Application and Optimization of Friction Stir Welding on ... · Application and Optimization of Friction Stir Welding on Electrical Transformers Components ... II. Key-Words FSW,

Application and Optimization of Friction Stir Welding

on Electrical Transformers Components

João Filipe Gomes Duarte Prior

Thesis to obtain the Master of Science Degree in

Materials Engineering

Supervisor: Professora Luísa Coutinho

Examination Committee

Chairperson: Professora Mª Fátima Vaz

Supervisor: Professora Luísa Coutinho

Members: Professora Rosa Miranda

Professor Rogério Colaço

Engenheiro Joel Mendes

March 2015

A toda a minha família e à Susana,

Por todo o apoio e paciência.

i

I. Abstract

This work intends to assist the industrial implementation of Friction Stir Welding (FSW)

process in components of electric power transformers. A methodology based on Taguchi method was

used to estimate the optimal parameters of butt welds in thin sheets of commercially pure aluminum,

AA1070, 1.6mm thick and 1.1mm thick, C11000, copper alloy. For this study three levels of the

parameters were considered: Axial Force (Fz), Travel Speed (Vx), and Probe Length (Lpin). The

optimum parameters were obtained through an analysis of variance (ANOVA) on three factors of

overall efficiency. GET, GEB and HARD coefficients were reached based on the results of tensile,

bending and hardness, respectively. Were also tested solutions for dissimilar welds with visually

satisfactory results. A preliminary feasibility study was made for the implementation of the process,

which shows a payback period of less than five years. Thus, it was concluded that the FSW process is

perfectly suited to the reality of SIEMENS FS because it allows significant improvements when

compared to the current process, Tungsten Inert Gas (TIG). This will lead to improvements on weld

quality, cost reduction and improved working environment.

II. Key-Words

FSW, Taguchi, Aluminum, Copper, Dissimilar Welding and Feasibility Study.

ii

III. Resumo

O presente trabalho foi desenvolvido com o intuito da aplicação industrial do processo de

soldadura por fricção linear (SFL) em componentes de transformadores eléctricos de potência. Assim,

de forma a estimar os parâmetros óptimos de soldaduras topo-a-topo em chapas finas de alumínio

(AA1070) e cobre (C11000) de 1.6mm e 1.1mm de espessura respectivamente, foi desenvolvido um

estudo com base no método de Taguchi. Para este estudo foram escolhidos três níveis diferentes

para os parâmetros de soldadura: força axial (Fz), velocidade de avanço (Vx) e comprimento do pino

(Lpin). Os parâmetros óptimos foram obtidos através da análise de variância (ANOVA) de três

factores de eficiência global, GET, GEB e HARD, desenvolvidos com base em resultados de tracção,

flexão e dureza respectivamente. Investigou-se ainda a ligação de materiais e geometrias

dissimilares, tendo sido realizadas soldaduras com características visualmente satisfatórias. Por

último foi realizado um estudo preliminar de viabilidade económica para a implementação do

processo, cujo período de retorno seria inferior a cinco anos. Conclui-se portanto que a SFL é um

processo perfeitamente adequado para a realidade da fábrica SIEMENS, pois permite melhorias

significativas em relação ao processo actual (TIG), nomeadamente na qualidade das soldaduras, no

custo unitário por soldadura e nas condições de segurança de trabalho dos operadores.

IV. Palavras-Chave

SFL, Taguchi, Aluminio, Cobre, Soldadura Dissimilar e Viabilidade Económica.

iii

V. Acknowledgements

I would like to express my deep gratitude to my supervisor, Professor Luísa Coutinho for all

commitment and personal interest in this dissertation.

A sincerely thank you to Professor Pedro Vilaça for inviting me to join SIEMENStir, such a

challenging and promising project. Thanks for all the trust and total freedom to develop my work and

all the technical support given.

Big thanks to my co-supervisor, Engineer Joel Mendes, for all the help, effort and interest. I

acknowledge the warm welcome within SIEMENS-FS and all the time spent for monitoring and

supervising the work.

My thanks to SIEMENS-FS for the confidence placed in me throughout the work. The material

and the funds invested in the project. I have to express my gratitude to Engineers António Silva and

Eugénio Santis.

I thank Professor Beatriz Silva for the help and time spent in the performance and

interpretation of the uniaxial tensile tests.

My thanks to Professor Rosa Miranda for the help and equipment provided to the hardness

tests, as well as to Professor Telmo Santos for supporting the conductivity tests.

I would like to thank Doctors João Gandra and Filipe Nascimento for the fellowship and

friendship, the training and help they always gave to me. I will never forget all the advices they gave to

me, in the many times I went desperate in their office. Thanks to Master André Oliveira for the

company and friendship in long hours writing and testing.

A special thanks to Mr. Daniel Pomiel and Mr. Carlos Farinha for all the assistance and

knowledge transmitted.

I also express my grateful to Mr. João Luís by the excellent work in the production of work

tools and other components designed, as well as to Mr. Lopes for machining of test specimens.

To my colleagues Nuno Ferreira, Daniel Pimentel, Mirela Lourenço, Tiago Gomes, Filipa

Baltazar, Francisco Sá, Bernardo Dias Miguel, Lucas Niven, João Nicolau, Teresa Gouveia e Tiago

Soares among many others, I express my deep regard for the strong friendship that has developed

throughout the course and systematic support during this work. Special thanks to Jacob Francisco for

reviewing and correcting the English.

iv

VI. Agradecimentos

Venho por este meio expressar a minha profunda gratidão à minha orientadora, Professora

Luísa Coutinho por todo o seu apoio e interesse na concepção desta dissertação.

Um sincero obrigado ao Professor Pedro Vilaça por me ter convidado para integrar um

projecto tão desafiante e promissor como o SIEMENStir, por toda a confiança que depositou em mim,

conferindo-me total liberdade para desenvolver o meu trabalho e por todo o apoio técnico dado.

Um enorme agradecimento ao meu co-orientador, Engenheiro Joel Mendes, por toda a sua

ajuda, empenho e interesse. Não posso deixar de agradecer a forma como me recebeu e me integrou

na SIEMENS-FS e por todo o tempo despendido para o acompanhamento e orientação do trabalho.

O meu muito obrigado à SIEMENS-FS pela confiança em mim depositada durante todo o

trabalho, pelo material cedido e pelos fundos investidos no projecto. Não posso deixar expressar a

minha gratidão aos Engenheiros António Silva e Eugénio Santis.

Agradeço à Professora Beatriz Silva pela ajuda e tempo despendido na realização e

interpretação dos ensaios de tracção uniaxial.

O meu muito obrigado à Professora Rosa Miranda pela ajuda e equipamento cedido para a

realização dos ensaios de dureza, assim como ao Professor Telmo Santos pelo apoio aos ensaios de

condutividade.

Gostaria de agradecer aos Doutores Filipe Nascimento e João Gandra pela camaradagem e

amizade com que sempre me trataram, pela formação e ajuda que me deram. Nunca esquecerei os

conselhos que me dirigiram nos muitos momentos em que entrei desesperado no gabinete deles. Ao

Mestre André Oliveira pela companhia e amizade em longas horas de escrita e ensaios.

Aos Srs. Daniel Pomiel e Carlos Farinha um especial obrigado por toda a assistência e

conhecimentos transmitidos.

Expresso também o meu apreço pelo Sr. João Luís por um excelente trabalho de produção

das ferramentas e de outros componentes projectados, assim como ao Sr. Lopes pela maquinação

de provetes.

Aos meus colegas Nuno Ferreira, Daniel Pimentel, Mirela Lourenço, Tiago Gomes, Filipa

Baltazar, Francisco Sá, Bernardo Dias Miguel, Lucas Niven, João Nicolau, Teresa Gouveia e Tiago

Soares entre muitos outros, expresso a minha profunda consideração pela forte amizade que se

desenvolveu ao longo do curso e todo o apoio sistemático durante a realização deste trabalho. Um

agradecimento especial ao Francisco Jacob pela revisão e correcção do Inglês.

v

VII. Contents

I. Abstract ..................................................................................................................................i

II. Key-Words .............................................................................................................................i

III. Resumo ................................................................................................................................ ii

IV. Palavras-Chave .................................................................................................................... ii

V. Acknowledgements ............................................................................................................. iii

VI. Agradecimentos ................................................................................................................... iv

VII. Contents ...............................................................................................................................v

VIII. List of Tables ..................................................................................................................... viii

IX. List of Figures .................................................................................................................... viii

X. List of Equations .................................................................................................................. ix

XI. Nomenclature .......................................................................................................................x

XII. List of Symbols .................................................................................................................... xi

1. Introduction .......................................................................................................................... 1

1.1. Scope ........................................................................................................................... 1

1.2. Problem Statement and Research Questions ............................................................. 1

1.3. Objectives .................................................................................................................... 2

1.4. Dissertation Structure .................................................................................................. 2

2. State of the Art ..................................................................................................................... 3

2.1. Introduction to Electrical Transformers ........................................................................ 3

2.1.1. Aluminum-Copper Comparison ............................................................................... 5

2.2. FSW ............................................................................................................................. 6

2.2.1. Basic concepts of the FSW process ....................................................................... 6

2.2.2. Parameters of the process ...................................................................................... 7

2.2.3. Microstructure obtained ........................................................................................... 8

2.2.4. Advantages and limitations of FSW ........................................................................ 9

2.3. Aluminum and its alloys ............................................................................................. 10

2.3.1. Properties and applications ................................................................................... 10

2.3.2. Alloys and temper designation .............................................................................. 10

2.3.3. Aluminum weldability ............................................................................................. 12

2.4. Copper and its Alloys ................................................................................................. 13

2.4.1. Properties and applications ................................................................................... 13

2.4.2. Alloys designation ................................................................................................. 13

2.4.3. Copper weldability ................................................................................................. 14

2.5. Statistic Method – Taguchi Method ........................................................................... 15

2.5.1. Methodology .......................................................................................................... 15

2.5.2. ANOVA .................................................................................................................. 17

2.6. Friction Stir Welding development on the study area ................................................ 18

2.6.1. FSW on thin sheets .............................................................................................. 18

2.6.2. Dissimilar Al/Cu welds .......................................................................................... 19

2.6.3. Taguchi on FSW ................................................................................................... 19

3. Equipment Characterization .............................................................................................. 21

3.1. Esab LegioTM

FSW 3U ............................................................................................... 21

3.2. FSW Tools ................................................................................................................. 23

4. Tests Characterization ....................................................................................................... 25

4.1. Tensile Test ............................................................................................................... 25

4.2. Bending Test .............................................................................................................. 27

vi

4.3. Micro-hardness .......................................................................................................... 29

5. Base Material Characterization ......................................................................................... 31

5.1. Aluminum Foil Characterization ................................................................................. 31

5.2. Copper Foil Characterization ..................................................................................... 33

6. Characterization of Taguchi on FSW................................................................................. 35

6.1. Evaluation Factors ..................................................................................................... 35

6.2. Control Parameters and their Levels ......................................................................... 36

6.3. Design of Experiments (DOE) ................................................................................... 36

7. Experimental Study Cases ................................................................................................ 37

7.1. Aluminum Butt Welding ............................................................................................. 37

7.1.1. Experimental Setup ............................................................................................... 37

7.1.2. Tool Geometry....................................................................................................... 37

7.1.3. Parameters ............................................................................................................ 38

7.1.4. Tensile Tests Results ............................................................................................ 40

7.1.5. Bending Tests Results .......................................................................................... 41

7.1.6. Hardness Tests Results ........................................................................................ 42

7.1.7. Analysis of Variance (ANOVA).............................................................................. 43

7.1.8. Optimum parameters identification ....................................................................... 44

7.1.9. Aluminum Butt Welding Results ............................................................................ 44

7.2. Copper Butt Welding ................................................................................................. 46

7.2.1. Experimental Setup ............................................................................................... 46

7.2.2. Tool Geometry....................................................................................................... 46

7.2.3. Parameters ............................................................................................................ 46

7.2.4. Tensile Tests Results ............................................................................................ 48

7.2.5. Bending Tests Results .......................................................................................... 49

7.2.6. Hardness Tests Results ........................................................................................ 50

7.2.7. Analysis of Variance .............................................................................................. 50

7.2.8. Optimum parameters identification ....................................................................... 51

7.2.9. Copper Butt Welding Results ................................................................................ 52

7.3. Overlap Foil-Bar weld ................................................................................................ 53

7.3.1. Aluminum Foil – Aluminum Bar Weld .................................................................. 54

7.3.2. Copper Foil – Copper Bar Weld ............................................................................ 54

7.3.3. Aluminum Foil – Copper Bar weld......................................................................... 55

7.3.4. Summary of Results .............................................................................................. 56

7.4. Other Geometries ...................................................................................................... 57

7.4.1. Aluminum-Copper Butt welding............................................................................. 57

7.4.2. Aluminum-Copper overlap welding ....................................................................... 57

7.4.3. Thin Copper-Copper butt weld ............................................................................. 57

8. Preliminary Feasibility Study ............................................................................................. 59

8.1. The Client Needs ....................................................................................................... 59

8.2. Operating costs.......................................................................................................... 60

8.3. Initial Investment and Payback .................................................................................. 60

8.4. Quality ........................................................................................................................ 62

8.5. Conclusions on Feasibility Study ............................................................................... 62

9. Conclusions ....................................................................................................................... 63

10. Future Work ....................................................................................................................... 65

10.1. Dissimilar butt welding ........................................................................................... 65

10.2. Foil-Bar Quality Tests ............................................................................................ 65

vii

10.3. Static Shoulder/Pinless Tool .................................................................................. 65

XIII. References ........................................................................................................................ 67

XIV. Annexes ............................................................................................................................... a

A. Experimental Procedures ................................................................................................ b

A1. Friction Stir Welding Procedures. ............................................................................ b

A2. Procedures for Metallographic Analysis. ..................................................................c

A3. Hardness Tests Procedures. ................................................................................... d

A4. Tensile Tests Procedures ........................................................................................ d

A5. Procedures for three point Bending Test. ................................................................ e

B. Results .............................................................................................................................. f

B1. Aluminum butt welds................................................................................................. f

B2. Copper Butt Welds ................................................................................................... h

C. Specimen Design...............................................................................................................j

C1. Tensile test specimen design ....................................................................................j

C2. Bending test specimen design ...................................................................................j

D. Bending structure...........................................................................................................j

E. Technical Sheets ..............................................................................................................k

E1. Support Table ............................................................................................................l

E2. Work Table ............................................................................................................. m

E3. Tool Body ................................................................................................................. n

E4. Probe – 4J3 ............................................................................................................. o

E5. Probe – 4I3 .............................................................................................................. o

E6. Shoulder 4P3 ........................................................................................................... p

E7. Shoulder 4O3 ........................................................................................................... q

F. Feasibility Study Calculation ............................................................................................. r

F1. TIG Cost ................................................................................................................... r

F2. FSW Cost .................................................................................................................s

G. Confidential Experiment Tables ................................... Error! Bookmark not defined.

G1. Aluminum Butt Welding ........................................... Error! Bookmark not defined.

G2. Copper Butt Welding................................................ Error! Bookmark not defined.

H. Confidential optimum parameters identification .......... Error! Bookmark not defined.

H1. Aluminum optimum parameters ............................... Error! Bookmark not defined.

H2. Copper optimum parameters ................................... Error! Bookmark not defined.

viii

VIII. List of Tables

Table 2.1 – Composition of the different series of wrought aluminum alloys. .......................... 11

Table 2.2 – Composition cast aluminum alloys series. ............................................................ 11

Table 2.3 – Specification for Cold Work alloys. ........................................................................ 12

Table 2.4 – Heat Treatment designation. ................................................................................. 12

Table 2.5 – Composition of the different families of wrought copper alloys. ............................ 14

Table 5.1 – Tensile tests results for 3 specimens of Aluminum Base Material. ....................... 32

Table 5.2 – Bending tests results for 3 specimens of Aluminum Base Material. ..................... 32

Table 5.3 – Tensile tests results for 2 specimens of Copper Base Material. ........................... 33

Table 5.4 – Bending tests results for 3 specimens of Copper Base Material. .......................... 33

Table 6.1 – GET weight for each property of tensile test. ........................................................ 35

Table 6.2 – GEB weight for each property of bending test. ...................................................... 36

Table 6.3 – Control Parameters and their Levels. .................................................................... 36

Table 6.4 – Taguchi L9 Orthogonal Array with 3 columns. ...................................................... 36

Table 7.1 – Visual analysis of aluminum welds. ....................................................................... 39

Table 7.2 – Summary of results for tensile tests of aluminum.................................................. 40

Table 7.3 – Summary of results for bending tests of aluminum. .............................................. 41

Table 7.4 – Summary of results for hardness tests of aluminum. ............................................ 42

Table 7.5 – Results of variance analysis for the three evaluation parameters for aluminum. .. 43

Table 7.6 – Visual analysis of copper welds. ........................................................................... 47

Table 7.7 – Summary of results for tensile tests of copper. ..................................................... 48

Table 7.8 – Summary of results for bending tests of copper .................................................... 49

Table 7.9 – Summary of results for hardness tests of copper .................................................. 50

Table 7.10 – Results of variance analysis for the three evaluation parameters for copper. .... 51

Table 7.11 – Aluminum Foil-Bar ............................................................................................... 54

Table 7.12 – Copper Foil-Bar weld ........................................................................................... 55

Table 7.13 – Cu-Al-Cu Sandwich like weld parameters range. ................................................ 56

Table 8.1 – Production variables for the winding manufacturing. ............................................ 59

Table 8.2 – Estimated costs per weld for both processes. ....................................................... 60

Table 8.3 – Estimated costs for the project. ............................................................................. 61

Table 8.4 – Comparison of non-quantifiable variables of both processes. .............................. 62

IX. List of Figures Figure 2.1 – Transformer principle basic scheme. Adapted from [5]. ........................................ 3

Figure 2.2 – Ideal Transformer equivalent circuit. ...................................................................... 4

Figure 2.3 – Different Winding assembly processes .................................................................. 4

Figure 2.4 – FSW Process Scheme adapted from Vilaça et al. [17]. ......................................... 6

Figure 2.5 – FSW hot and cold condition classification [15]. ...................................................... 8

Figure 2.6 – Typical macrograph scheme of a section transversal to the FSW direction [19]. .. 8

Figure 2.7 – Taguchi Method Flow Chart ................................................................................. 16

Figure 3.1 – Welding equipment LEGIOTM

FSW 3U of ESAB. ................................................. 21

Figure 3.2 – Representation of the different constituents of the equipment. ............................ 21

Figure 3.3 – Work table and fixing system used. ..................................................................... 22

Figure 3.4 – 3-D view of the iSTIRtool_v3 assembly. .............................................................. 23

Figure 3.5 – Different M4 shoulder geometries. ....................................................................... 24

Figure 3.6 – M4 Probes ............................................................................................................ 24

Figure 4.1 – Representation of the cuts made on butt welded sheets ..................................... 25

Figure 4.2 – Tensile testing machine, Instrom 4507, overall view at left, testing zone at right. 25

ix

Figure 4.3 – Magnification of Tensile test results of AA1070, BM1 trial, near elastic regime. . 26

Figure 4.4 – Tensile test results of AA1070, BM1 trial. ............................................................ 27

Figure 4.5 –Three Point Bending example of aluminum weld specimen. ................................ 27

Figure 4.6 – Representation of bending test setup for aluminum. ........................................... 28

Figure 4.7 – Bending test result for copper C11000. ................................................................ 28

Figure 4.8 – Mitutoyo HM-112 Vickers micro-hardness testing machine ................................. 29

Figure 5.1 – Aluminum foil, as received. .................................................................................. 31

Figure 5.2 – AA1070 composition certification. ........................................................................ 32

Figure 5.3 – Aluminum mechanical properties certification. ..................................................... 32

Figure 5.4 – Copper Foil, as received. ..................................................................................... 33

Figure 5.5 – Chemical composition certification for C11000. ................................................... 33

Figure 7.1 – Experimental setup for aluminum welds. ............................................................. 37

Figure 7.2 – Probe 4J3 and shoulder geometry 4P3 used in aluminum butt weld. .................. 37

Figure 7.3 – Overall look of the face and root sides of the nine trials of aluminum. ................ 38

Figure 7.4 – GET results for each trial (left) and parameter level (right). ................................. 40

Figure 7.5 – GEB results for each trial (left) and parameters level (right). ............................... 41

Figure 7.6 – Hardness coefficient results for each trial (left) and parameters level (right). ...... 42

Figure 7.7 – Average values of the nine hardness profiles of Aluminum welds. ...................... 43

