11
ORIGINAL ARTICLE Thermo-mechanical finite element model of friction stir welding of dissimilar alloys Fadi Al-Badour & Nesar Merah & Abdelrahman Shuaib & Abdelaziz Bazoune Received: 17 October 2013 /Accepted: 31 January 2014 /Published online: 21 February 2014 # Springer-Verlag London 2014 Abstract A thermo-mechanical finite element model is de- veloped based on Coupled Eulerian Lagrangian method to simulate the friction stir welding of dissimilar Al6061-T6 and Al5083-O aluminum alloys using different tool pin profiles. The model is validated using published measured tempera- tures and weld microstructure. The finite element results show that maximum temperatures at the weld joint were below the materialsmelting point. Placing the harder alloy (Al6061-T6) at advancing side led to a decrease in maximum process temperature and strain rate, but increased tool reaction loads. Featured tool pin produced better material mixing resulting in enhanced joint quality with reduced volumetric defects. Keywords Friction stir welding . Dissimilar metals . Finite element modeling . Coupled Eulerian Lagrangian analysis 1 Introduction Solid state welding, such as friction stir welding (FSW), has been recognized as a better process for joining similar and dissimilar metals and alloys with different physical, chemical, and mechanical properties compared to conventional fusion welding techniques [13]. Fusion welding has many draw- backs including high residual stresses, impurities, voids, and gas pockets that increase the risk of corrosion and weld failure; this is in addition to the need for consumables, welder, and process qualifications. FSW has found a wide range of technological applications, especially in aviation and automo- tive industry because of its capability in producing high- quality joints of dissimilar metals and alloys [4]. FSW is a multi-physics problem that involves excessive material deformation and flow around the pin tool. Finite element (FE) modeling of the FSW process leads to a better understanding of the effect of the process parameters such as the tool geometry, rotational speed, and welding speed as well as the types and locations of defect on the weldment. Existing FSW finite element models can be classified into three types: thermal, thermo-mechanical non-flow base, and thermo- mechanical flow-based models. In flow-based models, tradi- tional Lagrangian elements become highly distorted and re- sults may lose accuracy. Several modeling techniques such as adaptive re-meshing and Arbitrary Lagrangian Eulerian (ALE) are often used to avoid high mesh distortion. Flow- based models are also developed using computational fluid dynamics (CFD) as well as Coupled Eulerian Lagrangian (CEL) analysis. A brief review of the research work involving thermo-mechanical models in FSW of metals and alloys is presented below. Schmidt and Hattel [5] developed a localized thermo- mechanical model considering Johnson-Cooks material mod- el to study the steady state FSW of 2xxx aluminum alloy. The solution used coupled temperature-displacement dynamic ex- plicit with ALE techniques. The predictions showed that the cooling rate plays a significant role in defect formation. The authors found that higher cooling rate leads to faulty deposi- tion of material behind the tool pin. Zhang and Zhang [6] used an approach similar to that of Schmidt and Hattel [5] to study the effect of welding parameters on material flow and residual stresses in friction stir butt weld of Al 6061-T6. The material flow around the FSW tool was investigated using tracer nodes. F. Al-Badour (* ) : N. Merah : A. Shuaib : A. Bazoune Mechanical Engineering Department, King Fahd University of Petroleum and Minerals, Dhahran 31261, Saudi Arabia e-mail: [email protected] N. Merah e-mail: [email protected] A. Shuaib e-mail: [email protected] A. Bazoune e-mail: [email protected] Int J Adv Manuf Technol (2014) 72:607617 DOI 10.1007/s00170-014-5680-3

Thermo-mechanical finite element model of friction stir welding of dissimilar alloys

  • Upload
    asu

  • View
    0

  • Download
    0

Embed Size (px)

Citation preview

ORIGINAL ARTICLE

Thermo-mechanical finite element model of fr iction stir weldingof dissimilar alloys

Fadi Al-Badour &Nesar Merah &Abdelrahman Shuaib &

Abdelaziz Bazoune

Received: 17 October 2013 /Accepted: 31 January 2014 /Published online: 21 February 2014# Springer-Verlag London 2014

Abstract A thermo-mechanical finite element model is de-

veloped based on Coupled Eulerian Lagrangian method to

simulate the friction stir welding of dissimilar Al6061-T6 and

Al5083-O aluminum alloys using different tool pin profiles.The model is validated using published measured tempera-

tures and weld microstructure. The finite element results show

that maximum temperatures at the weld joint were below the

materials’ melting point. Placing the harder alloy (Al6061-T6)

at advancing side led to a decrease in maximum process

temperature and strain rate, but increased tool reaction loads.

Featured tool pin produced better material mixing resulting in

enhanced joint quality with reduced volumetric defects.

