Anda di halaman 1dari 14

Engineering Fracture Mechanics 87 (2012) 48–61

Contents lists available at SciVerse ScienceDirect

Engineering Fracture Mechanics


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

A finite element model for the prediction of Advanced High Strength


Steel spot welds fracture
S. Dancette a,⇑, D. Fabregue a, R. Estevez b, V. Massardier a, T. Dupuy c, M. Bouzekri c
a
Université de Lyon, INSA de Lyon, MATEIS-UMR CNRS 5510, Bât. Saint-Exupery, 25 Avenue J. Capelle, 69621 Villeurbanne Cedex, France
b
Laboratoire SIMaP, UMR CNRS 5266, Université de Grenoble, Grenoble-INP, UJF, 1130 rue de la piscine, B.P. 75, 38402 St. Martin d’Hères Cedex, France
c
ArcelorMittal R&D, Automotive Products, Voie Romaine, 57283 Maizières les Metz, France

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

Article history: A finite element model has been developed for the simulation of resistance spot welds
Received 7 February 2011 behavior in Cross Tension. The model is able to capture the competition between semi-brit-
Received in revised form 22 December 2011 tle fracture at the faying surface and ductile failure at the weld boundary. For this purpose,
Accepted 10 March 2012
ductile damage and fracture are considered in the different weld zones in addition to a
cohesive zone at the faying surface and an evaluation of the Rice integral (J) at the notch
tip. The model is run for different weld configurations made of TRIP780 steel. It reproduces
Keywords:
the proper level of load at the onset of cracking and predicts the governing damage mech-
Resistance spot welding
Cross Tension
anism leading to final fracture as a function of the geometrical features of the welds.
Finite elements Ó 2012 Elsevier Ltd. All rights reserved.
Damage
Fracture

1. Introduction

Advanced High Strength Steels (AHSSs) for automotive applications have been the focus of intense research and develop-
ment in the last decades. Their introduction in body in whites allows to reduce their weight and improve their behavior in
crash conditions. Nevertheless, the weldability of AHSS is an important issue since cars typically contain thousands of spot
welds. Consequently, the mechanical behavior of a body in white does not only depend on the sheets mechanical properties
but also on the joints strength. Being able to assess the latter is therefore of great importance in the automotive industry.
AHSS spot welds are often more prone to complex crack paths in the weld nugget than conventional mild steels. This can
give rise to so-called Partial or Full Interfacial Failures (PIF or FIF) rather than to the usual Button Pullout (BP) and can be
observed for example during quasi-static Cross Tension or Tensile Shear tests (Fig. 1), as usually done for the determination
of their weldability [1–3]. Understanding the mechanisms leading to the various failure types is therefore important in an
attempt to model the spot weld behavior and its load bearing capacity.
The geometrical features of the weld assembly are known to influence the mechanical behavior of spot welds [5–9],
among other factors related to the mode of loading and the metallurgical transformations occurring in the weld zones during
the welding process. These geometrical parameters include the weld size relative to the sheet thickness and to the speci-
men’s width, sharpness of the notch tip, distance between the machine grips, etc.
Ductile failure at the nugget boundary has been observed in the past in case of mild steel or HSLA spot welds subjected to
normal loads [10,11], while steels with a higher alloying level could present brittle (partial) interfacial failure under such
loading mode [4,6,12,13]. Large spot welds subjected to shear loads were shown to fail by strain localization in the Base

⇑ Corresponding author.
E-mail address: sylvain.dancette@insa-lyon.fr (S. Dancette).

0013-7944/$ - see front matter Ó 2012 Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.engfracmech.2012.03.004
S. Dancette et al. / Engineering Fracture Mechanics 87 (2012) 48–61 49

Nomenclature

BM Base Metal
HAZ Heat Affected Zone
CGHAZ Coarse Grained Heat Affected Zone
SCHAZ Subcritical Heat affected Zone
BP Button Pullout
FIF Full Interfacial Failure
PIF Partial Interfacial Failure
SIF Stress Intensity Factor
c spacing between the machine grips (mm)
d weld diameter (mm)
dA surface element, (mm2)
D ductile damage variable
E Young’s modulus (GPa)
F Cross Tension load (N)
J J-integral (kJ/m2)
Jc critical value of J at the onset of crack propagation
KCZ stiffness of the cohesive zone (MPa/mm)
L characteristic element length (mm)
n normal to dA
q direction of virtual crack extension
t sheet thickness (mm)
U crosshead displacement (mm)
 pl
u f
effective plastic displacement at fracture (mm)
W strain energy density (kJ/m3)
wD state variable for ductile damage initiation
d0 opening displacement at the onset of damage (mm)
dc opening displacement at failure (mm)
epl logarithmic plastic strain
epl
0 equivalent plastic strain at the onset of ductile damage
eexper
f
experimental ultimate fracture strain
b
C work of fracture of the cohesive zone (kJ/m2)
C0 elastic part of the work of fracture (kJ/m2)
CCZ cohesive part of the work of fracture (kJ/m2)
k(s) virtual crack advance (mm)
m Poisson’s ratio
r Cauchy stress (MPa)
r^ peak strength of the cohesive zone (MPa)
r undamaged equivalent stress (MPa)
h angle to loading direction

Metal (BM) close to the weld [10,11,14–17] and smaller ones would fail at the faying surface with sheared fracture surfaces
[14,18–21].
The weld failure zones and damage processes occurring during the Cross Tension testing of a range of High Strength and
Advanced High Strength Steels was summarized by the authors in [7,9]. These investigations were based on coupled X-ray
microtomographic, metallographic and fractographic analyses. The results are illustrated in Fig. 2.

Tensile Shear Cross Tension

Fig. 1. Sketch of the Cross Tension and Tensile Shear tests [4].
50 S. Dancette et al. / Engineering Fracture Mechanics 87 (2012) 48–61

Fig. 2. Failure zones and damage mechanisms during the Cross Tension testing of AHSS resistance spot welds [9].