Figure 7.8 – Contribution of each parameter for the three evaluation parameters. ................. 44

Figure 7.9 – Metallographic analysis of the confirmatory trial. ................................................. 45

Figure 7.10 – Eddy current conductivity test performed, at 250 kHz, ...................................... 45

Figure 7.11 – Probe 4J3 and shoulder 4O3 used in copper butt weld. .................................... 46

Figure 7.12 – Overall look of the face (left) and root (right) sides of the nine trials of copper. 47

Figure 7.13 – GET results for each trial (left) and parameters level (right). ............................. 48

Figure 7.14 – GEB results for each trial (left) and parameters level (right). ............................. 49

Figure 7.15 – Hardness Coefficient results for each trial (left) and parameters level (right). ... 50

Figure 7.16 – Contribution of each parameter for the three evaluation parameters. ............... 51

Figure 7.17 – Eddy current conductivity test performed, .......................................................... 52

Figure 7.18 – TIG Foil-Bar weld in aluminum at left, and in copper at right. ............................ 53

Figure 7.19 – Aluminum Foil-Bar weld example. ...................................................................... 54

Figure 7.20 – Copper Foil-Bar weld example. .......................................................................... 55

Figure 7.21 – Aluminum foil – copper bar weld instabilities. .................................................... 55

Figure 7.22 – Cu-Al-Cu Sandwich like weld example. ............................................................. 56

Figure 7.23 – Aluminum-Copper butt weld instabilities. ........................................................... 57

Figure 7.24 – Aluminum-Copper overlap weld example. ......................................................... 57

Figure 7.25 – Examples of thin copper trials. ........................................................................... 57

Figure 8.1 – Costs division for TIG and FSW weld. ................................................................. 60

Figure 8.2 – Estimated Payback Period. .................................................................................. 61

X. List of Equations (2.1) ............................................................................................................................................. 4

(2.2) ............................................................................................................................................. 5

(2.3) ............................................................................................................................................. 5

(2.4) ............................................................................................................................................. 5

(2.5) ............................................................................................................................................. 5

(2.6) ............................................................................................................................................. 5

(2.7) ............................................................................................................................................. 7

(2.8) ........................................................................................................................................... 17

(2.9) ........................................................................................................................................... 17

x

(2.10)......................................................................................................................................... 17

(2.11)......................................................................................................................................... 17

(2.12)......................................................................................................................................... 17

(4.1) ........................................................................................................................................... 26

(4.2) ........................................................................................................................................... 26

(6.1) ........................................................................................................................................... 35

(6.2) ........................................................................................................................................... 35

(6.3) ........................................................................................................................................... 36

(7.1) ........................................................................................................................................... 44

(7.2) ........................................................................................................................................... 44

(7.3) ........................................................................................................................................... 51

(7.4) ........................................................................................................................................... 51

XI. Nomenclature AA Aluminum Alloy

AC Alternating Current

ANOVA Analysis of Variance

BM Base Material

DC Direct Current

DOF Degrees Of Freedom

EN European Standard

FS Friction Surfacing

FSSW Friction Stir Spot Welding

FSW Friction Stir Welding

GEB Global Efficiency on Bending

GET Global Efficiency on Tensile

GOES Grain Oriented Electrical Steel

HARD Hardness Coefficient

HAZ Heat Affected Zone

HV Vickers Pyramid Number

IADS International Alloy Designation System

IDMEC Institute of Mechanical Engineering

IST Instituto Superior Técnico

iStir Friction Stir Investigation Group from IDMEC-IST

J Joule

kN kilo Newton

kV kilo Volt

LDT Large Distribution Transformer

LOP Lack Of complete Penetration of the weld seam in the weld joint thickness

MIG Metal Inert Gas

min Minutes

mm millimeters

OA Orthogonal Array

Pa Pascal

Qty Quantity

SIEMENS-FS SIEMENS Transformers Factory of Sabugo

SIEMENStir Investigation project between SIEMENS-FS and iStir

TMAZ Thermo-mechanically Affected Zone

TWI The Welding Institute

UNS Unified Numbering System

UTL Universidade Técnica de Lisboa

xi

XII. List of Symbols

𝛼 Tilt angle

𝛽 Side tilt angle

𝜀 Elongation

𝜌 Resistivity

𝜌𝑖(𝑗) Influence of parameter i on factor j

0.2 Offset yield strength

𝑚𝑎𝑥 Ultimate tensile strength

𝛺 Rotational speed or ohm

Ag Silver

Al Aluminum

Cr Chromium

Cu Copper

d Displacement

Dt Dwell time

DT Distribution Transformer

E Young’s modulus

En Energy

F Load

Fe Iron

Fi(j) F-test of parameter i on factor j

Fz Axial force

HCl Hydrogen chloride

HF Hydrogen fluoride

HNO3 Nitric acid

I Current

Li Lithium

Lpin Probe length

Mg Magnesium

Mn Manganese

MT Tenacity modulus

Ni Nickel

P Phosphorus

Pb Lead

Ps Plunge speed

PT Power Transformer

R Electric resistance

Si Silicon

Sn Tin

SSi(j) Sum of Squares of parameter i on factor j

SST(j) Total Sum of Squares on factor j

V Potential difference

Vi(j) Variance of parameter i on factor j

Vx Travel speed

Wp Weld position

Zn Zinc

xii

1

1. Introduction

1.1. Scope

SIEMENStir is a project between SIEMENS Transformers Factory, specialized in oil immersed

transformers, and iStir, a research group from IDMEC in IST. A first study was developed focusing on

the weldability of aluminum foil for electrical purposes [1]. This project aimed the joining of foil coils, a

raw component for LDT transformer windings. In this first study it was acknowledged that Friction Stir

Welding (FSW) could be a significant addition and leverage for the Factory (SIEMENS-FS) as an

upgrade of welding possibilities, for joining thin Aluminum sheets in butt welds. The higher quality of

FSW welding comparatively to the current arc welding process is also a significant improvement, from

the mechanical and electrical point of view.

The present work attempts to extend and strengthen the knowledge of friction stir welding

technique for electrical applications, using the core of the first SIEMENStir project as starting point.

Even though FSW welding technique is well known by iStir, the present application places new

challenges, conductors’ materials, such as thin sheets of pure Al and pure Cu, which were never tried

before in the group, and place particular constrains to the use of the process.

Recent demands for replacing the technically pure copper as conductive materials by new

materials, e.g. aluminium, introduced new manufacturing conditions that need feasible and reliable

new technological solutions focusing joining of components. The new solutions should allow an easy

integration in the existing production system and envisage higher productivity. Also dedicated non-

destructive testing techniques and solutions should be developed. Existing joining processes applied

in the manufacturing of the electrical transformers are based on fusion arc welding processes, mainly

GTAW. This results frequently in defective weld beads with porosity and hot cracking. The residual

deformations should also be reduced.

The Factory is also interested in evaluating the applicability of FSW from the technological and

economical point of view. Bearing in mind those goals, FSW was tested in several arrangements and

different materials, of aluminum and copper alloys. And a comparative preliminary feasibility study

between the current process and the FSW process was performed envisioning its possible

implementation.

1.2. Problem Statement and Research Questions

The search for new markets and opportunities require the use of the best possible solutions

alongside with innovative alternatives to develop products that stand out in the current market. The

current welding process, Tungsten Inert Gas (TIG), does not have this potential. TIG welding cannot

butt weld neither aluminum nor copper sheets because of their reduced thicknesses. The overlap

solution, used only for copper sheet, leads to both an increase in the size of the transformer and in the

2

consumption of material. TIG’s inability to weld dissimilar materials, due to the significant differences

of their chemical properties, is also a big disadvantage.

FSW seems the perfect replacing process for this application, if is found answer for some

questions. First of all, there is the need to prove that it’s possible to butt weld such pure alloys in such

thin sheets as the used 1.6mm thick AA1070 and 1.1mm thick C11000. Dissimilar aluminum-copper

butt weld has been done before, but, in order to enable the flow of the same current in both materials,

to avoid hotspots and material waste, the thickness of the two materials is different. Which make this a

very difficult and doubtful solution of thin sheet butt weld of dissimilar materials with dissimilar

thicknesses. The foil-end bar weld also raised some doubts, due to the slight discrepancy of

mechanical properties between the end-bar and foil, and it was uncertain if the former would support

the load necessary for a satisfactory weld.

1.3. Objectives

Evaluation of the FSW weldability on different materials and geometries to increase the ranges

of SIEMENS-FS applications and opportunities. Study the possibility of welding dissimilar foil butt

joints and foil-end bar overlapped joints. Finding the optimal parameters, through a Taguchi study, for

butt welding of Al and Cu thin sheets, in order to achieve and transfer the correct know-how and

technology to the factory. Analyzing the financial impact of the acquisition and implementation of FSW

equipment for SIEMENS FS based on a preliminary feasibility study.

1.4. Dissertation Structure

This section explains the structure of this essay, by addressing the main contents of each

chapter and their sections: The State of the Art, Chapter 2, is divided into six sections: 1) Introduction

to Electrical Transformers; which gives a quick overall view of transformers principals and their

features. 2) FSW; it explains the fundaments behind the process, such as its parameters,

microstructure obtained and the advantages and limitations. Sections 3) and 4) intend to be a general

guide of the materials used during the procedures, reviewing the main properties and applications.

They also refer to the alloy designations and weldability. 5) Statistic Method; which explains the

methodology used to the experimental study of butt welds. Finally, section 6) refers scientific work

developed in similar materials, technology and trials. Chapter 3 introduces the FSW equipment used

during the experimental work. Chapter 4 explains the mechanical tests performed. Chapter 5 presents

the base material characterization, in which are presented the geometry, microstructure and

mechanical properties of the material as received. Chapter 6 explain the methodology adopted for butt

weld tests. The evaluation factors, the control parameters and the design of experiments. Chapter 7

includes the experimental study case and it’s divided in four sections: Sections 1) and 2) address the

experimental work, results and variance analysis of both aluminum and copper butt welds. Section 3)

and 4) show foil-bar and dissimilar joints and the set of parameters used. Chapter 8 focuses on the

feasibility study to implement the process. Chapter 9 discusses the major results obtained in the thesis

and Chapter 10 describes the future work proposal.

3

2. State of the Art

The State of the Art addresses the main elements of this work, and the different approaches

done by several authors to the themes.

2.1. Introduction to Electrical Transformers

Transformers are important devices for the electric energy transmission and distribution

through the grid. These devices transfer energy from one circuit to another by a common magnetic

field [2]. Electric power transmission at long distances cannot be made at low voltage, because it has

great losses. On the other hand, the power generation and consumption can’t happen at high voltages

since they need pronounced insulation. For that reason step-up transformers are needed to raise the

voltage after the generation for transmission and then step-down transformers, named Distribution

Transformers (DT), take the voltage to an appropriate distribution level. The power generation is

usually obtained in a range from 11 to 33kV, then it is supplied to the transport network at 150, 220 or

400 kV[3]. At last, the consumption occurs at 230/420V as mono or tri phase voltage respectively.

The two principles in which the transformers rely on are electromagnetism and

electromagnetic induction. This is the capability of an electric current to produce a magnetic field, and

the varying magnetic field within a coil that induces a voltage across the extremities of that coil. This

induction only occurs in Alternating Current (AC) resulting from the reversal of electric flow.

Therefore, a transformer works in the following principle, an AC current is applied by an

energy source to the primary winding that induces the current to the secondary winding that will supply

the energy to the user [4]. Figure 2.1 shows a simple arrangement of this principle, the majority of the

transformers work in three-phase, so they have 3 sets of 2 bobbins each that are lagged 120º from

each other.

Figure 2.1 – Transformer principle basic scheme. Adapted from [5].

4

The relationship between the number of turns (n), the current intensity (I) and the voltage (V)

of primary (p) and secondary coils (s) in an ideal transformer is shown in Figure 2.2 and is given by

equation (2.1):

𝒏𝒑

𝒏𝒔=𝐈𝒔𝑰𝒑=𝐕𝒑

𝐕𝒔 (2.1)

In order to sustain the magnetic flux inside the coils high permeability silicon steel and/or grain

oriented electrical steel (GOES) is used as core. There are two main phenomena that reduce the

magnetic flux, and those are eddy current losses and hysteresis-losses, these are reduced by the

geometry of the core, generally, laminated steel of small thickness with insulated layers of

magnesium-silicate phosphate.

Generally HV windings are made of resin insulated wire, as shown in Figure 2.3 a). In

Distribution Transformers LV windings are usually assembled by concentric foils one above the other

and insulated by paper, Figure 2.3b). The joint between Foil windings and the connection bars is one

of the aims of this work.

Figure 2.2 – Ideal Transformer equivalent circuit.

Figure 2.3 – Different Winding assembly processes

a) wired winding made in copper b) foil winding in aluminum c) final assembly of an aluminum bobbin.

5

2.1.1. Aluminum-Copper Comparison

In the past years, aluminum has emerged in the electric conductors industry. The main reason

is the obvious economic factor because aluminum is between two to three times cheaper than copper.

However electric conductivity of aluminum is smaller, so to maintain the current density and voltage

drop the equivalent aluminum wire needs to be dimensioned.

Ohm’s Law states that in an electric circuit [5], the current (𝐼) through a conductor between

two points is directly proportional to the potential difference (𝑉) across the two points, and the constant

of proportionality is the electric resistance (𝑅), equation (2.2):

𝑰 =𝑽

𝑹 (2.2)

As the current and the potential difference is the same to both conductors, assumption can be

made that the copper resistance must be equal to aluminum resistance. Resistance can be written,

equation (2.3), in function of conductor material resistivity (𝜌), length (𝐿) and section (𝑆).

𝝆𝑪𝒖𝑳

𝑺𝑪𝒖=𝝆𝑨𝒍𝑳

𝑺𝑨𝒍 (2.3)

The relatively higher resistivity of commercial pure Aluminum (2.82x10-8

Ω.m) comparing to

annealed Copper (1.72x10-8

Ω.m) leads to a 1.64 times higher section for the same conductor length.

Then, for the same current density and voltage drop at a given length, copper wiring provides a

smaller nominal section leading to smaller conductor, which can be an advantage for certain

applications.

However for the majority of applications the weight factor is more significant. Assuming that

each wire has the shape of a perfect cylinder with cross section equal to 𝑆𝐶𝑢 and 𝑆𝐴𝑙, respectively for

Copper and Aluminum, the relationship between both is given by the equation (2.4).

𝑽𝑨𝒍 = 𝑺𝑨𝒍. 𝑳 = 𝟏. 𝟔𝟒𝑺𝒄𝒖. 𝑳 = 𝟏. 𝟔𝟒𝑽𝑪𝒖 (2.4)

Equation (2.5) expresses the relationship between the mass of both materials in function of

their density and respective volume:

𝒎𝑨𝒍

𝒎𝑪𝒖=𝒅𝒆𝒏𝒔𝑨𝒍

𝒅𝒆𝒏𝒔𝑪𝒖×𝑽𝑨𝒍

𝑽𝑪𝒖 (2.5)

Finally, using the known values of density for copper (8.89g/cm3) and for aluminum

(2.7g/cm3) the relation of mass is obtained.

𝑚𝐴𝑙 = 1.64 2.7

8.89= 0.498 𝑚𝐶𝑢 (2.6)

To sum up, for the same current and potential difference at a certain length, although

aluminum conductors exhibit bigger section and volume (164%), its mass and consequently its weight

is halved (49.8%) when compared to copper conductors, as shown in equation (2.6). Even considering

the lower resistance of aluminium to thermal, chemical and mechanical loadings, and all the issues

related with joining with terminals based on copper components. Aluminium is nowadays a significant

trend in the production of windings for transformers for power systems[6], [7] and opens the possibility

of developing new products and new processes.

6

2.2. FSW

2.2.1. Basic concepts of the FSW process

The Friction Stir Welding (FSW) process was firstly patented in 1991 by Wayne Thomas of

The Welding Institute (TWI) [8] and it might be the most important breakthrough of the last decade of

the XX century, regarding joining components’ technology. It has been through a rapid growth in

research, development and application since the early 2000s as shown in the 230 organizations

licensed by TWI whom submitted more than 3000 patent applications [9], in the numerous technic-

scientific published items and also in the several friction stir based technologies that are being

discovered and developed (Friction Stir Processing [10], Spot Weld [11], Channeling [12]). The FSW

technology has been subjected to the most demanding quality standard requirements and used in

challenging industrial applications over a wide range of structural and non-structural components

mainly in light alloys for transport industries. Such examples are naval [13], aerospace [14], railway

[15] and automotive [16] industries.

The FSW is a solid state joining process, which uses a non-consumable tool, made of a

material harder than the material being welded. Both the shoulder and the probe of the tool can be

designed depending on the thickness and type of material to be welded. The principles and the used

nomenclature of the process are shown in Figure 2.4.

Figure 2.4 – FSW Process Scheme adapted from Vilaça et al. [17].

The rotating tool is inserted (plunged) into the joint of two materials at a constant speed (Ps)

until it reaches the desired position (Wp), this position is critical to avoid root defects or thickness

reduction. After obtaining the appropriate thermal conditions by maintaining the rotation in the Wp

during the dwell time (Dt) the tool is animated of the linear movement (Vx). The weld control can

switch from the position control, which forces the tool to preserve the same height, to force control that

conserves the same axial force (Fz). At the end of the weld, the linear movement stops when the tool

reaches the final position and the tool is slowly removed, leaving a keyhole in the work piece. This is

one of the main drawbacks of the classic FSW process but, today, it can be avoided with a special

designed retreating probe [18].

7

The FSW process requires a tight constrain and this is obtained by the system, shoulder and

anvil, at top and bottom and by the cold base material at the sides. This is the well-known “third body

region”. The interfacial friction caused by the rotation and the axial force of the shoulder at the surface

of the material, origins an internal friction in a sub-superficial region inside the material, which is the

driving force of the process, the heat generated thermo mechanically softens the material and the tool

movement produces an extruded, forged and stirred weld. The material transportation is accomplished

because the probe splits the incoming flow of visco-plastic material that is then forge welded together

at the trailing edge of the probe. This process occurs continuously during the passage of the shoulder.

As generally happens in solid state weld processes the efficiency of FSW is high, because the

mechanical energy is converted in heat inside the work piece, as stated above, by the interaction

between the tool and the material. This heat is responsible for the reduction of material’s mechanical

resistance allowing this to flow around the pin.

2.2.2. Parameters of the process

The main FSW process parameters must be chosen according to the joint type, the materials

to weld and their geometry. Below, we found a brief description of the main parameters [19], followed

by their abbreviations (abr.) and currently used units [unit]:

• Tool geometry includes the probe and shoulder geometry, being the major concerns the

use of the adequate diameter and length accordingly to the thickness of the material;

• Plunge speed (Ps) [mm/s], the optimization of this parameter avoids increasingly the

appearance of defects in the starting phase of the process;

• Tool rotational speed (Ω) [rpm] and direction, rotational direction depends on the probe

geometry, left-handed screw must lead to a clockwise rotation and vice versa. Rotational speed is

generally associated with travel speed, together they define the hot-to-cold conditions that are

explained in the weld pitch ratio topic;

• Travel Speed (Vx) [mm/min] must be adjusted according to the superficial friction on the

material with the purpose of avoiding stick and slide movements;

• Axial force (Fz) [Kg] insufficient load result in poor weld conditions but excess axial force

increases the flash leading to a depression below the shoulder passage path;

• Dwell time (Dt) [s] is important to obtain the appropriate thermal conditions;

• Clamping system, material must be stiff enough to support, especially, the torsion forces

felt by the plates. It’s important that the system enables an easy extraction of the work piece;

• Tilt (α) and side tilt (β) angles, the first is very important to allow a good entrance of the

material below the shoulder and the second must be used in case of different thicknesses of the side

plates.

𝒘𝒆𝒍𝒅 𝒑𝒊𝒕𝒄𝒉 𝒓𝒂𝒕𝒊𝒐 [𝒓𝒆𝒗 𝒎𝒎⁄ ] = 𝜴 [𝒓𝒑𝒎]

𝒗[𝒎𝒎 𝒎𝒊𝒏]⁄ (2.7)

8

• Weld pitch ratio is a combination of two other parameters, see equation (2.7). Varying the

weld pitch ratio changes the heat input from the frictional internal and interfacial energy, being some

effects in the weld region behavior easily predicted. For aluminum alloys it is usual to consider the

value of weld pitch ratio = 4 as intermediate condition. For weld pitch ratio higher than 4 there is a hot

condition and for weld pitch ratio smaller than 4 there is a cold condition.[20] Concerning the influence

of the hot-to-cold conditions in the metallurgical features of a FSW weld joint, the classification is

established in Figure 2.5.

