184) (Fig. 4-185) (Fig. 4-186) (Fig. 4-187) (Fig. 4-188) (Fig. 4-189) (Fig. 4-190) (Fig. 4-191) (Fig. 4-192)

430


-aeof +-------^TI--

250

300

350

--.-r-------

400 Tenporoturo C * U

Fig. 4-184 DSC Pyrolysis of Bitumen B80 Heat Flow versus Temperature Parameter: Heating Rate p Atmosphere: Argon 5 cm3/min

450

r----i sw


43 1

I

0.8+---------7-

1.380

1.31s

1.370

---r----r-.-1--r---T-1.37s 1.3~0 1.385 1.380 1.3~s 1.400 1000/T

Fig. 4-185 DSC Pyrolysis of Bitumen B80 log Heating Rate versus 1000/T Kinetics according ASTM E 698-79

1.405

1.

10

432


Activation energy: 240.1 kJ/mole Log Cpreaxponential factor)s 17.051 l/mln 60 mln half-life tenporutvro8 387.4 OC Heat o f reaction8 -218.8 Jfg Convereion level: Peak (cureor)

Temperature Temperature

.C K -----------_------400.0 425.0 450.0 475.0 soo. 0 525.0 550.0

673.2 898.2 723.2 748.2 773.2 788.2 823.2

iO~O/Tamp.

1fK

.---------I. 488 1.432 1.383 1.337 1.293 1.253 i . 215

Half-L ife

-------slln

28.48 5.897 I. 363 0.3589

0.1030 0.031W 0.01085

Fig. 4-188 DSC Pyrolysis of Bitumen B80 Rate Constant k and Half Life Time versus Temperature (Table) Kinetics according ASTM E 698-79


-_

30 rl______.__l___

__________

433

__ .___ 1

I I _

Eactt 240.1 XJ/mla Log C f ) t 17.051 l/nln

,

60 nln 1/2 1lCa Entholpyt -218.8 Convm-miont Po&

i---,----t-----r------7--C-----.-7-3-T----i 400

420

440

480 480 500 Tanparaturn (93

Fig. 4-189 DSC Pyrolysis of Bitumen B80 Half Life Time versus Temperature Kinetics according ASTM E 698-79

520

540

560

434


7oT---.---.-.-

--

____I_-_.__I.

E n o t t 240.1 KJ/nola

L o g 0 8 17.051 l/mln 60 ain 1/Z ltfaa 387.4.C

Entholpys -218. B J/g Canvermlona Peak (cureor)

I

II

/

Fig. 4-190 DSC Pyrolysis of Bitumen B80 Specific Rate Constant k versus Temperature Kinetics according ASTM E 698-79


435 1

&-- .

-

,

0

-- T^I-Ic----)--------~-----~5

1s

10

Tina

a 17.0S1 l/mln Eocta 240.1

60 ain 1/2 life1 387.4-C Enthalpy: -218.8 J/g Convermiona Po& (cursor1

t - - - - - - - r - - - l - ~

380

400

420

440

480 480 Twporaturs C 0 O

500

Fig. 4-192 DSC Pyrolysis of Bitumen B80 Conversion Time versus Temperature (Isoconversion) Kinetics according ASTM E 698-79

