International Journal of Fatigue 32 (2010) 434–442

Contents lists available at ScienceDirect

International Journal of Fatigue

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

Simulation of dynamic cornering fatigue test of a steel passenger car wheel

Xiaofeng Wang a,*, Xiaoge Zhang b,1
a
State Key Laboratory of Automotive Safety and Energy, Department of Automotive Engineering, Tsinghua University, Beijing 100084, PR China
b
Dongfeng Automotive Wheel Co., Ltd., Shiyan 442042, PR China

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

Article history: This paper aims at seeking a practical and comprehensive method for simulating the dynamic cornering
Received 10 March 2009 fatigue test of the automotive wheels. The test of a steel passenger car wheel is simulated by combined
Received in revised form 10 August 2009 use of the linear transient dynamic finite element analysis and the local strain approach. A rotating force
Accepted 2 September 2009
of constant magnitude is applied to the moment arm tip to simulate the rotating bending effect on the
Available online 7 September 2009
wheel, with the wheel stationary. It is found that only a radial component of the rotating force is needed
to obtain the sufficiently accurate radial normal strain histories of the elements located along the radial
Keywords:
direction. The strain history of the element whose local stress–strain characteristic keeps linear and clos-
Passenger car wheel
Finite element analysis
est to the critical element is applied to predict the fatigue life of the critical element with Neuber’s rule
Transient strain history and local strain approach, which is quite close to the test results.
Radial normal strain Ó 2009 Elsevier Ltd. All rights reserved.
Fatigue life prediction

1. Introduction and introducing the method in such a way that it is easy for others
to make the simulation following the method introduced.
The dynamic cornering fatigue test is a standard SAE test [1],
which simulates cornering induced loads to the wheel. Fig. 1 shows
the test system applied by Dongfeng Automotive Wheel Co., Ltd., in 2. Static finite element model of the wheel
which the test wheel is mounted to the rotating table, the moment
arm is fixed to the wheel outer mounting pad with the bolts and a The wheel analyzed is a steel passenger car wheel produced by
constant force is applied at the tip of the moment arm by the load- Dongfeng Automotive Wheel Co., Ltd. The static finite element
ing actuator and bearing, thus imparting a constant rotating bend- model of the wheel is first constructed, as shown in Fig. 2. The fi-
ing moment to the wheel. If the wheel passes the dynamic nite element analysis software used in this paper is HyperWorks
cornering fatigue test, it has a good chance of passing all other re- 8.0. The rim and the disc of the wheel are modeled with the plate
quired durability tests [2]. So it is desirable to be able to estimate and shell elements (CQUAAD4 and CTRIA3) whose dimensions are
whether the wheel under design passes the dynamic cornering fa- about 2 mm. The thicknesses of the elements for the rim and the
tigue test or not and make design improvements necessary accord- disc are determined according to their nominal thicknesses and
ingly even before the first prototype wheel is made in the wheel are 2.8 mm and 3.5 mm, respectively. The thickness of the ele-
design stage. Therefore the method to simulate the dynamic cor- ments in the connection zones of the disc and the rim is 6.3 mm,
nering fatigue test of the wheel with finite element analysis meth- the sum of 2.8 mm and 3.5 mm. The moment arm is modeled with
od and local stress–strain fatigue life prediction was studied. the solid elements CTETRA whose dimensions are between 2 mm
Some papers dealing with the simulation of the dynamic cor- and about 25 mm. The forces and moments are transmitted
nering fatigue tests of automotive wheels have been published, through the common nodes in the interfaces of the disc and the
which, such as those by Riesner and DeVries [2], Karandikar and moment arm. The bolt connections are not modeled for simplicity
Fuchs [3], Shang, et al. [4], are valuable but generally do not pro- and the fact that comparatively small number of failures occurred
vide sufficient details for others to follow and make the simulation. at the bolt circle areas in the real dynamic cornering fatigue tests of
This paper aims at seeking a practical and comprehensive method the wheel studied. Although this simplified model is not realistic in
for simulating the test and predicting the wheel fatigue life in it the interfaces, it may only introduce minor errors in the critical
areas because they are sufficiently far away from the interfaces
of the moment arm and the wheel (as shown in Fig. 3). And it is
* Corresponding author. Tel.: +86 10 62781338; fax: +86 10 62784655.
E-mail addresses: wangxf60@mail.tsinghua.edu.cn (X. Wang), syxgchina@126.
beneficial for the calculation speed.
com (X. Zhang). The wheel is made of a material equivalent to St37, a steel,
1
Tel./fax: +86 719 8221399. whose fatigue properties are shown in Table 1 [5]. The material

