Received: 3 February 2021 Revised: 12 April 2021 Accepted: 14 April 2021

DOI: 10.1111/ffe.13482

ORIGINAL ARTICLE

Impact test behavior of aluminum alloys welded joints:

Experimental and numerical analysis

José A. Frutos Martínez1 | Ricardo R. Ambriz1 | Moussa Naït-Abdelaziz2 |

David Jaramillo1

1
Instituto Politécnico Nacional CIITEC-
IPN, Cerrada de Cecati S/N Col. Sta.
Abstract
Catarina, Azcapotzalco, Ciudad de The Charpy impact test behavior of base metal, weld metal, and heat-affected
México, 02250, Mexico zone for 6061-T6 and 7075-T651 aluminum alloy welds was analyzed. Force
2
University of Lille, Unité Mécanique de
versus time curves of welded joints were obtained by means of an
Lille, Av. Paul Langevin, 59650,
Villeneuve d'Ascq, France instrumented Charpy pendulum. An explicit finite element model was
implemented to evaluate the performance of the Cowper–Symonds constitu-
Correspondence
tive model. Parameters were obtained through a strain-rate-sensitivity analysis
Ricardo R. Ambriz, Instituto Politécnico
Nacional CIITEC-IPN, Cerrada de Cecati of base metals at high strain rates. Experimental results were compared with
S/N Col. Sta. Catarina, Azcapotzalco, those computed by finite element method (FEM) by assuming the same physi-
C.P. 02250 Mexico City, Mexico.
Email: [email protected]
cal and geometrical conditions of the dynamic-experimental tests. ANSYS-LS
DYNA was used to achieve the FEM simulations. The comparison between
experimental and FEM results in terms of force versus time data was
evaluated. Higher accuracies were observed in Al 7075-T651 alloy (roughly
98%). The 6061-T6 alloy has shown an approximation of about 90%. Discrepan-
cies are explained in terms of the effective plastic strain at fracture.

KEYWORDS
aluminum welds, Cowper–Symonds model, explicit finite element method, gas metal arc
welding, instrumented impact Charpy tests

1 | INTRODUCTION (HAZ).7,8 Thus, it would be expected that mechanical

properties (quasi-static and dynamic) will change
The use of aluminum alloys has increased due to the throughout the welded joints, producing different
combination between high resistance and low density, mechanical properties for each zone of the weld (base
which allows to reduce the weight of structures.1 Some of metal [BM], weld metal [WM], and HAZ).9–11 Mechani-
the most commonly used aluminum alloys are the cal properties of aluminum alloys welded joints have
7075-T651 and 6061-T6. The 7075-T651 material provides been studied. Temmar et al.12 reported a reduction
a yield and ultimate tensile strength of approximately around 50% of yield strength and 30% in elongation for
460 and 540 MPa, respectively, whereas the 6061-T6 alloy 7075-T6 welded, when compared to base material.
has values between 240 and 290 MPa.2 The use of arc 6061-T6 aluminum alloy joints are affected similarly with
welding processes tends to decrease the strength of the reductions of 30% in yield strength and 45% in elonga-
welded joints.3–6 That reduction is produced by the tion.4 This reduction on mechanical properties tends to
thermodynamic instability of the coherent η´ (7075 alloy) be of more severe at the HAZ.13,14 Microestructural
and β´´ (6061 alloy) phases, which produces a soft zone changes have also an effect on the fracture behavior of
formation (over-aging effect) in the heat-affected zone welded joints under dynamic conditions. The formation

Fatigue Fract Eng Mater Struct. 2021;1–16. wileyonlinelibrary.com/journal/ffe © 2021 John Wiley & Sons, Ltd. 1
2 FRUTOS MARTÍNEZ ET AL.