Zone 1 is the BM/Subcritical Heat Affected Zone (BM/SCHAZ) close to the weld, where strain localization is likely to occur
during the test due to its low yield strength compared to the hard bainitic–martensitic neighboring zones. Fracture in Zone 1
will lead to a BP failure type. Zone 2 is in the vicinity of the crack tip, usually between the Coarse Grained HAZ (CGHAZ) and
the weld nugget, where nugget pullout by ductile shear can be observed. Zone 3 is in the weld nugget, starting from the
notch tip, where PIF or FIF may occur. Interfacial failure in Cross Tension is triggered by mode I opening of the notch tip,
potentially leading to (semi-) brittle fracture.
The final failure type and load bearing capacity of a spot weld result from a competition between these main failure
mechanisms. Several of them may take place simultaneously, giving rise to a wide variety of spot weld fracture appearances,
as detailed for example by the American Welding Society [22].
Based on such understanding of spot weld fracture, different approaches can be adopted to model the load bearing capac-
ity under a given loading mode: statistical analysis or data mining approaches have been used to predict the weld strength
based on welding conditions and the sheets characteristics for example (thickness, mechanical properties, chemical compo-
sition, etc.) [12,23–26]. Another type of approach would try to develop analytical models based on elementary materials
properties (tensile strength, fracture toughness) and under the assumption of a simplified stress distribution in and around
the weld. This includes limit load analyses (in Cross Tension [27–29], Tensile Shear [11,28–30] or combined [11,31,32]) or
attempts to use the concepts of linear elastic fracture mechanics (LEFM) [27,33,34]. Nevertheless, the validity of LEFM is
questionable for the prediction of the overload failure of spot welds as the weld behavior may be significantly non-linear.
Finally, the Finite Element (FE) method is an attactive way to capture the non-linear weld response in the presence of large
deformations, as well as the detailed stress-deformation state in the vicinity of potential failure zones. The challenge in this
case lies in a adequate description of the gradient of mechanical properties from the BM to the weld nugget. In the absence of
local HAZ constitutive behavior, most authors would scale the BM response according to the evolution of a hardness profile
for example [35–38]. Different approaches have then been used to model damage and fracture in the FE simulations. These
include fracture strain failure criteria [39,37], Gurson-type modeling of ductile damage [35,36] potentially coupled with coa-
lescence criteria [38,40–43], introduction of cohesive zones at the potential failure zones [7,44–47] or the use of a critical
‘‘cleavage’’ stress [48]. A major difficulty in the use of these damage models consists in the identification of a ‘‘meaningful’’
set of the numerous model parameters (added to the multiplicity of the constitutive behaviors that have to be considered for
the description of the different weld HAZ), without resorting to extensive fitting procedures for specific test conditions.
In the present work, a FE model of resistance spot welds mechanical behavior is proposed. It is based on the insight gained
by the authors in [7,9] on welds behavior and relies on the local HAZ constitutive behaviors obtained in [7,8]. The model is
build so as to assess the competition between semi-brittle fracture from the notch tip at the faying surface and ductile failure
in the weld HAZ. Concerning brittle interfacial fracture, two approaches are investigated here, namely a finite strain estima-
tion of the Rice integral (J) at the notch tip and the introduction of a cohesive zone. Moreover, ductile damage and failure is
considered in the model for the assessment of ductile fracture in the different weld HAZ. Focus is given on a TRIP780 steel is
this study, while the model has been used elsewhere on other Advanced High Strength Steels [7,9]. The relevance of the
model predictions are discussed for different homogeneous and heterogeneous geometrical configurations of the Cross
Tension specimens.

2. Model

2.1. Spot weld model and constitutive behavior

Finite strain simulation of the weld mechanical response requires an appropriate description of the local HAZ constitutive
behavior in the weld. As introduced above, this is not an easy task, as both the discretization of the weld zones and the way
the constitutive behavior is obtained are debatable.
The weld model used is this work is depicted in Fig. 3. The partition of the weld zones is illustrated in Fig. 3(a), where the
weld dimensions for the different configurations were obtained from metallographic cross sections [7] and are listed in
Table 1. The weld nugget and the CGHAZ are assumed to exhibit similar bainitic/martensitic flow properties. Such simplified
description of the weld heterogeneity is a reliable and cost-effective alternative to a detailed discretization of the weld HAZ
(every 100 ° C) for the calculation of strain fields in the vicinity of potential failure zones [7].
S. Dancette et al. / Engineering Fracture Mechanics 87 (2012) 48–61 51

Fig. 3. Cross Tension model.

Table 1
Simulated weld configurations. Dimensions in mm.

Thick. t Weld diam. d HAZ diam. SCHAZ diam.


Homogeneous
1.0/ 1.0 5.0 7.0 8.5
1.5/ 1.5 5.0 7.0 8.5
1.5/ 1.5 7.0 8.0 9.5
Heterogeneous
1.0/ 2.0 5.0 5.3 6.8
1.0/ 2.0 5.5 5.8 7.3
1.0/ 2.0 6.5 6.8 8.3

Local mechanical properties in the weld were obtained from experimental Gleeble simulation of the resistance spot
welding thermal cycles, so as to reproduce the different HAZ microstructures. The specimens obtained were then tested
in tension at room temperature. This procedure is described in references [7,8] and an overview of the thermal cycles is given
in Table 2. The evolution of the flow stress in the weld zones is depicted in Fig. 4 for the TRIP780 steel investigated in this
study. A Swift-Voce law [49] was identified in order to extrapolate the tensile curves up to the large strain domain and was
used in conjunction with classical J2 isotropic plasticity for the description of the yield locus:

r ¼ ð1  aÞ½Kðe0 þ epl Þn  þ a½rsat  ðrsat  r0 Þ expðmepl Þ ð1Þ

where r and epl are respectively the Cauchy stress and the true logarithmic plastic strain, K,e0 and n are the parameters of the
Swift law, r0, rsat and m are the parameters of the Voce law and a is a fitted constant so as to fulfill the Considère criterion.
These parameters can be found in Table 3.