520

540

.0


437

The DSC only measures energetic data and does not supply any information on conversion. However, in ASTM E 698-79 the dependence of kinetic parameters on the level of conversion is not described (eq. 3-10 and eq. 3-13). Only the occurrence of the peak maxima at equal conversion levels is required, independent of the heating rate. But the DSCASTMKin v1.00 software includes a correction for the dependence of kinetic parameters on the conversion level, introduced for DSC tests by Ozawa [3-181. This correction is not present in the software version C 4.20 for the Simultaneous Thermal Analyzer, STA 790 (STA 1000). Since it is not possible to obtain information on the conversion level at peak maximum temperature, the total conversions were determined by weighing the sample cup after termination of the test runs. This value may be fed to the computer program, DSCASTMKin v l .OO, which offers three alteranatives for calculation of the kinetic parameters. One comprises the automatic detection of the peak maximum temperatures; one is the determination of peak maximum temperatures by setting the cursor; and the third permits calculation of the kinetic parameters at different pre-set conversion levels. Fig. 4-185 shows the regression line of the temperature data in the plot of log (Heat Rate) versus 1 OOO/T; these data were obtained by setting the cursor manually. A check by thermogravimetry proves that the temperatures of the DSC peak maxima correspond to those of the DTG peak maxima when the conversion level is 57 %. Fig. 4-193 shows the regression lines for five different conversion levels (25,50,57,75, and 80 %) in the plot of log (heating rate) versus 1 OOO/T. These data do not satisfactorily fit the regression line at lower conversion levels. At 25 % conversion the regression line has a positive slope, which is mathematically correct but physically irrational. Computation of the kinetic parameters gives a positive activation energy for 25 % conversion, a value of -360 kJ/Mol for 57 9% conversion, and -273 kJ/Mol for 80 % conversion, which are clearly incorrect. Therefore have we used a computation based on data obtained either by setting the cursor manually, or by feeding the total conversion after termination of the reaction into the program. The plot log (heating rate) versus 1 000/T presents a useful control of the accuracy of the peak maximum temperatures of each individual test run. Ideally the data of at least three test runs should fit a straight line in the diagram, otherwise the program will calculate a regression line using the method of least squares. The slope of the line is equivalent to the term EIR. That sometimes implies considerable deviation of the activation energy E (see chapter 4.10.3). No information is provided on the coefficient of correlation and therefore it is not possible to evaluate the scatter. The results of the DSC experiments are presented in Table 4-202. The software of the Simultaneous Thermal Analyzer, STA 780 (STA 1000)also allows the ASTM E 698-79 method to be used. Unfortunately there is no computer program to process data from simultaneous TGA/DTA (TGA/DSC) test runs. For computation according to ASTM E 698-79 only separately sampled DTA (DSC) data can be used, so the advantage of a simultaneous measuring instrumentation cannot be fully realized. The conversion levels at peak maximum temperatures are also unknown, and the total conversion can only be ascertained after termination of the test run. The installed software version C 4.20 offers plots similar to those of the DSCASTMKin v1.00:

438


1.

I.

3 8

dgr -I

lo 0.

0.

0.

0.

i0

Fig. 4-193 DSC Pyrolysis of Bitumen B80 log Heating Rate versus Temperature at Different Conversion Levels Kinetics according ASTM E 698-79

Table 4-202: Pyrolysis kinetics according to ASTM E 698-79 A) Results from DSC experiments No.

Substance

1 2 3 4 5

Bitumen B80 Bitumen B80 VR Kuwait VR Kuwait Dispersion medium from 3 Petroleum resins from 3 Asphaltenes from 3

6 7 ~~

~

E (kJ/Mol)

log A (min-')

Conversion (70)

273.8 240.1 211.7 229.1

19.249 17.051 15.016 16.231

80 57 83.7 81.4

239.3

16.877

89.2

223.1

15.579

80.2

220.6

15.42

50.9

~~

No. 1 to No. 3 from experiments in 1 bar argon No. 4 to No. 7 from experiments in 10 bar methane

439


Raw Date plot corresponds to Heat Flow versus Temperature In (Heating Rate) versus 1 000/K corresponds to log (Heat Rate) versus 1 000/T The following plots are identical: Half Life Time versus Temperature % Conversion versus Time (Isothermal) Conversion Time versus Temperature (Isoconversion) In the plot of Reaction Rate Constant k versus Temperature, the program version C 4.20 uses In k, whereas DSCASTMKin v1.00 uses the absolute value k. The kinetic parameters computed according to ASTM E 698-79 from DTA data of the Simultaneous Thermal Analyzer are presented in Table 4-203 No. 1 and No. 2. Table 4-203: Pyrolysis kinetics according to ASTM E 698-97, experiments with Simultaneous Thermal Analyzer. No.

