2198 AlLi plates joined by Friction Stir Welding: Mechanical

and microstructural behavior

P. Cavaliere
a,
*
, M. Cabibbo
c
, F. Panella
a
, A. Squillace
b
a
Department of Innovation Engineering, University of Salento, I-73100 Lecce, Italy
b
Department of Materials and Production Engineering, Engineering Faculty, University of Naples, Federico II, I-80125 Naples, Italy
c
Mechanical Department, Polytechnic University of Marche, I-63100, Italy
a r t i c l e i n f o
Article history:
Received 22 January 2009
Accepted 23 February 2009
Available online 1 March 2009
Keywords:
Friction Stir Welding
AlLi
Fatigue
Crack behavior
Thermoelastic stress analysis
a b s t r a c t
AlLi alloys are characterized by strong anisotropy. 2198 AlLi sheets were joined via Friction Stir Weld-
ing (FSW) in parallel and orthogonal direction with respect to the rolling one. The material microstruc-
ture and the different phases were individuated by means of TEM observations in different sections of
the produced joints; in addition, the mechanical properties were evaluated by means of tensile and
fatigue tests at room temperature; the fatigue tests were conducted in axial control mode with
R = r
min
/r
max
= 0.33 for different welding conditions. The crack initiation and propagation in the welded
zone was also studied by applying thermoelastic stress analysis (TSA) during cyclic fatigue tests, employ-
ing single edge notched specimens. Thermoelastic data were used to measure the principal stresses and
principal strains on the specimens surface around the crack tip, according to growth rate; all the results
were validated by employing nite element analysis (FEM) to model the crack evolution.
2009 Elsevier Ltd. All rights reserved.
1. Introduction
Friction Stir Welding technology, if compared to traditional
welding techniques, reduces the presence of distortions and resid-
ual stresses [14] and is being targeted by modern aerospace
industry for high performance structural applications. Based on
friction heating at the facing surfaces of two sheets to be joined,
in the FSW process a special tool with a properly designed rotating
probe along the contacting metal plates, produces a highly plasti-
cally deformed zone by the stirring action. The microstructure evo-
lution and the resulting mechanical properties depend strongly on
the variation of the processing parameters, leading to a wide range
of possible performances [7]. The thermo-mechanical affected zone
is produced by friction between the tool shoulder and the plate top
surface, as well as plastic deformation of the material in contact
with the tool [5,6]. The FSW process is a solid state process, there-
fore the problem related to the presence of brittle inter-dendritic
and eutectic phases due to solidication structures is eliminated
[7]. The low mechanical properties microstructure resulting from
melting and re-solidication are absent in FSW welds, leading to
improved mechanical properties such as ductility and strength al-
loys with low residual stresses [812]. The application of FSW
technology is in particular dependence on mechanical perfor-
mances affected by the processing parameters, since fatigue is
the principal cause of failure for welded joints; as James et al.
[13] showed the different fatigue behavior of two AlMg alloys
as a function of welding speed. In addition, Ericsson and Sandtrom
[14] and Dickerson and Przydatek [15] showed the variation of fa-
tigue life of AA6082 joints with the welding speed and AlMg and
AlMgSi alloys plates of different thickness, comparing the results
with conventional fusion welding techniques. Recent papers were
published regarding the microstructural and mechanical proper-
ties of friction stir welded AlLi alloys; in particular Hao et al. pre-
sented interesting results of tensile and bending mechanical
properties as a function of processing parameters in 1420 alloys
[16]. In the present work, 2198 AlLi sheets were joined via Fric-
tion Stir Welding (FSW) in parallel and orthogonal direction. The
mechanical properties were evaluated by means of tensile and fa-
tigue tests at room temperature for different welding conditions;
in addition the material microstructure and the different phases
were individuated by means of optical and TEM observations in
different sections of the produced joints. The crack propagation
in edge-cracked specimens beside the welded zone was also stud-
ied by applying thermoelastic stress analysis (TSA).
2. Experimental procedure
2.1. Materials and methods
The material under investigation was a 2198-T851 aluminium
lithium alloy produced by ALCAN (Toronto, Canada) under the
0261-3069/$ - see front matter 2009 Elsevier Ltd. All rights reserved.
doi:10.1016/j.matdes.2009.02.021
* Corresponding author. Tel.: +39 0832297357.
E-mail address: pasquale.cavaliere@unile.it (P. Cavaliere).