Table 2
Peak temperatures (Tmax) and cooling rates (Cr) in the Gleeble simulated thermal
cycles for the reproduction HAZ microstructures [8].

SCHAZ CGHAZ
Tmax (° C) 700 1200
Cr at 700 °C (°C/s) – 600
Cr at 450 ° C (°C/s) 400 350
52 S. Dancette et al. / Engineering Fracture Mechanics 87 (2012) 48–61

2000
1500
True stress, [MPa]
1000
500 BM
SCHAZ
martensite
0

0.0 0.2 0.4 0.6 0.8 1.0


Log. strain, [-]

Fig. 4. Constitutive behavior of the TRIP780 steel and its heat affected zones.

Table 3
Parameters of the Swift-Voce law for the TRIP780 steel microstructures.

BM SCHAZ CGHAZ
K (MPa) 1701 1799 2394
e0 (–) 0.02 0.0031 0.002
n (–) 0.312 0.254 0.114
r0 (MPa) 515 460 1074
rsat (MPa) 1155 1084 1612
m (–) 8.68 21.21 125.92
a 0.88 0.75 0.74

A typical weld mesh with Abaqus C3D8R elements [50] is presented in Fig. 3b, where the element size is in the order of
0.09 mm in the vicinity of the nugget and notch tip. This element type and mesh refinement was identified in [7] as a good
compromise between computational cost and accuracy. The quarter model of the Cross Tension specimen is depicted in
Fig. 3c. S4R shell elements [50] are used to transfer the load from the machine grips to the weld zone, which is an efficient
way to reduce the computational cost of the simulation without altering the response of the assembly [7].
Homogeneous and heterogeneous assemblies with regard to the sheet thickness were considered in this study (Table 1).
The configuration of the Cross Tension specimen is depicted in Fig. 5, where both Sheet 1 and Sheet 2 are either 1.0 mm or

Fig. 5. Cross Tension specimen, where both Sheet 1 and Sheet 2 are 1.0 mm or 1.5 mm thick in the homogeneous case, while Sheet 1 is 1.0 mm and Sheet 2
is 2.0 mm thick in the heterogeneous assembly. Angle h is 0° in the loading direction of Sheet 1 and 90° in the loading direction of Sheet 2.
S. Dancette et al. / Engineering Fracture Mechanics 87 (2012) 48–61 53

1.5 mm thick in the homogeneous case, while Sheet 1 and 2 are respectively 1.0 mm and 2.0 mm thick is the heterogeneous
case.

2.2. Failure criteria

2.2.1. J-integral evaluation at the notch tip


The J-integral is a widely accepted quasi-static fracture mechanics parameter extending the concept of the stress intensity
factor to cases where the condition of Small Scale Yielding is not met anymore. This is particularly the case during the over-
load analysis of an inelastic material such as quasi-static Cross Tension loading of a resistance spot weld.
Following Anderson [51] and Lee and Choi [35], the fracture parameter J carries dual meanings as an energy release rate
and an intensity of the crack-tip stress field. As an energy release rate, it assumes that an advancing crack in a homogeneous
material propagates in the same direction as the initial crack. As an intensity of the crack tip stress field, J at the nugget
boundary can be interpreted as a measure of local material damage, irrespective of crack growth direction and fracture loca-
tion. Both of these meanings can be relevant in case of interfacial fracture in a resistance spot weld, where the notch tip at
the nugget boundary acts as an external crack.
The J-integral along the notch tip in Cross Tension is evaluated in this work by the virtual crack extension/domain integral
methods implemented in Abaqus [50,52,53]:
Z  
@u
J¼ kðsÞn WI  r q dA ð2Þ
A @x
where k(s) is a virtual crack advance in the plane of fracture at the location s along the crack front, dA is a surface element
along a vanishing small tubular surface enclosing the crack tip, n is the outward normal to dA,q is the direction of virtual
Re
crack extension, @u
@x
is the spatial derivative of the local displacement, I is the unit matrix and W ¼ 0 r : d e is the strain en-
ergy density.
Its evolution from 0° (loading direction of Sheet 1, Fig. 5) to 90° (loading direction of Sheet 2) is considered. Mesh and
element type convergence with the present model was checked by the authors in [7]. Path independency of the J-integral
is ensured in this case by the simplified partition of welds zones (Fig. 3a) guarantying an homogeneous material along
the different contours. Given the geometry of the problem, with an infinite notch terminating as a crack at the faying surface,
noticeable stress triaxialities are expected at the crack tip and the J-integral may be used to assess the onset of crack prop-
agation, above a critical level.

2.2.2. Cohesive Zone model at the faying surface


A complementary modeling tool for the simulation of crack onset and propagation is the introduction of cohesive ele-
ments along the expected crack path, that is at the faying surface of the specimen. Cohesive zone models [47,54,55] allow
the description of decohesion at an interface both in terms of a peak strength and a decohesion energy. This may be partic-
ularly useful in this work where we want to model a semi-brittle crack propagation in the weld nugget at the interface [7,9].
Abaqus COH3D8 elements [50] (around 0.045 mm in size) were used in this study, with the bilinear, mode independent
constitutive behavior depicted in Fig. 6. The cohesive model implemented in Abaqus borrows the description due to
Camanho and Dávila [56]. The traction-opening profile accounts for a linear decay upon a maximum traction r ^ up to a
critical displacement jump dc. This corresponds to the damage process. In the normal mode, a linear response is considered
for (undamaged) traction or compression. In the tangential mode, a central symmetry with the origin is used. Failure with
the nucleation of a crack locally is observed when either the normal or tangential mode reaches the critical value dc

Fig. 6. Bilinear mode I cohesive zone model at the faying surface.


54 S. Dancette et al. / Engineering Fracture Mechanics 87 (2012) 48–61

(mode-independence). In the cases investigated here (cross tension), the loading of the faying surface exhibits a dominant,
mode I, normal load. Therefore, the tangential cohesive parameters (which are assumed identical to the normal ones for the
sake of simplicity) have no influence on the failure process.
The fracture behavior can be characterized by the energy dissipated C b , which is the area under the curve in Fig. 6 and is
obviously a function of r
^ and dc. It can be decomposed into its elastic C0 and cohesive CCZ contributions, while the former is
usually kept a small fraction of Cb by adjustment of the cohesive stiffness KCZ.

