Residual Stresses Due to Rigid

Cylinder Indentation and Rolling

M. Y. Ali
Department of Mechanical Engineering,
University of Michigan,
Ann Arbor, MI 48109-2125
e-mail: mdyusuf@umich.edu
J. Pan1
Department of Mechanical Engineering,
University of Michigan,
Ann Arbor, MI 48109-2125
e-mail: jwo@umich.edu
at a Very High Rolling Load
In this paper, residual stresses due to indentation and rolling of a rigid cylinder on a
finite plate at a very high rolling load with a relative peak pressure of 22 are examined
by two-dimensional plane strain finite element analyses using ABAQUS for the first time. In
the finite element analyses, the roller is modeled as rigid and has frictionless contact with
the finite plate. The geometry of the finite plate and its boundary conditions are assigned
to correspond to those of fillet rolling of crankshafts with the constraint in the rolling
direction. Finite element analyses with different meshes for single indentation on an elas-
tic flat plate under plane strain conditions are first carried out, and the results are bench-
marked with those of the elastic Hertzian solutions to establish the requirement of the
finite element meshes for acceptable numerical results. The results show that the accuracy
of computational results is limited by the discretization of the finite element analysis by a
plot of the contact width as a function of the load. For accurate peak pressure, a total of
at least eight linear elements are needed. Finite element analyses with different meshes
for single indentation on an elasticplastic flat plate under plane strain conditions are
then carried out. The plate material is modeled as an elasticplastic power-law strain
hardening material with a nonlinear kinematic hardening rule for loading and unloading.
The computational results are compared to establish the requirement of the finite element
meshes for acceptable numerical results within 4 mm distance to the rolling surface for
the crankshaft fatigue analyses. The computational results for rolling at the relative peak
pressure of 22 show that the symmetric Hertzian or modified Hertzian pressure distribu-
tion should not be used to represent the contact pressure distribution for rolling simula-
tion, while the computational results for rolling at the relative peak pressure of 5 show
that the symmetric Hertzian or modified Hertzian pressure distribution may be used to
represent the contact pressure distribution for rolling simulation. The computational
results for the rolling case also show a significantly higher longitudinal compressive re-
sidual stress and a lower out-of-plane compressive residual stress along the contact sur-
face when compared to those for the single indentation case. The results suggest that the
effects of rolling must be accounted for when two-dimensional finite element analyses of
crankshaft sections are used to investigate the residual stresses due to fillet rolling of the
crankshafts under the prescribed roller loads. Due to the boundary conditions of the finite
plate, the compressive residual stresses are larger when compared to those when the
boundary conditions of the finite plate are fully relaxed. [DOI: 10.1115/1.4031067]

Keywords: rigid roller, indentation, rolling, contact, residual stress, crankshaft fillet
rolling

1 Introduction crankshaft fillet rolling to estimate the residual stresses using the
commercial finite element code ABAQUS with the kinematic hard-
For automotive crankshafts, fatigue damages are initiated near
ening rule. They also investigated the behavior of crack initiation
the fillets due to stress concentration. The fillet rolling process has
and growth near crankshaft fillets with consideration of residual
been used for years to introduce compressive residual stresses
stresses based on linear elastic fracture mechanics. Chien et al. [2]
near crankshaft fillets and, consequently, to improve fatigue
used the rolling depth of crankshaft fillets after the unloading as
strengths of crankshafts [14]. During the rolling process, the
the calibration parameter in their two-dimensional finite element
material near the roller is loaded into plastic range. Once the roll-
analyses. The rolling depth was determined by the fillet surface
ing load is removed, compressive residual stresses are induced.
profiles obtained from the shadowgraphs taken before and after
The residual stresses can significantly affect the fatigue lives of
the fillet rolling process.
the crankshafts under high cycle fatigue loading conditions.
Spiteri et al. [3] examined the bending fatigue limit of crank-
Therefore, understanding the residual stresses near crankshaft fil-
shaft sections with consideration of residual stresses based on the
lets is important. Previously, researchers estimated the residual
results of three-dimensional finite element analyses. Spiteri et al.
stresses near crankshaft fillets either using two-dimensional or
[3] showed the subsurface axial stress distributions due to single
three-dimensional finite element analyses. Chien et al. [2] con-
indentation and rolling around a disk based on plane strain finite
ducted two-dimensional elasticplastic finite element analyses of
element analyses. However, the effects of single indentation and
rolling on the normal stress components in the circumferential and
1
Corresponding author. out-of-plane directions have not been explained in details. Choi
Contributed by the Manufacturing Engineering Division of ASME for publication
in the JOURNAL OF MANUFACTURING SCIENCE AND ENGINEERING. Manuscript received
and Pan [5] also conducted two-dimensional plane strain finite
October 7, 2014; final manuscript received July 5, 2015; published online September element analyses of a crankshaft section under fillet rolling and
4, 2015. Assoc. Editor: Yannis Korkolis. subsequent bending based on the anisotropic hardening rule of

Journal of Manufacturing Science and Engineering OCTOBER 2015, Vol. 137 / 051005-1
C 2015 by ASME
Copyright V

Downloaded From: http://manufacturingscience.asmedigitalcollection.asme.org/pdfaccess.ashx?url=/data/journals/jmsefk/934680/ on 04/23/2017 Terms of Use: http://www.asme

