You are on page 1of 11

See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/273388929

Prediction of fatigue life for multi-spot welded joints with different


arrangements using different multiaxial fatigue criteria

Article  in  Materials and Design · May 2015


DOI: 10.1016/j.matdes.2015.02.008

CITATIONS READS

17 4,242

4 authors:

Farrah Esmaeili Arash Rahmani


California State University, Northridge Buali Sina University
4 PUBLICATIONS   21 CITATIONS    10 PUBLICATIONS   158 CITATIONS   

SEE PROFILE SEE PROFILE

Sajjad barzegar Mohammadi Amir Afkar


University of Tabriz Iran University of Science and Technology
7 PUBLICATIONS   50 CITATIONS    42 PUBLICATIONS   262 CITATIONS   

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Biomechanical Modeling of a Seated Human Body Exposed to Vertical and Horizontal Vibrations Using Genetic Algorithms View project

All content following this page was uploaded by Sajjad barzegar Mohammadi on 10 January 2020.

The user has requested enhancement of the downloaded file.


Materials and Design 72 (2015) 21–30

Contents lists available at ScienceDirect

Materials and Design


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

Prediction of fatigue life for multi-spot welded joints with different


arrangements using different multiaxial fatigue criteria
F. Esmaeili a,⇑, A. Rahmani b, S. Barzegar c, A. Afkar d
a
Department of Mechanical Engineering, College of Engineering, Tabriz Branch, Islamic Azad University, Tabriz, Iran
b
Department of Engineering, Ajabshir Branch, Islamic Azad University, Ajabshir, Iran
c
Faculty of Mechanical Engineering, University of Tabriz, Tabriz, Iran
d
Faculty of Electrical, Mechanical and Construction Engineering, Department of Automotive Engineering, Standard Research Institute (SRI), Karaj, P.O. Box 31745-139, Iran

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

Article history: In this paper, the effects of weld arrangement on the fatigue behavior of the multi-spot welded joints
Received 24 September 2014 have been investigated via experimental and multiaxial fatigue analysis. To do so, three sets of the spot
Accepted 14 February 2015 welded specimens with different arrangements were prepared and fatigue tests were conducted under
Available online 21 February 2015
the various cyclic loads. Experimental tests revealed that the spot welded arrangement effect has a con-
siderable role in fatigue strength of multi-spot welded joints. A nonlinear finite element code was used to
Keywords: obtain the stress and strain distributions near the roots of the nuggets for the three kinds of the speci-
Multi-spot welded joints
mens. Fatigue lives of the specimens were predicted by means of six different multiaxial fatigue criteria
Finite element
Multiaxial fatigue
using the local stress and strain values obtained from the finite element simulations. It was found that the
SWT and Crossland criteria have the best accuracy for all types of the specimens among the applied
criteria.
Ó 2015 Elsevier Ltd. All rights reserved.

1. Introduction are proportional or non-proportional often occur at geometric dis-


continuities, e.g., notches or joints connections. The fatigue phe-
Resistance spot welding (RSW) is a jointing method, which is nomenon under these conditions, termed as multiaxial fatigue, is
widely used in the automotive industries in order to produce thin an important design consideration for a reliable operation and opti-
metal sheet components and structures due to its advantage in mization of many engineering components. Therefore, the multi-
welding efficiency and robotization. Among a variety design axial fatigue analysis becomes an essential tool for approximating
criteria such as yielding, buckling, fatigue, creep and corrosion, the fatigue lives of these components.
the fatigue phenomenon is the most important mode of the frac- Despite of existence of a large number of multiaxial fatigue cri-
ture in the spot-welded joints in vehicle bodies. Therefore, the dif- teria in the literature, engineers are often faced with some prob-
ferent models have been presented to estimate the fatigue lives of lems in the application of these criteria in the engineering
spot welded joints. Some parameters which affect the fatigue life of design. The application of these criteria for complex multiaxial
the spot welded joints can be classified into the following cate- loadings is one of the most important problems. Most of the mul-
gories; stress concentration factors, residual stresses, material tiaxial fatigue criteria have been discussed for several materials
characteristics, welding process parameters (welding time sched- and loading conditions by a number of authors such as You and
ule, welding current and electrode clamping pressure), weld qual- Lee [7] and Wang and Yao [8].
ity (weld nugget size, amount of sheet spacing), arrangement The multiaxial fatigue criteria for estimating the fatigue life of
effect, etc. Among the mentioned parameters, the spot-weld mechanical components may be classified into three main groups,
arrangement effect plays an important role on the fatigue strength including stress based criteria, strain based criteria and energy
of spot-welded joints [1–6]. based approaches [9]. As for the stress criteria proposed by Susmel
Many engineering components such as automotive bodies and and Lazzarin [10] and Crossland [11] and so on, are based only on
aircraft structures are subjected to complex states of stress. The the stress and are suitable for the high cycle fatigue when the
complex stress states in which the two or three principal stresses deformation is elastic or the plastic strain is small. The strain crite-
ria such as the Brown–Miller model [12], Fatemi–Socie [13], Li-
⇑ Corresponding author. Tel.: +98 411 3392492. Zhang [14] and Wang–Brown model [15], are appropriate for cases
E-mail address: f.esmaeili@iaut.ac.ir (F. Esmaeili). in which there is significant plasticity. The energy based multiaxial