2.2.3. Microstructure obtained

During the FSW process the work piece material suffers intense plastic deformation resulting,

generally, on smaller and rounder grain size than in the base material. This result from the thermo-

mechanical cycle, a feature of the FSW, in which the solid state heat leads to a stired cooling that

consists essentially in forging. This fine microstructure produces good mechanical properties in friction

stir welds. Better quality joints are associated with intense three-dimensional material flow.

The main zones in a FSW joint, showed in Figure 2.6, with distinct metallurgical properties are:

i) the thermo-mechanically affected central zone (TMAZ) that includes the dynamically recrystallized

zone ii) the nugget; iii) the heat affected zone (HAZ) and iv) the unaffected base material (BM).These

different zones result from the combined application of mechanical energy and heat energy from

frictional dissipation.

The typical characteristics of each of these zones for aluminum alloys are the following [21]:

i. The TMAZ grain preserves the characteristics of the HAZ however the grain presents

increased deformation, as they get closer to the interface with the nugget. This fact

results from the influence of the material flow prescribed by the movement of the tool

and the relatively high maximum temperature reached in this zone;

Figure 2.5 – FSW hot and cold condition classification [15].

Figure 2.6 – Typical macrograph scheme of a section transversal to the FSW direction [19].

9

ii. The nugget is the region of the TMAZ undergoing dynamic recrystallization with grain

size refined and homogenized. The TMAZ/nugget interface enhances the significant

difference between the structure of initial grain and the equiaxial grain resultant of the

dynamical recrystallization process, with fine dispersion of the precipitates in the solid

solution. The asymmetric geometry of the nugget region is due to the difference in

generated heat between the advance and retreating sides, resulting of the direct

superposition of rotational speed, and travel speed;

iii. The HAZ is only affected by the heat energy and typically presents some slight

coalescence of grain relatively to the original grain size but is subjected to internal

point and linear defects rearrangements. Thus, for the heat treatable wrought

aluminum alloys the HAZ may present some reduction in the distribution of

precipitates at grain boundaries;

iv. The BM is the region that was unaffected by the FSW process.

2.2.4. Advantages and limitations of FSW

The advantages claimed for the process result essentially of being a solid-state process that

allows similar metallurgical characteristics and wear and static mechanical resistance to the base

material. This avoids the degradation of material proprieties due to the low heat input of the process,

resulting also in lower distortion and smaller residual stress levels. It’s very environmental and human

friendly, as it does not need welding consumables, operates with the absence of welding fumes or UV

emissions and the noise it produces is almost non-existent. The training of the machine operator isn’t

so complex when compared to fusion weld, which probably results in a decrease of training expenses.

Being the FSW easily reproduced, it’s very suitable for automation repeatability, leaving a good

surface appearance right after the processing. This avoids subsequent surface treatment processes. It

can weld virtually any thicknesses with only one path (depending only on the machine capacity), in all

positions and in continuous mode (unlimited length, again depending on the machine configuration).

It’s not influenced by magnetic forces neither the environmental conditions.

The current limitations of the FSW process are the need of a backing anvil (except in bobbin-

tool). It also leaves a keyhole at the end of each weld, except when a run-off tab is introduced or when

is used a FSW tool with a retractable probe. Other limitation is the inability to start the weld joint from

the edges of the plates to be welded, except when using a run-on tab. The work piece requires rigid

clamping. The cost and size of the equipment are also a major drawback, as it has significant initial

investment and its size prevents it from being easily transported.

10

2.3. Aluminum and its alloys

Aluminum is the most abundant metal in the crust of the Earth, however it is always combined

with other elements, such as iron, oxygen or silicon [22]. After the development of the reduction Hall-

Héroult electrolytic process, in 1886, aluminum became an economic competitor in engineering

applications [23].

2.3.1. Properties and applications

Aluminum is already used in several industries and their use is expected to increase mostly

because of the following proprieties:

-Low density, 2,68g/cm3, about 1/3 of the density of steel or copper, giving it a good ratio

mass/volume that ensures a better carriage in relation to others metals that may be used to make

packages. The low density allied with its mechanical strength makes it a material to be considered in

aerospace and automotive industry because it provides good performance and low fuel consumption.

-Good electric and thermal conductivity, which is very desirable in the Transmission of Power

Industry and useful for heat exchangers.

-Nice formability that allows an easy conformation and deformation and enables its usage on

several production processes.

-Good corrosion resistance, when passived by an oxide layer, being used in civil engineering

ensuring preservation and easy maintenance. It also allows the manufacturing of hygienic and

contamination free packages.

-High variety of finishing, which allows anodization and painting, enabling another layer of

protection and corrosion barrier.

-Recyclable, grants a reduction of cost not only in production but also to the environment as it

can be reused, recovering part of the production costs and reduces the use of raw materials. The

recycling costs are much smaller than the production costs, as it can be done several times without

the loss of properties.

2.3.2. Alloys and temper designation

Aluminum alloys can be separated in wrought and cast aluminum, being the first class

responsible for 85% of the world’s production. The two classes use different identification systems.

Wrought aluminum is identified based on a four-digit number, assigned by the International Alloy

Designation System (IADS), in which the first digit identifies the alloying elements as shown in Table

2.1. The second digit is related with the modifications to the alloy, with the original alloy having this

digit equals to 0. The last two digits stand for the purity of the alloy in the case of the series 1xxx, for

example the alloy 1050 having 99.50% aluminum, and the alloy 1199 having 99.99%. For the others

series these two digits are just used to identify different alloys. When the alloy is in research it is used

the prefix X for its identification [22].

11

Table 2.1 – Composition of the different series of wrought aluminum alloys.

Series Main alloying element Others elements

1xxx Pure Aluminum -

2xxx Cu Mg, Li

3xxx Mn Mg

4xxx Si -

5xxx Mg -

6xxx Mg and Si -

7xxx Zn Cu, Mg, Cr, Zr

8xxx Li, Sn, Fe, Cu, e Mn -

9xxx -

The Aluminum Association (AA) has adopted a nomenclature for cast alloys similar to that of

wrought alloys. In the AA system, the second two digits reveal the minimum percentage of aluminum.

The digit after the decimal point takes a value of 0 or 1, denoting casting and ingot respectively. The

main alloying elements in the AA system are those shown in Table 2.2:

Table 2.2 – Composition cast aluminum alloys series.

Series Main alloying element

1xx.x Pure Aluminum

2xx.x Cu

3xx.x Si, Cu, Mg

4xx.x Si

5xx.x Mg

7xx.x Zn

8xx.x Li

Aluminum alloys can also be divided in two groups according to the ability to be heat treated,

alloys from the series 2xxx, 6xxx, and 7xxx are generally heat treatable and those from series 1xxx,

3xxx, 4xxx and 5xxx are considered non-heat treatable. The temper designation [23] follows the cast

or wrought designation number with a dash, a letter, and potentially a one to three digits number, e.g.

6061-T6. When the alloy does not suffer any temper process it receives the letter F (as fabricated),

when it was annealed for softening the letter O, and when it as the letter W it means that it suffered

only natural solution heat treatment. The letter H is used to cold work treatments and is usually

followed by two digits, the first is related to the treatment used and the second to the thickness

reduction leading to a variation in the hardness. Those two digit specifications are summarized in the

Table 2.3.

12

Table 2.3 – Specification for Cold Work alloys.

1st

Digit Cold Work 2nd

Digit Degree of Hardness

H1x Strain hardened without

thermal treatment Hx2 ¼ Hard

H2x Strain hardened and partially annealed

Hx4 ½ Hard

H3x Strain hardened and stabilized

by low temperature heating

Hx6 ¾ Hard

Hx8 Full Hard

Hx9 Extra Hard

Finally there are the alloys that receive heat treatments to enhance their properties, and which

are classified by the letter T followed by one or more digits. The temper designation of the first digit is

shown in Table 2.4, the second and following digits can be related to a thickness reduction by cold

work, with the type of stress relief or following heat treatments.

Table 2.4 – Heat Treatment designation.

Type Heat treatment

T1 Cooled from hot working and naturally aged (at room temperature)

T2 Cooled from hot working, cold-worked, and naturally aged

T3 Solution heat treated and cold worked

T4 Solution heat treated and naturally aged

T5 Cooled from hot working and artificially aged (at elevated temperature)

T6 Solution heat treated and artificially aged

T7 Solution heat treated and stabilized

T8 Solution heat treated, cold worked, and artificially aged

T9 Solution heat treated, artificially aged, and cold worked

T10 Cooled from hot working, cold-worked, and artificially aged

2.3.3. Aluminum weldability

In the 40’s aluminum was seen as unable to weld, but since then the evolution of the fusion

weld processes changed that belief. With the appearance of the Tungsten Inert Gas (TIG) and Metal

Inert Gas (MIG) it was possible to do good weld beads, with fine mechanical properties and industrially

competitive.

The weldability of aluminum alloys varies significantly, depending on the chemical composition

of the alloy used. Aluminum alloys must also be cleaned prior to welding, with the goal of removing all

oxides, oils, and loose particles from the surface to be welded. This is especially important because of

aluminum weld’s susceptibility to porosity due to hydrogen and dross due to oxygen [24].

Aluminum alloys are susceptible of hot cracking, and to address the problem, welders

increase the welding speed to lower the heat input. Preheating reduces the temperature gradient

across the weld zone and thus helps reduce hot cracking, but it can reduce the mechanical properties

of the base material and should not be used when the base material is restrained. The design of the

joint can be changed as well, and a more compatible filler alloy can be selected to decrease the odds

for hot cracking occurring.

13

The emergence of Friction Stir Welding resulted in an aluminum weld with high efficiency

without the need of consumable material, neither harmful noise and fumes, and avoiding the

undesired porosity and distortion caused by fusion techniques. To sum up, FSW can obtain excellent

and clean weld beads with high repeatability, boosting the aluminum alloys in the Metalworking

Industry.

2.4. Copper and its Alloys

Copper and its alloys have been used for thousands of years, as it can be found in native form

copper being the first metal to be extracted by the men. Copper smelting led to the second Era of the

human prehistory, the Bronze Age. During Bronze Age besides bronze (copper and tin alloy) the

ancients used native copper or brass (copper and zinc alloy) to produce their tools and weapons or as

a currency in their transactions[25].

2.4.1. Properties and applications

Pure copper is soft and malleable, its surface has a reddish-orange color when freshly

exposed, and after natural aging it forms a natural green patina very much appreciated in architecture

and design. Its high oxidation resistance to a wide variety of aqueous agents made it perfectly suited

for plumbing, and when it price was low its use became widespread. Nowadays, due to the price

increase it’s being replaced for cheaper materials such as polymers. This chemical resistance

combined with a good thermal conductivity led to a general use in heat exchangers and heating

systems.

Copper’s electric conductivity is one and a half times the aluminum conductivity and because

of that the major application of copper, about half of world’s production, is electrical wires; from power

generation, transmission and distribution to telecommunications and electronic circuits. The rest of

copper applications are mainly roofing and plumbing and industrial machinery, a small part is yet

combined with other elements to form an alloy such as cupronickel, used in low-denomination coins,

brass and bronze.

2.4.2. Alloys designation

Copper and copper alloys can be grouped in families [25]. Coppers with a minimum copper

content of 99.3%. High copper alloys, with at least 96% copper for wrought alloys and a minimum of

94% for cast alloys. Brasses, as stated above, are alloys that contain zinc as the principal alloying

element. Bronzes, originally described as alloys with tin as its only or principal alloying element are

now more accurately defined as copper alloys in which the major alloying element is not zinc or nickel.

Copper-nickels are, as implied, alloys with nickel as the principal alloying element, with or without

other designated alloying elements. Copper-nickel-Zinc alloys, commonly known as "nickel silvers",

are alloys that contain zinc and nickel as the principal and secondary alloying elements. Leaded

coppers, which comprise a series of cast alloys of copper with 20% or more lead, sometimes with a

small amount of silver, but without tin or zinc. And finally, those whose chemical compositions do not

fit into any of the above categories are combined in a group called "special alloys”.

14

In the Unified Numbering System (UNS), numbers from C10000 through C79999 denote

wrought alloys. Cast alloys are numbered from C80000 through C99999. In the Table 2.5 are

summarized the different families of wrought copper alloys, their groups and the UNS numbers

assigned to each one [26].

Table 2.5 – Composition of the different families of wrought copper alloys.

Family Main alloying

element Group

Secondary alloying elements

UNS Number

Copper - Coppers - C10100-C15999

High Copper Alloys - C16000-C19999

Brass Zinc (Zn)

Yellow Brasses - C2xxxx

Leaded Brasses Lead (Pb) C3xxxx

Tin Brasses Tin (Sn) C4xxxx

Bronzes

Tin (Sn) Phosphor Bronze Phosphorus (P) C50000-C52999

Leaded Phosphor Bronzes Lead (Pb), Phosphorus (P) C53000-C54999

- Brazing Alloys

Phosphorus (P), Silver (Ag)

C55000-C55299

Copper-Silver-Zinc-Alloys Silver (Ag), Zinc (Zn) C55300-C60799

Aluminum (Al) Aluminum Bronzes - C60800-C64699

Silicon (Si) Silicon Bronzes - C64700-C66199

Zinc (Zn) Copper-Zinc Alloys - C66200-C69999

Copper Nickel Nickel (Ni)- Copper Nickel - C70000-C73499

Nickel Silvers Zinc (Zn) Copper-Nickel-Zinc Alloys Nickel (Ni) C73500-C79999

2.4.3. Copper weldability

Copper weld is very susceptible to cracking during solidification and porosity. When welding

copper it is very important to take in consideration the oxygen content because it can lead to porosity

and/or discontinuities to the weld bead. Because of those problems, fusion processes do not easily

weld copper alloys. Therefore, FSW presents as a very suitable process and because the thermal

delivery is smaller, minimizing the HAZ, the problems with distortion and buckling are avoided [27].

FSW of copper is more difficult as compared to FSW of aluminum as copper has it has a significantly

higher melting point, thermal conductivity and flow stress.

An important aspect of linear friction welding of copper is the need of tool cooling, preventing

overheating and, therefore, the increase of temperature in the underlying base material. This can lead

to degradation of the quality of the welds due to penetration of sharp tool and burr formation.

Excessive heating of the base material may also cause grain growth in the heat affected zone, with

negative consequences for the mechanical properties of the joint. Despite the possibility of incidence

of these problems, it should however be noted that the cooling of the tool, if excessive, can also lead

to the formation of welding defects characteristic of execution of the process at low temperature [28].

15

2.5. Statistic Method – Taguchi Method

Genichi Taguchi developed a technique for designing and performing experiments in order to

investigate processes where the output depends on many factors, this methodology is called Taguchi

Method [29]. By systematically choosing certain combinations of variables it is possible to separate

their individual effects, and avoid the need of perform all possible combinations of values of those

variables.

Although similar to the design of experiments (DOE1), the Taguchi method only considers

balanced experimental combinations (orthogonal), which makes it even more effective than a

fractional factorial planning, which runs only a fraction of the total number of combinations of variables

process input. Using the Taguchi Method allows industries to significantly reduce the cycle time for

product development in the design and production, thereby reducing costs and increasing profits.

Taguchi had three main theories for quality management [30]:

1- Quality should be designed into a product, not inspected into it. Quality is designed into a

process through system design, parameter design, and tolerance design. Quality "inspected into" a

product means that the product is produced at random quality levels and those too far from the mean

are simply thrown out.

2- Quality is best achieved by minimizing the deviation from a target. The product should be

designed so that it is immune to uncontrollable environmental factors. In other words, the signal

(product quality) to noise (uncontrollable factors) ratio should be high.

3- The cost of quality should be measured as a function of deviation from the standard and the

losses should be measured system wide. This is the concept of the loss function, or the overall loss,

incurred upon the customer and society from a product of poor quality. Because the producer is also a

member of society and because customer dissatisfaction will discourage future patronage, this cost to

customer and society will come back to the producer.

2.5.1. Methodology

Application of Taguchi Method, Figure 2.7, begins with the definition of the problem or the

objective of the study to be performed. This means determining what performance to be optimized.

This may be a flow rate, temperature, weight, cost, roughness, thickness, and so on. With the choice

of the object is also necessary to choose which quality characteristic best suits the chosen object of

study. There are three types of quality feature: Larger the better (for example tensile strength), Smaller

the better (e.g. carbon emissions) or On-target with minimum variation (e.g. size).

The next step is to identify the operating parameters that have significant effects on the

process. Control parameters are those whose values will be controlled and changed. The number of

levels, their associated values for each test parameter defines the tests to be performed. Increasing

the number of levels to vary a parameter increases the number of experiments to be conducted, and

reducing the number of levels could lead to a non-conclusive test. Robust parameter designs consider

16

controllable and uncontrollable noise variables; they seek to exploit relationships and optimize settings

that minimize the effects of the noise variables.

After this process, it’s time to choose the Orthogonal Array (OA) for the parameter design

indicating the number and conditions for each experiment. The selection of OA is based on the

number of parameters and the levels of variation for each parameter.

Subsequently, the experiments indicated in the OA are conducted and the results are

collected. Here can be performed only one trial for each parametric combination or more repetitions

depending on the number of noise parameters.

After the tests performed, it is determined the optimal set of parameters. This optimum

condition is not necessarily provided in one of the tests performed during the trial, that is, may not

match any of the rows of the matrix of orthogonal vectors applied.

Once identified the optimum condition, follows the application of the analysis of variance. This

is a statistical tool used to interpret experimental results. Analysis of variance allows knowing the

influence of each parameter variation in the quality of the piece obtained, i.e. the importance of small

variations in the input parameters to the variance of output parameters.

Finally, after having found the optimal parametric adjustment and the subsequent response to

that adjustment, it’s advisable to conduct a confirmatory test. If the result of this test is satisfactory and

verifies an improvement in the process, the application of Taguchi method ends. Otherwise, it’s

necessary to repeat the method changing the experimental plan or use the same plan but reducing the

range of values of the levels of each parameter.

If n

ot

sati

sfa

cto

ry

Performance to be optimized

Objective

Matrix Selection

Conduct Experiments

Identify Optimum Condition

Confirmatory TRIAL

Number of parameters that will be changed

Number of levels of changes

Control Parameters

Figure 2.7 – Taguchi Method Flow Chart

17

2.5.2. ANOVA

In ANalysis Of VAriance (ANOVA) the variance in the response measurements is partitioned

into the components that correspond to different sources of variation. Those sources are the control

parameters, the interactions between parameters and a random variation (error) [31]. For a study of k

control parameters, varied for n levels, with no interactions between those parameters, the partition of

the Total Sum of Squares, 𝑆𝑆𝑇, is given by equation (2.8):

𝑺𝑺𝑻 = ∑𝑺𝑺𝒊

𝒊=𝒌

𝒊=𝟏

+ 𝑺𝑺𝒆𝒓𝒓𝒐𝒓 (2.8)

Where:

𝑆𝑆𝑇 = ∑ ∑ (𝑌𝑖𝑗 − )2𝑖=𝑘

𝑖=1𝑗=𝑛𝑗=1 , is the Sum of Squares of the deviation of all individual responses,

Y, and their average, ;

𝑆𝑆𝑖 = ∑ 𝑛(𝑌𝑖𝑗 − 𝑖)2𝑗=𝑛

𝑗=1 , is the Sum of squares of control parameter i:

In analogy, it’s possible to summarize the number of system degrees of freedom (DOF) [32] as

shown in equation (2.9):

𝑫𝑶𝑭𝑻 = ∑𝑫𝑶𝑭𝒊

𝒊=𝒌

𝒊=𝟏

+ 𝑫𝑶𝑭𝒆𝒓𝒓𝒐𝒓 (2.9)

Where:

𝐷𝑂𝐹𝑇 = (𝑛 × 𝑘) − 1

𝐷𝑂𝐹𝑖 = 𝑛 − 1

The Mean Square, MS, variation of each parameter or error is given by equation (2.10):

𝑴𝑺𝒊 =𝑺𝑺𝒊𝑫𝑶𝑭𝒊

(2.10)

F-test is used to check the null hypothesis, H0, that there is no deviation on the mean values of

each factor level. In order to accept or reject H0 it’s need to compare 𝐹𝑖 = 𝑀𝑆𝑖/𝑀𝑆𝑒𝑟𝑟𝑜𝑟 , with Fc.,

equation (2.11).

