You are on page 1of 8

See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/257345628

Coupled Eulerian Lagrangian finite element modeling of friction stir


welding processes

Article  in  Journal of Materials Processing Technology · August 2013


DOI: 10.1016/j.jmatprotec.2013.02.014

CITATIONS READS

32 620

4 authors:

Fadi Al-Badour Necar Merah


King Fahd University of Petroleum and Minerals King Fahd University of Petroleum and Minerals
25 PUBLICATIONS   208 CITATIONS    121 PUBLICATIONS   604 CITATIONS   

SEE PROFILE SEE PROFILE

A. N. Shuaib Abdelaziz Bazoune


Arizona State University King Fahd University of Petroleum and Minerals
53 PUBLICATIONS   313 CITATIONS    42 PUBLICATIONS   360 CITATIONS   

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Friction stir welding View project

Development and Evaluation of PVD Coated Friction Stir Welding (FSW) Tools View project

All content following this page was uploaded by Abdelaziz Bazoune on 22 October 2017.

The user has requested enhancement of the downloaded file.


Journal of Materials Processing Technology 213 (2013) 1433–1439

Contents lists available at SciVerse ScienceDirect

Journal of Materials Processing Technology


journal homepage: www.elsevier.com/locate/jmatprotec

Coupled Eulerian Lagrangian finite element modeling of friction stir welding


processes
Fadi Al-Badour ∗ , Nesar Merah, Abdelrahman Shuaib, Abdelaziz Bazoune
Mechanical Engineering Department, King Fahd University of Petroleum and Minerals (KFUPM), Dhahran 31261, Saudi Arabia

a r t i c l e i n f o a b s t r a c t

Article history: A 3-dimensional localized finite element model (FEM) is developed to predict likely conditions that
Received 30 November 2012 result in defect generation during friction stir welding (FSW). The workpiece is modeled using Eulerian
Received in revised form 8 February 2013 formulation, while the tool is modeled using Lagrangian. Coulomb’s frictional contact model is adopted to
Accepted 25 February 2013
define the tool workpiece interaction, while the welding speed is defined by material inflow and outflow
Available online 4 March 2013
velocities. The numerical results show that the coefficient of friction has a major effect on void formation;
the lower the friction coefficient is applied, the larger the void is formed. Furthermore, welding using
Keywords:
force control (FC) at lower welding speed results in smaller void size and wider plastic zone, leading to
Friction stir welding
Finite element modeling
higher quality weld.
Coupled Eulerian Lagrangian © 2013 Elsevier B.V. All rights reserved.
Void formation