http://dx.doi.org/10.1016/j.matdes.2015.02.008
0261-3069/Ó 2015 Elsevier Ltd. All rights reserved.
22 F. Esmaeili et al. / Materials and Design 72 (2015) 21–30

criteria such as the Smith–Watson–Topper model [16], Glinka et al. tests in accordance with the ASTM: E8M. The true stress–strain
[17] and Varani-Farahani [18] include both the stress and strain plot of the used material has been illustrated in Fig. 1. Tensile–s-
terms. From another point of view, the multiaxial fatigue criteria hear specimens were used for both of the static strength testing
can be categorized as those criteria that use the critical plane and fatigue testing of spot-welds.
concept and those do not use this concept. In the multiaxial fatigue Three kinds of spot-weld tensile–shear specimens with three
criteria based on the critical plane, initially a material plane on different arrangements including one-row four-spot parallel to
which a combination of some stress components has the maxi- the loading direction, one-row four-spot perpendicular to the load-
mum value has been determined. Then, the fatigue parameters as ing direction and two-row four-spot weld specimens were pro-
combinations of the maximum shear strain or stress and normal duced for this investigation. In addition, configurations and
strain or stress on the critical plane have been calculated. The cri- dimensions of the test specimens have been illustrated in Fig. 2.
teria will be checked then on this plane. Like to the classical mod- To do so, all resistance spot-welding process parameters such as
els, the critical plane models can be stress-, strain- or energy- electrode clamping force, welding time and electric current were
based. the same. A 5 kA current electric was used to generate the spot-
The energy criteria have been used in combination with the cri- welds and the spot-weld nugget diameters was 4 ± 0.1 mm.
tical plane approach, as proposed by Liu [19]. In the last years, For tensile–shear specimens, the stress range Ds is defined by
some of the international conferences have been allocated to the load range divided by nugget area. In the fatigue test, the load level
multiaxial fatigue. Some of the results of these conferences have is defined by the maximum load (Fmax) divided by the static frac-
been published as ASTM Special Technical Publications. Also, the ture load as follows:
results of a multiaxial fatigue testing and analysis program by
means of SAE Fatigue Design and Evaluation Committee were pub- F max smax Ds
¼ ¼ ð1Þ
lished as SAE [9]. FF sF ð1  RÞsF
Farahani et al. [20] in a study assessed the fatigue life of un-
where Fmax is the static fracture load, sF is the static fracture stress,
notched and notched components with different metallic materials
and R is the stress ratio (ratio of minimum stress to maximum stress
by means of some energy based critical plane fatigue criteria. In a
in one cycle of loading in a fatigue test) for the spot-welded joints.
separate investigation by Wang and Yao [8], the capabilities of sev-
In this study, the fatigue tests have been conducted with stress
eral multiaxial fatigue criteria have been evaluated for six different
ratio of 0.1 and frequency of 10 Hz using servo-hydraulic 250 kN
materials under various loading conditions. In previous studies,
Zwick/Roell fatigue testing machine. In each case, several fatigue
loading conditions were in-phase or standard out-of-phase ten-
tests were performed with different maximum applied longitudi-
sion–torsion where the local stress and strain could be easily calcu-
nal loads. The resulting average life from the constant amplitude
lated for the specimens. In addition, these criteria were validated
fatigue tests were plotted in accordance with the ASTM: E468-
for standard specimens like dog bone cylinder or thin walled tubu-
11, and displayed in Fig. 3.
lar. However, due to the complexity of this challenging problem
As it can be seen in Fig. 3, the B-type specimens have higher
and its practical application, additional studies are still needed to
fatigue lives comparing to the two other types of specimens while
evaluate the accuracy and reliability of the multiaxial fatigue crite-
of A-Type specimen have the lower fatigue lives. Furthermore, as it
ria in design, life approximation, and failure assessment especially
can be seen in this figure, the difference between the fatigue lives
for practical specimens.
of spot-weld specimens in the low cycle regime is greater than
In the current study, the spot-weld arrangement effects on the
those in the high cycle regime. Correspondingly, it can be detected
fatigue behavior of the multi-spot welded joints have been inves-
that, in the low cycle fatigue regime, the fatigue lives of B-Type
tigated using the experimental and multiaxial fatigue analysis. To
spot weld specimens were the longest, while those of A-Type spe-
do so, three sets of the four-spot welded joint specimens with dif-
cimen were the shortest.
ferent arrangements were manufactured and then fatigue tests
Figs. 4–6 show the fatigue fracture of all kinds of the specimens
were conducted under various cyclic longitudinal load levels. A
under different load levels. With attention to the fatigue fracture of
nonlinear finite element ANSYS code was used to obtain stress,
the mentioned spot-welded joints, it can be concluded that the
and strain distribution near the roots of the nuggets due to longi-
fatigue fracture mechanisms of the specimens be influenced not
tudinal applied loads. Fatigue lives of specimens were estimated
only by the factors such as sheet thickness and nugget diameter,
with six different multiaxial fatigue criteria using the local stress,
but also by the other important parameters such as the spot-welds
and strain distributions obtained from the finite element analysis.
arrangement and applied load levels.
As it can be revealed from these figures, for all kinds of speci-
2. Experimental procedures mens, under high load levels, nugget fracture across its interface
has occurred, and under low load levels, sheet fracture through
The details of the experimental tests were explained compre- the sheet thickness has happened. At intermediate load levels the
hensively in Ref. [6] for all types of the spot welded specimens with
different arrangements. However, in order to provide an overview,
600
a brief discussion is given here. The materials, which have been
used in this investigation, are the sheets with 1 mm thickness 500
Stress (MPa)

