Você está na página 1de 10

Fatigue & Fracture of Engineering Materials & Structures 1998; 21: 11231132

A CRACK GROWTH APPROACH TO LIFE PREDICTION OF


SPOT-WELDED LAP JOINTS
J. A. Nrvx:N and N. E. DoviiNc
Engineering Science and Mechanics, Virginia Polytechnic Institute and State University, Blacksburg, VA 24061, USA
AbstractFatigue behaviour of spot-welded lap joints is modelled as a single-degree-of-freedom crack
growth problem. Although such a model involves simplication of a complicated problem, predictions
are in good agreement with experimental results. The model developed here allows a design engineer to
analyse the fatigue behaviour of spot-welded steel sheets, which are commonly used in structures, without
knowledge of metallurgical and ne geometric details of spot-welds. Only fatigue properties of the sheet
metal are needed, so no laboratory facilities are required to generate fatigue data specic to spot-welds
or weld metal.
KeywordsSpot-weld; Fatigue; Stress ratio; Crack growth; Lap joint; Mixed-mode; Stress intensity factors.
NOMENCLATURE
a=crack length
a
f
=crack length at failure
C=constant for crack growth relation
d=tip diameter of spot-weld electrode
da/dN=crack growth rate
D=overall diameter of spot-weld electrode
F=spot-weld electrode force
HAZ=heat-aected-zone
k
f
=fatigue notch factor
K
I
=initial mode I stress intensity
K
II
=initial mode II stress intensity
K
Ieq
=equivalent mode I stress intensity
K
max=
maximum stress intensity
m=constant for crack growth relation
N
f
=total fatigue life
N
I
=crack initiation phase
N
II
=crack propagation phase
N
III
=large crack growth phase
P=applied load
P
max=
maximum applied load
r=spot-weld radius
R=load ratio, P
min
/P
max
t=sheet thickness
b=constant quantifying interaction of K
I
and K
II
DK=local stress intensity range (mode I)
DK
I
=initial mode I stress intensity range
DK
II
=initial mode II stress intensity range
DK=eective zero-to-max DK
DP=eective zero-to-max DP
c=constant for crack growth relation
h
o=
angle of crack growth
w=tip angle of spot-weld electrode
INTRODUCTION
Modern automobiles contain hundreds, or even thousands, of spot-welds. Spot-welding is an
inexpensive and eective way to join sheets of metal. Unfortunately, fatigue cracks tend to occur
1998 Blackwell Science Ltd 1123
1124 J. A. Nrvx:N and N. E. DoviiNc
around these welds, in the heat-aected-zone (HAZ), resulting in a loss of structural integrity and
eventual failure of the component in question. The study of spot-weld fatigue is complicated by
three-dimensional crack growth, residual stresses created by welding, and a complex metallurgical
structure produced during the welding process. The most common life prediction methods found
in the literature are briey introduced in this paper, followed by development of a simple model
that is shown to eectively estimate the fatigue lives of spot-welds. This simple model is intended
to be used by design engineers who might not understand intimate details of spot-weld fatigue,
but wish to make use of spot-welds in fatigue applications.
Four basic categories of life prediction methods have been discussed in the literature: reference
strain approach, local strain-based approach, stress intensity factor approach and crack growth
approach. These are briey described below.
Reference strain approach
In this approach, the strain at one or more particular locations, usually on the outer surface
near the edges of spot-welds, are monitored and used as reference values. Fatigue data, obtained
from small coupon specimens in a laboratory environment, and plotted in terms of this reference
strain amplitude, may then be used to monitor or predict the lives of spot-welds in structures
[13]. This method must be used with care because strain changes rapidly near spot-welds. Exact
placement of strain gauges, as well as the size of the strain gauges used, is critical and needs to be
standardized. Also, it is not clear if multiaxial loading eects can be completely characterized by
only one or two strain gauges.
