16

CHAPTER 2

LITERATURE REVIEW

2.1 INTRODUCTION

Sufficient literature is available on various aspects of modeling of

the conventional welding processes but only limited literature exists on

modeling and simulation of FSW. From the literature it is evident that there is

scope for parametric study of FSW of aluminium alloy AA2014 using a

validated thermo-mechanical model. The literature on FSW of various

materials like aluminium, copper and magnesium alloys, microstructural

issues of welded materials, analytical and numerical modeling of FSW

process to predict thermal cycles and residual stresses are reviewed in this

chapter. The effect of process parameters on thermal history, weld properties

and microstructures of FSW of various alloys is also reviewed.

2.2 NEED FOR FSW OF ALUMINIUM ALLOYS

Aluminium alloys are extensively used in aircraft and defence

industries because of their high strength to weight ratio and stiffness to weight

ratio. Most primary structural components in air frames are made by

mechanical fastening or by machining them from solid material. Mechanical

fastening suffers from a weight penalty, difficulty of automation and

problems due to corrosion. Machining is a costlier process in terms of time,

energy and raw material. But welding can provide cost savings upto 30% and

weight savings upto 10% for typical airframe structures (Stewert 2001).

17

Fusion welding of commercial aluminium alloys is difficult. Some

aluminium alloys can be resistance welded, but the surface preparation is

extensive with surface oxidation being a major problem. The difficulty of

making high-strength, fatigue and fracture resistant welds in aerospace

aluminium alloys, such as highly alloyed 2xxx and 7xxx series, has long

inhibited the use of welding for joining aerospace structures (Rhodes et al

1997). These alloys are generally classified as non-weldable because of the

poor solidification microstructure and porosity in the fusion zone (Mishra and

Ma 2005). FSW is relatively a new solid state welding process which can be

used to join most aluminium alloys and surface oxide is no deterrent to the

process. No special cleaning techniques are required prior to welding. Thomas

et al (1997) demonstrated the possibilities of joining the aluminium alloy

plates of 1 – 70 mm thickness by FSW.

2.3 MECHANICAL PROPERTIES OF FRICTION STIR

WELDED JOINTS

Reynolds et al (2003) studied the mechanical properties of 304L

stainless steel friction stir welds. Experimental results showed that the tensile

property of the welded material with tool rotation 300 rpm was greater than

that of one welded with 500 rpm which, in turn more than that of base metal.

Ericsson et al (2003) studied the influence of welding speed on

fatigue strength of friction stir (FS) welds and compared the results with that

of metal inert gas welding (MIG) and tungsten inert gas welding (TIG). It was

found that the welding speed had no influence on the mechanical properties

and fatigue properties of FS welds in the tested tool rotations of 2200 and

2500 rpm and welding speed range of 700-1400 mm/min.

18

Liu et al (2003a) investigated the influence of welding parameters

on tensile strength and fracture behaviour of the 2017-T351 material.

Relations were established between welding parameter, revolutionary pitch on

tensile properties of aluminium alloy 2017-T351 welded by FSW. The study

showed that the fracture occurred at the interface between the weld nugget

and the TMAZ on the advancing side. The fracture locations of the joints

changed with the revolutionary pitches. When the revolutionary pitch was

0.02 mm/rev the fracture location of the joint was 4.1 mm from the weld

center. It was observed that the fracture location moved towards weld center

when the revolutionary pitch was increased. Liu et al (2003b) also

investigated the influence of welding parameters on tensile strength and

fracture behavior of 6061-T6. The investigations revealed that the fracture

during tensile test occurred at retreating side of the weld.

Peel et al (2003) claimed that tool design, tool rotation and welding

speed were important parameters which could be controlled precisely, thus

controlling the energy input into the system in FSW. Results showed that

weld properties were dominated by the thermal input rather than the

mechanical deformation caused by the tool. It was also found that increased

welding speed (reduced heat input) narrowed the weld zone and the same was

observed by hardness survey across the weld zone. Further, increased welding

speed resulted in decreased plateau width hardness due to lower heat input per

unit distance travelled.