0142-1123/$ - see front matter Ó 2009 Elsevier Ltd. All rights reserved.
doi:10.1016/j.ijfatigue.2009.09.006
 X. Wang, X. Zhang / International Journal of Fatigue 32 (2010) 434–442 435

Fig. 1. Sketch of the dynamic cornering fatigue test system applied by Dongfeng Automotive Wheel Co., Ltd.

Fig. 2. Static finite element model of the wheel.

properties for the elements above are as follows: modulus of elas- of the wheel along its radial direction. And the local reference
ticity, E = 214 GPa; Poisson’s ratio, l = 0.3; density, q = 7900 kg/m3. frame of each strain rosette is attached to the wheel.
The translational degrees of freedom along X, Y and Z directions The strains of each strain rosette were measured for four posi-
of all element nodes contacting with the rotating table of the test tions by rotating the rotating table: (1) 0° position, in which the
rig are restrained and a force of 1562.4 N is applied to the center axis O1Y1 of the local reference frame coincides with the axis OY
node in the end plane of the moment arm of the length of of the fixed reference frame OXYZ (Fig. 4), which does not rotate
938 mm, as shown in Figs. 1 and 2. with the wheel; (2) 90° position, in which the axis O1Y1 coincides
The solver Optistruct is applied to process the wheel finite ele- with the axis OX of the fixed reference frame; (3) 180° position, in
ment model constructed and the results are shown in Fig. 3 (von which the axis O1Y1 coincides with the negative direction of axis
Mises stress distribution in the wheel disc), with the critical areas OY; (4) 270° position, in which the axis O1Y1 coincides with the
being mainly located near the ventilation windows and the hat of negative direction of the axis OX. Table 2 shows the strains mea-
the disc and the element No. 68225 having the maximum von sured and Von Mises stresses calculated with the strains and the
Mises stress 239 MPa. finite element model of the wheel. Table 3 shows the correspond-
ing normal stresses and shear stresses. The following formulas are
used for processing the strain data measured:
3. Static experimental stress analysis E
rx1 ¼ ðe0 þ le90 Þ ð1Þ
1  l2
The static experimental stress analysis was made to validate the E
static finite element model of the wheel constructed. Five 45° ry1 ¼ ðle0 þ e90 Þ ð2Þ
1  l2
strain rosettes were pasted onto the surface of the wheel disc, as
E
shown in Fig. 4, and the static bending test was done, in which sx1y1 ¼ ðe0 þ e90  2e45 Þ ð3Þ
the rotating table was held stationary and a constant force of 2ð1 þ lÞ
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
1562.4 N was applied at the tip of the moment arm. And the strains
rx1 þ ry1 ðrx1  ry1 Þ2 þ 4s2x1y1
were measured. Fig. 4 also shows the local reference frame of each r1 ¼ þ ð4Þ
2 2
strain rosette, in which axis O1Y1 points outwards from the center
436 X. Wang, X. Zhang / International Journal of Fatigue 32 (2010) 434–442

Fig. 3. Von Mises stress distribution in the outside surface of the wheel disc.

qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
ðrx1  ry1 Þ2 þ 4s2x1y1 It can be seen, in Table 2, that the percentage errors, Er’s, of rv,c
rx1 þ ry1 calculated with the finite element model of the wheel relative to rv
r2 ¼  ð5Þ
2 2 ffi
rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi based on the strains measured are less than 18% and most of them
1 2
rv ¼ ½r1 þ r22 þ ðr1  r2 Þ2  ð6Þ
 2 
rv ;c  rv  Table 1
Er ¼  
 ð7Þ St37 fatigue properties [4].
r v
Order no. Parameter Data
1 Cyclic modulus of elasticity, E, GPa 214
where rx1, ry1 and sx1y1 are normal stresses and shear stress in the 2 Cyclic Strength coefficient, K0 MPa 988
reference frame O1X1Y1; e0, e45 and e90 are 0° strain, 45° strain and 3 Cyclic strain hardening exponent, n0 0.207
90° strain of a strain rosette, respectively; r1 and r2 are principal 4 Fatigue strength coefficient, r0f , MPa 873
stresses; rv is the Von Mises stress calculated with the strains mea- 5 Fatigue strength exponent, b 0.10
6 Fatigue ductility coefficient, e0f 0.557
sured; rv,c is the Von Mises stress calculated with the finite element
7 Fatigue ductility exponent, c 0.518
model of the wheel; Er is the percentage error of rv,c relative to rv.
 X. Wang, X. Zhang / International Journal of Fatigue 32 (2010) 434–442 437

