FATIGUE BEHAVIOR AND STRUCTURAL STRESS ANALYSIS OF

COACH-PEEL AND LAP-SHEAR FRICTION STIR WELDED JOINTS OF

AZ31 MAGNESIUM ALLOY

by

MOHAMMED HAROON SHEIKH

J.B. JORDON, COMMITTEE CHAIR

M. BARKEY

Y. GUO

A THESIS

Submitted in partial fulfillment of the requirements for the degree of

Master of Science in the Department of Mechanical Engineering

in the

Graduate School of

The University of Alabama

TUSCALOOSA, ALABAMA

2014

Copyright Mohammed Haroon Sheikh 2014

ALL RIGHTS RESERVED

ii

ABSTRACT

In this work the fatigue behavior of coach-peel and lap-shear friction stir linear welded joints

of AZ31 magnesium alloy sheet were evaluated under different loads and the results were

compared using structural stress analysis. Lap-shear coupons of 30 mm in width were obtained

from a welded overlap configuration. However, since coach-peel configuration coupons were

not available, aluminum L-shaped brackets were adhesed weld to create the coach-peel

configured specimen. The experimental fatigue life results showed an inverse relationship

between applied load and the number of cycles to failure for both types of coupons. However,

due to differences in the configuration of the joint leading to higher applied stress, coach-peel

coupons failed at comparatively lower loads than lap-shear coupons. In order to correlate the

fatigue data with stresses developed at the weld, finite element analysis, using shell/plate

elements, was employed to model the joints under cyclic loading. The modeling effort focused

on several post-process techniques including equivalent stress and structural stress methods.

While the structural stress approach has merits over conventional nominal equivalent stress

approach for welded joints including accuracy of the results and mesh insensitivity, it could

not correlate the lap-shear and coach-peel fatigue results into a master design curve. As such,

a fatigue damage parameter (FDP) was introduced to correlate the fatigue results of both the

joints, thus establishing a basic relation to account for differences in stress at the weld for lap-

shear and coach-peel welded configurations.

iii

DEDICATION

This thesis is dedicated to my awesome family.

iv

LIST OF ABBREVIATIONS AND SYMBOLS

FSW Friction Stir Welding

FSSW Friction Stir Spot Welding

DOF Degree of Freedom

F,f Force

M Moment

σ Stress

τ Shear Stress

E Young’s Modulus of Elasticity

G Shear Modulus

Rho Density

Nu Poisson’s Ratio

SS Structural Stress

R Load Ratio

t Sheet Thickness

CP Coach Peel

LS Lap Shear

TIG Tungsten Inert Gas

MIG Metal Inert Gas

Mg Magnesium

Al Aluminum

Zn Zinc

Mn Manganese

v

ACKNOWLEDGMENTS

I would like to express my gracious gratitude to Dr. J. Brian Jordon, the chairman of the thesis

committee for selecting me to work on this project and for his continued guidance, support and

encouragement which played a phenomenal role in success of this project.

Additional appreciation is given to my committee which consists of Dr. Mark Barkey and Dr.

Yuebin Guo for guidance and for sharing their lab facility. Also I would like to thank all of the

graduate students working with Dr. Jordon: Harish, Bobby, Rogie and Joao for their help and

co-operation.

Also I would like to take this opportunity to thank all of my friends, UA students and faculty

for their extended support and necessary encouragement. Also Johnny Goodwin and Rob

Holler at UA central analytical facility deserve special thanks for their kind co-operation.

The greatest thanks of all goes to my family for their unconditional love, continuous support

and un-wearying faith in me from a distance of thousands of miles.

vi

CONTENTS

ABSTRACT ..............................................................................................................................ii

DEDICATION .........................................................................................................................iii

LIST OF ABBREVIATIONS AND SYMBOLS ....................................................................iv

ACKNOWLEDGMENTS..........................................................................................................v

LIST OF TABLES .................................................................................................................viii

LIST OF FIGURES ..................................................................................................................ix

1. INTRODUCTION

1.1 Need for Lightweight Materials ……...................................................................................1

1.2 Friction Stir Welding…………………….……………............................……...................1

1.3 Merits of Magnesium Alloy..................................................................................................3

1.4 Fatigue Behavior of Friction Stir Welding..……………............................……..................4

2. MATERIALS AND EXPERIMENTAL APPROACH

2.1 Materials and Specimen Preparation.....................................................................................7

2.2 Experimental Setup……………….......................................................................................9

3. FINITE ELEMENT ANALYSIS

3.1 Finite Element Analysis …………….................................................................................10

3.2 Structural Stress ……….....................................................................................................11

3.3 Finite Element Setup…………………...............................................................................15

3.4 Fatigue Data Correlation Using Equivalent Stress Approach .............................................17

4. RESULTS AND DISCUSSIONS

4.1 Microstructure of FSW …………….…………….............................................................19

4.2 Fatigue Test Results.……………………………..............................................................21

4.3 Fracture Behavior …………..............................................................................................22

vii

4.4 Structural Stress Results………………….........................................................................25

4.5 Conclusions………………................................................................................................29

REFERENCES………………………………………………….…………………….……...31

APPENDIX A1………………………………………………….……………………….…..36

APPENDIX A2………………………………………………….…………………………...37

viii

LIST OF TABLES

Table 2.1: Chemical Composition of AZ31 (mass%) ….…….…………………...…………..7

Table 3.1: Material Properties ……………..…...…………….…………………...…………17

ix

LIST OF FIGURES

Figure 1.1 Schematic Drawing of Friction Stir Welding Process………...………………….....2

Figure 1.2 Comparison of specific Young’s Modulus and Strength of Different Materials…....4

Figure 2.1 Preparation Steps: Coach Peel Specimen ………………..........……..….....……….8

Figure 2.2 Prepared: Coach Peel Specimen ……………………..……............……....…....…..8

Figure 2.3 Coach Peel Specimen before test…………………….....……...……………....…...9

Figure 2.4 Coach Peel Specimen after test…………………………….…...……………...…...9

Figure 3.1 Structural Stress Definition at the Weld Toe….........……………..…...……..…....12

Figure 3.2 Structural Stress Procedure for Shell/plate FE model………..……………..….…..14

Figure 3.3 Finite Element Model Set………………………..………….....……...……...……16

Figure 4.1 Macro and Micro- structure of FSW AZ31 Lap Joint...............................................20

Figure 4.2 Cycles to Failure variation with load applied for CP and LS Specimens.……...…..22

Figure.4.3 Failed lap shear specimens loaded on the advancing side........................................23

Figure.4.4 Coach Peel Failed specimens loaded on the advancing side.....................................24

Figure.4.5 Cycles to Failure variation with VonMises Stress....................................................25

Figure.4.6 Cycles to Failure variation with Equivalent Stress...................................................26

Figure.4.7 Bending Component variation along the weld line ..................................................27

Figrue.4.8 Membrane Component variation along the weld line..............................................27

Figrue.4.9 Maximum Shear Stress variation along the weld line..............................................28

Figure.4.10 FDP versus Cycles to Failure for Coach Peel.........................................................28

1

CHAPTER 1

INTRODUCTION

1.1 Need for Lightweight Materials

Due to continually increasing fuel prices and stringent pollution norms enforced by

governments all across the globe, the transportation industry, which includes automotive,

aerospace and ship-building is showing huge interest in manufacturing lightweight and

environment friendly products (e.g. cars, aircrafts, ships etc.). The process of light weighting

designs is essential to reducing fuel consumption and gas emission of vehicles [1,2]. This

motive has been further strengthened by the need to lower the cost of manufacturing without

compromising the strength and quality of the structures. The eco-friendly lightweight vehicles

can only be produced by replacing conventional iron-based metals and alloys with high strength

lightweight materials such as aluminum, magnesium and reinforced polymer composites [3-5].

These materials exhibit a combination of high strength properties with low density which

accounts for substantial weight reduction and improved vehicle performance by reducing

energy required for acceleration and rolling resistance, thus reducing the fuel consumption

which results in lower greenhouse emissions [6-8].

1.2 Friction Stir Welding

Some commonly employed metal joining techniques include fusion welding and resistance

spot welding which produce strong joints for ferrous based alloys. However to join non-ferrous

materials, these techniques have some limitations in the form of added process complexity,

drastic change in mechanical properties and poor mechanical performance [9-11]. In 1991 one

of the most significant developments in metal joining process took place in the form of a new

2

welding technique known as friction stir welding (FSW). FSW is a new versatile solid-state

joining technique with high energy efficiency and is environmentally friendly. In FSW, a

rotating tool pin is embedded into the interfacial material surfaces of plates to be joined with a

pre-determined tilt angle and then it is moved along the welded joint. Hence, materials are

joined by the frictional heat generated by the rotating tool at the interfacing material surfaces.