1. Introduction With the advancement in computational power, researchers


also used ALE re-meshing methodology with explicit solver to sim-
FSW problem is a multi-physics problem that includes excessive ulate steady state FSW. Deng and Xu (2004) modeled FSW and
material deformation and heat flow. Different modeling tech- simulated the metal flow pattern around the FSW tool assum-
niques such as computational fluid dynamics (CFD) and arbitrary ing plane strain conditions. The model considered experimentally
Lagrangian Eulerian formulation (ALE) have been used to simu- measured temperatures which were applied as body loads. The
late FSW process. Ulysse (2002) was the first to model FSW using authors used Abaqus Dynamic Explicit, to compare two tool
commercial CFD software FIDAP. He studied the effects of weld- pin-workpiece contact interaction models: modified Coulomb’s
ing and tool rotational speeds on temperature, loads on the tool frictional model and constant rate slip model. The comparison of
pin, and flow of particles near the rotating pin. Ignoring heat gen- the two models, based on tangential velocity, showed no large
erated by friction contact and heat loss to the backing plate, he difference between the two models. Schmidt and Hattel (2005)
found that the forces on the pin increase with increasing welding developed a localized thermo-mechanical model to study the
speed, and decrease with increasing rotational speed. Tempera- steady state FSW of 2xxx Aluminum alloy. Their model was gen-
tures predictions were overestimated compared to experimentally erated using commercial FEM package Abaqus, and solved using
measured ones. Colegrove and Shercliff Hugh (2005) used FLU- coupled temperature–displacement dynamic explicit where the
ENT (commercial CFD package) to study FSW process, considering material behavior was assumed to obey Johnson–Cook rule (1983).
rigid visco-plastic material behavior as in Ulysse (2002). But unlike The results of their investigation shows that the cooling rate plays
Ulysse, they considered tool pin threads with full sticking condi- a significant role in defect formation with higher cooling rate lead-
tion, and included heat dissipation to backing plate. The authors ing to faulty deposition of material behind the tool pin. Zhang et al.
addressed a number of drawbacks in the model, such as the pre- (2007), employed a model similar to that of Deng and Xu (2004),
dicted size of the deformation zone which was found to be much but considering 3D geometry. They found that the nugget zone
larger than observed experimentally, overestimation of the weld was much affected by the axial force than thermo-mechanical and
temperature, and underestimation of tool traversing force. Kim heat affected zones. Zhang (2008) used both classical and modified
et al. (2010) showed that the implementation of a proper thermal Coulomb’s frictional model conditions. He found that both models
boundary condition at the interface between the workpiece and the predicted similar results at low tool rotational speeds. The classical
backing plate is important for prediction of accurate results. Coulomb’s model failed to work for higher tool rotational speeds
due to the increase in the dynamic effect of the welding tool. Later,
Zhang and Zhang (2009) used an approach similar to that of Schmidt
∗ Corresponding author. Tel.: +966 569163321; fax: +966 3 860 2949. and Hattel (2005) to study the effect of welding parameters on
E-mail addresses: fbadour@kfupm.edu.sa, fbodour@hotmail.com (F. Al-Badour). material flow and residual stresses in friction stir butt welded Al

0924-0136/$ – see front matter © 2013 Elsevier B.V. All rights reserved.
http://dx.doi.org/10.1016/j.jmatprotec.2013.02.014
1434 F. Al-Badour et al. / Journal of Materials Processing Technology 213 (2013) 1433–1439

