You are on page 1of 86

TRANSIENT THERMOMECHANICAL INTERACTIONS OF SHAFT

-
BUSHING PAIR IN BEARINGS – A FINITE ELEMENT STUDY






A Thesis

Submitted to the Graduate Faculty of the
Louisiana State University and
Agricultural and Mechanical College
in partial fulfillment of the
requirements for the degree of
Master of Science in Mechanical Engineering

in

The Department of Mechanical Engineering








by
Rajesh Krithivasan
B.E., PSG College of Technology, Bharathiyar University, 1997
December 2002
ii
ACKNOWLEDGEMENTS
I would like to express my gratitude to my advisor, Dr. Khonsari, for his
invaluable guidance and encouragement throughout the course of this thesis work.
I am sincerely thankful to my committee members, Dr. Yitshak Ram and Dr.
Wenjin Meng, for their support and spending their valuable time in evaluating my thesis
work.
My special thanks goes to Mr. Haribabu Krishnamurthy for all the fruitful
discussions we had that were instrumental for the successful completion of this thesis.
I also thank all my friends in the tribology group (CeRoM), for their moral and
technical support, without which I would not have succeeded in finishing my thesis.
This work was supported in part by a grant from The Louisiana Board of Regents'
Industrial Ties Research Subprogram with Caterpillar Inc (Grant # ITRS-007B-2002).
iii
TABLE OF CONTENTS
ACKNOWLEDGEMENTS………………………………………………………………..ii
NOMENCLATURE………………………………………………………………………..v
ABSTRACT………………………………………………………………………………vii
CHAPTER 1. INTRODUCTION…………………………………………………………1
CHAPTER 2. LITERATURE SURVEY …………………………………………………2
CHAPTER 3. FINITE ELEMENT ANALYSIS: PROCEDURE AND OVERVIEW…...9
3.1 Thermal Analysis – Theory and Finite Element Formulation…………………9
3.2 Thermoelastic Analysis – Theory and Finite Element Formulation …………12
CHAPTER 4. THERMALLY INDUCED SEIZURE IN JOURNAL BEARINGS
DURING START UP ……………………………………………………15
4.1 Finite Element Modeling Procedure………………………………………….17
4.1.1 Transient Thermal Finite Element Analysis ……………………….18
4.1.2 Non-linear Transient Elastic Finite Element Model ……………….20
4.1.3 Seizure Criterion …………..……………………………………….22
4.2 Results and Discussion……………………………………………………….23
4.3 Model Parameters…………………………………………………………….26
4.4 Generalization………………………………………………………………...27
4.4.1 Influence of Operating Parameters…………………………………28
4.5 Non- Dimensionalization …………………………………………………….30
4.5.1 Statistical Treatment of Non-dimensional Operating Parameters….30
4.6 Verification and Analysis…………………………………………………….32
CHAPTER 5. TIS OF JOURNAL BEARINGS TRIGGERED BY A TRANSIENT
FLOW DISTURBANCE…………………………………………………37
5.1 Finite Element Analysis Modeling and Seizure Criteria……………………..40
5.2 Results and Discussion……………………………………………………….41
5.3 Model Parameters…………………………………………………………….43
5.4 Non-Dimensionalization and Generalization ………………………………...44
5.5 Analysis of Results…………………………………………………………...45
5.6 Verification…………………………………………………………………...47
CHAPTER 6. THERMOMECHANICAL ANALYSIS OF A HEAVILY LOADED
OSCILLATING PIN-BUSHING PAIR …………………………………49
6.1 Introduction and Problem Definition ………………………………………..49
6.2 Problem Description and Solution Methodology .…………………………..51
6.3 Theoretical Hertzian Contact Analysis ………………………………………51
6.4 Thermal Analysis …………………………………………………………….52
6.4.1 Element Types ……………………………………………………..53
iv
6.4.2 Calculation of Heat Flux …………………………………………...53
6.4.3 Implementation of Oscillating Heat Flux…………………………..53
6.4.4 Convection Film Transfer Coefficient between Bushing Outer
Radius and Atmosphere ……………………………………………55
6.4.5. Convective Coefficient in the Gap between Pin and Bushing …….57
6.4.6 Boundary Conditions for the Thermal Analysis …………………...59
6.5 Transient Thermomechanical Analysis ………………………………………60
6.5.1 Element Types ……………………………………………………..60
6.5.2 Meshing…………………………………………………………….61
6.5.3. Boundary Conditions ……………………………………………..61
6.6 Operating Parameters…………………………………………………………………….62
6.7 Results and Discussion……………………………………………………….63
6.8 Generalization………………………………………………………………...67
6.9 Verification…………………………………………………………………...70
6.10 Restrictions and Applications……………………………………………….71
CHAPTER 7. CONCLUSIONS………………………………………………………….72
7.1 TIS in Journal Bearings During Start Up …………………………………….72
7.2 TIS in Journal Bearings During Lubricant Flow Disturbance ……………….72
7.3 TIS and Thermal Galling in Oscillating Pin-Bushing Analysis……………...73
BIBLIOGRAPHY ………………………………………………………………………..74
APPENDIX - MAPLE PROGRAM TO CALCULATE CONVECTION
COEFFICIENT …………………………………………………………..76
VITA ……………………………………………………………………………………..79
v
NOMENCLATURE
A
s
= Area of the shaft that is in contact with the bushing, m
2