This generated heat along with the penetrating force applied on the tool causes plastic

deformation at the welding zone leading to joint formation. Fig. 1.1 [12] shows the general

arrangement of the FSW process. The amount of penetration of the tool varies according to the

thickness of the specimens, however for a lap-joint at full penetration, the tool shoulder makes

contact with the top surface of the bottom sheet of the specimen [13,14].

The FSW process has several advantages over conventional fusion welding techniques like

metal inert gas (MIG) and tungsten inert gas (TIG). Due to its low process temperature, below

the melting point, FSW is suited for joining thin or difficult to weld materials [15]. The most

important ability is the capacity to join dissimilar materials with high mechanical performance

[16,17]. With no melting issues regarding microstructure formation, issues like weld zone

shrinking from solidification and high risk of porosity in weld zone normally found in fusion

welds are thus avoided. The FSW joint is formed by friction heating with simultaneous severe

plastic deformation of the weld zone material. This stirring of the tool minimizes the risk of

Fig. 1.1. Schematic Drawing of Friction Stir Lap Linear Welding Process [12]

AS: Advancing Side

RS: Retreating Side

Pin

Tool Shoulder

3

having excessive local amounts of inclusions, resulting in a homogenous weld. The heat

distortions due to excessive heat normally found in fusion welding processes are avoided which

reduces the amount of residual stresses, and as such the deformation is easier to control. Thus

FSW requires lower resources, generates no emission, offers better controllability and produces

better quality joints when compared to conventional joining techniques like TIG and MIG

processes. Hence FSW has been gaining tremendous interest to replace conventional welding

techniques and has been used widely since its invention. With regard to the quality of the joints,

it is found that FSW specimens are usually stronger and have higher fatigue resistance than

specimens welded by TIG and MIG processes [18-23]. Also the welds formed using FSW

possess little distortion, low residual stresses and few weld defects [24]. This accounts in

general for higher life of friction stir welded joints.

Application of FSW ranges from small-scale component production such as cooling elements

and electric engines to welding to large panels used in ship building, in train wagons, offshore

structures and in bridge constructions.

1.3 Merits of Magnesium Alloy

As explained by factors like strict legislative rules, an increasing demand for fuel efficient

lightweight vehicles and lower manufacturing costs have driven automotive sector towards

stronger and lighter materials for vehicle manufacturing. The selection of an appropriate

material apart from weight and strength depends on economic feasibility. Mg alloys with its

good strength to weight ratio is a potential candidate offering a combination of lightweight

construction and desired strength. Hence, Mg alloys can be used in applications where low

mass and high specific properties are required. Also when Young’s modulus and high specific

strength magnitudes were compared, Mg alloys exhibited similar or even better values than

aluminum alloys and many commercial steels as shown in Fig. 1.2. Also, due to its increasing

4

uses, the cost per ton of Mg alloys are decreasing, thus making it more competitive from an

economic point of view [6].

1.4 Fatigue Behavior of Friction Stir Welding

The issue of fatigue in friction stir welded overlap joints is complex. This is in part due to

factors including variation in welding and loading configuration. Thus regarding overlap

friction stir welded fatigue results, literature primarily focuses on a variation of FSW: friction

stir spot welding (FSSW). In order to provide a thorough understand of fatigue in FSW in

magnesium alloys, a brief review of fatigue of FSSW is presented. Yin et al. [25,26] explained

that the hook formation and the overlap shear strength properties of AZ31 FSSW joints and

found that fatigue load strength of AZ31 is a function of dwell time of the tool. Also it is

concluded that stir zone microstructure and mechanical properties govern the fatigue life of the

AZ31 and AZ91 magnesium alloys. Furthermore, Jordon et al. [27] analyzed the fatigue

performance and failure mechanisms of FSSW in AZ31 alloy and presented the modeling

comparison with the experimental results. Mallick et al. [28] studied and discussed the fatigue

behaviors of FSSW joints in lap shear specimens of AM60 magnesium alloy and AA5754

aluminum alloy and analyzed the stress distribution and the location of maximum stresses in

Fig. 1.2. Comparison of specific Young’s modulus and strength of different materials [6]

5

FSSW joints in Mg-Mg specimens by using FEM. Luo et al. [29] investigated fatigue behavior

of FSSW specimens for AZ31 magnesium alloy sheet, and the fatigue strength, fatigue crack

propagation characteristic and failure mechanism is discussed in detail. It is reported that all

fatigue failures occurred in the stir zone of FSSW AZ31 magnesium alloy sheet joints. It was

also found that failure modes were different for high load and low load fatigue tests. The

magnitude of applied force also governed the crack initiation and propagation.

To the best of the author’s knowledge, no fatigue data exists for FSW overlap welds. However,

some work has been carried out on FSW butt-welded joints. Ericsson et al. and others

[18,30,31] have shown that the fatigue performance of FSW joints in aluminum alloys such as

5082/5083 Al alloys are far superior than that of MIG or TIG welded joints. Tsujikawa et al.

[32] studied and analyzed the factors affecting the fatigue strength of the welded commercial

magnesium alloys welded (FSW and TIG) joints. Joint efficiencies for static tensile strength

and fatigue strength were measured. They reported joint efficiencies value of 60% for fatigue

strength tests of AZ61 and 80% for AZ31 Mg alloy. The fatigue test results of welds prepared

using FSW is limited due to the recent invention of this welding method. Among the available

literature, the majority of the studies reporting fatigue behavior of the FSW have been limited

to constant amplitude loading [22]. Chowdhury et al. [33] evaluated lap shear strength and

fatigue properties of friction stir spot welded AZ31B-H24 Mg alloy and 5754-O Al alloy in

three different combinations: Mg-Mg, Al-Al and Al-Mg. They reported that both similar welds

had significantly higher lap shear strength, failure energy, and fatigue life compared to the Al-

Mg dissimilar weld. Furthermore, three different types of failure modes were observed. In

similar welds, higher cyclic loads resulted in nugget pullout due to fatigue crack propagation

circumferentially around the nugget, while at lower cyclic loads fatigue failure occurred

perpendicular to the loading direction caused by the opening of keyhole through crack

initiation.

6

While the primary focus of this thesis is related to FSW overlap welded joints, it is pertinent to

briefly discuss the fatigue behavior of coach-peel configured data for spot welds since data

does not exist for linear overlap joints. Coach peel specimens are primarily selected to evaluate

monotonic and fatigue strengths of joints (including riveted, welded) subjected to shear and

bending loads. Badarinarayan [34] has demonstrated fatigue failure of coach peel joint made

with different plate thickness in context with automotive closure panel. During fatigue tests it

was found that the loaded thin sheet of the coach-peel joint developed cracks which resulted in

pullout of thin sheet around circumference of nugget followed by joint failure where top sheet

broke free. When compared to lap-shear specimens, coach-peel specimens failed at lower

fatigue amplitudes for the same number of cycles to failure. In spite of the fact that the need of

coach-peel joint testing and evaluations exists, no investigation of deformation or fatigue

strength of friction stir welded joints of magnesium alloy in coach peel configuration exists.

Therefore the primary objective of this research is to investigate and evaluate fatigue strength

of coach-peel configuration of friction stir welded joints and determine appropriate FEA

modeling techniques.

7

CHAPTER 2

MATERIALS AND EXPERIMENTAL APPROACH

2.1 Materials and Specimen Preparation

In this research magnesium AZ31 alloy sheets of thickness 2 mm were used to prepare FSW

joints for fatigue testing. Table 2.1 presents the nominal chemical composition of AZ31

magnesium alloy. Prior to welding two magnesium sheets 700 mm long and 75 mm were

overlapped along the width and were tightly clamped on the worktable by the welding fixture.

The overlap along the longitudinal direction of the plate was 30 mm wide. The FSW tool used

was made of standard tool steel (H13) material and has shoulder diameter of 12 mm, pin length

of 3.2 mm with left hand threads and 10-degree concave shoulder. During FSW process tool

speed was maintained at 1000 RPM, tool plunge speed was kept at 20 mm/ min, shoulder

plunge depth was 0.1 mm.

The obtained lap-shear coupon was machined from the welded plate and contained the

following dimensions: total length of 120 mm, with overlap length 30 mm and width 30 mm.