of the softer zones in welded joints of aluminum alloys models are usually obtained by uniaxial tension or com-
hardened by precipitation produces ductile failures in pression tests at different strain rates or by using inverse
quasi-static testing as well as during the impact test. This methods.33,37 The efficiency of the models is often limited
aspect increases the energy absorption capability of the to a range of strain rates and the parameters must therefore
HAZ in comparison with the BM.12,15–18 However, due to be adapted to the studied strain rate range.38–40
the brittle microstructure produced during the solidifica- This article reports the numerical approach for
tion process in conjunction with the stress concentration V-Charpy impact tests on 6061-T6 and 7075-T651
factor promoted by the WM porosity, the impact aluminum alloy welds by considering the previously
resistance is greater than the BM but lower than the obtained results published by the authors.3,13 Also, some
HAZ.12,16,19–21 complementary experimental results were performed and
Charpy testing is an extended technique for analyzing reported. The Cowper–Symonds constitutive model was
materials under impact conditions, wherein the impact used in conjunction with an explicit finite element model
velocity (between 5 and 5.5 ms
1) is a function of the to determine the correlation between the computed and
height and mass of the striker, which could induce equiv- experimental results.
alent strain rates of 101 s
1.22 As a result, the mechanical
response produced by the impact becomes an
important issue when attempting to analyze the obtained 2 | METHODOLOG Y
results.23–25 The geometrical shape of the specimen com-
bined with the testing condition, increases the triaxiality 2.1 | Experiments
at the vicinity of the crack, inducing brittle fractures.26,27
The use of instrumented Charpy devices allows to Al–Si–Mg (6061-T6) and Al–Zn–Mg (7075-T651) alloys
obtain additional information of the mechanical response were used. Ambriz et al.15 previously reported the impact
of the materials under impact, allowing the understand- of Charpy behavior of the 7075-T651 joints. In the case of
ing of the impact phenomenon.9,15,16,28 For instance, the 6061-T6 alloy, force–time curves were experimentally
force versus time and force versus displacement curves determined and reported in this work. The welded joints
can be obtained, which allows to appreciate the material were obtained by using the welding parameters for the
behavior and its evolution during the contact of the 6061-T6 and 7075-T651 alloys reported previously.3,13
striker.29,30 In this context, some studies have been Standard Charpy V-notch specimens for the three
conducted on aluminum alloys9,17,30,31; however, not different zones of the welded joints (BM, WM, and HAZ)
enough studies in aluminum-welded joints have been were manufactured from the 6061-T6 aluminum-welded
performed.12 For instance, Ambriz et al. reported the plates according to ASTM E23 standard.26 For the HAZ
energy absorption for the different zones of a 7075-T651 and WM, the notches of the specimens were located
aluminum alloy welded by gas metal arc welding according to microhardness measurements reported in
(GMAW) process.15 These authors used experimental Figure 1A. The displacement of the sample s(t) was
force versus time curves to analyze the impact response determined by means of the following equation40:
behavior. They observed that HAZ have a better energy
ðt  ðt 
absorption capability than BM in the studied welds. 1
sðt Þ ¼ v0
F ðt Þdt dt ð1Þ
On the other hand, numerical models of the impact t0 m t0
process have proven to be a relevant method to predict
the response of structures wherein the capacity of
absorbing large amounts of energy over a very short time.32 Force versus time curves F(t) for each welding zone
For this purpose, researchers have gone from experimental were obtained with the following parameters: mass
tests in combination with numerical estimations to develop m = 18.8 kg and impact velocity vo = 5.24 m s
1, which
finite element models with some constitutive equations to provides an impact energy of 260 J.
reproduce the nonlinear visco-plastic behavior of materials
under dynamic loading.33 Those models, such as Johnson–
Cook or Cowper–Symonds, require to identify material 2.2 | Finite element calculation
parameters, which allows capturing their dynamic
mechanical response.34–36 Because Cowper–Symonds 2.2.1 | Numerical model
constitutive law is visco-plastic, it offers the capability to
capture the combined effects of strain rate and plastic An explicit finite element model was performed to corre-
deformation of the material response under impact late the experimental results with the numerical
testing.34–36 The parameters values used in constitutive approach. The model considers a standardized Charpy
FRUTOS MARTÍNEZ ET AL. 3

F I G U R E 1 (A) Microhardness
profiles for the 6061-T613 and
7075-T6513 welded joints,
(B) geometrical model, (C) finite
element model, and (D) mesh of the
standard Charpy specimen at the
notch zone [Colour figure can be
viewed at wileyonlinelibrary.com]

specimen (55  10  10 mm) as well as an anvil and a comparison between a homogeneous material and a
striker with defined geometries according to the impact multi-material approach was conducted. The homoge-
device. The initial impact conditions (kinetic energy and neous material considers a specimen with properties of
impact velocity) from the experimental tests were consid- the material of interest, whereas the multi-material con-
ered. A striker density value of 1.602  106 kg m
3 was siders the different zones of welds (BM, HAZ, and WM)
implemented to match the numerical impact energy with according to the microhardness measurement profiles
the experiments, that is, mass of the hammer of 18.8 kg performed previously by the authors (Figure 1A).3,13
and volume of the striker of 11,738.7  10
9 m3.
Figure 1B, illustrates the geometrical model for the
finite element calculation. Dynamic impact explicit 2.2.2 | Constitutive model
solution was performed by using full integration on
ANSYS-LS DYNA platform under its ANSYS parametric To consider the strain rate effects during the impact test,
design language (APDL) environment. A SOLID an isotropic/kinematic hardening model was used
164 element with nine degrees of freedom by node (Cowper–Symonds model), which considers the strain
(displacements, velocities, and accelerations in the nodal rate for scaling the yield flow stress σ Y by means of a
X, Y, and Z directions) and eight integration points was strain rate-dependent factor41:
used to mesh the specimen (Figure 1C,D).41 The anvil
"  1=P #
and striker were meshed by using SOLID 168 tetrahedral ε_ 
elements (Figure 1C). To capture the local displacements, σY ¼ 1 þ σ 0 þ βE p εeff
p ð2Þ
C
the notch tip mesh of the sample was refined by using
element size of 0.05 mm (Figure 1D).
A node to surface contact between the specimen and where σ 0 is the initial quasi-static yield strength, ε_ is the
the anvil was defined, whereas a surface-to-surface con- strain rate, C and P are the Cowper–Symonds strain rate
tact between the striker and the specimen was used. A parameters, εeff
p is the effective plastic strain, β is the
4 FRUTOS MARTÍNEZ ET AL.

strain hardening parameter (which adjusts the contribu- to match the finite element calculation with experimental
tion between isotropic and kinematic hardening), and Ep results.34 However, an analytical approach is proposed in
is the plastic hardening modulus, which is given by this work assuming that values of the yield stress of a mate-
rial under tension tests at different strain rates are known.
ET E A sensitivity analysis of the E, σ 0, ET, Ep, C, P, β, and
Ep ¼ ð3Þ
E
ET εeff
p,fracture parameters has been performed. These parame-
ters consider the yielding force, fracture initiation, maxi-
where E is the Young modulus and ET is the tangent mal force, force at the beginning of the unstable crack
modulus. growth, unstable crack growth, collapse energy, and the
E and ET were experimentally determined from crack failure criterion (fracture plastic strain). Thus, con-
quasi-static tensile tests (at ε_ = 10
3 s
1) for the different sidering the reported results for the 7075-T65142 and
zones of the 7075-T6513 and 6061-T613 aluminum alloy 6061-T643 aluminum alloys and the quasi-static yield
welds (Figure 2). Average tensile mechanical properties stress found by Ambriz et al.13 and Alatorre et al.,3 an
(Table 1) were used in the finite element model. estimation of the C and P Cowper–Symonds parameters
was performed by using Equation (2). When taking β = 0,
we obtain the following equation:
2.2.3 | Calibration parameters "  1=P #
σY ε_
In the literature, it is often found that Cowper–Symonds ¼ 1þ ð4Þ
σ0 C
empirical parameters (C and P) are stochastically modified

