5th International & 26th All India Manufacturing Technology, Design and Research Conference (AIMTDR 2014) December

12th–14th, 2014,
IIT Guwahati, Assam, India

Finite Element Simulation of Temperature and Strain Distribution in

Al2024 Aluminum Alloy by Friction Stir Welding

Rahul Jain1*, S.K. Pal2, S.B. Singh3

1
Department of Mechanical Engineering, IIT Kharagpur,
721302,[email protected]
2
Department of Mechanical Engineering, IIT Kharagpur,
721302,[email protected]
3
Department of Metallurgical and Material science, IIT Kharagpur,
721302,[email protected]

Abstract
Friction stir welding (FSW) is a solid state joining process and is handy for welding aluminum alloys.
Numerical simulation of FSW is highly complex due to non-linear contact interactions between tool and work
piece and interdependency of displacement and temperature. In the present paper a three dimensional finite
element model is proposed to study the thermal history, strain distribution and thermo-mechanical process in
butt welding of Aluminum alloy 2024 using DEFORM-3D software. Effect of tool rotational speed on plastic
strain is studied and insight is given on asymmetric nature of friction stir welding process. Lagrangian
incremental technique is used to model FSW process and sticking condition is defined between tool and work
piece.
Keywords: Friction stir welding, Finite element method, Temperature distribution

1 INTRODUCTION
Friction stir welding (FSW) is a solid state does not have any weld nugget zone. They have also
joining process invented by TWI England, Thomas et studied the plunging and dwelling periods in FSW.
al. (1991). In FSW, the temperature of material Effect of heat transfer from backing plate on
remains below the solidus temperature and hence all temperature distribution is also considered. Schmidt
defects arrested with solidification of the material are et al. (2004) have developed a mathematical equation
eliminated. In FSW a non consumable rotating tool is for heat generation and validated it by measuring
plunged in a work piece and stirred along welding power through experiments. They have considered
direction as shown in Figure 1. Heat generation heat generation through friction (both sticking and
between rotating tool and work-piece is responsible sliding friction) only but neglected the heat generation
for plasticizing and softening of the material and this by mechanical deformation. Arora et al. (2009) have
softened material is subjected to extrusion by traverse developed a three dimensional visco plastic model of
speed and tool pin rotation which leads to formation FSW to estimate temperature history and material
of weld nugget zone Mishra and Ma (2005). flow and validated the same with experimental data.
Simulation of FSW process is complex due to highly They have assumed a material as a non newtonian
non linear contact interaction between tool and work fluid and flow stress as a function of temperature and
piece, unknown boundary conditions and material strain rate. Buffa et al. (2006) have done a three
flow behavior during welding. Still over the last few dimensional analysis of FSW in DEFROM 3D to
years finite element analysis of FSW is under radar of predict temperature history and strain profile during
several researchers. Chao et al. (2003), Chen and welding. They have used flow stress as function of
Kovacevic (2003) have modeled FSW, by considering temperature, strain and strain rate. Regression
symmetry of the process along weld line and analysis is used to calculate material constants. Zhang
developed a heat flux equation by measuring the and Zhang (2007) have developed a fully coupled
temperature in FSW through experiment. They have three dimensional analysis to estimate temperature
not considered material deformation during the and strain profiles. Material flow is also simulated for
simulation and only sliding friction has been different time intervals of the simulation. They have
considered. Yu et al. (2012) have performed a three used arbitrary lagrangian and eulerian analyses to
dimensional finite element analysis of thermal history simulate FSW in ABAQUS/Explicit. Simple coulomb
in FSW by using Johnson cook material and damage law of friction is considered for the analysis. Gok and
model in ABAQUS/Explicit. Because of damage Aydin (2013) have used solid mechanics approach to
model, mesh nearby pin is deleted and the simulation model FSW for welding magnesium alloy. They have
486-1
Finite Element Simulation of Temperature and Strain Distribution in Al2024 Aluminum Alloy by Friction Stir Welding

simulated the thermal history. Similarly a number of mesh at the contact region to have better result and
researchers have also developed mathematical this refined mesh window will follow the tool during
expression for heat generation in FSW due to friction tool traverse movement. Boundary conditions
either by assuming constant contact pressure or assigned to FSW model are: bottom surface of work
neglecting heat generation due to mechanical piece is fixed by assigning zero velocity, with this all
deformation of work piece or by simplifying the movements of work piece is restricted. Rotational (Z-
contact by using coulomb friction law. Also, FSW is direction) and transverse (X-direction) velocities are
simulated by considering symmetry along weld line assigned to the tool.
which is not true in practical situation. In the present
paper a fully coupled three dimensional temperature
displacement analysis is used with lagrangian
incremental technique and to predict the strain and
temperature distribution and its variation with
rotational speed. Asymmetric nature of the process is
also explained with the help of strain distribution.