2.2.3. Ductile fracture


Beyond semi-brittle failure in the weld nugget, which is usually an undesired failure type, resistance spot welds in service
conditions often fail by the so-called ‘‘button pullout’’ exhibiting ductile fracture surfaces, either in the BM/SCHAZ or in the
nugget periphery [7,9]. Ductile fracture was thus considered in this work in the different weld zones. Experimental ultimate
fracture strains available from the tensile tests on the Gleeble simulated weld microstructures motivated the choice of the
progressive ductile damage and failure model in Abaqus [50].
Damage initiation is considered when the state variable wD reaches the value of 1:
Z epl
depl
wD ¼ ¼1 ð3Þ
0 epl
0

where epl
0 is the equivalent plastic strain at the onset of damage.
From this point, we adopt a linear damage law and the material stiffness is degraded through the damage variable D:

r ¼ ð1  DÞr ð4Þ
Z t _ pl
u
D¼ dt ð5Þ
0  pl
u f

where r is the effective (undamaged) stress, u  pl _ pl


f the effective plastic displacement at fracture, u ¼ 0 before damage initia-
_
 pl _ pl
tion and u ¼ e L during damage, L being a characteristic length associated with an integration point.
Following the work of Hillerborg et al. [57], the effective plastic displacement u  pl can be seen as the fracture work con-
jugate of the yield stress during damage. Its introduction in the definition of the damage variable D helps reducing the mesh
dependency once damage is initiated, creating a stress–displacement response of an element instead of a stress–strain one.

3. Procedure

3.1. Cross Tension tests

Resistance spot welds were produced out of 125  38 mm coupons of the TRIP780 sheets (bare), following the ISO 18278-
2 standard [1]. Homogeneous 1.0/1.0 mm and 1.5/1.5 mm as well as heterogeneous 1.0/2.0 mm configurations were
considered.
Welds were then subjected to the Cross Tension test (ISO 14272 [2]), with a crosshead speed of 15 mm/min. The Zwick
test machine used for this purpose was equipped with a 100 kN load cell and pneumatic machine grips maintaining the spec-
imen with a 300 bar pressure. The resulting load bearing capacity for the different weld configurations is depicted in Fig. 7a.
dF
Next, the derivatives dU of the Cross Tension loading curves were plotted as illustrated for example in Fig. 7b. Past the first
few thousand Newtons where small adjustments of the specimen in the machine grips are still on-going, the strong accidents
on these curves were shown by the authors in [7,9] to be correlated with the onset and propagation of the interfacial crack in
the weld nugget. This was demonstrated by a coupled X-ray microtomographic and metallographic analysis of interrupted
Cross Tension tests. Such curves will be used in the following in order to assess the onset of crack propagation at the interface
in the specimens.
Moreover, since the measured crosshead displacement contains the influence of the machine stiffness, the following pro-
cedure was adopted to correct it to the specimen deformation only.
Linear unidirectional spring elements were introduced in the model to transfer the load to the specimen (see Fig. 8a).
Springs stiffness was then fitted so as to reproduce the experimental Cross Tension loading curve with the overall crosshead
displacement. This was done once for the homogeneous t = 1.5 mm, d = 7.0 mm TRIP780 spot weld. Simulations with and
without springs could then be run for the different weld configurations. The numerical loading curves obtained in this
way are depicted in Fig. 8b in comparison to the experimental curve for the homogeneous t = 1.5 mm, d = 7.0 mm spot weld.
Next, a linear regression model was fitted to compute the correction in displacement to apply in order to retrieve the intrin-
sic specimen contribution to the displacement. Experimental loading curves could then be corrected as illustrated in Fig. 8b.

3.2. Identification of the model parameters

The different homogeneous and heterogeneous weld configurations listed in Table 1 (Section 2.1) were considered in this
study. As a first estimate of the welds behavior, simulations in the different configurations were run in order to estimate the
S. Dancette et al. / Engineering Fracture Mechanics 87 (2012) 48–61 55

Fig. 7. Cross Tension data for the different weld assemblies considered in this study.

Fig. 8. Removing the machine stiffness influence on the experimental crosshead displacement using spring elements.

J-integral at the notch tip, without ductile damage and failure. This was carried out with the Abaqus implicit solver (Newton–
Raphson integration). Next, the same simulations were run with the introduction of ductile damage and failure in the weld
zones, in competition with the cohesive zone model at the faying surface. The Abaqus explicit solver was then used in these
cases as it is best suited to deal with problems where the material stiffness may decrease due to damage and fracture. Usual
precautions were taken in order to ensure that the solutions remained quasi-static [50].
The cohesive zone parameters were first identified using the homogeneous t = 1.5 mm d = 7.0 mm assembly, fitting the
whole load–displacement curve within 2.5% error on Fmax. These welds were shown in [7,9] to fail by sudden semi-brittle
fracture at the faying surface (mixed dimple + cleavage fracture). This yielded the set of (r ^ ¼ 3500 MPa, d0 = 1.75 lm,
dc  d0 = 8.0 lm) for the cohesive parameters, which corresponds to a 17.1 kJ/m2 fracture toughness. The localized non-linear
failure process is thus represented by the bilinear traction–separation response. This is motivated by the geometry of such
local failure process zone, where both the opening (<10 lm) and the extent of the process zone (that amounts to a couple of
hundreds of microns) are significantly smaller than the other relevant dimensions of the problem. The cohesive element size
(around 0.045 mm) ensures that at least five cohesive elements are involved in the process zone. Such fracture toughness of
the weld nugget compares satisfactorily with the one estimated by Lacroix et al. [58] around 34 kJ/m2 for another (less
alloyed) automotive steel.
The parameters for the ductile damage and failure model were selected so as to reproduce the experimental ultimate frac-
ture strains measured for the different weld microstructures. They were obtained from the reduction in cross sectional area
of the corresponding tensile samples [7,9]. This procedure assumes implicitly that the level of stress triaxiality and strain rate
56 S. Dancette et al. / Engineering Fracture Mechanics 87 (2012) 48–61

Table 4
Experimental ultimate fracture strains for the TRIP780 microstructures and the corre-
sponding model parameters.