Lee et al (2003) demonstrated the improvements in mechanical

properties of cast aluminium alloy A356 jointed by FSW. It was claimed that

the mechanical properties of the weld zone were improved when compared to

that of base metal. In particular, the tensile strength of the weld zone was

120 % of that of base metal. The hardness of the weld zone was more uniform

19

than that of the base metal because some defects were reduced and eutectic

silicon particles were dispersed over the stir zone.

Boz and Kurt (2004) investigated the influence of stirrer geometry

on FSW of aluminium alloy 1080 by choosing five different stirrer

geometries. Micro-examination of the weld zone and tensile test results

showed that best bonding with tensile strength 110 MPa was obtained with

0.85 mm and 1.1 mm screw pitched stirrer.

The effects of friction stir processing (FSP) on mechanical

properties of the cast aluminium alloys A319 and A356 were studied by

Santella et al (2005). The investigation showed reduced porosity and more

uniform distribution of second phase particles. The ultimate tensile strength,

ductility and fatigue life of both alloys were increased by FSP (Lee et al

2003).

Zhao et al (2005) investigated the influence of tool geometry on

mechanical properties of aluminium alloy 2014 joined by FSW. It was

reported that screw pitched taper stir pin had given good bonding and tensile

strength.

Abbasi Gharacheh et al (2006) studied the influence of the ratio of

“rotation speed/welding speed” (ω/υ) on mechanical properties of AZ31

magnesium alloy welded by FSW. It was reported that increasing the ratio

(ω/υ) leads to a decrease in yield and ultimate strengths of stir and transitional

zones. It was also observed that increasing the ratio (ω/υ) increased the weld

nugget size and decreased the incomplete root penetration.

Minton and Mynors (2006) experimentally proved the capability of

a conventional milling machine for FSW of 6082-T6 aluminium alloy sheets

20

of thicknesses 4.6 mm and 6.3 mm. Cavaliere et al (2008) studied the

mechanical properties and microstructural issues of the same material welded

by FSW.

The effects of FSW welding parameters such as welding speed and

tool rotation on microstructure in stir zone were studied by Kim et al (2006)

in ADC 12 alloy by measuring Si particle distribution. It was found the size of

the Si particles was influenced by welding speed, but it was not affected by

tool rotation.

The effects of FSW welding parameters such as welding speed and

tool rotation on tensile properties and fracture behavior of 6061-T651were

studied by Ren et al (2007). It was reported that the welding speed appeared

to be dominating factor in determining the tensile properties and fracture

modes.

Prado et al (2001) examined tool wear for the FSW of an

aluminium alloy 6061 with 20 volume % alumina (Al2O3) particle additions

and showed that there was no measurable wear and essentially zero wear

when commercial 6061 aluminium alloy was welded by FSW.

2.4 MICROSTRUCTURAL ASPECTS OF FSW OF

ALUMINIUM ALLOYS

The microstructural changes in various zones have significant

effect on postweld mechanical properties. The microstructural evolution

during the FSW process has been studied by a number of researchers they are

discussed in this section.

21

The effects of FSW on microstructure of aluminium alloy AA7075

were studied by Rhodes et al (1997). It was observed that the weld nugget had

a recrystallized, fine equiaxed grain structure in the order of 2-4 μm in

diameter. In contrast to the parent metal the dislocation density in the weld

nugget was found low. The recrystallization of the weld nugget grains and the

redistribution of the precipitates indicated that the temperature excursion

during joining was above the solution temperature for the hardening

precipitates, but below the melting temperature of the alloy. It was also

reported that temperature ranged between 450°C and 480ºC. Tang et al (1998)

studied temperature distribution and claimed that temperature at weld zone is

0.8 Tm (where Tm = melting point of material).The transition zone between the

parent metal and the weld nugget was characterized by a highly deformed

structure. Transmission electron microscopy (TEM) revealed that these grains

have not recrystallised as occurred in the weld nugget.