Keywords Friction stir welding . Dissimilar metals . Finite

element modeling . Coupled Eulerian Lagrangian analysis

1 Introduction

Solid state welding, such as friction stir welding (FSW), has

been recognized as a better process for joining similar and

dissimilar metals and alloys with different physical, chemical,

and mechanical properties compared to conventional fusionwelding techniques [1–3]. Fusion welding has many draw-

backs including high residual stresses, impurities, voids, and

gas pockets that increase the risk of corrosion and weld

failure; this is in addition to the need for consumables, welder,

and process qualifications. FSW has found a wide range of

technological applications, especially in aviation and automo-tive industry because of its capability in producing high-

quality joints of dissimilar metals and alloys [4].

FSW is a multi-physics problem that involves excessive

material deformation and flow around the pin tool. Finite

element (FE) modeling of the FSW process leads to a better

understanding of the effect of the process parameters such as

the tool geometry, rotational speed, and welding speed as well

as the types and locations of defect on the weldment. ExistingFSW finite element models can be classified into three types:

thermal, thermo-mechanical non-flow base, and thermo-

mechanical flow-based models. In flow-based models, tradi-

tional Lagrangian elements become highly distorted and re-

sults may lose accuracy. Several modeling techniques such as

adaptive re-meshing and Arbitrary Lagrangian Eulerian

(ALE) are often used to avoid high mesh distortion. Flow-

based models are also developed using computational fluid

dynamics (CFD) as well as Coupled Eulerian Lagrangian(CEL) analysis. A brief review of the research work involving

thermo-mechanical models in FSW of metals and alloys is

presented below.

Schmidt and Hattel [5] developed a localized thermo-

mechanical model considering Johnson-Cook’s material mod-

el to study the steady state FSW of 2xxx aluminum alloy. The

solution used coupled temperature-displacement dynamic ex-

plicit with ALE techniques. The predictions showed that the

cooling rate plays a significant role in defect formation. Theauthors found that higher cooling rate leads to faulty deposi-

tion of material behind the tool pin. Zhang and Zhang [6] used

an approach similar to that of Schmidt and Hattel [5] to study

the effect of welding parameters on material flow and residual

stresses in friction stir butt weld of Al 6061-T6. The material

flow around the FSW tool was investigated using tracer nodes.

F. Al-Badour (* ) : N. Merah : A. Shuaib : A. BazouneMechanical Engineering Department, King Fahd University ofPetroleum and Minerals, Dhahran 31261, Saudi Arabiae-mail: [email protected]

N. Merahe-mail: [email protected]

A. Shuaibe-mail: [email protected]

A. Bazounee-mail: [email protected]

Int J Adv Manuf Technol (2014) 72:607–617

DOI 10.1007/s00170-014-5680-3

The results of their investigation showed that increasing the

tool rotational speed and decreasing the welding speed could

improve the weld quality. However, flash formation would be

more pronounced when rotational speed is increased. Their

results showed that longitudinal residual stresses are higher

than the transverse ones. Al-Badour et al. [7] developed a CEL

finite element model to simulate FSW and study the effect ofwelding conditions on weld quality. The model considered the

workpiece as localized Eulerian region, while the FSW tool as

Lagrangian. The study takes into account the effect of the

coefficient of friction, tool welding speed, and tool forging

control method. The outcomes revealed that force control

welding can enhance joint quality based on minimizing the

weld defect size.

FSW of dissimilar aluminum alloys has been intensively

investigated by experimental methods [8–13]. Only a fewnumerical models, which investigate the FSW of dissimilar

alloys or metals, were found in literature [10, 14].

Many research groups [8–13] have experimentally investi-

gated the effect of material position on the joint mechanical

and microstructural properties. Park et al. [8] showed that a

proper mixing of aluminum 5052-H32 and 6061-T6 was

formed when softer material was placed at the advancing side.

Using a similar combination of aluminum alloys, Hong et al.[9] observed limited mixing in the stir zone (SZ) regardless of

the position of the harder alloy. Jamshidi et al. [10] found that

the average process temperature was higher when the harder

material was placed at the advancing side of the tool. Peak

temperature and cooling rate in similar joints of AA 6061-T6

were found to be higher than other similar and dissimilar

material FSW for the same welding conditions. Khodir and

Shibayanagi [11] noticed that no mixing occurs at low rota-

tional speed whereas at high speed (i.e., 2,000 rpm), thematerial at the stir zone consisted mainly from the base mate-

rial located at advancing side of the tool. Lee et al. [12] found

that a wider bead was formed when softer alloy was placed at

the advancing side of the tool. Tensile tests of weldments of

different dissimilar aluminum alloys revealed that only small

improvements in joint strength were achieved by placing the

softer alloy at the advancing side of the tool [8–13].