C = Radial clearance, m
C
p
= Specific heat capacity, J/kgK
E = Young's modulus, N/m
2

f = Coefficient of friction
h = Convective heat transfer coefficient, W/m
2
K
k = Thermal conductivity, W/mK
L = Length of the bearing in the normal direction, m
N = Speed of the shaft, rpm
n = Heat partition factor
P = Contact force between the journal and the bearing, N
q
s
= Heat flux entering the shaft, W/m
2

Q = Heat generated, W
R
bi
= Inside radius of the bearing, m
R
bo
= Outer radius of the bearing, m
R
s
= Radius of the journal, m
R
eq
= Equivalent radius of contact, m
T

= Ambient temperature,
°
C
t
ss
= Seizure time during start-up, seconds
t
sp
= Seizure time when lubricant flow disturbance occurs, seconds
t
fo
= Failure time for oscillating pin-bushing, seconds

vi
ss
t = Non-dimensional seizure time during start-up
sp
t = Non-dimensional seizure time during lubricant flow disturbance
fo
t = Non-dimensional failure time for oscillating pin-bushing
t
ref
= Transition time for the bearing to go from fully lubricated conditions to
boundary lubrication conditions, seconds
u = Surface velocity of the journal, m/s
W = Load acting on the shaft, N
α = Coefficient of thermal expansion, m/mK
ε = Non-dimensional thermal strain
κ = Thermal diffusivity, m
2
/s
λ = Non-dimensional modified aspect ratio
ν = Poisson's ratio
ρ = Density of the material, kg/m
3

θ
c
= Contact angle, radians
τ = Frictional torque, Nm
ω = Angular velocity, radians/seconds
ω
o
= Oscillating frequency, radians/seconds

Subscripts
s = Shaft or journal or pin
b = Bearing
vii
ABSTRACT
This thesis presents a finite element parametric study of the transient
thermomechanical interactions of shaft-bushing pairs in bearings to gain insight into the
nature of two categories of failures encountered in practice. The first type of failure deals
with occurrence of the so-called thermally induced seizure (TIS) during the start-up
period followed by an investigation of TIS due to a transient flow-disturbance. The
second part deals with the thermomechanical interactions of pin-bushing assembly under
heavy oscillating loads where the failure is by TIS and/or thermal galling. An extensive
set of parametric simulations covering a wide range of loads, speeds, operating clearance,
bearing dimensions, friction coefficients and thermal expansion coefficients are
performed to gain insight into the phenomenon of TIS and thermal galling. A statistical
procedure is applied to the simulated results and an appropriate empirical relationship is
derived that predicts the time to failure for each category. Good agreement between the
empirical and published results attests to the capability of the model and its potential for
predicting bearing failure.
1
CHAPTER 1. INTRODUCTION