and 40 mm width, manufactured from ST37 steel whose mechan- 400


ical properties have been summarized in Table 1. The mechanical
300
properties of this material have been obtained from tension (static)
200

Table 1 100
Mechanical properties of ST37 steel. 0
0 0.05 0.1 0.15 0.2 0.25
Young’s Yield stress Tensile Poisson’s Elongation
modulus (GPa) (MPa) strength (MPa) ratio (%) Strain (mm/mm)
210 0.3 530 0.33 22
Fig. 1. True stress–strain curve of ST37 steel.
F. Esmaeili et al. / Materials and Design 72 (2015) 21–30 23

Fig. 2. Test specimens’ configuration and dimensions (in mm).

250
A type Specimens
Maximum Applied Stress

200 C Type Specimens


B Type Specimens
150
(MPa)

100

50

0
1.00E+02 1.00E+03 1.00E+04 1.00E+05 1.00E+06 1.00E+07
Number of cycles to failure (Nf)

Fig. 3. S–N curve for three different kinds of RSW joints.

situation is a little bit different, so that, for B-Type spot-weld speci- applied boundary conditions are illustrated in Figs. 7–9. As the fig-
mens button/plug nugget fracture has occurred, while this kind of ures shows, due to symmetry, only a half section of the spot-weld-
fracture cannot be seen in the C-Type spot-weld samples and for ed joint was modeled. The end surface of the upper sheet was
the A-Type specimens it has been observed in a very small scale constrained in all degree of freedom to prevent rigid body motion.
at the first spot-weld. Additionally, it was found that in the all To represent the steel alloy ST37 stress–strain behavior, an elastic–
types of specimens, cracks usually initiate from the edge of the plastic multi-linear kinematic hardening material model with Von
nugget roots due to stress concentration and propagated in a path Mises criterion was used.
perpendicular to the applied longitudinal load. Therefore, the In the finite element simulation, the solution was implemented
dominant failure mechanism in these types of specimens is due after the application of the value of maximum force in each cyclic
to tensile mode cracking. loading according to the experimental tests to the end of the lower
sheet in the model as a longitudinal static load. The longitudinal
stress distributions, for the three types of specimens under a typi-
3. Finite element analysis cal maximum remote stress have been illustrated in Fig. 10. As it
can be seen from this figure, the maximum longitudinal stress val-
As mentioned previously, in order to estimate the fatigue ues occur near the nuggets roots which are close to the loading
strength of RSW joints, the distribution of stress and strain should sides for all types of specimens due to high stress concentration.
be obtained near the nugget root, and finally, the number of cycles Additionally, it can be observed from these figures that the maxi-
to failure can be calculated from selected multiaxial fatigue crite- mum amount of longitudinal stress, rx, in A-Type specimens, was
ria. To do so, a 3D finite element analysis was performed by means higher in comparison to the other two types of specimens. In the
of ANSYS 11 finite element code in order to obtain the stress and case of the A-Type specimens a large amount of applied longitudi-
strain distribution in joint sheets. A detailed view of the meshed nal remote load is transmitted by the spot weld which is close to
models of the three different spot-welded specimens, and the the loading side for A-Type specimens. However, in the case of
24 F. Esmaeili et al. / Materials and Design 72 (2015) 21–30

