Engineering Fracture Mechanics 73 (2006) 2229–2249

[Link]/locate/engfracmech

Modeling of failure near spot welds in lap-shear specimens

based on a plane stress rigid inclusion analysis
P.-C. Lin, S.-H. Lin, J. Pan *

Mechanical Engineering, The University of Michigan, Ann Arbor, MI 48109-2125, USA

Received 29 April 2005; received in revised form 6 March 2006; accepted 14 March 2006
Available online 12 June 2006

Abstract

In this paper, the failure mechanism of resistance spot welds in dual-phase steel lap-shear specimens is investigated
based on experimental observations, two-dimensional elasticity theories and two-dimensional finite element analyses. Opti-
cal micrographs of the cross sections of spot welds in lap-shear specimens of a dual-phase steel before and after failure are
first examined to understand the failure mechanism. The experimental results suggest that under lap-shear loading condi-
tions, a necking failure is initiated near the middle of the nugget circumference in the base metal and then the failure prop-
agates along the nugget circumference in the sheet to final fracture. Based on the stress function approach of the elasticity
theory, an analytic solution for an infinite plate containing a rigid inclusion subjected to a resultant shear force is devel-
oped and used to investigate the stress and strain distributions near the nugget in lap-shear specimens. The results of the
elastic analytic solution and those of a two-dimensional elastic finite element analysis indicate that the initial yielding starts
on the two side edges of the inclusion in the sheet. However, the results of a two-dimensional elastic–plastic finite element
analysis indicate that as the applied displacement increases, the maximum equivalent plastic strain shifts from the two side
edges of the inclusion to the middle of the inclusion along the inclusion circumference in the sheet. The computational
results suggest that the location of the initial necking failure should occur near the middle of the nugget circumference
in the sheet as observed in experiments based on the forming limit diagram (FLD) for ductile sheet metals.
Ó 2006 Elsevier Ltd. All rights reserved.

Keywords: Spot weld; Rigid inclusion; Lap-shear specimen; Fatigue; Fracture

1. Introduction

Resistance spot welding is widely used to join sheet metals in the automotive industry. These spot welds are
subjected to multiaxial loads under cyclic or crash loading conditions. Various types of specimens have been
used to determine the strengths and fatigue lives of spot welds under different loading conditions. Since the
spot weld provides a natural crack or notch along the weld nugget circumference, the stress intensity factor
solutions at the critical locations of spot welds have been adopted to investigate the fatigue lives of spot welds

*
Corresponding author. Tel.: +1 734 764 9404; fax: +1 734 647 3170.
E-mail address: jwo@[Link] (J. Pan).

0013-7944/$ - see front matter Ó 2006 Elsevier Ltd. All rights reserved.
doi:10.1016/[Link].2006.03.017
2230 P.-C. Lin et al. / Engineering Fracture Mechanics 73 (2006) 2229–2249

in various types of specimens [1–9]. Recently, Wang et al. [10] conducted a computational study of the
stress intensity factor solutions for spot welds in lap-shear specimens. The computational results verify that
the stress intensity factor solutions of Zhang [7,8] based on the elastic stress solutions for a rigid inclusion
in thin plates and the J integral formulation can be used as the basis for the stress intensity factor solutions
of spot welds under lap-shear and other loading conditions. It seems that the stress and strain solutions near a
rigid inclusion under various loading conditions are very useful to understand the fatigue and fracture of spot
welds.
Based on elasticity theories, the nominal stress fields near spot weld nuggets under general loading condi-
tions can be approximated by those of rigid inclusions in plates, for example, see Zhang [7,9]. The elastic
analytic stress solutions for rigid inclusions in plates under shear, bending and opening loading conditions
were obtained by Muskhelishvili [11], Rießner [12] and Timoshenko and Woinowsky-Krieger [13], respec-
tively. A number of researchers also examined the stresses and strains near the nugget and their relations
to failure modes based on the elasticity theories. For example, Kan [14] and Oh [15] simplified a lap-shear
specimen as a two-dimensional sheet subjected to a uniformly distributed shear force on the weld nugget.
Their analytic and finite element solutions show that the initial yielding is located near the middle of the nug-
get circumference in the sheet. Note that the different material properties of the weld nugget, the heat affected
zone and the base metal were not considered in their model. Recently, Salvini et al. [16] proposed a simplified
model of a finite element assembly based on the elasticity solutions for a rigid inclusion in a finite plate under
shear, bending and opening loading conditions to represent the weld nugget and the neighboring sheet mate-
rial subjected to general loading conditions.
In order to investigate the detailed failure mechanism of spot welds, the plastic strain and stress distribu-
tions near the nugget in the sheet are needed. For lap-shear specimens, analytic elastic–plastic solutions for the
plastic strains and stresses near the nugget are difficult to obtain [17]. Therefore, researchers usually resorted to
numerical methods. For example, Kan [14] proposed a simplified model of a lap-shear specimen and used a
two-dimensional elastic–plastic finite element analysis based on the cyclic stress–strain curves for the sheet
material to obtain the local strain distribution near the spot weld to characterize the fatigue failure mode.
Satoh et al. [18] conducted three-dimensional elastic and elastic–plastic finite element analyses to investigate
the stress and strain distributions near spot welds along the symmetry plane in lap-shear specimens to identify
the fatigue crack initiation sites under high-cycle and low-cycle fatigue loading conditions. The results of the
three-dimensional elastic–plastic analysis show that the strains in the nugget along the symmetry plane are rel-
atively small and the maximum plastic strain occurs outside of the nugget with or without consideration of the
higher stress–strain curves in the nugget and the heat affected zone under low-cycle fatigue conditions. Deng
et al. [19] conducted three-dimensional elastic and elastic–plastic finite element analyses to investigate the
stress distributions in and near spot welds in lap-shear and symmetric coach-peel specimens to understand
the effects of the nugget size and the sheet thickness on the interfacial and nugget pull out failure modes.
Pan and Sheppard [20] also conducted a three-dimensional elastic–plastic finite element analysis to correlate
the fatigue lives of spot welds to the cyclic plastic strain ranges for the material elements near the main notch
in lap-shear and modified coach-peel specimens.
Failure mechanisms of spot welds have been studied based on experimental observations. For example,
Zuniga and Sheppard [21] studied the failure modes of coach-peel and lap-shear specimens using optical
micrographs and SEM fractographs. Lin et al. [22] studied the failure mechanisms of mild and HSLA steel
square-cup specimens using optical micrographs. Wung [23] and Wung et al. [24] proposed an empirical failure
criterion under combined loads based on their experimental results for spot welds in various specimens. From
a more rigorous plasticity viewpoint, VandenBossche [25] examined the failure mechanism of lap-shear spec-
imens based on a perfectly plastic limit load approach. Lin et al. [26] developed a mechanics-based failure cri-
terion for spot welds under combined opening and shear loads based on a perfectly plastic limit load approach
with the weld nugget being assumed as a rigid inclusion. More recently, Lin et al. [17] developed a mechanics-
based failure criterion for spot welds under combined three resultant forces and three resultant moments based
on a perfectly plastic limit load approach with the weld nugget being assumed as a rigid inclusion. The mul-
tiaxial failure criteria are validated by limited available experimental results.
In this paper, the failure mechanism of resistance spot welds in dual-phase steel lap-shear specimens is
investigated from the mechanics viewpoint based on experimental observations, two-dimensional elasticity
 P.-C. Lin et al. / Engineering Fracture Mechanics 73 (2006) 2229–2249 2231

theories and two-dimensional finite element analyses. Lap-shear specimens of a dual-phase steel were used in
the experiments. These lap-shear specimens were subjected to monotonically increasing loads to failure. Opti-
cal micrographs of the cross sections of spot welds in lap-shear specimens before and after failure are obtained
to understand the failure mechanism. Based on the stress function approach of the elasticity theory, an ana-
lytical solution based on a simplified model for an infinite plate containing a rigid inclusion subjected to a
resultant shear force is derived to understand the stress and strain distributions near the nugget in lap-shear
specimens. Based on the elastic analytic solution, the location of the initial yielding is determined and
compared with the experimental observations and the corresponding two-dimensional elastic finite element
computation. Then a two-dimensional elastic–plastic finite element analysis was conducted to understand
the development of plastic flow near the rigid inclusion. Finally, based on the experimental observation of
the necking failure mode near the nugget, the location of the initial necking failure is determined from the
computational results and the forming limit diagram (FLD) for sheet metals.