Murr et al (1998) have indicated that some of the precipitates were

not dissolved during welding and stated that the temperature rised to roughly

400°C in a friction stir welded aluminium alloy AA6061. The difference was

due to differences in soluabilities in aluminium alloys AA7075 and AA6061,

but the behavior of precipitate phenomenon and the temperature are not yet

well known.

Bussu and Irving (2003) investigated the role of residual stress and

heat affected zone properties on fatigue crack growth in friction stir welded

aluminium alloy 2024-T351. A comparative analysis of the results indicated

that crack growth behavior in the FSW joints was greatly dominated by the

weld residual stress and that microstructure and hardness changes in friction

stir welds had minor influence. Hence it was claimed that investigation of

residual stress in friction stir weld is necessary.

22

Su et al (2003) studied the microstructural aspects of aluminium

7050-T651 welded by FSW. It was reported that the microstructural

development in each region was a strong function of the local thermo-

mechanical cycle experienced during welding (Oertelt et al 2001). Various

zones like HAZ, TMAZ I, TMAZ II and nugget were identified in the weld

zone based on the differences in the microstructure (Su et al 2005).

Lienert et al (2003) demonstrated the feasibility of FSW of steels.

From the experiments it was concluded that tool rotation, welding speed, tool

geometry and downward force were important process parameters in FSW. It

was observed from the results that the stir zone and HAZ had greater yield

and tensile strength than the base metal.

Sato et al (1999) investigated the microstructure formation and

distribution, especially precipitate sequence, in friction stir welded aluminium

alloy AA6063, correlating to local thermal hysteresis and hardness. It was

also found that the shape of the weld zone may depend on the welding

parameters and thermal conductivity of the material. Simulated weld thermal

cycles, with different peak temperature and isothermal ageings were applied

to the base material to determine the local hysteresis and precipitation

sequence in various regions of the weld. The simulated thermal-cycles have

shown the effect of peak temperature on hardness of this alloy. The hardness

of the material did not change at temperatures lower than around 500 K

beyond which it decreased with the increase of the peak temperature. The

peak temperatures higher than 626 K produced roughly the same hardness

that of the solution treated base material. The precipitate distribution was not

effectively influenced by peak temperatures lower than 474 K. In temperature

range from 525 to 626 K, the density of the needle shaped precipitates

decreased with the increase in peak temperature. The peak temperatures

higher than 675 K (402ºC) lead to the dissolution of all precipitates. This

23

suggested that all precipitates may dissolve temporarily by rapid thermal

cycles. An isothermal ageing was carried out to observe reprecipitation at

these temperatures.

According to Threadgill (1997), the microstructure in a cross

section of a FSW joint is divided into several zones. The weld nugget in the

center of the weld was identified by the fine grains and characteristic ‘onion

ring’ structure. It was presented that the fine grain size was due to

recrystallisation process (Flores et al 1998, Benarides et al 1999). However, it

was noted that for aluminium alloys, recrystallisation was confined to the

nugget zone. Thus, in general, the nugget was considered as a part of the

TMAZ. For aluminium alloys, the TMAZ could be distinguished from the

nugget zone. In the TMAZ, the combination of high temperature and large

strains caused deformation of the grain structure, but no recrystallization took

place. Beyond this was the heat affected zone (HAZ), which was affected by

the heat but not by deformation. It was also reported that the hardness

variations across the weld joints due to microstructural changes (Jones et al

2003) had influenced on the ultimate tensile strength of these joints.

A detailed evaluation of the tensile properties of friction stir welded

aluminium alloy AA7075 was presented by Mahoney et al (1998). In the

transverse tensile tests, fracture occurred in the HAZ outside the weld nugget.

This was because the tool travel and rotation directions coincide on this side.