Tool rotation

Retreating
side

Shoulder
Figure 2 Geometric model of friction stir welding
Advancing side used in this work

Welding
Convectional heat transfer coefficient of 20W/m2/oC
direction is defined between work piece-environment and Tool-
environment. Convectional heat transfer coefficient
Figure 1 Schematic representation of friction stir between bottom surface of work piece and
welding environment is defined as 200W/m2/oC, to consider
backing plate effect.

2 Model Description 2.2 Finite element formulation

2.1 Geometric Modeling Finite element formulation of rigid viscoplastic
material is based on the variational approach in which
A three dimensional analysis of FSW is admissible velocities, ui should satisfy the conditions
performed in DEFORM-3D. Lagrangian implicit code of compatibility and incompressibility, and also the
is used along with adaptive re-meshing technique. velocity boundary conditions, which gives the
Fully coupled temperature-displacement analysis is following functional (function of a function) a
considered i.e. at every time increment, temperature stationary value
and displacement are calculated at each node
simultaneously. AA2024 aluminum alloy is taken as (1)
work piece material and is modeled as rigid visco-
plastic. Dimension of work piece is 80×60×5 mm3.
Work piece is meshed with 10339 nodes which forms where, is a work function and is the surface
44730 tetrahedral elements. Tool steel H13 is used as traction. A penalized form of the incompressibility is
the tool material which is modeled as a rigid body i.e. added to remove the incompressibility constraint on
stress and strain are not calculated on tool but heat admissible velocity fields. The actual velocity field is
transfer from work piece is considered. Since, yield now can be determined from the stationary value of
stress of tool steel is much higher than aluminum the variation equation which is stated as
alloy considering rigid tool is a valid assumption.
Tool is meshed with 9181 nodes which form 40247 (2)
tetrahedral elements. Shoulder diameter of tool is 24
mm. Tapered cylindrical pin is used with larger
where, and is the volumetric
diameter of 7 mm and included angle of 40o. Pin
height is 4.6 mm. Tool and work piece meshing is strain rate. δui is arbitrary variation; and are
shown in Figure 2. Mesh window is used to refine the the variations in the strain rate derived from δui. The

486-2
5th International & 26th All India Manufacturing Technology, Design and Research Conference (AIMTDR 2014) December 12th–14th, 2014,
IIT Guwahati, Assam, India

penalty constant, K is a penalty constant with very (N/mm2)

large positive value. Kobayashi et al. (1989)). Thermal conductivity
181 149
(N/soC)
2.3 Material model Heat capacity
2.43 2.78
Flow stress is a function of strain, strain rate, (N/mm2oC)
temperature as given in equation 3. Therefore to Thermal expansion
22 11.7
define flow stress, stress-strain behavior, at different (µmm/mmoC)
strain rates, temperature from DEFORM material Poisson’s ratio 0.33 0.3
library and isotropic hardening rule are considered.

σ = σ  ε , ε ,T 
.
(3) 3 Results and Discussions
  3.1 Temperature profile
. In FSW certain amount of heat is required to
where, σ is flow stress, ε is strain, ε is strain rate, T plasticise the material, which allows mixing of
is temperature. material to achieve a good weld. Heat is generated in
2.4 Thermal and frictional model FSW due to plastic deformation of material and
frictional heat generated between tool and work piece.
Temperature generated during FSW is the range The latter depends on shear friction factor and contact
of 0.5-0.6 Tm (where, Tm is the melting point in area. Figure 3 shows the calculated temperature
degree centigrade scale), it also changes the distribution along the weld line and top surface of the
mechanical response of the material. Fourier law of work piece. Maximum temperature is obtained at the
heat conduction is used to calculate temperature centre of the weld because rotation of shoulder and
distribution probe contributes towards highest heat generation in
this region. Also temperature profile along the
∂T thickness has attained V shape; this is because
k ∇ 2T + q& = ρ c p (2)
∂t convection heat transfer between bottom of work
piece and backing plate is much higher (high
where, k is thermal conductivity of material, T is convective heat transfer coefficient due to contact
temperature, q& is heat generation, ρ is material pressure between them) than convection heat transfer
density and c p is specific heat capacity. Thermal between top of work piece and atmosphere. This
difference of convective heat transfer coefficient leads
properties of material are independent of temperature to higher cooling at the bottom of work piece and
to avoid non-linearity in thermal calculation. Heat formation of V shaped temperature profile.
generation due to plastic deformation of the material
is given by
q& p = λσε& (3)