Fig. 4. Different fracture mechanisms of A-Type spot weld specimens at different


load levels; (A) Sheet fracture through the thickness at low load level; (B) Button
fracture in a very small scale in the first nugget at intermediate load level; (C) Fig. 5. Different fracture mechanisms of B-Type spot weld specimens at different
Nugget fracture across its interface at high load level. load levels; (A) Sheet fracture through the thickness at low load level; (B) Button
fracture at intermediate load level; (C) Nugget fracture across its interface at high
load level.
the B-Type specimens the applied load is divided between the four
spot welds. As a result, the local stress at the edge of the nugget
roots is lower, in comparison to the A-Type and B-Type specimens.
Then, to simulate the unloading at the second half cycle of the cyc-
lic loading, the value of minimum force according to experimental
tests in each cyclic loading was applied to the end of the sheet. This
longitudinal loading and unloading, was performed for two cycles
to consider the variation of stress and strain distributions during
cyclic loading. The obtained results from the non-linear finite ele-
ment simulations were then used for calculating the required para-
meters of the selected multiaxial fatigue criteria. The required
numerical data such as principal stresses and strains were record-
ed from the nodes of the region around the nugget, which have the
maximum tendency for crack initiation and propagation. The stress
history graphs for the typical critical node, near the nugget of RSW
joints are shown in Fig. 11 for all kinds of joints with different
arrangement under the application of maximum remote stress
equals to Smax = 125 MPa. As the figure shows the state of stresses
near the spot-weld nugget is multiaxial.

4. Multiaxial fatigue criteria


Fig. 6. Different fracture mechanisms of C-Type spot weld specimens at different
load levels; (A) Sheet fracture through the thickness at low load level; (B) Nugget
In this investigation, to predict the fatigue life of multi spot- fracture across its interface at intermediate and high load level.
welded joints with different arrangement, six multiaxial fatigue
criteria, i.e., SWT, Glinka, KBM, FS, Crossland and VF were consid-
ered. These criteria are discussed briefly in the following para- 4.1. Smith–Watson–Topper (SWT)
graphs. Table 2 lists the corresponding material properties used
in these criteria. It is important to note that, these values have been Smith et al. proposed an experimental damage parameter
obtained by means of approximated methods available in refer- which is evaluated at the plane of maximum normal strain. The
ences such as [21,22]. SWT multiaxial fatigue model is expressed as
F. Esmaeili et al. / Materials and Design 72 (2015) 21–30 25

Fig. 7. FE model of A-Type RSW specimens.

Fig. 8. FE model of B-Type RSW specimens.

0 2
De1 ðr fÞ
The critical plane of this parameter is the plane of maximum shear
rmax
n ¼ ð2N f Þ2b þ r0f e0f ð2Nf Þbþc ð2Þ strain, where Dcmax the maximum is shear strain range and Den is
2 E
the corresponding normal strain range at the critical plane. Also,
In the above equation rmax
n and De1 are the maximum normal stress Sk is a material dependent constant, which was assumed to be
and the maximum principal strain range at the critical plane. In this 0.3. These values can be determined with principal stresses and
 De1

paper the maximum value of the product, rmax n 2
in any node has strains obtained from the finite element analysis, and therefore,
been used. To do so, rmax
n (maximum normal stress) and De1 (first using the Eqs. (4) and (5) for the critical nodes near the nugget
principal strain range) have been calculated during cyclic loading root. In Eqs. (4) and (5), e1 and e2, are the first and third principal
in any node of the FE models and consequently the maximum strains respectively. In addition, h1 and h2 in these equations are
amount of the product of these two parameters has been employed indicating loading and unloading of a cycle. These parameters
in the Eq. (2) to calculate the estimated fatigue life. were determined for every node and the maximum value of the
left hand side of Eq. (2) was used to predict the fatigue life of
4.2. Kandil, Brown and Miller (KBM) the specimens.