As a result of the large deformation imposed by the stirring during welding,

the TMAZ had been rotated, but not recrystallised.

Svensson et al (2000) investigated the behavior of two dissimilar

aluminium alloys AA5083 and AA6082 and found that fracture never

occurred close to the original joint line. Instead, it occurred mostly close to

the line where the shoulder of the tool had touched the top side of the weld in

24

AA6082. These fracture surfaces were also inclined, so at the bottom side of

the weld the fracture surface was closer to original joint line, but still

displaced by about 7 mm from it. In this case, a few specimens also had a

different appearance, with the fracture surface lying at about 5 to 7 mm from

the original joint line at the top side. It was deduced that the weld travel speed

had a large influence on the peak temperature distances from the nugget

boundary. Detailed hardness examination revealed a difference between welds

in the two alloys. In AA5083 a relatively constant hardness was found across

the welded joints, while in AA6082 welds a minimum in hardness occurred in

the HAZ.

Liu et al (1997) studied the microstructural aspects of aluminium

alloy AA6061-T6. The characterization of friction stir welded AA6061– T6

showed a dynamic continuous recrystallisation microstructure in weld zone.

The researchers also examined the dislocation content in the different regions

and found that the nugget zone had a much lower dislocation density than the

base material. Weld zone hardness varied between 55 and 65 HV and the

workpiece hardness varied between 85 and 100 HV.

The contribution of intense plastic deformation and high

temperature exposure results in recrystallisation precipitate dissolution and

coarsening of grains and precipitates within the stirred zone during FSW

(Mishra and Ma 2005).

2.5 MODELING AND SIMULATION OF WELDING

PROCESSES

A complex state of thermal and residual stresses is developed in

welded structures as a direct consequence of the non-uniform heat supplied

and subsequent cooling process. These stresses reduce the load carrying

25

capacity of the structures. The high level of stresses in the neighbourhood of

the weld joint can increase the tendency to brittle fracture. Aforesaid factors

demand the designer to know about the state of the residual stresses state

quantitatively and qualitatively. This knowledge would help them to use

appropriate stress relieving techniques. As welding is a multi-physics

problem, evaluation of the temperature distribution and residual stresses is

complicated according to Canas et al (1995) and Teng et al (1998). Due to the

presence of such complexity simple mathematical solutions cannot address

the practical manufacturing processes. Furthermore, currently it is also

difficult to obtain a complete mapping of the residual stress distribution in a

general welded structure with an experiment technique. Computational

simulation thus plays an effective role in the integrity analysis of such welded

structures.

In nuclear power plants the damage of components caused by the

mechanism of intergranular corrosion cracking was triggered mainly by weld

induced residual stresses. One solution of this problem that has been used in

the past involves experimental measurements of residual stresses in

conjunction with weld optimization testing. However, the experimental

analysis of all relevant parameters is a tedious process. Numerical simulation

using the finite element method (FEM) not only supplements this method but,

in view of current digital computing capabilities, is also an equally valid

alternative in its own right (Fricke et al 2001).

Teng et al (1998) presented that the advances in the field of

computer and the capacity of numerical techniques such as finite element

methods had enhanced the quality of residual stress analysis in welded

structures. Vilaca et al (2005) claimed that a validated model has the potential

to produce reliable information about the deformation and mixing patterns

26

that are important when designing FSW tools and thus should be capable of

producing welds free of defects and voids.

Many investigators have brought out analytical and experimental

methods to predict welding residual stresses. However, with advances in

computer technology and such techniques as the finite element method, the

means of analyzing residual stresses in welded structures enhanced even

further.