The study of thermal effect on bearing performance has been considered to be an
important subject since the evolution of tribology as a field of study. The driving force
behind this is the frequent failure of tribological components due to metal-to-metal
contact and the associated rise in the frictional heating. Thermally induced seizure and
galling are examples of such failures relevant to this work. Thermally induced seizure
occurs when the rise in operating temperature causes a partial or complete loss in
operating clearance leading to the seizure of the journal in the bearing.
Galling failure, on the other hand refers to the condition whereby there is gross
damage on the sliding surfaces characterized by the formation of local welding of
surface.
The thesis concentrates on the transient thermomechanical interactions of journal
and bearings under varying operating conditions encountered in practice. The research
focused on the failure of journal bearings during start-up, failure of journal bearings
during a transient flow disturbance, and the failure of a heavily loaded pin-bushing pair
subject to oscillating heat fluxes.
2
CHAPTER 2. LITERATURE REVIEW
Ling and Saibel [1] performed a study of failure of bearings due to thermal galling
of sliding surfaces in contact. Galling was predicted to occur when the sliding surface
reached the recrystallization temperature of the metal. The surface temperatures were
calculated from Blok's criterion for predicting the flash temperature due to the asperity
contact. Galling criteria was expressed as a function of the surface velocity, load and
time. This criterion would be more appropriate for high energy sliding, like disc brakes
and clutches. This work provided the author with the idea to use a similar failure criterion
for bearings.
Dyson [2] presented a review of a type of failure and where gross damage
characterized by the formation of local welding spots occurs between the sliding surfaces.
This type of failure mode is sometimes referred to by technical terms like galling,
scuffing and scoring. According to Dyson [2], "the metallurgical reaction on a scuffed or
a galled surface is characterized by the formation of a hard, white, etch-resistant phase on
the sliding surfaces. The identity of the phase is commonly thought to be a mixture of
austenite, martensite and carbide. The presence of this transformed layer, often separated
from the original untransformed bulk of the material by a tempered layer suggests that the
scuffed surfaces have been subject to high temperatures and rapid cooling".
Gecim and Winer [3] performed a steady-state thermal analysis of a rotating
cylinder subjected to heating in a particular region on its surface while the rest of the
surface was subjected to convective cooling. The governing partial differential equation
and boundary conditions were solved using the finite Fourier transform. The analogy
between the analysis done by Gecim and Winer and a journal bearing undergoing
3
frictional heating is striking. A similar analysis was performed by Patula [4] to determine
the steady state temperature of a rotating roller used in rolling of hot ingots in metal
forming industry. Ulysee and Khonsari [5] generalized the conditions of a cylinder
undergoing heating and cooling by including non-uniform heating and cooling. They
obtained an analytical expression for the steady state solution using the Fourier transform
method.
Bishop and Ettles [6] analyzed the thermoelastic interaction of a journal in a
plastic bushing that was interference-fit with the shaft. The frictional heat generated was
calculated as the product of the coefficient of friction, contact pressure and the sliding
velocity. Bearing seizure was considered to have occurred when the temperature rise on
the surface of the shaft exceeded 360
ο
C. In that study, the shaft was considered to expand
radially outwards due to the rise in temperature. The expansion of the bushing was
ignored, as the temperature rise is not appreciable as the thermal conductivity of plastic
was low. Bishop and Ettles considered the temperature rise to be dependent primarily on
the load, speed and clearance and arrived at a critical PV/C number as a bearing seizure
number.
Dufrane and Kannel [7] performed a study of the thermoelastic interactions of a
journal bearing undergoing a catastrophic seizure leading to a complete loss in operating
clearance. The cause of this type of failure was identified to be dry metal-to-metal contact
during the bearing start-up and associated rise in the contact temperature. Bearings that
had been out of service for a relatively long time are particularly susceptible because of
the lack of adequate supply of the lubricant in the contact area. The work is of particular
importance as the encroachment of the shaft to the bushing and the concomitant reduction
4
in the operating clearance (on the order of microns) was found to occur rapidly. Dufrane
and Kannel performed a series of experiments to determine the effect of dry friction that
led to failure by seizure. The experimental results showed that typically seizure occurred
within 30 seconds in most journal bearings operating in dry conditions. A one-
dimensional thermal and thermoelastic analysis was performed to estimate the seizure
time as a function of the operating parameters. The shaft expanded radially outward and
seizure was assumed to be complete when the total operating clearance vanished. The
bearing thermal expansion was not considered in the theoretical analysis. A linear
analytical expression was derived relating the bearing operating conditions and the
seizure time. This linear equation holds good only for very short transient times.
Although it gave good results that matched experiments, it is somewhat restrictive and
does not accurately represent the behavior of the system. The linear variation of seizure
time with clearance implies that all bearings irrespective of the operating clearance would
eventually seize.
Khonsari and Kim [8], inspired by the work done by Dufrane and Kannel,
performed a 2-D finite element analysis to study the thermoelastic behavior of journal
bearings undergoing seizure during start-up. They developed two models, one with an
aligned journal and bearing and the other with an axial misalignment. The 2-D axial
model included the effect of heat transfer in the axial direction. A heat-partitioning factor
was calculated based on the ratio of the areas of contact of the journal and the bearing.
The results proved that the seizure time tended to "flatten out" with increasing time. This
non-linear expression closely captures the physics of the reduction in clearance with time.
If a misalignment were present in the axial direction, the shaft comes into contact with
5
the bearing on a very small area, further aggravating the thermomechanical interaction.
Khonsari and Kim approximated this as a point heat source and presented results reveled
that the misalignment of the journal in the axial direction led to faster seizure times.
Hazlett and Khonsari [9, 10] continued the research to gain further understanding
of the behavior of a journal bearing undergoing seizure. They developed a
thermomechanical finite element model using the finite element package ANSYS. The
thermal analysis was first done to study the effects of the frictional heating on the contact
area of the bushing and the entire area of the shaft. The thermal analysis incorporated the
partition of frictional heat generated at the contact area based on the ratio of the areas of
contact of the journal and the bearing. The thermal contact resistance at the contact area
between the shaft and the bushing was modeled using convection link elements. The
results of the thermal analysis were used as thermal loads for solving the thermoelastic