𝑭𝒊 > (𝑭𝒄 = 𝑭(𝜶,𝑫𝑶𝑭𝒊, 𝑫𝑶𝑭𝑻) (2.11)

The null hypothesis is rejected with a significance level of α when the above equation is

satisfied. This means that the factor is relevant to the final output. F-test is tabled for significance

levels of 1%, 5% and 10%, and generally the level used is the intermediate.

Finally it is used ρi, equation (2.12), to express the relative influence of each parameter and

error, to the total output of the study.

𝜌𝑖 = 𝑆𝑆𝑖/𝑆𝑆𝑇 (2.12)

18

2.6. Friction Stir Welding development on the study area

It was acknowledge that FSW is a high quality process in fast development, especially in the

transportation industry. Is perfectly suitable for transformers manufacture and can be a significant add

to this industry. SIEMENStir presents as an innovative and pioneer project in FSW to electrical

components. The versatility and easy reproducibility of the process, adding to, the improvement on the

weld quality and mechanical properties of the final joint make this an important study to SIEMENS-FS.

Several studies have been performed on aluminum FSW, but none has been done using such soft

alloy as the used AA1070 thin sheet. Electrolytic Though Pitch Copper FSWeldability is also relatively

unknown. This is why this is a challenging work and a possible breakthrough to the sector.

2.6.1. FSW on thin sheets

The FSW process diminishes some of the weldability problems usually associated with fusion

welding processes, due to its low heat input [33]. However, FSW process has limitations in butt-joining

thin sheets. The thickness reduction resulting from the forging effect of the shoulder can significantly

reduce the mechanical resistance in thin plates. The presence of micro defects, usually acceptable in

thick welds, also create serious problems in thin plate sheet welds [34]. FSW is primarily studied for

aluminum, after all those are the main applications of the process. However, typical thicknesses are

superior to 3mm and usual alloys are from heat treatable 2XXX, 6XXX or 7XXX series that were once

considered “unweldable”. Few studies have been found regarding the joint of thin sheets even for the

stated alloys and none of them used an annealed alloy.

Leitão et al. [35] used 1mm thick sheets of AA5182-H111 and AA6016-T4, with base

hardness’s of 71 and 66HV02, respectively, to study the mechanical behavior of similar and dissimilar

welds. Afterwards Leal et al. [36] studied the material flow in those dissimilar welds. And finally

Rodrigues et al. [37] used the same 1mm thick sheets of 6016-T4, to investigate the influence of tool

geometry and welding parameters on the material flow path during welding. Thin sheets of commercial

pure aluminum have been successfully join by FSW, although the used alloy, AA1050-T4, had

significant high hardness due to solution heat treatment. Topic et al. [38] studied the grain size

variations of accumulative roll bonding (ARB) sheets submitted to FSW. FSW is especially suitable to

weld ultra-fine grains produced by ARB because the softening of this alloys is reduced compared to

other joining processes [39].

In the first stage of this project [1] was shown that it is technologically feasible to weld the

aluminum sheet in study. It was also concluded that the maximum load capacity of the welded zone

even matches the base material properties for the tensile tests and overmatches in about 20% the

bending tests. Nevertheless the ductility of the weld zone is reduced.

Although copper’s mechanical resistance fits better in FSW needs, also few studies have been

found on thin copper sheets welded by FSW. Galvão et al. [40] studied the influence of shoulder

geometry on properties of FSW in 1mm sheet of copper-DHP (base hardness of 92HV). The welds

were produced using three different shoulder geometries: flat, conical and scrolled, and varying the

rotation (400, 750 and 1000 rpm) and travel speeds of the tool (160 and 250mm/min). It was

19

acknowledged that both scrolled and conical shoulders can be used to obtain defect free welds, but

both needed a minimum rotational speed to avoid internal defects, 1000rpm for conical shoulder and

750rpm for scrolled one. The same copper-DHP sheets have been successfully weld by Leal et al.[41]

using tools with different shoulder cavities. The study was done based on mechanical and

metallographic tests. It was concluded that the torque, the microstructure and hardness and the

formation of defects in the welds are influenced mainly by tool rotation speed and, to a lesser extent,

by the traverse speed and shoulder cavity.

2.6.2. Dissimilar Al/Cu welds

Copper and Aluminum are widely used in the electric power industry. However, it is difficult to

join them using fusion welding methods due to their large differences in physical and chemical

properties and tendency to form brittle intermetallic compounds [42].

Akinlabi [43] studied the effect of shoulder size on butt weld properties of AA5754 and C11000

dissimilar FSW of 3.175mm thick plates. Using 15, 18 and 25 mm shoulder diameter it was concluded

that 15 and 18 mm diameters were more suitable than 25mm shoulder. Galvão et al. [44] studied

copper over aluminum lap welding by FSW. Using 6mm thick plates of a heat-treatable (AA6082) and

a non-heat-treatable (AA5083) aluminum alloys and 1mm copper sheet, being stated that welds using

AA5083 resulted on excellent surface finishing but highly defective Al/Cu interfaces. It was also

registered an impressive hardness increase due to the formation of ultra-refined microstructure of

aluminum. Firouzdor and Kou [45] studied FSW of 6061 Al to commercially pure Cu. Conventional lap

FSW was modified by butt welding a small piece of Al to the top of Cu, with a slight pin penetration

into the bottom of Al. At travel speeds up to 127 mm/min the modified welds were about twice the joint

strength and five to nine times the ductility of the conventional lap welds.

2.6.3. Taguchi on FSW

Since it emerged, the Taguchi Method has been used in several areas of study, publications

can be found from medicine to robotics passing by food science or zoology. In the present work, it will

be addressed only works on Friction Stir related processes on metals [46]–[48], and other materials

[49]–[51].

Koilraj [46] used a Taguchi study for optimization of FSW process parameters to a dissimilar

aluminum alloys weld. Varying rotational speed, travel speed, tool geometry and ratio between tool

shoulder diameter and pin diameter. The optimum process parameters were determined with

reference to tensile strength of the joint and then confirmed by a last trial with the optimum

parameters. Vidal [47] optimized fatigue behavior of aerospace alloy AA2024-T351, based on tensile,

bending and hardness characterization. Heidarzadeh [48] used a larger DOE to study the tensile

behavior of friction stir welded 6061-T4. Bagheri [49] studied the effect of rotational speed, travel

speed and temperature to FSW of ABS. Bozkurt [50] developed the same study for FSW of

polyethylene varying rotational speed, travel speed and machine tilt angle. Dashatan et al. [51] used

similar approaches for optimized dwell time, plunge speed and rotational speed of FSSW of dissimilar

PMMA and ABS.

20

2.7. Summary of State of the Art

In the past years, aluminum has emerged in the electric conductors industry. The main reason

is the obvious economic factor because aluminum is between two to three times cheaper than copper.

Although aluminum conductors exhibit bigger section and volume, it’s mass and consequently its

weight is halved when compared to copper conductors. This leads to a significant trend in the

production of windings for transformers for power systems, and opens the possibility of developing

new products and new processes. One of the challenges of this change is the relatively low weldability

of aluminum sheets with conventional welding processes.

The Friction Stir Welding (FSW) process is a breakthrough on joining components’ technology

and has been through a rapid growth in research, development and application since the early 2000s.

It has been subjected to the most demanding quality standard requirements and used in challenging

industrial applications over a wide range of structural and non-structural components mainly in light

alloys for transport industries. But few tests have been found on electrical components.

The Taguchi Method is a technique for designing and performing experiments in order to

investigate processes where the output depends on many factors. By systematically choosing certain

combinations of variables it is possible to separate their individual effects, and avoid the need of

perform all possible combinations of values of those variables.

Several studies have been performed on aluminum FSW, but none has been done using such

soft alloy as the used AA1070 thin sheet. Electrolytic Though Pitch Copper FSWeldability is also

relatively unknown. So it’s important to develop process parameters and tools for such applications.

This is why this is a challenging work and a possible breakthrough to the sector.

21

3. Equipment Characterization

3.1. Esab LegioTM FSW 3U

All friction stir welds in this work have been made with LEGIOTM

FSW 3U manufactured by

ESAB (Figure 3.1), this is a FSW driven laboratorial equipment with computer numerical control that

can perform welds either in vertical position control or in forging force control. This machine enables

besides the force control, parameters monitoring and tool cooling system. Those are the main

advantages to conventional milling equipment also used to perform FSW.

Figure 3.1 – Welding equipment LEGIOTM

FSW 3U of ESAB.

The equipment has four degrees of freedom: one of head rotation (C-axis) and three linear

axis: (1) x-axis (1200 mm max. extent) granted by a rack and pinion system; (2) y-axis (amplitude 400

mm) by an endless screw; (3) z-axis (approx. 340 mm extent) done by a hydraulic cylinder with start

and end limit sensors.

Figure 3.2 – Representation of the different constituents of the equipment.

Caption 1. Welding Head 2. Welding Head Support 3. Overall Support Structure 4. Motor for X-axis Movement 5. Motor for rotation (C-axis) 6. Hydraulic Cylinder 7. Hydraulic Unit 8. Control Booth 9. Numerical Control Panel 10. Cooling System

22

In Figure 3.2 are represented the various components of the machinery. The support for the

welding head can be rotated ±5º in XZ plan to introduce tilt angle. When a side tilt angle is needed

there are two possible solutions. The first one is to slope the work table (rotation in X-axis), the second

solution is to weld in YY direction although this solution is inadvisable as it can damage the endless

screw transmission. More detailed information about the equipment and the parameters that can be

controlled can be found in the user guide.

Figure 3.3 – Work table and fixing system used.

The FSW machine is coupled to a support table (Annex E1) that has been designed to sustain

the load applied during the welding process. This table can be used directly as a work table, but since

most of the welds of this study needed strict restraining systems a second work table has been used.

The design of the work table can be seen in Annex E2, the table has been planned to allow the proper

fixing to the weld sheets, in an edge-to-edge setup. Figure 3.3 exhibits the fixing system used. This

work table presents a removable center piece, in the weld path, which can be replaced whenever it

gets so damaged that it could spoil the weld. This allows experimenting longer probe lengths or bigger

loads without concerns.

23

3.2. FSW Tools

FSW Tool geometry is one of the most important parameters in the process. Therefore, it is

essential not only the geometry of the shoulder and probe, that are in contact with the parts to be

welded, but also how the tool body resists to the mechanical efforts applied and dissipates the heat

generated during the process. The third version of the modular tool developed at IST, called

iSTIRtool_v3 [52], was used in the present work (Figure 3.4). This tool consists of three modules

(body, shoulder and probe) which are integrated to compose the final geometry, adjustable to the type

and thickness of the materials to be welded.

The body of the tool was built in Steel Ck45, shoulder and probe were made of tool steel, whit

a specification in accordance with the European Standard (EN) is X40CrMoV5-1, H13 in AISI

designation. This material was machined to its final form, annealed and subsequently subjected to

heat treatments of quenching and tempering under vacuum at a temperature of 400ºC, which gave it

an average hardness of 533HV. This hardness assures high resistance resulting in a rigid behavior

and non-consumable when in contact to the materials to weld. In order to increase the life of the tool

all the components were cleaned with glass beads, for surface oxides removal. Subsequently, it was

subjected to thermochemical treatment for surface hardening by nitriding to a depth of around 30

micrometers. In order to reduce the phenomenon of adherence of material to the shoulder and probe

was also promoted the oxidation treatment with steam at 500 ºC to a depth of about 3 micrometer.

Those modules, probe and shoulder are easily replaceable and matchable, allowing a wide

range of solutions. This tool allows the use of an internal forced cooling and incremental changes in

the length of the pin 1/12 mm without the need to dismantle or remove the tool from the machine.

Figure 3.4 – 3-D view of the iSTIRtool_v3 assembly. 1) Tool body; 2) Shoulder; 3) Probe; 4) Fixing Shoulder Screws; 5) Fixing Probe Screw.

24

For this work five different M4 shoulders and two M4 probes have been designed, combining

for 10 different tool geometries. Aluminum butt weld was the first study case of the project [1], and

after the first stage it was acknowledged that the 5mm diameter probe used, was probably too big,

causing problems with the material flow, and the 3mm diameter probe also tested was too small,

causing problems in the joint itself. To overcome this problem intermediate shoulders and probes with

4mm were designed as shown in Figure 3.5 and in Figure 3.6, respectively, the corresponding

drawings can be seen in Annex E. For each type of weld addressed in this report the chosen tool

combination is explained and justified in the beginning of the topic.

Figure 3.5 – Different M4 shoulder geometries. a) Concave Smooth; Flat scrolled with: b) one striate normal pace c) two striates double pace d) two striates

normal pace e) one striate double pace.

Figure 3.6 – M4 Probes a) with conical tips, final diameter 3mm, b) with cylindrical tips, 6 mm length.

25

4. Tests Characterization

In order to acknowledge the mechanical response of aluminum and copper butt welds, they

were submitted to tensile, bending and hardness tests. Figure 4.1 represents the cuts that were made

in the welded sheets to make the appropriated specimens for those tests. Each section has been

numbered, from X.0 to X.7, to indicate the position on the weld sheet. Tensile (T) and bending (B)

specimens were cut, according to standards, with 20mm width and try to represent, in both tests, all

the welding length. Specimens X.0 and X.7 (M) have been used for metallographic analysis and were

used to test the hardness profile of the transversal section of the weld. X character represents the trial

number of the Taguchi study that will be introduced further on.

4.1. Tensile Test

Tensile testing was performed using an Instrom 4507 equipment, Figure 4.2, in Secção de

Tecnologia Mecânica, Instituto Superior Técnico, Universidade Técnica de Lisboa with a 200kN load

cell and bi-axial extensometers of high resolution and following the experimental procedure in annex

A4.

Figure 4.2 – Tensile testing machine, Instrom 4507, overall view at left, testing zone at right.

Figure 4.1 – Representation of the cuts made on butt welded sheets of aluminum and copper for taguchi study.

26

Three tensile specimens in accordance to the NP-EN895:2002 standard [53] were machined

with the dimensions shown in annex C1, from each base material and experimental welding trial.

Tensile test results, load (𝐹) and respective displacement (𝑙), were properly treated in order to obtain

the Engineering Stress (σ) and Strain (ε) curve according to equation (4.1):

𝜺 =𝒍 − 𝒍𝟎𝒍𝟎

=𝒍

𝒍𝟎− 𝟏

𝝈 =𝑭

𝑨𝟎

(4.1)

After, they are transformed in the True Stress (𝜎) versus True Strain (𝜀) curve through the

equation (4.2):

𝒅 =𝒅𝒍

𝒍⇔ = 𝒍𝒏

𝒍

𝒍𝟎= 𝐥𝐧 (𝜺 + 𝟏)

=𝑭

𝑨=

𝑭 × 𝒍

𝑨𝟎 × 𝒍𝟎= 𝝈(𝜺 + 𝟏)

(4.2)

Values of five mechanical properties were found, Young’s modulus (𝐸), offset yield tensile

strength at 0.2% (𝜎0.2), ultimate tensile strength (𝜎𝑚𝑎𝑥), tenacity modulus (𝑀𝑇) and finally the maximum

elongation before rupture (𝜀) [54]. The Young’s modulus is obtained by the slope of the elastic regime

linear trendline based on Hooke’s law. The offset yield tensile strength is the tensile value that

generates a 0.2% elongation, after elastic recuperation, see Figure 4.3. The ultimate tensile strength is

found by corresponding the maximum engineering tensile with their true tensile value. Maximum

elongations before rupture is self-explained, and the tenacity modulus is the area below the curve until

rupture, see Figure 4.4.

Figure 4.3 – Magnification of Tensile test results of AA1070, BM1 trial, near elastic regime.

Schematic representation of 𝑬 and 𝟎.𝟐.

27

Figure 4.4 – Tensile test results of AA1070, BM1 trial.

Schematic representation of the properties 𝒎𝒂𝒙, 𝑴𝑻 and .

4.2. Bending Test

Three bending specimens in accordance to the NP EN 910:1996(E) standard [55] were

obtained from each trial and base material with dimensions of annex C2. These specimens were

tested in the ESAB LEGIOTM

FSW 3U machine using a 0.1mm/s Plunge Speed. The complete

experimental procedure can be seen in annex A5. Figure 4.5 shows the test apparatus using the

structure of annex D.

Figure 4.5 –Three Point Bending example of aluminum weld specimen.

28

The three point bending setup for aluminum was done with a mandrel of 9.5mm diameter (D)

and two 5mm rollers (R) spaced of 14mm (L) as shown in Figure 4.6. For copper the thickness (a) is

only of 1.1mm and the distance between the rollers (L) has been reduced to 13.0mm.

Bending test results represent the behavior of the material when submitted to a constant

descendent speed in a curve Load (F) vs displacement (d). From the bending results have been

found values of three mechanical properties, maximum force (Fmax), displacement at maximum force

(𝑑𝐹𝑚𝑎𝑥) and absorbed energy (En) as shown in the Figure 4.7.

Note that ESAB LEGIO reads data at each tenth of a second but the hydraulic cylinder that

generates the downward movement is not so sensitive and the decreasing speed is very slow,

0.1mm/s. This means that for each displacement there is, most of the times, more than one load read

as seen above. This graphical displacement can be corrected either by averaging the respective load

values or increasing the decreasing speed. Both solutions can be tricky leading to smaller

discrepancies in the results in the former case and to masked peaks in the latter.

Figure 4.6 – Representation of bending test setup for aluminum.

Figure 4.7 – Bending test result for copper C11000.

Schematic representation of Fmax, dFmax and En.

29

4.3. Micro-hardness

Vickers hardness testing was performed using a Mitutoyo HM-112 Micro-Vickers Hardness

Testing Machine available at Departamento de Engenharia Mecânica e Industrial, Faculdade de

Ciências e Tecnologia, Universidade Nova de Lisboa (Figure 4.8). Micro-hardness indentations have

been performed under a load of 200 g spaced 1.0 mm from each other according to the ISO 6507-

1:1997(E) standard [56]. Hardness profiles were taken at the average thickness of the transversal plan

to the welding direction. At least ten indentations have been made for each side of the sample, starting

from the center of the nugget region.

Figure 4.8 – Mitutoyo HM-112 Vickers micro-hardness testing machine a) Overall view of the equipment, control display at right, and data recording monitor at left b) View of the 50x lens

used to measure the hardness c) View of the indenter used to perform the test.

30

31

5. Base Material Characterization

The present chapter studies the base material used in this work, the material selection was

done together with SIEMENS FS and aims to represent not only the principal geometries used in

Sabugo´s products but also the more challenging combinations for the work. Having said that, foil

thicknesses slightly smaller than those that are usually studied were chosen, also, as one of the goals

of this work is dissimilar foil welding, and as shows in Chapter 2.1.1, aluminum thickness should be

around 1.64 times the copper thickness. The Aluminum foil had 1.6mm, and copper had 1.1mm of

thickness.

Foil characterization was done through metallographic analysis, and mechanical resistance

tests (micro-hardness, tensile and bending), to verify the supplier certification, when applicable. The

results were also compared and verified in the corresponding Standards for aluminum [57] and copper

[58] for electrical purposes.

5.1. Aluminum Foil Characterization

AA1070 aluminum foil was received from SIEMENS FS, in sheets with 1.6mm thick and

dimensions of 1100x800mm. Those sheets have been cut accordingly to the applications, for butt

welding samples with 200x100mm were used with the bigger dimension being cut perpendicular to

rolling direction. For foil-bar welding, the samples used were adjusted to the bar dimensions.

Figure 5.1 – Aluminum foil, as received.

Figure 5.1 shows the AA1070 cold hardened and partially annealed foil as received. This

sample has been prepared according to the procedure described in Annex A2, and etched to reveal

grain boundaries with Poulton reagent.

Figure 5.2 was obtained of a supplier certification document and indicates, in the first line, the

composition of the specified material of the client, SIEMENS FS, and the composition of two different

lots shipped to the factory. There is no guarantee that the material used came from any of those lots. It

was assumed that the material follows the client certification, so that aluminum foils have, at least

99.70% aluminum and, as it was explained in section 2.3.2, this can be classified as AA1070.

32

Figure 5.2 – AA1070 composition certification.

Figure 5.3 – Aluminum mechanical properties certification.

The table of Figure 5.3 was also obtained from the supplier technical sheet, and it was used to

verify the results of tensile tests done to base material. Those tests have been made in 3 specimens

extracted along the rolling direction with dimensions that can be seen in Annex C1. The results are

summarized in Table 5.1. The comparable values are the offset yield strength (at 0.2%), the ultimate

tensile strength, 𝜎𝑚𝑎𝑥, and elongation. All values seem to be correct and in accordance with the

standards that specify 𝜎𝑚𝑎𝑥 of 65-95 MPa, 𝜎0.2 of at least 20 MPa and minimum elongation of 33%

[57].