Data from

Substance

E (kJ/Mol)

log A (min-')

Conversion*

(%I ~~~~

8 9 10 11 12

~

Bitumen B80 VR Kuwait Bitumen B8O Bitumen B80 VR Kuwait VR Kuwait Dispersion medium from 2 Dispersion medium from 2 Petroleum resins from 2 Petroleum resins from 2 Asphaltenes from 2 Asphaltenes from 2

DTA DTA DTA DTG DTA DTG

214.5 212.7 195.2 193.5 209.4 209.4

15.003 14.986 13.365 13.452 15.073 15.073

53.9 55.5 52.3 52.0

DTA

188.1

12.243

57.7

DTG DTA DTG DTA DTG

190.5 190.7 206.8 211.6 224.1

13.366 13.161 14.252 14.782 15.567

59.7 36.0 34.0 23.7 23.0

-

No. 1 and No. 2 from separately collected DTA data No. 3 to No. 12 externally computed using peak maximum temperatures of simultaneous TGA/DTA measurements * conversion at the temperatures of peak maxima

4.10.1.1.2 Kinetics according to ASTM E 698-79 from simultaneous TGA/DTA experiments The standard evaluation program of the Simultaneous Thermal Analyzer STA 750 supplies the plots of TGA and DTA versus temperature. In addition, the curve of the first differential quotient of the weight loss with respect to time dGldt (DTG) may be plotted (Fig. 4-194). The peak maximum temperatures from both DTA and DTG may be determined quite easily and used to calculate the activation energy and frequency factor. There is no program to process such data, so the calculation is done on an external computer. The

440


I00

-

90

-

80

-

-n

70

-

>

60

-

1

*to

POINT TABLE

-

0

1

1.2

--

0

C

-

L

-

; 1.0 m

.a

--

k

0

-

.6--

-

+I 0

50

-

40

-

30

-

20

0

-I

4

L

o.

--

4J

4J

C

1.4

.4-

.2

463.8 1bO

2b0

3b0

4b0

c 5bO

7b0

8bO

-

0.0 --.a --

RESIDUE 22.59 L 6b0

-

gb0

Fig. 4-194 Pyrolysis of Petroleum Resins from Vacuum Residue Kuwait STA Diagram: TGA, DTA, DTG Heating Rate p: 10 K/min Atmosphere: Argon 30 + 20 cm3/min

advantage of this kind of experiment is the easy determination of the conversion levels at both the maximum heat flow (DTA) and the maximum reaction rate (DTG). The experiments prove that isoconversions at the corresponding DTA and DTG maxima are independent of the heating rates, so ASTM E 698-79 may be applied. The results of this investigation are given in Table 4-203 (No. 3 - No. 12).

4.10.1.2 Kinetics according to Borchardt and Daniels Software only for the Simultaneous Thermal Analyzer STA 780 was available for kinetics according to Borchardt and Daniels. Similarly to the kinetics program accordingto ASTM E 698-79, only separately sampled DTA data can be processed using the Borchardt and Daniels program. The DTA data from simultaneousTGA/DTA experiments cannot be

-6

-7 -8 -0


441

processed. Moreover the program can only process the data from positive (exothermic) peaks, but that can be overcome easily by exchanging the positions of the sample pan and the empty reference pan on the pan platform. Evaluation of the DTA curve is rather difficult. The main uncertainty is based on the correct selection of the integration limits. How should the DTA curve be interpreted? Experience tells us that no pyrolysis reactions occur in the temperature region below 400 O C , but the DTA pyrolysis curve of bitumen B80 demonstrates an onset point in the range of 250 "C (Fig. 4-195). Is that the start temperature of the pyrolysis reaction or does evaporation still take place up to 400 "C or 410 O C , where the DTA curve exhibis another point of inflexion? Using the temperatures of 400 "C and 500 "C as integration limits the calculation gives unrealistically high values of activation energy, E = 338.7 kJ/Mole, of frequency factor log A = 24.368 min-', and of reaction order n = 3.13. Pyrolysis reactions