In order to obtain coach-peel configuration coupons from the existing lap-shear coupons, the

following modification was made; 1) The legs of the overlap coupons were removed as shown

in Fig. 2.1; 2) 1/8 inch thick aluminum 6061 L-brackets were adhesed to the remaining FSW

joint. The aluminum brackets were joined to the FSW weld using 3M DP125 peel resistant

Al Zn Mn Fe Si Ni Cu Mg

2.92 0.85 0.46 0.002 0.006 <0.001 0.001 Bal.

Table 2.1 Chemical composition of AZ31 Alloy (mass %)

8

epoxy adhesive as shown in Fig. 2.1. (Please refer Appendix A1 for preparation steps detail).

Figure 2.2 shows a prepared coach peel sample before fatigue testing. Through trial-and-error,

it was determined that a minimum of 0.5 mm thick layer of adhesive was needed for optimal

strength. Prior to testing, the FSW joints were sectioned to characterize the microstructure and

geometrical features of the weld.

Fig. 2.2 Prepared Coach Peel Specimen

Weld Coupon

Epoxy

Glue

L-Bracket

CP Specimen Al-Brackets LS Specimen

Fig. 2.1 Preparation Steps: Coach Peel Specimen

9

2.2 Experimental Setup

The ultimate tensile load obtained during the tensile testing was subsequently used for

determining the maximum load to be used in the load controlled fatigue tests. Fatigue failure

tests for both types of specimen: coach-peel (CP) and lap-sheer (LS), were performed on fully

computerized MTS servo-hydraulic testing system with hydraulic grips, as shown in Fig. 2.3.

The grip-to-grip distance was 45 mm and 60 mm for coach-peel and lap-shear specimen,

respectively. For each load amplitude two specimens of same configuration and type were

tested. Load controlled constant amplitude fatigue tests were conducted according to ASTM

Standard E466 (“Standard Practice for Conducting Force Controlled Constant Amplitude Axial

Fatigue Tests of Metallic Materials”) where applicable. The load ratio (R=Pmin/Pmax) was kept

constant through the tests at 0.1. The applied cyclic loading waveform was sinusoidal and the

frequency of loading used varied from 2 Hz to 5 Hz, where most tests were conducted at 5 Hz.

Failure due to fatigue was defined as the complete separation of the specimens into two parts

as shown in Fig. 2.4.

Fig. 2.3. Coach-peel specimen before fatigue

testing.

Fig. 2.4 Coach-peel specimen after fatigue

testing.

10

CHAPTER 3

FINITE ELEMENT ANALYSIS: STURCTURAL STRESS DEFINITION

& FORMULATION

3.1 Finite Element Analysis

It is evident that while experimental work in the form of fatigue static testing etc., provides the

necessary physical insight about the behavior of the welded joints, predictive tasks such as

design, analysis and evaluation of the welded structures is often carried out by computational

methods. Stress estimation for a structure under a specific application is critical because a 10%

variation in stress value could translate into number of cycles to failure that varies by as much

as 100- 200%. Hence stress analysis is used to primarily ascertain structural viability and

integrity of structures such as ships, automobiles, aircrafts etc., under their respective operating

conditions. This type of analysis often involves use of an abstract or mathematical model for

the representation of the structural system and loads, e.g. partial differential equations are

employed to achieve classical analytical idealization. As complexity of the structural geometry

limits application of partial differential equations for structural mechanics analysis, numerical

approximation methods are often sought instead.

Similarly, analysis of any welding process includes a coupling of different mathematical

models used to address the static and dynamic behavior of the system. Numerical methods have

been used to implement these models. The use of finite element method (FEM) is a well-

established method for product development and manufacturing process improvement. FEM

offers efficient and economical product development thus aiding in productivity and quality of

products by offering better understanding of the influence of different process parameters.

11

Similarly FEM has been proved to be a powerful tool in welding analysis for the prediction of

welding deformations and residual stresses. Due to the versatility and economic benefits in

terms of costs and capital spent, FE has been used to predict fatigue life and failure

characteristics of the welded joints extensively. The FEM requires discretization of the domain

(structure) or boundary of interest into a large number of small elements connected via nodes.

By retrieving nodal forces and displacements from the model, these parameters can be used to

calculate different factors like structural stresses, principle stresses, and VonMises stresses etc.

to predict failure behavior of the structure or structural joint. This can be easily achieved while

using two and three-dimensional solid element models.

3.2 Structural Stress

Traditionally the fatigue life of a structural component (e.g. welded structures) is predicted

using nominal stress approach along with S-N curves generated using fatigue tests results. This

approach is relatively simple but sometimes is very difficult to apply for welded structures

because of the often challenge in determining nominal stress in complicated structures and

complex loading conditions. Furthermore, local stress concentration due to macro and micro

geometric effects of welded joints and residual stress effect due to different manufacturing

processes is not taken into account which can lead erroneous predictions [35,36].

Local stresses which occur in front of the weld toe or weld end notch (or on the inside of the

plate in front of the weld spot) is designated as structural stress [37]. The structural stress

approach is useful for calculating stresses where the structure contains geometric

discontinuities, e.g., welded joints, notches, ridges, bends, sharp corners, etc. As it considers

influence of all the global geometric parameters structural stress is also referred as “geometric

stress” and the maximum value of structural stress is known as ‘hot-spot stress’. This structural

stress can be considered as largely identical to the surface stress which can be calculated at the

12

hot spot in according with structural theories used in engineering by neglecting the notch effect

[38-40]. This type of stress approach has the capability to describe the macro-structural

behavior of welded structures and has been widely used to assess fatigue life of welded

structures. Because the stress concentration caused by a change in structural geometry is

included, the structural stress approach is considered more accurate than the nominal stress

approach [37]. Accordingly the structural stress in a localized fatigue-prone region of a

structure is defined in terms of the bending and membrane components. The stress values

within some distance from the weld toe can change significantly as the finite element mesh size

changes which is known as mesh-size sensitivity. The local stresses near a notch are mesh-size

sensitive due to the asymptotic singularity behavior as a notch position is approached. However

structural stress approach eliminates or minimizes the mesh-size sensitivity in stress

calculations due to the fact that the local stress concentration close to a notch is dominated by

self-equilibrating stress distribution [41-43].

The structural stress approach uses a definition of stress based on an idealized distribution in

the thickness of the members in proximity of the welds where the stress component orthogonal

to the weld line (which is also responsible for crack propagation) is reduced to a linearized

Fig. 3.1. Structural Stress Definition at the weld toe [44]

Weld Toe

13

distribution to find the weld stress levels. As a result of this linearization, the structural stress

σs at the weld toe is composed by a membrane component σm, constant in thickness and a

bending component σb, as depicted in Fig. 3.1 [44,45] . The remaining self-equilibrated non-

linear σnl is not considered, hence the structural stress includes only the effects of gross

structural discontinuities but disregards the local notch effect due to weld geometry.

In general, linear elastic finite element analysis cannot calculate the stresses at the edge of the

welded joints even if the finely meshed models are used mainly because of the considerable

plastic deformation at the edge of the weld nugget. In finite element modeling of a weld used

to calculate structural stresses, shell elements are used to represent the weld nugget and welded

elements. Based on the forces and moments obtained from the FE model, structural stresses are

calculated using linear elastic equation of beam and plate theory. The use of nodal forces and

moments has the distinctive advantage of providing a structural stress fairly insensitive to the

mesh element size and type in the areas corresponding to the weld toes. The typical mesh-

dependence is found in the traditional surface extrapolation method and partly in the through

thickness linearization is hence overcome. The stresses extrapolated to the nodes bear influence

from the element formulation and by the geometric characteristics of the FE, whereas nodal

forces are derived directly from the equilibrium of the structure. Using this approach the nodal

forces and moments for each element are calculated from the stiffness matrices and the nodal

displacements and rotations [42,46-49]. This structural stress is caused by the bending moment

along the weld line and by the normal force perpendicular to the weld line in the element plane

as shown in Fig. 3.2 below [42,46,50]. For the local through thickness stress distribution near

weld toe, it’s found that there exists a corresponding simple structural stress distribution, well

documented by Dong et al. [42]. This structural stress distribution has been represented by

14

membrane and bending components that are equilibrium-equivalent to the local stress

distribution.

Shell and plate element models of complex structures are widely used to perform stress and

fatigue analysis, where help of shell and plate theory is sought to obtain converged FE

solutions. In this study a shell and plate element model is used to represent coach-peel and lap-

shear joints and nodal forces and moments have been utilized to calculate the normal stress

acting on each FE element which is essentially a structural stress as for welded thin-sheet

structures cracks are most often located at the weld toes and is oriented along the weld line

[50]. Referring to Fig. 3.2, the weld line is oriented along Y-axis, while weld lies in plane XY.