KBM multiaxial theory is based on a physical interpretation of Dc e1  e3  e  e 


1 3
mechanisms of fatigue crack growth. The general form of KBM’s ¼  ð4Þ
2 2 h1 2 h2
parameter is expressed as:
Den e1 þ e3  e þ e 
1 3
Dcmax rf 0
2
¼
2

2
ð5Þ
þ Sk Den ¼ ð2Nf Þb þ e0f ð2Nf Þc ð3Þ h1 h2
2 E
26 F. Esmaeili et al. / Materials and Design 72 (2015) 21–30

Fig. 9. FE model of C-Type RSW specimens.

4.3. Glinka 4.5. Crossland

Glinka et al. [17] proposed a fatigue parameter by using the The Crossland [11] criterion is a stress based multiaxial fatigue
summation of elastic and plastic energy densities on the critical criterion which uses the second invariant of deviatoric stress ten-
shear plane. sor and maximum hydrostatic stress in its equation.

2
qffiffiffiffiffiffiffi
Dc Ds Den Drn ðr0f Þ Ee0f J 2;a þ krH;max ¼ r0f ð2N f Þb ð10Þ
: þ : ¼ ð2Nf Þ2b þ ð2Nf Þbþc ð6Þ
2 2 2 2 2E 2 qffiffiffiffiffiffiffi 1 h i 1=2
J 2;a ¼ pffiffiffi ðDr1  Dr2 Þ2 þ ðDr2  Dr3 Þ2 þ ðDr1  Dr3 Þ2 ð11Þ
where Dc; Ds; Den and Drn are the range of shear strain, shear stress 2 6
range, normal strain range and normal stress range on the critical
In the above equations, J2,a and rH,max are the amplitude of second
plane respectively. These values can be determined with principal
invariant of deviatoric stress tensor and the maximum value of the
stresses and strains obtained from the finite element analysis, and
hydrostatic stress respectively. Also, k is a material dependent con-
therefore, using the Eqs. (4), (5), (7) and (8) for nodes around the
stant which assumed to be unit for this research.
bolt hole. The maximum value of the left hand side of Eq. (6) among
the calculated values of the nodes can be used to estimate the fati-
gue life of specimens. 4.6. Jahed–Varvani method (JV)
r  r  r  r 
1 3 1 3
Ds ¼  ð7Þ The Jahed–Varvani [22] energy-based model, that considers the
2 h1 2 h2 sum of plastic and positive elastic strain energy densities as a mea-
r þ r  r þ r 
1 3 1 3 sure of fatigue damage, was also used for fatigue life predictions.
Drn ¼  ð8Þ
2 h1 2 h2 This criterion defines the critical plane as the plane where maxi-
mum amount of shear strain range occurs and uses normal stress
where r1, and r2 are the biggest and smallest principal stress val-
and strain perpendicular to the critical plane and introduces it as
ues respectively.
fatigue damage parameter. The maximum damage plane on which
the fatigue damage parameter assumes its maximum value was
4.4. Fatemi and Socie (FS) defined as the critical plane.
" #
0 
Proposed criterion by Fatemi and Socie is a strain-based para- 1 ð1 þ rm n =rf Þ Dc Ds
ðDrn  Den Þ þ :
meter that considers the plane of maximum shear strain amplitude r0f e 0 s0f c0f 2 2
max
as a critical plane. Fatemi–Socie fatigue model is expressed as
r0f b 0 c
s0f b0 0 c0
  ¼ ð2Nf Þ þ ef ð2N f Þ þ ð2Nf Þ þ cf ð2N f Þ ð12Þ
Dcmax rn;max s0f 0 0 E G
1þk ¼ ð2N f Þb þ c0f ð2Nf Þc ð9Þ
2 ry 2G where the normal mean stress rmn acting on the critical plane is
given as follows:
where Dcmax is the maximum shear strain and rn,max is the maxi-
mum normal on the plane that Dcmax occurs. Also k = 0.3 is the rmn ¼ 1=2ðrmax þ rmin
n n Þ ð13Þ
Fatemi–Socie constant and ry is the tensile yield strength. The coef-
ficients s0f ; c0f ; and G are the torsional fatigue strength and ductility In Eq. (12) Dc and Den are calculated from Eqs. (4) and (5), and rn is
coefficients, and the shear modulus respectively. The exponents b0 the normal stress that was assumed equal to first principal stress.
and c0 are the torsional fatigue strength and ductility exponents, Eq. (13) was solved for data of the critical node in which the left
respectively. hand side of the equation has achieved its maximum value.
F. Esmaeili et al. / Materials and Design 72 (2015) 21–30 27