6061-T6. The material flow around the FSW tool was investigated Table 1
Mechanical and thermal properties of AL6061-T6.
using tracer nodes. The results showed that the increase of the rota-
tional speed and the decrease of the welding speed can improve the T (◦ C) E (GPa)  Cp (J kg−1 ◦ C−1 )  (kg m−3 ) ˛ (␮m m−1 ◦ C−1 )
friction stir weld quality, but flash formation would be more obvi- 25 66.94 0.33 945 2690 23.5
ous when rotational speed is increased. Estimated residual stresses 100 63.21 0.334 978 2690 24.6
indicated that the longitudinal residual stresses are higher than the 149 61.32 0.335 1000 2670 25.7
transverse ones. 204 56.8 0.336 1030 2660 26.6
260 51.15 0.338 1052 2660 27.6
From the above, it can be concluded that among the drawbacks
316 47.17 0.36 1080 2630 28.5
in CFD simulations is its inability to include material hardening 371 43.51 0.4 1100 2630 29.6
as it only considers rigid-viscoplastic material behavior. Contact 427 28.77 0.41 1130 2600 30.7
conditions are usually assumed to be full sticking, leading to over- 482 20.2 0.42 1276 – –
estimation of weld peak temperature and tool reaction loads.
Furthermore, CFD models ignored material’s elasticity. On the other Table 2
hand, ALE technique can make use of sliding boundary conditions Johnson–Cook plasticity model constants for Al-6061-T6.
to define the tool pin workpiece interaction, assuming different val-
A (MPa) B (MPa) C n m Tref (◦ C) Tmelt (◦ C)
ues for the coefficient of friction or constant slip rate. ALE allows
also inclusion of material temperature and rate dependency as well 324 114 0.002 0.42 1.34 24 583
as material hardening. But, in Lagrangian implementation or ALE,
void formation cannot be simulated because Lagrangian elements
have to be completely filled with material and defect formation Elastic and thermal properties are considered as temperature
simulation using ALE can only show a lack of material deposition dependent while 90% of plastic work is assumed to be convert-
behind the FSW tool pin. ing into heat. The thermal and mechanical properties as well
In the current work, a coupled Eulerian Lagrangian (CEL) model as Johnson–Cook parameters of Al-6061-T6 used in this model
is developed using Abaqus (6.11-2, 2011) environment to simulate (Tables 1 and 2, respectively) can be found in (Soundararajan et al.,
the two phases of FSW process (plunging and welding). Eulerian 2005; Lesuer et al., 2001).
elements which can include multi-materials, in addition to void The AISI H13 FSW tool steel is modeled as rigid isothermal
are employed in the FE model. The model is used to predict vol- Lagrangian body and is constrained to a reference point (RP)
umetric defects and material flow during the FSW process and to in order to assign tool physical properties such as; mass, mass
estimate tool reaction loads. The investigation takes into account moment of inertia, as well as to apply process conditions (tool rota-
three parameters: (a) contact frictional coefficient , (b) welding tional speed and plunging conditions). The FSW tool geometry and
speed V, and (c) plunging control method (position vs. force). The dimensions are given in Fig. 1, considering a flat shoulder tool with
FEM is validated using experimentally measured forces and torque a cylindrical pin. Further details of the tool geometry and dimen-
as well as matching the estimated processed zone shape and void sions used in the experimental work can be found in Shuaib et al.
size with those obtained experimentally by Shuaib et al. (2012) and (2012).
Al-Badour (2012). The Eulerian domain has been defined as a cuboidal shape that
includes two regions: “full” and “void”. The lower region “full” is
assigned with the workpiece material (Al-6061-T6), using uniform
2. Problem idealization and implementation material assignment tool representing a localized part of the work-
piece. The upper region “void” is left with no material. The void
A localized three dimensional (CEL) FE model is developed and region on the upper side of the workpiece is created to visualize
solved using Abaqus explicit to simulate void formation for given flash formation during welding process.
process or welding conditions in addition to tool reaction forces, The dimensions of the Eulerian domain are considered based on
torques and the state of the processed material during and after inflow conditions, as stress free boundaries. Moreover, the domain
FSW. size has to compromise between accuracy and computational time.
Simulation of welding phase is performed by employing a con- It is therefore considered to be four times the FSW tool shoul-
trol volume approach, whereas the welding speed is defined as der diameter; similar to Schmidt and Hattel (2005) and Zhang and
inflow and outflow over Eulerian domain boundaries. In the FSW Zhang (2009). The domain thickness is kept equal to the plate thick-
process, heat is generated due to frictional contact and plastic ness (6 mm) used in the experimental tests (Shuaib et al., 2012), and
deformation. Only material softening due to inelastic heat genera- the void region thickness is taken as 1 mm.
tion is considered. Due to limitations of the used version of Abaqus
(2011), adiabatic heating effect is assumed and heat dissipation into
workpiece material or surrounding is not considered.
The material selected for this investigation is Al 6061-T6 where
the relationship between flow stress  0 , strain rate ε̇ and temper-
ature T, is governed by Johnson–Cook’s semi-empirical formula:
   m 
ε̄˙ pl T − Tref
0 = (A + Bε̄npl ) 1 + C ln 1− (1)
ε̇o Tmelt − Tref

where ε̄pl is the effective plastic strain, ε̄˙ pl the effective plastic strain
rate, ε̇0 the normalizing strain rate (typically 1.0 s−1 ) and A, B, C, n,
and m are material constants. The parameter n takes into account
the effect of strain hardening, the parameter m models the thermal
softening effect, and C represents strain rate sensitivity. Tref repre-
sents the temperature where parameters A, B and n are evaluated,
while Tmelt is material solidus temperature. Fig. 1. FSW tool geometry and dimensions in mm.
F. Al-Badour et al. / Journal of Materials Processing Technology 213 (2013) 1433–1439 1435

Fig. 3. Velocity boundary conditions.