2. Experimental observations and failure mechanisms

In this investigation, lap-shear spot weld specimens of a dual-phase steel were used. Fig. 1 shows a lap-shear
specimen of the dual-phase steel. Both the upper and lower sheets are the same dual-phase steel and have the
same thickness of 0.7 mm. The nugget has a nominal diameter of 6.4 mm. Note that lap-shear specimens were
commonly used to examine the fatigue, impact and static strengths of spot welds [14,21,25,27]. Fig. 2 shows

Fig. 1. A lap-shear spot weld specimen of a dual-phase steel.

Fig. 2. One piece of a failed dual-phase steel lap-shear specimen.

2232 P.-C. Lin et al. / Engineering Fracture Mechanics 73 (2006) 2229–2249

Fig. 3. A micrograph of the cross section of a spot weld before testing. The weld nugget and the base metal are visible. The heat affected
zone is between the weld nugget and the base metal.

one piece of a failed dual-phase steel lap-shear specimen. A pullout nugget can be seen in the figure. Also, the
sheet is bent near the nugget.
In order to understand the failure mechanism, micrographs of the cross sections of the weld nuggets in lap-
shear specimens are examined. Fig. 3 shows a micrograph of the cross section of a spot weld before testing.
The weld nugget and the base metal are indicated in the figure. The heat affected zone is between the weld
nugget and the base metal. Fig. 4 shows a micrograph of the cross section of the spot weld in a failed lap-shear
specimen before the final separation. In the figure, the arrows indicate the loading direction. Clearly, the spec-
imen is bent near the nugget. Near the upper left portion of the spot weld, a necking failure appears at point A
in leg 1 as marked. Near the lower right portion of the spot weld, a necking process appears to begin in leg 2.
The failure of the spot weld in the lap-shear specimen appears to be initiated near the middle of the nugget
circumference in the sheet, marked as point A and then propagates by necking/shear along the nugget circum-
ference in the sheet, marked as point B. Finally, the upper sheet is torn off. A similar failure mechanism was
also observed in HSLA steel lap-shear specimens [21] and in mild steel square-cup specimens under combined
opening and shear loading conditions [22].
In order to further investigate the development of the necking failure in lap-shear specimens, a rigid inclu-
sion model based on the two-dimensional elasticity theory is first considered here to understand the initial elas-
tic stress and strain distributions near the nugget in a lap-shear specimen and their implications to the failure
mechanism. Note that the failure of spot welds by plastic flow localization under axisymmetric and plane

Fig. 4. A micrograph of the cross section of the spot weld in a failed lap-shear specimen before the final separation.
 P.-C. Lin et al. / Engineering Fracture Mechanics 73 (2006) 2229–2249 2233

strain combined loading conditions was investigated by Wang et al. [28]. However, we intend to find out
whether the elastic analysis of a rigid inclusion subjected to a shear force in an infinite plate can help to under-
stand the development of plastic flow analytically.

3. Two-dimensional elastic analysis

Fig. 5(a) shows a schematic plot of a lap-shear specimen with a set of doublers as marked. The doublers are
usually used to align the applied load to avoid the initial realignment of the specimen under lap-shear loading
conditions. Here, the shear force F is the resultant force applied to the edges of the specimen. As shown in the
side view of the specimen, the shear force F passes through the interfacial surface of the nugget. Fig. 5(b)
shows a free body diagram of the lower sheet where the resultant shear force F is needed on the interfacial

F F

Doubler

Doubler

F F

(a)

F
x F

F F
(b)

Fig. 5. (a) A schematic plot of a lap-shear specimen. The shear force F is the resultant force applied to the edges of the specimen, (b) a free
body diagram of the lower sheet where the resultant shear force F is needed on the interfacial surface of the nugget to balance the resultant
shear force F applied to the lower edge of the specimen.
2234 P.-C. Lin et al. / Engineering Fracture Mechanics 73 (2006) 2229–2249

surface of the nugget to balance the resultant shear force F applied to the lower edge of the specimen. The
shear force appears to dominate the failure mechanism of a lap-shear specimen [21,22]. When the shear force
F becomes large, plastic deformation occurs near the weld nugget in lap-shear specimens with a large ratio of
the nugget diameter to the thickness. When plastic deformation becomes large, finite deformation of the mate-
rial near the nugget takes place and the specimen geometry changes significantly. It is obvious that a closed-
form analytical solution based on a plate or shell theory with the consideration of finite deformation and
plastic deformation is difficult to obtain. As shown in Fig. 4, the necking failure appears to be the dominant
failure mechanism. Note that two-dimensional plane stress analyses have been widely used to investigate the
necking failure of thin sheets, for example, see Hosford and Caddell [29] and Huang et al. [30]. We will take
advantage of the forming limit diagram (FLD) in the sheet forming industry to analyze the failure of the sheet
in the latter part of the paper. However, a two-dimensional elastic plane stress analysis for a rigid inclusion
under a resultant shear force in an infinite plate is first taken here to understand the initial elastic stress dis-
tribution near the spot weld nugget in a lap-shear specimen. The analytic solution derived in this paper will be
useful and convenient for investigation of fatigue failure of spot welds under lap-shear loading conditions.
Based on the top view of the lap-shear specimen shown in Fig. 5(b), a two-dimensional model of a rigid inclu-
sion in an infinite plate is considered as shown in Fig. 6. In this figure, the shaded circle represents a rigid
inclusion of radius a. As shown in the figure, a Cartesian coordinate system is centered at the center of the inclu-
sion. The shear force F, marked as an arrow in the positive y direction, represents the resultant force acting on
the inclusion, which causes the rigid inclusion to move a displacement d in the positive y direction. The rigid
inclusion represents the spot weld nugget and the infinite plate represents the sheet. Here, the inclusion is
assumed to be perfectly bonded to the plate at r = a and the stresses and strains are assumed to be zero at infin-
ity. The stress distributions along the circumference of the rigid nugget are not assumed in any specific forms, in
contrast to the uniform stress distribution assumed in Oh [15]. This rigid inclusion model can be considered as
an approximation of a spot weld in a large lap-shear specimen.
A polar coordinate system centered at the center of the rigid inclusion is also considered here for conve-
nience. The displacement boundary conditions along the inclusion circumference give the displacements at
r = a as
ur ¼ d sin h ð1Þ
uh ¼ d cos h ð2Þ
where ur and uh represent the displacements in the radial and tangential directions, respectively and d is the
rigid-body displacement in the positive y direction. The weak boundary condition along the circumference
of spot welds (r = a) gives the equilibrium equation as

r
θ
F x

2a

Fig. 6. A two-dimensional model of a rigid inclusion of radius a subjected to a resultant shear force F in an infinite plate.
 P.-C. Lin et al. / Engineering Fracture Mechanics 73 (2006) 2229–2249 2235

Z 2p
ðrrr sin h þ rrh cos hÞrt dh ¼ F ð3Þ
0

where t is the thickness of the plate.