Aidun et al. [14] presented a thermal model of dis-

similar FSW for joining Al-6061-T6 to Steel AISI 1018

using time varying functionality graded material to sim-

ulate the nugget zone thermal properties. The model

considered the FSW tool as a moving heat source.

The material properties in the weld nugget were simul-

taneously updated by user subroutines. Peak tempera-tures of the process were found in the steel side, and

the temperature history in that side was steeper than that

of aluminum, which was expected due to the lower

thermal conductivity and lower specific heat of AISI

1018.

In the simulation part, Jamshidi et al. [10] presented

a thermo-mechanical model similar to the one developed

by Schmidt and Hattel [5] to simulate similar and dis-

similar FSW of AA 6061-T6 with AA 5086-O. Fordissimilar material FSW, partitioning technique was

employed to assign material properties to each part of

the workpiece assuming Eulerian boundaries between

the dissimilar plates.

In finite element modeling (FEM), Lagrangian elements

cannot handle void or multi-materials. Such elements need to

be fully filled with single material in order to satisfy continu-

ity. On the other hand, Eulerian analysis is based on volume-of-fluid method where material is tracked as it flows through

the mesh by computing its Eulerian Volume Fraction (EVF)

within each element. Therefore, Eulerian elements can handle

multi-material as well as voids formation, giving an advantage

over Lagrangian and ALE [7, 15].

In the present work, a Coupled Eulerian Lagrangian

model is developed to simulate the friction stir welding

process of dissimilar aluminum alloys (Al 6061-T6 with

Al 5083-O). The proposed CEL model has a number ofadvantages over ALE; it can predict the material distri-

bution across the weldment as well as the formation of

voids. Different weld configurations (materials position)

and FSW tool pin profiles are considered in the analy-

sis. The model is validated using published measured

temperatures, at similar welding conditions, and by

qualitative comparison of weld microstructure.

Fig. 1 a Schematic representation of FSW of dissimilar butt joint; b idealization of localized FSW model for dissimilar materials

608 Int J Adv Manuf Technol (2014) 72:607–617

2 Finite element model

2.1 Problem idealization

The butt welding process of dissimilar materials using FSW is

modeled by CEL technique and analyzed using coupled

temperature-displacement, explicit method available inAbaqus 6.11-3 [16]. The model is an extension of an earlier

work on FSW of similar materials by the authors [7]. The

geometrical model considers a localized region incorporating

both tool and processed zone shown by the dashed line in

Fig. 1a. The welding phase is simulated by employing a

control volume approach, whereas the welding speed is de-

fined as inflow and outflow over Eulerian domain boundaries

as described in Fig. 1b at speeds equal to those of tool welding

speeds. On the other hand, the FSW tool is considered as arigid Lagrangian body. The tool workpiece interaction is

governed by Coulomb’s frictional law and is implemented

using general explicit contact.

During FSW process, heat is generated by frictional

heating and plastic deformation. Heat is assumed to be purely

produced by inelastic work during sticking condition while

frictional heat during slipping is ignored. The generated heat

dissipates into the workpiece by conduction, while dissipa-

tions to surroundings are considered in the form of free con-vection and radiation. Temperature distribution is thus obtain-

ed by solving the governing heat transfer equation taking into

account the above three modes of heat transfer.

2.2 Geometrical model

As mentioned above, a localized region of the workpiece is

included in the analysis. The size of the Eulerian domain is

considered based on stress-free inflow conditions. The

domain size is considered to be four times larger than the tool

shoulder diameter [5–7, 10]. This size is a compromise be-

tween accuracy and computational time. The Eulerian domain

is generated in a cuboidal shape, which is partitioned initially

into three zones: void, workpiece 1, and workpiece 2 (shown

in Fig. 1b). The domain volume is taken as 80×80×6 mm3.

The actual workpiece thickness is 5 mm; the domain is con-sidered to be thicker than the physical workpiece by 1 mm,

which was left initially empty (void). This empty layer or

difference in thickness is included for visualizing flash forma-

tion during welding.

The FSW tool shape and size are similar to those used in

the work of Jamshidi et al. [10]. The FSW tool has a concave

shoulder and tapered pin. Two different tool pin profiles are

used to study the effect of tool pin features on developed joint.

Simulation is conducted for tri-flutes of 0.5 mm depth pin andsmooth tapered pin tool. Figure 2a illustrates the FSW tool

geometrical properties, while Fig. 2b presents the featured tool

pin in 3D.

2.3 Processed materials model

Material plasticity is governed by Johnson-Cook [17] model,

which is selected to include the effects of temperature, strain,

and strain rate. The expression that relates flow stress σo,strain rate ε̇ , and temperature T is given by Johnson-Cook’s

semi-empirical formula (1):

σo ¼ A þ Bεn

pl 1 þ Clnε̇ pl

ε̇ o

!