2.5.1 Thermal Modeling

Gould et al (1996) developed an analytical model based on

Rosenthal equation considering only the heat generated at the tool shoulder to

study how the heat was conducted into the plate. In this work, to estimate the

frictional heating, a line contact in the form of a ring was assumed between

the shoulder and workpiece. It was found that the temperature distribution

was asymmetric, with the leading edge considerably colder than the trailing

edge. This was due to the reason that the leading edge supplied heat to cold

material, while the trailing edge supplied heat to material already preheated

by the leading edge of the tool. In the program a relatively simple model for

the FSW process was developed. This model used a point heat sources,

integrated around the periphery of the local shoulder. Results of this model

were compared in a preliminary way with experiments. The model was basic,

and did not predict such features as stir zone shape.

Stewart et al (1998) developed two different models, (mixed zone

and the single slip surface models) to study the temperature and plastic flow

of the material. Mixed zone model used a concept of finite region of

continuous gradients of deforming material surrounding the pin tool. It

indicated that the actual deforming region might be more restricted than the

27

mixed zone model. Using a limited region slip, predictions of the shape of the

weld plug, the energy input, the forces and the maximum temperature were

found and all were in agreement with measures.

Little and Kamtekar (1998), studied the effects of thermal

conductivity on the computed temperature distributions. It was reported that

the transient temperature in a welded plate was significantly affected by the

value chosen for the thermal conductivity. A higher value of thermal

conductivity also leads to a more rapid fall in the temperature after the peak

temperature had been reached.

Chao and Qi (1998) published a 3-D heat transfer model, a trial and

error procedure was used to adjust the heat input until all the calculated

temperature matches with the measured.

Chao et al (2003) in the study found that only 5% of the heat

generated by the friction flows into the tool and rest flows to the workpiece.

Frigaard et al (2001) developed a model for FSW in which the heat

input was adjusted in such a way that the temperature at weld zone did not

exceed the melting point of the material.

Reynolds et al (2003), made the assessment of the tensile

properties, optical microstructure, and residual stress state of 304L stainless

steel and found that the specific weld energy for the weld made at 300 rpm is

1158 J/mm and for the 500 rpm weld, 1438 J/mm. In each case, it can be seen

that the higher tool rotation results in a higher rpm results in a higher

temperature. From the data presented, it can be assumed that the maximum

temperature and the time spent above any given temperature is greater for the

weld made at 500 rpm than for the weld made at 300 rpm.

28

Song and Kovecevic (2003) developed a thermal model to find the

temperature distribution in the workpiece. In the investigation the peak

temperature at weld center, TMAZ and HAZ were predicted as 820 K,

721-612 K and 439-612 K respectively.

Vilaca et al (2005) developed a thermal analytical model to

simulate asymmetric heat field developed below the tool shoulder due to the

composition of the rotation and linear speeds. The welding condition was

classified based on the value of the ratio between tool rotation and welding

speed. The difference that arises in heat flow between hot and cold welds are

that for cold welds the heat is mostly derived from viscous dissipation

(internal friction) due to large plastic flow deformation by the pin profile as

material is transported around the pin, and dissipated at the retreating side.

For hot welds, most of the plastic flow deformation is localised nearest to the

pin and the heat generated by the interfacial friction between the tool and the

parts is higher. The heat generated is almost equally distributed for both

advancing and retreating side. It was reported that hot welds allowed much

greater time for the temperature field to distribute throughout the weld zone.

Zhang et al (2006) developed a model to study the preheating period in FSW.

2.5.2 Thermo-Mechanical Modeling

Tekriwal and Mazumder (1988) found that the influence of

Poisson’s ratio was usually not significant. Free and Goff (1989) investigated

that a simplified modeling of the welding process of mild steel could lead to a

reasonable approximation of the final state of residual stress. It was suggested

that the factors like dependence of thermal properties on temperature, phase

transformation and the variation of the mechanical properties with the

temperature, (except the yield), did not appreciably affect the results. But it

was suggested that these suggestions could not be validated for the case of

stainless steel or aluminium.