Also it should be noted that in the present analysis, global and local coordinate systems have

the same orientation, hence no transformation of nodal forces and moments is required as

suggested by Dong et al. and others [42,51,52]. The procedure presented by Dong et al. to

calculate structural stress has been proven to provide consistent characterization of stress

concentration effects at the weld toe while achieving a good mesh-size and element type low

Fz

my

Fig. 3.2 Structural Stress Procedure for Shell/Plate FE Model

15

sensitivity for continuous welds. Hence it’s been widely used to determine structural stress.

Using the analytical model presented by Fermer et al. [50] and the structural stress model for

shell and plate elements using nodal forces and moments established by Dong et al., the method

to calculate structural (normal) stresses for the weld joint shown in Fig. 3.2 can be presented

as follows:

𝜎𝑛 = 𝜎𝑚 + 𝜎𝑏.....................................................(3.1)

𝜎𝑛 =𝑓𝑧

𝑡± 6

𝑚𝑦

𝑡2…………………………………………………(3.2)

3.3 Finite Element Setup

In the presented FE analysis of the lap-Shear and coach-Peel joints, welded joints are modeled

as a combination of shell elements where continuous linear friction stir weld is modeled as

shell element with thickness (13.5 mm) pertaining to the width of the weld nugget aligned in

the direction of weld line (along y-axis) as shown in Fig. 3.3. The weld plates are assigned a

thickness value of 2 mm. The end-to-end distance for lap-shear and coach-peel specimens was

kept at 60 mm and 45 mm respectively with the bottom end completely fixed (zero DOF) and

the top end free to move along z-axis (1 DOF) which is consistent with the experimental fatigue

test setup. As explained earlier, coach-peel specimens were prepared using lap-shear joints with

the aid of epoxy. Hence the epoxy joint is modeled with shell elements with a thickness of 0.5

mm along with corresponding material properties to represent thickness of epoxy layer. The

modeling and simulation was performed using Hypermesh 11.0 and optistruct solver was used

for calculating nodal forces, moments and stresses. Table 3.1 shows the material properties

used for the FE calculations.

16

Coach Peel

Lap-Shear

Fig. 3.3 Finite Element Model set

17

* Refer appendix A2 for source details

3.4 Fatigue Data Correlation Using Equivalent Stress Approach

In order to correlate the fatigue test data with the stress experienced by the specimens, an

equivalent structural stress (𝜎𝑒𝑞) approach is employed, where a combination of maximum

shear stress (𝜏𝑚𝑎𝑥) and corresponding normal stress (𝜎𝑛) are included [53]

𝜎𝑒𝑞 = 𝑘1𝜎𝑛 + 𝜏𝑚𝑎𝑥 (3.3)

𝜎𝑛 = 𝜎𝑚 + 𝜎𝑏 (3.4)

where,

𝜎𝑛 =𝑓𝑥

𝑡± 6

𝑚𝑦

𝑡2 For coach-peel specimen…………………………………. (3.5)

𝜎𝑛 =𝑓𝑧

𝑡± 6

𝑚𝑦

𝑡2 For lap-shear specimen………………………………….... (3.6)

According to the forces transferred by weld nugget for lap-shear specimen, stresses at the

nugget include both in-plane tensile loading and bending. However for the coach-peel

specimen, stresses at the nugget is dominated by bending. Also for the welded thin-sheet

structures, cracks are most often located at the weld toes and oriented along the weld line.

Hence in order to incorporate the stress induced on the welded joint due to different forces,

normal stress 𝜎𝑛 is defined as a combination of tensile and bending loads as shown in equation

Property Aluminum 6061 Magnesium AZ31 DP 125

E (MPa) 6.9 x 104 4.5 x 104 18.96

G (MPa) 7 x 104 5 x 104 17.56

Nu 0.3 0.35 0.32

Rho(t/mm3) 2.72 x 10-9 1.77 x 10-9 1.05 x 10-9

Table 3.1 Material Properties *

18

3.5 and 3.6, where bending moment along the weld line and normal force perpendicular to the

weld line in the plane of element were considered.

In order to correlate fatigue results of coach-peel and lap-shear tests a parameter is defined

which is referred as fatigue damage parameter (FDP),

𝐹𝐷𝑃 = 𝜎𝑒𝑞𝑒𝑘2𝜎𝑚𝑎𝑥

𝜎𝑚𝑖𝑛

(3.7)

where , 𝜎𝑚𝑖𝑛 Minimum normal stress

𝜎𝑚𝑎𝑥 Maximum normal stress

The FDP is calculated for specimens with two different geometries i.e. coach –peel and lap-

shear involving different loading conditions and fatigue loads are calculated and applied to

correlate experimental data. The model parameters k1 and k2 are optimized to obtain a good

correlation between the coach-peel and lap-shear test results. The equivalent structural stress

is calculated along the weld line at the weld nugget edge which bears the maximum stress as

evident from the FE model. The maximum equivalent stress magnitude is determined by

comparing the stress values for all the elements at mentioned location. The element with a

maximum 𝜎𝑒𝑞 is considered to be the critical location for crack initiation and the corresponding

𝜎𝑒𝑞 is used for FDP calculation.

19

CHAPTER 4

RESULTS AND DISCUSSIONS

4.1 Microstructure of FSW

Using optical imaging, microstructural variation in different zones of a FSW joint can be

identified which exists primarily due to differences in plastic deformation and frictional heat

generated during FSW process [54,55]. A macroscale view of cross-sectional plane of the FSW

is shown in Fig. 4.1 (a) Higher magnification images at various locations of the top and bottom

weld sheet in a FSW AZ31 lap joint shown in Fig. 4.1(a) are presented in Fig. 4.1 (b-g).

As shown in Fig.4.1 the microstructure of the weld can be generally classified into three distinct

regions namely: stir zone (SZ), the thermo mechanically affected zone (TMAZ) and the heat

affected zone (HAZ). The microstructure of the cross section in the retreating side and

advancing side is different, which is relative to the flow field of weld. The tunnel defect usually

can be seen near the RS zone [56]. As reported [57-59], due to FSW coarse grains in the base

metal turns into equiaxed and fine grains in the SZ resulting from the dynamic recrystallization

as a result of severe plastic deformation and frictional heat generated during FSW process. It

is also reported that pin geometry of the tool greatly affects the grain formation in SZ,

discussion of which is out of scope of this study. Typically the TMAZ, which is adjacent to

nugget, has deformed grains mainly due to high temperature during FSW. However the grain

structure in HAZ remains same as that of BM.

20

In several previous researches [60,61], authors have concluded that during FSW process, the

SZ gets exposed to a temperature higher than the recrystallization temperature but lower than

the melting point. This effect in association with the stirring effect of the tool, produces fine

and equiaxed grains in SZ as evident from Fig. 4.1 (d), hence it can be concluded that the SZ

underwent dynamic recrystallization (DRX). Furthermore, it is found that within nugget, the

grain size differs significantly, where grains in the SZ are finer than the ones at the top part of

the nugget next to top weld surface (Fig. 4.1 (d) and (e)) Fig. 4.1 (c) can be identified as HAZ

where grain structure is slightly different than that of BM. The microstructure of BM is shown

in Fig. 4.1 (b) and (g) it is mainly composed of granular magnesium solid solution grains.

TMAZ can be identified at location Fig. 4.1 (f) where nearly equiaxed grains are found which

HAZ TMAZ FLOW ARM HAZ TMAZ

AS RS

WELD NUGGET

(SZ)

Fig. 4.1 Macro and Micro- structure of FSW AZ31 Lap Joint

(a)

(b)

(c) (d)

(e) (g)

(f)

21

are finer than HAZ but coarser than SZ. Therefore based on the grain microstructures in the

TMAZs of the presented joint, it can be concluded that DRX occurred in TMAZs as well under

the combined effects of heating and deformation which resulted in the generation of nearly

equiaxed grains.

Fig. 4.1 (g) presents the bottom sheet weld microstructure at location which can be classified

as zone outside of the HAZ. The images presented can be used to compare grain size and

orientation at different locations as one moves away from the center of the weld. A close

observation shows that grain size increases from SZ to BM, which is in accordance with the

results published elsewhere [54-56,62-64].