1−T−Tref

Tmelt−Tref

m

ð1Þ

Fig. 2 a FSW tool geometricproperties, dimensions in (mm);(b) 3D representation of thefeatured tool pin

Table 1 Johnson-Cook plasticity model constants for Al-6061-T6 andAl-5083-O

Alloy A[MPa]

B[MPa]

C n M Tref

[K]Tmelt

[K]

Al-6061-T6 324 114 0.002 0.42 1.34 297 856

Al-5083-O 170 425 0.0335 0.42 1.225 297 913

Table 2 Simulated weld conditions with tool rotational speed, N=900 rpm, and welding speed, V=150 mm/min

Case no. Advancingside material

Retreating sidematerial

FSW tool

1 6061-T6 6061-T6 Featured

2 6061-T6 5083-O Featured

3 5083-O 6061-T6 Featured

4 6061-T6 5083-O Smooth pin

5 5083-O 6061-T6 Smooth pin

Int J Adv Manuf Technol (2014) 72:607–617 609

where εpl is the effective plastic strain, ε⋅

pl is the effective

plastic strain rate, ε̇ o is the normalizing strain rate (typically

1.0 s−1), and A, B, C, n, and m are material constants. The

exponent n takes into account the effect of strain hardening

while m models the thermal softening effect. C representsstrain rate sensitivity. Tref represents the temperature where

parameters A, B, and n are evaluated, while Tmelt is the

material solidus temperature. Johnson-Cook parameters of

Al-6061-T6 and Al 5083-O that were used in this analysis

are shown in Table 1. The material model also takes into

consideration the effect of temperature on elastic and thermal

properties of both alloys. Thermal and mechanical properties

as well as Johnson-Cook parameters of Al-6061-T6 and Al

5083-O, used in this analysis, were taken from references[18–20].

2.4 Interactions, loads, and boundary conditions

Since Eulerian mesh is rigid, boundary displacement con-

straints are not required, but velocity constraints along the

Eulerian domain are needed to prevent the material from

flowing out of the boundaries. As mentioned earlier, the

welding phase is simulated using control volume approach,

where the tool rotates around its centroidal axis, while thematerial inflow across the Eulerian boundary simulates the

tool advancing along the seam line. The rigid tool is

constrained to a reference point to assign tool plunging feedand rotational speed. The simulation assumes displacement

control FSW. Therefore, during welding phase, the tool is

constrained in axial and radial directions.

The following welding conditions are considered in this

investigation: plunging feed, f=100 mm/min; tool rotational

speed, N=900 rpm; and welding speed, V=150 mm/min.These conditions are selected from the experimental work by

Jamshidi et al. [10] for the validation of the present model. The

tool tilting angle is taken to be 2°. Emissivity of 0.09 isemployed for commercial aluminum [21]. The maximum val-

ue of the convective heat transfer coefficient, which ranges

between 5 and 25 W m−2°K−1[22] is assumed. Heat conducted

to backing plate is included in the analysis by employing a

fictitious convective heat transfer coefficient of

1,000 W m−2°K−1 [5, 6, 10]. The interaction between the

Lagrangian tool and the Eulerian material is governed by

Coulomb’s friction contact law and a coefficient of friction of

0.8 is used [7], representing maximum coefficient of frictionbetween tool steel and aluminum. According to the above

Fig. 3 a Material assignment inthe Eulerian domain; b numericalmesh

440

490

540

590

640

690

740

40 42 44 46 48 50 52 54 56 58

Tem

pera

ture

[K

]

Measured [10] Simulated

Fig. 4 Temperature history at10 mm from weld centerline atadvancing side of case (1)

610 Int J Adv Manuf Technol (2014) 72:607–617

model properties, five different welding conditions were sim-

ulated as indicated in Table 2.

2.5 Mesh generation

The Eulerian domain is meshed with 82,800 multi-

material, thermally coupled 8-node Eulerian elements

(EC3D8RT). Each element has 4 degrees of freedom

at each node. In order to keep the material from pene-trating the tool surface and to enhance the computation-

al efficiency, bias mesh technique is applied to generate

fine mesh at the tool-workpiece interaction zone andcoarse mesh at the sides. To simulate dissimilar material

FSW, each of the three zones, described earlier in Fig. 1b, is

initially assigned with one material using uniform material

assignment tool. Figure 3a, b shows the material assignment

(by color) and the numerical mesh, respectively.

3 Validation

Validation of the present model is performed by comparing

estimated steady-state temperatures with those measured byJamshidi et al. [10] for the same FSW processing conditions at

10 mm from the weld centerline at both advancing and

retreating sides. Furthermore, the macrostructure analysis per-

formed by Silva et al. [25] is used to validate the estimated

distribution of materials along the processed zone.