It is straightforward to solve the two-dimensional elastic boundary valued problem by the Airy stress func-
tion approach using the general solution of Michell [31]. Based on the boundary conditions in Eqs. (1) and (2)
and the equilibrium equation in Eq. (3), the Airy stress function / in terms of the polar coordinates r and h can
be derived as
F f4r2 h cos h  ½a2 ð1 þ mÞ þ 2r2 ð1 þ mÞ ln r sin hg
/¼ ð4Þ
8prt
The stresses can be derived from the stress function /(r, h) in Eq. (4) as
 
ð1 þ mÞa2 ð3 þ mÞ
rrr ¼ F  sin h ð5aÞ
4pr3 t 4prt
 
ð1 þ mÞa2 ð1 þ mÞ
rrh ¼ F þ cos h ð5bÞ
4pr3 t 4prt
 
ð1 þ mÞa2 ð1 þ mÞ
rhh ¼ F  sin h ð5cÞ
4pr3 t 4prt
where t is the thickness of the plate. Note that the effective stress re based on the Mises yield criterion under
plane stress conditions is expressed as
1
re ¼ ðr2rr  rrr rhh þ r2hh þ 3r2rh Þ2 ð5dÞ
Substitution Eqs. (5a), (5b) and (5c) into Eq. (5d) gives the expression of re in terms of F, m, r, t and h. The
expression for re is quite long and therefore is not listed here. The displacements can then be derived in terms
of F as
F ð1 þ mÞ½ða2  r2 Þð1 þ mÞ  2r2 ð3 þ mÞ ln r
ur ¼  sin h ð6aÞ
8pr2 tE
F ð1 þ mÞ½ða2  r2 Þð1 þ mÞ þ 2r2 ð3 þ mÞ ln r
uh ¼ cos h ð6bÞ
8pr2 tE
The normalized stresses r~rr , r
~rh and r
~hh can be defined as
 
rrr ð1 þ mÞa3 ð3 þ mÞa
~rr ¼
r ¼  sin h ð7aÞ
F =2pat 2r3 2r
 
rrh ð1 þ mÞa3 ð1 þ mÞa
~rh ¼
r ¼ þ cos h ð7bÞ
F =2pat 2r3 2r
 
rhh ð1 þ mÞa3 ð1 þ mÞa
~hh ¼
r ¼  sin h ð7cÞ
F =2pat 2r3 2r

Note that we normalize the stresses such that the maximum magnitude of the normalized radial stress r ~rr
equals 1.0 since the radial stress rrr is the most important stress for derivation of the maximum mode II stress
intensity factor for spot welds under lap-shear loading conditions, see Zhang [7,9]. The normalized effective
stress r
~e can be defined as
1
r r2rr  r
~e ¼ ð~ ~2hh þ 3~
~hh þ r
~rr r r2rh Þ2 ð7dÞ
The stress distributions in the infinite plate now can be obtained from Eq. (5) or (7). Note that the stress dis-
tributions listed in Eq. (5) are important and useful for derivation of the distributions of the stress intensity
factor solutions based on the J integral formulation [7,9].
Fig. 7 shows the angular distributions of the normalized stresses along the inclusion circumference at r = a.
In this investigation, the Poisson’s ratio m is taken as 0.3. Fig. 7(a) shows the angular distribution of the nor-
2236 P.-C. Lin et al. / Engineering Fracture Mechanics 73 (2006) 2229–2249

(max 1.00) (max 1.00)

1.5 1.5

1 1

0.5 0.5
~
σ 0 ~
rr σ 0
rθ

-0.5 -0.5

-1 -1

-1.5 -1.5
0 90 180 270 360 0 90 180 270 360
Angle θ (degree) Angle θ (degree)
(a) (b)

(max 0.30) (max 1.73)

1.5 2

1
1.5
0.5

~
σ 0 ~ 1
θθ σ e

-0.5
0.5
-1

-1.5 0
0 90 180 270 360 0 90 180 270 360
Angle θ (degree) Angle θ (degree)
(c) (d)

Fig. 7. Angular distributions of the normalized stresses along the inclusion circumference at r = a, (a) the normalized radial stress r
~rr ,
(b) the normalized shear stress r
~rh , (c) the normalized hoop stress r
~hh , (d) the normalized effective stress r
~e .

malized radial stress r ~rr at r = a. Since the resultant force F is applied in the positive y direction, the normal-
ized radial stress r~rr is negative in the plate for the upper part of the inclusion (0° < h < 180°) and is positive in
the plate for the lower part of the inclusion (180° < h < 360°). The maximum magnitude of the normalized
radial stress r
~rr is located at the middle of the inclusion (h = 90° and 270°) where their directions are parallel
to the resultant force F. At the locations on the two side edges of the inclusion (h = 0° and 180°), the normal-
ized radial stress vanishes.
Fig. 7(b) shows the angular distribution of the normalized shear stress r ~rh along the inclusion circumfer-
ence. The normalized shear stress r ~rh is negative in the plate on the right hand side of the inclusion
(90° < h < 90°) and is positive on the left hand side of the inclusion (90° < h < 270°). The shear stress is
anti-symmetric about the y axis (h = 90°) because of the requirement of symmetry with respect to the y axis.
In this figure, the maximum magnitude of the normalized shear stress r ~rh is located in the plate on the two side
edges of the inclusion (h = 0° and 180°). At the locations along the symmetry (y) axis (h = 90° and 270°), the
normalized shear stress vanishes.
Fig. 7(c) shows the angular distribution of the normalized hoop stress r ~hh along the inclusion circumference.
~hh has the same trend as r
r ~rr , which is negative in the plate for the upper part of the inclusion (0° < h < 180°)
and is positive in the plate for the lower part of the inclusion (180° < h < 360°). This same trend comes from
the rigid displacement boundary conditions along the circumference of the rigid inclusion. In Eqs. (1), (2) and
(6), we know the displacements along the circumference (at r = a) depend on h only. The definition of the hoop
strain ehh and its elastic relation to rhh and rrr can be written as
1 ouh ur 1
ehh ¼ þ ¼ ðrhh  mrrr Þ ð8Þ
r oh r E
 P.-C. Lin et al. / Engineering Fracture Mechanics 73 (2006) 2229–2249 2237