L ocal strain-based approach
In this method, the edges of spot-welds, where fatigue cracks occur, are considered to be
geometric notches with some notch radius at the root [48]. Analysis is similar to the strain-
based approach used elsewhere [9], where the stressstrain and strainlife curves, for the given
material, are used with Neubers rule. Since no well-dened notch exists, the fatigue notch factor,
k
f
, must be obtained by empirical means. The total life of a given spot-welded specimen is divided
into three phases: crack initiation, crack propagation through the sheet metal to the free surface,
and large crack growth resulting in structural failure. The number of cycles dedicated to each
phase are N
I
, N
II
and N
III
, respectively. The total life is the sum of these three phases, as shown
in Eq. ( 1).
N
f
=N
I
+N
II
+N
III
(1)
Sharply notched members in general are observed to have their fatigue lives dominated by crack
growth [1013]. Spot weld connections are a form of a sharply notched member, and also have
behaviour dominated by crack growth, with crack initiation lives appearing to be short or non-
existent [14]. However, a strain-based approach is based on strains that occur locally at the tip
of a notch having a denite radius, and so seems inappropriate for situations dominated by
crack growth.
Stress intensity factor approach
An alternative means of applying fracture mechanics is to use stress intensity factors to charac-
terize the loading. The stress intensity factors for spot-welds in the lap shear geometry have been
calculated [15], and both mode I and II stress intensities exist initially at the location of failure,
point A in Fig. 1. Due to symmetry, a crack can form on either side of a given spot-weld; however,
failure almost always results from a single large crack, as shown. Both sides of a spot-welded
Fatigue & Fracture of Engineering Materials & Structures, 21, 11231132 1998 Blackwell Science Ltd
Life prediction of spot-welded lap joints 1125
(a) (b)
Fig. 1. Geometry of a lap-shear spot-weld specimen and typical angled thumbnail crack. (a) Side view
(centreline). ( b) Front view (outer surface).
specimen need monitoring as a result of this observation. An equivalent mode I stress intensity
factor, K
Ieq
, is dened, typically in the form of Eq. (2), where b is a constant quantifying the
interaction between K
I
and K
II
[1618].
K
Ieq
=

K2
I
+bK2
II
(2)
This parameter, K
Ieq
, may be plotted versus fatigue life and serves as an empirical life prediction
method. This method is used to consolidate data from various geometric and loading congurations
of spot-welds.
Crack growth approach
Since the crack initiation phase in spot-welds is a small portion of total fatigue life, if it exists
at all, a more accurate way to describe the fatigue process in spot-welds is by using a crack growth
approach. A great deal of literature exists that supports the belief that spot-weld edges behave as
sharp cracks [14,1926]. Recall that stress intensity factors for a lap joint conguration are known
and an equivalent mode I stress intensity, K
Ieq
, can be calculated. A crack growth model may be
employed using a da/dN versus DK relation to predict life. Sheppard derived such a model where
only mode I stress intensities, initially acting at the location of failure, were considered [27]. The
approach taken in this paper is more closely related to the crack growth portion of Ref. [27] than
to any of the other various approaches found in the published literature.
MODEL DEVELOPMENT
The purpose of this work is to develop a simple single-degree-of-freedom model that a design
engineer can use to estimate fatigue lives of three-dimensional spot-welded lap joints when only
the crack growth behaviour of the base metal is known. A crack growth approach is used so the
initial spot-weld connection is treated as a sharp crack. Furthermore, failure is dened to occur
when a crack propagates completely through at least one steel sheet. In terms of Eq. (1), N
I
=0
and N
III
=0. Aside from the da/dN versus DK relation, no empirical means are used; however,
some simplifying assumptions are made. The stress state, created by both mode I and II stress
intensities at the failure location (point A in Fig. 1), is used to calculate an eective mode I stress
intensity factor and to predict the direction of crack propagation.
