Materials Science and Engineering A 403 (2005) 340–348

The finite element simulation of the friction stir welding process

H.W. Zhang ∗ , Z. Zhang, J.T. Chen
Department of Engineering Mechanics, State Key Laboratory of Structural Analysis for Industrial Equipment,
Dalian University of Technology, Dalian 116024, PR China

Received in revised form 11 May 2005; accepted 19 May 2005

Abstract

Solid mechanics-based finite element models and computational procedures are developed to study the flow patterns and the residual stresses
in frictional stir welding (FSW). Two-dimensional results of the material flow patterns and the residual stresses are presented. The flow of
metal during FSW is investigated using tracer particles. It is shown that the flows on the advancing side and retreating side are different.
After several rotations the material which is rotating around the nib sloughs off in its wake of the pin, primarily on the advancing side. The
residual stresses of the welded plate are investigated in this analysis. The distribution of the longitudinal residual stress along the direction
perpendicular to the welding line is a double feature curve. With the increase of the translational velocity, the maximum longitudinal residual
stress can be increased.
© 2005 Elsevier B.V. All rights reserved.

Keywords: Friction stir welding; Finite element simulation; Residual stress; Material flow; Equivalent plastic strain

1. Introduction welds [3,4]. Colligan [5] followed material flow in 6061 and
7075 aluminum by imbedding small steel balls as tracers
Friction stir welding (FSW) [1] is a new solid-state join- into grooves cut into the work piece parallel to the weld line.
ing technology invented at the welding institute (TWI) in Murr et al. [6] and Li et al. [7] showed that dynamic recrys-
1991. It has become a unique technique well suited for join- tallization plays a key role in material movement and that
ing many hard-to-weld metals, especially the high strength for many similar examples of extreme deformation at very
aluminum alloys and some dissimilar metals. Compared with high strain rates, solid-state flow accommodates the process,
the conventional welding processes, FSW possesses many which include shaped charges, hypervelocity impact crater
advantages over traditional fusion joining techniques, such formation, and thick metal or alloy target penetration by bal-
as no melting, low defect and low distortion, and FSW can listic energy rods.
even join thin and thick sections. This new technique is being Another fundamental aspect of such thermal and thermo-
successfully applied to the aerospace, automobile, and ship- mechanical processes is still not completely understood. It
building industries and opens up a broad avenue of joining is well known that the residual stress of the weld affects the
technology development and research. fatigue properties [8]. Webster et al. [9] reported the measure-
Like any new technology many unanswered questions ment of residual stress in FSW by X-ray technique, which
remain, but even such a fundamental question is how the shows that the longitudinal residual stress varies in the range
material flows from the front to back around the pin remains from 60 to 140 MPa, and also shows a correlation between
the subject of various investigations. For aluminum welds, the detailed residual stress feature and the heat flow in the
temperatures near the rotating pin never reach the melting weld. Sutton et al. [10] investigated the residual stress in
point [2], but quite complicated flow has been observed in 2024-T3 aluminum friction stir butt welds using the neutron
diffraction technique, and the results show that the highest
∗ Corresponding author. Tel.: +86 411 84706249; fax: +86 411 84708769.
stresses occur near the crown side of the weld over the entire
E-mail address: [email protected] (H.W. Zhang). FSW region.

0921-5093/$ – see front matter © 2005 Elsevier B.V. All rights reserved.
doi:10.1016/j.msea.2005.05.052
 H.W. Zhang et al. / Materials Science and Engineering A 403 (2005) 340–348 341