where E represents the Young’s modulus. Substituting Eq. (6) into Eq. (8) at r = a, the hoop strain ehh is zero
along the circumference. In fact, we can argue that ehh is zero at r = a due to the rigid displacement boundary
conditions at r = a. Since ehh = 0 at r = a, rhh = mrrr. As shown in Fig. 7(a) and (c), the angular distributions of
~rr and r
r ~hh are similar and the ratio of r ~hh to r~rr is equal to the Poisson’s ratio m (=0.3). As indicated in
Fig. 7(a)–(c), the material in the plate along the symmetry (y) axis is under biaxial tensile or biaxial compres-
sive stress conditions with no shear stress due to the symmetry. On the other hand, the material in the plate on
the two side edges of the inclusion is under pure shear conditions.
Finally, Fig. 7(d) shows the normalized effective stress r ~e along the inclusion circumference based on Eq.
(7d). The maximum normalized effective stress r ~e is located on the two side edges of the inclusion (h = 0°
and 180°) in the plate, which coincide with the locations of the maximum normalized shear stresses r ~rh as
shown in Fig. 7(b). As shown in Fig. 7(a), (b) and (c), the maximum magnitudes of r ~rh and r ~rr are identical
but the maximum magnitude of r ~hh is 30% of that of r ~rr . As indicated in Eq. (7d), the contribution of the nor-
malized shear stress, 3~ r2rh , is larger than the contribution of the normalized radial and hoop stresses,
~2rr  r
r ~hh þ r
~rr r ~2hh . Therefore, the maximum magnitude of the normalized shear stress r ~rh determines the loca-
tions of the maximum normalized effective stress r ~e and also the locations of the initial yielding near the rigid
inclusion. Note that we do not make any specific assumption for the distributions of the stresses along the
inclusion circumference. Our results show that the initial yielding occurs on the two side edges of the inclusion.
Ken [14] and Oh [15] assumed a uniform shear stress distribution on the inclusion and their results indicate
that the initial yielding occurs near the middle of the inclusion circumference.
Based on the discussions earlier, the radial distributions of the normalized stresses r ~rr , r
~rh and r ~hh at
h = 180° and 270° are selected to understand the stress distributions in the plate. Fig. 8 shows the radial dis-
tributions of the normalized stresses as functions of the normalized radial distance r/a based on Eq. (7) in the
leftward direction (h = 180°) and downward direction (h = 270°) of the inclusion. All the stresses approach to
zero at infinity based on the given assumptions for the infinite plate. For the leftward direction (h = 180°), the
normalized stresses r ~rr and r ~hh are identically zero regardless of the radius. Fig. 8(a) shows the normalized
shear stresses r ~rh as a function of the normalized radial distance r/a. In Fig. 8(a), the maximum normalized
shear stress r ~rh is located on the inclusion circumference. When the normalized radius increases, the normal-
ized stress sharply decreases. The normalized stress should approach to zero as the normalized radius goes to
infinity.
For the downward direction (h = 270°), the normalized shear stress r ~rh is identically zero due to the sym-
metry. Fig. 8(b) shows the radial distribution of the normalized radial stress r ~rr . When the normalized radius
increases, the normalized radial stress r ~rr slightly increases and then gradually decreases. The normalized stress
should approach to zero as the normalized radius goes to infinity. The maximum r ~rr is located at r/a = 1.087
in the plate. Note that the location of the maximum r ~rr is also very close to the inclusion. The value of the
maximum r ~rr is 1.014 and is very close 1.0 for r ~rr at r = a. Fig. 8(c) show the normalized hoop stress r ~hh as

θ=180° θ=270° θ=270°

1.2 1.2 0.6

0.4
0.8 0.8
~ ~ ~
σ rθ σ rr σ θθ 0.2
0.4 0.4
0

0 0 -0.2
1 3 5 7 9 11 1 3 5 7 9 11 1 3 5 7 9 11
r/a r/a r/a
(a) (b) (c)

Fig. 8. Radial distributions of the normalized stresses as functions of normalized radial distance r/a in the leftward direction (h = 180°)
and downward direction (h = 270°) of the inclusion, (a) the normalized shear stress r
~rh at h = 180°, (b) the normalized radial stress r
~rr at
h = 270°, (c) the normalized hoop stress r
~hh at h = 270°.
2238 P.-C. Lin et al. / Engineering Fracture Mechanics 73 (2006) 2229–2249

a function of the normalized radial distance r/a. Although the normalized hoop stress r ~hh is proportional to
the normalized radial stress r ~rr along the inclusion circumference, the radial stress distributions of the normal-
ized hoop stress r ~hh and the normalized radial stress r ~rr are not proportional to each other. The radial distri-
bution of the normalized hoop stress r ~hh in Fig. 8(c) is quite different from that of the normalized radial stress
~rr in Fig. 8(b). Fig. 8(c) shows that the maximum normalized hoop stress r
r ~hh (=0.3) is located on the inclusion
circumference. As the normalized radial distance increases, the normalized hoop stress r ~hh steeply falls to a
minimum point and then smoothly increases. The normalized stress should approach to zero as the normal-
ized radius goes to infinity.
As shown by the radial stress distributions in Fig. 8 and the angular stress distributions in Fig. 7, the mate-
rial in the plate along the upward and downward directions (h = 90° and 270°) close to the inclusion is under
biaxial compressive and biaxial tensile stress conditions, respectively, due to the symmetry. On the other hand,
the material in the plate along the leftward and rightward (h = 0° and 180°) directions of the inclusion is under
pure shear conditions.
Since the elastic analytic solution shows that the locations of the initial yielding are not in agreement with
the location of the initial necking failure observed in the experiments, an elastic–plastic analysis should be con-
sidered. Unfortunately, the geometry and loading conditions of lap-shear specimens are quite difficult to
derive an elastic–plastic analytic solution as discussed earlier. We here consider an elastic–plastic finite element
analysis as many other researchers, for example, see Kan [14], Satoh et al. [18], Deng et al. [19] and Pan and
Sheppard [24]. It should be mentioned that the necking failure of the base metal shown in Fig. 4 is very similar
to the necking failure of thin sheets under biaxial stretching conditions. Under stretching dominant conditions
in sheet metal forming operations, a two-dimensional elastic–plastic membrane approach can be used to deter-
mine the strain distribution in the sheet [32]. The failure of the sheet can then be determined by the forming
limit diagram [33] which is widely adopted in the sheet metal forming industry. In this paper, we take a two-
dimensional plane stress elastic–plastic approach in order to be consistent with the corresponding two-dimen-
sional elastic analysis as discussed earlier.

4. Two-dimensional finite element analyses

In this investigation, a two-dimensional plane stress elastic finite element analysis is first conducted to
benchmark our numerical solution to the elastic analytic solution as discussed earlier. Then, a two-dimen-
sional plane stress elastic–plastic finite element analysis is conducted to understand the development of plastic
flow near the inclusion under the resultant shear force. Here, the commercial finite element program ABAQUS
is used to perform the computations. A finite element model for a rigid inclusion in a finite square plate is
developed as shown in Fig. 9. Eight-node plane stress quadrilateral continuum elements are used. The diam-
eter of the inclusion is taken as 6.4 mm based on the dimension of the nugget in the lap-shear specimen that we
investigated. The plate width is selected to be 128 mm in order to have the ratio of the plate width to the inclu-
sion diameter equal to 20 to approximate an inclusion in an infinite plate.
In the elastic inclusion analysis discussed earlier, the rigid inclusion is subjected to a rigid-body displacement,
and the stress and the deformation of the infinite plate tend toward zero at infinity. In our finite element model,
the inclusion circumference is fixed and a uniform displacement of d is imposed in the negative y direction along
the upper and lower edges of the plate as schematically shown in Fig. 10. Note that the displacements of the
upper and lower edges are also constrained in the x direction to represent the clamped conditions near the upper
and lower edges of the lap-shear specimen in experiments. The left and right edges of the plate represent the side
edges of the lap-shear specimen and are set as traction-free. As indicated in Zhang [7,9] and Lin et al. [34], the
load applied to the lap-shear specimen can be decomposed into four types of loads under elastic deformation.
The boundary condition considered here is to represent the primary type of load for lap-shear specimens. The
finite element model considered here is used to investigate the stress and strain distributions near the weld nug-
get in the lower sheet of the specimen by considering the elastic–plastic behavior of the plate material. Note that
in this investigation, our goal is to investigate the effect of elastic–plastic material behavior on the plastic flow
near the inclusion qualitatively. Therefore, we consider the elastic perfectly plastic behavior for the plate mate-
rial for simplicity. In the finite element analysis, the Young’s module E is taken as 200 GPa, the Poisson’s ratio m
is taken as 0.3, and the yield stress rY is taken as 523 MPa for the dual-phase steel.
 P.-C. Lin et al. / Engineering Fracture Mechanics 73 (2006) 2229–2249 2239

X
Z

Fig. 9. A two-dimensional plane stress finite element model.