From numerical work done previously by Pook [15], the stress intensity factors for a three-
dimensional spot-welded lap joint conguration are known. Since a lap joint always fails at the
Fatigue & Fracture of Engineering Materials & Structures, 21, 11231132 1998 Blackwell Science Ltd
1126 J. A. Nrvx:N and N. E. DoviiNc
same location, the stress intensities are usually only given for this point, specically point A in
Fig. 1. At point A, only mode I and II stress intensities exist, and are given as Eqs (3) and (4),
where r is the radius of the weld, t is the thickness of the steel sheets (assumed to be of equal
thickness), and P is the force applied to the lap joint. These equations apply for specimens at least
ve spot-weld radii wide (w5r) and average stress on the weld nugget no more than 80% of the
yield stress.
K
I
=
P
r3/2 C
0.341
A
2r
t B
0.397
D
(3)
K
II
=
P
r3/2 C
0.282+0.162
A
2r
t B
0.710
D
(4)
It is assumed that a crack in a spot-weld, at the failure location, can be modelled as a planar
crack. This approximation becomes valid as the ratio t/r is decreased. For the welds tested in this
paper, t/r#0.28, which is typical of well-designed spot-welds. Cracks grow from the interface of
the two steel sheets at the edge of the weld, in a shape similar to a semi-ellipse, but curved slightly
out of plane to give it a thumbnail shape (Fig. 1). If the weld radius is large compared to the
sheet thickness, this thumbnail crack is nearly planar. Although not truly two-dimensional, this
crack can be described by a single degree of freedom, a. Thus, Eqs (3) and (4) are used for this
single-degree-of-freedom model.
Now that this spot-weld model has been reduced to a single degree of freedom, fatigue life is
modelled as starting from a two-dimensional crack, which exists initially under both mode I and
II loading. If we assume that the crack propagates perpendicular to the direction of the maximum
principal normal stress at the crack tip, it will initially grow at an angle h
o
, as shown in Fig. 2.
This angle satises Eq. (5), where K
I
and K
II
are the stress intensity factors initially acting at point
A in Fig. 1 before crack growth occurs [15]. This equation is obtained by combining well-known
crack-tip singularity stress eld equations for modes I and II, and nding the principal stress
directions [28]. Note that Eq. ( 5) is valid only in the case where K
III
is zero.
K
I
sin h
o
=K
II
(3 cos h
o
1) (5)
It has been noted in the literature that the crack growth rate in spot-welds remains nearly constant
as the test proceeds for constant amplitude loading and for a small ratio of t/r [24]. This suggests
that the stress intensity factor also remains nearly constant. These assumptions are not valid for
thick sheets and/or small welds, which are of reduced interest, since such welds are generally
inferior in both static and fatigue loading [25,26].
Fig. 2. Geometry of a propagating crack. Fig. 3. Geometry of lap-shear spot-weld specimens, as
tested.
Fatigue & Fracture of Engineering Materials & Structures, 21, 11231132 1998 Blackwell Science Ltd
Life prediction of spot-welded lap joints 1127
Since the stress intensity factor is assumed to be constant, h
o
also remains constant throughout
the test, which is consistent with experimental observations. After solving for h
o
, an eective mode I
stress intensity factor needs to be calculated. This was obtained using equations for stress around
large two-dimensional cracks subject to mode I and II loading with an innitesimally small kinked
crack growing from the original crack tip at an angle [29,30]. The innitesimal kink is assumed
not to disturb the original stress eld. The small kink grows in principal stress directions, so no
local mode II loading exists at the kinked crack tip. The local mode I stress intensity factor, acting
on the innitesimal kinked crack, is given in Eq. (6), and is obtained by examining the principal
normal stresses acting on this new kinked crack tip [28]. This is the formulation for eective
mode I stress intensity used for this model.