-6

-

-e

-

-10

4

6

Taperrturc Limits: 400

I

100

'\\

- 500 'C

I

200

1

300

I

I

400

OOQ

Fig. 4-195 Pyrolysis of Bitumen B80 STA Diagram: DTA Heating Rate p: 10 K/min Atmosphere: Argon 30 + 20 cm3/min Kinetics according Borchardt and Daniels Integration Limits: 400 "C and 500 "C

c

500

I

600

I

700

e

442


normally obey a first order law with n = 1. The choice of 250 "C and 5 10 "C as integration limits gives E = 75.4 kJ/Mole, log A = 4.795 min-', and II = 1.17. Those values favor an evaporation process. It is possible to find reaction orders in the vicinity of the unity combined with acceptable values of the kinetic parameters by trial and error (Table 4-204). Since the lack of thermogravimetric data increases the uncertainity of the interpretation of the DTA curves, we have decided not to use the Borchardt and Daniels method. Table 4-204: Pyrolysis kinetics of bitumen B80. Evaluation of DTA curves according to Borchardt and Daniels. Experiments with Simultaneous Thermal Analyzer. Temperature limits ("C)

Reaction order n

E (kJ/Mol)

log A (min-')

260 - 510 300 - 510 330 - 510 375 - 500 372.5 - 505 376.5 - 515 377.5 - 500 380 - 510

1.17 1.076 0.77 0.92 1.13 1.16 1.17 1.41

75.435 93.171 102.330 158.87 164.77 176.36 175.81 196.55

4.7946 6.1196 6.7403 10.9354 11.4045 12.2645 12.2210 13.7758

4.10.1.3 Kinetics according to Flynn and Wall This method is based on the evaluation of the curves of at ,:ast three TGA experiments at different heating rates. Fig. 4-196 shows the plot of the pyrolysis of bitumen B80 at a heating rate p = 10 K/min. There is only a small difference between the DTA and DTG maximum temperatures within the limits of the method. At peak maximum the conversion is still incomplete and amounts to only 57 % whereas the total conversion is 80 %. For evaluation according to Flynn and Wall, sections of the curve comprising the regions of special interest may be called up or the whole curve may be evaluated. Fig. 4-197 represents the raw data plot of three pyrolysis experiments on bitumen B80 at different heating rates. The lines of isoconversion may be set independently by random entry or selected manually, so any points of special interest in the thermogram may be marked. This is particularly important if the TGA curve consists of different regions possessing different slopes, for example in the oxidation test (see chapter 4.10.2.3). The evaluation program presents the following plots: In Heating Rate versus 1 000/Kelvin Arrhenius Plot: In k versus 1 000/Kelvin Specific Rate Constant versus Temperature Half Life Time versus Temperature

(Fig. 4-198) (Fig. 4-199) (Fig. 4-200) (Fig. 4-201)


Isothermal: Percent Conversion versus Time Isoconversion: Time versus Temperature Activation Energy versus % Conversion Table of Results

443

(Fig. 4-202) (Fig. 4-203) (Fig. 4-204) (Fig. 4-205)

I10

9

100

E

90

7

70

5 ;

Q

0

4

k I

.?I

50

3

40

2

30

1

20

0

10

-1

Fig. 4-196 Pyrolysis of Bitumen B80 STA Diagram: TGA, DTA, DTG Heating Rate p: 10 K/min Atmosphere: Argon 30 + 20 cm3/min

444


100

-

.4

90

-

.3

80

-

.2

70

-

.l

60

-

0.0

a

U

iu

> 0

U

k a

;

50

-

-.:

40

-

-.2

30

-

-.3

20

-

-.4

10 200

250

3b0

I 350

A0

4bO -g

-.5 I 500

c

Fig. 4-197 Pyrolysis of Bitumen B80 STA-Diagram: Raw Data Plot, Lines of lsoconversion Heating Rate p : 5, 10, 20 K/min Atmosphere: Argon 20 + 30 cm3/dn Kinetics according Flynn and Wall

5bO

&I0

650

k

...I

..