4.2 Fatigue Test Results

Figure 4.2 presents the fatigue test results for coach-peel and lap-shear specimens where cycles

to failure is plotted against load applied (N). For comparison purpose fatigue tests were

designed in a way that, under the application of load for both types of specimen, the number of

cycles to failure fell in the range of 103 to 106. Initial fatigue tests of coach-peel specimens

revealed failure of joint occurred at lower loads when compared to loads applied during fatigue

tests of lap-shear specimens. Due to their geometry, coach-peel specimens were required to be

subjected to lower magnitude of loads as compared to lap-shear specimen which can bear

higher loads. Hence different load magnitudes were selected to test coach-peel and lap-shear

specimens. This is evident from Fig. 4.2, which shows that the coach-peel specimen failed at

lower loads for the same number of cycles to failure as that of lap-shear specimens.

22

4.3 Fracture Behavior of Lap-shear and Coach-peel Specimens

Figure. 4.3 shows cross-section view of two failed lap-shear samples which were loaded at

advancing side: (a) low cycle loading, where the specimen was subjected to maximum load of

2069 N which resulted in fatigue failure after 503 cycles, and (b) high cycle loading, where the

specimen was subjected to maximum load of 371 N which caused fatigue failure after 353,589

cycles. It is clear that the failure resulted along the line of the interfacial hook that existed

between the two plates which terminated in the form of upward facing hook as shown in Fig.

4.3 (marked with a circle). At a higher magnitude of fatigue loading the upward facing hook at

the nugget boundary travels approximately in a normal direction to the top weld surface in the

form of crack as shown in Fig. 4.3a. This resulted in crack propagation through the weld nugget

edge, which later resulted in final rupture through the top of weld thickness, resulting in weld

failure. However in case of lower magnitude of fatigue loading, the crack initiated (marked

with a circle) at the end of interfacial hook and travels along the nugget boundary as shown in

Fig. 4.3b. It is anticipated that due to lower load application, the developed cracks propagated

rather slowly when compared to the higher fatigue load crack propagation. It should be noted

Fig. 4.2 Cycles to failure variation with load applied for coach-peel (CP) and lap-shear (LS) specimens.

1

10

100

1,000

10,000

1.0E+02 1.0E+03 1.0E+04 1.0E+05 1.0E+06 1.0E+07 1.0E+08

LO

AD

(N

)

CYCLES TO FAILURE

CP LS

23

that in this case the crack propagation followed a path along the nugget boundary without

rupturing the nugget, as this is primarily due to the lower magnitude of the fatigue load. It

important to note that the weld nugget in Mg alloys has the highest hardness and strength level

compared to the TMAZ, HAZ and BM [56,65-67]. Hence it takes higher stresses to rupture

weld nugget as observed in low cycle higher fatigue load test.

Figure 4.4 summarizes the crack propagation and final failure for high cycle and low cycles

coach-peel specimen during fatigue testing. Figure 4.4a shows high cycle (low load) fatigue

failure where the crack initiates at half weld thickness location next to nugget boundary

(marked with circle) where two sheets merge due to FSW. It should be noted that, this is also

the location of upward facing interfacial hook. After initiation and crossing the interfacial hook

location, the crack propagates along the nugget boundary away from the weld center in an

upward-right direction and final failure occurred due to the travel of the crack through the weld

Fig. 4.3 Failed specimens loaded on the advancing side: a) fatigue at 503 cycles and

maximum load of 2069 N, b) fatigue at 353589 cycles and maximum load of 371 N.

AS RS

(a)

RS AS

(b)

24

top surface where, due to lower thickness of the weld the crack changed the propagation

direction to an upward-left direction. Due to the higher number of cycles to failure it can be

concluded that, in this case, the crack propagated at a slower rate till it reached the minimum

weld thickness, where it the traveled faster. In the case of the low cycle (high load) shown in

Fig. 4.4b, the crack initiated at the location of interfacial hook, hence the initiation of the crack

was the same as that of high cycle case. However unlike in the high cycle failure case, the crack

changed direction of propagation towards the center of the weld. During each successive cycle

to failure, the crack propagated more closely to the nugget center and follows an upward and

left direction of propagation as shown. It should also be noted that the crack appears to follow

a steeper leftward path just before final failure which is mainly due to lower thickness of weld

at the center, where under application of same magnitude of fatigue load, the crack took less

energy to travel. As the failure occurred at lower number of cycles, it can be deduced that the

crack propagation was faster when compared to low cycle application primarily due to higher

load magnitude. Furthermore, the difference in crack propagation and final failure can be

attributed to the magnitude of applied force and difference in microstructures at various zones.

Similar failure modes have been reported by Yang et al. for AZ31 linear FSW lap joints [68].

(a)

(b)

Fig.4.4 Failed coach-peel specimens loaded on the advancing side: a) fatigue life at 193,330 cycles

for a maximum load of 178 N., b) fatigue life at 4,548 cycles for a maximum load of 400 N.

(AS)

(AS)

(RS)

(RS)

25

4.4 Structural Stress Results

Figure 4.5 shows the variation of number of cycles to failure with maximum von Mises stress

acting on the lap-shear and coach-peel joints during the fatigue tests. It is evident that coach-

peel specimen experienced higher von Mises stress compared to the lap-shear specimens.

Figure 4.6 shows the result of the structural stress determined from the FEA associated with

the applied remote loading during the experimental fatigue tests. Similar to the von Mises stress

analysis, the equivalent structural stress for the coach-peel specimen is of higher magnitude

when compared to lap-shear specimen. As equivalent SS is a combination of normal stress and

shear stress, it can be deduced that stress levels due to normal and shear forces in the coach-

peel specimen are higher. This is primarily attributed to the different orientation of coach-peel

weld with respect to line of loading during the fatigue tests. In case of the lap-shear fatigue

tests, the weld plane is parallel and in the line of action of the fatigue loads. However in the

case of the coach-peel tests, the fatigue load acts normal to the weld plane. Hence it can be

concluded that weld orientation and load application of the coach-peel joints underwent higher

Fig. 4.5 Cycles to Failure variation with VonMises Stress for Coach Peel (CP) and

Lap-Shear (LS) Specimen

1

10

100

1,000

10,000

1.0E+02 1.0E+03 1.0E+04 1.0E+05 1.0E+06 1.0E+07 1.0E+08

MA

X V

ON

MIS

ES

(M

PA

)

CYCLES TO FAILURE

CP

LS

26

stresses for a comparatively lower magnitudes of applied load. This also explains the higher

equivalent SS in coach-peel specimen and lower equivalent SS levels in lap-shear welds.

In order to quantify the lower fatigue life of the coach-peal specimen and higher stress levels,

it is imperative to analyze the induced maximum stress level. As explained earlier, equations

3.5 and 3.6, as per definition, the structural stress or normal stress consists of a bending

component (BC) and a membrane component (MC). Hence the effect of the BC and MC along

the weld line at the maximum stress location (i.e. adjacent to weld nugget) requires an

investigation. Being a contributing factor for equivalent stress (equation 3.3), maximum shear

stress variation along the weld line is also studied. Figure 4.7, 4.8 and 4.9 show variation of

BC and MC and shear stress respectively at a maximum stress location along the weld line for

coach-peel and lap-shear specimens. Comparing the BC components for both types of

specimens it is found that coach-peel specimen underwent large positive bending stresses

whereas the lap-shear specimen experienced lower negative bending stresses. However it is

found that the MC played dominant role in lap-shear specimen deformation and had a positive

magnitude unlike the coach-peel specimen which experienced a negative membrane/normal

Fig.4.6 Cycles to failure variation with equivalent stress for coach-peel (CP) and lap-shear (LS) specimen

1

10

100

1,000

10,000

1.0E+02 1.0E+03 1.0E+04 1.0E+05 1.0E+06 1.0E+07 1.0E+08

EQ

_S

S (

MP

A)

CYCLES TO FAILURE

CP

27

stress of lower magnitude. As anticipated, the coach-peel specimen experienced greater shear

stress compared to the lap-shear specimen as indicated in Fig. 4.9. Hence it can be concluded

that deformation of coach-peel specimen includes contribution from both the shear and bending

stresses, however the bending stress is of higher magnitude which suggests, fatigue failure is

thus largely dominated by bending moments. Similarly from Fig. 4.8 it can be concluded that

deformation in lap-shear specimen is dominated by the MC or force component of the structural

stress definition. The higher magnitudes of the BC and shear stress in coach-peel specimen also

leads to higher equivalent stress developed as shown in Fig. 4.6.