are less than 10%. Thus the finite element model of the wheel con-
structed is quite accurate.
The strain rosette B, C, D and E were pasted almost in the same
circumferential circle (Fig. 4). And it can be seen, in Table 2, that
the rv’s in the 0° and 180° positions are quite close to each other,
which means that the bolt holes and ventilation windows only
have minor influence on the stresses in those two positions, in
which the stresses reach their extreme values.
It can be seen, in Table 3, that the shear stresses, sx1y1’s in the 0°
and 180° positions are generally much smaller than the corre-
sponding normal stresses, rx1’s and ry1’s. And the principal stres-
ses are quite close to the corresponding normal stresses. The
stress states on those points are basically tensile normal stress in
both O1X1 and O1Y1 direction in the 0° position and compressive
normal stress in both O1X1 and O1Y1 direction in the 180° position.
And the normal stress ranges in O1Y1 direction, Dry1’s, are larger
than the corresponding normal stress ranges in O1X1 direction,
Drx1’s, which indicates that the fatigue crack may occur first in
the circumferential direction of the wheel, which is parallel to
Fig. 4. Five 45° strain rosettes on the outside surface of the disc and local reference
the O1X1 direction. And most fatigue cracks did occur in the cir-
frame of each strain rosette. cumferential direction in the dynamic cornering fatigue tests made
by Dongfeng Automotive Wheel Co., Ltd. Fig. 5 shows an example
of the wheel after the test, in which the circumferential fatigue
cracks were formed in the test. Thus the fatigue crack of the wheel
Table 2
Measured stains of each strain rosette and corresponding Von Mises stresses.
disc may be mainly caused by the stress–strain cycles in the O1Y1
direction or radial direction.
Strain Position(°) 0° 45° 90° rv rv,c Er
rosette strain strain strain (MPa) (MPa) (%)
(le) (le) (le)
A 0 358 569 737 176 197 12
4. Normal modes analysis of the wheel and moment arm
90 9 269 62 68 63 7.4 combination
180 399 568 704 174 196 13
270 8 266 56 68 68 0 The modal transient dynamic analysis module of the Optistruct
B 0 322 581 756 176 187 6.3 will be used to simulate the dynamic cornering fatigue test of the
90 37 309 35 79 84 6.3 wheel, which is based on the normal modes analysis of the wheel.
180 282 553 763 174 179 2.9
Thus the normal modes analysis of the wheel is done first with the
270 27 311 22 82 84 2.4
normal modes analysis module of the Optistruct, in which the fi-
C 0 114 240 717 151 171 13
nite element model is the same as the static model introduced
90 36 109 139 34 31 8.8
180 149 256 847 180 173 3.9 above, except for the loading condition. Table 4 shows the modal
270 31 111 137 33 31 6.1
D 0 305 640 644 160 172 7.5
90 129 37 371 104 85 18
180 337 672 639 164 172 4.9
270 121 13 346 100 85 15
E 0 266 456 714 163 176 8.0
90 112 436 181 94 91 3.2
180 301 473 728 169 177 4.7
270 118 446 201 96 91 5.2

Table 3
Stress states of five points for different positions based on the strains measured.

Point Position(°) r1 r2 rx1 Drx1 ry1 Dry1 sx1y1

(MPa) (MPa) (MPa) (MPa) (MPa) (MPa) (MPa)
A 0 199 136 136 199 3.5
180 143 194 144 280 194 393 2.7
B 0 201 128 129 201 6.9 Fig. 5. Fatigue cracks in the wheel formed in the dynamic cornering fatigue test.
180 120 200 120 249 199 400 5.0
C 0 23 161 24 161 10.1
180 190 23 25 49 189 350 15.3 Table 4
Modal frequencies lower than 1000 Hz.
D 0 106 184 117 173 27.2
180 188 110 124 241 174 347 30.3 Order 1 2 3 4 5 6 7 8
E 0 113 187 113 187 5.6 Frequency, 22.2 22.2 104.3 269.4 271.2 506.8 509.7 804.0
180 193 121 122 235 192 379 6.8 Hz
438 X. Wang, X. Zhang / International Journal of Fatigue 32 (2010) 434–442

frequencies lower than 1000 Hz. And Fig. 6 shows the 1st and the
5th order of the modes (eigenvectors).