Technology Inc. (MTI) is used to perform bead on plates and record


tool loads during the process. The welding conditions considered
for validation are:
Fig. 2. FSW geometric model and Eulerian domain (a) idealization (b) zoomed and
sectioned numerical meshed model with material assignment. (a) Plunging phase: tool rotational speed N = 1000 rpm, and plung-
ing feed fp = 20 mm/min.
The Eulerian domain is meshed with 64,896 multi-material (b) Welding phase: N = 1500 rpm, and welding speed
thermally coupled 8-node (EC3D8RT) Eulerian elements. The ele- V = 175 mm/min.
ment has four degrees of freedom at each node. In order to avoid
Eulerian material flowing through Lagrangian mesh (tool) and to During plunging stage, the estimated axial force profile (Fig. 4a)
enhance the computational efficiency, bias meshing technique has does not exactly match the experimental measured one, while the
been used to generate fine mesh at the tool-workpiece interaction estimated and measured torque profiles (Fig. 4b) are found to be
zone with coarse mesh at the sides. Fig. 2a and b illustrates the ide- matching. Furthermore, compared to maximum experimentally
alization of the proposed model and mesh and material assignment, recorded values, the model was found to be underestimating tool
respectively. torque (11.9 N m) by about 10%, while the maximum plunging force
(10 kN) is overestimated by about 6%.
Welding phase simulation is validated using average recorded
3. Interaction, loads and boundary conditions
torque, axial, transverse, and crossfeed loads, measured during
welding under position control. The estimated forces and torque
Eulerian body is coupled to Lagrangian through contact inter-
during the welding phase showed fluctuations that are a result of
action. Tool-workpiece contact is defined using general contact
stick-slip phenomena. Therefore, the mean values are compared
formulation with Coulomb’s frictional law. Both modified and clas-
to the ones measured experimentally and presented in Table 3.
sical forms of Coulomb’s law were tried. The classical model has
These results show that the model tends to overestimate all tool
shown better estimation of tool torque. Three different coefficients
reaction forces and underestimate the tool torque. It is worth men-
of friction ( = 0.3, 0.58 and 0.8) were used to show their effect
tioning that the tool used in experiments has a profiled pin and
on void formation. These values represent the minimum, the esti-
shoulder, while in the FEM the tool has a cylindrical pin and a flat
mated value based on torque measurement, and maximum. The
shoulder. Scrolled shoulders are known to reduce axial tool load,
values obtained respectively from references (Schmidt and Hattel,
and threaded tool pin reduces the transverse load as well (Fuller,
2005; Al-Badour, 2012; Javadi and Tajdari, 2006).
2007). On the other hand, for similar tool dimensions scroll and
Force control welding condition is simulated by applying a con-
thread features on the FSW tool produce a larger area of contact
centrated force on the tool reference point, while for position
between the tool and the processed material, leading to an increase
control displacement constraints are used. Because the Eulerian
in tool torque. Assumptions of rigid body FSW tool and adiabatic
mesh is rigid, velocity constraints at the boundaries are applied
heat effect are expected to lead to overestimations of reaction loads.
in order to avoid material escaping from sides and bottom of the
Further validation is performed by comparing the equiva-
Eulerian domain, while welding speed is applied as inflow veloc-
lent plastic strain zone at the conditions used above to the
ity. Fig. 3 illustrates schematically the velocity boundary conditions
microstructure of the processed zone obtained experimentally. The
used in the present model.

4. FEA validation Table 3


Comparison between numerically estimated and experimentally measured tool
loads during welding.
As mentioned earlier, the finite element model is validated
using experimentally measured tool forces and torque as well as Load Plunging (N) Traverse (N) Crossfeed (N) Torque (N m)

by matching the estimated processed zone shape and void size Exp. 2974.6 722.2 50 8.64
with those obtained experimentally by Shuaib et al. (2012). In the FEM 3386 898 869 6.25
Error % +13.8 +24.3 – −27.7
experimental work, RM-1 FS welder developed by Manufacturing
1436 F. Al-Badour et al. / Journal of Materials Processing Technology 213 (2013) 1433–1439

Fig. 4. Comparison between experimental and numerical, (a) plunging force, and (b) torque.

