Int J Adv Manuf Technol (2009) 41:498–503
DOI 10.1007/s00170-008-1498-1
ORIGINAL ARTICLE
Mode I failure modeling of friction stir welding joints
Guido Borino & Livan Fratini & Francesco Parrinello
Received: 11 December 2007 / Accepted: 26 March 2008 / Published online: 29 April 2008
# Springer-Verlag London Limited 2008
Abstract This paper analyzes mechanical response by
finite element method up to the decohesion failure in
fracture mode I for joints of friction stir welding (FSW) of
an aluminum alloy. It first describes experimental inves-
tigations on specimens with FSW embedded, subjected to
uniform traction and local punch tests used to characterize
local elastic and plastic material parameters. The heteroge-
neity of the mechanical properties induced by the FSW
process is taken into account for the elastic-plastic finite
element simulation. The growing damage and the opening
failure of the welding zone are described by the adoption
of a cohesive interface model with specific mechanical
properties.
Keywords Friction stir welding . Finite elements .
Cohesive interface . Failure analysis
G. Borino : F. Parrinello
Dipartimento di Ingegneria Strutturale e Geotecnica,
Università di Palermo,
Viale delle Scienze,
90128 Palermo, Italy
G. Borino
e-mail: borino@unipa.it
F. Parrinello
e-mail: parrinel@diseg.unipa.it
L. Fratini (*)
Dipartimento di Tecnologia Meccanica,
Produzione ed Ingegneria Gestionale, Università di Palermo,
Viale delle Scienze,
90128 Palermo, Italy
e-mail: abaqus@dtpm.unipa.it
1 Introduction
Friction stir welding is a quite recent welding process
employed to join hard-to-weld metals and light-weight
alloys like aluminum [1–3]. Friction stir welding is a solid
state welding process in which a specially designed rotating
pin is first inserted into the adjoining edges of the sheets to
be welded with a proper nuting angle and then moved all
along the joint. Such pin produces frictional and plastic
deformation heating in the welding zone; actually no
melting of material is observed during the process. Recent
studies [1] have proved that the FSW technique is a
relevant alternative to traditional fusion-based welding and
gives, as a result, welded plates free from solidification
microcracks, porosity, oxidation and other shortcomings
produced by standard fusion welding methods.
During the process, the tool rotation speed (R) and feed
rate (Vf), determining the specific thermal contribution
conferred to the joint, are combined in a way that an
asymmetric metal flow is obtained. In particular, an
advancing side and a retreating side are observed: the
former being characterized by the “positive” combination
of the tool feed rate and of the peripheral tool velocity
while the latter having velocity vectors of feed and rotation
opposite to each other. A detailed observation of the
material microstructure in the joint section by Nelson et al.
[4] indicated that there exists an area located at the core of
the welding, called “nugget”, where the original grain and
subgrain boundaries appear to be replaced with fine,
equiaxed recrystallized grains characterized by a nominal
dimension of a few μm. Furthermore, it is observed that
the FSW process produces a few different zones along the
welding line. Moving from the welding line towards the
parent material of the blanks a thermo-mechanically
affected zone (TMAZ) is highlighted, where the stirring
Int J Adv Manuf Technol (2009) 41:498–503
action caused by the tool pin determines a distortion of the
material grains. Beside this central zone, there are two
layers which are heat affected zones (HAZ). In both TMAZ
and HAZ zones, one can observe a different microstructure
(grain size, defect density, etc.); thus, different mechanical
properties are expected and need to be carefully investigat-
ed. The effectiveness of the obtained joint is strongly
dependent on several operating parameters [5]. First of all,
the geometric characteristics of the tool and in particular of
its pin determine the stirring action on the flowing material;
what is more, either the force superimposed on the rotating
tool during the process or the tool sinking in the sheet has
to be properly chosen, since the pressure generated on the
tool shoulder surface and under the pin end determines the
heat generated during the process. Finally both the rotating
speed and the feed rate determine the heat flux during the
welding process.
Whereas the FSW process has been extensively studied
[6–8] to the authors knowledge few studies have been done
in order to mechanically characterize the FSW with the
focus to make nonlinear structural analysis [9]. Since it
seems clear that the FSW zone is the weakest part of
welded plates, it is of paramount relevance to be able to
characterize the mechanical response of this zone embed-
ded in the continuous metal plate. It should be observed
that the proper characterization of the material in the
different zones of the joint would also allow one to develop
effective numerical simulation of subsequent forming
processes developed on the jointed blanks, i.e., bending or
stamping of the so called tailored blanks [10].
In turn, in the recent scientific literature two approaches
have been proposed aimed to the numerical simulation of
the FSW processes. First of all thermal models, taking into
account the heat generated by both friction forces work and
the material deformation work, have been proposed for
instance by Song et al. [11] and Schmidt et al. [12] trying to
highlight the temperature distributions nearby the rotating
pin. Also, finite element thermo-mechanical models have
been presented also by Chen et al. [13], Deng et al. [14] and
Zhang et al. [15] presented the results of both 2D and 3D
thermo-mechanical analyses based on the finite element
method aimed at the investigation of the material flow
during FSW processes and of the distributions of the most
relevant field variables such as temperature, strain and so
on.
The aim of the present paper is to investigate by means
of experimental tests the material mechanical properties in
the zones of the FSW, with the intent to obtain a useful
finite element representation of the system Aluminum plate