Materials and Design 30 (2009) 36223631
Contents lists available at ScienceDirect
Materials and Design
j our nal homepage: www. el sevi er . com/ l ocat e/ mat des
form of rolled sheets of 5 mm thickness with the following compo-
sition (wt%): Si 0.03, Fe 0.04, Cu 3.3, Mn 0.01, Mg 0.32, Cr 0.01, Ni
0.01, Zn 0.02, Ti 0.03, Zr 0.11, Pb 0.01, Li 1.0, Al bal. Rectangular
plates 200 mm length 80 mm width were welded in perpendicu-
lar and parallel direction with respect to the rolling one. The em-
ployed rotating velocities (in clockwise direction) of the
cylindrical threaded tool was 1000 rpm, while the advancing se-
lected speed was 80 mm/min with tilt angle set equal to 2. The at
tool geometry dimensions are reported in Table 1. Some sheets
were instrumented with thermocouples embedded in the parent
material at different position, in order to monitor the temperature
proles as a function of the distance from the weld line; K-type
thermocouples with a sheath diameter of 1 mm were employed;
different thermocouples congurations were used in order to
map the different thermal history of the material as shown in
Fig. 1. Tensile tests were performed in order to evaluate the
mechanical properties obtained in the different welding condi-
tions. The Residual Stresses were also calculated in longitudinal
direction respect to the loading one, by employing the sin
2
W
method [17]. The RS were measured in longitudinal direction,
being the one affecting the crack tip stress eld. The tensile tests
were carried out at room temperature using a MTS 810 testing ma-
chine with initial strain rate of 10
3
s
1
. Specimens were sectioned
in the perpendicular direction to the weld line by employing an
electrical discharge machine (EDM), the tensile specimens mea-
sured 12 mm width, 80 mm length for a gauge length of 25 mm.
Endurance fatigue tests were performed by a resonant electro-
mechanical testing machine under constant loading control up to
80 Hz, with sine wave loading (TESTRONIC
TM
25 25 kN, produced
by RUMUL, SUI) in high cycles regimes. The cyclic fatigue tests
were conducted under axial stress amplitude control mode with
loading ratio R = r
min
/r
max
= 0.33. The fatigue specimens gauge
dimensions measured 12.5-mm width, 50-mm length and tests
were performed up to failure. Specimens for the microstructural
analyses were prepared by standard metallographic techniques
and etched with Kellers reagent to reveal the grain structure.
The microstructure and the different phases were individuated
by means of TEM observations in the center of the joints and at dis-
tances of 2.5, 5. 7.5 and 10 mm from the weld center. For TEM
investigations, thin foils were prepared by means of double jet
electro-polishing, using a solution of 20% HNO3 in methanol
(18 V and 35 C). The fatigue crack propagation experiments
were performed by employing single edge (1 mm) notched speci-
mens obtained on the advancing side. The FSW, in fact, does not
produce a symmetric deformation respect to the center line of
the advancing tool [18]. Due to such inhomogeneities, when a
clock wise direction rotation is employed, the less resistant zone
results the one on the advancing side of the tool. From previous
studies performed by the authors, it was demonstrated for 6082
aluminium alloy the fractured zone for tensile and fatigue speci-
mens resulted the one on the advancing side of the tool. In such
experiments a clock wise direction rotation was employed in a
wide range of processing speeds [19]. The use of TSA has been opti-
mized at tracking the crack initiation and growth.
2.2. Principles of TSA
Not many papers exist on the evaluation of fatigue behavior of
materials using instead thermographic infrared techniques [20
24]. These techniques are contactless and able to investigate areas
rather than just single surface points; analysis of the stress elds
around a crack tip is performed starting from 2D images which
show the temperature distribution over the sample surface. The
more recently developed DeltaTherm system makes possible to
actually perform real-time investigation of the crack propagation,
based on a much faster data collection [2528]. The use of TSA is
therefore aimed at tracking the crack initiation and growth in rela-
tion to the different areas in which it could take place; areas which
are affected in various ways by the weld presence. In particular, the
stress intensity factor was numerically evaluated taking into ac-
count a term including the residual stresses effect (which was cal-
culated apart using a weight function) and subsequently compared
to that inferred from TSA [2931].
Many papers describing the principles of TSA methods and
instruments are reported in literature [2936].
