Scientific Research and Essays Vol. 8(3), pp.

128 –138, 18 January, 2013

Available online at http://www.academicjournals.org/SRE
DOI: 10.5897/SRE12.536
ISSN 1992-2248 ©2013 Academic Journals

Full Length Research Paper

Finite element simulation of non- linear deformation

behaviour in large diameter angular contact thrust
bearing
M. S. Starvin1*, S. Babu2, Sriramachandra Aithal3, K. Manisekar4 and P. Chellapandi3
1
Department of Mechanical Engineering, Anna University Tirunlelveli, Tirunelveli, TamilNadu, India.
2
School of Energy, PSG College of Technology, Coimbatore, TamilNadu, India.
3
Nuclear and Safety Engineering Group, Indira Gandhi Atomic Research Centre, Kalpakkam, TamilNadu, India.
4
Centre for Manufacturing Sciences, National Engineering College, Kovilpatti, TamilNadu, India.
Accepted 24 October, 2012

The objective of the present study is to obtain relevant contact stiffness for a single ball-raceway
contact which can be used in load distribution analysis of full bearing. Towards this, an elastic
constitutive model is used in the 3D finite element analysis of load deformation characteristic of large
diameter bearing ball - raceway contact and the finite element results are validated by conducting
experiment on the linear mock up model bearing. The FE results show a good agreement with the
values calculated using Hertz’s relations for deformation.

Key words: Fast breeder reactor, rotatable plugs, slewing bearing, load-deformation, finite element analysis,
rolling element, bearings, experimental analysis.

INTRODUCTION

In typical sodium, cooled fast breeder reactor of 0.012 with no permanent indentation during idle condition
Prototype Fast Breeder Reactor (PFBR) type which is of the bearings.
under construction at Kalpakkam, near Chennai, in India, ii) Functioning at a temperature of 120°C, handling
two rotatable plugs namely large rotatable plug (LRP) thermal expansion problem as the temperature difference
and small rotatable plug (SRP) are provided along with between top and bottom races is of 20°C.
transfer arm (TA) to facilitate fuel handling operation of iii) Withstanding dead weight of components of the order
the reactor. The large and small rotatable plugs are of 500 t for a 7 m bearing supported on them as well as
supported on by a large diameter slewing ring bearings to dynamic load and seismic load.
facilitate their rotation during fuel handling while iv) Working under dry/solid lubrication and expectation of
withstanding the dead weight of rotatable plugs and design life is for 60 years.
components supported on the plugs along with seismic
loads during seismic events. The bearing (~7 m) Considering the duty, constraints due to differential
consists of top and bottom races through which the thermal expansion between races and ease of
component thrust loads are transmitted. These bearings manufacture and assembly, an angular contact thrust
require certain most important design features such as: bearing is selected for the current application, which can
take heavy axial load in addition to the radial load.
i) Running friction factor of 0.006 and start-up friction of These large diameter bearings are generally known as
slewing ring bearings. The application of slewing
bearings ranges from cranes to wind-power generators,
excavation machinery, capital goods machinery, nuclear
*Corresponding author. E-mail: [email protected]. Tel: reactors and where the operational safety is essential. It
+919486865043. is therefore required to design a bearing with specific
 Starvin et al. 129

Figure 1. Geometry of angular contact bearing.