Table 5.1 – Tensile tests results for 3 specimens of Aluminum Base Material.

E [GPa] 𝟎.𝟐 [MPa] 𝒎𝒂𝒙 [MPa] 𝑴𝑻 [J/mm3] [%]

BM1 81.19 34.77 77.74 20.25 36.00

BM2 71.17 34.53 76.01 19.90 36.00

BM3 69.80 35.42 79.54 19.01 34.00

Average 74.05 34.91 77.76 19.72 35.33

Bending tests results are summarized in

Table 5.2, again, they were obtained by testing 3 specimens and the average value of the 3

tests was considered the standard for the following work.

Table 5.2 – Bending tests results for 3 specimens of Aluminum Base Material.

Fmax [kN] dFmax [mm] 𝑬𝒏 [J]

BM1 1.03 6.19 530.07

BM2 1.07 5.96 518.50

BM3 1.03 6.04 513.94

Average 1.04 6.06 520.84

Finally the Base Material was tested for measurement of it hardness, 2 sets of 9 points were

measured, and the average was 27.9HV.

33

5.2. Copper Foil Characterization

C11000 fully annealed copper sheet was received in a 101Kg winding, with 1.1mm thick and

690mm width. The metallographic analysis of copper, Figure 5.4, revealed a matrix-lamellae

microstructure with anisotropic arrangement of the grains. Figure 5.5 shows the supplier chemical

composition certificate.

Figure 5.4 – Copper Foil, as received.

The same approach used to characterize aluminum, was used here. Three specimens were

produced for tensile testing and three for bending. A sample was used to analyze microstructure and

test the hardness. The hardness test revealed a value of 65.1HV. The results of the tensile and

bending tests are shown in Table 5.3 and Table 5.4, respectively. All these results are very similar and

are in accordance to those found in the Standards [58] for the same alloy and manufacture process.

The Standards recommend a 𝜎𝑚𝑎𝑥 of at least 250 MPa, an yield strength with a maximum of 120MPa,

an elongation of 25-35% and an Hardness of 40-65HV.

Table 5.3 – Tensile tests results for 2 specimens of Copper Base Material.

E [GPa] 𝟎.𝟐 [MPa] 𝒎𝒂𝒙 [MPa] 𝑴𝑻 [J/mm3] [%]

BM1 175.48 76.62 306.34 64.15 35.00

BM2 103.53 73.93 305.60 64.76 36.00

Average 139.50 75.28 305.97 64.45 35.50

Note: There was a problem with the extensometer in the third trial, so the values were discarded.

Table 5.4 – Bending tests results for 3 specimens of Copper Base Material.

Fmax [kN] dFmax [mm] 𝑬𝒏 [J]

BM1 1.16 6.53 645.99

BM2 1.16 6.62 997.89

BM3 1.17 7.73 996.38

Average 1.16 6.96 880.09

Figure 5.5 – Chemical composition certification for C11000.

34

35

6. Characterization of Taguchi on FSW

In this work, two Taguchi studies were made, one for aluminum and the other for copper butt-

welding. In both studies the evaluation and control parameters, as well as the orthogonal array

applied, were the same. The nominal values of those parameters were obviously adjusted to each

condition. The concepts and theory explained in this section are general and will be used in each of

the sections corresponding to both tests, for aluminum and copper, in section 7.1 and 7.2 respectively.

6.1. Evaluation Factors

Three factors were used to evaluate the overall quality of the different trials, based in the

mechanical response of the welds. The first of those global analysis factors, Global Efficiency on

Tensile (GET) has been developed by Vilaça, P. [59]. Representing in percentage the performance of

the trial, according to the base material values weighted by the importance between the various

properties obtained. Equation (6.1)defines GET factor:

𝑮𝑬𝑻 = 𝑪𝑬 ×𝑬𝒊𝑬𝑩𝑴

+ 𝑪𝟎,𝟐 ×𝟎.𝟐𝒊𝟎.𝟐𝑩𝑴

+ 𝑪𝒎𝒂𝒙 ×𝒎𝒂𝒙𝒊𝒎𝒂𝒙𝑩𝑴

+ 𝑪𝑴𝑻×

𝑴𝑻𝒊

𝑴𝑻𝑩𝑴

+ 𝑪 ×𝒊𝑩𝑴

(6.1)

Where:

𝐸𝑖 ; 𝜎0.2𝑖 ; 𝜎𝑚𝑎𝑥𝑖 ; 𝑀𝑇𝑖 ; 𝜀 – represent the properties measured in tensile test of trial, 𝑖 ;

𝐸𝐵𝑀 ; 𝜎0.2𝐵𝑀 ; 𝜎𝑚𝑎𝑥𝐵𝑀 ; 𝑀𝑇𝐵𝑀 ; 𝜀𝑀 – these are the properties measured in tensile test

of base material used in trial, 𝑖 ; 𝐶𝐸 ; 𝐶0.2 ; 𝐶𝑚𝑎𝑥 ; 𝐶𝑀𝑇 ; 𝐶 – indicate the weight factors for each property of GET

factor. Those are specified in Table 6.1.

GET weight factors of the five properties measured in tensile test were chosen based on the

know-how and experience of the author, although lacking any scientific support. For this application,

the main objective is maintaining the elastic behavior. Therefore, the most important parameter is the

offset yield strength (𝜎0.2) as Young’s Modulus (𝐸) variance is generally small or related to the trial

initial instability. Maximum elongation (𝜀) is also an important factor to allow higher deformations,

Ultimate tensile strength (𝜎𝑚𝑎𝑥) has average importance and Tenacity Modulus (𝑀𝑇) is not relevant

due to the absence of impact loads.

Table 6.1 – GET weight for each property of tensile test.

𝑪𝑬 𝑪𝟎.𝟐 𝑪𝒎𝒂𝒙 𝑪𝑴𝑻 𝑪

10% 30% 20% 15% 25%

In analogy to GET, Vidal et al. [47] developed Global Efficiency on Bending (GEB), shown in

equation (6.2), and the Hardness Coefficient (HARD) that appears in equation (6.3). These summarize

the response of the welds to bending and hardness tests, respectively.

𝑮𝑬𝑩 = 𝑪𝑭 ×𝑭𝒊𝑭𝑩𝑴

+ 𝑪𝒅 ×𝒅𝒊𝒅𝑩𝑴

+ 𝑪𝑬𝒏 ×𝑬𝒏𝒊𝑬𝒏𝑩𝑴

(6.2)

Where:

𝐹𝑖 ; 𝑑𝑖 ; 𝐸𝑛𝑖 – represent the properties measured in bending test of trial, 𝑖 ; 𝐹𝐵𝑀 ; 𝑑𝐵𝑀 ; 𝐸𝑛𝐵𝑀 – are the properties measured in bending test of base material used

in trial, 𝑖 ; 𝐶𝐹 ; 𝐶𝑑 ; 𝐶𝐸𝑛 – indicate the weight factors for each property of GEB factor. Those are

specified in Table 6.2.

36

The GEB weight factors were, once again, just like in the GET case, chosen without scientific

support but representing an effort to establish a global criterion based on authors experience and in

the product application.

Table 6.2 – GEB weight for each property of bending test.

𝑪F 𝑪d 𝑪En

50% 25% 25%

𝑯𝑨𝑹𝑫 = 𝒎𝒊𝒏𝒊𝒎𝒖𝒎 𝒉𝒂𝒓𝒅𝒏𝒆𝒔𝒔

𝒃𝒂𝒔𝒆 𝒎𝒂𝒕𝒆𝒓𝒊𝒂𝒍 𝒉𝒂𝒓𝒅𝒏𝒆𝒔𝒔 (6.3)

Where the numerator is the smaller value of hardness read at half thickness of weld cross

section and the denominator the average hardness measured at half thickness of the base material.

6.2. Control Parameters and their Levels

The control parameters have been chosen from the FSW operation parameters (Section 2.2.2)

based on expertise and previous works [47]. The selected control parameters were Axial Force, Travel

Speed and Probe Length as they characterize all welding possibilities and avoid or introduce LOP

defects. Three levels have been selected for each of the three control parameters.

Table 6.3 – Control Parameters and their Levels.

Control Parameters Level 1 Level 2 Level 3

Vertical Force (Fz) F1 F2 F3

Travel Speed (Vx) V1 V2 V3

Probe Length (Lpin) L1 L2 L3

Table 6.3 represents the different parameter levels that are nominal values, in Kg, mm/min

and mm for Fz, Vx and Lpin parameters, respectively, equally spaced from the level before. As the

goal is to study the influence of the control parameters on the quality of the join, the remaining

parameters were kept fixed in all trials. Those parameters and respective values will be addressed

later.

6.3. Design of Experiments (DOE)

As the aim is to find the best set of parameters for three different control parameters (F, V and

L) over a window of three defined values (Level 1,2 and 3) it is used the Taguchi L9 orthogonal array

(OA), with just three columns, shown in Table 6.4. This OA has a total of 8 DOF (number of trials - 1).

So it’s possible with this OA to test up to four different parameters, with no interactions, or two

parameters, with interaction. For this study each factor will have 2 DOF (number of levels – 1)

summing a total of 6 DOF. This leaves 2 free degrees for error.

Table 6.4 – Taguchi L9 Orthogonal Array with 3 columns.

Trials Vertical Force (Fz) Travel speed (Vx) Probe Length (Lpin) 1 F1 V1 L1 2 F1 V2 L2 3 F1 V3 L3 4 F2 V1 L2 5 F2 V2 L3 6 F2 V3 L1 7 F3 V1 L3 8 F3 V2 L1 9 F3 V3 L2

37

7. Experimental Study Cases

7.1. Aluminum Butt Welding

7.1.1. Experimental Setup

The experimental setup is one of the most important factors when studying the interference of

parameters to the quality of the welds, because if the experimental conditions are different from trial to

trial it is not possible to establish a relation between results and those trials.

To avoid adhesion of the weld specimen to the worktable a copper bar with 10mm thick was

used, below the sheets to weld. As shown in Figure 7.1, the copper plate was placed above the steel

worktable, and fixed by four clamps. The aluminum sheets are then placed above and fixed by a

secondary and more local restrictive fixing system: a five screw portico in each side.

Figure 7.1 – Experimental setup for aluminum welds.

7.1.2. Tool Geometry

Figure 7.2 – Probe 4I3 and shoulder geometry 4P3 used in aluminum butt weld.

After testing different combinations of shoulders and probes, based on experience [1], visual

analyses of the resulting welds and the cleaning of the tool after the weld, the 4I3 probes (Annex E7)

were chosen with the 4P3 shoulder (Annex E4) to this aluminum butt welds (Figure 7.2). A cylindrical

probe was required because of the small thickness of the welding sheets, avoiding the accumulation

of material in the gap between probe and shoulder. The one-striate shoulder was used because in

38

double scrolled ones, aluminum adheres to the surface, similarly to what happens with the concave

shoulder. This phenomenon causes instabilities through the weld path, causing the need to clean the

shoulder and the probe after each weld and consequently decreasing the productivity, especially when

the aim is to create a more efficient industrial solution.

7.1.3. Parameters

Once the experimental setup and the tools were defined, the following step was to choose the

fixed and the control parameters of the study. The following parameters were fixed, using the know-

how of a previous work [1]:

Ω = 800 rpm

Wp: 0.1 mm

α = 0.5º

Dt = 3s

Ps = 0.1mm/s

Tool Rotation Speed (Ω) was fixed according to preliminary trials, Welding Position (Wp) was

measured from the anvil copper plate, tilt angle (α) was chosen in compliance with the tool geometry

and Dwell Time (Dt) and Plunge Speed (Ps) were settled based on the thermal response of the

material.

Axial Force, Travel Speed and Probe Length have been chosen as the control parameters, as

explained in Section 6.2. The three levels of Vertical Force have been chosen so that they cover the

most of the welding spectrum of aluminum, from almost underrated to almost overrated force, Travel

speed levels aimed to represent both cold and hot welds and finally the Probe Length levels try to

represent undersized and oversized lengths. Table G.1 of the confidential annex G shows the Taguchi

matrix of trials with the nominal values of the control parameters.

Figure 7.3 shows both face and root sides of the nine trials performed on aluminum. In Annex

B1 can be seen a closer look of each weld.

Figure 7.3 – Overall look of the face and root sides of the nine trials of aluminum.

39

Table 7.1 summarizes the visual analysis of such welds. The amount of burr was not

significant in none of the welds although it was present in some. The Weld Bead Hollow in face side

and the Depression on root side seem to be directly related to the Axial Force of the Tool. Doubtful

Bond was found in more than half the welds.

Table 7.1 – Visual analysis of aluminum welds.

Trial

Burr Weld Bead (face) Weld Bead (root)

Quantity Type Width Striate Hollow Depression Bond

None

Few

A lo

t

Contin

uous

Gra

nula

r

Regu

lar

Varia

ble

Regu

lar

Varia

ble

None

Sm

ooth

None

Sm

ooth

Pro

no

unced

Yes

Doubtfu

l

1 X NA NA X X X X X

2 X X X X X X X

3 X NA NA X X X X X

4 X X X X X X X

5 X NA NA X X X X X

6 X NA NA X X X X X

7 X NA NA X X X X X

8 X NA NA X X X X X

9 X NA NA X X X X X

40

7.1.4. Tensile Tests Results

The results for aluminum welds are summarized in Table 7.2, and are weighted to the same

properties of base material. Finally the GET factor was calculated as explained in Section 6.1. The

results can be seen graphically in Figure 7.4.

Table 7.2 – Summary of results for tensile tests of aluminum.

Base Material

𝑬𝑩𝑴 [GPa] 𝟎.𝟐𝑩𝑴[MPa] 𝒎𝒂𝒙𝑩𝑴[MPa] 𝑴𝑻𝑩𝑴

[J/mm3]

𝑩𝑴 (%) 𝑮𝑬𝑻𝑴𝑩

74.05 34.91 77.76 19.72 35.33 1.00

Trials 𝑬𝒊 𝑬𝒊𝑬𝑩𝑴

𝟎.𝟐𝒊 𝟎.𝟐𝒊𝟎.𝟐𝑩𝑴

𝒎𝒂𝒙𝒊 𝒎𝒂𝒙𝒊𝒎𝒂𝒙𝑩𝑴

𝑴𝑻𝒊

𝑴𝑻𝒊

𝑴𝑻𝑩𝑴

𝒊 𝒊𝑩𝑴

𝑮𝑬𝑻𝒊

1 61.78 0.83 34.70 0.99 65.73 0.85 8.79 0.45 17.50 0.50 0.74

2 67.43 0.91 35.03 1.00 68.55 0.88 10.44 0.53 20.00 0.57 0.79

3 61.21 0.83 34.89 1.00 61.66 0.79 5.85 0.30 12.00 0.34 0.67

4 63.71 0.86 36.00 1.03 69.80 0.90 14.29 0.72 27.00 0.76 0.87

5 55.19 0.75 35.75 1.02 57.90 0.74 4.22 0.21 9.00 0.25 0.63

6 73.21 0.99 36.75 1.05 69.51 0.89 14.35 0.73 27.00 0.76 0.89

7 56.21 0.76 35.25 1.04 54.27 0.70 3.51 0.18 7.75 0.22 0.61

8 54.27 0.73 35.51 1.02 63.91 0.82 8.20 0.42 16.50 0.47 0.72

9 70.37 0.95 36.76 1.05 70.50 0.91 17.70 0.90 33.00 0.93 0.96

In the first graph of Figure 7.4 it’s possible to see that Trials 4, 6 and 9 have factors of more

than 0.8, which means that those parameters are well adjusted to this material. Trials 3, 5 and 7, the

trials with smaller GET factor, have the longest probe length. This means that this probe length was

too long and induced instabilities on the movement, such as scratching the base plate. To evaluate the

influence of the parameters in GET factor, the mean values of each level were calculated and plotted

in the right graph of Figure 7.4. This means that, for example, value V1 corresponds to the average of

trials 1, 4 and 7 GET value. From the analysis of the graph it is possible to conclude that the optimum

values for GET factor of the axial force, the travel speed and the probe length are F2, V3 and L2,

respectively. The value of the Level 3 of probe length is the smallest as expected from the previous

analysis of the individual trial results. All three values of Axial Force are very similar, which indicates

that variations in this parameter do not affect significantly this factor.

0.74 0.79

0.67

0.87

0.63

0.89

0.61

0.72

0.96

0.00

0.10

0.20

0.30

0.40

0.50

0.60

0.70

0.80

0.90

1.00

1 2 3 4 5 6 7 8 9

Trial

GET - Trials

0.73 0.80 0.76 0.74 0.71

0.84 0.79

0.87

0.64

0.00

0.10

0.20

0.30

0.40

0.50

0.60

0.70

0.80

0.90

1.00

F1 F2 F3 V1 V2 V3 L1 L2 L3

Parameter Level

GET - Parameters

Figure 7.4 – GET results for each trial (left) and parameter level (right).

41

7.1.5. Bending Tests Results

From the bending results, values of three mechanical properties were found, being those the

maximum load (𝐹), displacement (𝑑) and absorbed energy (En). Table 7.3 summarizes the values

obtained by the average of two valid results, from the three performed. Those results, as for tensile,

have been weighted with those of base material, and the global efficiency factor has been calculated

and plotted in the left graph of Figure 7.5.

Table 7.3 – Summary of results for bending tests of aluminum.

Base Material 𝑭𝑩𝑴 [kN] 𝒅𝑩𝑴[mm] 𝑬𝒏𝑩𝑴[J] 𝑮𝑬𝑩𝑴𝑩

1.04 6.08 520.84 1.00

Trials 𝑭𝒊 𝑭𝒊𝑭𝑩𝑴

𝒅𝒊 𝒅𝒊𝒅𝑩𝑴

𝑬𝒏𝒊 𝑬𝒏𝒊𝑬𝒏𝑩𝑴

𝑮𝑬𝑩𝒊

1 0.84 0.81 4.50 0.74 322.19 0.62 0.75

2 0.89 0.85 5.66 0.93 410.10 0.79 0.86

3 0.91 0.88 5.93 0.98 458.42 0.88 0.90

4 0.92 0.89 4.75 0.78 365.03 0.70 0.81

5 0.89 0.86 5.02 0.83 375.45 0.72 0.82

6 0.87 0.84 5.35 0.88 381.77 0.73 0.82

7 0.89 0.85 4.78 0.79 355.36 0.68 0.79

8 0.91 0.88 3.78 0.62 275.60 0.53 0.73

9 0.89 0.86 3.69 0.61 269.80 0.52 0.71

Figure 7.5 illustrates the GEB factor trials for aluminum welding, showing a good overall result

for all trials. Global Efficiency on Bending illustrate very consistent results for almost all trials, being

the exception trials 8 and 9 which have a reduced value, probably due to excessive force. Trials 2 and

3 have above average results, which would indicate better parameters selection. The parameters’

influence in GEB is shown in the right figure and demonstrate that this factor is very well behaved, with

increase of force and decrease of speed and probe length, the GEB factor increases. So the optimum

levels for this factor are F1, V3 and L3. Speed levels look very similar, indicating that this parameter

has little influence; however the force shows a big amplitude.

0.75

0.86 0.90

0.81 0.82 0.82 0.79 0.73 0.71

0.00

0.10

0.20

0.30

0.40

0.50

0.60

0.70

0.80

0.90

1.00

1 2 3 4 5 6 7 8 9

Trial

GEB - Trials

0.84 0.82 0.74 0.76 0.79

0.84 0.78 0.80 0.81

0.00

0.10

0.20

0.30

0.40

0.50

0.60

0.70

0.80

0.90

1.00

F1 F2 F3 V1 V2 V3 L1 L2 L3

Parameter Level

GEB - Parameters

Figure 7.5 – GEB results for each trial (left) and parameters level (right).

42

7.1.6. Hardness Tests Results

Table 7.4 summarizes the results of hardness tests. For each trial, a profile of at least 21

points was done. Those were summarized into three comparable values, minimum, average and

maximum hardness values. According to (6.3), the Hardness coefficient (Hard) for each trial was

calculated.

Table 7.4 – Summary of results for hardness tests of aluminum.