2.3. K
I
numerical evaluation for crack analyses
The evaluation of effective J-integral is largely recognized as an
important method for the analysis of materials response in fracture
mechanics problems. It is related to the energy release associated
with the crack growth and gives the measure of the deformation
at the crack tip. In the case of linear materials it can be related to
the stress intensity factor. In the quasi-statically loaded stationary
cracks the J-integral can be dened as:
J lim
C!0
_
C
n H qdC; 1
where Cis a contour starting on the bottom surface of the crack and
ending at the top surface in anti clockwise direction. The limit
C ?0 indicates that C dimension decreases at the crack tip, q is a
vector in the direction of the crack growth and n is the vector per-
pendicular to the C contour. The factor H is described by:
H WI r
@u
@x
; 2
Table 1
Tool geometry.
Tool parts Dimensions (mm)
Pin length 4.64
Pin diameter 5.67
Shoulder diameter 9.5
Fig. 1. Thermocouples measurements points used for the temperatures proles
during the joining process.
P. Cavaliere et al. / Materials and Design 30 (2009) 36223631 3623
where I is the identity matrix, r is the stress tensor, u is the dis-
placement in the x direction. For an elastic behavior of the material
W is the strain elastic energy, while for an elasto-plastic behavior W
is the strain elastic energy plus the plastic dissipation.
Therefore, the J-integral can be expressed as:
J
_
C
Wdy S
i
@u
@x
ds
_ _
; 3
where S
i
= r
ij
n
j
is the tensile vector perpendicular to C.
It was demonstrated that the J-integral dened along a contour
surrounding the crack tip is the variation of the potential energy
for a virtual extension of the crack:
J
@V
@a
; 4
where V is the potential energy and a is the crack length.
Fig. 2. Specimen geometry and the nite element mesh for small scale yielding problem solving for the single edge notch.
Fig. 3. Temperature proles, in the T and L congurations, on both the advancing and the retreating sides of the tool.
3624 P. Cavaliere et al. / Materials and Design 30 (2009) 36223631
For Mode I the characteristic stress intensity factor can be calcu-
lated by [36]:
K
I

EG
1 am
2
_ _1
2
5
m is the Poisson ratio and G tangential modulus. The evaluation of
the characteristic fracture parameters for the studied joints was
performed by employing nite element simulations by ABAQUS
software.
The mesh used for the simulations was built up by using 4-
nodes elements, different numbers of elements were used to model
the small scale yielding problem for the different studied congu-
rations. The crack propagation zone simulation was performed by
employing 1558 elements. The specimen geometry and the nite
element mesh for small scale yielding problem solving is shown
in Fig. 2 (all the dimensions are in mm).
The materials behavior was considered to be governed by the
rate-independent theory of plasticity with isotropic hardening.
The loads were applied in an incremental way, and the solution
to the non linear boundary value problem at each increment was
obtained by a fully implicit update using the NewtonRapson
method.
3. Results and discussion
3.1. Thermo-mechanical joints behavior
The temperature variation as a function of the distance from the
welds centerline for both the joining congurations is shown in
Fig. 3. For both the cases, the maximum measured temperature
(at a distance of 10 mm from the centerline) was close to 250 C
and then decreasing by increasing the distance; for both the weld-
ing congurations the temperature behavior is similar. In all the
measurements, the temperatures in the retreating sides resulted
lower with respect to the advancing one, by increasing the distance
from the welds centerline. For the L welding conguration, the
absolute temperatures resulted higher (in the same zone) than
the T conguration.
Such different thermal history for the two welding congura-
tions led to different proles in the microhardness measurements.
In Fig. 4 the microhardness proles are plot for the L and T welds;
for comparison the mechanical properties are plot in the same
graphs with the provisional thermal proles obtained by the poly-
nomial t from the measured temperatures [36]. In the case of L
conguration it seems that the material is less dissipating the ther-
mal energy developed by the tool movement; in fact, for both the
congurations the highest temperature is identical (close to 320 C
in the center of the nugget zone) but the prole describes that the
temperature decreases much faster in the T conguration with re-
spect to the L one. Such behavior is conrmed by the microhard-
ness values; for example it can be noted at 5 mm from the
centerline on the advancing side the classical overaging behavior
of FSW material with strong softening due to the coupled effect
of temperature and forging-extrusion of the material; the hardness
Fig. 4. Temperature proles coupled with the microhardness values, in the T and L congurations, on both the advancing and the retreating sides of the tool.
Fig. 5. Optical micrographs of the cross sections of the studied joints in L (a) and T
conguration (b).
P. Cavaliere et al. / Materials and Design 30 (2009) 36223631 3625
value is very close (9095 HV) for both the welds. By increasing the
distance from the centerline in the case of T conguration, the tem-
perature starts to decrease faster than L one and in this case the
softening zone appears broader.