composition to suit a particular application (Josu et al., possible. This calls for a complete 3D FE analysis which
2012; Hirania, 2009). There are no standard procedures is difficult and time consuming in considering the large
to define the load carrying capacity, service life and number of bearing elements and diameter of the bearing.
rotational resistance due to the special construction of However, this can be overcome by using super elements
large rolling bearings, geometry of raceways, bearing ring for which stiffness of the bearing element–raceway
materials, mechanical and heat treatment technologies contact need to be evaluated a priory (Alain et al., 2008;
and working conditions (Zaretsky, 1992; Harris and Ludwik, 2006). This paper presents a 3D FE model used
Kotzalas, 2007; Sague, 1978). Unlike in classical to study the load-deformation characteristics of single
bearings, slewing ring bearings are less firmly supported rolling element-raceway contact model.
by their mounting structures and the elasticity of the Further, to experimentally verify the load deformation
upper and support structures cause an uneven relationship of angular contact slewing ring bearing, a
distribution of external loads over the rolling bearing linear mock up model bearing with hardened raceway is
diameter (Prebil et al., 1995). The unexpected excessive fabricated. Finally, the FE results are validated by
load on the ball due to varying load distribution causes experimental results as well as a Hertz contact model.
local deformation of a bearing race way and subsequent
yielding of case/core interface (Sague, 1978; Pallini and
Sague, 1985). GEOMETRICAL STUDY OF AN ANGULAR CONTACT
The actual load acting on a bearing is different from the BALL BEARING
assumed conditions in determining static load rating of
the bearing. Therefore, it is essential to determine the In this paper, the large diameter angular contact bearing
maximum load acting on the rolling element of the has been considered for the study, which is a two point
bearing. For this, it is required to know the load contact bearing having single curvature for each raceway
distribution and angle of contact of rolling element of the (Figure 1). Each raceway of the bearing rings is defined
bearing. After that, it is necessary to convert the actual by toroidal surfaces with centers C1 & C2. The diameters
load into an equivalent static load which would cause the of curvature of the raceways and the rolling elements are
same contact stress at the centre of the most heavily related by osculation ratio:
loaded rolling element–race way contact as that which
occurs under the actual load conditions. This equivalent Dw
load will be more useful for calculating the dynamic
s
D profile
capacity of the slewing bearing from the particular loads
of each rolling element (American National Standard.
ANSI/AFMA Std. 9, 1990). In the slewing ring bearings, Where, Dw is the diameter of the rolling element and
prediction of life and uniform distribution of load among
the rolling elements are tough task due to the flexibility of
D profile is the diameter of the raceway.
the supporting structures and rings (Pallini and Bearing manufacturers recommended a value between
Rumbarger, 1979). Analyzing the bearing for these 0.92 and 0.98 (Harris and Kotzalas, 2007; Zupan and
conditions is a complex task, and the determination of Prebil, 2001) for the osculation ratio. The contact
rolling elements loading under this condition is feasible between the rolling element and the raceway is a single
only by the numerical methods. A sufficiently accurate FE point contact at the equilibrium. The contact angle α is
bearing model which can take into account the support defined as the angle between R axis of the bearing and
structures (bearing rings), rolling element-raceway the line passing through the point of contact of ball-
contact and the phenomena which occur in it, can make it raceway and the centre of curvature of a raceway. The
130 Sci. Res. Essays

Figure 2. Illustration of contact parameters.

change in contact angle due to variation in load can be

F   
k 2

 1  k   2 F k 
determined from the shift in the centre of curvatures.  
k 2  1  k  (1)
1
MODELING OF CONTACTING SURFACES   KQ m
(2)

Hertz considered the stress and deformation in the Where K is the flexibility of the rolling elements, Q is the
perfectly smooth, ellipsoidal, contacting elastic solids. contact load in N, δ is the deflection in mm, and m is the
The application of Hertz theory can be used to model the Hertzian constant exponent (m = 1.5 for point contact).
contact between the raceways and the rolling elements This relation relates the force supported by each ball
(Thomas and Hoersch, 1990). The ball deformations are bearing and the overall displacements of the bearing.
mainly concentrated in the vicinity of the contact points. However, the flexibility K is a function of the osculation
The point contact between the race and ball develops ratio s. Since the osculation ratio is an important
into an area contact, which has the shape of an ellipse industrial parameter for bearing manufacturers, it is
with ‘a’ and ‘b’ as the semi-major and semi-minor axes required to include its effect in the deflection relationship
respectively (Figure 2). The curvature sum and difference (Table 1). The objective is not to develop a new
are needed in order to obtain the contact force of the ball. relationship but to derive an analytical relationship from
The ellipticity parameter (k) is related to the geometrical works existing in literature (Alain et al., 2008). In this
parameter F (ρ) by the means of an implicit Equation 1, framework, the analytical relationship given by Houpert
where  (k ) and F (k ) are the complete integrals of the (2001) can be used to analytically derive the effect of
osculation ratio on the deflection. This relationship is
first and second kind, respectively and k is the constant given by the following equation:
for Hertzian contact elastic deformation. By assuming the
elliptical eccentricity parameter k, it is possible to 2
3
  8.97 *10 1  s 
Q
calculate the corresponding values of F (  ) by 4 0.1946
1
(3)
3
numerical means. The elliptic integrals are commonly D W