Base Material

𝑴𝒊𝒏 [HV] 𝑨𝒗𝒈[HV] 𝑴𝒂𝒙[HV] 𝑯𝒂𝒓𝒅𝑴𝑩

21.9 27.9 34.0 1.00

Trials 𝑴𝒊𝒏𝒊 𝑨𝒗𝒈𝒊 𝑴𝒂𝒙𝒊 𝑯𝒂𝒓𝒅𝒊

1 27.8 32.9 44.7 1.00

2 26.7 30.7 37.7 0.96

3 21.8 28.6 40.3 0.78

4 28.2 32.2 35.5 1.01

5 27.9 33.8 43.6 1.00

6 24.9 32.7 41.5 0.89

7 31.2 35.5 39.3 1.12

8 27.2 33.5 43.3 0.97

9 26.1 31.2 38.1 0.94

From the analysis of the average values it’s obvious that FSW strengthens aluminum in the

nugget region, being this event predictable. The doubt was if it would weaken the joint near the

transition between TMAZ and HAZ. Apparently this does not occur, as the hardness coefficients are

very close to the unit, for almost all trials, except for trials 3 and 6.

Figure 7.6 – Hardness coefficient results for each trial (left) and parameters level (right).

The hardness coefficient seems significantly much influenced by the parameter levels,

increasing when force and probe length increases and decreasing when the speed decreases.

However, it’s more dependent on Speed and Force that on probe length. The optimum parameters are

F3, V1 and L3 for axial force, travel speed and probe length.

1.00 0.96

0.78

1.01 1.00 0.89

1.12

0.97 0.94

0.00

0.20

0.40

0.60

0.80

1.00

1.20

1 2 3 4 5 6 7 8 9

Trial

HARD - Trials

0.95 0.97 0.97 1.04

0.98

0.87 0.91

0.97 1.01

0.00

0.20

0.40

0.60

0.80

1.00

1.20

F1 F2 F3 V1 V2 V3 L1 L2 L3

Parameter Level

HARD - Parameters

43

Figure 7.7 – Average values of the nine hardness profiles of Aluminum welds.

From advancing side at left, to retreating side at right.

Figure 7.7 shows the average hardness of all nine trials by position, in order to emphasize the

increase of hardness resulting from the FSW. Due to the grain refinement of the process, it’s more

intense in the region between the TMAZ and the nugget, about 3 mm to the advancing side.

7.1.7. Analysis of Variance (ANOVA)

ANOVA has been used to quantify the contribution (ρ) of each parameter to each factor.

Table 7.5 summarizes the results, identifying the degrees of freedom (DOF), the sum of

squared deviations (SS), variance (V) and F-test (F) for each parameter of each evaluation factor. As

this test was made without replication isn’t possible to assign error contribution to random errors or

lack-of-fit errors.

Table 7.5 – Results of variance analysis for the three evaluation parameters for aluminum.

Parameter

DO

F SS V F ρ(%)

GET GEB Hard GET GEB Hard GET GEB Hard GET GEB Hard

Fz 2 0.0063 0.0141 0.0142 0.0031 0.0070 0.0071 9.9570 1.7698 1.7859 5.11 50.58 45.00

Vx 2 0.0274 0.0012 0.0452 0.0137 0.0006 0.0226 43.6702 0.1465 5.7003 22.43 3.10 3.72

Lpin 2 0.0880 0.0081 0.0004 0.0440 0.0040 0.0002 140.0725 1.0170 0.0479 71.94 24.36 25.86

Error 2 0.0006 0.0079 0.0079 0.0003 0.0040 0.0040 - - - 0.51 21.96 25.42

Total 8 0.1224 0.0313 0.0677 - - - - - - 100 100 100

As shown in Figure 7.8 the contribution of each parameter is especially important to each one

of the factors. Being the probe length the main contributor to the GET variance, as well as the travel

speed for Hardness coefficient and the axial force for GEB. Error contribution is small for each factor,

which indicates that neither pure errors nor lack-of-fit errors are big contributors to this test.

Consequently, one could argue that this design of experiments was appropriated.

20

25

30

35

40

-10 -8 -6 -4 -2 0 2 4 6 8 10

Ha

rdn

es

s (

HV

)

Distance to center (mm)

Average Hardness

Average Base Material

44

7.1.8. Optimum parameters identification

To fulfill the main objective of improvement of the mechanical behavior of friction stir welded

aluminum butt welds, it was used an algorithm that would encompass the results of Taguchi method to

obtain the most robust set of parameters to improve globally the properties of welded joints [47].

Using the optimum values for the evaluation factors, described in sections 7.1.4 to 7.1.6, the

first matrix of equation (7.1) was build. The second matrix was made by weighting the contribution of

each parameter from section 7.1.7. Identity of former product matrixes represents the optimum values

for each parameter, considering the same weight for the three test factors.

[

𝑭𝑮𝑬𝑻 𝑭𝑮𝑬𝑩 𝑭𝑯𝒂𝒓𝒅𝑽𝑮𝑬𝑻 𝑽𝑮𝑬𝑩 𝑽𝑯𝒂𝒓𝒅𝑳𝑮𝑬𝑻 𝑳𝑮𝑬𝑩 𝑳𝑯𝒂𝒓𝒅

]

[ 𝝆𝑭(𝑮𝑬𝑻)

𝝆𝑭(𝑻)

𝝆𝑽(𝑮𝑬𝑻)

𝝆𝑽(𝑻)

𝝆𝑳(𝑮𝑬𝑻)

𝝆𝑳(𝑻)𝝆𝑭(𝑮𝑬𝑩)

𝝆𝑭(𝑻)

𝝆𝑽(𝑮𝑬𝑩)

𝝆𝑽(𝑻)

𝝆𝑳(𝑮𝑬𝑩)

𝝆𝑳(𝑻)𝝆𝑭(𝑯𝒂𝒓𝒅)

𝝆𝑭(𝑻)

𝝆𝑽(𝑯𝒂𝒓𝒅)

𝝆𝑽(𝑻)

𝝆𝑳(𝑯𝒂𝒓𝒅)

𝝆𝑳(𝑻) ]

= [𝑳𝒐𝒂𝒅 𝒙 𝒙𝒙 𝑽𝒆𝒍𝒐𝒄𝒊𝒕𝒚 𝒙𝒙 𝒙 𝑷𝒓𝒐𝒃𝒆

] (7.1)

Equation (7.2) represents the results:

[𝑭𝟐 𝑭𝟏 𝑭𝟑𝑽𝟑 𝑽𝟑 𝑽𝟏𝑳𝟐 𝑳𝟑 𝑳𝟐

] [𝟎. 𝟎𝟕 𝟎. 𝟐𝟒 𝟎. 𝟕𝟓𝟎. 𝟔𝟑 𝟎. 𝟎𝟒 𝟎. 𝟐𝟔𝟎. 𝟐𝟗 𝟎. 𝟕𝟐 𝟎. 𝟎𝟏

] ≅ [𝑭𝟏. 𝟓 𝒙 𝒙𝒙 𝑽𝟏. 𝟓 𝒙𝒙 𝒙 𝑳𝟐

] (7.2)

The nominal values of the parameters can be found in the confidential annex H.

7.1.9. Summary of Aluminum Butt Welding Results

The application of Taguchi method to the aluminum butt weld was developed defining three

levels for the three main parameters and performing nine trials according to Taguchi L9 orthogonal

array. Given the small contribution of Error for the variation of the results (GET 0.47%, GEB 21.96%

and HARD 11.72%), it’s possible to assume that the design of experiments was correctly defined and

that inherent errors had a small influence.

5.11%

22.43%

71.94%

0.51%

45.00%

3.72%

25.86%

25.42%

20.93%

66.79%

0.56%

11.72%

Force

Speed

Probe

Error

Hardness

GEB

GET

Figure 7.8 – Contribution of each parameter for the three evaluation parameters.

45

Tensile results revealed that the optimum set of parameters for Global Efficiency on Tensile

were F2, V3 and L2, with the probe length being the most influencing parameter and Axial Force the

less significant. The Three Point Bending results showed a large dependence on force and little

relation to travel speed. The optimum parameters were F1, V3 and L3. The Hardness coefficient was

mainly dependent on travel speed variation and indifferent to probe length. The best results were

obtained with F3, V1 and L3 parameters levels. This study concluded that the optimum set of

parameters equally weighted by the three factors were the mean value between F1 and F2 for Axial

Force, the mean value of V1 and V2 for travel speed and L2 for probe length.

Confirmatory trials have been made using the optimum parameters and submitted to

mechanical, metallographic and electrical conductivity characterization tests. The mechanical results

on the aluminum confirmatory trials were 0.79 GET, 0.80 GEB and 0.99 HARD, which gives a

reduction of only 14% from the base material. Those results confirmed the legitimacy of this statistic

model and proving that this analysis can be used to define welding parameters. The metallographic

analysis, Figure 7.9, evidences the transition, Heat Affected Zone, between the Base Material and

Thermo-Mechanical Affected Zone in which the Nugget can be found.

Electrical conductivity results, Figure7.1, express the very low impact of the FSW weld beads

on conductivity of commercial pure aluminum. The decay from the base material electrical conductivity

is about 4%. This reduction is due to the substantial reduction of the grain size inside the TMAZ of the

weld bead, shown in Figure 7.9 and as concluded by Santos et al. [60].

Figure 7.10 – Eddy current conductivity test performed, at 250 kHz,

in three arbitrary paths (L1, L2, L3) of aluminum’s confirmatory trial.

BM

HAZ

TMAZ

Nugget BM

Figure 7.9 – Metallographic analysis of the confirmatory trial.

46

7.2. Copper Butt Welding

7.2.1. Experimental Setup

For copper’s butt welding was used the work table and the fixing system that were addressed

in Chapter 3.1. Again, it is important to mention that the system constrains are very important in the

FSW process. Without the proper constraint, conditions are not fulfilled for correct welding. Bigger

support plates with 250mm were produced. This allowed the use of all 3 screws of the portico, causing

better fitting and constraining. Another very important detail of a correct setup is the alignment and

flattening of the foil edges. To achieve this the foils were flatten, using a milling machine, the edges

parallel to the welding direction.

7.2.2. Tool Geometry

Probe 4I3 and Shoulder 4O3 were used in copper welding, based on preparatory trials and

expertise. Figure 7.11 shows the mentioned tools and their drawings can be seen in Annex E4 and

E7, respectively. Shoulder cavity has about 3º but has little influence in final weld quality as stated by

Leal [41]. The use of the concave shoulder lead to a tilt angle of two and an half degrees to promote

the correct material feed. Due to the small thickness of the foil a cylindrical probe was used once the

vertical flow of the material is not very useful.

7.2.3. Parameters

For copper welding the following parameters were used, according to the material properties,

the tool geometry and based on the expertise and preliminary trials:

Ω = 1000 rpm

Wp = 0.1 mm

α = 2.5º

Dt = 5s

Ps = 0.1mm/s

Tool Rotation Speed (Ω) was fixed according to preliminary trials and based on the study of

Galvão [40]. Welding Position (Wp) was measured from the back-anvil plate, tilt angle (α) was chosen

in compliance with the tool geometry and Dwell Time (Dt) and Plunge Speed (Ps) were settled based

on the thermal response of the material.

Figure 7.11 – Probe 4I3 and shoulder 4O3 used in copper butt weld.

47

Axial Force, Travel Speed and Probe Length have been chosen as the control parameters, as

explained in Section 6.2. The three levels of Vertical Force have been chosen so that they cover the

most of the welding spectrum of copper, from almost underrated to almost overrated force, Travel

speed levels aimed to represent both cold and hot welds and finally the Probe Length levels try to

represent undersized and oversized lengths. Nominal values can be found in Table G.2 of the

confidential annex G.

Figure 7.12 shows both face and root sides of the nine trials performed on copper. In Annex

B2 can be seen a closer look of each weld. Table 7.6 summarizes the visual analysis of such welds.

Table 7.6 – Visual analysis of copper welds.

Trial

Burr Weld Bead (face) Weld Bead (root)

Quantity Type Width Striate Hollow Depression Bond

None

Few

A lo

t

Contin

uous

Gra

nula

r

Regu

lar

Varia

ble

Regu

lar

Varia

ble

None

Sm

ooth

None

Sm

ooth

Pro

no

unced

Yes

Doubtfu

l

1 X X X X X X X

2 X X X X X X X

3 X X X X X X X

4 X X X X X X X

5 X X X X X X X

6 X X X X X X X

7 X X X X X X X

8 X X X X X X X

9 X X X X X X X

Figure 7.12 – Overall look of the face (left) and root (right) sides of the nine trials of copper.

48

7.2.4. Tensile Tests Results

Copper’s tensile test results were handled similarly to what was explained in section 4. The

five physical properties’ values as well as their weighing of base material have been compiled into

Table 7.7. In the last column it is shown the corresponding values of the GET factor plotted in the left

chart of Figure 7.13.

Table 7.7 – Summary of results for tensile tests of copper.

Base Material 𝑬𝑩𝑴 [GPa] 𝟎.𝟐𝑩𝑴[MPa] 𝒎𝒂𝒙𝑩𝑴[MPa] 𝑴𝑻𝑩𝑴 [J/mm

3] 𝑩𝑴 (%) 𝑮𝑬𝑻𝑴𝑩

139.50 75.28 305.97 64.45 35.50 1.00

Trials 𝑬𝒊 𝑬𝒊𝑬𝑩𝑴

𝟎.𝟐𝒊 𝟎.𝟐𝒊𝟎.𝟐𝑩𝑴

𝒎𝒂𝒙𝒊 𝒎𝒂𝒙𝒊𝒎𝒂𝒙𝑩𝑴

𝑴𝑻𝒊 𝑴𝑻𝒊

𝑴𝑻𝑩𝑴

𝒊 𝒊𝑩𝑴

𝑮𝑬𝑻𝒊

1 108.40 0.78 95.77 1.27 262.26 0.86 60.39 0.94 34.00 0.96 1.01

2 104.87 0.75 90.29 1.20 272.07 0.89 67.03 1.04 37.50 1.06 1.03

3 84.62 0.61 94.51 1.26 148.92 0.49 8.22 0.13 6.75 0.19 0.60

4 107.88 0.77 95.83 1.27 270.44 0.88 67.49 1.05 37.25 1.05 1.06

5 128.76 0.92 90.26 1.20 100.04 0.33 3.35 0.05 3.50 0.10 0.55

6 115.08 0.82 94.33 1.25 257.33 0.84 50.07 0.78 28.50 0.80 0.94

7 126.38 0.91 92.29 1.23 265.59 0.87 67.14 1.04 38.00 1.07 1.06

8 105.34 0.76 92.77 1.23 216.95 0.71 25.60 0.40 16.50 0.46 0.76

9 109.65 0.79 93.23 1.24 171.52 0.56 12.59 0.20 9.50 0.27 0.66

The right chart of Figure 7.13 displays the GET factor per parameter level (right chart). From

its analysis it is clear that Force has little influence in the tensile results of copper weld. However it is

possible to identify some relations from the other two parameters. It seems that only the first level of

travel speed is able to perform consistent welds, because the remaining leads to cold conditions

welds. Level V3 leads to unstable welds probably because of anvil scratching. The optimum

parameters for maximizing GET factor are F1, V1 and L2. This factor demonstrates an inverse

proportionality with the axial force, travel speed and the second and third values of probe length.

1.01 1.03

0.60

1.06

0.55

0.94 1.06

0.76 0.66

0.00

0.20

0.40

0.60

0.80

1.00

1.20

1 2 3 4 5 6 7 8 9Trial

GET - Trials

0.88 0.85 0.83

1.04

0.78 0.73

0.91 0.92

0.74

0.00

0.20

0.40

0.60

0.80

1.00

1.20

F1 F2 F3 V1 V2 V3 L1 L2 L3

Parameter Level

GET - Parameters

Figure 7.13 – GET results for each trial (left) and parameters level (right).

49

7.2.5. Bending Tests Results

The three points bending results for copper where summarized in Table 7.8 and they reveal

very close results, when compared with the base material, with maximum load achieved in almost all

trials. The displacement values also show good results, being higher than 0.79 for eight of the nine

trials executed.

Table 7.8 – Summary of results for bending tests of copper

Base Material

𝑭𝑩𝑴 [kN] 𝒅𝑩𝑴[mm] 𝑬𝒏𝑩𝑴[J] 𝑮𝑬𝑩𝑴𝑩

1.16 6.96 880.09 1.00

Trials 𝑭𝒊 𝑭𝒊𝑭𝑩𝑴

𝒅𝒊 𝒅𝒊𝒅𝑩𝑴

𝑬𝒏𝒊 𝑬𝒏𝒊𝑬𝒏𝑩𝑴

𝑮𝑬𝑩𝒊

1 1.20 1.03 6.02 0.84 594.05 0.60 0.87

2 1.16 1.00 5.93 0.83 589.61 0.59 0.85

3 0.94 0.81 3.93 0.55 385.92 0.39 0.64

4 1.09 0.94 5.79 0.81 527.46 0.53 0.80

5 1.10 0.95 5.98 0.83 556.88 0.56 0.82

6 1.16 1.00 6.43 0.90 673.53 0.68 0.89

7 1.15 0.99 6.08 0.85 625.85 0.63 0.86

8 1.09 0.94 5.45 0.76 513.05 0.51 0.79

9 1.23 1.06 6.84 0.95 689.55 0.69 0.94

The GEB values per trial (left chart of Figure 7.14) show a very good consistency, except for

trial 3. Optimum parameters for this evaluation factor are the maximum values of each parameter

contribution in the right chart of Figure 7.14 to be precise F3, V1 and L2. In all parameters and levels

the results are very similar, varying only from 0.78 to 0.87. However it is possible to notice the

proportionality of the GEB to the axial force, and again the poor resistance of L3.

0.87 0.85

0.64

0.80 0.82 0.89 0.86

0.79

0.94

0.00

0.10

0.20

0.30

0.40

0.50

0.60

0.70

0.80

0.90

1.00

1 2 3 4 5 6 7 8 9Trial

GEB - Trials

0.79 0.84 0.86 0.85 0.87

0.78 0.85 0.82 0.82

0.000.100.200.300.400.500.600.700.800.901.00

F1 F2 F3 V1 V2 V3 L1 L2 L3

Parameter Level

GEB - Parameters

Figure 7.14 – GEB results for each trial (left) and parameters level (right).

50

7.2.6. Hardness Tests Results

Hardness Tests of copper butt welds are summarized in Table 7.9 and plotted in the left chart

of Figure 7.15 and reveal unexpected values for trial 1 in which all comparative values were much

smaller than the remaining ones. The remaining trials resulted in very homogeneous hardness

coefficients, varying just from 0.97 to 1.05 of base material. However, the influence of each parameter

level on the hardness coefficient showed a very good behavior being proportional to each level,

resulting in F3, V3 and L3 optimum values, as shown in the right chart of Figure 7.15.

Table 7.9 – Summary of results for hardness tests of copper

Base Material 𝑴𝒊𝒏 [HV] 𝑨𝒗𝒈[HV] 𝑴𝒂𝒙[HV] 𝑯𝒂𝒓𝒅𝑴𝑩

93.2 103.6 110.6 1.00

Trials 𝑴𝒊𝒏𝒊 𝑨𝒗𝒈𝒊 𝑴𝒂𝒙𝒊 𝑯𝒂𝒓𝒅𝒊

1 76.3 87.7 95.3 0.85

2 97.4 104.7 114.5 1.01

3 97.8 105.7 113.9 1.02

4 98.5 105.5 111.3 1.02

5 94.0 100.3 109.4 0.97

6 102.5 107.5 115.1 1.04

7 103.1 109.2 119.7 1.05

8 99.9 104.6 110.2 1.01

9 94.0 100.5 107.7 0.97

7.2.7. Analysis of Variance

From the results of the three mechanical tests performed and which have been addressed

previously, it was expected that the variance analysis would result in large values for error, due

especially to unexpected values of trial 3 on bending, and trial 1 on hardness. Table 7.10 shows the

degrees of freedom (DOF), the sum of squared deviations (SS), the means square variance (MS) and

the F-test (F) for each evaluation factor and contribution (ρ) of each process parameter to the final

outcome of each test.

0.85

1.01 1.02 1.02 0.97

1.04 1.05 1.01 0.97

0.00

0.20

0.40

0.60

0.80

1.00

1.20

1 2 3 4 5 6 7 8 9

Trial

Hardness - Trials

0.96 1.00 1.01 0.97 1.00 1.01 0.96 1.01 1.01

0

0.2

0.4

0.6

0.8

1

1.2

F1 F2 F3 V1 V2 V3 L1 L2 L3

Parameter Level

Hardness - Parameters

Figure 7.15 – Hardness Coefficient results for each trial (left) and parameters level (right).

51

1.36%

46.71%

17.60%

34.33%

15.00%

2.09%

24.30%

58.60%

16.91%

6.66%

12.86%

63.56%

Force

Speed

Probe

Error

Hardness

GEB

GET

Table 7.10 – Results of variance analysis for the three evaluation parameters for copper.