 Choi and Pan [6] and the nonlinear kinematic hardening rule of
ABAQUS. The results indicated that the compressive hoop stress
based on the anisotropic hardening rule is larger than that based
on the nonlinear kinematic hardening rule within the depth of
2 mm from the fillet surface. Most of the previous analyses for the
fatigue behavior of crankshaft sections are based on the residual
stresses from single indentation of rollers due to the intensive
computational requirements of two- and three-dimensional rolling
simulations using commercially available finite element codes.
Numerous studies were conducted to understand the stress dis-
tributions and deformation patterns near rolling contact surfaces.
Researchers mainly conducted indentation and rolling contact
analyses to examine the pressure distribution on the contact
surface and the stressstrain histories near the contact surfaces for
fatigue and durability analyses of gears, bearings, and railroad
rails at the rolling loads close the elastic and plastic shakedown
limits. Several previous elasticplastic analyses were on the
indentation of an infinitely long cylinder and a sphere on elastic
plastic half spaces. The results of the finite element analyses of
Dumas and Baronet [7], Hardy et al. [8], Lee et al. [9], Kral et al.
[10], and Sinclair et al. [11] showed that the pressure distributions
are quite flat and deviated from the Hertzian distribution for a
cylinder or a sphere indented on a half space of elastic perfectly
plastic materials, elastic linear hardening materials, and actual
nonlinear stressstrain curves from the tensile tests.
Many elasticplastic analyses of rolling were also based on
translating a Hertzian and modified Hertzian pressure distributions
across the surface of an elasticplastic half space. Merwin and
Johnson [12] developed approximate solutions for a rigid cylinder
rolling on a half space of an elastic perfectly plastic material.
They determined the deformation characteristics of the rolling
contact with a p0 =k ratio less than 3.1 for elastic deformation, a
p0 =k ratio less than 4 for shakedown, and a p0 =k ratio larger than
4.0 for continuous plastic deformation, where p0 is the maximum
Hertzian pressure and k is the shear yield stress of the elastic
plastic material. Bhargava et al. [13,14] conducted finite element
analyses of a cylinder rolling on a half space of elastic perfectly
plastic materials for both single and multiple passes. The loads of
the cylinder are represented by a translating Hertzian and modi-
fied Hertzian pressure distribution for p0 =k ratios of 3.5, 4.35, and
5. Bhargava et al. [15] also conducted finite element analyses of a
cylinder rolling on a half space of a rail steel idealized as a linear
kinematic hardening material. The loads from the cylinder are rep-
resented by a translating Hertzian pressure distribution for p0 =k
ratios of 4, 4.5, and 6.2 (which represents the rail wheel load).
Kulkarni et al. [16,17] conducted three-dimensional finite element
analyses of a sphere rolling on a half space of elastic perfectly
plastic materials and elastic linear kinematic hardening plastic
materials. The loads from the sphere are represented by a translat-
ing Hertzian pressure distribution for p0 =k ratios of 4.68 and 6, at
and above the shakedown limit. Kulkarni et al. [18] also
conducted finite element analyses of an elliptical contact loading,
representing a rail wheel load, moving on a half space of a rail
steel idealized as an elastic linear kinematic hardening plastic
material with a p0 =k ratio of 9.2.
Jiang et al. [19] conducted finite element analyses of a sphere
rolling on a steel surface with the Hertzian pressure distributions
to represent the normal and tangential loads with a p0 =k ratio of 6.
A cyclic plasticity model was adopted in Ref. [19] to obtain stabi-
lized residual stresses and strains. It should be mentioned that the
semi-analytical methods were used by Hearle and Johnson [20],
Bower and Johnson [21], McDowell and Moyar [22], and Jiang
and Sehitoglu [23,24] to investigate the cumulative plastic defor-
mation due to rolling and sliding line contact with different cyclic
plasticity theories. The Hertzian pressure distributions were used
in these investigations to represent the rolling loads. Yu et al.
[25,26] presented a direct finite element method for elasticplastic
rolling contact, and demonstrated its application to obtain steady-
state solutions for two- and three-dimensional repeated rolling
contact problems. They also used a moving Hertzian pressure
051005-2 / Vol. 137, OCTOBER 2015
Downloaded From: http://manufacturingscience.asmedigitalcollection.asme.org/pdfaccess.ashx?url=/data/journals/jmsefk/934680/ on 04/23/2017 Terms of Use: http://www.asme
distribution to simulate the rolling process for p0 =k ratios from
4 to 6.
Most researchers used the moving Hertzian pressure distribu-
tion to represent the contact pressure for rolling with p0 =k ratios
ranging from 5 to 9. One exception of these previous works was
from Ref. [27] who used elastic rollers with spring contact
elements for rolling simulations. They modeled the process of a
cylinder running on the strip by finite element analyses, using a
program SEGLA developed by the authors. Their work gives good
insight on the residual stress distribution in the strip due to rolling.
However, the results were obtained for a relatively large wheel
radius (R 457 mm) and the normal force corresponding to the
average load values of the train wheel on the rail, giving a p0 =k
ratio of 4. For the crankshaft rolling, the roller radius is much
smaller and the rolling load is high, with a p0 =k ratio of 22 based
on a finite element analysis of the indentation of the roller on the
crankshaft fillet as reported in Appendix A. It should be noted that
no computational results using a rigid roller for rolling with a high
p0 =k ratio as 22 have been presented in the literature. As indicated
in Refs. [14], determination of residual stresses due to crankshaft
fillet rolling for fatigue life correlation is a critical enabling
technology for developing high power density engines.
This investigation is aimed to quantify the differences between
the residual stresses distributions due to single indentation and
rolling with a very large contact pressure for initial design itera-
tions of crankshaft fillet rolling of a large relative peak pressure of
22 for high power density engines. Crankshafts can be made from
cast iron [2] and forged steel [28]. In this investigation, a typical
steel for crankshafts is considered for finite element analyses. This
investigation also serves as a preliminary study for simulations of
rolling of rectangular bars for residual stresses for fatigue analyses
in the roller burnishing experiments conducted at Caterpillar. The
rolling model and its boundary conditions selected for this study
are based on the geometry of the crankshaft to identify whether
the results of the finite element analyses for single indentation of
crankshaft sections conducted by the previous researchers are suf-
ficient enough to represent the residual stresses distributions for
further strength/fatigue analyses. While a three-dimensional finite
element analyses are more appropriate to obtain the residual
stresses due to fillet rolling, a two-dimensional model is useful for
the initial design iterations. The final goal of this investigation is
to establish the computational scheme for three-dimensional finite
element analyses of fillet rolling at very high loads using commer-
cially available finite element codes. It should be mentioned that
semi-analytical methods with incorporation of the plastic strains
due to contact with efficient contact models and different plastic-
ity models have been developed by a number of researchers
recently for relatively smaller relative peak pressures where the
plastic deformation is not as extensive as those considered in this
investigation. A long list of these papers can be found in Ref. [29]
and will not be discussed here. Furthermore, the effects of the
boundary conditions on the residual stresses are investigated. This
could be of particular interest during crankshaft design process,
for example, lightning holes to reduce crankshafts weight or oil
drilling holes passing very close to the undercut fillets due to a
tighter design space may influence the resulting residual stresses
that would impact crankshafts fatigue lives.
In this paper, a crankshaft fillet rolling along the circumference
of the crankshaft journal is idealized to a two-dimensional flat
plate rolling under plane strain conditions with relevant boundary
conditions. It should be noted that the two-dimensional plate roll-
ing model in the current investigation is different from the thin
sheet rolling process [30,31] in a sense that the plastic zone
extends to a small fraction of the plate thickness under the roller
for the former case, whereas the entire thin sheet section is plasti-
cally deformed under the roller for the latter case. The roller is
assumed to be rigid, and the contact between the roller and the
plate is frictionless. Frictionless condition is assumed since the
crankshaft is rolled in a heavily lubricated environment. The roller
load per unit width for the two-dimensional model is based on
Transactions of the ASME
 that for the corresponding three-dimensional finite element analy-
sis for single indentation. First, elastic plane strain finite element
analyses of single indentation on a flat plate are carried out. For
this case, the plate is modeled as linear elastic. The results of the
elastic finite element analyses are then compared with that of the
elastic Hertzian solutions in order to establish the guideline for
the finite element mesh requirement for acceptable contact pres-
sure and subsurface stress distributions. Next, the plate material is
modeled as an elasticplastic power-law strain hardening material
with the nonlinear kinematic hardening rule of ABAQUS [32] for
loading and unloading. Elasticplastic plane strain finite element
analyses of single indentation of a rigid cylinder on a flat plate are
carried out for the development of the computational model for
rolling simulation. Only a single pass rolling is simulated. Then,
the results of elasticplastic plane strain finite element analyses of
single indentation and rolling are presented and compared. The
plastic zone development and the residual stress distributions on
the plate upper surface and in the subsurface under the roller cen-
ter induced by the single indentation and the rolling are also pre-
sented and compared. In addition, the effects of the boundary
conditions on the residual stresses are presented and discussed.
Finally, some conclusions are made based on the results of this
investigation.
2 Material Model
A typical steel for crankshafts is considered for the plate. The
steel has the Youngs modulus of 207 GPa and the Poissons ratio
of 0.29 as discussed above. The tensile stressplastic strain curve
of the steel is shown in Fig. 1. The nonlinear kinematic hardening
rule of the ABAQUS is assumed based on the tensile stressplastic
strain curve shown in Fig. 1. It should be noted that the steel mate-
rial stressplastic strain curve shown in Fig. 1 is an example of a
Mises material. A general concept and its results are presented in
this paper. Using the guidelines provided in this paper and com-
mercial finite element software packages, users can readily obtain
solutions for cases with very large contact pressures.
The nonlinear kinematic hardening rule in ABAQUS is based on
the Mises yield function. For the loading/unloading/reloading pro-
cess, the yield function can be expressed as
   1=2
 0
 3 0 0 0 0
f rij  aij ; r r  aij rij  aij r0 0 (1)
2 ij
Fig. 1 The tensile stressplastic strain curve of the plate mate-
rial used in the finite element analyses
Journal of Manufacturing Science and Engineering
Downloaded From: http://manufacturingscience.asmedigitalcollection.asme.org/pdfaccess.ashx?url=/data/journals/jmsefk/934680/ on 04/23/2017 Terms of Use: http://www.asme
where rij is the stress tensor, r0ij is the deviatoric stress tensor, aij
represents the center of the yield surface, a0ij is the deviatoric part
of aij , and r0 represents the size of the yield surface. Here, r0ij and
a0ij are defined, respectively, as
1
r0ij rij  rkk dij (2)
3
1
a0ij aij  akk dij (3)
3
where dij is the Kronecker delta.
The associated plastic flow is expressed as
 