3.2. Friction Stir Welding microstructure analysis
FSW has become a very effective tool in solving the joining
problems of proled sheets with material continuity, without
Fig. 6. TEM pictures of the studied joints in T conguration at a distance of 0, 2.5, 5, 7.5, 10 mm from the weld center, respectively.
3626 P. Cavaliere et al. / Materials and Design 30 (2009) 36223631
using different joining methods; particularly in case of aerospace
industry, where high ductility and tensile strength are required.
In the present work, different FSW butt welds of AA2198 sheets
were successfully obtained by varying the welding orientation re-
spect to the rolling one; no supercial or macroscopic defects
(Fig. 5) were observed on the surfaces and the welded joints where
mechanically and microstructurally characterized. The TEM obser-
vations showed the grain mean grain size variation from the weld
center to the TMAZ and the HAZ (Fig. 6). The different precipitated
phases, analyzed through SAEDP, indicated the h
0
(Al
2
Cu), d
0
(Al
3
Li)
phases formation in the nugget zone and h
0
(Al
2
Cu), d
0
(Al
3
Li), T
1
(Al
2
CuLi) phase in the TMAZ and HAZ. The rst observation of such
analysis was that the T1 precipitates, under the form of ne plates,
are the main strengthening phases active in the base material; be-
low 300 C many of such precipitates dissolve and the main
strengthening identied phases are the h
0
(Al
2
Cu), d
0
(Al
3
Li) ones.
Also the shape of the precipitates is strongly different, in the HAZ
the particles appear round shaped and uniformly distributed in
the grain interior and at the grain boundaries; at 5 mm from the
weld center it can be observed an overageing producing a drop
in the measured microhardness and this mechanical behavior coin-
cides with the observation that the TEM images show precipitates
free large zones. In the TMAZ the same precipitates appear elon-
gated and distributed in the grain interior. In the nugget zone (less
than 3 mm from the weld center) the precipitates return to be
round shaped and much more ne with respect to the HAZ, the
T1 precipitates dissolve. The mean grain size varied from the
HAZ (56 lm) to the nugget zone (1, 1.5 lm).
3.3. Mechanical properties
The base metal tensile properties of AA2198-T851 aluminium
alloy sheets are described in Table 2 for both the T and L directions.
The mechanical strength of the base material results much more
higher in the L direction, accompanied with a reduction in ductility
and an increase in the necking value, demonstrating the strong
anisotropy of such kind of AlLi alloys. The fatigue properties of
the base materials in L and T direction can be observed in SN
curves as shown in Fig. 7. The fatigue specimens have been cut
from the rolled plates in a number of 25 for each condition. Both
the materials show similar fatigue behavior with higher number
of cycles to failure exhibited by the material in L direction at the
same stress level. Such behavior produces a higher fatigue limit
(estimated in 220 MPa for the L direction, 180 MPa for the T direc-
tion in terms of stress amplitude). The total strain fatigue behavior
for low cycle numbers was also studied for the rolled sheets in L
and T conguration. By the CofnManson law, relating the total
fatigue life to the plastic strain amplitude e
pl
/2, and the Basquin
equation, relating the stress amplitude to the numbers of cycles
to failure, the fatigue life under a xed total strain value is shown
in Fig. 8. The alloy exhibits two characteristic strain amplitude
ranges, having different slopes in the cyclic stressplastic strain
curve; in the L direction the material fractures at a higher number
of cycles with respect to the plastic strain amplitude. Representa-
tive R-curves for the base material in longitudinal and transversal
directions have been evaluated and are shown in Fig. 9. The curves
exhibit a rising behavior and substantial crack extension before the
occurrence of plastic instability. The AA2198 sheets showed again
Table 2
Mechanical properties of the base metal in both directions.
Alloy ry (MPa) UTS (MPa) E (GPa) e A (%)
AA2198-T 347 431 74.3 0.2 18.3
AA2198-L 436.3 490.7 76.7 0.16 13.7
Fig. 7. Fatigue endurance curves of the 2198 AlLi plates in both T and L directions.
Fig. 8. Plastic strain amplitude as a function of the number of cycles of the 2198 Al
Li plates in both T and L directions.
Fig. 9. Crack growth behavior of the 2198 AlLi plates in both T and L directions.
P. Cavaliere et al. / Materials and Design 30 (2009) 36223631 3627
a higher crack resistance in the longitudinal direction if compared
with the transversal one.