and consequently,
 
σY 1 1
log
1 ¼ log_ε
logC ð5Þ
σ0 P P

The expression 5 predicts a linear evolution of the

flow stress when plotted against the strain rate in log–log
coordinates. Thus, the C and P parameters can be there-
fore adjusted by using a least square method, as shown in
Figure 3.
Using the best fitting of the C and P parameters above
obtained, the hardening parameter β can be adjusted in a
true stress–true strain curve. It is worth nothing that β
parameter is a constant independent of the strain rate. It
means, that the Cowper–Symonds model is restrictive
because the strain hardening is linear (it is not possible to
capture the nonlinearity of the stress–strain mechanical
response in the plastic domain). Additionally, since β is con-
stant, the hardening is the same whatever the strain rate.
Figure 4 shows a comparison between the experimen-
tal results and that obtained with the model. The identifi-
cation of the εeffp,fracture parameter, which controls the
fracture initiation, is quite more difficult. Theoretically, it
must be determined by tension tests at the same strain
rate than that of the impact. However, the problem is the
triaxiality at the vicinity of the notch, which also affects
the effective plastic strain. To get a close value of this
parameter, it is necessary to perform tensile tests on
F I G U R E 2 True stress–true strain behavior for base metal, notched specimen exhibiting a triaxiality ratio close to
weld metal, and heat-affected zone, (A) 6061-T613 and that observed near the crack tip of the Charpy specimen
(B) 7075-T6513 [Colour figure can be viewed at wileyonlinelibrary. at the same strain rate.44 In our case, the value εeff p,fracture
com] was fitted taking into account the force–time curve of the
FRUTOS MARTÍNEZ ET AL. 5

TABLE 1 Cowper–Symonds empirical parameters for 6061-T6 and 7075-T651 aluminum alloys welded joints

eff
Parameters E (GPa) σ 0 (MPa) ET (MPa) EP (MPa) β C P εp,fracture
6061-T6
Base metal (BM) 70 280 875 886 0.45 3.71E22 20.28 0.28
Weld metal (WM) 70 120 100 1001 6 3.71E21 20.28 0.15
Heat-affected zone (HAZ) 70 100 950 963 1.5 3.71E20 20.28 0.29
7075-T651
Base metal (BM) 72 520 2000 2057 1.5 1.337E6 6.43 0.03
Weld metal (WM) 72 110 1570 1605 1.7 1.337E15 6.43 0.11
Heat affected zone (HAZ) 72 230 1150 1170 1.0 1.337E12 6.43 0.24

Note: E = Young modulus; σ 0 = yield strength; ET = tangent modulus; EP = plastic hardening modulus; β = strain hardening parameter; C and P = Cowper–
p = effective plastic strain.
Symonds strain rate parameters; εeff

F I G U R E 4 True stress–true strain curves (10
3 s
1) and its

comparison with the proposed Cowper–Symonds approach for the
6061-T6 and 7075-T651 aluminum alloys [Colour figure can be
viewed at wileyonlinelibrary.com]

impact tests. This value was adjusted considering, from a

meshing sensitivity analysis, that there is a close
relationship between the failure criterion εeff
p,fracture and
the element size. This affects the time duration of the
simulated force–time curves.

3 | RESULTS A ND DISCUSSION

For the HAZ and WM of both 6061-T6 and 7075-T651

alloys, the C and P parameters were adjusted to the experi-
F I G U R E 3 Experimental strain rate sensitivity for the C and P mental data, taking into account the values obtained by
Cowper–Symonds parameters for (A) 6061-T6 and (B) 7075-T651, the analytical results of the corresponding BM for each
experimental values were taken from literature42,43 [Colour figure joint (Figure 3). Table 1 shows the values, which were con-
can be viewed at wileyonlinelibrary.com] sidered the most convenient to determine the force–time
6 FRUTOS MARTÍNEZ ET AL.

behavior for the 6061-T6 and 7075-T651 alloys. In contrast, 6061-T6 and 7075-T651 alloys for the different zones of
due to the unavailable data of HAZ and WM of joints the joints.15
tested at different strain rates and considering the compli- A ductile behavior for the 6061-T6 alloy can be
cations involved to determine accurately the local mechan- inferred from Figure 5A, reaching a maximal impact
ical properties for each zone, the empirical parameters of force Fmax of about 7.5 kN. This force corresponds with
those zones were adjusted to fit their corresponding exper- an impact time of approximately 0.5 ms. For the
imental force–time curve (Figure 3). 7075-T651 alloy (Figure 5B), the Fmax value is higher than
It should be pointed out that the adjustments of the the 6061-T6 alloy (more than 17%). However, a brittle
different zones were realized assuming the P parameter behavior (impact time of 0.13 ms) is observed. Thus, the
of the BM as reference. According to Equation 5 and 6061-T6 alloy shows an impact energy 6.5 times greater
Figure 3, this term describes the slope of the strain rate than the 7075-T651 alloy.
sensitivity of materials, which has been considered as the In the case of the WM and HAZ, a ductile behavior
same for the different zones. For this reason, the WM was observed, decreasing the maximal impact force in
and HAZ approaches can be applicable if it is considered comparison with its respective base material. For WM,
only the loading conditions modeled. the Fmax value tends to decrease roughly 67% (6061-T6)
and 58% (7075-T651) in comparison with the BM before
welding. This behavior was previously attributed to the
3.1 | Experimental Charpy impact tests welding defects (porosity) as well as the microstructural
results characteristics of the welded joints,3,4,6,13,15 which
promote the crack growth rate propagation during
Figure 5 shows the experimental force–time behavior impact.15,31 On the other hand, the fracture time for the
obtained from the instrumented Charpy device for a HAZ increases considerably in relation with the BM

