<<

Comparison of Mechanical Properties of Joints in AA7050-T7451 Using Different FSW Tools

Daniel André Sequeira de Sousa

Thesis to obtain the Master of Science Degree in Mechanical Engineering

Supervisor: Prof. Maria Luísa Coutinho Gomes de Almeida

Examination Committee

Chairperson: Prof. Rui Manuel dos Santos Oliveira Baptista Members of the Committee: Prof. Virginia Isabel Monteiro Nabais Infante Dr. Jorge Fernandez dos Santos

April 2016

i

Acknowledgement

I would like to express my profound gratitude to:

Prof. Luísa Coutinho, my supervisor at IST, for giving me the opportunity to do my master thesis with her and helping me with all my doubts and problems during my internship at Helmholtz Zentrum

Geesthacht (HZG), Germany.

Dr. Jorge dos Santos, co-supervisor at HZG, Germany, for letting me do my internship at Helmholtz

Zentrum Geesthacht (HZG) and also for the help during my experimental work.

Eng. Luciano Bergmann, technical supervisor at HZG, Germany, for the help and cooperation with my experimental work.

Eduardo Feistauer, PhD candidate at HZG, for the prompt availability and for helping me with the laser microscope and tool wear assessment.

Dr. Stephanie Hanke, Co-worker at HZG, for helping me with the metallography characterization of the SEM images of the fatigue fractures.

Menno Peters, Co-worker at HZG, for the prompt availability with all my technical issues and questions.

My parents and family for all the love and support during my hardest times.

All the people, colleagues at HZG and friends that even though are not mentioned here contributed to my work and well-being.

i

To my family, mother, father, sister and nieces…

ii

Resumo

O processo de Soldadura por Fricção Linear (SFL), “Friction Stir Welding (FSW)”, tornou-se uma boa alternativa à junção de chapas por elementos mecânicos e por soldadura de fusão. As boas características metalúrgicas deste novo processo permitiram novos métodos de ligação de ligas de alumínio como as ligas da série 2000 ou 7000. Este trabalho consiste na análise das propriedades mecânicas de juntas soldadas por fricção linear numa liga de alumínio AA7050-T7451 com 8 mm de espessura. Foi efectuado um estudo prévio de forma a determinar a combinação de parâmetros com as melhores propriedades estáticas e de dureza. Esta combinação de parâmetros foi utilizada para realizar as soldaduras para a análise de distorção e fadiga, incluindo uma avaliação de desgaste da ferramenta. As propriedades mecânicas foram comparadas por duas abordagens diferentes, soldadura por fricção linear em raiz de passe simples, e de passe duplo. Contrariamente ao esperado, as distorções de passe duplo foram maiores que as de passe simples. O procedimento utilizado para o processo de soldadura de passe duplo e a perda de propriedades mecânicas após o primeiro passe são as causas para as maiores distorções do processo de passe duplo. Não houve diferença na resistência à fadiga entre as duas abordagens. O desgaste da ferramenta não foi a principal causa para a falha da mesma. As forças envolvidas durante o processo e fadiga podem ser as razões para a falha da ferramenta.

Palavras Chave

Soldadura por Fricção Linear, AA7050-T7451, Soldadura de passe simples (FSW), Soldadura de passe duplo (DS-FSW), Distorção, Fadiga, Desgaste da Ferramenta

iii

Abstract

Friction Stir Welding, (FSW), has become a good alternative for mechanical joining and fusion welding processes. The positive metallurgical features of this relatively new process allowed new possibilities of joining alloys like the 2000 or the 7000 series. This research work is based on the analysis of the mechanical properties of friction stir welded joints of the aluminium AA7050-T7451 with 8 mm of thickness. A previous parameter study was done in order to determine the best set of parameters with the best results of static and hardness properties. These were used to perform the welds for distortion and fatigue assessments as well as the tool wear assessment. The mechanical properties were compared by two different approaches, single side FSW (FSW) and double side FSW (DS-FSW). Unexpectedly, the distortions of double side FSW were higher than those of single side FSW. The welding procedure chosen for the double side process and the loss of properties after the first pass are the main causes for bigger distortions in the double side FSW. There was no big difference in the fatigue performance of both approaches. Tool wear is not the main cause for tool failure. The forces involved during the process and fatigue could be the reasons for tool failure.

Keywords

Friction Stir Welding (FSW), AA7050-T7451, Single side FSW (FSW), Double side FSW (DS-FSW) , Distortion, Fatigue, Tool Wear

iv

Table of Contents

Acknowledgement...... i

Resumo ...... iii

Abstract ...... iv

1. Introduction ...... 1

1.1. Background ...... 1

1.2. Objectives ...... 2

1.3. Dissertation Structure ...... 3

2. Literature Review ...... 3

2.1. Aluminium and the Aerospace Industry ...... 4

2.1.1. of Aluminium ...... 6

2.2. Friction Stir Welding ...... 7

2.2.1. Fundamentals ...... 7

2.2.2. Parameters ...... 8

2.2.3. Joint Configurations and Properties ...... 11

2.2.4. Defects ...... 12

2.2.5. FSW Variants ...... 13

2.2.6. FSW Applications ...... 14

2.3. Distortion ...... 15

2.4. Fatigue ...... 16

2.4.1. Characterization of Fatigue Process ...... 16

2.4.2. Stress Cycles ...... 17

2.4.3. Fatigue in Aluminum ...... 19

2.4.4. Fatigue in FSW welds ...... 20

2.5. Tool Wear ...... 21

2.6. Summary of Literature Review ...... 21

3. Experimental Procedure ...... 22

v

3.1. FSW System ...... 23

3.2. Welding Parameters ...... 24

3.2.1. AA7050-T7451 – Single Side FSW (FSW) ...... 24

3.2.2. AA7050-T7451 – Double Side FSW (DS-FSW) ...... 25

3.3. Weld Analysis ...... 26

3.3.1. Metallography analysis ...... 26

3.3.2. Distortion Assessment ...... 27

3.3.3. Fatigue Test ...... 28

3.3.4. Fracture Surface Analysis ...... 29

3.3.5. Tool Wear Assessment ...... 30

4. Results and Discussion ...... 31

4.1. AA7050-T7451 - Base Material Characterization ...... 31

4.1.1. Chemical Analysis ...... 31

4.1.2. Metallography Characterization ...... 32

4.1.3. Mechanical Properties ...... 32

4.2. AA7050-T7451 – Single Side FSW (FSW) ...... 33

4.2.1. Summary of Previous Analysis [66] ...... 33

4.2.2. Distortion Assessment ...... 35

4.2.3. Fatigue Assessment ...... 38

4.2.4. Tool Wear Assessment ...... 46

4.3. AA7050-T7451 – Double Side FSW (DS-FSW) ...... 58

4.3.1. Summary of Previous Analysis [66] ...... 58

4.3.2. Distortion Assessment ...... 60

4.3.3. Fatigue Assessment ...... 62

4.4. AA7050-T7451 – Single Side FSW Vs. Double Side FSW ...... 69

4.4.1. Summary of Previous Analysis [66] ...... 69

4.4.2. Distortion Assessment ...... 69

4.4.3. Fatigue Assessment ...... 71

vi

5. Summary and Conclusions ...... 72

6. Future Work ...... 73

7. References ...... 73

vii

1. Introduction

1.1. Background

Aluminium is one of the most common metals on earth and is widely used for engineering structures and components in many industries such as, aeronautical, automotive, rail and ship industries [1]. It is known for its low , high corrosion resistance, excellent machining properties and high thermal and electrical properties [1]. Nevertheless, many aluminium alloys, such as the 7000 and 2000 series are also known for their low weldability capabilities and are often classified as non-weldable materials when using arc electric processes [2]. The use of arc welding processes in these type of alloys results in a low quality joint due to the poor solidification microstructure and porosity in the fusion zone. Normally, to overcome these metallurgical defects the joining of these alloys is usually done by rivets and/or fasteners which leads to an increase in weight and cost production.

The Friction Stir Welding (FSW) process is a relatively new joining technique invented and patented by Thomas Wayne at TWI, Ltd in 1991 [3]. The main metallurgical advantage of this new process is that the welding is carried out in the solid state not reaching the melting point of the base material thus leading to less distortion, lower residual stresses and fewer weld defects in comparison to other welding techniques [4], [5]. With this new welding technique materials that were once considered as non-weldable are now weldable using FSW. For this reason, this new process is very well received in the industry, especially in the aeronautical sector which still uses rivets and fasteners in many structural components [6]. Large amounts of research are being done in this matter in order to study the applicability of this new welding process in industrial applications [7].

The scope of this work is to understand and evaluate the application of the Friction Stir Welding process in the AA7050-T7451 with 8 mm thickness with the intention of producing structural elements for the Embraer’s next generation of aircrafts. Several mechanical tests such as, tensile, distortion, fatigue and tool wear, were done in order to quantify the mechanical performance of the present material. Two different approaches (single and double side welds) were carried out in order to compare and establish which one is the most appropriate for industrial application regarding productivity and joint quality. The tensile test results were used to determine the optimum parameters that were used afterwards for the distortion, X-ray, fatigue and tool wear assessments. The main focus of this work is about these last four topics.

Welding induced distortion is inevitable and this is an actual problem that manufacturers want to avoid at all costs [8]. The distortion problem creates dimensional inaccuracies that lead to an increase in fabrication costs and to the poor quality of welded components [8]. It is a known fact that FSW process

1 leads to lower distortion and lower residual stresses than arc welding processes, however, there is very little information about residual stresses and distortion assessment of thick plates [9]. The present work will address this topic in order to quantify the amount of deformation that FSW process generates.

The study of fatigue has become very important for the industry since approximately 90% of all mechanical service failures are due to this phenomenon [10]. The relevance of fatigue testing is related with the increasing use of high strength materials and the need for higher performance of these materials [10]. The fatigue data of the welded specimens will be presented in form of S-N diagrams and the respective fracture surfaces will be analyzed.

Although friction stir welding process has been used in other materials like and steels, FSW was originally intended for softer materials like aluminium alloys [11]. The tool wear when welding soft materials is particularly small or even negligible and the wear in materials like titanium and steels is expected to be bigger [12], [13]. The study of tool wear is very important for productivity factors and also for the quality of the joint with the purpose of keeping FSW as a competitive process in relation to other welding processes. The FSW process is relatively new and therefore there is very little investigation in tool wear in thick aluminium plates [13]. A preliminary study of tool wear was carried out with the intention of filling this knowledge gap.

In conclusion, Friction Stir Welding offers great possibilities for welding thick materials opening the technology to other applications [14]. Many research works have been using thin plates limited to thicknesses of 5 to 6 mm and there is little information of thicker plates with several mechanical tests [14], [15]. This research work will expand the knowledge of thicker plates and of its mechanical behavior in terms of distortion, fatigue and tool wear.

1.2. Objectives

The FSW research works have been limited to the study of thin plates. The main goals of this research work is:

 Expand the knowledge of welding thicker materials  Evaluate and compare the mechanical properties of two different approaches, single and double side welds  Define which method is the most suitable for an industrial application

2

Some specifications for the present work were defined by Embraer in order to maintain the competitiveness with other welding processes in terms of quality and productivity, e.g. laser welding. These goals are defined as follows:

 Sound welds  Welding speed ≥ 5 mm/s  Weld efficiency ≥ 80 % (mechanical properties of the welds above 80% of the base material properties)  Tool Life Performance ≥ 25 m

1.3. Dissertation Structure

This work is divided into 5 big chapters such as: Introduction, Literature Review, Experimental Procedure, Results and Discussion and Summary and Conclusions. The literature review serves as a theoretical base for the scientific content of the current paper. The literature review (chapter 2) addresses all the subjects exhibited in this document as well as many research works done by other authors. Chapter 2 is divided into 6 sub-chapters such as: Aluminium and its Alloys, Friction Stir Welding, Distortion, Fatigue, Tool Wear and Summary of Literature Review. The following chapter (Experimental Procedure) displays all the standards, procedures, equipment and tools used in this work. Then, the Results and Discussion chapter is divided into 4 sub-chapters: AA7050-T7451 – Base material characterization, AA7050-T7451 – Single Side FSW (FSW), AA7050-T7451 – Double Side FSW (DS-FSW) and AA7050-T7451 – Single Side FSW Vs. Double Side FSW. In the first sub-chapter, a small description of the chemical, metallography and mechanical properties is done in the AA7050-T7451 aluminium alloy with 8 mm of thickness. The following two sub-chapters illustrates a summary of a previous analysis done in this material as well as complementary tests, such as distortion, fatigue and tool wear evaluation. The previous analysis was done by another student and one of the main goals of his study was to define the optimum parameters that were afterwards used for the other tests already mentioned. The last sub-chapter compares the results from both approaches. The Summary and Conclusions are presented in chapter 5.

2. Literature Review

This chapter presents a small literature review about the friction stir welding process and all of the mechanical tests that are addressed in this work.

3

2.1. Aluminium and the Aerospace Industry

Pure aluminium is relatively soft, so to overcome this, the metal can be alloyed with other materials to increase its mechanical properties. To identify them, a numeric designation of 4 digits was created. The first one identifies the group of alloys that contains specific elements. The second one represents the level of impurities. The last two identify the alloy within each group [1], [16]. Table 1 presents the AA numbering system for wrought aluminium alloys used by the Aluminium Association [17].

Table 1 – Numeric designation for wrought aluminium alloys

Alloy Series Alloying element 1xxx 99% minimum Aluminium 2xxx 3xxx 4xxx 5xxx 6xxx Magnesium and Silicon 7xxx 8xxx Other elements

The Aluminium Association also uses a numbering system for casting alloys represented in Table 2. The first digit refers to the major alloying element and the second digits serve to identify a particular composition. The zero after the decimal point identifies the product as a casting [17].

Table 2 – Numeric designation for casting aluminium alloys

Alloy Series Alloying element 1xx.0 99% minimum Aluminium 2xx.0 Copper 3xx.0 Silicon plus copper and/or magnesium 4xx.0 Silicon 5xx.0 Magnesium 6xx.0 Unused series 7xx.0 Zinc 8xx.0 Tin 9xx.0 Other elements

The aluminium alloys can be divided into two categories: non-heat treatable and heat treatable alloys. The series 1xxx, 3xxx, 4xxx and 5xxx are non-heat treatable alloys and they are usually hardened by cold working. The series 2xxx, 6xxx, 7xxx are heat treatable alloys [1], [18], [19]. Aluminium alloys have the heat treatment designations presented after the designation of the alloy, separated by a dash, e.g. AA7050-T7 [17] (see Table 3). For Hxy, the digit x designates heat treatment after the strain hardening and the digit y designates the relative amount of strain hardening.

4

Table 3 – Heat treatment designations

Designation Heat treatment F As fabricated Y=1 – Quarter X=1 – Strain hardened; hard X=2 – Strain hardened and partially Hxy Strain hardened Y=2 – Half hard annealed; X=3 – Strain hardened and stabilized; Y=3 – Full hard Y=4 – Extra hard O Annealed and recrystallized T Solution heat treatment plus age hardening

The solution heat treatment plus age hardening (T) also have a number associated, represented in Table 4.

