Structural Connections for Lightweight Metallic Structures

Advanced Structured Materials Volume 8 Series Editors Andreas Öchsner Lucas F. M. da Silva Holm Altenbach For further...

Author: Pedro M.G.P. Moreira | Lucas F. M. da Silva | Paulo M.S.T. de Castro

265 downloads 3318 Views 8MB 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

Advanced Structured Materials Volume 8

Series Editors Andreas Öchsner Lucas F. M. da Silva Holm Altenbach

For further volumes: http://www.springer.com/series/8611

Pedro M. G. P. Moreira Lucas F. M. da Silva Paulo M. S. T. de Castro •

Editors

Structural Connections for Lightweight Metallic Structures

123

Dr. Pedro M. G. P. Moreira INEGI Rua Dr. Roberto Frias 400 4200-465 Porto Portugal Prof. Dr. Lucas F. M. da Silva Departamento de Engenharia Mecânica Faculdade de Engenharia Universidade do Porto Rua Dr. Roberto Frias s/n 4200-465 Porto Portugal

ISSN 1869-8433 ISBN 978-3-642-18186-3 DOI 10.1007/978-3-642-18187-0

Prof. Paulo M. S. T. de Castro Departamento de Engenharia Mecânica Faculdade de Engenharia Universidade do Porto Rua Dr. Roberto Frias s/n 4200-465 Porto Portugal

e-ISSN 1869-8441 e-ISBN 978-3-642-18187-0

Springer Heidelberg New York Dordrecht London Library of Congress Control Number: 2012930377 Springer-Verlag Berlin Heidelberg 2012 This work is subject to copyright. All rights are reserved by the Publisher, whether the whole or part of the material is concerned, specifically the rights of translation, reprinting, reuse of illustrations, recitation, broadcasting, reproduction on microfilms or in any other physical way, and transmission or information storage and retrieval, electronic adaptation, computer software, or by similar or dissimilar methodology now known or hereafter developed. Exempted from this legal reservation are brief excerpts in connection with reviews or scholarly analysis or material supplied specifically for the purpose of being entered and executed on a computer system, for exclusive use by the purchaser of the work. Duplication of this publication or parts thereof is permitted only under the provisions of the Copyright Law of the Publisher’s location, in its current version, and permission for use must always be obtained from Springer. Permissions for use may be obtained through RightsLink at the Copyright Clearance Center. Violations are liable to prosecution under the respective Copyright Law. The use of general descriptive names, registered names, trademarks, service marks, etc. in this publication does not imply, even in the absence of a specific statement, that such names are exempt from the relevant protective laws and regulations and therefore free for general use. While the advice and information in this book are believed to be true and accurate at the date of publication, neither the authors nor the editors nor the publisher can accept any legal responsibility for any errors or omissions that may be made. The publisher makes no warranty, express or implied, with respect to the material contained herein. Printed on acid-free paper Springer is part of Springer Science+Business Media (www.springer.com)

Preface

Increasing concern with fuel consumption leads to widespread interest in lightweight structures for transportation vehicles. Several competing technologies are available for the structural connections of these structures, namely welding, mechanical fastening/riveting, and adhesive technologies. Arranged in a single volume, the intention of this book is to present state-of-theart discussions of those aspects and processes presenting greater novelty whilst simultaneously keeping wide applicability potential and interest. It is therefore not intended to substitute the existing comprehensive treatments dealing with each technology, as (da Silva et al. 2011) or (Grote and Antonsson 2009). Riveting is the traditional aeronautical process for joining Aluminium alloy fuselage components. Although there is growing interest in the use of composites for aerostructures, it is likely that Aluminium alloy fuselage construction will still be used at least for certain sizes of aircraft, and certainly knowledge of the ageing behaviour of the vast existing fleet will continue to be needed in the next decades. For these reasons, the book will address specific problems of riveted joints, including the statistical treatment of multiple site damage, and the behaviour of cracks growing from fastener holes. Great interest is dedicated to welding of lightweight structures, particularly because of economic reasons derived of part count reduction, faster and cheaper fabrication, and possible weight gains. However, drawbacks of welding include variation of properties in the weldment area, intrinsic metallurgical difficulties possibly leading to unacceptable defects, as well as, from a mechanical design point of view, possibly detrimental fatigue behaviour features associated with the continuous path for crack propagation. Currently, the technological processes receiving more widespread interest in metallic lightweight structures are laser beam welding (LBW) and friction stir welding (FSW) and the book discusses both thoroughly, including novel structural designs and applications. Also, a chapter concerning manufacturing cost assessment of friction stir welded structures in the aeronautical context, compared with the traditional riveting process, is included.

v

vi

Preface

Aeronautical and automotive manufacturers dedicate increasing attention to adhesive joining technologies, because of potential cost and weight savings, associated with excellent capability for geometrical precision because, unlike in welding, no great temperature variations are involved in the fabrication of adhesive bonded components. Among current drawbacks preventing more widespread application in industry that will be discussed in detail, are durability concerns as well as specific problems of stress concentration at the end of overlap joints. The topics chosen for coverage in this book have the common feature of being currently applied in lightweight metallic structures, and one of the characteristics of this work is bringing together relevant state-of-the-art information usually presented in separate publications specializing in a single technology. New aircraft designs and new propulsion systems are under continuous development, with new materials and new manufacturing processes playing an important role in the improvement of air transportation efficiency. This continuous development leads to significant fuel savings, and lower operational costs. Design concepts for aircraft must also pass stringent performance criteria that impose minimum values of weight. Aeronautical designs rely on damage tolerance principles, e.g. (Schmidt 2003). There has always been a drive to reduce aircraft weight, and, simultaneously, an appreciation that lightweight designs can be too costly. Indeed, as noted by (Emero 1967) minimum-weight designs are frequently too costly to manufacture, whereas less expensive and easy to fabricate and assemble designs are often much heavier. The most efficient design on the basis of both cost and weight often lies between these two extremes. The present book gives a contribution to these everlasting discussions, of which some past steps are documented in publications of archival or historical significance. In the case of automobile engineering, an early paper (O’Gorman 1908) wisely states that ‘when …. we come to consider particular parts, pistons, connecting rods, …. or body weight, we shall find that every single item is so full of suggestiveness that no man can write more than a brief essay on motor car weights if he has anything else to do ’. Later, discussing weight savings in automobiles, (Ferrier Brown 1923) notes that ‘the keynote of success in weight reduction is attention to detail. Every ounce that can be spared should be cut out, provided it is a straightforward machining operation. Weight saving may not mean economy in the sense that first cost will be substantially reduced, but it will certainly prove a true and continuous economy in running costs and extended life of the vehicle’. After this early manifestation of concern with the life-cycle cost of a product, the mentioned paper goes on to note that ‘considerable saving in weight can be effected by using aluminium for the panels instead of steel, though they are not so robust as steel panels, have not the same high physical properties, and do not offer the same facilities for welding’. As shown by (Pomeroy 1939) discussing the use of aluminium in automobile design, cost is aubiquitous object of concern. The abundance of literature on the subject precludes further reference to the past, but, nevertheless, (Kewley 1987) and (Allen 1994) may be cited for comprehensively

Preface

vii

discussing cost implications of Al and adhesive technology in automobile manufacturing, and of the use of Ti as means to save weight in automotive engines, respectively. In the case of railways, (Thring 1954) and (Brockway 1959) discuss the implications of the use of Al instead of steel in railway passenger cars, whereas (Mills 1939) ‘describes a new (welded) design for a steam locomotive frame of plate type, and draws comparisons with the corresponding design for riveted construction, showing important economies in cost and in weight’. Further on in that paper, the author states that ‘much attention has been given of late years to the question of residual stresses in welded structures and methods of stress relief’, cogently illustrating that certain subjects keep resurfacing, albeit in different circumstances and environments, as the present book will once again show. This book is composed by ten chapters covering the topics described above. A brief description of each chapter is now presented: Chapter 1—Assessment of Multiple Site Damage in Riveted Aircraft Joints Multiple Site Damage (MSD) and Widespread Fatigue Damage (WFD) are typical effects which may occur in structures of aging aircraft. They consist in the development of scenarios of sets of small cracks, which interact at a certain stage and may suddenly burst into one large crack. The chapter summarizes the approach used in a number of European companies for the assessment of the susceptibility to WFD as well as by some European researchers today. In addition, an approach to interpret data from these probabilistic analyses with respect to the criticality of designs is proposed. Chapter 2—Laser Welding of Structural Aluminium The chapter presents an overview, characteristics and progresses of the fusion welding processes used in aluminium welding. Laser sources for welding have been available for a few decades but new concepts are coming to the market. The chapter addresses the most commonly used lasers for materials processing and their interaction with aluminium alloys in welding applications. More recent laser types are also included, namely fibre lasers and disc lasers as, though only more recently available in the market, their potential is foreseen as being interesting for welding of aluminium. Chapter 3—Laser Beam Welding and Automotive Engineering This chapter gives an overview of the use of laser welding on automotive industry, which is the main industrial sector for its application. Laser welding permits the use of significant improvements in lightweight structures manufacturing for the automotive industry, as the use of tailor welded blanks. The main features of the laser welding process and breakthroughs achieved by the use of laser welding in the automotive industry are summarized.

viii

Preface

Chapter 4—State-of-the-Art in FSW Technologies The Friction Stir Welding (FSW) is currently considered an important development in welding technology, saving costs and weight for a steadily expanding range of applications of lightweight metallic structures. Evidences of the disruptive character of the FSW process are the prompt adoption by world-wide industry of the significant advantages of FSW and the numerous technic-scientific papers and patents published. In this chapter, some of the basic fundamentals underpinning the invention of FSW technology are presented with emphasis on the concept of the third-body region. The Non-Destructive Testing assessment of the most relevant imperfections in FSW is discussed for butt and lap joints. Chapter 5—FSW of Lap and T-Joints Even if in the last years several researches have studied the Friction Stir Welding (FSW) process, it should be observed that most of these studies are concerned with the butt joint and just a few of them extend to more complex geometries. The acquired knowledge on FSW process of butt joints is not immediately extendable to lap and T-joints. The chapter is focused in the development of FSW process for its application to lap and T-joints. As an example, in lap and T-joints a major vertical component of the material flow is required to obtain sound joints. Furthermore, in the FSW of T-parts a proper clamping fixture is needed in order to fix the stringer during the process. The tool geometries together with the tool feed rate and rotating speed must be determined in order to get an effective material flow and bonding conditions during the FSW process. Chapter 6—Lightweight Stiffened Panels Fabricated Using Emerging Fabrication Technologies: Fatigue Behaviour The need for lower cost and the emergence of new welding technologies has brought interest in large integral metallic structures for aircraft applications; however, new problems must be addressed, e.g. in integral structures, a crack approaching a stiffener propagates simultaneously in the skin and into the stiffener and breaks it. The use of manufacturing techniques such as high speed machining (HSM), laser beam welding (LBW) and friction stir welding (FSW) requires further experimental and numerical work concerning the fatigue behaviour of panels manufactured using those processes. The chapter is focused on an experimental test programme including fatigue crack growth rate characterization in panels fabricated using HSM, LBW and FSW.

Preface

ix

Chapter 7—Damage Tolerance of Aircraft Panels Taking into Account Residual Stress The chapter seeks to summarize the ideas of damage tolerance of aircraft panels, keeping in mind the effect of residual stresses. First, concepts and techniques are briefly reviewed, and afterwards their application to experimental work is discussed. As a result of the traditional usage of riveted joints in the aluminium alloy fuselage of civil aircraft, one advantage of this type of joints is the existing experience concerning their design and maintenance. Alternatives to riveting are being considered aiming at economies in fabrication time, cost and weight. However, open issues concerning the use of integral structures in aeronautics include the damage tolerance problem, since the integral nature of the structure provides a continuous path for crack growth. A topic of the chapter is the fatigue behaviour of integral stiffened panels focusing on the influence of residual stress fields. Chapter 8—Multi Material Adhesive Joining in the Automotive Sector Technological progress over the last decade has provided the automotive industry with more flexibility in terms of design and manufacturing. Automotive manufacturers face conflicting requirements regarding environmental legislation and an increasing demand for safety and on-board equipment which tends to increase the overall vehicle weight and fuel consumption. This challenge has led to the increased use of lightweight materials to replace conventional steel parts and structures. The chapter is intended to provide an overview of the materials, technologies and associated processes used to manufacture Aston Martin’s vehicle bodies. Chapter 9—Welded Aeronautical Structures: Cost and Weight Considerations Product development is limited by engineering design capabilities. Engineering design is one of the most important phases during the development of a new product, particularly in the case of complex and safety-critical systems, in order to consider all safety concerns, e.g. in aircraft and nuclear power plants. In these cases, the introduction of new design concepts and solutions is tightly tapered by existing materials and manufacturing processes. In this chapter, a breakthrough joining process—friction stir welding (FSW) is discussed from the point of view of manufacturing costs. Chapter 10—Materials Selection for Airframes: Assessment Based on the Specific Fatigue Behavior Structural weight reduction is a major driver to improve the transportation efficiency. However, minimum-weight designs are frequently too costly to manufacture, whereas less expensive and easy to fabricate and assemble designs are often much heavier. Composite materials have high specific strength, are less

x

Preface

prone to fatigue crack initiation and provide enhanced flexibility for structural optimization compared to the aluminum alloys. On the other hand, aluminum alloys display higher toughness and better damage tolerance in the presence of defects. In order to improve the material selection and the comparison of airframe materials, Chap. 10 presents a weight assessment based on the specific weight for different damage scenarios.

References Allen, P. G.: Technical and commercial considerations for the application of titanium in automotive engines. Proceedings of the Institution of Mechanical Engineers. Part D: J. Automob. Eng. 208(D1), 25-32 (1994) Brockway, K. P.: Aluminium technology and railway rolling stock’. J. Inst. Locomot. Eng. 49, 665–769 (1959) da Silva, L. F. M., Öchsner, A., Adams, R. D. (Eds): Handbook of Adhesion Technology. Springer, Heidelberg (2011) Emero, D. H., Alvey, V. W.: Structures cost effectiveness. J. Aircr. 4(3), 218–223 (1967) Ferrier Brown, W.: Automobile body engineering. Proc. Inst. Automob. Eng. 18, 172–220 (1923) Grote, K.-H., Antonsson, E. K.: Springer Handbook of Mechanical Engineering. Springer, Heidelberg (2009) Kewley, D.: Aluminium alloy body structures for future vehicles. Proceedings of the Institution of Mechanical Engineers. Part D: Transp. Eng. 201(D2), 129–134 (1987) Mills, F.: The fabrication of the locomotive frame by arc welding. J. Inst. Locomot. Eng. 30, 13–66 (1939) O’Gorman, M.: The weight of motor cars and motor car parts. Proc. Inst. Automob. Eng. 3, 106–154 (1908) Pomeroy, L. H.: Automobile design in terms of aluminium. Proc. Inst. Automob. Eng. 33, 476–502 (1939) Schmidt, H-J., Schmidt-Brandecker, B.: Damage tolerant design and analysis of current and future aircraft structure. AIAA/ICAS International Air and Space Symposium and Exposition: ‘The Next 100 Y’, Dayton; paper AIAA 2003–2784, 14–17 July 2003 Thring, F.: The design of light alloy coaches for East African railways. J. Inst. Locomot Eng. 44, 495–540 (1954)

Contents

Assessment of Multiple Site Damage in Riveted Aircraft Joints. . . . . . Peter Horst

1

Laser Welding of Structural Aluminium . . . . . . . . . . . . . . . . . . . . . . L. Quintino, R. Miranda, U. Dilthey, D. Iordachescu, M. Banasik and S. Stano

33

Laser Beam Welding and Automotive Engineering . . . . . . . . . . . . . . . Eva Vaamonde Couso and Joaquín Vázquez Gómez

59

Friction Stir Welding Technology . . . . . . . . . . . . . . . . . . . . . . . . . . . Pedro Vilaça and Wayne Thomas

85

FSW of Lap and T-Joints . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . L. Fratini

125

Lightweight Stiffened Panels Fabricated Using Emerging Fabrication Technologies: Fatigue Behaviour . . . . . . . . . . . . . . . . . . . P. M. G. P. Moreira, V. Richter-Trummer and P. M. S. T. de Castro

151

Damage Tolerance of Aircraft Panels Taking into Account Residual Stress . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . V. Richter-Trummer, P. M. G. P. Moreira and P. M. S. T. de Castro

173

Multi-Material Adhesive Joining in the Automotive Sector . . . . . . . . . Sylvain Pujol

195

xi

xii

Contents

Welded Aeronautical Structures: Cost and Weight Considerations . . . S. M. O. Tavares Materials Selection for Airframes: Assessment Based on the Specific Fatigue Behavior . . . . . . . . . . . . . . . . . . . . . . . . . . . . S. M. O. Tavares, P. P. Camanho and P. M. S. T. de. Castro

219

239

Assessment of Multiple Site Damage in Riveted Aircraft Joints Peter Horst

Abstract Multiple Site Damage (MSD) and Widespread Fatigue Damage (WFD) became a concern after the Aloha incident in 1988. Both, Multiple Site Damage and Widespread Fatigue Damage are typical effects which may occur in structures of aging aircraft. They consist in the development of scenarios of sets of small cracks, which interact at a certain stage and may suddenly burst into one large crack. The chapter consists of five topics, namely the description of the relevance of such an assessment method from an industrial point of view, the fast calculation method, which is used to account for the interaction of cracks in riveted joints, a Monte Carlo Simulation (MCS) method based on the model mentioned above, typical results of such a MCS procedure and its validation and in the end some considerations of methods which may be used to improve the calculation effort needed for a MCS.

1 Introduction The subject of aging aircraft is established at least since it gained a wider public interest after the Aloha incident in 1988. The special problem treated in this chapter is the classical MSD (multiple site damage) and to a certain extent the WFD (widespread fatigue damage) problem. This problem may be divided into different sub-problems, which are not all treated in the same depth in this chapter. Some of these problems are:

P. Horst (&) Institute of Aircraft Design and Lightweight Structures, TU Braunschweig, Hermann-Blenk-Strasse 35, 38108 Braunschweig, Germany e-mail: [email protected]

Adv Struct Mater (2012) 8: 1–32 DOI: 10.1007/8611_2010_50 Springer-Verlag Berlin Heidelberg 2011 Published Online: 28 January 2011

1

2

P. Horst

• how multiple site damage occurs • the development of cracks of the same size • the residual strength problem in a WFD scenario The first two questions are mainly treated in this chapter, while the third one is not deeply looked at. For this third question many publications are known for some years, see e.g. Newman and Dawicke [1], Atluri [2], Nilsson and Hutchinson [3], Horst [4]. The approach used in this chapter follows a line, which has been kept since the mid to the end 1990s by a group of European researchers as well as some industrial partners in two European projects, namely the GARTEUR Structures & Materials Action Group SM18 and the EU-funded project SMAAC (Structural Maintenance of Ageing Aircraft). The approach consists primarily in a probabilistic view of the process of the evolution of MSD. This approach tries to circumvent expensive full scale tests or very restrictive damage scenarios, since the MCS will yield less drastic scenarios. The special problem treated in this chapter is visualized in Fig. 1, it is the classical MSD problem of a riveted longitudinal lap joint in a pressurized fuselage. The stress level is taken as widely constant for at least a set of rivets in a row, and it is taken to act only in circumferential direction. The general procedure is not limited to this set-up, it is applicable in a much wider sense to other items, which are susceptible to WFD, as e.g. circumferential joints, cordwise splices at wings etc. The chapter tries to summarize the approach used in a number of European companies for the assessment of the susceptibility to WFD as well as by some European researchers today (e.g. Balzano et al. [5]). In addition, it is tried to find an approach to interpret data from these probabilistic analyses with respect to the criticality of designs. This last approach is tried via wavelets as a kind of feature detection. The fracture mechanical method to assess multiple crack propagation in riveted joints is the compounding method. The basic principle is well documented, but it has been extended for the purpose needed in this case, especially with regard to loaded rivet holes. The MCS method in itself is very simple, and may be used for the given purpose in a well-known direct way, provided the calculation method

Fig. 1 Typical lap-joint problem

Assessment of Multiple Site Damage in Riveted Aircraft Joints

3

behind it is fast enough. Typical results show that many parameters have an essential influence, namely the fatigue behavior at the rivet holes, the probability of detection and the material data, as well as geometrical parameters. The validation of the model is done by comparison with a large test, which already has been used for validation purposes years ago. It shows that also the relatively simple calculation model behind the MCS is well capable to assess MSD in such cases. An attempt to improve the computational effort needed for the MCS is not a simple task. The well-known FORM or SORM (first or second order reliability method) approaches do not really help in this special case, since the highdimensional space of probabilistic parameters exhibits critical states only in small, but unlinked areas. For the same reason, directional or importance sampling do not work well. The method proposed here is via pattern analysis, which is based on wavelet transforms.

2 Theoretical Background and Model 2.1 Overall Approach The overall approach is based on the Monte Carlo Simulation method, which is widely used in different technical and other areas (see Melchers [6]). This procedure is in a way quite simple, although the actual theoretical and practical background of the deterministic part of the method may be complex and also subject to different discussions. The general procedure may shortly be described by Eq. 1. A set of uncorrelated random variables z is used by an algorithm Algðx; y; z; . . .Þ to find some statistical distribution of variables y: Then, an other algorithm uses a limit state function gðyÞ indicating improper states by values smaller or equal to zero is used to define the ratio of failure states to the number of all states tested, which is called the probability of failure pf : z 2 Rm

Algðx;y;z;...Þ

!

y 2 Rn

...

! gðyÞ

...

! pf

ð1Þ

In this case, the scatter in SN-data from small coupon tests is used to define randomly distributed damage scenarios as initial state for a deterministic calculation, i.e. z in the sense of Eq. 1, assuming that the initiation of cracks at adjacent rivets is an uncorrelated event. The algorithm Alg is a deterministic method, which in this case calculates damage accumulation and crack growth, as well as residual strength and compares it with a given value, e.g. for the inspection interval (gðyÞ). The overall model is shown in the following flow-diagram

4

P. Horst

The initial damage scenario is defined for a simplified set-up of fatigue critical locations l; this is e.g. the perimeter of a rivet hole etc. It is of course well-known that the actual fatigue initiation location is not always—in an ideal sense—at the perimeter of the hole, but may be located away from the hole at the faying surfaces, see e.g. Schijve [7]. But this point is not essential in this case. What really is needed is a well-defined and statistically sound coupon test of the joint in order to have a well-based fatigue model and an appropriate estimation of the scatter in the data. It is quite obvious that many parameters may influence both the fatigue behavior as well as the scatter in this parameter, see e.g. Müller [8]. It is quite clear that especially these parameters may be the key to the occurrence of MSD, if certain features of the fatigue life, i.e. the initiation of cracks, are deteriorated. The information on this point fully relies on basic experimental data. It is not the intention of the model to provide any synthetic fatigue data. It will turn out that the scatter of the fatigue data is really the most relevant parameter for the occurrence of MSD. Therefore, this type of data has to be considered in the preparation of the modeling. Any additional deteriorating effect, as e.g. corrosion, debonding, badly manufactured rivet rows etc. have to be included into the fatigue data before-hand. They are not explicitly included in the deterministic model. Surely the question arises, why an approach has been used, which is based on initiation of cracks and a subsequent deterministic crack propagation. The reason mainly lies in the fact that the model for the calculation of the crack growth is surely not prepared to cover extremely small cracks, i.e. cracks with a length lower than a few tenth of a millimeter. Since this would be necessary for an approach, which is based on some equivalent flaw type of model, this has not been used. In a way, it seems to be of no relevance, whether an equivalent flaw size approach is used, which is based on an insufficient crack growth model or an initiation approach is used, which starts the deterministic modeling at a few tenth of a millimeter crack length. For the description of the fatigue data and their scatter, a log-normal distribution has been chosen. This is in a way a compromise. It seems as if the parameters for a


5

log-normal distribution are much more reliably defined by a limited number of fatigue tests than the parameters of a 3-parameter Weibull distribution. The 3-parameter Weibull distribution would be preferred, since it offers much better results for high numbers of fatigue critical locations, if the size-effect is taken into account, i.e. that a high number of fatigue critical locations are present. The point is that the log-normal distribution has no lower bound and therefore tends to zero, or even negative fatigue life for high numbers of fatigue critical locations, while the Weibull distribution does not show this, since it has a lower limit. This point should be remembered, if really high numbers of fatigue critical locations have to be covered. In general, the damage accumulation law that has been used is the PalmgrenMiner law, i.e. a simple linear accumulation criterion. It is well-known that this law is not flawless, but it provides good data for the more or less constant stress cases, which have been treated in this chapter. Although the basic data on fatigue are taken from (simple coupon) experiments, it is not always simple to find accumulation data for cases, where cracks are already present in the vicinity of the fatigue critical location. Here, a simple model for the influence of higher stresses according to stress concentration ak -data has been used. As written in the flowdiagram above, each of the n different scenarios starts with a random distribution of damage at each of the fatigue critical locations. This random process is a deterministic random process, which assigns a random value in the interval ½0; 1 for the value FðzÞ; which is the accumulated probability of failure for each location. The random process is taken from Press et al. [9]. It seems to be reasonable to use a deterministic random process, since this allows to repeat runs with the same scenario, if this is needed, e.g. during development of the code or for justification purposes. There are algorithms at hand, which would already provide normally distributed values for z; but these are not used here. One of the well-known approximations for the mapping (see e.g. Hastings [10]) z ¼ f ðFðzÞÞ

ð2Þ

is used in this code, i.e. the random process picks out a real number FðzÞ in the interval ½0; 1 and by means of the Hastings method the appropriate value z is calculated. It is the general procedure of a Monte Carlo Simulation to use a deterministic model based on the random starting scenario. This is used here. The deterministic model is described in Sect. 2.2. It may be asked, why a Monte Carlo Simulation has been chosen instead of one of the well-established methods which are often used in structural reliability, like e.g. FORM or SORM etc., see e.g. Melchers [6], since they usually provide high accuracy with less computational effort. The reason is that it seems not to be possible to define a conceivable parameter space, which takes into account the relevant data of MSD and on the other hand to put this into one of the methods mentioned above. The Monte Carlo Simulation is much more flexible in this case, although there are some problems arising from the computational effort that is needed for high reliability indices (see Sect. 3.1.2).

6

P. Horst

2.2 The Deterministic Model The type of deterministic model behind this approach is not fixed in any way, apart from the fact that reasonable computational effort is needed in order to achieve a significant number of different simulations, as just stated. In this chapter, the fatigue and crack scenario depicted in Fig. 1 is idealized in the well-known way of a riveted joint. The three types of loading of the rivet hole, i.e. remote stress, point load and possibly bending may be taken into account. In the present case, a 2D-approach has been used. This is to say that bending has been neglected. This may be a bit astonishing, but results are relatively good, as shown in Horst et al. [11] for different cases, where no thick doubler is used. Surely other, more sophisticated methods are at hand to do more reliable and more appropriate assessments, as e.g. the finite element method in a 3D version, see e.g. Fawaz and Andersson [12] for p-elements or others for boundary element approaches. But it is questionable, whether these approaches are able to provide enough data for complex set-ups in a Monte Carlo Simulation. After all, data from such methods are very useful for the check of the 2D results, or may be used for the calculation of individual crack propagation of cracks without interaction (see Sect. 4). The basic method is the compounding method, which is well documented in many publications, e.g. Rooke [13]. The basic method is very well described in the ESDU data sheet [14]. The method uses simple single or double crack solutions for the assessment of complex problems, i.e. multiple interactions of cracks and other boundaries. The basic equation is Kr ¼ K0 þ

m X ðKn0 K0 Þ þ Ke

ð3Þ

n¼1

where Kr is the stress intensity factor of the r-th crack tip, K0 is the stress intensity factor of a crack without interaction with a boundary of equal size, Kn0 is the stress intensity factor which occurs from the interaction of crack tip r and boundary n: Ke is an additional factor, which occurs from multiple interaction. This factor is usually neglected. It may have a considerable impact, if e.g. a specimen of limited width faces large cracks. The prime idea of the method is that the stress in the vicinity of the crack tip r is raised by the interaction with the other boundaries and that this is simulated by an additional approach for many boundaries interacting. This fact has been used to extend the method to partly point loaded rivet holes, since the effect of the stress increase due to other boundaries may also be covered for this case. The basic solutions needed for this task are: • • • •

unequal cracks emanating from a hole interacting unequal cracks crack with non-central point load ...


7

The many simple solutions needed for this method have been taken from a set of publications, namely, Rooke and Cartwright [15], Tada et al. [16] and Rooke and Tweed [17]. For speed-up reasons these data have been used as interpolated data-sets, equations etc. What else is needed in this case is a link-up criterion for two adjacent cracks. For crack propagation link-up the criterion proposed by Swift [18] proved to be useful. This criterion predicts link-up, if the plastic zones in front of the approaching crack tips touch each other. The Irwin criterion for the assessment of the plastic zone

rp ¼

KI2 pr2y

ð4Þ

has been used for this purpose, with KI as the mode I stress intensity factor and ry as yield stress. It is obvious that the criterion largely depends on the definition of the yield stress. For the crack growth rate da=dN a simple Forman law has been used, i.e. da cf DK nf ¼ dN ð1 RÞKf DK

ð5Þ

where cf ; nf and Kf are material parameters and R is the stress ratio. The start crack length of the deterministic calculation has been set to 0.1 mm. This is of course a relatively small size, and it is questionable, whether the model is able to calculate this region reliably, but it surely is not far away from the size, which is well defined. The relation of the point load from load transfer to the remote stress has been calculated according to the simple models which are given in many publications for the load transfer, as e.g. Niu [19]. It is not the intention to repeat the verification of the method by comparison of a complex test result with the method presented here. In fact, such a test is quite expensive. Such an expensive test has been performed by Airbus for a longitudinal lap joint within a full scale test, which showed a deteriorated fatigue behavior. The size of the area included a set of frame-bays. Data and the comparison with the compounding method in a MCS have been presented some time ago by Horst and Schmidt [20]. Other more simple cases have benn compared in [11] and also show good results for the method used here. The results have been quite promising. Therefore, it seems to be reasonable to follow this way. What is essential for a good prediction is to have reliable data on the initiation of small cracks. It is not sufficient to use the standard data for the failure of the complete coupon specimen. In order to achieve this, either special inspection methods for quite small cracks are needed during testing, or the failure life data must be used to back-calculate the crack propagation phase and to assess the initiation life by this indirect method.

8

P. Horst

3 MSD Scenarios Some insight into the evolution of MSD scenarios may be shown by means of a few examples. The basic design of the lap joint is given as follows: two aluminum 2024 T3 sheets, thickness 1.6 mm are riveted by NAS1097 (countersunk) rivets in a three rivet row lap joint. The diameter of the rivets is 4 mm, the rivet pitch is 20 mm. It is assumed that the critical row of the lap joint takes a load transfer of 37%. The remote circumferential stress is taken to 84 MPa, which seems to be a typical stress level for such an item. The basic material data used in these examples are always the same Forman factors: • cf ¼ 2:01 108 pffiffiffiffiffiffiffiffi • Kf ¼ 2256 MPa mm • nf ¼ 2:7 In all cases, the critical crack length has been put to 500 mm, which is not completely out of the scope of practical examples (it is more on the smaller side). This means that the critical crack length is not different for all basic examples, since the crack tips are already outside the area of the holes, before the long lead crack becomes critical. The actual critical crack length is not very essential in the scope of this chapter, since the few additional load cycles which would be possible for a longer critical crack length are not changing the quality of the results considerably.

3.1 Basic Influences The first section deals with some basic influences like the number of fatigue critical rivet locations, the number of required MCS scenarios to be simulated and the basic stochastic parameters like mean life and standard deviation of the fatigue life.

3.1.1 Number of Fatigue Critical Locations Figure 2 shows the influence of the number of fatigue critical locations on the overall result of a Monte Carlo Simulation. The figure uses two output parameters of the Monte Carlo Simulation, which are essential for the determination of the MSD behavior, namely the number of cycles up to the point in time, when the first crack reaches a detectable crack length of 5 mm. This is called threshold in this chapter. The second parameter is the inspection interval, i.e. the number of cycles between first detectability and the critical state. Both parameters are not affected by anything like a scatter or safety factor.

Assessment of Multiple Site Damage in Riveted Aircraft Joints Fig. 2 Influence of the number of fatigue critical locations

9

50000

interval / cycles

8 rivets 16 rivets

25000

0 100000

125000

150000

175000

threshold up to detectable crack length / cycles

A number of 250 scenarios per distribution has been used in this case. The mean value for the fatigue life was 115,000 cycles in both cases. The standard deviation is 0.05 in a log-normal scale. The size effect is clearly visible in this case in two ways: first, the mean value of the distributions visible in Fig. 2 is decreasing slightly due to the number of possible fatigue critical locations, where the first cracks could start (this is 16 fatigue critical locations in the 8 rivets case and 32 fatigue critical locations in the 16 rivets case). Second, a drop in the interval values. This set-up with 8–16 rivets may be seen as a typical set, which typically occurs in a single frame-bay of a longitudinal lap joint. Normally such a frame-bay is about 500 mm wide, which means more than 16 rivets in a bay, but due to the influence of the frames, the stress in the outer rivets is considerably lower than in the center. Therefore, it is not unrealistic to restrict the focus on this set-up. From an in-service point of view, the outer rivets may be more fatigue critical, although they do not face such high stresses, but they may be more prone to manufacturing problems. This aspect is neglected in this model at this point. The drop in the threshold values due to the size effect would be much larger if only a small number of fatigue critical locations is taken into account, but this would not be very interesting with respect to WFD and MSD. The drop in the inspection interval is mainly explained by the fact that criticality is much earlier reached, if in the case of 16 rivets, a wider area is damaged to a certain extent. This drop may be slighter if even larger areas are taken into account.

3.1.2 Influence of the Number of MSC Scenarios The number of necessary Monte Carlo Simulation scenarios (MSC scenarios) may be assessed by means of an equation, which has been proposed by Broding et al. [21] Nscenarios [

lnð1 CÞ pf

ð6Þ

Fig. 3 Required number of simulated MSC scenarios

P. Horst

number of MCS scenarios / -

10

C = 0.95 C = 0.99 10

5

104

10

-5

10

-4

10

-3

pf / -

with C as the confidence level and pf as the required probability of failure. Figure 3 visualizes this equation for two different confidence levels, namely 0:99 and 0:95: Supposing that a commercial transport aircraft has a design goal of 40,000 flight hours and a required probability of failure per flight hour of 109 ; the total probability of failure is pf \4 105 : The curve for the lower confidence level (0:95) in Fig. 3 indicates that the required number of scenarios is larger than 74; 893 for this case. This is surely a very high number to be simulated even with this very fast and simple simulation method. Figure 4 shows the results of three different numbers of scenarios in a threshold versus interval diagram. The results are not easy to distinguish. It has to be pointed out that the random numbers used in this case are deterministic, i.e. that the first n runs of a simulation are the same for each of the cases. The number of simulation runs in the diagram are: 250, 1000 and 2500. This is of course quite low, in the case of 2,500 runs, the probability of failure achieved is approximately pf ¼ 103 : The computational time used for these 2500 runs has been about 51 min on a 2 GHz PC for the case of 16 fatigue critical rivets in a row. Figure 4 also includes a linear fit of the data. This linear fit indicates two points. First, the difference between the 1000 and the 2500 scenarios is so small that the two fits are almost not to be distinguished. Second, there is a clear dependency between both output parameters: threshold and interval. This second fact will be better understood in the next sections. What can be seen in Fig. 4 is mainly that with increasing numbers of scenarios simulated, extreme values are slightly increasing. Therefore, it surely is necessary for a reliable assessment to simulate a high number as required, but for the evaluation of the method itself in this chapter a lower number must be sufficient.

3.1.3 Extreme Values The values found in Fig. 4 may partly be interpreted by means of some crack growth data. For this purpose, different results indicated in Fig. 5 may be used.

Assessment of Multiple Site Damage in Riveted Aircraft Joints Fig. 4 Different numbers of simulations in the case of 16 fatigue critical rivets

11

30000 x

x

interval / cycles

x

20000

x x x x xx x x x x x xx xx x x x xxx xx xxx x x xx x x x x xxxx x x x xx x x x x x xx x xx x x xx xx x xx xxx x x xxx xx xxxxxx xx xx x xx x xx xx xxxxxx x xx xxxxx x x x x x xxx xxxxxxx xxx x x xx xx x xxx x xxxx xx xx xxx x x xxxxx x x xx xxxxx x xxxxxxxx xxxxx x xxxxxx x xx xxx x x xxx xx x x xx xx xx xxx xxxxxxxxx x xxx x x x x x x x x x x x x x x x x x x x x xxxxx xxxxxxxx xxx x x x x xxxxx x xxx xx x xx xx x x x x xx xx x xxxxxxxxxxx xxx xx xxxxx xxxx x xxxxx xxxxxxxxxxx x xx xx xxxxxx xxx xxxxxxx xxx x x x x x x x x x x x x xxxxxx xxxxxxxxx xxxxxxx xxx x x x xxx x x xxx x xxxx x x x xxx xxxxx x xx x x xxx xxxx xx x xx xxxxx xx xxxxxxxxxxxxxxxxxxxxxx x x x x x x x x x x x x x x x x x x x x x x x x x xxxxxx xxxx xxxxxxxxxxxxxxx xx x xx xxx xxxxxxx x xx xx xx x x xxxxx x xx x x xxxxxx xxxxxxxxxxxxxxxx x x xx x x xx xxxx x xxxxxxx xxxxxxxxxx x x x x xxxxx xx xxxxxxx x x x x x x x x x x x x x x xxx x xxxxxxx xx xx xxxxx x x xx xxx xx x xx x x x xx xxx xxxxxxxxx xxxxxxxxxxxxxxxxxxxxx x xxx xx xxx xxxxx xxx xxxxx x xxx x x x x x x x x x x x x x x x x x xxxx xxx x x xxxx xxxxxxxxxxxxxxxxx x x xx xx xx xxxxxxxx xxxxxxxxx xx xx x x xx x x xx x xxx x xxxxxxxxxx xxx x xx x x x xx x xxx x x x x x x x x xx

x

10000 x

250 1000 2500

0

140000

160000


Fig. 5 Extreme values of the MCS

50000 16 rivets

interval / cycles

linear fit

min. threshold

max. interval

25000

max. life max. threshold min. life min. interval

0 100000

125000

150000

175000


These data are both the minimum and maximum interval as well as the minimum and maximum threshold (again in the sense of fatigue life plus crack growth up to 5 mm detectable crack length). In addition, minimum and maximum life, i.e. the total of all cycles until criticality is given. Figures 6 and 7 present the crack growth behavior for the case of minimum and maximum interval. All data for the example are exactly like in Sect. 3.1.1. The number of cycles is given on the ordinate, while the x-direction indicates the x-position, i.e. the horizontal position of the perimeter of the rivet hole or the crack tip at this perimeter. In order to keep the information of the two figures in a reasonable scale, the ordinate starts at 100; 000 cycles, i.e. that the number of cycles for fatigue and crack growth up to the detectable crack length are not completely covered by the diagrams. The two figures are showing how the cracks are growing at different holes —at first more or less independently — before at least some are joining to build larger cracks. It is very obvious that this process is quite different in the two cases. Obviously, the minimum interval case produces a set of adjacent cracks of nearly equal size, while in the maximum interval case just a few cracks are growing adjacent to each other. This results in a fairly lower crack growth rate in the final state.

12

P. Horst

Fig. 6 Crack growth for the minimum interval case

Fig. 7 Crack growth for the minimum interval case

The two figures on the minimum and maximum threshold, Figs. 8 and 9 are showing different scenarios, but the basic influences are the same again. The maximum threshold figure is not meaning too much with respect to the MSD/WFD problem, apart from the fact that it indicates again that scenarios similar to the minimum interval scenario are leading to similar results. Figure 10 may look a bit strange at first sight, but the final crack scenario results in a critical state, which would join up completely in the next step. Since all diagrams do not really indicate all three phases of the total life, this is given in Table 1 for the six cases given in the crack growth diagrams. From all this, the linear dependency of interval and threshold is relatively simple to explain: cases where—due to the size effect—a single crack initiates


13

Fig. 8 Crack growth for the minimum threshold case

Fig. 9 Crack growth for the maximum threshold case

early, will not have many cracks initiated in the same period. This early, single crack initiation leads to longer crack growth periods, i.e. intervals (Fig. 11). On the other hand, if the first crack initiates lately, the chance that other cracks initiate more or less shortly afterwards is large. This means that a more MSD-like scenario will occur, which results in a much shorter interval. This aspect will be treated again in more detail in Sect. 4 on feature detection.

14

P. Horst

Fig. 10 Crack growth for the minimum life case

Table 1 Results for the six extreme values

Fig. 11 Crack growth for the maximum life case

Case

Crack growth \5 mm cycles

Fatigue cycles

Threshold cycles

Min. interval Max. interval Min. threshold Max. threshold Min. life Max. life

86,315 77,729 78,495 97,592 85,059 93,100

65,800 61,000 56,300 58,100 54,400 59,400

13,100 25,500 25,200 16,100 15,100 20,000


15

3.1.4 Influence of the Scatter Value of the Fatigue Data It is obvious from the statements made in Sect. 3.1.3 that both, scatter and mean value of the fatigue data will have a considerable influence on the MSD scenario creation. This point of view is discussed in the present section. Figure 12 shows the influence of a change in the standard deviation from 0:05; as in the previous examples, to 0:10: The changes are significant. It is quite obvious that the standard deviation has an impact on both, the threshold as well as the interval. Interpretation of these results is again not too hard to find. Obviously, the higher standard deviation results in a more scattered crack scenario, which in turn results in longer crack growth periods, since less MSD-like scenarios are found. In addition, since also the threshold up to detectable crack length is influenced, also the linear fit, i.e. the dependency of the two parameters, is changed significantly. This may be seen in Fig. 13, which is parallel to Fig. 6 in all respects, apart from the different standard deviation. Also this is the crack growth curve for the minimum interval case. Compared with Figs. 6, 13 shows an even more pronounced MSD case of nearly equally sized cracks in the area between x ¼ ½40; þ100 mm. But in Fig. 6 the area with nearly similar sized cracks is even more widespread, although the outer cracks are not quite as large. This leads to a difference in the interval of 17; 100 cycles for sdev = 0.1 and 13; 100 for sdev ¼ 0:05: It can easily be understood that small standard deviations lead to more critical scenarios with respect to MSD/WFD. 3.1.5 Influence of the Mean Fatigue Life The influence of a decreasing mean fatigue life is shown in Fig. 14. The basic mean life of 115; 000 cycles is compared to a mean life of 80; 000 cycles. Both Fig. 12 Influence of the standard deviation of the fatigue data

16

P. Horst

Fig. 13 Crack growth for the minimum interval case (sdev = 0.1)

Fig. 14 Influence of the fatigue mean value

numbers are low for well designed joints, but may occur, if something extraordinary happens, as this is the case in MSD/WFD situations. It can easily be seen in Fig. 14 that the linear fit lines of the two distributions have almost the same gradient. This means that the dependency between threshold and interval remains the same, even if the threshold value itself drops considerably. The influences which have been discussed in Sect. 3.1 are surely only a part of all conceivable parameters. Other parameters could be: scattered SN-curves, scattered da/dN-DK-curves, scatter in the position of the rivet holes etc. Some of these influencing parameters have been tested in Horst and Schmidt [20] as well as [22].


17

The papers show that, although these parameters are interesting for the individual result of a simulation, the overall MCS results are not deeply affected. For this reason, these parameters are left out in this chapter.

3.2 More Realistic Problems This section is dealing with a more realistic, and also more MSD prone situation. The problem consists of 4 times 16 rivets, which are build in a way that is given in Fig. 15. The four parts are slightly separated, as this would be the case for four sets of rivets in adjacent frame-bays, which are all highly loaded compared with the other outer rivets. Only the rivets indicated by the filled symbols are taken into account as critical, while the other two rows are only included due to the load transfer ratio used in the calculations. Figure 16 gives the well-known interval versus threshold distribution for the two cases of the single 16 rivet set-up and the 4 times 16 rivets set-up. The results are not extremely different. Primarily, the threshold values drop, which can easily be interpreted as a size effect, while the interval values almost remain. This is a very interesting result, because it provides a very simple method to conclude from a few simulated sites to a larger set-up. This effect is surely linked to the fact that the critical crack length is in the order of magnitude of the frame spacing, which means that no interaction of the cracks in different frame-bays is needed to reach the residual strength limit.

Fig. 15 The more realistic problem of 4 frame-bays (only the critical rivet row shown)

18

P. Horst

Fig. 16 Comparison of the 1 9 16 and the 4 9 16 rivets set-up

30000 1 x 16rivets 4 x 16rivets

interval / cycles

25000

20000

15000

10000 140000

160000


The same effect is also visible in the two crack growth diagrams presented in Figs. 17 and 18 for the minimum and maximum interval case. Both diagrams are only giving a rough impression since the scale of the x-axis does not allow to distinguish too many crack growth curves, but the impression of the type of crack growth in both cases also becomes clear from this overview type of diagram. Only the last few thousand load cycles are shown in order to make the impression clearer. The computational effort for 250 scenarios of this type of scenario has been approximately 75 min on a 2 GHz PC. Since Fig. 16 showed an almost unchanged interval distribution, it would be interesting to try to deduce from single frame-bay scenarios to multiple frame-bay scenarios in the case of the initiation, and therefore the threshold numbers. The Fig. 17 Crack growth up to minimum interval


19

Fig. 18 Crack growth up to maximum interval

way the size effect may be calculated for the initiation phase is quite simple: the size effect with respect to fatigue may be assessed by means of the equation F ðnÞ ðzÞ ¼ ð1 ð1 F ð1Þ ðzÞÞn

ð7Þ

where F ð1Þ is the accumulated probability that a single location is facing a fatigue damage at time z; while F ðnÞ is the probability that one out of n locations is damaged. This method can be used in this case. The arithmetic mean and standard deviation for the fatigue life in the two cases with 16 and 4 times 16 rivets, i.e. 32 and 128 fatigue critical locations from the MCS are given in Table 2. The data for both, the complete crack growth as well as the crack growth up to 5 mm crack length in these cases are almost not changing, as may be seen in Table 3

Table 2 Fatigue life from calculations via MCS

Case unit

Mean value cycles

Standard deviation cycles

1 16 4 16

89,778 84,260

5,123 3,827

Table 3 Total crack growth and crack growth up to 5 mm via MCS Crack growth Crack growth Crack growth Crack growth \5 mm Case mean cycles standard cycles \5 mm mean cycles standard cycles value unit 1 16 4 16

75,439 75,080

4,843 4,021

56,856 56,947

3,911 3,391

20

P. Horst

Fig. 19 Data from MCS for 1 9 16 and 4 9 16 rivet cases, plus prediction via Eq. 9

1 0.9 0.8 0.7

F(x) / -

0.6 0.5 0.4 0.3 (32)

F MCS (128) predicted F eqn. 9 (128) F MCS

0.2 0.1 0 50000

75000

100000

125000

x / cycles

The data from Table 2 have been used to try to predict the data for the initiation of the 4 9 16 rivets case by using the 1 9 16 rivet data and Eq. 7. From this equation the following relation is found F ð1Þ ðzÞ ¼ 1 ð1 F ð32Þ ðzÞÞ1=32

ð8Þ

F ð128Þ ðzÞ ¼ 1 ð1 F ð1Þ ðzÞÞ128

ð9Þ

F ð128Þ ðzÞ ¼ 1 ð1 ð1 ð1 F ð32Þ ðzÞÞ1=32 ÞÞ128

ð10Þ

By using Eq. 10, the data in Fig. 19 have been found. A simple normal distribution has been used in this case, but a log-normal distribution could even be more appropriate. By looking at the data, an extremely good agreement between Eq. 8 and the result from the Monte Carlo Simulation is found. This shows, if the fact that the data in Table 3 for crack growth and crack growth until 5 mm crack length are almost the same for both cases is taken into account, that the overall prediction of complex set-ups is possible by using more fundamental, simpler example cases. This gives hope that reliable predictions are possible for such complex cases without exhausting computational sources, and to stay in a frame, which is still interesting for the industrial applier.

4 Possibilities of Feature Detection It is surely one aim of the work on MSD to judge from data, which are not needing a complete MCS, whether a scenario is likely to result in some sort of multiple site damage, or whether it is far away from this type of damage. Modern ways of


21

feature detection seem a good way to try to achieve this goal. Two different states of scenarios will be examined in this section for this purpose, namely • the initial damage scenario from the random process described above • the crack scenario at the point in time, when the first crack reaches 5 mm. Both types seem to be attractive. The first obviously does not need the deterministic crack growth calculation, the second only needs the deterministic calculation up to the first crack reaching 5 mm, which means that the cracks (for a rivet pitch of 20 mm) do nearly not interact. This in turn would reduce the computational effort dramatically, allowing even to use more complex 3D-models for the non-interacting crack growth. The type of diagram used for this purpose is shortly described by means of Fig. 20. This figure shows on the left-hand side the distribution of the damage indicated by a normalized number, i.e. damage D ¼ 1 indicates the initiation of a crack. The figure shows exactly the situation given by Fig. 6 for the crack growth of the minimum interval case in the basic example. Parallel to Fig. 7, the damage distribution is given in Fig. 21 for the maximum case. Figures 20 and 21 have to be explained: The left-hand diagram in both figures shows the distribution of the damage over the 32 fatigue critical locations. The thinner line marks the case which would yield ideal MSD, i.e. the same damage everywhere. The other diagrams are showing a form of a wavelet transform, namely the Haar transform of the distribution. This is a kind of information reduction scheme. A good introduction for this subject is given in Walker [23]. The Haar transform of the compressed signal works in different levels. This is from left to right, a rising level. The compressed data are getting smaller, while the fluctuations, which are the right tail of the graphs are getting longer. In essence, the Haar transform consists in the compression

Fig. 20 Damage distribution for the minimum interval case, including wavelet transform

22

P. Horst

Fig. 21 Damage distribution for the maximum interval case, including wavelet transform

am ¼

f2m1 þ f2m pffiffiffi 2

ð11Þ

dm ¼

f2m1 f2m pffiffiffi 2

ð12Þ

for the transform and

for the fluctuation, if f is the vector of the original signal and fi an element of this vector. This process is carried out from one level to the other. The rising level of the data is based on the fact that this method comes from signal processing, where a reduction of the data requires a higher level in order to achieve the same energy of the reduced set. The difference in the damage scenarios as well as in the Haar transforms is not very large, although the two cases are the extremes of the MCS. Therefore, it seems as if the initial damage scenario is not a very good indication for the MSD-likeness of scenarios. In Figs. 22 and 23 the distributions of the crack scenario, i.e. the crack length, of all cracks is shown at the point in time, when the first crack reaches 5 mm. These diagrams are again parallel to the crack growth scenarios in Figs. 6 and 7. What may be seen at first sight is that the distributions (level 0) in the two figures are quite different now, and this also seems to be the case for the Haar transforms. In this case the thin line is again the ideal MSD case, i.e. a 5 mm crack at each fatigue critical location. It is not possible to show all 250 different scenarios in this way, but there is a certain ranking in the different distributions, which may be used for the assessment of the criticality of a scenario with respect to MSD. A small insight in this line may be given by the two Figs. 24 and 25 for the minimum and maximum threshold, as given in Figs. 8 and 9. As explained before, in the discussion of Figs. 8 and 9, the scenarios are near to the ones on the extreme values for the interval, but, due to the negative gradient of the linear fit in Fig. 5


23

Fig. 22 Crack distribution at the point in time, when the first crack reaches 5 mm, minimum interval case, including Haar transform

Fig. 23 Crack distribution at the point in time, when the first crack reaches 5 mm, maximum interval case, including Haar transform

Fig. 24 Crack distribution at the point in time, when the first crack reaches 5 mm, minimum threshold case, including Haar transform

24

P. Horst

Fig. 25 Crack distribution at the point in time, when the first crack reaches 5 mm, maximum threshold case, including Haar transform

etc. the minimum and maximum scenarios are exchanged. Gradually, the two distributions are quite near to the extreme values. What is missing for the assessment of MSD-criticality is something like a single value-criterion. One way to try this is the normalized correlation between the ideal distribution (or a truncated version of this), which is indicated by the thin line and the actual distribution. The normalized correlation between two discrete distributions f and g is given by [23] as h f : gi ¼

f1 gk þ f2 gkþ1 þ þ fN gkþN1 f12 þ f22 þ þ fN2

ð13Þ

where f is one distribution ranging from ½1; N and g is an other one, which may have an other length and is possibly shifted by a value k: The ideal single value of a normalized correlation would be 1 in this case. This equation may be used in the determination of a criterion, first for the 1 9 16 rivets case discussed in Sect. 4 up to now, and later on for the 4 9 16 rivets case.

4.1 Correlation in the 1 Times 16 Rivets Case The results of the normalized correlation of the two signals, the damage scenario, as it is given in Fig. 22 for the minimum interval case, and the reference signal of constant crack size 5 mm, has been calculated for all 250 scenarios according to equation 13. The results have been plotted versus the interval, which is e.g. given in Fig. 5. The result is shown in Fig. 26.

Assessment of Multiple Site Damage in Riveted Aircraft Joints Fig. 26 Normalized correlation of the 250 scenarios in the 1 9 16 rivets case versus interval

25

2 normalized by reference normalized by signal

1.75

norm. correlation / -

1.5 1.25 1 0.75 0.5 0.25 0

0

5000

10000

15000

20000

25000

interval / cycles

The correlation has been normalized in two ways, once by the reference as vector f in Eq. 13, and in addition with the actual damage scenario as f: It seems to be obvious that none of the two ways of normalization yields more information than the other one. Therefore, it seems to be more appropriate to use the reference as normalizing vector. What is found in Fig. 26 is that the correlation obviously offers a good insight into the criticality of a scenario.

Fig. 27 Correlation for the 1 9 16 rivets case, including a constant 5 mm initial scenario

1


0.75

0.5

initial damage 250 scenarios 1000 scenarios / exponential fit constant a = 5 mm

0.25

0

0

5000

10000

15000

interval / cycles

20000

25000

26

P. Horst

One interesting information is that the normalized correlation between any level of the Haar transform and the corresponding Haar transform of the reference results in the same number. This can be used for larger systems if necessary. The same input as in Fig. 26 has been used for Fig. 27, apart from the fact that one additional input has been made. This is the interval for the case of equally sized cracks of 5 mm. The result does not meet the linear fit, but it is in an acceptable distance to the line. In addition to the data on the crack sizes at the point in time, when the first crack hits 5 mm, the correlation with the initial damage, as it is e.g. shown in Fig. 20 is given. The optical impression is that the cloud of results is less focused. Therefore, it seems to be right to rely on the crack scenario and not the damage scenario. In addition to the 250 scenario cloud, a 1; 000 scenario cloud has also been added. It might be a bit tricky to keep in mind that this cloud is a bit more scattered. What seems to be essential is that the linear fit is nearly exactly the same as for 250 scenarios. If a non-linear fit is used, as e.g. an exponential fit in Fig. 27, the result for the equally sized cracks gets nearer to this line. There is of course the idea pending that the correlation with the 5 mm ideal solution must not be required over the full length of the rivet row. This point is examined in Fig. 28. The number of fatigue critical locations which have been used for the correlation analysis are: 4, 8, 16 and 32 (as before). This correlation analysis somehow corresponds to the criterion of how near one of the values in the Haar transform of level 2, 3, 4 and 5 comes to the ideal transform value. It is not exactly the same, since in the Haar transform the blocks of fatigue critical locations

Fig. 28 Selective correlation over n fatigue critical locations

1


0.9

0.8

0.7

0.6

0.5

0.4

0.3 10000

4 8 16 32 equal size 5 mm 15000

20000

interval / cycles

25000

30000


27

are equidistant, while in this case the correlation has been made at steps of one single fatigue critical location. Surely, it is hard to distinguish between the different clouds. But what is visible is the fact that the clouds have different linear fits and are also a bit different in the scatter around the fit. This may be used for the examinations in Sect. 4.2 A linear correlation coefficient r\ 0:91 for both, the 8 and the 16 fatigue critical location shows that these criteria via normalized correlation are working quite well.

4.2 Correlation in the 4 Times 16 Rivets Case The same procedure has now been used for the case of 4 times 16 rivets, i.e. 128 fatigue critical locations. A Haar transform looks a bit more like signal processing in this case. Figures 29 and 30 show the minimum and maximum inspection interval case, as in the preceding sections. On the left-hand side, the original distribution of cracks is shown, while to the right the Haar Transforms are shown. In this more complicated case, the two original crack distributions do not look so differently, but on the right-hand side, the level 4 Haar transform is quite different, where 16 fatigue locations are condensed to one value. Clearly, the height of this transform is quite different in the two figures (Figs. 29 and 30). Therefore, this seems to be a good criterion to detect MSD-like situations. In Fig. 31, this is expressed in another way. Again the normalized correlation (selective over 16 sites) is plotted versus the interval. The figure supports the impression that this criterion, and this also means the criterion on the level 4 Haar transform, is a good way to assess the MSD-likeness of scenarios. Again, the linear correlation coefficient is r\ 0:91 for this case.

Fig. 29 Haar transform of the 4 9 16 rivet case, minimum inspection

28

P. Horst

Fig. 30 Haar transform of the 4 9 16 rivet case, maximum inspection Fig. 31 Normalized (selective) correlation over 16 fatigue critical locations for the 4 times 16 rivet case normalized correlation / -

1

0.75

0.5

0.25 16 equally sized 5 mm

0 0

5000

10000

15000

20000

25000

interval / cycles

5 Outlook The use of the wavelet transform with respect to a reduction of necessary calculations in the MCS will be shown by means of the same single frame-bay example as above, but in this case with many more scenarios. In this case, 25; 000 scenarios have been calculated. The question is whether the level x of the Haar transform can be used to indicate, whether a scenario is interesting with respect to critical values or not, i.e. how many scenarios have to be calculated in order to cover all possible scenarios below a certain value for the inspection interval (II). This number will be indicated by Nreq : Figure 32 illustrates the problem by means of one example. The shaded area covers those points, which are not interesting in the case that Nreq ¼ 16; 000 cycles.