Numerical simulations have been done to investigate

the mechanism of material flow in FSW process. Since
1999 there have been several interesting studies focused
on observing material flow patterns in the FSW process.
Smith et al. [11] reported an effort to determine material
properties for fluid dynamics simulations. Seidel and
Reynolds [12] described a two-dimensional simulation
study based on fluid dynamics, which predicted material
flow patterns that compare well with experimental results.
Because their difficulties, solid mechanics-based models
and simulations that include material flow effects appeared
in the literature later than thermal or thermal mechanical
models and fluid dynamics models, which usually does
not include consideration of elastic responses of the work
piece. A two-dimensional modeling and simulation effort,
based on the Arbitrary Lagrangian–Eulerian (ALE) finite
element formulation and with material flow effects, was
reported by Deng and Xu [13–14], which serves as the basis
of the current paper. The above effort has been extended to
three-dimensional material flow by Xu and Deng [15].
In this paper, a two-dimensional model based on the Arbi- Fig. 1. The geometry model and the boundary conditions of FSW.
trary Lagrangian–Eulerian finite element method is proposed
to study the evolution of the stresses in the FSW process and temperature on yielding is considered explicitly in this anal-
the residual stresses of the welded plates. The entire weld- ysis. The properties of the material at different temperatures
ing process is simulated using the commercial finite element are shown in Table 1 [16]. T represents temperature, E elastic
package ABAQUS. modulus, σ u the current yield stress and ν the Poissons ratio.
Different with the work mentioned above, the whole pro- The material flow and the strain and stress distributions
cess including the flow patterns of the material, the strain and are modeled in this paper to help understand the mechanism
stress distributions, the residual stresses can be modeled in of FSW. The residual stress of the plate will be investigated
this research work. This model is based on solid mechanics, in after the FSW process is finished.
which the elastic responses of the work pieces are considered.
2.1. Material and the associated flow
2. Model description
Isotropic material behavior with isotropic hardening is
considered. The constitutive relations are given by von Mises
In FSW the work piece material in front of the tool will
yield criterion and the associated flow rule.
be pushed aside by the pin and forced to flow around the pin
The flow rule can be written as:
by the action of the pin, so large plastic flow develops around
the pin. In order to model the contacting surface properly
∂gi
dεpl = dλi (1)
and handle mesh distortion during large plastic deformation, ∂σ
a thermo-mechanical finite element based on the Arbitrary
where gi (σ, θ, Hi,α ) is the is the temperature-dependent flow
Lagrangian–Eulerian formulation and adaptive remeshing
potential and dλi is a scalar measuring the amount of the plas-
are employed in this study.
tic flow rate, whose value is determined by the requirement
The radius of the pin is R = 3 mm, and the dimensions
of the two plates are 100 mm in length (along the welding
line), 30 mm in width. The geometric schematic is shown in Table 1
Temperature-dependent material properties for AL6061-T6
Fig. 1. Three cases are modeled to investigate the effect of
the variations of process parameters on the flow behaviors, T(◦ C) E (GPa) σ u (MPa) ν
equivalent plastic strain, residual stress, etc. 25.00 66.94 278.12 0.330
The two-dimensional geometry described above is divided 100.00 63.21 260.68 0.334
148.89 61.32 251.24 0.335
into four-node quadrilateral elements. Reduced integration 204.44 56.80 221.01 0.336
with hourglass control is used to avoid the mesh-locking 260.00 51.15 152.26 0.338
problems associated with large incompressible plastic defor- 315.56 47.17 73.87 0.360
mation. The mesh consists of 11,717 elements and 11,986 371.11 43.51 36.84 0.400
nodes. The material of the plate is AL6061-T6, as a rate 426.67 28.77 21.58 0.410
482.22 20.20 10.49 0.420
independent elastic–plastic material. However, the effect of
342 H.W. Zhang et al. / Materials Science and Engineering A 403 (2005) 340–348

to satisfy the consistency condition:

fi = 0 (2)
For some rate independent plasticity models the direction of
the flow is the same as the direction of the outward normal
to the yield surface:
∂gi ∂fi
= ci (3)
∂σ ∂σ
where ci is a scalar. Such models are called “associated flow”
plastic models.
To accelerate the computation, the tool rotation and the Fig. 2. Modified Coulomb Law.
translation speeds are both increased 1000 times in this anal-
ysis, so that the ratio ν/ωR stays the same. This acceleration The Newton–Raphson iteration method is applied until the
is necessary because the slow speed in practical FSW pro- flow equation and yield constraints are satisfied.
cess cannot be completed in a realistic period of time using
ABAQUS code. To minimize the effect of this change, rate
2.3. The definition of contact
independent elastic–plastic material is used in this study and
the effect of temperature on yielding is considered, as shown
In this paper the friction control model is used. The
in Table 1. The yielding stress can be written as:
interface may experience frictional contact described by
q = σ0 (ε̄pl , θ) (4) a modified Coulomb friction law, as shown in Fig. 2. The
coordinate system x1 o1 y1 is fixed on the pin. The Coulomb
2.2. Integration method friction law is modified so that there exists a maximum
critical frictional stress, above which the frictional stress
The simplest operator that provides unconditional stability stays constant and is no longer equal to the product of friction
for integration equations is the backward Euler method. The coefficient and the contact pressure. It is necessary to use
strain rate decomposition is integrated over a time increment this frictional contact interface in order to model the plastic
as: shear flow behavior of the plate material when the applied
shear stress is near the material’s shear failure stress
√ [17]. In
ε = εel + εpl (5) this paper, a reasonable upper bound τ max = σ s / 3 is used,
where σ s means the initial yield strength of the material.
where ε is defined by central difference operator:
 