Table 4 – Numeric designation of solution heat treatment plus age hardening

Designation Heat treatment T1 Naturally aged T2 Annealed T3 Solution heat treated, cold worked and then naturally aged T4 Solution heat treated and then naturally aged T5 Artificially aged T6 Solution heat treated and then artificially aged T7 Solution heat treated and then artificially over-aged T8 Solution heat treated, cold worked and then artificially aged T9 Solution heat treated, artificially aged and then cold worked T10 Artificially aged and then cold worked

The Aluminium Association designation system is now being used in many European countries because of its simplicity but the designation in some countries may differ according to the country’s regulations [20].

The development of new materials combining high strength and lightweight properties is nowadays one of the biggest requirements of the aerospace industry. New concepts and materials need to be developed to face the new challenges of the next century’s mass air transportation [4].

A. Heinz et al. [4] are developing new high strength thick plates, new large fuselage skin sheets with improved damage tolerance properties and lightweight materials which are tailored to advance joining technologies. For example, the floor of a Boeing military transport, C17, was converted from riveted and joined to a machined structure. The conversion to machined structure led to a dramatic reduction in the weight of the aircraft, reduction of the number of joints by a factor of 4 and a significant cost savings for the company [4].

5

The most critical issue in aerospace industry is the weight of the aircraft, which needs to be as low as possible in order to have more fuel efficiency. Polymer matrix composites are beginning to be used in commercial aircrafts, e.g. 50% of Boeing’s 787 weight is made by composite materials [21]. However, aluminium continues to be the primary material of choice for the fuselage, wings and supporting structure of commercial and military aircrafts [22].

The 2000 series (e.g. duraluminium, AA2017) are the most common alloy series used in the aerospace industry due to the high damage tolerance. The 7000 series are also quite used in aircrafts when strength is the primary requirement [22]. Early developments are being taken in the 6000 series to replace the 2000 series for cost and weight reduction [4].

2.1.1. Weldability of Aluminium

Aluminium is high corrosion resistant due to the existence of a natural layer of aluminium oxide also called alumina. This thin layer is more stable than pure aluminium and adheres to the surface and thus protects the material [1]. Due to this oxide, it is very difficult to weld this material as the alumina has a very high melting point of 2050 °C in comparison to the aluminium, 660 °C [23].

Aluminium can be arc welded without cracking related problems; however, if the welding procedure or welding operation is not appropriate, defects may appear in the welded area therefore compromising the product. It may appear problems like porosities, hot cracking, lack of penetration or lack of fusion [24], [25].

Aluminium has high so high amounts of energy are required, thus making the material more vulnerable for the hot cracking which is the most common problem of all. It is possible to avoid cracking situations using the appropriate filler material, energy inputs or adding other elements [25], [26]. The coefficient of is approximately the double of steel leading to undesirable distortions and buckling during welding [26].

The usual welding processes used in aluminium alloys are often TIG (Tungsten Inert Gas) and MIG (Metal Inert Gas) due to their easier applicability and better economy. Laser and electron beam welding are other welding processes used for aluminium alloys [27]. The high temperatures involved in arc welding processes result in a loss of strength due to the melting and quick re-solidification in the heat affected zone [28].

7000 series

The 7000-series are known for high strength properties therefore extensively used in aircraft primary structures [5]. These aluminium alloys are generally classified as non-weldable using arc welding

6 processes because of the poor solidification microstructure and porosity in the fusion zone [2]. Therefore, the joining of these alloys is usually done by rivets and fasteners leading to an increase in weight and costs. These alloys can be divided into two different groups according to the amount of copper percentage [1], [24], [29]:

 Al-Zn-Mg-Cu alloys are the highest strength alloys after heat treatment reaching stresses around 500 MPa. These alloys have low weldability and are more vulnerable for hot cracking.

 Al-Zn-Mg alloys have lower strength than the previous ones (휎푦=200 MPa) but on the other hand, they are easier to weld using the appropriate filler material. Therefore, they are particularly useful in structural applications. This type of alloys resists better to hot cracking.

Aluminium alloys such as 2000 and 7000 series are difficult to join by conventional fusion welding techniques because of the dendrite structure formed in the fusion zone which can seriously compromise the mechanical properties of the joint [14]. Friction Stir Welding is a new solid state joining technique and has been showing better results comparing to the arc welding processes [28], [30].

2.2. Friction Stir Welding

Friction Stir Welding (FSW) process is a relatively new joining technique invented and patented by Thomas Wayne at TWI, Ltd in 1991 [3]. The main metallurgical advantage of this new process is that the welding is carried out in the solid state not reaching the melting point of the base material thus leading to less distortion, lower residual stresses and fewer weld defects in comparison to other welding processes [4], [5].

The FSW process is energy efficient, environmentally friendly and can produce low cost and high quality joints [5]. These characteristics are very well received in the industry, thus becomes the process of choice for manufacturing components for ships, aircrafts, cars and many others. FSW was originally intended for aluminium alloys, but recent studies have managed to successfully weld other materials like steel, magnesium, copper and titanium alloys [11]. However, there are some disadvantages such as high initial investment, low versatility regarding to the weld joints which is easier to achieve with manual processes and also extensive clamping devices are needed in order to fix the material during the process [31].

2.2.1. Fundamentals

Although there are many process variations, all of them use the same principle with small modifications. A rotating tool containing a probe and a shoulder plunges into the workpiece

7 generating enough heat to soften it at a temperature lower than the melting point, then the material flows around the probe and plasticizes as long the tool is passing through the joint line [32]. Figure 1 shows the basic principle of the conventional friction stir welding process.

Vertical Force FSW Tool Rotational Direction Travel Direction

Joint Advancing Line Side

Retreating Stir Side Zone

Figure 1 – Friction Stir Welding Process

Many researchers, cited by Gibson et al. [32], report that FSW process can be considered as an process due to the plastic deformation of the joint line and the flow characteristics of the material. Colligan, cited by Gibson et al. [32], conducted experiments with AA6061 and AA7075 alloys and determined that the material is transported in a chaotic mixing and part of the material is extruded on the retreating side of the probe.

2.2.2. Parameters

There is a considerable number of input variables needed to produce a welded joint such as, rotational speed, welding speed, axial force, plunge depth, tilt angle, shoulder and probe geometries and features [32]. The mechanical properties and microstructure of the weld joint are dependent on these factors and if the optimum parameters are not well selected then the quality of the welds is compromised [33].

Angular and Linear Velocity

Angular and Linear velocity are the most important parameters of all since they are responsible for the mixture of the material. The angular and linear velocities have an important role on the weld quality. An optimum ratio between rotational and linear velocity is needed to guarantee a good material flow and heat input throughout the joint line [33]. This ratio is often denominated as weld pitch and is intimately related with the heat input of the weld [34]. There are several heat input models but J. W.

8

Pew et al. [34], suggested a torque weld based power model where the weld pitch is directly proportional to the heat input evolution [34]. According to P. Vilaça et al. [35], the ratio between rotational speed (Ω) and welding speed (푣) (weld pitch) can be classified as hot to cold condition according to a specific range defined as follows [35]:

Ω > 4, Hot weld condition, 푣

Ω 2 ≤ ≤ 4, Intermediate FSW condition, 푣

Ω < 2, Cold FSW condition 푣

The heat in cold FSW condition is resultant from viscous dissipation (internal friction) due to large plastic flow deformation. The heat in hot FSW condition is caused because of most of the plastic flow deformation is located near the probe and the heat generation is caused by interfacial friction [35]. The welding conditions affect the size of the stir zone and heat affected zones [35], [36].

T. Saeid et al. [37] studied the effect of the welding speed (50, 100, 150, 200 mm/min) on the microstructure and mechanical properties of a duplex and concluded that increasing the welding speed, the heat input would therefore decrease. The lower the heat input, smaller is the grain size of the microstructure leading to an enhancement of the strength and hardness of the material. However, low heat inputs could lead to groove-like defects along the joint line due to the defective flow of the material [37].

The influence of the angular and linear velocities are very important to achieve sound joints, however other parameters like vertical force or tool design also have an important role in the quality of the weld. Thus, test experiments need to be conducted in order to well understand the behavior of the macro and microstructure of the material and mechanical properties using different combination of parameters.

Tilt Angle

The influence of the tilt angle has effects on the flow characteristics of the material. The tool shoulder induces an additional compression force onto the workpiece which results in more heat input. This fact can be beneficial due to the fact that the additional force could remove defects like voids or wormholes [33], [38].

9

Moneer Tolephih et al. [38], investigated the mechanical properties and microstructure using different offsets and tilt angles (0° and 2°) on an aluminium alloy 2024 and commercial pure copper. Results show higher strength using 2° of tilt angle rather than 0°.

Nevertheless, the modification of the tilt angle must be analyzed wisely because changing the tilt angle may not always be beneficial. It depends on many factors, like joint configuration, material and load application. Kittipong Kimapong et al. [39], investigated the effect of the tilt angle on FSW lap joints on AA5083 and SS400 steel and concluded that increasing the tilt angle decreased the shear strength due to the increase of the thickness of brittleness intermetallic compound (IMC) on the interface of the joint [39].

Tool Design

Tool design has an important role in FSW process. The tool is responsible for many factors such as heat generation, mixing, creating pressure to maintain the material in the joint line (prevention or minimization of surface weld flash) and prevent the formation of defects such as wormholes, sheet- thinning, or hooking defects [32].

The tool contains a probe and a shoulder and the selection of these are very important because they are responsible for generating frictional heat on the weld joint and for mixing the material. Figure 2 displays many different combinations of tool geometries [40]. There are various geometry profiles for the shoulder and the probe, they are responsible for the flow of the material and also effect on the microstructure of the material which define the size of the grain and mechanical properties [41].

Figure 2 – Different geometry profiles for the shoulder (right) and probe (left) [40]

M. Ilangovan et al. [42], studied the effect of the tool pin profile on microstructure and tensile properties of FSW of AA6061 and AA5086 aluminium alloy joints. They used three different pin profiles like straight cylindrical, taper cylindrical and threaded cylindrical and concluded that the first one resulted in cross sectional macro level defects in the stir zone and the other pin profiles resulted in defect free joints with similar tensile properties [42].

10

Axial Force

The axial force is another important parameter for producing high quality welds. However, this parameter needs to be properly selected. Axial force influences on surface defects and on the appearance of flash on the weld joint [43].

P. Upadhyay et al. [43] reported the effects of the axial force on welds of AA6056 and verified that relatively low forge forces resulted in surface defects and relatively high forge forces led to a considerable level of flash. The range of acceptable axial forces is small and depends on the material and the tool geometry [43].

2.2.3. Joint Configurations and Properties

There are many joint configurations in FSW; the most common are square butt joints, lap joints and T- Joints. It is possible to obtain other joint configurations depending on the type of machine and with some modifications on the fixation system [32].

Figure 3 – Different joint configurations for FSW: (a) square butt, (b) edge butt, (c) T butt joint, (d) lap joint, (e) multiple lap joint, (f) T lap joint and (g) fillet joint [2]

Very little joint preparation is needed; however, adequate clamping system is required to guarantee the fixation of the workpiece and to prevent misalignments of the joint line. The FSW joint properties are similar to the joints of fusion welding. Both have a region denominated as heat affected zone (HAZ) and the unaffected zone known as the base material (BM). In FSW there are two particular regions denominated as stirred zone (SZ) and thermo-mechanically affected zone (TMAZ). Figure 4 displays the different regions of a FSW microstructure.

11

TMAZ SZ TMAZ

BM HAZ HAZ BM

Figure 4 – Typical cross section of FSW joint

During the process the rotating tool causes severe plastic deformation and also frictional heat that leads to recrystallization of the grains and precipitate dissolution and coarsening of the material in the stirred zone (SZ) [2], [44]. On the outer areas around the stirred zone, there is a transition zone called thermo-mechanically affected zone (TMAZ) that experiences both heat and plastic deformation. Normally for aluminium alloys there is no occurrence of recrystallization in the TMAZ region unlike other materials like titanium, stainless steels and copper [2], [40]. Outside the TMAZ there is a heat affected zone (HAZ) where no plastic deformation occurs and where only heat is experienced [2].

Another characteristic feature of a FSW microstructure is the circular pattern around the center of the stirred zone. The origin and effect of these on the properties of welded components are not very well understood [45]. Krishnan, K. N. [45], came up with an attempt of an explanation of the formation of onion rings. He suggested that the formation of onion rings are related to a geometric effect given by the extrusion nature of the process that creates cylindrical sheets in each rotation [45].

The advancing-retreating phenomenon is another representative feature of a FSW microstructure. Both are intimately related to the way the material flows around the probe. The advancing side is characterized by a sharp boundary between the stirred zone and TMAZ whereas the retreating side has a more complex microstructure with no clear boundary between the stirred zone and the TMAZ [46].

2.2.4. Defects

It is very important to choose the right combination of parameters in order to achieve welds with no defects and with the best mechanical properties. The welding parameters and tool features and geometries have major influence in the quality of the weld, they control the way the material flows around the probe and also the amount of energy that is being given to the workpiece. The most common defects are presented as follows.

Although there are a couple similarities with some aspects of arc welding processes, the defect formation in FSW is very different. FSW defects can be defined as excessive flash, excessive concavity

12

(sheet-thinning), impurities (grease, oil, dirt), voids, wormholes, lack of penetration (LOP) in the root and kissing bond defects which may occur in the root or in the interior of the weld [32].

Sanding and cleaning the joint line before welding is advised in order to avoid impurities within the weld region and also to remove oxide layers to prevent kissing bond defects [32]. Root defects such as lack of penetration (LOP) and kissing bond defects are caused by improper tool design, plunge depth or poor stirring near the root of the base material. The definitions of LOP and kissing bond in the literature are not clear because both defects indicate lack of bonding of the material. However, LOP is referred when the defect is more pronounced and the kissing bond is referred for less severe cases. Lack of penetration in the root has a major influence on mechanical properties such as fatigue life, impact strength, root bend tests and through-thickness load-bearing capacity. Normally, as a predictor of root flaws, bending tests are done to foresee this type of defects [32].

Voids and wormholes have a major impact in mechanical properties of the welded material. This type of defect could appear due to an insufficient flow of the material around the probe, excessive welding speed, improper forging pressure, improper tool design or excessive wear of the tool [32].

2.2.5. FSW Variants

The FSW process has evolved and expanded into numerous variations. Some includes tool modifications and even changes in the methodology of joining technique. Friction Spot Welding (FSpW), Friction Riveting (FricRiveting), Friction Stir Processing (FSP), Friction Surfacing (FS), Bobbin- tool Friction Stir Welding (BTFSW) and Stationary Shoulder Friction Stir Welding (SSFSW) are some examples of new developed variants [32], [47].

The Stationary Shoulder is similar to the standard FSW process using a rotating probe and a non- rotating shoulder. All the frictional heat is done by the probe and the non-rotating shoulder slides above the surface providing forging pressure to the weld joint. The heat input is nearly linear throughout the workpiece thickness, resulting in narrower welds and a reduced HAZ and smooth surface finish [32], [48].