29

Fig. 32 Visualization of one case of required inspection interval II

At a closer look, it becomes obvious that, in order to cover all points in the non-shaded area, the criterion would be to use a Haar transform level 5 value of slightly more than 15. This would cover the single point which is visible on the lower edge of the cloud near the shaded area. The amount of scenarios needed in this case is high, if all scenarios above Haar transform level 5 are calculated. Obviously, this would change quite a bit, if a lower Nreq -value is searched for. Figure 33 indicates the lowest Haar transform level 5 of each inspection interval or Nreq ; respectively, where the line including the symbols varies from inspection Fig. 33 Minimum Haar transform levels per inspection interval

30

P. Horst

interval to inspection interval, while the real minimum dotted line just indicates the lowest value starting with the lowest inspection interval from the left. It is then the point to find the number of scenarios to be analyzed in the sampling case which covers all necessary calculations, but no unnecessary ones. In a way, this value is linked to the dotted line of minimum Haar transform levels just discussed, because the Haar transform value given in this line has to be exceeded. Figure 34 shows the required scenarios to be examined for a given inspection interval II. It clearly shows that a considerable gain compared to the 25,000 scenarios calculated for the direct MC simulation is only found for small inspection intervals, i.e. low probabilities of failure. This number is the minimum number required, since it has been taken from a known MC simulation, where all analyses have been performed. This means that in reality the gain from this sampling method will be less good. There are ways to assess the necessary level 5 value for the prediction of critical and non-critical scenarios, as e.g. to calculate a low number of scenarios first and then to predict the necessary level. After this first step of analyzing only a certain percentage of the entire number of envisaged scenarios with a full analysis of the damage and crack growth as a trial set, only those further scenarios will fully be analyzed, if they exceed the level found above. In our reference case, 25; 000 scenarios have been analyzed. The results of these M percent full trial analyses are then used to extrapolate the minimum line of Fig. 35. If this line would be known, the gain in Fig. 34 would be met exactly. Surely, it does not make sense to choose the number M too high, since this would spoil the whole gain of the method. Figure 35 shows the line of the minimum values of 1%, 5%, 10% and 100% as initial trial set. In addition, the results of the complete 1% trial set are shown. It is visible that the global tendencies are already visible for a 1% trial set. Obviously, the upper left area is the most problematic area, which at the same time is the most interesting one. The problem of a reliable and efficient algorithm will be to find a conservative, but not

Fig. 34 Required scenarios and gain in calculation effort

25000

1

0.75

15000 req. scenarios gain

0.5

gain

Number of scenarios / -

20000

10000

0.25 5000

0

0 0

10000

20000

II / cycles

30000

Assessment of Multiple Site Damage in Riveted Aircraft Joints Fig. 35 Minimum values for different percentages of scenarios

31

25 1% results 1% Minimum-values 5% Minimum-values 10% Minimum-values 100% Minimum-values

Level 5 / -

20

15

10

20000

30000

II / cycles

extremely conservative prediction of from the 1% minimum values curve to the 100%-curve, especially for the low II and therefore pf values Acknowledgments A part of this chapter is based on the work which has been performed in the frame of the EU project SMAAC-Structural Maintenance of Ageing Aircraft.

References 1. Newman, J.C., Dawicke, D.S.: Fracture Analysis of Stiffened Panels Under Biaxial Loading with Widespread Cracking. In: Widespread Fatigue Damage in Military Aircraft, AGARDCP-568 (1995) 2. Atluri, S.N.: Energetic approaches and path-independent integrals in fracture mechanics. In: Atluri, S.N. (ed.) Computational Methods in the Mechanics of Fracture, Chapter 5, pp. 121–165 (1986) 3. Nilsson, K.F., Hutchinson, J.W.: Interaction between a major crack and small crack damage in aircraft sheet material. Int. J. Solids Struct. 31, 2331–2346 (1994) 4. Horst, P.: Multiple site damage in integrally stiffened structures. In: Proceedings of the 6th Joint FAA/DoD/NASA Conference on Aging Aircraft. San Francisco, USA (2002) 5. Balzano, M., Beaufils, J.-Y., Santgerma, A.: An engineering approach for the assessment of widespread fatigue damage in aircraft structures. In: Proceedings of the Second Joint NASA/FAA/DoD Conference on Aging Aircraft, NASA/CP-1999-208982/PART1, pp. 124–131 (1999) 6. Melchers, R.E.: Structural Reliability—Analysis and Prediction, 2nd edn. Wiley, England (1999) 7. Schijve, J.: Fatigue of Structures and Materials. Kluwer Academic Publishers, Dordrecht (2001) 8. Müller, R.P.G.: An experimental and analytical investigation on the fatigue behaviour of fuselage riveted lap joints. PhD Thesis, TU Delft (1995)

32

P. Horst

9. Press, W.H., Flannery, B.P., Teukolsky, S.A., Vetterling, W.T.: Numerical Recipes: The Art of Scientific Computing, 3rd edn. Cambridge Univerity Press, New York (2007) 10. Hastings, C.: Approximations for Digital Computers, p. 201. Princeton University Press, Princeton (1955) 11. Horst, P., Collins, R.A., Balzano, M., Santgerma, A., Cook, R., Young, A., Nilsson, K.F., Ottens, H.H., ten Hoeve, H.J.: Numerical ‘Round Robin’ tests on the assessment of MSD. In: Proceedings of the ICAF’97 Congress, vol. 1. EMAS, West Midlands (1997) 12. Fawaz, S.A., Andersson, B.: Accurate Stress Intensity Factor Solutions for Unsymmetric Corner Cracks at a Hole, FFA-Report, Sweden (2000) 13. Rooke, D.P.: Stress intensity factors for cracked holes in the presence of other boundaries. In: Stanley, P. (ed.) Fracture Mechanics in Engineering Practice, pp. 149–163. Applied Science Publishers, London (1977) 14. ESDU data sheet: The Compounding Method of Estimating Stress Intensity Factors for Cracks in Complex Configurations Using Solutions from Simple Configurations, ESDU Item Number 78036 (1978) 15. Rooke, D.P., Cartwright, D.J.: Compendium of Stress Intensity Factors. HSMO, London (1976) 16. Tada, H., Paris, P., Irwin, G.: Handbook of Stress Intensity Factors for Researchers and Engineers, 3rd edn. ASME, New York (2000) 17. Rooke, D.P., Tweed, J.: Opening-Mode Stress Intensity Factors for Two Unequal Cracks at a Hole, RAE Technical Report 79105 (1979) 18. Swift, T.: Damage tolerance capacity. In: Beukers, A., deJong, Th., Sinke, J., Vlot, A., Vogelesang, L.B. (eds.) Fatigue of Aircraft Materials—Proceedings of the Specialists’ Conference, dedicated to the 65th Birthday of J. Schijve, pp. 351–387. Delft University Press, Netherland (1992) 19. Niu, M.C.Y.: Airframe Structural Design. Commilit Press Ltd., Hong Kong (1988) 20. Horst, P., Schmidt, H.-J.: A Concept for the Evaluation of MSD Based on Probabilistic Assumptions. In: Widespread Fatigue Damage in Military Aircraft, AGARD-CP-568 (1995) 21. Broding, W.C., Diederich, F.W., Parker, P.S.: Structural optimization and design based on a reliability design criterion. J. Spacecr. 1, 56–61 (1964) 22. Horst, P.: Widespread Fatigue Damage — An Issue for Aging and New Aircraft. Keynote Lecture: 5th International Conference on Fracture and Damage Mechanics held in Harbin, China, 13–15 September 2006. In: Key Engineering Materials, vols. 324–325, pp. 1–8 (2006) 23. Walker, J.S.: Wavelets and their Scientific Applications. Chapman & Hall, Boca Raton (1999)

Laser Welding of Structural Aluminium L. Quintino, R. Miranda, U. Dilthey, D. Iordachescu, M. Banasik and S. Stano

Abstract This chapter starts with an overview of the fusion welding processes used in aluminium welding and further progresses by analysing in detail the characteristics of laser welding of aluminium. Laser sources for welding are available for a few decades but new concepts are coming to the market. The chapter addresses the most commonly used lasers for materials processing, CO2 and Nd-YAG (neodymium–yttrim aluminium garnet) and their interaction with aluminium alloys in welding applications. More recent laser types are also included, namely fibre lasers and disc lasers as, though only more recently available in the market, their potential is foreseen as being interesting for welding of aluminium. Hybrid laser MAG (Metal Active Gas) welding has proven to lead to good results in welding aluminium plates namely for long seam welding.

L. Quintino (&) IDMEC, Institute of Mechanical Engineering, TULISBON, Lisbon, Portugal e-mail: [email protected] R. Miranda UNIDEMI, Departamento Engenharia Mecânica e Industrial, FCT, Universidade Nova de Lisboa, Caparica, Portugal U. Dilthey Aachen University, Aachen, Germany D. Iordachescu UPM Laser Centre, Universidad Politécnica de Madrid, Madrid University, Madrid, Spain e-mail: [email protected] M. Banasik S. Stano Department of Welding Technology, Instytut Spawalnictwa, Gliwice, Poland e-mail: [email protected] S. Stano e-mail: [email protected]

Adv Struct Mater (2012) 8: 33–57 DOI: 10.1007/8611_2010_46 Springer-Verlag Berlin Heidelberg 2011 Published Online: 26 May 2011

33

34

L. Quintino et al.

1 Introduction Welding of aluminum can be done using several different processes from fusion arc welding to solid state processes or brazing and soldering. The present chapter focuses on the use of laser in welding of aluminum. It starts with an overview of the fusion welding processes and further progresses with an analysis of the potential of different types of lasers used for welding of aluminum alloys. Welding includes the joining process for metals and plastics where both the work pieces to be joined as well as the filler material used experience melting. A common method for welding metals is fusion welding, where a heat source is used to bring the parts to fusion temperature. One of the heat sources most commonly used in fusion welding is an electric arc created by an electrical discharge between an electrode and the work pieces to be welded together generating enough heat to melt the material under the arc. The solidification of the melted material forms the weld. Arc welding processes can use a consumable or a non-consumable electrode. In the first case, the molten metal from the electrode and the molten base metal mix together, solidifying to form a joint upon cooling. In order to protect the molten material from contamination or the surrounding atmosphere a flux or a shielding gas are used. In the second case, the joint is constituted by the base metal that melts and solidifies. High temperatures generated by the welding process alter the microstructure in the welded areas creating a fusion zone associated with the molten metal and a heat affected zone (known as HAZ) which undergoes metallurgical transformations. This can change the mechanical behavior of the material. In aluminum alloys, these metallurgical transformations can lead to softening of the material in the HAZ, cracking and porosity. The process of fusion and solidification also generates residual stresses that can lead to distortion. These are important in aluminum welding due to the high thermal conductivity of this material and linear expansion coefficient which leads to large fusion and heat affected zones. For these reasons, the welding process must be optimized (heat input, metal composition and cooling rate) aiming at minimizing microstructural changes and residual stresses in welded joints. Examples of fusion welding processes that can be used in aluminum are Oxygas, TIG/GTAW (Tungsten Inert Gas/Gas Tungsten Arc Welding) and MIG/MAG/GMAW (Metal Inert Gas/Metal Active Gas/Gas Metal Arc Welding). Without a doubt, the use of aluminum as a welded structural material improved with the introduction in the 1940s of the inert gas welding processes. With a welding process that uses an inert gas to protect the molten aluminum during welding, it became possible to make high quality, high strength welds at high speeds and in all positions.

Laser Welding of Structural Aluminium

35

2 Arc Welding Processes 2.1 Oxygas Welding Oxyfuel gas welding is a gas welding process where the heat is generated by an oxygen-fuel gas flame. An active flux is used when welding aluminum to remove the oxide and shield the weld pool. This was one of the earliest welding processes used for welding aluminum around 25 years prior to the development of the inert gas welding processes (GTAW and GMAW). The problem with this welding process is that the flux used during the process is hydroscopic, absorbing moisture from the surrounding atmosphere and the flux becomes corrosive to aluminum. Therefore, after welding, the flux must be removed to minimize the chance of corrosion. Other disadvantages of using this process for welding aluminum are the wide heat affected zones created and respective deterioration of mechanical properties. Welding is only practical in the flat and vertical positions, but distortion is always significant. The oxyfuel gas welding process was widely used for welding aluminum prior to the development of the inert gas welding process, but has limited use today.

2.2 Shielded Metal Arc Welding Shielded metal arc welding (SMAW), also known as manual metal arc (MMA) welding is a manual arc welding process that uses an electric arc created through the flow of current between a consumable electrode coated in flux and the parent material to be welded (Fig. 1). The electric current, either alternative current or direct current, is provided by a welding power supply. As the weld is deposited, the flux coating of the electrode disintegrates, giving off vapors that serve as a shielding gas and provide a layer of slag, both of which protect the weld area from atmospheric contamination [1]. The versatility and simplicity of the process determine its wide use around the world, namely in the maintenance and repair. Prior to the development of the inert gas welding process (GTAW and GMAW) the arc welding of aluminum was mainly restricted to the Shielded Metal Arc Process (SMAW). The core of the electrodes is in an aluminum alloy with a composition similar to that of the base material. The flux acts to dissolve the aluminum oxide on both the base alloy and the rod during welding. Some of the flux components vaporize in the arc to form shielding gases that help to stabilize the arc and shield the weld pool from the surrounding atmosphere. One of the main problems with this welding process was corrosion caused by flux entrapment, particularly in fillet welds where the flux could be trapped behind the weld and promote corrosion from the back of the weld. Other problems were that welds from this process are prone to gross porosity.

36

L. Quintino et al.

Fig. 1 Shielded metal gas welding

The limitations of availability of electrodes with some alloys compositions as for the high magnesium content base alloys and the fact that the electrodes, once exposed to the air, begin to absorb moisture into the flux, which eventually corrodes the aluminum core and produces excessive porosity problems, has led to its substitution by gas shielded welding processes. Current welding codes and standards for aluminum structures do not recognize this welding process as being suitable for production welding applications. For repair welding, SMAW is quite often the solution due to limitations of the case, like accessibility. When this is the case, careful welding procedure specifications need to be developed to minimise the problems referred above.

2.3 Tungsten Inert Gas (TIG) Welding Tungsten Inert Gas (TIG) or Gas Tungsten Arc Welding (GTAW) process is quite often a viable option for welding aluminum. The processes uses an electric arc like in SMAW but established between a non-consumable electrode in tungsten and the material to be welded, under an inert shielding gas to protect the melted metal from atmospheric contamination (Fig. 2). This process is widely used to successfully weld aluminum alloys. Aluminum alloys form refractory surface oxides which absorb the heat of the arc preventing the melting of the underneath material and thus creating difficulties for welding. In order to overcome this, alternating current (AC) is used for most applications. AC current provides a surface cleaning action while the electrode is positive (DCEP), due to the flow of electrons from the negative pole to the plate and reduced overheating of the tungsten electrode by dividing the arc heat about evenly between electrode and base material (Fig. 3).


37

Fig. 2 Gas tungsten arc welding- GTAW/TIG

Fig. 3 a Schematic representation of the effect of alternative current in GTAW of aluminum. b Positive electrode. c Negative electrode

GTAW power sources for AC welding, which allow for adjustment of the balance between polarities, enable the user to choose either enhanced arc cleaning or greater penetration capabilities. Direct current (DC) power is employed for some specialized applications, as thin sheet welding because risks for burn through reduce since the plate heats less. Argon shielding is generally used for welding aluminum with alternative current due to a better arc starting, cleaning action and weld quality when compared to helium. The cost of helium is also usually higher than argon.

38

L. Quintino et al.

Fig. 4 Gas metal arc welding (MIG/MAG, GMAW)

2.4 Metal Inert Gas/Metal Active Gas Welding (MIG/MAG) With MIG (Metal Inert Gas) or MAG (Metal Active Gas) welding also called GasShielded Metal Arc Welding (GMAW), an arc is maintained between a continuous solid wire electrode and the base material of the work piece. The arc and the weld pool are shielded by a inert or active gas (Fig. 4). A constant voltage, direct current power source is most commonly used in order to obtain a correcting arc. The MIG/MAG welding process operates on D.C. (direct current) usually with the wire electrode positive. This is known as ‘‘reverse’’ polarity. ‘‘Straight’’ polarity is seldom used because of unstable transfer modes of molten metal from the wire electrode to the work piece. There are four primary methods of metal transfer in MIG/MAG, called globular, short-circuiting, spray and pulsed-spray, each of which has distinct properties and corresponding advantages and limitations. Short circuiting transfer occurs for low currents and low voltages and is desirable to weld thin plates and in positional welding due to the small dimension of the molten pool and the balance of forces actuating in the metal transfer, which allow for transfer against gravity (Fig. 5). Globular transfer is obtained with medium current and voltage levels and is characterized by big droplets detaching form the wire due to gravity force. This is usually associated with spatter and instability in the electric arc (Fig. 5). Spray transfer is characteristic of currents above the transition current and high voltages and is used in high productivity welding for thick plates in the down hand position (Fig. 5).


39

Fig. 5 Spray (left) and globular (right) transfer modes

Pulsed spray is obtained with the use of pulsed current in MIG/MAG. Processes like GMAW-P (Pulse) also introduce additional complexity (with the need of defining supplementary variables such as peak current, base current, peak time, base time, frequency and duty cycle) which needs intense process knowledge for correct definition. For achieving good weld quality, one droplet should detach in every pulse to produce best weld quality with minimal defects and spatter. There are many power supplies in the market with embedded software for the most common ranges of applications, thus facilitating the choice of the welding procedure. Aluminum can be welded with GMAW and GMAW-P with good results.

3 Laser Welding 3.1 Basic Considerations Laser beam welding is a fusion welding process where radiant energy is used to produce the heat required to melt the materials to be joined [2]. A concentrated beam of coherent, monochromatic light is directed by optical devices and focused to a small spot, for higher power density, on the abutting surfaces of the parts being joined (Fig. 6). Gas shielding is generally used to prevent oxidation of the melted material. Different parts and even different metals can be joined in a noncontact process. The required accessibility to the work piece being from one side only and the opportunity to abandon filling material completely are the main advantages of laser beam welding. Laser joining can be performed using either pulsed or continuous wave mode lasers. Lasers are playing an important role in the joining of materials since the invention of high power solid state and gas lasers in 1964. In the first years, the main applications were resistance trimming of electronic circuits but with the

40

L. Quintino et al.

Fig. 6 Laser welding

introduction in the market of a reliable high performance laser in the late 1970s, welding of sheet metal parts was possible. Several types of lasers can be used in materials processing though the most common have wavelengths of 10.6 lm—CO2 lasers (gas lasers), or 1.06 lm— Nd-YAG lasers (solid state lasers). The last decade has seen the rise of diode lasers and diode pumped solid state lasers. More recently, high power diode pumped fiber lasers and disk lasers were developed. These new lasers have been stated to be a serious alternative to solid state and carbon dioxide lasers for different materials processing applications. Beam quality is of major relevance in materials processing with high power laser. The beam quality is defined by ISO 11146:1999 by the beam parameter product (BPP) or the M2 factor, and these assess the ability to focus a laser beam. The beam quality of a solid state laser is often specified by the beam parameter product (BPP) defined as follows: BPP ¼ a:x0

mm:mrad

ð1Þ

where x0 is the radius of the beam waist and a is half the total divergence angle. Low values of BPP express good quality beams or the ability to focus a beam to a small spot. Beam quality is also dependent on the laser radiation wavelength since the beam diameter at the focus is directly related to the wavelength (Eq. 2). do ¼

4k f pD

ð2Þ

where do is the beam diameter at the focus, D the beam diameter before the focusing system with a focal distance f and k the wavelength.

Laser Welding of Structural Aluminium Table 1 Beam quality for different laser systems

41

Laser type

Typical beam quality mm.mrad

Wavelength nm

CO2 Lamp pumped Nd:YAG Diode pumped Nd:YAG Yb Fibre laser (7,000 W) Yb Fibre laser (IPG YLR 17,000 W) Thin disc Yb YAG (1,500) Yb Disc Lasers (40,000 N)

3.7 25 12 18.5 11.7

10,600 1,060 1,060 1,070 1,070

7 10

1,030 1,030

Fig. 7 Estimated laser operating costs [6]

Generally, CO2 lasers have very good beam quality, while solid state lasers are usually worse. However, fibre lasers have achieved significantly better beam quality, as shown in Table 1 [3–5]. It is well know that the use of laser welding requires high investment costs but it is wise not to forget that the running costs will also play an important part in the production costs. Figure 7 give a comparison of operating costs for different laser systems [6]. Laser welding of aluminium is widely used in industry, but there are nevertheless limitations which need to be addressed for each application. First, there is a poor coupling of the laser energy due in part to the high density of free electrons in the solid, making aluminium one of the best reflectors of light. In addition, many aluminium alloys contain magnesium or zinc, which are easily vaporized forming a plasma that blocks the incident beam. Other issues that need to be considered when laser welding aluminium are low power absorption, alloy compositional differences and the importance of surface preparation. The power absorption changes dramatically at times, producing an unstable process with poor penetration control and a rough bead surface (it is commonly believed that the difference in the fraction of absorbed power is caused by melting of the metal) [7].

42

L. Quintino et al.

Fig. 8 Sketches of conduction laser welding (a), and of keyhole laser welding (b) [7] Fig. 9 The radiation absorption of metals as a function of wavelengths of electromagnetic radiation

Welding aluminium alloys can be done both by conduction mode and keyhole mode (Fig. 8). Weld pool shapes in aluminium depend on the mean intensity of the laser beam and the laser pulse time. The start of the laser welding process of aluminum is difficult in itself because of high reflection coefficient of laser radiation, Fig. 9. When the liquid metal appears, the radiation absorption increases (the reflection coefficient decreases), although it has remained on relatively low level. The keyhole technique is more widely used on aluminium. To this end, it is necessary to provide suitable power density of the laser beam, i.e., to reach the threshold value of power density, above which is formed the capillary filled with gases and metal vapors and surrounded with a thin film of molten metal. The laser radiation is absorbed by metal vapors in the capillary and by molten metal as a result of multiple reflections of the laser beam from the capillary walls covered with the liquid metal. For iron based materials, the threshold value of power density is about 106 (W/cm2). For aluminium and its alloys, on account of much


43

higher thermal conductivity and high coefficient of laser radiation reflection, the threshold value of power density is about 1.5 9 106 (W/cm2), while for a stable process it should be not less than 2 9 106 (W/cm2) [8]. The above requirements bring about the necessity to use more powerful lasers than in case of welding of steel. In case of standard optical system enabling to focus the laser beam e.g., to the spot about 0.6 mm in diameter, the minimum power of the solid state laser of the Nd:YAG type should be more than 3 kW. It should be kept in mind, however, that during welding with use of the low– power lasers there is danger of reflection of the laser beam from the workpiece surface back to the laser optical system. Thus, the welding head and the laser light cable can be damaged. The reflected radiation can also come back through the laser head and the laser light cable to the resonator. In extreme cases, the laser resonator can be damaged due to excessive thermal load that is due to exceeding the permissible power density inside the laser active element. Laser welding of aluminium is used in industry with appropriate welding procedures that prevent the occurrence of defects. In case of aluminium, large trapped vapour pores near the root of keyhole-mode welds are common at higher power density. Due to the important difference (20:1) between hydrogen solubility values in liquid and solid aluminium, respectively, hydrogen remains trapped. The use of shielding gas is the common solution. The pores are mainly due to hydrogen that can be significantly eliminated by surface milling and vacuum annealing [3]. Pores may also be produced by the vaporisation of the alloying elements, especially magnesium. An unstable keyhole with tendency to form ‘‘bottle neck’’ will increase the level of porosity. But the real danger generated by the presence of hydrogen is hot cracking susceptibility. Solidification cracks are more common in case of aluminium alloys containing copper. The partially melted zone is defined as the HAZ sub-region in which a peak temperature between the liquidus and solidus was attained during welding. Localized melting occurs, accompanied by segregation at grain boundaries. A microstructure is produced that is unable to withstand the contraction stresses generated when the weld metal solidifies, rendering it susceptible to solidification cracking. Heat treatable aluminium alloys of the 6,000 series, for example, are susceptible to this type of cracking. These aspects need to be taken into consideration when developing laser welding procedures.

3.2 CO2 and Nd-YAG Lasers The two most industrially used lasers are carbon dioxide and Nd/YAG lasers. The former is a gas laser emitting at a wavelength of 10.6 l with higher output powers, efficiencies around 20% (higher than Nd:YAG lasers), with a good beam quality, easy to focus but requiring complex and robust manipulation systems since the light cannot be transmitted via optical fiber. These lasers have high output powers

44

L. Quintino et al.

Fig. 10 Variation of absorption over wavelength and type of material (TRUMPF courtesy) [24]

(up to 50 kW) which allows for keyhole welding modes in thick Al plate, despite the poor absorption of its wavelength [9]. Nd:YAG lasers are solid state lasers emitting at 1.06 l in a range of wavelengths where the metallic materials are most absorbent, but have poor energy conversion efficiency (around 5%), limited output powers and high running costs specially due to maintenance. The possibility to be transported via optical fiber improves the system flexibility, but the most relevant advantage is the high absorption of Al alloys of its laser wavelength. That is, for the same weld penetration, the required power is lower for Nd:YAG laser than for CO2 laser. Diode lasers and diode pumped lasers have been the subject of innovative developments in the last decades. They use the same principles as a diode, so they are classified as solid state lasers and in recent years several developments have been performed by manufacturers. The resonator cavity is made of two coated bars to have the desired optical properties. They have good energy efficiencies in the range of 30–50% and can be transported via an optical fiber. Together with the low implementation area, they are easily integrated in manufacturing lines. Diode lasers are particularly adequate for laser welding of Al since they emit in the range of 800–900 nm (less than Nd/YAG lasers). CO2, Nd:YAG and diode lasers can be used for conduction mode welding (Fig. 8a) of metals and alloys; the absorptivity of the majority of metals increases as the wavelength decreases (Fig. 10). The use of CO2 lasers will involve more absorption issues, since its wavelength is 109 higher than Nd-YAG beams. Conduction welding is normally used with relatively small components, with the beam delivered to the workpiece via a small number of optics; the use of optics instead of optical fibers is compulsory for CO2 laser beam wavelength. Simple beam defocusing to a projected diameter that corresponds to the size of weld to be made, allowing for any gaps in the joint, is normally sufficient.


45

As referred before, keyhole mode welding (Fig. 8) is the most common for joining metallic materials by laser, aluminium included. Energy is absorbed by the material through two mechanisms, which determine the overall energy transfer efficiency. Inverse Bremsstrahlung absorption (transfer of energy from photons to electrons) takes place in the partially ionized plasma formed in and above the keyhole; it is the dominant mechanism at low welding speeds. Fresnel absorption by multiple reflections at the walls of the keyhole dominates at high welding speeds, and is dependent on the polarization of the beam. Plasma (ionized vapour) and plume (vaporized material) facilitate energy transfer from the beam to the material, but they also defocus and absorb the laser beam, reducing its power density. This absorption is more evident and must be avoided in case of higher wavelength (case of CO2 laser, 10.6 lm), than in the case of wavelengths about 1 lm (Nd:YAG—1.064 lm). The deleterious influence of the plasma and vapours plume can be drastically reduced by using appropriate gas flow. The same gas is also used for shielding. The keyhole is surrounded by molten material having a conical shape. The cavity is maintained through equilibrium between opening forces arising from material ablation and plasma formation, and forces caused by the surface tension and hydrostatic pressure of the molten pool, which act to close it. The requirement to maintain this balance leads to practical minimum and maximum travel speeds for keyhole welding—excessive speed causes the keyhole to collapse, whereas insufficient speed results in a wide weld bead that sags. Due to the phenomenon evolution inside the keyhole, the shape and size of the keyhole have a certain dynamic and fluctuate during welding and may negatively influence the constancy of penetration value. During aluminium welding in keyhole mode, the absorption process is much more unstable than in the case of steel welding. The molten metal conical wall of the keyhole and the resulting seam are wider than in case of steel. Even if the vaporisation temperatures (of the keyhole walls) for steel and aluminium are close, the fusion temperature of aluminium is about 900 K, whilst for steel is about 1,800 K, while thermal conductivity of aluminium is higher than steel. Due to the increased volume of molten metal and to its lower viscosity, the aluminium welding process is characterised by important flow of molten metal. Moreover, there is a trend of flowing through the root, generating excess of penetration and ‘‘drop’’, whilst the lack of material at surface may generate undercut. Root flaws may be prevented by assuring a support (gas, flux, or other) at root or by an appropriate design of the groove.

3.3 Fiber Lasers Fiber lasers date back from the early 1960s, in low power applications [10, 11]. In 2000, the first 100 W fiber laser was produced [12] and in past years increasingly multi-kilowatt fibre lasers have been introduced for materials processing up to 30 kW, and the technology is available to produce systems of higher power.

46

L. Quintino et al.

Fig. 11 Double-clad fiber laser [3]

Fig. 12 Schematic presentation of a fiber laser architecture [15]

High power fiber lasers have multiple advantages, including: high efficiency; very compact design with minimum footprint area; good beam quality with small beam focus diameter; and a robust setup for mobile applications [13]. The active medium is the core of a silica fiber doped with a rare earth: erbium, ytterbium, neodymium or thallium. The pumping process uses multimode diodes rather than diode bars. The laser beam is emitted longitudinally along the fiber. Two Bragg gratings inscribed on the fiber limit the wavelength of the emitted beam. The resonator cavity is thus constituted by the fiber itself, either the core, as in single-mode lasers or the inner cladding around this core as in double-clad fiber laser, as shown in Fig. 11 [14]. The outer cladding is made of a polymeric material with low refraction index to minimise attenuation (Fig. 12) [15]. The resulting laser beam is essentially diffraction limited and when fitted with an integral collimator, produces a beam that is extremely parallel. Figure 13 shows the beam wavelength for different rare earth doping agents. Typically, ytterbium-doped multi-clad fiber has an emission wavelength of 1.07– 1.08 l. The diode pumped energy is delivered to the active medium via multimode fibers that are spliced to the multi-clad coil. Fiber lasers are modular so, for example, a 1 kW unit can be made up of ten individual fiber lasers integrated into a common cabinet. Although the beam is no longer single-mode, the resulting beam quality is better than for the same power in Nd-YAG lasers. A 6 kW laser can deliver a beam via a 200–300 lm fiber.


47

Fig. 13 Fiber laser spectral ranges [15]

Ytterbium fiber laser has a wall plug efficiency of 16–20%. Though erbium and thulium fiber lasers demonstrate lower wall plug efficiencies, they are still more efficient than typical YAG lasers. Additionally, the lifetime of the pumping diodes exceeds the lifetime of other diode pumped lasers [13], reducing costs. Fiber laser welding of aluminium has been developed and, as with CO2 and NdYAG lasers, provided a correct welding procedure is used, sound welds with no porosity or cracking are obtained [25].

3.4 Disk Lasers Disk lasers are of the newest generations of solid state lasers. The scheme of the disk laser is shown in Fig. 14. The active element of disk lasers is YAG crystal in the form of a disk, 150– 300 lm in thickness and approximately 12–15 mm in diameter. Usually, the crystal is ytterbium–doped (Yb:YAG). The laser disk is mounted on the copper radiator carrying away the heat produced during generation of the laser beam. The active element –Yb:YAG disk—is optically excited by lighting it with diode laser radiation focused on the disk to form a spot, several millimeters in diameter. The application of diode lasers makes it possible to match closely the wavelength of pumping radiation to the Yb:YAG crystal pumping absorption. Because the disk is very thin, only a part of the pump radiation passing the disk is absorbed. The remainder is reflected by the rare side of the disk and is directed back to the disk through the mirror system. When it is focused again, a part of its energy is turned over to the disk in order to excite it. Nowadays, in case of one disk, up to 20 optical paths are used for pump radiation enabling to achieve optical efficiency of the laser on the level of above 60%. This, in turn, means that the wall plug

48

L. Quintino et al.

Fig. 14 Scheme of Yb:YAG disk laser (on the left) and resonator of the TruDisk Laser (on the right) (source Trumpf GmbH ? Co. KG)

Fig. 15 Temperature profile on the cross-section surface of the active element of conventional Nd:YAG lamp-pumped laser (on the left) and disk laser (on the right) (source Trumpf GmbH ? Co. KG)

efficiency is about 25%. On account of the crystal geometry, the heat flow occurs in one direction only, i.e., along the resonator optical axis. The temperature on the disk cross-section surface is constant—isothermal lines are perpendicular to the laser optical axis. Lack of temperature differences on the disk cross-section surface results practically in elimination of unfavourable phenomenon of crystal deformation on this surface which happens in conventional, lamp-pumped Nd:YAG lasers, Fig. 15. Owing to that, the laser beam quality depends only to a low degree on the power of emitted radiation. The rear side of the disk adjacent to the heat


49

Fig. 16 Scheme of the coupling of several disks in one resonator (on the left) and an example of disk laser resonator composed of four disks (on the right) (source Trumpf GmbH ? Co. KG) [24]

Fig. 17 Relation between welding speed and penetration depth in disk laser welding (source Trumpf GmbH ? Co. KG) [24]

exchanger, to which the crystal is mounted, has a reflective surface, which reflects laser beam and pump light. Disk laser fabricators are incessantly striving after the increase in laser beam power from single disk. Now (in 2010), it is possible to obtain about 5.5 kW laser beam power per disk. Calculations show that there is no fundamental limitation up to 30 kW per disk [16, 17]. The laser beam power depends directly on that of the pump radiation, i.e., on the power of diode lasers generating the pump radiation. A further increase in laser power can be obtained by setting-up of several disks on one optical path. Theoretically, it is possible to set-up any number of disks. In practice, more than four disks are not used in one resonator, (Fig. 16). The time of failure-free operation between replacements of pumping elements has been extended significantly in comparison to lamp—pumped solid state lasers—the life of diode lasers used for pumping is estimated at about 20,000 h. The designs used at present enable to exchange the single diode pack in 3 min

50

L. Quintino et al.

without the necessity of aligning of the pumping system. Planned replacements of pumping systems have not an essential influence on the production continuity. Nevertheless, in order to protect the welding head and the laser light cable transmitting the laser radiation, it is recommended to incline the head about 5–15 in relation to the axis perpendicular to the material under welding so that the reflected laser beam, if any, does not meet again the optical system. The welding process can be carried out with both positive and negative incident angle of the beam. In many cases, when welding by conventional single beam laser technique, it is possible to obtain quality joints while high welding speed is achieved simultaneously, (Fig. 17). As with fiber lasers, disc laser welding of aluminium alloys has been tested with promising results.

3.5 Hybrid Laser MAG Welding If two welding processes, such as laser welding and MIG welding are coupled in one common process zone this is called a ‘‘Hybrid Process’’, a Laser-MIG-Hybrid welding process, (Fig. 18). The laser beam and the MIG arc are interacting in the process zone; they are coupled via the molten metal and in the majority of cases via the common process plasma. Generally, the use of filler material may work out not only bead geometry issues, but also metallurgical and joint strength problems. Besides the bad appearance of the bead, an inadequate geometry introduces stress picks that multiply the local loading several times. On the other hand, the losses caused by burning of alloying elements (mainly those with low vaporisation temperature, namely Mg and Si) may be appropriately compensated by using filler metal. Reports have also indicated the stabilisation of the keyhole dynamic when filler metal is used. Definitely, the best solution becomes the use of laser-GMA hybrid welding, that provides very stable process and combines in a holistic way the advantages of these two welding processes. The coupling of both methods results in process-specific advantages which are called synergistic effects: deep penetration, high welding speed, low heat input and low distortion, Figs. 19 and 20. In Laser-MIG-Hybrid welding, the addition of filler material is carried out via the MIG arc process. The application of the MIG technology entails the advantage that the filler wire can be molten with relatively inexpensive energy while the high-quality laser beam energy is maintained in an almost undiminished form for the penetration and the material transport into the depth. The positioning of the wire to the laser beam is relatively easy compared to cold filler wire feeding as by the common plasma a guidance of the arc towards the keyhole can be observed. The Laser-MIG-Hybrid process allows avoiding the substantial disadvantages of laser beam welding without filler material. The process offers excellent gap bridging ability and allows to influence the weld pool metallurgy without


51

Fig. 18 Laser MIG hybrid welding process

Fig. 19 Laser MIG hybrid welding of aluminum alloys, contribution of laser and MIG process

abandoning the main advantages of laser beam welding—the deep penetration and the high welding speed. The MIG arc is either pulsed arc or is triggered by direct current arc, in the most cases a spray arc via commercial MIG welding power sources. The shielding of the welding zone is realised through pure helium or helium-argon gas mixtures. Low argon contents in the shielding gas exert positive influence on the material

52

L. Quintino et al.

Fig. 20 Laser MIG hybrid welding, increase of welding speed and reduction of heat input

transition and the arc stability. The applicable gas mixture is a compromise between high plasma shielding and arc stability. Welding filler materials are mainly commercially available solid wires which are also used for MIG welding of the aluminium alloys. The advantages of the Laser-MIG-Hybrid welding processes can be summarised as followed: • • • • • •

High process speed Low heat input Low thermal distortion Good gap bridging ability Excellent weld properties Good process stability

Despite all the advantages of Laser-MIG-Hybrid welding the fact still remains that, caused by two coupled welding processes, the number of setting variables increases. For a reliable and stable hybrid welding process and excellent weld properties, the exact setting of the parameters is mandatory.

4 Advantages, Limitations and Trends The main advantages of laser welding reside in the shape of the weld and good penetration, high precision, high mechanical properties of the weld, high welding speed, low heat input, high flexibility and possibility of automation and robotising. In general, there is no need to machine the weld/joint after welding. The case of beam transmission and directional control permits multi-station operation. Nearly 100% duty cycle is possible by switching the beam from station to station. The high energy density beam enables narrow welds and narrow heat


53

affected zones, and therefore good retention of metal properties and relatively low distortion of the work piece. The main disadvantages are the relatively high cost of investment and the important requirements related to the machining of the parts to assure a precise groove (reduced dimensional tolerances). Most of the multi-kilowatt laser systems operate at only about 10–20% efficiency for converting electrical power into a focused infrared laser beam. Higher productivity is possible in laser welding through: a reduction in processing time because of the low inertia of laser welding; a reduction in the use of materials; the use of a wide range of materials; a reduction in labour costs; a reduction in scrap; a reduction in process cycle time through the minimization of post treatment working; a reduction in response time to orders; the high equipment availability time. Higher product quality, in terms of improved in-service properties, can be achieved through: improved tolerances; accurate control of process parameters; selection of new materials; and product redesign. Laser welding is environmentally friendly; it uses clean energy and few contaminating materials [2]. Hybrid welding processes combine the characteristics of their constituent techniques: a laser beam is highly penetrating, but intolerant to variations along the joint line; these can be accommodated by more forgiving arc fusion processes. Hybrid processing enables technical problems to be overcome as well as reducing the requirements placed on the laser, leading to a reduction in capital investment. This type of reasoning can be applied to most thermal mechanisms of laser processing, presenting opportunities for novel application of an existing technology [18]. Laser welding of aluminium is commonly used on many applications through procedures for keyhole welding, and quality standards must be generated and approved by classification societies, particularly for thick section materials. Both are currently obstacles to greater application. Equipment for on-line process monitoring and quality control is continually being sophisticated. This will enable rugged systems to be made, which will increase process automation, and make the process more attractive to a wider range of industry sectors. Weldability of aluminium alloys depends in great part on the alloy grade, i.e., on the alloy chemical composition. Some of them, due to increased susceptibility to hot cracking, need to be welded with the use of additional material modifying the chemical composition of welds. Three joint designs (butt, tap, and flange) are best suited for laser welding of aluminium [7]. Butt joints with sheared edges are acceptable provided they are square and straight. A fit-up tolerance of 15% of the material thickness is desirable. Misalignment and out-of flatness of parts should be less than 25% of the metal thickness. In lap joints, air gaps between pieces to be welded severely limit weld penetration and/or weld travel speed. For round welds in aluminium, no gap can be tolerated unless inert gas coverage can be maintained over the entire weld area.

54

L. Quintino et al.

Flange joint geometry is especially suited for aluminium because of the aluminium’s high shrinkage rate. Square edge and good fit-up are also necessary. Shipbuilding and aerospace are industries where the application of laser welding of aluminium can be often found. Although the first aircraft made by Boeing was partially welded, mechanical fastening methods such as riveting have dominated because of the difficulties associated with fusion welding of the common aluminium aerospace alloys 2024 and 7075. Laser welding can reduce joining costs relatively to adhesive bonding and riveting by about 20%. EADS Airbus has invested heavily in laser welding as a replacement for riveting in non-critical applications. One application involves joining stiffening stringers to the skin of the fuselage. The damage-tolerant alloy 6013 is the base material and 4047 the filler material. The stringers are welded from two sides at 10 m/min, using two 2.5 kW CO2 laser beams. The joint is designed such that the HAZ is contained in the stringer, and does not impinge on the skin. The process was first used in series production of the Airbus 318, and was then implemented in other aircraft types. High power fiber lasers can be used for deep penetration welding in a diversity of materials since the low wavelength that characterizes these lasers allows its absorption by almost all metals and alloys and the fiber delivery system provides the necessary flexibility for beam manipulation and positioning especially when 3D is required for processing. Micro-welding was done in medical equipments [19]. The Bremer Institute in Germany studied fiber laser welding in AA6056 aluminium alloy 6 mm thick, with a 6.9 kW laser and a welding speed of 50 mm/s [20] with excellent results. Quintino et al. performed successful experiments in high power laser welding of AA7150 aluminium alloy, for pipeline application [21, 22] with a 8 kW fiber laser installed in Cranfield Institute in UK. Laser welding of aluminium has important applications in automotive industry since aluminium started to be used not only for gear box, but also for structural parts. The new concept of aluminium integrated car body involves the use of extruded high strength elements jointed in nodal points. Both laser welding and laser-GMA hybrid welding are used, besides classical arc welding and brazing processes. Laser-GMA hybrid welding has much more volume of applications in industry when compared with laser welding [23]; e.g., from about 500 km of weld in case of a cruiser, about 250 km are made by laser-GMA welding, but in this case the major part of the structure is made of naval steel. Laser-MIG-Hybrid applications of aluminium-alloys in the manufacturing industry, especially in the car industry, show the high potential of this method, Figs. 21 and 22. Through the development of innovative laser systems, as, for example, fibre laser and disc laser, which excel with their compact design, high efficiency, high precision and high flexibility new possibilities for the Laser-MIGHybrid welding methods present themselves. The increasing power and beam quality of solid state lasers will emerge new possibilities especially for welding light metals.


55

Fig. 21 Laser-MIG-hybrid welding, industrial application, reduction of distortion, audi A8 roof rail

Fig. 22 Laser-MIG-hybrid welding, industrial application, lightweight aluminium car door, VW phaeton

56

L. Quintino et al.

5 Conclusions Lasers have a good potential for welding aluminium namely in deep penetration welds, where the high power of the beam combined with a sharp focus, can penetrate comparatively thick joints, while producing a very narrow weld, and a correspondingly narrow HAZ, when compared to arc welding processes. The current understanding of the important physical processes occurring during laser welding of these alloys such as energy absorption, fluid flow and heat transfer in the weld pool, and alloying element vaporisation create a sound platform to use this process. The market presents quite a few options of different types of lasers and combination of laser and arc welding for the user to choose from. Whichever the part or construction to be welded and whichever the laser equipment to be used, care must be put in the definition of the welding procedure to assure sound welds with mechanical properties matching the base material.

References 1. Welding Handbook: Welding Processes, Luisa tens esta vol. 2, 8th edn. (1991) 2. Ion, J.C.: Laser Processing of Engineering Materials. Elsevier Butterworth-Heinemann, Amsterdam (2005), GB, ISBN 0 7506 6079 1, pp. 327–344, 395–448 3. Weckman, D.C., Kerr, H.W., Liu, J.T.: Metallurgical and mechanical characterization of laser spot welded low carbon steel sheets. Metall. Mater. Trans. B28, 687–700 (1997) 4. Seefeld, T., Thomy, C., Kohn, H., Vollertsen, F., Jasnau, U., Seyffarth, P.: Erste Erfahrungen und Anwendungs—untersuchungen mit einem 10 kW—Faserlaser. 5. Workshop ‘‘Industrielle Anwendungen von Hochleistungs—Diodenlasern’’, 21./22. Oktober 2004, Fraunhofer IWS, Dresden 5. Assunção, E., Quintino, L., Miranda, R.M.: Comparative study of laser welding of tailor blanks for the automotive industry. Int. J. Adv. Manuf. Technol. (2010), 49(1–4), 123–131 doi:10.1007/s00170-009-2385-0 6. Vaidya, W.V., Horstmann, M., Ventzke, V., Petrovski, B., Kocak, R., Kocik, R., Tempus, G.: Structure-property investigations on a laser beam welded dissimilar joint of aluminium AA6056 and titanium Ti6Al4 V for aeronautical applications Part I: local gradients in microstructure, hardness and strength. Materialwiss. Werkstofftech. 40(8), 623–633 7. The Different Modes of Laser Welding, http://laserwelding.me 8. Ion, J.C.: Laser beam welding of wrought aluminium alloys. Sci. Technol. Weld. Join. 5(5), 265–276 (2000) 9. Rofin-Sinar Laser GmbH. http://www.rofin.com/en/products/solid_state_lasers/disc_lasers/ ds_series/. Accessed 10 May 2010 10. Miyamoto, I., Park, S., Ooie, T.: Ultrafine keyhole welding processes using single-mode fiber laser. LMP Section A, pp. 203–212 (2003) 11. Miyamoto, I., Kosumi, T., Seo-jeong, P., Huragishi, H., Watanabe, K., Ooie, T.: Applications of single mode fiber lasers to novel micromachining. In: Proc. LMP 2004, Osaka May 2004 12. Hill, P.: Fiber laser hits 2 kW record mark. Opto and Laser Europe (OLE), July/August 2002, p. 9 13. Thomy, C., Seefeld, T., Vollersten, F.: Application of high power fibre lasers in laser and MIG welding of steel and aluminium, pp 88–98. In: Junek, L. (ed.) Proceedings of the IIW


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

57

conference on benefits of new methods and tends in welding to economy, productivity and quality, 10–15 July, Prague, Czech Republic A new Generation of Lasers for Industrial Applications: Fiber Lasers. http://www.zugo.com.sg Shiner, B.: High power fiber lasers impact material processing. Industrial Laser Solutions, February 2003, pp 9–11. http://ils.pennnet.com Brockmann, R., Havrilla, D.: Third generation of disk lasers lasers enable industrial manufacturing. Laser Tech. J. 6(3), 26–31 (2009) Brockmann, R., Mann, K.: Disk lasers enable industrial manufacturing—What was achieved and what are the limits? Laser Tech. J. 4(3), 50–53 (2007) Dawas, C. (ed.): Laser Welding. McGraw-Hill, NewYork (1992) Shiner, B., IPG Photonics Inc.: Fiber Lasers for Material Processing. LIA Today, April 2004 Grupp, M., Seefeld, T., Vollertsen, F.: BIAS, Bremen, Germany. Industrializing fiber lasers. Industrial Laser Solutions, March 2004 Quintino, L., Costa, A., Miranda, R.M., Yapp, D., Kumar, V., Kong, C.J.: Welding with high power fiber lasers—a preliminary study. Mater. Des. 28(4), 1231–1237 (2007) Costa, A., Miranda, R.M., Quintino, L., Yapp, D.: Analysis of beam material interaction in welding of Ti with fiber lasers. Mater. Manuf. Processes 22(7) 798–803 (2007) Dilthey, U., Olschok, S.: Robotic Fibre-Laser-GMA Hybrid Welding in Shipbuilding, IIW doc. 950-07 Trumpf GmbH ? Co. KG. http://www.de.trumpf.com/produkte/lasertechnik/produkte/ festkoerperlaser/scheibenlaser/trudisk.html. Accessed 10 May 2010 Miranda, R.M., Lopes, G., Quintino, L., Rodrigues, J.P.: Rapid prototyping with high power fiber lasers. Mater. Des. 29, 2072–2075 (2008)

Laser Beam Welding and Automotive Engineering Eva Vaamonde Couso and Joaquín Vázquez Gómez

Abstract Since extensive literature is published on the topic of laser welding in different bibliographic resources, this chapter is intended to present only brief discussions on important aspects of the laser welding process. Taking into account that the automotive industry is the main industrial sector where laser welding process is widely used, the impact of the laser welding process on this industry is analyzed compiling the fundamental applications of this welding process into car manufacturing process. That includes one of the most significant improvements in lightweight structures manufacturing for the automotive industry, the tailor welded blanks. This chapter, besides introducing the main features of the laser welding process, shows the breakthrough that this welding process involved in the automotive industry.

1 Introduction After the first working laser system was built in 1960 by Theodore H Maiman [1], technology advantages for materials processing were early identified, although progresses in this field were subjected to different factors and the use of lasers for welding applications has increased at different rates. The firsts studies about the applicability and uses of lasers for welding were published through the 1960s. Works were mainly focused on laser welding of small components since low power sources

E. Vaamonde Couso (&) J. Vázquez Gómez AIMEN Technology Centre, c/Relva, 27 A. Torneiros, 36410 Porriño, Pontevedra, España, Spain e-mail: [email protected] J. Vázquez Gómez e-mail: [email protected]

Adv Struct Mater (2012) 8: 59–84 DOI: 10.1007/8611_2011_57 Springer-Verlag Berlin Heidelberg 2011 Published Online: 2 November 2011

59

60

E. Vaamonde Couso and J. Vázquez Gómez

were available. Initial works on the laser welding process involved butt and edge joints in thin materials, where joining of small components for the electronics industry was the main application. The first industrial application, around 1965, was to repair broken connectors inside assembled television tubes using a pulsed Nd: YAG laser. Several laser-based joining processes such as soldering and brazing were also developed in response of microelectronic industry. By the 1970s, the development of multi-kilowatt CO2 laser sources favoured deep penetration welding allowing emergence of new industrial applications for laser welding [2]. The most significant growth in the field of laser welding applications has taken place over the past few decades driven by several factors: the development and availability of enabling technologies, which include improvement in process understanding, progress in laser sources and systems, and constant development and progression in process technology for laser beam welding of macro and micro components. With the developments in the high-power laser technology, laser welding became capable of joining thicker sections with higher processing speed and better weld quality. Specifically, advancements in solid state laser technology, that offers the capability for fibre optic beam delivery, have also provided higher flexibility for 3D welding applications. Laser welding has evolved as an important industrial manufacturing process for joining a variety of metallic and non-metallic materials, improving efficiency and reducing costs in a wide range of industries. One of the main examples was automotive industry. Since new features in cars manufacturing are demanded in legislation related to environment (reduction of CO2 emissions and fuel consumption) and safety aspects, automotive industry is enforced to introduce modifications in its manufacturing process. In this sense, manufacturing of lightweight structures for body car parts has represented one of most significant breakthroughs for the automotive industry. Manufacturing of lightweight structures was directly associated with development of new materials, new joints designs and also with new welding process. In 1964 the concept of tailor welded blank was introduced as a new way of lightweight structures manufacturing, but it was not until the 1980s when it reached special significance due to the introduction of the laser process as a welding process [3]. Since then, the use of laser welding for tailor blanks applications has been growing from year to year becoming one of the main laser welding applications into the car industry. Along this chapter, main characteristics of the laser beam welding process are discussed, with special attention to different concepts of the laser welding process and principal advantages and limitations. Different factors such as the influence of welding parameters, the type of materials that can be welded and main defects that can take place in laser weld beads are addressed. In addition, special laser welding process as laser brazing and laser-arc welding are also detailed. At the end of the chapter, laser welding applications into the automotive industry are discussed, focussing on main applications like tailor welded blanks, assembly parts manufacturing and body in white applications. Other applications of the laser welding process in the automotive industry are mentioned like laser-arc welding applications and laser welding of plastic components.

Laser Beam Welding and Automotive Engineering

61

Fig. 1 Representation of conduction welding mode

2 The Laser Beam Welding Process 2.1 Laser Welding Process Laser beam welding is defined as a welding process where melting is produced by the heat obtained from the application of a concentrated coherent light beam impinging the surfaces to be joined. Laser beam can be focused to a small spot on the work piece surface that results in high irradiance whose energy is absorbed by the piece and transformed into heat. Increasing the beam intensity, energy density of the spot is increased and the material is heated up to reach melting point so fusion welding takes place. Although the bulk laser power is essential, interactive mechanisms between the beam and the work piece are mainly dependant on the beam power density (power per unit area). Depending on power density of the laser beam on the joint surface two different welding modes can take place: conduction mode, or melt-in mode, and deep penetration mode or keyhole mode [2–4]. Conduction welding involves absorption of energy from laser beam by the material surface and consequent energy transfer into the surrounding material by conduction mechanisms. Usually this welding mode is related to low energy density of the spot usually lower than 106 W mm-2 that is, low beam intensity or high spot diameter. An illustration of the conduction welding mode is shown in Fig. 1. The macrograph from Fig. 2 shows the geometry of the weld bead and the heat affected zone, both with a hemispheric profile. Weld beads carried out by laser conduction welding are characterized by a low depth to width ratio, which is required in applications with limited penetration conditions. This mode is good for joining thin sections (e.g., 0.025–1.5 mm), for making fine welds at low power, or for joining thicker sections, up to 3 mm, at higher power. In deep penetration welding, the beam is focused to generate a high energy density at the surface of the work piece able to start material vaporization. A narrow and deeply vapour cavity, or keyhole, is formed by multiple internal reflection of the beam. The principle of deep penetration welding is represented in Fig. 3. This keyhole is surrounded by molten material and it is kept stable during the welding process by equilibrium between opening forces arising from

62


Fig. 2 Macrograph of welded joint Fig. 3 Laser welding process: deep penetration mode

material ablation and plasma formation and forces caused by surface tension and hydrostatic pressure of the molten pool. The weld bead is formed moving the beam relative to the work piece. The molten metal surrounding the intensely hot vapour cavity is drawn by surface tension or capillary forces to fill the cavity at the trailing edge of the weld pool. To keep this equilibrium involves practical minimum and maximum traverse rates for keyhole welding, excessive speed causes keyhole collapse whereas low speed results in a wide weld bead with sagging defects. As a result, a narrow weld bead is achieved which is characterized by a high aspect ratio depth–width. In full penetration welds, the heat affected zone (HAZ) is narrow and parallel to weld bead, while in partial penetration welds it resembles a conduction weld. Keyhole welding is excellent for welding applications requiring deep penetration and it is the most used laser welding mode. Figure 4 shows the general profile of a keyhole laser weld and the macrograph of a welded joint.

2.2 Process Advantages and Limitations Laser beam welding has many advantages in relation to other welding processes but also presents some limitations [4–10]. The main advantages and limitations are detailed below. Some key advantages are:


63

Fig. 4 Weld geometry (top) and weld bead macrograph (bottom)

• Main characteristic of laser beam, that is, its high energy focusability makes available precision and clean welds reaching high penetration depth. Availability of higher laser powers also results in high rate between penetration depth and welding speed achieving single pass welds without edge preparation even in thin materials at high welding speeds. • As a consequence of high energy focussing, heat input during laser welding is lower compared with conventional welding processes providing narrow heat affected zones. This condition makes the process suitable for joining heat sensitive components such as electronic devices. Since low heat input is used, heat induced distortion of the workpiece is minimized which involves a reduction of post welding machining. • Laser welding is a non contact welding process and that, coupled to the fact that it is easy to automate, makes the process highly flexible providing higher workpiece accessibility than other welding processes such as e.g. resistance and arc welding processes. In addition, it must be taken into account that one laser can be shared for different applications on several workstations increasing productivity. High welding speed, flexibility and repeatability make the laser welding process suitable for manufacturing applications with high volume production, such as automotive applications. • A wide range of materials and joint configurations can be laser welded, even dissimilar joints. • Versatility is another main advantage of laser welding process due to its suitability for welding large structures as well as micro welds in microelectronic industry. On the other hand laser welding also presents some disadvantages. • Laser equipment and also laser welding workstations require high investment costs. High power laser sources are expensive and safety conditions such as watertight installations to laser radiation increases the final costs. • Although new laser sources present higher energy conversion efficiency compared to old sources, (30 vs. 4% of the diode pumped Nd:YAG lasers) being more efficient, energy consumption is high compared to other welding process.

64


Fig. 5 Variation of penetration depth for different laser power

Influence of laser power on penetration depth Diode pumped Nd:YAG Laser Penetration Depth (mm)

4 3 2 1 0 0

1

2

3

4

5

Laser Power (kW)

• The small focused spot size of laser beam limits the laser welding tolerance to joint positioning. Joint fit up and therefore fixturing, are critical for laser welding applications. In some cases, filler material must be added in order to ensure an adequate weld bead. • The high cooling rate of laser welds related to a high welding speed can lead to the formation of welding defects such as cracking or formation of brittle phases.