Many different specimens in the two friction stir welded cong-
urations were then tested under fatigue to study the behavior of
the welded material. The fatigue properties of the FS Welded joints
can be observed in SN curves for the toughest studied joints, as
shown in Fig. 10. The fatigue specimens have been cut from the
welded plates for each condition and polished, according to stan-
dard procedures and avoiding supercial defects. The bead mate-
rial in all the welding congurations shows similar fatigue
behavior in the low cycle regime, since extremely narrow data dif-
ferences are produced in terms of stresses, especially in the inter-
mediate region. In the high cycle regime, the Whler curves did not
show clear differences, Fig. 10. It can be observed decreasing stress
amplitude limits as function of life endurance up to the fatigue lim-
it (4 10
6
cycles to failure) and such value is similar for both the
congurations at a value of the maximum stress equal to 200 MPa.
In different conguration and with the same welding parame-
ters it was reached a good combination between the material plas-
ticization during tool rotation and the mixing affect produced by
the advancing movement leading to the disappearance of the ef-
fects due to the rolling process. The tool advancing action (inclined
at a certain angle) is extremely similar to an extrusion process
which requires optimal temperature conditions for the better qual-
ity of the material in terms of microstructure; consequently, since
the resultant fatigue behavior for butt welded joints is directly re-
lated to the microstructure, provided that porosities occurrence is
avoided and micro-cracks formation is absent, and considering that
stress concentrators are missing, the studied FS Welded joints offer
the best fatigue performances only when optimal microstructure
congurations are reached. With a revolutionary pitch in the range
of 0.070.1 [19], the process is in the optimal temperature and
strain rates conditions to produce good microstructure quality
without defects for butt joints and therefore sound welds are
achieved. The longitudinal residual stresses were measured in
the cross sections of the welds by X-ray diffraction using the sin
2
W
method [38]. Residual stress proles of the two studied joints con-
guration is shown in Fig. 11. It must be observed that the residual
stresses can be relaxed for the cut of the mechanical testing spec-
imens, for this reason the residual stresses showed in the present
work are related to the measurements performed on the cut mate-
rial. As a general behavior, the residual stresses have compressive
character by approaching the weld line, changing to a tensile char-
acter in the weld zone from the heat affected one. It can be ob-
served that the higher values of residual stresses are achieved in
the advancing side of the tool, the proles show a very similar
behavior in both the congurations with higher values experienced
in the case of T joints. The residual stresses values differences de-
pend on the asymmetry of the FSW process; it is demonstrated by
several nite element calculations that the higher deformation
across the weld line are achieved in the retreating side of the tool
when a clockwise direction is employed for the rotation [3942].
Such a higher deformation produces an increase in the tempera-
ture respect to the advancing side leading to a softening during
the process. The small difference observed in the T and L congu-
ration can be attributed to the fact that in the T conguration the
main deformation acts perpendicularly to the main resistant direc-
tion of the sheets leading to a hardening respect to the L
conguration.
3.4. Crack behavior
All the fatigue crack tests were monitored via thermoelastic
stress analysis. Thermoelastic stress analysis (TSA) provided the
fulleld stress maps from the surface of the specimens subjected
to cyclic loading. In order to achieve a higher resolution in the
stress spatial distribution, the investigating area was restricted
using suitable optics around the region where crack initiation
Fig. 11. Residual stress proles for the FSW joints in both the studied congurations.
Fig. 10. Fatigue endurance curves of the friction stir welded 2198 AlLi plates in
both T and L directions.
3628 P. Cavaliere et al. / Materials and Design 30 (2009) 36223631
was observed during preliminary tests. In all the test conditions,
the crack initiation leading to the samples failure was observed
in real-time from the thermal images. As already mentioned, when
the material is loaded at high frequency in the elastic region, the
thermoelastic data relative to the conduction are linearly propor-
tional to the sum of principal stresses. Crack formation and propa-
gation can be followed in real-time providing continuously the real
stress elds around the crack tip, including closure effects.
All the stress maps were obtained using the scale factor K calcu-
lated following Eq. (2) together with the output signal V from the
measurement system in each measurement point and considering
the material as isotropic and homogeneous. Eq. (2) requires also
adiabatic conditions to hold and it is therefore important to nd
a way to verify that these are fullled, examining carefully the
thermal images around the area of crack initiation. In the sur-
roundings of this zone, adiabatic conditions may fail to verify.