Bobbin Tool FSW has 2 shoulders: one upper shoulder and a lower shoulder which replaces the backing plate of the conventional FSW. One of the main advantages is the reduction of the vertical forces in comparison to the conventional process [32].

The automotive industry frequently uses Resistance Spot Welding (RSW) to weld many components. However, the increase of the use of lighter materials, like aluminium, has become a challenge for the car manufacturers because the RSW process needs high currents and proper pressure in order to

13 provide enough heat to melt the aluminium and create a sound weld. Friction Spot Welding has become a good alternative to RSW since it is possible to weld aluminium joints with better mechanical properties and high energy and cost savings [49].

The joining of composite structural materials has also become a challenge. The most common joining method used is riveted or bolted joints that leads to an increase in costs and weight. Friction Riveting has been tested in GFRP lightweight bridge construction and has been showing promising results with comparable mechanical behavior and strength as bolted joints [50].

2.2.6. FSW Applications

Friction Stir Welding Process has been well received and high demanded by many manufacturers. There is a wide variety of applications for Friction Stir Welding but as the process is still relatively new, lot of research is necessary to analyze and evaluate the applicability and performance of FSW process in new components.

The first applications of the FSW process were done in aluminium panels for deep freezing of fish on fishing boats. FSW is also used to join panels for deck and wall construction for catamarans which improves the weight and cost of the structure since there is no need to use any fasteners or rivets [51].

FSW process has also been used in the aerospace industry and has replaced the use of rivets and fasteners e.g. the Eclipse 500 has weld joints equivalent to 136 m in length and has replaced the use of 7378 conventional fasteners, leading to a stronger and safer structure. Delta rockets and the Airbus A380 are using the same technique [32], [52].

The automotive industry has been implementing FSW technology in many structures by many different car manufacturers. Mazda has been using a process variant, FSSW, to join the aluminium rear door structure of the model RX-8 since 2003 which has improved the protection from side-impact and also has granted a five-star rollover protection [2]. More recently, in 2013 Honda used FSW to weld aluminium and steel to manufacture a lightweight engine cradle which contributed to a 25% weight reduction [32].

FSW is also being used in maritime applications as a replacement of (GMAW). The low heat input in FSW is very appealing for the present industry since it is possible to manufacture large components with little distortions and reduced residual stresses [2], [32].

14

2.3. Distortion

Low distortion and low residual stresses are some of the interesting advantages of the FSW process that makes this technology very well appreciated by many manufacturers [53]. The distortion phenomena creates dimensional inaccuracies that lead to an increase in fabrication costs and to the poor quality of welded components [8].

Distortion and residual stresses are intimately related because the type of stresses in the plates determine the behaviour of the distortion in terms of shape and mode. The residual stresses can reduce the fatigue life of welded components, therefore its study must be acquainted [54].

S. R. Bhide et al. [55], compared the distortions in HSLA 65 plates (711.2 x 228.6 x 6.35 mm) using three different welding processes, submerged arc welding (SAW), gas metal arc welding (GMAW) and friction stir welding (FSW). There are three different modes of distortions namely bowing, angular and buckling distortions (see Figure 5). They have concluded that FSW resulted in a buckling distortion and SAW and GMAW have angular and bowing distortions [55]. The higher compressive stresses in the outer edges of the plates welded by FSW resulted in the buckling shape distortion [55].

Figure 5 – Types of welding distortion [55]

Q.-Y. Shi et al. [9], reported that the distortion of AA6013 using FSW process is in saddle shape which is a combination of convex bending in a longitudinal direction and concave bending in a transverse direction. They have obtained the same distortion shape as S. R. Bhide et al. [9]. Q.-Y. Shi et al. [9], thus concluded that the bigger the length of the panels, the bigger is the deformation in longitudinal direction. They also concluded that the distortions are overall much smaller than arc welding process [9].

The forces of the extensive clamping system of FSW technology have not been taken in too much consideration so far as well as their influence on residual stress and distortion [53]. V. Richter-Trummer et al. [53], studied the influence of FSW clamping force on the final distortion and residual stress field

15 and concluded that high clamping forces contributed to a lower distortion, a more uniform residual stress field and better joint properties. On the other hand, higher clamping forces result in higher residual stresses [53].

N. A. McPherson et al. [56], did a comparison between single and double sided friction stir welded 8 mm thick DH36 steel plate. They concluded that both had longitudinal distortion but with different levels of deformation. The single side FSW plates ranged between 5 to -5 mm and the double side FSW plates ranged between 5 to -25 mm. They have included SAW (Submerged Arc Welding) process to compare with the FSW results and determined that the distortions for the SAW were much higher [56].

Welding induced distortion is inevitable and this is an actual problem that manufacturers want to avoid at all costs. S. Larose et al. [54], reported that the use of needle peening is a suitable post-process for reducing the FSW induced distortions. They have managed to remove 37% of the FSW induced transversal distortion and 82% of the longitudinal distortion in AA2024-T3 material [54]. Dean Deng et al. [8], developed a numerical model that can predict the welding distortion of a skin plate with longitudinal and transversal stiffeners. This could lead to a great production cost savings and better quality of the welded components [8].

2.4. Fatigue

Fatigue failure occurs in a material submitted to dynamic loading conditions that could result in a crack propagation causing failure after a sufficient number of cycles of loading. Fatigue life is mostly dependent on the crack growth rate [57]. The study of fatigue has become very important for the industry since approximately 90% of all mechanical service failures are due to this phenomenon. Fatigue testing has become very significant as a result of the increasing use of high strength materials and the need for higher performance of these materials [10].

2.4.1. Characterization of Fatigue Process

Fatigue failure occurs due to the nucleation and propagation of one or more cracks causing the fracture of the component. Fatigue process can be divided into three different stages [10], [57].

16

A-Nucleation (Stage I) – Initiation of one or more micro cracks. The micro cracks usually start at a notch or other surface discontinuity that leads to stress concentrations. The notch is strongly affected by a slip movement that leads to intrusions and extrusions at the surface of the material, eventually leading to the formation of a crack. B-Stable crack propagation (Stage II) – Creation of beach marks. These are present as a result of stress modifications during fatigue.

Figure 6 – Fatigue fracture Each beach mark contains thousands of microstructural details surface [57] called striations. Each striation represents one cycle of fatigue.

C-Unstable propagation (Stage III) – A final fracture occurs when the remaining material cannot hold the applied load.

Today, the study of fracture mechanics can predict the number of cycles of a certain component thus preventing the crack to grow into a critical size or complete failure [57]. It is possible to predict the service life of a component or do inspection intervals under a certain amount of loading conditions and service environment and proceed to the necessary maintenance or replacement [10].

2.4.2. Stress Cycles

Fatigue failure occurs when a material is being submitted to a dynamic load and when a sufficiently large number of cycles are reached. There are many types of dynamic loads and the most common follow a sinusoidal behavior as shown in Figure 7.

Figure 7 – Fully reversed loading, (흈풎 = ퟎ)

Many fatigue tests are done using a dynamic load similar to Figure 7 where there is no mean stress value and maximum and minimum stresses are equal. Another common stress cycle can be done by applying tensile and compressive stresses. In this case the mean stress value is different than zero (see

17

Figure 8). The effect of mean stress value is very important because has much influence in fatigue strength. Tensile mean stress has a degradative effect and compressive mean stress has a beneficial effect in terms of fatigue strength [10], [58]. There are mathematical models that can predict the effect of mean stress value from fully reversed loading data [10].

Figure 8 – Repeated stress cycle, (흈풎 ≠ ퟎ)

In real type situations the dynamic loads are different and can have an irregular behavior in which the part is subjected to random loads during service, e.g. dynamic load submitted in an aircraft. There are cumulative damage models to predict the amount of damage a component can withstand. Miner’s Rule is an usual method for estimating the fatigue life under variable amplitude loading [59].

Figure 9 – Random loading submitted to a an aircraft [57]

The important elements that define a dynamic loading are the mean stress value and the alternating stress, which are defined as follows.

휎푚푎푥 + 휎푚푖푛 휎 = 2. 1 푚 2

The alternating stress value, 휎푎 is defined by equation 2. 2,

18

휎푚푎푥 − 휎푚푖푛 휎 = 2. 2 푎 2 and the stress range is defined by equation 2. 3.

휎푟 = 휎푚푎푥 − 휎푚푖푛 2. 3

The stress ratio is also an important element regarding fatigue testing and is given by Equation 2. 4. Stress ratio is used as a representation of the mean stress value.

휎푚푖푛 푅 = 2. 4 휎푚푎푥

The stress cycles are divided into two groups, namely the high cycle fatigue and low cycle fatigue. The first implicates a large number of cycles (푁 > 105 cycles) and an elastically applied stress. Although, the applied stress is elastic, plastic deformation can occur in the crack tip. Low cycle fatigue is characterized by the small number of cycles (푁 < 103 cycles) and also by the plastic strain dominant region.

2.4.3. Fatigue in Aluminum

The good mechanical properties of aluminum are very well received in many industries, thus making this material widely used in many structural components. Fatigue failure is one of the main causes responsible for most of the mechanical service failures. For this reason many research work has been done in order to guarantee that aluminum has a good fatigue performance [10].

The substitution of assemblies and built-up structures for integral structures became very important as a result of the constant need for lightweight and improved damage tolerance properties [4]. The aerospace industry is developing new high strength thick plates for machined structures in order to achieve great cost savings and improved service life of the component. Studies have been made for the 7050 aluminum alloy, and results show that the fatigue performance decreases with the increase of its thickness. This poor performance is due to micro-porosities located near the surface. However, Hoogovens Aluminium Rolled Products has developed a process that improves the fatigue performance of thick 7050 plates [4].

The fatigue strengths of welded structures are normally low in comparison to their static strengths. There are 2 main reasons responsible for this fact [59], [60]:

 Stress concentration factors

 Micro cracks close to surface area

19

There are some fatigue life improvement methods that allow to enhance the service life of the welded component. In order to reduce the stress concentration factors, any sharp edges or discontinuities in the weld must be eliminated. Machining, grinding or TIG remelting are some techniques used. Other improvement techniques such as hammer and shot peening are good examples to decrease the micro cracks present in the surface of the material. Although there are many research work done in this area, there are still many uncertainties regarding this subject [59].

Libor Trsko et al. [61], studied the effect of severe shot peening treatment on fatigue life in an AW 7075 aluminum alloy and have managed to achieve good results using 9.6 N of intensity. This parameter increased the fatigue life up to 9%. However, increasing the intensity up to 14.9 N the fatigue life decreased to 21% [61].

Normally, welding is the primary joining method and welded components have lower fatigue properties. Although there are many improvement methods, it is unknown if in fact it can extend the service life of a real component or structure. There are no regulated specifications for the use of this type of improvements, in spite of many research works done in this area with good results [59].

As said before, fatigue failure is the most common cause for mechanical failure and with the increase of new materials in many industries, this phenomenon must be a major design criterion and many tests have to be done prior to its use.

2.4.4. Fatigue in FSW welds

Friction Stir Welding is a new solid state process that transformed many industries due to its promising results in comparison to arc welding processes, as described in chapter 2.2. This new welding technique offers better fatigue performance than those welded by fusion processes [59]. There is very little information about direct comparisons of fatigue performance between the same material and different methods [62].

Caizhi Zhou et al. [62], made a comparative study of fatigue properties of friction stir and MIG pulse welded joints in 5083 Al-Mg alloy and results show that the friction stir welds had better appearance and better fatigue performance than the MIG pulse welds. The fatigue life of FS welds was 18-26 times longer than the MIG pulse welds under a stress ratio of 0.1 [62].

Teng Zhang et al. [63], studied the fatigue properties of friction stir welding joints and riveted lap joints and concluded that the fatigue strength of FSW joint is better than that of the riveted lap joint. The stress concentration factor of the riveted lap joint is the main cause for fatigue failure [63].

20

2.5. Tool Wear

The study of tool wear is very important for productivity factors and also for the quality of the joint with the purpose of keeping FSW as a competitive process in relation to other welding processes. The FSW process is relatively new and therefore there is very little investigation in wear of thick aluminium plates [13].

The geometrical feature of the probe is intimately related to the microstructure and mechanical properties [64]. Zeng, W. M. et al. [64], studied the effect of tool wear on microstructure, mechanical properties and acoustic emission of friction stir welded 6061 aluminium alloy. They used different tools with different wear levels and concluded that wear contributed to the appearance of voids in the microstructure and therefore wear compromises the mechanical properties of the welded material [64].

Although friction stir welding process has been used in other materials like titanium, steels and metal matrix composites, FSW was originally intended for softer materials like aluminium alloys [11], [64]. The wear expected in these softer materials is particularly small or even negligible and the wear in materials like titanium, steels or metal matrix composites is expected to be bigger [12], [13], [32]. Tool wear in FSW of steel is more pronounced, preheating the workpiece is one approach to reduce the tool wear [32]. Prater, T. et al. [65], did a comparative study of the wear resistance of various tool materials in friction stir welding of metal matrix composites (MMC) with 20 and 30 % of reinforcement. They used three different materials for the tool, namely 01 steel, tungsten carbide (WC) and tungsten carbide coated with diamond. They have concluded that the use of harder materials in FSW prolongs the tool life. The tool that evidenced better results was the one with the diamond coating as already expected [65].

The metal matrix composites (MMC) are quite interesting materials for many industries due to their high strength-to-weight ratio and high mechanical properties. However, the joining of these materials by fusion welding create precipitates due to reactions of the molten aluminium and the ceramic reinforcements that could be beneficial or detrimental to the material’s properties. FSW has proven to be a good alternative due to the solid state condition [65].

2.6. Summary of Literature Review

Aluminium is known for its good mechanical properties and is very well appreciated in many industries and it will continue to be used in many industrial applications [1]. The 2000 and 7000 series have low weldability properties and the joining of these materials is usually done by rivets or fasteners that contribute to an increase in weight and fabrication costs [2]. FSW has become a great alternative due

21 to the ability of welding the material with temperatures below the fusion temperature resulting in a well bonded microstructure without any loss of mechanical properties in contrary to fusion welding that have detrimental effects in these type of materials [2], [4].

This work will expand the knowledge about the mechanical performance of thick plates of AA7050- T7451 with 8 mm of thickness and open the possibility of introducing this new welding process in next generation aircrafts. The focus of this work is the mechanical performance in three key topics of the aeronautic industry namely, distortion, fatigue and tool wear assessments.

3. Experimental Procedure

In this section will be presented all the experimental procedures used during each step of the project as well as the equipment and standards. As said before the focus of this work is distortion, fatigue and tool wear assessment. These mechanical tests were done using an optimum parameter that was identified through an extensive parameter study that was done by another student in a previous analysis [66]. The optimum parameter was achieved taking into consideration the microstructure of the joint and the microhardness and tensile test results. These last test results had to answer to the specifications mentioned in chapter 1.2 in order to combine good mechanical properties and good productivity. Figure 12 shows a diagram of the work plan of this work.