BM SCHAZ CGHAZ
exper
ef
0.6–0.8 0.4–0.6 0.2–0.3
pl
e 0.5 0.3 0.15
0
 pl
u (mm) 0.027 0.018 0.009
f

in the potential ductile failure zones of the weld is approximately the same as the one experienced in the necking region of
the tensile samples. This was indeed verified in [7].
Table 4 summarizes the experimental ultimate fracture strains obtained for the TRIP780 microstructures and the corre-
sponding ductile failure model parameters for the weld mesh presented in Section 2.1 and Fig. 3b. Note that the fusion zone
does not have intrinsic ductile fracture parameters in this approach. It is assumed that its failure will mainly occur through
the interfacial cracking of the cohesive zone instead, as observed experimentally for this steel [9].
Based on these identified parameters for the cohesive zone and ductile fracture, competition between semi-brittle frac-
ture at the faying surface and ductile failure could then be simulated for the different weld configurations.

4. Results and discussion

4.1. J-integral at the notch tip

Fig. 9a depicts the evolution of the J-integral at increasing loading, for the homogeneous t = 1.5 mm d = 5.0 mm and the
heterogeneous d = 5.0 mm welds. For comparison purpose, the evolution of J as calculated by the analytical expression of
Zhang [34] in the homogeneous case is also plotted. It is calculated under the hypothesis of linear elastic fracture mechanics
and writes:

117ð1  m2 Þc2 F 2
J¼ 2
ð6Þ
1024p2 d t3 E
where m is the Poisson’s ratio, E is the Young modulus and c is the spacing between the machine grips.
The accordance with the FE estimate obtained in this work is good up to a load of approximately 4000 N, where the influ-
ence of the non-linearities in the vicinity of the notch tip begins to be more sensible.
In the homogeneous case of Fig. 9a, it can be seen that J is constant along the crack tip from 0° (Sheet 1 direction) to 90°
(Sheet 2 direction), which is in accordance with former linear elastic studies of Radaj et al. [59] for example. The situation is
different in the heterogeneous case, where J in the direction of Sheet 2 (90°) is higher than in the direction of Sheet 1, past
about 2500 N. This is obviously related to the difference in sheet thickness and can be explained by the fact that plastic yield-
ing outside the weld occurs sooner in the thinner sheet than in the thicker one. Indeed, once extensive straining starts to
occur in the BM/SCHAZ of the thinner sheet due to the bending stress, the notch tip in the direction of Sheet 1 tends to
be relaxed, while it is still extensively loaded at 90° in the direction of the thicker sheet.
50

50

homo. 1.0mm d5.0mm


homo. 1.5mm d5.0mm
homo. 1.5mm d7.0mm
het. d5.0mm
40

40

het. d5.5mm
het. d6.5mm
J, [kJ / m2]

J, [kJ / m2]
30

30
20

20
10

10
0

0 2000 4000 6000 8000 0 2000 4000 6000 8000


F, [N] F, [N]

Fig. 9. Evaluation of the J-integral at the notch tip.


S. Dancette et al. / Engineering Fracture Mechanics 87 (2012) 48–61 57

Based on these informations on the distribution of the stress intensity along the notch tip, the evolution of J is plotted at
90° for the different weld configurations in Fig. 9b. The following observations can be made:

 For a given sheet thickness configuration and at a given load, J increases with decreasing weld diameter d. This is in accor-
dance with usual experimental observations regarding the increased sensitivity of small welds to (partial) interfacial
failure during the determination of the welding range for a given steel sheet [5–7].
 For a given weld diameter in homogeneous assemblies and at a given load, J increases while the sheet thickness decreases.
This can be related to the fact that a given load level results in a higher straining around the notch tip for a thin homo-
geneous assembly than for a thicker one.
 For a given weld diameter and at a given load, J is lower in the heterogeneous assembly than in the homogeneous one, as
discussed in Fig. 9a. This is related to the relaxation of the notch tip once plastic yielding starts to develop outside the
weld in the thinner sheet.

Finally, the horizontal line at J = 22.5 kJ/m2 in Fig. 9b corresponds to the critical level of J correlated with the onset of crack
propagation in the experimental specimens, as will be discussed in the next section. Re-examining Fig. 9b from the point of
view of such critical Jc, one can note that the corresponding critical force increases with the sheet thickness and the weld
diameter, as expected from common experimental observation of interfacial failures [5–7].

4.2. Competition between interfacial failure and button pullout

The model is run here in the different weld configurations with the competing failure criteria. The predictions of the onset
of cracking at the interface, of the governing damage mechanism leading to final fracture and of the failure type are com-
pared to the experiments. 3000
3000

2000
2000
dF / dU, [N/mm]

dF / dU, [N/mm]
1000
1000

0
0

exper., FIF exper., FIF


J=22.5 kJ/m2 J=22.5 kJ/m2
onset CZ onset CZ
-1000
-1000

failure CZ failure CZ

0 2000 4000 6000 8000 0 2000 4000 6000 8000


F, [N] F, [N]
3000
3000

exper., BP
J=22.5 kJ/m2
onset CZ
failure SCHAZ
2000
2000
dF / dU, [N/mm]

dF / dU, [N/mm]
1000
1000

0
0

exper., BP
J=22.5 kJ/m2
onset CZ
-1000
-1000

failure SCHAZ

0 2000 4000 6000 8000 0 2000 4000 6000 8000


F, [N] F, [N]

dF
Fig. 10. Onset of crack propagation and fracture as highlighted by the derivative of the Cross Tension loading curve dU
. FIF = Full Interfacial Failure.
BP = Button Pullout.
58 S. Dancette et al. / Engineering Fracture Mechanics 87 (2012) 48–61