∂x 2.4. Boundary conditions
ε = sym
∂(xt + 1/2x)
During the solution, the elasticity relationship and the inte- At the boundaries of the work plates to be joined, AB,
grated rate decomposition are satisfied exactly, so that: BC, CD, and DA, material particles move with a constant
speed u, relative to the pin, in the direction opposite to
cσ = −Del : cε (6) the translation movement of pin. Pin rotates with a certain
angular velocity ω, as shown in Fig. 1. To obtain a more
where cσ is the correction to the stress, c␧ the correction to general conclusion, three cases of different process param-
the plastic strain increments and Del = ∂2 U/∂εel ∂εel is the eters are modeled, u = 2 mm and ω = 390 rpm, u = 4 mm and
tangent elasticity matrix. ω = 390 rpm, u = 2 mm and ω = 290 rpm, respectively.
If hardening laws are considered, then:
 
∂hα ∂hα 2.5. Temperature field
cα = hα cλ + λ : cσ + cβ (7)
∂σ ∂Hβ
The limitation of the PC computing power makes a
where cα is the correction to Hα , and cλ is the correction to thermo-mechanical finite element analysis of FSW impracti-
λ. cal. To compensate the lack of a predicted temperature field,
The flow rule is not satisfied exactly until the solution has actual temperature values from practical FSW test [18] will
been found: be used to construct an approximate temperature field for
 2  the FSW process simulation, as shown in Fig. 3. In this test,
∂g ∂ g ∂2 g
cε − cλ − λ : cσ + cα the rotational speed of the pin is 400 rpm and the moving
∂σ ∂σ∂σ ∂σ∂Hα
speed of the pin is 2 mm/s. The maximum temperature
∂g created by FSW ranges from 80 to 90% of the melting
= λ − εpl (8)
∂σ temperature of the welding material [2]. Due to the limitation
 H.W. Zhang et al. / Materials Science and Engineering A 403 (2005) 340–348 343

3. Results and discussions

The movement of the material points on the advancing side

is shown in Fig. 4 when u = 2 mm/s and ω = 390 rpm. Fig. 4(a)
shows the initial location of material points. Fig. 4(b) and (c)
show the movement of material points on the retreating side
at t = 7.5 × 10−4 s and at t = 1.35 × 10−3 s, respectively. It can
be seen in Fig. 4(b) the material points start to contact the pin
and some of the material points start to rotate with the pin. The
monitored material points form a fluidized bed around the pin
in Fig. 4(c). After several revolutions the material points start
to slough off in the wake behind the pin in Fig. 4(d). It also
can be seen that the speed of the material points in the wake is
Fig. 3. Fitted temperature history values at various distances from the weld
faster than those out of the wake, i.e. when the material points
line.
leave the rotational zone near the pin, the speed is higher than
the translational one u.
of the melting point of the material, the temperature fields The initial locations of the material points on the retreating
near the pin under different cases do not have significant side are shown in Fig. 5(a) when u = 2 mm/s and ω = 390 rpm.
differences. So the same temperature field is used to model Fig. 5(b) shows the movement of the material points on the
the friction stir welding process under different cases in this retreating side at t = 7.5 × 10−4 s. The material points start
analysis. to contact the pin and some of the material points start to
be pushed due to the rotation of the pin. In Fig. 5(c) the
2.6. Analysis of residual stress monitored material points are pushed into the wake behind
the pin but do not enter the rotational zone around the pin. All
The stress will form in the weld during welding since the monitored points are leaving the pin without any rotation
the expansion of material occurs during the heating of the around it in Fig. 5(d).
welded plates, followed by the contraction during the cooling To obtain the more general conclusions, the material
of the welds. Furthermore, the rotational and the transverse behaviors under different process parameters are also dis-
movements of the tool will cause additional stress in the cussed below.
weld due to the mechanical constraint of the plates by the When the angular velocity of the pin is decreased from
fixture. The temperature of the plate will be reduced to 25 ◦ C 390 to 290 rpm, the results of material behavior are shown in
and then the fixture will be removed to obtain residual stress Figs. 6 and 7. The translational velocity of the pin is 2 mm/s
distributions. in this case. It can be seen that the variation of the angular