445

3.4

3.2

;3 . 0 4J

W

a m

2.8

C

. I

2.6

m

I 2.4

.-I

2.2 2.0

1.E

1.6

0

I

1.35

1.I40

I

1.45

I

1.W

1000/Kelvin

1.?i5

I

1.60

1

1.65

Fig. 4-198 Pyrolysis of Bitumen B80 STA Diagram: In Heating Rate versus 1000,lKelvin at Different Conversion Levels Kinetics according Flynn and Wall

1

446


3.0 2.5 2.0

-t . I

Y

1.0

-s

C H

0.0

-.I -1.0

-1.5 -2.0

Fig. 4-199 Pyrolysis of Bitumen B80 STA Arrhenius Plot: In k versus 1 000/Kelvin Kinetics according Flynn and Wall

441


16

-

14

-

iz

-

10

-

----

1

ill

- - - - ~

/

/

8 -

64-

2-

0-

+

340

'

I

360

I

I

400

4h0

I

440

neg c

I

460

Fig. 4-200 Pyrolysis of Bitumen B80 STA Diagram: Specific Rate Constant k versus Temperature Kinetics according Flynn and Wall

4b0

I

500

!O

448

4 Thermoanalyticul Investigations on Petroleum und Petroleum Products

3.3

3.0

-5

2.5

E

Y

m

y.

2.0

J y.

n

2

1.5

1.0

.5

0.0 3i0

3kO

!

I

380

400

I

460

440

Fig. 4-201 Pyrolysis of Bitumen B80 STA Diagram: Half Life Time versus Temperature Kinetics according Flynn and Wall

4k0

I

I

480

500


Fig. 4-202 Pyrolysis of Bitumen I380 STA Diagram: Percent Conversion versus Time (Isothermal at 450 "C) Kinetics according Flynn and Wall

449

450


3.5

3.0

i!

-

j

I II

2.1 E ."

L-

'D

2.0

-

I

-

1

I

4

m

L

-

1.5 0

0

1.0

-

.5

-

0.0

380

i

j I

1

!

702

4bO

I

420

I

440

I

460

I

480

Fig. 4-203 Pyrolysis of Bitumen B80 STA Diagram: Conversion Time versus Temperature (Isoconversion, Conversion Level 70 %) Kinetics according Rynn and Wall

I

500

I

520

540


C ~ m n x mw x

I m QDeR

ACTIVATION -I PFII!+xm”lIAL

-

451

1 1

Z RJ/uelm J :

CACTaR l L / u l n l :

XOC.67SS

P. 1 ~ B a . c + i 4

Fig. 4-205 Pyrolysis of Bitumen B80 STA Diagram: Table of Kinetic Results Kinetics according Flynn and Wall

Table 4-205: Pyrolysis kinetics according to Flynn and Wall. Experiments with Simultaneous Thermal Analyzer. No.

1 2 3 4 5 6 7 8

9 10 11 12

Substance Bitumen B80 BitumenB80 Bitumen B80 VRKuwait VRKuwait VRKuwait Dispersion medium from 4 Dispersion medium from 4 Petroleum resins from 4 Petroleum resins from 4 Asphaltenes from 4 Asphaltenes from 4

E (kJ/Mol)

log A (min-I) Conversion (%)

213.23 202.77 190.85 202.53 180.30 182.31

14.708 14.160 14.050 13.980 12.610 12.786

70 (Max.Conversion) 56 (DTG Maximum ) 54 (DTA Maximum) 83 (Max. Conversion) 52 (DTG Maximum) 51 (DTA Maximum)

213.30

14.643

94 (Max. Conversion)

198.15 329.72 230.77 253.89 218.75

14.025 20.060 15.975 16.873 15,955

60 (DTG Maximum ) 75 (Max. Conversion) 34 (DTG Maximum) 53 (Max. Conversion) 23 (DTG Maximum)

In experiments No. 8, 10 and 12 the temperature data of DTG and DTA give identical values.

452


-m

-

220

-

210

-

200

-

r (

\ 3 Y

-

----

-~

,-__-_--------

230

1

z

: 0

140

C

W

J

160

-

0

4

160

-

150

-

170

140 I

10

I

20

1

30

I

I

I

60 50 40 P e r c e n t Conversion '

Fig. 4-204 Pyrolysis of Bitumen B80 STA Diagram: Activation Energy versus Percent Conversion Kinetics according Flynn and Wall

f0

1

80

I

9(


453

The plot of In Heating Rate versus 1 000/Kelvin is a useful control of the precision of the individual test runs. The plot of Activation Energy versus % Conversion and the Table of Results both enable comparison of this kind of evaluation with the results according to ASTM E 698-79, which were calculated for defined conversion levels. Table 4-205 shows the kinetic parameters determined according to Flynn and Wall. The values for maximum conversion and for conversion levels corresponding to the DTG maxima (which also equal the DTA maxima) are presented. The kinetic parameters for conversions at the DTG (DTA) maxima can be compared with the corresponding values found according to ASTM E 698-79 and given in Table 4-203.