model. The operating clearance was modeled by contact elements available in the
ANSYS element library. These elements come into effect only when there is a contact
between the journal and the bearing. The status of these elements was used to find the
variation of clearance and contact forces over time. Frictional torque was introduced as
the basis for seizure. When the frictional exceeded a limiting value, the bearing is
assumed to be seized. Based on the results of the thermomechanical analysis, Hazlett and
Khonsari found that the encroachment of the shaft into the bearing was non-linear with
time. The results established that TIS is triggered by the ovalization of the bushing and
formation of new contact patches at the top of the bearing. The establishment of new
contact patches accelerated the seizure process by increasing the contact forces and thus
6
increasing the frictional torque. This may be visualized as a positive feedback of thermal
energy into the system accelerating the process of seizure.
Wang, Conry and Cusano [11,12] performed a finite element study of the burn-off
and force stack-up in a railroad axle. An axially symmetric finite element and a radially
symmetric finite element analyses were performed simultaneously to simulate the 3D
analysis. The thermal analysis was similar to the analysis procedure performed by Hazlett
and Khonsari. The partition of heat between the journal and the bearing was done by
iteratively comparing the surface temperatures of the radial and axisymmetric models.
This model of heat partitioning was found to match closely with the results of the
partitioning method used by Hazlett and Khonsari. Radiation was also considered in the
thermal analysis. This is of importance as the temperature rise in the journal-bearing
system was very large. Temperature dependent material properties and friction
coefficients were used because of the high temperatures involved. Failure of the railroad
followed two patterns depending on the clearances involved. When the clearances are
low (0.05 mm - 0.5 mm), the failure was due to seizure of the axle in the bearing cone.
The seizure patterns were similar to those reported by Hazlett and Khonsari. New contact
areas were established that triggered seizure. When the clearances are sufficiently large
(2 mm – 5 mm), the failure was due to axle burn-off. The bearings operated for a long
time subjected to frictional heating continuously and the temperatures reached
recrystallization temperatures when the failure happened due to axle burn-off at high
temperatures.
More recently, Wang [13] performed a review of published results on TIS in
conformal contacts. It revealed that seizure in unlubricated conformal contacts was
7
primarily due to a thermal ratcheting effect in a positive feedback of increases of
interfacial pressure and heat. In lubricated journal bearings, starvation of the oil was
found to be a direct cause for seizure. The starvation led to wear and temperature rise in
the contact area that caused the onset of seizure. Other factors that affect the seizure
process are the surface quality and chemical composition, surface roughness and
microstructure of the shaft and the bushing.
Thermally induced Seizure (TIS) can also occur in circumstances where there is a
disturbance in the steady-state operation of the journal bearing. Lacey and Kawamura
[14] performed a study on the effects of lubricant flow disturbance in aircraft gas turbine
engine bearings. These bearings are required to survive with little or no oil under severe
operating conditions such as those encountered during flight take-off, landing and sudden
maneuvering. The oil flow interruption typically lasts for 15 to 30 seconds. The aircraft
engine bearings are not only required to survive these operating conditions but also to
resume normal operation once the lubricant flow is re-established. Although this study
was done to improve the performance of ball bearings in aircraft engines, the inspiration
to perform a similar study on journal bearings was developed. The effects of the oil flow
interruption in lubricant supply took place in either of the two mechanisms described
below:
(1) Oil flow interruption ⇒ Adverse ∆T ⇒ Reduction in bearing clearance ⇒
Excessive Hertzian stresses ⇒ High heat generation ⇒ Bearing seizure.
(2) Oil flow interruption ⇒ Surface damage (Wear) ⇒ High heat generation ⇒
Reduction in bearing clearance ⇒ Seizure.
8
This study provided the author the inspiration to study the effects of flow
disturbance in journal bearings that may occur due to supply line blockage, contaminant
blockage, airlock etc.
Thermally induced seizure is not a phenomenon restricted to bearings operating
under dry or boundary lubricated conditions. It can also occur in bearings running in fully
lubricated condition. This phenomenon was discussed by Pascovici, Khonsari and Jang
[15] and an analytical model was formulated for this transient analysis. Performing an
energy balance on a lumped system, the thermal response of the journal bearing is
obtained. A limiting temperature is taken as the condition for onset of seizure. This
limiting temperature is when the clearance between the journal and the bearing is
completely lost. A "no-seizure" condition is also derived based on the limiting
temperature.
Jang, Khonsari and Pascovici [16] performed a comprehensive analytical and
experimental study on the thermohydrodynamic (THD) seizure in journal bearings. The
experimental set-up studied THD seizure in an unloaded bearing that is operating in fully
lubricated conditions. The journal was made of polyamide and the bearing was made of
glass-reinforced epoxy resin. The seizure time was determined when the driving torque
requirement exceeded a certain range. A theoretical THD analysis was done to determine
the thermal response of the lubricating oil. The viscous dissipation caused the oil
temperature to rise. The thermal expansion of the shaft and its encroachment into the
bushing was determined using a standard 2-D heat conduction equation. The boundary
conditions at the interface of the shaft and the fluid film were determined from the THD
analysis. Good agreement of theoretical and experimental seizure time was observed.
9
CHAPTER 3. FINITE ELEMENT ANALYSIS: PROCEDURE AND OVERVIEW
The objective of this thesis is to study the thermomechanical interactions of
journal-nearing systems subject to different types of boundary conditions like
unlubricated bearing start-up, a fully-lubricated bearing undergoing a disturbance in the
lubricant oil supply and a journal and bearing subject to oscillating heating. The finite
element analysis is a simple and handy tool that is used with good accuracy in
engineering. The commercial FEM software package ANSYS 5.7 was utilized to perform
a detailed analysis of the thermoelastic interactions of journal and bearing. In this
chapter, the finite element analysis procedure, the type of elements used and the
boundary conditions involved are presented with the theoretical background [17, 18, and
19].
3.1 Thermal Analysis – Theory and Finite Element Formulation
Heat conduction analysis must be performed to determine the material
temperatures and the heat flow rates. The temperature distribution is also needed in order
to perform an analysis for thermally induced stresses. Fortunately, it is possible to devise
a single mesh layout for both problems: a computer program can read a single data file,
compute the temperatures at the nodes, then use these temperatures in a
thermomechanical analysis to compute the displacement, stresses etc.
For an isotropic material with no internal heat generation, Equation 3.1 and 3.2 give the
governing equation for heat conduction under transient condition in rectangular and polar
coordinates respectively [18].
t k
C
y
k
x
k
p
2
2
2
2