2.3 Welding Parameters Suitable laser welding requires optimization of different welding parameters. In addition to the laser type, the laser welding process depends on different parameters that can be classified into laser processing parameters, joint geometry factors and shielding gas conditions. Optimization of these welding parameters identifies the welding condition for each application under certain rules or standards. The main welding parameters affecting the laser welding process are laser power, welding speed and energy density onto the work piece surface. These welding parameters are interconnected, that is, high energy density provides higher welding speeds keeping laser power constant. All parameters have influence on the weld bead geometry and penetration depth. In general, increasing the laser power results in increased penetration depth, as it is observed in Fig. 5, where the bead on plate made of low carbon steel at different laser powers reveals a linear increment of penetration depth behaviour as the laser power increases. An increased penetration can be also achieved decreasing welding speed, but in this case the behaviour is relatively different as it is shown in Fig. 6. Initially (from 100 to 60 mm/s) the penetration depth increases rapidly as the welding speed decreases, nevertheless, at lower welding speeds, the penetration does not increase at the same high rate. In this case, the energy from the laser beam is absorbed by


65

Penetration depth vs. welding speed Diode pumped Nd:YAG Laser

Penetration depth (mm)

6

Penetration depth D/W= Ratio penetration depth - width

5 D/W= 0,43

4

D/W= 0,43

D/W= 0,54

3 2

D/W= 0,43

D/W= 0,40

1 0 0

20

40

60

80

100

120

Welding Speed (mm/s) Fig. 6 Variation of penetration depth as function of welding speed

the molten metal causing fusion of the surrounding material by conduction. As a result, the weld bead is wider and the penetration depth does not increase so much. On the other hand, the energy density is represented by the spot diameter on the work piece. For a constant laser power, the energy density is higher the smaller the beam diameter on the work piece is and consequently the penetration depth increases. In fibre delivered lasers, the spot diameter on the work piece depends on the fibre diameter, the focus length of the optical components of the welding system and the working distance. Laser welding process is very sensitive to joint fit-up, which depends on material preparation, mainly edge finishing of the sheets and gaps between the sheets, and joint fit-up such as the relative position between the laser beam and the joint. If the final joint assembly is not the correct one, even the optimal welding parameters may result in weld beads with various imperfections, affecting to weld process repeatability. Edge quality, mainly dependant on cutting process, has influence on weld bead geometry causing undercut, root concavity, sagging or lack of fusion. Figures 7 and 8 show macrographs of a butt joint carried out with different joint assemblies. The first (Fig. 7) case presents irregular edges, i.e., the cut is not at right angles, and weld bead has defects at the surface, showing undercutting and root concavity. The second case (Fig. 8), where joint preparation is correct, shows a homogeneous weld bead. Positioning of sheets is also a factor to take into account for laser welding. Due to the small spot size of the laser beam, laser welding is characterized by a low tolerance to gaps, especially in butt joints, tailored weld joints or T-joints. For these joint geometries, or similar, when the root opening increases, the melting material decreases and defects as undercut and sagging take place, or even fusion

66


Fig. 7 Left image shows bevelled edges in a butt joint and right figure shows the weld bead instability associated with the irregular cut of the sheets

Fig. 8 Weld bead macrograph of a correct joint assembly

may not occur if gap size is too large. Although laser welding sensitivity to gaps depends on weld speed, beam diameter and beam quality, generally as the material thickness increases, tolerance to gaps also increases and the effect of weld undercutting of the joint surface becomes less pronounced than in thin materials. For materials up to 13 mm thick, the gap tolerance is be up to 3% of the material thickness. Overlap joints show higher tolerance to gaps, even if in some cases introducing a specific gap between the sheets turn out to be beneficial. The main example is overlap welding of galvanized steels for automotive applications. Vaporization temperature of zinc (907C) is lower than that reached in the laser welding process to melt steel (higher than 1400C) so, during laser welding, zinc from joint interface vaporizes and it enters into the molten pool. As solidification rate is very high zinc vapour remains entrapped into the weld beads causing a lack of material in the weld bead or porosity defects. Introducing a gap between both plates, degassing of the evaporated zinc is allowed and porosity does not occurs. Figure 9 illustrates the difference between a weld bead carried out without gap and with


67

Fig. 9 Overlap welds of galvanized steel: Difference between using close fit-up and gap

gap. However, in overlap joints where full penetration is not required, gaps too large may result in no weld defects. Numerous studies show that optimum gap size to avoid porosity defects in overlap joints of galvanized steels is between 0.1 and 0.2 mm [11–14]. Laser beam position regarding to joint alignment is also a critical factor for the laser welding process. It should be taken into account that in laser welding applications the spot diameter of the laser beam on the work piece surface is usually quite small (0.2–0.6 mm) so minimal deviations in the beam positioning can result in welding imperfections such as a lack of penetration or a lack of fusion. In different laser welding applications, especially when sufficiently high power is used, ionization of vapour from work piece surface can result in plasma generation, especially for CO2 lasers. This plasma absorbs the laser radiation and can influence the keyhole stability, so in order to avoid plasma formation a shielding gas is used to remove vapour from beam incident point. Also, for welding active materials such as titanium (in molten state titanium reacts strongly with oxygen causing metal oxidation and consequently loss of material properties), a protection gas is needed to avoid weld bead oxidation. This also applies to applications where high quality weld beads are required such as specific nuclear and aeronautic components. In both cases, the type of gas, the relative shielding position regarding welding direction and torch angle are the main factors affecting the laser welding process. The most common shielding gases used for laser welding applications are argon, helium (mixture He ? Ar), nitrogen and carbon dioxide. Usually, argon is the most used shielding gas due to its high ratio between quality of weld beads and cost. Helium is more expensive than argon so it is used in applications where high quality is needed, mainly for aluminium and titanium welding, although, depending on requirements, a mixture of He ? Ar is also used. Nitrogen is used specially for steel welding in automotive applications. It is also adequate to weld austenitic stainless steel to avoid porosity formation at low welding speeds. Figure 10 shows the variation of penetration depth in function of the shielding gas used for laser welds carried out on steel plates using in all cases the same welding parameters. As it can be observed in the macrographs of Fig. 10, the

68


Penetration Depth (mm)

Shielding gas influence 8 6 4

4,4

4,6

Argon

Helium

4,1 3,1

2 0 Without gas

Nitrogen Gas

Fig. 10 Influence of shielding gas on penetration depth

shielding gas also affects the surface quality. In this case, helium provides higher penetration depth and a smoother surface than the other shielding gases used.

2.4 Materials and Joint Configurations One of the main advantages of the laser welding process is its suitability for welding a wide range of materials, metallic or non metallic, although thermophysical properties of materials will provide a different weld quality. Traditionally, metallic materials are the most studied materials for laser welding. Generally, all conventionally weldable metals are also laser weldable. Steel and aluminium welding are the main applications for the laser welding process but there are also some titanium or magnesium alloys that present good laser weldability. Laser weldability of steels depends, as for other conventional welding processes, on carbon content. Steels with low carbon content, those are the most used in many industrial applications, present good laser weldability. In stainless steels, welding problems can be associated with the changes in the composition of the weld metal due to the large vaporization of the alloying elements from the weld pool. This vaporization can favour formation of intermetallic compounds or certain metallographic phases that influence the corrosion and mechanical properties of stainless steels. Austenitic stainless steels are those that present the highest laser weldability, while cases of martensitic and ferritic stainless steels present restricted weldability requiring in some cases additional processing such as pre or post-welding heat treatments. In recent years, laser welding is also used for welding of aluminum alloys in industries such as aerospace and automotive. Laser welding of aluminium alloys


69

can be difficult because of high reflectivity (that depends on laser wavelength) and high thermal conductivity of aluminium that contribute to the presence of defects on weld beads. Defects on laser welded joints are similar to those founded in welds made by other fusion welding processes. The most common ones are porosity, hot cracking and the problem of drop through [15–17]. These problems affect directly to mechanical properties of weld beads. Porosity can be caused by: – Hydrogen that can be dissolved in the base metal, the filler wire, in the oxide film, or coming from environment humidity; – Instability of the keyhole; – Vaporization of alloying elements of low boiling point. Cracking can take place at the fusion zone or at the heat affected zone. In the first case cracks formation is associated with loss of alloying elements or a coarse solidification microstructure, while in the second case liquation of alloying elements with a lower melting point that aluminium may result in micro-cracks formation. In overlap joints with full penetration, the problem of drop through is caused by the low viscosity of the molten weld metal, which causes imperfections on the root surface of the weld beads. It depends on the sheet thickness being the limit 2 mm. As in other fusion welding processes, for laser welding of aluminium, a shielding gas is needed, being argon, helium or a mixture of both the most common. Titanium alloys, because of their excellent properties such as good ratio strength/weight or excellent corrosion resistance, find wide applications in aerospace, chemical and petrochemical industries. Due to high chemical activity, conventional welding of titanium is often associated with serious weld defects. Pure titanium and most of the alpha alloys are either easily welded, or at least weldable by the laser welding process. Much like other metals, such as aluminium, the cleaning of titanium before welding is extremely important. Any dirt, oil, paint, or any other foreign substance must be removed prior to welding. Moreover, due to the fact that titanium is very reactive with nitrogen and oxygen, almost all welding processes require a trailing shield and the laser welding process too. This shield needs to be large enough to allow the welded joint to cool below 425C. The type of shielding gas used is argon, helium, or a mixture of the two. Laser welding is a non contact welding process that provides high flexibility and so on there is a wide range of opportunities for welding materials in a variety of joint geometries. In the field of metallic materials welding, laser welding process is used for welding different joint configurations such as butt joints, overlap joints, edge joints, T-joints or fillet joints (Fig. 11). In addition to welding parameters, quality of weld beads are influenced mainly by joint fit up and edge preparation. In addition to metallic materials, the laser welding process can be applied to other materials. In the case of polymers welding in an overlap joint geometry, the laser welding process is based on different absorption of the laser beam

70


Fig. 11 Different joint configurations used for laser welding applications

wavelength from polymers to be welded [18–20]. The laser beam penetrates the upper material (that must be transparent to laser beam) and is absorbed by the lower component thus heating up the lower layer directly. This layer transports the heat indirectly via heat expansion and conduction to the upper layer so that both materials are simultaneously heated up and melted. The principle of overlap welding of polymers is shown in Fig. 12. The main advantage of transmission welding is that the weld is inside the component and thus the surface is not harmed and no micro particles are generated. For polymer welding using the transparent-absorbing overlap method, diode lasers (808, 940 nm) and also cw Nd:YAG lasers (1064 nm) are the most suitable lasers. Fusion welding of ceramics presents different problems being the main one cracking, caused by thermal stresses induced in the heat affected zone and the weld bead. Therefore, laser welding for ceramic materials is limited by this problem. Moreover, the rapid solidification during laser processing leads to a sharp thermal gradient into the workpiece enhancing thermal stresses which results in development of extended cracks in the weld zone [21, 22]. Other severe defects, such as porosity and grain growth makes welding ineffective. Most information about laser welding of ceramics is related to zirconia [23] and alumina materials [21, 22, 24–26]. In case of alumina, ceramic was welded by laser process using preheating treatment to reduce thermal shock that minimizes the probability of cracks formation. Laser welding of composites is a relatively new welding process. There is not too much documentation reported on laser welding of composites and works are mainly focused on metal matrix composites [7]. Nevertheless, laser welding of these kind of materials seems to show some problems that may come from different ways:


71

Fig. 12 Laser welding of polymers by heat conduction

• Overheating or dissolution of one component associated to a preferential absorption, resulting in precipitation of detrimental phases; • Cracking caused by differences in thermal properties of components. Most cases of studies on laser welding of composites are focused on welding aluminium matrix composites reinforced with SiC particles or whiskers [27, 28]. Results show that, welding under the optimum conditions, it is possible to avoid excessive and large carbides formation and the amount of defects is greatly reduced. Finally, welding dissimilar materials represents a wide field of work for laser welding since several configurations may be possible. The main problem for laser welding of dissimilar materials lies on the difference of physical properties of the materials to be welded, mainly thermal conductivity. This difference may generate cracking due to different cooling rates of materials or high loss of mechanical properties due to formation of intermetallic compounds or brittle metallurgical phases. The addition of filler materials, with laser brazing or laser soldering, can improve dissimilar joints weldability [29–31].

2.5 Weld Imperfections As it was commented above, different factors have influence on the laser welding process and, therefore, the quality of welded joints. During laser welding of metallic materials different types of defects can appear during laser welding [32], being the main ones listed in Table 1. To detect these weld bead imperfections, non destructive testing must be performed and results will define the weld bead quality. There are international standards that define quality levels for different laser welding imperfections such as ISO 13919-1:1996, ‘‘Welding. Electron and laser beam welded joints. Guidance on quality levels for imperfections. Part1: Steel’’ and ISO 13919-2:2001, ‘‘Welding. Electron and laser beam welded joints. Guidance on quality levels for imperfections. Part 2: Aluminium and its weldable alloys’’. These standards define acceptance limits for each imperfection as a function of its size and repeatability, resulting in three quality levels designed as a function of permissiveness as Moderate (D), Intermediate (C) and Stringent (B) levels.

72 Table 1 Main defects of laser weld beads Defect


Scheme

Cracking

Porosity

Lack of penetration

Lack of fusion

Misalignment

Undercut/Sagging

Reinforcement/Root concavity

Surface crater

In addition, destructive tests are needed in order to determine the presence of metallurgical defects (metallographic tests, micro-hardness) and to evaluate mechanical properties of weld beads (bending tests, tensile tests, fatigue tests, etc.). In this case, the final application and, therefore, end users enforce the final requirements of the laser welds.

2.6 Laser Welding Processes Using Filler Materials Although, by definition, laser welding is a process that does not need filler materials, and autogenous laser welding is used for most applications, there are welding applications or materials where addition of filler materials is required. In this field, two main welding processes become relevant, laser brazing and laser-arc welding (hybrid welding), both detailed below. 2.6.1 Laser Brazing Brazing is a method of joining materials by heating them in the presence of a filler material that has a melting point lower than that of the base materials. The heating process may be carried out with different heating sources. When a laser beam is used as the heat source, the process is known as laser brazing. Laser welding


73

processes such as laser brazing and laser soldering are considered as conduction welding process. Both processes involve addition of filler material that is melted by the laser beam. During the heating cycle, laser beam melts the filler material which fills the joint by capillary action bonding the materials to be joined by wetting action [33–35]. The welding process takes place without fusion of the base material. Laser brazing is different from soldering because of the melting point of the filler material. Usually, filler materials with a melting point under 450C are used for soldering (tin–lead alloys) while filler materials used for the brazing process have higher melting point (copper, silver, nickel alloys). In this case, laser brazing does not necessarily require the laser energy to be focused into a small diameter spot, as is the case in laser cutting or keyhole mode welding. Therefore, usually a laser source with low beam quality (that means high spot diameter in the workpiece, usually higher than 1 mm) is used for this welding application, e.g. diode lasers, which involve a lower capital investment in the laser source regarding other laser sources with high beam quality, such as disc or fibre lasers. These laser sources can be also used for laser brazing applications, although in these cases, as low power densities are required, the laser beam is defocused in order to ensure filler material melting and to avoid fusion of the base material. Compared to other brazing processes, laser brazing allows welding materials that are difficult to braze or to use filler materials with problematic compositions. One of the main advantages of the laser brazing process is the capability of heat focussing in a small area, which avoids heating the materials to be joined that acquires significance for thermal sensitive materials. The macrograph of Fig. 13 shows the weld bead profile of an edge joint of galvanized steel 1 mm thick used in automotive applications. A diode laser was used and the filler material was Cu–Si wire of 1.2 mm diameter. As it is observed in the macrograph, the welding process provides a small heat affected zone on both sheets which means minimal thermal damage in the base material.

2.6.2 Laser-Arc Welding Laser arc welding process, or hybrid laser-arc welding, consists of matching an electric arc (MAG, TIG or plasma) with the laser keyhole or the laser-material interaction zone produced during laser welding (CO2 or solid state laser). Initial studies about this process were carried out in the 1970s [36], but interest in this welding process raised with development of high power lasers. The laser arc welding process combines advantages from laser and arc welding, increasing welding productivity regarding both processes. As a result, high penetration depth at high welding speed provided by laser welding is combined with the high tolerance to joint fit-up of the arc welding process [37–41]. The main advantages of laser-arc welding regarding to arc welding processes are: • Increment of productivity due to higher welding speed; • Higher penetration depth even without edge joint preparation;

74


Fig. 13 Weld bead geometry of a laser brazed joint (edge joint)

• Lower heat input which means a reduction of thermal distortions into the workpiece and reduction on rework activities; • Reduction on filler material consumption. On the other hand, hybrid welding also provides advantages regarding to laser welding process such as: • Reduction of edge positioning accuracy; • Low metallurgical defects associated to higher cooling rates that avoid formation of brittle phases; • Control of metallurgical variables through the addition of filler wire. By contrast, laser-arc welding also involves some disadvantages such as: • The process set-up is complex due to a wide range of welding parameters must be controlled, which includes parameters of each welding process and those that result from process combination, such as laser-arc distance or relative position regarding welding direction. • The profile of hybrid weld beads is wide at top surface and narrow at the root, so defects like lack of fusion are probable. Figure 14 shows two macrographs of overlap joints welded with laser-MAG process with and without full penetration.

3 Laser Welding in the Automotive Industry Last years trend in the automotive industry is orientated to development and application of technology in order to achieve weight and cost reductions, minimising the energy consumption and environmental impact of vehicles. The use of new and lighter materials like advanced high strength steels and aluminium or


75

Fig. 14 Laser-arc welds with full penetration (left) and partial penetration (right)

magnesium alloys, high productivity joining processes, such as laser welding, and new design concepts are the main manufacturing methods that have been applied to achieve those goals. Regarding the laser welding process, the automotive industry was one of the first to introduce this process in production initially for welding small parts. Advantages displayed by this welding process promptly aroused the interest of automakers on the laser welding process leading to the development of new applications. Over the years, the industry has remained as the most important user of the laser technology and as such, it has been one of the driving forces behind the development of new industrial laser sources and innovative laser applications. In conjunction with cutting, welding and marking, lasers offer considerable benefits compared to conventional techniques, but improved flexibility, high production rate and ease of automation were the central requirements of this sector. There are numerous and diverse applications of laser welding into the automotive industry, being the main ones: • Multi-thickness welded blanks; – Tailored welded blanks: Blanks of relatively simple geometry with linear weld seams or welded blanks of complex shape with non-linear weld seams, for high productivity, weight optimization and resistance improvements; – Welded patchwork blanks for components requiring local reinforcement. • Car assembly parts, • Body in white elements, • Mechanical elements.

3.1 Tailored Welded Blanks One of the main changes in the automobile manufacturing process was reached with the introduction of tailor welded blanks (TWB) for car parts manufacturing.

76


Fig. 15 Schematic view of TWB joints

Tailor blanks are composed of two or more dissimilar sheet metals, with different thicknesses, shapes, strengths, or materials that are butt-welded together before being formed. Figure 15 shows the joint configuration of a tailor welded blank. Initially, TWBs were used mainly to reduce material consumption and cost. Thus, for example, pieces of stamping scrap can be jointed together, stamped again and reused as a new part for reducing material consumption. The classic manufacturing sequence, i.e., first form and then join, is inverted by this innovative product. While in the conventional parts manufacturing process, two or more stamped parts are spot-welded together to form a part, in the TWB stamping process, sheets are first welded together, and then integrally stamped into a part. In this way, the numbers of parts, welding steps, and press dies are reduced, and costs are reduced accordingly. This manufacturing process allows the use of thicker or stronger materials in the critical regions of a component, so as to increase the local stiffness, while thinner or lighter materials are used in other regions to reduce the component overall weight [42–45]. In the last years, the use of tailor welded blanks has increased in the automotive industry due to the need of obtaining lighter automotive vehicles keeping or improving safety issues. In addition, other aspects that have contributed to this increment are cost reduction or reduction of the number of parts without compromising the strength or stiffness of the final part. Although different welding process have been used in the past for tailor welded blanks manufacturing, such as Tungsten Inert Gas (TIG) welding process, Metal Inert/Active Gas (MIG/MAG) process or mash seam welding, currently, most of the tailor blanks are welded using the laser welding process, especially since high power lasers are commercially available. Common laser sources for this application are Nd:YAG, (lately replaced by disc and fibre lasers) and CO2 lasers. Laser welding is considered especially suitable, because of its low heat-input, flexibility through optical fibre


77

Fig. 16 TWB laser welded

delivery, ease of automation and, in particular, its fast processing speeds suitable for a high-volume production environment like automotive production. The use of laser welding to make up a single blank from an appropriate patchwork of different grades and thicknesses of sheet steel prior to pressing was developed by Thyssen Stahl AG in 1985 [46]. The application consisted of welding two sheets of galvanized steel to form a blank for the floor pan of an Audi car model. Although the use of tailor welded blanks at a mass production level was started in Europe, thereafter the use of this technology expanded rapidly all over the world. At present, there are many applications of the tailor-welded blanks for automotive components. Some examples are the A, B and C pillars situated at both sides of the doors, inner door panels, longitudinals, cross rail bumpers, floor panels, wheel housings, inner panel tail gates, etc. Concerning materials, there has been increasing use of the welding together aluminium, magnesium and high strength steel blanks to produce tailor-welded blanks in order to obtain the maximum weight and strength advantage. Figure 16 shows the macrograph of a tailor blank joint welded by laser. In this case, the joint consist of a Dual Phase Steel DP780 (high strength steel) of 1.7 mm thickness and steel E275 of 1 mm thickness. An alternative to the concept of tailored blanks is the so called patchwork blank technique. The principle of patchwork blanks is to add a flat piece of sheet metal onto the main blank in the areas where reinforcements are required, in an overlap joint, Fig. 17. The concept of tailored blank is extended in the sense that, even the smallest of areas can be reinforced easily. Main welding processes for these applications are resistance spot welding and the laser welding. As an example, a macrograph of a laser welded patchwork is shown in Fig. 18. Ferritic Steel Grade ES of 0.7 mm thickness was used as base material and a high strength steel (DOCOL 1200) was used as reinforcement. The main applications of patchwork blanks are: hood inner panels, hinges, gas shocks, locks, etc.

78


Fig. 17 General patchwork joint configuration

Fig. 18 Weld bead geometry of the patchwork joint ES/ DOCOL1200 welded by laser

3.2 Assembly Parts Traditionally, car assembly parts such as roof, door reinforcements, or seat backs were welded by resistance spot welding, MAG or MIG brazing processes. Although the use of laser welding technology for these applications has been undergoing a significant increase, particularly since the emergence of high power solid state lasers, the main achievement has been reached with the concept of remote laser welding. In the remote laser welding process the laser beam is focused over the workpiece from a distance of about 0.5 m or more. The combination of mirrors and mechanical movement of the laser delivery system results in very fast beam positioning, in fact, weld-to-weld repositioning may be in the order of milliseconds. This is more efficient than traditional spot welding or even


79

than robotic laser welding due to the time required to move the robot from one weld to another is almost removed [47, 48]. Remote laser welding takes advantage of the technical and economic benefits of single-sided, non-contact laser welding and combines them with the benefits associated with high-speed scanning optics, which considerably reduces weld times to increase overall productivity in the welding process. Initial laser remote welding systems worked with a CO2 using a scanning mirror that deflects the beam, at very high speeds, to multiple welding locations on a body component. The main limitation in this case is that for complex geometries the workpiece must be reoriented so positioning systems can be complex and in some cases, accessibility can be limited. High beam quality of disk and fiber lasers along with the development of galvanometric optics (scanner systems) made possible to take advantage of CO2 systems, since solid state lasers can be used with robot systems providing greater flexibility for welding operations, leading to the so called welding in the fly. Thus, robot movement is combined with movement of the scanner mirrors increasing accessibility for welding complex components and also decreasing the time needed to proceed from the end of one welded joint to the beginning of the next. By contrast, it must be taken into account that remote laser welding requires accessibility in the sense that the path of the laser beam must be unblocked, and that has influence on fixturing and external clamping design. In addition, the process requiring shielding gas involves incorporating delivery nozzles into the welding head. Remote laser welding is faster than resistance welding and has lower operational costs to higher volume production, making it suitable for those applications requiring a large amount of spot welds in one assembly. For instance, a remote laser welding system can perform 46 welds on a rear seat back assembly in 12 s [49]. The most common applications are door assemblies, instrument panels, seat backs, and side impact structures.

3.3 Body in White Elements Remote laser welding is also used for body in white welding applications, that is, for welding different elements to form the final car structure (Chassis) [50–54]. Initially, for these applications resistance spot welding was the most used welding process. Taking into account the complexity of pieces to be welded and that the first high power lasers were CO2 lasers with limitations on 3D applications, it was not until the emergence of high power solid state lasers over the 1990s, that laser welding of body in white components interested the main car makers. Roof to body side welding is one of the most important applications of laser welding in the manufacturing of body in white. At first, autogenous laser welding to attach the roof to body sides was performed using CO2 lasers, but lately autogenous laser has been replaced by laser brazing for this welding application [55, 56]. Smooth

80


geometry of laser brazed weld beads makes laser brazing suitable for applications where stringent visual requirements are needed, but not mechanical requirements. Therefore, the connection between the upper portion of the A-pillar and the front header and trunk lid components are also laser-brazed. Among the advantages and opportunities offered by BIW (Body In White) laser welding are: significant product improvements such as increases in vehicle stiffness, design flexibility, dimensional control, weight reductions and also improvements in productivity and cost reduction. Compared to the Resistance Spot Welding (RSW) process, the economic feasibility of laser welding has been proved by the continuously growing number of BIW laser welding applications.

3.4 Hybrid Welding Process Applications One of the first applications of the laser hybrid welding in automotive industry was for welding the doors of the Volkswagen Phaeton, where 48 joints 3,570 mm long were welded with this process [57]. Laser-hybrid welding was used for welding the extruded sections, castings, and sheets made from aluminum mainly in two joint configurations, lap joints in fillet welds and butt joints. The hybrid welding process was selected in order to avoid using heavy casting material. In addition, doors requirements involved the use of GMAW (Gas Metal Arc Welding) and laser welding process autogenously, so that it can be performed with the same welding installation, switching off one of the two processes and optimizing welding parameters in each case. Audi also has used the laser hybrid system for welding the lateral roof frame of the Audi A8 model, comprising a total of 4.5 m of welded joints [58]. Features of laser hybrid welding such as high welding speed, low heat input, and good penetration also aroused a great interest in Mercedes Benz for welding the rear subframe. Investigations on hybrid welding process for this application revealed an increase of welding speed by a factor of 2.5 and a heat input reduction of 25% with regarding to GMAW (Gas Metal Arc Welding) process [59].

3.5 Laser Welding Applications on Plastic Parts Lastly, in the automotive industry, the use of moulded plastics is rapidly increasing in different car components. Although most of these components seem to be simple components, those like headlights and tail-lights [60], can have complex geometrical shapes, and this geometrical complexity may be a problem for the lighting unit assembly. Laser welding process is then suitable for these components due to its high flexibility to access zones that are not accessible with conventional plastic welding processes like ultrasonic welding. As it was discussed in a previous section, the main requirement for laser welding of plastic materials in


81

an overlap configuration is that the upper material must be transparent to laser radiation while the lower must be absorbent. One of the most well-known applications of the laser welding process in the automotive industry for plastic components is car keys, that are common today [61]. Other applications of laser welding for plastic components in the automotive industry are: sensor housings, underhood components or instrument panels. Attaching ancillary components within the vehicle cabin is also another application for this welding process. Laser welding of plastics also favours incorporation of modern components, such as hands-free mobile phone microphones, that can be fitted even upon the seatbelt.

4 Conclusions Along this chapter laser welding process characteristics have been commented taking into account the influence of different factors. The advantages of laser welding process such as high productivity related to high welding speeds and high process flexibility, especially for solid state lasers, confirmed the suitability of this welding process for a high-volume production environment like automotive production. In automotive industry, laser welding has been increasingly used, not only for producing tailor welded blanks, that is one of the main applications in this industrial sector, but also as an alternative to resistance spot welding in BIW assembly. This continuous welding process provides structures with increased rigidity, allows thickness reductions and, therefore, greater potential for lightweight car body designs. So, laser welding process seems to be a well established process into the automotive industry. Nevertheless, taking into account that changes in automotive industry are so frequent in terms of new car designs, new joint geometries and/or the use of advanced materials, new challenges for laser welding process will emerge.

References 1. Maiman, T.H.: Stimulated optical radiation in ruby. Nature 187(4736), 493–494 (1960) 2. Ion, J.C.: Laser Processing of Engineering Materials: Principles, Procedure and Industrial Application. Elsevier Butterworth-Heinemann, Oxford (2005) 3. Duley, W.W.: Laser Welding. Wiley Interscience, Toronto (1999) 4. Kannatey-Asibu, E.: Principles of Laser Material Processing. Wiley, New Jersey (2009) 5. Meijer, J.: Laser beam machining (LBM), state of the art and new opportunities. J. Mater. Process. Technol. 149, 2–17 (June 2004) 6. Dutta Majumdar, J.I., Manna, I.: Laser processing of materials. Sadhana. 28(3&4), 495–562 (2003) 7. Dahore, N.B., Harimkar, S.P.: Laser Fabrication and Machining of Materials, pp. 412–447. Springer, London (2008)

82


8. Welding Handbook, Volume 3 Welding Processes, Part 2, American Welding Society, pp. 504–530 9. Steen, W.M., Mazumfer, J.: Laser Material Processing, pp. 199–250. Springer, London (2010) 10. Schlueter, H.: Laser beam welding: benefits, strategies and applications. Weld. J. 86(5), 37–39 (2007) 11. Wilden, J., Bergmann, J.P., Dolles, M., Reich, S.: An innovative joining strategy in order to join zinc coated steels with minimized damaging of the coating, 24th International Congress on Applications of Lasers & Electro–Optics, ICALEO, paper P550, Oct 31–Nov 3, Miami (2005) 12. Gualini, M.M.S.: Laser welding of zinc coated steel sheets. An old problem with a possible solution, 20th International Congress on Applications of Lasers & Electro–Optics, ICALEO, paper P511, Jacksonville–Florida, 15–18 Oct (2001) 13. Mäkikangas, J., Mäntyjärvi, K., Keskitalo, M., Karjalainen, J.A., Niemela, J., Ojala, J.: Laser welding of coated sheet metal constructions, 11th NOLAMP Conference in Laser Processing of Materials, Lappeenranta (2007) 14. Pan, Y., Richardson, I.M.: Study of laser welding of zinc coated steel sheets in overlap configuration, 11th NOLAMP Conference in Laser Processing of Materials, Lappeenranta (2007) 15. Vázquez, J., Vandewynckèle, A., Vaamonde, E., Gesto, D.: Welding of aerospace aluminium alloys using FSW and LBW, 27th International Congress on Applications of Lasers & Electro–Optics, ICALEO, paper 1803, Temecula, 20–23 Oct (2008) 16. Vandewynckèle, A., Vaamonde, E., Guin, A.: Evaluación de procesos de soldeo por láser orientados a minimizar la formación de macroporosidad en uniones de aluminio soldadas a transparencia, Actas del IV Taller Nacional–Procesado Materiales con Láser, Valencia, Octubre (2007) 17. Zhao, H., White, D.R., DebRoy, T.: Current issues and problems in laser welding of automotive aluminium alloys. Int. Mater. Rev. 44(6), 238–266 (1999) 18. Troughton, M.J.: Handbook of Plastics Joining: A Practical Guide. William Andrew Inc., New York (2008) 19. Russek, U.A.: Laser beam welding of polymers with high power diode lasers, joining innovation for micro and macro technologies, 20th International Congress on Applications of Lasers & Electro–Optics, ICALEO, paper P509, Oct 15–18, Jacksonville–Florida (2001) 20. Ilie, M., Kneip, J.C., Mattei, S., Nichici, A.: Through transmission welding of polymers: effects of particles on laser beam scattering. 25th International Congress on Applications of Lasers & Electro–Optics, ICALEO, paper 601, Oct 30–Nov 2, Scottsdale, Arizona (2006) 21. Riviere, C., Robin, M., Fantozzi, G.: Comparison between two techniques in laser welding of ceramics. Journal de Physique IV, Vol. 4, Avril (1994) 22. Exner, H., Nagel, A.M.: Laser welding of functional and constructional ceramics for microelectronics, Laserinstitut Mittelsachsen e.V., Mittweida–Germany 23. Maruo, H., Miyamoto, I., Arata, Y.: Laser welding of sintered zirconia: CO2 laser welding of ceramics III. Q. J. Jpn. Weld. Soc. 4, 470–476 (1986) 24. Harris, J., Akarapu, R., Segall, E.: Welding of alumina using a pulsed dual beam CO2 laser. J. Manuf. Sci. Eng. (Feb 2011) 25. Tomie, M., Abe, N., Noguchi, S., Arata, Y., Oda, T.: Weld bead and joint strength characteristics during laser welding of 87% AI2O3 ceramics–High-power CO2 laser welding of 87% Al2O3 ceramics (1st report). Weld. J. 9, 615–620 (1995) 26. Noguchi, S., Abe, N.: High power CO2 laser welding of alumina ceramics, Reports of Industrial Technology Research Institute (1999) 27. Niu, J., Pan, L., Wang, M., Fu, C., Meng, X.: Research on laser welding of aluminum matrix composite SiCw/6061. Vacuum 80, 1396–1399 (2006) 28. Durmus, H.: Weldability of AL99—SiC composites by CO2 laser welding. J. Compos. Mater. 43, 1435–1450 (2009) 29. Mathieu, A., Matteï, S., Viala, J.C., Grevey, D.: Laser braze welding using hot (88%-aluminum, 12%-Silicon) filler material to join steel with aluminium, 24th International Congress on Applications of Lasers & Electro–Optics, ICALEO, paper 2005, Oct 31–Nov 3, Miami (2005)


83

30. Vandewynckèle, A., Romero, P., Arias, J.L., Vázquez, J.: Laser welding techniques and hybrid laser-plasma welding for dissimilar materials, Minutes of II National Workshops Materials processing by laser, Paterna (Valencia), (2005) 31. Engelbrecht, L., Meier, O., Ostendorf, A., Haferkamp, H.: Laser beam brazing of steel aluminium tailored hybrid in automotive industries, 25th International Congress on Applications of Lasers & Electro–Optics, ICALEO, paper 1702, Oct 30–Nov 2, Scottsdale, Arizona (2006) 32. Norman, P., Karlsson, J., Kaplan, A.F.H.: Monitoring undercut, blowouts and root sagging during laser beam welding, Proceedings of the Fifth International WLT-Conference on Lasers in Manufacturing 2009, Munich, (2009) 33. Hoffmann, P., Kugler, P., Schwab, J.: Laser brazing with high power solid state laser systems and applications in automotive industry, Proceedings of the Second International WLT Conference on Lasers in Manufacturing 2003, Munich (2003) 34. Larsson, J.K.: Laser Brazing–A new technology for cosmetic joints in the body structure, 11th NOLAMP Conference in Laser Processing of Materials, Lappeenranta (2007) 35. Wirth, A., Laukant, H., Thomy, C., Glatzel, U., Vollertsen, F.: Laser brazing or high strength steels, 26th International Congress on Applications of Lasers & Electro–Optics, ICALEO, paper 1003, Oct 29–Nov 1, Orlando (2007) 36. Eboo, M., Steen, W.M., Clark, J.: Arc-augmented laser welding. Proceedings of 4th International Conference On advances in Welding Processes, 257–265, United Kingdom, 9–11 May (1978) 37. Olsen, F.O.: Hybrid Laser-Arc Welding. Woodhead Publishing Limited, Cambridge (2009) 38. Petring, D., Fuhrmann, C., Wolf, N., Poprawe, R.: Investigations and applications of laser-arc hybrid welding from thin sheets up to heavy sections components, 22nd International Congress on Applications of Lasers & Electro–Optics, ICALEO, Oct 13–16, Jacksonville, Florida (2003) 39. Downs, D.L., Mulligan, S.J.: Hybrid CO2 laser-MAG welding of carbon steel- a literature review and initial study, TWI (2002) 40. Wouters, M.: Hybrid laser MIG welding: An investigation of geometrical considerations. Lulea University of Technology, Sweden (2005) 41. Vandewynckèle, A., Vaamonde, E., Arias, J.L., Pérez, M., Quintáns, G.: Laser-Arc welding of duplex stainless steel, 26th International Congress on Applications of Lasers & Electro– Optics, ICALEO, paper 603, Oct 29–Nov 1, Orlando (2007) 42. Arias, R., Prada, A., Vaamonde, E., Vandewynckèle, A., Gutierrez, D., Lara, A., García, M.: Laser welding applied to advanced high strength steels for automotive applications, 29th International Congress on Applications of Lasers & Electro–Optics, ICALEO, paper 1002, Sept 26–30, Anaheim (2010) 43. Naeem, M., Jesset, R.: Welding aluminum tailored blanks with Nd:YAG lasers for automotive applications 44. Ono, M., Yoshitake, A., Ohmura, M.: Laser weldability of high strength steel sheets in fabrication of tailor welded blanks, NKK Technical Review, No. 86 (2002) 45. Assunçao, E., Quintino, L., Miranda, R.: Comparative study of laser welding in Taylor blanks for the automotive industry. Indus. J. Adv. Manufact. Technol. Oct 2009 46. Spoettl, M.: High performance laser welding systems for the production of innovative laser welded automotive components, ThyssenKrupp Lasertechnik GmbH 47. Busuttil, P.: Remote laser welding. Indus. Laser. Solut. (Sept 2009) 48. Bembenek, M.: Welding from a distance, Indus. Laser. Solut. (Mar 2006) 49. Mueller, R.: Getting close to remote laser welding, www.thefabricator.com (2009) 50. Oikawa, M., Minamida, K., Kumehara, H.: Development of the laser spot welded stainless steel panel for vehicle body, 25th International Congress on Applications of Lasers & Electro–Optics, ICALEO, paper 1706, Oct 30–Nov 2, Scottsdale, Arizona (2006) 51. Tarui, T., Hasegawa, T., Mori, K.: Latest laser welding applications for Nissan Body in White, Nissan Motor Co Ldt

84


52. Ribolla, A., Damoulis, G.L., Batalha, G.F.: The use of Nd:YAG laser weld for large scale volume assembly of body in white. J. Mater. Process. Technol. 164–165, 1120–1127 (2005) 53. Park, H.S., Choi, H.W.: Development of digital laser welding system for car side panels, University of Ulsan, South Korea 54. Drexler, F.: Roof laser welding, European Automotive Laser Applications, EALA 2007, Bad Nauheim (2007) 55. Diguet, A.: Latest laser application at Renault: flesible roof brazing station, European Automotive Laser Applications, EALA 2009, Bad Nauheim, 3–4 Feb (2009) 56. Bailly, N., Cretteur, L., Koltsov, A., Kelley, S.: Laser brazing ability of different products for automotive applications, Proceedings of Advanced Laser Applications Conference & Exposition, ALAC, Minneapolis Oct 1–3, (2008) 57. Staufer, H.: Laser hybrid welding in the automotive industry. Weld. J. 86(10), 36–40 (2007) 58. Staufer, H., Rührnößl, M., Miessbacher, G. Laser hybrid welding and laser brazing: State of the art in technology and practice by the examples of the Audi A8 and VW-Phaeton, Fronius Internacional GMBH, Wels, Austria 59. Steinmetz, H., Höfer, T.: Laser hybrid welding of safety relevant chassis structures, European Automotive Laser Applications, EALA 2009, Bad Nauheim, 3–4 Feb (2009) 60. Chang, I.S.: Plastic laser welding for automotive lamps at Hyundai, European Automotive Laser Applications, EALA 2011, Bad Nauheim, 9–10 Feb (2011) 61. Flowers, S.T.: The growth of laser welding of plastics. Indus. Laser. Solut. (2011)

Friction Stir Welding Technology Pedro Vilaça and Wayne Thomas

Abstract The friction stir welding (FSW) process was invented in 1991 by Wayne Thomas et al., one of the authors of this chapter. This machine tool based process is currently considered an important development in welding technology, saving costs and weight for a steadily expanding range of applications of lightweight metallic structures. Evidences of the disruptive character of the FSW process are the prompt adoption by world-wide industry of the significant advantages of FSW and the numerous technic-scientific papers and patents published. The FSW technology has been subjected to the most demanding quality standard requirements and used in challenging industrial applications over a wide range of structural and non-structural components. In this chapter, some of the basic fundamentals underpinning the invention of FSW technology are presented with emphasis for the concept of the third-body region. The state-of-the-art concerning tooling in FSW for conventional and bobbin stir welding approaches are introduced. The nondestructive testing assessment of the most relevant imperfections in FSW is also discussed for butt and lap joints. In summary, the FSW is a key joining technology for lightweight metallic structures. The international organization for standardization standard for welding aluminium alloys by FSW is available and the most recent European standards for design of structures—Eurocodes, already include guidelines for the application FSW process.

P. Vilaça (&) Instituto Superior Técnico, Technical University of Lisbon, Avenida Rovisco Pais, 1049-001 Lisbon, Portugal e-mail: [email protected] W. Thomas The Welding Institute, Granta Park, Great Abington, Cambridge CB21 6AL, UK e-mail: [email protected]

Adv Struct Mater (2012) 8: 85–124 DOI: 10.1007/8611_2011_56 Springer-Verlag Berlin Heidelberg 2011 Published Online: 10 April 2011

85

86

P. Vilaça and W. Thomas

Fig. 1 FSW within some of the most relevant friction based mechanical processing technologies

1 Introduction Friction stir welding (FSW) process was invented by Wayne Thomas et al. and patented by The Welding Institute (TWI) on 6th December 1991 [1]. This patent has since become one of the most widely referenced documents in welding and joining technology. The FSW process has become one major milestone in the welding technology history, especially in the joining of lightweight metallic structures. TWI’s first patent describes the process in Claim 1 as: ‘‘A method of joining workpieces defining a joint region there between, the method comprising carrying out the following steps without causing relative bodily movement between the workpieces: causing a probe of material harder than the workpiece material to enter the joint region and opposed portions of the workpieces on either side of the joint region while causing relative cyclic movement between the probe and the workpieces whereby frictional heat is generated to cause the opposed portions to take up a plasticised condition; removing the probe; and allowing the plasticised portions to solidify and join the workpieces together’’ [1]. The earliest reference to the use of frictional heat for solid-phase welding and forming appeared over a century ago in a United States patent [2]. A period of about 50 years then passed before any significant advancement in friction technology took place, namely a British patent in 1941 that introduced what is known as friction surfacing [3]. FSW derived from continuing research on friction processes and, more directly, after theoretical studies of friction surfacing [4–6]. A further 50 years passed before FSW was invented [1]. Figure 1 presents the FSW among some of the most relevant friction based mechanical processing technologies.

Friction Stir Welding Technology

87

Fig. 2 Identification of friction stir based mechanical processing technologies

Fig. 3 Identification of FSW variants

The disruptive nature of FSW technology has led to many developments from the research and technological development (RTD) institutions and industry. Since the original patent has been issued, many other friction stir based technologies and FSW variants have been and are still being developed [7–12]. Some of these developments are identified in Figs. 2 and 3. For the manufacturing industry, FSW technology represents an opportunity for innovation; consequently it has had the effect of spawning many patent applications from early adopters of the process. Patents are often filed to protect the competitive advantage gained by being an early adopter for a given application. Over 2,400 patent filings exist, although not all have succeeded in obtaining a patent. The filings made cover the application of FSW to particular circumstances, new tooling ideas, products made by FSW, and improvements to the process. As of April 2008 [13], TWI had issued 200 FSW licences to a range of organizations including end users, equipment suppliers, academia and RTD institutes. The growth in the number of licenses issued is shown in Fig. 4. Friction stir welding has been applied to many lightweight metals and alloys, for example aluminium [14, 15], copper and zinc [16, 17], magnesium [18], and titanium [19], with excellent results once the operational parameters have been optimised. Applications to non lightweight metals, e.g. steel and stainless steel have also shown to be feasible [7, 20], but are outside the scope of this chapter. The FSW process enables significant developments in mechanical structural design, where integral structures can now be produced integrating new and already existent advanced engineering materials. Moreover, the FS welded structures can attain very appreciated spatial distribution of properties, unfeasible before its invention.

88


Fig. 4 Cumulative number of licences issued with time [13]

Joining does not usually involve the use of any filler metal and therefore any lightweight metal or alloy of similar composition can be joined without concern for the compatibility and composition of the weld metal, which would be an issue in fusion welding. In addition, the hydrogen contents of FSW joints tend to be low, which is important in welding alloys susceptible to hydrogen damage. When desirable, certain dissimilar lightweight metal alloys and metal matrix composites can also be joined. The principal advantages of FSW, essentially a solid-phase process, are low distortion [21], absence of melt-related imperfections and high joint strength, even in those alloys that are considered non-weldable by conventional fusion techniques. Although incipient melting during welding has been observed for certain materials, FSW is regarded as a solid-phase autogenous keyhole joining technique. The bulk temperature remains below the melting point of the major phases in the material. However, certain alloys can form minor phases which melt significantly below the bulk solidus of the material. The forging operation and thermal management during manufacture of wrought products and a similar forging action during FSW usually eliminates any melt related flaws. An example of lower melting constituent parts can be made with conventional friction welding of carbon steels inoculated with lead or stainless steel inoculated with selenium to improve the machining operation. During the FS processing the low melting points can smear between the flaying surfaces and make the friction welding of these materials extremely difficult if not impossible. FSW has matured to the point where it is used in applications such as commercial and military aircraft where welding has never before been allowed. The ever growing list of FSW users includes Boeing, Airbus, Eclipse, BEA, Lockheed Martin, NASA, US Navy, Mitsubishi, Kawasaki, Fokker as well as other industrial concerns throughout the world in transport structural applications [13]. The uptake of FSW by industry has been very rapid, especially when compared with other innovations in materials joining. By 1995 FSW was in industrial production fabricating panels for refrigeration purposes. Many other applications rapidly followed world-wide including its use in deck structures for ships, railway


89

carriages, airframes and secondary structures, heat sinks, nuclear waste containment vessels, launch aerospace vehicle fuel tanks, wheels, automotive body parts, electrical distribution assemblies and small pressure vessels [7]. It has also been employed to make large-scale blanks for subsequent machining or forming, when stock material sizes are too small for the application. Although still regarded as a novel process, FSW has made a significant contribution to the joining of lightweight metallic components and has become the first choice welding technique for certain applications. This chapter starts by presenting the process fundaments and basic nomenclature addressing the mechanical and geometric parameters and metallurgical properties of the FS welds. Emphasis will be put on the concept behind FSW technology: transient ‘third-body region’, process asymmetry, coupled thermal-structural-metallurgical character of the process, heat efficiency and tests for assessing the mechanical performance of the friction stir welds. The basics of joint design are introduced. Moreover, various tool types that can be used in FSW for conventional and bobbin stir welding approaches are described. The NDT assessment of the most relevant imperfections in FSW is also included for butt and lap joints.

2 Process Fundaments 2.1 Fundaments 2.1.1 Basics and Nomenclature FSW is a process for joining workpieces in the solid-phase, using an intermediate non-consumable tool, with a suitably profiled shoulder and probe, made of material that is harder than the workpiece material being welded. FSW can be regarded as an autogenous keyhole joining technique, essentially, without the creation of liquid metal. The rotating tool is plunged into the weld joint and forced to traverse along the joint line, heating the abutting components by interfacial and internal friction, thus producing a weld joint by extruding, forging and stirring the materials from the workpieces in the vicinity of the tool. The basic principles of the process and some nomenclature are represented in Fig. 5. Essentially, the shoulder and the probe thermomechanically soften and then separate the material being processed by the passage of the probe through the material. The material flows around the probe and is then forge welded together at the trailing edge of the probe. This separation and welding together occur continuously by backfilling from the probe and compaction/containment from the shoulder. This transient separation/rewelding operation happens during and before the trailing edge of the shoulder moves away from the processed/weldtrack. The transient plasticised region immediately coalesces and forms a solid-phase bond as the tool moves away.

90


Fig. 5 Representation of the main parameters and nomenclature of FSW joints

During the FSW process, the material undergoes intense plastic deformation at elevated temperature, as a rule resulting in the generation of fine and equiaxed recrystallized grains. This fine microstructure produces good mechanical properties in friction stir welds. Better quality joints are associated with intense threedimensional (3D) material flow. The typical metallurgical structures present in the processed zone of FS welds are established and classified in Fig. 6. Beside the thermomechanically affected central zone (TMAZ), there is the heat affected zone (HAZ) and the unaffected parent material or base material (BM). Previous descriptions regarding the so-called onion ring feature within the recrystallised zone of the TMAZ (the nugget) in macrographs of transversal sections of the weld seam (e.g. Fig. 7) have been a misnomer as the contour features run the length of the weld and are more akin to an elongate swiss roll than the finite diameter of an onion. The composition of the nugget is unchanged from that of the parent material and there is no measurable segregation of alloying elements but grain size varies across the flow contours [22].

2.1.2 Advantages and Disadvantages The advantages claimed for the process include: • Solid-phase nature of the process; • Capability of welding materials whose structure and properties would be degraded by melting; • Minimal edge preparation required;


91

Fig. 6 Typical macrostructures in FSW of lightweight metallic materials [39]

Fig. 7 Macrostructural features in the weld region. FSW of AA6082-T6, 25 mm thick, using a WhorlTM probe, at a travel speed of 240 mm/min [39, 40]

• • • • • • •

Machine tool technology, simple to use with good surface appearance; Minimal distortion; Hot forged microstructure; Low residual stress levels, compared with arc welding processes; Environmentally friendly with absence of welding fume and excessive noise; Suitability for automation; Good mechanical properties;

92


Fig. 8 Friction processing technologies and concepts supporting the FSW development [41]: a Friction surfacing; b Rotary friction welding

• Welding consumables not required, with exception for inserts that can be used and gas shielding for reactive materials such as titanium and its alloys; • Not influenced by magnetic forces; • Continuous—unlimited length; • Joint can be produced from one side and in all positions. The current limitations of the FSW process are: • Backing anvil required (except bobbin stir); • Keyhole at the end of each weld (except when a tool with a retractable probe is used); • Workpiece requires rigid clamping (except when the Twin-stirTM variant is used); • Application not as flexible as certain arc welding processes.

2.2 The Third-Body Region Concept in FSW FSW is more or less an inverse 3D version of friction surfacing. In both friction surfacing (Fig. 8a) and FSW, the time and temperature parameters are very similar, but it is the different deformation modes experienced that defines each process. During friction surfacing, the visco-plasticised layer is constrained only in one dimension (tool rotation axis), whilst in FSW this visco-plasticised layer (generally referred to as the third-body) is subjected to 3D constraint. Under these conditions, the highly plasticised third-body flows in a 3D pattern almost as a liquid, mixing and blending, whilst remaining solid. Both these techniques, therefore, rely on producing suitable temperature and shear conditions within the


93

third-body transient region that exists in friction surfacing between the consumable bar and the substrate, and between the tool and the workpiece in FSW. The equal or unequal consumption and temperature distribution in a friction system has been put to good use in the development of many friction based processing technologies, e.g. surfacing and rotary welding (Fig. 8). FSW is a further development in that only a small workpiece weld region is processed, without any macroscopic geometry changes to form a solid-phase welded joint. The marked difference between the elevated temperature properties of the tool and the workpieces, together with suitable cyclic movement between the tool and the workpieces, generates sufficient heat to cause plasticised third-body conditions in the weld zone of the workpiece material [23]. There is a volumetric contribution to heat generation from adiabatic heating owing to deformation within the thirdbody region surrounding the probe and part of the shoulder. This heating is mainly accomplished by plastic deformation of the workpiece but also by interfacial friction between the tool and the workpieces. Thus, FSW is a continuous hot shear process. This stirring motion breaks up and fragments oxides from the abutting plates allowing metallurgical bonding to occur between clean/chemically active surfaces. The contacting surface of the shoulder of the tool and the length of the probe below the shoulder essentially allow the probe to maintain penetration to the required through-thickness depth. The shoulder region of the tool provides an additional friction heating and pressure treatment to the workpiece top surface. The plasticised third-body is confined in a close volume condition confined at the top by the shoulder, at the back by the anvil and at the sides by the cold regions of the workpieces. Thus, the correct diameter of the shoulder is the minimum size that prevents the hot/plasticised material of the workpieces from being expelled.

2.3 Main FSW Process Parameters One important advantage of FSW is the high quality and repeatability of the results once all the process parameters are correctly established and monitored during the welding/processing operation. The process parameters are relatively easy to assess because FSW is mostly a mechanical welding process where the results do not depend on difficult control conditions such as environmental conditions or operator skills. For achieving the total quality assurance conditions, one important requirement is a strong, stiff machine and clamping system, able to react and apply the necessary load onto the workpiece via the tool in order to create and maintain the correct conditions. In order to optimize the performance of the resultant FSW joint, and considering that the FSW results can be sensitive to variations of some welding parameters, it is important to identify and understand possible interactions between the welding parameters.

94


The main FSW process parameters are the following (Fig. 5): • Tool geometry: The shoulder and probe geometry is the most important feature that influences the final properties and overall integrity of the weld seam. This parameter will be addressed later in Sect. 4. – – – – –

Shoulder: outer diameter ? shape ? features; Probe: length +diameter along the length ? shape ? features; Oblique angle between probe axis and tool rotation axis; Ratio between static volume and dynamic volume of the probe; Tool axis eccentricity relatively to real tool rotation axis.

• Clamping system: The FSW process is a mechanical process and that requires a rigid and reliable clamping system that holds the workpieces in correct place and confines the third-body region. The correct design of this system will control the time for loading and unloading the workpieces and thus may become one major issue concerning the process productivity. The distortion can also be minimized with a correct solution of the clamping system. • Axial load, Fz: This is a key parameter for applying the correct contact pressure along the z-axis of the shoulder and probe over the workpieces. This load will always depend on the tool geometry (mainly regarding the shoulder diameter) and material of the workpieces. The correct range of values for this parameter provide the correct visco-plasticized zone without excessive flash or residual defects in the weld seam. • Tool rotational direction [CW or CCW]: The rotation direction should consider the tool special features in order to promote in the visco-plasticized zone of the workpieces a inwards material flow, at the shoulder vicinity, and downwards material flow, at the probe vicinity. • Plunge depth of probe in workpieces: This parameter correspond to the penetration of the probe in the workpieces thickness. • Plunge speed of the probe in the workpieces at the start position: This speed should decrease with the increase of the hardness of the workpieces. For very hard materials, a pre-drilled start hole with dimensions that are the same or slightly smaller than the probe diameter may be necessary. • Dwell time at start of the weld: The period of time after the rotating tool has reached the final registered contact with the workpieces at the start position. During this period the heat produced should enable the correct temperature field within the workpieces to be reached, to commence travel speed of the rotating tool over the weld path. • Tilt angle: Angle between tool axis and vertical direction to weld seam plane. This angle facilitates the entrance of material from workpieces into the close volume being FS processed. The shoulder features should be designed to enable 08 tilt angle. • Side tilt angle: Angle between the tool’s axis and the surface of the parts, measured in a plane perpendicular to the weld. This parameter is applied when


95

Fig. 9 The shoulder contact leaves in its wake a sequence of almost semi-circular ripples that points towards the start position in the weldtrack

joining workpieces with different thickness. This is only relevant for butt joints in anvil-flat arrangement. • Control during plunge, dwell and weld periods: Force control (Fz) versus position control. Typically it is applied position control during the plunge and dwell periods, shifting to load control at the start of the travel period. Nevertheless, FS welds are feasible made using position control as well as force control. • Preheating/interpass temperature of workpieces: These thermal management parameters can affect the productivity via the increase of the weld travel speed and cooling rates with consequent metallurgical impact. One example is the application of hybrid joints where the FSW tool is preceded by a heat source with or without filler material. The pre-heating of workpieces in the vicinity of FS processing zone may be necessary in the FSW of polymers, mainly because their very low heat conductivity result in inadequate plastic material flow. • Welding speed, v versus rotation speed, X: The relation between the rotation speed and the welding travel speed is known as weld pitch ratio and is established in (1). The inverse value of the weld pitch ratio is the distance between ripples, and is established in (2) and represented in Fig. 9. weld pitch ratio ½rev=mm ¼ dripples ½mm=rev ¼

X½rpm v½mm= min

v½mm= min X½rpm

ð1Þ ð2Þ

The weld pitch ratio provides a classification for the flow pattern within the thirdbody zone: effectively establishing a hot-to-cold FSW condition [24]. Although it may vary, depending on some FSW technological conditions parameters, such as tool geometry and material thermo-physical properties and thickness, for aluminium alloys it is usual to consider hot-to-cold condition criterion established in (3).

96


Fig. 10 Sample of a macrograph for hot (left) and cold (right) FSW conditions from a transversal section of a FSW butt joint of AA6056, thickness = 3.9 mm [42]

8 X½rpm > [ 4 , hot condition > > < v½mm=mm X½rpm 2 v½mm=mm 4 , intermediate condition > > > : X½rpm v½mm=mm\2 , cold condition

ð3Þ

In Fig. 10, a typical macrostructure from both hot and cold FSW conditions is shown. The above classification for aluminium alloys result from experience gained during metallurgical analysis and temperature measurements while performing friction stir welds. The differences that arise in heat and material flow between hot and cold welds are the following: • For cold welds, the HAZ is smaller and the TMAZ is larger. Because of the lower viscosity of the third-body region, the drag effect is higher and thus the TMAZ is larger and the heat is more concentrated at the retreating side (the flow side). This FSW condition allows more productive welds. • For hot welds, the HAZ is larger and the TMAZ is smaller. Most of the plastic flow deformation is localized near the probe and the heat generated by interfacial friction between the tool and the workpieces is higher. In global terms, the heat generated is almost uniformly distributed at both advancing and retreating sides. Thus, temperature field is axissymmetrically distributed throughout the weld zone. This FSW condition typically results in good superficial appearance, due to the very small distance between the ripples at the weld seam. • Intermediate weld conditions usually have the best joint overall properties.

2.4 Other Key Issues of FSW Technology 2.4.1 Process Asymmetry The inherent lack of symmetry of travelling rotating member is an important factor that should not be overlooked; it is a contributing factor to the position of certain imperfections. The influence of asymmetry is noticeable when non-optimised conditions prevail and can lead to plate thinning imperfections in lap welds and running voids (usually on the advancing side when welding aluminium alloys).


97

Fig. 11 Scheme of the velocity field at the periphery of the shoulder for a FSW tool with the following parameters [42]: Rshoulder = 7.5 mm; X = 1000 rpm; v = 300 mm/min. The ‘‘+ hot’’ and ‘‘- hot’’ areas indentified under the shoulder are established relatively to a condition where the heat was generated exclusively from the rotational speed

To better understand the asymmetric effect of the composition of velocities of the tool over the workpieces, the velocity field under the shoulder surface contacting the workpieces is represented in Fig. 11. Considering the orthogonal coordinate system (x, y) represented in Fig. 11, and its relation with the polar coordinates system and other conventions present in Fig. 11, the velocity field can be expressed by (4). vy ðr; hÞ ¼ X 2p y ¼ r cos h 60 r sin h ) ð4Þ r x ¼ r sin h vx ðr; hÞ ¼ X 2p 60 cos h þ v The relative velocity field plotted in Fig. 11 is establish in expression (5). qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ffi 2p 2 2p 2 r þ2vX r cos h þ v X vXþv ðr; hÞ 60 60 vrel ðr; hÞ ¼ ¼ ð5Þ vX ðr; hÞ X 2p r 60

98


Fig. 12 Schematic representation of typical inter-relations in the complete coupled thermostructural-metallurgical character of the FSW process [43]

2.4.2 Coupled Thermal-Structural-Metallurgical Nature of FSW Although the technological principles behind the FSW may look somewhat simple, the FSW physical phenomena supporting this solid state-process is rather complex, mainly due to the intrinsically coupled character of the process (see Fig. 12): • Structural and fluid mechanical phenomena: The viscous-plastic material flow behaviour, depending on local temperature and strain rate, interacts with (i) a dynamic rigid tool; (ii) rigid anvil supporting the joint; (iii) and the cold zone of the workpieces, with an elastic–plastic behaviour. The original and resultant residual stresses also play an important role in the overall properties of the joint. • Thermal phenomena: The temperature field depends on the heat generated during the extensive material deformation in the region containing fully-plasticized material (internal friction) and friction between the shoulder and probe surfaces sliding over the outer layers of the material workpieces. The friction coefficient also varies depending on temperature and friction mechanisms. The thermal history affects significantly the metallurgical transformations and the material flow properties.


99

• Metallurgical phenomena: The evolution of the metallurgy of the material workpieces depends on the temperature history and strain rates during material deformation. In the absence of imperfections within the weld seam, the final mechanical properties of the weld joint will depend on the final metallurgical characteristics. Many of the physical properties governing the material flow and thermal field depends on the local metallurgical characteristics. A full understanding of the coupled nature of the FSW is important for developing correct parameters for the modelling of the FSW. Modelling of FSW may complement the many experimental analysis developed by research centres worldwide, by assessing phenomena difficult or even impossible to assess experimentally, e.g. local heat generation, stress and strain rates in the vicinity of the tool, and imperfection formation mechanisms. Thus, a validated model for FSW could generate valuable information, which could be used to improve tool design and process parameters. This, in turn, would help attain defect-free joints, increased resistance to fatigue crack growth and corrosion. Nevertheless, modelling of FSW process is rather complex and there are no commercially available software with the capacity to fully simulate all the process features. Owing to the coupled character of the process, it is necessary to consider all factors of the process phenomena for the model. Any partial results are limited in value. The main difficulties in modelling the FSW process can be summarized as follows: (i) Bulk material deformation in the region containing fully-plasticized material makes numerical modelling of FSW computationally demanding and complex. Such complexity is associated with its highly non-linear character both in geometry (large deformations), material behaviour and physical formulation; (ii) The viscous-plastic flow of the materials near the tool and the elastic–plastic behaviour of the remaining material, demand a hybrid formulation, coupling fluid and solid mechanics; (iii) The model should also consider the heat flow from the weld bead into the rigid surfaces of the tool and anvil because it significantly affects the thermal field in the materials; (iv) The correct modelling of material behaviour depending on strain rates and temperature development is also of key importance in FSW modelling; (v) FSW process modelling does not allow geometric simplification because it deals with an asymmetric complex 3D material flow around the probe and shoulder. Thus, there is no symmetry plan or line to be considered; (vi) The rotating tool (shoulder and probe) has, typically, a complex geometric profile (e.g. threaded with shallow flutes), which is rather difficult to consider for most of the numerical methods available. The challenge is then to create a model able to fully describe the process, as illustrated in Fig. 12, which represents the coupled thermal-structuralmetallurgical nature of the process. Some promising attempts are published [25–27].