4.10.1.4 Kinetics according to McCarty and Green The method of McCarty and Green is most useful since only one TGA experiment is required, but selection of the correct limits for the integration raises problems. If the limits are set to the two horizontal branches of the TGA curve, for example to 250 "C and 520 OC, then the values of the kinetic parameters of bitumen B80 will be far too small as shown in Fig. 4-206 with E = 120.18 kJ/Mol and log A = 8.160 m i d . We must consider which physical and/or chemical processes occur during the rise in temperature. Evaluation according to Borchardt and Daniels (chapter 4.10.1.3) revealed a heat flow in the temperature region far below 400 OC, as a consequence of some reaction. The TGA curve in Fig. 2-206 shows that weight loss begins below 400 "C, which is generally considered as the lower limit of the pyrolysis reaction. An experiment starting with an isothermal run at a temperature below 400 "C, followed by a dynamic run up to high temperatures, may clarify the behavior. For example, an isothermal run at 375 "C for 15 minutes is followed by a dynamic run using a heating rate p = 10 K/min up to 525 "C has been performed. At the end of the isothermal step a weight loss of 15 % is recorded, caused by evaporation. Subsequent integration using the limits 375 "C and 500 "C gives E = 173.72 kJ/Mol and log A = 12.086 min-l. These values are still a little bit low. The results most comparable to the other methods are obtained by selecting the onset and offset point temperatures as integration limits. From Fig. 4-207 it can be seen that about 15 % of the sample is evaporated at the onset point temperature (400.4 "C) while the residue at the offset point temperature (482.5 "C) is approx 24 %. Calculation using 250 "C and 400 "C as limits gives the kinetic parameters of evaporation, i.e. E = 109.82 kJ/Mol and log A = 8.360 min-'. Integration using the temperature of the onset point (400.4 "C) and offset point (482.5 "C) as limits gives E = 202.81 kJ/Mol and log A = 14.258 min-', which represent values for a pyrolysis reaction. When applying low heating rates the kinetic parameters do not show any significant dependence on the heating rates (Table 4-206, bitumen BSO). Table 4-207 shows the kinetic parameters of VR Kuwait and its components. The values for evaporation of the asphaltenes are already in the range of a pyrolysis reaction. The curve of pyrolysis of the petroleum resins shows two reaction regions with different slopes (Fig. 4-208). The first onset appears at 432 "C. The offset point of this branch of the curve is at the same time the

454 110

4 Thermoanalytical Investigations on Petroleum and Petroleum Products L i n i t r PL~O t o

SOp r d c

100

90 80 *I

m U L

a

70 60 90

nmw

40

30 20 10 I

100

2bO

1

300

I

400

Deg C

Fig. 4-206 Pyrolysis of Bitumen €380 STA Diagram: TGA Heating Rate p: 10 K/min Atmosphere: Argon 30 + 20 cm3/min Kinetics according McCarty and Green Integration Limits: 250 "C and 520 "C

5bO

I

600

I

700

I

0

455

4.10 Comparison of commercial computer programs 130

100

90

80 70 60

50 40

30 20

'

10

'

0

I

100

2b0

I

900

4b0

Fig. 4-207 Pyrolysis of Bitumen B80 STA Diagram: TGA Heating Rate p: 10 Klmin Atmosphere: Argon 30 + 20 cm3/min Onset Point and Offset Point Temperatures

5b0

6bO

7b0

456

100

C

m L 0

a

! i

-

go=-RERlJs

10

-

70

-

u


432

- 475 'C

I

i I I 1

REALTS 476

I

- 620 *C

i

50 60

!