There are two basic phenomena that lead to a lack of adiabatic con-
ditions and hence a change in phase: heat generation due to plastic
work and the presence of high stress gradients. Both are conditions
occurring near the crack tip area. The effect of these phenomena on
the thermoelastic images is a blurring around the crack tip area
which makes it difcult to determine crack tip only from observa-
tion. The exact location of the crack tip is important when, for
example, the stress intensity factor is inferred from thermoelastic
measured data. The loss of adiabaticity can be identied from the
phase maps and from the examination of the different phase pro-
les. When adiabatic conditions are achieved the phase is constant
and equal to zero. Observing the phase prole three different zones
can be identied approaching the crack tip area (see also [27]): (I) a
region where the phase is constant and therefore adiabatic condi-
tions are achieved; (II) a region where the phase starts assuming
positive values indicating a loss of adiabaticity due to plasticity
and high stress gradients; (III) a region where the phase changes
to negative until the crack tip is reached, due probably to reverse
plasticity effects. From the crack tip onwards the phase sign
changes continuously, this fact being attributed probably to con-
tact between the crack faces and background reections coming
from inside the crack. In this way it is possible to calculate the
Fig. 12. Examples of different phase proles, phase variation along lines at 32,500
(a) and 33,500 (b) cycles.
Fig. 13. Stress elds maps for the FSW joints welded in T conguration at 32,500 (a) and 33,500 (b) cycles; Von Mises stress distribution around the crack tip for the specimen
welded in the T conguration (c) and the corresponding stress eld map for a crack length of 2 mm (d).
P. Cavaliere et al. / Materials and Design 30 (2009) 36223631 3629
position of the end of region III during the fatigue tests in all the
loading conditions relating it to the number of cycles. This data
is used to infer the crack length and rate (Fig. 12).
The change in the shape of the curves related to the stress elds
is therefore due to the lack of adiabatic conditions and then to the
different heat generation and conduction due to the plastic defor-
mation of the material in the crack zone subjected to various cyclic
loadings.
The principal stresses distribution directly measured and nor-
malized around the crack site for different number of loading cy-
cles is shown in Fig. 13a and b. An increase of the stress values
around the crack site was observed as increasing the number of cy-
cles as expected. The broadening of stress proles around the crack
by increasing the cyclic loading reveals that the stress concentra-
tion zone increase also with stress.
The results of ABAQUS simulations were used to calculate the
theoretical K
I
values at different crack lengths. In Fig. 13c and d
it is shown the Von Mises stress distribution around the crack tip
for the specimen welded in the T conguration and the corre-
sponding stress-eld map for a crack length of 2 mm. From the
previous calculations the stress intensity factor (SIF) values as a
function of the crack lengths were obtained. In such calculations
it was not considered the effect on residual stresses on the
stress-eld maps and consequently on the effective values of the
stress intensity factors. To take into account such residual stresses
effects it was calculated their contribution to the K
I
as follow:
K
I

_
a
0
S
0
xmx; adx; 6
where
mx; a
1 m
1
1
x
a
_ _
m
2
1
x
a
_ _
2
_ _

2pa x
_ ;
m
1
0:6147 17:1844
a
W
_ _
2
8:7822
a
W
_ _
6
;
m
2
0:2502 3:2889
a
W
_ _
2
70:0444
a
W
_ _
6
:
7
With S
0
is the residual stresses prole along the crack path, m(x, a) is
a weight function and m
1
and m
2
its coefcients.
The contribution due to the effect of crack closure was esti-
mated by the following equation:
K
aperture
Cr
aperture

pa
p
;
C 1:12 0:231
a
W
_ _
10:55
a
W
_ _
2
21:72
a
W
_ _
3
30:39
a
W
_ _
4
:
8
In such way it was possible to take into account all the phenomena
affecting the deviation from the theoretical SIF value. The corrected
K
I
behavior as a function of the crack length compared with those
belonging to the calculations performed with CCD camera observa-
tions and TSA is plotted in Fig. 14a. The crack initiation and growth
behavior, for both the studied congurations is shown in Fig. 14b.
The fatigue crack growth rate results are shown in the curves pro-
cessed at a load ratio of R = 0.33. Both the L and T welds show a sim-
ilar behavior in the crack initiation and growth leading to the
conclusion that the FSW process eliminates the effect of rolling
direction on the fatigue properties of the welds. As a general behav-
ior the experimental curves tend to underestimate the K
I
value at
the same crack length, after few mm of crack propagation the TSA
measurements are coincident with the theoretical model while
the CCD camera calculations continue to underestimate the SIF.