=


+

∂ θ
ρ
θ θ
(3.1)
10
t k
C
r r
1
r
p
2
2
2
2


=


+


+

∂ θ
ρ
ϕ
θ θ θ
(3.2)
where θ = Rise in temperature,
ο
C
k = thermal conductivity of the material, W/mK
ρ = Density of the material, kg/m
3

C
p
= Specific heat capacity of the material, J/kgK
t = time, seconds
Equation 3.1 and 3.2 can be solved analytically if we know the geometry of the
solid and its material properties and four boundary conditions (2 in the x/r-direction and 2
in the y/φ-direction) and one initial condition. There are normally two types of boundary
conditions to solve a thermal analysis: (i) the Dirichlet boundary condition, where the
temperature boundary conditions are specified and (ii) the Neumann boundary
conditions, where the surface boundary conditions like heat flux or convection are
specified.
Equation 3.3 gives the equivalent finite element formulation for the transient
thermal analysis. Note that Equation 3.3 is the formulation for a single element [17]. The
elements are arranged in a global matrix formulation to arrive at the complete solution.
The finite element formulation does not depend on the coordinate system.
[ ] [ ] ( ){ } [ ] { } { }
h q
R R
t
T
C T H k + =








+ + (3.3)
where [k] = element thermal stiffness matrix = [ ] [ ][ ]dV B B
e
V
T

Κ
[H] = [ ] [ ][ ]dS N h N
e
S
T


11
[C] = [ ] [ ]dV N C N
e
V
P
T

ρ
{R
q
} = [ ] dS q N
e
S
b
T

, Boundary condition matrix on surface, S
e
{R
h
} = [ ] dS hT N
e
S
f
T

, Boundary condition matrix on surface, S
e

[Κ] = Thermal conductivity matrix of the material
[N] is the shape function of the element used in the finite element analysis. [B] is
the strain displacement matrix, which is a spatial derivative of the shape function matrix,
[N]. The shape function of a finite element is defined as the interpolation function that
describes the distribution of the degree of freedom (temperature, displacements etc.)
across the element. Four-noded elements as shown in Figure 3.1 were used in the
analyses in this thesis for modeling thermal solids. The temperature field across the
element is modeled as a linear function of the nodal temperatures. For such a 4-noded
element, the shape function matrix and the strain displacement matrix are given by the
following equations 3.4 and 3.5.