I

40

I

!

30

20

0

1bo

260

I

300

1

400

dl0

nag c

Fig. 4-208 Pyrolysis of Petroleum resins from VR Kuwait STA Diagram: TGA Heating Rate p: 10 K/min Atmosphere: Argon 30 + 20 cm3/min Integration Limits: 432 "C and 475 "C 475 "C and 620 "C

s60

I

700

1bO

900


457

Table 4-206: Pyrolysis kinetics of bitumen B80 according to McCarty and Green. Experiments with Simultaneous Thermal Analyzer

E (kJ/Mol)

Heating rate Temperature limits (K/min)

log A (min-I)

("C)

Weight loss at the upper temperature limit

(%I Evaporation 250. . . 389 (onset 1) 250. . ,400 (onset 1) 250. . .419 (onset 1)

5 10 20

Pyrolysis 389 (onset 1). . . 466 (onset 2) 400 (onset 1). . . 483 (onset 2) 419 (onset 1). . . 496 (onset 2)

5 10 20

109.71 109.82 116.51

8.1865 8.3602 8.9421

16.9 15.3 14.7

200.69

14.1014

76.9

202.81

14.2579

76.5

221.26

15.5999

76.1

Table 4-207: Pyrolysis kinetics according to McCarty and Green. Experiments with Simultaneous Thermal Analyzer. Heating Rate p= 10 K/min ~

Substance Temperature limits (OC)

Evaporation E (kJ/Mol)

log A (min-')

Temperature limits ("C)

Pyrolysis E (W/Mol)

log A (min-')

VR Kuwait Dispersion medium Petroleum resins

250.. ,395

118.44

9.1506

395.. ,490

187.88

13.2199

250. ..395 250 . . . 432

113.09 146.51

Asphaltenes

250 ... 430

214.20

8.6946 395 . . . 485 10.6836 432 . . . 475 475.. ,620 15.9212 430 ... 505

186.54 367.53 149.49 258.75

13.1157 26.0533 8.7684 17.9624

onset point of the second reaction and appears at 475 "C.The last offset point occurs at 620 "C.The two regions of reaction can be integrated separately:

1st reaction 2nd reaction

E (kJ/Mol) 367.53 149.49

log A (min-') 26.053 8.768

458


4.10.2 Oxidation reaction The oxidation reaction is a more complicated process than pyrolysis. In DTA or DSC the oxidation of pure chemicals always reveals three or more maxima at different temperatures in the plot of heat flow versus temperature. The TGA curves display more than one point of inflexion and the DTG and DSC (DTA) curves more than one peak maximum (Fig. 4-209 and 4-210). Although oxidation is not a first order reaction, it can be treated mathematically as a first order reaction with regard to the consumption of fuel, provided there is an excess of air (oxygen). Therefore the current computer programs can be used to examine the kinetics of oxidation.

110

------

I----

t

i POINT TABLE

c I (

IPt

I

70

6

536.D7

t

L

POXNT

I 0

TABLE

3a.9621 444.24 -1.0701 5 530.30-362.267 3

425.n.3

4

Lir-0

t

300

-

200

5.5051

U

a

'0°

$0

RESIDUE 1.71 I 3bo

Dog C

Fig. 4-209 Oxidation in Air of Bitumen B80 STA Diagram: TGA, DTA Heating Rate p: 10 K/min Atmosphere: Air 30 + 20 cm3/min

4bO

-

100

-

50

-

0

150


I

d

ib0

260

Fig. 4-210 Oxidation in Air of Bitumen B80 STA Diagram: TGA, DTG Heating Rate p: 10 K/min Atmosphere: Air 30 + 20 cm3/min

3bO

Dog c

4bO

d0

6bO

459

0

460


4.10.2.1 Kinetics according to ASTM E 698-79 4.10.2.1.1 DSC (DTA) experiments The DSC oxidation of bitumen B80 in air at a heating rate p = 5 K/min shown in Fig. 4-211 is taken as an example. The three regions of reaction are defined by distinct peaks. The LTO region is represented by only one peak between 303 "C (5 K/min) and 359 "C (50 K/min). Between 394 "C (5 K/min) and 478 "C (50 K/min) there are several more or less distinct peaks in the region of fuel deposition. The peaks of experiments at different heating rates are not easy to associate for evaluation. Unfortunately this can only be effected by trial and error. The safest but also the most time consuming method consists in the selection of the data which best fit the regression line in the plot of log (heating rate) versus 1 OOO/T. The evaluation is also more difficult because altering the heating rates

-20

!