with welding embedded. For this purpose, beside the
diffuse plastic strains produced by stress states exceeding
the yield point, simulated by standard elastic-plastic finite
elements, the localized damage and the subsequent forma-
499
tion of a macrocrack is reproduced by the presence of a
cohesive elastic-damage interface model disposed along the
weakest line of the welding. The analysis carried out
regards as material the Al 6082 T6 aluminum alloy and the
obtained results, even difference in the through thickness
material behavior are neglected, permitting one to follow
the plastic behavior of the joint during the tensile test.
2 Experimental investigation
The considered joints were developed by FSW on AA6082-
T6 3 mm thick specimens, using properly designed
clamping fixture to fix the specimens to be welded on a
milling machine. As far as the utilized tool is regarded, it
was made in H13 steel quenched at 1020°C, characterized
by a 52 HRc hardness; a cylindrical pin was used with the
following geometrical characteristics: pin diameter was
equal to 4.00 mm and pin height was equal to 2.90 mm;
finally the shoulder diameter was equal to 10 mm. As the
process parameters are regarded, a tool rotation speed of
1040 rpm., a tool feed rate of 100 mm/min, a tilt angle of 2°
and a tool sinking of 2.9 mm were used on the basis of a
previous experimental campaign developed by the authors
[16].
In order to characterize the mechanical behavior of the
FSW joints two series of tests have been conducted. The
fist series is a quite standard uniaxial tension test, whereas
the second series are micro punch realized in the unaltered
metal and at points in the welding zones (TMAZ and
HAZ).
2.1 Tensile tension tests
This is a quite simple and standard test, carried out with a
universal tensile machine in displacement control regime
and with a very low constant displacement rate. It is useful
for the observation the overall behavior of the specimen.
The welding line is perpendicular to the loading direction.
The results in terms of engineering stress-strain curves are
reported in Fig. 1. It should be observed that the strain is an
average value, because the strain is not uniform along the
base length of measure which is 30 mm. In the same figure
the stress-strain response of the uniform material is reported
for comparison. It can be observed that the presence of the
FSW produce an overall decrement of the global strength of
about 30%. Moreover, a significant reduction of the
ductility is also clearly observable. A series of five tests
were developed both for the parent material and for the
welded joints in order to assess the validity of the obtained
results: in particular for the welded joints a maximum
scattering of about 4% was observed in the true stress vs.
strain curves.
500
300
stress [Mpa]
200
Tensile tests
100
0
0 4 8 12
Fig. 1 Stress-strain relation for uniaxial tension test
The results of the uniaxial test do not give direct data for
the material characterization being the sample not homo-
geneous; however some general information on the overall
behavior and on the modality of failure can be gained.
2.2 Local constitutive characterization of the FSW zones
The method adopted is called FIMEC, and it is a non-
destructive technique for measurement of mechanical
properties of materials [17], like yield strength, and load-
displacement curve, only a small zone of material is
involved in the test in each trial, that permit one to execute
many investigation in a relatively small testing surface;
using a flat punch, we can also obtain ultimate tensile
strength (UTS) and strain hardening exponent [18].
This kind of investigation was made in different part of
the joint, in central zone, nugget, in heat-affected zone and
in unaltered zone. Namely, seven points have been
experimentally tested, one in the center and three for each
side. The distance between each point has been chosen
4 mm. In this way an investigation in the different material
zones determined by the FSW in the transverse section of
the welded joints was developed.
Fig. 2 MTS Systems corporation—Alliance RT-50
Int J Adv Manuf Technol (2009) 41:498–503
True stress is correlated to the pressure at the contact
surface between flat punch and material through a value
called constraint factor, it is the base of many techniques for
measurement of tensile properties from indentations tests
[17–19]. Constraint factor is
Al 6082-T6
welded joint C ¼ 4F=πD2 σ ð1Þ
welded joint
where F is load applied by tensile machines, D diameter of
punch, σ stress under punch. A 0.2% offset constraint
factor of three, and a constraint factor for ultimate tensile
16 strength of 3.7, were utilized.
Indentation load at tensile strength is found using
equation
cF=cd ¼ F=+D ð2Þ
where + is a constant, d displacement, D the punch
diameter; results is relatively insensitive to + like observed
by Scibetta [18].
The test is a load controlled indentation, in which
indentation data are continuously measured during experi-
ment; tensile machine MTS Systems Corporation—Alliance
RT-50, with a load cell of 10 kN was used in this test (Fig. 2).
Diameter of flat punch used was 1 mm. Some results of
tests and tensile machine are shown in following image
(Fig. 3).
In Fig. 3 are reported the results (average values of five
developed measures on different specimens) of six inden-
tation tests, made in symmetric point of welded joint. The
curves are almost two by two are superimposed. Higher
curves correspond at point E and F in Fig. 5, middle curves
at point A and B, lower curves at point C and D. These
point are distributed on joint section with distance greater
than distance of depletion of boundary effects. It should be
observed that, again, a good overlapping was observed
between the five repetitions of the measurements; in
particular a maximum scattering of 5% was measured.
The punch trace after unload and zone of depletion of
boundary effects is very little respect distance between two
test points. From previous curves are found ultimate tensile
strength load corresponding, FTS (like explained before),
Fig. 3 Results load-displacements of indentation tests
Int J Adv Manuf Technol (2009) 41:498–503 501