-0.10

0.00

0.10

0.20

0.30

0.40

0.50

-15 -10 -5 0 5 10 15

Ben

din

g C

om

p

Weld Line Dist (mm)

CP

LS

Fig. 4.7 Bending Component variation along the weld line for CP and LS specimen

Fig. 4.8 Membrane Component variation along the weld line for CP and LS specimen

-4.0E-03

0.0E+00

4.0E-03

8.0E-03

1.2E-02

1.6E-02

2.0E-02

-15 -10 -5 0 5 10 15

Mem

bra

ne

Co

mp

Weld Line Dist (mm)

CP

LS

28

Thus, the large variation of BC and MC of the FSW overlap joint suggests that the structural

stress approach alone cannot collapse the fatigue data into a single line without significant

modifications. Hence, the use of a FDP parameter that accounts for the large discrepancy has

been shown to provide a similitude approach for welded joints. The FDP correlation to number

of cycles to failure is presented in Fig.4.10. Using optimized values of k1 and k2 of 0.94 and

1.35, respectively, the FDP values for the coach-peel and lap-shear specimen data collapse onto

0.0E+00

3.0E-02

6.0E-02

9.0E-02

1.2E-01

1.5E-01

1.8E-01

-15 -10 -5 0 5 10 15

Max

Shea

r S

tres

s

Weld Line Dist (mm)

CP

LS

Fig. 4.9 Maximum Shear Stress variation along the weld line for CP and LS specimen

Fig. 4.10 FDP versus Cycles to Failure for Coach Peel (CP) and Lap-Shear (LS) Specimen

1

10

100

1,000

10,000

1.0E+02 1.0E+03 1.0E+04 1.0E+05 1.0E+06 1.0E+07 1.0E+08

FD

P

CYCLES TO FAILURE

CP LS

29

a single curve. This kind of behavior primarily exist due to the formulation of FDP where a

relation between equivalent stress and maximum and minimum induced stress is expressed

exponentially. This result suggests that the FDP parameter can be used to establish the relation

between fatigue data and cycles to failure for different types of loading configurations for the

FSW fabricated from Mg AZ31 alloy sheets.

The fatigue prone location in FE model for coach-peel and lap-shear coupons, calculations and

all the relevant data used to plot the results is presented in Appendix A2.

4.5 Conclusions

In this study fatigue performance of linear friction stir welded coach-peel and lap-shear

specimens of AZ31 magnesium alloy sheets of 2 mm thickness are experimentally and

numerically investigated. An equivalent stress approach is utilized to calculate stresses and a

fatigue parameter referred as FDP is used to take into account the effect of specimen geometry

and loading types. By analyzing the results, the following conclusions are drawn:

1. For the same number of cycles to failure the coach-peel specimen failed at lower loads

when compared to lap-shear specimen. This is primarily attributed to the higher level

of induced shear and bending stress components. Due to the geometry of the specimen

and loading conditions, the friction stir weld nugget in the coach-peel is subjected to

higher bending stress and shear stress which led to its lower fatigue strength when

compared to lap-shear specimen.

2. A new procedure is developed to convert and test lap-shear welds into coach-peel

configuration. This is achieved by gluing L-shaped aluminum brackets to the

overlapped weld plates.

3. Failure analysis revealed that for lap-shear specimen at lower fatigue loads, cracks

propagated slowly and followed a path along the periphery of the weld nugget. This

30

resulted in higher number of cycles to failure and failure occurred by fracture through

weld top sheet. The presence of the initial upward facing hook and lower thickness of

weld side sheet is concluded as the reason for this type of failure behavior. At higher

loads (low cycles), the same type of failure took place, however due to higher fatigue

load, the crack propagated through part of the nugget. Similar behavior is observed for

coach-peel specimen.

4. FE analysis was done to calculate structural stress in coach-peel and lap-sheer

specimens. Based on the SS magnitude calculated using nodal forces and moments it

was found that coach-peel specimen bears higher SS values than lap-sheer specimen.

Further investigation revealed stresses due to bending and shear in the coach-peel

specimen is higher compared to the lap-shear specimen. However due to the bending

and shear stresses acting on the lap-shear specimen, the stresses are of very low

magnitude and the dominant component of the stress (membrane stress) has a

comparatively low magnitude. This resulted in the coach-peel specimens failing

quickly whereas the lap-shear specimen was capable of enduring higher loads for longer

time period resulting in higher number of fatigue cycles.

5. Fatigue data of both the specimen types was correlated using an equivalent stress based

FDP. The FDP considered in this study took into account differences in specimen

geometry and stresses developed, and hence was found to be effective in consolidating

a large variation of fatigue data into a narrow band. The defined FDP related forces and

moments acting on weld nugget which were transferred to a single FDP-N curve for

two different specimens. Hence based on the FDP model, the fatigue life of the FSW

overlap joints under consideration can be predicted in advance.

31

REFERENCES

(1) DeCicco J, An F., Ross M. (2001)Technical options for improving the fuel economy of

U.S. cars and light trucks by 2010-2015. ;T012.

(2) Duflou J, R., De Moor J, Verpoest I, Dewulf W. Environmental Impact Analysis of

Composity Use in Car Manufacturing. CIRP Annals – Manufacturing Technology (2009);58:9-

12.

(3) Merklein M, Geiger M. New materials and production technologies for innovative

lightweight constructions. Journal of Materials Processing Technology, (2002);125–126

(9):532-536.

(4) Tharumarajah A, Koltun A. Is there an environmental advantage of using mangesium

components for light-weighting cars? Journal of Cleaner Production (2007);15 (11–12):1007-

1013.

(5) Kochan A. Audi moves forward with all aluminum cars. Assembly Automat (2000);20

(20):132-135.

(6) Blawert C, Hort N, Kainer KU. Automotive applications of magnesium and its alloys.

Transactions of the Indian Institute of Metals (2004);57(4):397-408.

(7) Davies G. Future trends in automotive body materials. In: Davies G, editor. Materials for

Automobile Bodies Oxford: Butterworth-Heinemann; (2003). p. 252-269.

(8) Hassanifard S, Ahmadi S,R., Pour M, M. Weld arrangement effects on the fatigue behavior

of multi friction stir spot welded joints. Mater Des (2013) 2;44(0):291-302.

(9) Gean A, Westgate SA, Kucza JC, Ehrstrom JC. Static and fatigue behavior of spot-welded

5182-O aluminum alloy sheet. J Weld (1999);78:80-86.

(10) Cavaliere P, Nobile R, Panella F,W., Squillace A. Mechanical and microstructural

behaviour of 2024–7075 aluminium alloy sheets joined by friction stir welding. Int J Mach

Tools Manuf (2006) 5;46(6):588-594.

(11) Thornton P, H., Krause A, R., Davies R, G. The aluminum spot weld. J Weld

(1996);75:101-108.

(12) Yang Q, Li X, Chen K, Shi YJ. Effect of tool geometry and process condition on static

strength of a magnesium friction stir lap linear weld. Materials Science and Engineering: A

(2011) 3/15;528(6):2463-2478.

(13) Gemme F, Verreman Y, Dubourg L, Jahazi M. Numerical analysis of the dwell phase in

friction stir welding and comparison with experimental data. Materials Science and

Engineering (2010) 6/25;527(16–17):4152-4160.

32

(14) Rao H,M., Jordon J,B., Barkey M,E., Guo Y,B., Su X, Badarinarayan H. Influence of

structural integrity on fatigue behavior of friction stir spot welded AZ31 Mg alloy. Materials

Science and Engineering: A (2013) 3/1;564(0):369-380.

(15) Knipstrom K, E. New Welding Method for Aluminium. Welding Technology and Fatigue

Properties (1995);50.

(16) Su P, Gerlich A, North T,H., Bendzsak G,J. Intermixing in Dissimilar Friction Stir Spot

Welds. Metallurgical and Materials Transactions (2007);38(3)(- 3):584-595.

(17) Lee C, Lee J, Park D, Na S. Mechanical and Microstructural Properties of Friction Spot

Joined 5052 and 6111 Al Alloys. Material Science Forum (2008);580-582:435-438.

(18) Ericsson M, Sandström R. Influence of welding speed on the fatigue of friction stir welds,

and comparison with MIG and TIG. Int J Fatigue (2003) 12;25(12):1379-1387.

(19) Moreira,P.,M.,G.,P., de Jesus,A.,M.,P., Ribeiro AS, de Castro,P.,M.,S.,T. Fatigue crack

growth in friction stir welds of 6082-T6 and 6061-T6 aluminium alloys: A comparison. Theor

Appl Fract Mech (2008) 10;50(2):81-91.