4.1. Two-dimensional elastic finite element analysis

The two-dimensional elastic finite element analysis gives the stress and strain distributions in the plate and
the resultant force along the circumference of the rigid inclusion. In order to compare the stresses obtained
from the finite element analysis with the elastic analytic solution, the elastic analytic solution is obtained
by substituting the value of the resultant force along the circumference of the rigid inclusion from the finite
element analysis into Eq. (5). Fig. 11 shows a comparison of the normalized stresses from the finite element
analysis and the elastic analytic solution at the displacement d = 0.003 mm. The normalized stresses rij and the
normalized effective stress re for the elastic–plastic analysis are defined as
rij
ij ¼
r ð9Þ
rY
re
e ¼
r ð10Þ
rY
where i and j represent r and h, and rY represents the yield stress of the plate material.
Fig. 11(a) shows the normalized effective stress r e along the circumference of the rigid inclusion from
the finite element analysis and the elastic analytic solution. As shown in the figure, the results are in a good
2240 P.-C. Lin et al. / Engineering Fracture Mechanics 73 (2006) 2229–2249

128 mm x

6.4 mm

128 mm

Fig. 10. A schematic plot of the dimensions of the two-dimensional plane stress finite element model. Here, d is the uniform displacement
in the negative y direction along the upper and lower edges. The shaded area represents the fixed displacement boundary along the
inclusion circumference.

agreement. Both results indicate that the initial yielding should start on the two side edges of the inclusion in the
plate. Fig. 11(b) shows the normalized radial stress r rr in the downward direction of the inclusion (h = 270°)
from the finite element analysis and the elastic analytic solution. The results of the finite element analysis
are similar to those of the elastic analytic solution for 1 < r/a < 5. However, as the normalized radial distance
r/a increases, the difference of the results increases. The finite size of the plate in our numerical model gives a
finite value of the normalized radial stress r rr at r/a = 20 determined by the applied displacement boundary
conditions, whereas the elastic analytic solution gives a vanishing value of the normalized radial stress r rr at
infinity. Therefore, the stresses of the finite element analysis are different from those of the elastic analytic solu-
tion at large r/a’s. It should be noted that the elastic analytic solution gives the maximum radial stress at r/
a = 1.087. The maximum radial stress is not significantly larger than the radial stress at r = a. The finite element
size closest to the rigid boundary r = a is about one order less than the inclusion radius. It is not our intention to
catch the maximum radial stress here. Note that when the ratio of the plate width to the inclusion diameter is set
as 10, the difference between the normalized radial stress r rr at r = a and h = 270° based on the elastic analytic
solution and the two-dimensional elastic finite element analysis is less than 5%. The finite element model with
the selection of the ratio of the plate width to the inclusion diameter of 20 here seems to reasonably represent the
model of an inclusion in an infinite plate. When the ratio is less than 10, the difference of the stresses near the
inclusion based on the elastic finite element analysis and the elastic analytic solution become large.

4.2. Two-dimensional elastic–plastic finite element analysis

It is clear now that the stress distributions of the elastic analytic solution and the elastic finite element anal-
ysis near the inclusion agree with each other. However, the elastic analysis cannot lead to any useful informa-
tion to understand the failure mechanism of lap-shear specimen under monotonic increasing loading
conditions. However, we need to conduct the elastic analytic analysis to give us some preliminary understand-
ing of the stress and strain distributions near the inclusion. Based on the von Mises yield criterion, when the
effective stress re reaches the yield stress rY, the plastic deformation should be initiated. Based on the linearity
of the linear elasticity, the displacement d that can initiate plastic deformation on the two side edges of the
rigid inclusion is calculated as 0.025 mm.
 P.-C. Lin et al. / Engineering Fracture Mechanics 73 (2006) 2229–2249 2241

0.2
Analytic solution
0.18 ν=0.3

Normalized effective stress σe (σe /σY)

FEA result
0.16

0.14

0.12

0.1

0.08

0.06

0.04

0.02

0
0 45 90 135 180 225 270 315 360
(a) Angle θ (degree)

0.08
Analytic solution
ν=0.3
0.07
Normalized radial stress σrr (σrr/σY)

FEA result
0.06

0.05

0.04

0.03

0.02

0.01

0
1 2 4 6 8 10 12 14 16 18 20
(b) Normalized radial distance (r/a)

Fig. 11. Comparisons of the results of the elastic finite element analysis and the elastic analytic solution at the displacement d = 0.003 mm,
(a) the angular distributions of the normalized effective stress r e along the circumference of the rigid inclusion at r = a, (b) the radial
distributions of the normalized radial stress r
rr in the downward direction of the inclusion at h = 270°.

Based on the von Mises yield criterion, the equivalent plastic strain rate e_ p is defined as
rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
2 p p
_ep ¼ e_ : e_ ð11Þ
3
where e_ p represents the plastic strain rate tensor. The equivalent plastic strain ep is obtained from the integra-
tion of e_ p over the deformation history as
Z
e ¼ e_ p dt
p
ð12Þ

The equivalent plastic strain ep can be used to indicate the extent of the plastic deformation. In our finite ele-
ment analysis, a total displacement of 0.12 mm, divided into 40 increments, is applied along the upper and
lower edges of the plate. Now we define dY = 0.025 mm at which the plastic deformation is initiated. We
can define the normalized displacements  d as

 d
d¼ ð13Þ
dY
2242 P.-C. Lin et al. / Engineering Fracture Mechanics 73 (2006) 2229–2249

We will present the computational results in terms of the normalized displacement d.
Fig. 12 shows the angular distributions of the equivalent plastic strains ep for several values of the normal-
ized displacement  d. As shown in Fig. 12(a), when the normalized displacement d increases to 1.08, a small
amount of the equivalent plastic strain ep is initiated near the two side edges of the inclusion (h = 0° and
180°). These locations agree with those of the initial yielding according to the elastic analytic solution. When
normalized displacement  d increases to 1.68, the material in the plate along the inclusion circumference
becomes fully plastic and the minimum equivalent plastic strain ep is located at the middle of the inclusion
(h = 90° and 270°). When the normalized displacement d increases to 1.80, the maximum equivalent plastic
strain ep shifts from the locations on the two side edges of the inclusion to the locations near h = 45°, 135°,
225°, and 315° along the inclusion circumference, as shown in Fig. 12(b). As the normalized displacement d
continues to increase to 3.24, the maximum equivalent plastic strain ep along the inclusion circumference grad-
ually shifts to the middle of the inclusion (h = 90° and 270°).
In order to determine the location of the initial necking failure, the strains and the principal strains along
the inclusion circumference (at r = a) are considered. Fig. 13(a) and (b) show the angular distributions of the
radial strain err, the shear strain erh, the major principal strain e1 and the minor principal strain e2 along the
inclusion circumference at the normalized displacement d ¼ 3:24. In Fig. 13(a), the maximum radial strain err