100


Fig. 13 Forces applied by the FSW tool into the workpieces. Note: the directions in the drawing are the effective (positive) direction of the loads and torque concerning the travel and rotation speed directions represented

2.4.3 Mechanical Characterization and Heat Efficiency The control of the FSW process is purely mechanical. Thus, understanding the forces during the FSW process becomes of fundamental importance. Figure 13 presents the direction of the loads and torque applied by the tool into the workpieces during the steady phase of FSW. The downward load (Fz) has already been addressed before, during the analysis of the welding parameters. This load plays an important role in FSW and it should be kept constant to an optimised value The loss of the downward (Fz) load entails loss in joint and surface texture that may lead to uncontrollable oscillation of the tool caused by the traverse load (Fy). Such tool oscillation may increase rapidly due to lack of registered contact between the shoulder and the workpiece surface. The traverse load (Fy) results from the asymmetrical material flow from the workpieces, around the FSW tool, owing to the combination of rotation and travel speeds (Fig. 13). In fact, the traversal load (Fy) is a reaction load resulting from the action that the material flow applies on the FSW tool. The traversal load (Fy) applied by the FSW tool into the workpieces has always the direction from the advancing side to the retreating side. As the value of the weld pitch ratio gets lower, the value of the traverse load (Fy) gets higher. Sometimes the traverse load (Fy) can even change the position of the tool from the desired path. Other relevant variable is the load in the direction of the welding joint (Fx) because if this load increases too much, the tool could be damaged seriously or even break due to combination of bending and torsional stresses. Thus, successful FSW requires the use of load cells to monitor and guarantee constant loads prescribed for the process, or the machine must be sufficiently robust as to prevent deflection in position control.


101

The total mechanical power delivered by the FSW tool into the parts being welded is expressed in (6). Then, the heat efficiency (gheat) and heat input (HI) are established in (7) and (8), respectively. Pmech ½W ¼ ½Mz ½N:mm 2pp½rpm þ Fx ½N v½mm= min gheat ¼

ð6Þ

Pheat 100% Pmech

ð7Þ

Pheat ½W 60 v½mm= min

ð8Þ

1

HI½J=mm ¼

1000 60

Note that Pheat is the part of the total Pmech that is dissipated/lost in the form of heat into the workpieces producing the thermal field (and correspondent HAZ), to the anvil, and to the tool. Typically, colder FSW condition release higher Pheat and minor HI, than hotter FSW condition [24, 28].

2.4.4 Quality Assessment with Destructive Static Loading Tests The quality of FSW joints needs to be assessed and compared with other methods. Certain destructive tests and respective analysis of the results will now be presented. Butt and overlap joint welds need particular consideration owing to the process asymmetry, eventual reduction of hardness in TMAZ and HAZ or possible internal imperfections, with emphasis for root imperfections, remnant oxide layers and alignment of any oxide layers. Methods to assess the quality of the FSW joints, in terms of mechanical resistance efficiency relatively to the base material properties, is via tensile and bending static tests of representative specimens. These tests, however, deliver many outputs which may become difficult to analyse. In order to overcome this difficulty and simplify the analysis of the results, a coefficient called global efficiency to tensile strength, GETS (9) is used where E—Young modulus; ry—yield stress; rUTS—ultimate stress; A—elongation; UT—toughness [29]. GETS ¼ CE

Ei ry i rUTS i Ai UT i þ Cry þ CrUTS þ CA þ CUT ð9Þ EBM ry BM rUTS BM ABM UT BM

In analogy with the tensile tests and the GETS coefficient, a global efficiency to bending, GEB (10) was also considered where F—maximum load; d—displacement at maximum load and UB—energy consumed until a minimum specified load is reached after the maximum load is exceeded [29]. GEB ¼ CF

Fi di UB i þ Cd þ CUB FBM dBM UB BM

ð10Þ

102


Fig. 14 Unrestrained hammer ‘S’ bend method [7]

Note that the weights CE; Cry; CrUTS; CA; CUT and CF; Cd; CUB in (9) and (10) consider the relative importance level of the mechanical properties in the mechanical design of the lightweight metallic structure. For overlap joints, a very feasible alternative to the typical three-point bending test is the hammer ‘S’ bend test. Details of the hammer ‘S’ bend test for lap welds are shown in Figs. 14 and 15. Bend testing was carried out with the weld region unrestrained. This form of bend test has proved a discerning method for establishing basic weld integrity and freedom from weakness caused by plate thinning and gave correlation with good fatigue properties [7, 30]. The bend procedure established was based on 6 mm thick 5083-O condition aluminium alloy plate with a 25 mm overlap and a set-up gap equal to the plate thickness (see Fig. 14). For more ductile materials it is suggested that the set-up gap would be \0.5 plate thickness and for less ductile materials the set-up gap would be [1.5 plate thickness. Owing to process asymmetry, the hammer ‘S’ bends tests need to be carried out in both retreating and advancing joint configurations, see Fig. 15.

3 Developments in Joint Design Early in the development of FSW it was clear that the process could be applied to a wide variety of joint types, thickness and configurations with minor alterations to typical joint design for conventional fusion welding technologies. FSW is not restricted to any kind of raw material shape (plate, tube, etc.) and can be applied to produce many different types of joint geometries by combining the butt and overlap weld seam within the workpieces arrangement (Fig. 16).


103

Fig. 15 Typical unrestrained hammer ‘S’ bends carried out in both retreating and advancing joint configurations [7]

Fig. 16 Different joints configurations by producing butt and overlap weld seam [44]: a butt; b Combination of a lap and butt; c lap joint; d lap joint; e butt to produce a T-joint; f lap to produce a T-joint; g butt to produce a L-joint; h butt to produce a corner-joint

In the case of the overlap weld, the process asymmetry should be considered because the geometry of notch tip is typically different at both advancing and retreating sides. Figure 17 presents the typical classification for overlap joints near the edge, of any of the workpieces. The obvious advantage of solid-phase friction stir welding process is the absence of gravitation effects, thereby enabling the process to be used in all positions, as shown in Fig. 18 [31].

4 Developments in Tooling 4.1 Conventional Stir Welding 4.1.1 History of Tooling At the heart of the FSW process is the welding tool. In fact, for many applications, tooling design is the critical operational parameter. Typically, the tool is designed

104


Fig. 17 Regions of tensile stress for simple lap welds [7, 8]: a Advancing side near the top sheet edge (ANE); b Retreating side near the top sheet edge (RNE); c Stress flow lines (ANE); d Stress flow lines (RNE); e Unrestrained lap welds after loading (ANE); f Unrestrained lap welds after loading (RNE)

to efficiently plasticize the joint material while minimizing the forces required (such as torque, travel and reactive loads) and increase the stirring effect of the workpieces contributing towards a better fragmentation and dispersal of all original existent layers of particles (e.g. oxide layer), within the recrystalised zone of the TMAZ. To create a weld seam, the probe part of the tooling is driven downwards into the joint until the shoulder contacts the workpiece and is then traversed along the weld seam. It is the role of the probe to create the plasticised layer and thoroughly stir both sides of the joint region together. During joining, the shoulder remains firmly in contact with the joint and provides both extra friction heat and constraint to the flow of the third-body plasticised material. Moreover, the welding tool must be sufficiently rugged to withstand the applied combinations of mechanical torsion, bending, and vibration under extreme wear and temperature conditions. A patent filed by Thomas et al. [32] covered a wide range of tool designs including textures, shapes, projections, profiles, grooves and scrolls on pins and shoulders as well as many non-tool related features such as consolidating rollers and underwater FSW. To fulfill these requirements, especially when welding 25–100 mm thick plates, two different tool forms have been developed; the WhorlTM and the MX TrifluteTM, both of which generate sufficient third-body plasticisation and provide


105

Fig. 18 Examples of positional friction stir welding (Thomas and Dolby [31]: a Welding of hollow component; b Welding of hollow component; c Welding stationary horizontal pipe; d Welding stationary 45 angled pipe; e Welding a corrugated component; Note: The G designations refer to the AWS convention for positional welding [45]

satisfactory flow path to minimise the power required to carry out the welding operation. Both tools have a comparatively high swept volume to static volume ratio when compared to ‘pin’ style tools. This is particularly important as plate thickness increases or welding travel rate is increased if weld quality is to be maintained. The WhorlTM tools were designed such that the probes were not parallel sided but frustrum shaped. Such a shape displaces substantially less material during

106


Fig. 19 Basic variants for the WhorlTM type probes [32]

Fig. 20 Prototype WhorlTM tool superimposed on a transverse section of a weld [39]

welding than a comparable conventional cylindrical pin type tool. Furthermore, a coarse auger thread is employed which further reduces the displaced volume to about 60% and which again facilitates flow of material past the tool during welding and which maintains a downward force on the weld zone during welding. Figure 19 shows the first concepts of such tool used and Fig. 20 presents the prototype WhorlTM tool superimposed on a transverse section of a weld. One of the most revolutionary tool designs to appear following this WhorlTM was the MX TrifluteTM see Fig. 21. The TrifluteTM tools were designed to give an even more efficient flow path than the WhorlTM tools. This tool design enabled a threefold increase in joining speed, in part enabled by the tapered pin shape and action of the tool features. The MX-TrifluteTM probe displaces substantially less material during welding (approximately 70%) than the conventional cylindrical pin type probe. TrifluteTM type probes can be designed with any combination of neutral, left or righthanded flute or ridge groves to suit the material and joint geometry being welded. Moreover, the individual ridges on the probe can be


107

Fig. 21 Concept of MX-TrifluteTM probe [33]

regarded as independent features. This effectively enables neutral, left or right hand inclined ridge groves to deflect plasticised material and fragmented oxides upward or downward as required with every 120 part rotation of the probe. The MX-Triflute is able to weld plate material of 75 mm in thickness in one pass. TWI filed a patent for this design in 1998 [33], but chose not to progress it as placing this design in the public domain should benefit the process at large. Since then, very few patents have been granted on the basic principles of tool design. 4.1.2 Shoulders Profiles The first FSW tools had a flat shoulder with the tool perpendicular to the workpiece [1]. In addition, concave and convex shoulders as well as flat shoulders were used with the tool slightly tilted, as first described by Thomas [32]. For welding aluminium alloys, nowadays shoulder design includes scrolled design (Fig. 22). Concerning the shoulder design, the following features should be considered: • The minimum shoulder outer diameter has to be establish to allow closing the plasticized zone being stirred by the tool, minimizing the amount of extruded flash; • On the other hand, the shoulder outer diameter should be minimized, in order to minimize the total vertical downward forging load, Fz; • When using non-scrolled shoulders, the tilt angle up to 3 is typically necessary;

108


Fig. 22 Example of geometries for the shoulder of the FSW tool: a Non-scrolled concave; b Planar with concentric scrolls; c Planar with spiral scrolls Table 1 Graphical representation speed field simulation of planar shoulders with 2 scrolls (spiral striates) for different Scroll Pitch = {0.5; 1; 2}. Constant Values: Number of scrolls (striates) = 2; Rext = 20 mm; Rint = 5 mm; X = 800 rpm Scroll Pitch = 0.5 Scroll Pitch = 1 Scroll Pitch = 2

• When using non-scrolled shoulders, a concave shape is the most frequent solution in order to reduce the amount of extruded flash from the workpieces during the weld. But it should also be considered that a convex shoulder allows for variation in plunge depth and material thickness. This concept was presented in the TWI patent filings, but was combined with other features in later filings [32]. Colligan and Pickens [34] showed a typical example of this type of tool, which has significant benefits in use. One of the primary advantages of this type of tool is the variable effective diameter of the shoulder, meaning that variations in penetration and shoulder contact zone can be made during welding. The planar scrolled shoulder for certain applications presents some advantages over the non-scrolled shoulder such as the absence or reduction of undercut and extruded flash of material left at the advancing side of the tool, and allows a tilt angle null or very small. This enables easier travel of the FSW tool over more complex paths (two-dimensional non-linear and 3D). Note that the workpieces material flowing into the space between the shoulder scrolls is compressed as it flows from outer to inner shoulder diameter. Thus, considering the incompressibility of the workpieces material, some of that material will be pushed downwards as it flows from outer to inner shoulder diameter (see Table 1).


109

Fig. 23 Compact concept of FSW tools

A design criterion for producing scrolled shoulders is established in (11) and (12) in polar coordinates. Table 1, represents the scrolls for different Scroll Pitchs. Scroll Pitch / Scroll Position ¼

NScr v X

PitchbðRshoulder Rint Þ R ¼ Rshoulder Scroll 2p b ¼ 0::Scroll2pPitch

ð11Þ ð12Þ

where: NScr—Number (integer) of scrolls (this value is set to a maximum of NScr = 4); Rshoulder—Exterior radius of the shoulder (mm); Rint—Interior radius of the shoulder (mm); b—Angle starting with the scroll on the exterior diameter of the shoulder and ending where the scroll ends at the interior radius of the shoulder (rad).

4.1.3 Tools Architecture The main concepts for the construction of a FSW tool are represented in the next three figures. Among these, the conventional compact tool (see Fig. 23) is preferred for high productive industrial environments after correct development of welding parameters This architecture allow better cooling efficiency and do not require assembling time. The modular concept is represented in Fig. 24. These tools enable probe length fine tuning and combination between different shoulders and probes. This solution is more appropriate for development of parameters. It also enables easy replacement of damaged/wear components. Mechanised tool, like the one presented in Fig. 25, is a modular tool at the ultimate level of development and control of some important tool parameters such as the length of the probe. The heat generated during the FSW processing of the workpieces is dissipated mainly via conduction of the heat flow into the clamping system ? anvil ? throughout the FSW tool (eventually flowing into the equipment via tool clamping device). Therefore, the cooling capacity of the FSW tool system (see alternatives in Fig. 26) assumes significant importance, controlling namely:

110


Fig. 24 Modular concept of FSW tools developed and patented [46]. This tool is based on 3 main components: a body; b shoulder and c probe. The drawings depict interior solutions for cooling and fine tuning of the probe length. The photographs show the tool components pre and post assembly

Fig. 25 Adjustable probe tool design architecture with in-weld continuous adjustable probe length [47]


111

Fig. 26 FSW tools with forced convection cooling: a with the exterior air; b via forced convection with a continuous flowing refrigeration liquid in the interior of the FSW tool system, showing in detail c the top tip of the refrigeration channel with O-ring for sealing

• Cooling rate of the workpieces, and thus final metallurgical properties of the joint; • Life of the FSW tool, because the tool wear increases with the mean temperature attained during the stationary stage of FSW process; • Life of the welding equipment, mainly of the bearing system nearest the tool clamping device.

4.2 Bobbin Stir Welding Bobbin tools were described in the first TWI patent [1] but have been the subject of many subsequent filings that improved on the basic idea, e.g. [35]. A bobbin tool has a second shoulder connected to the end of the probe, providing clamping and consolidation to the underside of the workpiece. This lessens the amount of clamping required and removes the need to provide support during welding. Some of the patent filings deal with developments of such tools with variable force, position control and independent shoulder rotation. FSW using a self-reacting bobbin tool has been shown to be effective for joining hollow extrusions and lap joints (Fig. 27). Essentially, there are two main types of self-reacting techniques. One is known as the fixed-gap ‘bobbin tool’ [34] and the other one as the adjustable [36] or ‘adaptive technique’ (AdAPT) [37].

112


Fig. 27 Self-reacting bobbin stir welding, showing near and far side shoulders [41]

Fig. 28 Bobbin tool showing self-contained reactive forces [41]

A derivative of the fixed-gap is the ‘floating bobbin’ tool, which is a fixed-gap tool that has been designed to float in the direction perpendicular to the workpiece. The bobbin techniques provide a fixed-gap between two shoulders, while the adaptive technique enables adjustment of the gap between the shoulders during the welding operation. The self-reacting principle of the bobbin technique means that the normal down force required by conventional FSW is reduced or eliminated. The reactive forces within the weld are contained between the bobbin shoulders (Fig. 28). Trials in 25 mm thick AA6082-T6 aluminium using the above arrangement produced good quality welds. Figure 29 shows a metallurgical transversal section of the widths of the larger diameter (drive side) shoulder and the smaller opposed shoulder in the weld area. Unlike single-sided stir welds, the weld profile is narrower in the mid-thickness than at the shoulder regions. Several flow features within the TMAZ can be seen in Fig. 29. Bobbin type tools are similar to other standard FSW tools that are driven from one side in that the tool behaves as a rotating cantilever. The use of a tapered probe for a simple (non-floating) bobbin tool provides for a more uniformly stressed tool which displaces substantially less material during welding than a cylindrical pin type probe. The use of a tapered probe for the bobbin tool enables a proportional


113

Fig. 29 Macrosection of bobbin weld in 25 mm thick AA6082-T6 aluminium, made with a no-floating fixed-gap bobbin tool [41]

reduction in the diameter of the lower shoulder of the bobbin tool. A reduction in the lower shoulder diameter results in lower frictional contact and resistance, therefore less torque and bending moment on the tool. The additional frictional contact provided by the lower shoulder and the absence of a backing anvil, which acts as a heat sink, means that the operating temperature will be higher than that of a similar conventional weld. Moreover, owing to the limited thermal conduction path from the shoulder furthest away from the drive side, this shoulder will run slightly hotter. In some situations, thermal management techniques such as cooling the shoulder by an air blast are used. Tool design and process conditions will need to be adjusted to allow for the welding travel speed to be increased benefiting from such additional heat generation. Bobbin welds essentially eliminate partial penetration, lack of penetration and root defects. Preliminary trials have also shown that lap welds produced by the bobbin technique have fewer problems with the adverse orientation of the notch at the edge of the weld. Moreover, the use of bobbin techniques typically causes less distortion than conventional FSW due to a more balanced heat input. Moreover, the low welding forces in the Z axis may eliminate the need for heavy-duty FSW machines and clamping system. The floating bobbin fixed-gap concept is shown in Fig. 30. It is self-positioning in the axis perpendicular to the workpiece. The geometry and features of the floating bobbin tool are symmetrical and are designed to produce a balanced material flow equalising the opposing reactive forces on the upper and lower tool shoulders. The bobbin tool operates within a sleeve which provides vertical guidance and the rotational drive via a keyway [36]. The instrumentation chart shown in Fig. 31 provides clear evidence of the very low axial (Fz) force, well balanced around the zero-force datum line. For certain applications, bobbin tools that are driven from both ends can be envisaged (Fig. 32). The concept of a double-driven bobbin also includes the use of a double-adaptive technique whereby both shoulders can be adjusted and a load applied from both ends, see Fig. 32b. The latter arrangement will reduce the

114


Fig. 30 Fixed-gap floating bobbin [41]

Fig. 31 Instrumentation chart from a typical fixed-gap floating bobbin weld of 25 mm thick AA6082-T6 [41]

reactive forces transmitted through the probe and enable FSW to tackle thicker plate material than currently possible. This concept is expected to increase the welding speed significantly above that which is possible using conventional bobbin techniques and may even provide welding speeds faster than conventional FSW for thick plate welding.


115

Fig. 32 Bobbin tool [41]: a Driven from both ends; b Driven from both ends and reactive force applied from both ends

With both sides of a fixed-gap bobbin tool driven, the probe part of the tool no longer behaves as a rotating cantilever. A bobbin tool that is driven from both ends and designed for uniform stress, means that the aspect ratio of the probe can be altered (decrease in cross-section area and/or increase in length) to provide an improved flow path. However, while the torque and bending forces can be shared between both ends, the cross-section of the probe must be able to accommodate the reactive forces that tend to push the shoulders apart. The use of bobbin type techniques requires run-on and run-off regions for the tool. Bobbin techniques are best suited to flat two-dimensional applications but could be developed for more complex shapes.

4.3 Material for the Tooling To correctly select the material, heat treatment and superficial hardening condition for the tooling in FSW, is technologically and economically critical. The thermophysical properties of the selected material should comprise with all the forecast applications conditions. This is a somewhat easy task for welding aluminium alloys, but it may become rather complex and expensive when joining workpieces with high strength and/or fusion temperature such as titanium alloys. Concerning the architecture of the tool, for the case of a modular tool, it is common to find different components in different materials. The tool wear can be reduced by promoting a lower working temperature, e.g. via internal refrigeration of the tool. The most common solutions for FSW of aluminium and its alloys are the following: • Plain carbon steel AISI 1045 or DIN Ck45 K: This material can be selected for components of modular FSW tools not contacting the workpieces;

116


• Tool steel for hot working conditions EN X40CrMoV5-1 or AISI H13: This material is used for compact FSW tools or for components contacting the workpieces of modular FSW tools. This material can be easily machined and is sensible to heat treatment and superficial hardening, e.g. the hardness of the H13 tool steel can be increased from the original 240 HV to about 1200 HV after the ionic nitruration. An oxidation treatment can be applied to the surfaces, e.g. under an environment of water vapor at 5008C, in order to reduce the adhesion between the FSW tool and the workpieces. For FSW of workpieces of high toughness and/or temperature resistant materials, a proper tool material must be found which can withstand the high temperatures (easily above 900C) and high mechanical cycling loads experienced during welding. The following materials are possibilities to consider: • MMC Densimet 185: 90% W ? 10% (Ni ? Fe). This material has good corrosion resistance and good mechanical properties for the range of temperature from 500 to 1000 8C, but it is not recommended for long runs; • Polycrystalline cubic boron nitride (PCBN); • MMC TZM (trademark): Mo alloy with 0.5% Ti ? 0.08% Zr ? 0.02% C. This material has good combination of metallurgical properties for applications at high temperature; • Carburizing or nitriding refractory metal such as Lanthanated tungsten (La/W), in order to lower the ductile–brittle transition temperature of the W below room temperature. • Tungsten–Rhenium alloys. These alloys cannot be carburized. Unlike pure tungsten, tungsten–rhenium (W–Re) maintains a much greater ductility due to its rhenium content.

5 Imperfections and NDT Techniques for FSW 5.1 Imperfections in FSW Joints FS welds can be produced free of imperfections and reproduced with total quality assurance, however, some imperfections may arise owing to poor tool design, improper development and establishment of the process parameters or mismonitoring of the key-parameters during the welding or processing operation. Namely: insufficient 3D bulk stirring of the third-body region, inadequate surface preparation, incomplete penetration of the probe, or inadequate axial load (force control) or lack of contact (positional control). The FSW is essentially a mechanical process involving the application of some relatively high forces into the workpieces by the tool and clamping system. The results from the application of the process can be rather sensitive to some of these conditions, and as result of badly implemented conditions, many imperfections can


117

Fig. 33 Representation of possible imperfections in FSW butt joints [48]

Fig. 34 Representation of possible imperfections in FSW lap joints: a 2 plates in overlap joint with particles alignment and voids in recrystalised zone of TMAZ; b 3 plates in overlap joint with severe thickness reduction resultant from the material flow; c fracture due to thickness reduction, and d Overlap weld seam to produce a T-joint with internal and geometrical imperfections

occur. A representation of the imperfections possible to develop is shown in Fig. 33 for butt joints and in Fig. 34 for lap joints. Certain bobbin welds can reveal a mid-thickness ‘‘blip’’ that appears on the advancing side. Non-optimised welds can also be characterised by imperfections that appear in the mid-thickness near the ‘blip region’ of the weld on the advancing

118


Fig. 35 Non-optimised bobbin welds in 25 mm thick 6082-T6 aluminium alloy showing a mid-thickness ‘blip’ and imperfections on the advancing side [41]

side, see Fig. 35. These imperfections are usually caused by insufficient static to dynamic volume ratio of the probe to provide an adequate flow path. The action of two shoulders causes a bifurcation in the plastic flow which in turn leads to an increase in the number of cyclic flows than would otherwise be observed when the conventional FSW technique is used. Some of most relevant potential effects of these imperfections in the performance of the friction stir welded structures are the following: • Corrosion resistance: In general, corrosion concentrates at geometric variations and follows particles alignment. At the top surface, corrosion mechanisms start at geometric discontinuities, e.g. ripples and beneath the flash (if existing). At the bottom surface, the corrosion mechanisms concentrate and follow the root imperfections; • Mechanical resistance: – Static loading: Uniaxial tensile condition is only sensitive to incomplete penetration and voids. Root bends (root in tension) is sensitive to the presence of root flaws controls and incomplete penetration. Face bends (top surface in tension) assess weld surface defects. Side bends can be useful for addressing both above mentioned defects localization. – Fatigue loading: Root imperfections plays a fundamental role in the fatigue efficiency. The fracture mechanism of FSW under fatigue load is mainly determined by the size of the imperfection at the root of weld bead. The tests of FSW beads with root imperfections higher than a critical level, always results in less number of cycles when compared to the base material. However, the fatigue performance of the FSW joints is higher when compared to the other welding process like GTAW or synergic GMAW. Moreover, reducing in-depth size of the root imperfections via surface smoothing brings the fatigue life of the FSW very close to the one observed on the base material.


119

Fig. 36 Representation of the only meaningful FS weld imperfections that should be considered for NDT inspection of high performance components of lightweight metallic structures [48]

5.2 Detection of FSW Imperfections by NDT NDT is a vital function in achieving the goals of efficiency and quality at an acceptable cost. The consequences of failure of engineering materials, components and structures are well known and can be disastrous. Among all the possible imperfections in FSW, established previously, the only technologically meaningful FS weld imperfections that should be considered for NDT inspection are the following: (i) root imperfections (weak or intermittent welding); (ii) incomplete penetration (incorrectly called, kissing-bond); (iii) voids mainly located at the advancing side of the weld but also possible at retreating side, and (iv) second phase particles and oxides aligned under the shoulder (Fig. 36). The geometry, location, and microstructural nature of the friction stir weld imperfections bear no resemblance to the imperfections typically found in fusion welds of lightweight metallic materials. As an example, these imperfection characterization parameters will now be established for one of the most difficult to detect imperfection: particles alignment at the root of FSW joints in zones with severe plastic deformation, resulting in weak or intermittent welding. The size is typically within the range of 20 lm to 500 lm. The location is root superficial layer. The morphology presents no physical discontinuity (even in incomplete penetration) with weak effect of energy changes (acoustic and electromagnetic), in a modified zone (recrystalised zone of TMAZ) with significant local gradients of material physical properties. In general, the data available for the application of NDT to FSW joints is still scarce. Nevertheless international cooperative projects are being undertaken on

120


Fig. 37 The IOnic Probe prototype: a geometric concept and b implementation with 2 sensitive coils and c 4 sensitive coils, also in d and e printed on a flexible substrate and f the basic principle emphasizing the 3D character of the eddy currents enabling to detect imperfections with different orientation [38]

this subject. These projects are endorsed by the industrial total quality management concepts based on very high quality standards and total quality assurance paradigms. At the top of the requirements, one can find the aeronautic industry. Moreover, the typical NDT techniques, such as: visual inspection, magnetic particles, liquid penetrant and X-rays, do not enable the detection and/or quantification of the meaningful FS weld imperfections. Even considering the most recent form of evolution of the NDT techniques, namely: Pulsed Eddy Currents (PEC), Eddy Currents (EC) Arrays, Meandering Winding Magnetometer (MWM), Superconducting Quantum Interference Device (SQUID), Giant Magnetoresistance (GMR), Ultrasonic (US) Phased Array, Time of Flight Diffraction (ToFD), and digital radiography, none of these existing NDT techniques used independently allow the assessment of the size, morphological diversity and localisation of all possible FSW defects. This difficulty, among others factors, is due to the fact the techniques based on EC and US are very sensitive to coupling and lift-off conditions between the probes and the surfaces under inspection. Thus, there is a need for a new dedicated NDT technique and/or data fusion algorithm specifically developed for the NDT of friction stir welds. Nevertheless, in conductivity materials, when high reliability is required as mentioned before, Eddy Currents is the most common used technique. A new NDT system was recently developed based in new patented eddy current probes (IOnic Probes) [38]. The new patented design allows: (i) 3D induced eddy currents


121

in the material; (ii) deeper penetration; (iii) independence of the deviation between the probe and the material surface (planar lift-off); (iv) easy interpretation of the output signal based not on the absolute value but on a comprehensible characteristic change. In addition, this probe can be manufactured on flexible substrate, allowing non-planar surfaces inspection. The IOnic probe is constituted by one excitation filament in the middle of two sensitive planar coils, in a symmetric configuration (Fig. 37). The experimental results reported by Santos et al. [38] conclude that the IOnic probe is able to identify different levels of FSW root imperfections down to a size of about 50 lm, by a distinctive perturbation on the output signal. It was also shown that there is a good proportionality between the defects size and the signal perturbation.

6 Conclusions The state-of-the-art collected in the present chapter enables the following conclusions to be made: • The basic fundaments of the FSW technology have been presented and will continue to be further investigated. These basic fundamentals have enabled the invention and development of many variants of FSW and friction stir processing. It is expected that there are still many more FSW inventions to be discovered and developed based on the highly plasticised third-body region concept. • FSW is a very complex process and difficult to be assessed based on computational modelling; nevertheless, there are a few successful attempts which should be pursued in order to benefit the ongoing development of FSW technology. • The FSW tooling concepts and design solutions presented in this chapter are able to be selected in order to support any application concerning the welding of lightweight metallic structures. • New joint design concepts, different from the ones typically applied in fusion welding, should be considered for more effective transfer of FSW technology to production lines. Moreover, FSW with the correct joint design can provide significant economic savings, reduce environmental impact, and increase overall joint properties. • The destructive tests and non-destructive testing technologies for quality assessment of the FSW joints have developed to a point where industry can rely on the results found. This will help establish confidence in the FSW of lightweight metallic components. • FSW is a mature and reliable technology with guidelines in many construction codes, mainly focusing on products made from aluminium and its alloys. The ISO standard 25239 Friction stir welding—Aluminium alloys, has been published comprising the following parts: Part 1-Vocabulary; Part 2-Design of weld joints; Part 3-Qualification of welding operators; Part 4-Specification and

122


qualification of welding procedures; Part 5-Quality and inspection requirements. This ISO standard will help with the implementation of FSW technology in light metal fabrication industries. Acknowledgments The authors would like to acknowledge the support of Dr. R. Fenn (formerly Brunel University), Dr. C. S. Wiesner (TWI) and Dr. T. Santos (New University of Lisbon). Certain Figures—‘Courtesy of TWI’ and any other photos from the public domain appropriate reference and courtesy is made.

References 1. Thomas, W.M., Nicholas, E.D., Needham, J.C., Murch, M.G, Temple-Smith, P., Dawes, C.J.: Improvements relating to friction stir welding. US Patent No. 5,460,317 (1991) 2. Bevington, J.: Spinning tubes mode of welding the ends of wire, rods etc., and mode of making tubes. US patent 463134 (1891) 3. Klopstock, H., Neelands, R.: An improved method of joining or welding metals. British Patent specification 572789 (1941) 4. Nicholas, E.D., Thomas, W.M.: Metal deposition by friction welding. Weld. J. 8, 17–27 (1986) 5. Bedford, G.M.: Friction surfacing for wear applications. Met. Maters. 6, 702–705 (1990) 6. Thomas, W.M.: The manufacture of metal matrix composite clad layers by friction-surfacing. MPhil thesis, Brunel University (1992) 7. Thomas, W.M.: An investigation and study into the friction stir welding of ferrous-based material. Ph.D. Thesis, CMRI University of Bolton (2009) 8. Thomas, W.M., Nicholas, D.E., Staines, D.G., Tubby, P.J., Gittos, M.F.: FSW Process Variants and Mechanical Properties. Doc IIW-1664–04. Weld. World 49(3/4), 4–11 (2005) 9. Thomas, W.M.: Friction stir welding and related friction process characteristics. 7th International Conference Joints in Aluminium, Inalco, TWI, Cambridge (1998) 10. Thomas, W.M., Sylva, G.: Developments of friction stir welding. ASM Materials Solutions 2003, Conference and Exposition, Pittsburgh, Pennsylvania, USA (2003) 11. Thomas, W.M., Norris, I.M., Staines, D.G., Clarke, P.J., Horrex, N.L.: Friction stir welding— variants and process techniques. The First International Conference ‘Joining of Aluminium Structures’ Moscow, Russian (2007) 12. Mazzaferro, J.A.E., Rosendo, T.S., Mazzaferro, C.C.P., Ramos, F.D., Tier, M.A.D., Strohaecker, T.R., dos Santos, J.F.: Preliminary study on the mechanical behavior of friction spot welds. Soldagem e Inspeção 14(3), 238–247 (2009) 13. Smith, I.J., Lord, D.D.R.: FSW Patents—A stirring Story. 7th International Friction Stir Welding Symposium, Awaji Island, Japan (2008) 14. Vidal, C., Infante, V., Vilaça, P.: Fatigue behaviour in friction stir welded joints of AA2024 treated by improvement techniques. Welding in the World (ISSN 0043-2288), vol. 53, special issue, pp. 241–246 (2009) 15. Pépe, N., Vilaça, P., Quintino, L., Reis, L., Freitas, M.: Fatigue behavior of shipbuilding N. aluminium alloy welded by Friction Stir Welding. Session 3: Welds, 9th International Fatigue Congress, Atlanta, USA (2006) 16. Leal, R.M., Leitão, C., Loureiro, A., Rodrigues, D.M., Vilaça, P.: Microstructure and Hardness of Friction Stir Welds in Pure Copper. J. Mat. Sci. Forum 636–637, 637–642 (2010) 17. Çam, G., Mistikoglu, S., Pakdil, M.: Microstructural and mechanical characterization of friction stir butt joint welded 63% Cu-37%. Zn brass plate Weld. J. 88(11), 225S–232S (2009)


123

18. Kannan, B.M., Dietzel, W., Zeng, R., Zettler, R., dos Santos, J.F.: A study on the SCC susceptibility of friction stir welded AZ31 Mg sheet. Mat. Sci. Eng. A 460–461, 243–250 (2007) 19. Fujii, H., Sun, Y., Kato, H., Nakata, K.: Investigation of welding parameter dependent microstructure and mechanical properties in friction stir welded pure Ti joints. Mat. Sci. Eng. A 527(15), 3386–3391 (2010) 20. Thomas, W.M., Wiesner, C.S., Marks, D.J., Staines, D.G.: Conventional and Bobbin friction stir welding of 12% chromium alloy steel using composite refractory tool materials. Sci. Technol. Weld. J. 14(3), 247–253 (2009) 21. Shi, Q-Y., Silvanus, J., Liu, Y., Yan, D-Y., Li, H-K.: Experimental study on distortion of Al6013 plate after friction stir welding. Sci. Technol. Weld. J. 13(5), 472–478 (2008) 22. Mahoney, M.W., Rhodes, C.G., Flintoff, J.G., Spurling, R.A., Bingel, W.H.: Properties of friction-stir-welded 7075 T651 aluminium. Metall. Mat. Trans. A 29A (1998) 23. Godet, M.: The third-body approach: A mechanical view of wear. Wear 100, 437–452 (1984) 24. Vilaça, P., Quintino, L., dos Santos, J.F., Zettler, R., Sheikhi, S.: Quality assessment of friction stir welding joints via analytical thermal model, iSTIR. Int. J. Mat. Sci. Eng. A 445–446, 501–508 (2007) 25. Askari, A., Silling, S., London, B., Mahoney, M.: Modeling and analysis of friction stir welding processes. The Minerals, Metals and Materials Society, Indianapolis, USA, pp. 43–54 (2001) 26. Colegrove, P.A., Shercliff, H.R.: 3-Dimensional CFD modelling of flow round a threaded friction stir welding tool profile. J. Mat. Process. Technol. 169(2), 320–327 (2005) 27. Schmidt, H., Hattel, J.: Modelling heat flow around tool probe in friction stir welding. Sci. Technol. Weld. J. 10(2), 176–186 (2005) 28. Neumann, T., Zettler, R., Vilaça, P., dos Santos, J.F., Luísa Quintino: Analysis of SelfReacting Friction Stir Welds in a 2024-T351 Alloy. Friction Stir Welding and Processing IV, TMS, pp. 55–72 (2007) 29. Silva, M.B., Skjoedt, M., Vilaça, P., Bay, N., Martins, P.A.F.: Friction Stir Welding of Aluminium Tailored Blanks Processed by Single Point Incremental Forming. IRF’2009—3rd International Conference on Integrity, Reliability & Failure, Chapter XXI, pp. 599–600 (2009) 30. Thomas, W.M., Pisarski, H.G., Norris, I.M., Marks, J.D., Godden, J.C.: An investigation of ‘Through-hole’ impact testing of weld root imperfections in friction stir welded 12% chromium alloy steel. J. Eng. Manuf. 222, Part B (2008) 31. Thomas, W.M., Dolby, R.E.: Friction Stir Welding Developments. 6th International Conference on Trends in Welding Research, Georgia, USA (2002) 32. Thomas, W.M., Nicholas, E.D., Needham, J.C., Temple-Smith, P., Kallee, S.W.K.W., Dawes, J.C.: Friction stir welding. UK Patent application GB 2306366A 17/10/1996 (1996) 33. Thomas, W.M., et al.: High performance tools for Friction Stir Welding. Patent filing WO 99/ 52669 (1999) 34. Colligan, K.J., Pickens, J.R.: Friction Stir Welding of Aluminium Using a Tapered Shoulder Tool’. ‘Friction Stir Welding and Processing III’. Jata K.V., Mohoney M., Mishra R.S., Lienert T.J. (eds) TMS Annual Meeting, San Francisco, pp. 161–170 (2005) 35. Otsuka, D., Sakai, Y.: Self Reacting Pin Tool Application for Railway Car Body Assembly. 7th International Symposium on FSW, Awaji Yumebutai Conference centre, Japan (2008) 36. Thomas, W.M., Nicholas, E.D., Needham, J.C., Murch, M.G., Temple-Smith, P., Dawes, C.J.: Improvements relating to friction welding. European Patent Specifications 0615 480 B1 37. Thomas, W.M., Sylva, G.: Developments of Friction Stir Welding. ASM Materials Solutions 2003, USA (2003) 38. Santos, T., Vilaça, P., Dos Santos, J., Quintino, L.: A new NDT system for micro imperfections detection: Application to FSW and FSpW. Weld. World 53, special issue, pp. 361–366 (2009)

124


39. Fenn, R., Thomas, W.M.: The Friction Stir Welding Process. Light Metal Age, pp. 28–33 (2001) 40. Thomas, W.M., Staines, D.G., Norris, I.M., de Frias, R.: Friction Stir Welding—Tools and Developments. FSW Seminar IST Porto, Portugal (2002) 41. Thomas, W.M., Martin, J., Wiesner, C.S.: Discovery invention and innovation of friction technologies—for the aluminium industries. INALCO 2010, Netherlands (2010) 42. Vilaça, P., Quintino, L., dos Santos, J.F.: iSTIR—analytical thermal model for friction stir welding. J. Mat. Process. Technol. 169(3), 452–465 (2005) 43. Santos, T., Vilaça, P., Quintino, L., dos Santos, J.: Computational tools for modelling FSW and an improved tool for NDT. Weld. World 53(5–6), R99–R108 (2009) 44. Dawes, C., Thomas, W.: Friction stir process welds aluminium alloys. Weld. J. 3, 41–45 (1996) 45. Kearns, W.H.: Engineering, Costs, Quality and Safety. Welding Handbook, AWS, vol. 5, 7th edn, Miami, FL 33135 (1984) 46. Vilaça, P., Santos, T.: Ferramenta não consumível modular ajustável e refrigerável para soldadura e processamento por fricção linear. National Portuguese patent application PT 104072 (2008) 47. Ding, R., Oelgoetz, P.: Mechanical Property Analysis in the Retracted Pin-Tool Region of Friction Stir Welded Aluminium-Lithium 2195. 1st International Symposium on Friction Stir Welding, Thousand Oaks, California, USA (1999) 48. Santos, T., Vilaça, P., Quintino, L.: Developments in NDT for detecting imperfections in friction stir welds in aluminium alloys. Weld. World 52(9/10), 30–37 (2008)

FSW of Lap and T-Joints L. Fratini

Abstract Even if in the last years several researches have studied the Friction Stir Welding (FSW) process, it should be observed that most of these studies are concerned with the butt joint and just a few of them extend to more complex geometries. It is worthy to notice that the acquired knowledge on FSW process of butt joints is not immediately extendable to lap and T-joints. The first observation is that in butt joints the surface to be welded is vertical, while in lap and T-joints it is horizontal and placed at the bottom of the top blank to be welded; in this way a major vertical component of the material flow is required to obtain sound joints. In the FSW of lap-joints four different geometrical configurations are possible— actually reducible to two—on the basis of the combination of the mutual position of the sheets to be welded and of the tool rotation direction, strongly affecting the process mechanics and the effectiveness of the final part. Furthermore, in the FSW of T-parts a proper clamping fixture is needed in order to fix the stringer during the process; such fixture is characterized by two radii, one for each side of the joints, corresponding to the radii between skin and stringer in the final welded part (corner fillets). Actually during the FSW process such radii must be filled by the flowing material. Consequently, an actual forging operation is required to force the sheet and the stringer material in fulfilling the radii of the clamping fixture, resulting in the radii of the T-joint. In other words, the material flow induced by the tool in the FSW process must be effective enough to get both the bonding of the two blanks and the fulfillment of the fixture radii. On the basis of the above observations, once the material of the blanks to be welded is selected, the most effective set of operating and geometrical parameters that optimize the FSW of butt joints will not, in all probability, work for the lap or T-joints. In particular, the

L. Fratini (&) Dipartimento di Ingegneria Industriale, University of Palermo, Palermo, Italy e-mail: [email protected]

Adv Struct Mater (2012) 8: 125–149 DOI: 10.1007/8611_2010_48 Ó Springer-Verlag Berlin Heidelberg 2011 Published Online: 11 January 2011

125

126

L. Fratini

tool geometries together with the tool feed rate and rotating speed must be redetermined in order to get an effective material flow and bonding conditions during the FSW process since the plastomechanics of the two processes are completely different. The specific peculiarities of the two processes must be properly investigated and the correlations between the characteristics of the materials to be welded and the mechanics of the welding configurations must be highlighted.

1 Introduction Since the invention and first proposition of FSW [1] a large number of researches, papers and reports have been proposed by the scientific community. Actually, almost all the reported research activities have been developed on simple joints and specimens, namely butt joints developed on undeformed flat blanks, along simple straight welding lines [2]. In other words, the developed researches can be classified as fundamental ones: on one hand they have been aimed to investigate relevant aspects as the FSW process mechanics, the material flow, the metallurgical evolution of the material and the actual bonding conditions. On the other hand they have been focused on the determination of the most effective process parameters for different materials in order to maximize the mechanical performances of the obtained joints [3]. In this way, most of the acquired knowledge on FSW is referred to the basics of such technology. It should be observed that there is still a lack of knowledge has to be covered regarding materials (innovative aluminum alloys, titanium, steel, Metal Matrix Composites and so on) and processes (mixed joints, thin sheets, spot FSW, Friction Stir Processing). For these reasons the next step in the development of FSW technology is probably an investigation of the actual industrial sustainability of the process also regarding joints configurations different from the traditional butt one. It is well known that the most common configurations of welded joints—apart from the butt joint—are the lap and the T ones. As far as the formers are regarded, FSW is obtained fixing the overlapped adjoining blank edges on a properly designed fixture, inserting a specially designed rotating pin into the edges of the sheets to be welded and then moving it all along the welding line. In Fig. 1 a sketch of the process for linear lap joints is reported: in particular, a proper support has to be given during the process to the joining edges in order to avoid the forward extrusion of the blanks material due to the tool pin action, as in the welding of butt joints, but also bending mechanics of the upper blank must be prevented with a proper element of the fixture. A few papers can be found out in literature on FSW of lap joints focusing on different topics. First of all the relevance of the used tool was considered: Thomas et al. [4] and also Colegrove et al. [5] used complex geometry tool pins such as the TrivexTM, the Flared-TrifluteTM, the Re-stirTM and others. The effects on the shape and extension of the nugget were considered through micro and macro observations. What is more, in the mentioned

FSW of Lap and T-Joints

127

Fig. 1 Sketch of the FSW process of lap joints

paper, the importance of the enlargement of the stirring effect determined by action of such tools on the overlapped blanks was highlighted. Another relevant topic which has been investigated in the last years regards specific aspects concerning the welding of dissimilar materials. In other words the material micro-structural evolutions have been taken into account with specific reference to the cases in which two different materials are welded together by FSW in a lap joint. In this way, Elrefaey et al. [6, 7] took into account lap joints of pure aluminum to steel and zinc coated steel plates. In particular it was found that the performance of the joint depended strongly on the depth of the pin tip of the FSW tool into the steel plate surface. The referred research overall deals with a very relevant topic regarding FSW processes of lap joints, namely the level of tool interaction with the bottom blank of the joint. The authors specify that without a relevant tool-blank interaction, i.e. penetration, the obtained joint would show low and ineffective mechanical performances in peel tests. Moreover, the authors also showed that the zinc coating induced a beneficial effect on the joint resistance if compared with joints of the same blanks for which an uncoated steel sheet is used. Finally the SEM investigation revealed the layered structures at the interface between the two welded blanks; the chemical composition of the latter structure depended on the used process parameters. Similar joints were investigated by Chen et al. [8, 9] through optical microscopy, scanning electron microscopy and X-ray diffraction with the aim to highlight the interface microstructure. Four different material zones were found out, namely the stirred aluminum alloy, the new intermetallic layers, the zinc coating and the steel base material. Again relations between the process parameters and the thickness of the inter-metallic compounds were established leading to different joints mechanical performances. In particular it was clarified that increasing the compound thickness the joints strength decreases. Another interesting and discussed kind of lap joint is the one made of aluminum and copper blanks. Again, material micro-structural issues were investigated [10],

128

L. Fratini

a so called dark area was detected and several intermetallic compounds were highlighted characterizing the interface between the two welded blanks. Overall, proper process parameters were obtained in order to get effective mechanical performances of the joints. Furthermore, in a more recent research work [11] a basic guideline was assessed in order to get effective welded joints between aluminum and copper blanks: actually lowering the amount of the heat flux during the FSW process can lead to better results in terms of joint strength since a limited formation of inter-metallic compounds is observed. It should be observed that a sort of process windows in terms of process parameters ranges was determined by the authors allowing to avoid or reduce the formation of micro-cracks in the dark area at the aluminum-copper interface. Interesting information is also derived from lap joints of two different aluminum alloys [12]. Pull up and pull down mechanics were highlighted varying the process parameters, namely tool feed rate and tool rotation speed. Also, the mutual position of the two blanks was taken into account; such latter issue will be clarified in the further sections of the present chapter. It should be observed that, on the basis of the mutual position of the two blanks to be welded and of the tool rotation speed versus, different locations for advancing side and retreating side are identified in the joint. Lee et al. highlighted the material movement in the vertical direction of the joint through a detailed macro and micro analysis at the varying of the most relevant process parameters. Finally a few guidelines were outlined regarding the process mechanics and the effect of the process variables. As far as fatigue properties of lap joints are regarded, a few researches can be found out in literature in the last years. Ericsson et al. [13] gave an interesting overview of the matters regarding fracture initiation and propagation focusing their attention on aluminum alloys. The authors stressed the importance of the width of the weld at the interface between the sheets with reference to the mechanical resistance of the joints. Another relevant aspect highlighted in the mentioned research regards the tensile stress distribution in the joint as it is subjected for instance just to an ideal shear test. The latter issue, as it will be shown in the next sections of the present chapter, has to be carefully considered in relation to the loci of the FSW joint, namely retreating side and advancing one. The carried out research led to the definition of interesting relations between the joints mechanical properties and the tool geometrical features, namely the tool shoulder diameter and the tool pin shape and dimensions. In particular Ericsson et al. found out that a larger tool shoulder diameter leads to better fatigue performances due to the increased heat energy supplied by a larger contact area. It is well known that in FSW processes, the frictional forces work at the tool-workpiece interface decays into heat both at the tool shoulder and tool pin surfaces. Moreover, the plastic deformation work determined on the material by the tool action decays, for about 95%, into heat determining a further moving heat flux on the material. Schmidt et al. [14] set an analytical model for the heat generation in FSW processes and assessed that the mechanical power due to the traverse movement of the tool is negligible compared to the rotational power originating the frictional resistances. More precisely, the frictional forces work, calculated at the tool


129

shoulder interface, represents about the 86% of the total heat. Such relevant contribution can be then considered the major heat source occurring in FSW processes. this very important result was confirmed by all the investigations made by Woo et al. [15]. In this way it looks reasonable that in a lap-joint configuration since a larger heat source is needed in order to get proper welding conditions at the sheet blanks interface, a larger tool shoulder diameter is required. Finally, crack propagation rate and surface hardness were also investigated by Woo et al.. The characterization of the fatigue behavior of FSW lap-joints of 2XXX series aluminum alloys were considered in a few further papers [16, 17]. In particular, the stress intensity factor at the tip of the propagation crack and the fatigue crack path were investigated using shareware Finite Element code Franc 2D, and the lifetime of the joints was estimated by integrating the material propagation law. So called hook effects were highlighted in the joints, being the initiations of the cracks at the interface between the two welded blanks. Moreover, mixed mode I/II conditions were highlighted and investigated at the crack tip as the joint is loaded in shear. In particular, an approach founded on numerical simulation based on the finite element method and experimental observation was followed, the latter being based on the so called ‘‘fatigue failure mechanism map’’ [17]. Further interesting aspects to be considered in the design of FSW process of lap joints are related, as already mentioned, to the tool geometry. Chen and Nakata (2009) investigated the effects of such issues on the microstructure and the mechanical properties of magnesium alloy and steel joints. In particular the authors investigated the effects of the tool pin length and showed that as a longer tool pin was used—i.e. reaching also the lower steel blank during the welding process and in this way extending to such blank the stirring effect—a better interface between Mg and steel is provided, promoting the formation of diffusion joint with a thin interface. In turn, with a shorter tool pin only a mechanical fastening effect between the two blanks was obtained, and as a consequence the joints showed lower mechanical resistance. Another relevant topic is the surface state at the interface of the blanks to be welded [18]. Actually, as different surface finishing or coatings are investigated, totally different results in terms of mechanical resistance of the joints are obtained and for very low values of surface roughness quite poor joint strength are obtained. Obviously, the process parameters effects must also be considered in order to carry out an effective process design for FSW of lap joints. Kulekci et al. [19] for instance, investigated the effect of tool rotation and pin diameter on the fatigue properties of 5XXX aluminum alloys lap joints. In particular, the authors determined an optimal pin diameter for a given tool rotational speed in order to maximize the fatigue properties of the joints. Such phenomenon was explained on the basis, on one side, of the necessity to obtain a wide welding region which requires large values of the tool pin diameter. In turn, on the other side, a too large tool pin would cause an excessive increase in the heat input which deteriorates the local micro-structural properties of the welded material. In this way the effective optimization of the process should take into account both geometrical and process parameters causing the best effect of material stirring and heat input.

130

L. Fratini

As far as the author of the present chapter is regarded, at least two contributions to the research debate on lap joints were given in the last years. First of all, process parameters optimization through gradient techniques was proposed [20]. A 3D lap joint made of blanks for automotive applications was considered and the mentioned optimization technique permitted to maximize the joint strength up to 85.5% of the base material resistance. Moreover, the intrinsic variability of the considered process was measured in order to estimate the optimization technique resolution, i.e. the step width of the used algorithm. Furthermore, a systematic study of the process mechanics was recently proposed [21]: the influence of the mutual position of the blanks with respect of the loci of the joint was considered and an insight into the process mechanics was given. Several of the obtained results are recalled in the next sections of the present chapter. Regarding FSW of lap joints, spot welding must also be considered [22]. In Fig. 2 the subsequent steps of the friction stir spot welding process (FSSW) are recalled: as the rotating tool is inserted into the sheets (Fig. 2a) a local backward extrusion mechanics is observed and a full contact between the upper sheet and the tool shoulder is reached. Then, holding the tool into the sheets, a heat flux is generated by the friction forces work and plastic deformation work decaying into heat (Fig. 2b). Finally, after a proper dwelling time, the tool is moved up and the joint is released (Fig. 2c). On this topic several researches were carried out and a large number of papers can be found in literature. Starting from the investigation on the influence of the most relevant process parameters (see for instance [23]) several studies have been focused on the importance of the tool rotating speed and plunging speed but also of the welding time which, in such variation of the FSW technology, assumes a relevant role as it determines the amount of heat conferred to the material. Also, the possibility to use a tool characterized by a retractile pin (see for instance [24]) was investigated. In this way, the typical exit hole left by the tool pin at the end of the process is avoided with obvious benefits to the joint resistance. The author of the present chapter also proposed a modified FSSW process, in which a tool path is given to the tool itself after the completion of the sinking phase (Fig. 2b); in other words, the tool during the process moves around its starting position following a trajectory and causing a definitively larger stirred zone in the material. In this way larger joint strength values are reached [25].

Fig. 2 The FSSW process


131

In aeronautical and aerospace applications—but more, in general, in parts and components of the transportation industries—it is often required to locally enhance the stiffness of the structure through so called skin & stringer joining. In this way, for the development of T joints, two welding seams are usually carried out on the two sides of the stringer part close to the interface with the skin panel. In the last years, T joints were investigated from a few different points of view; for instance, results were presented regarding the fatigue resistance of welded T joints in light weight alloys, in stainless steels and also in composite materials [26–28]. Typically, fusion welding processes were considered and the effects of the localized heating and subsequent rapid cooling on residual stresses and distortions were investigated through different approaches. It should be observed that in such former studies it was highlighted that the failure of the joints was often due to the presence of a wide heat affected zone resulting into local strong metallurgical modifications in the material and discontinuities and defects occurring due to the used fusion welding processes. In this way, on the basis of its peculiarities, once again the application of FSW seems very promising in order to improve the joints effectiveness and mechanical performances. As far as FSW process is concerned the idea is to carry out a welding in transparency, i.e. inserting the tool and in particular its pin from the external surface of the skin (Fig. 3). It should be observed that in order to develop the FSW operations, a properly designed clamping fixture must be used in order to fix the blanks to be welded. As it will be shown in the next paragraph, the role and the geometry of the clamping fixture are very important in terms of the FSW process mechanics since, in this case, a strong forging action is exerted on the material in order to fulfill the die obtaining the final shape of the joint and in particular its fillets. Analyzing the international scientific literature just a few examples of investigation focusing on FSW of T joints are found. The author of the present chapter first carried out some fundamental investigations aimed to compare the FSW technology to the most commonly used fusion welding processes such as MIG [29]. The evolution of the micro-structural changes occurring in the material due to

Fig. 3 The FSW of T joints

132

L. Fratini

the melting and solidifications stages were highlighted in macro image analyses. Furthermore, considering bending actions, the mechanical performances of welded joints, obtained both through MIG and FSW processes, were compared to the ones of extruded profiles of similar material. In this way, the behaviors of the parts were compared both in the elastic and in the plastic fields. Finally, the relevance of the tilt angle was clearly highlighted. It was shown that larger values of the tilt angle with respect to the butt joints conditions must be chosen in the case of T joints in order to avoid flow defects during the process, i.e. tunnels or un-effective bonding phenomena. Further fundamental studies were carried out focusing on the material flow occurring during FSW of T joints [30, 31]. In the next section, the latter topic will be discussed in detail. Here it must be just stressed that the vertical material flow induced by the tool pin is strongly relevant for the effectiveness of the weld carried out. In this way, the highlighting of the actual bonding phenomena occurring at the skin & stringer interface is a key factor in order to determine the set of process parameters able to maximize the mechanical performances of the joint. Furthermore, a research activity was aimed to an electrochemical analysis of the FSW T joints in comparison with the parts obtained through laser welding [32] with specific reference to the cathodic activity of the joints. An important topic regarding the FSW of T joint is the correlation between the mechanical characteristics of the blanks to be welded and the process parameters, basically the tool rotational speed and feed rate, in order to maximize the mechanical performance of the joint. It should be observed that due to the forging mechanics shown by the process, the materials behavior strongly determine the material flow and then the actual bonding between the skin and the stringer [33]. It should be observed that, depending on the mechanical behavior of the involved materials as a function of the temperature, strain and strain rate values, different material flows are observed and as a consequence effective or un-effective FSW processes are obtained. Moreover, real case studies of industrial interest must be considered since, often, different and peculiar process conditions are found [34]. As coated blanks are used in FSW process, the interaction of the coating with the material flow and with the tool pin stirring action has to be properly considered. Actually, some negative effects could be found with a sort of hooking effect of the coating into the flowing material which can give rise to cracks and failures. As far as the automatic seam tracking of FS welded T joints are regarded, a further interesting application is proposed by Fleming et al. [35]. The authors proposed a control technique for the process which makes the tool weave back and forth during the process to maintain effective process conditions, i.e. the location where axial force is greatest, namely the centre of the weld where the stringer is exactly under the tool pin axis. Another recent quite relevant field of interest is the behavior of FS welded structures and panels under fatigue loads or in bucking conditions. Such kind of analyses are definitively of industrial importance in order to predict the behavior of the welded parts and components. Yoon et al. [36], for instance, investigated the effect of the FS weld on the buckling behavior of an integrally stiffened panel


133

using the Riks method based on geometrical imperfection and the RambergOsgood deformation plasticity model for the elasto-plastic material behavior. Different load cases were taken into account. The authors showed that the presence of a FSW reduces the buckling load relative to the non weld case. Also, fatigue behavior of the welded profiles is often investigated (see for instance [37]) considering the actual geometrical features of the real panels of industrial interest. In order to properly introduce some of the results and information shown in the next section of the present chapter, some words must be spent for the most recent advances in the numerical simulation of the FSW process. It should be observed that, first of all, thermal models were proposed, based on analytical heuristic expressions and taking into account the heat generated by both friction force work and the material deformation work [38]. Such models were aimed to highlight the temperature distributions nearby the rotating pin. On the other hand, finite element thermo-mechanical models were presented [39–42] with the aim to investigate the stress and strain distribution during the FSW process and the overall distribution in the workpieces of the most relevant process field variables, namely temperature, strain and strain rate. The author of the present chapter, together with some colleagues and coworkers [43, 44], proposed a continuum based FEM model for FSW process, that is 3D Lagrangian implicit, coupled, rigid-viscoplastic. The model was calibrated by comparison with experimental results of force and temperature distribution; then, it was used to investigate the distribution of the field variables in the heat affect zone and the weld nugget but also the influence of the tool pin shape and of the process parameters on the induced material flow. The numerical results shown in the present chapter were obtained through the latter model. In the next sections of the present chapter, first of all, the macroscopic differences between the process mechanics occurring in FSW of butt joints and in FSW of lap and T ones will be highlighted, also with reference to the induced material flow. Furthermore, some indications will be provided regarding the most relevant geometrical features which must characterize the tools to be used for lap and T joints. Finally, information regarding the plastomechanics of the FSW process in lap and T joints will be added.