Table 1 Numerical results related to the indentation test

Distance [mm] −12 −8 −4 0 4 8 12

σy [Mpa] 225 159 160 164 161 156 228

σTS [MPA] 249 177 201 197 202 179 249
N 0.0959 0.0827 0.1331 0.1172 0.1324 0.089 0.091

Yield stress, ultimate stress and exponential hardening are evaluated at different distance from the welding center.

and the corresponding yield stress load FY, and then, the material [7], but in the TMAZ-HAZ zone where
relative tension using opportune constraint factor. From actually fractures occur during tensile tests.
ultimate tensile strength σTS and yield stress σy, hardening
exponent is obtained using a simple analytic equation

 n . 3 Finite element simulation
σTS σy ¼ nE=σy expðnÞ ð3Þ
The finite element method has been adopted as an effective
where n is the hardening exponent, E Young modulus, σTS
numerical tool capable to reproduce the detailed mechanical
and σy were explained before. As far as the Young modulus
behavior of structures with FSW lines [20]. The first
is regarded the value measured for the parent material was
attempt, which is useful to calibrate the model and the
taken into account all over the calculations, namely
material constants, is the reproduction of the simple tension
69.000 MPa.
test. The structure has been discretized by 430 elastic-
Average values of experiment’s results, for each point
plastic finite elements (Fig. 4).
taken on surface as before shown in Fig. 3, in different zone
The constitutive relations adopted here are standard J2
of welded joint are stored in Table 1 (point 0 corresponds to
associative plasticity law (von Mises) with linear isotropic
the centre of welding line, numbers at head of table are
strain hardening. The material constants (elastic modulus,
indicative of distance of each point from centre of welding
yield stress and hardening modulus) are spatially variable
line).
and they have been deducted by the local punch test
It should be observed that the obtained results highlight
described in the previous Section. In particular 7 zones
the softening action determined by the FSW process on the
have been considered. The elastic modulus E=69 GPa and
processed material because of the generated heat flux; the
the linear isotropic hardening modulus H=530 MPa has
strongest effect is not observed in the nugget at the core of
been selected equal at every point, whereas the yield stress
the welding, due to dynamic recrystallization phenomena
which relatively improve the mechanical performances of

240
[MPa]

200
y

160

b) 120
-16 -12 -8 -4 0 4 8 12 16
Interfase [mm]

FSW axis
a)

Fig. 5 a) The indentation points. b) Yield stress vs. the distance from
the welding center (the dots are the experimental values and the solid
Fig. 4 Finite element mesh adopted for the analysis line are the value assigned for the finite element analysis)
502 Int J Adv Manuf Technol (2009) 41:498–503

Fig. 6 Average stress — aver- 250

age strain curve obtained by the
finite element simulation com-
200

Average stress [Mpa]

pared with the experimental
results
150

100 Experimental and numerical results

experimental
experimental
numerical
50

0 2 4 6 8 10
Average strain [%]