4.2.1. Onset of crack propagation


dF
Fig. 10a–d depicts the evolution of the derivative of the experimental loading curve, dU , for different homogeneous and
heterogeneous assemblies. The accidents on these curves, past the first few thousand Newtons, highlight crack propagation
in the weld nugget as introduced in Section 3.1. The vertical lines superimposed in Fig. 10 correspond to the load levels ob-
tained respectively at the critical J = Jc = 22.5 kJ/m2, at the onset of interfacial cracking with the cohesive zone model and at
the maximum load in the simulation. The latter corresponds to either complete fracture of the cohesive zone or ductile frac-
ture in the weld HAZ. It can be seen from Fig. 10a–d that:

 The load at the critical Jc = 22.5 kJ/m2 is very close to the load at the onset of the cohesive zone fracture.
 Both of these two numerical estimates of the onset of interfacial fracture are closely related to the experimental onset of
the interfacial crack.
 The predicted load bearing capacities Fmax and failure types are in accordance to the experimental ones, as will be further
discussed in Section 4.2.2. That is FIF for the homogeneous assemblies and ductile BP for the heterogeneous assemblies
(Table 5).
 The progressive crack propagation in fits and starts over a load range of about 1500 N for the homogeneous t = 1.5 mm
d = 5.0 mm configuration is properly reproduced, and the quite sudden interfacial fracture of the homogeneous
t = 1.5 mm d = 7.0 mm too.

One can note here that the level of Jc = 22.5 kJ/m2 is higher than the decohesion energy of the cohesive zone identified
above (17.1 kJ/m2, Section 3.2). Following Hutchinson and Evans [60], such situation can be explained by the fact that the
peak strength of the interface (r
^ ¼ 3500 MPa) is larger than the yield strength of the surrounding bulk material, which in-
duces a plastic zone around the fracture process zone (represented by the cohesive zone). Consequently, the total (macro-
scopic) work of fracture exceeds the intrinsic work of separation in the cohesive zone.

4.2.2. Load bearing capacity and failure type


Figs. 11 and 12 illustrate the experimental and numerical loading curves for the homogeneous and the heterogeneous
cases respectively. Close agreement between the experimental and predicted load bearing capacity and failure type can
be observed, as already introduced in Fig. 10. In the case of the heterogeneous assemblies, the difference in initial stiffness

Table 5
Failure types and load bearing capacity of the different welds.

Thick. t Weld diam. d Failure type (exper.) Failure type (simu.) Fmax, [N] (exper.) Fmax, [N] (simu.)
Homogeneous
1.0/ 1.0 5.0 FIF FIF 4030 3860
1.5/ 1.5 5.0 FIF FIF 5350 5530
1.5/ 1.5 7.0 FIF FIF 8040 8240
Heterogeneous
1.0/ 2.0 5.0 FIF FIF 5460 5395
1.0/ 2.0 5.5 BP BP 6340 6200
1.0/ 2.0 6.5 BP BP 7450 7200
10000

exper. d5
exper. d7
o FEA d5
8000

x
x FEA d7 xxx
xxx
x
xxx
xx
xx
xx
6000

xxx
xoxx
F, [N]

xxxo
xxxoxoxoooo
xxxooo
xxxoooo
4000

xxxoooo
x
xxxoxoxoxoooo
o
x
xxoo
x
xoxoxoxoxoooo
xo
xo
xoxoxoxooo
xo
xo
xxxoo
2000

xoxoxooo
xo
xo
xx
x
oo
xxoo
xxoxooo
xo
xoxoxo
xoxo
x
o
x
oo
xo
o
0

0 2 4 6 8
U, [mm]

Fig. 11. Loading curves and failure type for the homogeneous 1.5 mm weld assembly. FIF = Full Interfacial Failure.
S. Dancette et al. / Engineering Fracture Mechanics 87 (2012) 48–61 59

10000
exper. d5.5
exper. d6.5
o FEA d5.5

8000
x FEA d6.5

xxx
xxx
6000
x xxxox x
x oo o
xxxooooo o
xoxoxoooo
F, [N]

xx o
xxxoooo
xxxooo
xxxoooo
4000

xxxxoxoxoooo
o
xoxoxoooo
xxo o
xoxoxoxooo
xo
xxo
x
xx
oxoxoxoxooo
o
o
xo
xoxoxoxoo
2000

xo
xo
xxoo
xo
x
o xoxoxoxoo
xo
x
o
x
xooo
xxoxoxo
xoxo
xoxo
xo
xo
xo
o xoo
xo
0

o
0 2 4 6 8
U, [mm]

Fig. 12. Loading curves and failure type for the heterogeneous 1.0/2.0 mm weld assembly. BP = Button Pullout.
5000
4000
3000
Fmax, [N]
2000
1000
0

homo. exper. homo. FEA het. exper. het. FEA

Fig. 13. Experimental and predicted load bearing capacity of homogeneous 1.0 mm VS heterogeneous 1.0/ 2.0 mm welds for a 5.0 mm weld diameter.
Welds fail by interfacial failure.

that can be seen in Fig. 12 between the experimental and predicted curves may be related to the occurrence of some sliding
in the machine grips, due to the difficulty of maintaining the heterogeneous sheets with equal pressure.
Further agreement was observed between the experimental and predicted load bearing capacity and failure types for the
homogeneous t = 1.0 mm, d = 5.0 mm and heterogeneous d = 5.0 mm assemblies, as illustrated in Fig. 13. It can be noted here
that the heterogeneous d = 5.0 weld fails by interfacial failure (FIF), while the transition from FIF to BP has already been
reached for the heterogeneous d = 5.5 mm and d = 6.5 mm welds (Fig. 12). However, careful examination of Fig. 10c and
of the cohesive zone for the heterogeneous d = 5.5 mm weld indicates that significant crack propagation already occurred
at the interface before the maximum load, despite the final fracture occurring in the SCHAZ by ductile BP, both in the model
and in the experiment.
Moreover, Fig. 13 illustrates that the heterogeneous 1.0/2.0 mm weld has a better load bearing capacity than the corre-
sponding homogeneous t = 1.0 mm one, for the same weld diameter (d = 5.0 mm) and interfacial failure type. This can be ex-
plained by the higher stress intensity at the notch tip in the homogeneous weld, as evaluated by the J-integral in Fig. 9b. The
critical Jc for the onset of crack propagation at the interface is then reached sooner, leading to the lower load bearing capacity
of the homogeneous weld.