Fig. 4. The material flow at advancing side when u = 2 mm/s, ω = 390 rpm.
344 H.W. Zhang et al. / Materials Science and Engineering A 403 (2005) 340–348

Fig. 5. The material flow at retreating side when u = 2 mm/s, ω = 390 rpm.

velocity has little influence on the material behavior whether case, the angular velocity is 390 rpm. The same phenomenon
on the advancing side or on the retreating side. The possible of material behavior can be observed. It takes less time for
reason for it is that the shear stress in the pin–plate interface the material points on the advancing side to slough off in the
is limited by the critical shear stress τ max . The change of wake than those in the case u = 2mm/s, ω = 390 rpm.
the angular velocity does not change the value of the criti- From the above three figures, it can be observed that
cal shear stress significantly. In these two figures, the same the flow of the plate material on the advancing side and the
phenomenon of material flow can be observed. The material retreating side are different. The material on the retreating
on the retreating side does not rotate with the pin but the one side never enters into the rotational zone near the pin, but
on the advancing side forms a rotational zone around the pin. the material on the advancing side forms the fluidized bed
After several revolutions the material points start to slough near the pin and rotates around it. After several revolutions
off in the wake behind the pin. the material on the advancing side starts to slough off in
When the translational velocity of the pin is increased from the wake behind the pin. Compared with the experimen-
2 to 4 mm/s, the results are shown in Figs. 8 and 9. In this tal data [19], the flow of the plate material is shown to

Fig. 6. The material flow at advancing side when u = 2 mm/s, ω = 290 rpm.
 H.W. Zhang et al. / Materials Science and Engineering A 403 (2005) 340–348 345

Fig. 7. The material flow at retreating side when u = 2 mm/s, ω = 290 rpm.

demonstrate the validity of the model developed in this become flatter, as shown in Fig. 11. It means that the material
paper. can hold the pin tightly to keep the FSW process continuing
Fig. 10 shows the distributions of radial and circumfer- successfully.
ential stress on the circle which is 0.25 mm away from the It is shown that time has little effect on the distributions
pin–plate interface in the FSW process at t = 0.00375 s. The of shear stress on the circle which is 0.25 mm away from the
distribution of the radial stress is similar to the distribution pin–plate interface in Fig. 12. The distribution laws are very
of the circumferential one. The minimum of radial stress similar at different time t. The shear stress is limited by the
is near θ = 320◦ and the maximum near θ = 200◦ . The critical shear stress and is certainly less than it. In fact, the
minimum of circumferential stress is near θ = 360◦ and material around the pin is very soft due to the increase of the
the maximum near θ = 140◦ . When t = 0.0075 s, the curves temperature. So the material cannot bear larger shear force,

Fig. 8. The material flow at advancing side when u = 4 mm/s, ω = 390 rpm.
346 H.W. Zhang et al. / Materials Science and Engineering A 403 (2005) 340–348

Fig. 9. The material flow at retreating side when u = 4 mm/s, ω = 390 rpm.

which is the real cause for the shear stress to be equal or less of the pin, 90◦ < θ < 180◦ , the equivalent plastic strain is
than the critical shear stress τ max . much smaller than that behind the pin. It means that the
Fig. 13 shows the distributions of equivalent plastic strain material behind the pin is highly deformed especially in the
on the circles which are 0.25 and 0.84 mm away from the regions of 0◦ < θ < 30◦ and 330◦ < θ < 360◦ . The region can be
pin–plate interface when u = 2 mm/s, ω = 390 rpm. In front

Fig. 10. The radial stress and the circumferential stress near the pin–plate Fig. 12. The shear stress near the pin–plate interface in FSW process at
interface in FSW process at t = 0.00375 s when u = 2 mm/s, ω = 390 rpm. different time when u = 2 mm/s, ω = 390 rpm.