comparison reveals that the present model describes quite accu- 5. Results and discussion
rately the shape of the plasticized zone as well as the presence of
void in the weld as shown in Fig. 5. However, because of the rea- 5.1. Influence of coefficient of friction on void formation
sons explained, the model is found to overestimate the void size
with a defect diameter around 0.29 mm compared the measured Fig. 6 illustrates the effect of coefficient of friction on the devel-
one of about 0.11 mm. Many reasons may lead to this discrepancy. oped volumetric defect. Here, the welding conditions (N = 1500 rpm
Firstly, material flow, which is affected by tool profile. It is well and V = 175 mm/min) are held constant while the friction coeffi-
known that threaded tools produce better material flow and more cient is varied ( = 0.30, 0.58 and 0.80). The results show that a
axial flow which leads to smaller void like defects. Secondly, adia- defect is generated for all three conditions. Using  of 0.3 leads to
batic heating effect is assumed; ignoring the cooling rate which has an unsuccessful weld as no material is deposited behind the tool
great effect on developing voids as reported by Schmidt and Hattel pin and the FSW tool has produced a key like channel in the work-
(2005). Moreover, the assumption of constant coefficient of friction piece. On the other hand, results obtained using a coefficient of 0.58
in the contact model and the limitation of material surface visu- and 0.80 show defected welds with different void sizes and shapes
alization in CEL post processing tool as well as other phenomena (Fig. 7). While only an internal defect is formed for  = 0.80, and
such as grain size, grain recrystallization and microstructure trans- a surface defect in addition to an internal void are developed for
formation may have contributed to such discrepancy.  = 0.58. From Coulomb’s law, the larger the value of  the greater
Finally, the estimated slip rate ˙ is found to be about 0.33 at the sticking zone is formed. Stick-slip phenomenon is known to be
the tool pin and 0.13 at the tool shoulder. The above values were responsible for defect formation. Thus, in order to have a sound
obtained from the predicted average material velocity around the weld, tool design and welding conditions should guarantee having
FSW tool (0.05 m/s) the maximum expected velocity near the pin more sticking than slipping.
assuming sticking condition (0.16 m/s) and the maximum velocity It is also found that the average value of the axial force for
at tool shoulder (0.41 m/s). Schmidt et al. (2006) reported experi-  = 0.58 is higher than that for  = 0.8, but the latter resulted in
mental values that are in the range of 0.1 to 0.3. greater force oscillation. The estimated average axial force for 
0.58 and 0.80 are around 5200 N and 3400 N, respectively. Sim-
ilar observations were found in estimated torque. The average
values of the torque for  = 0.58 and  = 0.80 are 5.66 N m and
6.25 N m, respectively. Results for  = 0.3 are excluded from the
comparison because the condition resulted in an unsuccessful
weld.

5.2. Effect of welding parameters

The welding speed and the plunging control method are the two
welding parameters investigated using this model, and both con-
ditions are simulated with a frictional contact of  = 0.80.Effect of
tool welding speed: two welding speeds of 125 and 175 mm/min
have been selected to assess the speed effect on developed plas-
tic zone and void formation. As can be seen in Fig. 8, the CEL FE
results show that the welding speed has a limited effect on devel-
oped void size as the void size is found to decrease by about 20% for
low welding speed. The void shape is more affected by the welding
speed. The void formed at 175 mm/min has elliptical shape while
that formed at 125 mm/min is more circular. Moreover, welding at
175 mm/min, produces a greater void under the tool shoulder. The
Fig. 5. Matching estimated equivalent plastic strain and void with experimentally
estimated equivalent plastic strain values showed a slight increase
found. when the welding speed is reduced. The plasticized zones are found
F. Al-Badour et al. / Journal of Materials Processing Technology 213 (2013) 1433–1439 1437

Fig. 6. Effect of coefficient of friction on void size; (a)  = 0.3, (b)  = 0.58 and (c)  = 0.8, (top view).