@f rij  aij ; r0
e_pij e_ p (4)
@rij
where e_pij represents the plastic strain rate tensor and e_ p is the
equivalent plastic strain rate defined as
 1=2
2 p p
e_ p e_ e_ (5)
3 ij ij
In ABAQUS, the nonlinear kinematic hardening rule is defined to
be an additive combination of a purely kinematic term of the Zie-
gler linear hardening law and a relaxation term which introduces
the nonlinearity. The evolution of aij is expressed as
C p 
a_ ij e_ rij  aij  ce_ p aij (6)
r0
where r0 is constant and represents the size of the initial yield sur-
face, aij represents the center of the yield surface, e_ p is again the
equivalent plastic strain rate, and C and c are material parameters
that must be calibrated from cyclic test data. Here, C is the initial
kinematic hardening modulus and c determines the rate at which
the kinematic hardening modulus decreases with increasing plas-
tic deformation. When C and c are zero, the model reduces to the
isotropic hardening rule. When c is zero, the Ziegler linear hard-
ening law is recovered. Note that, for the nonlinear kinematic
hardening rule of ABAQUS, the option of half cycle for data type
was adopted in order to use the cyclic stressstrain curve of the
material as the input stressstrain data.
3 Rolling Load
The crankshaft rolling process of interest involves 12 rotations
of a crankshaft in a rolling machine where the load is increased
linearly up to the full rolling load for the first three rotations of the
crankshaft. The load is then kept as a constant level for the next
six rotations. Finally, the load is decreased linearly to zero for the
final three rotations. However, in this investigation, only one pass
of the rolling load is considered for a two-dimensional model of a
rigid cylinder rolling on a finite flat plate, where the roller is
moved down with a full rolling load at the starting position, rolled
a full rolling length to the end point, and finally moved up to
release the load. The details of the simplification from the three-
dimensional rollercrankshaft model to the two-dimensional
roller-plate model are discussed in Sec. 4. The full rolling load for
the crankshaft fillet rolling is assumed to be 5700 N applied to the
primary roller with the direction that is 34 deg inclined from the
vertical direction. The details of the geometric arrangement of the
primary and secondary rollers and crankshafts can be found in
Refs. [33] and [34]. The details to convert the 5700 N load for the
crankshaft fillet rolling to the load per unit thickness for plane
strain plate rolling are outlined in Appendix A. After the conver-
sion, a load per unit thickness of 3847.0 (N/mm) was used in the
OCTOBER 2015, Vol. 137 / 051005-3
 plane strain plate rolling model as the roller load for single inden-
tation and rolling.
4 Computational Model Development for Indentation
and Rolling
Figure 2(a) shows a schematic view of the arrangement of a
crankshaft journal, a primary roller, a portion of a secondary
roller, and a support roller during the rolling process. The dashed
line HB shown in Fig. 2(a) represents the primary roller angle h
with respect to the vertical line. Along the line HB, the primary
roller diameter is the largest and represents the roller diameter. A
line AB is chosen from the line HB to represent the height of the
crankshaft material. At the instant shown in Fig. 2(a), a plane par-
allel to the flat bottom surface of the primary roller and passing
through the line HB is selected as a representative roller plane.
Along the perpendicular direction to this plane, the contact
pressure distribution between the roller and the crankshaft fillet
surfaces is almost flat as shown and explained in Appendix A.
It should be noted that the fatigue crack near the fillet in the
crankshaft sections under bending loads typically occurs near
angle h from the vertical line. In an actual rolling process, the roll-
ers are held in the roller cartridges while the crankshaft journal
rotates. This is equivalent to consider that the primary roller is
moving and steering along the crankshaft undercut with respect to
the fixed crankshaft. At any instant, due to the relative motion of
the primary roller with respect to the crankshaft journal, it can be
considered that the dashed line AB is revolving with the primary
roller and would essentially form a conical surface upon one revo-
lution of the crankshaft journal as illustrated by the conical sur-
face in Fig. 2(b). In the figure, the primary roller is illustrated and
the crankshaft is illustrated by lighter lines.
Figure 2(c) shows an auxiliary view of the primary roller and
the conical surface shown in Fig. 2(b). It should be noted that
Figs. 2(b) and 2(c) are for illustrative purpose only and not for a
true engineering drawing. At any instant when viewed at the pri-
mary roller plane as shown in Fig. 2(c), the contact surface
between the roller and the edge of the conical surface would be
Fig. 2 (a) A schematic view of the arrangement of a crankshaft journal, a primary roller, a portion of a secondary roller,
and a support roller during the rolling process, (b) an illustration of the conical surface generated by the dashed line AB
upon one revolution of the crankshaft journal, and (c) an auxiliary view of the primary roller and the conical surface
shown in (b)
051005-4 / Vol. 137, OCTOBER 2015
Downloaded From: http://manufacturingscience.asmedigitalcollection.asme.org/pdfaccess.ashx?url=/data/journals/jmsefk/934680/ on 04/23/2017 Terms of Use: http://www.asme
approximately perpendicular to the dashed line AB. Upon
unwrapping, the conical surface would project on to a two-
dimensional plane as a strip with two curved edges where one
edge contacting the roller. During the rolling, the shaft material is
fed under the roller. From the roller perspective, the curved edge
is approximated as a flat plate for this study since the circumfer-
ence of the curved edge is quite large compared to the roller
radius. For simplicity, the length AB is assumed to be the radius
of the crankshaft journal. The crankshaft journals are supported
by the support rollers during rolling as schematically shown in
Fig. 2(a). As a result, the translational degrees-of-freedom of the
central portion of crankshaft journals can be considered negligi-
ble. This zone is marked in Fig. 2(c). This assumption will allow
fixing the boundary conditions at the bottom of the flat plate as
considered in Sec. 5.
5 Finite Element Model
The finite element analyses were performed using the implicit
solver of commercial finite element code ABAQUS. Figure 3(a)
shows a two-dimensional finite element model of a rigid cylinder
rolling on a finite flat plate. The Cartesian coordinate system is
also shown in the figure. Two-dimensional plane strain four-
noded CPE4R elements of size 0.4 mm  0.4 mm are used in this
model. The plane strain model considered has unit length in the
out-of-plane direction. The selection of the sizes of the finite ele-
ments to simulate single indentation and rolling follows the guide-
lines that will be presented later. The bottom of the plate is fixed,
and the displacements on both end surfaces AB and CD of the
plate in the x direction are constrained. The length of the plate is
300 mm with the rolling start point on the upper surface at
x 62 mm and the rolling end point at x 216.264 mm to elimi-
nate possible end effects. One goal of this investigation is to use
the results obtained here to examine the residual stresses in crank-
shafts after fillet rolling. The diameter of the rigid cylinder is
14 mm which represents the diameter of the primary roller for the
crankshaft rolling as discussed in Appendix A. The total rolling
length 154.264 mm and the plate thickness 24.552 mm are
Transactions of the ASME
 corresponding to the circumference and the radius of a crankshaft
journal of interest, respectively.