Fig. 10. Distribution of longitudinal normal stress rx in MPa under the application of maximum remote stress equals to Smax = 125 MPa for; (A) A-Type, (B) B-Type and (C) C-
Type specimens.
28 F. Esmaeili et al. / Materials and Design 72 (2015) 21–30

500 The solid line indicates that the estimated results are equal with
S1-1st Principal Stress
experimental tests results. Also, two other bounds are also plotted
400 S2-2st Principal Stress
with dashed lines. The inner bound is in accordance with the life
S3-3st Principal Stress
300 factor of 2. The outer bound is in accordance with the life factor
Stress (MPa)

200 of 10. Figs. 12–14 show that in all specimens, most criteria under-
estimate the fatigue lives. It should be noted that, the multiaxial
100 fatigue criteria are generally based on the fatigue crack initiation
0 stage and they do not include the fatigue crack growth stage. In
other words, the multiaxial criteria use only the stress and strains
-100
in critical region (at the edge of the nugget root) and they do not
First Cycle Second Cycle
-200 consider fatigue crack growth. Additionally, it should be noted that,
(A) the differences among the experimental results and fatigue lives
obtained from the multiaxial criteria can be related to the fact that
500 in the numerical simulation some of the parameters such as resi-
S1-1st Principal Stress
dual stress, imperfections, and also the effects of varying mechan-
400 S2-2st Principal Stress
ical properties on the heat affected zone which have not been
Stress (MPa)

S3-3st Principal Stress


300 considered in this study.
200 It can be seen from Fig. 12 that for A-Type specimens, the Cross-
land and SWT criteria agree with the experimental tests results
100 very well. As it can be seen from this figure, more of the points esti-
0 mated via SWT and Crossland criteria fall in the range of life factor
2. Also, it can be seen that the estimation made by Glinka criterion
-100
mostly fall near the range of life factor 2. However, estimated fati-
First Cycle Second Cycle
-200 gue lives by means of KBM criterion were underestimated. Addi-
(B) tionally, the fatigue life of A-Type specimens underestimated in
low cycle region and overestimated in high cycle region by VF
500 and FS criteria.
S1-1st Principal
Stress
400 S2-2st Principal The comparison between the estimated fatigue lives using mul-
Stress
tiaxial fatigue criteria and experimental fatigue lives for B-Type
Stress (MPa)

S3-3st Principal
300 Stress
specimens are shown in Fig. 13. In these specimens, the estimated
200 results using the SWT criterion fall into the range of life factor 2.
Therefore, this criterion estimates the fatigue life very accurately
100
for these specimens. Also, the results of Crossland criterion for
0 more of the points are within the range of life factor 2 so it esti-
-100
mates fatigue life fairly accurate. But, the other criteria do not pre-
First Cycle Second Cycle dict the fatigue life as accurate as SWT and Crossland criteria for
-200 this specimen.
(C) The estimated fatigue lives using multiaxial fatigue criteria ver-
sus experimental fatigue lives for C-Type specimens are illustrated
Fig. 11. The history of principal stresses for the typical critical node near the nugget
in Fig. 14. According to the Fig. 14, the estimated results using the
root under the application of maximum remote stress equals to Smax = 125 MPa for;
(A) A-Type, (B) B-Type and (C) C-Type specimens. SWT criterion fall into the range of life factor 2. Furthermore, in
these specimens, the estimated results using the Crossland criteria
except for two points fall into the range of life factor 2. Therefore,
5. Results and discussion these criteria are able to estimate fatigue lives with sufficient level
of accuracy for these specimens. Additionally the results of Glinka
A series of experimental tests were conducted to investigate the and KBM criterion for more of the points are within the range of
arrangement effects in the multi-spot welded joints on the fatigue life factor 2 so it estimates fatigue life fairly accurate. As the figure
strength of RSW joints subjected to longitudinal load. Three kinds shows, the other criteria underestimate the fatigue lives.
of spot welded tensile–shear specimens with three different In accordance with Figs. 12–14, most of the experimental
arrangements were produced for this investigation. It was shown results are located below the solid line where predicted and
that the arrangement effect has a considerable role in fatigue experimental cycles correspond. This means that experimental
strength of multi-spot welded joints. In addition, finite element fatigue lives are less than predicted ones. It must be mentioned
analysis was used to obtain the stress distribution in the RSW that, from design point of view, the KBM, and Crossland multiaxial
joints due to longitudinal applied loads in order to estimate the fatigue criteria are better than the other criteria, because they do
fatigue lives based on selected multiaxial fatigue criteria and final- not nearly overestimate the life for all types of the specimens.
ly to compare with the experimentally obtained fatigue lives. In order to compare more quantitatively the results of multi-
The comparison among the estimated fatigue lives and the axial fatigue criteria with experimental test results an error index
experimental lives has been illustrated in Figs. 12–14 for multi- has been used. The error index and mean absolute values of errors
spot welded specimens with different arrangements, respectively. have been used as [8]:

Table 2
ST37 mechanical properties used in multiaxial fatigue criteria.

ry t E r0f b e0f c c0f s0f b0 c0

298 MPa 0.3 210 GPa 795 MPa 0.11 0.45 0.59 0.779 458 MPa 0.11 0.59
F. Esmaeili et al. / Materials and Design 72 (2015) 21–30 29

1.E+07

1.E+06

Esmated Fague Life


1.E+05

1.E+04 SWT
Crossland
KBM
1.E+03 FS
Glinka
VF
1.E+02
1.E+02 1.E+03 1.E+04 1.E+05 1.E+06 1.E+07
Experimental Fague Life

Fig. 12. Comparison among the estimated and experimental fatigue lives for A-Type specimens.

1.E+07

1.E+06
Esmated Fague Life

1.E+05

SWT
1.E+04
Crossland
KBM
1.E+03 FS
Glinka
VF
1.E+02
1.E+02 1.E+03 1.E+04 1.E+05 1.E+06 1.E+07
Experimental Fague Life

Fig. 13. Comparison among the estimated and experimental fatigue lives for B-Type specimens.

1.E+07

1.E+06
Esmated Fague Life

1.E+05

SWT
1.E+04
Crossland
KBM
1.E+03 FS
Glinka
VF
1.E+02
1.E+02 1.E+03 1.E+04 1.E+05 1.E+06 1.E+07
Experimental Fague Life

Fig. 14. Comparison among the estimated and experimental fatigue lives for C-Type specimens.

 
Nexp X Table 3
E ¼ log ð14Þ The average absolute values of errors for selected criteria (in %).
N est
1 Xn SWT Glinka KBM FS Crossland VF
E¼ jEi j ð15Þ
n i¼1 A-Type specimens 6 38 18 30 24 60
B-Type specimens 13 36 26 32 25 59
where Nexp is the experimental life and Nest is the estimated life. C-Type specimens 7 34 29 36 26 60
The average absolute values of errors in different employed
multiaxial fatigue criteria are presented in Table 3.
The error index shows that the SWT criterion has the best accu- failure occurs primarily due to tensile mode cracking [16]. In these
racy for all types of the specimens. This is related to the fact that, materials, cracks nucleate in shear but early life is controlled by
the SWT fatigue can be successfully used in the analysis of both crack grow on planes perpendicular to the maximum principal
proportionally and non-proportionally loaded components that stress and strain. Also, it can be seen from the Table 3 that, the
30 F. Esmaeili et al. / Materials and Design 72 (2015) 21–30