4. Conclusions
The effect of rolling direction respect to the welding one on the
mechanical behavior of FSW AlLi plates was studied and the re-
sults reported in the present paper. The thermal proles of the
material during welds was very similar for both the congurations
with a temperature of 250 C at 10 mm from the centerline to
320 C in the center of the welds. The TEM observations in different
zones of the weld allowed the identication of the different rein-
forcing phases coupled with the grain size variation, the tempera-
ture evolution and the microhardness prole. The mechanical
strength of the base material results much more higher in the L
direction respect to the T one accompanied with a reduction in
ductility and an increase in the necking value, demonstrating the
strong anisotropy of such kind of AlLi alloys with respect to the
rolling direction. In the high cycle fatigue regime, the Whler
curves did not show clear differences. It was observed decreasing
stress amplitude limits as function of life endurance up to the fati-
gue limit (4 10
6
cycles to failure) and such value was very sim-
ilar for both the congurations at a value of the cyclic maximum
stress of 200 MPa.
Both the L and T welds show a similar behavior in the crack ini-
tiation and growth leading to the conclusion that the FSW process
eliminates the effect of rolling direction on the fatigue properties of
the welds.
Fig. 14. K
I
as a function of crack length for the specimen welded in T conguration
calculated with FEM and measured with TSA and CCD observations (a); Crack
growth behavior for the 2198 AlLi plates friction stir welded in T and L
congurations (b).
3630 P. Cavaliere et al. / Materials and Design 30 (2009) 36223631
References
[1] Mishra RS, Ma ZY. Friction stir welding and processing. Mater Sci Eng
2005;R50:178.
[2] Bussu G, Irving PE. The role of residual stress and heat affected zone properties
on fatigue crack propagation in friction stir welded 2024-T351 aluminium
joints. Int J Fatigue 2003;25:7788.
[3] John R, Jata KV, Sadananda K. Residual stress effects on near threshold fatigue
crack growth in friction stir welded aerospace alloys. Int J Fatigue
2003;25:93948.
[4] Jata KV, Sankaran KK, Ruschau J. Friction-stir welding effects on microstructure
and fatigue of aluminum alloy 7050-T7451. Met Mater Trans
2000;31A:218192.
[5] Guerra M, Schmidt C, McClure JC, Murr LE, Nunes AC. Flow patterns during
friction stir welding. Mater Charact 2003;49:95101.
[6] Ulysse P. Three-dimensional modeling of the friction stir-welding process. Int J
Machine Tools Manuf 2002;42:154957.
[7] Cavaliere P, Squillace A, Campanile G, Panella F. Effect of welding parameters
on mechanical and microstructural properties of AA6056 joints produced by
friction stir welding. J Mater Process Technol 2006;180:26370.
[8] Rhodes CG, Mahoney MW, Bingel WH. Effects of friction stir welding on
microstructure of 7075 aluminium. Scripta Mater 1997;36:6975.
[9] Sato YS, Urata M, Kokawa H, Ikeda K. Hall-/Petch relationship in friction stir
welds of equal channel angular-pressed aluminium alloys. Mater Sci Eng
2003;A354:298305.
[10] Berbon PB, Bingel WH, Mishra RS, Bampton CC, Mahoney MW. Friction stir
processing: a tool to homogenize nanocomposites aluminum alloys. Scripta
Mater 2001;44:616.
[11] Cavaliere P, Nobile R, Panella F, Squillace A. Mechanical and microstructural
behaviour of 2024-7075 aluminium alloy sheets joined by friction stir
welding. Int J Mach Tools Manuf 2006;46:58894.
[12] Cavaliere P, Squillace A. Effect of welding parameters on mechanical and
microstructural properties of dissimilar AA6082-AA2024 joints produced by
friction stir welding. Mater Sci Forum 2006;11631168:51921.
[13] James MN, Bradley R, Lombard H, Hattingh DG. The relationship between
process mechanisms and crack paths in friction stir welded 5083-H321 and
5383-H321 aluminium alloys. Fatigue Fract Eng Mater Struct 2005;28:24556.
[14] Ericsson M, Sandstrom R. Inuence of welding speed on the fatigue of friction
stir welds, and comparison with MIG and TIG. Int J Fatigue 2003;25:137987.
[15] Dickerson TL, Przydatek J. Fatigue of friction stir welds in aluminium alloys
that contain root aws. Int J Fatigue 2003;25:1399409.