calculated based on arithmetic/geometric mean

algorithm, as described in Abramovitz and Stegun (1972). Numerical model
Once the value of k related to F(ρ) is obtained, the
dimensions of the contact area (Harris and Kotzalas, The finite element method (FEM) has been successfully
2007), the displacement (mutual approach of the bodies used to evaluate contact problems for the ball and roller
centers) and maximum stress can be calculated. bearings with complex model and nonlinear capability.
The relative displacement can be mathematically This numerical modeling of the rolling contact was carried
modeled by the following equation: out with Abaqus FEA-Static General commercial
 Starvin et al. 131

y
x

Figure 3. Bearing ball and race way contact model (dimensions are in meters).

software. The finite element model represents the actual analysis of contact problems. The method of digitization
geometry of the rolling contact between the ball, and the of the model and the selection of suitable dimensions for
bearing ring raceways and it is used for the determination the meshing is based on the following presumptions: the
of load deformation relationship. For the purpose of primary objective of analysis is to obtain the dependency
analysis, a simplified model is needed to save the time in between deformation and the load whereas determination
determining the ball deformation characteristic of the of contact area and distribution of contact pressure are
bearing. In this analysis, a bearing with N balls, the FE secondary. Similarly, increasing the density of the mesh
model as shown in Figure 3 is constructed, which in 3D FE model increases significantly the time needed to
consist of a half ball contacted by the half of 1/N top perform the analysis due to increase in number of
raceway ring and the bottom raceway ring. The model elements. Table 2 includes the effect of the density of the
was constructed with a groove radius of 16.51 mm and mesh on the parameters of the contact zone for a 2D
ball diameter of 31.75 mm. The numerical model of the model of the contact zone of the roller with the bearing
rolling contact comprises the measured geometric and raceway. Ultimately, the mesh dimensions were finalized
material characteristics of the bearing raceways. It was for further calculations in the contact zone.
used to simulate a contact of the rolling element and a The following boundary conditions are included for the
bearing raceway at a varying contact force from 1 to 30 simulation of a bearing model:
kN.
Selection of the suitable element-type for modeling has a) all nodal degrees of freedom (d.o.f.) on bottom
a significant influence on the accuracy and efficiency of a surfaces of the bottom raceway are constrained to move
FE simulation. Several simulations have been conducted in the direction of Z;
to test FE models with a view to identify finest modeling b) since one half of the 1/N of the ball bearing assembly
scheme. In the simulations, 8-node 3D first-order was taken for analysis, symmetrical conditions are
reduced-integrated elements were used for meshing the applied along the axial (x- direction) of the model; c) the
contact surfaces of ball and raceways. The edges of the outer surface of the bottom and top race ways are
contact surfaces in the direct contact zone of the model constraint to move in the out directions (direction of Y).
were divided to obtain the mesh dimensions of the order
0.1 × 0.1 mm and the other regions were meshed with EXPERIMENTAL VERIFICATION
coarse mesh (Qin and Balendra, 2000). The fine meshes
at the contact region provide better solution for the In order to verify the stiffness characteristics of large
132 Sci. Res. Essays

Table 1. Global deformation relations for ball bearing (Thomas and Hoersch, 1990).

Rolling element SKF INA

4
4.5.10 2 4.84.104 2 3
Balls, m = 3/2  Q 3
 Q
3 D 3 D
w w

Table 2. Comparative calculation results of 2D models of the contact zone of rollers with a slewing bearing raceway (d = 20
mm, l = 18 mm, Flim = 35,713 kN) (Gibezyńska and Marciniec, 1995).