(20) Moreira, P.,M.,G.,P., A. M,P. deJesus, Ribeiro A,S., deCastro,P.,M.,S.,T. Assessment of

the Fatigue Behaviour of Friction Stir Welded Joints: Aluminium Alloy 6082-T6. Key

Engineering Materials (2007);348-349:209-212.

(21) Kobayashi Y, Sakuma M, Tanaka Y, Matsuoka K. Fatigue strength of friction stir welding

joints of aluminium alloy 6082 extruded shape. - Welding International (2007);21(- 1):18-24.

(22) Costa J,D., Ferreira, J.,A.,M., Borrego L,P. Influence of spectrum loading on fatigue

resistance of AA6082 friction stir welds. International Journal of Structural Integrity

(2011);2:122-134.

(23) Fatigue performance of friction stir butt welds in a 6000 series aluminum alloy. Computer

Methods and Experimental Measurements for Surface Treatment Effects; (1995).

(24) Thomas M, Nicholas J, Needham J, Murch M, Templesmith P, Dawes C, inventors.

Anonymous Friction stir welding. . (1991) .

(25) Yin Y,H., Sun N, North T,H., Hu S,S. Microstructures and mechanical properties in

dissimilar AZ91/AZ31 spot welds. Mater Charact (2010) 10;61(10):1018-1028.

(26) Yin Y,H., Sun N, North T,H., Hu S,S. Hook formation and mechanical properties in AZ31

friction stir spot welds. J Mater Process Technol (2010) 11/1;210(14):2062-2070.

(27) Jordon J,B., Horstemever M,F., Daniewicz S,R., Badarinarayan H, Grantham J. Fatigue

Characterization and modeling of friction stir spot welds in magnesium AZ31 alloy. Journal of

Engineering Materials Technology (2010);132:1-10.

(28) Mallick P, Agarwal L. Fatigue of spot friction welded joints of Mg-Mg, Al-Al and Al-Mg

alloys. SAE Technical Paper (2009);1:24-27.

33

(29) LUO T, SHI B, DUAN Q, FU J, YANG Y. Fatigue behavior of friction stir spot welded

AZ31 Mg alloy sheet joints. Transactions of Nonferrous Metals Society of China (2013)

7;23(7):1949-1956.

(30) Zhou C, Yang X, Luan G. Fatigue Properties of Friction Stir Welds in Al 5083 Alloy.

Scripta Material (2005);53:1187-1191.

(31) James M,N., Hattingh D, Bradley G. Weld Tool Travel Speed Effects on Fatigue Life of

Friction Stir Welds in 5083 Aluminum. Internaltional Journal of Fatigue (2003);25:1389-1398.

(32) Tsujikawa M, Somekawa H, Higashi K, Iwasaki H, Hasegawa T, Mizuta A. Fatigue of

Welded Magnesium Alloy Joints. Materials Transactions (2004);45:419-422.

(33) Chowdhury S,H., Chen D,L., Bhole S,D., Cao X, Wanjara P. Lap shear strength and

fatigue life of friction stir spot welded AZ31 magnesium and 5754 aluminum alloys. Materials

Science and Engineering: A (2012) 10/30;556(0):500-509.

(34) H. Badarinarayan. Fundamentals of friction stir spot welding. Missouri: Missouri

University of Science and Technology; (2009).

(35) Kang H, Lee Y. Chapter 10 - Fatigue Life Prediction Methods of Seam-Welded Joints. In:

Lee Y, Barkey ME, Kang H, editors. Metal Fatigue Analysis Handbook Boston: Butterworth-

Heinemann; (2012). p. 383-427.

(36) Pan J, Sripichai K, Lin P, Wang D, Lin S. 1 - Mechanics modeling of spot welds under

general loading conditions and applications to fatigue life predictions. In: Sun X, editor. Failure

Mechanisms of Advanced Welding Processes: Woodhead Publishing; (2010). p. 1-23.

(37) Radaj D. Design codes, assessment of nominal and structural stress. Design and analysis

fo fatigue resistant welded structures: Woodhead Publishing; (1990). p. 163-207.

(38) Radaj D. Stress singularity, notch stress and structural stress at spot-welded joints. Eng

Fract Mech (1989);34(2):495-506.

(39) Fricke W, Kahl A. Comparison of different structural stress approaches for fatigue

assessment of welded ship structures. Mar Struct (2005) 0;18(7–8):473-488.

(40) Radaj D, Sonsino CM, Fricke W. Recent developments in local concepts of fatigue

assessment of welded joints. Int J Fatigue (2009) 1;31(1):2-11.

(41) Rupp A, Störzel K, and Grubisic V. Computer Aided Dimensioning of Spot-Welded

Automotive Structures. SAE Technical Paper (1995):635-645.

(42) Dong P. A structural stress definition and numerical implementation for fatigue analysis

of welded joints. Int J Fatigue (2001) 11;23(10):865-876.

(43) Sheppard S,D. Further Refinement of a Methodology for Fatiuge Life Estimation in

Resistance Spot Welding Connections. Advances in Fatigue Lifetime Predictive Techniques

(1996);3:265-282.

34

(44) Marin T, Nicoletto G. Fatigue Design of Weleded Joints Using The Finite Element Mehtod

and The 2007 ASME Div.2 Master Curve. Frattura ed Integrità Strutturale (2009);3(9):76-86.

(45) Hong J,K., Forte T. Fatiuge Evalution of Notched Plate Specimen By the Battelle

Sturcutral Stress Method. SAE International Journal of Materials and Manufacturing

(2013);6(2):262-270.

(46) Dong P. A Robust Structural Stress Method for Fatigue Analysis of Offshore/Marine

Structures. Journal of Offshore Mechanics and Arctic Engineering (2005) March 23;127(1):68-

74.

(47) Dong P. A Robust Structural Stress Procedure for Characterizing Fatigue Behavior of

Welded Joints. SAE 2001 World Congress (2001).

(48) Kyuba H, Dong P. Equilibrium-equivalent structural stress approach to fatigue analysis of

a rectangular hollow section joint. Int J Fatigue (2005) 1;27(1):85-94.

(49) Xi P, Xia L, Fei P, Zhang D, Cheng B. Preparation and performance of a novel

thermoplastics polyurethane solid–solid phase change materials for energy storage. Solar

Energy Mater Solar Cells (2012) 7;102(0):36-43.

(50) Fermer M, Andreasson M. Fatigue Life Prediction of MAG-Welded Thin-Sheet

Structures. International Body Engineering Conference & Exposition (1998);2:1-7.

(51) Kim M,H., Kang S,W., Kim J, H., Kim K,S., Kang J, K., Heo J,H. An experimental study

on the fatigue strength assessment of longi-web connections in ship structures using structural

stress. Int J Fatigue (2010) 2;32(2):318-329.

(52) Selvakumar P, Hong JK. Robust Mesh Insensitive Structural Stress Method for Fatigue

Analysis of Welded Structures. Procedia Engineering (2013);55(0):374-379.

(53) Shi Y, Guo H. Fatigue Performance and Fatigue Damage Parameter Estimation of Spot

Welded Joints of Aluminium Alloys 611-T4 and 5754. Fatigue & Fracture of Engineering

Materials & Structures (2013);36(0):1081-1090.

(54) Yang J, Xiao BL, Wang D, Ma ZY. Effects of heat input on tensile properties and fracture

behavior of friction stir welded Mg–3Al–1Zn alloy. Materials Science and Engineering: A

(2010) 1/15;527(3):708-714.

(55) Woo W, Choo H, Prime M,B., Feng Z, Clausen B. Microstructure, texture and residual

stress in a friction-stir-processed AZ31B magnesium alloy. Acta Materialia (2008)

5;56(8):1701-1711.

(56) Xunhong W, Kuaishe W. Microstructure and properties of friction stir butt-welded AZ31

magnesium alloy. Materials Science and Engineering: A (2006) 9/15;431(1–2):114-117.

(57) Motalleb-nejad P, Saeid T, Heidarzadeh A, Darzi K, Ashjari M. Effect of tool pin profile

on microstructure and mechanical properties of friction stir welded AZ31B magnesium alloy.

Mater Des (2014) 7;59(0):221-226.

35

(58) Microstructure and mechanical properties of friction stir welded AZ31 MG alloy. TMS

Annual Meeting; Feb 21 2002; ; (2002).

(59) Zhang h, Lin S,B., Wu L, Feng J,C. Microstructure studies of friction stir welded AZ31

Magnesium Alloy. ACTA METALLURGICA SINICA (ENGLISH LETTERS)

(2003);17(5):747-753.