An aluminium alloy AA7050-T7451 with 8 mm of thickness was used to perform all the welds. Due to the thickness of the workpiece two different approaches were taken; Single and Double side welds. The first is done in one single pass where the probe welds through the entire thickness of the workpiece and the second, a first pass is welded through half of the thickness and a second pass is welded after rotation of the plate (see Figure 10 and Figure 11). After the first pass, the excessive flash must be withdrawn in order not to damage the backing bar during the second pass.

Rotational Direction Tilt Angle

Travel Direction

Figure 10 – Single Side FSW representation

22

First Pass Second Pass

Rotational Direction Rotational Direction

Travel Direction Travel Direction

Figure 11 – Double side FSW representation (left – first pass; right – second pass)

AA 7050-T7451 8 mm

Single Side FSW Double Side FSW

Optimum

Parameters

Distortion Assessment

X-ray Assessment

Fatigue Assessment

Tool wear Assessment

Figure 12 – Work plan diagram 3.1. FSW System

The welds were carried out in the FSW gantry system with a stiff clamping system presented in Figure 13. The clamping forces applied to the specimens were done in two different directions namely the vertical direction (Fz=15 ± 2 kN) and the horizontal direction (Fy=22 ± 2 kN). The gantry system has four degrees of freedom and two different control modes, position and force control. In addition, it is possible to define an angular position (Tilt Angle) of the tool (see Figure 10). The machine allows for precise monitoring during the welding process, such as forces across x, y and z directions, torque monitoring and rotational and transverse speed monitoring. The shaft of the machine has a cooling system in order to lubricate and also cool down the internal system. The forces involved during the

23 process of Friction Stir Welding are high and in order to achieve high quality welds, a stiff welding machine and clamping system are required to assure a stable process throughout the weld (see Figure 13).

Figure 13 – FSW Portal equipment and clamping system

3.2. Welding Parameters

Before welding, all plates were submitted to fine grinding in order to remove the oxide layer present on the surface of the material. All welds were performed at room temperature and butt joint configuration was chosen for this project. The tilt angle was fixed at 1 ° for all weld experiments. The welding procedure was done according to the Friction Stir Welding Standard ISO 25239 [67].

3.2.1. AA7050-T7451 – Single Side FSW (FSW)

Single side welds were done in AA7050-T7451 plates with 8 mm of thickness. Figure 10 shows a representation of this approach where the probe welds through the entire thickness. This approach has great advantages regarding productivity, however, it offers higher forces. An 8 mm Triflat probe with conical and threaded features and a 20 mm diameter shoulder were used for this work (see Figure 14).

24

Figure 14 – Tool for single side FSW of AA7050-T7451

3.2.2. AA7050-T7451 – Double Side FSW (DS-FSW)

Double side welds were done in AA7050-T7451 plates with 8 mm of thickness. Figure 11 shows a representation of this approach where the probe welds through half of the thickness during each pass. After the first pass the plates are rotated 180 ° to complete the second side of the double side FSW. This approach has lower forces involved however implicates more preparation thus lowering productivity levels. A 6 mm Triflat probe with conical and threaded features and a 15 mm diameter shoulder were chosen for this work (see Figure 15). The welding direction for each pass was done in the same direction (see Figure 16).

Figure 15 – Tool for double side FSW of AA7050-T7451

25

First Pass

Welding Direction

180 ° rotation

Second Pass

Welding Direction

Figure 16 – Representation of double side process

3.3. Weld Analysis

3.3.1. Metallography analysis

Samples for metallography analysis were cut in order to evaluate the microstructure and also for assessment of any defects caused by the FSW process. The samples were embedded in a cold mounting resin (Demotec 30) for post metallographic analysis. The grinding and polishing was carried out in a Struers Tegramin-30 polishing machine and the experimental procedure is described as follows, in Table 5.

Table 5 – Grinding and polishing procedure

Disks Lubricant Silicon carbide #320 water Silicon carbide #800 water Grinding Silicon carbide #1200 water Silicon carbide #2500 water Floc Diamond suspension – 3 µm Polishing Nap Diamond suspension – 1 µm

26

After grinding and polishing, the samples were submitted to etching using Barker’s electrolytic reactant for 120 seconds. Macrographs and micrographs were obtained using an optical microscope LeicaTM DM IRM (see Figure 17).

Figure 17 – Optical microscope LeicaTM DM IRM

3.3.2. Distortion Assessment

In order to evaluate the effect of the welding process in terms of distortion, an optical measurement system called PONTOS was used. The measuring system detects markers randomly disposed on the plates and information regarding coordinates and displacement can be withdrawn from each marker. Two static photos were taken; the first was taken before welding and served as a reference stage and another photo was taken immediately after welding. The welds for distortion analysis were done using the optimum parameters obtained in the previous analysis [66]. Figure 18 shows a representation of the distortion measurement system.

Figure 18 – Representation of distortion measurement using Pontos

27

3.3.3. Fatigue Test

The welds for fatigue tests were done using the optimum parameters obtained in the previous analysis [66]. Before fatigue testing, an X-ray analysis was performed in the plates as-welded to ensure that there were no types of defects in the welded area. The plates were machined in order to reduce the thickness of the plates to 6.5 mm and also to produce the fatigue specimens. Axial fatigue tests were done according to the standards ASTM E466-96, E467-98a, E468-90 and E739-10 [68]–[71]. Testing was conducted at room temperature using servo-hydraulic machines with a horizontal configuration with a maximum load capacity of 125 kN. The stress ratio and frequency used for fatigue testing was, respectively, 0.1 and 10 Hz.

Figure 19 – Fatigue testing machine for AA7050-T7451

The clamping system used for the machines consists of 2 steel plates and three bolts which are tightened with 121 N.m of torque (see Figure 20).

Figure 20 - Clamping system representation for AA7050-T7451

A two parameter Weibull distribution was used to determine the reliability curves and the fatigue life of the specimens. The probability density function (PDF) for a two parameter Weibull distribution is given by [72], [73],

훽−1 푥 훽 훽 푥 −( ) 푓(푥) = ( ) 푒 훼 훼 ≥ 0, 훽 ≥ 0 3. 1 훼 훼

28 where 훼 and 훽 are the characteristic life and shape parameter, respectively. Integrating the PDF equation, cumulative density function (CDF) is obtained (see Equation 3. 2) [72], [73].

푥 훽 −( ) 퐹(푥) = 1 − 푒 훼 3. 2

This equation can be used to fit the fatigue data by means of a Weibull plot (see Equation 3. 3) in order to determine the characteristic life, 훼 and the shape parameter, 훽. The Bernard’s Median rank empirical estimator (see Equation 3. 4) is used to determine 퐹(푥) [72], [73].

ln(− ln(1 − 퐹(푥))) = 훽 ln(푥) − 훽ln (훼) 3. 3

푖 − 0.3 푀푅 = 3. 4 푛 + 0.4 where 푖 is the failure serial number and 푛 is the number of specimens. After computing the Weibull parameters, 훼 and 훽, it is possible to determine the mean time to failure or Weibull mean life for each stress level defined by equation 3. 5.

1 푀푇푇퐹 = 훼Γ (1 + ) 3. 5 훽 where Γ( ) is the gamma function [72], [73]. A reliability study was also done in the current work. A reliability level of 90% was drawn in the fatigue S-N diagrams for both approaches. The experimental data using Weibull distribution is equivalent to 36.8 % of reliability. This study was done with the quantile function of the Weibull distribution given by equation 3. 6 [72], [73].

1 훽 푁푃 = 훼 ((−ln (푅푥)) ) 3. 6

3.3.4. Fracture Surface Analysis

Single and double side welds were submitted to fracture surface analysis in order to analyse the fracture surface of some fatigue specimens. The analysis was conducted using a scanning electron microscope (SEM) QuantaTM 650 FEG.

29

Figure 21 – Scanning electron microscope (SEM) QuantaTM 650 FEG

3.3.5. Tool Wear Assessment

The machine used for tool wear analysis was a KeyenceTM VK-9700 Laser scanning microscope (see Figure 22). The shoulder and probe were analyzed separately due to limitations of the existing equipment. The scans were done in order to assess the wear caused by the process. The scanning of the probe and shoulder was done every 2.4 m of weld length, i.e. after two set of plates, until tool failure. Only one of the sides of the probe was analyzed for wear profiles. The tool wear assessment was only performed in the 8 mm probe and 20 mm shoulder (see Figure 23). Samples were cut in order to analyze the presence of defects in the microstructure.

Figure 22 – Laser scanning microscope KeyenceTM VK-9700

30

Figure 23 – 20 mm shoulder and 8 mm probe for tool wear assessment

4. Results and Discussion

In order to enable a better understanding of the experimental results, the analysis will be divided into four big chapters. Firstly, a characterization of the AA7050-T7451 base material will be presented, secondly the results obtained for single side welds will be discussed, thirdly the results obtained for double side welds will be presented and ultimately a comparison between both approaches will be discussed.

4.1. AA7050-T7451 - Base Material Characterization

The main purpose for the base material characterization is to evaluate the microstructure and mechanical properties of the material to ensure that the proper material is being used.

4.1.1. Chemical Analysis

Chemical composition analysis was carried out and the results are shown in Table 6 as well as the upper and lower limits defined in the literature [74]. The experimental results were according to the literature results.

Table 6 – Chemical composition of AA7050-T7451

Literature[74], Element Chemical composition, % % Al 87.3 – 90.3 89.3 Zn 5.7 – 6.7 5.99 Cu 2.0 – 2.6 2.37 Mg 1.9 – 2.6 2.05 Zr 0.08 – 0.15 0.11 Fe ≤0.15 0.07 Si ≤0.12 0.04 Ti ≤0.06 0.04 Mn ≤0.10 0.008

31

4.1.2. Metallography Characterization

The microstructure of the AA7050-T7451 hot rolled plate is presented in Figure 24 with the different grain orientations namely rolling direction (RD), long transverse direction (LTD) and short transverse direction (STD). STD LTD

RD

Figure 24 – Microstructure of AA7050-T7451 (RD – Rolling Direction, LTD – Long Transverse Direction and STD – Short Transverse Direction)

The microstructure of the material shows a significant elongation of the grains along the rolling direction as well as an anisotropic structure along the different directions. The anisotropy of the material and grain boundaries has a great influence in the mechanical properties.

4.1.3. Mechanical Properties

Tensile tests were performed in the different directions of the base material (RD) and LTD) as shown in Table 7. Theoretical results are also presented in the same table [74]. The tensile tests experiments were done according to the standard DIN 50125:2009 and ASTM E8-09 [75], [76]. Testing was conducted at room temperature using a servo-hydraulic machine at a constant crosshead speed of 1 mm/min. Strain was measured using a mechanical MTS axial extensometer with an initial length (Lo) of 75 mm.

Table 7 – Mechanical properties of AA7050-T7451 8 mm thick

Mechanical Properties Property Literature[74] RD Direction LTD Direction Yield Tensile Strength, MPa 469 451.0 ± 0.8 449.8 ± 0.5 Ultimate Tensile Strength, MPa 524 507.1 ± 0.8 513.0 ± 0.3 Elongation at break, % 11 16.4 ± 0.5 15.9 ± 0.7

32

In addition, microhardness tests for all directions (RD, LTD and STD) as well as the literature hardness value are presented in Table 8 [74]. The microhardness tests were done according to the standard ASTM E92-82 and DIN ISO 9015-1:2001 [77], [78]. Testing was conducted in a conventional Zwick/Roell ZHV microhardness machine at room temperature. Hardness indentations were done with a spacing of 0.5 mm using 2 kg of load (HV 0.2) and a dwelling time of 10 seconds.

Table 8 – Microhardness values of AA7050-T7451 8 mm thick

Microhardness Literature, HV [74] Hardness, HV

Rolling Direction, RD 157.3 ± 3.0

Long Transverse Direction, LTD 162 158.5 ± 3.0

Short transverse Direction, STD 159.2 ± 2.0 4.2. AA7050-T7451 – Single Side FSW (FSW)

In this section, a summary of the previous analysis with the optimum parameter and the results of the distortion, fatigue and tool wear assessments for single side FSW will be presented.

4.2.1. Summary of Previous Analysis [66]

To define the optimum parameter for single side FSW, several weld experiments were done with variations in the welding parameters (rotational and welding speeds and vertical force) in order to achieve a good surface appearance and also to correspond to the specifications mentioned in chapter 1.2 namely welding speeds higher than 5 mm/s and weld efficiencies above 80 %. In general, all welds had good visual appearance with no significant defects on the surface of the workpiece. Bending tests were done in order to evaluate the ductility of the specimen and also identify defects like lack of penetration in the weld root. The ones that evidenced lack of penetration were immediately discarded.

The optimum parameter was defined after metallography, microhardness and tensile evaluation. The specimen with good microstructure (no defects), hardness and static strength above 80 % of the base material properties was selected as the optimum parameter. The set of parameters that enabled a good microstructure and good mechanical performance were 700 rpm, 5 mm/s and 25 kN (FSW-PA- LB-1851). Figure 25 shows the microstructure of the optimum parameter with no defects in the microstructure. Overall, welds with higher rotational and welding speeds than those used for weld 1851 showed volumetric defects in the upper area of the stir zone. This could be originated due to insufficient heat to soften the material thus making the flow of the material inadequate to fill the gaps left throughout the joint.

33

TMAZ TMAZ SZ

BM BM HAZ HAZ

Figure 25 – Microstructure of weld FSW-PA-LB-1851 (700 rpm, 5 mm/s, 25 kN)

In Figure 25, it is possible to identify the typical microstructural evolution caused by the friction stir welding process that is the base material (BM), heat affected zone (HAZ), thermo-mechanically

34 affected zone (TMAZ) and the stirred zone (SZ). It is possible to observe the difference of the size and shapes of the grains between the different regions as well as the circular features in the stirred zone so called “onion rings”. The frictional heat and plastic deformation led to the recrystallization of the grains and to precipitate dissolution and coarsening in the stirred zone. In TMAZ, it is evident the plastic deformation of the grains due to the heat generation and also due to the chaotic mixing done by the probe. In relation to the HAZ, the size and shape of the grains are slightly modified. The size is slightly increased and also the characteristic elongated grain feature caused by the manufacture process is lost. This region (HAZ) is the most critical one since there is a loss of strength (softening) caused by the frictional heat, therefore, the materials tend to fail in most cases in the HAZ.

The specimens of the tensile tests that had a good microstructure tended to fail in the HAZ evidencing a ductile fracture type and the ones that had volumetric defects in the microstructure failed through the stir zone and TMAZ in a brittle manner. The tensile specimens tended to fail in the retreating side which could be related to the higher temperatures registered in this area that led to a loss of mechanical properties (retreating side ≈ 299.1 ± 2.5 °C and advancing side ≈ 271.8 ± 7.7 °C). Table 9 illustrates all the mechanical properties of the optimum parameter.

Table 9 – Summary of mechanical test results for single side FSW of AA7050-T7451 8 mm thick [66]

Designation Welding Min. Yield Rotational Vertical UTS, Heat (FSW-PA-LB- speed, Hardness, Strength, Elongation, % speed, rpm force, kN MPa Input, XXXX) mm/s HV MPa kJ/mm 362.7 ± 475.0 ± 127.1 ± 2.5 7.1 ± 0.4 1851 700 5 25 2.5 2.4 0.60 (80.2%) (43.3%) (80.4%) (93.7%)

The set of parameters identified in Table 9 resulted in particularly good results answering to the main specifications mentioned in chapter 1.2 granting both good productivity and good mechanical performance. This set of parameters was defined by another student after metallography, hardness and tensile evaluation.