DK=
DK
I
4 C
3 cos
h
o
2
+cos
3h
o
2 D

DK
II
4 C
3 sin
h
o
2
+3 sin
3h
o
2 D
(6)
A Paris crack growth relation, with an adjustment for the R-ratio eect as suggested by Walker
[31], is assumed to describe crack growth, as in Eq. (7). With C, m and c as material constants
from fatigue test data of the base metal, this relation is intended only for pure mode I loading.
da
dN
=C
C
DK
(1R)(1c)D
m
(7)
Since no mode II or III loading is present at the small kinked crack of length a, Eq. (7) can be
used in this model. With reference to Fig. 2, the crack grows a distance a
f
=t/sin h
o
before
penetrating the outer surface of the plate, which is considered failure in this study. In the case of
constant crack growth, Eq. (7) can be integrated to obtain an equation for fatigue life, Eq. (8).
N
f
=
t
C sin h
o
C
DK
(1R)(1c)D
m
(8)
By substituting Eq. (6) into Eq. (8), we are now able to express fatigue life in closed form for a
known h
o
. Note that h
o
must be known a priori, and thus Eq. (5) must rst be solved for h
o
using
Eqs (3) and (4) for K
I
and K
II
. The resulting equation is the desired single-degree-of-freedom
approximation for fatigue life, and is given by Eq. (9).
N
f
=
t
C sin h
o
C
DK
I C
3 cos
h
o
2
+cos
3h
o
2 D
DK
II C
3 sin
h
o
2
+3 sin
3h
o
2 D
4(1R)(1c)
D
m
(9)
COMPARISONWITH EXPERIMENTAL RESULTS
Experimental method
Experimental data were generated using small coupon specimens with a single weld, as shown
in Fig. 3. Coupon specimens were made by spot-welding two strips of steel sheet metal together.
Each of the steel strips was 75 mm long, 25 mm wide and 1 mm thick. These strips were overlapped
by 25 mm and welded in the centre of the overlapped region with a weld #7 mm in diameter, so
that the overall length of a specimen was 125 mm. Fatigue tests were performed under constant
amplitude loading on closed loop servo-hydraulic testing machines operated at various frequencies
(130 Hz) and load amplitudes. No eect on fatigue life was observed to occur due to test frequency.
Three dierent load ratios, R=P
max
/P
min
, were tested; R=0.1, 0.5 and 0.75. Both ends of each
Fatigue & Fracture of Engineering Materials & Structures, 21, 11231132 1998 Blackwell Science Ltd
1128 J. A. Nrvx:N and N. E. DoviiNc
specimen had a spacer attached, so no initial bending occurred before testing. Each specimen end
was placed #20 mm deep into the grips of a testing machine. Again, failure was dened to occur
when a fatigue crack was observed to have penetrated one or both of the steel sheets. All cracks
were found to propagate from the spot-weld at point A in Fig. 1. All the tests were run in lab air
and at ambient temperature.
Data interpretation
The R-ratio adjustment suggested by Walker [31] allows an eective zero-to-maximum stress
intensity range to be dened, as given in Eq. (10). This allows the driving force on the crack to be
expressed in terms of both DK and R, and allows tests performed at dierent load ratios and load
amplitudes to be compared.
DK=
DK
(1R)(1c)
=K
max
(1R)c (10)
Since load and stress intensity are linearly related, as in Eqs (3) and (4), an eective zero-to-
maximum load can also be dened [29] and is given by Eq. (11). This allows fatigue life data for
dierent load ratios to be shown on the same plot.
DP=P
max
(1R)c (11)
Two base metals were used in this study, SAE 1010 and SAE 950X steels. The dierence in
fatigue performance between these two steels was small, and in light of this fact, the same constants
for Eq. (7) were used for both steels. Although some limited crack growth data exist in the literature
for the weld metal and the HAZ [32,33], where cracks around spot-welds propagate, such data
are rare. It has been shown that fatigue crack growth behaviour of HAZ metal is nearly identical
to that of the base metal for another steel [34]. It is assumed that this observation is typical of
weldable steels. Crack growth data for the base metal are used, instead of those for the HAZ,
because they are readily available for most engineering materials and deliver nearly identical
results. Crack growth data for the base metal are used because they are readily available for most
engineering materials. Due to the small quantity of the available material, no crack growth data
for the two base metals, SAE 1010 and SAE 950X steels, were generated in this study. Thus, the
constants C, m and c, were chosen from established data for similar ferritepearlite steels [35,36].