0

I

100

I

200

4d0 T~peroturo(.C)

300

Fig. 4-211 DSC Oxidation in Air of Bitumen B80

General Data Plot Heating Rate p: 5 K/min Atmosphere: Air 5 cm3/min

I

500

.

800

7


461

changes the position and shape of the peaks and can cause peaks to disappear completely. The last region, representing fuel combustion, is marked by a very clear peak in the temperature range from 490 "C (5 Kimin) to 572 "C (50 Kimin). In Fig. 4-212 the plot of log (heating rate) versus 1 OOOlT for the LTO region shows that all data from the experiments fit the regression line perfectly. In the fuel deposition region only two sets of the peaks from four heating rates can be identified and provide two pairs of kinetic parameters. The last region (fuel combustion) also shows a good fit of the experimental data to the regression line. The values of the kinetic parameters for bitumen €380 and vacuum residue Kirkuk are presented in Table 4-208. The computer program of the STA 780 only processes separately sampled DTA data. Nevertheless the plot of heat flow versus temperature is virtually the same as the diagram produced using DSC (Fig. 4-213). Here again the LTO region exhibits only one peak, the

..- ,

P

Thermal methods in petroleum analysis

Thermal Methods of Petroleum Production (Developments in Petroleum Science)

Thermal and Catalytic Processes in Petroleum Refining

Thermal and Catalytic Processes in Petroleum Refining

Thermal Analysis

Thermal Analysis

Thermal Analysis in the Geosciences

BEST ESTIMATE METHODS IN THERMAL HYDRAULIC SAFETY ANALYSIS

Thermal Analysis, 3rd Edition

Thermal Analysis of Materials

Thermal Analysis of Materials

Thermal Analysis, 3rd Edition

Thermal Analysis of Materials

Handbook of Thermal Analysis

Thermal Analysis of Materials

Thermal Analysis of Pharmaceuticals

Handbook of thermal analysis

Handbook of Petroleum Product Analysis

Methods in Nonlinear Analysis

Ramsey methods in analysis

Asymptotic Methods in Analysis

Asymptotic methods in analysis

Methods in nonlinear analysis

Methods in Algorithmic Analysis

Ramsey Methods in Analysis

Matrix Methods in Analysis

Thermal Power Plant Performance Analysis

Thermal Analysis of Polymeric Materials

Functional Analysis Methods in Numerical Analysis

Petroleum Geology (Developments in Petroleum Science)

Thermal methods in petroleum analysis

Thermal Methods of Petroleum Production (Developments in Petroleum Science)

Thermal and Catalytic Processes in Petroleum Refining

Thermal and Catalytic Processes in Petroleum Refining

Thermal Analysis

Thermal Analysis

Thermal Analysis in the Geosciences

BEST ESTIMATE METHODS IN THERMAL HYDRAULIC SAFETY ANALYSIS

Thermal Analysis, 3rd Edition

Thermal Analysis of Materials

Thermal Analysis of Materials

Thermal Analysis, 3rd Edition

Thermal Analysis of Materials

Handbook of Thermal Analysis

Thermal Analysis of Materials

Thermal Analysis of Pharmaceuticals

Handbook of thermal analysis

Handbook of Petroleum Product Analysis

Methods in Nonlinear Analysis

Ramsey methods in analysis

Asymptotic Methods in Analysis

Asymptotic methods in analysis

Methods in nonlinear analysis

Methods in Algorithmic Analysis

Ramsey Methods in Analysis

Matrix Methods in Analysis

Thermal Power Plant Performance Analysis

Thermal Analysis of Polymeric Materials

Functional Analysis Methods in Numerical Analysis

Petroleum Geology (Developments in Petroleum Science)

Recommend Documents