5. Transient dynamic analysis of the dynamic cornering fatigue

test of the wheel

The wheel rotates and the force is constant in both magnitude

and direction in the dynamic cornering fatigue test (Fig. 1). But
in order to simplify the construction of the finite element model,
the wheel is held fixed and the force of constant magnitude Fo ro-
tates, which is applied at the tip of the moment arm, as shown in
Fig. 7. The rotating force can be resolved into the components
along OX and OY axis of the fixed reference frame OXYZ,

F x ¼ F o cos xt ð8Þ
F y ¼ F o sin xt ð9Þ

where Fx is the force component along OX axis; Fy is the force com-

ponent along OY axis; x is the angular speed of the wheel in the
test; t is time.
The modal transient dynamic analysis module of the Optistruct Fig. 7. The wheel is held fixed and the force of a constant magnitude Fo rotates.
is used to simulate the dynamic cornering fatigue test of the wheel,
in which the finite element model is the same as the static model up the stresses and strains at the same point, which are obtained
introduced above, except for the loading condition. Two finite ele- with those two models, respectively. Thus a linear hypothesis is
ment models are constructed. Only Fy is applied at the tip of the made. And this hypothesis is reasonable because the finite element
moment arm in the Model 1, as shown in Fig. 8, and only Fx is ap- model constructed is a linear one and the wheel structure basically
plied at the tip in the Model 2, as shown in Fig. 9. The stress and remains linear in the test except for some very small areas of high
strain at any point in the wheel structure are obtained by adding stress concentration.

Fig. 6. The first five orders of the modes (eigenvectors).

 X. Wang, X. Zhang / International Journal of Fatigue 32 (2010) 434–442 439

Table 5
Results of the accelerated fatigue tests of the wheel.

Test moment Fatigue life or test time 104 Failure mode

(Nm) cycles
2199 4.8 Circumferential crack in the
disc
2199 5.6 Circumferential crack in the
disc
2199 4.2 Crack at the bolt circle in
the disc
2199 4.3 Circumferential crack in the
disc
2199 4.0 No failure (suspended)
2199 4.0 No failure (suspended)

valley values of the transient responses. And 800 transient re-

Fig. 8. Model 1 (only Fy is applied at the tip of the moment arm). sponse points are calculated, which corresponds to the time length
of 5.6 s. The modes whose modal frequencies are below 1000 Hz
are used to calculate the transient responses. A constant damping
Dongfeng Automotive wheel Co., Ltd. found that the test time value of 0.02 is set for each mode in the above frequency range.
for initiating a fatigue crack in the wheel was too long if a constant Fig. 10 shows the transient displacement history of the center node
force of 1562.4 N (refer to 2) was applied. Thus they did acceler- in the end surface of the moment arm in the OY direction (refer to
ated fatigue tests by applying a larger force of 2344.4 N to the Fig. 7), which is obtained with the Model 1 shown in Fig. 8. And it
wheels. The overall number of the wheels tested was six and Table can be seen that the transient displacement basically reaches its
5 shows the results. It can be seen, in the table, that there were four steady state after about 3 s. The similar situation occurs for the
wheels failed out of the six and the three wheels failed had the cir- transient displacement history of the same center node in the OX
cumferential fatigue cracks in their discs, as shown in Fig. 5. Thus it direction, which is obtained with the Model 2 shown in Fig. 9.
is decided to focus on studying the method for predicting the fati- The reason for choosing a constant damping value of 0.02 is that
gue life of the wheel for this failure mode. The simulation of the the ratio of the excitation frequency 7.143 Hz to the first order
accelerated test mentioned above is studied as an example. Thus modal frequency 22.2 Hz is smaller than 0.75, with the damping ra-
the constant force Fo is 2344.4 N in the formulas (8) and (9). And tios having little effect on the steady response amplitudes. In view
the angular speed x in the formulas should be the same as that of the test system (Fig. 1), the damping ratios are small and a com-
of the rotating table of the test rig (Fig. 1) in the dynamic cornering paratively small damping ratio 0.02 is chosen. The simulations with
fatigue tests, which is about 429 rpm. So. the model show that the steady displacement amplitudes of the
moment arm tip are 10.025 mm, 10.020 mm, 9.996 mm and
429
f ¼ ¼ 7:143 Hz 9.955 mm, respectively, for the damping ratios of 0.02, 0.04, 0.08
60 and 0.12, with the maximum relative error being 0.7%.
1 1
T¼ ¼ ¼ 0:14 s
f 7:143
x ¼ 2pf 6. Fatigue life prediction of the wheel in the dynamic cornering
fatigue test
where f is the frequency of the rotating table; T is the periodic time
of it. It is decided to make use of the local strain approach [6–8] to
The time interval selected is 0.007 s or one twentieth of T, in or- predict the fatigue life of the wheel, whose basic equations are
der to obtain sufficiently good calculation accuracy of the peak and the cyclic stress–strain curve,