5.1 Mesh Sensitivity for Elastic Finite Element Analyses. For

elastic contact analyses, only the results of two finite element
models are presented here. The finite element model mesh-1
(coarse mesh) has the element size of 0.4 mm, and the finite ele-
ment model mesh-2 (fine mesh) has element size of 0.1 mm. It
should be mentioned that the subsurface stresses within 4 mm dis-
tance to the contact surface are of interest for residual stress deter-
mination and the fatigue analyses as detailed in Ref. [2]. This
length scale is the major factor to select the mesh size for the cur-
rent computational study. Many mesh designs were considered in
the initial stage of the research. Only the results for the two repre-
sentative mesh designs are presented here in details. In Appendix
B, the results based on finer meshes with the element sizes of
0.05 mm and 0.025 mm are compared with the results based on
mesh-1 and mesh-2 for interested readers. It should be noted that
the fine mesh results are compared with the Hertzian solution to Fig. 4 A comparison of the contact widths between the flat
establish the acceptability of the fine mesh adopted. Then the plate and the rigid roller obtained by the elastic Hertzian solu-
elasticplastic results between the fine mesh and coarse mesh are tion and the elastic finite element analysis as functions of the
compared to establish the acceptability of the coarse mesh used indentation load up to the maximum load of 3847 N/mm based
for the study. Magnified views of the finite element models for on the fine mesh-2
mesh-1 and mesh-2 are shown in Figs. 3(b) and 3(c), respectively.
The results based on the finite element model mesh-2 showed bet- the in-plane stresses r11 and r22 as r33 r11 r22 .
ter correlations with the elastic Hertzian solutions and the results The von Mises stress rMises is expressed as rMises
are presented here. Figure 4 shows a comparison of the contact q
widths (2a) between the flat plate and the rigid roller obtained by r11  r22 2 r22  r33 2 r33  r11 2 6r212 =2. The
the elastic Hertzian solution and the elastic finite element analysis in-plane stresses based on the elastic Hertzian solution are plotted
as a function of the indentation load up to the maximum value of in Fig. 5. The elastic Hertzian solutions will be used to benchmark
3847 N/mm used in this investigation.p Note that the Hertzian solu-
the results of the elastic finite element analyses. At the contact
tion for the half contact width is a 4PR1   2 =pE, where P point, the values of the stresses r11 and r22 obtained from the
is the load, R is the roller radius, E and  are the Youngs modulus finite element analysis are 23% and 11% lower than the elastic
and the Poissons ratio, respectively, for the elastic plate. As the Hertzian solution. This may be attributed to the mesh discretiza-
indentation proceeds with the increasing load, the contact width tion as discussed in Ref. [37]. Away from the contact surface, the
increases. However, until the next node makes contact, the contact stress r22 obtained from the finite element analysis is within 3%
width (2a) remains the same in the elastic finite element analysis of the elastic Hertzian solution. The shear stress component r12
as shown in the figure. should be zero directly under the roller center. However, the
Figure 5 shows the distributions of the stresses directly under stresses are obtained from the nodal points which are not exactly
the roller in the y direction from the top surface for the top under the center of the roller. Therefore, very small values of r12
5 mm of the plate for indentation by the rigid roller at the full are shown in Fig. 5 due to a small offset of these nodal points
rolling load of 3847 N/mm. It should be mentioned again that the
plate is modeled to be linear elastic. For the elastic Hertzian
solution, at the contact interface r11 r22 px. The expres-
sion for the contact pressure distribution px can be obtained
according to Ref. [35]. The in-plane stresses r11 , r22 , and r12 at a
point x; y can be obtained according to Ref. [36]. Under plane
strain conditions, the out-of-plane normal stress r33 is related to

Fig. 5 A comparison of the results from the finite element anal-

Fig. 3 (a) A two-dimensional finite element model of a rigid ysis and the elastic Hertzian solution for contact between the
cylinder rolling on a finite flat plate and magnified views of the elastic plate and the rigid roller directly under the roller. The
finite element models for (b) mesh-1 with the element size of stress distributions in the 2y direction from the top surface for
0.4 mm and (c) mesh-2 with the element size of 0.1 mm the top 5 mm of the plate for single indentation.

Journal of Manufacturing Science and Engineering OCTOBER 2015, Vol. 137 / 051005-5

Downloaded From: http://manufacturingscience.asmedigitalcollection.asme.org/pdfaccess.ashx?url=/data/journals/jmsefk/934680/ on 04/23/2017 Terms of Use: http://www.asme

 from the centerline of the roller. Accordingly, the elastic Hertzian distributions directly under the roller center in the y direction
solutions based on the appropriate value of the x coordinate for from the top surface of the elasticplastic plate when the roller is
these nodal points are also plotted in Fig. 5 for comparison. How- down for the cases of mesh-1 and mesh-2. For the indentation
ever, at the contact point, the values of the stresses r11 and r22 case, the peak values of the compressive stress components r11 ,
obtained from the finite element analysis based on mesh-1 are
68% and 43% lower than the elastic Hertzian solution. Thus, finite
element model mesh-2 with the element size of 0.1 mm is found
reasonable for the elastic single indentation with the given high
indentation load. Figure 6 shows the contact pressure distributions
on the plate surface based on the elastic Hertzian solution and the
results of the elastic finite element analysis due to indentation by
the rigid roller based on the finite element model mesh-2 with the
element size of 0.1 mm at the load of 3847 N/mm. The contact
pressure distribution obtained by the elastic finite element analysis
is in reasonably good agreement with the elastic Hertzian solution
except near the end of the contact zone due to the mesh discretiza-
tion. It should also be noted that a slight asymmetry of the contact
pressure distribution is not due to boundary effects, since the
length of the plate and indentation or rolling start point is chosen
to eliminate possible end effects but rather due to the roller center
being not directly aligned to a particular node. It should be noted
that for the elastic finite element analysis the results based on both
linear element CPE4 and CPE4R were identical. Based on this
parametric study of the contact stress distribution at the different
load levels, it is suggested that for the linear element CPE4 or
CPE4R a total of four to five finite elements for the half contact
width are needed to obtain acceptable contact pressure and subsur-
face stress distributions.

5.2 Mesh Sensitivity for ElasticPlastic Finite Element

Analyses. Similar to the elastic finite element analyses, a mesh
sensitivity study was conducted for simulation of the rigid cylin-
der indenting on the elasticplastic finite plate. The finite element
models mesh-1 and mesh-2 with the element sizes of 0.4 mm and
0.1 mm are again used in this study. In Appendix B, the results
based on the finer mesh with the element sizes of 0.025 mm are
compared with the results of mesh-2 for interested readers. The
results based on mesh-1 and mesh-2 are compared and presented
here. Figure 7(a) shows the contact pressure distributions for both
mesh-1 and mesh-2. The elastic Hertzian solution is also shown
for reference. As shown in the figure, the contact pressure distri-
butions for both mesh-1 and mesh-2 are in agreement with each
other except near the boundary of the contact region due to the
difference in mesh size, and the difference of the areas under the
curves is within 9%. For mesh-1, there are four contact elements
for the half contact width. Figure 7(b) shows the stress

Fig. 6 The contact pressure distributions on the plate surface

based on the elastic Hertzian solution and the results of the Fig. 7 (a) The contact pressure distributions, (b) the stress dis-
elastic finite element analysis based on the fine mesh-2 with tributions, and (c) the residual stress distributions in the 2y
the element size of 0.1 mm due to single indentation by the rigid direction from the top surface based on mesh-1 and mesh-2 for
roller single indentation on the elasticplastic plate

051005-6 / Vol. 137, OCTOBER 2015 Transactions of the ASME

Downloaded From: http://manufacturingscience.asmedigitalcollection.asme.org/pdfaccess.ashx?url=/data/journals/jmsefk/934680/ on 04/23/2017 Terms of Use: http://www.asme