29

Zhu and Cho (2002) studied the effect of thermal conductivity on

the distribution of transient temperature field during welding. The material

density and specific heat had negligible effect on temperature field. The yield

stress was the key mechanical property in welding simulation, since its value

had significant effect on the residual stress and distortion. If the room

temperature value of the yield stress was taken, the FEA computation

predicted zero residual stress and no plastic strain, occurred under this

circumstance. The temperature dependent yield stress property must be

considered in a welding process to simulate and obtain correct results.

Young’s modulus and the thermal expansion co-efficient had small effects on

the residual stress and distortion respectively in welding deformation

simulation. It was found that the numerical results obtained by using the room

temperature value of Young’s modulus were much better than those using

average value over the temperature history.

Peel et al (2003) observed that weld zone was in tension in both

longitudinal and transverse direction. Experimental results showed that

longitudinal stress increased with traverse speed. This increase was probably

due to steeper thermal gradients during welding and the reduced time for

stress relaxation to occur.

Ulysee (2003) parametrically studied the effect of process

parameters on temperature, axial load and flow stress. The support table,

located underneath of the workpiece was not included in the analysis in order

to reduce the size of the numerical model. The model of the workpiece region

was actually small when compared to that of the samples used in the

experiments. The tool pin having left handed threads is suitable for clockwise

tool rotation and vice versa. The maximum measured and predicted

temperature decreased when welding speed was increased (Bartier et al 2003).

Increase in temperature was observed while tool rotational speed was

30

increased. It was also observed that increase in welding speed, regardless of

TRS, had the effect of increasing the axial thrust and shear force on the pin. In

addition for a fixed welding speed, increasing the tool rotational speed had the

effect of decreasing the force acting on the pin. From the parametric study, it

was found that increasing the welding speed had the effect of increasing

magnitude of the forces, while increasing the rotational speed had the

opposite effect.

Chen and Kovecevic (2003) studied the relationship between the

calculated residual stresses of the weld and tool traverse speed. It was claimed

that the model could be extended to optimize the FSW process in order to

minimize the residual stress of the weld. Mechanical effect by the shoulder

was incorporated in the mechanical model, as the relatively large contact

region of the shoulder and workpiece was expected to contribute a larger part

of the mechanical stress, especially in the upper half part of the weld. The

prediction revealed that the release of the welded plates from the fixture

would affect the stress distribution of the weld.

Zhu and Chao (2004) conducted an inverse analysis for predicting

thermal cycles. The residual stresses in the welded plate were then calculated

using a three-dimensional elasto plastic thermo - mechanical simulation. The

difference of residual stress between the two cases (tool rotational speeds of

300 and 500 rpm) was small. Therefore the fixture release in the FSW should

be considered in the computer simulation for the determination of residual

stresses.

Chang and Teng (2004) developed a thermal elasto-plastic analysis,

using finite element techniques, to analyse the thermo- mechanical behavior

and evaluate the residual stresses in butt welded joints. The welding process

lead to a non-uniform temperature distribution associated with thermal strains

31

and localized plastic deformation. A large tensile longitudinal residual stress

occurred near the weld toe, and a compressive stress appeared away from the

weld bead. A high transverse residual tensile stress was produced near the

weld toe. Meanwhile, the stress approached zero as the distance from the weld

toe increased.

Vijay et al (2005) developed a thermo-mechanical model to predict

the transient temperature field and active stresses developed during FSW of

aluminium alloy AA6061. The thermal stresses constituted a major portion of

the total stress developed during the process. Boundary conditions in the

thermal modeling of process played a vital role in the final temperature profile

and thermal stresses. An attempt was made to predict realistic temperature of

the aluminium workpiece by applying adaptive boundary conditions. Contact

conductance between workpiece and back plate depends on the pressure at the

interface and has non-uniform variation. A finite element thermo-mechanical

model with mechanical tool loading was developed considering uniform value

of contact conductance for predicting the active stresses at the workpiece and

back plate interface. This pressure distribution contours were used for

defining the non-uniform adaptive contact conductance used in the model for

predicting the thermal history in the workpiece. The thermo-mechanical

model was then used in predicting the stress development in FSW.