The estimated temperature histories at advancing side,10 mm from tool center, is compared to one measured by

[10] for similar welding of case 1 and shown in Fig. 4. It can

be seen that the model overestimates the peak temperature.Table 3 presents a comparison between estimated peak tem-

peratures, 10 mm from tool centerline at both advancing and

retreating sides, with those measured by [10]. The estimated

values are found to be within the range of 8–15 %. Part of the

discrepancy between estimated and measured values may be

Table 3 Comparison between measured and estimated temperatures at 10 mm from the weld centerline

Advancing Retreating

Workpiece Present work estimated temp.[K]

Measured temp. [K][10]

Error (%) Present workestimated temp.[K]

Measured temp.[K][10]

Error (%)

6061-6061 734 678 8.25 695 640 8.6

6061-5083 736 646 13.9 715 621 15.1

(b)

(a)

6061Adv. (6061,7075)

Ret. (5083, 2024)

Section line

Tool

(c)

(d)

5083

6061

Section line

Tool

Adv. (6061,7075)

Ret. (5083, 2024)

(e)

(f)

5083Tool advancing

direction

Fig. 5 Material distribution: a present model section on advancing side, b macrostructure analysis [25] advancing side, c present model section onretreating side, d macrostructure analysis [25] retreating side, e and f location of sections along the welding direction

Int J Adv Manuf Technol (2014) 72:607–617 611

due to the precision in measuring temperatures, as the authors

in [10] reported poor thermocouple connectivity.

The material distribution within the processed zone is

computed and presented using EVF. These results are com-

pared with weld microstructure reported by Silva et al. [25]

who experimentally investigated the effect of welding condi-

tions on material flow of FSWof dissimilar butt weld (7xxx to

2xxx). Figure 5a shows the estimated material distribution

along the weld at an offset from the seam line into theadvancing side. The simulation results revealed that the ma-

terial in the retreating side (soft) moved to the upper region of

the processed zone: behind the tool, while the material in the

advancing side (hard) moved to the lower region. This result

qualitatively matches the distribution revealed by Silva et al.

for a friction stir welded Al 7075-Al 2024 at a tool rotational

speed of 1,000 rpm (see Fig. 5b). On the other hand, the

section along the welding with an offset to the retreating side

(Fig. 5c) shows that the material at the advancing side flows

around the FSW tool and moves to the upper region of the

processed zone. Although the distribution of material ahead of

the tool pin matches macrostructure reported by Silva et al. in

Fig. 5d, the material distribution behind the tool does not. The

macrostructure in [25] showed that the material at the advanc-

ing side (Al 7075) will flow around the tool and settles at the

bottom of the processed zone, while the one predicted by the

model (Fig. 5c), the material at advancing side, was found toflow around the tool and settle at the top. The apparent

contradiction between the experimental results and the esti-

mated ones could be attributed to the difference in the tool pin

profile. Silva et al. [25] used a threaded tool pin, which can

produce a larger axial flow when compared to the tool used in

the present model (tapered pin with tri flutes). Figure 5e, f

shows a schematic diagram of the welded workpiece with

sections locations used in the validation process.

Fig. 6 Estimated temperaturedistributions across the weldingzone, a case (3) and b case (2);temperature in Kelvin

Fig. 7 Estimated mechanical strain rate ε̇ , top view of processed zone beneath FSW tool; a case (3) and b case (2), section view across the weld center; ccase (3) and d case (2)

612 Int J Adv Manuf Technol (2014) 72:607–617

4 Results and discussion

The CEL technique, described in Section 2 above, is used to

study the effects of the position of the harder material (Al-

6061) relative to the softer material and of the tool pin profile

on the quality of weld joints produced by FSW. The parame-

ters under investigation include temperatures distributions,

mechanical strain rate, plastic deformation, materials distribu-

tion, von Mises stresses across the welded joint, and material

velocity profiles developed around the FSW tool. Results are

presented and discussed in the following sections.

4.1 Effect of Al-6061-T6 position

The estimated temperature distributions during FSW using

featured tool pin, depicted in Fig. 6, show that the temperaturegradient in the soft material (Al 5083) is higher than that in the

hard material (Al 6061), irrespective of Al 6061-T6 position.

It is worth mentioning that the maximum predicted tempera-

ture about 900 K, in cases (2) and (3) is found to be below Al-

5083 solidus temperature of 913 K, which means the joint is

formed in a solid state.

Placing the hard material at the advancing side of case (2)

resulted in slightly lower process temperatures, about 20–

50 °C, when compared to estimated temperature in case (3),

as it can be shown from Fig. 6. These results are in agreementwith the observations of Lee et al. [12] and Amancio et al.