In order to complement this study, the previous parameters were used to perform the welds for distortion and fatigue assessments which are crucial mechanical tests for the aeronautical industry. A tool wear assessment was also included in order to determine the amount of weld length that was manageable using these parameters without tool failure.

4.2.2. Distortion Assessment

Distortion is a major design criterion because it could lead to dimensional errors and to an increase in fabrication costs. Therefore, the study of this problem is of major importance in order to quantify,

35 understand and possibly reduce the distortion problem. Distortion and residual stresses are intimately related to each other but regrettably the study of residual stresses was not performed due to time limitations. The deformations obtained for single side FSW were particularly small.

The experimental procedure for the distortion assessment is discussed in chapter 3.3.2. Figure 26 and Figure 27 show the marker distribution along the plates and the reference plate obtained from Pontos system, respectively.

Figure 26 – Marker distribution of AA7050-T7451 8 mm thick

Although there is no information about the residual stresses that the plates may have due to the manufacture process, these were used as the reference stage and the displacements were computed in relation to them. The reference plate (un-welded) shows some small distortions which could be due to measuring errors or due to the residual stresses from the manufacture process that resulted in the deformations evidenced in Figure 27. Even though the Pontos system evidences slight deformations throughout the workpiece, these are negligible to the naked eye. Figure 28 illustrates the relative displacement caused by the process.

36

Figure 27 – Reference plate (un-welded) obtained from Pontos system

WD

Figure 28 – Distortion assessment for single side FSW of AA7050-T7451 8 mm thick (WD – Welding Direction)

37

The plates after welding have their maximum deformations in the outer edges. Table 10 represents the maximum and minimum displacement values obtained from the Pontos optical system. The distortion measured in both x and y directions are small and can be neglected. The maximum displacement measured was around 1 mm on the outer edges of the plate in relation to the reference plate.

Table 10 – Maximum and minimum relative displacement values for single side FSW of AA7050-T7451 8 mm thick

Single side FSW of AA7050-T7451 8 mm Max. Min. dx [mm] 0.30 -0.01 dy [mm] 0.33 -0.22 dz [mm] 1.03 -0.56

There is no design criteria defined by Embraer, however taking into consideration the dimensions of the plates (1200 mm X 400 mm), the distortions can be considered as acceptable. As discussed in chapter 2.3, most researchers use thin plates to assess the distortion behaviour and their shape is very similar to the shape of the distortion illustrated in Figure 28. Both have a combination of convex bending in longitudinal direction and concave bending in transverse direction which is also called as saddle shape distortion or buckling distortion. Regrettably, there is no information regarding the residual stresses from the manufacture process, clamping system or after welding. The internal stresses give information about the way the plates are going to deform and due to the buckling distortion behaviour (saddle shape distortion) they have most likely compressive stresses in the outer boundaries of the plate. There is very little information of distortion assessment in similar plates using other welding processes but is expected that the distortion of aluminium using FSW is much lower than the distortion using other welding processes.

4.2.3. Fatigue Assessment

Aproximately 90% of all mechanical service failures are due to fatigue therefore the study of fatigue of welded components is of great importance for the industry. Fatigue evaluation was performed using the set of parameters that had the best mechanical properties (700 rpm, 5 mm/s and 25 kN) [66]. Particularly good results were obtained having almost the same fatigue strength of the base material. A fracture surface analysis was done in order to conclude why some specimens with the same load conditions were breaking in the base material and others in the welded area.

38

Before the fatigue test, a non-destructive test was done to ensure that there were no defects throughout the as-welded plates. Figure 29 illustrates an X-ray image of one single side FSW plate.

Figure 29 – X-ray image of one single side FSW plate of AA7050-T7451 8 mm thick (as-welded)

According to Figure 29, the welded plates are free of defects throughout the entire joint and therefore can be submitted to fatigue testing. The defects could lead to stress concentrations which would decrease the fatigue resistance. Table 11 displays the fatigue data of single side FSW of AA7050-T7451 8 mm thick with different loads with the respective cycles. A Weibull distribution was chosen to analyse the data and obtain S-N diagrams for different reliabilities.

39

Table 11 – Fatigue data for single side FSW of AA7050-T7451 6.5 mm thick

Weibull Max. Min. Mean Mean Cycles σmáx, Amp., (36.8% Failure Specimen Load, Load, Load, Cycles MPa kN Location kN kN kN reliability)

FSW-SN1 60.709 6.071 33.390 27.319 68 880 BM FSW-SN2 310.3 60.718 6.072 33.395 27.323 25 407 44 835 Weld FSW-SN3 60.678 6.068 33.373 27.305 24 937 Weld FSW-SN4 54.000 5.400 29.700 24.300 138 693 Weld FSW-SN5 54.083 5.408 29.746 24.337 146 916 BM 275.8 105 657 FSW-SN6 54.025 5.403 29.714 24.311 55 463 Weld FSW-SN7 54.025 5.403 29.714 24.311 51 244 Weld FSW-SN8 50.598 5.060 27.829 22.769 271 334 BM FSW-SN9 50.632 5.063 27.848 22.784 450 169 BM 258.6 374 388 FSW-SN10 50.606 5.061 27.833 22.773 65 124 Weld FSW-SN11 50.589 5.059 27.824 22.765 429 083 BM FSW-SN12 47.306 4.731 26.018 21.288 2 687 370 Weld FSW-SN13 241.3 47.168 4.717 25.942 21.226 1 582 106 2 808 515 Weld FSW-SN14 47.137 4.714 25.925 21.212 140 337 Weld FSW-SN15 40.506 4.051 22.278 18.227 4 000 000 Run-Out FSW-SN16 40.486 4.049 22.267 18.219 4 000 000 Run-Out 206.9 - FSW-SN17 40.437 4.044 22.240 18.197 4 000 000 Run-Out FSW-SN18 40.443 4.044 22.244 18.199 4 000 000 Run-Out Stress ratio – 0.1, Frequency – 10 Hz, Run-Out criteria – 4 000 000 cycles

According to the literature the fatigue strength of the welded specimens is close to the base material’s [79]. The fatigue strength of the welded specimens is around 206.9 MPa. Figure 30 illustrates the S-N diagram with different reliability curves.

40

S-N Diagram for Single Side FSW of AA7050-T7451 400

350 y = 546,91x-0,056

[MPa] 300 R² = 0,8999 36,8%

max 250 σ Run-Outt 200 (4 M) 90% 150 Axial force fatigue tests in 6.5 mm AA7050-T7451, Single Side FSW σ =363 MPa [Yield strength of welded specimen] 100 y Max. Stress, Stress, Max. Tests performed at room temperature, R=0.1, f=10 Hz 50 1000 10000 100000 1000000 10000000 Fatigue Life, N, cycles

Figure 30 – S-N Diagram for single side FSW of AA7050-T7451 6.5 mm thick

The specimens highlighted in blue and green in Table 11 broke in the base material and in the weld. SEM observations were done with the intention of knowing the reason why the specimens were breaking in different locations even though the loads were the same for each pair of specimens. Figure 31 shows the specimen FSW-SN1 that broke in the base material with 310.3 MPa of load after 68 880 cycles.

BM

A

B

41

FSW-SN1, 68 880 cycles

A A1 A1

B

Figure 31 – SEM observations of fatigue fracture surface (FSW-SN1; 68 880 cycles; Failure location - BM)

It is clear where the crack begins (see Figure 31 – Zone A). Using a different detector lens of SEM it can be seen an agglomerate of particles near the surface of the specimen. This is thought to be an inclusion that resulted in stress concentrations leading to the crack propagation in this area. The inclusions act as obstacles to dislocation movement and therefore after a certain amount of cycles a crack starts propagating through this area. Zone B illustrates the unstable part of the fracture and is characterised by the dimpled surfaces which is a normal feature of ductile materials. Figure 32 shows the specimen FSW-SN2 that broke in the stir zone with the same load condition as the previous specimen.

42

FSW-SN2, 25 407 cycles

Weld

B A

A A1

A1

B

Figure 32 – SEM observations of fatigue fracture surface (FSW-SN2; 25 407 cycles; Failure location - Weld)

The surface finish marks caused by the milling process may be the cause for the crack propagation. Zone A1 illustrates a combination of intergranular and transgranular fracture type. The crack has the tendency to propagate through and around the small recrystallized grains of the stir zone. Zone B evidences dimpled surfaces of the unstable part of the fracture. Figure 33 illustrates specimen FSW- SN4 that broke in the stir zone after 138 693 cycles with 275.8 MPa of load.

43

FSW-SN4, 138 693 cycles

Weld

A

B

A A1

A1

B

Figure 33 – SEM observations of fatigue fracture surface (FSW-SN4; 138 693 cycles; Failure location - Weld)

The crack propagation is not clear in this specimen. Zone A has the same failure mechanism as identified for the previous specimen (FSW-SN2), there is a combination of intergranular and transgranular fracture type. Transgranular fracture is also a characteristic of ductile fracture type. In Zone B, it can be seen a combination of striations and dimpled surfaces. Striations are characteristic features of fatigue fractures. This combination means that the material is in fatigue mode (striations)

44 and then a localized ductile fracture (appearance of dimples) occurs and afterwards the material enters again in fatigue mode (appearance of striations). This is a clear evidence of the natural tendency of this material, AA7050-T7451, to resist to fatigue. Figure 34 shows the fracture surface of specimen FSW- SN5 after 146 916 cycles with 275.8 MPa of load.

FSW-SN5, 146 916 cycles

BM

B

A

A A1

A1

B

Figure 34 – SEM observations of fatigue fracture surface (FSW-SN5; 146 916 cycles; Failure location - BM)

The crack propagated in an agglomerate of particles near the surface of the base material. An Energy Dispersive Spectroscopy (EDS) analysis was done in order to know which type of inclusion is it (see

45

Figure 35). This is a semi-quantitative analysis that gives a good indication of which inclusion is. For a

more accurate analysis, an X-ray diffraction or TEM analysis should be done.

Counts

Energy (keV) Figure 35 – Semi-quantitative analysis by EDS (Energy Dispersive Spectroscopy)

Table 12 – Weight percentage of and Copper

Element Weight, % Error, % Fe 11.49 3.7 Cu 26.64 3.4

The EDS analysis shows that there are abnormal quantities of Iron and copper in that specific area giving a good hint that is an Iron-Copper inclusion that is causing stress concentrations near the surface of the base material leading to the crack propagation (see Figure 35 and Table 12). Zone B has also the same feature of specimen FSW-SN4, there is a combination of striations and dimples.

The specimens that fractured in the stir zone and the base material have the same fracture mechanism which means that the process has little influence in fatigue performance and the presence of inclusions and the quality of the surface are determining factors to fatigue resistance. Comparing both pairs of specimens, there are no significant differences regarding the mechanisms of fracture in spite of the difference of the number of cycles or the different load conditions.

4.2.4. Tool Wear Assessment

The wear of the tool is important in order to maintain the quality of the weld and also to be competitive with other welding processes in terms of productivity. One of the main objectives is to perform more than 25 m of weld length without tool failure and guarantee good process repeatability (good mechanical properties and good microstructure) (see chapter 1.2). The tool wear assessment was done in order to conclude if wear is the main cause for tool failure. The probe normally breaks after a certain amount of length depending on the process parameters and forces during the process. This is a

46 preliminary study since this analysis was never made before in HZG. One of the conclusions of this study is that the wear is not the main cause for tool failure. However, there is a need of validation tests in the future in order to reach a more solid conclusion.

Tool wear assessment was done using the same parameters as those used for distortion and fatigue evaluation (700 rpm, 5 mm/s, 25 kN) [66]. The tool used for this analysis was the one used for the single side welds namely the 20 mm shoulder and 8 mm probe. It was possible to weld around 3.8 m until the complete failure of the tool which is quite far from the primary objective (see chapter 1.2). Figure 36 shows the tool before and after tool failure.

Figure 36 – Tool before and after failure (20 mm shoulder and 8 mm probe)

The probe breaks in the plane of the shoulder where the radial configuration lies. The tool is still rotating after breaking which results in the complete destruction of the fracture surface making impossible to evaluate it and withdraw more conclusions about the failure mechanism. Laser scans in the shoulder and probe were done before welding (reference stage) and after 2.4 m in order to quantify the amount of wear revealed. Figure 37 shows the 3D aspect of the probe taken from the laser scans before and after welding. The analysis of the probe was done only on one of the threaded sides.

47

Reference stage 2.4 m stage

Figure 37 - 3D representations of the 8 mm probe before and after welding (left – reference stage and right – 2.4 m stage)

Overall, there are no great visual modifications of the wear of the probe. In order to quantify the amount of wear, profiles were taken in the regions identified by the blue and green lines in Figure 37. The relative height between peak and valley was measured in different locations in order to assess the amount of wear observed in the respective locations (see Figure 38 and Figure 39). The height differences were compared between the reference stage and after 2.4 m of weld. Figure 38 represents the profile of the probe before and after welding (reference stage and after 2.4 m of weld).

Probe Profile Probe Profile Tool 1 TW - 3.8 m Tool 1 TW - 3.8 m Reference Stage 2.4 m 3,0 3,0 1 2,5 2 2,5 3 2,0 4 2,0

[mm] 5

1,5 1,5 6 1,0 1,0

Height [mm]

Height

0,5 0,5

0,0 0,0 0 2 4 6 8 0 2 4 6 8 Length [mm] Length [mm]

Figure 38 – Probe profile before and after welding (left – reference stage; right – 2.4 m stage)

48

Relative Height between Peak and Valley 0,9 0.01 0,8 0.007 0.06 0.01 0,7 0.01 0.03 0,6 0,5 0,4 Probe - Reference Stage 0,3 Probe - After 2.4 m 0,2 Relative Height [mm] Relative 0,1 0,0 1 2 3 4 5 6

Figure 39 – Relative height between peak and valley of the probe in different locations

As it can be seen from Figure 39 the amount of wear in the probe after 2.4 m is almost inexistent leading to the possibility that the wear may not be the main cause for tool failure. The same analysis was done in the shoulder and there are also not great visual modifications between the reference and after 2.4 m of weld (see Figure 40). There is however a small imperfection on the radial configuration near the interface between the probe and the inner diameter of the shoulder.

A A

B B

Reference stage 2.4 m stage

Figure 40 - 3D representations of the 20 mm shoulder before and after welding (left – reference stage; right – 2.4 m stage)

49

Two different profiles were taken before and after welding (Profile A and Profile B). These profiles were taken according to the locations identified by the green and blue lines in Figure 40. The relative height between peak and valley was measured in different locations in order to assess the amount of wear observed in the respective locations (see Figure 41 to Figure 44). The height differences were compared between the reference stage and after 2.4 m of weld for both profiles A and B.