where, λ is inelastic heat fraction and its value is 0.9

Buffa et al. (2006). q& p is the heat generated due to
plastic deformation. Contact interaction between tool
and work piece is very critical and to simulate this,
sticking condition (shear stress at the interaction
remains constant) is considered.
60 mm
τ = mτ max (4)
where, τ is frictional shear stress at the interface, m 80 mm
is shear friction factor and its value is 1, τ max is
equivalent shear stress of material which is 0.577
Figure 3 Temperature (oC) profile in longitudinal
times of the yield stress of material.
direction and top surface of work piece for 1000
Table 1 Physical properties of material (DEFORM rpm and 60mm/s traverse speed
manual)
Tool steel 3.2 Strain profile
Properties AA2024
H13 Plastic deformation of material plays a vital role
Young’s modulus 68900 210000 in material movement and hence in formation of good
486-3
Finite Element Simulation of Temperature and Strain Distribution in Al2024 Aluminum Alloy by Friction Stir Welding

weld. It also affects the microstructure of the weld Figure 6 Strain variations in FSW as a function of
zone along with thermal history. Therefore it is rotational speed
important to study the strain distribution to understand
the deformation of material. Figure 4 shows the
effective strain distribution in transverse direction of 4 Conclusion
the weld for 1000 rpm and 60 mm/s traverse speed. A three dimensional fully coupled temperature-
displacement analysis is done to understand the
thermo-mechanical process of FSW process for
AS RS AA2024 aluminum alloy. Model is capable of
calculating temperature, strain, stress and other
parameters which give insight on FSW process.
Maximum temperature is attained in the nugget zone
and temperature profile attains a V shape. Higher
strain is observed on the top surface of the work piece
Figure 4 Strain (mm/mm) profile along y-z section as compared to the bottom surface and also strain
for 1000 rpm and 60 mm/s distribution is not uniform which indicates
asymmetrical nature of FSW process. Maximum
strain reduces with reduction in tool rotational speed.
AS RS
References
Arora, A., Nandan, R., Reymolds, A.P. and Debroy,
T. (2009), Torque, power requirement and stir zone
geometry in friction stir welding through modeling
and experiments. Scripta Materialia, Vol. 60(1),
Figure 5 Strain (mm/mm) profile along y-z section pp.13–16.
for 600 rpm and 60 mm/s Buffa, G., Hua, J., Shivpuri, R., and Fratini, L.
(2006), A continuum based fem model for friction stir
Figure 4 clearly depicts that strain is higher on the top welding-model development, Materials Science and
surface of the material as compared to bottom. This is Engineering, Vol. 419(1), pp.389–396.
due to the fact that rotating shoulder contributes to the Chao, Y.J., Qi, X. and Tang, W., (2003), Heat
deformation of material more than the pin. Figure 5 Transfer in Friction Stir Welding-Experimental and
shows the contour plot for 600 rpm rotational speed. Numerical Studies, Journal of Manufacturing Science
Maximum strain value is reduced due to reduction in and Engineering, Vol. 125(1), pp.138-145.
rotational speed which leads to reduced deformation Chen, C.M. and Kovacevic, R., (2003), Finite element
of material. Figure 6 clearly indicates that strain in the modeling of friction stir welding-thermal and
advancing side (A.S.) of the welding is higher as thermomechanical analysis, International Journal of
compared with the retreating side (R.S.). This proves Machine Tools and Manufacture, Vol. 43(13),
the asymmetric nature of the friction stir welding pp.1319–1326.
process. Higher strain on advancing side as compared Gök, K. and Aydin, M., (2013), Investigations of
with retreating side is due to positive impact of friction stir welding process using finite element
rotational and traverse velocity of the tool. Similar method. The International Journal of Advanced
pattern is obtained for 600 rpm but the values are Manufacturing Technology, Vol. 68(1), pp.775–780.
reduced. Kobayashi, S., Soo ik oh, Altan, T., (1989), Metal
forming and Finite elment method, Oxford series of
advance manufacturing, Oxford university press.
Mishra, R.S. and Ma, Z.Y., (2005), Friction stir
welding and processing. Materials Science and
Engineering: R: Reports, Vol. 50(1), pp.1–78.
Schmidt, H., Hattel, J. and Wert, J., (2004), An
analytical model for the heat generation in friction stir
welding. Modelling and Simulation in Materials
RS AS Science and Engineering, Vol. 12(1), pp.143–157.
Wayne M. Thomas, Edward D. Nicholas, James C.
Needham, Michael G.Murch, Peter Temple Smith,
C.J.D., Patent_FrictionWelding_ThomasTWI.pdf. U.
Yu, M., Li, W.Y., Li, J.L. and Chao, Y.J. (2012),
Modelling of entire friction stir welding process by
486-4
5th International & 26th All India Manufacturing Technology, Design and Research Conference (AIMTDR 2014) December 12th–14th, 2014,
IIT Guwahati, Assam, India

explicit finite element method. Materials Science and

Technology, Vol. 28(7), pp.812–817.
Zhang, Z. and Zhang, H.W., (2007), A fully coupled
thermo-mechanical model of friction stir welding, The
International Journal of Advanced Manufacturing
Technology, 37(3), pp.279–293.

486-5