[13]. Lee et al. [12] found that when Al 6061-T6 was placed at

advancing and A356 at retreating side, recrystallization in the

stir zone (SZ) was driven by dynamic recrystallization pro-

cess, which indicates low-process temperature. On the other

hand, placing softer material (A356) at advancing side led to

disappearing of precipitates in the SZ as a result of reaching

material resolving temperature. Amancio et al. [13] investi-

gated FSW of dissimilar Al alloys, 2024-T351 to Al 6056-T4.Formation of lamellae was observed in the SZ when 2xxx

alloy was placed on the advancing side, limited chemical

mixing was also reported. These findings indicate a low

process temperature when harder alloy is placed at advancing

side. On the other hand, Jamshidi et al. [10] and Firouzdor and

Kou [24] recorded higher welding temperatures when harder

material was placed at the advancing side. Temperature sur-

faces were measured at the top [10] and bottom [24] of the

workpiece.The controversy among the above measurements and the

present FE results are mainly due to neglecting frictional heat

in the model, which is a result of the limitation of the used

software [16]. Placing the harder alloy at the advancing side

Fig. 8 Estimated average equivalent plastic strain (PEEQAVG) across the weld of dissimilar alloy a case (2) and b case (3)

6061 (Ret.)5083 (Adv.)

6061 (Adv.)5083 (Ret.)

(a)

(b)

Fig. 9 Estimated von Mises stress distribution, a case (3), b case (2); stress in Pascal

Int J Adv Manuf Technol (2014) 72:607–617 613

would create a greater tool-workpiece slippage, as a result of

material flow resistance, which leads to domination of fric-tional heat over plastic work heat. Therefore, higher temper-

ature would be expected to develop at the surface [10].

Moreover, placing the softer material at advancing side would

lead to less slippage; therefore, less frictional heat and more

plastic work are produced.

When harder alloy is placed at advancing side, large mate-

rial resistance can be supported by tool reaction loads.

Variation in tool reaction forces and torque were detected by

the developed model. When harder material is placed at theadvancing side case (2), the tool axial force and torque were

found to increase by 15 and 30 %, respectively. These findings

are in agreement with Cavaliere et al. [23] and Firouzdor and

Kou [24].The increase in process temperature predicted by the model

in case (3) is due to the increase in plastic work. This can be

indicated by the mechanical strain rates values depicted in

Fig. 7a, b, and the larger equivalent plastic strain is shown in

Fig. 8a, b. Higher strain rates and larger plastic deformations

were found when the softer material (Al 5083) was placed at

the advancing side case (3). This resulted in an asymmetrical

strain rate distribution across the weld. The strain rates at the

advancing side were much larger than those of the retreatingside (Fig. 7a), indicating less tool slip rate. Furthermore, strain

rate distribution in case (2) was more uniform, as strain rate at

Al 5083-O Ret.Al 6061 Adv.

Al 5083-O Adv.Al 6061 Ret.

(a)

(b)

Tool central axis

Fig. 10 Material distributions across the weld, a case (2) and b case (3), black 6061 and gray 5083

Void

(a)

(b)

Fig. 11 Estimated void formation during FSWof dissimilar Al alloys, a Al 5083-O (adv.)-Al 6061-T6 (ret.) case (5), b 6061-T6 (adv.)-5083-O (ret.) case(4)

614 Int J Adv Manuf Technol (2014) 72:607–617

advancing and retreating sides were of same level of magni-

tude, while a thicker flow arm in Al-6061 side was formed

when compared to case (3) (Fig. 7c, d). These results were

found to be in agreement with those of Lee et al. [12] whofound that wider beads were formed when softer material was

placed at advancing side.

Estimated von Misses stress distributions across thewelding direction for cases (2) and (3) showed that high

stresses were developed at the retreating side of the harder

plate case (3) as shown in Fig. 9a. Similar to the strain rates,

stresses developed in case (2) across the plate were found to be

smaller compared to the ones developed in case (3), except at

localized regions under the circumference of the tool shoulder(Fig. 9b), high localized stresses were observed and may have

been brought by lower process temperature or tool axial force.

In addition, stress distribution profiles show that material incontact with the tool has almost zero flow stress, which is a

result of using Johnson-Cook material model, as the material

approaches solidus temperature, the flow stress approaches

zero.

Material distributions across the processed zone using EVF

are shown in Fig. 10a, b for cases (2) and (3), respectively. In

case (2), mixing happened in a thin layer below the tool

shoulder, and penetration of harder material into retreating

side was found to be limited. Similar observations were made

by Silva et al. [25] who explained the cause of limited mixingin the SZ to excessive material flow under the tool shoulder.