Fig. 11. The radial stress and the circumferential stress near the pin–plate Fig. 13. The equivalent plastic strain near pin–plate interface when
interface in FSW process at t = 0.0075 s when u = 2 mm/s, ω = 390 rpm. u = 2 mm/s, ω = 390 rpm.
 H.W. Zhang et al. / Materials Science and Engineering A 403 (2005) 340–348 347

Fig. 16. The distributions of the longitudinal residual stress in the direction
Fig. 14. The equivalent plastic strain near pin–plate interface when perpendicular to the welding line when u = 2 mm/s, ω = 390 rpm.
u = 2 mm/s, ω = 290 rpm.
occurs near (R + 1∼2) mm (R is the radius of the pin) away
approximately corresponded with the wake behind the pin. from the welding line. The longitudinal residual stress ranges
When being away from the pin–plate interface, the equiva- from −100 MPa to about 100 Mpa. The range of the magni-
lent plastic strain is decreased except for the one in the region tude of the longitudinal residual stress can be fitted approx-
of 0◦ < θ < 30◦ and 330◦ < θ < 360◦ . The equivalent plastic imately with the one in Ref. [9]. The distribution of the
strain distributions are believed to correlate well with the longitudinal residual stress is a double feature curve. The
microstructure zones [15]. So it is studied in details in this experimental data on residual stress distributions in refer-
paper. When u = 2 mm/s, ω = 290 rpm, the equivalent plastic ences [20–22] can also show that the longitudinal residual
strain distributions are shown in Fig. 14. Compared with the stress distributions along the direction perpendicular to the
results in Fig. 13, the equivalent plastic strain in the region welding line are double feature curves, which shows the valid-
of 0◦ < θ < 30◦ and 330◦ < θ < 360◦ is decreased slightly. The ity of the model established in this paper. It seems that the
maximum of the equivalent plastic strain occurs in the loca- maximum longitudinal residual stress occurs in the location
tion where theta is approximately 15◦ and is decreased with where the equivalent plastic strain is decreased dramatically,
the increase of the angular velocity of the pin. When the trans- as shown in Fig. 17. The region of the large tensile residual
lational velocity is increased, the equivalent plastic strain stress can correlate approximately with the one of the large
is increased, as shown in Fig. 15, especially in the region equivalent plastic strain. Where the equivalent plastic strain
of 0◦ < θ < 30◦ and 330◦ < θ < 360◦ . It means that with the is decreasing, the longitudinal residual stress is decreasing.
increase of the translational velocity of the pin, the material When being away from the exaction point of the pin, the lon-
in the wake becomes more highly deformed. The equivalent gitudinal residual stress in the welding line is increasing. But
plastic strain distributions can be fitted well with the ones after about 2 mm, the distribution of the longitudinal resid-
obtained in Ref. [15], which shows the validity of the model ual stress keeps steady. It means that the distributions of the
used in this paper. longitudinal residual stress are not affected by the exaction
Fig. 16 shows the distribution of residual stresses in the of the pin after 2 mm away from the extraction point.
direction perpendicular to the welding line at y1 = 1mm, Fig. 18 shows the distribution of the longitudinal residual
y1 = 2 mm, and y1 = 3 mm (x1 o1 y1 is a coordinate system stress in the direction perpendicular to the welding line
fixed on the pin, as shown in Fig. 1) when u = 2 mm/s, when u = 2 mm/s, ω = 290 rpm. Compared the results in
ω = 390 rpm. The maximum of the longitudinal residual stress Fig. 16, the variation of the angular velocity has little effect
on the longitudinal residual stress distribution. When the

Fig. 15. The equivalent plastic strain near pin–plate interface when Fig. 17. The distributions of the equivalent plastic strain in the direction
u = 4 mm/s, ω = 390 rpm. perpendicular to the welding line when u = 2 mm/s, ω = 390 rpm.
348 H.W. Zhang et al. / Materials Science and Engineering A 403 (2005) 340–348

Acknowledgements

This work was supported by the National Natural Sci-

ence Foundation (10302007, 10225212, 10421002) and
the National Key Basic Research Special Foundation
(G1999032805). The authors would like to thank Prof.
Xiaomin Deng at University of South Carolina for his invalu-
able suggestions and discussions on this work.