Values for the constant C were chosen so that they made upper and lower bounds on established
data within the Paris regime of fatigue crack growth, and were C=7109 and 2109,
respectively. Note that the units used are SI, specically, MPam for K and mm/cycle for da/dN.
The chosen value of m is typical for ferritepearlite steels [35], and c is estimated from the limited
R-ratio data on a steel similar to SAE 1010, specically ASTM A36 [S. J. Hudak (1979) unpublished
data]. The constants used were m=3 and c=0.75. Care was taken not to choose constants for
Eq. ( 7) to simply best t this spot-weld fatigue data and, thus, force this model to work, since
design engineers would likely not have the benet of such prejudices.
As seen in Figs 4 and 5, fatigue data for the three load ratios tested, R=0.1, 0.5 and 0.75,
collapse such that a linear trend is seen on a loglog plot when Eq. (11) is used, allowing typical
scatter associated with fatigue data. The dashed and dotted lines on these loadlife plots are
predicted lines using the theoretical model previously outlined. The dashed lines are for
C=2109 and the dotted lines are for C=7109. Two solid horizontal lines also appear in
both Figs 4 and 5. The upper line corresponds to the average static strength of a specimen made
from the corresponding material. The lower horizontal line corresponds to an observed fatigue
threshold condition. The average static failure loads are 8000 N and 8800 N, and DK corresponding
Fatigue & Fracture of Engineering Materials & Structures, 21, 11231132 1998 Blackwell Science Ltd
Life prediction of spot-welded lap joints 1129
10
2
10
3
10
4
10
5
10
6
10
7
10
2
10
3
10
4
10
5
10
6
10
7
Fig. 4. Fatigue data for SAE 1010 lap-shear spot-welds. Fig. 5. Fatigue data for SAE 950X lap-shear spot-welds.
to the fatigue thresholds indicated in Figs 4 and 5 are 10.0 MPam and 7.6 MPam for SAE
1010 and SAE 950X steels, respectively.
DISCUSSION
In order to analyse three-dimensional behaviour as a single-degree-of-freedom model, several
simplifying assumptions have to be made. The assumptions made to develop this model are
as follows.
$ The weld is circular, the edge of which behaves as a sharp crack, so that no crack initiation
phase exists. The weld and specimen are symmetric.
$ The ratio t/r is small. This produces a nearly planar crack, and allows Eqs (3) and (4) to be
used for this single-degree-of-freedom model.
$ The rate of crack growth is nearly constant. This allows Eq. (7) to be integrated and makes
a closed form solution for fatigue life possible.
$ Equation (7) is a good descriptor of crack growth. This allows the fatigue lives to be calculated
and compared for various load ratios, R, and load amplitudes, DP.
$ The thickness of both steel sheets connected by the spot-weld are equal. Relations similar to
Eqs (3) and (4) are available for sheets of diering thickness, but are more complicated and
not used here.
$ The specimen is suciently wide, i.e. at least ve spot-weld radii, and the average stress on
the weld nugget does not exceed 80% of the yield stress [15]. These are restrictions placed
on Eqs (3) and (4).