F I G U R E 5 Force versus time behavior obtained from instrumented Charpy impact tests, (A) 6061-T6, (B) 7075-T65115, and
(C) schematic curve of test [Colour figure can be viewed at wileyonlinelibrary.com]
FRUTOS MARTÍNEZ ET AL. 7

and WM. This phenomenon is attributed to the the specimens, the ductile fracture for the 6061-T6 BM,
microstructural transformation (over-aging process)6–9,12–14 and HAZ for both materials is observed, which is in
generated by the heat input during the welding process agreement with the Charpy impact records. However, for
that decreases the hardening of the HAZ but increases the 7075-T651 BM and WM, the fracture seems to be brit-
ductility. In both materials, the microstructural transfor- tle confirming the observations highlighted by the impact
mation is generated by the thermodynamic instability of tests data. Figures 7 and 8 show fractographic images of
the coherent precipitates. In the case of the 6061-T6 alloy, the specimens. Images were taken by using scanning
the metastable β´´ transforms into the incoherent β´ electron microscopy through a TM3030Plus Tabletop
precipitate,4,6,8,13 whereas for the 7075-T651 alloy, the microscope with an acceleration voltage of 15 kV and
over-aging is produced by the transformation of the η´ to η secondary-electron mode observation. The lateral expan-
precipitates.2,3,9,14,15 Even if the over-aging phenomenon sion provides information on the nature of the fracture,
decreases the mechanical properties in welds of which is ductile for BM of the 6061-T6 alloy (Figure 7A)
aluminum alloys, the porosity formation in the WM zone and HAZ for both alloys (Figures 7E and 8E). This is con-
is recognized as the main cause of crack nucleation and its firmed by dimple observation at the fracture surface,
corresponding cause of failure.4,6,15 which are generated by the coalescence of microvoids ini-
In general, the relation between the material tiated by decohesion and cracking of brittle particles dur-
hardness and the impact period is due to strain rate sensi- ing the impact process (Figures 7B,F and 8F). In contrast,
tivity. A brittle material will have short displacements the flat fracture surface observed in the 7075-T651 alloy
before it breaks and will take less time to fracture even if allows to deduce a brittle behavior, that is, no dimple
a higher force is reached. In contrast, a ductile material formation. For the WM (Figures 7C,D and 8C,D), a com-
could reach a lower force than a brittle material. How- bination of brittle–ductile fracture can be observed. Brit-
ever, it exhibits a stable crack growth, which increases tle characteristics are promoted by the inherent porosity
the displacement and consequently the fracture energy. generated by the welding process,3,13 whereas the
presence of dimples is attributed to a ductile behavior.
The fracture surface findings can be associated with
3.2 | Fracture surfaces the energy absorption capacity experimentally deter-
mined for the different materials. It is means that brittle
Lateral view of the fractured Charpy specimens for BM, fracture is characterized by higher impact force, short
WM, and HAZ are shown in Figure 6. Considering the impact times, and vice versa for a ductile behavior.
crack direction pattern as well as the deformation of Figure 5 and Table 2 present that maximal impact force

F I G U R E 6 Lateral view of the fractured

Charpy specimens [Colour figure can be
viewed at wileyonlinelibrary.com]
8 FRUTOS MARTÍNEZ ET AL.

FIGURE 7 Fracture surfaces, (A,B) 6061-T6 aluminum alloy, (C,D) weld metal, and (E,F) heat-affected zone

FIGURE 8 Fracture surfaces, (A,B) 7075-T651 aluminum alloy, (C,D) weld metal, and (E,F) heat-affected zone

for 6061-T6 and 7075-T651 is higher than WM and HAZ. over-aging process, which produces incoherency with the
However, the impact fracture energy is lower than its aluminum matrix of the fine, coarse, and distributed pre-
respective HAZ. It means that microstructural transfor- cipitates. Finally, the combination of welding defects
mation produced by the welding heat increases the such as porosity as well as the dendritic microstructure13
energy absorption capacity of the HAZ, due to the produced by the solidification process reduced the
FRUTOS MARTÍNEZ ET AL. 9

TABLE 2 Comparison between experimental and finite element results for Charpy impact tests of a 6061-T6 and 7075-T651 welded
joints

Material Fmax (kN) Fdy (kN) Fug (kN) t (ms) Fracture energy (J) s (mm) FEM accuracy (%)
6061-T6
BM (experimental) 8.00 6.5 6.70 1.46 34.70 7.41 89.36
BM (FEM) 8.50 7.50 7.75 1.93 38.83 —
WM (experimental) 2.60 2.20 2.40 0.56 4.72 2.72 68.90
WM (FEM) 3.35 2.24 2.65 0.92 6.85 —
HAZ (experimental) 5.36 3.71 4.00 2.90 38.67 16.60 91.46
HAZ (FEM) 4.82 3.20 4.30 2.78 42.28 —
7075-T651a
BM (experimental) 8.76 7.09 7.50 0.26 4.72 1.25 94.50
BM (FEM) 11.50 8.20 8.60 0.24 4.46 —
WM (experimental) 4.04 2.92 3.60 0.90 7.88 5.00 91.20
WM (FEM) 4.18 2.45 3.19 0.92 8.64 —
HAZ (experimental) 6.90 5.38 5.90 2.05 31.13 10.0 95.43
HAZ (FEM) 7.40 5.81 6.14 2.28 32.62 —
a
Results were obtained considering a homogeneous material for the Charpy sample. Fmax, Fdy, and Fug were determined according to the ASTM E2298.27