2 Differences Between Butt and Other FSW Joints In order to highlight the different process mechanics characterizing FSW of lap and T joints, the material flow occurring in butt joints FSW will be briefly recalled. It should be observed that several studies were carried out in the last years on such topic since it is definitively fundamental for the full comprehension of the FSW process (see for instance [45–47]). In FSW of butt joints, the material, during the tool rotating and feeding movement, flows at the back of the pin from the retreating side towards the advancing side. It should be observed that the material velocity is larger in the upper layers of the blanks due to the action of the tool shoulder.

134

L. Fratini

Fig. 4 The material and thermal flow in FSW of a butt joints, b lap joints and c T joints. The advancing sides (AS) and the retreating ones (RS) are highlighted in the joints [48]

Furthermore, a vertical movement of the material towards the bottom of the joint is observed due to both the tool pin shape and the tilt angle. The material bonds with the undeformed one living in the advancing side along a bonding surface whose trace in a transverse section is represented by an inclined line closer to the welding line at the bottom of the joint. It should be observed that the acquired knowledge on FSW process of butt joints is not immediately extendable to lap and T-joints. The first observation is that in butt joints the surface to be welded is vertical, while in lap and T-joints it is horizontal and placed at the bottom of the top blank to be welded; in this way a totally different material flow is required in order to get effective process conditions (Fig. 4a) [48]. It should be observed then that a proper vertical flow of the material is required in order to obtain an effective bonding between the blanks to be welded, otherwise the needed conditions of pressure and overall metal flow would not be reached at the interface of the blanks to be welded. A further consideration should be carried out referring to the heat source in the FSW process. As it is shown in Fig. 4a–c, while in butt joints the welding line is close to the major heat source, i.e. the contact interface between the tool shoulder and the blanks, both in lap and T-joints the welding line is not strictly close to the heat source. In this way, a stronger heat flux is required in order to locally reach effective bonding conditions in terms of temperature and material plasto-mechanic conditions. In Fig. 5a–c, with reference to AA7075-T6 blanks, the temperature distributions in the transverse sections of a butt, a lap and a T-joint, respectively are shown; in all the three processes a tool rotating speed R of 700 r.p.m. and a tool feed rate of 100 mm/min were used. It should be observed how, for a given


135

Fig. 5 Temperature distribution in the transverse section of a butt joints, b lap joints and c T joints (AA7075-T6, R = 700 r.p.m., Vf = 100 mm/min)

Fig. 6 Sketch of a clamping fixture for FSW of skin & stringer parts

material different temperature distributions are definitively obtained changing the joint configuration. In this way, different bonding conditions are expected for the two blanks to be welded and as a consequence the joints will show different mechanical performances. On the basis of the shown results, it can be argued that the process engineering of FSW strongly depends on the considered joint configuration; in other words, different sets of process parameters are determined maximizing the mechanical performance of butt, lap and T-joints respectively. Furthermore, in the FSW of lap and T-joints, peculiar clamping fixtures are needed in order to assure the mutual positions the blanks to be welded. Such fixtures are quite simple in the case of lap joints since the two blanks must be just clamped in order to avoid any movement during the welding process. In turn, fixtures for FSW of T-joints are made of a die, which is usually used to fix the stringer, and of an upper part in which the skin is clamped. As far as the die is regarded, it is characterized by two radii, one for each side of the joint, corresponding to the radii between skin and stringer in the final welded part (corner fillets). In Fig. 6, the sketch of a fixture to weld prototype parts made of a skin and

136

L. Fratini

a stringer is reported. It should be observed that often such kind of parts are obtained welding an extruded profile, i.e. the stringer, to a flat blank, i.e. the skin. In this way, a properly designed fixture is required in order to allocate the extruded part inside the die during the welding process. Moreover, during the FSW process, the die radii must be filled by the flowing material. Consequently, an actual forging operation is required to force the sheet and the stringer material in fulfilling the radii of the clamping fixture, resulting in the radii of the T-joint. In other words, the material flow induced by the tool in the FSW process must be effective enough to get both the bonding of the two blanks and the fulfillment of the fixture radii. The latter mechanics will be described in details in the next sections of this chapter also with respect to the actual material flow occurring during the process.

3 Tool Geometries for Lap and T Joints As already mentioned in Sect. 1 of the present chapter, the tool pin shape strongly determines the material flow during FSW processes and in this way, the proper choice of the tool geometry strongly affects the final properties of the welded joint and the effectiveness of the material bonding. In the present section, simple shape tool pins will be considered for sake of simplicity even if in Sect. 1 more complex shapes such as the TrivexTM, the Flared-TrifluteTM, the Re-stirTM ones were already cited. It should be observed that even if such complex shapes definitively determine a strong and effective stirring effect on the processed material, they can be scarcely considered in numerical simulations just due to their complex shape which would require highly dense meshes and as a consequence very long computing times. Moreover, considering basic geometries, the most relevant effect of the tool geometry on the process mechanics and then on the effectiveness of the obtained joints is highlighted. Starting from lap joints, several different authors already stressed the importance of the tool and in particular of the tool pin shape for the mechanical resistance of the welded joints [13, 17]. The authors reported importance of a large shoulder in order to obtain a sufficient heat flow and of a threaded pin with the aim to improve the stirring effect. Moreover, a large pin diameter must be used to enlarge the width of the welding seam improving the mechanical performances of the joint. A few different tests were carried out [21] on 3.2 mm thick AA2198-T4 sheets with the aim to highlight the effects of the tool shape on the process mechanics and on the mechanical performances of the obtained joints. For all the welds, an advancing speed equal to 100 mm/min, a 2° tilt angle and a tool shoulder sinking on the top sheet of 0.2 mm were selected. The rotational speed ranged between 500 rpm and 1000 rpm. Every weld was repeated three times and, from each obtained lap joint, specimens were cut for shear load tests. Moreover, from each


137

Fig. 7 Tools used for FSW of AA2198-T4 lap joints [21]

weld further specimens were derived, embedded by hot compression mounting, polished, etched with Keller reagent for 45 s and observed by a light microscope. In particular, three different tool geometries, characterized by a shoulder diameter equal to 20 mm, were tested varying the process parameters: namely a cylindrical (T1), a conical (T2) and a cylindrical-conical (T3) pin tool, were used, as shown in Fig. 7. It should be observed that the geometry of the used tools was chosen on the basis of former experimental activities aimed to investigate each tool configuration. As typical example of the obtained results, in Fig. 8 the macro analysis of the transverse sections of the obtained joints for a rotating speed of 500 r.p.m. is shown. These process parameters were found to be the best, permitting to maximize the mechanical resistance of the joints with the described process conditions. First of all, in Fig. 8a the considered configuration for the carried out joints is shown. In Fig. 8b, corresponding to the weld obtained with the cylindrical pin tool, even if a quite large nugget area (La distance) is observed, a typical tunnel defect is highlighted too, showing the ineffectiveness of the material flow during the welding process. Furthermore, looking at Fig. 8c no macroscopic defects are observed, denoting a significant improvement in the material flow since in this case the conical pin determines a stronger vertical flow of the material which allows a more effective bonding between the two blanks to be welded. In turn, because of its shape, the conical pin is characterized by a smaller section at the sheet-sheet interface, hence the resulting resistant section is smaller, as highlighted in Fig. 8c (Lb \ La). According to these considerations, the T3 tool was used, able to combine the beneficial effects of the T1 and T2 tools (distance Lc & La, Fig. 8d). The above observations are further explained by the carried out numerical analyses. In Fig. 9, the strain rate distribution in a joint transverse section, taken right behind the tool once the process has reached the steady state, is shown again for the R = 500 r.p.m. with reference to T2 and T3 tools. Figure 9 shows that the

138

L. Fratini

Fig. 8 Macro analyses of the transverse sections of AA2198-T4 lap joints (R = 700 r.p.m.) [21]. a The sketch of the FSW of lap joints. b–d Macro images of the transverse sections of the welded joints obtained with T1, T2 and T3 tools, respectively

Fig. 9 Strain rate distribution in a transverse section for a T2 and b T3 case studies (R = 700 rpm) [21]

maximum values are similar for the T2 and T3 tests. Actually, a wider area characterized by large deformations is found when the T3 tool is used due to the wider pin section near the bottom of the joint. In this way, a larger material volume is involved in the material flow resulting in a more effective and homogeneous weld.


139

Fig. 10 Shear tests results, expressed as maximum failure load per mm of welding, as a function of the used tool (T1, T2 and T3) for R = 700 r.p.m

All the reported information and the deduced reasoning were confirmed as the mechanical performances of the welded joints were investigated through shear tests. Figure 10 reports the obtained results in terms of maximum load per mm of welding. It should be observed that T1 tool is significantly less performing than the other two due to the tunnel defect shown in Fig. 8b. Moreover, a strong improvement in the mechanical resistance of the joints was observed using T3; such behavior is fully justified looking again at Fig. 8. As already observed, a relevant difference between the actual welded area was observed between the joints obtained with T2 and the ones welded with T3: the latter pin permits to merge positive effects of the conical pin which increases the material vertical flow and allows to obtain a large nugget area, i.e. a larger welding area. Moreover, a larger material volume is involved in the plastic flow induced by T3 tool pin allowing a more effective bonding mechanics. Overall, the obtained results permit to assess that in order to get mechanical efficiency for lap joints, a proper vertical material flow must be activated and what is more a pin characterized by a large diameter value must be used with the aim to extend as much as possible the width of the actual welding seam. As far as T joints are considered, a few different considerations must be carried out observing the actual process mechanics. It was already underlined that in this case, the FSW process is characterized also by a relevant forging action leading the material to fulfill the die radii forming the final corners between the welded sin and stringer parts. A new family of tools can then be used just aimed to solve the former problem (Fig. 11). In Figs. 12 and 13 the effects, of a classic tool with a conical pin at its end are compared with the action of the double pin tool. It should be observed that with the latter tool the heat flux is generated closer to the interface between the two blanks to be welded improving the bonding conditions during the FSW process. Moreover, the double pin tool allows obtaining a specific increase in the mutual pressure between the blanks which improves the material flow fulfilling the die radii during the process. In this way, a better final surface between the skin and stringer parts can be obtained. It should be observed that the same kind of tool can also be used for FSW of lap joints since the benefits that it offers can be useful also for that configuration.

140

L. Fratini

Fig. 11 Double pin tool

Fig. 12 Sketch of the FSW process for T-joints with a traditional tool (a) and a ‘‘double pin’’ one (b) (transversal section). Retreating side (RS) and advancing one (AS) are highlighted

In Fig. 14 the transverse section of a T joint made of 4 mm thick AA6056 blanks is shown with macro and micro images (200X). The joint was welded with a double pin tool with a 24 mm shoulder diameter, a first cylindrical pin (12 mm diameter, 3.15 mm height) and a second conical pin (3.95 mm height, 5 mm major diameter, 3.8 mm minor diameter), using a tool


141

Fig. 13 Sketch of the FSW process for T-joints with a traditional tool (a) and a ‘‘double pin’’ one (b) (longitudinal section). Trailing edge (TE) and leading one (LE) are highlighted Fig. 14 Transverse section of AA6056 T joint with micro views of the typical process zones: base material (BM), Thermo Mechanical Affected Zone (TMAZ) and upper and lower nugget

142

L. Fratini

rotating speed of 500 r.p.m. and a tool feed rate of 50 mm/s. It should be observed that an effective material flow is highlighted with no flow defects such as tunnels.

4 Plastomechanics of the FSW Process in Lap and T Joints Material flow during FSW processes is probably the most relevant aspect in order to get effective mechanical properties of the welded joint. A detailed observation of the material microstructure in the joint section of two 3.2 mm thick AA2198-T4 sheets allows to highlight the different areas typically characterizing the transverse section of FSW joints (Fig. 15a). It should be observed that, as far as lap joints are regarded, four different geometrical configurations are possible on the basis of the combination of the mutual position of the sheets to be welded and of the tool rotation direction. It should be observed that the four possible configurations can be clustered into

Fig. 15 a Material flow in a typical transverse section of FSW lap joints (T3, 500 r.p.m.); sketch of stresses generated during shear tests: b configuration A and c configuration B. Advancing side (AS) and retreating one (RS) are highlighted [21]


143

two, (A and B in Fig. 15b, c, respectively) on the basis of the plane of symmetry of the joint [21]. What is more, the typical material flow evolutions occurring in the transverse sections of a lap joint are highlighted in the figures. It should be observed that in lap joints, even if due to the mutual position of the two blanks to be welded the welding surface is horizontal, the actual bonding surface after the FSW process is not perfectly horizontal and flat. The material flow rather causes a wavy bonding surface characterized by two separation lines on the two sides of the joints. Such effect is often referred to as hooking effect [13, 17]: as the lap joints undergo shear tests, the terminal parts of the cited separation lines determine the failure of the welding since cracks grow up till the joint loses its load carrying capability. A further consideration can be carried out comparing the two different configurations A and B. In the two cases, the separation lines show the same geometry: in the advancing side of the joint it curves downwards towards the bottom sheet while, and in the retreating side, the separation line moves upwards towards the top sheet. Actually, looking at the mechanical resistance of the lap joints in the two configurations A and B some additional aspect must be underlined (see Fig. 16). In Fig. 15b and c, the qualitative distributions of the normal stress acting on the two blanks of the joints, as they undergo to shear tests, are also highlighted. It should be observed that different stress levels, depending on the joint configuration, act on the same area of a given sheet. In other words, the same section of each blank experiences different levels of stress depending on the joint configuration. The hooking effect must also be considered: in configuration A the hooking effect is so that thicker sections of the blanks are subjected to the largest values of stress. In turn, in configuration B the most relevant mechanical actions occur in the area corresponding to the thinner sections of the blanks. In this way the strong reduction in the mechanical performances of the B configuration lap joints shown in Fig. 16 is justified. The above observations make clear why the mutual position of the blanks to be welded and the process parameters determining the positioning of the advancing and retreating sides in the joint must be carefully selected in order to get the most effective process conditions. As far as T joints are regarded, preliminary tests showed the relevance of the tilt angle for such configuration [29]. Actually, with tool inclinations lower than 3°

600 B-700

A-700

500

[N/mm]

Fig. 16 Shear tests results, expressed as maximum failure load per mm of welding, as a function of the used tool (T1, T2 and T3) for R = 700 r.p.m. considering both configurations A and B

400 300 200 100 0 T1

T2

T3

144

L. Fratini

Fig. 17 Experimental (a) and numerical (b) marker positioning of the advancing and retreating side [31]

often ineffective material flow was observed with an incomplete filling of the die radius in the advancing side of the joint. Furthermore, a few investigations were carried out with the aim to highlight the material flow occurring during the FSW process, using 0.1 mm thick foils of commercially pure copper (99.95%) placed in the joints between the stringer edge and the skin sheet to be welded, as described in [39] for the butt joints. It should be observed that on the basis of the actual material flow induced by the tool action on the blanks, three-dimensional shift of the marker was observed in the transverse section of the joints [31]. First of all, some results regarding the FSW of T joints made of 3 mm thick AA6082-T6 blanks are reported on the basis of both experimental tests and numerical simulations. In Fig. 17, a comparison between the experimental and numerical positions of the marker, i.e. of the boundary nodes in the numerical model, is shown. The reported results are referred to tests carried out with a tool characterized by a conical pin (20°) using a tool rotating speed of 500 r.p.m. and a tool feed rate of 100 mm/min. In the above figure, an important first information on the material flow occurring in the FSW of T-joints is highlighted: at the end of the process, the boundary surface delimiting the two blanks, i.e. the actual bonding surface, is no longer flat. The latter is characterized by a wave with an upper shift of the material of the stringer in the retreating side and a sort of corresponding symmetric downwards shifting of the skin material in the advancing side. What is more, in the advancing side the most evident turbulence effects are observed. The above results


145

Fig. 18 AA2024-T4 and AA7075-T6 T-joint obtained with a conical tool pin and tool rotation speed 340 r.p.m. and tool feed rate of 50 mm/ min [48, 49]. Advancing side (AS) and retreating one (RS) are highlighted

Fig. 19 Positions of two referring nodes, P1 and P2, before and after the tool pin action. Advancing side (AS) and retreating one (RS) are highlighted

become more and more evident if skin and stringer of two different materials are welded together. In Fig. 18, the etched transverse section of a T weld of a AA7075-T6 stringer and a AA2024-T4 skin is shown [49]. Another relevant matter regarding the FSW of T-joints and in particular the material flow occurring during the process is shown in Fig. 19. The positions of two referring nodes of a FEM model are indicated just before and after the tool pin action. The used tool was a very simple one, characterized by a conical pin (4 mm in height and major diameter). The numerical simulation is again an useful tool in order to highlight the material evolutions during the FSW process. The investigated nodes were placed exactly at the corners of the used ‘‘single block’’ FE model on two different planes of the model itself. The points at the corner in the retreating side are pushed away till they come in contact with the fixture fillet. The same mechanics is observed in the advancing side even if the perfect contact is not reached. It should be observed that the tool forging action determining such material flow which fulfills the die radii is fundamental in order to obtain sound components without any defect such as tunnels or underfilling, especially in the advancing side of the joint. In this way, the tool geometry must be

146

L. Fratini

Fig. 20 The nugget area a AA6082-T6, b AA2024-T4 (conical tool pin, tool rotation speed 715 p.m. and tool feed rate of 100 mm/min). Bonding defect is highlighted in the latter picture [33]

chosen and designed with the aim to force as much as possible the material towards the die radii and at the same time to reach the most effective bonding conditions in terms of pressure and temperature at the horizontal interface between the skin and the stringer to be welded. From the above consideration derives that the most effective bonding conditions to be reached at the interface of the blanks strongly depend on the materials to be welded and in particular on their rheological behavior as a function of strain, strain rate and temperature. In Fratini et al. [33], such aspect was investigated in detail through both experiments and numerical simulations considering two different aluminum alloys, namely AA6082-T6 and AA2024-T4, characterized by quite different mechanical response at high temperature, at least in the range of strain rates of interest for the FSW processes. The obtained results showed that, starting from the same process parameters, defects were highlighted for the AA2024-T4 joints while sound welds were obtained with AA6082-T6 blanks. Furthermore, a different position and size for the nugget area was observed (Fig. 20a, b): in the AA6082-T6 joint the recrystallization phenomena resulted in a nugget area extending to the stringer. In turn, in the AA2024-T4 joints transverse section, a smaller recrystallyzed area was found by metallographic inspection with a very limited interaction with the stringer blank, denoting ineffective bonding mechanics and conditions.

5 Conclusions In the present chapter, joint configurations different from the butt one were considered with reference to FSW operations. It should be observed that the process mechanics occurring in FSW processes referred to lap and T geometries are definitively different from the ones characterizing FSW of butt joints. The material flow determining the actual bonding between the two blanks to be welded must be properly investigated in order to


147

obtain sound welded parts. In the present chapter, some general indications were then given with reference to aluminum alloys. It should be observed that such lightweight alloys are characterized by wide range of mechanical rheological behaviors and, in this way a proper engineering of the FSW processes is required especially as lap and T joints are regarded. At the moment the FSW of T-joints out of steel or titanium blanks appears very challenging due to the tools to be used and to the complexity of the clamping fixtures. Further research activity is also aimed to the development of hybrid joints between aluminum alloys and titanium ones or steel grades. Acknowledgments The author wish to express his grateful thanks to Dr. Gianluca Buffa for all his work made in experiments and numerical simulations, to Prof. Rajiv Shivpuri for his help and ideas and finally to Prof. Fabrizio Micari for his constant encouragement and supervising action.

References 1. Thomas, W.M., Nicholas, E.D., Needham, J.C., Murch, M.G., Temple-Smith, P., Dawes, C.J.: Friction Stir Butt Welding (1991). International Patent Application No. PCT/GB92/02203 2. Mishra, R.S., Ma, Z.Y.: Friction stir welding and processing. Mater. Sci. Eng. R Rep. 50, 1–78 (2005) 3. Nandan, R., DebRoy, T., Bhadeshia, H.K.D.H.: Recent advances in friction-stir welding— process, weldment structure and properties. Prog. Mater. Sci. 53, 980–1023 (2008) 4. Thomas, W.M., Johnson, K.I., Wiesner, C.S.: Friction stir welding-recent developments in tool and process technologies. Adv. Eng. Mater. 5(7), 485–490 (2003) 5. Colegrove, P.A., Shercliff, H.R., Hyoe, T.: Development of the TrivexTM friction stir welding tool for making lap welds. In: 5th International Symposium on Friction Stir Welding (2004) 6. Elrefaey, A., Gouda, M., Takahashi, M., Ikeuchi, K.: Characterization of aluminum/steel lap joint by friction stir welding. J. Mater. Eng. Perform. 14(1), 10–17 (2005) 7. Elrefaey, A., Takahashi, M., Ikeuchi, K.: Friction-stir-welded lap joint of aluminum to zinccoated steel. Yosetsu Gakkai Ronbunshu/Q. J. Jpn. Weld. Soc. 23(2), 186–193 (2005) 8. Chen, Y.C., Komazaki, T., Kim, Y.G., Tsumura, T., Nakata, K.: Interface microstructure study of friction stir lap joint of AC4C cast aluminum alloy and zinc-coated steel. Mater. Chem. Phys. 111(2–3), 375–380 (2008) 9. Chen, Y.C., Nakata, K.: Effect of the surface state of steel on the microstructure and mechanical properties of dissimilar metal lap joints of aluminum and steel by friction stir welding. Metall. Mater. Trans. A Phys. Metall. Mater. Sci. 39(8), 1985–1992 (2008) 10. Abdollah-Zadeh, A., Saeid, T., Sazgari, B.: Microstructural and mechanical properties of friction stir welded aluminum/copper lap joints. J. Alloy. Compd. 460(1–2), 535–538 (2008) 11. Saeid, T., Abdollah-zadeh, A., Sazgari, B.: Weldability and mechanical properties of dissimilar aluminum-copper lap joints made by friction stir welding. J. Alloy. Compd. (2009). doi:10.1016/j.jallcom.2009.10.127 12. Lee, C.-Y., Lee, W.-B., Kim, J.-W., Choi, D.-H., Yeon, Y.-M., Jung, S.-B.: Lap joint properties of FSWed dissimilar formed 5052 Al and 6061 Al alloys with different thickness. J. Mater. Sci. 43(9), 3296–3304 (2008) 13. Ericsson, M., Jin, L.-Z., Sandström, R.: Fatigue properties of friction stir overlap welds. Int. J. Fatigue 29, 57–68 (2007) 14. Schmidt, H., Hattel, J., Wert, J.: An analytical model for the heat generation in friction stir welding. Model. Simul. Mater. Sci. Eng. 12, 143–147 (2004)

148

L. Fratini

15. Woo, W., Choo, H., Brown, D.W., Bourke, M.A.M., Feng, Z., David, S.A., Hubbard, C.R., Liaw, P.K.: Deconvoluting the influence of heat and plastic deformation on internal strain generated by friction stir processing. Appl. Phys. Lett. 86, 1–4 (2005) 16. Fersini, D., Pirondi, A.: Fatigue behavior of Al2024-T3 friction stir welded lap joints. Eng. Fract. Mech. 74(4), 468–480 (2007) 17. Fersini, D., Pirondi, A.: Analysis and modeling of fatigue failure of friction stir welded aluminum alloy single-lap joints. Eng. Fract. Mech. 75, 790–803 (2008) 18. Chen, Y.C., Nakata, K.: Effect of tool geometry on microstructure and mechanical properties of friction stir lap welded magnesium alloy and steel. Mater. Design 30(9), 3913–3919 (2008) 19. Kulekci, M.K., Sik, A., Kaluç, E.: Effects of tool rotation and pin diameter on fatigue properties of friction stir welded lap joints. Int. J. Adv. Manuf. Technol. 36, 877–882 (2008) 20. Fratini, L., Corona, V.: Friction stir welding lap joint resistance optimization through gradient techniques. J. Manuf. Sci. Eng. 129, 1–6 (2007) 21. Buffa, G., Campanile, G., Fratini, L., Prisco, A.: Friction stir welding of lap joints: influence of process parameters on the metallurgical and mechanical properties. Mater. Sci. Eng. A 519, 19–26 (2009) 22. Tran, V.-X., Pan, J., Pan, T.: Effects of processing time on strengths and failure modes of dissimilar spot friction welds between aluminum 5754-O and 7075-T6 sheets. J. Mater. Process. Technol. 209(8), 3724–3739 (2009) 23. Pan, T.Y., Joaquin. A., Wilkosz, D.E., Reatherford, L., Nicholson, J.M., Feng, Z., Santella, M.L.: Spot friction welding for aluminum joining. In: Proceedings of 5th International Symposium on Friction Stir Welding, Metz, France (2004), ISBN 1-903761-04-2 24. Alléhaux, D., Marie, F.: Mechanical and corrosion behavior of the 2139 aluminum-copper alloy welded by the Friction Stir Welding using the bobbin tool technique. Mater. Sci. Forum 519–521(2), 1131–1138 (2006) 25. Buffa, G., Fratini, L., Piacentini, M.: On the influence of tool path in friction stir spot welding of aluminum alloys. J. Mater. Process. Technol. 208, 309–317 (2008) 26. Mashiri, F.R., Zhao, X.L., Grundy, P.: Stress concentration factors and fatigue behaviour of welded thin-walled CHS-SHS T-joints in-plane bending. Eng. Struct. 26, 1861–1865 (2004) 27. Carpinteri, A., Birghenti, R., Huth, H., Vantadori, S.: Fatigue growth of a surface crack in a welded T-joint. Int. J. Fatigue 27, 59–69 (2005) 28. Shaikh, H., Khatak, H.S., Mahendran, N., Sethi, V.K.: Failure analysis of a T-joint of AISI type 316L stainless steel. Eng. Fail. Anal. 10, 113–118 (2003) 29. Fratini, L., Buffa, G., Filice, L., Gagliardi, F.: FSW of AA6082-T6 T-joints: process engineering and performance measurement. J. Eng. Manuf. B 220(5), 669–676 (2006) 30. Fratini, L., Micari, F., Squillace, A., Giorleo, G. Experimental characterization of FSW T-joints of light alloys. Key Eng. Mater. 344, 751–758 (2007) 31. Fratini, L., Buffa, G., Micari, F., Shivpuri. R.: On the material flow in FSW of T-joints: influence of geometrical and technological parameters. Int. J. Adv. Manuf. Technol. 44, 570–578 (2009) 32. Padovani, C., Fratini, L., Squillace, A., Bellucci, F.: Electrochemical analysis on friction stir welded and laser welded 6XXX aluminium alloys T-joints. Corros. Rev. 25(3–4), 475–489 (2007) 33. Fratini, L., Buffa, G., Shivpuri, R.: Influence of material characteristics on plastomechanics of the FSW process for T-joints. Mater. Design 30, 2435–2445 (2009) 34. Acerra, F., Buffa, G., Fratini, L., Troiano, G.: On the FSW of AA2024-T4 and AA7075-T6 Tjoints: an industrial case study. Int. J. Adv. Manuf. Technol. (2009). doi: 10.1007/ s00170-009-2344-9 35. Fleming, P.A., Hendricks, C.E., Wilkes, D.M., Cook, G.E., Strauss, A.M.: Automatic seamtracking of friction stir welded T-joints. Int. J. Adv. Manuf. Technol. (2009). doi: 10.1007/ s00170-009-1990-2 36. Yoon, J.W., Bray, G.H., Valente, R.A.F., Childs, T.E.R.: Buckling analysis for an integrally stiffened panel structure with a friction stir weld. Thin Walled Struct. 47(12), 1608–1622 (2009)


149

37. Shahri, M.M., Sandström, R.: Fatigue analysis of friction stir welded aluminum profile using critical distance. Int. J. Fatigue 32, 302–309 (2010) 38. Song, M., Kovacevic, R.: Thermal modeling of friction stir welding in a moving coordinate system and its validation. Int. J. Mach. Tools Manuf. 43, 605–615 (2003) 39. Zhang, Z., Chen, J.T.: The simulation of material behaviors in friction stir welding process by using rate-dependent constitutive model. J. Mater. Sci. 43(1), 222–232 (2008) 40. Lorrain, O., Serri, J., Favier, V., Zahrouni, H., Hadrouz, M.: A contribution to a critical review of friction stir welding numerical simulation. J. Mech. Mater. Struct. 4(2), 351–369 (2009) 41. Assidi, M., Fourment, L., Guerdoux, S., Nelson, T.: Friction model for friction stir welding process simulation: Calibrations from welding experiments. Int. J. Mach. Tools Manuf. 50(2), 143–155 (2010) 42. Butan, D., Monaghan, J.: Thermomechanical modelling friction stir welding aluminium 2024-T3. Int. J. Comput. Mater. Sci. Surf. Eng. 2(1–2), 63–72 (2009) 43. Buffa, G., Hua, J., Shivpuri, R., Fratini, L.: A continuum based FEM model for friction stir welding—model development. Mater. Sci. Eng. A 419(1–2), 389–396 (2006) 44. Buffa, G., Hua, J., Shivpuri, R., Fratini, L.: Design of the friction stir welding tool using the continuum based FEM model. Mater. Sci. Eng. A 419(1–2), 381–388 (2006) 45. Fratini, L., Buffa, G., Palmeri, D., Hua, J., Shivpuri, R.: Material flow in FSW of AA7075-T6 butt joints: numerical simulations and experimental verifications. Sci. Technol. Weld. Join. 11(4), 412–421 (2006) 46. Xu, S., Deng, X.: A study of texture patterns in friction stir welds. Acta Mater. 56, 1326–1341 (2008) 47. Hamilton, C., Dymek, S., Blicharski, M.: A model of material flow during friction stir welding. Mater. Charact. 59(9), 1206–1214 (2008) 48. Buffa, G., Fratini, L., Ruisi V.F.: Friction stir welding of tailored joints for industrial applications. In: Proceedings of Esaform 2009 Conference, Lyon, France (2009) 49. Buffa, G., Fratini, L., Lo Monaco, L.: Improved FE model for the simulation of friction stir welding of different materials. Sci. Technol. Weld. Join. 15(3), 199–207 (2010)

Lightweight Stiffened Panels Fabricated Using Emerging Fabrication Technologies: Fatigue Behaviour P. M. G. P. Moreira, V. Richter-Trummer and P. M. S. T. de Castro

Abstract The need for lower cost and the emergence of new welding technologies has brought interest in large integral metallic structures for aircraft applications; however, new problems must be addressed, e.g. in integral structures, a crack approaching a stiffener propagates simultaneously in the skin and into the stiffener and breaks it. The use of manufacturing techniques such as high speed machining (HSM), laser beam welding (LBW) and friction stir welding (FSW) requires further experimental and numerical work concerning the fatigue behaviour of panels manufactured using those processes. This chapter is focused on an experimental test programme including fatigue crack growth rate characterization in panels fabricated using HSM, LBW and FSW. The work was developed in the frame of the European Union DaToN project. Data was obtained for panels tested in mode I crack propagation under load ratios (R) of 0.1 and 0.5. It was found that welded panels presented longer lives up to rupture. This result is associated to the residual stress fields existing in the welded panels, and also to the location of the initial artificial defect, placed in the skin midway the specimen’s two stiffeners.

P. M. G. P. Moreira (&) INEGI, Institute of Mechanical Engineering and Industrial Management, Rua Dr. Roberto Frias 400, 4200-465 Porto, Portugal e-mail: [email protected] V. Richter-Trummer P. M. S. T. de Castro Faculdade de Engenharia da Universidade do Porto, and IDMEC-Porto, Rua Dr. Roberto Frias, 4200-465 Porto, Portugal

Adv Struct Mater (2012) 8: 151–172 DOI: 10.1007/8611_2010_49 Springer-Verlag Berlin Heidelberg 2011 Published Online: 12 January 2011

151

152

P. M. G. P. Moreira et al.

1 Introduction Minimum weight is a major concern in aircraft design, [1]. Because of interest in integral structures, it has become increasingly important to develop methodologies to predict failure in fatigue damaged fuselage structures, [2], since the fuselage is supposed to sustain cracks safely until it is repaired or its economic service life has expired—damage tolerant design principle. Strength assessment of the structures is necessary for their in-service inspection, repair and health monitoring, [3]. Recently, studies are being conducted to validate monolithic designs aiming at equal or better performance than conventional designs with regard to weight and structural integrity, while achieving a significant reduction in manufacturing costs, [4]. An aircraft fuselage structure includes, among other parts, the external skin and longitudinal stiffeners (stringers and longerons) [5]. Stiffened panels are light and highly resistant metal sheets reinforced by stringers designed to cope with a variety of loading conditions. Stiffeners improve the strength and stability of the structure and provide a means of slowing down or arresting the growth of cracks in the panel. Most common stiffener cross-sections are bulb, flat bar or T- and L-sections, that can be bonded, extruded, connected by means of fasteners, machined or welded to form a panel. When experimentally testing stiffened panels, attention should be given to the loading and boundary conditions to ensure that the behaviour of the panel in the complete structure is reproduced, [6]. The skin structure of a pressurized fuselage for transport aircraft is fatigue sensitive. The residual strength concept permits the determination of the maximum crack length that can be safely sustained. With this information and the characterization of the crack growth behaviour of the material, the number of loading cycles that will be necessary for the crack to grow up to its critical length can be estimated in order to ensure safe operation [7]. The development of numerical methodologies with the help of small laboratory coupon test results should be used to predict the residual strength of complex built-up aircraft fuselage structures [8]. Riveted and bolted stiffeners tend to remain intact as the crack propagates under them, providing an alternative path for the panel load to pass. Also, riveted stiffeners continue to limit crack growth after the crack propagates past the stiffener since a crack cannot propagate directly into the stiffener. The permanent need for low cost and the emergence of new technologies has brought interest in large integral metallic structures for aircraft applications. Evaluative programs for replacement of traditional fastening with these new emerging technologies have been carried out all over the aircraft sector, e.g. [9]. In an integral stiffener (machined, extruded or welded) a crack propagates simultaneously in the stiffener and in the skin beyond the stiffener at approximately the same rate. Work in monolithic integral structures, albeit in steel [10–12], indicates that the crack may propagate into and break the stiffener. In [13] it was observed that the rate of crack growth is significantly reduced in the presence of stiffeners. There is an urgent pressure from the manufacturing side in the aerospace industry to apply advanced structural concepts, since they promise considerable

Lightweight Stiffened Panels Fabricated Using Emerging Fabrication Technologies

153

cost and production time benefits, producing, in addition, a smaller number of fatigue and corrosion critical locations. The main drawback of integrally stiffened structures is the damage tolerance behaviour. Such design behaves totally different from the differential designs created by using riveted stiffeners. The prime problem is the crack arresting capability of the stiffeners both in fatigue crack growth as well as in residual strength. In the literature, only a few studies on the mechanical behaviour of welded aeronautical stiffened panels can be found. As an example, in a work by Murphy et al. [14, 15] the static strength performance of friction stir welding assemble methods on stiffened panels was studied. It was observed that welding can introduce undesirable effects due to residual stress. Uz et al. [16], presented a work based on fatigue crack propagation tests of LBW stiffened panels compared with panels with crenellations to improve damage tolerance behaviour. The present chapter presents work which is part of the European Union FP6 project DaToN. This project aimed to increase the understanding of the fatigue phenomena of integrally stiffened structures produced by means of three novel production methods: Laser Beam and Friction Stir Welding (LBW and FSW) and High Speed Machining (HSM). Reviews of these technologies can be found in [17–21] concerning FSW [22] concerning LBW and [23] concerning HSM. The main focus of this work was the development of an experimental test programme, which included fatigue crack growth rate of HSM, LBW and FSW stiffened panels. Also, a preliminary numerical analysis of the integral specimen was performed in order to understand the mechanical behaviour of this type of specimen geometry.

2 Finite Element Analysis of the Selected Panel Geometry The first task in this study was the finite element method (FEM) analysis of the stiffened panel geometry that was defined by the project guidelines. The geometry of the specimen studied in the experimental component of the DaToN project is presented in Fig. 1. This specimen geometry is not intended to represent a particular location of an aircraft structure; it intends to be a generic specimen allowing the study of new manufacturing techniques for stiffened aeronautical joints. A three-dimensional (3D) stress analysis of the specimen was done using the FEM in ABAQUS [24]. The centre of gravity of the crack plane cross section is located 2.76 mm above the specimen front face. In all the analyses carried out the load was applied aligned with the centre of gravity. The remote load chosen for all analyses corresponds to a 100 MPa uniformly distributed nominal stress. Two distinct situations were analysed: stiffened panel without and with a central crack. In the following analyses x (and 1) is the coordinate axis in the thickness direction, y (and 2) is the coordinate axis in the loading (longitudinal) direction, and z (and 3) is the coordinate axis in the transversal direction. The specimen side

154


Fig. 1 Specimen geometry used in the experimental component of the DaToN project, [25]

containing the stiffeners is named back side, whereas the opposite side is named front side. 8-nodes brick elements (C3D8) and 6-nodes brick elements (C3D6) were used to model the specimen. These elements use linear interpolation in each direction and are often called linear or first-order elements. A total of 60,083 elements was used to model half of the symmetric stiffened panel. The deformed 3D FEM model, that presents the stress in the load direction, ry throughout the un-cracked stiffened panel, is shown in Fig. 2. In this figure, displacements were enlarged (deformation scale factor of 20) in post-processing of the FEM analysis. In order to assess the specimen overall behaviour when cracked, particularly as concerns stress distribution and displacements especially those perpendicular to the specimen plane, a crack with length 2a = 55.39 mm was also modelled in the centre of the specimen. The detail of ry stress distribution in the cracked specimen middle cross section is presented in Fig. 3. The ry stress distribution and displacements in the x direction were analysed along the nodes on the side of the plate containing the stiffeners and on the opposite side, in the direction of the arrow (b) plotted in this figure. The evolution of ry stress along the nodes that lay on the arrow (a), for the case of the un-cracked and cracked panels, is presented in Fig. 4. For the un-cracked panel, the stress values are higher in the plate and decrease through the stiffener moving away from the plate, as indicated by the arrow. The highest and lowest ry stress values along this line are 112.2 and 17.5 MPa respectively. The introduction of a crack in the stiffened panel leads to an increase of the stress values in the plate and a decrease at the top of the stiffener.


155

Fig. 2 Deformed model and stress distribution on the un-cracked stiffened plate, stress in the load direction

The ry stress distribution in the specimen middle cross section, with and without a crack, in the front and back layers (i.e., side containing the stiffener, and opposite side), is presented in Fig. 5. In the lateral areas ry stress is similar in the panel with and without a crack. It is also possible to identify the change in the stress distribution due to the crack. It was also verified that the maximum displacement value found in the cracked panel in the perpendicular direction to the

Fig. 3 Detail of specimen middle cross section, stiffened panel with a crack

156


Fig. 4 Stress distribution along the nodes that lay on the arrow (a) presented in Fig. 3

Fig. 5 ry distribution in the specimen middle cross section in the front and back layer nodes

specimen surface is 2.95 mm and occurs at the middle of the panel. In the case of an un-cracked panel the corresponding displacement was 2.77 mm. The ry stress distribution along the specimen longitudinal direction is presented in Fig. 2. In this figure three lines—(a) to (c)—are marked. The ry stress and displacements in the x direction were determined along these lines. Results presented for lines b) and c) were obtained in the plate side opposite to the stiffeners (front side). The ry stress distribution throughout the specimen longitudinal direction in the stiffener top surface (a), panel lateral surface (b) and panel longitudinal central line (c); as presented in Fig. 2 is presented in Fig. 6. For the un-cracked panel, the higher stress values are found in the centre of the plate, but they are of the same magnitude as those found in the side layer. Stress values on the stiffener top surface are near 18 MPa, except in the stiffener end where some low compressive values are found. When a crack is present there is a


157

Fig. 6 ry distribution through the specimen longitudinal direction in stiffener top surface (a), lateral surface (b) and centre layer (c), as presented in Fig. 2. Stiffened panel with and without a crack

decrease of the ry stress value in the stiffener near the crack, but in the remaining stiffener ry stress has similar values as in the un-cracked panel. In the case of the cracked panel, ry stress is similar to the un-cracked panel, except in the middle plane near the crack face where ry has zero or very low values as expected.

3 DaToN Stiffened Panels Fatigue Life; Experimental Measurements The present work is mainly focused on the analysis of fatigue test results of twostiffener specimens (Fig. 1) manufactured by three different processes, HSM (High Speed Machining), LBW (Laser Beam Welding) and FSW (Friction Stir Welding). The stiffened panels were manufactured using AA6056, a modified variant of AA6013, which is considered a promising airframe material candidate for processing by fusion laser beam welding and solid-state friction stir welding. AA6056 is an Al–Mg–Si–Cu alloy that can be heat treated to different strength levels by precipitation hardening and has a good corrosion resistance. Further to the HSM specimens, two different welding techniques were used to manufacture the remaining panels. LBW panels were welded in two different configurations (LBW1 and LBW2), as presented in Fig. 7a and b. The main difference between LBW1 and LBW2 is the position of the weld bead. In the LBW1 configuration the weldment is at the junction of the skin with the blade (T-joint); in the LBW2 configuration the weldment is at the lower part of the stringer web, 1 mm above the skin (butt-joint). FSW panels welding configuration, where the tool shoulder was positioned in the specimen opposite surface to the stiffener, is presented in Fig. 7c.

158 Fig. 7 Welding configurations for the welded specimens. a LBW1 configuration, b LBW2 configuration, c FSW configuration


Lightweight Stiffened Panels Fabricated Using Emerging Fabrication Technologies Table 1 Specimens tested Process Material HSM

AA6056-T651

LBW1

AA6056 PWHT–T6

LBW2

AA6056 PWHT–T6 AA6056-T6 as-welded

FSW

AA6056-T4 PWHT–T6

159

Maximum stress [MPa]

R

Specimen

80 110 80 110 80 110 80 110 80 110

0.1 0.5 0.1 0.5 0.1 0.5 0.1 0.5 0.1 0.5

HSM01 HSM02 LBW03 LBW04 LBW05 LBW06 LBW07 LBW08 FSW09 FSW10

A total of ten specimens was tested, Table 1. Measurements of crack length in all specimens were performed according to the scheme presented in Fig. 8. The specimen front and back sides are the sides of the plate without and with stiffeners, respectively. For all specimens tested, the strain distribution for static and fatigue loading was recorded and the crack growth rate was measured. An initial central through the thickness notch 20 mm long and 0.2 mm wide was created by electrodischarge machining (EDM). The two tips of this notch were the crack initiation locations. This initial crack location allows studying the influence of the stiffener, and its manufacturing processes, on the crack growth behaviour, when a crack initiates midway between the stiffeners. This 2-stiffener geometry is not intended for studying the effects of a crack initiating in the stiffener region.

3.1 Base Material Tensile and Crack Growth Tests Tensile and compact tension (CT) specimens were machined from the 4 mm thick part of the stiffened panels, see Fig. 1. Base material characterization was presented in [26]. Table 2 presents basic tensile test data.

Fig. 8 Fatigue crack measurement scheme

160


Table 2 Tensile test results of DaToN specimens base material ryield [MPa] Elongation [%] rrupt [MPa]

E [GPa]

HSM 6056-T651 LBW 6056 PWHT-T6 LBW 6056-T6 as-welded FSW 6056 PWHT-T6

70.8 67.0 72.4 69.7

381.3 346.9 350.3 352.2

357.5 335.3 341.3 341.3

9.0 11.6 11.1 10.3

CT specimens were used for characterization of the crack growth law for AA6056 material in both conditions for both R values. Results are shown in Fig. 9. The results obtained are in accordance with values found by Vaidya et al. [27] testing AA6056-T6 specimens with a thickness of 4 mm. The results could be successfully fitted with a power law with their coefficients C and m being the parameters of the Paris law, da ¼ C DK m dN

ð1Þ

where da/dN is the fatigue crack growth rate, DK is the stress intensity factor range (DK = Kmax - Kmin), and C and m are material constants dependent upon environment, frequency, temperature and stress ratio among other factors. The parameters C and m obtained for each type of material are presented in Table 3. It was verified that for both material conditions the specimens with R = 0.5 presented a higher crack growth rate for the same DK value, as expected. Also, when tested at the same R value the specimens extracted from the LBW panels have higher crack growth rate than those extracted from HSM panels. This may be explained by the origin of the specimens. While the LBW specimens were rolled plates, the HSM plates where machined from 40 mm thick blocks.

Fig. 9 Crack propagation data obtained with CT specimens from the HSM AA6056-T651 and from the LBW AA6056-T6 panels, [26]


161

Table 3 C and m Paris law parameters for the AA6056 material from the HSM and LBW panels h i Material from the HSM panel Material from the LBW panel Nmm1;5 mm=cycle AA6056-T651 AA6056-T6 C m

R = 0.1

R = 0.5

R = 0.1

R = 0.5

2.32E-12 2.92

9.98E-10 2.06

1.37E-11 2.74

6.36E-11 2.57

3.2 High Speed Machining AA6056 Panels Since the objective of this work was to evaluate the influence of the new welding processes on the fatigue behaviour of stiffened panels, the HSM specimens were used as reference. HSM is a result of advances in cutting tool and machining tool technologies, being associated with high-speed spindles (15,000–40,000 rpm) and higher feed rates when compared to most common milling operations. Its major characteristics are high metal removal rates, low cutting forces, minimal workpiece distortion, and the ability to machine thin walled constructions. Two stiffened HSM panels of aluminium AA6056-T651 were fatigue tested at a maximum stress of 80 MPa with R = 0.1 (specimen HSM01) and at a maximum stress of 110 MPa with R = 0.5 (specimen HSM02). In this text, more complete details of the test procedure are given in the case of these HSM specimens. Testing of the remaining specimens, although also covering all aspects dealt with for HSM, is recorded here giving only the a vs. N data, which was indeed the main testing objective. In order to evaluate the quality of the gripping system used, two panels with a central notch of 20 mm length were instrumented and loaded at five incremental load levels to acquire the stress distribution at specific sites of the specimen. The strain gauges were distributed in the specimen according to the scheme presented in Fig. 10: two couples (C1, C5, C6 and C7), on both faces of the panel, bonded on

Fig. 10 DaToN panel strain gauges location (front gauges in brackets, see Fig. 8)

162


the skin at the centre of each bay (spaced 225 mm in horizontal direction), on the horizontal symmetry plane; another couple (C3 and C8), placed on the longitudinal axis of the panel, 200 mm above the horizontal symmetry plane; two couples (C2, C4, C7 and C9), placed in correspondence of a stringer, with a strain gauge bonded on top of the stringer and the other one on the skin. An accurate symmetry of load distribution along the specimen width was identified. At the specimen horizontal middle line, the stiffener top surface has the lower stress values. A value of 70 GPa was used for the Young modulus in order to convert strain to stress. Fatigue crack propagation tests were carried out and strain was measured in order to understand the different load transfer stages. For the specimen HSM01, strain gauge values were measured in periodic stops of the fatigue test at the average fatigue load (44 MPa, 47.98 kN). The strain distribution on the stiffened panel during the fatigue test is presented in Fig. 11. Taking into account that C2 and C4 correspond to gauges at the top of the stiffener and C7 and C9 to gauges at the same width but placed in the plate front side it is easily identified that when the crack is near and reaches the stiffener most of the load is transmitted through the plate front side. Fatigue crack propagation tests were carried out and crack length was measured at periodic stops of the fatigue test. Specimen HSM01, tested at a maximum stress of 80 MPa and R = 0.1, had a fatigue life of 113,784 cycles. The fatigue crack growth in the stiffened panel during the fatigue test is presented in Fig. 12. The first fatigue crack was only detected at 15,000 cycles. The crack started to grow through the stiffener at 109,800 cycles, 96.5% of the total fatigue life. The crack in the left stiffener bifurcated at nearly 113,000 cycles. For specimen HSM02 (rmax = 110 MPa, R = 0.5) the strain gauges’ values were measured in periodic stops of the fatigue test at the maximum fatigue load (119.95 kN). Again, when the crack grows through the stiffener most of the load is carried by the plate front side. This specimen had a fatigue life of 117,744 cycles. The first fatigue crack was first detected at 7,500 cycles. The crack started to grow through the stiffener at 113,000 cycles, 96.0% of the total fatigue life.

Fig. 11 Strain distribution during the fatigue crack growth test, HSM01 (rmax = 80 MPa, R = 0.1)


163

Fig. 12 Fatigue crack growth test, specimen HSM01

After 115,000 cycles the fatigue crack in both stiffeners bifurcated and started to propagate parallel to the panel along the stiffener. Photographs were taken to record particular features during the tests. The complete test setup for specimens HSM01 and HSM02 is presented Fig. 13. It is possible to identify the gripping system, the instrumented specimen, and the travelling microscopes used for crack length measurements. Fig. 13 Test setup, specimen HSM02

164 Fig. 14 Cracks at the stiffeners (113,700 cycles), specimen HSM01. a Bifurcation of the crack at the left side stiffener, b fatigue crack propagation at the right stiffener



165

Fig. 15 Fatigue crack surface for HSM01

In Fig. 14a the bifurcation of the crack at the left side stiffener is presented. The fatigue crack propagation at the right stiffener is presented in Fig. 14b. The same fatigue crack bifurcation appeared in the HSM02 left and right stiffeners. In the right stiffener the bifurcation occurred for a shorter crack length than in the left stiffener. It should be emphasized that this bifurcation phenomenon occurred at the very end of the cyclic test. A detail of the fatigue crack surface in a stiffener of specimen HSM01 is presented in Fig. 15. A rough surface on the entire fractured area was found.

3.3 Laser Beam Welded AA6056 2-Stiffener Panels Six laser beam welded panels of aluminium AA6056 were fatigue tested. Half of the panels were tested at a maximum stress of 80 MPa with R = 0.1 and the remaining at 110 MPa and R = 0.5. Panels with two different heat treatment conditions were tested: 1. after the welding procedure panels were submitted to an aging treatment T6 (PWHT) which corresponds to 4 h at 190C. The machining of the panels has been performed on T4 tempered 5 mm thickness sheet; 2. another set of panels was previously heat treated to the condition T6 and than tested (as-welded). Two different welding configurations were analyzed (LBW1 and LBW2), as presented in Fig. 7. The main difference between LBW1 and LBW2 is the position of the weld bead. In the LBW1 configuration the weldment is at the junction of the skin with the blade (T-joint); in the LBW2 configuration the weldment is at the lower part of the stringer web, 1 mm above the skin (butt-joint). In subsequent sections of this paper, overall comparative plots of the crack growth data obtained for the remaining specimens will be given.

166


3.4 Friction Stir Welding AA6056 2-Stiffener Panels Two stiffened friction stir welded panels of aluminium AA6056 PWHT-T6 were fatigue tested: (1) at a maximum stress of 80 MPa with R = 0.1; (2) and at a maximum stress of 110 MPa with R = 0.5. Panels were welded in the T4 condition, and after welding were aged to achieve the T6 condition. These panels presented the highest distortion when compared with the HSM and LBW panels, Fig. 16. Again, a vs. N data for all specimens tested is given in subsequent sections, allowing for comparisons to be made.

3.5 Discussion The crack length measurements at the stiffeners location were difficult to execute. A solution was found using a photographic camera which permitted crack growth reconstitution. In all specimens the crack grew symmetrically at both sides, as shown by the plots of the crack size at both sides and both surfaces of the stiffener. This feature validates the geometry of the gripping system developed for these tests, see Fig. 17. A comparison of the specimens tested with R = 0.1 and R = 0.5 (HSM, LBW1 PWHT-T6. LBW2 (as-welded and PWHT-T6) and FSW PWHT-T6) is presented in Fig. 18. In these figures the initiation of fatigue cracks from the initial electro discharge machined notch is not considered, and a vs. N data for all specimens is plotted starting at the same crack length (a = 12 mm). For both R values, the

Fig. 16 Distortion detail of the HSM, LBW and FSW panels. a HSM panel, b LBW panel, c FSW panel


167

Fig. 17 Specimen grip system

HSM specimen presented the lower fatigue lives. At the opposite, the PWHT-T6 specimens tested in the LBW2 configuration (butt joint) presented the higher fatigue lives for both R ratios. The FSW specimen tested at R = 0.1 presented a fatigue life similar to the LBW2 as-welded specimen. For R = 0.5 the FSW specimen performed better than the LBW1 PWHT-T6 specimen and worse than the LBW2 as-welded specimen. In all specimens tested it was found that the crack arrest feature (decrease of crack growth rate) introduced by the stiffener was not significant, probably due to the low width of these specimens. Nevertheless, when the stiffeners are fractured the remaining life of the specimen is marginal.

4 Discussion Mechanically fastened panels are desirable in terms of fail-safety criterion since the stringers are effective crack stoppers. However, whereas riveted stiffeners continue to limit crack growth after the crack propagates past the stiffener, the effect of a welded or integral stiffener is quite different, since the crack may propagate directly into an integral stiffener and completely break it, e.g. [11, 12, 28, 29]. Welded panels presented longer lives up to rupture, implying that during most of their fatigue testing the crack growth rates were smaller than with HSM panels. This somewhat unexpected result—fatigue lives of welded components longer than fatigue lives of monolithic machined similar components—is certainly associated to the residual stress fields existing in the welded panels, and also to the location of the initial artificial defect, placed in the skin precisely at mid distance between the two stiffeners.

168


Fig. 18 Comparison of a-N for all specimens tested. a Specimens tested at R = 0.1, b specimens tested at R = 0.5

As stated in [12], welded stiffeners affect skin crack growth in a unique way because of residual stresses. In the vicinity of the stiffeners, welding is expected to create predominantly tensile residual stresses that may be of the order of the material yield stress, whereas lower-level compressive stress will occur in the skin between the stiffeners in order to guarantee the necessary residual stress


169

equilibrium on the specimen cross section. Therefore, such a schematic residual stress distribution will increase crack growth rates near the stiffeners, and decrease crack growth rates between them. A contribution to explain the slower crack growth observed in the LBW welded stiffened panels may therefore consist in the crack growth retardation caused by the compressive residual stress field in the region between the stiffeners. Of course the reverse trend is expected when the crack tip finds the high positive residual stresses close to the stiffeners. There, the crack growth rate is expected to accelerate. This justifies the fact that more than 85% of a specimen fatigue life occurs until the crack reaches the 3 mm thickness zone. If the starting artificial crack was located in a region of positive residual stress—skin under a stiffener—the reverse effect should be expected and a lower fatigue life would result from accelerated crack growth. The complex testing of the DaToN panels raised a question concerning the choice of their geometry. It was found that, with these two-stiffener HSM and welded panels, the crack growth rate when the crack tip reaches the stiffener is already very large, implying that there is almost no noticeable crack slow down effect due to the stiffeners. As suggested in [30], a systematic study of this ratio should be carried out in order to identify its optimum value. Also, the shape of stiffener cross section is likely to influence the fatigue behaviour, and related research efforts are already documented, [28], continuing the exploration of failsafety design options for integral/welded stringer panels.

5 Developments Studies exist in literature which are developments of the experimental work presented in this text. In [31, 32], the stress intensity factor calibration of the DaToN stiffened panel geometry in the absence of residual stresses was obtained using a finite element model. That result shows a local decrease of stress intensity factor (SIF) in the region of the stiffener. Nevertheless, when cracks approach the stiffeners the SIF decreases, but there is no clear slowing down of the crack propagation. A reasonable agreement between measured and predicted a vs. N is shown in the case of HSM specimens, for which the available SIF calibration is applicable. Since a vs. N predictions for welded specimens would require consideration of residual stress fields, in [26, 33] further developments were made in the model to take into account the residual stress fields due to the welding processes. Reasonable agreement between the experimental work and numerical predictions for LBW and FSW specimens was found. Scanning electron-microscopy (SEM) analysis of these specimens was conducted in [34, 35] in order to perform a fractographic reconstitution of the fatigue crack growth rate during the fatigue tests. Good agreement between fatigue striation analysis and the macrographic fatigue crack measurements presented in this chapter was found.

170


In [36], panels were reinforced with carbon-epoxy or boron-epoxy strips in order to analyse the crack arrest feature of the stiffeners. A large potential for increasing the damage tolerance life of the panels was identified.

6 Conclusions As a result of this work: • Welded panels presented longer lives up to rupture, implying that during most of their fatigue testing the crack growth rates were smaller than with HSM panels. This somewhat unexpected result is certainly associated with the residual stress fields existing in the welded panels, and also with the location of the initial artificial defect, placed in the skin precisely in the mid distance between both stiffeners. • When fatigue testing the DaToN stiffened panels it was found that, although when cracks approach the stiffeners the stress intensity factor decreases, there is no clear slowing down of the crack propagation. Acknowledgments This research is part of IDMEC-Porto contribution to the EU project DaToN contract FP6-516053. The authors acknowledge the collaboration of Mr. Miguel A. V. de Figueiredo and Mr. Rui Silva in the experiments. Dr. Moreira acknowledges POPH—QRENTipologia 4.2—Promotion of scientific employment funded by the ESF and MCTES.