Profile A

Shoulder Profile Shoulder Profile Profile A Tool 1 TW - 3.8 m Profile A Tool 1 TW - 3.8 m

Reference Stage 2.4 m 1,50 1,50

1,25 1 1,25 2 3 4 1,00 1,00

0,75 0,75

0,50 0,50

Height [mm]

Height [mm]

0,25 0,25

0,00 0,00 0 5 10 15 20 0 5 10 15 20 Length [mm] Length [mm]

Figure 41 – Shoulder profile before and after welding (Profile A) (left – reference stage; right – 2.4 m stage)

Profile A Relative Height between Peak and Valley 0,8

0.003 0.01 0.002 0,6 0.01

0,4 Shoulder - Reference Stage Shoulder - After 2.4 m

0,2

Relative Height [mm] Relative

0,0 1 2 3 4

Figure 42 - Relative height between peak and valley of the shoulder in different locations (Profile A)

50

Profile B

Shoulder Profile Shoulder Profile Profile B Profile B Tool 1 TW - 3.8 m Tool 1 TW - 3.8 m

Reference Stage 2.4 m 1,50 1,50

1,25 3 4 1,25 1 2 1,00 1,00

0,75 0,75

0,50 0,50

Height [mm]

Height [mm]

0,25 0,25

0,00 0,00 0 5 10 15 20 0 5 10 15 20 Length [mm] Length [mm]

Figure 43 - Shoulder profile before and after welding (Profile B) (left – reference stage; right – 2.4 m stage)

Profile B Relative Height between

Peak and Valley 0,8

0.004 0.004 0.006 0.003 0,6

Shoulder - Reference Stage

0,4 Shoulder - After 2.4 m

0,2

Relative Height [mm] Relative

0,0 1 2 3 4

Figure 44 - Relative height between peak and valley of the shoulder in different locations (Profile B)

As concluded for the probe, the amount of wear on the shoulder is also almost inexistent confirming the hypothesis that the wear is not the main cause for tool failure. The material of the probe is MP159 which is a nickel-cobalt based alloy and is a high strength and very though material and the shoulder is Hotvar, a molybdenum-vanadium based alloy. The hardness of both is, respectively, above 200 and 500 HV [80], [81] and are much higher than the hardness of the AA7050-T7451 (≈158 HV) so it is expected that the wear of the tool to be almost inexistent.

In order to ensure that the microstructure was not affected by the tool wear, metallography analysis was done after the beginning of the weld and also before the tool failure. Figure 45 illustrates the microstructure after the beginning of the process and Figure 46 shows the microstructure before the tool failure. Both macrographs clarify that there are no defects in the weld confirming that there is no influence of the tool wear during the process.

51

Figure 45 –Microstructure of weld FSW-PA-LB-1971 after beginning of the process

Figure 46 – Microstructure of weld FSW-PA-LB-1977 before tool failure

As it can be concluded the small wear verified in both the probe and the shoulder has no influence on the microstructure of the weld. From the results presented above it can be concluded that the wear is not the main cause for tool failure. There could be other reasons for tool failure such as fatigue or even defects in the material of the probe, either way these are mere assumptions. The forces involved during the process could also be a viable explanation for the failure of the tool. These forces induces a bending action in the probe causing it to fail in the plane of the shoulder. Figure 47 shows a schematic of the forces involved during the process.

52

P F

Bending Figure 47 – Schematic of the forces involved during the process

During the linear movement of the tool there is a reaction force, F, exerted by the cold material. The fixed probe counter reacts this force (P) and creates a bending stress in the probe. The region identified by the red line in Figure 47 is an area with a considerable amount of stress concentrations due to the threads of the probe. Following this statement, the probe tends to break in this critical area. It is important to mention that Figure 47 is a mere representation of the forces involved during the process and these in a real type situation would be characterized by distributed forces. Since the tool does not break instantly, there could be eventually a combination of failure mechanisms such as fatigue and forces involved during the process. The fact that the tool is completely destroyed (see Figure 36) makes the study of the fracture surface under SEM observation impossible. This would contribute to take more information about the mechanism of fracture and reach more solid conclusions about the reason for tool failure.

As an attempt to improve the tool life performance and also to improve the strength in the upper part of the probe (reducing sharp edges in this area), a small modification was made. The depth and angle of the threads was reduced and modified, respectively, with the purpose of making the tool more resistant and therefore live for a longer period of time. Figure 48 shows the different aspects of the new probe configuration. The same 20 mm shoulder was used for all welds.

Figure 48 – Original probe (left) and new configuration probe (right)

53

The same parameters (700 rpm, 5 mm/s and 25 kN) were used for this new probe configuration during welding. The tool survived for around 13 m which is a great improvement comparing to the original tool configuration. Regrettably, all welds presented volumetric defects in the microstructure which compromises the mechanical behaviour. As mentioned in the literature review there are several works that used different probe geometries and these have huge influence in the microstructure (see chapter 2.2.2). The following results confirm the great influence that the slight tool modification has on the microstructure. Figure 49 shows the microstructure after 2.4 m of weld with volumetric defects in the advancing side.

Figure 49 – Microstructure of weld FSW-PA-LB-1934 with defects using the new probe configuration and the same parameters (700 rpm, 5 mm/s and 25 kN)

In order to improve the microstructure behaviour the forging pressure was increased to 30 kN with the intention of eliminating the defects observed before in Figure 49. This slight parameter change is undesirable since all the fatigue and distortion assessments were already done with the optimum parameter (700 rpm, 5 mm/s and 25 kN) and if by chance this new process parameter works all the previous mechanical tests should be repeated in order to satisfy both the tool life performance (25 m of continuous weld length) and mechanical properties above 80 % of the base material properties.

Still, the new process parameters (700 rpm, 5 mm/s and 30 kN) were used to see if the defects would disappear and reach tool life performances higher than the initial 3.8 m. Regrettably, volumetric defects were still detected in the stir zone (advancing side) suggesting that the new probe configuration results in an inadequate material flow on the upper part of the weld which is the area where the depth of the thread is less pronounced (see Figure 50). Perhaps with a small modification

54 on the process parameters the defects could be eliminated but as said before that would be undesirable due to the need to repeat the mechanical tests in order to ensure that these new hypothetical set of parameters would satisfy the 80% efficiency goal in the previous mechanical tests.

Figure 50 – Microstructure of weld FSW-PA-LB-1973 with defects using the new probe configuration and an increase on the vertical force (700 rpm, 5 mm/s and 30 kN)

Although there were defects in the microstructure, the new probe configuration and the increase of the vertical force resulted in a tool life performance of around 11.5 m. Since there were defects in the microstructure it makes no sense to do a quantitative analysis as it was done for the first study. The same metallography condition as the first analysis, i.e. no defects in the microstructure, must be achieved in order to better compare both performances and features of the tool. Therefore, a qualitative analysis was done for the probe and shoulder. Figure 51 and Figure 52 show the aspect of the tool (probe + shoulder) after 9.6 m of weld which was the last scan taken before the tool failure at 11.5 m.

55

0 m (reference stage) 9.6 m stage

Figure 51 – 3D representations of the new 8 mm probe before and after welding (left – reference stage and right – 9.6 m stage)

0 m (reference stage) 9.6 m stage

Figure 52 – 3D representations of the 20 mm shoulder before and after welding (left – reference stage and right – 9.6 m stage)

There is not a significant visual modification regarding the wear of the probe or the shoulder that could lead to tool failure. However, it can be observed that after 9.6 m there are some cracks propagating

56 along the radius of the shoulder which influences on the surface appearance of the weld. The cracks in the shoulder are a clear indication that perhaps the shoulder cannot withstand the 25 m of weld. The same type of discontinuity in the radial configuration of the shoulder was observed as it occurred for the first study (see Figure 52). During welding the probe broke inside the inner shaft of the shoulder and the same was still rotating on top of the probe leaving the shoulder completely destroyed (see Figure 53). The high deformation of the tool makes the fracture surface analysis impossible.

Figure 53 - Tool after failure

Although there were defects in the microstructure, the new probe configuration showed a great improvement in performance. The new configuration would contribute to a stronger probe being capable to withstand higher forces during the process. The new process parameter should be such that would contribute to lower forces and better microstructure free of defects. This must be an iterative process because both the mechanical and tool life performance must be taken into consideration. The use of different materials in the probe or shoulder could be an interesting variable to be studied since wear is not the main cause for tool failure. Off course, the materials must be resistant enough in order to wear not become a factor in the quality of the weld.

57

4.3. AA7050-T7451 – Double Side FSW (DS-FSW)

In this section a summary of the previous analysis with the optimum parameter and the results of the distortion and fatigue assessments for double side FSW will be presented.

4.3.1. Summary of Previous Analysis [66]

Several weld experiments were done with variations in the welding parameters (rotational and welding speeds and vertical force) with the aim of determining the optimum parameters. The parameters were selected based on the experience obtained in single side FSW. All welds presented defect free microstructures.

The optimum parameter was chosen taking into consideration the microhardness and tensile evaluation. The set of parameters that offered the best mechanical properties was 1100 rpm, 9 mm/s and 15 kN (DS-FSW-PA-LB-1857). Figure 54 shows the microstructure of the optimum parameter with no defects in the microstructure. The microstructure of double side FSW has the same specific characteristics than the ones in single side FSW presented and discussed in chapter 4.2.1 with the only difference of the shape of the stir zone which is composed by two passes.

Since all welds presented a defect free microstructure all tensile test specimens fractured in the most vulnerable area of all which is the HAZ evidencing a slight necking (ductile fracture). There was no failure side preference like in single side FSW. The tensile test and microhardness results had better performance than single side FSW. The temperature measurements revealed that there was no tendency for higher temperatures in one of the sides, advancing or retreating side. The use of a smaller tool and the same position for the thermocouple holes may be the reason for the lower temperatures (retreating side ≈ 188.9 ± 9.8 °C and advancing side ≈ 182.3 ± 9.2 °C). Moreover, the temperatures were much lower than those verified in single side FSW which was expected since a smaller tool was being used and less energy was being delivered to the workpiece in each pass. Table 13 illustrates all the mechanical properties of the optimum parameter.

Table 13 – Summary of mechanical test results for double side FSW of AA7050-T7451 8 mm thick [66]

Designation Rotational Welding Min. Yield Heat Vertical UTS, Elongation, (DS-FSW-PA- speed, speed, Hardness, Strength, Input force, kN MPa % LB-XXXX) rpm mm/s HV MPa A/B, kJ/mm 421.5 ± 510.6 ± 137.4 ± 2.5 10.7 ± 0.2 0.22/ 1857 1100 9 15 0.7 1.3 (86.2%) (65.2%) 0.23 (93.5%) (≈100%)

58

TMAZ SZ BM BM HAZ HAZ

SZ TMAZ

Figure 54 – Microstructure of the weld DS-FSW-PA-LB-1857 (1100 rpm, 9 mm/s, 15 kN)

59

The set of parameters identified in Table 13 resulted in particularly good results answering to the 5 mm/s and 80 % efficiency goals. This set of parameters was defined by another student after metallography, hardness and tensile evaluation.

In order to complement this study, the parameters were used to perform the welds for distortion and fatigue assessments which are crucial mechanical tests for the aeronautical industry.

4.3.2. Distortion Assessment

The study of distortion of welded plates is important in order to control and limit the deformation caused by the welding process. Distortion assessment was performed using the parameters that had the best mechanical performance (1100 rpm, 9 mm/s and 15 kN) as acknowledged in chapter 4.3.1. The deformations obtained for double side FSW were higher than those obtained for single side FSW.

The same experimental procedure as single side welds was used for the distortion assessment of double side welds and can be found in chapter 3.3.2. Figure 55 shows the reference plate obtained from Pontos system, respectively. Although there is no information about the residual stresses that the plates may have due to the manufacture process, these were used as the reference stage and the displacements were computed in relation to them. Figure 56 and Figure 57 illustrate the relative displacement caused by the process for the first and second passes, respectively. The displacements of both passes were computed in relation to the un-welded plates (see Figure 55).

Figure 55 – Reference plate (un-welded) obtained from Pontos System

60

WD

Figure 56 - Distortion assessment for double side FSW of AA7050-T7451 8 mm thick (WD – Welding Direction) – First pass

WD

Figure 57 - Distortion assessment for double side FSW of AA7050-T7451 8 mm thick (WD – Welding Direction) – Second pass

61

The plates after the first pass tend to deform on the outer corners due to the fact that only half of the thickness is welded, causing the plate somehow to deform more in the corner areas. The other half of the plate is a free surface and therefore the workpiece is more prone to distortion. After the second pass, there is an excess of deformation in the end of the plate that can be explained due to the heat that is being given in the end of the weld during each pass. The welding direction is done in the same way in both passes causing the end of the plate to be submitted to higher heat during both passes thus causing higher deformations in the end of the plate. The loss of properties (softening) due to heat input is also on the basis for the higher deformations. Table 14 represents the maximum and minimum displacement values obtained from the Pontos optical system for the first and second passes. The distortion measured in both x and y directions are small and can be neglected. The maximum displacement measured was around 1.5 and 3.3 mm for the first and second passes, respectively.

Table 14 - Maximum and minimum displacement values for double side FSW of AA7050-T7451 8 mm thick

Double side FSW of AA7050-T7451 8 mm Max. Min. dx [mm] 0.03 -0.09 First Pass dy [mm] -0.03 -0.35 dz [mm] 1.49 0.03 dx [mm] 0.15 -0.09 Second dy [mm] 0.43 -0.25 Pass dz [mm] 3.33 -0.15

There is no information available about the distortion shape behaviour of similar aluminium plates using the double side FSW method. There are some studies of distortion using double side FSW method but all of them use steel and the shape distortion behaviour of steel is different than aluminium (see chapter 2.3) making impossible to compare these results to other welding processes or to similar plates. Either way, it is expected that the distortions of double side FSW are still lower than those of other arc welding processes.

4.3.3. Fatigue Assessment

Fatigue evaluation was performed using the set of parameters that had the best mechanical properties (1100 rpm, 9 mm/s and 15 kN) [66]. Particularly good results were obtained having almost the same fatigue strength of the base material. A fracture surface analysis was done in order to conclude why some specimens with the same load conditions were breaking in the base material and others in the welded area.

62

Before the fatigue test, a non-destructive test was done to ensure that there were no defects throughout the as-welded plates. Figure 58 illustrates an X-ray image of double side FSW plate.

Figure 58 – X-ray image of one double side FSW plate of AA7050-T7451 8 mm thick (as welded)

According to Figure 58, the welded plates are free of defects throughout the entire joint and therefore can be submitted to fatigue testing. The defects could lead to stress concentrations which would decrease the fatigue resistance. Table 15 displays the fatigue data of double side FSW of AA7050-T7451 8 mm thick with different loads with the respective cycles. A Weibull distribution was chosen to analyse the data and obtain S-N diagrams for different reliabilities.