F I G U R E 9 Plastic strain and stress distribution during the impact for a 6061-T6 alloy, (A) von Mises plastic strain at the starting
contact, (B) von Mises plastic strain at 0.05 ms, (C) von Mises stress contours (units in Pa), and (D) fracture appearance [Colour figure can
be viewed at wileyonlinelibrary.com]
10 FRUTOS MARTÍNEZ ET AL.

fracture energy for the 6061-T6 WM by approximately is in agreement with the fracture mechanics theory for
85% in comparison with the BM. Even if the 7075-T651 ductile materials. In contrast, for the 7075-T651 alloy,
WM has poor metallurgical characteristics, it was due to its brittle behavior, no plastic zone is developed.
observed that fracture energy of the metal increases Thus, the stress concentration is extending through the
approximately 47% in comparison with the BM. width of the material from the notch tip. This is probably
the result of the stress triaxiality, which magnifies the
brittle characteristics of the fracture. For instance, the tri-
3.3 | Finite element results axiality factor (TF) is roughly 0.78 and 0.82 for the
6061-T6 and 7075-T651, respectively. These values were
Finite element simulations were performed according to determined by means of the hydrostatic stress (σ m) and
the methodology described above in Section 2.2. The the equivalent von Mises stress (σ eq) according to the
force versus time response was determined for the differ- relation: TF ¼ σ m=σeq .
ent zones of the welded joints. A comparison of the force versus time curves between
Figures 9 and 10 show the von Mises plastic strain the experimental and the simulated results for BM, WM,
contours as well as the stress distribution for 6061-T6 and and HAZ for both aluminum alloys are shown in
7075-T651 alloys obtained from the finite element calcu- Figures 11 and 12. These results were obtained consider-
lation. Real fractured sample appearance is shown in ing homogeneous materials for the Charpy samples, that
Figures 9D and 10D. is, the material for the WM and HAZ was considered for
When comparing the stress distribution in Figures 9A the entire specimen.
and 10A, a noticeable difference is observed. Indeed, for According to Figures 11 and 12, it is possible to
the ductile 6061-T6 alloy, the stress concentration at the observe that numerical results have in general an
notch tip generates a plastic zone (zone marked in yel- acceptable approximation with those experimentally
low) forming a shape like butterfly wings. This geometry obtained (Figure 5). However, some discrepancies between

F I G U R E 1 0 Plastic strain and stress distribution during the impact for a 7075-T651 alloy, (A) von Mises plastic strain at the starting
contact, (B) von Mises plastic strain at 0.05 ms, (C) von Mises stress contours (units in Pa), and (D) fracture appearance [Colour figure can
be viewed at wileyonlinelibrary.com]
FRUTOS MARTÍNEZ ET AL. 11

F I G U R E 1 1 Force versus
time curves obtained for the
explicit finite element
calculation and their respective
comparison with experimental
curves of the 6061-T6 aluminum
alloy, (A) base metal, (B) weld
metal, and (C) heat-affected
zone [Colour figure can be
viewed at wileyonlinelibrary.
com]

F I G U R E 1 2 Force versus
time curves obtained for the
explicit finite element
calculation and their respective
comparison with experimental
curves of the 7075-T651
aluminum alloy, (A) base metal,
(B) weld metal, and (C) heat-
affected zone [Colour figure can
be viewed at wileyonlinelibrary.
com]
12 FRUTOS MARTÍNEZ ET AL.