σy varies in a piecewise constant form as shown in Fig. 5b). Y ¼ 1=2½uT K½u; ð4bÞ
Figure 5 shows the measured value of the yield stress and
the value adopted in the analysis.
 n
As shown by the experimental results of Fig. 1, there is a c
load threshold after which strain softening appears. This # ¼ k ln ð4cÞ
1
ξ
softening regime is produced by localized damage which
starts to be produced as microcracks in the weakest FSW where K is a diagonal matrix collecting the interface elastic
zones. Microcracks increase porosity of the material and stiffness coefficients. The damage activation function has
reduce the stiffness of the sound material as well as the load been selected as
carry capacity. Usually this kind of behavior is modeled by 7ðt; # Þ ¼ Y
#  0  ð5Þ
Gurson plasticity model, which anyway fails to predict a
mesh objective solution as strain tends to localize along a The interface damage laws are then completed by the
line of fracture. An accepted solution is to modify the damage flow rules, which are of the following associa-
standard local plasticity models by including strain gradient tive type, together with the damage loading/unloading
effects which ensure that the problem is well posed [21]. condition.
Here, because of the specific welded problem, we adopt
 @7 
an interface element elastic-damage approach. Interface 5¼ 1 ð6aÞ
@Y
elements are disposed along the weakest line. They have
zero thickness and describe in an average sense the elastic-
damage properties which characterize the localized damage  @7 
ξ¼
1 ð6bÞ
process zone. The introduction of the interface elements @#
allow one to follow the material response in the post peak
softening regime, up to the complete failure, i.e., the full

separation of the specimen in two pieces. l0 ð6cÞ

3.1 Interface model

The interface model adopted is a quite simple elastic-
damage interface model. The following are the state
variables of the model: displacement discontinuity jump
vector [u], isotropic damage ω, and a scalar internal
variable ξ, which represents the damage hardening evolu-
tion. The following are the work conjugate variables: the
interface traction vector t, the energy release rate Y, and a
static-like internal variable χ. The state equation which
relate statics to kinematics state variables are
t ¼ ð1
5ÞK½u;