VF’s criterion shows more inaccuracy among the employed multi- [5] S. Hassanifard, M. Zehsaz, K. Tohgo, The effects of electrode force on the
mechanical behaviour of resistance spot-welded 5083-O aluminium alloy
axial fatigue criteria for RSW joints considered in this investigation.
joints, Strain 47 (2011) e196–e204.
[6] S. Hassanifard, M. Zehsaz, F. Esmaeili, Spot weld arrangement effects on the
6. Conclusions fatigue behavior of multi-spot welded joints, J. Mech. Sci. Technol. 25 (2011)
647–653.
[7] B.-R. You, S.-B. Lee, A critical review on multiaxial fatigue assessments of
In this paper, the arrangement effects in the multi-spot welded metals, Int. J. Fatigue 18 (1996) 235–244.
joints on the fatigue behavior of the RSW joints have been investi- [8] Y.-Y. Wang, W.-X. Yao, Evaluation and comparison of several multiaxial fatigue
criteria, Int. J. Fatigue 26 (2004) 17–25.
gated via experimental and multiaxial fatigue analyses. Based on [9] N. Shamsaei, A. Fatemi, Effect of hardness on multiaxial fatigue behaviour and
the obtained results, the following conclusions can be drawn from some simple approximations for steels, Fatigue Fract. Eng. Mater. Struct. 32
the experimental and multiaxial fatigue analyses: (2009) 631–646.
[10] L. Susmel, P. Lazzarin, A bi-parametric Wöhler curve for high cycle multiaxial
fatigue assessment, Fatigue Fract. Eng. Mater. Struct. 25 (2002) 63–78.
 It was revealed that the spot welded arrangement effect has a [11] B. Crossland, Effect of large hydrostatic pressures on the torsional fatigue
considerable role in fatigue strength of multi-spot welded strength of an alloy steel, Proc. Int. Conf. Fatigue of Metals (1956) 138–149.
[12] M.W. Brown, K.J. Miller, A theory for fatigue failure under multiaxial stress–
joints.
strain conditions, Proc. Inst. Mech. Eng. 187 (1973) 745–755.
 The results based on the employed multiaxial fatigue criteria [13] A. Fatemi, D.F. Socie, A critical plane approach to multiaxial fatigue damage
have been compared with those obtained from experimental including out-of-phase loading, Fatigue Fract. Eng. Mater. Struct. 11 (1988)
fatigue test. The results showed that there is a relatively good 149–165.
[14] J. Li, Z-p Zhang, Q. Sun, C-w Li, Y-j Qiao, A new multiaxial fatigue damage
agreement among the employed multiaxial fatigue criteria pre- model for various metallic materials under the combination of tension and
dictions and experimental results. torsion loadings, Int. J. Fatigue 31 (2009) 776–781.
 Among the applied criteria, the SWT criterion has the best accu- [15] C.H. Wang, M.W. Brown, A path-independent parameter for fatigue under
proportional and non-proportional loading, Fatigue Fract. Eng. Mater. Struct.
racy for all types of the specimens. 16 (1993) 1285–1297.
 The VF’s criterion shows rather inaccurate prediction of fatigue [16] R.N. Smith, P. Watson, T.H. Topper, A stress strain function for the fatigue of
life among the employed multiaxial fatigue criteria for fatigue metals, J. Mater. JMLSA 5 (1970) 767–778.
[17] G. Glinka, G. Shen, A. Plumtree, A multiaxial fatigue strain energy density
strength of multi-spot welded joints studied in this parameter related to the critical fracture plane, Fatigue Fract. Eng. Mater.
investigation. Struct. 18 (1995) 37–46.
[18] A. Varvani-Farahani, A new energy-critical plane parameter for fatigue life
assessment of various metallic materials subjected to in-phase and out-of-
phase multiaxial fatigue loading conditions, Int. J. Fatigue 22 (2000) 295–305.
References [19] K.C. Liu, A method based on virtual strain-energy parameters for multiaxial
fatigue, in: D.L. McDowell, R. Ellis (Eds.), Advances in multiaxial fatigue,
[1] X. Long, S.K. Khanna, Fatigue properties and failure characterization of spot American Society for Testing and Materials, Philadelphia, 1993, pp. 67–84.
welded high strength steel sheet, Int. J. Fatigue 29 (2007) 879–886. [20] A. Varvani-Farahani, T. Kodric, A. Ghahramani, A method of fatigue life
[2] S.K. Khanna, X. Long, Residual stresses in resistance spot welded steel joints, prediction in notched and un-notched components, J. Mater. Process. Technol.
Sci. Technol. Weld. Joining 13 (2008) 278–288. 169 (2005) 94–102.
[3] Y.P. Yang, J. Gould, W. Peterson, F. Orth, P. Zelenak, W. Al-Fakir, Development [21] D. McClaflin, A. Fatemi, Torsional deformation and fatigue of hardened steel
of spot weld failure parameters for full vehicle crash modelling, Sci. Technol. including mean stress and stress gradient effects, Int. J. Fatigue 26 (2004) 773–
Weld. Joining 18 (2013) 222–231. 784.
[4] H. Adib, J. Gilgert, G. Pluvinage, Fatigue life duration prediction for welded [22] H. Jahed, A. Varvani-Farahani, Upper and lower fatigue life limits model using
spots by volumetric method, Int. J. Fatigue 26 (2004) 81–94. energy-based fatigue properties, Int. J. Fatigue 28 (2006) 467–473.

View publication stats

You might also like