Fig. 7. Eulerian volume fraction-void (a)  = 0.58, and (b)  = 0.8.

to be larger extending more under the tool shoulder toward the Force vs. position plunging control: in this comparison a welding
retreating side (Fig. 8). These findings are similar to the experi- speed of 125 mm/min is used to study the effect of force control and
mental ones reported by Gharacheh et al. (2006). The presence of position control on the void and plastic zone sizes. In both simula-
larger plastic zone and smaller void size at lower welding speed tions tool rotational speed of 1500 rpm is employed and the contact
could be attributed to the increase in heat input leading to easier model assumes a coefficient of friction of 0.80.
material flow. The effect of plunging control method on dependant process
In terms of dependent process parameters; axial force and parameters is presented in Table 4. It is found that using force con-
torque, the welding speed seems to have a negligible effect on both trol plunging method increases the maximum tool torque. This is
loads; reducing the welding speed has reduced both axial force and believed to be due to the increase of tool penetration depth dur-
torque by an average of 5%. ing force control leading to greater tool workpiece contact area.

Fig. 8. Equivalent plastic strain for (a) V = 125 mm/min, (b) V = 175 mm/min, weld zone zoom in.
1438 F. Al-Badour et al. / Journal of Materials Processing Technology 213 (2013) 1433–1439

Table 4
Maximum estimated loads and temperatures at investigated welding conditions.

Welding conditions Maximum axial load (N) Maximum torque (N m) Maximum temperature (◦ C)

Speed V (mm/min) Plunging control method

175 Displacement 3800 8 583


125 Displacement 3600 7.5 583
125 Force 5500 12.5 583

Fig. 9. Equivalent plastic strain: (a) position control and (b) force control.

Fig. 10. Microstructural analyses from two locations of beads performed at a speed of 150 mm/min in (a and b) position control and (c and d) force control.

Because the model considers adiabatic heating effect, the maxi- for the decrease in the force, leading to forging and closing the void
mum estimated temperature remains equal to the material solidus defect. Soundararajan et al. (2006) studied the effect of tool over
temperature (583 ◦ C) set in the Johnson–Cook material model. plunging depth on surface defect formation and showed that an
Fig. 9 qualitatively compares the void size developed at force intermediate (0.4 mm) tool over plunging produced zero surface
and position control after 4.2 s of welding. The illustrated weld sec- defects while both small (0.2 mm) and large (0.6 mm) over plung-
tions, taken directly behind the FSW tool shoulder, indicate that ings produced welds with surface defects. This is an indication that
the generated void cross-sectional area during welding using posi- greater tool over plunging may not necessarily lead to defect free
tion control is larger (0.16 mm2 ) than the one developed using weld.
force control (0.02 mm2 ). These results also indicate that using
force control develops larger processed zone, as the equivalent 6. Conclusions
plastic strains are 1.5 times higher than those developed with posi-
tion control. These observations are supported by the experimental In this study a localized Coupled Eulerian Lagrangian model
results obtained by Al-Badour (2012) and Shuaib et al. (2012). The with adiabatic heat effect was implemented to simulate friction
microstructure analysis of beads produced at 150 mm/min (Fig. 10) stir welding of Al-6061-T6. The experimental validation of the CEL
clearly shows that force control produced no volumetric defects. It model showed that:
is believed that during force control welding the material resistance
to axial load drops when the void is formed close to the tool pin. • In tool plunging phase, the estimated maximum plunging force
As a result, the tool over plunging depth increases to compensate and torque were in close agreement with the experimental
F. Al-Badour et al. / Journal of Materials Processing Technology 213 (2013) 1433–1439 1439