0.06
−
δ = 1.80
−
δ = 1.68
−
0.05 δ = 1.08
p
Equivalent plastic strain ε

0.04

0.03

0.02

0.01
−
δ = 1.08

0
0 45 90 135 180 225 270 315 360
(a) Angle θ (degree)

0.5

0.4
p
Equivalent plastic strain ε

0.3

0.2

0.1

0
0 45 90 135 180 225 270 315 360
(b) Angle θ (degree)

Fig. 12. Angular distributions of the equivalent plastic strain ep along the inclusion circumference as functions of the normalized
displacements 
d, (a) for d ¼ 1:08–1:80, (b) for d ¼ 1:08–3:24.
 P.-C. Lin et al. / Engineering Fracture Mechanics 73 (2006) 2229–2249 2243

0.6
εrθ
εrr
−
0.4 δ = 3.24

0.2

Strains 0

-0.2

-0.4

-0.6
0 45 90 135 180 225 270 315 360
(a) Angle θ (degree)

0.6
ε1
ε2
−
0.4 δ = 3.24

0.2
Principal strains

-0.2

-0.4

-0.6
0 45 90 135 180 225 270 315 360
(b) Angle θ (degree)

Fig. 13. (a) Angular distributions of the radial strain err and the shear strain erh along the inclusion circumference at d ¼ 3:24, (b) angular
distributions of the major principal strain e1 and the minor principal strain e2 along the inclusion circumference at d ¼ 3:24.

at 
d ¼ 3:24 appears at the middle of the inclusion (h = 270°) where the shear strain erh equals to zero. The max-
imum shear strain erh appears on the two side edges of the inclusion (h = 0° and 180°) where the radial strain
err equals to zero. Note again that due to the rigid displacement boundary condition at r = a, the hoop strain
ehh is zero. In Fig. 13(b), the curve of the major principal strain e1 has a nearly flat hump near the middle part
of the inclusion (h = 270°), where the minor principal strain e2 reaches its maximum value and equals to zero.
Along the inclusion circumference, the major principal strain e1 is positive and the minor principal strain e2 is
negative or zero.

5. Failure analysis

It is very difficult to determine the location of the initial necking failure based on the strains with respect to
the polar coordinates because both the tensile and the shear strains exist along the inclusion circumference.
Here, we consider the necking or thinning failure criterion based on the forming limit diagram (FLD) widely
used in the sheet metal forming industry. Discussions on the forming limit diagram can be found, for example,
in the ASM handbook [33]. Necking or thinning failure of thin sheets can be modeled by considering thin
2244 P.-C. Lin et al. / Engineering Fracture Mechanics 73 (2006) 2229–2249

sheets subjected to in-plane biaxial stretching conditions [35]. A schematic forming limit diagram in terms of
the in-plane major and minor strains is shown in Fig. 14. The schematic forming limit diagram is based on the
testing data of Yan et al. [36]. The major and minor strains represent the in-plane principal strains of the
sheets. In the figure, the heavy solid line represents the forming limit under biaxial stretching conditions.
The lowest point of the forming limit is usually located at or near the plane strain tension state where the
minor principal strain is zero as shown in Fig. 14. For the failure analysis of a sheet under a forming opera-
tion, when the strain state of a sheet element is below the forming limit, the sheet element is said to be safe.
When the strain state of a sheet element is above the forming limit, the sheet element is said to be unsafe. The
general trend of a typical forming limit diagram for ductile sheet metals is that on the right (stretching) side,
the increase of the major strain for the forming limit is generally small as the minor strain increases. However,
on the left (drawing) side, the increase of the major strain for the forming limit is quite steep as the minor
strain decreases. The strain state can be defined based on the ratio of the in-plane minor strain to the in-plane
major strain. For example, when the minor strain is equal to the major strain, the sheet is under equal biaxial
tension. When the minor strain is zero and the major strain is positive, the sheet is under plane strain tension.
When the ratio of the minor strain to the major strain is 0.5, the sheet is under uniaxial tension with the
assumption of incompressibility. When the ratio of the minor strain to the major strain is 1, the sheet is
under pure shear. When the major strain is zero and the minor strain is negative, the sheet is under plane strain
compression. When a sheet element is under monotonically increasing straining conditions, the strain paths
for several of these strain states are marked in Fig. 14. The loading path of plane strain compression, not
marked in the figure, is a horizontal line emanating from the origin to the left side of the diagram.
As shown in Fig. 4, a localized neck occurs in the upper left leg and the lower right leg, marked as leg 1 and
leg 2, respectively. Therefore, we consider the principal strains along the inclusion circumference and the typ-
ical forming limit diagram as shown in Fig. 14 to determine the location of the necking failure along the cir-
cumference. Fig. 14 shows the principal strains along the inclusion circumference at d ¼ 3:24. In Fig. 14, the
principal strains for the left hand side of the inclusion from h = 90° to h = 270° are considered because of the
symmetry about the y axis. The strain states of the upper half part of the inclusion from h = 90° to h = 180°
are from plane strain compression to pure shear and these strain states should not result in a necking failure.

1
Forming limit
−
0.9 Strains at δ = 3.24

0.8

0.7

0.6
Major strain ε1

0.5 Plane strain

tension

0.4
Uniaxial
tension
0.3 Pure o θ=270
o

Shear θ=202.5 Equal biaxial

tension
0.2 θ=180
o

0.1
o
θ=90
0
-0.5 -0.4 -0.3 -0.2 -0.1 0 0.1 0.2 0.3 0.4 0.5
Minor strain ε2

Fig. 14. A comparison of the principal strains along the inclusion circumference for the left part of the inclusion from the top (h = 90°) to
the bottom (h = 270°) and a schematic forming limit diagram (FLD) for ductile sheet metals.
 P.-C. Lin et al. / Engineering Fracture Mechanics 73 (2006) 2229–2249 2245

o
90
o o
135 45

o o
180 0 x
F
o o
225 315
o
270

Fig. 15. A schematic plot of the shift of the maximum equivalent plastic strain points along the circumference of the inclusion.

The strain states of the lower part of the inclusion from h = 202.5° to h = 270° are from uniaxial tension to
plane strain tension and these strain states can result in a necking failure. As shown in Fig. 14, the location of
the major strain e1 for h = 270° under plane strain conditions meets the lowest point of the typical forming
limit diagram for ductile sheet metals. In this case, a necking failure should initially occur at h = 270°. It
should be noted that we just use the general trend of a typical forming limit diagram for sheet metals where
the lowest forming limit is located at or near the plane strain tension state to determine the location of the
initial necking failure.
Fig. 15 shows a schematic plot of the shift of the locations of the maximum equivalent plastic strain ep along
the inclusion circumference as the applied displacement increases. The maximum equivalent plastic strain ep
occurs initially on the two side edges of the inclusion (h = 0° and 180°) as the displacement increases. When
the displacement continues to increase, the material near inclusion becomes fully plastic and the locations of
the maximum equivalent plastic strain ep shift to the locations near h = 45°, 135°, 225°, and 315° along the
inclusion circumference. As the displacement further increases, the maximum equivalent plastic strain ep
occurs at the top and bottom of the inclusion (h = 90° and 270°). The material at the top of the inclusion
is under plane strain compression and the material at the bottom of the inclusion is under plane strain tension.
Therefore, the material at the bottom of the inclusion (h = 270°) is the possible location of the initial necking
failure based on the typical forming limit diagram (FLD) as shown in Fig. 14.