References
Fig. 18. The distribution of the longitudinal residual stress in the direction [1] W.M. Thomas, E.D. Nicholas, J.C. Need ham, M.G. Murch, P. Tem-
perpendicular to the welding line when u = 2 mm/s, ω = 290 rpm. plesmith, C.J. Dawes, Friction Stir Welding, International Patent
Application No. PCT/GB92102203 and Great Britain Patent Appli-
cation No. 9125978.8, 1991.
[2] W. Tang, X. Guo, J.C. McClure, L.E. Murr, A. Nunes, J. Mater.
Process. Manuf. Sci. 7 (1998) 163–172.
[3] S. Xu, X. Deng, A.P. Reynolds, T.U. Seidel, Sci. Technol. Weld.
Joining 6 (2001) 191–193.
[4] L.E. Murr, Y. Li, R.D. Flores, Mater. Res. Innov. 2 (1998) 150–
153.
[5] K. Colligan, Weld. J. 78 (7) (1999) 229–237.
[6] L.E. Murr, E.A. Trillo, Y. Li, R.D. Flores, B.M. Nowack, J.C.
McClure, Min. Met. Mater. Soc. (1999) 31–37.
[7] Y. Li, L.E. Murr, J.C. McCluure, Mater. Sci. Eng. A 271 (1999)
213–271.
[8] D. Dalle, G. Biallas, T. Ghidini, Second International Sympo-
Fig. 19. The distribution of the longitudinal residual stress in the direction sium on Friction Stir Welding, Gothenburg, Sweden, 2000, pp. 26–
perpendicular to the welding line when u = 4 mm/s, ω = 390 rpm. 28.
[9] P.J. Webster, L.D. Oosterkamp, P.A. Brown, J. Strain Anal. 36 (1)
translational velocity is increased, the maximum longitu- (2000) 61–70.
[10] M.A. Sutton, A.P. Reynolds, D.Q. Wang, C.R. Hubbard, J. Eng.
dinal residual stress is increased from 100 to 110 Mpa, as
Mater. Technol. 124 (4) (2002) 215–221.
shown in Fig. 19. The longitudinal residual stress is always [11] C.B. Smith, G.B. Bendzsak, T.H. North, J.F. Hinrichs, J.S. Noruk,
positive near the welding line but negative out of the region R.J. Heideman, in: T. Siewert, C. Pollock (Eds.), Proceedings of the
−10 mm < x1 < 10 mm. 9th International Conference in Computer Technology in Welding,
Detroit, 2000.
[12] T.U. Seidel, A.P. Reynolds, The 12th Annual Advanced Aerospace
Materials and Processes Conference and Exposition, Long Beach,
4. Conclusions California, 2001, pp. 11–14.
[13] X.M. Deng, S.W. Xu, Transactions of NAMRI/SME (2001) 631–
A two-dimensional model is developed for FSW of an Al- 638.
alloy, in order to build qualitative framework to understand [14] X.M. Deng, S.W. Xu, J. Manuf. Process. 6 (2) (2004) 125–
133.
the mechanism of FSW process and the distributions of the
[15] S.W. Xu, X.M. Deng, CD-ROM Proceedings of the 21st Southeast-
residual stress of the welding plate. ern Conference on Theoretical and Applied Mechanics, 2002, pp.
699–704.
1) Material on the advancing front side of a weld forms a
[16] W.F. Brown Jr., H. Mindlin, C.Y. Ho, Aerospace Structural Metals
fluidized bed around the pin and slough off in the wake Handbook, CINDAS/procedure university, 1993.
after several revolutions but the material on the retreating [17] ABAQUS Theory Manual, Version 6.4, Hibbit, Karlsson, and
side never rotates with the pin. Sorensen Inc., 2003.
2) The material behind the pin is highly deformed especially [18] J.C. McClure, Z. Feng, W. Tang, J.E. Gould, L.E. Murr, X. Guo, in:
in the regions of 0◦ < θ < 30◦ and 330◦ < θ < 360◦ . J.M. Virtek, et al. (Eds.), Trends in Welding Research, Proceedings
of the 5th International Conference on Trends in Welding Research,
3) The maximum longitudinal residual stress occurs near Pine Mountain, GA, 1998, pp. 590–595.
(R + 1∼2) mm away from the welding line. [19] M. Guerra, C. Schmidt, J.C. McClure, L.E. Murr, A.C. Nunes, Mater.
4) The region of the large tensile residual stress can correlate Charact. 49 (2003) 95–101.
approximately with the one of the large equivalent plastic [20] M. Ya, F.L. Dai, J. Lv, Weld. J. 23 (5) (2002) 53–56.
[21] P. Staron, M. Kocak, S. Williams, A. Wescott, Phys. B 350 (2004)
strain.
491–493.
5) With the increase of the translational velocity of the pin, [22] A.P. Reynolds, W. Tang, T. Gnaupel-Herold, H. Prask, Scripta Mater.
the maximum longitudinal residual stress is increased. 48 (2003) 1289–1294.