 r22 , and r33 occur at the contact surface directly under the roller 6 Stress Distributions Due to Plate Rolling
center. With the value of the stress component r33 between those
Based on the coarse mesh-1, the entire rolling simulation was
of the stress components r11 and r22 , the distributions of r11 , r22 ,
carried out. However, most computational results presented are
and r33 are nearly parallel until the end of the plastic zone. Figure
still based on the fine mesh-2. Figure 8(a) shows the Mises stress
7(c) shows a comparison of the residual stresses distributions in
distribution and the plastic zone size and shape as the roller is
the y direction from the top surface of the plate for both mesh-1
moved down at the full rolling load of 3847 N/mm based on
and mesh-2 after the roller load for single indentation is removed.
mesh-1. The darker inner circular area represents the material that
The previous study of Chien et al. [2] indicates that the stress dis-
is yielded, and hence the dark red area represents the plastic zone
tribution within 4 mm distance to the fillet surface is quite impor-
size and shape. The total contact width is about 6a for the rolling.
tant to establish the crack initiation, growth, and arrest during the
As the rolling continues, the Mises stress contours reach almost a
typical crankshaft bending load and residual stress conditions.
steady state at a distance of nearly 6:5a from the start point, as
Therefore, the subsurface residual stresses are very important for
compared to 6:4a obtained by Bhargava et al. [13] and the steady-
calculating crankshafts bending fatigue life. The results based on
state plastic zone extends to approximately 5a. Here, a is the half
mesh-1 and mesh-2 show that the subsurface stresses and residual
contact width based on the elastic Hertzian theory of contact. For
stresses based on both meshes were in good agreement except the
the given roller radius, the half contact width a is obtained as
stresses on the contact surface with acceptable differences. For the
0.3895 mm. Figure 8(b) shows the Mises stress contours and plas-
elasticplastic plate material, as the load increases and the mate-
tic zone size and shape (shown as darker zone closer to the rolled
rial under the contact surface becomes plastic, the material acts
surface) when the roller is rolled to the half rolling length. Figure
much softer and the contact width increases. Therefore, more
8(c) shows the Mises stress contours and plastic zone size and
nodes come in contact with the roller and the contact load is dis-
shape (shown as darker zone closer to the rolled surface) after the
tributed to more elements. Therefore, mesh-1 can give reasonable
roller is moved up and the rolling is completed.
contact pressure and subsurface stress distributions and was used
Figure 9 shows again the contact pressure distribution between
for the subsequent rolling simulation for saving computational
the roller and the plate based on the fine mesh-2 from the
time. It seems that the guideline of four linear elements for a half
elasticplastic finite element analysis at the maximum load of
contact width is still applicable to the elasticplastic finite element
3847 N/mm due to single indentation. The elastic Hertzian solu-
analyses.
tion is also plotted for comparison. The results of the
elasticplastic finite element analysis show that the contact pres-
sure distribution is no longer Hertzian pressure distribution but
rather flatter and wider, and they are in general agreement with
the findings of Ref. [7]. This is due to the subsurface plastic flow
that leads to a flatter and wider pressure distribution. Figure 9 also
shows a representative contact pressure distribution between the
roller and the plate during rolling after the steady state of the con-
tact pressure distribution is reached based on the fine mesh-2. The
results confirm that the contact pressure distribution is neither
Hertzian pressure distribution nor symmetric. A finer mesh size of
0.1 mm was used for this purpose to capture the detailed pressure
distribution. It should be noted that the computational time for the
finite element analysis based on the 0.1 mm mesh is prohibitively
high to conduct a full rolling analysis.
Figure 10 shows the stress distributions of the upper surface
during rolling at the half rolling length. During the rolling process,
depending on the rolling depth, a zone of material in front of the
roller is being pushed and hence results in a nonsymmetric contact
pressure distribution. At the same time, a small zone of material
behind the roller is being pulled by the roller and is less compres-
sive shown as a peak of the longitudinal stress r11 in Fig. 10. As
the roller moves further, the very thin top layer with the less

Fig. 8 The Mises stress distributions and the plastic zone sizes
and shapes represented by the darker zone closer to the rolled
surface due to the rolling when (a) the roller is moved down at the
full rolling load, (b) the roller is rolled to the half rolling length, and Fig. 9 Contact pressure distributions for single indentation
(c) the roller is moved up after the rolling is completed and during rolling based on a finer mesh size of 0.1 mm

Journal of Manufacturing Science and Engineering OCTOBER 2015, Vol. 137 / 051005-7

Downloaded From: http://manufacturingscience.asmedigitalcollection.asme.org/pdfaccess.ashx?url=/data/journals/jmsefk/934680/ on 04/23/2017 Terms of Use: http://www.asme

 compressive stress returns to a more compressive stress state due
to a relatively larger subsurface zone that is in a compressive
stress state (Fig. 12(b)). The roller needs to always push the rise-
up material in front of it during rolling. Therefore, a very thin top
layer on the rolling surface passes through a large compressive-
less compressive-more compressive loading cycle. This loading
pattern causes a relatively less compressive longitudinal stress in
the thin top layer than that of the subsurface (Fig. 12(b)). Also this
loading pattern is different from the single indentation case where
the material is loaded by only a compressive loading.
Figure 11(a) shows the stress distributions directly under the
roller center in the y direction from the top surface when the
roller is down for the cases of single indentation and during roll-
ing as the roller is rolled to the half rolling length based on the
coarse mesh-1. For the single indentation, the peak compressive
values of the stress components r11 , r22 , and r33 occur at the con-
tact surface directly under the roller center. During rolling, the
peak compressive values of the stress components r11 , r22 , and
r33 still occur at the surface but are located in front of the roller.
In the rolling case, directly under the roller center, the peak com-
pressive value of the stress component r11 occurs at a depth to the
contact surface and can be explained in the following. Figure
11(b) shows the contours of the longitudinal stress component r11
when the roller load is applied for both the single indentation and
rolling cases. For the single indentation, the r11 contour lines are
symmetric about the roller center, whereas for the rolling the com-
pressive zone is moving to the front of the roller such that it is
leaving a compressive stress trail zone behind it and is no longer
symmetric about the roller center line. Therefore, the peak value
of the compressive stress r11 for the rolling case occurs in the sub-
surface directly under the roller center. The distribution of the
stress component r11 under the roller center during the rolling is
different from that of the single indentation case. The peak value
of the compressive stresses r22 and r33 for the rolling case occurs
at the contact surface directly under the roller center as in the sin-
gle indentation case.
Note also that the stresses are obtained from the nodal points
which are not exactly under the roller center. Therefore, for the
single indentation case, very small values of r12 exist due to a
small offset of these nodal points from the roller centerline as
shown in Fig. 11(a). The results of the finite element analysis
show that for the rolling case the zero value of r12 occurs at a dis-
tance approximately 2:16a in front of the roller. The shear stress
component r12 shows the peak value at the location of the peak
compressive stress r11 , and thus indicating the rolling effect on
the subsurface layer. Directly under the roller center, the magni-
tudes of the stress components r11 , r22 , and r33 during rolling are
Fig. 10 The stress distributions of the upper surface during
rolling at the half rolling length
051005-8 / Vol. 137, OCTOBER 2015
Downloaded From: http://manufacturingscience.asmedigitalcollection.asme.org/pdfaccess.ashx?url=/data/journals/jmsefk/934680/ on 04/23/2017 Terms of Use: http://www.asme
24%, 10%, and 9% lower at the contact surface than those of the
single indentation case.
Figure 12(a) shows a comparison of the residual stresses of the
single indentation and rolling cases after the roller load is
removed. The results for the single indentation case show a small
tensile longitudinal residual stress (r11 ) at the contact surface
(Figs. 12(a) and 12(b)). The results for the rolling case show a
compressive residual stress component r11 at the contact surface.
The results also showed a significantly higher subsurface
compressive residual stress component r11 after rolling when
compared to that for single indentation. Overall, the distributions
of the out-of-plane residual stress components r33 for the single
indentation and the rolling cases are similar, except within a few
millimeters of the contact surface where the residual stress r33 on
the surface for the rolling case is lower than that of the single
indentation case. The residual stress component r12 has small
values near the surface for the single indentation case, and the re-
sidual stress components r12 and r22 have small values near the
surface for the rolling case. These small values could not be
shown clearly in Fig. 12(a) due to the scale of the figure. Figure
12(b) shows the distributions of the residual longitudinal stress
r11 after the roller load is removed for the single indentation and
rolling cases. Figure 13 shows the residual stress distributions of
the upper surface after the rolling is completed. These plots show
that residual stress component r11 (along the plate length) is the
dominant component, and the out-of-plane residual stress compo-
nent r33 is also significant.
Now consider the scenario of using a two-dimensional plane
strain finite element analysis of a crankshaft section under single
indentation with loading/unloading phases similar to those in Ref.
[2]. Note that r11 and r33 of the flat plate single indentation/
rolling from the current study can be related to the out-of-plane
stress rzz and the hoop stress rhh , respectively, for the crankshaft
section based on the cylindrical coordinate system used by Chien
et al. [2]. Since the residual out-of-plane stress r33 for rolling is
less than that for single indentation within 1 mm from the surface
as shown in this investigation, the use of the hoop stress rhh
(which corresponds to the out-of-plane stress r33 ) based on the
coordinate system of the roller for the single fillet indentation of
the roller to the crankshaft section as an approximation for the fil-
let rolling is not conservative under prescribed loading conditions.
Note that Chien et al. [2] matched their two-dimensional finite
element rolling depth after the unloading of the roller with the fil-
let surface profiles measured by the shadowgraphs taken before
and after the fillet rolling process without direct use of the actual
rolling load and thus may have obtained a good correlation. How-
ever, using the fillet surface profiles to calibrate the two-
dimensional finite element analysis may not be available upfront
in many actual design iteration processes. The current study sug-
gests that the effect of rolling should be taken into consideration
to interpret the results of finite element analyses based on the two-
dimensional crankshaft sections under single indentation with
loading/unloading phases using the rolling loads.
7 Effects of Boundary Conditions on Residual Stresses
The rolling gives a relatively longer longitudinal length of the
plate that is plastically deformed when compared to that under
the single indentation. The compatibility and equilibrium between
the plastic and the elastic layers in the plate result in a compres-
sive stress in the top deformed (rolled) layer after the roller load is
removed. For a plate with a finite dimension in this investigation,
the boundary conditions have significant influence on the residual
stresses and are examined here. It should be mentioned that many
previous investigations on rolling and indentation have their
boundary conditions setting up for those of an elastic half space.
The physical reasons for the constrained displacement in the x
direction for the boundary surfaces AB and CD and the con-
strained displacements in the x and y directions for the boundary
Transactions of the ASME
 Fig. 11 (a) The stress distributions directly under the roller center in the 2y direction from the
top surface for single indentation and during rolling and (b) the contours of the longitudinal
stress component r11 for single indentation and during rolling