Fig. 9. Model 2 (only Fx is applied at the tip of the moment arm).

440 X. Wang, X. Zhang / International Journal of Fatigue 32 (2010) 434–442

  10 The strains of the critical zones in the wheel structure should be

De Dr Dr n
¼ þ ð10Þ determined to predict the wheel fatigue life. Fig. 11 shows the
2 2E 2K 0
critical element of the wheel, Element A, and other two elements,
and the strain–life equation, Element B and Element C, which are located along the OY axis.
Fig. 12 shows the normal Y strain history of the Element A with
De r0f  r0 the amplitude larger than those of other elements. It should be
¼ ð2Nf Þb þ e0f ð2N f Þc ð11Þ noted that the Element A is the element No. 68225 having the max-
2 E
imum von Mises stress 239 MPa, as shown in Fig. 3. So, the static
where E is cyclic modulus of elasticity; K0 is cyclic strength coeffi- finite element model is effective in searching the critical areas of
cient; n0 is cyclic strain hardening exponent; r0f is fatigue strength the wheel tested.
coefficient; b is fatigue strength exponent; e0f is fatigue ductility The amplitudes of the normal Y strain histories of the Element
coefficient; c is fatigue ductility exponent. Table 1 shows the values A, which are caused only by Fy and by both Fx and Fy, are
of these parameters of the wheel material (st37). De is the strain 1925.0 le and 1924.8 le, respectively. So the contribution of the
range of the stress–strain hysteresis loop caused by the load history force F x ¼ F o cos xt along the OX axis to the amplitude of the resul-
in the material; Dr is the stress range of the loop; r0 is the mean tant normal Y strain history only account for 0.010%. The similar
stress of the loop. De, Dr and r0 for each stress–strain hysteresis situations occur for the Element B and the Element C. Thus the
loop in the material can be obtained by the local stress–strain re- amplitudes of the resultant normal Y strain histories of the ele-
sponse analysis [7]. ments which are located along the OY axis can be accurately

Fig. 10. Transient displacement history of the center node in the end surface of the moment arm in the OY direction.

Fig. 11. Critical elements of the wheel.

 X. Wang, X. Zhang / International Journal of Fatigue 32 (2010) 434–442 441

obtained only with the Model 1 (Fig. 8), in which only the force
F y ¼ F o sin xt is applied. So only the Model 1 is needed to calculate
the transient normal Y strain history of an element by rotating the
wheel to make the element be located along the OY axis.
A special software based on the local strain approach [7] is
made and used to process the strain histories of the Element A, Ele-
ment B and Element C, which are obtained with the Model 1
(Fig. 8) and have the amplitudes of 1925 le, 564.40 le and
332.00 le, respectively. And Fig. 13 shows the local stress–strain
hysteresis loops formed. It indicates that the local stress versus
strain characteristic of the Element A is highly non-linear. But
the modal transient dynamic analysis applied is based on linear
hypothesis. Thus the finite element model used can not give an
accurate and realistic transient strain history of the Element A. It
is necessary to find an element along the OY axis, which is close
to the critical element (the Element A) and whose local stress–
strain characteristic remains basically linear, in order for the Model
1 (Fig. 8) be able to obtain accurate transient strain history of it.
It can be seen in Fig. 13 that the Element C is the element closest
to the critical element (the Element A), in which the local stress
versus strain characteristic keeps nearly linear. And Neuber’s rule
is applied to predict the local stress–strain response in the critical
Fig. 13. Local stress–strain hysteresis loops formed in the Element A, Element B and
Element C.