63

Table 15 – Fatigue data for double side FSW of AA7050-T7451 6.5 mm thick

Weibull Max. Min. Mean Mean Cycles σmáx, Amp., (36.8% Failure Specimen Load, Load, Load, Cycles MPa kN Location kN kN kN reliability)

DS-FSW-SN1 60.606 6.061 33.333 27.273 72 819 BM DS-FSW-SN2 310.3 60.657 6.066 33.362 27.296 67 533 72 412 Weld DS-FSW-SN3 60.606 6.061 33.333 27.273 78 162 BM DS-FSW-SN4 54.138 5.414 29.776 24.362 88 237 Weld DS-FSW-SN5 275.8 53.999 5.400 29.700 24.300 107 522 87 509 BM DS-FSW-SN6 54.083 5.408 29.746 24.337 67 615 Weld DS-FSW-SN7 50.754 5.075 27.915 22.839 567 536 Weld DS-FSW-SN8 50.789 5.079 27.934 22.855 779 842 BM 258.6 481 771 DS-FSW-SN9 50.641 5.064 27.853 22.788 181 120 BM DS-FSW-SN10 50.754 5.075 27.915 22.839 81 074 Weld DS-FSW-SN11 47.307 4.731 26.019 21.288 336 683 Weld DS-FSW-SN12 241.3 47.232 4.723 25.978 21.255 493 269 1 296 890 BM DS-FSW-SN13 47.192 4.719 25.956 21.237 2 258 135 BM DS-FSW-SN14 40.549 4.055 22.302 18.247 4 345 833 Run-Out DS-FSW-SN15 40.590 4.059 22.325 18.266 4 219 637 Run-Out DS-FSW-SN16 206.9 40.458 4.046 22.252 18.206 4 000 000 - Run-Out DS-FSW-SN17 40.465 4.047 22.256 18.209 4 000 000 Run-Out DS-FSW-SN18 40.492 4.049 22.271 18.221 4 000 000 Run-Out Stress ratio – 0.1, Frequency – 10 Hz, Run-Out criteria – 4 000 000 cycles

According to the literature the fatigue strength of the welded specimens is close to the base material’s [79]. The fatigue strength of the welded specimens is around 206.9 MPa. Figure 59 illustrates the S-N diagram with different reliability curves.

64

S-N Diagram for Double Side FSW of AA7050-T7451 400

350 y = 650,41x-0,071 300 [MPa] R² = 0,8403 36,8 %

max 250 σ Run-Out 200 (4 M) 90 % 150 Axial force fatigue tests in 6.5 mm AA7050-T7451, Double Side FSW σ =422 MPa [Yield strength of welded specimen] Max. Stress, Stress, Max. 100 y Tests performed at room temperature, R=0.1, f=10 Hz 50 10000 100000 1000000 10000000 Fatigue Life, N, cycles

Figure 59 – S-N Diagram for Double Side FSW of AA7050-T7451 6.5 mm thick

The specimens highlighted in blue and green in Table 15 broke in the base material and in the weld. SEM observations were done with the intention of knowing the reason why the specimens were breaking in different locations even though the loads were the same for each pair of specimens. Figure 60 shows the specimen DS-FSW-SN1 that broke in the base material with 310.3 MPa of load after 72 819 cycles.

DS-FSW-SN1, 72 819 cycles

BM

A

65

A A1

A1

Figure 60 – SEM observations of fatigue fracture surface (DS-FSW-SN1; 72 819 cycles; Failure location - BM) The agglomerate of particles located near the surface of the base material caused stress concentrations leading eventually to the complete failure of the specimen. Figure 61 illustrates specimen DS-FSW-SN2 with the same load condition as the previous specimen and broke in the stir zone after 67 533 cycles.

DS-FSW-SN2, 67 533 cycles

Weld

A B

A B

Figure 61 – SEM observations of fatigue fracture surface (DS-FSW-SN2; 67 533 cycles; Failure location - Weld)

Zone A shows the crack propagation that may have resulted from one of the surface finish marks resulting from the milling process. Zone B indicates the intergranular and transgranular fracture type

66 as already seen in single side FSW specimens. Figure 62 shows specimen DS-FSW-SN4 that broke in the stir zone with 275.8 MPa of load after 88 237 cycles.

DS-FSW-SN4, 88 237 cycles

Weld

B C

A

A B

C

Figure 62 – SEM observations of fatigue fracture surface (DS-FSW-SN4; 88 237 cycles; Failure location - Weld)

The combination of striations and dimples is also present in the double side FSW specimens. Zone B evidences this combination as well as a couple of micro-cracks due to the closeness to the unstable part of the fracture. Zone C demonstrates a clear view of the dimpled surfaces in the unstable part of the fracture that characterizes the plate as being a ductile material.

67

DS-FSW-SN5, 107 522 cycles

BM

B

A

A A1

A1

B

Figure 63 – SEM observations of fatigue fracture surface (DS-FSW-SN5; 107 522 cycles; Failure location - BM)

Zone B demonstrates also a combination of striations and dimples which confirms that the different approach of FSW process has no influence in the mechanism of fatigue failure. The material AA7050- T7451 has a natural tendency to resist to fatigue and that is one of the reasons why this material is widely used in many industries and applications. Comparing both pairs of specimens there are no significant differences regarding the mechanisms of fracture in spite of the difference of the number of cycles or the different load conditions.

68

4.4. AA7050-T7451 – Single Side FSW Vs. Double Side FSW

In this section a comparison of the results of both single and double side FSW will be discussed.

4.4.1. Summary of Previous Analysis [66]

Both approaches resulted in particularly good results answering to the 5 mm/s and 80 % efficiency goals. The mechanical test results of the double side FSW were much higher than those of the single side FSW (see Table 16).

Table 16 – Summary of mechanical test results for single and double side FSW and base material AA7050- T7451 8 mm thick [66]

Welding Min. Yield Rotational Vertical UTS, Elongation, Heat Designation speed, Hardness, Strength, speed, rpm force, kN MPa % Input, mm/s HV MPa kJ/mm 451.0 ± 507.1 ± Base Material - - - 158.5 ± 3.0 16.4 ± 0.5 0.8 0.8 - 362.7 ± 475.0 ± FSW-PA-LB- 127.1 ± 2.5 7.1 ± 0.4 700 5 25 2.5 2.4 0.60 1851 (80.2%) (43.3%) (80.4%) (93.7%) 421.5 ± 510.6 ± DS-FSW-PA- 137.4 ± 2.5 10.7 ± 0.2 0.22/ 1100 9 15 0.7 1.3 LB-1857 (86.2%) (65.2%) 0.23 (93.5%) (≈100%)

The lower heat input delivered in double side FSW contributed to a similar mechanical performance than the base material. However, the need for more preparation in double side approach could be a negative factor in terms of productivity.

In order to complement this work and to decide which approach is the most appropriate, complementary mechanical tests such as distortion and fatigue assessments were performed. The following chapters compares both approaches in order to sum up all the results and also to better understand the final conclusions.

4.4.2. Distortion Assessment

The relative displacements of double side FSW were higher than those achieved for single side FSW. It was expected that the distortions of double side FSW would be smaller than single side FSW since the heat input is less in each pass and is delivered in a more phased manner (see Table 16). Table 17 shows a summary of the relative displacements of both approaches.

69

Table 17 – Maximum and minimum displacement values for single and double side FSW of AA7050-T7451 8 mm thick

Relative Displacements Max. Min. dx [mm] 0.30 -0.01 Single Side FSW dy [mm] 0.33 -0.22 dz [mm] 1.03 -0.56 dx [mm] 0.03 -0.09 First Pass dy [mm] -0.03 -0.35 Double Side dz [mm] 1.49 0.03 FSW dx [mm] 0.15 -0.09 Second dy [mm] 0.43 -0.25 Pass dz [mm] 3.33 -0.15

However, the free surface left during the double side process and the welding direction done in the same way in both passes were the causes for the bigger distortions in the double side FSW. Contrary to fusion processes, there was not a compensation after the second pass, instead, the deformation after the second pass was even higher. This may be related to the experimental procedure used during the double side process. The welding direction was done in the same way for both passes, causing the end of the plates to be submitted to higher heat inputs during each pass. In order to control the distortion of the plate, an alternate direction can be used during each pass as a first solution. The loss of properties (softening) due to heat input after the first pass is also a factor for the susceptibility to higher deformations. As discussed in chapter 2.3, Richter-Trummer et al. [53], concluded that the higher the clamping forces, the more restrained the material is and, therefore, the lower the distortions [53]. Following this idea, another solution could be the use of an efficient clamping system that would allow a good fixation of the workpiece. Naturally, there would be an increase in the residual stresses in the workpiece that could be detrimental in terms of performance. For this reason, a heat treatment would be necessary for stress relaxation of the workpiece. Either way, despite there is no design criteria for the distortion measurements, the results for both approaches can be considered as acceptable taking into consideration the dimensions of the plate (1200 mm x 400 mm). As mentioned before, there is no information about the residual stresses of the plates, therefore it is not possible to conclude the effect of the process in terms of distortion. It would be interesting to evaluate the residual stresses from the manufacture process and compare with the residual stresses from the clamping system and the process itself in order to conclude which influences the most in terms of distortion.

70

4.4.3. Fatigue Assessment

The S-N diagram of double side FSW is very similar to single side FSW (see Figure 64) which is an indication that the different approaches have little influence in the fatigue performance.

S-N Diagram for Single and Double Side FSW of AA7050-T7451 400

350 DS-FSW y = 650,41x-0,071 300 Double Side FSW [MPa] R² = 0,8403 FSW max 250 -0,056 σ y = 546,91x Run-Out (4 M) R² = 0,8999 200 Single Side FSW 150 Axial force fatigue tests in 6.5 mm AA7050-T7451, Single Side FSW

Max. Stress, Stress, Max. 100 and Double Side FSW Tests performed at room temperature, R=0.1, f=10 Hz 50 10000 100000 1000000 10000000 Fatigue Life, N, cycles

Figure 64 - S-N diagram of double and single side FSW of AA7050-T7451 6.5 mm thick

According to the literature the fatigue strength of the welded specimens is close to the base material’s [79]. The fatigue strength of both single and double side FSW specimens is around 206.9 MPa. The same fracture mechanism occurred in both approaches; the specimens that broke in the base material evidenced an agglomerate of particles on the surface of the specimen that caused stress concentrations leading eventually to the complete failure after a certain amount of cycles. The specimens that broke in the stir zone demonstrated a mixture of intergranular and transgranular fracture which is a characteristic feature of a ductile material. In addition, for both specimens that broke in the stir zone and base material, a combination of striations and dimples were observed indicating that the material, AA7050-T7451, has a natural tendency to resist to fatigue and also the process has little influence in fatigue failure. This was observed for both approaches.

Overall, the different approaches have little influence in fatigue performance in spite of the better static properties of double side FSW. The presence of inclusions and the quality of the surface are determining factors to fatigue behaviour. Given the fact that the presence of inclusions in the base material caused the premature failure of the specimens, it is of most importance to verify and control the quality of the base material after the manufacture process. On top of that situation, other

71 specimens broke in the base material (see Table 11 and Table 15), so there is a great possibility that the inclusions may be affecting the fatigue performance of the other samples.

5. Summary and Conclusions

As discussed before, FSW offers great mechanical properties for heat treatable alloys and is a viable alternative for mechanical joining and fusion welding. The 8 mm aluminium alloy AA7050-T7451 was successfully welded with remarkably good mechanical results. The challenging requirements mentioned in chapter 1.2 were achieved with exception of the tool wear performance. Two different approaches were taken, single and double side FSW, both turned out to have overall similar results in terms of mechanical properties. It is important to mention that the main focus of this work was distortion, fatigue and tool wear assessment and according to the experimental work carried out on AA7050-T7451 8 mm thick aluminium alloy it can be concluded:

1. The distortions of double side FSW were higher than in single side FSW. The maximum relative displacement measured in double side FSW was 1.5 mm for the first pass and 3.3 mm for the second pass and the maximum relative displacement for single side FSW was 1 mm on the outer edges of the plates. It was expected that the distortions of double side FSW would be smaller than single side FSW due to the less energy delivered during each pass. However, the distortions of either single or double side FSW can be considered as small.

2. The material AA7050-T7451 showed a natural tendency to resist to fatigue. Both fatigue S-N diagrams for single and double side FSW were very similar (see Figure 64) which indicates that the different approaches have no influence in fatigue performance. The inclusions and the surface finish marks are responsible for stress concentrations leading to crack propagation and then eventually to complete failure.

3. The tool wear measured in the probe and shoulder was considerably small and therefore the wear is not the main cause for tool failure. Additionally, there was no influence on the microstructure of the weld. The tool life performance was around 3.8 m which is much lower than the 25 m requirement. Although the trials with the new probe configuration evidenced volumetric defects in the microstructure, the new probe showed a great improvement in tool life performance, around 13 m and 11.5 m.

Overall, both approaches evidenced similar results. Taking into consideration that nowadays production costs and productivity are crucial factors in the industry, single side FSW seems to be the most appropriate approach to be used in the future given the need for less preparation and less time

72 to weld the same amount of length. This research work will contribute for a better understanding about the mechanical behavior of thicker materials and will serve as a base for future works.

6. Future Work

In this section, some suggestions are presented for future projects that will contribute to improve the results already presented in this work.