Mesh size Number of Number of Calculation time indicator

S. No Model Po[MPa] δ(mm)
(mm) elements nodes (for model A = 100)
1 A 0.025 6060 6172 2690.34 0.092554 100
2 B 0.05 1565 1622 2687.45 0.092267 21.6
3 C 0.1 496 638 2679.46 0.092056 5.58

diameter angular contact bearings determined strain measurement techniques, high accuracy micrometers (least
analytically, a Linear Model Mockup (Angular contact) count of 1 µm) are used to measure the relative approach of the
two races; that is, the deformation of balls and races combined as
Bearing (LMMB) was fabricated at IGCAR, Kalpakkam as shown in Figure 5. The two micrometers each are placed at
part of development program for large diameter bearings opposite corners of each raceway. The tests were conducted on
to be used in FBRs. The cost of making a complete two sets of linear bearing mock up pieces by placing different
bearing was prohibitive and also it will be challenging to number of balls that is, two, three and four balls between the
conduct experiments due to large size of set-up and huge raceways under loads, varying from 200 N to 31 kN per ball. The
loads required for loading. Hence, sections of complete loads were chosen in such a way that there are no indentations
formed during the tests.
races were represented by small rectangular steel blocks Care was taken to ensure the following conditions during these
(Whittemore and Petrenko, 1921). The linear model tests:
angular contact bearing is shown in Figure 4.
1) The surfaces of the raceways and rolling elements were cleaned
and degreased before assembly.
Experimental setup 2) All bearing surfaces were parallel to each other and also
perpendicular to the line of action of the load.
Each block has a cylindrical groove on one face, parallel to the 3) The balls were placed symmetrically with respect to the line of
opposite face, having a radius slightly greater than that of the ball action of the load.
with which it was to be used (curvature co-efficient of 0.52) so that 4) A cage was provided to restrict the balls from moving along the
the contact between ball and raceway will be elliptical. These races length of the cylindrical raceways before and during the test so that
were of length 150 mm with semi circular raceway radius of 16.51 the symmetric spacing is maintained.
mm. The horizontal offset between the axes of the cylindrical 5) The load was applied gradually to the balls through a hydraulic
grooves were 1.27 mm. The raceways were made from Medium Jack.
Carbon Chromium Steel (AISI 4140) which is through hardened to
50 HRc. The raceway grooves were ground to the required radius The deformation of contact surfaces were evaluated by subtracting
and surface finish. In an actual ball bearing, the axis of the groove average readings of the lower raceway from the average readings
would be an arc of a circle about the axis of rotation. The rolling of two micrometers fixed at the opposite corners of the top raceway.
elements (balls of standard sizes of 31.75 mm) were purchased Due to this, the deformations in the support structure are not added
from commercial manufacturers made of through hardened (62 with final contact deformation. Deformation at lower loads are
HRc) high-Carbon Chromium Bearing Steel (AISI 52100). The neglected as the same are highly non-linear due to geometrical
variations in diameter of balls are 2.5 µm. A grinding surface finish deviations in the loading structure including gaps which need to be
of less than 0.4 µm on bearing raceways was achieved. overcome prior to establishing the ball-raceway contact. The load–
As the primary goal of the experiment was to obtain load- deformation characteristics of various tests conducted by placing
deformation characteristic of the ball-raceway contact, the length of different number of balls under varying loading conditions are
the linear mock-up bearing is selected to accommodate 3 to 4 balls, plotted in Figure 6. The experiments are repeated three times to
though theoretically one number of ball is sufficient to evaluate the ensure repeatability of the results.
load-deformation characteristics. The set-up is designed to
accommodate higher number of balls (upto 4) so that it will be easy
to align the set-up with respect to the loading point to ensure RESULTS AND DISCUSSION
stability and possible to make a parametric study with respect to
number of balls. Towards studying the load-deformation
characteristics of linear angular contact bearing, a special purpose
The estimation of the load–deformation characteristics for
test rig was fabricated and the setup is shown in Figure 5. Since, it the contact zone of the ball and raceways have been
is difficult to measure strain in the bearing ball through conventional carried out using Hertz theory and FE analysis. The
 Starvin et al. 133

Figure 4. Photo of linear model mock up bearing assembly.