References 1. Paik, J., van der Veen, S., Duran, A., Collette, M.: Ultimate compressive strength design methods of aluminum welded stiffened panel structures for aerospace, marine and land-based applications: a benchmark study. Thin Walled Struct. 43(10), 1550–1566 (2005) 2. Wen, P., Aliabadi, M., Young, A.: Fracture mechanics analysis of curved stiffened panels using BEM. Int. J. Solids Struct. 40(1), 219–236 (2003) 3. Murthy, A., Palani, G., Iyer, N.: Remaining life prediction of cracked stiffened panels under constant and variable amplitude loading. Int. J. Fatigue. 29(6), 1125–1139 (2007) 4. Pettit, R., Wang, J., Toh, C.: Validated feasibility study of integrally stiffened metallic fuselage panels for reducing manufacturing costs. In NASA/CR-2000-209342 (2000) 5. Murphy, A., Price, M., Lynch, C., Gibson, A.: The computational post-buckling analysis of fuselage stiffened panels loaded in shear. Thin Walled Struct. 43(9), 1455–1474 (2005) 6. Aalberg, A., Langseth, M., Larsen, P.: Stiffened aluminium panels subjected to axial compression. Thin Walled Struct. 39(10), 861–885 (2001) 7. Salgado, N., Aliabadi, M.: The application of the dual boundary element method to the analysis of cracked stiffened panels. Eng. Fract. Mech. 54(1), 91–105 (1996) 8. Seshadri, B., Newman, J., Dawicke, D.: Residual strength analyses of stiffened and unstiffened panels—Part II: wide panels. Eng. Fract. Mech. 70(3–4), 509–524 (2003) 9. Hoffman, E., Harley, R., Wagner, J., Jegley, D., Pecquet, R., Blum, C., Arbegast, W.: Compression buckling behavior of large-scale friction stir welded and riveted 2090-T83 A1Li alloy skin-stiffener panels. In: NASA/TM-2002-211770 (2002)


171

10. Mahmoud, H., Dexter, R.: Propagation rate of large cracks in stiffened panels under tension loading. Marine Struct. 18(3), 265–288 (2005) 11. Dexter, R.J., Pilarski, P.J., Mahmoud, H.N.: Analysis of crack propagation in welded stiffened panels. Int. J. Fatigue 25(9–11), 1169–1174 (2003) 12. Dexter, R.J., Pilarski, P.J.: Crack propagation in welded stiffened panels. J. Construct. Steel Res. 58(5–8), 1081–1102 (2002) 13. Mellings, S., Baynham, J., Adey, R., Curtin, T.: Durability prediction using automatic crack growth simulation in stiffened panel structures, in http://www.beasy.com (2002) 14. Murphy, A., McCune, W., Quinn, D., Price, M.: The characterisation of friction stir welding process effects on stiffened panel buckling performance. Thin Walled Struct. 45(3), 339–351 (2007) 15. Murphy, A., Lynch, F., Price, M., Gibson, A.: Modified stiffened panel analysis methods for laser beam and friction stir welded aircraft panels. Proc. Inst. Mech. Eng. Part G J. Aerospace Eng. 220(G4), 267–278 (2006) 16. Uz, M.V., Kocak, M., Lemaitre, F., Ehrstrom, J.C., Kempa, S., Bron, F.: Improvement of damage tolerance of laser beam welded stiffened panels for airframes via local engineering. Int. J. Fatigue 31(5), 916–926 (2009) 17. Lohwasser, D., Chen, Z.: Friction Stir Welding: From Basics to Applications. CRC and Woodhead Publishing Ltd, Boca Raton (2009) 18. Mishra, R.S., Ma, Z. Y.: Friction stir welding and processing. Mat. Sci. Eng. R Rep. 50(1–2), 1–78 (2005) 19. Mishra, R.S., Mahoney, M.W.: Friction Stir Welding and Processing. ASM International, New York (2007) 20. Nandan, R., DebRoy, T., Bhadeshia, H.K.D.H.: Recent advances in friction-stir welding— Process, weldment structure and properties. Prog. Mat. Sci. 53(6), 980–1023 (2008) 21. Threadgill, P.L., Leonard, A.J., Shercliff, H.R., Withers, P.J.: Friction stir welding of aluminium alloys. Int. Mat. Rev. 54(2), 49–93 (2009) 22. Cam, G., dos Santos, J.F., Kocak, M.: Laser and electron beam welding of Al-alloys: literature review. In: GKSS report (1997) 23. Byrne, G., Dornfeld, D., Denkena, B.: Advancing cutting technology. CIRP Ann. Manufact. Technol. 52(2), 483–507 (2003) 24. Hibbit, D., Karlsson, B., Sorenson, P.: ABAQUS users manual, Karlsson Sorenson Inc., USA (2006) 25. DaToN.: Innovative fatigue and damage tolerance methods for the application of new structural concepts. Strengthening the competitiveness, specific targeted research project: a proposal for the 6th European framework program (2004) 26. Moreira, P.M.G.P., Richter-Trummer, V., Tavares, S.M.O., de Castro, P.M.S.T.: Characterization of fatigue crack growth rate of AA6056 T651 and T6: application to predict fatigue behaviour of stiffened panels. Mat. Sci. Forum 636–637, 1511–1517 (2010) 27. Vaidya, W., Angamuthu, K., Kocak, M.: Effect of load ratio and temper on fatigue crack propagation behaviour of Al-Alloys 6056. In: 8th International fatigue congress—FATIGUE 2002. Stockholm, Sweden (2002) 28. Llopart, L., Kurz, B., Wellhausen, C., Anglada, M., Drechsler, K., Wolf, K.: Investigation of fatigue crack growth and crack turning on integral stiffened structures under mode I loading. Eng. Fract. Mech. 73(15), 2139–2152 (2006) 29. Edwards, L., Fitzpatrick, M., Irving, P., Sinclair, I., Zhang, X., Yapp, D.: An integrated approach to the determination and consequences of residual stress on the fatigue performance of welded aircraft structures. J. ASTM Int. 3(2), 1–17 (2006) 30. Brot, A., Peleg-Wolfin, Y.: The damage-tolerance behaviour of integrally stiffened metallic structures. In: 48th Israel annual conference on aerospace sciences (2008) 31. Augustin, P.: Simulation of fatigue crack growth in the high speed machined panel under the constant amplitude and spectrum loading. In: 25th ICAF Symposium. Rotterdam (2009) 32. Augustin, P.: Prediction of crack growth in integrally stiffened panels. Letecky´ Zpravodaj’ 2, 5–7 (2007)

172


33. Tavares, S.M.O., Richter-Trummer, V., Moreira, P.M.G.P., de Castro, P.M.S.T.: Fatigue behavior of lightweight integral panels. In: 7th EUROMECH Solid Mechanics Conference. Lisbon, Portugal (2009) 34. Kunz, J., Kovárik, O., Lauschmanna, H., Siegla, J., Augustin, P.: Fractographic reconstitution of fatigue crack growth in integrally stiffened panels. Procedia Eng. 2, 1711–1720 (2010) 35. Moreira, P.M.G.P., de Castro, P.M.S.T.: Fractographic analysis of fatigue crack growth in lightweight integral stiffened panels. Int. J. Struct. Integ. (in press) 36. Brot, A., Peleg-Wolfin, Y., Kressel, I., Yosef, Z.: The use of composite material strips to extend the damage-tolerance life of integrally stiffened aluminum panels. In: 25th ICAF Symposium. Rotterdam (2009)

Damage Tolerance of Aircraft Panels Taking into Account Residual Stress V. Richter-Trummer, P. M. G. P. Moreira and P. M. S. T. de Castro

Abstract The present chapter seeks to summarize the ideas of damage tolerance of aircraft panels, keeping in mind the effect of residual stresses. First concepts and techniques are briefly reviewed, and afterwards their application to experimental work is discussed.

1 Introduction The basic concepts of damage tolerance criteria for civil aircraft are briefly reviewed in this chapter. As a result of the traditional usage of riveted joints in the aluminium alloy fuselage of civil aircraft, one advantage of this type of joints is the existing experience concerning their design and maintenance. When subjected to cyclic loading, riveted joints suffer fatigue damage, including multiple site damage (MSD). Thus a second item of this chapter concerns modelling of MSD and of residual strength in riveted structures. Means to improve the strength of those joints are presented and discussed. As an example the cold expansion process as a means of improving fatigue strength in open hole specimens is presented. Alternatives to riveting are being considered aiming at economies in fabrication time, cost and weight. One such alternative is welding, particularly laser or friction stir welding (LBW or FSW). However, open issues concerning the use of integral V. Richter-Trummer (&) P. M. S. T. de Castro Faculdade de Engenharia da Universidade do Porto, IDMEC-FEUP, Porto, Portugal e-mail: [email protected] P. M. G. P. Moreira INEGI-Institute of Mechanical Engineering and Industrial Management, FEUP, Porto, Portugal

Adv Struct Mater (2012) 8: 173–194 DOI: 10.1007/8611_2011_55 Springer-Verlag Berlin Heidelberg 2011 Published Online: 25 March 2011

173

174

V. Richter-Trummer et al.

structures in aeronautics include the damage tolerance problem, since the integral nature of the structure provides a continuous path for crack growth. Furthermore, high magnitude residual stresses may arise in parts of these production processes and have therefore to be considered. The third topic of the chapter therefore centres on the fatigue behaviour of integral stiffened panels focusing on the influence of residual stress fields.

1.1 Damage Tolerance As in many other engineering specialities, aircraft accidents may be sources of knowledge contributing to the advancement of the field. Organizations as the USA National Transportation Safety Board (NTSB) document in detail aircraft failure analyses. Concise presentations of some particularly exemplary cases are given by Wanhill [1] namely the Comet losses and the Aloha Airlines Boeing 737-200 1988 failure, both mentioned here because of their importance in the advancement of design criteria. Comet I, developed by the UK Havilland Aircraft Company was the first commercial jet. Designed for high altitude service, it entered service in 1951. In 1954 tragic accidents occurred with planes with 1286 and 903 flights, originated by fatigue crack propagation leading to the disintegration of the pressurized fuselage. Failure analysis revealed that although designed and tested for the conditions found in service, the design was defective concerning crack arrest capabilities. A second problem concerned the tests carried out by the manufacturer [2, 3]: due to economic reasons, a fuselage that was subjected to pressure testing up to ultimate design load, was subsequently fatigue tested. It is now understood that the results of the fatigue test were improved by compressive residual stresses originated by the previous pressure proof loading: indeed, compressive residual stresses retard crack propagation and as a consequence the manufacturer had no indication of the real structure weaknesses in service. The Comet catastrophes indicated the need for a design with crack arrest capability and contributed decisively towards a new design philosophy based upon damage tolerance. Historical facts are linked with the current research on alternative joining solutions, cheaper and faster than riveting, like FSW or LBW, and identifying possible fatigue crack propagation problems associated with these current developments including the effect of residual stress. Also yielding important lessons concerning fatigue behavior was the 1988 accident of the Aloha flight 243 [4]. At an altitude of 7300 m, part of the fuselage of the Boeing 737 was suddenly detached from the aircraft, with sudden decompression of the cabin. The upper fuselage loss was up to 5.5 m long extending from the aircraft main entrance door. Failure analyses identified multiple site damage along a substantial portion of a riveted joint as a key contribution to the accident. Together with increasing recognition of fatigue phenomena, aircraft were initially designed with the intention of avoiding fatigue crack initiation during

Damage Tolerance of Aircraft Panels Taking into Account Residual Stress

175

their life-time. The Comet accidents and progress of fatigue knowledge lead to a damage tolerant approach, where damage is considered unavoidable and measures are taken for its control along the aircraft life cycle. This leads to weight savings but also to increased maintenance costs, particularly those related with periodical inspections; however, the balance between the benefit of a lighter structure, and cost increase due to periodical inspection, is positive from the viewpoint of direct operating costs [5]. The adopted design philosophies are: • ‘fail-safe design’ where it is assumed that the component or structure may be safely operated even in the presence of some damage, that may grow up to a limit value before a component or structure replacement is required. In these structures a crack may grow in service, but will not reach critical size before its detection. Useful life is defined by a critical crack size (defined by the material toughness or other applicable criteria). Essential ingredients of these approaches are the knowledge of crack propagation as related with applied loading, and periodical inspections with a frequency ensuring that undetected damage in one inspection will not grow up to critical size before next inspection. • ‘safe-life design’ seeking to ensure that the component or structure will not develop fatigue cracks in service. This design philosophy is therefore based upon fatigue crack initiation avoidance during lifetime. The concept of safety factor in safe life design or in damage tolerant design is different. The first implies evaluations of uncertainties associated with life estimation, whereas the second implies monitoring processes to ensure continued structural integrity [6]. Safe life is used in components that cannot be duplicated, for example the landing gear; fail-safe is used for major structural components such as the fuselage. Fail-safe implies the definition of tolerable damage in the aircraft. The resistance of the structure may decrease below ultimate loads but not below design limit loads, defined as ‘‘JAR 25.301 loads’’ ‘‘…(a) Strength requirements are specified in terms of loading conditions that give rise to limit loads (… maximum loads to be expected in service) and ultimate loads (… limit loads multiplied by prescribed factors of safety)’’ [6, 7]. Figure 1 presents this concept schematically. With damage tolerant aircraft, inspection intervals are defined on the basis of acceptable damage propagation. Starting from an initial flaw size corresponding to the greatest undetectable size using conventional inspection tools, the number of cycles up to its critical dimensions is estimated. The corresponding time interval is then divided by a suitable safety factor (SF), leading to the selected interval between inspections, see Fig. 1, where the inspection interval would be (Nc-Nd)/ SF. The already mentioned flight 243 accident underlined the need to take into account initiation in more than one location, i.e. multiple site damage (MSD), and lead to a revision of the procedure presented above, taking into account shorter inspection intervals, given the consequent lower number of cycles between damage detection and critical conditions [5, 8], see Figs. 2 and 3.

176


Fig. 1 Damage tolerance concept

Fig. 2 Shorter life under MSD conditions

2 Residual Stress Residual stresses play an important role in damage tolerance considerations of modern integral structures [9]. Economic and energy efficiency factors required by today’s structures lead to the need for a superior behaviour of applied materials. This is partly due to the reduced safety coefficient used for saving weight and partly due to the production processes used, such as high speed machining and welding. Non of the savings obtained may come at the cost of reduced safety in the


177

Fig. 3 Effect of MSD

final product. Therefore the stress state in the lightweight integral structures has to be precisely known for accurate fatigue life and service interval estimation. Even in traditional lightweight structures, for example produced by riveting, compressive residual stresses around the rivet holes play an important role in the fatigue life of this kind of structures. Residual stresses may be of three types. Type III stresses are intra-granular stresses acting within a grain, type II stresses are inter-granular stresses acting through the grain boundaries at a small scale and type I stresses are usually called macroscopic residual stresses due to their influence over large number of grains [10]. Several thorough reviews of residual stress measurement techniques are available [11–13]. In the present work, only macroscopic residual stresses are considered due to their obvious effect on the damage tolerance of lightweight structures. Such macroscopic residual stresses may arise from heat treatment, machining or welding operations, or even by the assembly process of complex structures [14]. Residual stress measurement techniques are available in non-destructive, destructive and intermediate semi-destructive variants. Non-destructive techniques are generally based on diffraction of beams from different sources of energy. Semidestructive techniques are based on the partial relaxation of stresses by removal of small parts of the material which does not in general influence the structural integrity of the component, and destructive methods are based on the complete relaxation, for example by cutting a component into several parts, measuring the resulting deformations [13]. Figure 4 shows a synthesis of different available techniques for measuring residual stresses. It should be noted that residual stresses may not be measured directly—only their effect may be measured, most commonly their relaxation due to material removal. The techniques referenced in Fig. 4 are the non-invasive Neutron, X-ray and synchrotron X-ray diffraction methods, the semi-destructive X-ray diffraction (XRD) technique with removal of surface layers, the incremental hole drilling technique and the deep hole drilling technique, and finally the invasive crack compliance, sectioning, contour and layer removal techniques. This is not a complete list of available residual stress measurement techniques, but represents the most used techniques to cover a large range of depths.

178

sectioning

invasive

Fig. 4 Some residual stress measurement techniques and their penetration depth [15– 22]


crack compliance

contour

semi−invasive

layer removal XRD − removal DHD

non−invasive

iHDT Neutrons Synchrotron XRD XRD −3

10

−2

10

−1

0

1

10 10 10 penetration depth [mm]

2

10

3

10

2.1 Measurement Techniques Some details about the measurement techniques used in the present chapter are given below. The incremental hole drilling technique is based on the standard hole drilling technique, but due to the special data reduction applied, it is able to determine information on stress variations through the thickness of a component. The ASTM standard E837 [23] and the Vishay Technical Note TN-503-6 [24] should be followed for best results. The HDT is one of the most widely used techniques for measuring residual stress, since it is relatively simple, inexpensive, quick and versatile. Its functioning principle is based around the introduction of a hole into a body with residual stresses, which relaxes the stresses at that location. The hole drilling method for residual stress measurements was first proposed by Mathar in 1934 [25]. Presently, this method is a widely accepted technique for measuring residual stresses. It is a semi-destructive technique where a tolerable small volume of material is removed. The basic hole drilling procedure, involves drilling a small hole into the surface of a component at the centre of a special strain gage rosette and measuring the relieved strains. Figure 5 shows an example of an experimental setup. The method is very versatile and can be performed either in the laboratory or the field, on different materials, and on components in a wide range of sizes and shapes. The hole is typically 0.8–4.8 mm in both diameter and depth. Nevertheless, achieving accurate results is not trivial; a meticulous measurement practice and the adequate choice of data analysis method are crucial for obtaining good results. The sectioning method is based on the surface measurement of the relaxation of a specimen due to a cut by a high number of strain gauges, as is shown in Fig. 6.


179

Fig. 5 Picture of the hole drilling process in a fusionwelded plate

Fig. 6 Instrumented specimen being cut for residual stress relaxation and determination [26]

Since strain gauges only measure surface strains, this method does not give good information of through the thickness variations of the residual stress. The best though the thickness information approximation that may be obtained by this method is a linear variation between top and bottom surface measurements. Even though this method was used before, one of the first publications directly mentioning the method appeared in 1997 [27] for the measurement of welding residual stresses in a 7 mm thick plate. The contour method for residual stress determination has been developed by Prime in 2001 [28]. It consists in the measurement and analysis of a cut-surface in the plane where residual stress is to be determined. The relaxation due to the cut is measured and then used to infer the residual stresses that were present at the plane of interest before the cut was made. Only stresses perpendicular to the cut surface can be determined by this method. The following procedure is used to obtain an accurate model of the residual stresses by the contour method. First the specimen is cut along the surface where stresses are to be determined as is shown in Fig. 7a for a wire electro discharge-machining (EDM) cut.

180


(a)

(b)

Fig. 7 Contour method application. a Wire EDM cut of a FSW specimen for residual stress relaxation for application of the contour method. b Schematic representation of the symmetric free surface deformation due to the residual stress relaxation after the cut

Figure 7b shows a scheme of the free surface deformation after the cut on both plate halves. Afterwards the obtained free surfaces are digitized. The obtained relaxation data is treated and applied to a FEM model from where the stresses originally present before the cut can be retrieved. The X-ray diffraction (XRD) technique is a non-destructive technique applicable in laboratory but also outside. It allows the determination of residual stresses in all three axes. Laboratory X-ray sources have wavelengths of typically around 1.58A, are limited to penetration depths below 100 lm, and therefore only produce data for surface stresses. Successively removing layers of material allows the measurement of sub-surface stresses, but turns this technique into a semi destructive technique. This method is especially recommended for works related to surface treatments. Further details may be seen in reference [18] for example. The layer removal method for residual stress measurement is a fully destructive method based on the measurement of the distortion of a specimen due to the removal of layers of material containing residual stresses. Schajer recommends the use of the layer removal technique when stresses are known to vary through the thickness, but are uniform parallel to the surface [11]. This method is capable of providing information for a relatively low magnitude of residual stresses, since the distortion (most easily the curvature, k) of a specimen due to the removal of layers may be determined very accurately, see Fig. 8a. While mostly the recommended layer removal technique is by chemical etching, a milling process such as shown in Fig. 8b at a high speed and in small steps does also lead to good measurement results. The literature includes several comprehensive reviews of residual stress measurement techniques, e.g. [9–11, 13, 15, 29], including several techniques not dealt with in this section.


(a)

181

(b)

Fig. 8 Layer removal process. a Scheme of the layer removal process. The removal of a thin layer of material leads to measurable distortion when residual stresses were present in this layer. b Milling of a layer in the layer removal technique for residual stress determination

3 Riveted Joints Riveting is so far the standard method of joining aluminium fuselage parts. This is likely to change, given the current interest of aircraft manufacturers in composite fuselages for some aircraft sizes (e.g. Boeing 787), but also to efforts to find more economic aluminium solutions using welding technologies. Nevertheless, according to ATAG [30], some 2000 airlines around the world operate a total fleet of 23,000 aircrafts. These approximate figures underline the importance of riveting technology and associated maintenance practices, since most of these riveted aircraft are expected to still have long useful lives. As a very first approximation, the open hole specimen subjected to remote tensile cyclic loads gives relevant information on a riveted joints’ behaviour. It allows, for example to identify the importance of fatigue improvement technologies such as cold expansion, see Fig. 9 taken from [31]. Nominal cross section was 25 mm 9 2 mm, with nominal hole diameter 5 mm. The Boeing Company originally developed the basic cold expansion process in the late 1960s. Cold expansion of rivet holes consists of expanding a hole (usually 3–5%) using a mandrel. This creates an annular region of plastically deformed material, which gives rise to a residual compressive stress field when the mandrel is removed. Figure 10 shows the residual stress around a cold worked hole measured by XRD [32]. The resulting residual stress field improves fatigue performance by delaying nucleation and retarding crack propagation [33–37]. The riveted lap joint presents a complex behaviour including bending and contact between the rivet and the hole, and between sheets, leading to some specific features of fatigue cracking as the initiation of cracks above the hole and not through a plane containing the centre of the hole as might be expected [38–40].

182


Fig. 9 AA2024-T3 Alclad: 24 specimens with and 21 specimens without residual stresses) tested at R = rmin/ rmax = 0.1 [31]

Fig. 10 Residual stresses around a cold worked hole as measured by XRD [32]

The simplest tests of actual riveted lap joints concern one column specimens, as shown in Fig. 11, outcome of a round robin exercise intended to generate input data (crack length a versus number of cycles N) for the equivalent initial flaw size (EIFS) distribution determination and to generate a probabilistic stress-life (P–S–N) curve for safe-life analysis [41]. The SN data of the 300 specimens tested is shown in Fig. 12. It is to be noticed that the diameter of open hole specimens reported in Fig. 9 is 5 mm, and their thickness 2 mm, whereas the specimens of Fig. 11 are made of 1.2 mm thick sheets and the rivet hole diameter is 3.2 mm. Also, R = 0.1 in the case of the open hole specimens, and 0.05 in the case of the one rivet column lap joints. Because of these reasons, no direct comparisons can be made between the data of Figs. 9 and 12. However, the data suggests an increase in strength of the lap joint as compared to the normal non-cold worked open-hole specimen.


183

Fig. 11 Riveted lap joint: one column of three rivets [31]

Fig. 12 Fatigue test data for 2024-T3 Alclad

170

Maximum stress [MPa]

160 150 140 130 120 110 100

Univ Napoli Federico II

90

Univ Pisa

80

IDMEC-FEUP, Porto

70 1

10

100

1000

10000

number of cycles [x103]

That should be related to the load distribution among the 3 rivets in the case of the lap joint specimens and to the presence of compressive residual stresses around the hole. Moreira et al. [42] discussed the distribution of load among the three rivets in the context of crack propagation. It suffices to mention here that for the un-cracked situation, the overall load is distributed approximately 38% for the first and last rivet, and the remaining (23%) for the middle one. A serious form of damage in riveted Al fuselage is MSD, consisting of the simultaneous presence of several active fatigue cracks in a single joint. MSD lead to the widely discussed 1988 Aloha Boeing 737 accident. A discussion of this type of damage is given in [43, 44]. The problem of MSD may be studied from the initiation and propagation viewpoints. As far as initiation is concerned, statistical techniques are used to model the initiation behaviour, as discussed in detail by

184


Cycles

Crack configuration 1

2

3

4

5

6

7

8

9

10

11

12

13

14

12.30

8.33

10.62

11.40

16.52

5.20

11.70

12.10

12.42

12.68

14.63

13.36

15.27

14.17

14.53

13.36

15

16

17

18

19

20

797500

799500

800500

801500

802500

804000 16.08

16.08

16.39

804179 8.81

804399 linkup

Fig. 13 MSD in a riveted lap joint, showing crack length as a function of number of cycles; details in [47]

Horst [45, 46] concerning propagation, test results are shown plotting number of cycles in vertical axis versus crack growth at relevant locations, in horizontal axis, e.g. Fig. 13 taken from Silva et al. [47].

4 Welded Joints Part count reduction and possible economies in cost and fabrication time generated an interest in alternative joining processes for metallic fuselage structures. Conventional fusion welding techniques applied to typical aluminium fuselage alloys generate weldments with unacceptable defects. This problem was overcome with laser beam welding (LBW) and friction stir welding (FSW), both already applied to parts of fuselages of Airbus aircrafts. FSW was adopted for the fabrication of the fuselage of the Eclipse 500. FSW joints present sound metallurgical properties and using adequate welding parameters defects may be substantially avoided. Heat input and residual stresses are comparatively low. Figure 14 shows the longitudinal residual stress distribution measured by the contour method on a plane perpendicular to the FSW weldment of AA6082 as an example.

Damage Tolerance of Aircraft Panels Taking into Account Residual Stress −200

−150

−100

−50

0

50

100

150

200

0.5 plate thickness direction [mm]

Fig. 14 Longitudinal residual stresses measured by the contour method on a plane perpendicular to the FSW weldment [48]

185

0 −0.5 −1 −1.5 −2 −2.5 −3

−100

−50

0

50

100

plate width direction [mm]

As can be seen, tensile residual stresses almost reaching 200 MPa are present at the border of the TMAZ, and compressive residual stresses guarantee the necessary equilibrium outside of this region. For a fusion MIG weld, an equivalent stress measurement on AA6082 shows higher tensile residual stresses in the weld bead, almost reaching the yield strength of the material and compressive stresses outside of this region, see Fig. 15. It may therefore be concluded, that this kind of fusion weld leads to both, higher stress magnitudes, and higher stress gradients. Some relevant comparisons between MIG (metal inert gas), LBW and FSW applied to aluminium alloys are available [50, 51]. FSW joints present a decrease in hardness in the thermo-mechanically affected zone (TMAZ, essentially the nugget and the region below the tool shoulder), see for example Fig. 16 [52], concerning butt joints of AA 6082-T6. In most applications, particularly in aeronautics, fatigue behaviour is a critical concern. In case there is not a pre-existing crack, the fatigue process is composed of crack initiation, growth, and final rupture. It is therefore important to characterize the fatigue behaviour of FSW joints using a variety of tests. Initiation behaviour may be assessed by using SN tests, since in this type of test, once a crack initiates the specimen final rupture occurs very soon after. It should be noted that this kind of small test coupon generally has no significant residual stresses due to relaxation during the specimen production. Basic SN data is reported in [53], for the aluminium alloys AA6061 and AA6082, showing the reduction of fatigue lives of as welded FSW joints compared with base material specimens, but also their improved behaviour when compared to MIG welded specimens. For a maximum stress corresponding to 105 cycles, reductions of approximately 45% for LBW, 53% for FSW and 63% for MIG are found, see Fig. 17. These facts are not surprising; more noteworthy is what happens when there is a notch within the TMAZ. Load controlled SN (stress versus number of cycles) tests of AA6063 specimens loaded perpendicularly to the weldment, containing a small circular open hole in the TMAZ have been discussed [54]. In this case, it was

186


c) b)

(a)

−200

−150

−100

−50

0

50

100

150

200

(b)

(c) Fig. 15 Longitudinal residual stresses (in MPa) measured by the contour method on a plane perpendicular to the MIG weldment [49]. a Cross section. b Detail of the weld bead area. c Detail of the area from the weld bead region

Fig. 16 Typical FSW hardness profile [52]

verified that such specimens presented slightly higher fatigue lives than base material specimens with a similar geometry, possibly due to the microstructural changes undergone in the TMAZ. Another interesting feature of FSW fatigue behaviour is reported in [55], where fatigue crack propagation tests were


187

Fig. 17 SN behaviour, base and welded materials [53]

Fig. 18 Welded CT specimen types [53]

conducted on compact tension specimens with the crack presenting three different orientations in relation to the weldment: crack growing along the heat affected zone, crack in the plane of symmetry of the weldment and crack growing perpendicularly to the weldment, see Fig. 18. For comparison, base material specimens of identical geometry were also tested. The results, discussed in detail in [50, 55], are presented in Fig. 19. The remarks made concerning the various types of tests and behaviours mentioned—SN tests on un-notched and notched specimens, fatigue crack propagation for several locations of the crack plane on CT specimens, concern a few different base material aluminium alloys (AA6061, AA6063 and AA6082). The trends identified should not be generalized to all families of aluminium alloys. Nevertheless, for the materials and conditions studied, it is clear that FSW joints present good or very good fatigue behaviour, concerning initiation of cracks in notches located in the weldment and propagation of fatigue cracks in the weldment. Good fatigue behaviour was also found testing FSW specimens made of AA2195-T8X, an Al–Li alloy being considered for the cryogenic tanks of the next generation of

188


Fig. 19 Comparison of base and FSW welded material [50, 55]

Fig. 20 FSW AA 2195T8X, SN tests at room temperature [56]

the ARIANE launcher, see Fig. 20 [56, 57]. In this case, when testing base material at R = 0.1, maximal stress corresponding to 105 cycles is of the order of 350–400 MPa [58] whereas for FSW joints, tested under R = 0.1 with the weldment perpendicular to the load, maximal stress is in the order of 260–280 MPa for the same number of cycles, a reduction of approximately 30%.

5 Stiffened Panels A damage tolerant fuselage is supposed to sustain cracks safely until it is repaired or its economic service life has expired. Strength assessment of the structures is necessary for their in-service inspection, repair, retrofitting and health monitoring;


189

therefore, damage tolerance analysis should provide information about the effect of cracks on the strength of the structure. Recently, studies are being conducted to explore designs with equal or better performance than conventional designs with regard to weight and structural integrity, while achieving a significant reduction in manufacturing cost; For example the Boeing/NASA work, e.g. [59], or the EU DATON project, which ran from 2005 to 2008 (see synopsis in [60]). The skin structure of a pressurized fuselage for transport aircraft is fatigue sensitive, which is a problem often, requiring frequent repairs. Under cyclic loading, a flaw can develop into a fatigue crack and propagate until fracture occurs. The residual strength concept permits the determination of the maximum crack length that can be safely sustained. With this information and the characterization of the crack growth behaviour of the material, the number of loading cycles that will be necessary for the crack to grow up to its critical length can be estimated in order to ensure safe operation. The development of numerical methodologies with the help of small laboratory coupon test results should be used to predict the residual strength of complex built-up aircraft fuselage structures. Fracture mechanics concepts in conjunction with fatigue crack propagation laws are widely used to analyze and predict crack growth and fracture behaviour of aircraft panels. The DATON project aimed to provide assessment tools for the damage tolerance of integrally stiffened structures produced by laser beam welding (LBW), friction stir welding (FSW) and high speed machining (HSM). There is an urgent pressure from the manufacturing side in the aerospace industry to apply these advanced structural concepts, since they promise considerable cost and production time benefits, together with a smaller number of fatigue and corrosion critical locations. The main drawback of integrally stiffened structures is the damage tolerance behaviour. Such an integrally stiffened design behaves totally different from the differential designs, which are usually created by using riveted stiffeners. The prime problem is the crack arresting capability of the stiffeners both in fatigue crack growth as well as in residual strength. Fatigue and damage tolerance is therefore one of the main drivers in innovative aerospace structural designs as well as one of the main concerns on the safety of aging aircraft. A test program which included fatigue crack growth rate of HSM, LBW, and FSW stiffened panels was performed. This test program was complemented by a scanning electron microscopy analysis of the fractured specimens. Details of these experiments may be found in chapter ‘‘Lightweight stiffened panels fabricated using emerging fabrication technologies: fatigue behaviour’’. Since the stiffeners were welded in some of the specimens, these specimens exhibit tensile residual stresses in the stiffener region, and compressive residual stresses at some distance to them guaranteeing the necessary stress equilibrium. While HSM specimens do not include such a residual stress distribution, as can be seen from Fig. 21, residual stresses originating from the rolling and other production processes may still be found. Due to their lower magnitude, normally these stresses tend to lead to shape distortion but less to variations in the damage tolerance behaviour. Since these residual stresses are normally constant along the

190

specimen A specimen B applied to FEM

30

residual stress [MPa]

Fig. 21 Residual stress distribution through the thickness measured on two comparable specimens by the layer removal technique of a rolled aluminium plate used for aeronautical applications


20 10 0 −10 −20 −30 0

1

2

3

4

5

6

thickness [mm]

Fig. 22 Comparison of a versus N for all specimens tested at R = 0.1 [50]

plate but vary only though the thickness, the influence in the crack growth behaviour of thin plates is normally negligible. When fatigue tested at R = 0.1 the behaviour found for specimens of the 3 types studied—LBW, FSW and HSM—is shown in Fig. 22, where welded panels always display greater fatigue lives than machined (HSM) panels. The work mentioned above illustrates again the importance of residual stresses in fatigue performance. In the case of DATON panels, because the initial crack was considered in the skin midway between stiffeners (were the welding residual stresses are tensile), the initial stages of fatigue crack propagation took place in a compressive residual stress field, thus retarding crack growth, which explains the behaviour presented in Fig. 22. Open issues concerning the use of monolithic integral structures in aeronautics include the damage tolerance problem, since the integral nature of the structure provides a continuous path for crack growth. Possible solutions under consideration


191

include systematic thickness variations (crenellations) [61], or specific surface treatments as supersonic particle deposition (SPD) desirably leading to some form of control of the crack growth path. In parallel, alloys with improved strength but also good toughness and fatigue behaviour are considered, as the current generation of Al–Li alloys that overcomes the drawbacks of earlier generations characterized by high strength but poor toughness.

6 Concluding Remarks Open issues concerning the use of monolithic integral structures in aeronautics include the continuous path for crack growth. Possible solutions under consideration include crenellations or specific surface treatments as supersonic particle deposition (SPD) desirably leading to some form of control of crack growth path. In parallel, alloys with improved strength but also good toughness and fatigue behaviour are considered, as the current generation of Al–Li alloys that overcomes the drawbacks of earlier generations characterized by high strength but poor toughness. The influence and importance of considering residual stresses in damage tolerance considerations, especially of welded lightweight structures, has been shown. Acknowledgments Valentin Richter-Trummer acknowledges the support of the Ph.D. scholarship SFRH/BD/41061/2007 of the Portuguese Fundação para a Ciência e Tecnologia. The support of colleagues of the University of Porto, particularly M. A. V. Figueiredo is acknowledged. R&D projects DATON, ADMIRE and DATOL of IDMEC-Porto and INEGI have provided interesting results presented in this chapter. R&D project IMAGINE (contract of GKSS) has provided access to material used for the layer removal technique for residual stress measurement. Dr Pedro Moreira acknowledges POPH-QREN-Tipologia 4.2 promotion of scientific employment funded by the ESF and MCTES.

References 1. Wanhill, R.J.H.: Milestone case histories in aircraft structural integrity. In: Milne, I., Ritchie, R.O., Karihaloo, B. (eds.) Comprehensive Structural Integrity, vol. 1, pp. 61–72. Elsevier. ISBN:008-044157-2 2. Schijve, J.: Fatigue damage in aircraft structures, not wanted, but tolerated? Int. J. Fatigue 31, 998–1011 (2009) 3. Brot, A., Peleg-Wolfin, Y.: The damage-tolerance behavior of intregrally stiffened metallic structures. In: 48th Israel Annual Conference on Aerospace Sciences IACAS Conference (2008) 4. National Transportation Safety Board – NTSB. Aircraft Accident Report, Aloha Airlines, Flight 243, Boeing 737-200, N73711, Near Maui, Hawaii, April 28, 1988. National Transport Safety Board, Washington, DC (1989) 5. Boller, C., Buderath, M.: Fatigue in aerostructures—where structural health monitoring can contribute to a complex subject. Philos. Trans. R. Soc. A 365, 561–587 (2007)

192


6. Bristow, J.W., Irving, P.E.: Safety factors in civil aircraft design requirements. Eng. Fail. Anal. 14, 459–470 (2007) 7. Joint Aviation Authorities, JAR-25, Airworthiness, Large Aeroplanes, JAR 25.301 loads http://www.jaa.nl/publications/publications.html 8. Schijve, J.: Multiple site damage in aircraft fuselage structures. Fatigue Fract. Eng. Mater. Struct. 18(3), 329–344 (1995) 9. Withers, P.J.: Residual stress and its role in failure. Rep. Prog. Phys. 70(12), 2211–2264 (2007) 10. Withers, P., Bhadeshia, H.: Overview—residual stress part 2—nature and origins. Mater. Sci. Technol. 17(4), 366–375 (2001) 11. Schajer, G.: Relaxation methods for measuring residual stresses: techniques and opportunities. In: Experimental Mechanics, pp. 1–11 (2010). ISSN:0014-4851. doi:10.1007/ s11340-010-9386-7, http://dx.doi.org/10.1007/s11340-010-9386-7 12. Radaj, D.: Welding Residual Stresses and Distortion—Calculation and Measurement, revised edn. DVS-Verlag GmbH (2003) 13. Withers, P., Bhadeshia, H.: Residual stress. Part 1: measurement techniques. Mater. Sci. Technol. 17(4), 355–365 (2001) 14. Venkitakrishnan, P.V., Philip, J., Krishnamurthy, R.: An Assessment of stresses in thin walled welded tubes through hole drilling and sectioning methods. J. Mater. Process. Technol. 185, 228–232 (2007) 15. Pardowska, A.M., Price, J.W.H., Finlayson, T.R.: Efficient use of available techniques to measure residual stresses in welded components. In: International Workshop on Thermal Forming and Welding Distortion, Bremen, 22–23 April 2008 16. Grant, P.V., Lord, J.D., Whitehead, P.S.: The measurement of residual stresses by the incremental hole drilling technique. A National Measurement Good Practice Guide 53, National Physics Laboratory; Stresscraft Ltd, Teddington, Middlesex TW11 0LW, UK, August 2002 17. Prime, M.: Cross-sectional mapping of residual stresses by measuring the surface contour after a cut. J. Eng. Mater. Technol. Trans. ASME 123, 162–168 (2001) 18. Fitzpatrick, M.E., Fry, A.T., Holdway, P., Kandil, F.A., Shackleton, J., Suominen, L.: Determination of residual stresses by X-ray diffraction. A national measurement good practice guide 52, National Physics Laboratory, Teddington, Middlesex TW11 0LW, UK (2005) 19. http://www.veqter.co.uk/dhd-technology, http://www.veqter.co.uk Accessed: 10.03.2011VEQTER Ltd, University Gate East, Park Row, Bristol BS1 5UB, UK (2010) 20. Webster, P., Oosterkamp, L., Browne, P., Hughes, D., Kang, W., Withers, P., Vaughan, G.: Synchrotron X-ray residual strain scanning of a friction stir weld. J. Strain Anal. Eng. Des. 36, 61–70 (2001) 21. Woo, W., Feng, Z., Wang, X.-L., Brown, D.W., Clausen, B., An, K., Choo, H., Hubbard, C.R., David, S.A.: In situ neutron diffraction measurements of temperature and stresses during friction stir welding of 6061-T6 aluminium alloy. Sci. Technol. Weld. Join. 12, 298–303 (2007) 22. Hospers, F., Vogelsang, L.: Determination of residual-stresses in aluminum-alloy sheet material. Exp. Mech. 15(3), 107–110 (1975) 23. ASTM E837.: Standard Test Method for Determining Residual Stresses by the Hole-Drilling Strain-Gage Method 1. ASTM Standard (2007) 24. Vishay.: Measurement of residual stresses by the hole-drilling* strain gage method. Technical note TN-503-6, Vishay, August 2007 25. Mathar, J.: Determination of initial stresses by measuring the deformation around drilled holes. ASME Trans. 56(4), 249–254 (1934) 26. Moreira, P.M.G.P., Richter-Trummer, V., da Silva, R.A.M., de Figueiredo, M.A.V., de Castro, P.M.S.T.: Residual stress evaluation of a MIG butt welded aluminium alloy plate. Mecânica Exp. 17, 29–39 (2009) 27. Galatolo, R., Lanciotti, A.: Fatigue crack propagation in residual stress fields of welded plates. Int. J. Fatigue 19, 43–49 (1997)


193

28. Prime, M.B.: Cross-sectional mapping of residual stresses by measuring the surface contour after a cut. J. Eng. Mater. Technol.-Trans. ASME 123(2), 162–168 (2001) 29. Ruud, C.O.: Residual stress measurement. In: Analytical Characterization of Aluminum, Steel, and Superalloys, 1st edn. Taylor & Francis Ltd. 10 October 2005 30. ATAG.: The Economic and Social Benefits of Air Transport 2008. Air Transport Action Group, Geneva. April 2008 31. de Castro, P.M.S.T., de Matos, P.F.P., Moreira, P.M.G.P., da Silva, L.F.M.: An overview on fatigue analysis of aeronautical structural details: open hole, single rivet lap-joint, and lapjoint panel. Mater. Sci. Eng. A 468–470, 144–157 (2007) 32. Pina, J.C.P., Dias, A.M., de Matos, P.F.P., Moreira, P.M.G.P., deCastro, P.M.S.T.: Residual stress analysis near a cold expanded hole in a textured alclad sheet using X-ray diffraction. Exp. Mech. 45, 83–88 (2005) 33. Matos, P., Moreira, P.M.G.P., Pina, J., Dias, A., de Castro, P.: Residual stresses around an expanded hole in an aluminum clad sheet. Mater. Sci. Forum 490–491, 41–46 (2005) 34. de Matos, P.F.P., McEvily, A., Moreira, P.M.G.P., de Castro, P.M.S.T.: Analysis of the effect of cold-working of rivet holes on the fatigue life of an aluminum alloy. Int. J. Fatigue 29, 575–586 (2007) 35. de Matos, P.F.P.: Plasticity induced fatigue crack closure: modelling and experiments. DPhil thesis, University of Oxford, UK (2008) 36. de Matos, P.F.P., Nowell, D.: Analytical and numerical modelling of plasticity-induced crack closure in cold-expanded holes. Fatigue Fract. Eng. Mater. Struct. 31, 488–503 (2008) 37. Pasta, S.: Fatigue crack propagation from a cold-worked hole. Eng. Fract. Mech. 74, 1525– 1538 (2007) 38. Silva, L.F.M., Gonçalves, J.P.M., Oliveira, F.M.F., de Castro, P.M.S.T.: Multiple-site damage in riveted lap-joints: experimental simulation and finite element prediction. Int. J. Fatigue 22, 319–338 (2000) 39. de Rijck, J.J.M., Homan, J.J., Schijve, J., Benedictus, R.: The driven rivet head dimensions as an indication of the fatigue performance of aircraft lap joints. Int. J. Fatigue 29, 2208–2218 (2007) 40. Skorupa, M., Skorupa, A., Machniewicz, T., Korbel, A.: Effect of production variables on the fatigue behaviour of riveted lap joints. Int. J. Fatigue 32(7), 996–1003 (2010) 41. Koolloos, M.F.J., de Castro, P.M.S.T., Esposito, R., Cavallini, G.: Fatigue testing of singlerivet lap joint specimens. Technical report ADMIRE-TR-3.1-06-3.1/NLR CR-2003-281 (2003) 42. Moreira, P.M.G.P., de Matos, P.F.P., Camanho, P.P., Pastrama, S.D., de Castro, P.M.S.T.: Stress intensity factor and load transfer analysis of a cracked riveted lap joint. Mater. Des. 28, 1263–1270 (2007) 43. Jones, R., Molent, L., Pitt, S.: Understanding crack growth in fuselage lap joints. Theor. Appl. Fract. Mech. 49, 38–50 (2008) 44. Jones, R., Molent, L., Pitt, S.: Study of multi-site damage of fuselage lap joints. Theor. Appl. Fract. Mech. 32, 81–100 (1999) 45. Horst, P.: Widespread fatigue damage—an issue for aging and new aircraft. Key Eng. Mater. 324–325, 1–8 (2006) 46. Horst, P.: Assessment of Multiple Site Damage in Riveted Aircraft Joints. In: Moreira, P.M.G.P et al (eds.) Structural Connections for Lightweight Metallic Structures, Springer, (inpreparation) 47. Silva, L.F.M., Gonçalves, J.P.M., Oliveira, F.M.F., de Castro, P.M.S.T.: Multiple-site damage in riveted lap-joints: experimental simulation and finite element prediction. Int. J. Fatigue 22, 319–338 (2000) 48. Richter-Trummer, V., Moreira, P.M.G.P., Ribeiro, J., de Castro, P.M.S.T.: The contour method for residual stress determination applied to an AA6082-T6 friction stir butt weld. In: ECRS8’ 2010, June 26–28, 2010

194


49. Richter-Trummer, V.: Characterization of different aluminium alloys of the series 6000 and of their joining processes. Mestrado integrado em engenharia mecânica, Faculdade de Engenharia da Universidade do Porto, Porto, Portugal, February 2008 50. Moreira, P.M.G.P.: Lightweight stiffened panels: Mechanical characterization of emerging fabrication technologies. Ph.D. thesis, Faculty of Engineering of the University of Porto, Portugal (2008) 51. Richter-Trummer, V., Silva, S.O., Peixoto, D.F.C., Frazão, O., Moreira, P.M.G.P., Santos, J.L., de Castro, P.M.S.T.: Fibre bragg grating sensors for monitoring the metal inert gas and friction stir welding processes. Meas. Sci. Technol. 21, 085105 (2010) 52. Moreira, P.M.G.P., de Jesus, A.M.P., Ribeiro, A.S., de Castro, P.M.S.T.: Assessment of the fatigue behaviour of friction stir welded joints: aluminium alloy 6082-T6. Key Eng. Mater. 348–349, 209–212 (2007) 53. Moreira, P.M.G.P., Richter-Trummer, V., de Castro, P.M.S.T.: Multiscale fatigue crack initiation and propagation of engineering materials—structural integrity and microstructural worthiness. In: Sih, G.C. (ed.) Fatigue Behaviour of FS, LB and MIG Welds of AA6061-T6 and AA6082-T6, pp. 85–111. Springer, New York (2008) 54. Moreira, P.M.G.P., de Oliveira, F.M.F., de Castro, P.M.S.T.: Fatigue behaviour of notched specimens of friction stir welded aluminium alloy 6063-T6. J. Mater. Process. Technol. 207(1–3), 283–292 (2008) 55. Moreira, P.M.G.P., de Jesus, A.M.P., Ribeiro, A.S., de Castro, P.M.S.T.: Fatigue crack growth in friction stir welds of 6082-T6 and 6061-T6 aluminium alloys: a comparison. Theor. Appl. Fract. Mech. 50(2), 81–91 (2008) 56. Windisch, M., Glaser, U, Trautmann, K.H., de Castro, P.M.S.T., Sattler, E.: Damage tolerance characterization of 2195 base material and friction stir welds. In: 11th European Spacecraft Structures, Materials and Mechanical Testing Conference (SSMMT), Toulouse, 15–17 Sept 2009 57. Eigen, N., Glaser, U., Sinnema, G., Moreira, P.M.G.P., Shneider, J., Sattler, E.: Microstructure and properties of AA 2195 base material and friction stir welds. In: 11th European Spacecraft Structures, Materials and Mechanical Testing Conference (SSMMT), Toulouse, 15–17 Sept 2009 58. Bonnafé, J.P., Gabard, D., Grosjean, E.: Aluminium lithium alloys use for reusable future launcher cryogenic metallic tanks. In: 5th Conference on Aerospace Materials, Processes, and Environmental Technology, Huntsville, Alabama. NASA/CP-2003-212931, Sept 2002 59. Pettit, R., Wang, J., Toh, C.: Validated feasibility study of integrally stiffened metallic fuselage panels for reducing manufacturing costs. NASA/CR-2000-209342 (2000) 60. European Communities.: Aeronautics Research: 2003–2006 projects. Project synopses, vol 1. p 167. ISBN 92-79-00643-6 61. Uz, M.-V., Koçak, M., Lemaitre, F., Ehrström, J.-C., Kempa, S., Bron, F.: Improvement of damage tolerance of laser beam welded stiffened panels for airframes via local engineering. Int. J. Fatigue 31, 916–926 (2009)

Multi-Material Adhesive Joining in the Automotive Sector Sylvain Pujol

Abstract Aston Martin uses lightweight materials in combination with adhesive bonding technology to manufacture its vehicles Body-in-White and this Chapter gives an overview of the motivations, advantages and challenges behind this choice. A detailed material breakdown and associated joining processes is presented, showing how Aston Martin succeeded in manufacturing a leading performance multi-material automotive body structure. It is also highlighted how joint design and materials selection were carried out for structural and semi-structural applications. It is demonstrated how adhesive bonding technology and processes were developed and optimised to overcome some of the challenges associated with low volume manufacturing of lightweight, multi-material automotive structures.

1 Introduction Technological progress over the last decade has provided the automotive industry with more flexibility in terms of design and manufacturing. Materials and joining techniques now allow car body structures to be built economically that suit both the volumes and performance of the final product. Automotive manufacturers face conflicting requirements regarding environmental legislation and an increasing demand for safety and on-board equipment which tends to increase the overall vehicle weight and fuel consumption. This challenge has led to the increased use of lightweight materials to replace conventional steel parts and structures. This new generation of materials typically includes thermoset composites for body panels, fibre reinforced thermoplastics for semi-structural applications and lightweight alloys such as magnesium and aluminium for body structures. S. Pujol (&) Aston Martin, Banbury Road, Gaydon CV35 0DB, UK e-mail: [email protected] Adv Struct Mater (2012) 8: 195–217 DOI: 10.1007/8611_2010_51 Springer-Verlag Berlin Heidelberg 2011 Published Online: 12 February 2011

195

196

S. Pujol

Table 1 Cost associated with different vehicle architectures [3] Steel Aluminium monocoque monocoque Low volume (\20K/year) Medium volume (\100K/ year) High volume ([100K/year)

Aluminium spaceframe

$5800 $2500

$7200 $3600

$4500 $2800

$1400

$2000

$2400

This Chapter is intended to provide an overview of the materials, technologies and associated processes used to manufacture Aston Martin’s vehicle bodies.

1.1 Different Types of Vehicles Architectures Stamped sheet monocoque, extrusion intensive structures and spaceframe assemblies are the three main vehicle architectures developed by OEMs (Original Equipment Manufacturers). The car underbody, usually a stamped steel monocoque, accounts for 20–30% [1] of the total weight of a vehicle. Investigations were carried out to replace the steel structure with an alternative lightweight solution while keeping an equivalent or improved level of strength and stiffness. Aluminium in the form of sheet, castings and extrusions offers a lightweight alternative to stamped steel panels and a variety of design options such as aluminium monocoque, spaceframe or extrusion intensive design. A comparison between steel monocoque and aluminium structures is complex and depends upon production volumes [2]. Due to the increased cost of aluminium compared to steel, at higher volumes the automotive industry usually prefers a steel monocoque whereas aluminium structures have found applications on specialist or high performance cars where the increased raw material cost can be offset more easily. Table 1 below shows the approximate cost associated with different vehicle architectures versus production volumes [3]. As the production volume reduces, the monocoque option, whether in steel or in aluminium, looks less favourable compared to spaceframes. For volumes lower than 10K/year, a monocoque is difficult to justify due to high investment costs related to tooling and parts manufacturing.

1.2 The Aston Martin VH Architecture The need for flexibility led to the development of the Vertical-Horizontal (VH) architecture which could be modified both in length and width to offer a wide variety of packaging options for different vehicle models. The VH architecture is an example of extrusion intensive design. The Aston Martin Vantage V8 (2005) presented in Fig. 1 is produced at 5K/year and benefits from an aluminium bonded

Multi-Material Adhesive Joining in the Automotive Sector

197

Fig. 1 Aston Martin Vantage V8 VH underbody architecture

Fig. 2 Aston Martin V8 Vantage Body-in-White

underbody made of 172 parts for a total weight of 224 kg (excluding body panels and glass). The Body-in-White (BiW) refers to the stage where body panels and closures are assembled. Figure 2 shows the BiW components adhesively bonded at Aston Martin in the Gaydon facilities for a V8 Vantage. Body skin panels and other components are manufactured from steel and from lightweight materials such as aluminium, magnesium, and glass polyester Sheet Moulding Compound (SMC) and Resin Transfer Moulding (RTM) composites as depicted in Fig. 2. Fenders and closures such as bonnet, doors and decklid are

198

S. Pujol

bolt-on parts and are not presented in Fig. 2. The total weight of the BiW (including closures but without glass) is 329 kg.

1.3 Why Adhesive Bonding Several limitations arise with resistance spot welding of aluminium structures. Access to both sides of the assembly is required during the spot welding process which can be difficult when manufacturing structures based on extruded box sections. Spot-welding remains a capital intensive process which represents little interest for low volume production using aluminium. Compared to steel, additional complications arise when spot welding aluminium such as aluminium oxide stability, modified aluminium properties at the spot weld or fit-up of parts [4]. Extrusions will not bend and it is therefore important to find a joining technique capable of filling gaps. In addition the tip life of the welding equipment can also be an issue for resistance spot welding which could lead to increased cost, down time and redressing to prevent build up and sticking. Adhesive bonding is the preferred joining method to assemble such a wide variety of materials. Unlike mechanical fasteners or spot-welds, the adhesive is applied over large surfaces, which improves the stress distribution, the stiffness, and the noise, vibration and harshness (NVH) properties of the joints. The adhesive also acts as a sealant and prevents moisture ingress between the parts. Adhesives can be used to bond dissimilar materials by applying the appropriate surface pre-treatment to the substrates and as there is no direct contact between the parts, galvanic corrosion can be avoided. The uniform stress distribution in the joint improves the fatigue performance compared to spot-welded structures. The overall outcome is an improved body stiffness and durability [5].

2 Body Structure Manufacturing Process The body structure manufacturing process is divided into two stages: aluminium underbody and Body-in-White assembly.

2.1 Aluminium Underbody Manufacturing Process The underbody bonding process comprises several stages as presented in Fig. 3 and this section aims at presenting the adhesive bonding process used at Aston Martin to manufacture aluminium underbodies. Aluminium extrusions, castings, folded and pressed sheets and super plastically formed panels bonded together with a single component hot cure, toughened epoxy adhesive is the preferred

Multi-Material Adhesive Joining in the Automotive Sector Fig. 3 Aluminium underbody manufacturing process

Anodise aluminium parts

Oven cure adhesive

199 Apply adhesive

Assemble structure

combination of materials for car manufacturers using adhesive bonding technology such as Aston Martin. 2.1.1 Anodising Anodising is the preferred surface pre-treatment for structural bonding of low volume aluminium based automotive structures. This stage is intended to protect the parts from corrosion by growing a controlled and stable oxide layer at the surface of the substrate. The aluminium parts are anodised using a standard Direct Current—Sulphuric Acid Anodising (DC-SAA) process, which produces an average oxide thickness of 2–10 lm [6]. 2.1.2 Adhesive Application The adhesive bead is applied onto the parts using robotic dispensing to ensure consistent adhesive bead geometry and correct placement. The robot nozzle follows a sequence of adhesive bond paths, specific to the geometry of each part, which ensures a maximum coverage after assembly.

2.1.3 Structure Assembly The parts are placed onto various fixtures and the underbody is built in different sub-assemblies. At this stage the adhesive is still uncured and mechanical fasteners are used to hold the structure in shape until the adhesive curing stage. Bond line thickness is controlled by 0.2 mm ribs extruded directly onto the parts.

2.1.4 Adhesive Curing During the last stage of the bonding process the thermal cycle presented in Fig. 4 is applied to the underbody structure to achieve full crosslinking of the adhesive. The underbody passes through a batch oven made of 3 different zones: Zone 1—Pre-heating: This first stage is intended to heat the underbody structure up to 130C without starting the cure of the adhesive. Zone 2—Curing: The temperature is gradually ramped up to 190C to initiate and achieve full cure of the adhesive.

200

S. Pujol 250 Zone 1 Pre- heating

Zone 2 Curing

Zone 3 Cooling

Temperature (degC)

200

150

100

Underbody structure 50

Oven Temperature

0 0

10

20

30

40

50

60

70

80

90

100

Time (min)

Fig. 4 Oven and aluminium underbody thermal profile

Zone 3—Cooling: The structure is cooled down using airflow before being released on the production line.

2.2 The BiW Manufacturing Process Composite parts, aluminium and steel body panels are bonded onto the aluminium underbody using a low modulus, cold cure two-component (2K) polyurethane (PU) based adhesive. Such adhesives generally have large gap filling properties and are ideal for tolerance compensation and panel adjustment. The risk of bondline read-through is minimized due to the low modulus and high elongation properties, and they allow different thermal expansion of the parts during the later paint bake cycle. The different BiW components are assembled following the sequence depicted in Fig. 5.

2.2.1 Parts and Surface Preparation Unlike monocoque bodies, Aston Martin vehicles are not subjected to an electrocoating (e-coat) paint cycle after assembly. Aluminium and steel body panels are

Fig. 5 The BiW assembly process

Prepare surface

Accelerated curing

Apply adhesive

Assemble structure


201

e-coated individually to give a class-A surface finish before paint and to protect the panels against corrosion. Composite parts are abraded over the bonded areas to ensure a consistent surface finish and to avoid adhesion issues due to the presence of release agents. Each bonded area is cleaned using isopropyl alcohol and primed using an adhesion promoter prior to adhesive application to ensure durability of the joint throughout the life of the vehicle. 2.2.2 Adhesive Application and Assembly The adhesive is applied robotically to ensure correct positioning of the bead and consistent bead diameter. Dispensing of both components is monitored carefully to ensure the mix ratio is correct and consistent. Unlike the single component epoxy adhesive used for aluminium underbody manufacturing, the two-component polyurethane formulation starts to crosslink as soon as the components are mixed together and therefore it is critical to dispense the adhesive, place the parts onto the fixtures and close the joints within the maximum adhesive open time. The maximum open time permissible varies between 6 and 10 min depending on temperature and relative humidity (RH). When exposed for more than 10 min, the adhesive viscosity increases which could affect the wettability, the bond gap thickness and the coverage onto the substrate and reduce joint performance.

2.2.3 Adhesive Curing Although two-component PU adhesives cure with environmental humidity and do not require heat curing, the crosslinking process can be accelerated using localised hot air impingement or infrared heating systems. Aston Martin uses a hot air impingement system which consists of blowing hot air (*90C) onto the adhesive joints. Accelerated curing needs to be considered carefully when bonding dissimilar materials as differential thermal expansion causes the panels to expand and contract at different rates which could lead to the formation of residual stresses in the joint and subsequent bondline read-through. A typical hot air impingement cycle is presented in Fig. 6.

3 Multi Material Joint Design It is important to differentiate structural joints used for load bearing structures from semi-structural applications where the performance of the joint is secondary provided the components remain bonded during service. Aluminium components of the underbody are required to maintain body stiffness and structural integrity throughout the life of the vehicle and therefore they fall into the structural

202

S. Pujol

Fig. 6 Typical hot air impingement cycle for BiW joints

Table 2 Typical properties of body structure adhesives [7] Adhesive E-Modulus (MPa) Tensile Elongation at break (%) Glass transition Strength (MPa) temperature Tg (C) 1K epoxy 2500 2K PU 5

51 7

5 230

120 -40

category, as opposed to composites parts and body panels which are part of the semi-structural applications, although the roof panel contributes significantly to torsional stiffness. Adhesives are selected based on both engineering and manufacturing requirements. High performance structural epoxy adhesives systems are generally used for load bearing applications such as underbody manufacturing. Low modulus two-component PU adhesives are preferred for semi-structural applications such as body panels bonding. Typical properties of epoxy and PU adhesives are summarised in Table 2 [7], and Fig. 7 shows where they are used on the car.