Figure 3.1 – 4-noded bilinear element

[N] =





 + − + + − + − −
ab
y b x a
ab
y b x a
ab
y b x a
ab
y b x a
4
) )( (
,
4
) )( (
,
4
) )( (
,
4
) )( (
(3.4)

x
y
2 (a, -b)
3 (a, b) 4 (-a, b)
1 (-a, -b)
4-noded Bilinear
Element
12
[B] = [ ] N
x y
y
x






















0
0
(3.5)

The shape functions in finite element packages like ANSYS have a robust solver
that can solve for elements that are not necessarily a regular rectangular shape. A
parametric formulation effectively solves for complex geometries. The shape functions
are modified accordingly but the bilinearity is maintained. If T
1
, T
2
, T
3
and T
4
are the
nodal temperatures of the element shown in Figure 3.1, then the temperature field across
the element is given by:
T(x, y) =
ab
y b x a
4
) )( ( − −
T
1
+
ab
y b x a
4
) )( ( − +
T
2
+
ab
y b x a
4
) )( ( + +
T
3
+
ab
y b x a
4
) )( ( + −
T
4
3.2 Thermoelastic Analysis – Theory and Finite Element Formulation
The thermoelastic analysis is done to determine the displacements at nodes,
stresses, strains, etc. The results of the thermal analysis are input as nodal loads to
determine the temperature effects on the elastic behavior. Stress on an element can be
calculated when its nodal degrees of freedom (d.o.f.s) are known. These nodal d.o.f.s are
available after the structural governing equation (3.6) has been solved [17].
[ ]{ } { } R D K = (3.6)
where [K] = Structural stiffness matrix of the bilinear element, as shown in Fig 3.1
= [ ] [ ][ ]
∫ ∫
− −
b
b
a
a
T
dxdy t B E B
{D} = Nodal displacement matrix.
{R} = Load matrix.
13
[E] = Young's modulus matrix =










E
E
E
0 0
0 0
0 0
, for isotropic materials
The stresses in a finite element can be solved from the solution of Equation 3.6
and imposing the temperatures as nodal loads acting at the nodes. The temperature at
nodes is found by performing a thermal analysis that is solved before running the
thermomechanical analysis. If the nodal temperatures of an element is given by the
following matrix.
{T
e
} = [T
1
, T
2
, T
3
, T
4
]
T

then the thermal nodal loads for solving the thermomechanical analysis are given by the
following matrix [17].
Thermal Load Matrix: {R
e
} = εA
e
α [T
1
, T
2
, T
3
, T
4
]
T

Equation 3.7 and 3.8 are used to solve for the stresses in plane stress conditions
and plane strain conditions respectively.
For a general case, the generalized Hooke’s law is given by the following
equation:
{ } [ ] { } { } ( )
0
E ε ε σ − =
where {ε
0
} represents the strains due to the thermal expansion caused by the nodal loads
exported from the thermal analysis.
The stress field for plane stress problems is given by Equation 3.7 [17].
[ ]{ }
|
|
|
¹
|

\
|
¹
)
¹
`
¦
¹
¹
¹
´
¦

1
1
1
1
]
1

¸



=
¹
)
¹
`
¦
¹
¹
¹
´
¦
0
2
1
0 0
0 1
0 1
1
2
T
T
D B
E
yx
y
x
α
α
ν
ν
ν
ν
τ
σ
σ
(3.7)
The stress field for plane strain problems is given by Equation 3.8 [17].
14
[ ]{ }
|
|
|
¹
|

\
|
¹
)
¹
`
¦
¹
¹
¹
´
¦

1
1
1
1
]
1

¸




+ −
=
¹
)
¹
`
¦
¹
¹
¹
´
¦
0
2
2 1
0 0
0 1
0 1
) 1 )( 2 1 (
T
T
D B
E
yx
y
x
α
α
ν
ν ν
ν ν
ν ν
τ
σ
σ
(3.8)

The specific boundary conditions, element types and loading for the various
analyses are described in brief in Sections 4, 5 and 6 as per the specific problem.
15
CHAPTER 4. THERMALLY INDUCED SEIZURE IN JOURNAL BEARINGS
DURING START UP

Thermally induced seizure (TIS) in journal bearings is a mode of failure that can
occur quite suddenly and end up with a catastrophic damage to the system. Although TIS
can take place in lubricated bearings, it is predominant when a hydrodynamic bearing
happens to operate in the boundary or mixed lubrication regimes. These conditions occur
during start-up or in an event of lubricant supply blockage.
A significant amount of work has been reported that analyzed the
thermomechanical interactions in stationary loaded bearings susceptible to TIS. Bishop
and Ettles [6] analyzed the thermoelastic interaction of a journal in a plastic bushing that
is interference-fit with the shaft. The seizure criterion was based on a cut-off temperature.
When the surface temperature of the shaft or the bushing reaches 320
ο
above the ambient
temperature, the shaft is assumed to seize in the bearing. A critical PV/C number was
proposed to be an influential parameter for assessing the seizure time. Dufrane and
Kannel