71 ¼ 0 ð6dÞ
Equations (4a–6) have been integrated stepwise in a
fully implicit method and implemented in a FE interface
element. The date selected for the simulation are: a1 =10e-
9 μm 2/ μN, a 2 = 10e-6, K N = K T = 50e5 μN/μm 3 , Ka =
10e5 μN/μm3, c=3, n=15,99, k=12e-11
3.2 Results
Figure 6 shows the average stress—average strain curves of
the tension test compared with the analogous result
ð4aÞ obtained in the experimental test of Fig. 1. The average
Int J Adv Manuf Technol (2009) 41:498–503
strain is obtained dividing the relative displacement by the
initial measure length (L=30 mm).
4 Discussion and conclusions
The friction stir welding technique was applied to joint
Al6082-T6, and experimental tests were made. Two kinds
of measurements were made: conventional tensile tests and
local indentation tests to investigate mechanical properties
of dissimilar zones of FSW welded joint.
The mechanical properties collected by tests were used
to build a finite element model constituted by different
plastic properties material zones, which inserted a zero
thickness interface. The obtained numerical results were
compared with tensile test ones, to verify correctness of the
overall proposed model and showed a satisfying agreement.
It should be observed that the developed procedure
neglects any variation of the material mechanical properties
throughout the blank thickness, since the measurement
points for the FIMEC tests were taken just on the equatorial
plane of the transverse section of the welded joints. The
simulation, for instance, of bending mechanics-based
operations developed on the basis of the presented material
characterization would then lead to poor results.
Anyway, the proposed approach seems promising in
order to develop local material characterizations aimed to
get necessary information in order to allow effective
numerical analyses. In particular, with reference to FSW
processes, on the basis of the quite complex material flow
occurring during the welding, it is necessary to get a proper
map of the material mechanical characteristics after welding
covering all the different “material zones”. As further
development of the present research the authors are going
to take into account also the different material character-
istics all along the sheet thickness with the final aim to
develop a complete procedure able to give indications and
material data to be used in FEM simulations of sheet metal
forming operations involving welded blanks.
Acknowledgements This work has been made using MURST
(Italian Ministry for University and Scientific Research) funds. The
authors wish to thank Prof. V. Tagliaferri’s research group at
University of Rome “Tor Vergata” for their fruitful support in the
FIMEC measurements.
References
1. Liu HJ, Fujii H, Maeda M, Nogi K (2003) Tensile properties and
fracture locations of friction-stir-welded joints of 2017-T351
aluminium alloy. J Mater Process Technol 142:692–696 DOI
10.1016/S0924-0136(03)00806-9
503
2. Rhodes CG, Mahoney MW, Bingel WH, Spurling RA, Bampton
CC (1999) Effects of friction stir welding on microstructure of
7075 aluminum. Scripta Mater 36(1):69–75 DOI 10.1016/S1359-
6462(96)00344-2
3. Guerra M, Schmidt C, McClure LC, Murr LE, Nunes AC (2003)
Flow patterns during friction stir welding. Mater Charact 49:95–
101 DOI 10.1016/S1044-5803(02)00362-5
4. Su JQ, Nelson TW, Mishra R, Mahoney M (2003) Microstructural
investigation of friction stir welded 7050-T654 aluminium. Acta
Mater 51:713–729 DOI 10.1016/S1359-6454(02)00449-4
5. Lee WB, Yeon YM, Jung SB (2003) The improvement of
mechanical properties of friction-stir-welded A356 Al alloy. Mater
Sci Eng A355:154–159
6. Mishra RS, Ma ZY (2005) Friction stir welding and processing.
Mater Sci Eng R Rep 50:1–78
7. Peel M, Steuwer A, Preuss M, Withers PJ (2003) Microstructure,
mechanical properties and residual stresses as a function of
welding speed in aluminium AA5083 friction stir welds. Acta
Mater 51(16):4791–4801 DOI 10.1016/S1359-6454(03)00319-7
8. Prime MB, Gnaupel-Herold T, Baumann JA, Lederich RJ,
Bowden DM, Sebring RJ (2006) Residual stress measurements
in a thick, dissimilar aluminium alloy friction stir weld. Acta
Mater 54:4013–4021 DOI 10.1016/j.actamat.2006.04.034
9. Liu S, Chao YJ (2005) Determination of global mechanical
response of friction stir welded plates using local constitutive
properties. Model Simul Mater Sci Eng 13:1–15 DOI 10.1088/
0965-0393/13/1/001
10. Fratini L, Buffa G, Hua J, Shivpuri R (2006) Friction stir welding
of tailored blanks: investigation on process feasibility. Ann CIRP
55/1:279–282
11. Song M, Kovacevic R (2003) Thermal modeling of friction stir
welding in a moving coordinate system and its validation. Int J
Mach Tools Manuf 43:605–615 DOI 10.1016/S0890-6955(03)
00022-1
12. Schmidt H, Hattel J, Wert J (2004) An analytical model for the
heat generation in friction stir welding. Model Simul Mater Sci
Eng 12:143–157 DOI 10.1088/0965-0393/12/1/013
13. Chen CM, Kovacevic R (2003) Finite element modeling of
friction stir welding - thermal and thermomechanical analysis. Int
J Mach Tools Manuf 43:1319–1326 DOI 10.1016/S0890-6955
(03)00158-5
14. Deng X, Xu S (2001) Solid mechanics simulation of friction stir
welding process. Transaction of NAMRI/SME, SME 29:631–638
15. Zhang HW, Zhang Z, Chen JT (2005) The finite element
simulation of the friction stir welding process. Mater Sci Eng A
403:340–348 DOI 10.1016/j.msea.2005.05.052
16. Barcellona A, Buffa G, Fratini L (2004) Process parameters
analysis in friction stir welding of AA6082-T6 sheets, keynote
paper of the VII ESAFORM Conference, Trondhaim, pp. 371–374
17. Ricciardi B, Montanari R, Moreschi LF, Sili A, Storai S (2001)
Mechanical characterisation of fusion materials by indentation
test. Fusion Eng Des 58–59:755–759 DOI 10.1016/S0920-3796
(01)00561-0
18. Scibetta M, Lucon E, Chaouadi R, van Walle E (2003)
Instrumented hardness testing using a flat punch. Int J Press
Vessels Piping 80:345–349 DOI 10.1016/S0308-0161(03)00077-2
19. Taljat B, Zacharia T (1998) New analytical procedure to determine
stress-strain curve from spherical indentation data. Int J Solids
Struct 35:4411–4426 DOI 10.1016/S0020-7683(97)00249-7
20. Zienkiewicz OC, Taylor RL (eds), The finite element Method,
fourth ed., McGraw-Hill, UK, 1991
21. Borino G, Fuschi P, Polizzotto C (1996) Elastic-plastic-damage
constitutive models with coupling internal variables. Mech Res
Commun 23(1):19–28 DOI 10.1016/0093-6413(95)00072-0