Parameter D

OF

SS V F ρ(%)

GET GEB Hard GET GEB Hard GET GEB Hard GET GEB Hard

Fz 2 0.0047 0.0088 0.0051 0.0024 0.0044 0.0026 0.0395 0.2560 0.2661 1.36 15.00 16.91

Vx 2 0.1629 0.0012 0.0020 0.0815 0.0006 0.0010 1.3604 0.0357 0.1049 46.71 2.09 6.66

Lpin 2 0.0614 0.0142 0.0039 0.0307 0.0071 0.0019 0.5127 0.4147 0.2024 17.60 24.30 12.86

Error 2 0.1198 0.0343 0.0192 0.0599 0.0171 0.0096 - - - 34.3 58.60 63.56

Total 8 0.3488 0.0585 0.0302 - - - - - - 100 100 100

The contribution of each parameter for the evaluation parameters is shown in Figure 7.16. As

expected the contribution of Error is higher than the contribution of the process parameters both for

GEB and for Hardness Coefficient. For the GET factor, despite the fact that travel speed has a higher

contribution with 46.71%, the error still has a very significant contribution with 34.33% exceeding both

probe length and force together.

7.2.8. Optimum parameters identification

As in aluminum, in order to obtain the set of parameters that maximize the mechanical

response of the three evaluating parameters it was used the algorithm of equation (7.3). With the

weighing factors and the optimum values for each parameter calculated in the previous sections.

[

𝑭𝑮𝑬𝑻 𝑭𝑮𝑬𝑩 𝑭𝑯𝒂𝒓𝒅𝑽𝑮𝑬𝑻 𝑽𝑮𝑬𝑩 𝑽𝑯𝒂𝒓𝒅𝑳𝑮𝑬𝑻 𝑳𝑮𝑬𝑩 𝑳𝑯𝒂𝒓𝒅

]

[ 𝝆𝑭(𝑮𝑬𝑻)

𝝆𝑭(𝑻)

𝝆𝑽(𝑮𝑬𝑻)

𝝆𝑽(𝑻)

𝝆𝑳(𝑮𝑬𝑻)

𝝆𝑳(𝑻)𝝆𝑭(𝑮𝑬𝑩)

𝝆𝑭(𝑻)

𝝆𝑽(𝑮𝑬𝑩)

𝝆𝑽(𝑻)

𝝆𝑳(𝑮𝑬𝑩)

𝝆𝑳(𝑻)𝝆𝑭(𝑯𝒂𝒓𝒅)

𝝆𝑭(𝑻)

𝝆𝑽(𝑯𝒂𝒓𝒅)

𝝆𝑽(𝑻)

𝝆𝑳(𝑯𝒂𝒓𝒅)

𝝆𝑳(𝑻) ]

= [𝑳𝒐𝒂𝒅 𝒙 𝒙𝒙 𝑽𝒆𝒍𝒐𝒄𝒊𝒕𝒚 𝒙𝒙 𝒙 𝑷𝒓𝒐𝒃𝒆

] (7.3)

Equation (7.4) expresses the results:

[𝑭𝟏 𝑭𝟑 𝑭𝟑𝑽𝟏 𝑽𝟏 𝑽𝟑𝑳𝟐 𝑳𝟐 𝑳𝟑

] [𝟎. 𝟎𝟒 𝟎. 𝟖𝟒 𝟎. 𝟑𝟐𝟎. 𝟒𝟓 𝟎. 𝟎𝟒 𝟎. 𝟒𝟒𝟎. 𝟓𝟏 𝟎. 𝟏𝟐 𝟎. 𝟐𝟑

] ≅ [𝑭𝟑 𝒙 𝒙𝒙 𝑽𝟏 𝒙𝒙 𝒙 𝑳𝟐

] (7.4)

The nominal values of the parameters can be found in the confidential annex H.

Figure 7.16 – Contribution of each parameter for the three evaluation parameters.

52

7.2.9. Summary of Copper Butt Welding Results

The optimum set of values for each evaluation factor was defined, the Global Efficiency on

Tensile, the Global Efficiency on Bending and the Hardness Coefficient and the set of values that

maximize the global response to all those factors. Since the inherent error to this results looks too

high, so it is mandatory to conduct a confirmatory trial.

Tensile results revealed that the optimum set of parameters for Global Efficiency on Tensile

were F1, V1 and L2. With the travel speed being the most important parameter followed by error. The

three Point Bending results showed a much larger dependence of error than from the tested

parameters, the optimum parameters were F3, V1 and L2. The hardness coefficient had, again, an

error value superior than for the remaining factors the axial force, the travel speed and the probe

length. Anyway the best results were obtained with F3, V3 and L3 parameters levels.

This study concluded that the optimum set of parameters equally weighted by the three factors

were Axial Force level F3, V1 of travel speed and a probe length with L2.

Confirmatory trials have been made using the optimum parameters and submitted to

mechanical and electrical conductivity characterization tests. The mechanical results on copper

confirmatory trials were 0.67 GET, 0.77 GEB and 0.97 HARD, which gives a reduction of about 20%

from the base material. Copper’s conductivity test, Figure 7.17, revealed a slight increment of 0.5%

IACS in the region between TMAZ and HAZ in the advancing side of the weld. This increase can be

explained by a possible growth of the grain size in this region.

Figure 7.17 – Eddy current conductivity test performed,

in two arbitrary paths (L1, L2),at 250 kHz, of copper’s confirmatory trial.

53

7.3. Overlap Foil-Bar weld

Foil-Bar welding was another geometry required from the client, those welds are essential and

irreplaceable in the present foil winding manufacture. For each foil winding, two mandatory foil-bar

welds are performed, to the entrance bar and end-bar.

Figure 7.18 shows both macro and micrographic photos of the two similar welds presently

performed in SIEMENS-FS in TIG weld. As shown in macro, the HAZ is very large. It can also be seen

that big differences exist in the microstructure with larger grain size in the affected zone. Both welds

evidence initial cracking sites that were highlighted.

Preliminary trials were done to find an adequate range of parameters for these welds. As

stated in Section 2.1 those components are not subjected to high mechanical stresses. So there was

no necessity to perform any optimization, neither any characterization test. As the objective of this

preliminary study was only to achieve a range of parameters that could perform acceptable welds. The

geometry used was overlapped. In theory this weld could also be done in a corner joint in order to

save material, using a side tilt angle.

Figure 7.18 – TIG Foil-Bar weld in aluminum at left, and in copper at right.

Full macros at the top and micrographic photo of initial cracking sites, at bottom.

54

7.3.1. Aluminum Foil – Aluminum Bar Weld

Table 7.11 – Aluminum Foil-Bar

weld parameters range.

Several tool geometries were tested, such as double and single scrolled or smooth concave

shoulders, with conical and cylindrical probes. Tilt angles have been varied from 0º with scrolled

shoulders to a maximum of 2.5º with smooth concave. Different probe lengths and diameters were

also tested. Substantial probe lengths were used with almost the double of the foil thickness (3mm

probe length to a 1.6mm thickness). Because of this generous length M3 probes were automatically

discarded, both M4 and M5 probes were tested.

Finally a range of welding parameters was obtained as summarized in Table 7.11. Those

parameters were obtained for a conical M5 probe that proved to be steadier than cylindrical or M4

ones. The shoulder used was a double scrolled with 0.5 pace. A 0º tilt angle was used, Dt was set to

3s and the Ps to 0.2 mm/s. Figure 7.19 shows a weld performed with those parameters.

7.3.2. Copper Foil – Copper Bar Weld

After experiencing no technical difficulties with the smooth concave shoulder and a tilt angle of

2.0º in copper’s butt welding, copper’s Foil-Bar weld was also made with this geometry, using the

same cylindrical probe. Again as for aluminum, the goal was to achieve material mixture in the joint, so

a probe length with almost the double of the foil thickness (2mm probe length for 1.1mm thick) was

used. The M4 probe was used with suitable results. Table 7.12 shows the parameter window that led

to acceptable outcomes from the visual point of view. An example of such welds is shown in Figure

7.20.

Parameter Min Max

Fz (Kg) 200 300

Vx (mm/min) 150 300

Ω (rpm) 1000 1200

Figure 7.19 – Aluminum Foil-Bar weld example.

55

Table 7.12 – Copper Foil-Bar weld

parameters range.

7.3.3. Aluminum Foil – Copper Bar weld

The aluminum foil – copper bar weld was by far the more challenging combination of all foil-

bar welds performed. The softness of aluminum foil is too high to support the load needed to weld

copper, regardless of the constraining system tightness, the load applied or the heat generated. Figure

7.21 evidence some of those superficial defects. No satisfactory trials were performed although it

seems that it can be done with a more intensive study, with different tool geometries. Increasing

shoulder diameter to avoid aluminum blow-off or shorter pin lengths to prevent vertical instabilities can

be tested for solve this problem.

Figure 7.22 demonstrate the approach used to achieve satisfactory welds to this joint. This

approach involved overlap a copper foil over the aluminum, and then weld together both foils and bar

in a sandwich like form. The parameters of Table 7.13 and a Tool with M5 conical probe (3.5mm

length) and a smooth concave shoulder were used to achieve some good welds from the visual point

of view. A tilt angle with two degrees was used alongside with a 5s dwell time.

Parameter Min Max

Fz (Kg) 700 1000

Vx (mm/min) 150 250

Ω (rpm) 800 1200

Figure 7.21 – Aluminum foil – copper bar weld instabilities.

Figure 7.20 – Copper Foil-Bar weld example.

56

Table 7.13 – Cu-Al-Cu Sandwich like weld parameters range.

7.3.4. Summary of Results

The overlap Foil-Bar weld was performed in similar and dissimilar materials with satisfactory

results, despite the significant differences between physical properties of foil and bars. Several tool

geometries and welding parameters have been tested and defined.

It was acknowledged that M5 cylindrical probe with a double scrolled shoulder is the most

suitable of the available geometries for Aluminum Foil – Aluminum Bar weld. For Copper Foil –

Copper Bar weld was used a smooth concave shoulder with a M4 cylindrical probe. And finally for

dissimilar materials was used a M5 conical probe and a smooth concave shoulder.

The main difficulties of those welds lay on the significant differences in the hardness values

between foil and bars. Those differences are even more pronounced in the aluminum foil – copper bar

joint. A sandwich like setup was used to constrain aluminum foil and enable a good weld.

A deeper study must be done on Foil-Bar welds, in order to quantify the quality of the

performed welds. Mechanical tests need to be adjusted to the geometry of the samples and

conductivity tests must be run.

Parameter Min Max

Fz (Kg) 450 600

Vx (mm/min) 180 210

Ω (rpm) 800 1000

Figure 7.22 – Cu-Al-Cu Sandwich like weld example.

57

7.4. Other Geometries

7.4.1. Aluminum-Copper Butt welding

As explained in section 2.6.2 this geometry, aluminum-copper butt weld with such thin foils

and especially with dissimilar thicknesses could not be found in bibliography. Theoretically it’s possible

to perform this weld. The problems rely on the difference in the two materials which prevents a good

mixture and the thickness difference that avoids a good constraining to the system (Figure 7.23). The

latter problem can be solved with a correct side tilt angle. However the existing work table is not able

to perform small angle variations. Some new designs have been studied and can be seen in Annex Z,

yet none have been produced, needing further study on feasibility.

According to expertise and

bibliography the harder material should be

placed in the advancing side of the weld.

During this study side tilt angles of 2.5º, 3º and

3.7º were tested and the best results seem to

be found in the latter case. It’s also possible to

confine the weldable parameters in the ranges

of 200-450 Kg, 150-250 mm/min and 800-1200

rpm respectively for axial force, travel speed

and rotational speed.

7.4.2. Aluminum-Copper overlap welding

After experiencing the technical

difficulties inherent to aluminum-copper butt

welding, another possibility was tested.

Aluminum-Copper overlap welding was

performed with relative success. Placing

copper on top of aluminum allowed a better

constraining of aluminum, Figure 7.24.

7.4.3. Thin Copper-Copper butt weld

A thinner copper foil with just 0.4mm thick was also tested. With 0.5mm the shoulder scrolls

were thicker than the foil himself. Besides that, due to the small section any small instability could lead

to a disruption. Small loads and speeds were tested, hot and cold conditions also. However, the

outcome was always the same, Figure 7.25, unavoidable breakdown and not once proper welding

conditions were achieved.

Figure 7.24 – Aluminum-Copper overlap weld example.

Figure 7.25 – Examples of thin copper trials.

Figure 7.23 – Aluminum-Copper butt weld instabilities.

58

59

8. Preliminary Feasibility Study

This study is divided in 4 separated parts. In the first part, the production needs of SIEMENS-

FS are listed. The second part explains the operating costs of both processes (TIG and FSW). The

third part shows the initial investment to purchase, optimize and implement the FSW machine. The

fourth part enumerates the non-quantifiable variables related to the Quality of the process.

8.1. The Client Needs

In this section are shown the actual needs of the client, SIEMENS-FS, in Table 8.1. The

number of weld paths has been estimated from the expected number of transformers. For an

estimated annual production of 500 transformers a total of 3000 weld paths are mandatory. This

because there are two welds per bobbin and each transformer has three bobbins.

The weld length and the thickness vary from about 850 mm and 1.1 mm thick for copper welds

to 1100 mm and 1.6 mm thick for aluminum welds. The current production of aluminum coil

transformer in SIEMENS-FS is small but with the increase of copper’s price and the improvement of

aluminum products it is expected that aluminum demand will exceed copper’s. It was used a ratio of 1

aluminum transformer for each 5 produced.

Table 8.1 – Production variables for the winding manufacturing.

Needs Qty units

Transformers/year 500 unit

Unit/year 1500 unit

Welds/year 3000 unit

Weld length variable m

Thickness of base sheets variable mm

Number of weld passes 1 unit

Ratio aluminum/copper weld 0.20 -

60

8.2. Operating costs

In order to estimate the operating costs, the models introduced by Tipaji [61] to calculation and

comparison of FSW and TIG costs. All calculations and assumptions used in this work can be

consulted in the annex F.

Table 8.2 – Estimated costs per weld for both processes.

Operating Cost TIG FSW

Labor cost 9.23 € 5.07 €

Machine cost 0.33 € 5.00 €

Tool cost 0.20 € 0.44 €

Gas cost 8.40 € 0.00 €

Power cost 1.67 € 0.27 €

Total cost 19.83 € 10.78 €

As it can be seen in the summary Table 8.2 TIG operating costs per weld is almost double the

cost of FSW. This is due to the high cost of the shielding gas and a slight reduction on weld time. Pie

diagrams in Figure 8.1 show the comparative costs of the current solution and the FSW process. Each

process has two big contributors, Gas and Labor costs for TIG and Machine and Labor costs for FSW.

Figure 8.1 – Costs division for TIG and FSW weld.

8.3. Initial Investment and Payback

The initial investment, Table 8.3, is the area of this study that lacks more confirmation, due to

all the values being estimated by excess. The first phase is already in motion. SIEMENStir is a

partnership between SIEMENS-FS and IDMEC-IST and involves two Master Degree proposers, so

the money invested for this preliminary study is acknowledged as sunk cost and it’s assumed to be nil.

Design and Development was envisioned to support the wage and travel expenses of someone to

follow all the Design and Development of the project, for a minimum of 12 months. Machinery costs

and Locksmith costs are self-explained, and are the biggest part of the investment. This particular

investment cannot be neglected since this is a precision process.

Labor cost 47%

Machine cost 2%

Tool cost 1%

Gas cost 42%

Power cost 8%

TIG

Labor cost 47%

Machine cost 46%

Tool cost 4%

Power cost 3%

FSW

61

Table 8.3 – Estimated costs for the project.

Initial Investment

Study 0 €

Design and Development 18,000.00 €

Machine cost 150,000.00 €

Locksmith cost 25,000.00 €

Tests 5,000.00 €

Staff training 7,000.00 €

Total cost 205,000.00 €

The Payback period, Figure 8.2, was estimated to about four and an half years. It was used

the simplest approach to obtain this estimation. The cost of each process was multiplied by the

production and added the initial investment. The intersection of both curves is considered the payback

period, that means the moment FSW cost equals the cost of TIG.

This study only takes into account the welds that the present process is able to do. If it’s taken

in consideration the amount of welds that FSW can add to production, and all the versatility that it

gives, it’s clear that the real payback period is smaller. Also with good use of financial mathematics it

is possible to decrease even more this period.

Figure 8.2 – Estimated Payback Period.

0

100000

200000

300000

400000

500000

600000

700000

0 1 2 3 4 5 6 7 8 9

Co

st

(€)

Years

Payback Period

TIG FSW

62

8.4. Quality

In this final section have been exposed some proprieties that couldn’t be quantifiable. The

purpose of Table 8.4 is to help the display of the real advantages that FSW will bring, when compared

with the current process. Some properties are related to the environment and safety, others show the

versatility of the process and last but not least there are the mechanical properties, which have been

previously reported.

Table 8.4 – Comparison of non-quantifiable variables of both processes.

Property TIG FSW

Environment risk High Low

smoke /toxic gas Very High Very Low

UV emissions Very High X

Noise High Low

Fire hazard √ X

Protection Gear √ X

Operator Training High Low

2D work-plain CAD-CAM Available Implicit

Different geometries X √

Welding of dissimilar alloys X √

Repeatability High Very High

Date log X √

Porosity High Low

Distortion High Low

Contamination of joints High Low

Mechanical Proprieties of joints Base Mat. Stronger

Energy consumption High Low

Recycle possibility Low High

Heating - Achieved temp. Very High Low

Al welding thin sheets X √

Initial investment Low High

Constrains precision Low High

8.5. Conclusions on Feasibility Study

In this preliminary study, it was analyzed the payback of the investment needed and the fixed

costs of each solution, the standard being the TIG Welding and proposed solution, the use of FSW.

It’s clear that from the quality, energy consumption and work condition aspects, the FSW

process is much better than the standard process used nowadays. The initial investment is high, albeit

this opens the opportunity to improve the quality of the process and creates the possibility of welding

dissimilar materials together, as Al and Cu.

63

9. Conclusions

This essay intended to extend knowledge about friction stir welding of electrical components in

pure aluminum and copper, by analyzing different geometries and optimizing their process

parameters. This was accomplished through the Taguchi analysis of butt welds in aluminum and

copper thin sheets, of 1.6 and 1.1mm thick, respectively. Parameters and tools geometry have also

been investigated for Foil-Bar welds and Al/Cu dissimilar weld.

The Taguchi Method has been used for butt welding supported by mechanical tests results of

tensile, three point bending and hardness. Using the evaluation factors GET, GEB and Hardness and

a variance analysis it was possible to conclude the optimum parameters to weld each material.

Confirmatory trials in Aluminum Butt welds revealed a slight decrease of mechanical properties from

the base material of 14% and a decrease in conductivity of 4%, the model fitted perfectly with small

error influence. Copper Butt welds revealed a decrease in mechanical properties of 20% and a slight

increase in conductivity. Also copper welds were much more influenced by external errors.

Overlapped Foil-Bar welds have been performed with satisfactory results, tool geometry and

welding parameters have been defined with success. Similar welds have been achieved without any

particular challenge. On the other hand, dissimilar weld of aluminum foil to copper bar was hard to

achieve due to the components hardness’ difference, a sandwich like weld has been attempted with

success. Dissimilar butt welds have been also study and a similar approach has been used to avoid

the discontinuity in the weld bead. This discontinuity is mainly due to the different thicknesses of the

two materials.

A preliminary feasibility study was built based on the actual needs of the client to estimate the

operating cost of TIG and FSW. It was acknowledged that the main contributors to the TIG cost per

weld were the gas and labor cost. In the case of FSW, machine cost and labor cost represent the

bigger part of a cost that still represents only half of copper’s. Based on the cost per weld of both

processes and using the estimate production (per year) of the factory, it was also calculated the

payback period of a project, for implementation of FSW solution. The estimated time of payback was

of about four and an half years.

The final conclusion is that the implementation of the FSW is perfectly suitable for the

SIEMENS FS status quo, because FSW can perform the same amount of welds currently done using

the TIG with the surplus of enabling the production of more profitable joints, with no loss of reliability

and quality standards.

64

65

10. Future Work

During the development of this work, several ideas occurred from the need to adjust the

technical approach and move towards the best solution, mainly due to the complex and unknown field

of FSW application on conductive materials. Most of them were could not be performed along this

work, however were documented in this report as hints and described in this chapter.