[7] analyzed the catastrophic seizure of bearings due to dry friction by a simple
1D equation relating the seizure time to the bearing operating parameters and material
properties. A series of experiments were also conducted to determine the seizure time
( )
1
]
1

¸

+
|
¹
|

\
|
− −

+
=

n
R
R
R
R
1 n
1
q 1 2
C C
t
1
bo
s
s
bo
s
p
ss
α ν
ρ
) (
(4.1)
Equation (4.1) predicts a linear relation between the seizure time and the
operating clearance. This means that the bearing will seize even if the clearance is very
large. Khonsari and Kim [8] performed a comprehensive 2D numerical analysis of a
bearing undergoing seizure during start-up. The analysis included axial length of the
16
bearing as well as the effects of a misaligned shaft on the thermoelastic interactions. This
study established a non-linear nature for the clearance loss as a function of time. Hazlett
and Khonsari [9,10] performed a detailed finite element analysis to gain insight into the
nature of the contact forces and encroachment of the mating pair leading to TIS of a dry
bearing during start up. Their analysis revealed that the thermoelastic deformation
between the journal and the bearing led to a reduction in clearance in a non-linear
fashion. The effect of bearing parameters on the seizure time was also performed as a
sensitivity study. More recently, Wang et.al. [11,12] analyzed the axle burn-off and
stack-up force of a railroad roller bearing using the finite element method and
independently verified the results of Hazlett and Khonsari's findings. More recently,
Wang

[13] performed a review of published results on TIS in conformal contacts. It
revealed that seizure in unlubricated conformal contacts was primarily due to a thermal
ratcheting effect in a positive feedback of increases of interfacial pressure and heat. In
lubricated journal bearings, starvation of the oil was found to be a direct cause for
seizure. The starvation led to wear and temperature rise in the contact area that caused the
onset of seizure. Other factors that affect the seizure process are the surface quality and
chemical composition, surface roughness and microstructure of the shaft and the bushing.
The objective of this work is to perform a comprehensive study of seizure in
bearings during start-up and arrive at a seizure time evaluation formula that is a function
of the various operating parameters. The aim of the work is to provide a simple empirical
relation for practicing engineers and lubrication system designers to predict the seizure
time, which can help them to design instrumentation systems and warning devices to take
necessary precautions.
17
4.1 Finite Element Modeling Procedure
The finite element modeling is done using ANSYS 5.7. [19] First, the analysis
done by Hazlett [9,10] was recreated. The finite element model of the present work
employs a finer mesh than the mesh used by Hazlett and Khonsari to evaluate the contact
forces with more accuracy. A simplified 2-dimensional analysis is performed. The
analysis assumes that the contact pressure is uniform in the axial direction and no
crowning or misalignment is present in the system. The effect of bearing length is
analyzed in the 2-D analysis by taking into account the change in the contact width and
change in the heat flux generated with change in bearing length.
The analysis of a bearing undergoing TIS during start up is done by the following
steps:
1. A 2-D static contact analysis was performed to determine the contact forces and
the contact angle.
2. A transient heat transfer analysis was done to model thermal effects of dry
frictional heating on the journal and the bearing.
3. A transient thermoelastic analysis was performed to study the interactions of the
journal-bearing pair during bearing start-up. The variation of radial clearance,
contact forces and ovalization of the bearing were studied in this analysis.
Analysis Model
The model consists of a shaft rubbing on the inner surface of the bushing as
shown in Figure 4.1. The contact forces results in the generation of frictional heat on the
entire surface of the shaft and in the area where it contacts the bushing inner radius. Due
to the rise in temperature, the shaft expands and its encroachment to the bushing leads to
18
a loss of clearance. At some point in time, the bearing clearance reduces to a minimum
and shaft starts to encroach the bearing. Analyses show that typically during TIS, the
following three phenomena occur: (i) The contact forces increase, increasing the heat
generated. (ii) The contact angle increases causing a higher percentage of heat entering
the bush. (iii) New areas of contacts are established resulting in a chain reaction of events
leading to a rapid loss in the operating clearance. In the simulations presented in this
paper, these processes were implemented by performing a thermal analysis and a
thermoelastic analysis in a stepwise linear fashion. The model utilized a one-half
symmetry and neglected the heat conduction in the axial direction.