Distortion Assessment

 Residual stresses analysis to complement the distortion assessment

Fatigue Assessment

 Fatigue tests for the base material for a more accurate comparison with the FSW process  More SEM observations to conclude if the inclusions are also causing the fatigue failure in other specimens

Tool Wear Assessment

 Design of a new clamping device for the laser microscope for the tool wear assessment in order to reduce the human factor in the measuring process and guarantee that the tool is secured in the same position in every stage of the procedure. The position of the current measurements was done with the use of gauge blocks which could still lead to inaccuracies.  Validation tests for tool wear assessment

7. References

[1] F. Mazzolani, Aluminium Alloy Structures, Second Edition. CRC Press, 1994. [2] R. S. Mishra and Z. Y. Ma, “Friction stir welding and processing,” Mater. Sci. Eng. R Reports, vol. 50, no. 1–2, pp. 1–78, 2005. [3] TWI, “Friction Stir Welding,” TWI. [Online]. Available: http://www.twi- global.com/capabilities/joining-technologies/friction-processes/friction-stir-welding/benefits- and-advantages/. [Accessed: 01-Mar-2015]. [4] A. Heinz, A. Haszler, C. Keidel, S. Moldenhauer, R. Benedictus, and W. . Miller, “Recent development in aluminium alloys for aerospace applications,” Mater. Sci. Eng. A, vol. 280, no. 1, pp. 102–107, 2000. [5] J. Q. Su, T. W. Nelson, R. Mishra, and M. Mahoney, “Microstructural investigation of friction stir welded 7050-T651 aluminium,” Acta Mater., vol. 51, no. 3, pp. 713–729, 2003. [6] B. J. Dracup and W. J. Arbegast, “Friction Stir Welding as a Rivet Replacement Technology,” in Automated Fastening Conference & Exposition, 1999. [7] P. L. Threadgill, a J. Leonard, H. R. Shercliff, and P. J. Withers, “Friction stir welding of aluminium

73

alloys,” Int. Mater. Rev., vol. 54, no. 2, pp. 49–93, 2009. [8] D. Deng, H. Murakawa, and W. Liang, “Numerical simulation of welding distortion in large structures,” Comput. Methods Appl. Mech. Eng., vol. 196, no. 45–48, pp. 4613–4627, Sep. 2007. [9] Q.-Y. Shi, J. Silvanus, Y. Liu, D.-Y. Yan, and H.-K. Li, “Experimental study on distortion of Al-6013 plate after friction stir welding,” Sci. Technol. Weld. Join., Dec. 2013. [10] F. C. Campbell, “Elements of Metallurgy and Engineering Alloys,” 1st ed., USA: ASM International, 2008, pp. 243–265. [11] R. Nandan, T. DebRoy, and H. K. D. H. Bhadeshia, “Recent advances in friction-stir welding - Process, weldment structure and properties,” Prog. Mater. Sci., vol. 53, no. 6, pp. 980–1023, 2008. [12] R. A. Prado, L. E. Murr, K. F. Soto, and J. C. McClure, “Self-optimization in tool wear for friction- stir welding of Al 6061+20% Al2O3 MMC,” Mater. Sci. Eng. A, vol. 349, no. 1–2, pp. 156–165, May 2003. [13] P. Fleming, T. Bloodworth, T. Prater, and et. al, “An investigation in Tool Wear Detection in Friction Stir Welding,” Vanderbilt School of Engineering. [14] J. G. Perret, J. Martin, P. L. Threadgill, and M. M. Z. Ahmed, “Recent developments in friction stir welding of thick section aluminium alloys,” in Paper presented at 6th World Congress, Aluminium Two Thousand, Florence, Italy, 2007. [15] D. A. Kumar, P. Biswas, S. Tikader, M. M. Mahapatra, and N. R. Mandal, “A study on friction stir welding of 12mm thick aluminum alloy plates,” J. Mar. Sci. Appl., vol. 12, no. 4, pp. 493–499, Jan. 2014. [16] E. M. D. Lopes and R. M. Miranda, “Metalurgia da Soldadura,” in Edições Técnicas ISQ, Edições Técnicas ISQ, Ed. p. 228. [17] C. Gilmore, Materials Science and Engineering Properties. Cengage Learning, 2014. [18] C. M. Branco, A. A. Fernandes, and P. M. S. T. Castro, Fadiga de estruturas soldadas. Fundação Calouste Gulbenkian, 1986. [19] C. Vidal, “Análise da Melhoria do Comportamento à Fadiga em Juntas Soldadas por Fricção Linear de Ligas de Aluminio para a Industria Aeronáutica,” MSc Thesis, IST, 2009. [20] The Aluminium Automotive Manual, “Materials - Designation System,” 2002. [21] J. Hale, “BOEING 787 - FROM THE GROUND UP,” BOEING. [Online]. Available: http://www.boeing.com/commercial/aeromagazine/articles/qtr_4_06/article_04_2.html. [22] E. A. Starke and J. T. Staleyt, “Application of Modern Aluminium Alloys to Aircraft,” vol. 32, no. 95, pp. 131–172, 1996. [23] C. Bray, “Dictionary of Glass: Materials and Techniques,” vol. 5, University of Pennsylvania Press, 2001, p. 179. [24] ESAB, “How to Avoid Cracking in Aluminium Alloys,” ESAB Knowledge Center, 2014. [Online]. Available: http://www.esab.ca/ca/en/education/blog/how-to-avoid-cracking-in-aluminum- alloys.cfm. [Accessed: 20-May-2002]. [25] P. a. Molian and T. S. Srivatsan, “Weldability of aluminium-lithium alloy 2090 using laser welding,” J. Mater. Sci., vol. 25, no. 7, pp. 3347–3358, 1990. [26] G. Mathers, The Welding of Aluminium and its Alloys. Woodhead Publishing, 2002.

74

[27] Q. Yunlian, D. Ju, H. Quan, and Z. Liying, “Electron beam welding, laser beam welding and of titanium sheet,” Mater. Sci. Eng. A, vol. 280, no. 1, pp. 177–181, Mar. 2000. [28] S. Malarvizhi and V. Balasubramanian, “Effect of welding processes on AA2219 aluminium alloy joint properties,” Trans. Nonferrous Met. Soc. China (English Ed., vol. 21, no. 5, pp. 962–973, 2011. [29] F. C. Campbell, Manufacturing Technology for Aerospace Structural Materials. Elsevier, 2006. [30] P. Vilaça, J. P. Santos, a Góis, and L. Quintino, “Joining Aluminium Alloys Dissimilar in Thickness By Friction Stir Welding and Fusion Processes,” vol. 49, pp. 56–62, 2005. [31] B. Carter, “Introduction to Friction Stir Welding ( FSW ),” NASA. [32] B. T. Gibson, D. H. Lammlein, T. J. Prater, W. R. Longhurst, C. D. Cox, M. C. Ballun, K. J. Dharmaraj, G. E. Cook, and a. M. Strauss, “Friction stir welding: Process, automation, and control,” J. Manuf. Process., vol. 16, no. 1, pp. 56–73, 2014. [33] G. H. Payganeh, N. B. M. Arab, Y. D. Asl, F. a. Ghasemi, and M. S. Boroujeni, “Effects of friction stir welding process parameters on appearance and strength of polypropylene composite welds,” Int. J. Phys. Sci., vol. 6, no. 19, pp. 4595–4601, 2011. [34] J. W. Pew, T. W. Nelson, and C. D. Sorensen, “Torque based weld power model for friction stir welding,” Sci. Technol. Weld. Join., vol. 12, no. 4, pp. 341–347, 2007. [35] P. Vilaça, L. Quintino, and J. F. dos Santos, “iSTIR—Analytical thermal model for friction stir welding,” J. Mater. Process. Technol., vol. 169, no. 3, pp. 452–465, 2005. [36] I. Radisavljevic, a. Zivkovic, N. Radovic, and V. Grabulov, “Influence of FSW parameters on formation quality and mechanical properties of Al 2024-T351 butt welded joints,” Trans. Nonferrous Met. Soc. China (English Ed., vol. 23, no. 12, pp. 3525–3539, 2013. [37] T. Saeid, a. Abdollah-zadeh, H. Assadi, and F. Malek Ghaini, “Effect of friction stir welding speed on the microstructure and mechanical properties of a duplex stainless steel,” Mater. Sci. Eng. A, vol. 496, no. 1–2, pp. 262–268, 2008. [38] M. H. Tolephih, H. M. Mahmood, and H. H. Esam, “Effect of tool offset and tilt angle on weld strength of butt joint friction stir welded specimens of AA2024 aluminum alloy welded to commercial pure copper,” IISTE, vol. 3, no. 4, 2013. [39] K. Kimapong and T. Watanabe, “Effect of Welding Process Parameters on Mechanical Property of FSW Lap Joint between Aluminum Alloy and Steel,” Mater. Trans., vol. 46, no. 10, pp. 2211– 2217, 2005. [40] P. Podrzaj, B. Jerman, and D. Klobcar, “Welding defects at friction stir welding,” Metalurgija, vol. 54, no. 2, pp. 387–389, 2015. [41] J. Mohammadi, Y. Behnamian, a. Mostafaei, and a. P. Gerlich, “Tool geometry, rotation and travel speeds effects on the properties of dissimilar magnesium/aluminum friction stir welded lap joints,” Mater. Des., vol. 75, pp. 95–112, 2015. [42] M. Ilangovan, S. Rajendra Boopathy, and V. Balasubramanian, “Effect of tool pin profile on microstructure and tensile properties of friction stir welded dissimilar AA 6061–AA joints,” Def. Technol., vol. 11, no. 2, pp. 174–184, Jun. 2015. [43] P. Upadhyay and a. P. Reynolds, “Effects of forge axis force and backing plate thermal diffusivity on FSW of AA6056,” Mater. Sci. Eng. A, vol. 558, pp. 394–402, 2012. [44] U. A. Mercado, “Microstructure, mechanical behavior and corrosion properties of friction stir

75

welded aluminum alloys used in the aerospace industry,” in Vortrag Dissertation, 2011. [45] K. N. Krishnan, “On the formation of onion rings in friction stir welds,” Mater. Sci. Eng. A, vol. 327, no. 2, pp. 246–251, Apr. 2002. [46] Oxford, “Characteristic texture variations in a friction stir welded aluminium alloy,” Oxford - Instruments, The Business of Science. [47] W. M. Thomas, S. W. Kallee, D. G. Staines, and P. J. Oakley, “Friction Stir Welding – Process Variants and Developments in the Automotive Industry,” SAE 2006 World Congress & Exibition, Apr. 2006. [48] H. Wu, Y.-C. Chen, D. Strong, and P. Prangnell, “Stationary shoulder FSW for joining high strength aluminum alloys,” J. Mater. Process. Technol., vol. 221, pp. 187–196, 2015. [49] M. K. Kulekci, U. Esme, and O. Er, “Experimental comparison of resistance spot welding and friction-stir spot welding processes for the en aw 5005 aluminum alloy,” Mater. Tehnol., vol. 45, no. 5, pp. 395–399, 2011. [50] L. Blaga, S. T. Amancio-Filho, J. F. dos Santos, and R. Bancila, “FricRiveting as a new joining technique in GFRP lightweight bridge construction,” Constr. Build. Mater., pp. 1–33, 2015. [51] S. Kalle, “Application of friction stir welding in the shipbuilding industry,” The Royal Institution of Naval Architects, 2000. [Online]. Available: http://www.twi-global.com/technical- knowledge/published-papers/application-of-friction-stir-welding-in-the-shipbuilding-industry- february-2000/. [Accessed: 01-Mar-2015]. [52] P. Mendez and T. Eagar, “New trends in welding in the aeronautic industry,” Manuf. Aeronaut. Ind., pp. 1–15, 2000. [53] V. Richter-Trummer, E. Suzano, M. Beltrão, a. Roos, J. F. dos Santos, and P. M. S. T. de Castro, “Influence of the FSW clamping force on the final distortion and residual stress field,” Mater. Sci. Eng. A, vol. 538, pp. 81–88, 2012. [54] S. Larose, L. Dubourg, C. Perron, M. Jahazi, and P. Wanjara, “Limitation of Distortion in Friction Stir Welded (FSW) Panels Using Needle Peening,” Mater. Sci. Forum, vol. 638–642, pp. 1203– 1208, Jan. 2010. [55] S. R. Bhide, P. Michaleris, M. Posada, and J. DeLoach, “Comparison of Buckling Distortion Propensity for SAW, GMAW and FSW,” Weld. J., no. American Welding Society and Welding Research Council, 2006. [56] N. McPherson, A. Galloway, S. R. Cater, and M. M. Osman, “A comparison between single sided and double sided friction stir welded 8mm thick DH36 steel plate,” 9th International Conference on Trends in Welding Research. 04-Jun-2012. [57] L. Reis and V. Infante, “Mechanical Behaviour of Materials.” IST, Lisbon, 2014. [58] Y.-L. Lee, J. Pan, R. Hathaway, and M. Barkey, “Fatigue Testing and Analysis,” in Fatigue Testing and Analysis, Elsevier, 2005. [59] S. Maddox, “Review of fatigue assessment procedures for welded aluminium structures,” Int. J. Fatigue, vol. 25, no. 12, pp. 1359–1378, Dec. 2003. [60] T. R. Gurney, Fatigue of Welded Structures. CUP Archive, 1979. [61] L. Trško, O. Bokůvka, F. Nový, and M. Guagliano, “Effect of severe shot peening on ultra-high- cycle fatigue of a low-alloy steel,” Mater. Des., vol. 57, pp. 103–113, May 2014. [62] C. Zhou, X. Yang, and G. Luan, “Comparative study on fatigue properties of friction stir and MIG-

76

pulse welded joints in 5083 Al-Mg alloy,” Trans. Nonferrous Met. Soc. China, vol. 15, 2005. [63] T. Zhang, Y. He, Q. Shao, H. Zhang, and L. Wu, “Comparative Study on Fatigue Properties of Friction Stir Welding Joint and Lap Joint,” in 13th International Conference on Fracture, 2013, pp. 1–10. [64] W. ZENG, “EFFECT OF TOOL WEAR ON MICROSTRUCTURE, MECHANICAL PROPERTIES AND ACOUSTIC EMISSION OF FRICTION STIR WELDED 6061 Al ALLOY,” Acta Metall. Sin. (English Lett., vol. 19, no. 1, pp. 9–19, Feb. 2006. [65] T. Prater, A. Strauss, G. Cook, B. Gibson, and C. Cox, “A Comparative Evaluation of the Wear Resistance of Various Tool Materials in Friction Stir Welding of Metal Matrix Composites,” J. Mater. Eng. Perform., vol. 22, no. 6, pp. 1807–1813, Jan. 2013. [66] V. Correia, “Friction stir welding processes in AA7050 thick plates,” MSc Thesis, University of Porto (FEUP), 2015. [67] ISO 25239, “Friction Stir Welding.” ISO, 2011. [68] ASTM E466-96, “Standard Practice for Conducting Force Controlled Constant Amplitude Axial Fatigue Tests of Metallic Materials.” ASTM International, 2002. [69] ASTM E467-98a, “Standard Practice for Verification of Constant Amplitude Dynamic Forces in an Axial Fatigue Testing System.” ASTM International, 8004. [70] ASTM E468-90, “Standard Practice for Presentation of Constant Amplitude Fatigue Test Results for Metallic Materials.” ASTM International, 2004. [71] ASTM E739-10, “Standard Practice for Statistical Analysis of Linear or Linearized Stress-Life (S- N) and Strain-Life (ε-N) Fatigue Data.” ASTM International. [72] R. Sakin and İ. Ay, “Statistical analysis of bending fatigue life data using Weibull distribution in glass-fiber reinforced polyester composites,” Mater. Des., vol. 29, no. 6, pp. 1170–1181, Jan. 2008. [73] P. S. Effertz, V. Infante, L. Quintino, U. Suhuddin, S. Hanke, and J. F. dos Santos, “Fatigue life assessment of friction spot welded 7050-T76 aluminium alloy using Weibull distribution,” Int. J. Fatigue, vol. 87, pp. 381–390, Jun. 2016. [74] “Matweb - Material Property Data.” [Online]. Available: http://www.matweb.com/. [75] DIN 50125, “Prufung metallischer Werkstoffe - Zugproben.” DIN, 2009. [76] ASTM E8/E8M-09, “Standard test Methods for Tension Testing of Metallic Materials,” ASTM Int., pp. 1–27, 2010. [77] ASTM E92-82, “Standard Test Method for Vickers Hardness of Metallic Materials,” ASTM, vol. 82, no. Reapproved 2003. ASTM International, pp. 1–9, 2010. [78] DIN EN ISO 9015, “Destructive tests on welds in metallic materials - Hardness Testing.” DIN, 2011. [79] Department of Defence of United States of America, “Military Handbook - Metallic Materials and Elements for Aerospace Vehicle Structures,” 1998. [80] “Aircraft Materials - Alloy MP159 (AMS 5841).” [Online]. Available: http://www.aircraftmaterials.com/data/nickel/mp159.html. [81] “Uddeholm Hotvar,” 2011.

77