results while the axial, transverse and cross feed forces were Deng, X., Xu, S., 2004. Two-dimensional finite element simulation of material flow in
overestimated by the CEL model. the friction stir welding process. Journal of Manufacturing Processes 6, 125–133.
Fuller, C.B., 2007. Friction stir tooling: tool materials and designs. In: Mishra, R.S.,
• The FE model seemed to underestimate tool torque in the welding
Mahoney, M.W. (Eds.), Friction Stir Welding and Processing. ASM International,
phase. Materials Park, OH, USA, pp. 7–35.
• The CEL model described quite accurately the shape of the plas- Gharacheh, M., Abbasi, Kokabi, A.H., Daneshi, G.H., Shalchi, B., Sarrafi, R., 2006. The
Influence of the ratio of rotational speed/traverse speed (ω/V) on mechanical
ticized zone as well as the presence of void in the weld. properties of AZ31 friction stir welds. International Journal of Machine Tools
and Manufacture 46, 1983–1987.
The analysis of FSW process parameters lead to the following Javadi, M., Tajdari, M., 2006. Experimental investigation of the friction coefficient
between aluminium and steel. Materials Science – Poland 24, 305–310.
conclusions:
Johnson, G.R., Cook, W.H., 1983. A constitutive model and data for metals subjected
to large strains, high strain rates and high temperatures. In: 7th International
1. The developed void size was directly affected by the frictional Symposium on Ballistics, Netherlands, pp. 541–547.
Kim, D., Badarinarayan, H., Kim, J.H., Kim, C., Okamoto, K., Wagoner, R.H., Chung, K.,
contact value and tool features; the higher the coefficient of
2010. Numerical simulation of friction stir butt welding process for AA 5083-H18
friction the smaller the produced void size is formed. sheets. European Journal of Mechanics – A/Solids 29, 204–215.
2. Because of tool over plunging, FSW using force control produced Lesuer, D.R., Kay, G.J., Leblanc, M.M., 2001. Modeling Large-strain, High-rate Defor-
smaller void defects. mation in Metals. Report no. ucrl-jc-134118. Lawrence Livermore National
Laboratory, Livermore, Canada.
3. The welding speed had more effect on void shape than on void Schmidt, H., Hattel, J., 2005. A local model for the thermomechanical conditions in
size. friction stir welding. Modelling and Simulation in Materials Science and Engi-
neering 13, 77–93.
Schmidt, H., Dickerson, T.L., Hattel, J., 2006. Material flow in butt friction stir welds
Acknowledgements in AA 2024-T3. Acta Materialia 54, 1199–1209.
Shuaib, A., Merah, N., Bazoune, A., 2012. Friction Stir welding of tube-tubesheet
The authors would like to acknowledge the support provided by joints. KACST, NSTP Project no. 08-adv66-04. King Fahd University of Petroleum
and Minerals, Dhahran, Saudi Arabia.
King Abdulaziz City for Science and Technology (KACST) through Ulysse, P., 2002. Three-dimensional modeling of the friction stir-welding process.
the Science and Technology Unit at King Fahd University of International Journal of Machine Tools and Manufacture 42, 1549–1557.
Petroleum and Minerals (KFUPM) for funding this work through Soundararajan, V., Zekovic, S., Kovacevic, R., 2005. Thermo-mechanical model with
adaptive boundary conditions for friction stir welding of Al 6061. International
project No. NSTIP (080ADV66-04) as part of the National Science, Journal of Machine Tools and Manufacture 45, 1577–1587.
Technology and Innovation Plan. Soundararajan, V., Atharifar, H., Kovacevic, R., 2006. Monitoring and processing the
acoustic emission signals from the friction-stir-welding process. In: Proceedings
of the Institution of Mechanical Engineers, Part B: Journal of Engineering Man-
References ufacture, 220, pp. 1673–2168.
Zhang, H.W., Zhang, Z., Chen, J.T., 2007. 3D modeling of material flow in friction
Abaqus ‘software’ 6, 2011, pp. 11–12. stir welding under different process parameters. Journal of Materials Processing
Al-Badour, F., 2012 Numerical and experimental investigations of friction stir weld- Technology 83, 62–70.
ing of tube-tubesheet joints. Ph.D. thesis, King Fahd University of Petroleum and Zhang, Z., 2008. Comparison of two contact models in the simulation of friction stir
Minerals, Dhahran, KSA. welding process. Journal of Materials Science 43, 5867–5877.
Colegrove, P., Shercliff Hugh, R., 2005. 3-Dimensional CFD modeling of flow round a Zhang, Z., Zhang, H.W., 2009. Numerical studies on controlling of process param-
threaded friction stir welding tool profile. Journal of Materials Processing Tech- eters in friction stir welding. Journal of Materials Processing Technology 209,
nology 169, 320–327. 241–270.

View publication stats

You might also like