3.1 Joint Design for Structural Applications Unlike high volume manufacturers using adhesives to complement mechanical joining (typically spot-welding or riveting), Aston Martin mainly relies on adhesive bonding to ensure the structural integrity of its vehicles. Monobolts and selfpiercing rivets (SPRs) are only used during the assembly process to hold the


203

Fig. 7 1K epoxy and 2K PU used on V8 Vantage body structure

Fig. 8 Typical aluminium underbody joint between two extruded parts

geometry of the structure prior to high temperature curing of the adhesive. Figure 8 shows a typical joint between two aluminium parts. 3.1.1 Joint Stiffness Bonded areas are usually designed over large and flat surfaces to improve stress distribution and to ensure maximum joint strength and stiffness. Joint stiffness is dictated by the combination of several parameters: substrate thickness, adhesive modulus, adhesive coverage and bond gap thickness [8–10]. Bond gap thickness and adhesive coverage are the main parameters affecting joint stiffness and joints should be designed to allow large overlaps with minimum gap between the parts. The nominal bond gap thickness is 0.2 mm, however, during production, bond gaps were observed to vary by up to 1 mm due to dimensional variability in the parts and fixturing [11]. Bond gap sensitivity should be considered during the early stage of design to minimise tolerances during manufacturing of the parts and assembly of the structure. Tolerance stack up due to parts or fixtures variability is inevitable and the structure

204 Fig. 9 The different toughening mechanisms [12]

S. Pujol

Adhesive

Plastic yield zone

Crack 2 - Cavitated rubber 3 - Rubber bridging

1 - Localised shear bands

needs to be designed to minimise sensitivity to bond gap thickness. With variable bond gaps, joint stiffness becomes much more sensitive to the adhesive modulus, and therefore it is desirable to have a high modulus material to provide a consistent level of performance across the entire range of bond gaps. Adhesive joints are generally designed to work primarily in shear, however in the event of a crash, a wide range of loading conditions will arise and it is also necessary to ensure that the joint will withstand peel loading in both quasi-static and dynamic conditions. 3.1.2 Fracture Toughness Under impact or high strain rate conditions adhesives are required either to maintain sufficient strength such that failure does not occur or sufficient toughness such that if failure does occur the energy is absorbed by the surrounding structure. The mechanical and physical properties of both cured and uncured epoxy adhesives can be modified by the addition of a toughening agent in the initial formulation. Kinloch [12] suggests different toughening mechanisms to explain how the second phase particles can help improve the fracture toughness of adhesives. The presence of rubber particles in the adhesive can improve the toughness in different ways as depicted in Fig. 9. 1. Shear yielding or shear banding between rubber particles; 2. Void growth by cavitation or debonding of the rubber particles; 3. Bridging of the crack surfaces. These toughening mechanisms result in improved peel and impact properties. The toughening effect will be a function of the intrinsic properties of the particles, the size and distribution, and the adhesion between the second phase particles and the adhesive.

3.2 Joint Design for Semi-Structural Applications Although they contribute to the overall vehicle stiffness, semi-structural joints are not intended to fulfil a load bearing role and the main requirement is to keep the parts attached throughout the life of the vehicle. Therefore, adhesive modulus and joint stiffness are secondary requirements although toughness is required to avoid body panels separation in the event of a crash.


205

Fig. 10 Typical BiW joint between the aluminium underbody, a SMC composite part and a steel body panel

Semi-structural joints can be used to compensate tolerance stack up arising from underbody manufacturing which results in a wide range of bond gap thicknesses, generally between 3 and 8 mm. Such a design allows for the body panels to be positioned consistently car-to-car, while dimensional variability of the underbody and body panels can be absorbed in the bond gaps. Figure 10 shows a typical BiW joint between the aluminium underbody and a SMC composite part and a steel body panel. Mechanical joining is usually not possible in BiW due to the wide range of materials and thicknesses seen across the structure, but also due to the lack of access and for obvious cosmetic reasons. Semi-structural joints are important for sealing purposes as they will prevent air and water ingress in the vehicle. Complete or partial joint failure could lead to air and/or moisture ingress into the vehicle and affect the overall NVH performance.

3.3 Corrosion Durability It is a fundamental requirement to understand the long-term corrosion performance of adhesively bonded structures and although it is straightforward to measure the initial performance of adhesive joints, one major issue is to evaluate how this performance will be affected over the product lifetime. Water is the main source of problems when considering adhesive joint stability. Moisture ingress may affect the adhesive bulk properties such as the elastic modulus or the glass transition temperature (Tg) [13] but the most deleterious effect of water occurs at the interface between the adhesive and the substrate where hydration of the aluminium oxide occurs, leading to the formation of pseudobohemite, weakening of the interface and premature failure of the joint. Accelerated ageing tests were developed based upon a time–temperature superposition principle to shorten the duration of the tests. Most of the durability tests consist of placing the joints in a warm humid and corrosive environment for several weeks in order to replicate severe environmental conditions. Usually, the corrosion performance of an adhesive joint geometry is measured in terms of

206

S. Pujol

initial strength retention after ageing. Aston Martin uses test procedures such as the Sea Water Accelerated Ageing Test (SWAAT) [14], the Stress-Durability test [15] or the Arizona Proving Ground (APG) vehicle accelerated corrosion test [16] for durability assessment of structural joints. For semi-structural applications based on low modulus adhesives, corrosion durability is measured in terms of failure mode changes after exposure to different environments. An adhesive bead is applied onto the relevant substrates, preliminary prepared as per a standard production part, and failure mode is measured by peeling different sections of the bead after each environmental cycle. Samples are exposed to 30 APGE cycles (adhesion measured every 10 cycles) and to a 4-weeks environmental test [17] comprising four exposures, respectively 7 days at room temperature, 7 days water immersion at 23C, 7 days at 90C and 7 days at 70C/100% RH ? 16 h at -20C (adhesion is measured at the end of each cycle).

4 Manufacturing Constraints Several constraints arise from the underbody and Body-in-White manufacturing processes. This section provides two typical examples of constraints, their impact on joint performance and how they were addressed within Aston Martin.

4.1 Effect of Processing Parameters on Epoxy Joints On a plant scale, it is often difficult to control the relative humidity and the ambient temperature, thus large seasonal variations can be expected throughout the year. Once the adhesive has been applied onto the parts, the joints are left open during the assembly of the structure. Both the assembly time and the cure cycle of the adhesive are required to be compatible with the rest of the manufacturing process. The physical and chemical properties of both the adhesive and the substrate are sensitive to moisture uptake when exposed to humid environments [18]. The porous aluminium oxide layer present at the surface of the substrate is particularly sensitive to moisture uptake. The moisture entrapped in the oxide layer and the adhesive during the assembly of the structure is released into the adhesive during the high temperature cure cycle, which can create porosity and cracks within the adhesive joint as depicted in Fig. 11.

Fig. 11 Different levels of voids within the bonded area


207

1

Fracture toughness 0.9

Fracture toughness (kJ/m2)

0.8 0.7 0.6 - 60% 0.5 0.4 0.3 0.2 0.1 0 5

10

16

34

Void area (%)

Fig. 12 Effect of different void areas on fracture toughness of aluminium bonded joints

4.1.1 Effect of Voids on Joint Integrity The presence of voids within the bonded area could have a detrimental effect on the joint integrity. Figure 12 shows the influence of different void areas on the fracture toughness in peel mode of aluminium bonded samples [19]. Samples were prepared and tested using the relevant standard [20]. The different void areas were achieved by applying different exposures (time and relative humidity levels) to the samples prior to assembly and cure. Increasing the amount of voids up to 34% reduced the fracture toughness of the adhesive by 60%. Other joint properties such as lap shear strength or T-Peel strength were also observed to be sensitive to the presence of voids within the bonded joint [10, 19]. In order to minimize the presence of moisture in the joint and subsequent void formation, the original process used at Aston Martin included a high temperature thermal cycle (so called ‘‘degas’’) applied to the parts prior to adhesive application and a maximum joint open time of 4 h after adhesive application.

4.1.2 Pre-Bond Moisture Uptake in the Adhesive Joint Moisture uptake was measured in 6060 aluminium plates anodised using a DC-SAA process to the standard condition (*4–5 lm) during exposure at 22%, 35%, 65%, 80% and 95% relative humidity. The plates were subjected to a 30 min degas cycle at 190C and allowed to cool down at ambient temperature prior to humidity exposure and gravimetric analysis. Figure 13 [19] shows the mass uptake measured for the different samples during humidity exposure.

208

S. Pujol 300 RH = 22%

RH = 35%

RH = 65%

RH = 80%

RH = 95%

250

Mass uptake (mg/m2)

200

150

100

50

0 0

20

40

60

80

100

120

140

Time (min)

Fig. 13 Mass uptake for anodised samples at different RH levels [19]

The mass uptake recorded for the samples generally followed the same trend. The mass increased sharply within the first 20 min of exposure then reached an equilibrium plateau after 30–60 min depending on the RH level. The maximum moisture uptake at equilibrium was found to be a function of the humidity level and aluminium oxide layer thickness: samples exposed to high humidity levels or samples with a thick oxide layer absorbed more moisture. Further investigations [19] revealed that moisture uptake kinetics was very fast and 80% of the maximum mass uptake was achieved within the first 30 min of exposure, which questioned the efficiency of the degas cycle applied prior to adhesive application. Adhesive beads were applied onto glass slides and exposed to different humidity environments. The bead diameter was calibrated at 8 mm to reflect the standard production conditions. Figure 14 [19] shows the results obtained after exposure between 16% and 90% RH. When exposed to 16% RH the adhesive lost weight which indicated that moisture already present in the adhesive was released into the environment. Generally, moisture uptake followed a slow exponential trend for the first 20 min of exposure with the slope dictated by the humidity gradient between the sample and the environment. Once the adhesive bead surface was fully covered with a monolayer of moisture, water molecules were allowed to diffuse into the adhesive film which translates to the slower absorption rates after 20 min. Moisture sorption was faster in samples conditioned in high RH environments.

4.1.3 Alternative Pre-Treatments and Void Formation In an attempt to reduce the amount of moisture in the adhesive joint, samples were anodised with the standard DC-SAA process to produce thin oxide layers (B2 lm), however the corrosion durability performance of the joint with a thin


209

1.2 RH = 16%

RH = 35%

RH = 67%

RH = 75%

RH = 90%

1

Mass uptake (mg/g)

0.8

0.6

0.4 Problem with micro -balance

0.2

Moisture desorption from adhesive bead

0 0

20

40

60

80

100

120

140

160

180

200

-0.2

Time (min)

Fig. 14 Mass uptake for the epoxy adhesive bead at different RH levels [19]

Table 3 Parameters used to evaluate the moisture sensitivity of the various surface pre-treatments Alloy Factors Levels 6060

Pre-treatment/coating thickness

Degas/conditioning (aluminium substrate ? adhesive bead)

Hot AC-SAA/\1 lm ACDC-PSAA/2–3 lm DC-SAA/5–6 lm Keronite/5 lm Cr3+/\1 lm Cr6+/\1 lm No degas No conditioning 30 min at 190C 4 h at 75% RH

anodic layer could be dramatically reduced for some alloys and decreasing the oxide thickness was not considered as a viable solution. Alternative pre-treatments were investigated to provide the aluminium substrate with a thin and corrosion resistant coating. The alternative pre-treatment processes provided the aluminium samples with coating thicknesses between 1 and 5 lm, and the sensitivity to moisture uptake of the coatings obtained after the different pre-treatment processes was evaluated by applying different degas and pre-bond humidity exposures to the substrates as listed in Table 3. Hot Alternating Current Sulphuric Acid Anodising (Hot AC-SAA) and combined ACDC Phosphoric-Sulphuric Acid Anodising (ACDC-PSAA) were benchmarked against the current DC-SAA process. Other pre-treatments such as Keronite, hexavalent chromium (Cr6+) and trivalent chromium (Cr3+) were also included in this study. Although chromium-based pre-treatments are out of the scope of this study because of environmental regulations applying to the

210

S. Pujol 12 No degas + No RH

11

190C / 30 min + 4hrs / 75% RH

10 9

Void area (%)

8 7 6 5 4 3 2 1 0 ACDC-PSAA

Hot AC-SAA

DC-SAA

Keronite®

Cr (3+)

Cr (6+)

Fig. 15 Void area in the samples prepared with different pre-treatments

automotive industry, they were considered as reference points for corrosion durability and moisture sensitivity. Figure 15 shows the amount of voids measured in the various samples after different degas and humidity exposure conditions were applied. Generally, samples prepared with alternative processes were not sensitive to the degas cycle or humidity exposure and the maximum amount of voids measured was 4.5% for the Cr6+ process. The low moisture sensitivity of these samples was attributed to the thinner coatings obtained by the alternative pre-treatments combined to the nature of the chemistry itself. The corrosion durability performance of the various pre-treatments was evaluated by measuring the lap shear strength loss of the various samples after SWAAT. Figure 16 shows how the different samples were affected after the 6-week corrosion cycle. With a maximum lap shear strength loss of 25%, the Keronite process showed similar corrosion durability performance to the standard DC-SAA process. Other alternative anodising processes generally performed better than the standard DC-SAA process. The ACDC-PSAA process performed equally with the trivalent chromium process. The thin oxide layer generated during the ACDC-PSAA process outperformed oxides obtained by the standard DC-SAA process which showed a lap shear reduction of 15% after SWAAT. Alternative anodising processes were observed to deliver thin oxide layers (\3.5 lm) and to limit moisture uptake and void formation after conditioning while maintaining a sufficient level of corrosion protection. Independently of the degas or humidity exposure applied to the samples, the amount of voids measured within the samples prepared with hot AC-SAA or ACDC-PSAA was below 2%. The presence of hydrophobic phosphates within the film obtained after ACDCPSAA [21] also accounted for the low moisture sensitivity during humidity conditioning and thus the low amount of voids measured within the joints. The ACDC-PSAA process had all the attributes required to avoid moisture uptake and


211

30

No Degas + No RH 190C / 30 min + 4h / 75%RH 25

LSS Loss (%)

20

15

10

5

0 ACDC-PSA

Hot AC-SAA

DC-SAA

Keronite®

Cr (3+)

Cr (6+)

Fig. 16 Lap shear strength loss after SWAAT for samples prepared with the various pretreatments

void formation while offering significant initial strength and protection against corrosion. However, the compatibility of this process with other alloys remains to be demonstrated. In addition to technical constraints, the ACDC-PSAA process is not widely available commercially, limiting the number of potential suppliers which represents a risky business strategy for Aston Martin.

4.1.4 Alternative Adhesive Formulation and Void Formation The results presented above show that it was possible to reduce void formation by using specialist pre-treatments. Such pre-treatments might not be a viable solution and therefore modified adhesive formulations were investigated as a potential way to reduce void formation. The formulation of the current one-component structural epoxy adhesive was modified by the supplier to reduce its moisture sensitivity while keeping the adhesive performance at the same level. Three different alternative formulations were investigated: F001, F002, and F003. The three alternative formulations were benchmarked against the standard adhesive. Aluminium samples were anodised with a 10 lm oxide layer to reflect the worst conditions in terms of substrate sensitivity to moisture. Both the aluminium samples and the adhesive were exposed to 75% RH for 4 h without high temperature degas cycle prior to adhesive application. Samples prepared with the alternative formulations F001, F002 and F003 showed negligible amount of voids with the maximum void area measured at 0.5%, compared to 21% for samples prepared with the standard formulation. Figure 17 shows the samples after testing. The adhesive formulation was observed to be the most sensitive factor affecting the presence of voids within the joint. Modification of the standard formulation

212

S. Pujol

Fig. 17 Samples prepared with the different adhesive formulations after testing

helped in preventing void formation independently of the exposure time, the humidity level or the thickness of the oxide layer generated during the anodising process. Formulation F002 was further developed in collaboration with the adhesive supplier and submitted to the full Aston Martin Design Verification Plan (DVP) to ensure compliance with the other engineering and manufacturing requirements. The alternative formulation was introduced into production and the manufacturing process was modified to remove the degas process prior to adhesive application, which resulted in significant cost saving, increased production flexibility and improved product robustness.

4.2 BiW Adhesive and Cycle Time Other manufacturing constraints arise from the BiW area with conflicting requirements such as adhesive open time and fixture time. A long open time is desirable to help during adhesive application while short fixture time is necessary to reduce cycle time and improve manufacturing flexibility. The open time is the time frame between adhesive dispensing and joint closure and it is required to allow dispensing of the adhesive over large panels and for positioning the panels into the fixtures. Open time is critical when the adhesive is applied manually due to the reduced dispensing speed compared to robotic application. Excessive open time could result in dramatic loss of adhesion due to reduced wettability, limited adhesive coverage and increased joint thickness due to viscosity build-up and therefore reduced squeeze-out. Open time is usually dictated by several parameters such as adhesive chemistry, bead diameter, temperature and relative humidity. Typical open time for BiW two-component PU adhesives used at Aston Martin varies between 6 and 25 min depending on application and adhesive formulation [7]. Open time requirements are conflicting with fixture time. Due to the lack of mechanical fasteners, body panels have to remain clamped in the fixtures until the adhesive has developed enough strength to prevent the panels from springing back or from sagging under their own weight upon removing of the clamping force. Body panels are large and of complex geometry which can lead to tolerance issues during manufacturing and pre-constraints in the joint during the later bonding process. The strength required to hold the panels safely into position when indexing the car onto the next station, so called ‘‘de-jig’’ strength, is a complex


213

interaction between adhesive crosslinking and modulus, adhesion developed onto the substrate, joint geometry, panel position and dimensional quality. Additional operations such as rework, or surface preparation occurring after the car has been indexed will also dictate the minimum de-jig strength permissible. Such a number of variables make the determination of the minimum de-jig strength difficult and application specific which is why it is often determined empirically. Adhesive crosslinking and adhesion onto the substrate are time, temperature and relative humidity dependent, although humidity was found to have a limited influence when compared to time and temperature. Thus, minimum cycle time will be limited by the time required to achieve the minimum de-jig strength in various temperature and humidity scenarios. Figure 18 shows the effect of cure temperature on the lap shear strength achieved after 40 min on typical anodised aluminium to e-coated aluminium joint. The data presented above show that seasonal temperature variations could affect dramatically the lap shear strength and that a minimum temperature of 30C is necessary to achieve the minimum de-jig strength recommended by the supplier after 40 min. In order to achieve the de-jig strength, the adhesive must fulfil two requirements: • Crosslinking to develop a modulus high enough to minimise spring back and to prevent the panels from sagging under their own weight; • Develop adhesion onto the substrate to ensure a strong interface. Fast curing products may appear more attractive, however they do not always offer the best manufacturing alternative. Figure 19 shows the peel strength achieved on e-coated steel samples for both a slow and a fast curing polyurethane adhesive formulation of similar elastic modulus after several de-jig time at 21C.

0.400 Lap shear strength after 40min (kN)

0.350

Lap shear strength (kN)

0.300

0.250

0.200

0.150 De-jig strength from supplier

0.100

0.050

0.000 15

25

30

Temperature (degC)

Fig. 18 Effect of temperature on lap shear strength for a typical BiW joint

35

214

S. Pujol 45 Fast curing PU Slow curing PU

40

35

Peel strength (N)

30

25

20

15

10

5

0 40

50

60

De-jig time (min)

Fig. 19 Peel strength versus de-jig time for a slow and fast curing PU adhesive

E-coat failure

Interfacial failure

Cohesive failure within the adhesive

SLOW PU

FAST PU

100 90 80

Failure mode %

70 60 50 40 30 20 10 0 40 min

50 min

60 min

7 days

40 min

50 min

60 min

7 days

De-jig time

Fig. 20 Failure mode versus de-jig time for the two PU adhesive formulations

Although the fast curing adhesive formulation was expected to develop strength quicker, results presented in Fig. 19 show the opposite. The analysis of the failure modes observed for the different samples is presented in Fig. 20, and it suggests that although the fast curing formulation cures more quickly, its adhesion onto the substrate develops more slowly. After 40 min, neither of the two adhesives has


215

400

350 Peel strength for fast PU 300

Peel strength (N)

250

200

150

100

50

0 40min RT

7days RT

60minRT

12min 80°C

Cure cycle

Fig. 21 Influence of hot air impingement on peel strength of bonded joints

achieved sufficient crosslinking and therefore the peel load applied during testing results in 100% cohesive failure within the adhesive. At this stage the adhesive is still in a paste-like state and fails within itself. After 50 and 60 min, the slow curing PU adhesive has only built up a limited modulus due to its slower curing kinetics, and therefore peel strength increases slightly and the failure mode remains mainly cohesive within the adhesive. In contrast, quick modulus build up for the fast curing formulation has increased significantly but adhesion onto the substrate is still weak, and any stress applied onto the joint will concentrate at the interface and initiate interfacial failure, which results in reduced joint performance. The failure modes obtained after seven days at ambient temperature show that both adhesives were compatible with the substrate and could develop a good adhesion over time. In order to minimise the impact of environmental parameters and to reduce cycle time, Aston Martin uses a hot air impingement system to speed-up both the adhesive crosslinking and the adhesion onto the substrate. The hot air is blown directly onto the joint which generates joint temperatures between 70 and 90C as depicted in Fig. 6. Figure 21 shows the influence of a 12 min hot air impingement cycle at 80C on the peel strength for the fast curing PU adhesive. It should be noted that the failure modes were 100% cohesive within the adhesive. The data presented in Fig. 21 above show a dramatic improvement in joint performance combined with a massive de-jig time reduction when using hot air impingement. Hot air impingement offers a suitable solution to improve process robustness and flexibility, and although it represents a capital investment, it

216

S. Pujol

remains a cost effective and convenient solution for niche market vehicles manufacturers compared to other systems such as induction curing.

5 Conclusions and Future Trends This chapter gave an overview of the materials and associated joining technologies used at Aston Martin to manufacture lightweight high performing automotive body structures. Some of the constraints related to the manufacturing of adhesively bonded structures for automotive applications were highlighted and it was demonstrated how they were addressed within Aston Martin. Joining is probably one of the biggest challenges the automotive industry is facing at the moment. Environmental legislation for fuel consumption and CO2 emissions are putting more emphasis on weight reduction which pushes designers to use lightweight materials that cannot be joined using traditional methods. Adhesive bonding is a potential alternative to spot-welding. However, structural crash-resistant epoxy adhesives generally require curing at elevated temperatures, which limits both material selection and design flexibility due to differential thermal expansion. Furthermore, such adhesive bonding processes are time and energy consuming and although they are suitable for niche market vehicle manufacturers such as Aston Martin, they would need further developments to use for higher volume production. Two-component cold cure structural adhesives could be an option to use more dissimilar materials and improve vehicle weight and performance. Nevertheless, such adhesives often suffer from a loss of performance [22] when compared directly to high temperature curing systems and designers have to deal with limitations in terms of elastic modulus, toughness, glass transition temperature and corrosion durability. Further work is required in collaboration with adhesive manufacturers to develop robust and high performance two-component cold cure systems that can match the performance of one component high temperature curing epoxy toughened adhesives. Alternatively, another way to introduce two-component adhesives and increase the material mix in the body structure would be to use finite element modelling to optimise the joint design. Current adhesive modelling techniques used at vehicle level are usually simplified to reduce calculation resources and further progress in this area is also desirable to understand the impact of a lower performance adhesive on both the static and dynamic performance of the joint. Surface pre-treatment is paramount to achieve a strong and durable joint and it should be considered carefully when joining dissimilar materials due to the risk of galvanic corrosion. While standard OEM processes include an electro-coat paint stage at the end of the assembly process to provide the entire body with corrosion protection, Aston Martin components are pre-treated individually and bonded directly without any further corrosion protection. It is therefore essential to select and develop suitable pre-treatments for individual joint configuration.


217

References 1. http://www.eaa.net/aam—October (2007) 2. Personal discussion with Dr John Hill, Adhesive Technical Leader, Ford Motor Company, Research and Innovation Centre, PO Box 2053, MD3135, Dearborn MI48124-2053, USA, October (2007) 3. http://www.ussautomotive.com—October (2007) 4. Barnes, T.A., Pashby, I.R.: Joining techniques for aluminium spaceframes used in automobiles Part I—solid and liquid phase welding. J. Mater. Process. Technol. 99, 62–71 (2000) 5. Barnes, T.A., Pashby, I.R.: Joining techniques for aluminium spaceframes used in automobiles Part II—adhesive bonding and mechanical fasteners. J. Mater. Process. Technol. 99, 72–79 (2000) 6. Hill, D.J.: Anodising of aluminium. Engineering Specification AMES-005-Anodising (2005) 7. Data provided by Dow Automotive 8. Personal discussion with Roland Snell, CAE Manager, Body Structures, Aston Martin, Banbury Road, Gaydon, CV35 0DB, UK, January (2008) 9. Grant, L.D.R., Adams, R.D., da Silva, L.F.M.: Experimental and numerical modeling of single-lap joints for the automotive industry. Int. J. Adhesion Adhesives 29, 405–413 (2009) 10. Grant, L.D.R, Adams, R.D., da Silva, L.F.M.: Experimental and numerical analysis of T-peel joints for the automotive industry. J. Adhesion Sci. Technol. 23, 317–338 (2009) 11. Tressel, X.: Bond gap thickness variability study on 30 V8 Vantage underbodies. Aston Martin internal report (2007) 12. Kinloch, A.J.: Adhesives in engineering. In: Proceedings of the Institution of Mechanical Engineers, 1997, 211, Part G 13. Comyn, J.H.: Adhesive Bonding. The Royal Society of Chemistry/Information Services, Herts, UK (1997) 14. ASTM G85-02, Standard practice for modified salt spray (fog) testing 15. Ford laboratory test method BV 101-07, Stress durability test for adhesive lap-shear bonds (2002) 16. Ford laboratory test method BI 123-01, Painted sheet metal corrosion test (2005) 17. Aston Martin Engineering Specification 006, Test specification for pumpable sealers (2005) 18. Pujol, S., Johnson, M.S., Warrior, N.A., Kendall, K.N., Hill, D.J.: The effect of processing parameters on reactive epoxy adhesives. In: International Conference on Composite Materials 16 proceedings, Kyoto (2007) 19. Pujol, S.: Optimisation of adhesive bonding for aluminium automotive structures. Doctoral Thesis, School of Materials, Mechanical and Manufacturing Engineering, University of Nottingham (2009) 20. Ford Standard Test Method, Impact peel tests for adhesive bonds, BU121-01 21. Cartwright, T.: Non-standard Anodising Processes for the pre-treatment of structurally bonded aluminium alloys. Doctoral Thesis, Institute of Polymer Technology and Materials Engineering, Loughborough University (2005) 22. Norton, T.W., Pujol, S., Johnson, M.S., Turner, T.A.: Cold-Cure Adhesives, for the Use in Structural Aluminium Bonding. Advanced Computational Engineering and Experimenting, Paris (2010)

Welded Aeronautical Structures: Cost and Weight Considerations S. M. O. Tavares

Abstract Product development is limited by engineering design capabilities. Engineering design is one of the most important phases during the development of a new product, particularly in the case of complex and safety-critical systems, in order to consider all safety concerns, e.g. in aircraft and nuclear power plants. In these cases, the introduction of new design concepts and solutions is tightly tapered by existing materials and manufacturing processes. In this chapter, a breakthrough joining process—friction stir welding (FSW)—is discussed from the point of view of manufacturing costs. Friction stir welding is a solid state welding process, widely considered one of the most relevant advances in welding technologies in the last decades. The final cost of the manufacturing processes has a fundamental role in the success of its infusion and massification since this is one of the most important drivers in almost all industries. The application of FSW in new product development for aerospace components allows large weight and cost savings.

1 Introduction Civil aviation, as other transportation sectors, has been experiencing along its history vast improvements in efficiency and effectiveness. These improvements have been following different objectives due to the diverse market needs along the time. At the beginning of the aviation era, this sector was driven by ‘‘Faster, Higher and Farther’’ goals in order to take the advantages of the air transportation

S. M. O. Tavares (&) Faculdade de Engenharia da Universidade do Porto, Rua Dr. Roberto Frias, s/n, Porto, Portugal e-mail: [email protected]

Adv Struct Mater (2012) 8: 219–237 DOI: 10.1007/8611_2011_58 Springer-Verlag Berlin Heidelberg 2011 Published Online: 4 August 2011

219

220

S. M. O. Tavares

to medium and long range travels. The ‘‘Faster, Better and Cheaper’’ (FBC) drivers emerged after the Cold War due to the increased resources scarcity and due to the globalization and massification in the sector. More recently, ‘‘Quieter, Cleaner and Greener’’ or ‘‘More Affordable, Cleaner and Quieter’’, [29], have been trying to replace the ‘‘Better, Faster and Cheaper’’ drivers due to the impact of the transportation in the environment changes, which should be reduced drastically in order to achieve more sustainable transportation systems. For instance, the European Union has been concerned with the impact of aviation in the global environment, and aims to achieve several goals for 2020 proposed by the Advisory Council for Aeronautics Research in Europe originating the Clean Sky initiative [26]. This initiative has the ambition to reduce the perceived noise to one-half of the actual values, and aims at 50% reduction in CO2 emissions and 80% nitrogen oxide ðNOx Þ emissions by 2020, [26]. International organizations, as the (CAEP), have also been concerned with minimizing the aviation’s effects on the environment regarding to emissions and noise, [20]. Undoubtedly, safety is the major concern, overriding any design driver. Historically, the main motivation for safety improvement was economical, linked to insurance companies policies, and not an intrinsic social requirement. Nowadays, all aeronautical systems require to be airworthy, which means that they are required to guarantee safe conditions during all the flight phases. All aircraft and related systems comply with necessary requirements so that the aircraft flies in safe conditions and the allowable limits are respected during all phases (i.e. maximum weight or maximum speeds), [11]. Behind the assumption that the safety must be maximized for all aircraft and the severity of the aeronautical certification, the economical point of view should be also considered in order to achieve an economically viable product. Figure 1 shows schematically a theoretical trend of safety as a function of cost. A 100% safe airplane is unrealistic, although it is certainly possible to invest indefinitely in safety, improving it. For instance, it is possible to invest indefinitely in the design of redundancies of the airplane systems improving the safety, however the improvement will be very low compared with the increase of cost. A trade-off between cost and safety is always required in order to have and economically feasible product. Aircraft designers as well as airline operators have to find a compromise between economy and safety. Nevertheless, all aeronautical sector is driven by cost, as observed by Murman et al. [36]. It is observable that the main advances over the years in this sector have been been focused into reduction of the global costs, increasing simultaneously reliability and performance. These costs might be considered a transversal target since most of the drivers can be converted in costs (as the environmental or performance costs). The cost reduction in the aircraft might include optimization in development, manufacturing, operational and disposal costs. For instance, in the development of a new aircraft, it is a common option for the constructors to try to share the design and characteristics of parts in order to reduce development costs [35]. One way to do this is to conceive a family of aircraft who share common parts and characteristics, such as platforms and systems, but each one satisfying a

Welded Aeronautical Structures: Cost and Weight Considerations 100% Safe

Economically feasible

Safety

Fig. 1 The safety and cost trade-off

221

Cost

different mission requirement (as the Boeing 757 and 767 or the Airbus A330 and A340 or the Embraer E170 and E190). The cost of any product or component is significantly committed at the conceptual design stage. For more successful programs of new aircraft development or even new components development with infusion of new manufacturing processes, the cost evaluation is fundamental since the major fraction of the total life cycle cost for a product is committed in the early stages of design, see e.g. [37]. Not with standing the major drivers of aircraft industry mentioned above, the aircraft manufacturing is a highly cost-driven sector. The tools used for multidisciplinary design cycle analysis have also taken into account the cost analysis, as in the tool proposed by van der Laan and van Tooren [48]. These cost analyses can be done through different techniques, each one with its own specific characteristics, advantages and disadvantages. Curran et al. review these techniques in [7] based in four major groups: • Analogous cost estimation: product cost is estimated by comparing it with previously produced similar products. Taking into account technical differences between the product of which the cost is to be estimated and similar products that provide the analogous data can refine the cost estimation; • Parametric cost estimation: the cost of a product is linked to technical parameters such as weight, size or part count. To link cost to the technical parameters, relations are developed based on historical data, using statistical techniques; • Bottom-up cost estimation: in this approach, the cost of all entities in the workbreakdown structure of the product are determined; for instance, the cost of producing a joint based on all actions and materials involved in the process of producing the joint. The main disadvantage of this method is that it is very information intensive; • Genetic causal estimation: the genetic causal cost modeling methodology imposes a breakdown of the cost into a number of cost elements, including material cost, fabrication cost and assembly cost; so that cost can be formulated into semi-empirical equations to be linked to the same design variables as considered in the structural analysis.

222 100,000

Rocket

Cost per pound [$/lb]

Fig. 2 Costs per pound versus transportation speed, illustrative trend

S. M. O. Tavares

10,000 Military Jet 1,000 Aircraft 100 10 Car

1 10

100

1,000

10,000

100,000

Velocity [km/h]

As concerns weight, it has a direct impact in most of the goals as it is detrimental to most of the variables in the aircraft design goals. For instance, Manufacturer’s Empty Weight (MEW) reduction leads to improvements in the performance, reducing the fuel consumption and contributing to cleaner and greener aircraft. It is expected that a reduction of 1% in the MEW will reduce fuel consumption between 0.7% for larger aircraft and 0.75% for smaller aircraft [25]. Figure 2 shows the trend proposed by Mendez and Eagar [31], which correlates the vehicle speed with the transportation cost per pound, assuming the fuel cost between one to two dollars per gallon ( 3:8 liters). This trend has been changed due to the increase of oil costs and increase of the transportation efficiency. Nevertheless, is is expected that this exponential trend keeps unaltered nowadays. Babikian et al. [3] noticed that structural efficiency have decreased between 10% and 25% for regional and large aircrafts between 1959 and 2000 even after all efforts to improve this factor. The authors defend in this study that this efficiency reduction is related to the structural changes to improve aerodynamics, reinforcements to integrated new inflight systems and to accommodate increased engine weights. This chapter discusses and assesses the input of the adoption a welding process in the weight reduction and its cost compared with the riveting process. This assessment is based in generalist assumptions for medium and large civil aircrafts in order to achieve generic comparisons and to determine some of the advantages and disadvantages in the application of welding process such as friction stir welding (FSW) in this kind of structures. The long overlaped and riveted joints that are intensively used in the current aeronautical structures will be the main focus of this analysis, since a simplification of the joint design is straightforward and it can add significant value to the final structure.

2 Joining Processes for in Aeronautical Structures Joining processes are essential components of the manufacturing and assembly effort in most of engineered mechanical parts. For instance, data published in 2008 states that material joining technologies are responsible for approximately 7.1% of

Welded Aeronautical Structures: Cost and Weight Considerations

223

the manufacturing added value in the vehicle transportation sector in Germany, [14]. These joining processes are fundamental in the design and fabrication of large civil aircraft due to the semi-monocoque construction, requiring thousands of meters of structural joints in order to join all parts. Usually, these joints produce weakness points due to stress concentrations, areas exposed to corrosion and prone to manufacturing defects during the joining operations. From an idealistic (and from a Lean philosophy) point of view, structural joints should be avoided since they do not originate direct added value and they increase the structural weight. Also, the reduction of the number of parts is one of the guidelines of the design for assembly, to decrease the productions costs and lead times. Nevertheless, it is completely unfeasible to create large reinforced structures composed by just one piece. In addition, from the manufacturing and assembly side, large structures can create difficulties in their manufacturing and in tolerance management, and from maintenance side, the replacement of parts is more unsuitable. Three main joining processes are used to join the structural aeronautical parts: fastener joining, bonding and welding, fastener joining being the predominant process.

2.1 Fastener Joining Fastener joining is widely used to manufacture aeronautical structural joints due to a number of advantages: it is a low cost process which does not require highly skilled operators; it can join entirely dissimilar materials, it does not change material microstructure, it does not require special joint preparations, and structural parts can be disassembled an replaced without damaging the remaining parts. In addition, this joining method allows material physical discontinuity, i.e., the damage propagation through the joint may be arrested due to the absence of a continuous material path. Hundreds of different types of rivets and fasteners can be found in a single aircraft due to the different structural and aerodynamics requirements. Along the airframe joints, the countersunk head rivets or the blind head rivets are the most common rivets, although more complex geometries can be found, [4]. The countersunk head rivets were essential in the fuselage skin in order to not deteriorate the aerodynamics. Hi-shear rivets are also common in this structure to improve the shear strength, where instead a solid shank is used a collar to plugging improving its strength. In situations of removable panels for maintenance or other operations, is applied threaded fasteners, as the Hi-Lok fasteners composed by a threaded pin and threaded locking collar. The application of these fasteners still is a labor intensive process, requiring specialized workers to perform the several tasks as hole drilling, countersinking, deburring, riveting, shaving and sealing, consistently. The aircraft manufacturers have been attempting to carry out this work automatically, however in some components this is difficult due to the physical access to both parts of the joints and due to the accuracies required, [43]. Nevertheless, automated riveting has been

224

S. M. O. Tavares

used successfully and has been replacing the manual riveting in many parts, as the A380 wing box assembly at Airbus, [43], A320 wing panels, [19], and in skin panels and covers.

2.2 Adhesive Bonding Early application of adhesive bonding were in primary aeronautical structures since the origins of the aircraft due to the use of wood parts. After the transition of wood to metals materials, bonding fall out of use. In 1941 a vinyl-phenolic adhesive was developed and was applied in the Havilland Hornet fighter for stringer-skin joints, [9]. The application of adhesives in primary structures has been found in metallic and non-metallic joints, as airframes joints joining the stringers to the skin in the fuselage and wings or the metallic sheets to the honeycomb cores (applied for instance in the elevators and spoilers), [17]. The joint strength under shear loading is satisfactory. However, in joints where the major stresses are tensile, adhesive bonding is avoided due to the possibility of peeling failures. The proclaimed advantages of the bonding joints are related to the integral joint (the joint is joined continuously), when compared with riveted joints: points of stress concentration are reduced, and in joints of two metallic parts, they do not contact due the adhesive layer, avoiding fretting, [8, 44]. The most typical adhesives in aeronautical structural joints are based on the phenolic or epoxy resins with hot or cold cure. In the hot cure, the adhesive reaction is based on condensation reactions (as the Hexcel Redux adhesives) and in the cold cure, a catalytic reaction occurs to cure the adhesive, requiring a catalyst that is mixed with the adhesive (the most common in these group are the Araldite adhesives), [17]. The application of adhesives has been growing in the composite materials structures, as in the Boeing 787 where epoxy resins are applied to bond various laminates and fuselage parts, [39]. The adhesives can be also applied as films and pre-impregnated to produce and join hybrid panels (as the aluminium composite laminates or GLARE). In critical applications, adhesives are applied together with other joining processes, as riveting or fastening, improving the joint strength by 1.5–2 times, and enhancing the joint reliability and durability, [40]. When adhesives are exposed to extreme environments, their static and fatigue strength can be compromised. For instance, the humidity exposure of adhesive bonded aluminium structures can severely affect their performance, reducing its strength, [2]. In this study, Ashcroft et al. showed also that in adhesive bonded composite structures the temperature has a considerable effect on mechanical strength and it was responsible for the failure in most of tested cases.


225

2.3 Welding Welding is a joining alternative mainly applied in metallic structures, however it can also be used to join ceramic and thermoplastic polymer components. This process is widely applied in many different sectors due to the high joint efficiency without substantial weight penalty. This joining process can be easily fully automated and in most of applications is an inexpensive process when compared with the fastened applications. An important drawback is the disassembly of this joints since it cannot be done without the destruction of the weld, [32]. Its application in aeronautical structures is an attractive option since it allows joints with less stress concentration points and might be applied efficiently without overlapping the two joining parts (with a butt-joint configuration) reducing the joint weight. This weight reduction can have small impact in the production costs, but has huge impact in the life cycle costs. However, the application of welding process has been limited due to two major reasons: low crack arrest in welded joints compromising the structural integrity and the weldability of the aluminium alloys used in the airframes. Several welding processes can now deal with the weldability of the hardened precipitated aluminium alloys, although the low crack arrest persists and is an obstacle for massive adoption of this joining process. Nevertheless, the application of the welding process has been growing to join metallic structures. Electron Beam Welding (EBW) was been adopted to join titanium parts in military aircraft, as in Lockheed Martin F/A-22, where the manufacturer GKN used EBW to join the different parts of the aft boom, reducing by approximately 75 percent the use of fasteners, [5]. Laser Beam Welding (LBW) was used by Airbus to join the fuselage stringers to the skin in the A318, A340 and A380 eliminating thousands of fasteners, however only the lower panels of the fuselage since that the stresses are lower due to the compression, being less exposed to fatigue cracks. These two welding processes are based on high concentrated energy beams which originate small heat affected zones and the distortion. Other welding processes had been adopted by aeronautical manufacturers as the Gas Tungsten Arc Welding (GTAW), Plasma Arc Welding (PAW) or Variable Polarity Plasma Arc Welding (VPPA) and diffusion welding, but just for specific applications, [30, 31]. FSW has constituted a breakthrough welding process since the welding occurs without fusion. It has been experiencing a huge growth since it was invented, with applications in multiple domains such as ship, automotive, train, aeronautical and aerospace structures. Most of these applications are linked to aluminium alloys, as a result of the process readiness and reliability in aluminium alloys joining. Precipitation hardening (pptn) aluminium alloys (mainly 2XXX and 7XXX series) are widely used in aeronautical structures, but their low weldability with conventional fusion welding processes due to the hot cracking phenomenon, makes them impracticable for structural parts when high strength joints are required. Therefore, the application of FSW in aeronautical structures is promising and has been raising the interest of aircraft structure manufacturers.

226

S. M. O. Tavares conventional fuselage

10 panels

future fuselage concepts

8 panels

4 panels

2 panels

1 panel

Fig. 3 Reduction of the number of fuselage panels (adapted from [35])

3 Integral Structures Reinforced panels composed by multiples parts (as skin, stiffeners, frames, doublers and others), are the major structural components of an aircraft. Traditionally, most of these components are joined by riveting. In order to optimize these panels from the manufacturing and operational points of view, new joining processes have been investigated to reduce structural weight without compromising the structural integrity during the life cycle of an aircraft. The integral and differential panels are two design options to produce reinforced structures using different joining processes. Differential panels are the conventional solution, where the parts that compose the structure are joined differentially (wich means that the joint is joined in discrete points with rivets, fasteners or screws), whereas in the integral solution the joints are integrally joined, i.e. the different parts are joined continuously. These integral panels have been the object of considerable interest by different aircraft manufactures and research organizations due to the possibilities to reduce manufacturing and operational costs, structural weight and complexity. An example of the research programs carried out is the Integral Airframe Structures (IAS) Program managed by (NASA), where several techniques to create reinforced stiffened panels were investigated, [34]. In this program, it was pointed out that, compared to conventional riveting processes, the high-speed machining of an exemplary stiffened panel, from an aluminum plate, would yield a recurring cost savings of about 61%. In addition, the number of parts that compose the exemplary panel dropped from 78 to 7 parts for machined panels, a significant reduction with repercussions in all product life-cycle. Other advantage of the application of the integral joints, such as welding processes, in the reinforced stiffened panels is the capability of the production larger panels for the final assembly of the fuselage reducing substantially the assembly lead times. The increase fuselage panels in order to create ‘superpanels’ has been investigated by several aircraft manufactures and is reported by Munroe et al. in [34], reducing the number of panels required to make a fuselage barrel, Fig. 3. Other alternative is to increase the panel length, this option was adopted by Airbus for the fuselage panels in the A350 XWB, [22]. Some rough numbers of cost and weight reduction due to the reduction of the number of panels


227

Table 1 Cost reduction due the decrease the number of fuselage panels, [35] No. of Panels/Superpanels 10 (Basis) 8 (%) 6 (%) Engineering Cost Material Cost Part Fabrication Cost Assembly Cost Weight

(a) Avionics

Systems Integration

10 5 20 15 2

0 0 0 0 0

20 10 35 30 4

4 (%) 30 15 50 50 6

(b) Structures 35-40% Empennage, 15%

Gear

Wing, 30%

Fuselage, 55%

Propulsion Fixed Equipment

Fig. 4 Significance of fuselage costs (adapted from [34]. a Recurring production costs. b Structures costs)

circumferentially required for the whole barrel are presented in Table 1. This table shows that a huge reduction of part fabrication costs and assembly costs can be achieved just increasing the size of fuselage panels.

4 Weight and Manufacturing Costs Significance in Aeronautical Structures The importance of the structures’ manufacturing costs in aircraft is illustrated in Fig. 4.1 The largest slice in the production costs of an aircraft, Fig. 4a, corresponds to the structures, where in turn, the fuselage corresponds to about 55% of the total structure cost. Consequently, optimization in the structural manufacturing processes has large repercussions in the final aircraft cost and should be a main concern for the aircraft manufacturers. For instance, a huge amount of manual labor tasks are found in riveting structures due to the complexity to automate some of the procedures. It is always a challenge to reduce simultaneously weight and costs, since new materials, adopted for weight reduction, are usually more expensive and require new design and manufacturing processes increasing the final cost. The case of the adoption of carbon fiber reinforced polymers is an example where weight reductions are 1

These values are illustrative and dependent on the aircraft model.

228 Fig. 5 The variation of weight and cost in function of aircraft performance

S. M. O. Tavares

Cost Weight

Performance

expected, although its price and manufacturing prices are more expensive than aluminium solutions [18]. Therefore, a tradeoff between these two variables is always required in order to achieve a good compromise. Figure 5 illustrates this tradeoff that is usually associated to aeronautical components, requiring a compromise solution. As the MEW has impact in the complete life cycle of the aircraft, relations between costs and weight are frequently used to optimize the structures. For instance, Kaufmann et al. [23] proposed an objective function for Direct Opening Cost (DOC) based on the manufacturing and assembly costs—Cman ; the costs of the lifetime fuel burnt per each kg of aircraft—p and the part weight—W: DOC ¼ Cman þ pW

ð1Þ

For a complete evaluation of the costs, it is required to take into account, further to the manufacturing costs and fuel burnt costs, many other costs, as service, maintenance, repair and disposal. However, it is assumed that these ones are predominant for the design of new parts and can be used for structural optimization in order to find the best tradeoff between weight and cost, as in the example presented by Kaufmann et al. [23] or by Hailian and Xiongqing for a composite structure [15].

5 Weight Reduction Several initiatives have been proposed to reduce the structural weight and improve the structural efficiency, mainly focused in two groups: (i) the application of new materials with new aluminium alloys or with carbon fiber reinforced polymers or even GLARE and (ii) with new structural concepts as the integral panels produced by FSW. The reduction of the weight through new design concepts is a function of the material selected which will confine the design alternatives and manufacturing processes. Concerning the aluminium alloys materials, the application of new design and manufacturing concepts has been done in different directions, with new machining operations (as high speed machining) or with new treatments (as the shot peening) or with new joining process to create integral structures without fasteners.


229

An example of weight reduction using new design and manufacturing solutions is the case of adoption of LBW to join the stringers in the fuselage panels. The application of LBW to produce these panels has been adopted intensively by Airbus, mainly in three models of aircraft: A318 (2 panels, corresponding to more than 50 m of welds in each aircraft), A380 (8 panels, corresponding to more than 300 m of welds) and A340 (14 panels, corresponding to more than 400 m of welds in each aircraft), [45]. Rendigs and Knwer in [42] pointed out that more than 1200 welded plates were produced until 2010 for Airbus aircrafts. The weight savings of applying (LBW) to join the stringers to the fuselage skin are significative compared with riveting. Pacchione and Telgkamp show that the replacement of riveting stringer by welded stringers allows savings 0.18 kg per meter of joint, [38]. This saving corresponds to savings of 9 kg in the A318, 54 kg in the A380 and 72 kg in the A340. These values can be considered modest weight savings, although they corresponds to approximately 10% of weight savings in these panels, [42]. The application of FSW can also be used to produce these reinforced panels, with T-joint configuration. Equivalent weight savings compared to the LBW stiffened panels are expected, since the geometry of the joint is similar. In the case of the small business jet Eclipse 500, which was the first aeronautical application where the FSW is the major joining process, approximately 7378 fasteners, corresponding to 65% of the riveted joint were replaced by FSW representing a weight saving of about 22 kg, [16]. The weight savings in this case are modest since the joint geometry was not adapted to this joining process. The joints skinstringers were welded in the overlap positioning. This configuration has a weak fatigue strength due to the interface defects. However, these aircraft have lower fatigue cycles than the commercial aircraft and Eclipse reduced the interface defects with new tools, in order to not compromise the structural integrity, [6]. The replacement of longitudinal joints, as the ones applied to join fuselage panels has higher potential for weight savings since these joints require 2 or 3 rows of rivets for the load distribution. Pacchione and Telgkamp in [38] indicate that a joint with 3 rows of rivets requiring 75 mm of overlap leads to an additional weight up to 0.8 kg per meter of joint (for a plate thickness about 3.8 mm). Assuming that the Airbus A380 has at least 8 longitudinal joints along its 73 m length, about 500 kg could be saved just in these joints.

6 Manufacturing Costs The major target of this section is the cost reduction concerning the manufacturing and assembly of aircraft structures. As mentioned above, these costs are an important part of the total aircraft costs. Due to the complexity of the automation of some tasks to manufacture these structures, as riveting, [49], their manufacturing still requires substantial amount of manual work. Nevertheless, several initiatives to develop new solutions to reduce the fabrication costs of metallic

230

S. M. O. Tavares

structures have been proposed. An example of the development of new manufacturing processes with the aim of cost reduction is disclosed by Pechiney Rhenalu, currently Rio Tinto, in [27]. These solutions comprise: • Machining: – More near-to-the final shape products; – Low residual stress materials; – New machining sequences for reduced distortion. • Forming: – Stretch forming with high formability qualities; – New age forming alloys and dedicated ageing practices; – Stringer alloys adapted to severe joggling. • Heat treating: – Avoid heat treating due to high formability sheets; – Adapt aging treatments to the customer capabilities (shorter equivalent ageing practices). • Assembly: – New solutions to reduce/suppress riveting, with weldable solutions as FSW; or integral machining of heavy gauge plates; – Improved alloys for increased stringer pitches; – New design alternatives (as cast doors). The replacement or reduction of riveting has been a major goal in the aeronautical industry to reduce the manufacturing costs, since it presents several drawbacks from the manufacturing point of view such as being an expensive and time consuming process, still requiring hand intensive work because its automation is not always feasible. The adoption of integral structures to reduce the number of riveted joints has been the solution adopted for carbon fiber reinforced polymer fuselage panels, [41], and for the aluminium fuselages. In the case of aluminium fuselage, additionally to the weight reduction, also the manufacturing costs decrease since more automation is used. The fuselage panels where the stringers are welded with LBW or in the Eclipse 500 where FSW is adopted are examples where the manufacturing costs were substantially reduced. In the case of the Eclipse 500 this adoption allowed to produce a business jet with one quarter of the costs of comparable competitive aircraft [1].

6.1 Main FSW Costs A cost estimation comparing the FSW with the riveted joints based on deterministic and empirical assumptions is analyzed in this chapter. Regarding FSW, the cost evaluation can be based in the methodology for cost evaluation of other

Welded Aeronautical Structures: Cost and Weight Considerations Fig. 6 Major manufacturing costs of a joining process

231

Equipment (% of utilization) Consumables

Materials

Labour

Manufacturing Process Cost

Energy

welding processes. In this case, it is assumed that the major costs are the ones presented in Fig. 6. For FSW process, equipment and labor (with considerable lead times) are the main costs. However, compared with other welding processes, as LBW, consumables contribute also to the manufacturing cost. A first cost estimation tool for FSW, the E-Tool, was developed by Tipaji, [46]. This tool estimates the total manufacturing cost, CT ; based on: CT ¼ CL þ CM þ CP þ CTools

ð2Þ

where CL is the cost related to weld preparation time plus the actual weld time, CM is the machine cost, CP is the power cost and CTools is the tooling cost. This E-Tool did not include the costs related to the fixture of the welding parts, which could be a substantial cost in large parts due to the tolerance management. For the detailed example presented in [46], it was concluded that the machine cost represents 49% and labor cost 46% of the total costs. Di Lorenzo and Fratini presented in [13] a cost model based on machine availability and maintenance, patent royalties, power supply, human resources, fixture and tooling and process set-up and tool change. From this study, it was observed that the multiple T-joints can decrease significantly the welding costs, i.e., the cost is reduced to half if the number of T-joint increases from 2 to 6 joints. Another conclusion is the cost reduction with the weld length increase and/or thickness decrease. The cost per meter of a joint with 1 m is more than a double of a joint with 3 m and the cost per meter of a joint with 3 mm thick is nearly half of a joint with 25 mm thick. Another cost model for manufacturing costs estimation of FSW is proposed in [47]. In this study, a stiffened panel with 7 stringers with 2.05 m length joined by riveting and by FSW were modeled and compared. The authors of this study decomposed the riveting and the FSW process in three main phases. In the case of FSW included: set-up and pre weld tasks, the welding and post welding operations. Figure 7 presents the comparison of the time and the cost required to manufacture the 7 stringer panels. They concluded that with FSW the manufacturing time is reduced by 60% and the cost is reduced by 18.5%.

232

S. M. O. Tavares

Riveting FSW

9

Process Time [Hours]

8

200

7 150

6 5

100

4 3

50

2 1

Assembly Costs [Pounds]

250

10

0

0

Process Time

Cost

Fig. 7 Process time and costs, comparison between riveting and FSW (adapted from [47])

Table 2 FSW equipments and order of magnitude of their prices Company Model

Price (€)

Transformation Technologies ESAB Nova-Tech Engineering (NTE)

300; 000 150; 000 250; 000

RM1 LEGIO FSW 3UL V30K

The welding and setup equipment are the highest fixed costs for the adoption of FSW, and are a function of the application type, panels dimensions and geometries. Due to the specific requirements of aeronautics, the equipment for FSW should be multipurpose in order to be adaptable to different types of structures. For cost evaluation purposes, Table 2 shows three FSW machines for 2D welds, for welds about 1–2 m long, in force or displacement control mode and numerically controlled. The order of magnitude of their prices prices may present some variation due to other characteristics as the possible automatic variation of the tilt angle, refrigeration and other characteristics or circumstances. For the industrialization of the FSW to join large panels, as the reinforced fuselage panels with lengths higher than 5 m and sometimes with curvatures in two axis, larger machines with more than 3 axis are required. An example of a multipurpose FSW equipment is proposed by MTS Corporation, the I-STIR 10, [33], Fig. 8. This FSW has 5 degrees of freedom (3 linear and 2 angular) and can apply a maximum forging load of about 100 kN. Airbus in Bremen has been using this FSW model, [28] with a retractable pin tool. Comparisons between riveting and FSW costs can be done in a generic way considering illustrative costs. An illustrative comparison is done considering the replacement of riveting by FSW. For this purpose longitudinal joints were considered. With riveting, this joint has double rows of rivets and when welded it has a single FSW pass. It is known that the manual riveting does not require heavy initial investments but is heavily labor consuming; on the other hand, the


233

Fig. 8 MTS ISTIR 10 friction stir welding equipment (Courtesy of the I-STIR Technology Business)

1.00 Riveting - Manual

0.90

Riveting - Automatic

Cost per joint

0.80

FSW

0.70 0.60 0.50 0.40 0.30 0.20 0

500,00

1,000,000

1,500,000

Joint length [m]

Fig. 9 Cost analysis comparing FSW with riveting process

automated riveting and the FSW process require significant initial investments but this is amortized along time with the low cost of operation. A base of 100 mm joint length was considered, that with a riveting pitch of 20 mm, [10], corresponds to 10 rivets. Considering the initial investment of $50; 000 USD for the manual riveting equipment, $2; 500; 000 USD for automatic riveting and $2; 000; 000 USD for the FSW equipment, the break-even points was determined for these processes considering a cost of $0.04 USD per rivet with manual application, $0:02 USD for automatic riveting and $0:17 USD per 100 mm of FSW length, [12]. The cost per joint vs. the joint length for these three joining processes is presented in Fig. 9. These results should be considered empirical since the cost data is highly variable in the joint types and length, although this can be used as a basis for comparison. In aeronautical applications the management of tolerances will have higher cost impact since with overlap joints it is reasonably easy to manage the tolerances, but

234

S. M. O. Tavares

in the case of FSW butt joints, the required tolerances might be a manufacturing challenge.

7 Inclusive Value Engineering Assessment Value engineering is a generic methodo logy suitable to evaluate the added value in products, processes and projects where is assigned a fraction of the final cost to each function or impact in order to find where is feasible the reduction of the global cost (manufacturing or others). Mainly, the value engineering is decomposed in four activities, [24]: • • • •

Identify the function of a product or service; Establish a worth for that function; Generate alternatives through the use of creative thinking; Provide the required functions to accomplish the original purpose of the project at the lowest life-cycle cost without sacrificing safety, necessary quality; and/or environmental attributes of the project.

The engineering value in aeronautical sector needs to address all costs during the life-cycle as discussed in previous Chapter. Nevertheless, Murman et al. [36] reported a generic functional relationship for value as: Value ¼

fp ðperformanceÞ fc ðcostÞ ft ðtimeÞ

ð3Þ

This definition is based on the ‘‘Better, Cheaper and Faster’’ drivers which can be transposed into improved performance, lower cost and shorter times, respectively. With the tendency based on ‘‘More Affordable, Cleaner and Quieter’’ aircraft the value function needs to be changed taking into account the life-cycle cost, noise, and emissions. A functional relationship can be: Value ¼

1 flcc ðlife-cycle costsÞ fn ðnoiseÞ fn ðemissionsÞ

ð4Þ

These value functions can be considered in he value engineering analysis considering other aspects that are not only the cost. However, most of the value engineering still established on the cost. Most of these variables can be converted in cost (for instance environmental costs or social costs), although their quantification in complex and unobjective. Based on published data and analyses, the replacement of riveted joints by FSW will add value to the aeronautical structures and subsequently to the aircraft, since it decreases the manufacturing costs, increases the performance due to the decrease of structural weight and decrease the lead time. The quantification of the value engineering added due to the FSW or other manufacturing processes can be estimated and assessed in the early phases of the product development resulting in a more successfully projects. The International Air Transport Association (IATA)


235

already identified the FSW as a technology ready and suitable to new aircraft designs prior to 2020, [21], concerning to short range aircraft. A technology readiness level of 7 (system prototype demonstrated in an operational environment) was pointed out by this association confirming the FSW applicability in aeronautical structures. In addition, a reduction of about 1% of fuel burn is reported, which means a considerable reduction of life-cycle costs. From the value engineering point of view the replacement of riveting by welding processes as FSW requires a considerable investment for equipment and resources although this investment is amortized in a short period and allows several manufacturing improvements and reduction of the global lead times. This adoption is not absent of risk, since the structural integrity needs to be meticulously analysed in the product and process development and the joint quality needs to be completely ensured and inspected during all service life with non destructive techniques. The value added with this replacement is substantial considering the ‘‘Better, Cheaper and Faster’’ motto, nevertheless it also allows a cleaner aircraft structure due to the weight reduction.

8 Conclusions The systems engineering approach for new product and process development in the space industry moved from its traditional focus on performance (‘‘Higher, Faster speed, Farther’’) and risk avoidance, towards the consideration of development lead time, life cycle cost, and risk management with product performance still meeting requirements - or, using buzzwords, ‘‘better, faster to the market, and cheaper’’. This implies that cost issues are subject of growing attention. The application of new manufacturing processes can have an important role in the reduction of costs. This reduction is mainly done through reduction of the manufacturing costs and through life cycle costs. The life cycle costs are highly influenced by the structure weight, therefore the reduction of weight with new design concepts has a high impact in the cost reduction. New joining processes, as the FSW, allows the application of integral structures in airframes, reducing the manufacturing costs and the global weight. Comparing the FSW with the classic joining process in aircraft, riveting, the initial investment might be higher, although the higher automatization, the reduction of the number of tasks to perform the joint and the absence of fasteners, allows a fast amortization of initial investments, even for small joint lengths, with considerable cost and weight savings during the complete component life-cycle.