5. Conclusion

A FE model for the mechanical behavior of AHSS resistance spot welds in Cross Tension was developed in this study. It is
able to capture the competition between semi-brittle fracture at the interface and ductile failure by button pullout at the
weld boundary. Ductile damage and fracture was considered in the different weld zones for this purpose, in addition to a
cohesive zone at the faying surface and an evaluation of the J-integral at the notch tip.
The model was run for different homogeneous and heterogeneous configurations of TRIP780 spot welds and could repro-
duce the appropriate failure type and the load bearing capacity of the different assemblies. The onset of crack propagation at
60 S. Dancette et al. / Engineering Fracture Mechanics 87 (2012) 48–61

the interface and the governing damage mechanism leading to final fracture could be properly predicted as a function of the
geometrical features of the welds.

Acknowledgement

Prof. J. Merlin is gratefully acknowledged for his comments on the paper.

References

[1] ISO standard, ISO 18278-2. Resistance welding – weldability – Part 2: alternative procedures for the assessment of sheet steels for spot welding; 2004.
[2] ISO standard, ISO 14272. Specimen dimensions and procedure for cross tension testing resistance spot and embossed projection welds; 2001.
[3] ISO standard, ISO 14273. Specimen dimensions and procedure for shear testing resistance spot, seam and embossed projection welds; 2002.
[4] Nishi T, Saito T, Yamada A, Takahashi Y. Evaluation of spot weldability of high-strength sheet steels for automobile use. Nippon Steel Tech Rep
1982;20:37–44.
[5] Williams N, Jones T. Spot weld size and fracture mode in low carbon mild steel. Metal Construct 1979;11:541–6.
[6] Gould J, Workman D. Fracture morphologies of resistance spot welds exhibiting hold time sensitivity behavior. In: AWS Detroit Section. Sheet metal
welding conference VIII; 1998.
[7] Dancette S. Comportement mécanique des soudures par points: mécanismes et stratégies de prédiction dans le cas des tôles en acier pour automobile.
Ph.D. thesis. Lyon: INSA-Lyon; 2009.
[8] Dancette S, Massardier V, Fabrègue D, Merlin J, Dupuy T, Bouzekri M. HAZ microstructures and local mechanical properties of high strength steels
resistance spot welds. ISIJ Int 2011;51:99–107.
[9] Dancette S, Fabrègue D, Massardier V, Merlin J, Dupuy T, Bouzekri M. Experimental and modeling investigation of the failure resistance of advanced
high strength steels spot welds. Engng Fract Mech 2011;78(10):2259–72.
[10] Zuniga S, Sheppard S. Resistance spot weld failure loads and modes in overload conditions. In: Fatigue and fracture mechanics, vol. 27. West
Conshohocken (PA); 1997. p. 469–89.
[11] Chao Y. Ultimate strength and failure mechanism of resistance spot weld subjected to tensile, shear, or combined tensile/shear loads. J Engng Mater
Technol 2003;125:125–32.
[12] Moore L, Jaffrey D, Kerr H. Failure mechanisms in rephosphorised sheet steel spot welds. Mater Sci Technol 1989;5:492–8.
[13] Nait-Oultit B, Lens A, Klocker H. Mechanical behaviour of advanced high-strength steel resistance spot welds. In: AWS Detroit Section. Sheet metal
welding conference XIII, Livonia, MI; 2008.
[14] Pouranvari M, Asgari H, Mosavizadch S, Marashi P, Goodarzi M. Effect of weld nugget size on overload failure mode of resistance spot welds. Sci
Technol Weld Join 2007;12:217–25.
[15] Hernandez VB, Kuntz M, Khan M, Zhou Y. Influence of microstructure and weld size on the mechanical behaviour of dissimilar AHSS resistance spot
welds. Sci Technol Weld. Join 2008;13:769–76.
[16] Khan M, Kuntz M, Zhou Y. Effects of weld microstructure on static and impact performance of resistance spot welded joints in advanced high strength
steels. Sci Technol Weld Join 2008;13:49–59.
[17] Marya M, Wang K, Hector L, Gayden X. Tensile-shear forces and fracture modes in single and multiple weld specimens in Dual-Phase steels. J Manuf Sci
E 2006;128:287–98.
[18] Marya M, Gayden X. Development of requirements for resistance spot welding Dual-Phase (DP600) steels. Part 1 – The causes of interfacial fracture.
Weld. J. 2005;84-s:172–82.
[19] Milititsky M, Pakalnins E, Jiang C, et al. On characteristics of DP600 resistance spot welds. Soc Autom Engineers 2003;01:244–51.
[20] Tao W, Li L, Chen Y, Wu L. Joint strength and failure mechanism of laser spot weld of mild steel sheets under lap shear loading. Sci Technol Weld Join
2008;13:754–9.
[21] Nikoosohbat F, Kheirandish S, Goodarzi M, Pouranvari M, Marashi S. Microstructure and failure behaviour of resistance spot welded DP980 dual phase
steel. Mater Sci Technol 2010;26:738–44.
[22] American Welding Society. AWS D8.1. Specification for automotive weld quality - resistance spot welding of steel; 2007.
[23] Ferrasse S, Verrier P, Meesemaecker F. Resistance spot weldability of high strength steels for use in car industry. Weld World 1998;41:177–95.
[24] Peterson W, Accorsi I, Coon T. Review of weld mechanical property specification requirements in AWS D8.1 (Proposed). In: AWS Detroit Section. Sheet
metal welding conference XII, Livonia, MI; 2006.
[25] Yamazaki K, Sato K, Tokunaga Y. Static and fatigue strength of spot welded joints in ultra-high- strength, cold rolled steel sheets. Weld Int
2000;14:533–41.
[26] Kabasawa M, Funakawa Y, Ogawa K, et al. Estimation of tensile shear strength of spot welded joint of steel sheets. Yosetsu Gakkai Ronbunshu
1996;14:754–61.
[27] Chao Y. Failure mode of spot welds: interfacial versus pullout. Sci Technol Weld Join 2003;8:133–7.
[28] Sakuma Y, Oikawa H. Factors to determine static strengths of spot weld for high strength steel sheets and developments of high strength steel sheets
with strong and stable welding characteristics. Nippon Steel Tech Rep 2003;88:33–8.
[29] Kuo M, Chiang J. Weldability study of resistance spot welds and minimum weld button size methodology development for DP steel. Int J Mater Manuf
2004;113:67–77.
[30] Satonaka S, Kaieda K, Okamoto S. Prediction of tensile-shear strength of spot welds based on fracture modes. Weld World 2004;48:39–45.
[31] Lin S, Pan J, Wu S, Tyan T, Wung P. Failure loads of spot welds under combined opening and shear static loading conditions. Int J Solids Struct
2002;39:19–39.
[32] Lin S, Pan J, Tyan T, Prasad P. A general failure criterion for spot welds under combined loading conditions. Int J Solids Struct 2003;40:5539–64.
[33] Rokhlin S, Adler L. Ultrasonic method for shear strength prediction of spot welds. J Appl Phys 1984;56:726–31.
[34] Zhang S. Stress intensities at spot welds. Int J Fracture 1997;88:167–85.
[35] Lee H, Choi J. Overload analysis and fatigue life prediction of spot-welded specimens using an effective J-integral. Mech Mater 2005;37:19–32.
[36] Kong X, Yang Q, Li B, Rothwell G, English R, Re X. Numerical study of strengths of spot-welded joints of steel. Mater Des 2008;29:1554–61.
[37] Yang Y, Babu S, Orth F, Peterson W. Integrated computational model to predict mechanical behaviour of spot weld. Sci Technol Weld Join
2008;13:232–9.
[38] Nielsen K. 3D modelling of plug failure in resistance spot welded shear-lab specimens (DP600-steel). Int J Fracture 2008;153:125–39.
[39] Wang J, Xia Y, Zhou Q, et al., Simulation of spot weld pullout by modeling failure around nugget. In: SAE. SAE 2006 world congress and exhibition,
Detroit, MI; 2006.
[40] Markiewicz E, Drazetic P. Exprimentation et simulation numrique locale/globale de la tenue mcanique des assemblages souds par points. Méc Ind
2003;4:17–27.
[41] Lamouroux E, Coutellier D, Doelle N, Kuemmerlen P. Detailed model of spot-welded joints to simulate the failure of car assemblies. Int J Interact Des
Manuf 2007;1:33–40.
[42] Nielsen K, Tvergaard V. Ductile shear failure or plug failure of spot welds modelled by modified Gurson model. Engng Fract Mech 2010;77:1031–47.
[43] Nielsen K. Predicting failure response of spot welded joints using recent extensions to the Gurson model. Comput Mater Sci 2010;48:71–82.
S. Dancette et al. / Engineering Fracture Mechanics 87 (2012) 48–61 61