10.1. Dissimilar butt welding

As described on Chapter 7.4.1 the tests have shown that the welding of dissimilar materials

was difficult. This is due mainly to the soft aluminum that cannot receive the same load as Copper.

Additionally, the different and small thickness increase the complexity of the welding. Several

procedures were tried to reach a reasonable weld bead, with different tools geometries and several tilt

angles tested. One possibility was left out due to the equipment limitation of using tilt and side tilt

angles simultaneously. To solve this problem it was created a concept for an oscillating table that

needs to guarantee its stability and stiffness during the welding process.

An alternative solution was tested to enable the connection of these different materials, an

overlapped arrangement where the copper was placed above, supporting most of the load and

restricting the aluminum below. This is a technical approach that must be studied and tested in detail.

10.2. Foil-Bar Quality Tests

The weld quality is generally proven by mechanical tests, such as Tensile and Bending, as

done in this work for butt welds. However for Foil-Bar geometry, standard tests could not be used. In

this case, the significant differences between the thickness of the components leads to asymmetry,

which means that the axial strength test will be challenging, and there will be a distortion and constrain

difficulty. For instance it could be applied a Taguchi study adapted to tests that can be performed such

as hardness, conductivity or fatigue for example.

10.3. Static Shoulder/Pinless Tool

Because of the reduced thickness of the foils used, especially those with 0.4mm in copper,

some more exoteric and new solutions can be tested. A static shoulder could enhance the joint quality

of similar butt welds and a pinless tool with a shallower striate shoulder could potentiate the welding of

thinner foils.

66

67

XIII. References

[1] P. Vilaca, J. Mendes, E. Santis, J. Prior, and L. Quintino, “Application of FSW to join aluminium foil windings for electrical transformers,” presented at the 6th IIW Annual Assembly, Dusseldorf, Germany, 2013, p. 18.

[2] D. Allen and H. Moore, “Theory and Principles,” in Electric Power Transformer Engineering, CRC Press, 2004, p. 10.

[3] REN - Rede Eléctrica Nacional S.A., “Technical Report on Electricity 2012.” 02-Apr-2013.

[4] DOE, DOE Fundamentals Handbook - Electrical Science, vol. Volume 4. Washington, 1992.

[5] J. A. B. Faria, Electromagnetic Foundations of Electrical Engineering. John Wiley & Sons, 2008.

[6] C. R. Sullivan, “Aluminum Windings and Other Strategies forHigh-Frequency Magnetics Design in anEra of High Copper and Energy Costs,” IEEE Transactions on Power Electronics, vol. 23, no. 4, pp. 2044–2051, 2008.

[7] P. Paul, M. Sacotte, and F. Walter, “Large aluminium distribution transformer,” in 2012 1st International Conference on Power and Energy in NERIST (ICPEN), 2012, pp. 1–5.

[8] W. M. Thomas, E. D. Nicholas, J. C. Needham, M. G. Murch, and C. J. Temple-Smith, “Improvements relating to friction stir welding,” US Patent N

o.54603171991.

[9] TWI, “List of FSW patents & applications - Intellectual Property Rights,” 2013. [Online]. Available: http://www.twi.co.uk/services/intellectual-property-licensing/friction-stir-welding/intellectual-property-rights/. [Accessed: 13-Aug-2013].

[10] M. P. Scott, A. F. Richard, J. S. Russell, E. R. Monte, and E. T. Brian, “Solid State Processing of Materials Through Friction Stir Processing and Friction Stir Mixing.,” WO Patent N

o.

0945412005.

[11] J. Bersaas, A. Oosterkamp, and L. Oosterkamp, “Friction Stir Spot Welding Method and Apparatus,” WO0128732 (A1)26-Apr-2001.

[12] R. S. Mishra, “Integral channels in metal components and fabrication thereof,” US Patent N

o.69233622005.

[13] S. W. Kallee, “Application of friction stir welding in the shipbuilding industry,” presented at the Lightweight Construction - Latest Developments, London, UK, 2000.

[14] S. W. Kallee, E. D. Nicholas, and W. M. Thomas, “Industrialisation of friction stir welding for aerospace structures,” presented at the Structures and Technologies - Challenges for Future Launchers Third European Conference, Strasbourg France, 2001.

[15] S. W. Kallee, J. Davenport, and E. D. Nicholas, “Railway rolling stock manufacturers implement friction stir welding,” Welding Journal, Oct-2002.

[16] W. M. Thomas, S. W. Kallee, D. G. Staines, and P J Oakley, “Friction Stir Welding - Process variants and developments in the automotive industry,” presented at the 2006 SAE World Congress, Cobo Center, Detroit, Michigan, USA., 2006.

[17] P. Vilaca, J. Gandra, and C. Vidal, “Linear Friction Based Processing Technologies for Aluminum Alloys: Surfacing, Stir Welding and Stir Channeling,” in Aluminium Alloys - New Trends in Fabrication and Applications, Z. Ahmad, Ed. InTech, 2012.

[18] C. B. Fuller, “Friction Stir Tooling: Tool Materials and Designs,” in Friction Stir Welding and Processing, R. S. Mishra and M. W. Mahoney, Eds. ASM International, 2007, pp. 7–35.

[19] P. Vilaca and W. Thomas, “Friction Stir Welding Technology,” in Structural Connections for Lightweight Metallic Structures, Springer, 2012.

[20] P. Vilaca, L. Quintino, and J. dos Santos, “iSTIR - Analytical thermal model for friction stir welding,” Journal of Materials Processing Technology, vol. 169, no. 3, pp. 452–465, 2005.

[21] R. S. Mishra and M. W. Mahoney, “Introduction,” in Friction Stir Welding and Processing, ASM International, 2007, pp. 1–5.

68

[22] W. F. Smith, “Ligas Metálicas de Engenharia,” in Princípios de ciência e engenharia de materiais, 3rd ed., McGraw-Hill, 1998.

[23] J. R. Davis, Aluminum and Aluminum Alloys. ASM International, 1993.

[24] AWS, “Aluminum and Aluminum Alloys,” in Welding Handbook, Volume 3 - Materials and Applications, Part 1, 8th ed., 1996, pp. 1–120.

[25] J. R. Davis, Copper and Copper Alloys. ASM International, 2001.

[26] “UNS Standard Designation for Wrought and Cast Copper.” [Online]. Available: http://www.copper.org/resources/properties/standard-designations/introduction.html. [Accessed: 21-May-2013].

[27] AWS, “Copper and Copper Alloys,” in Welding Handbook, Volume 3 - Materials and Applications, Part 1, 8th ed., 1996, pp. 163–216.

[28] R. M. Leal, “Soldadura por fricção linear de elementos de baixa espessura - ligações homogéneas e heterogéneas,” Coimbra, 2010.

[29] G. Taguchi, Taguchi methods, research and development. American Supplier Institute, 1992.

[30] R. K. Roy, Design of Experiments Using The Taguchi Approach. John Wiley & Sons, 2001.

[31] J. Nahas, “Statistics and Design of Experiments,” presented at the Short Course Fall 2012, University of Notre Dame, 2012.

[32] D. C. Montgomery, Design and analysis of experiments. Hoboken, NJ: John Wiley & Sons, Inc., 2013.

[33] W. M. Thomas and E. D. Nicholas, “Friction stir welding for the transportation industries,” Materials & Design, vol. 18, no. 4, pp. 269–273, 1997.

[34] A. Scialpi, M. De Giorgi, L. A. C. De Filippis, R. Nobile, and F. W. Panella, “Mechanical analysis of ultra-thin friction stir welding joined sheets with dissimilar and similar materials,” Materials & Design, vol. 29, no. 5, pp. 928–936, 2008.

[35] C. Leitao, R. M. Leal, D. M. Rodrigues, A. Loureiro, and P. Vilaça, “Mechanical behaviour of similar and dissimilar AA5182-H111 and AA6016-T4 thin friction stir welds,” Materials & Design, vol. 30, no. 1, pp. 101–108, Jan. 2009.

[36] R. M. Leal, C. Leitao, A. Loureiro, D. M. Rodrigues, and P. Vilaca, “Material flow in heterogeneous friction stir welding of thin aluminium sheets: Effect of shoulder geometry,” Materials Science and Engineering a-Structural Materials Properties Microstructure and Processing, vol. 498, no. 1–2, pp. 384–391, 2008.

[37] D. M. Rodrigues, A. Loureiro, C. Leitao, R. M. Leal, B. M. Chaparro, and P. Vilaca, “Influence of friction stir welding parameters on the microstructural and mechanical properties of AA 6016-T4 thin welds,” Materials & Design, vol. 30, no. 6, pp. 1913–1921, 2009.

[38] I. Topic, H. W. Höppel, and M. Göken, “Friction stir welding of accumulative roll-bonded commercial-purity aluminium AA1050 and aluminium alloy AA6016,” Materials Science and Engineering: A, vol. 503, no. 1–2, pp. 163–166, Mar. 2009.

[39] Y. S. Sato, Y. Kurihara, S. H. C. Park, H. Kokawa, and N. Tsuji, “Friction stir welding of ultrafine grained Al alloy 1100 produced by accumulative roll-bonding,” Scripta Materialia, vol. 50, no. 1, pp. 57–60, Jan. 2004.

[40] I. Galvão, R. M. Leal, D. M. Rodrigues, and A. Loureiro, “Influence of tool shoulder geometry on properties of friction stir welds in thin copper sheets,” Journal of Materials Processing Technology, vol. 213, no. 2, pp. 129–135, Feb. 2013.

[41] R. M. Leal, N. Sakharova, P. Vilaca, D. M. Rodrigues, and A. Loureiro, “Effect of shoulder cavity and welding parameters on friction stir welding of thin copper sheets,” Science and Technology of Welding & Joining, vol. 16, no. 2, pp. 146–152, 2011.

[42] K. Savolainen, “Friction Stir Welding of Copper and Microstructure and Properties of the Welds,” PhD, Aalto, 2012.

69

[43] E. T. Akinlabi, “Effect of Shoulder Size on Weld Properties of Dissimilar Metal Friction Stir Welds,” JOURNAL OF MATERIALS ENGINEERING AND PERFORMANCE, vol. 21, no. 7, pp. 1514–1519, Jul. 2012.

[44] I. Galvão, D. Verdera, D. Gesto, A. Loureiro, and D. M. Rodrigues, “Influence of aluminium alloy type on dissimilar friction stir lap welding of aluminium to copper,” Journal of Materials Processing Technology, vol. 213, no. 11, pp. 1920–1928, Nov. 2013.

[45] V. Firouzdor and S. Kou, “Al-to-Cu Friction Stir Lap Welding,” Metall and Mat Trans A, vol. 43, no. 1, pp. 303–315, Jan. 2012.

[46] M. Koilraj, V. Sundareswaran, S. Vijayan, and S. R. Koteswara Rao, “Friction stir welding of dissimilar aluminum alloys AA2219 to AA5083 – Optimization of process parameters using Taguchi technique,” Materials & Design, vol. 42, pp. 1–7, Dec. 2012.

[47] C. Vidal, V. Infante, and P. Vilaca, “Assessment of Improvement Techniques Effect on Fatigue Behaviour of Friction Stir Welded Aerospace Aluminium Alloys,” Fatigue 2010, vol. 2, no. 1, pp. 1605–1616, 2010.

[48] A. Heidarzadeh, H. Khodaverdizadeh, A. Mahmoudi, and E. Nazari, “Tensile behavior of friction stir welded AA 6061-T4 aluminum alloy joints,” Materials & Design, vol. 37, pp. 166–173, May 2012.

[49] A. Bagheri, T. Azdast, and A. Doniavi, “An experimental study on mechanical properties of friction stir welded ABS sheets,” Materials & Design, vol. 43, pp. 402–409, Jan. 2013.

[50] Y. Bozkurt, “The optimization of friction stir welding process parameters to achieve maximum tensile strength in polyethylene sheets,” Materials & Design, vol. 35, pp. 440–445, Mar. 2012.

[51] S. H. Dashatan, T. Azdast, S. R. Ahmadi, and A. Bagheri, “Friction stir spot welding of dissimilar polymethyl methacrylate and acrylonitrile butadiene styrene sheets,” Materials & Design, vol. 45, pp. 135–141, Mar. 2013.

[52] P. Vilaca and T. G. Santos, “Ferramenta não Consumível Modular Ajustável e Refrigerável para Soldadura e Processamento por Fricção Linear,” PT Patent N

o.1040722008.

[53] NP EN 895, “NP EN 895:2002 - Ensaios destrutivos de soldaduras em materiais metálicos Ensaio de tracção transversal.” ISQ, 2002.

[54] W. F. Smith, “Propriedades Mecânicas dos Materiais Metálicos,” in Princípios de ciência e engenharia de materiais, 3rd ed., McGraw-Hill, 1998.

[55] DIN 40501-1, “DIN 40501-1:1985-06 - Aluminium for electrical purposes - E-Al plate, sheet and strip - Technical delivery conditions.” Deutsches Institut für Normung, 1985.

[56] ISO 6507-1, “ISO 6507-1:2005 - Metallic materials - Vickers hardness test - Part 1: Test method.” International Organization for Standardization, 2005.

[57] ISO 6361-2, “ISO 6361-2:2011 - Wrought aluminium and aluminium alloys - Sheets, strips and plates - Part 2: Mechanical properties.” International Organization for Standardization, 2011.

[58] BS EN 13601, “BS EN 13601:2002 - Copper and copper alloys - Copper rod, bar and wire for general electrical purposes.” British Standards, 2002.

[59] P. Vilaca, “Fundamentos do Processo de Soldadura por Fricção Linear: Análise Experimental e Modelação Analítica,” IST, 2003.

[60] T. G. Santos, R. M. Miranda, P. Vilaca, and J. P. Teixeira, “Modification of electrical conductivity by friction stir processing of aluminum alloys,” International Journal of Advanced Manufacturing Technology, vol. 57, no. 5–8, pp. 511–519, 2011.

[61] P. K. Tipaji, “E-design tools for friction stir welding: cost estimation tool,” 2007.

70

a

XIV. Annexes

b

A. Experimental Procedures

A1. Friction Stir Welding Procedures.

In this section the procedure adopted for the laboratory testing of friction stir welding is

described. These tests require an initial phase of preparation to the experimental work that will follow:

i. Acquire the material to test in sufficient quantity and at the same time.

ii. Cutting the foils to obtain sheets with the predetermined final dimensions, for the

welds described in Cap. 4 were used sheets with 200x100mm.

iii. Machining at least one side of the sheet to ensure complete perpendicularity between

the sides in order to perfect adjustment of the sheets, either between themselves or in

relation to the base plate of the fastening system. Followed by a slight grinding of

those surfaces to remove excess material.

iv. Clean equipment, including support desk where will be placed the fixing system.

v. Turn on and check the connectivity of both FSW machine and the computer of data

acquisition.

Performed the preparatory work described, follows the test procedure adopted:

1. Cleaning modular tool components, in particular the base and the pin that will be

used. Coating the bolts thread portions with Teflon, mount the tool and adjust the

length of the pin.

2. Fit and tight the tool in the equipment.

3. Start the cooling system and check its correct functioning.

4. Set the reference point in the z-axis.

5. Clean the sheets to be welded with acetone to degrease and remove any particles.

6. Place the sheets in the fixing system, ensuring the perfect fit with each other and

orienting joint line in accordance with the linear forward movement of the tool.

7. Tightening the clamping system to ensure the constraining of all movements of the

sheets.

8. Insert the parameters in the software of tool control.

9. Set the coordinates (x, y) of the start and end points of the weld path.

10. Start the software of data acquisition.

11. Perform the weld, by starting the tool movement.

12. Follow the weld process to ensure that everything is ok. Respect a secure distance to

the set until all movement cease.

13. Check the data read by the software and save it to further consultation

14. Extract the weld plate, mark and book it, with used parameters and observations of

defects. Classify the specimen and photograph them.

15. If this was not the last trial, repeat the procedure with a new pair of sheets.

c

A2. Procedures for Metallographic Analysis.

The experimental procedure followed for mounting metallographic samples were:

1. Cut and machine a sample from the welded plate with suitable, pre-determined

dimensions.

2. Grind sample edges to remove the burr.

3. Identifying each sample.

4. Mount samples in molds with a diameter appropriate to the size of the sample, with

cold forming resin.

5. Grind the samples with Sandpaper with following particle size: 600, 1000, 2400 and

4000 lubricated with water.

Notes: Advance in sandpaper after homogenizes the entire sample surface. Don't

reuse sandpaper of grinding harder materials. After each sandpaper, rinse the

sample, pass through alcohol and dry it.

6. Apply the samples to ultrasound, immersed in alcohol for a period of 5 minutes to

remove surface impurities.

7. Continue polishing the samples with cloths impregnated lubricant and diamond

powder with a grain size of 3 micrometer and 1 micrometer, for 5 minutes.

Notes: Whenever there is contamination of the cloth wash and brush it with detergent.

The time stated above must be seen as an approximated value, we must homogenize

the sample before move forward.

8. Repeat step 6.

9. Finish polishing, in a cloth impregnated with an OP-S solution for 1 minute.

10. Repeat step 6.

11. Contrasting the samples the appropriated reagent previously prepared. Note: For the

same reagent the time of contrasting is lesser for micrograph than for macrograph

analysis.

12. Use an optical microscope to capture several photomicrographs, with different

magnifications to illustrate the relevant aspects of each sample.

Reagents used:

For aluminum: 60% HCl + 30%HNO3 + 5% HF + 5% H2O (Poulton)

For copper: 50% HNO3 + 50% H2O (no specified name)

d

A3. Hardness Tests Procedures.

The test procedures adopted in the measurement of the hardness profile were as follows:

1. Adjust the sample dimensions to the sample holder of the equipment.

2. Flatten the sample through polishing its base.

3. Prepare the sample as defined in section B.2 until point 5.

4. Connect the equipment and check if it is calibrated.

5. Determine the minimum distance between indentation centers, according to ISO

6507-2.

6. Establish a plan of indentations.

7. Select, in the control panel of the equipment the option for the type of Vickers

hardness. Choose the appropriated load and indentation time for the material to test.

8. Identify and mark the center of the nugget. Choose a testing direction.

9. Focus the middle vertical distance of the surface with a 40x lens.

10. Change the lens to the indenter, and proceed with the indentation.

11. Shift the sample to the side in the direction chosen in point 8, at least the minimum

distance determined in point 5.

12. Repeat points 9 to 11, until all samples is traveled, since the nugget through the

processed zone until base material.

13. If only one direction as been cover, go back to point 8 shifting direction. If not, skip this

point.

14. Switch to the 40X objective and measure the two diagonals of each indentation.

15. Record the hardness values obtained.

16. If it is the last sample, remove sample and turn-off the equipment. If not, re-start the

procedure.

A4. Tensile Tests Procedures

There were tested three specimens for each condition of welding and three for the base

material analysis. The procedure used was the follow:

1. Cut and machine the specimens with dimensions according to the ISO 6892-1.

2. Grind the edges to remove the burr, but avoid reducing the thickness.

3. Use a marker to define the initial length of the specimens.

4. Introduce the geometric parameters of the specimen in the equipment control

software.

5. Place and grip the specimen to the testing machine. Adjust the extensometer, with a

starting distance of 50mm to the specimen gauge section.

6. Start the tensile test.

7. Wait until the rupture of the specimen.

8. Record the data acquired for further processing and analysis.

9. Measure the final extension.

10. Return to point 5 if there are more specimens to test.

e

A5. Procedures for three point Bending Test.

The test procedure used for the testing of three point bending was as follows:

1. Cut and machine all the specimens from the welded sheets, with dimensions specified

in the standard.

2. Grind the edges to remove the burr.

3. Determine and prepare the support with the distance indicated by the standard.

4. Place the specimen on the supports, centering the weld with the line of action of the

mandrel.

5. Start data acquisition software.

6. Start the test at a constant speed of advancement of the mandrel, Plunge Speed.

7. Keep testing until the specimens bow to an angle of about 90 degrees.

8. Record the data acquired for further processing and analysis.

9. Remove and photograph the sample and visually check the final state.

10. Back to the point 5 if there are more specimens to be tested.

f

B. Results

B1. Aluminum butt welds

g

h

B2. Copper Butt Welds

i

j

C. Specimen Design

C1. Tensile test specimen design

C2. Bending test specimen design

D. Bending structure

k

E. Technical Sheets

l

E1. Support Table

m

E2. Work Table

n

E3. Tool Body

o

E4. Probe – 4J3

E5. Probe – 4I3

p

E6. Shoulder 4P3

q

E7. Shoulder 4O3

r

F. Feasibility Study Calculation

F1. TIG Cost

s

F2. FSW Cos

t