References 1. Aronson, R.: A new look at aircraft assembly. Manuf. Eng. 132(3), 101–108 (2004) 2. Ashcroft, I., Hughes, D., Shaw, S.: Adhesive bonding of fibre reinforced polymer composite materials. Assembly Autom. 20(2), 150–161 (2000)

236

S. M. O. Tavares

3. Babikian, R., Lukachko, S., Waitz, I.: The historical fuel efficiency characteristics of regional aircraft from technological, operational, and cost perspectives. J. Air Transp. Manag. 8(6), 389–400 (2002) 4. Biel, E., Landy, J., Humrichouser, G., Ervin, B.: Aviation Structural Mechanic (H & S) 3 & 2. Naval Education and Training Professional Development and Technology Center, NAVEDTRA 14018 (1993) 5. Boeing: Boeing To Deliver Major Section For First F-22 Fighter. http://www.boeing.com/ news/releases/1996/news.release.961015.html, Seatle, WA, USA (1996) 6. Christner, B., McCoury, J., Higgins, S.: Development and testing of friction stir welding (FSW) as a joining method for primary aircraft structure. In: 23rd International Council of the Aeronautical Sciences (ICAS 2002). Toronto, Canada (2002) 7. Curran, R., Raghunathan, S., Price, M.: Review of aerospace engineering cost modelling: the genetic causal approach. Prog. Aerosp. Sci. 40(8), 487–534 (2004) 8. da Silva, L., Öchsner, A., Adams, R. (eds.): Handbook of Adhesion Technology. Springer, Heidelberg (2011) 9. de Bruyne, N.: Joint design for primary structures. J. Appl. Polym. Sci. 6(20), 122–129 (1962) 10. de Castro, P., Oliveira, F.: Advanced Design Concepts and Maintenance by Integrated Risk Evaluation for Aerostructures, WP3: Testing, Testing Programme of IDMEC, ADMIRETR-3-43-1.1/IDMEC (2001) 11. De Florio, F.: Airworthiness: An Introduction to Aircraft Certification: A Guide to Understanding JAA, EASA, and FAA Standards. Butterworth-Heinemann, Woburn (2006) 12. Defalco, J.: Friction stir welding vs. fusion welding. Weld. J. 85(3), 42–44 (2006) 13. Di Lorenzo, R., Fratini, L.: A cost model for the friction stir welding process. In: Proceedings of ESAFORM 2005, 8th International. Conference of the European Scientific Association for Metal Forming, pp. 1073–1076. Cluj, Romania (2005) 14. Grote, K.H., Antonsson, E. (eds.): Assembly, disassembly, joining techniques. In: Springer Handbook of Mechanical Engineering, pp. 656–732. Springer, Heidelberg (2009) 15. Hailian, Y., Xiongqing, Y.: Integration of manufacturing cost into structural optimization of composite wings. Chin. J. Aeronaut. 23(6), 670–676 (2010) 16. Harris, D., Norman, A.: Properties of friction stir welded joints: a review of the literature. In: TWI The Welding Institute; presented at EUROSTIR, Progress report presented at the 6th PSG Meeting (2003) 17. Higgins, A.: Adhesive bonding of aircraft structures. Int. J. Adhes. Adhes. 20(5), 367–376 (2000) 18. Hinrichsen, J., Bautista, C.: The challenge of reducing both airframe weight and manufacturing cost. Air Space Eur. 3(3–4), 119–121 (2001) 19. Holden, R., Haworth, P., Kendrick, I., Smith, A.: Automated riveting cell for A320 wing panels with improved throughput and reliability (SA2). In: AeroTech Congress & Exhibition. SAE International, Los Angeles, CA, USA (2007) 20. Hupe, J.: CAEP/8—delivering substantial proposals to help minimize aviation’s effects on the environment. Int. Civil Aviat. Organ. J. 65(3), 6–12 (2010) 21. International Air Transport Association (IATA): The IATA Technology Roadmap Report, 3rd edn (2009) 22. Kaminski-Morrow, D.: German plant produces largest a350 panel. Flightglogal, http://www. flightglobal.com/articles/2011/03/18/354540/picture-german-plant-produces-largest-a350panel.html (2011) 23. Kaufmann, M., Czumanski, T., Zenkert, D.: Manufacturing process adaptation for integrated cost/weight optimisation of aircraft structures. Plast. Rubber Compos. 38(2), 162–166 (2009) 24. KPIT Cummins: Value Engineering, White Paper. http://www.kpitcummins.com/downloads/ value-engineering-whitepaper.pdf (2011) 25. Lee, J., Lukachko, S., Waitz, I., Schafer, A.: Historical and future trends in aircraft performance, cost, and emissions. Annu. Rev. Energy Environ. 26, 167–200 (2001) 26. Lehmann, M.: Clean sky initiative. Aeros. Test. Int. Swocase 2010, 36–38 (2010)


237

27. Lequeu, P., Lassince, P., Warner, T., Raynaud, G.: Engineering for the future: weight saving and cost reduction initiatives. Aircr. Eng. Aeros. Technol. 73(2), 147–159 (2001) 28. Lohwasser, D., Chen, Z.: Friction Stir Welding: From Basics to Applications. Woodhead Publishing in Materials. CRC Press, New York (2010) 29. Martinez-Val, R., Perez, E.: Aeronautics and astronautics: recent progress and future trends. Proceedings of the Institution of Mechanical Engineers, Part C: Journal of Mechanical Engineering Science 223(12), 2767–2820 (2009) 30. Mendez P., Eagar T.: Welding processes for aeronautics. Advanced Materials and Processes 159(5), 39–43 (2001) 31. Mendez P., Eagar T.: New trends in welding in the aeronautic industry. In: 2nd Conference ‘‘New Trends for the Manufacturing in the Aeronautic Industry’’. HEGAN and INASMET Bilbao, Spain (2002) 32. Messler R.: Joining of Materials and Structures: From Pragmatic Process to Enabling Technology. Elsevier Butterworth-Heinemann, Burlington (2004) 33. MTS Systems Corporation: ISTIRTM Friction Stir Welding Solutions. Eden Prairie, MN, USA (2003) 34. Munroe, J., Wilkins, K., Gruber, M.: Integral Airframe Structures (IAS)—Validated Feasibility Study of Integrally Stiffened Metallic Fuselage Panels for Reducing Manufacturing Costs. Technical Report: NASA/CR-2000-209337, Prepared by Boeing for Langley Research Center under Contracts NAS1-20014 and NAS1-20267 (2000) 35. Murman, E. A value perspective on aerospace innovation. In: 24th Congress of International Council of the Aeronautical Sciences (ICAS 2004), Yokohama, Japan (2004) 36. Murman, E., Walton, M., Rebentisch, E.: Challenges in the better, faster, cheaper era of aeronautical design, engineering and manufacturing. Aeronaut. J. 104(1040), 481–489 (2000) 37. National Research Council (US). Committee on Engineering Design Theory: Improving Engineering Design: Designing for Competitive Advantage. National Academies Press, Washington (1991) 38. Pacchione, M., Telgkamp, J.: Challenges of the metallic fuselage. In: 25th International Congress of Aeronautical Science (ICAS 2006). Hamburg, Germany (2006) 39. Petrie, E.: Adhesives for the assembly of aircraft structures and components: decades of performance improvement, with the new applications of the horizon. Met. Finish. 106(2), 26–31 (2008) 40. Petrova, A., Lukina, N.: Adhesive technologies in aircraft construction. Polym. Sci. Ser. D 1(2), 83–90 (2008) 41. Rakow, J., Pettinger, A.: Failure analysis of composites: laminate behavior. Adv. Mater. Process. 167(7), 16–18 (2009) 42. Rendigs, K.H., Knwer, M.: Metal materials in airbus A380. In: 2nd Izmir Global Aerospace and Offset Conference. Gaziemir-Izmir, Turkey (2010) 43. Rooks, B.: Automatic wing box assembly developments. Ind. Robot Int. J. 28(4), 297–302 (2001) 44. Schijve, J.: Fatigue of Structures and Materials, 2nd edn. Springer, Netherlands (2001) 45. Schumacher, J.: Laserstrahlschweien im Flugzeugbau. In: Neueste Entwicklungen der Industriellen Lasertechnik. Wolfsburg, Germany (2005) 46. Tipaji, P.: E-design Tools for Friction Stir Welding: Cost Estimation Tool. Master’s thesis, Missouri University of Science and Technology (2007) 47. van der Laan, A., Curran, R., van Tooren, M., van Ritchie, C.: Integration of friction stir welding into a multi-disciplinary aerospace design framework. Aeronaut. J. 11, 759–766 (2006) 48. van der Laan, A., van Tooren, M.: Incorporating cost analysis in a multi-disciplinary design environment for aircraft movables. J. Eng. Des. 19(2), 131–144 (2008) 49. Webb, P., Eastwood, S., Jayaweera, N., Chen, Y.: Automated aerostructure assembly. Ind. Robot Int. J. 32(5), 383–387 (2005)

Materials Selection for Airframes: Assessment Based on the Specific Fatigue Behavior S. M. O. Tavares, P. P. Camanho and P. M. S. T. de Castro

Abstract Structural weight reduction is a major driver to improve the transportation efficiency particulary in aeronautics. However the lightweight structural designs can be too costly. Indeed, minimum-weight designs are frequently too costly to manufacture, whereas less expensive and easy to fabricate and assemble designs are often much heavier. The most efficient design on the basis of both cost and weight often lies between these two extremes. The current trend in structural materials selection, by the principal commercial aircraft producers, consists of the extensive use of composite materials in the airframe, as in the last generation of twin aisle aircrafts. Composite materials have high specific strength, are less prone to fatigue crack initiation and provide enhanced flexibility for structural optimization compared to the aluminum alloys. On the other hand, aluminum alloys display higher toughness and better damage tolerance in the presence of defects. In order to improve the material selection and the comparison of airframe materials, this chapter presents an weight assessment based on the specific weight for different damage scenarios taking into account their damage tolerant properties.

S. M. O. Tavares (&) P. P. Camanho P. M. S. T. de Castro Faculdade de Engenharia da Universidade do Porto, Rua Dr. Roberto Frias, s/n, Porto, Portugal e-mail: [email protected] P. P. Camanho e-mail: [email protected] P. M. S. T. de Castro e-mail: [email protected]

Adv Struct Mater (2012) 8: 239–261 DOI: 10.1007/8611_2011_59 Springer-Verlag Berlin Heidelberg 2011 Published Online: 8 December 2011

239

240

S. M. O. Tavares et al.

1 Introduction Engineering systems, and in particular the transportation system, are permanently facing a fundamental challenge related to increasing efficiency. Efficiency improvement commonly increases engineering value and reduces general costs including those related to environmental impact. ‘‘greener design’’,‘‘sustainable aviation’’ and ‘‘eco-efficiency’’ are some of the keywords associated to the future trends for the conception of new products in civil aeronautics. From the structural point of view, weight reduction is one comfortable solution to achieve these goals, because it is immediately associated to the optimization of the structure and consequently to lower operational costs, fuel consumption and emissions, instigating more efficient and environmentally friendly transportation systems. New aircraft structural designs and new propulsion systems are in continuous development, with new materials and new processes having an important role in the improvement of air transportation efficiency. This continuous development of new materials and processes has achieved weight and drag enhancements providing significant fuel savings, and lower operational costs. Nonetheless, the adoption by commercial aircraft manufacturers of new concepts using new materials solutions and new advanced manufacturing processes is limited, mainly due to the requirements needed to achieve the highest technology readiness level before technology implementation. In addition, when considering new technology adoption, the civil aeronautical sector tends to be more conservative and risk adverse than other sectors, due to the high safety standards and specifications. From the structural point of the view, the aircraft structures require the lowest weight configurations with enough strength to support all operational loads with high reliability. Since the aircraft airframe is the largest and most important structure, composed mainly by primary parts, it is the one with the most demanding design requirements requiring deep investigations before the adoption of new concepts, although, more efficient structural concepts allow huge operational cost savings and less CO2 emissions considering the large life cycle of these structures.

2 Airframe Structure Truss, monocoque, and the semi-monocoque are the solutions found for the design of this structure. Truss or framework types of construction have wood, steel or aluminum tube, or other cross sectional shapes which may be bolted, welded, bonded, pinned, riveted or machined into a rigid assembly. The vertical and diagonal cross-members are arranged to withstand both tension and compression loads. This type of fuselage has been in use for about 80 years, [1]. It is a very strong structural concept and of relatively light weight (high specific strength). Both the monocoque and semi-monocoque fuselage structures use their skin as an integral structural or load carrying member. Monocoque (single shell) structure is a thin walled tube or shell which may have rings, bulkheads or formers installed

Materials Selection for Airframes

241

within. It can carry loads effectively, particularly when the tubes are of small diameter. The stresses in the monocoque fuselage are transmitted primarily by the strength of the skin. As its diameter increases to form the internal cavity necessary for a fuselage, the weight-to-strength ratio becomes more efficient, and longitudinal stiffeners or stringers are added to it. This progression leads to a semimonocoque fuselage, which depends primarily on bulkheads, frames and formers for vertical strength, and longerons and stringers for longitudinal strength. Semimonocoque is the most popular type of structure used in aircraft nowadays, composed of a long tube shape with a large number of longitudinal reinforcements (stringers) and circumferential reinforcements (frames) which carry all stresses, [2]. Semi-monocoque airframe configuration are applied in all current civil aircraft structures, as in the popular Boeing 737 or in the Airbus A320 aircrafts. The fuselage structure is required to withstand multiple loads which mainly promote tension stress on the upper parts, shear stress on the fuselage laterals and compression stress on the bottom. These stress conditions are developed by the cabin pressurization and by fuselage bending. During the aircraft taxing the top and bottom stresses are inverted, but with lower amplitude due to the absence of pressurization. Fuselage materials need to have high specific strength, high specific Young’s modulus, good fatigue properties and toughness and corrosion resistance. The material fracture toughness is often the higher limitation in the structural design of the components in tension, [3]. The manufacturing of the fuselage in large aircraft is done through the assembly of the reinforced panels, that already have assembled the skin, stringers and frames. The design of these reinforced panels can vary as a function of their position in the aircraft, since the loads are different in each point of the aircraft. This allows the adoption of different materials and cross sections for the reinforced panels, tailoring its strength and giving opportunity for better optimization.

3 Airframe Materials The material selection is one major driver in the aircraft design, considering that it will stipulate the processes (joining, manufacturing and assembly) that will be involved. The traditional fuselages are based on aluminium alloys. For this structure, the most widely used aluminium alloy is the AA2024 that was introduced by Alcoa in 1931 and supplied as an Alclad sheet with good corrosion resistance. However, a new tendency for the use of composite materials started to emerge firstly in military applications and has now reached the major civil aircraft constructors for medium and long-range civil aircraft. The most recent examples of composite materials application at a large scale in airframes are the Airbus A350 XWB that is foreseen to fly in 2013 and the Boeing 787 that is going through its testing program for final certification expected during 2011. For instance, in the Airbus A350 XWB, [4], 52% of airframe materials are composites, to be compared with, for example, the Airbus A320 that has 65.5% of aluminium alloys and 12.5%

242


Fig. 1 Evolution of the materials percentage in the airframe (Source: Manufacturers data). a Airbus A320, 1988. b Boeing 787, 2011. c Airbus A350 XWB, 2013

composite materials, [5]. The titanium percentage has been also growing, since some aluminium parts need to be replaced by titanium parts due to the galvanic corrosion with composite materials and for a higher thermal expansion compatibility. This trend, presented in Fig. 1, implies a significant change in the global characteristics of the fuselage. The decision to select composite materials as the main material for airframe material, instead of aluminium alloys, is not consensual and represents a radical innovation, when compared with other alternatives. Since the aircraft is a complex system, many different variables are not easily quantified and measured, as the material behavior during the life cycle, the impact of the environmental conditions, the real maintenance costs and many other variables. In addition, costs forecasting with new materials is complex, since the variation of materials’ quotations during the complete project extension (more than 20 years in the case of civil aircraft) is unpredictable in the early phases. The adoption of aluminium as main material in airframes had also presented problems; the structural behavior including fatigue aspects was not completely understood, which led to several casualties, for example in the de Havilland Comet or the Aloha Boeing 737 accidents, [6]. The development of advanced aircraft airframes is currently one of the important improvements and challenges of the air transportation sector. In the development of primary structures, the industry started to investigate smart solutions and smart structures with intelligent characteristics, [7]. The use of these solutions requires innovative materials allied to optimized designs. The prospect materials require clear technology readiness level valuation, with well understood mechanical properties and mechanical behavior and with well stabilized manufacturing processes, which means that only mature materials (technological ready) are suitable for primary and secondary structures. Taking into account the mature materials for airframes, presently, three main materials groups are applied: • Aluminium alloys • Fibre reinforced composites • Fibre metal laminates


243

The aluminium alloys, the carbon fiber reinforced polymers and the GLARE (a fiber metal laminated) are, within of these groups of materials, the materials that that have been applied to produce airframes, as a result of their lightweight and high specific strength (static and dynamic). For better decisions on the best materials and processes selection, the largest number of variables during the design phase should be considered, for a wide-ranging and more complete comparison. As regards material selection, Vermeulen and van Tooren, [8], analyzed the performance of different aerospace materials summarizing that the fatigue damage growth and residual strength are the main design drivers for a fuselage. The major fatigue properties of materials suitable for fuselages were examined, estimating the specific weight taking into account material toughness and damage tolerant properties. Commonly, the static strength is applied to assess materials in terms of specific weight; however the static strength is not a primary property for the engineering design of an airframe. Due to the different fatigue behaviors of these materials, direct comparison of their properties is not realistic since their failure is based in different phenomena.

3.1 Aluminium Alloys Aluminium alloys have been the dominant material for airframes since the 1940s, at that time because of their strength-to-weight ratio or their specific strength and reasonable costs, [9]. After the first applications in aeronautics, Alcoa developed a higher strength alloy by increasing the alloys content and developed new heat treatments creating the most popular aircraft aluminium alloy, AA2024 with heat treatment T3 that is still in use in most aeronautical structures. Corrosion in the Al alloys was a concern in these applications, and to improve the corrosion resistance the Alclad concept was created consisting in the application of a small layer of pure aluminium that has good corrosion resistance and protects the aluminium alloy structure. Aluminium alloys still represent a competitive solution as a material for airframe structures. However, they are challenged by composite and hybrid materials with higher specific strengths. In order to face this competition the aluminium alloys producers increased the R&D activities to improve the properties of their alloys. Figure 2 shows the evolution of two different types of aluminium alloys optimized for the application in skin and stiffeners of the airframes. For the skin the main mechanical property is the apparent fracture toughness, [10] (noted in the cited reference as Kapp) and for the stringers the main property is yield strength (TYS in Fig. 2). It is noticeable that until the middle of 1990s, no significant evolution took place. Facing the competition of composites, the producers invested in the R&D of new alloys and improvements of 40% higher fracture toughness and 20% higher tensile strength were achieved.

244

S. M. O. Tavares et al. 180% Relative Variation [%]

Skin, Kapp T-L 160%

Stringer, TYS L

2139 2198

140%

6156 2524

120% 100%

7150

2024

Al-Li 7349 PA766

7075 80% 1950

1970

1990 Year of registration

2010

2030

Fig. 2 Aluminium alloys evolution regarding to the fundamental properties, [11]. (Reprinted with permission)

Aluminium-lithium alloys are examples of currently successful materials (for instance, the alloys AA2195 and AA2198). Lithium (Li) is less dense than the aluminium, 0:53 kg/m3 compared to 2:7 kg=m3 ; reducing the density of the alloy and improving the mechanical properties. Nonetheless, pure lithium is not abundant on earth, its extraction is very expensive and its application in electrical batteries inflates its price, making these alloys costly. Due to the high material cost, the cost benefit for airframes is low. Nevertheless, these alloys are already applied for space applications since in these application the weight costs are higher. Beginning in the later 1990s, these Al-Li alloys have been applied in different aerospace structures, as in the Space Shuttle external tanks, where the Al-Li alloys AA2195-T8 was used in order to reduce the total weight and enable the shuttle to carry out more payload, [12] and are being considered for the next generation of space launchers [13]. The Airbus A350 XWB will use Al-Li for some frames of the fuselage (whereas the skin and stiffeners will be in carbon fibre reinforce polymer), [14]. Scandium-reinforced aluminium is another new type of aluminium alloy, under development, with interest for aeronautical primary and secondary structures due to its mechanical performance. These alloys are stronger than other high strength alloys, having a significant grain refinement, and exhibit a good resistance to corrosion, [15]. Some of these alloys are developed by Alcoa, as the alloy C557 and AA7X11 that NASA is developing for the Hypersonic-X fuel tanks, [16]. In this case the price is also high, due to the scandium scarcity, [17]. Aluminiumberyllium alloys have also interest in airframe structures, due to the lower density of the beryllium (1:85 kg/m3 ) and the high stiffness of these alloys, although its strength still lower that the Al-Li alloys, [18] and the application of beryllium could generate health problems. One of the limitations in the development of new aluminium alloys is the difficulty in the improvement of one property without degradation of another one. New aluminium alloys emerging recently are the aluminium-copper with vanadium patented by Alcoa, [19]. Although the alloy density is similar to the common alloy AA2024, these new alloys have improved combined mechanical properties.


245 Boeing

Airbus

% weight of composites materials

60% A350WXB 50% 787 40% 30% A380

20%

A330

A320 10% 747 0% 1970

A300 767 1975

1980

A310 757 737-300 1985 1990

777

1995

2000

2005

2010

2015

Fig. 3 Carbon fiber reinforced composites evolution

3.2 Fibre Reinforced Composites Fibre reinforced composites materials are composed primarily by fibres, such as glass fibres, carbon fibres or others, impregnated in a matrix that transfers the stresses between the reinforcing fibres. This matrix can be made by different types of polymeric resins such as epoxy or polyurethane. The most attractive composite material, widely used in aeronautical primary and secondary structures, is the (CFRP). Structures designed with this type of material present an attractive alternative to structures using the more conventional materials due to its high specific strength (ratio of strength per density) in compression and tension, good thermal conductivity and dielectric constant, good toughness and wear properties, [20]. Drawbacks are manufacturing and processing costs, low damage tolerance properties, [21], difficulty in inspection and repair, dimensional tolerances and less knowledge about the material behavior in service during its life cycle. Nevertheless, CFRPs are an increasingly popular material for aeronautical applications; their percentage of application in airframes is growing significantly and it evolved to be the main material in some new civil airframes, as shown in Fig. 1. Early applications of these materials were found in military aircraft and in a few parts of civil transport aeronautical structures. Progressively, they have been applied in civil aircraft with continuously growing weight percentage of civil airframes, as exhibited in Fig. 3. The major progress of these materials in civil airframes is found in the new Boeing 787 (Dreamliner), with a fuselage entirely in CFRP materials, [22]. The solution adopted by Boeing represents a dramatic shift from traditional airframe philosophy and creates a considerable number of new challenges. This solution was selected by Boeing in order to achieve the goal of 20% less fuel consumption per passenger compared with the Boeing 767 or the Airbus A320. However according Ostrower in [23], the weight savings, changing the technology from

246


aluminium to CFRP, will be less significant due to multiple modifications in the initial design, although just after the final design maturation it will be possible to estimate the real weight savings. Further advantages have been possible with the replacement of aluminium alloys by CFRPs such as reduction in the number or parts (rivets and other fasteners to join the stiffeners to the skin), an expected reduction in the maintenance costs in 30%, the increase of the passenger comfort and Leaner manufacturing processes, [24]. Airbus adopted the same approach in materials selection as Boeing in order to improve the efficiency of its twin aisle long range aircraft. The A350 XWB, presently in design and development phases, will use intensively composites in primary structures, (52% of the airframe weight), Fig. 1. The fuselage design approach is distinct from the one used in the Boeing 787. Boeing uses a monolithic barrel construction. The barrel is constructed integrally, where the composite preimpregnated tapes are applied to a spinning barrel using multiple robotic tapes laying heads, [25]. In the design of the Airbus A350 XWB fuselage, each barrel is composed by four reinforced panels in CFRP, (4 shells concept: top, bottom and laterals), each one attached to metallic frames, in aluminium-lithium alloy or in titanium. The A350 XWB fuselage is a hybrid solution (skin, doublers, joints and stringers are in CFRP) that will save weight via optimization of the fibre lay-up and thickness of the different skin panels, which can be tailored to the local load requirements of each individual airframe part, [4].

3.3 Fibre Metal Laminates Fibre metal laminates are hybrid materials developed focusing on aeronautical applications, particularly for fuselages. The development of these materials started in 1945, with cooperation between Delft University and Fokker in the Netherlands. The material concept consists on aluminium sheets bonded to sheets of embedded fibres, [26]. After few years of product development, a patent with the concept of fibre metal laminate was submitted in 1982, [27]. Firstly, ARALL (aramid aluminium laminate) composed by aluminium plates and tough aramid fibres was used to improve the specific strength, protecting the fibre/epoxy layers by aluminium in order to allow water permeation. This material presented drawbacks concerning fatigue behavior and was ten times more expensive than the aluminium. After some research it was demonstrated that the weakness of this material was the difficulty of adhesion between aramid fibres and epoxy resin since the fibres can easily split, [28]. Therefore, several other alternatives were studied. Carbon fibres are not suitable material to combine with aluminium plates, due to the galvanic corrosion; however glass fibres, while still being a strong material, do not interact with aluminium. In this way, a new fibre metal laminate with aluminium and high strength glass fibres, denominated Glare, glass-reinforced aluminium laminate, was developed and patented in 1991, [29].


247

Fig. 4 Glare (glass-reinforced aluminium laminate) concept. a Glare concept, example of 1.4 mm thick sheet, [30]. b Glare crack bridging, [31]. (Reprinted with permission from Elsevier)

Figure 4a shows schematically the Glare concept; with the orientation of the glass fibre it is possible to optimize the structure. The University of Delft demonstrated that this type of material presents an exceptional fatigue resistance as a result of the crack bridging effect of the glass fibres in the cracks located in the aluminium plate, as represented in Fig. 4. Some of the relevant fatigue properties are also improved compared with the fibre reinforced composites, mainly the damage tolerance to the low-energy impacts, an important issue in fibre reinforced composites. A recent application of Glare is in the upper fuselage shell of the Airbus A380, offering 15–30% weight savings over aluminium panels with improvement in fatigue properties, [32]. Taking advantage of the mechanical characteristics of this material, the application of Glare panels on this aircraft is done mainly in the upper fuselage part, Fig. 5, since the major stress in these areas are tensile stress.

4 Materials Fatigue Assessment Most of failures in structures occur due to fatigue; some structural engineers indicate that 80% to 90% of all structural failures are due to fatigue, [33]. In aircraft structures most of the parts are subject to cyclic loads and the fatigue behavior of the components is an important issue, considered during the design phase using damage tolerant philosophies. The damage tolerance criterion is used for structures subjected to dynamic loads and assumes that the structure is tolerant to flaws with a maximum size defined during the design phase. These flaws and damages are controlled by periodic inspections during the life cycle of the structure using non-destructive inspections. The maximum flaw size and the growth rates are estimated previously supported by multiple analyses in order to guarantee the structure safety.

248


Fig. 5 Glare panels in the Airbus A380 [34]. (Courtesy of Flightglobal)

Fig. 6 Phases in the fatigue life of a component

Phase I Crack Nucleation External Damage

Phase II

Phase III

Crack Growth

Final Failure

The fatigue life of structural components can be decomposed in three distinct phases. The first phase is characterized by the appearance of the flaw. This flaw can be a result of fatigue of the structure or be generated by an external damage source. A second phase is related to the stable and slow growth of the crack under the fatigue loads. Depending on the stress state in the crack tip and the properties of the material, the structure will tolerate a given number of cycles before the flaw reaches a critical value. The third phase corresponds to the failure of the structure, which in the case of thin shells (as found in fuselages) is associated to a R-curve residual strength behavior before the complete rupture. Figure 6 represents the different phases for a component subjected to high cycle fatigue. Although these modes are interconnected, crack growth can occur without nucleation and the fracture of the structure can occur without previous crack growth. The materials presently applied in fuselages have distinct fatigue crack propagation behaviors, as presented in Fig. 7. With these different behaviors, it is not reasonable to carry out direct comparisons of the materials based on the fatigue performance. In the case of composite materials, different modes of fracture can occur as delamination between layers, internal delamination, matrix cracks and


249

Fig. 7 Fatigue crack propagation modes for metals, fibre reinforced composites and fibre metal laminates, respectively, [35]. (Reprinted with permission of the American Institute of Aeronautics and Astronautics)

fibre fracture. These modes are linked to distinct crack propagation behaviors. Nevertheless, it is feasible to make comparisons with equivalent damage scenarios. Composites fatigue also presents a strong non-isotropic nature, where each layer presents different properties in different directions. These materials require a multi-layer layout to be comparable with isotropic materials, although it creates opportunity for structural strength optimization in the different directions using multiple layers.

4.1 Crack Nucleation Crack nucleation is a phenomenon that occurs in ductile materials, mainly metallic structures, when exposed to cyclic loads above a minimum value, which is a function of their mechanical properties. The crack nucleation in fatigue is associated to the slip of grains. Slip lines appear at early stages of fatigue, and with the continuation of application of cyclic loads these slip lines broaden into bands in which fatigue cracks ultimately form, [36]. Fatigue strength characterization is done with standard specimens that are tested under constant cyclic loading, usually below the yield strength, measuring the number of cycles until the complete failure. The standard representation of fatigue strength is commonly done with SN (or Whler) curves where the stress amplitude or the maximum stress is represented

250


(a)

(b) R=0.5 R=0.06 R=-0.41 R=-1.0

600

600

R=0.43 R=0.00 R=-0.50

400 300 200

R=0.74 R=0.46

400

R=0.06

300

R=-0.37 R=-1.00

200 100

100 0 1.E+03

R=0.85

500 σ max [MPa]

σ max [MPa]

500

1.E+04

1.E+05

1.E+06

0 1.E+03

1.E+07 1.E+08

1.E+04

# Number of Cycles

1.E+05

1.E+06

1.E+07

1.E+08

# Number of Cycles

Fig. 8 SN curves for different load ratios and two stress concentration factors, based on the data presented at [10]. a Stress concentration factor Kt = 1. b Stress concentration factor Kt = 3.4

(a)

(b)

1800

[45/-45]2s [0/45/90/-45/90/45/0]S [0/45/02/-45/02/45/02/-45/0]S

1200

1600 1000

1200

[90]8T [45/-45]2s [0/45/-45/90]2s [0]8T

1000 800 600

σ f [MPa]

σ f [MPa]

1400

800 600 400

400 200

200 0 1000

0 10000

100000 # Number of Cycles

1000000

10000000

1

100

10000

1000000

# Number of Cycles

Fig. 9 SN curves for CFRP in different configurations, [37]. a Unidirectional layers. b Unidirectional layers. (Reprinted with permission from Elsevier)

as a function of the number of cycles. Two different factors are generally considered for these curves, the load ratio (R ¼ rmin =rmax ) and stress concentration factor (Kt ). Figure 8 shows two SN curves, with the fatigue strength in MPa, of the common aluminium alloy AA2024 with heat treatment T4, for different load ratios and for two stress concentration factors (Kt ¼ 1 and Kt ¼ 3:4). It is noticeable that the notch with a stress concentration factor equal to 3.4 reduces significantly the fatigue strength of the material. SN curves for carbon fibre reinforced polymers present a slightly different behavior with lower fatigue strength drop, as shown in Fig. 9, since this material is less sensitive to the crack nucleation phenomena. This is one of the advantages of composites compared with aluminium alloys and with other lightweight metals. Figure 9 presents SN curves for different configurations of fibre orientation, which has a significant influence in the fatigue strength. Combining different layers orientations it is possible to design structures in order to obtain higher resistance to fatigue. Regarding Glare materials, the fatigue nucleation is induced by the aluminium layers reducing the fatigue strength. Figure 10 shows a comparison of the SN


251

Fig. 10 SN curves for Glare and comparison with the aluminium alloy AA2024-T3, [38]. (Reprinted with permission from Elsevier)

Fig. 11 Typical da=dN curves for different groups of materials

Tough Composites

Metals

da/dN

Brittle Materials

B 1

DG

curves between the Glare and the aluminium alloy AA2024-T3. Stresses in Glare are related to the overall applied stresses as well as to the stresses in the aluminium layers.

4.2 Fatigue Crack Growth The stage linked to stable crack propagation, where the crack grows progressively up to a certain limit preceding final rupture, follows crack nucleation. This stage has significant importance in damage tolerant design, since the crack growth rate is related to the tolerance of the material in the presence of damages or cracks. The damage tolerant design of a structure estimates the maximum number of cycles (or the remaining life) that the structure can withstand after detection of a defect. This is done using the material characterization, particulary da=dN versus DK curves, which measure the crack propagation rate (a is the crack length, N is the number of cycles and DK is the amplitude of the stress intensity factor, function of the load amplitude, geometry of the structure and crack size). Figure 11 presents, schematically, the typical shape and behavior of the curves and their comparison for metals and brittle materials as composite materials. The metallic materials have advantage in this point and that is the main reason why these materials are called damage tolerant materials.

252

S. M. O. Tavares et al. Basic Fit for M2EA11AB1 2024-T3 Clad & Bare Sht; L-T; LA & HHA 0.01 0.001

da/dN [in/cycle]

0.0001 1e–05 1e–06

R Data ID Thk Form Orien Env Ref 0.5 M2EA01AB01A1 0.039 0.04" SHT 0.5 M2EA01AB01B2 0.126 0.125" SHT 0 M2EA01AB01C2 0.15 0.15" SHT 0.8 M2EA01AB01D1 0.09 .09" SHT 0.8 M2EA01AB01E1 0.09 .09" SHT 0.5 M2EA11AB01A2 0.09 0.09" SHT 0 M2EA11AB01B1 0.125 0.125" SHT -1 M2EA11AB01I2 0.09 0.09" SHT Fit for R = 0.5 Fit for R = 0 Fit for R = 0.8 Fit for R = -1

L-T L-T L-T L-T L-T L-T L-T L-T

LA LA LA LA LA LA LA LA

181 189 189 361 362 A 73 C 81 189

Curve parameters: C = 8e–09, n = 3.2 p = 0.25, q = 1 Smax/So = 0.3 Yield = 53, alpha = 2 Kc = 74.3041, K1c = 30 Ak = 1, Bk = 1.5 DK1 = 1.22, Cth = 1.21

1e–07 1e–08 1e–09 0.1

1

10

100

Delta K [ksi*sqrt(in)]

Fig. 12 Crack growth curve, da/dN, for AA2024-T4, at different load ratios, [39]. (Reprinted with permission of the NASGRO/Southwest Research Institute)

The stress intensity factor (K) concept and the amplitude of this factor (DK ¼ Kmax Kmin ) are used to characterize crack growth rates (da=dN) mainly in metals, considering linear elastic principles, since that the plasticity in these phenomena is negligible in most of the fatigue life. The Linear Elastic Fracture Mechanics (LEFM) philosophy is applied in most metal structures due to this fact, contributing to a better estimation of the fatigue life. In fibre reinforced composites, the plasticized area in front of the crack tip has influence in the crack growth and needs to be considered for a correct fatigue life estimation. In fibre reinforced composites and in fibre metal laminates the crack growth rate curves are presented as function of dA=dN (where A is the crack or damage area) and energy release rate (G). Nevertheless, the stress intensity factor can be converted in energy release rate considering LEFM assumptions, and for plane stress (as the case of thin shells) can be estimated by: G¼

K2 E

ð1Þ

where E is the Young’s modulus of the material. For metal alloys a database with different curves of da=dN for different conditions is provided by Southwest Research Institute, NASGRO, [39]. Figure 12 shows the da=dN curves for the same alloy considered above, AA2024. These curves correspond to several experimental tests for different load ratios and the continuous lines are best fits to the experimental points using the NASGRO law. Figure 13 shows experimental points of a delaminated CFRP HTA/6376C carbon/epoxy material, one curve obtained using a numerical method, and other curve with a curve fitting of the Paris law. At the horizontal axis the energy release rate is normalized by the fracture toughness of the material in mode I. Due to the


253

Fig. 13 Crack growth curve, da/dN, for CFRP HTA/6376C, numerical, experimental and Paris law curve fitting, [41]. (Reprinted with permission from Elsevier)

anisotropy of composite materials the crack propagation is analyzed in different directions and in different modes of crack tip deformation, tension (mode I), shear in plane (mode II) and shear out plane (mode III). [40] measured experimental da/dN curves of cracked Glare 3-6/5 panels composed by 6 aluminium sheets and 5 glass fibre sheets (½Al=0=90=Al=0=90=Al=90=0=Al). The experimental results were evaluated for three different stress range with maximum stress 80, 100 and 120 and with a load ratio R ¼ 0:05 (R ¼ rmin =rmax ). Figure 14 presents the experimental data points measured. It is noticeable that the da/dN values are nearly constant with the increase of the stress intensity factor. This is one of the main advantages of Glare materials, since they tolerate high damage size with small reduction in the structural integrity, being a good material for damage tolerant design. Considering the different da=dN data and converting K to G with Equation 1, plane stress formulation, and considering a reference of 1 mm thick aluminium sheet in order to convert da=dN to dA=dN; the different crack growth curves can be compared. Additionally, to take into account the material density for elucidation of the weight penalty for equivalent crack growth rates, a specific energy release rate (G=q) was used to compare the different materials. This comparison is presented in Fig. 15. As expected, CFRP presents the lower damage tolerant properties under fatigue loading and Glare presents lower crack propagation rates for higher energy release rates. Nevertheless, the experimental data found in literature for Glare does not include the crack growth rates in the threshold area, possibly due to the difficulty of measuring the crack/damage length in this type of material.

254


da/dN [mm/cycle]

1.E-03

1.E-04

1.E-05

σ =80 MPa σ =100 MPa σ =120 MPa

1.E-06 200

400

600

800

1000

Δ K [MPa.mm0.5 ] Fig. 14 Experimental fatigue crack growth rates for Glare-3, under maximum stresses of 80, 100, and 120 MPa, [40]. (Reprinted with permission from Elsevier) 1.E+00 AA2024-T3 1.E-01 CFRP HTA/6376

dA/dN [mm 2 /cycle]

1.E-02

GLARE 3-6/5 (σ =100 MPa)

1.E-03

1.E-04

1.E-05

1.E-06

1.E-07

1.E-08 0.010

0.100

1.000

10.000

Δ G/ρ [J.m/kg] Fig. 15 Comparison of fatigue crack growth curves for different materials

Despite the lower damage tolerant properties of the CFRP, the selection of these materials is due to the higher specific strength properties (static and fatigue), as noticeable in Figs. 8 and 9. As explained above, CFRP is still a less understood material when compared to aluminium alloys, and it is expected that its


255

Table 1 Damage uncertainties in metals and composites in aeronautical structures Fatigue damage, metal Impact damage, composites Type of uncertainty

Predominantly: fatigue crack

Location of uncertainty

Predominantly: stress concentration locations Limited to the non destructive techniques, although the design accepts it and can be stopped. Well developed, the fatigue life can be predicted with reasonable accuracy Predominantly: should be enough to detect any crack before to reach a critical size

Size of uncertainty

Predictive methods

Inspection interval

Predominantly: external impact damages All structure, mostly on exposed parts to impacts Created instantly, then usually doesn’t grow.

Poor prediction due to lack of appropriate statistical data Uncertain: no deterministic criteria to follow

predominant damage is due to external impacts causing delaminations contrasting with metallic structures, where critical damage is mainly due to fatigue. Therefore, several types of uncertainties have been identified and compared in damaged aeronautical structures. Table 1 lists several of these uncertainties in the structural aeronautical design with aluminium alloys and CFRP, based on the reliability damage tolerant design research done by [42].

5 Structural Weight Assessment The selection of structural materials for airframes are based on different multiple criteria, with a major goal that is the minimum weight for a required structural strength. Since the damage sources are different for different materials, an example of weight assessment of fuselage panels manufactured with Al-alloys and manufactured with CFRPs (the more common fuselage materials) is detailed. This estimation can be based on different assumptions, but for the present case, it is considered that the panel needs to withstand a cabin pressurization of 60 kPa (p) for an airframe with 4 m diameter (d). For the fatigue estimation the cycles of pressurization and depressurization (corresponding to one flight) are considered. Figure 16 shows one aluminum panel with a central crack of length 2a and another panel in composite material with a damage of the size 2c; typically created by an external impact. Both panels are 500 mm width (common frame pitch). For both panels the thickness required for the structure to tolerate these damages without compromising integrity was estimated. Typical damage sizes were considered for both situations, as detailed below.

256 Fig. 16 Typical damages in fuselage aluminium panels and composite panels. a Central crack in analuminium panel. b Damage in a composite panel


(a)

(b) 500 mm

500 mm 2c

2a

5.1 Aluminium Panel A first and more conservative approach consists of assuming that the crack does not propagate, which means that the stress intensity factor should be lower that the fatigue threshold (DKth ). The stress intensity factor for a finite plate with a central crack (KI ) can be estimated by, [43]: rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi pa ffi a 2 a 4 pffiffiffiffiffiffi KI ¼ r pa sec þ 0:06 1 0:025 ð2Þ 2w w w where a is the half crack length and w is the half plate length and r is the remote applied stress perpendicular to the crack direction, that in this case is equivalent to the hoop stress, estimated by: r¼

pd 2t

ð3Þ

For aluminium AA2024-T351, an initial flaw of 2a = 2.54 mm (0.1 in) was considered since this is a reference value for initial flaw size in airframe structures as shown in [44]. If the structure is designed to withstand this flaw without growth (no crack propagation scenario), the minimum thickness of the skin can be determined considering the material fatigue threshold and the Equation 2. For the AA2024-T351 the fatigue threshold is DKth ¼ 42 MPa:mm0:5 , [39] and for the presented conditions, the minimum thickness value should be 5.71 mm. However, this scenario is highly conservative since it does not take advantage of the damage tolerant properties of the metals. Considering damage tolerant design of the panel and taking into account the material toughness of aluminium, the panel weight can be optimized considering the fatigue life of it in the presence of a defect as a crack (2a ¼ 2:54 mm) length as assumed above. This estimation was done using a fatigue crack growth law, the NASGRO law, [39], and the fatigue material properties of AA2024-T351.


257

Crack Length a [mm]

180 160 t=0.8 mm t=0.85 mm t=0.9 mm t=0.95 mm t=1.0 mm t=1.05 mm t=1.1 mm t=1.1 5mm t=1.2 mm

140 120 100 80 60 40 20 0 0

20000

40000

60000

80000

100000

120000

Number of Cycles N

Fig. 17 Fatigue life Al panels with different thickness and an initial crack of 2a ¼ 2:54 mm

The NASGRO law is defined a comprehensive fatigue crack growth law, defined as: DKth p nN 1 da 1f DK ð4Þ ¼ CN DK Kmax q dN 1 Reff 1 KC where CN ; nN ; p and q are empirically derived constants of the material and f is the crack opening function for plasticity-induced crack closure, that in this case is not considered since a positive load ratio is used. The stress intensity factors solutions are calculated for a multiple crack lengths using the Equation (2) which are used for the integration of the NASGRO law, Equation (4) resulting in the fatigue life of the panel (number of cycles vs. crack length). For the conditions presented above fatigue crack growth curves are presented in Fig. 17 taking into account multiple panel thicknesses. From Fig. 17, it is observed that considering just the pressurization and depressurization cycles, a life of to rupture of approximately 100,000 cycles (i:e; flights), is achieved with a minimum thickness of approximately 1:2 mm:

5.2 Fibre Reinforced Composites Damage tolerance is the main design driver for the composite structures used in the aeronautical industry. Accidental damage caused by low-velocity impacts perpendicular to the mid-surface of a composite laminate drastically reduces its low-carrying capability: delamination damage as a result of impact results in over a 50% reduction in the compressive strength at the barely visible impact damage level with respect to an undamaged laminate.

258


This fact has severe consequences in the design and certification of composite structures. Current certification rules impose that the composite structures must sustain the limit load (the maximum load expected in service) after the introduction of credible impact damage and damage up to the detection threshold. This means that the operating strain of the structure is usually less than half the undamaged strength of the same configuration. The simulation of impact and tension/compression after impact remains a formidable task that requires advanced computational models implemented using the non-linear Finite Element method. A practical approach to predict the strength of damaged laminates consists in assuming that the damaged region may be represented using an equivalent hole. For the credible impact energy level used in the certification of composite components, 50 J; experimental data obtained by [45] shows that the damage area is approximately 3000 mm2 : Therefore, a circular notch with a diameter of 62 mm is considered here. Taking into account the design rules that impose that the composite is symmetric and that it has plies in the 0; þ45; 45; and 90 fibre directions, the following lay-up is selected: ½þ45= 45=90=0s: Using the IM7-8552 carbon-epoxy material system, the nominal thickness of the laminate is 1 mm; resulting in a remote stress of 120 MPa: The verification of the integrity of the damaged laminate subjected to a remote stress of 120 MPa is therefore required. The method proposed by [46] for the prediction of the strength of notched composites is used here. It is considered that the non-critical damage mechanisms occurring before the ultimate failure of a notched composite laminate can be lumped into inherent flaws of length a that emanate from the hole. The critical value of the stress intensity factor of a plate with a cracked hole of radius R is: pffiffiffiffiffiffi KIc ¼ f ða; RÞr1 pa ð5Þ with: f ða; RÞ ¼ 0:5 3

a Rþa

1 þ 1:243 1

a Rþa

3 ! ð6Þ

Reference [46] considered that the strength of an unnotched specimen can be predicted by taking into account that the hole radius tends to zero, in which case the critical value of the stress intensity factor reads: pffiffiffiffiffiffi ð7Þ KIC ¼ XTL pa where XTL is the tensile strength of the unnotched laminate. From the previous equations, the notched strength is given as: r1 ¼

XTL f ða; RÞ

ð8Þ


259

Table 2 Weight comparison between Al and CFRP panels Thickness [mm] Density [kg/m3 ]

Weight [kg]

AA2024/T351 - 25 000 cycles AA2024/T351 - 50 000 cycles AA2024/T351 - 100 000 cycles CFRP IM7-8552

1.14 1.362 1.626 0.785

0.82 0.98 1.17 1

2780 2780 2780 1570

The strength of the damaged laminate containing an open-hole is predicted using the previous equations, where the tensile strength of the unnotched laminate, XTL ; and the length of the inherent flaw, a, were obtained in [47] for IM7-8552 as: XTL ¼ 845 MPa and a ¼ 1:3 mm: Therefore, the notched strength of the laminate results in r1 ¼ 272 MPa; a value higher than the applied 120 MPa: This result validates the lay-up selected for the composite laminate.

5.3 Weight Comparison With the several assumptions made and the minimum thicknesses estimated above for both materials, a methodology to estimate the difference in weight between the two structural solutions can be adopted to design structures considering damage tolerant properties. In this case, taking into account the density (q) of the aluminum AA2024-T351, which is approximately 2780 kg/m3 and the density of the CFRP IM7-8552, 1570 kg/m3 ; the weight for panels 500 mm wide and 1000 mm long is calculated, as presented in the Table 2, showing an advantage to the CFRP materials, since higher damages will not affect the structure. The initial estimations of weight savings regarding to selection of new materials in complex products as aircraft structures, could be underestimated or overestimated since the behavior of new materials under life-cycle and service conditions are not yet completely understood, namely from the environmental, scale and aging points of view. In this case, carbon fibre reinforced polymers present the advantage that, unlike in metals, the effect of fatigue crack initiation is not substantial. However, metallic structures present greater damage tolerance, with low crack propagation rates. Fibre metal laminates incorporate the advantage of both materials and show crack growth rates nearly insensitive to the stress intensity factor range. This weight assessment analysis was done using an unstiffened panel as a simplified example representing part of the fuselage skin. Assuming typical possible defects and using the relevant current approaches, the minimum thickness required for the panel in order to guarantee its integrity can be estimated considering the fatigue and external damages effects in the structure integrity giving a more comprehensive analysis of the material selection problem.

260


References 1. Katz, P.: Stresses and strains on aircraft. Aviat. Mon. (1997) 2. Megson, T.: Aircraft Structures for Engineering Students, 3rd edn. Butterworth Heinemann, Oxford (1999) 3. Starke, E. et al.: Application of modern aluminum alloys to aircraft. Prog. Aerosp. Sci. 32 (2–3), 131–172 (1996) 4. Airbus: Taking the Lead: the A350 XWB Xtra Wide Body (Airbus A350 XWB Presentation) (2006) 5. Tober, G., Schiller, D.: NDT in Aerospace—State of Art. In: 15th World Conference on Nondestructive Testing (2000) 6. Wanhill, R.: Milestone case histories in aircraft structural integrity. In: Milne, I., Ritchie, R.O., Karihaloo, B. (eds.) Comprehensive Structural Integrity, Pergamon, Oxford (2003).doi:10.1016/B0-08-043749-4/01002-8 7. Rösner, H., Jockel-Miranda, K.: Airbus airframe—New technologies and management aspects. Materialwissenschaft und Werkstofftechnik 37(9), 768–772 (2006) 8. Vermeulen, B., van Tooren, M.: Design case study for a comparative performance analysis of aerospace materials. Mater. Des. 27(1), 10–20 (2006) 9. Staley, J.T., Hunt, W.H. Jr.: Needs of the aircraft industry for aluminum products. In: 12th Annual NCMS Technical Conference (1998) 10. United States Department of Defense: MIL-HDBK-5H: Military Standardization Handbook: Metallic Materials and Elements for Aerospace Vehicle Structures. Knovel Interactive Editor (2003) 11. Bordesoules, I., Ehrstrom, J., Warner, T., Lequeu, P., Eberl, F.: Trends in developments of aluminum solutions for aerospace applications solutions applications. In: European Workshop on Short Distance Welding Concepts for Airframes—WELAIR. Geesthacht (Hamburg), Germany (2007) 12. Bickley, F., Schwinghamer, R.J.: NASA Experience with the shuttle External Tank. National Manufacturing Week, NASA Marshall Space Flight Center (1999) 13. Moreira, P., de Jesus, A., de Figueiredo, M., Windisch, M., Sinnema, G., de Castro, P.: Fatigue crack growth behaviour of friction stir welded aluminium-lithium alloy 2195 T8X. In: Iberian Conference on Fracture and Structural Integrity—CIFIE 2010. Porto, Portugal (2010) 14. Béral B.: Airbus composites technologies & structures. In: Colloque Composite—Toulouse. Toulouse, France (2007) 15. Ahmad, Z.: The properties and application of Scandium–Reinforced aluminum. J. Minerals Metals Mater. Soc. 55(2), 35–39 (2003) 16. Lee, J., Chen, P.: Aluminum-Scandium alloys: Material. Characterization, Friction Stir Welding, and Compatibility With Hydrogen Peroxide (2004) 17. Røyset, J., Ryum, N.: Scandium in aluminium alloys. Int. Mater. Rev. 50(1), 19–44 (2005) 18. Williams, J., Starke, E.: Progress in structural materials for aerospace systems. Acta Materialia 51(19), 5775–5799 (2003) 19. Lin, J., Sawtell, R., Bray, G., Giummarra, C., Wilson, A., Venema, G.: Aluminum–Copper Alloys Containing Vanadium. United States Patent and Trademark Office: Patent Application US 2010/0183474 A1 (2010) 20. Bhagyashekar, M.S., Rao, R.: Characterization of mechanical behavior of metallic and nonmetallic particulate filled epoxy matrix composites. J. Reinforced Plastics Compos. 29(1), 30–42 (2010) 21. Soutis, C.: Carbon fiber reinforced plastics in aircraft construction. Mater. Sci. Eng. A 412 (1–2), 171–176 (2005) 22. Griffiths, B.: Boeing sets pace for composite usage in large civil aircraft. High Performance Composites pp. 68–71 (2005) 23. Ostrower, J.: Boeing 787-8 Weight examined. Flight Blogger (Flightglobal Blogs) (2009) http:// www.flightglobal.com/blogs/flightblogger/2009/05/analysis-787-8-weight-examined.html


261

24. Baldwin, H.: Boeing 787 dreamliner: Unmatched economics, performance and passenger Appeal. Aviation Week & Space Technology Market Supplement pp. S1–S30 (2005) 25. Boeing. Design Approaches: A Milestone for Composites in Aviation. J. Failure Anal. Prevention 5(4), 5–7 (2005) 26. Vlot, A.: Glare: History of the Development of a New Aircraft Material. Springer, Netherlands (2001) 27. Schijve, J., Vogelesang, L., Marissen, R.: Laminate of Metal Sheet Material and Threads Bonded Thereto, as Well as Processes for the Manufacture Thereof. United States Patent and Trademark Office: US 4,489,123 (1984) 28. Vlot, A., Vogelesang, L.B., De Vries, T.J.: Towards application of fibre metal laminates in large aircraft. Aircraft Eng. Aerospace Technol. 71(6), 558–570 (1999) 29. Vogelesang, L., Roebroeks, G.: Metal-Resin Laminate Reinforced with S2-Glass Fibres. United States Patent and Trademark Office: US 5,039,571 (1992) 30. Alderliesten, R.C.: Fatigue crack propagation and delamination growth in glare. PhD Thesis, Technische Universiteit Delft (2005) 31. Alderliesten, R., Homan, J.: Fatigue and damage tolerance issues of glare in aircraft structures. Int. J. Fatigue 28(10), 1116–1123 (2006) 32. Norris, G., Kingsley-Jones, M., Learmount, D., Phelan, M.: Europe’s giant: A special A380 supplement. Flight Int. Suppl. 20(26), 1–16 (2003) 33. Pun, A.: How to predict fatigue life: Three methods of calculating total life, crack initiation, and crack growth. Design News. December (2001) 34. Norris, G.: Creating a titan, Flightglobal. http://www.flightglobal.com/news/articles/creatinga-titan-199071/ (2005). Accessed 14 June 2005 35. Alderliesten, R., Benedictus, R.: Fiber/Metal composite technology for future primary aircraft structures. J. Aircraft 45(4), 1182–1189 (2008) 36. Lin, T., Ito, Y.: Fatigue crack nucleation in metals. Proc. Natl. Acad. Sci. USA 62(3), 631 (1969) 37. Sihn, S., Tsai, S.: Prediction of fatigue SN curves of composite laminates by Super Mic-Mac. Composites Part A Appl. Sci. Manufact. 36(10), 1381–1388 (2005) 38. Homan, J.: Fatigue initiation in fibre metal laminates. Int. J. Fatigue 28(4), 366–374 (2006) 39. NASGRO: Fracture mechanics and fatigue crack growth analysis software—Reference Manual, Version 4.02. Southwest Research Institute (SwRI) (2002) 40. Shim, D., Alderliesten, R., Spearing, S., Burianek, D.: Fatigue crack growth prediction in GLARE hybrid laminates. Composites Sci. Technol. 63(12), 1759–1767 (2003) 41. Turon, A., Costa, J., Camanho, P., Dávila, C.: Simulation of delamination in composites under high-cycle fatigue. Composites Part A Appl. Sci. Manufact. 38(11), 2270–2282 (2007) 42. Lin, K.Y., Styuart, A., Cheung C.H.E.: Development of reliability-based damage tolerant structural design methodology. In: Advanced Materials in Transport Aircraft Structures (AMTAS), Spring 2007 Meeting (2007) 43. Tada, H., Paris, P., Irwin, G.: The Stress Analysis of Cracks Handbook. ASME Press, New York (2000) 44. Wood, H.: Application of fracture mechanics to aircraft structural safety. Eng. Fract. Mech. 7(3), 557–558 (1975) 45. Lopes, C.: Damage and failure of non-conventional composite laminates PhD thesis, TU Delft, Faculteit Luchtvaart-en Ruimtevaarttechniek (2009) 46. Waddoups, M., Eisenmann, J., Kaminski, B.: Macroscopic fracture mechanics of advanced composite materials. J. Composite Mater. 5(4), 446 (1971) 47. Camanho, P., Maimí, P., Dávila, C.: Prediction of size effects in notched laminates using continuum damage mechanics. Composites Sci. Technol. 67(13), 2715–2727 (2007)

Structural Connections for Lightweight Metallic Structures

Structural Lightweight Aggregate Concrete

Connections in Steel Structures III

Metallic Materials with High Structural Efficiency

Lightweight aggregate concrete

AISC - Design of Structural Connections 4th 1994

Materials, Design and Manufacturing for Lightweight Vehicles

Soil retaining structures: development of models for structural analysis

Thin-Walled Structures with Structural Imperfections

Connections

Connections

Connections

Unusual Structures and Physical Properties in Organo Metallic Chemistry

Multifunctional Metallic Hollow Sphere Structures: Manufacturing, Properties and Application

Lightweight Enterprise Architectures

Connections

Electromagnetic Modeling of Composite Metallic and Dielectric Structures

Metallic Nanomaterials

Electromagnetic modeling of composite metallic and dielectric structures

Electromagnetic Modeling of Composite Metallic and Dielectric Structures

Structural Steelwork for Students

Structural system for architects

Structural Design for Architecture

Hollow Structural Section Connections and Trusses - a Design Guide

Nano-metallic optics for waveguides and photodetectors

Metallic Nanomaterials (Nanomaterials for Life Sciences (VCH))

Half-metallic Alloys

Lightweight Sandwich Construction

Strong Metallic Arm

New Structures for Physics

Engineering Mechanics for Structures

Structural Connections for Lightweight Metallic Structures

Structural Lightweight Aggregate Concrete

Connections in Steel Structures III

Metallic Materials with High Structural Efficiency

Lightweight aggregate concrete

AISC - Design of Structural Connections 4th 1994

Materials, Design and Manufacturing for Lightweight Vehicles

Soil retaining structures: development of models for structural analysis

Thin-Walled Structures with Structural Imperfections

Connections

Connections

Connections

Unusual Structures and Physical Properties in Organo Metallic Chemistry

Multifunctional Metallic Hollow Sphere Structures: Manufacturing, Properties and Application

Lightweight Enterprise Architectures

Connections

Electromagnetic Modeling of Composite Metallic and Dielectric Structures

Metallic Nanomaterials

Electromagnetic modeling of composite metallic and dielectric structures

Electromagnetic Modeling of Composite Metallic and Dielectric Structures

Structural Steelwork for Students

Structural system for architects

Structural Design for Architecture

Hollow Structural Section Connections and Trusses - a Design Guide

Nano-metallic optics for waveguides and photodetectors

Metallic Nanomaterials (Nanomaterials for Life Sciences (VCH))

Half-metallic Alloys

Lightweight Sandwich Construction

Strong Metallic Arm

New Structures for Physics

Engineering Mechanics for Structures

Recommend Documents