Zhang et al (2005) developed a solid mechanics based 2D finite

element models to study the flow patterns and the residual stresses. It was

shown that the material flow on the advancing side and retreating side were

different. The distribution of the longitudinal residual stress along the

direction perpendicular to the weld line was a double feature curve. With the

increase of the translational velocity, the maximum longitudinal residual

stress could be increased. Furthermore, the rotational and transverse

movements of the tool would cause additional stress in the weld due to the

32

mechanical constraints of the plates by the fixture. The temperature of the

plate would be reduced to 25ºC and then the fixture would be removed to

obtain residual stress distributions.

The residual stresses were predicted during one-pass arc welding in

a steel plate using ANSYS finite element techniques by Teng et al (1998).

The effects of travel speed, specimen size, external mechanical constraints

and preheating on residual stresses were also discussed. Notably, the high

tensile stresses in the central region decreased with increasing length of the

specimen.

Zahedul et al (2006) studied the residual stresses caused by the

thermal cycles during FSW of metal without considered the plastic

deformation by sequentially coupled FE model. So the results of the model

deviated from the experimental data.

2.6 RESIDUAL STRESS DISTRIBUTION IN WELDMENTS

In welded structures, the regions which are subjected to tensile

stresses are prone to cracking. So it is important to understand the stress

distribution pattern in weldments. In this section, a brief account of various

components that will constitute the residual stress and the effects of these

individual components on the stress distribution pattern are discussed. In

welding, residual stresses are built during weld pool cooling. Residual stresses

built up in any welded construction include three components, namely

shrinkage stress, quenching stress and phase transformation stress (Balusamy

2001).

33

2.6.1 Residual Stresses due to Shrinkage

Residual stresses are developed when the shrinkage in the weld

zone and HAZ is prevented by the adjacent cold regions. As a consequence,

tensile residual stresses develop at the weld centre and compressive stress

elsewhere.

The tensile stresses in the weld grow with progressive cooling. Even if

the tensile stresses are diminished by relaxation, a stress distribution as shown

in Figure 2.1 (a) is observed for the residual stresses in the weld direction.

The equilibrium conditions ensure that the stress amplitude zero points occur

at the ends of the weldment. This distribution is typical for butt-welds in

unalloyed steels.

The shrinkage hindered in the lengthwise direction induces residual

stresses perpendicular to the weld (Figure 2.1 (b)). Due to shrinkage in the

direction perpendicular to the weld, tensile stresses occur at the centre of the

weld, which are in equilibrium with the compressive stresses near the end of

the weld. For long plates, the middle of the plate will be free of residual

stresses in this direction. Generally the maximum amplitude of the stress in

the transverse direction is similar than that in the longitudinal direction.

The amplitude of shrinkage residual stress grows with the ratio of

plate to weld thickness. The tensile residual stress is larger for smaller widths

of the weld, for higher thermal coefficient of expansion, and for higher

Young’s modulus. From the knowledge of shrinkage in welds, a ‘golden rule’

is that tensile stress occurs in the regions which are the last to cool.

34

Figure 2.1 Residual stress (a) Longitudinal (b) Transverse

2.6.2 Residual Stresses due to Quenching

The cooling rate has no influence on the residual stresses if

homogeneous cooling conditions are assured throughout plate thickness. In

welding, the zones near to the surfaces of the weld metal and HAZs cool

much faster than the other regions. If the thermal stresses become higher than

the yield strength, after cooling to room temperature, compressive residual

stress occurs in the near surface zone of the weld and the HAZ, with tensile

residual stress in the inner region of the weld.

Residual stress due to quenching increases with both decreasing the

tensile yield strength and increasing temperature gradients. They also increase

with plate thickness and the rate of cooling.

(a) (b)

35

2.6.3 Residual Stresses due to Phase Transformation

Phase transformation of alloys contributes to these stresses. This

transformation does not occur at the same time in near surface zones and in

the inner regions, nor does it occur at the same time in the weld and in HAZ.