element (the Element A) from the transient normal Y strain history

of the Element C, which is obtained with the Model 1 (Fig. 8). The
Neuber’s rule applied is

K 2f DrC DeC ¼ DrB DeB ð12Þ

where DrC and DeC are the local stress range and strain range of the
hysteresis loop formed in the Element C; DrA and DeA are the local
stress range and strain range of the corresponding hysteresis loop in
the Element A; Kf is the fatigue notch factor. The fatigue notch factor
of the Element A relative to the Element C, Kf, is estimated with the
following formula:
AA
Kf ¼ ð13Þ
AC
where AA and AC are the amplitudes of the normal Y strain histories
of the Element A and the Element C, respectively, which are ob-
tained with the finite element model. Thus Kf = 1925/332 = 5.80.
The special software based on the local strain approach [7] is
applied to process the transient normal Y strain history of the Ele-
ment C to predict the local stress–strain response in the Element A
and predict the fatigue life of it. The predicted fatigue life of the
Element A is 38,271 cycles. As shown in Table 5, there are three
wheels having the failure mode of circumferential crack in the disc
in the accelerated fatigue tests. The fatigue lives of these wheels
are 43,000, 48,000 and 56,000 cycles, respectively. The average life
is 49,000 cycles. Thus the percent error of the predicted life relative
to this average test life is (38,271  49,000)/49,000 = 22%. So this
life prediction of the wheel in the dynamic cornering fatigue tests
has the engineering significance, which indicates that the method
introduced and the assumption that the circumferential crack in
the wheel disc is mainly caused by the radial normal strain history
are reasonable.

7. Conclusions

The linear transient dynamic finite element analysis is applied

here to obtain the transient strain histories of the wheel structure
in the dynamic cornering fatigue test. Thus accurate responses
can only be obtained for the elements whose local stress–strain
characteristics keep linear in the test. It is important to find the lin-
Fig. 12. Normal Y strain histories of the Element A. ear element which is closest to the critical element. And Neuber’s
442 X. Wang, X. Zhang / International Journal of Fatigue 32 (2010) 434–442

rule can be applied to predict the local stress–strain response in References

the critical element from the strain history of the linear element,
based on which the local strain approach can be used to predict [1] SAE handbook. Warrendale, PA: Society of Automotive Engineers, Inc.; 1982. p.
30.05.
the fatigue life of the critical element. And the assumption that [2] Riesner M, I.’DeVries R. Finite element analysis and structural optimization of
the circumferential crack of the wheel is mainly caused by the ra- vehicle wheels. SAE technical paper, No. 830133; 1983.
dial normal strain history is proved reasonable. [3] Karandikar H, Fuchs W. Fatigue life prediction for wheels by simulation of the
rotating bending test. SAE technical paper, No. 900147; 1990.
The amplitudes of the resultant normal Y strain (a form of radial [4] Shang R, Altenhof W, Hu H, Li N. Rotary fatigue analysis of forged magnesium
normal strain) histories of the elements which are located along road wheels. SAE technical paper, No. 2008-01-0211; 2008.
the OY axis can be accurately obtained by only applying the dy- [5] Radaj D. Design and analysis of fatigue resistant welded structures. Abington,
Cambridge: Woodhead Publishing Ltd.; 1990.
namic force component along the axis. So only the Model 1
[6] Dowling N. A discussion of methods for estimating fatigue life. SAE technical
(Fig. 8) is needed to calculate the transient normal Y strain history paper, No. 820691; 1982.
of an element by rotating the wheel to make the element be lo- [7] Reemsnyder H. Constant amplitude fatigue life assessment models. SAE
cated along the OY axis. technical paper, No. 820688; 1982.
[8] Socie D. Variable amplitude fatigue life estimation models. SAE technical paper,
It should be pointed out that this paper does not take the man- No. 820689; 1982.
ufacturing effects into account, which may have significant effects
on the wheel life.