On the other hand, more material mixing took place through-

out the plate thickness in case (3) in which more material at the

advancing side moved to the retreating side (Fig. 10b).

Moreover, the stir zone was mainly composed of material at

the advancing side. Similar observations are reported by a

number of researchers [8, 9, 11]. The effect of harder alloy

location (Al 6061) was also studied in cases (4) and (5). The

welding parameters were kept similar to those in cases (2) and(3), except that a tapered smooth pin was used. The numerical

simulations showed that larger voids developed behind the

FSW tool when the harder material was placed at theretreating side case (5), as shown in Fig. 11.

It can be inferred from the above results that placing the

harder material at the advancing side may result in a better

weld quality, regardless of tool profile used. This is due to the

observed lower process temperatures, uniform stress distribu-

tions, symmetric strain rates, and smaller void formation. On

(a) (b)

void

Fig. 12 Equivalent plastic strain distributions for a case (4) and b case (2)

Fig. 13 Weld material velocity profiles for a case (2), b case (4); velocity in meters per second

Int J Adv Manuf Technol (2014) 72:607–617 615

the other hand, placing the softer alloy on the advancing side

resulted in a better material penetration between the welded

plates. Therefore, better bonding between the welded plates is

expected. The slightly larger process temperature also reduced

the reaction loads on the FSW tool, which can be counted as

an advantage.

4.2 Effect of tool profile

It is known that pin tool features such as threads, flats, or flutesmay enhance the material flow and lead to better material

mixing and eventually better weld strength. The effects of tool

features on developed equivalent plastic strain and material

velocity are shown in Figs. 12 and 13, respectively. The

equivalent plastic strain values for welds performed with a

featureless tool pin (Fig. 12a) are smaller than those developed

using featured tool pins (Fig. 12b). Moreover, when a featured

tool was used, no sign of the nucleation of voids or volumetric

defects were found in all simulation results; however, usingfeatureless tool pin led to volumetric defect formation as can

be shown in Fig. 12a. Material velocity around the FSW tool

is found to be higher for welds performed with featured tool

(Fig. 13). Estimated material velocity results show that more

sticking occurs when featured tool is used, as average slip rate

is found to be 0.2 in case (2) and around 0.7 in case (4). The

slip was calculated using equation (2):

δ̇ ¼ 1−Vmat=V tool ð2Þ

where Vmat is material velocity and Vtool is the maximum toolvelocity. It is clearly visible from the flow results (Fig. 13) that

the velocity at the trailer side vanishes when featureless tool

pin was used, which is responsible for void formation.

5 Conclusions

A thermo-mechanical finite element model is developed based

on Coupled Eulerian Lagrangian method to simulate FSW

process of dissimilar Al 6061-T6 and Al 5083-O materials.

The model is implemented in Abaqus Explicit and validatedusing published measured temperatures and microscopic anal-

ysis of material flow. The effects of hard Al 6061 material

position and tool pin profile on weld and material behaviorwere investigated and the following conclusions are drawn:

& The maximum operating temperature of about 900 K is

below the minimum melting temperature of welded mate-

rials for both similar and dissimilar FSW, which is of the

order of 914 K.& Placing the harder material at the advancing side reduces

the maximum process temperature and mechanical strain

rate and produces stress distributions that are relatively

similar in both advancing and retreating sides.

& The flow arm thickness was larger when harder material is

placed at the advancing side.

& Better material mixing is obtained using a featured tool

when softer material is placed at the advancing side.

& Featured tool pin produces more plastic flow; therefore,enhanced joint quality can be achieved by eliminating

volumetric defects.

& The use of featureless pins may result in volumetric de-

fects at the trailer side as proved by the vanishing velocity

profiles.

& Welds performed with featureless tool pins are found to be

more sensitive to the location of the hard material, placing

the harder material at the advancing side may enhance the

material flow.

Acknowledgments The authors acknowledge the support provided byKing Abdulaziz City for Science and Technology (KACST) through theScience and Technology Unit at King Fahd University of Petroleum andMinerals (KFUPM) for funding this work through Project No. NSTP 11-ADV2130-04 as part of the National Science, Technology and InnovationPlan.