6. Discussion

As shown in Fig. 15, the development of plastic zone near the rigid inclusion is from the side edges to the
middle. In the two-dimensional finite element analysis of Kan [14], the plastic flow starts from the middle. The
different development patterns of plastic flow are possibly due to different assumptions for the material prop-
erties of the weld nugget and the uniform distribution of the applied shear stress on the nugget in Kan [14].
The results of the three-dimensional elastic–plastic finite element analysis of Satoh et al. [18] show that the
maximum plastic strain is located at some distance in the order of the sheet thickness away from the nug-
get along the symmetric plane, which corresponds to the location of the initial failure of low-cycle fatigue
in experiments. Note that the yield stress variation in the weld nugget and heat affective zone is considered
in their finite element model based on the micro-Vickers hardness distribution. In the three-dimensional elastic
and elastic–plastic finite element analyses of Deng et al. [19], the stresses at a very small critical distance near
the crack tip are considered to determine the nugget shear fracture mode or the necking failure mode of the
base metal.
In fact, recent axisymmetric and plane strain finite element analyses of the failure modes of spot welds by
Wang et al. [28] based on the experimental observations of Lin et al. [22] for specimens with the thickness of
2246 P.-C. Lin et al. / Engineering Fracture Mechanics 73 (2006) 2229–2249

1 mm suggests that the necking failure should occur at the distance in the order of the thickness away from the
notch tip due to the plastic flow pattern near the nugget. Fig. 16(a) shows a schematic plot of the right half of
the cross section near the spot weld nugget. Here, F represents the resultant shear force. Fig. 16(b) shows a
contour plot of the equivalent plastic strain ep for the right half of the deformed cross section near the weld
nugget based on the two-dimensional plane strain elastic–plastic finite element analysis of Wang et al. [28].
Because of the geometric constraint conditions of the nugget and the material constraint conditions of the
higher stress–strain curves of the nugget and the heat affected zone, a localized neck develops at the distance
in the order of one thickness away from the nugget in the base metal. When the ductility of the material near
the notch or crack along the nugget circumference is high, the necking failure develops before kinked cracks
can be initiated at the critical locations of the notch or crack. The necking failure is initiated by localized shear
bands from the surface of the sheet along the 45° lines as schematically shown in Fig. 16(a). The locations of
the 45° lines depend upon the geometric and material constraints near the nugget. As the resultant shear force
increases, the shear bands continue to develop and finally a necking failure occurs away from the nugget near
the intersection of the two 45° bands, which is about one thickness away from the nugget when the finite defor-
mation of the sheet or the elongation of the neck region is accounted for as shown in Fig. 16(b). The location
of necking at the distance about one thickness away from the nugget cannot be modeled by the plane stress

Fig. 16. (a) A schematic plot of the right half of the cross section near the spot weld nugget, (b) a contour plot of the equivalent plastic
strain ep for the right half of the deformed cross section near the spot weld nugget based on the two-dimensional plane strain elastic–plastic
finite element analysis of Wang et al. [28].
 P.-C. Lin et al. / Engineering Fracture Mechanics 73 (2006) 2229–2249 2247

elastic–plastic finite element analysis as reported in this paper. A detailed three-dimensional elastic–plastic
finite element analysis, which is out of the scope of this investigation, can be performed to model this necking
failure accurately in the future. In general, the simplified two-dimensional analysis could be valid for the cases
of thin sheets where the ratios of the nugget radius to the thickness are large. We have also conducted an
elastic–plastic finite element analysis where the top edge of the square plate is stress-free. The general trend
for the development of plastic flow for the lower part of the inclusion is quite similar to the results presented
here. Therefore, the results will not be presented.
In general, for mild steel and dual-phase steel spot welds, the yield stress for the weld nugget should be
higher than that of the base metal based on the higher hardness for the weld nugget, for example, see Satoh
et al. [18] and Milititsky et al. [37]. Therefore, when we investigate the failure of the spot weld specimen with
consideration of plastic deformation under monotonically increasing loading conditions, the assumption of
the rigid inclusion for the weld nugget that we adopt here seems to be reasonable. On the other hand, the elas-
tic stress and strain distributions near the nugget from the two-dimensional plane stress analysis may not be
very accurate when compared to those based on the results of three-dimensional finite element analyses where
the nearly linear stress and strain distributions through the thickness or in the remaining ligament of kinked
cracks due to the bending are observed, see Deng et al. [19] and Wang et al. [38]. However, the elastic stress
solutions based on the assumption of the rigid nugget as detailed here are very useful to obtain the analytical
KI, KII and KIII solutions to determine the fatigue lives of spot welded structures [4,7,9]. From this viewpoint,
the presentation of the detailed elasticity solution procedure is justified here. Note that fatigue crack growth
models have been proposed to predict fatigue lives in various types of specimens based on the elastic stress
intensity factor solutions for kinked cracks [38–41]. The elastic solution for the rigid inclusion under shear
in an infinite plate was presented in Muskhelishvili [11] based on the complex variable approach. Here, we
take a stress function of Michell [31] to solve the rigid inclusion under shear in an infinite plate. Note that
the stress function approach for the inclusion under shear in a finite circular plate was taken by Salvini
et al. [16] for their finite element assembly to represent the weld nugget and the neighboring sheet. The elastic
solutions for thin plates with rigid inclusions under various loading conditions are the key results to develop
the closed-form stress intensity factor solutions for spot welds, for example, see Zhang [7,8] and Lin et al. [34].

7. Conclusion

The failure mechanism of resistance spot welds in dual-phase steel lap-shear specimens was investigated
based on experimental observations. Both the upper sheet and lower sheet have the same thickness. For
the lap-shear specimens, the nugget pullout failure mode was observed. The micrographs of the cross sections
of the failed spot welds suggest that under lap-shear loading conditions, the necking failure is initiated at
the middle of the nugget circumference in the base metal and the failure propagates by necking/shear along
the nugget circumference in the sheet to final fracture.
In the two-dimensional elastic analysis, an infinite plate containing a rigid inclusion subjected to a resultant
shear force is considered to represent the lower sheet of a lap-shear specimen under loads. Based on the Airy
stress function of the two-dimensional elasticity theory, the stress distributions in the infinite plate are derived.
The elastic analytic solution shows that the maximum effective stress is located in the plate on the two side
edges of the inclusion. The locations of the maximum effective stress suggest the locations of the initial yielding
in lap-shear specimens. The detailed presentation of the stress distributions near the inclusion can be useful to
determine the stress intensity factor solutions along the spot weld circumference.
Both two-dimensional plane stress elastic and elastic–plastic finite element analyses of a square plate with a
rigid inclusion were conducted. The results of the two-dimensional elastic finite element analysis near the
inclusion are in a good agreement with those of the elastic analytic solution. The initial yielding is located
on the two side edges of the inclusion. As the applied displacement increases, the material near the inclusion
becomes fully plastic. Once the material along the inclusion circumference becomes fully plastic, the maximum
equivalent plastic strain ep along the inclusion circumference gradually shifts from the locations on the two
side edges of the inclusion to the locations at the middle of the inclusion. With considerations of the forming
limit diagram and the experimental observations, the location of the initial necking failure can be determined
by comparing the principal strains along the inclusion circumference with a typical forming limit diagram
2248 P.-C. Lin et al. / Engineering Fracture Mechanics 73 (2006) 2229–2249

(FLD) for ductile sheet metals. The results of the comparison suggest the location of the initial necking failure
is at the middle of the inclusion (h = 270°) as observed in experiments. In the future, a three-dimensional elas-
tic–plastic finite element analysis can be used to validate the qualitative trends of the two-dimensional elastic–
plastic finite element analysis presented in this paper.