[16] Wei S, Hao C, Chen J. Study of friction stir welding of 01420 aluminumlithium
alloy. Mater Sci Eng 2007;A452453:1707.
[17] Chan RW. Concise encyclopedia of materials characterization. Elsevier Science
Publisher; 2005.
[18] Buffa G, Donati L, Fratini L, Tomesani L. Solid state bonding in extrusion and
FSW: process mechanics and analogies. J Mater Proc Tech 2006;177:3447.
[19] Cavaliere P, Squillace A, Panella F. Effect of welding parameters on mechanical
and microstructural properties of AA6082 joints produced by friction stir
welding. J Mater Process Technol 2008;200:36472.
[20] Luong MP. Thermography: a new nondestructive evaluation method in fatigue
damage. Mech Mater 1998;28:15563.
[21] Hermanson KS, Sandor BI. Corrosion fatigue modeling via differential infrared
thermography. Exp Tech 1998;22:1921.
[22] Yang B, Liaw PK, Wang H, Huang JY, Kuo RC, Huang JG. Thermography: a new
nondestructive evaluation method in fatigue damage. JOM 2003:e55.
[23] Zhang D, Sandor BI. ASTM STP, vol. 1122; 1991. p. 34153.
[24] Liaw PK. Nondestructive evaluation (NDE) and materials properties
IV. Warrendale, PA: TMS; 1999.
[25] Tomlison RA, Olden EJ. Thermoelasticity for the analysis of crack tip stress
elds a review. Strain 1999;35:4955.
[26] Daz FA, Patterson EA, Tomlison RA, Yates JR. Measuring stress intensity factors
during fatigue crack growth using thermoelasticity. Fatigue Fract Eng Mater
Struct 2004;27:57183.
[27] Harwood N, Cummings WM. Thermoelastic stress analysis. Adam Hilger;
1991.
[28] Shiratori M, Miyoshi T, Nakanishi T, Noda T, Hanada A. Detection of cracks and
measurement of stress intensity factors by infrared video systems. JSME Int J
Series 1990;I 33:4008.
[29] Belgen MH. Structural stress measurements with an infrared radiometer. ISA
Trans 1967;6:16477.
[30] Cavaliere P, Rossi GL, Di Sante R, Moretti M. Thermoelasticity for the
evaluation of fatigue behavior of 7005/Al2O3/10p metal matrix composite
sheets joined by FSW. Int J Fatigue 2008;30:198206.
[31] Cavaliere P, De Santis A, Panella F, Squillace A. Thermoelasticity and CCD
analysis of crack propagation in AA6082 friction stir welded joints. Int J
Fatigue. doi:10.1016/j.ijfatigue.2008.07.016.
[32] Huang YM, Abdel Mohsen HH, Rowlands RE. Determination of individual
stresses thermoelastically. Exp Mech 1990;30:8894.
[33] Lesniak JR. Thermoelastic data improvements. In: Proceedings of SEM spring
conference on experimental mechanics, Dearborn, Michigan; 1993. p. 7219.
[34] Offermann S, Beaudoin JL, Bissieux C, Frick H. Thermoelastic stress analysis
under nonadiabatic conditions. Exp Mech 1997;37:40913.
[35] John R, Jata KV, Sadananda K. Residual stress effects on near-threshold fatigue
crack growth in friction stir welds in aerospace alloys. Int J Fatigue
2003;25:93948.
[36] Kong HS, Ashby MF. MRS Bull 1991;41.
[38] Cavaliere P, Panella F. Effect of tool position on the fatigue properties of
dissimilar 2024-7075 sheets joined by Friction Stir Welding. J Mater Process
Technol 2008;206:24955.
[39] Liechty BC, Webb BW. Modeling the frictional boundary condition in friction
stir welding. Int J Mach Tools Manuf. doi:10.1016/j.ijmachtools.2008.04.005.
[40] Bastier A, Maitournam MH, Roger F, Dang Van K. Modelling of the residual
state of friction stir welded plates. J Mater Process Technol 2008;200:2537.
[41] Zhang Z, Zhang HW. Numerical studies on controlling of process parameters in
friction stir welding. J Mater Process Technol. doi:10.1016/j.jmatprotec.
2008.01.044.
[42] Rajesh SR, Bang HS, Chang WS, Kim HJ, Oh CI, Chu JS. Numerical determination
of residual stress in friction stir weld using 3D-analytical model of stir zone. J
Mater Process Technol 2007;187188:2246.
P. Cavaliere et al. / Materials and Design 30 (2009) 36223631 3631