surface BC come from the fact that the plate is used to represent
the cross section of a crankshaft journal (referring to Fig. 3).
For a half space subjected to a single indentation, one might
expect that the longitudinal residual stress component directly
under the roller center may be in self equilibrium without any
effect from the boundary. However, for a finite plate with the
dimensions relatively large compared to the roller dimension and
the contact width, the boundary conditions have effects on the
residual stresses. It is found that the distribution of the longitudi-
nal residual stress component r11 along the roller centerline (Fig.
12(a)) is no longer in self equilibrium by itself due to the change
in the length of the top layer of the plate by rolling but the
equilibrium is achieved together with the reaction forces at the
boundary surfaces. It should be noted that the finite element model
is definitely in equilibrium.
For a finite plate, the residual stresses in the plastically
deformed zone are also affected by the boundary conditions. For
the single indentation, consider a hypothetical condition where all
the boundary conditions are released. Figure 14(a) shows the
effects of the boundary conditions on the residual stresses in the
plastically deformed zone for the single indentation case. After
the boundary conditions are released, the changes in the residual
stresses are recorded for a few nodes marked as node-1, node-2,
and node-3 in the top portion of the plate located directly under

Fig. 12 (a) The residual stress distributions in the 2y direction from the top surface after the
roller load is removed for the single indentation and rolling cases and (b) the distributions of
the residual longitudinal stress r11 after the roller load is removed for the single indentation and
rolling cases

Journal of Manufacturing Science and Engineering OCTOBER 2015, Vol. 137 / 051005-9

Downloaded From: http://manufacturingscience.asmedigitalcollection.asme.org/pdfaccess.ashx?url=/data/journals/jmsefk/934680/ on 04/23/2017 Terms of Use: http://www.asme


Fig. 13 The residual stress distributions of the upper surface
after the rolling is completed
the roller center and closer to the indentation center as shown in
Fig. 14(a). Note that the rigid roller is shown in the figure to show
the locations of these nodes. After the boundary conditions are
released, the value of the residual longitudinal stress r11 becomes
around 2.7 times more tensile at node-1 on the top surface. As
shown in the figure, the changes are not so drastic at node-2 and
node-3.
Figure 14(b) shows the effects of the boundary conditions on
the residual stresses in the plastically deformed zone for the roll-
ing case. For the rolling case, the boundary conditions are also
important, since any constraint will restrict the plate to take any
deformed shape due to the compatibility and equilibrium of the
plastic and the elastic layers after the roller load is removed. For
Fig. 14 The effects of the boundary conditions on the residual stresses at the marked locations (a) for the sin-
gle indentation case and (b) at the half rolling length for the rolling case
051005-10 / Vol. 137, OCTOBER 2015
Downloaded From: http://manufacturingscience.asmedigitalcollection.asme.org/pdfaccess.ashx?url=/data/journals/jmsefk/934680/ on 04/23/2017 Terms of Use: http://www.asme
the rolling case, after the rolling load is removed and the displace-
ment boundary conditions on both the boundary surfaces AB and
CD are released successively, the changes in the residual stresses
are recorded for a few nodes marked as node-1, node-2, and node-
3 in the top portion of the plate located at the half rolling length as
shown in Fig. 14(b). Note that the rigid roller is shown to show
the locations of these nodes relative to the starting of the rolling.
At node-1 located on the top surface, the absolute value of the
compressive residual longitudinal stress (r11 ) is reduced by 17%
and the stress becomes less compressive. Next, the fixed boundary
conditions on the bottom surface BC are also removed so that the
plate can deform freely in the longitudinal direction. The value of
the compressive residual longitudinal stress is further reduced by
54% at node-1 on the top surface at the location of the half rolling
length. That results in a total drop of 62% in the value of the com-
pressive longitudinal stress component due to the release of all the
boundary conditions. For the rolling case, after all the boundary
conditions are released, the plate takes a convex shape and the
values of the compressive residual stresses reduce
8 Discussions
In this paper, residual stresses due to single indentation and
rolling on a finite plate at a very high rolling load with a p0 =k ratio
of 22 are investigated by two-dimensional plane strain finite ele-
ment analyses. The rolling loads in many previous studies were
set at much lower p0 =k ratios, and the rolling loads were repre-
sented by the Hertzian pressure distributions. Therefore, bench-
marking finite element analyses of indentation and rolling were
also carried out for a p0 =k ratio of 5 to show the effects of using a
rigid roller on the contract pressure distributions. Figure 15 shows
the contact pressure distributions for single indentation and during
rolling for the p0 =k ratio of 5. The elastic Hertzian solution is also
plotted for comparison. As shown in the figure, the contact pres-
sure distributions are not symmetric. The contact pressure distri-
bution for single indentation is quite different from the elastic
Transactions of the ASME
 when the boundary conditions of the finite plate are fully
relaxed.

Acknowledgment
The encouragement of Dr. Jagadish Sorab of Ford Motor Com-
pany to conduct this research work is greatly appreciated. Helpful
discussions with Simon Ho of Chrysler LLC are also greatly
appreciated. The support of this work by Ford Motor Company
and Chrysler LLC is greatly appreciated.

Appendix A: Conversion of the Roller Load for the

Fig. 15 Contact pressure distributions for single indentation
and during rolling for a p0 =k ratio of 5
Plane Strain Model
Figure 16(a) shows a portion of a three-dimensional finite ele-
ment model for a crankshaft section under fillet rolling. A portion
of the secondary roller on one side of the symmetry line is also
modeled by finite elements. Figure 16(b) shows a magnified view
near the contact zone between the crankshaft section and the pri-
mary roller. As shown in Fig. 16(b), very fine elements are used

Hertzian solution and is also quite different from that for rolling.
In fact, the contact pressure distribution for rigid cylinder rolling
appears to be more similar to the elastic Hertzian distribution with
some asymmetry. Although this near Hertzian distribution for roll-
ing is on a finite plate but can serve as a validation using the
Hertzian pressure to represent the load of a rigid cylinder rolling
on a half space at this p0 =k ratio of 5. The asymmetry of the near
Hertzian pressure distribution appears to be much more than that
reported in Ref. [27].