Figure 5. Experimental set-up.

contacting bodies are made of materials defined by their contacting materials (American National Standard
5
modulus of elasticity E(E1 = E2 = 2.06 × 10 MPa) and ANSI/AFMA Std. 9, 1990). The initial clearance between
Poisson’s ratio μ ( 0.3) and the equivalent modulus of the contacting bodies in the numerical model is made
elasticity E* is defined in terms of properties of both zero. The friction coefficient between the rolling element
134 Sci. Res. Essays

35000

30000

25000
LOAD (N)

20000

15000

10000 Four Ball

5000 Three Ball
Two Balls
0
0 0.05 0.1 0.15 0.2 0.25 0.3
DEFORMATION (mm)
Figure 6. Load–deformation characteristics.

and the raceways is taken to be 0.002. The three- element and analytical results demonstrates the accuracy
dimensional contact model developed and its accuracy of the finite element model and appropriateness of the
was validated in light of favorable comparisons between boundary conditions.
finite element results and analytical solutions obtained
within the elastic region. To compare these results with
the experimental results, experiments were conducted on Implementation of FE model
a linear angular contact mock up bearing as discussed in
the previous section. The load deformation Proper modeling and accurate representation of rolling
characteristics of ball-raceway combinations under elements is the key factor in solving the large diameter
varying load conditions with different number of balls are bearings numerically. Modelling of large number of rolling
shown in Figure 6. From graph shown in Figure 6, it can elements, shape of the rolling elements and its modeling
be seen that the load-deformation characteristics are of the contact problem in a full bearing is a difficult one.
non-linear in nature which is as expected. Further, the Hence, by considering the complexity of modeling, it is
differences in the deformation observed with two, three better to apply a revolutionary approach like super
and four balls are very interesting and this needs further element techniques to model the full bearing (Alain et al.,
study. Several possible reasons exist for these 2008). This type of simplification allows replacing the
differences which include the variation in the actual ball rolling elements under compression by a non-linear
diameters and deviation in the raceway curvature along spring working in tension. In the FE model, a rolling
the length of the cylindrical groove, the asymmetric in element in a sector is defined by two nodes representing
loading results in non uniform load distribution etc. the centers of curvature of race ways. Then, each node is
The comparison of the load – deformation linked to the corresponding opposite node by non-linear
characteristics of FE results, Hertz theory and the traction spring as shown in Figure 10. Top and bottom
experimental value for four balls are shown in Figure 7. raceway zones of contact between the rolling element
From the plot, it can be seen that all the three results and raceway in a sector are modeled by rigid shells.
closely match with each other. It is observed that Further, these rigid shells are connected to the
homogenized four ball case very closely follows the FEM corresponding centers of curvatures to establish elliptical
and Hertz theory compared to two and three balls case. contact. The super element was synthesized (non-linear
This may be due to more uniformity and symmetry in load spring) based on the results obtained in the ball raceway
application in the case of four balls compared to two and deformation characteristics and it is substituted for rolling
three balls case. Figure 8 shows the agreement between elements in the global model as shown in Figure 11.
the load and contact stresses of FEM result and Hertz The FE analysis shows a uniform load distribution on the
theory. Figure 9 shows the variation of contact area with 2 m model bearing (Figure 11). From Figure 12, it is clear
the applied load. The good agreement between finite that the loads are equally distributed on all the
 Starvin et al. 135

35000

30000

25000
Load (N)

20000

15000

10000 Hertz
5000
FEM
Experimental
0
0 0.05 0.1 0.15 0.2

δ Deformation (mm)
Figure 7. Comparison of load-deformation characteristics.

Load Vs Contact Pressure

3500

3000
Contact Pressure (MPa)

2500

2000

1500
(MPa)

1000 HERTZ
FEM
500

0
0 5000 10000 15000 20000 25000 30000 35000
Load (N)

Figure 8. Plot between load and contact stress.

rolling elements of the bearing for the applied uniform angular contact bearing were determined using FE
load. This implies that the simplified model validates the Models. The FE results were validated by conducting
actual bearing model and it can be used for further experiments on a linear contact mock up bearing
analysis. fabricated at IGCAR, Kalpakkam. The results of FE
experiments are compared with the Hertz relation. The
results show a good agreement between them. The other
CONCLUSION contact parameters such as contact area and contact
pressure measured using FEA are in good agreement
The non-linear load-deformation characteristics of with analytical results. The study of load–deformation
136 Sci. Res. Essays

Load Vs Contact Area

20
Contact area (sq mm)

18
16
14
12
10
mm)

8 Hertz
6 FEA at Top race
4
FEA at outer race
2
0
0 5000 10000 15000 20000 25000 30000 35000
Load (N)
Figure 9. Graph between load and contact area.