As a consequence, tensile residual stress is produced in regions where the

transformation occurs first and compressive stress in regions where the

transformation occurs later.

2.6.4 Distribution of Residual Stresses

All the three effects contribute to residual stresses in the

weldments. In practice, the form of the residual stress distribution is

determined by the dominant of the three effects described above. In general,

the residual stress distribution in a weld has a W-form or an M-form. The

most conventional pattern of residual stresses takes the W-form, with the

tensile stresses along the weld centre line. This occurs when the shrinkage

residual stress is dominant compared to phase transformation residual stress.

The M-form of residual stress pattern occurs when the phase transformation

stresses are dominant compared to shrinkage stresses.

2.6.5 Reduction of Residual Stresses

Residual stresses during welding are unavoidable and their effects

on welded structures cannot be disregarded. Design and fabrication

conditions, such as the structure thickness, joint design, welding conditions

and welding sequence, must be altered so that the adverse effects of residual

stresses can be reduced to acceptable levels. Teng et al (1998) found that a

higher welding speed not only reduced the amount of adjacent material

affected by the heat of the arc, but also progressively reduced the residual

36

stresses. The important difference was in the fact that the high speed welding

technique produced a slightly narrower isotherm. This isotherm’s width

influenced the transverse shrinkage of butt welds. This was because the faster

welding speeds generally resulted in less residual stresses. Moreover, the

residual stresses depend on the final equilibrium temperature of the

temperature-stress cycle. Preheating treatments were used primarily to

influence the cooling rates within the weldment, thereby reducing the residual

stresses. Herein, the specimen was preheated homogeneously up to 200ºC,

300ºC and 400ºC. For the residual stresses distribution in a butt weld, the

middle weld bead was in tension and the magnitude of this stress was equal to

the yield stress. The ends of the weld were in compression. Owing to the

preheating treatment, the weldment significantly reduced the residual stresses

(Teng et al 1998).

Staron et al (2004) demonstrated the reduction of residual stresses

in friction stir welds by mechanical tensioning. Sheets were mechanically

tensioned to 70% of the tensile yield strength prior to welding. After welding,

the sheets were released. It was found that in the untensioned sheets, there

were tensile stresses in the longitudinal direction with peak values of about

130 MPa, but no significant stresses were present in the transverse direction.

Results indicated that mechanical tensioning during welding has introduced a

compressive strength in the weld zone. The width of the compressive stress

zone was approximately the same as the width of the stresses zone in the

untensioned reference sheet. The results reveal that it is possible to avoid

tensile residual stresses in the weld zone of FSW joints, which can have a

negative influence on mechanical properties of the welds under service

conditions.

37

2.7 SUMMARY

While there is a sufficient amount of literature available on the

various aspects of FSW of various alloys, only a limited amount of literature

exists on modeling and simulation of FSW for thermal cycles and residual

stresses and correlation of thermal history, mechanical properties and

microstructures with process parameters. Previous models have been built

with several assumptions as to how the material should be modeled, which

meshing schemes are better to use, how the temperature evolves, how heat

escapes from the welding area, and how this affects bonding. The material

response within the weld, as well as the post-weld microstructure that

develops, depend on how the material is heated, cooled, deformed and the

duration of these effects. This makes it imperative that an improved model to

be developed which includes changes in temperature, stress and strain

experienced in the FSW process as well as being able do so in a reasonable

amount of time. However, from the literature survey it is observed that the

investigations on prediction of thermal cycles and residual stresses during

FSW of aluminium alloy AA2014-T6 using validated numerical models

incorporating realistic boundary conditions have not been addressed in detail.

Systematic investigation on effects of FSW parameters such as

welding speed and tool rotation on thermal history, residual stresses and

mechanical properties of aluminium alloy AA2014-T6 welded by FSW has

not been addressed adequately.