[44] Niordson C. Analysis of steady-state ductile crack growth along a laser weld. Int J Fracture 2001;111:53–69.
[45] Cavalli M, Thouless M, Yang Q. Cohesive-zone modelling of the deformation and fracture of spot-welded joints. Fatigue Fract Engng Mater
2005;28:861–74.
[46] Zhou B, Thouless M, Ward S. Determining mode-I cohesive parameters for nugget fracture in ultrasonic spot welds. Int J Fracture 2005;136:309–26.
[47] Zhou B, Thouless M, Ward S. Predicting the failure of ultrasonic spot welds by pull-out from sheet metal. Int J Solids Struct. 2006;43:7482–500.
[48] Needleman A, Tvergaard V. A micromechanical analysis of the ductile–brittle transition at a weld. Engng Fract Mech 1999;62:317–38.
[49] Lemoine X. Behavior laws and their influences on numerical prediction. In: 10th ESAFORM conference on material forming, Zaragoza, Spain; 2007. p.
269.
[50] Hibbit, Karlsson, Sorensen. Abaqus users manual. Version 6.7; 2007.
[51] Anderson TL. Fracture mechanics – fundamentals and applications. 2nd ed. CRC Press; 1995.
[52] Parks DM. The virtual crack extension method for nonlinear material behavior. Comput Methods Appl Mech Engng 1977;12:353–64.
[53] Shih C, Moran B, Nakamura T. Energy release rate along a three-dimensional crack front in a thermo-mechanical field. Int J Fracture 1986;30:79–102.
[54] Needleman A. A continuum model for void nucleation by inclusion debonding. J Appl Mech 1987;54:525–31.
[55] Tvergaard V, Hutchinson J. The relation between crack growth resistance and fracture process parameters in elastic–plastic solids. J Mech Phys Solids
1992;40:1377–97.
[56] Camanho P, Davila C. Mixed-mode decohesion finite elements for the simulation of delamination in composite materials. NASA-TM-2002-211737;
2002.
[57] Hillerborg A, Modeer M, Petersson P. Analysis of crack formation and crack growth in concrete by means of fracture mechanics and finite elements.
Cem Concr Res 1976;6(6):773–81.
[58] Lacroix R, Monatte J, Lens A, Kermouche G, Bergheau J, Klocker H. Spot weld strength determination using the wedge test: in-situ observations and
coupled simulations. Appl Mech Mater 2010;24–25:299–304.
[59] Radaj D, Zhaoyun Z, Mohrmann W. Local stress parameters at the weld spot of various specimens. Engng Fract Mech 1990;37:933–51.
[60] Hutchinson J, Evans A. Mechanics of materials: top-down approaches to fracture. Acta Mater 2000;48:125–35.

Anda mungkin juga menyukai