For lives in the range of 103106 cycles, experimental data agree well with the predictions of this
model. Since the dashed and dotted lines in Figs 4 and 5 are bounds on the fatigue data (for the
base metal ), the predicted values are somewhat conservative, which is desirable from a design
standpoint. Deviations between experimental data and prediction occur at both high and low
applied load amplitudes. This model assumes that fatigue crack growth is described by Eq. (7),
but this description is not valid near the threshold or as P
max
approaches the average static strength
of the specimens. At lives less than 103 cycles, the peak load approaches the maximum loads
observed in static tests, as seen in Figs 4 and 5. In these large load amplitude tests, the spot-weld
nugget rotates a signicant amount and large deformations exist about the spot-weld. Large
Fatigue & Fracture of Engineering Materials & Structures, 21, 11231132 1998 Blackwell Science Ltd
1130 J. A. Nrvx:N and N. E. DoviiNc
deformations indicate that Eqs (3) and (4) may be invalid, because the average shear stress on the
spot-weld nugget may exceed 80% of the yield stress. For a design engineer, such behaviour would
likely be considered undesirable, avoided, and thus, is of limited interest. At long lives, greater
than 106 cycles, a threshold condition is observed where further reduction in load produces a
much longer life than predicted using this method. The fatigue threshold shown in Figs 4 and 5
was suggested by examination of spot-weld fatigue data and determining a value of DP below
which no crack growth was observed. The fatigue thresholds indicated correspond to 10.0 and
7.6 MPam for SAE 1010 and SAE 950X steels, respectively. For both materials, the observed
fatigue threshold is #40% larger than that found in similar steels for R=0 [33]. This model can
easily be programmed into a computer and modied to output parameters, e.g. K
max
and DK, to
gauge the validity of Eq. (7), as was done for this study. For the material and geometry tested,
this method has been shown to predict experimental data well, with a slight conservative bias.
Because this method is straightforward, and essentially only consists of Eq. (9), it is a potentially
useful and easy to use tool for design engineers that may have little detailed information about
fatigue behaviour specic to spot-welds.
Although this life prediction model has limitations, it overcomes problems of the other major
methods, e.g. reference strain approaches, local strain-based approaches, stress intensity factor
approaches, and even other crack growth approaches. The reference strain approach, although
easy to use because only outer surface measurements are required, is sensitive to placement and
dimensions of the strain gauge(s) used. Furthermore, empirical trends are used to predict fatigue
lives, which may be specic for the material and geometry tested. Local strain-based approaches
appear not to be good descriptors of actual behaviour. From a design standpoint, if crack initiation
lives are negligibly small, ignoring this would simplify fatigue analysis and lead to slightly conserva-
tive predictions. Stress intensity factor approaches are also based on empirical relations and do
not directly address questions about crack driving forces or the direction of crack growth. Even
with other crack growth approaches, such questions go unanswered.
This method denes the direction of crack propagation as well as the crack driving force in
terms of the mixed mode stress eld initially present at the edge of the spot-weld, modelled here
as a planar crack. No empirical relations are used, other than Eq. (7), and failure is predicted
based on the experimentally observed failure mechanism, fatigue crack growth. The manner in
which these details are treated is unique to this model.
CONCLUSIONS
In this study, a life prediction model for spot-welded lap joints was developed and veried with
experimental data. Although complicated three-dimensional behaviour is modelled as a single-
degree-of-freedom crack growth problem, good agreement with experimental data is obtained. The
major strength of this method is its simplicity. No specic knowledge regarding metallurgy or ne
geometry at the edge of a spot-weld is required. With only standard fatigue crack growth data for
the base metal, fatigue life estimations are possible. Thus, specic details of the weld metal and
additional laboratory facilities to generate crack growth data specic to weld metal, or HAZ, are
not needed. Furthermore, this method is mechanism based (fatigue crack growth) and does not
rely on empirical means for life prediction. The method outlined in this paper is intended for a
design engineer who needs to estimate fatigue lives of spot-welds.
AcknowledgementsPortions of this work were supported by Ford Motor Co., with Yuting Rui as technical monitor.
Specimens used in this study were provided by Ford. This support is appreciated.
Fatigue & Fracture of Engineering Materials & Structures, 21, 11231132 1998 Blackwell Science Ltd
Life prediction of spot-welded lap joints 1131
REFERENCES
1. Y. Rui, R. S. Borsos, R. Gopalakrishnan, H. N. Agrawal and C. Rivard (1993) The fatigue life prediction
method for multi-spot-welded structures. SAE Paper no. 930571.
2. A. Mabuchi, J. Niisawa and N. Tomioka (1986) Fatigue life prediction of spot-welded box-section beams
under repeated torsion. SAE Paper no. 860603.