Acknowledgements

The support of an Army/Ford IMPACT project, a Ford University Research Program, and a NSF grant
under Grant No. DMI-0456755 is greatly appreciated. Helpful discussions with Dr. D.-A. Wang of the
National Chung Hsing University are appreciated.

References

[1] Pook LP. Fracture mechanics analysis of the fatigue behaviour of spot welds. Int J Fract 1975;11:173–6.
[2] Pook LP. Approximate stress intensity factors obtained from simple plate bending theory. Engng Fract Mech 1979;12:505–22.
[3] Radaj D, Zhang S. Stress intensity factors for spot welds between plates of unequal thickness. Engng Fract Mech 1991;39:391–413.
[4] Radaj D, Zhang S. Simplified formulae for stress intensity factors of spot welds. Engng Fract Mech 1991;40:233–6.
[5] Swellam MH, Kurath P, Lawrence FV. Electric-potential-drop studies of fatigue crack development in tensile-shear spot welds. In:
Mitchell MR, Landgraf RW, editors. Advances in fatigue lifetime predictive techniques. ASTM STP, vol. 1122; 1992. p. 383–401.
[6] Swellam MH, Banas G, Lawrence FV. A fatigue design parameter for spot welds. Fatigue Fract Engng Mater Struct
1994;17:1197–204.
[7] Zhang S. Stress intensities at spot welds. Int J Fract 1997;88:167–85.
[8] Zhang S. Approximate stress intensity factors and notch stresses for common spot-welded specimens. Weld J 1999;78:173s–9s.
[9] Zhang S. Fracture mechanics solutions to spot welds. Int J Fract 2001;112:247–74.
[10] Wang D-A, Lin P-C, Pan J. Geometric functions of stress intensity factor solutions for spot welds in lap-shear specimens. Int J Solids
Struct 2005;42:6299–318.
[11] Muskhelishvili NI. Some basic problems of the mathematical theory of elasticity. Groningen: Noordhoff; 1953.
[12] Rießner Von H. Über die unsymmetrische biegung dünner kreisringplatten. Ing Archiv 1929;1:72–83.
[13] Timoshenko S, Woinowsky-Krieger S. Theory of plates and shells. 2nd ed. New York: McGrow-Hill; 1959.
[14] Kan YR. Fatigue resistance of spotwelds—an analytical study. Met Engng Quart 1976;16:26–36.
[15] Oh HL. Fatigue-life prediction for spotweld using Neuber’s rule. In: Abelkis PR, Hudson CM, editors. Design of fatigue and fracture
resistant structures. ASTM STP, vol 761; 1982. p. 296–309.
[16] Salvini P, Vivio F, Vullo V. A spot weld finite element for structural modelling. Int J Fatigue 2000;22:645–56.
[17] Lin S-H, 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.
[18] Satoh T, Abe H, Nishikawa K, Morita M. On three-dimensional elastic–plastic stress analysis of spot-welded joint under tensile shear
load. Trans Jpn Weld Soc 1991;22:46–51.
[19] Deng X, Chen W, Shi G. Three-dimensional finite element analysis of the mechanical behavior of spot welds. Finite Elem Anal Des
2000;35:17–39.
[20] Pan N, Sheppard S. Spot welds fatigue life prediction with cyclic strain range. Int J Fatigue 2002;24:519–28.
[21] Zuniga S, Sheppard SD. Resistance spot weld failure loads and modes in overload conditions. In: Piascik RS, Newman JC, Dowling
NE, editors. Fatigue and fracture mechanics: 27th volume. ASTM STP, vol. 1296; 1997. p. 469–89.
[22] Lin S-H, Pan J, Wu S, Tyan T, Wung P. Spot weld failure loads under combined mode loading conditions. SAE Technical Paper no.
2001-01-0428. Warrendale: Society of Automotive Engineering; 2001.
[23] Wung P. A force-based failure criterion for spot weld design. J Exp Mech 2001;41:107–13.
[24] Wung P, Walsh T, Ourchane A, Stwart W, Jie M. Failure of spot welds under in-plane static loading. J Exp Mech 2001;41:100–6.
[25] VandenBossche DJ. Ultimate strength and failure mode of spot welds in high strength steels. SAE Technical Paper no. 770214.
Warrendale: Society of Automotive Engineering; 1977.
[26] Lin S-H, Pan J, Wu S-R, 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.
[27] Zhang S. A strain gauge method for spot weld testing. SAE Technical Paper no. 2003-01-0977. Warrendale: Society of Automotive
Engineering; 2003.
[28] Wang D-A, Lin P-C, Pan J. Failure mechanisms of spot welds under axisymmetric and plane-strain combined loading conditions, to
be published.
[29] Hosford WF, Caddell RM. Metal forming: mechanics and metallurgy. Englewood Cliffs: Prentice Hall; 1993.
[30] Huang H-M, Pan J, Tang SC. Failure prediction in anisotropic sheet metals under forming operations with consideration of rotating
principal stretch directions. Int J Plast 2000;16:611–33.
[31] Michell JH. On the direct determination of stress in an elastic solid, with application to the theory of plates. Proc London Math Soc
1899;31:100–24.
 P.-C. Lin et al. / Engineering Fracture Mechanics 73 (2006) 2229–2249 2249

[32] Wang N-M, Budiansky B. Analysis of sheet metal stamping by a finite element method. J Appl Mech 1978;45:73–82.
[33] ASM handbook volume 14: forming and forging. Materials Park, Ohio: ASM International; 1988.
[34] Lin P-C, Wang D-A, Pan J. Mode I stress intensity factor solutions for spot welds in lap-shear specimens. Int J Solids Struct,
submitted for publication.
[35] Hill R. The mathematical theory of plasticity. Oxford: Clarendon Press; 1950.
[36] Yan B, Laurin K, Sriram S, Huang M, Chintamani J, Lalam SH. A new dual phase steel for automotive body panels. SAE Technical
Paper no. 2003-01-0518. Warrendale: Society of Automotive Engineering; 2003.
[37] Milititsky M, Pakalnins E, Jiang C, Thompson AK. On characteristics of DP600 resistance spot welds. SAE Technical Paper no.
2003-01-0520. Warrendale: Society of Automotive Engineering; 2003.
[38] Wang D-A, Lin S-H, Pan J. Stress intensity factors for spot welds and associated kinked cracks in cup specimens. Int J Fatigue
2005;27:581–98.
[39] Lin S-H, Pan J. Fatigue life prediction of spot welds in coach-peel and lap-shear specimens with consideration of kinked crack
behavior. Int J Mater Product Technol 2004;20:31–50.
[40] Wang D-A, Pan J. A computational study of local stress intensity factor solutions for kinked cracks near spot welds in lap-shear
specimens. Int J Solids Struct 2005;42:6277–98.
[41] Lin S-H, Pan J, Wung P, Chiang J. A fatigue crack growth model for spot welds in various types of specimens under cyclic loading
conditions. Int J Fatigue 2006;28:792–803.