Figure 10. Geometrical approach.

characteristics of a single ball with the raceways can be technique is the replacement of the rolling elements by
used as a basis to model a full bearing by sub structuring non linear traction springs. The characteristics of the
technique. The key element for the sub structuring springs can be defined by the load-deformation
 Starvin et al. 137

No. of Rolling Elements: 70

Dia of Rolling Element: 31.75 mm

Figure 11. FE model of 2 m slewing ring bearing.

25000

20000
Contact Load (N)

15000

10000

5000

0
0 5 10 15 20 25 30 35 40
Ball Num ber

Figure 12. Uniform contact load distribution under axial loading.

characteristics derived from the Numerical/Experimental American National Standard ANSI/AFMA Std 9-1990. Load Ratings and
Fatigue Life for Ball Bearings.
method. This type of simplification allows for avoiding the
Alain D, Zouhair C, Antoine G (2008). 3D Simplified finite elements
multiple modeling of the contact problems in large analysis of load and contact angle in a slewing ball bearing. ASME J.
diameter bearings. Mech. Des. 130:082601-1-8.
Gibezyńska T, Marciniec A (1995). Influences des parameters
geometriques de la couronne d’orientation sur les efforts de conact,
ACKNOWLEDGEMENT In:IASTED International Conference Modelling,
Indentification,Control,Igls pp. 85-88.
The authors thank BRNS, Mumbai which provided Harris TA, Kotzalas MN (2007). Essential concepts of bearing
technology. Taylor & Francis. New York pp. 174-175.
financial support for carrying out this research work at Hirania H (2009). Root cause failure analysis of outer ring fracture of
Structural Mechanics Laboratory at IGCAR, Kalpakkam, four-row cylindrical roller bearing. Tribol. Trans. 52(2):180-190.
India. Houpert L (2001). An Engineering Approach to Hertzian Contact
Elasticity - Part 1. ASME J. Tribol. 123(3):582-588.
Josu A, Mikel A, Rafael A, Igor FB (2012). Theoretical calculation of
REFERENCES general static load-carrying capacity for the design and selection of
three row roller slewing bearings. Mechanism Mach. Theory 48:52-
Abramovitz M, Stegun I (1972). Handbook of mathematical functions 61.
with formulas, graphs, and mathematical tables. Dover Publications Ludwik K (2006). Modelling of rollers in calculation of slewing bearing
New York. eds. with the use of finite elements. Mechanism and Machine theory.
138 Sci. Res. Essays

41:1359–1376. Thomas HR, Hoersch VA (1990). Stresses due to the pressure of one
Pallini RA, Sague JE (1985). Computing core-yield limits for case- elastic solid upon another. University of Illinois Bulletin 212.
hardened rolling bearings. ASLE Trans. 28(1):91–96. Whittemore HL, Petrenko SN (1921). Friction and carrying capacity of
Pallini R, Rumbarger J (1979). Failure Analysis and Redesign of Large ball and roller bearings. No.201, Technologic Papers of the Bureau of
diameter stacker Reclaimer Slew Bearing. Lubr. Eng. 12:692-697. Standards, Washington.
Prebil I, Zupan S, Lučič P (1995). Lastverteilung auf Wälzkörper der Zaretsky EV (1992). STLE life factors for rolling bearings, chapter 4 –
Drehverbindungen Konstruktion. 47(11):339–345. Bearing Load. Illinois. STLE Publications.
Qin Y, Balendra R (2000). A method for the simulation of temperature Zupan S, Prebil I (2001). Carrying angle and carrying capacity of a large
stabilization during multi-cycle cold forging operations. Int. J. Mater. single row ball bearing as a function of geometry parameters of
Process. Technol. 107(1–3):252–259. rolling contact and supporting structure stiffness. Mech. Mach.
Sague JE (1978). The special way big bearings can fail. Mach. Des. Theory 36(10):1087–1103.
21:113–120.