experimental and numerical results were found. For difference between the finite element calculation and the
instance, in the finite element model, it is possible to experimental results. The difference between simulation
observe a perturbation of the force versus time curves at and experimental results in terms of fracture energy for
the beginning of the impact (first oscillation mode materials with ductile behavior (BM and HAZ for 6061-T6
reported by Sahraoui and Lataillade24), which are alloy, HAZ and WM for 7075-T651 alloy) was up to roughly
associated to the natural oscillation of the Charpy sample. 10%, whereas for a brittle material (7075-T651 alloy) values,
This phenomenon is not clearly observed in the it is around 5%. The more significant differences were
experimental data, which could be attributed to the data found for the 6061-T6 alloy, a high tensile strength alloy
acquisition rate at which the experiments were carried out with ductile behavior (Figure 5A).
(100 kHz). It was observed that by using C and P values The complications to evaluate numerically large
of high order, it is possible to reduce those numerical displacements-shear problems were reported by
oscillations. Another option is to decrease the σ 0 value in Nagtegaal and Jong.46 These authors found that when
the constitutive model; however, this option was not kinematic hardening is implemented for materials which
chosen because it would mean that materials yielding harden monotonically in tension, an oscillatory response
decreased at high strain rates, which is in discordance with is observed. Johnson and Bammann47 attributed this
literature.23,25 A second perturbation is shown at the behavior to the nature of the rate type constitutive model
beginning of the unstable crack growth force, which is employed, that is, this effect can occur even if plasticity is
attributed to the formation of the new surfaces caused by considered or hardening mechanisms are introduced.
the evolution of the specimen fracture at a given moment. Nevertheless, the method developed in this paper allows
It is important to remark that this effect could not be to obtain satisfactory results at a macro-scale level and
reduced by increasing the number of elements. leads to have an adequate qualitative trend even if some
Taking into account the experimental and finite ele- divergences are observed. This methodology could be
ment results in conjunction with the schematic force– enhanced by introducing a nonlinear hardening model to
time curve shown in Figure 5C, it is possible to observe improve the limitation of the Cowper–Symonds constitu-
that E, σ 0, ET, and Ep tend to modify the dynamic yield- tive model in terms of hardening evolution.
ing force (Fdy), as well as the hardening behavior and
fracture initiation of the material (Fin). On the other
hand, changes in C, P, β, and εeff
p,fracture parameters tend to
3.4 | Heterogeneity of the welded joint
affect the maximal force (Fmax), the force at the begin-
ning of the unstable crack growth (Fug), the unstable Additionally, the material heterogeneity effect on the
crack growth, the collapse energy (UCGE + CE), and quality of the finite element results was evaluated. This
aspect was analyzed by using a multi-material modeling
crack failure, respectively. The failure criterion εeff
p,fracture is of the Charpy sample for the welded joints according to
controlled by the effective plastic strain εeff p, , which the impact zone. Distances from the hardness profiles
is computed by the computational algorithm.41 given previously by the authors3,13 (Figure 1A) were
Table 2 shows a comparison between the experimen- taken into account to divide the Charpy specimen into
tal and the finite element results. The accuracy of the different volumes (Figure 13). The mechanical properties
finite element model was calculated by means of the rela- of each zone (BM, HAZ, and WM) were given according
tionship between the energies from the force versus time to the data reported in Table 1.
curves compared with those experimentally obtained. Figure 14 shows slightly different results when com-
The Cowper–Symonds constitutive model is acceptable paring the multi-material with the homogenous
to obtain the force–time behavior during the Charpy approach. As seen previously in Figures 9 and 10, the
impact test in aluminum-welded joints. However, it was evolution of the fracture and consequently the major
observed that for the 6061-T6 alloy it is more complicated energy consumption takes place on the volume, which
to balance the empirical model coefficients (C and P) to surrounds the notch for the entire fracture event. This
match the proposed approach with the experimental data. confirms that fracture process is a localized phenomenon,
This could be explained considering that constitutive model and it is not necessary to consider all the zones involved
tends to overestimate the displacement calculations,45 in an aluminum-welded joint for a Charpy specimen.
increasing the complications to reach acceptable results in Thus, it is not necessary to include the mechanical
terms of impact time. It means that the model is not suit- parameters for the entire thermal affected regions of the
able to capture the real local behavior in terms of plastic 6061-T6 and 7075-T651 welded joints in the same finite
flow (hardening parameters). For this reason, the fracture element model of a Charpy impact. Similarity in terms of
energy of ductile materials (6061-T6 alloy) shows a bigger energy between homogeneous and multi-material
FRUTOS MARTÍNEZ ET AL. 13

F I G U R E 1 3 Multi-material
configuration of a Charpy
specimen for (A) heat-affected
zone and (B) weld metal [Colour
figure can be viewed at
wileyonlinelibrary.com]

F I G U R E 1 4 Force versus time curves comparison for homogeneous material and multi-material approach for the Charpy specimen,
(A) 6061-T6 weld metal, (B) 6061-T6 heat-affected zone, (C) 7075-T651 weld metal, and (D) 7075-T651 heat-affected zone [Colour figure can
be viewed at wileyonlinelibrary.com]

approaches was 99.95% for HAZ and 99.93% for WM in BM. Consequently, the fracture process is concentrated
the case of the 6061-T6 alloy, whereas for the HAZ in the weakest material and plasticity is not involved in
and WM of the 7075-T651 alloy, values of 91.41% and the BM.
93.64% were obtained. This is probably because the Moreover, a multi-material approach may produce
mechanical properties of HAZ and WM in terms of yield discrepancies in numerical models due to the mismatch
strength are lower than that of their corresponding of the mechanical properties. The flexural response of the
14 FRUTOS MARTÍNEZ ET AL.

entire specimen induced a stress concentration in Such approach allows computing the local mechanical
the zones with the lowest εeff p,fracture values, leading to a quantities more accurately and thus better controlling
rigid behavior of the specimen, producing a penetration the fracture process under impact. In this case, specific
of the striker into the material. To eliminate this effect, experiments must be conducted on notched specimen in
the failure criterion of the impacted volume was reduced tension under different strain rates and different notch
by readjusting the value of εeff
p,fracture to 0.01 for both cases radii to capture the stress triaxiality effects and therefore
(6061-T6 WM and 7075-T651 HAZ). This value allows to extrapolate the results to V notch on Charpy specimen.
concentrate the fracture zone within the notch volume. It
is to say that failure criterion increment of the adjacent ACKNOWLEDGMENTS
zones does not affect the energy absorption of the speci- The authors thank CONACyT-México (project
men. However, a reduction of the failure criterion can A1-S-27474) and SIP-IPN for supporting this research as
induce fracture outside the notched zone. Another option well as the support of the Université de Lille (Unité
could be to establish an equivalent failure criterion for Mécanique).
the entire specimen. This would be considered for a
future work. CONFLICT OF INTEREST
The authors declare no conflict of interest.