(60) Park,S.,H.,C., Sato Y,S., Kokawa H. Effect of micro-texture on fracture location in friction

stir weld of Mg alloy AZ61 during tensile test. Scr Mater (2003) 7;49(2):161-166.

(61) Darras B,M., Khraisheh M,K., Abu-Farha F,K., Omar M,A. Friction stir processing of

commercial AZ31 magnesium alloy. J Mater Process Technol (2007) 8/1;191(1–3):77-81.

(62) Liu D, Xin R, Zheng X, Zhou Z, Liu Q. Microstructure and mechanical properties of

friction stir welded dissimilar Mg alloys of ZK60–AZ31. Materials Science and Engineering:

A (2013) 1/20;561(0):419-426.

(63) Woo W, Choo H, Brown D,W., Liaw P,K., Feng Z. Texture variation and its influence on

the tensile behavior of a friction-stir processed magnesium alloy. Scr Mater (2006)

6;54(11):1859-1864.

(64) Forcellese A, Gabrielli F, Simoncini M. Mechanical properties and microstructure of joints

in AZ31 thin sheets obtained by friction stir welding using “pin” and “pinless” tool

configurations. Mater Des (2012) 2;34(0):219-229.

(65) Fukumoto S, Tanaka S, Ono T, Tsubakino H, Tomita T, Aritoshi M, et al. Microstructural

Development in Friction Welded AZ31 Magnesium Alloy. Materials Transactions

(2006);47(4):1071-1076.

(66) Commin L, Dumont M, Masse J-, Barrallier L. Friction stir welding of AZ31 magnesium

alloy rolled sheets: Influence of processing parameters. Acta Materialia (2009) 1;57(2):326-

334.

(67) Motalleb-nejad P, Saeid T, Heidarzadeh A, Darzi K, Ashjari M. Effect of tool pin profile

on microstructure and mechanical properties of friction stir welded AZ31B magnesium alloy.

Mater Des (2014) 7;59(0):221-226.

(68) Yang Q, Li X, Chen K, Shi YJ. Effect of tool geometry and process condition on static

strength of a magnesium friction stir lap linear weld. Materials Science and Engineering: A

(2011) 3/15;528(6):2463-2478.

36

APPENDIX A1

Coach Peel Coupon Preparation

The weld piece was obtained by removing legs of from the overlap weld. The contacting

surfaces of the weld and aluminum L-bracket to be joined were cleaned using sand paper and

ethanol. The advancing side (AS) edge of the weld piece was preserved using adhesive tape to

prevent epoxy glue from joining the sheets leading to erroneous results, as shown in Fig. A1.

A layer (approx. 0.5 mm thick) of DP 125 epoxy was applied on contacting surfaces of L-

brackets and weld-piece. After 2 minutes, the weld piece was aligned between the two L-

brackets forming a “coach - peel” configuration. Special care was taken to maintain the mutual

alignment of the L-brackets and their alignment with respect to weld piece as well. In order to

ensure correct alignment of all three pieces, supports were employed throughout the curing

period (7 days) as shown in Fig. A2.

Fig. A1. Preparation Steps: Coach Peel Specimen

AS

RS

Adhesive tape Edge

Fig. A2. Coach Peel Specimen with supports

AS RS

37

APPENDIX A2

FE Data & Calculation Tables

Element 20

Element 1

Coach Peel Specimen: Failure Location

Element 1 Element 20

Lap Shear Specimen: Failure Location

38

Element No.

Element Force (Fz)

El. Moment (My)

Min σn Max σn Eq.Stress MAX

Eq.Stress FDP

Max FDP

1 3.41E-02 -1.51E-02 -0.006 0.040 0.062

0.074

0.052

0.052

2 2.77E-02 -2.46E-02 -0.023 0.051 0.071 0.038

3 2.65E-02 -2.70E-02 -0.027 0.054 0.074 0.037

4 2.59E-02 -2.73E-02 -0.028 0.054 0.074 0.037

5 2.56E-02 -2.71E-02 -0.028 0.054 0.073 0.036

6 2.55E-02 -2.69E-02 -0.028 0.053 0.073 0.036

7 2.53E-02 -2.66E-02 -0.027 0.053 0.073 0.036

8 2.52E-02 -2.64E-02 -0.027 0.052 0.072 0.036

9 2.50E-02 -2.63E-02 -0.027 0.052 0.072 0.036

10 2.48E-02 -2.62E-02 -0.027 0.052 0.072 0.036

11 2.46E-02 -2.62E-02 -0.027 0.052 0.072 0.035

12 2.44E-02 -2.61E-02 -0.027 0.051 0.071 0.035

13 2.41E-02 -2.62E-02 -0.027 0.051 0.071 0.035

14 2.38E-02 -2.62E-02 -0.027 0.051 0.071 0.035

15 2.35E-02 -2.63E-02 -0.028 0.051 0.071 0.034

16 2.31E-02 -2.64E-02 -0.028 0.051 0.071 0.034

17 2.26E-02 -2.63E-02 -0.028 0.051 0.071 0.034

18 2.23E-02 -2.56E-02 -0.027 0.050 0.070 0.033

19 2.19E-02 -2.28E-02 -0.023 0.045 0.065 0.033

20 3.41E-02 -1.51E-02 -0.006 0.040 0.062 0.052

Table A1: Fatigue Data for Lap Shear Specimens

39

Element No.

Element Force (Fx)

El. Moment (My)

Min σn Max σn Eq.Stress MAX

Eq.Stress FDP

Max FDP

1 4.36E-03 1.78E-01 -0.264 0.269 0.418

0.539

0.111

0.140

2 1.90E-03 2.52E-01 -0.377 0.379 0.515 0.135

3 2.96E-04 2.68E-01 -0.401 0.402 0.538 0.140

4 -2.21E-04 2.68E-01 -0.402 0.402 0.539 0.140

5 -4.30E-04 2.66E-01 -0.400 0.399 0.536 0.139

6 -4.99E-04 2.65E-01 -0.397 0.397 0.533 0.138

7 -5.09E-04 2.63E-01 -0.395 0.394 0.530 0.137

8 -4.96E-04 2.62E-01 -0.393 0.393 0.529 0.137

9 -4.77E-04 2.61E-01 -0.392 0.392 0.527 0.136

10 -4.61E-04 2.61E-01 -0.392 0.391 0.527 0.136

11 -4.52E-04 2.61E-01 -0.392 0.391 0.527 0.136

12 -4.54E-04 2.61E-01 -0.392 0.392 0.527 0.137

13 -4.67E-04 2.62E-01 -0.393 0.393 0.529 0.137

14 -4.91E-04 2.63E-01 -0.395 0.394 0.530 0.137

15 -5.23E-04 2.65E-01 -0.397 0.396 0.533 0.138

16 -5.61E-04 2.66E-01 -0.399 0.399 0.535 0.139

17 -5.92E-04 2.67E-01 -0.401 0.400 0.537 0.139

18 -6.23E-04 2.65E-01 -0.398 0.397 0.533 0.138

19 -4.51E-04 2.46E-01 -0.370 0.369 0.506 0.131

20 4.36E-03 1.78E-01 -0.264 0.269 0.418 0.111

Table A2 Fatigue Data for Coach Peel Specimens

40

Source: Material Properties

www.matweb.com

www.makeitfrom.com

3M Scotch-Weld Epoxy Adhesives DP125 Translucent and Gray Data Sheet

(www.datasheetlib.com)

Table A4 Test Data & FDP Values for Coach Peel

Specimens

Load (N) Cycles FDP

400.4 4548 55.86

355.9 7144 49.67

355.9 9209 49.67

333.7 8691 46.56

333.7 9207 46.56

311.4 8556 43.46

266.9 10949 37.25

266.9 14827 37.25

244.7 15913 34.15

222.4 30306 31.04

200.2 35259 27.94

178.0 193330 24.83

133.5 1680000 18.63

Table A3 Test Data & FDP Values for Lap Shear

Specimens

Load (N) Cycles FDP

2069 503 106.60

2069 499 106.60

1856 814 95.62

1856 975 95.62

1392 2377 71.72

1392 2015 71.72

1392 2354 71.72

1021 6965 52.60

928 12203 47.81

928 8180 47.81

928 8529 47.81

696 25872 35.86

696 22472 35.86

580 43368 29.88

510.4 109069 26.30

464 181578 23.91

464 211253 23.91

464 212422 23.91

440.8 255165 22.71

417.6 165454 21.51

394.4 444567 20.32

394.4 298807 20.32

371.2 353589 19.12

371.2 537190 19.12

324.8 10000576 16.73