9 Conclusions
In this paper, two-dimensional plane strain finite element analy-
ses are used to investigate the resulting residual stresses due to
very high rolling load with a relative peak pressure of 22 for the
first time. The Hertzian solutions are used as a guidance to
develop suitable finite element models. Then the computational
models are used to quantify the effects of indentation and rolling
on the resulting residual stresses due to very high loads. The
following conclusions were made based on this investigation:
(1) The results show that the accuracy of computational results
is limited by the discretization of the finite element analysis
by a plot of the contact width as a function of the load. For
accurate peak pressure, a total of at least eight linear ele-
ments are needed. For the rolling case, the computational
results show that the stress distributions under the roller
reached the steady-state conditions after rolling about three
times the contact width distance.
(2) The computational results for rolling at the relative peak
pressure of 22 show that the symmetric Hertzian or modi-
fied Hertzian pressure distribution should not be used to
represent the contact pressure distribution for rolling simu-
lation, while the computational results for rolling at the rel-
ative peak pressure of 5 show that the symmetric Hertzian
or modified Hertzian pressure distribution to represent the
roller may be used to represent the contact pressure distri-
bution for rolling simulation.
(3) The rolling case shows a significantly higher longitudinal
compressive residual stress and a lower out-of-plane com-
pressive residual stress along the contact surface than the
single indentation case.
(4) The results suggest that the effects of rolling must be
accounted for to investigate the residual stresses due to fil-
let rolling of the crankshafts under prescribed loading Fig. 16 (a) A portion of a three-dimensional finite element
conditions. model of a crankshaft section under fillet rolling and (b) a mag-
(5) Due to the boundary conditions of the finite plate, the com- nified view near the contract zone between the crankshaft sec-
pressive residual stresses are larger when compared to those tion and the primary roller

Journal of Manufacturing Science and Engineering OCTOBER 2015, Vol. 137 / 051005-11

Downloaded From: http://manufacturingscience.asmedigitalcollection.asme.org/pdfaccess.ashx?url=/data/journals/jmsefk/934680/ on 04/23/2017 Terms of Use: http://www.asme

 near the fillet region of the crankshaft section. The crankshaft sec-
tion and rollers are mainly modeled by hexagonal elements
(C3D8) in ABAQUS. The primary and secondary rollers also have
fine elements near the contact zone. The nonlinear kinematic hard-
ening material model of ABAQUS based on the tensile stressplastic
strain curve shown in Fig. 1 is used for the crankshaft section in
the finite element analyses. The diameters of the primary and sec-
ondary rollers are taken as 14 mm and 80 mm, respectively. The
primary roller load is taken as 5700 N. The material properties of
the primary and secondary rollers are obtained from Refs. [33]
and [34].
Figure 17(a) shows the contact pressure distribution between
the crankshaft section and the primary roller from the three- Fig. 18 Magnified views of the finite element models near the
dimensional finite element analysis when the roller load is rolling surfaces for (a) mesh-1 with the element size of 0.4 mm,
applied. Figure 17(b) shows the contact pressure distributions (b) mesh-2 with the element size of 0.1 mm, (c) mesh-3 with the
along the two circumferential paths, paths 1 and 2, as marked in element size of 0.05 mm, and (d) mesh-4 with the element size
Fig. 17(a). The results in Fig. 17(b) indicate that the contact pres- of 0.025 mm
sure distributions are quite similar along the two circumferential
paths. Based on the area under the contact pressure distribution of
path 2, the load per unit thickness is calculated as 3847.0 (N/mm) which is the rolling load used for the simulations of indentation
and rolling under plane strain conditions in this investigation.
Figure 17(c) shows the contact pressure distributions along the
path 3 that is perpendicular to the rolling direction, as marked in
Fig. 17(a). The results in Fig. 17(c) indicate that the contact pres-
sure distribution along the path is quite flat in the middle portion
between u and v as marked in the figure and the inset. With the
near constant contact pressure distribution in the middle portion
for path 3 as shown in Fig. 17(c) and the similar contact pressure
distributions as shown in Fig. 17(b), the load for the three-
dimensional rollercrankshaft model can be converted approxi-
mately to the rolling load for the two-dimensional plane strain
roller-plate model.

Appendix B: Additional Results for Mesh Sensitivity

Study
For completeness of the mesh sensitivity study, in addition to
the finite element model mesh-1 with the element size of 0.4 mm
and mesh-2 with the element size of 0.1 mm, the results for mesh-
3 with the element size 0.05 mm and mesh-4 with the element size
0.025 mm are presented here. Magnified views of the finite ele-
ment models for mesh-1, mesh-2, mesh-3, and mesh-4 near the
rolling surfaces are shown in Figs. 18(a), 18(b), 18(c), and 18(d),

Fig. 19 A comparison of the results from the finite element

Fig. 17 (a) The contact pressure distribution between the analyses and the elastic Hertzian solution for contact between
crankshaft section and the primary roller, (b) the contact pres- the elastic plate and the rigid roller directly under the roller. The
sure distributions for paths 1 and 2 as marked in (a), and (c) the stress distributions in the 2y direction from the top surface for
contact pressure distributions for path 3 as marked in (a) the top 2 mm of the plate for single indentation.

051005-12 / Vol. 137, OCTOBER 2015 Transactions of the ASME

Downloaded From: http://manufacturingscience.asmedigitalcollection.asme.org/pdfaccess.ashx?url=/data/journals/jmsefk/934680/ on 04/23/2017 Terms of Use: http://www.asme