References

1. Dubourg L, Merati A, Jahazi M (2010) Process optimisation andmechanical properties of friction stir lap welds of 7075-T6 stringerson 2024-T3 skin. Mater Des 31:3324–3330

2. Ericsson M, Sandstrom R (2003) Influence of welding speed on thefatigue of friction stir welds, and comparison with MIG and TIG. Int JFatigue 25:1379–1387

3. Moreira PMGP, de Figueiredo MAV, de Castro PMST (2007) Fatiguebehaviour of FSW and MIG weldments for two aluminium alloys.Theor Appl Fract Mech 48:169–177

4. Mishra RS, Mahoney MW (2007) Friction stir welding and process-ing. ASM International Materials Park, Ohio 44073–0002

5. Schmidt H, Hattel J (2005) A local model for the thermomechanicalconditions in friction stir welding. Model Simul Mater Sci Eng 13:77–93

6. Zhang Z, Zhang HW (2009) Numerical studies on controlling ofprocess parameters in friction stir welding. J Mater Process Technol209:241–270

7. Al-Badour F, Merah N, Shuaib A, Bazoune A (2013) CoupledEulerian Lagrangian finite element modeling of friction stir weldingprocesses. J Mater Process Technol 213(8):1433–1439

8. Park SK, Hong ST, Park JH, Park KY, Kwon YJ, Son HJ (2010) Theeffect of material locations on the properties of friction stir weldingjoints of dissimilar aluminium alloys. Sci Technol Weld Join 15:331–336

9. Hong ST, Kwon YJ, Son HJ (2008) The mechanical properties offriction stir welding (FSW) joints of dissimilar aluminum alloys.Proc. 1st Int. Symp. on Hybrid materials and processing, Busan,South Korea, October 2008, Paper 69

10. Jamshidi AH, Serajzadeh S, Kokabi AH (2009) Evolution of micro-structures and mechanical properties in similar and dissimilar frictionstir welding of AA5086 and AA6061. Mater Sci Eng A 528:8071–8083

616 Int J Adv Manuf Technol (2014) 72:607–617

11. Khodir SA, Shibayanagi T (2007) Microstructure and mechanicalproperties of friction stir welded dissimilar aluminum joints ofAA2024-T3 and AA7075-T6. Jpn Inst Metals Mater-T 48–7:1928–1937

12. Lee WB, Yeon YM, Jung SB (2003) The mechanical propertiesrelated to the dominant microstructure in the weld zone of dissimilarformed Al alloy joints by friction stir welding. J Mater Sci 38:4183–4191

13. Amancio-Filho ST, Sheikhi S, dos Santos JF, Bolfarini C (2008)Preliminary study on the microstructure and mechanical propertiesof dissimilar friction stir welds in aircraft aluminium alloys 2024-T351 and 6056-T4. J Mater Process Technol 206:132–142

14. Aidun K, Li D, Marzocca P (2009) Time-varying functionally gradedmaterial thermal modeling of friction stir welding joint of dissimilarmetals, trends in welding research. Proceedings of the 8thInternational Conference, Stan A. David, Tarasankar DebRoy, JohnN. DuPont, Toshihiko Koseki, Herschel B. Smartt, editors, 731–735.

15. Fadi Al-Badour (2012) Numerical and experimental investigations offriction stir welding of tube-tubesheet joints. Ph. D. Thesis, KingFahd University of Petroleum and Minerals, Dhahran, KSA

16. (2011) Abaqus ‘software’ 6.11-317. Johnson GR, Cook WH (1983) A constitutive model and data for

metals subjected to large strains, high strain rates and high tempera-tures. 7th International Symposium on Ballistics, Netherlands, pp.541–547

18. Soundararajan V, Zekovic S, Kovacevic R (2005) Thermo-mechanical model with adaptive boundary conditions for friction stirwelding of Al 6061. Int J Mach Tool Manuf 45:1577–1587

19. Gray GT III, Chen SR, Wright W, Lopez MF (1994) Constitutiveequations for annealed metals under compression at high strain ratesand high temperatures. Los Alamos National Laboratory, LosAlamos

20. Lesuer DR, Kay GJ, Leblanc MM (2001) Modeling large-strain,high-rate deformation in metals. Report no. UCRL-JC-134118,Lawrence Livermore National Laboratory, Livermore, Canada

21. (2013) Omega, Table of Total Emissivity, http://www.omega.com/temperature/Z/pdf/z088-089.pdf. Accessed Oct 12 2013

22. Incropera and Dewitt (2001) Fundamentals of heat and mass transfer,6th Ed. Wiley

23. Cavaliere P, De Santos A, Panella F, Squillace A (2009) Effect ofwelding parameters on mechanical and microstructural properties ofdissimilar AA6082–AA2024 joints produced by friction stir welding.Mater Des 30:609–616

24. Firouzdor V, Kou S (2010) Al-to-Mg friction stir welding: effect ofmaterials position, travel speed and rotational speed. Metall MaterTrans A 41A:2914–2935

25. da Silva AAM, Arruti E, Janeiro G, Aldanondo E, Alvarez P,Echeverria A (2011) Material flow and mechanical behaviour ofdissimilar AA2024-T3 and AA7075-T6 aluminium alloys frictionstir welds. Mater Des 32:2021–2027

Int J Adv Manuf Technol (2014) 72:607–617 617