4 | C ON C L U S I ON S DATA AVAILABILITY STATEMENT

Data available on request from the authors.
By means of the Cowper–Symonds constitutive model, it
is possible to approximate the force versus time curves ORCID
obtained from Charpy impact tests for the different zones José A. Frutos Martínez https://orcid.org/0000-0001-
of the 6061-T6 and 7075-T651 aluminum alloys welded 7290-9537
joints. Ricardo R. Ambriz https://orcid.org/0000-0001-8255-
The accuracy of the fracture energy absorbed by the 0484
materials between the experiments and the finite element
Moussa Naït-Abdelaziz https://orcid.org/0000-0001-
calculation reaches up to 93.46% for the 6061-T6 alloy 6541-2899
(WM), whereas for the 7075-T651 alloy, a maximum David Jaramillo https://orcid.org/0000-0003-2309-2545
accuracy of 97.98% (HAZ) was obtained. The lowest
accuracies were observed when simulating the response RE FER EN CES
of BM and WM of the 6061-T6 alloy. Even the Cowper–
Symonds model does not consider the plasticity damage 1. Ducker worldwide, "Aluminum content in North American
light vehicles 2016 to 2028," Troy, MI 48084, 2017.
of materials, the results show that in terms of global
2. ASTM B209, Standard specification for aluminum and
quantities, it has an acceptable accuracy. aluminum-alloy sheet and plate [M], West Conshohocken:
Even if the multi-material approach was rough, a ASTM, 2010.
comparison with the homogeneous results shown that, at 3. Alatorre N, Ambriz RR, Noureddine B, Amrouche A, Talha A,
the macro-scale, multi-material model is close to the Jaramillo D. Tensile properties and fusion zones hardening for
homogeneous model. But, at the micro-scale, it is obvious GMAW and MIIEA welds of a 7075-T651 alluminum alloy.
that strong differences will be pointed out when Acta Metall Sin. 2014;27(4):694-704.
4. Ambriz RR, Barrera G, García R, L opez VH. The microstruc-
analyzing the mechanical fields around the notch tip.
ture and mechanical strength of Al-6061-T6 GMA welds
To improve the local performance of the finite
obtained with the modified indirect electric arc joint. Mater
element model close to the crack tip, it is necessary to Des. 2010;31(6):2978-2986.
use a more robust constitutive model accounting for the 5. Bardel D, Perez M, Nelias D, Dancette S, Chaudet P,
nonlinearity of the stress–strain curve as well as the Massardier V. Cyclic behaviour of a 6061 aluminium alloy:
strain rate effects. Also, it is necessary to identify the frac- coupling precipitation and elastoplastic modelling. Acta Mater.
ture strain more accurately according to the comments 2015;83(1):256-268.
given in Section 2.2.3. This is an important issue to be 6. Yongxian H, Xiangchen M, Zongliang L, et al. Microstructures
and mechanical properties of micro friction stir welding
addressed because this parameter plays an important role
(μFSW) of 6061-T4 aluminum alloy. J Mater Res Technol. 2018;
in the fracture mechanism modeling, and its dependence
8(1):1084-1091.
on strain rate and stress triaxiality must be assessed. 7. Ringer SP, Hono K. Microstructural evolution and age harden-
Another framework to deal with the problem of ing in aluminum alloys: atom probe field-ion microscopy and
impact fracture could be a damage approach such as that transmission electron microscopy studies. Mater Charact. 2000;
developed by Gurson–Tvergaard48,49 for ductile materials. 44(1):101-131.
FRUTOS MARTÍNEZ ET AL. 15