Fig. 20 The residual stress distributions in the 2y direction
from the top surface for the top 10 mm of the plate based on
mesh-2 and mesh-4 for single indentation on the elasticplastic
plate
respectively. In order to keep the model sizes manageable for the
finer meshes, uniform element sizes are used for the top layers up
to a depth of about 1 mm and 0.6 mm for mesh-3 and mesh-4,
respectively. Then the element sizes are gradually increased to-
ward the bottom surfaces. It should be noted that the element size
progression is kept similar for mesh-3 and mesh-4 in the gradually
increased element size zones.
For elastic indentation analyses, the Hertzian solution and the
results based on mesh-2, mesh-3, and mesh-4 are shown in
Fig. 19. The stress distributions of only three stress components in
the y direction from the top surface for the top 2 mm of the plate
are shown for visual clarity. The results show that the accuracy of
the stresses along the depth is only affected up to the length corre-
sponding to the size of the element of the first layer. Below the
first layer, the results are fairly comparable.
Similarly, for the elasticplastic indentation analyses, the resid-
ual stresses are compared. For the elasticplastic cases, the results
based on mesh-3 and mesh-4 are nearly identical and are not
shown here. Figure 20 shows a comparison of the residual stresses
based on mesh-2 and mesh-4. The results are nearly identical for
the two cases up to a depth of 1.6 mm. The results show some
deviation below 1.6 mm depth. That can be attributed to the larger
element size progression for mesh-4. It should be mentioned that
a comparison for the results based on mesh-1 and mesh-2 is shown
in Fig. 7.
References
[1] Love, R. J., and Waistall, D. N., 1954, The Improvement in the Bending
Fatigue Strength of Production Crankshafts by Cold Rolling, M.I.R.A. Report
No. 1954/1, pp. 18.
[2] Chien, W. Y., Pan, J., Close, D., and Ho, S., 2005, Fatigue Analysis of Crank-
shaft Sections Under Bending With Consideration of Residual Stresses, Int. J.
Fatigue, 27(1), pp. 119.
[3] Spiteri, P., Ho, S., and Lee, Y. L., 2007, Assessment of Bending Fatigue Limit
for Crankshaft Sections With Inclusion of Residual Stresses, Int. J. Fatigue,
29(2), pp. 318329.
[4] Ho, S., Lee, Y.-L., Kang, H.-T., and Wang, C. J., 2009, Optimization of a
Crankshaft Rolling Process for Durability, Int. J. Fatigue, 31(5), pp. 799808.
[5] Choi, K. S., and Pan, J., 2009, Simulations of Stress Distributions in Crank-
shaft Sections Under Fillet Rolling and Bending Fatigue Tests, Int. J. Fatigue,
31(3), pp. 544557.
[6] Choi, K. S., and Pan, J., 2009, A Generalized Anisotropic Hardening Rule
Based on the Mroz Multi-Yield-Surface Model for Pressure Insensitive and
Sensitive Materials, Int. J. Plast., 25(7), pp. 13251358.
Journal of Manufacturing Science and Engineering
Downloaded From: http://manufacturingscience.asmedigitalcollection.asme.org/pdfaccess.ashx?url=/data/journals/jmsefk/934680/ on 04/23/2017 Terms of Use: http://www.asme
[7] Dumas, G., and Baronet, C. N., 1971, Elastoplastic Indentation of a Half-
Space by an Infinitely Long Rigid Circular Cylinder, Int. J. Mech. Sci., 13(6),
pp. 519530.
[8] Hardy, C., Baronet, C. N., and Tordion, G. V., 1971, The Elasto-Plastic Inden-
tation of a Half-Space by a Rigid Sphere, Int. J. Numer. Methods Eng., 3(4),
pp. 451462.
[9] Lee, C. H., Masaki, S., and Kobayashi, S., 1972, Analysis of Ball Indentation,
Int. J. Mech. Sci., 14(7), pp. 417426.
[10] Kral, E. R., Komvopoulos, K., and Bogy, D. B., 1993, Elastic-Plastic Finite
Element Analysis of Repeated Indentation of Half-Space by a Rigid Sphere,
ASME J. Appl. Mech., 60(4), pp. 829841.
[11] Sinclair, G. B., Follansbee, P. S., and Johnson, K. L., 1985, Quasi-Static Nor-
mal Indentation of an Elasto-Plastic Half-Space by a Rigid Sphere-II. Results,
Int. J. Solids Struct., 21(8), pp. 865888.
[12] Merwin, J. E., and Johnson, K. L., 1963, An Analysis of Plastic Deformation
in Rolling Contact, Proc. Inst. Mech. Eng., 177(1), pp. 676690.
[13] Bhargava, V., Hahn, G. T., and Rubin, C. A., 1985, An Elastic-Plastic Finite
Element Model of Rolling Contact Part 1: Analysis of Single Contacts, ASME
J. Appl. Mech., 52(1), pp. 6774.
[14] Bhargava, V., Hahn, G. T., and Rubin, C. A., 1985, An Elastic-Plastic Finite
Element Model of Rolling Contact Part 2: Analysis of Repeated Contacts,
ASME J. Appl. Mech., 52(1), pp. 7582.
[15] Bhargava, V., Hahn, G. T., and Rubin, C. A., 1988, Analysis of Rolling
Contact With Kinematic Hardening for Rail Steel Properties, Wear, 122(3),
pp. 267283.
[16] Kulkarni, S. M., Hahn, G. T., Rubin, C. A., and Bhargava, V., 1990,
Elastoplastic Finite Element Analysis of Three-Dimensional, Pure Roll-
ing Contact at the Shakedown Limit, ASME J. Appl. Mech., 57(1),
pp. 5765.
[17] Kulkarni, S. M., Hahn, G. T., Rubin, C. A., and Bhargava, V., 1991, Elasto-
Plastic Finite Element Analysis of Three-Dimensional Pure Rolling Contact
Above the Shakedown Limit, ASME J. Appl. Mech., 58(2), pp. 347353.
[18] Kulkarni, S., Hahn, G. T., Rubin, C. A., and Bhargava, V., 1991, Elasto-
Plastic Finite Element Analysis of Repeated Three-Dimensional, Elliptical
Rolling Contact With Rail Wheel Properties, ASME J. Tribol., 113(3),
pp. 434441.
[19] Jiang, Y., Xu, B., and Sehitoglu, H., 2002, Three-Dimensional Elastic-Plastic
Stress Analysis of Rolling Contact, ASME J. Tribol., 124(4), pp. 699708.
[20] Hearle, A. D., and Johnson, K. L., 1987, Cumulative Plastic Flow in Rolling
and Sliding Line Contact, ASME J. Appl. Mech., 54(1), pp. 17.
[21] Bower, A. F., and Johnson, K. L., 1989, The Influence of Strain Hardening on
Cumulative Plastic Deformation in Rolling and Sliding Line Contact, J. Mech.
Phys. Solids, 37(4), pp. 471493.
[22] McDowell, D. L., and Moyar, G. J., 1991, Effects of Non-Linear Kinematic
Hardening on Plastic Deformation and Residual Stresses in Rolling Line Con-
tact, Wear, 144(12), pp. 1937.
[23] Jiang, Y., and Sehitoglu, H., 1994, An Analytical Approach to Elastic-Plastic
Stress Analysis of Rolling Contact, ASME J. Tribol., 116(3), pp. 577587.
[24] Jiang, Y., and Sehitoglu, H., 1996, Rolling Contact Stress Analysis With the
Application of a New Plasticity Model, Wear, 191(12), pp. 3544.
[25] Yu, M., Moran, B., and Keer, L. M., 1993, A Direct Analysis of Two-
Dimensional Elastic-Plastic Rolling Contact, ASME J. Tribol., 115(2),
pp. 227236.
[26] Yu, M. M. H., Moran, B., and Keer, L. M., 1995, A Direct Analysis of Three-
Dimensional Elastic-Plastic Rolling Contact, ASME J. Tribol., 117(2),
pp. 234243.
[27] Bijak-Zochowski, M., and Marek, P., 1997, Residual Stress in Some Elasto-
Plastic Problems of Rolling Contact With Friction, Int. J. Mech. Sci., 39(1),
pp. 1532.
[28] Park, H.-S., and Dang, X., 2015, Multiobjective Optimization of the Heating
Process for Forging Automotive Crankshaft, ASME J. Manuf. Sci. Eng.,
137(3), p. 031011.
[29] Chaise, T., and Nelias, D., 2011, Contact Pressure and Residual Strain in 3D
Elasto-Plastic Rolling Contact for a Circular or Elliptical Point Contact,
ASME J. Tribol., 133(4), p. 041402.
[30] Nelson, A. W., Malik, A. S., Wendel, J. C., and Zipf, M. E., 2014,
Probabilistic Force Prediction in Cold Sheet Rolling by Bayesian Inference,
ASME J. Manuf. Sci. Eng., 136(4), p. 041006.
[31] Sun, Q., Chen, J., and Pan, H., 2015, Prediction of Edge Crack in Cold Rolling
of Silicon Steel Strip Based on an Extended GursonTvergaardNeedleman
Damage Model, ASME J. Manuf. Sci. Eng., 137(2), p. 021003.
[32] ABAQUS Version 6.10 User Manual, 2010, SIMULIA, Providence, RI.
[33] Choi, K. S., Pan, J., and Ho, S., 2004, Fatigue Failures of Rollers in Crankshaft
Fillet Rolling, SAE Technical Paper No. 2004-01-1498.
[34] Choi, K. S., Pan, J., and Ho, S., 2005, Effects of Roller Geometry on Contact
Pressure and Residual Stress in Crankshaft Fillet Rolling, SAE Technical
Paper No. 2005-01-1908.
[35] Johnson, K. L., 1987, Contact Mechanics, Cambridge University Press, Cam-
bridge, UK.
[36] McEwen, E., 1949, Stress in Elastic Cylinders in Contact Along a Generatrix,
Philos. Mag., 40, pp. 454459.
[37] Mesarovic, S. Dj., and Fleck, N. A., 1999, Spherical Indentation of Elastic-
Plastic Solids, Proc. R. Soc. London, Ser. A, 455(1987), pp. 27072728.
OCTOBER 2015, Vol. 137 / 051005-13