3. M. Mizui, T. Sekine, A. Tsujimura, T. Takishima and Y. Shimazaki (1988) An evaluation of fatigue
strength for various kinds of spot-welded test specimens. SAE Paper no. 880375.
4. F. V. Lawrence, N.-J. Ho and P. K. Mazumdar (1981) Predicting the fatigue resistance of welds. Ann.
Rev. Mater. Sci. 11, 401425.
5. F. V. Lawrence, P. C. Wang and H. T. Corten (1983) An empirical method for estimating the fatigue
resistance of tensile-shear spot-welds. SAE Paper no. 830228.
6. J. C. McMahon, G. A. Smith and F. V. Lawrence (1990) Fatigue crack initiation and growth in tensile-
shear spot weldments. In: Fatigue and Fracture T esting of Weldments (Edited by H. I. McHenry and
J. M. Potter), ASTM STP 1058, ASTM, Philadelphia, PA, pp. 4777.
7. S. D. Sheppard and M. Strange (1992) Fatigue life estimation in resistance spot welds: initiation and
early growth phase. Fatigue Fract. Engng Mater. Struct. 15, 531549.
8. S. D. Sheppard (1993) Estimation of fatigue propagation life in resistance spot welds. In: Advances in
Fatigue L ifetime Predictive T echniques, Second Volume (Edited by M. R. Mitchell and R. W. Landgraf ),
ASTM STP 1292, ASTM, Philadelphia, PA, pp. 169185.
9. N. E. Dowling (1993) Mechanical Behavior of Materials, Prentice Hall, Englewood Clis, New Jersey.
10. N. E. Dowling (1982) A discussion of methods for estimating fatigue life. SAE Paper no. 820691.
11. N. E. Dowling (1979) Notched member fatigue life predictions combining crack initiation and propagation.
Fatigue Fract. Engng Mater. Struct. 2, 129138.
12. N. E. Frost (1960) Notch eects and the critical alternating stress required to propagate a crack in an
aluminum alloy subject to fatigue loading. J. Mech. Engng Sci. 2, 109119.
13. N. E. Frost and D. S. Dugdale ( 1957) Fatigue tests on notched mild steel plates with measurements of
fatigue cracks. J. Mech. Phys. Solids 5, 182192.
14. J. F. Cooper and R. A. Smith (1986) Fatigue crack propagation at spot welds. Metal Construction
June, 383386.
15. L. P. Pook (1975) Fracture mechanics analysis of the fatigue behavior of spot welds. Int. J. Fracture
11, 173176.
16. M. H. Swellam, P. Kurath and F. V. Lawrence (1992) Electric-potential-drop studies of fatigue crack
development in tensile-shear spot welds. In: Advances in Fatigue L ifetime Predictive Techniques (Edited
by M. R. Mitchell and R. W. Landgraf ), ASTM STP 1122, ASTM, Philadelphia, PA, pp. 383401.
17. M. H. M. Swellam (1991) A fatigue design parameter for spot welds. Fracture Control Program Report
no. 157, University of Illinois.
18. P. Kurath (1992) Multiaxial fatigue criteria for spot welds. SAE Paper no. 920668.
19. J. L. Overbeeke and J. Draisma (1978) Inuence of stress relieving on fatigue of heavy duty spot welded
lap joints. Metal Construction, Sept., 433434.
20. J. L . Overbeeke and J. Draisma (1974) Fatigue characteristics of heavy-duty spot-welded lap joints. Metal
Construction British Welding J., Sept., 213219.
21. J. L . Overbeeke (1976) Fatigue of spot welded lap joints. Metal Construction, July, 212215.
22. G. Matsoukas, G. P. Steven and Y. W. Mai (1984) Fatigue of spot-welded lap joints. Int. J. Fatigue 6, 5557.
23. B. Pollard (1982) Fatigue strength of spot welds in titanium-bearing HSLA steels. SAE Paper no. 820284.
24. J. F. Cooper and R. A. Smith (1985) The measurement of fatigue cracks at spot-welds. Int. J. Fatigue
7, 137140.