8. Vargas JA, Torres JE, Pacheco JA, Hernandez RJ. Analysis of 24. Sahraoui S, Lataillade JL. Analysis of load oscillations in
heat input effect on the mechanical properties of Al-6061-T6 instrumented impact testing. Eng Fract Mech. 1998;60(4):
alloy weld joints. Mater Des. 2013;52:556-564. 437-446.
9. Alexopoulos ND, Stylianos A, Campbell J. Dynamic fracture 25. Jones N. Structural Impact. Cambridge, Reino Unido:
toughness of Al–7Si–Mg (A357) aluminum alloy. Mech Mater. Cambridge University Press; 2003.
2013;58:55-68. 26. ASTM E23, Standard test methods for notched bar impact test-
10. Gutierrez I. Effect of microstructure on the impact toughness ing of metallic materials, ASTM: West Conshohocken, United
of Nb-microalloyed steel: generalisation of existing relations States, 2018.
from ferrite–pearlite to high strength microstructures. Mater 27. ASTM E2298, Standard test method for instrumented impact
Sci Eng A. 2013;571:57-67. testing of metallic materials, ASTM: West Conshokochen,
11. Pallaspuro S, Kaikilainen A, Mehtonen S, Kömi J, Zhang Z, United States, 2018.
Porter D. Effect of microstructure on the impact toughness 28. Xing M-Z, Wang Y, Jiang Z-X. Dynamic fracture behaviors of
transition temperature of direct-quenched steels. Mater Sci selected aluminum alloys under three-point bending. Def
Eng. 2018;712(17):671-680. Technol. 2013;9(4):193-200.
12. Temmar M, Hadji M, Sahraoui T. Effect of post-weld aging 29. Tronskar JP, Mannan MA, Lai MO. Measurement of
treatment on mechanical properties of tungsten inert gas fracture initiation toughness and crack resistance in
welded low thickness 7075 aluminium alloy joints. Mater Des. instrumented Charpy impact testing. Eng Fract Mech. 2002;
2011;32(6):3532-3536. 69(3):321-338.
13. Ambriz RR, Mesmacque G, Ruiz A, Amrouche A, 30. Perez-Martin MJ, Erice B, Galvez F. Experimental determina-
Lopez VH. Effect of the welding profile generated by the tion of the dynamic fracture-initiation toughness of high-
modified indirect electric arc technique on the fatigue strength metals. Eng Fract Mech. 2019;205:498-510.
behavior of 6061-T6 aluminum alloy. Mater Sci Eng A. 2010; 31. Frutos A, Montealegre EG, Ambriz RR, Plascencia G,
527(7-8):2057-2064. Jaramillo D. Experimental and numerical approach on the
14. Alatorre N, Ambriz RR, Amrouche A, García C, Jaramillo D. impact behavior of aluminum welds. In: New trends in fatigue
Fatigue crack growth in Al–Zn–Mg (7075-T651) welds obtained and fracture—NTF2F18. Portugal: Lisbon; 2018.
by modified indirect and gas metal arc welding techniques. 32. Lim HK, Lee J-S. On the structural behavior of ship's shell
J Mater Process Technol. 2017;248:207-217. structures due to impact loading. Int J Nav Archit. 2018;10(1):
15. Ambriz RR, Jaramillo D, García C, Curiel F. Fracture energy 103-118.
evaluation on 7075-T651 aluminum alloy weld determined by 33. Jones N. Quasi-static analysis of structural impact damage.
instrumented impact pendulum. T Nonferr Metal Soc. 2016;26 J Constr Steel Res. 1995;33(3):151-177.
(4):974-983. 34. Hernandez C, Maranon A, Ashcroft A, Casas-Rodriguez JP. A
16. Montealegre E, Ambriz RR. Simulacion numérica y validaci on computational determination of the Cowper–Symonds parame-
experimental del comportamiento al impacto Charpy en una ters from a single Taylor test. Appl Math Model. 2013;37(7):
aleacion 7075-T651. In: XXI Congreso Internacional Anual de la 4698-4708.
Sociedad Mexicana de Ingeniería Mec anica. México: Veracruz; 35. Jones N. Some recent developments in the dynamic
2015. inelastic behaviour of structures. Ships Offshore Struct. 2006;
17. Chen Y, Pedersen KO, Clausen AH, Hopperstad OS. An experi- 1(1):37-44.
mental study on the dynamic fracture of extruded AA6xxx and 36. Cui P, Liu Y, Huo J. Dynamic behaviour of square tubular
AA7xxx aluminium alloys. Mater Sci Eng A. 2009;523(1-2): T-joints under impact loadings. J Constr Steel Res. 2018;143:
253-262. 208-222.
18. Chauhan A, Kumar A. Impact strength of joints of 37. Liua S, Kouadri-Henni A, Gavrusa A. DP600 dual phase steel
aluminium matrix composite formed using friction stir thermo-elasto-plastic constitutive model considering strain rate
welding technique. Mater Today: Proceedings. 2021;38(1): and temperature influence of FEM residual stress analysis of
234-236. laser welding. J Manuf Process. 2018;35:407-419.
19. Zhang S, Wang Y, Zhu M, Zhang Z, Nie P, Li Z. Relationships 38. Ambriz RR, Froustey C, Mesmacque G. Determination of the
among Charpy impact toughness, microstructure and fracture tensile behavior at middle strain rate of AA6061-T6 aluminum
behavior in 10CrNi3MoV steel weld joint. Mater Lett. 2020; alloy welds. Int J of Imp Eng. 2013;60:107-119.
281(15):1-4, 128328. 39. Stoffel M, Schmidt R, Weichert D. Validation of constitutive
20. Sudin VV, Kantor MM, Solnttsev KA. Features of weld metal laws for simulations of the dynamic response of plates. PAMM
brittle fracture in Charpy tests. Procedia Struct Integr. 2020;28: Proceedings of Appl Mech. 2003;3(1):210-211.
1637-1643. 40. Landrein P, Lorriot T, Guillaumat L. Influence of some test
21. Jebaraj A, Aditya KVV, Kumar TS, Ajaykumar L, Deepak CR. parameters on specimen loading determination methods in
Mechanical and corrosion behaviour of aluminum alloy 5083 instrumented Charpy impact tests. Eng Fract Mech. 2001;
and its weldment for marine applications. Mater Today: Pro- 68(15):1631-1645.
ceedings. 2020;22(4):1470-1478. 41. ANSYS. LS-DYNA User's Guide; 2019.
22. Lucon E. Estimating dynamic ultimate tensile strength from 42. Sasso M, Forcellese A, Simoncini M. Strain rate behaviour
instrumented Charpy data. Mater Des. 2016;97:437-443. of AA7075 aluminum alloy at different initial temper states.
23. Meyers MA. Dynamic Behavior of Materials. New York: John In: Key Engineering Materials. Vol.651–653. Switzerland:
Wiley & Sons Inc; 1994. Eindhoven, Trans Tech Publications; 2015:114-119.
16 FRUTOS MARTÍNEZ ET AL.

43. Scapin M, Manes A. Behaviour of Al6061-T6 alloy at different 48. Yildiz RA, Yilmaz S. Experimental investigation of GTN model
temperatures and strain-rates: experimental characterization parameters of 6061 Al alloy. Eur J of Mech/a Solids. 2020;
and material modelling. Mater Sci Eng A. 2018;734:318-328. 83(9-10):1-10, 104040.
44. Wang B, Xiao X, Astakhov VP, Liu Z. The effects of stress triax- 49. Wen J, Huang Y, Hwang KC, Liu C, Li M. The modified
iality and strain rate on the fracture strain of Ti6Al4V. Eng Gurson model accounting for the void size effect. Int J Plast.
Fract Mech. 2019;219(10):1-15, 106627. 2005;21(2):381-395.
45. Alizadeh Dehkharghani A. Tuning Johnson–Cook Material
Model Parameters for Impact of High Velocity, Micron Scale
Aluminum Particles. Boston, Massachusetts: Northeastern
University; 2016. How to cite this article: Frutos Martínez JA,
46. Nagtegaal J. C. and Jong J. E., “Some aspects of nonisotropic Ambriz RR, Naït-Abdelaziz M, Jaramillo D. Impact
work-hardening in finite stress plasticity”, Numerical Analysis test behavior of aluminum alloys welded joints:
of Metalforming Processes: Aplication and Experiment Verifica- Experimental and numerical analysis. Fatigue
tion, Eindhoven, Technische Hogeschool Eindhoven, 1982;
Fract Eng Mater Struct. 2021;1–16. https://doi.org/
65–103. https://doi.org/10.6100/IR107405
47. Johnson GC, Bammann DJ. A discussion of stress rates in finite
10.1111/ffe.13482
deformation problems. Inr J Solids Structures. 1984;20(8):
715-737.