25. D. J. VandenBossche (1977) Ultimate strength and failure mode of spot welds in high strength steels.
SAE Paper no. 770214.
26. J. A. Davidson and E. J. Imhof Jr (1983) A review of the fatigue properties of spot-welded sheet steels.
SAE Paper no. 830033.
27. S. D. Sheppard (1996) Further renement of a methodology for fatigue life estimation in resistance spot
weld connections. In: Advances in Fatigue L ifetime Predictive Techniques, 3rd Volume (Edited by M. R.
Mitchell and R. W. Landgraf ), ASTM STP 1292, ASTM, Philadelphia, PA, pp. 265282.
28. H. Tada, P. Paris and G. Irwin (1985) T he Stress Analysis of Cracks Handbook, Del Research Corp.
29. J. A. Newman (1996) Life prediction of spot-welds: a fatigue crack growth approach. M.S. thesis,
Fatigue & Fracture of Engineering Materials & Structures, 21, 11231132 1998 Blackwell Science Ltd
1132 J. A. Nrvx:N and N. E. DoviiNc
Engineering Science and Mechanics Department, Virginia Polytechnic Institute and State University,
Blacksburg, VA, USA.
30. T. L. Anderson (1991) Fracture Mechanics: Fundamentals and Applications, 2nd edition, CRC Press, Boca
Raton, FL, pp. 9195.
31. K. Walker (1970) The eect of stress ratio during crack propagation and fatigue for 2024-T3 and 7075-T6
aluminum. In: EVects of Environment and Complex L oad History on Fatigue L ife, ASTM STP 462, ASTM,
Philadelphia, PA, pp. 114.
32. B. M. Kapadia and E. J. Imhof Jr (1977) Fatigue-crack propagation in electroslag weldments. In: Flaw
Growth and Fracture, ASTM STP 631, ASTM, Philadelphia, PA, pp. 159173.
33. D. Taylor and L. Jianchun (1993) Sourcebook on Fatigue Crack Propagation: T hresholds and Crack
Closure, Chameleon Press, London, UK.
34. D. Aurich, K. Wobst and H. Krafka (1987) Some experimental results concerning cleavage fracture and
crack resistance in heat aected zones and comparison with base material. In: T he Fracture Mechanics
of Welds (Edited by J. G. Blauel and K.-H. Schwalbe), Mechanical Engineering Publications, London,
pp. 125153.
35. J. M. Barsom and S. T. Rolfe (1987) Fracture and Fatigue Control in Structures, 2nd edition, Prentice
Hall, Englewood Clis, NJ, p. 290.
36. S. J. Hudak (1979) unpublished data.
APPENDIX
Spot-welds are created when two electrodes, as shown in Fig. A1, compress two or more sheets of metal together, pass
an electrical current through the sheet metal to be welded, and hold the sheets together while the liqueed weld metal
quickly resolidies. These stages are called squeeze time, weld time and hold time, respectively. The electrical current melts
a portion of metal between the welding electrodes from all the sheets involved, which resolidies and joins the sheets
together. The current, electrode force, electrode dimensions and length of time for each stage of welding are carefully
prescribed for a given sheet thickness and type of metal. For both SAE 1010 and SAE 950X steels used in this study, the
electrode dimensions, as shown in Fig. A1, are as follows: D=15.75 mm, d=6.35 mm and w=45. All other information
regarding the welding schedule is shown in Table A1.
Table 1. Welding schedule
SAE 1010 steel SAE 950X steel
Electrode force, F (N) 3100 4300
Squeeze time (cycles) 6 6
Weld time (cycles) 10 13
Hold time (cycles) 7 7
Weld current (amperes) 12 500 10 000
Fig. A1. Geometry of spot-welding electrodes.
Fatigue & Fracture of Engineering Materials & Structures, 21, 11231132 1998 Blackwell Science Ltd

Você também pode gostar