Figure 4.1 – Schematic of a journal supported on a pillow block
4.1.1 Transient Thermal Finite Element Analysis
Thermal Elements - The thermal analysis is done to determine the temperature
distribution in the journal-bushing pair. The journal and bushing were modeled as 4-
noded solid thermal elements viz. PLANE55. PLANE55 element has a single degree of
Load, W
Journal
Pillow Block
R
s
R
bi
L
2C
19
freedom, namely temperature. This element is compatible with the 4-noded structural
solid element used in the thermomechanical analysis. This means that the results of the
thermal analysis can be successfully exported to perform the thermoelastic analysis
Loading - The loading applied in the thermal analysis consists of the heat
generated by the frictional contact at the shaft-bushing interface. The total heat generated
in the contact zone is:

=
=
n
i
i
P fu Q
1
(4.2)
where f is the coefficient of friction of the rubbing surfaces, ∑P
i
is the summation of the
contact forces; n the total number of elements in contact and u is the surface velocity of
the shaft. The summation of the contact forces was equal to the total load acting on the
system, W. The frictional heat generated heats the entire surface area of the shaft in an
on-off mode. Thus the surface of the shaft is intermittently heated in the contact area and
cooled in the clearance area. The frictional heat generated heats only the surface of the
bushing that is in contact. Temperature and flux continuity exists in the contact patch of
contact of the journal and the bearing. The heat flux due to the frictional heating is
computed using the following formulae.
L R 2
P u f
A
P u f
q
s
i
s
i
s
π

=

=
. . . .
(4.3)
Boundary Conditions - The rotating shaft is heated periodically when it makes
contact with the bushing. This can be thought of as an on-off type heating. It was shown
by Hazlett [10] that the on-off heating could be modeled as an average heat flux on the
entire surface. Also, there is dissipation of heat by convective cooling by the air within
the clearance of the journal and the bushing
20










Figure 4.2 - Finite Element Model for Thermal Analysis, with loads and boundary
conditions. (Not to scale)

To represent the periodic heat dissipation in the finite element model, the nodes
on the surface of the shaft are coupled. The temperature on the surface of the journal and
the bushing at the interface is constant and is modeled by coupling the temperatures at the
nodes on the interface. The outer surface of the bushing is subject to natural convection.
The finite element mesh, thermal boundary conditions and the thermal loads are
schematically represented in Figure 4.2. For clarity, the operating clearance in Figure 4.2
is scaled by a factor of approximately 300.
4.1.2 Non-linear Transient Elastic Finite Element Model
Elastic/Thermoelastic Elements - The steady state analysis to find the contact
forces and the transient thermoelastic analysis use two types of elements in the Finite
Element Program ANSYS 5.7. The solid element PLANE42 is used to model the journal
Convection on
inner surface, h
Heat Flux at interface, q
Convective Cooling
on outer surface h
Shaft radius, R
s
Bearing inner radius, R
bi
21
and the bush. This element is a 2D bilinear element with the displacements in the x and y
directions as the degrees of freedom. The radial clearance between the journal and the
bush is modeled using two-noded contact elements, namely CONTAC52. Contact
elements are used to model gap and they come into effect only when the two nodes that
make the element come into contact. The element properties include a normal stiffness
value that governs the resistance to normal load. The finite element programmer assigns
the stiffness value for the contact element. To find the stiffness value, the theoretical
Hertzian contact width [20] must be determined. The stiffness value is then fixed at a
value by trial and error so that the Hertzian contact width and the contact width found by
Finite Element method are the same. The ANSYS 5.7 manual suggests that the normal
stiffness value for the contact element CONTAC52 can vary in the range of 0.01 to 100
times the stiffness of the underlying solid element material. The manual also suggests the
programmer to use his judgment in the trial and error method used to finalize the stiffness
value. The verification of the stiffness value against the Hertzian theory [20] was
followed to establish validity.
Loading - The loading for the non-linear thermoelastic analysis consists of the
thermal loads applied as nodal temperatures and the radial force acting on the journal.
The time dependent thermal load is obtained from the results of the transient thermal
analysis. The static load, W is applied to act in the negative y-direction on the shaft. As
the model utilizes half-symmetry, a load of W/2 is applied.
Boundary Conditions - Symmetry boundary conditions are used to model the
one-half symmetry as shown in Figure 4.3. The constraint of the bearing on its outer
surface is modeled by fixing the bearing at the node under the shaft on the outer edge of
22
the bearing on the symmetry plane. This constraint approximates the boundary conditions
that could be expected from a pillow block type of bearing as shown in Figures 4.1 and
4.3.









Figure 4.3 - ANSYS Model for Elastic Analysis, with loads and boundary conditions
4.1.3 Seizure Criterion
Frictional torque is the torque resisting the driving torque exerted by the motor.
When the frictional torque increases beyond the extent of the driving torque capability, it
can be concluded that the journal has seized in the bearing. The present model assumes
that TIS is complete when the frictional torque reaches at least 50 times the driving
torque. The contact forces acting on the gap elements at any instant of time determine the
frictional torque at any time.

=
=
m
i
i s
t P R f t
1
) ( . . . 2 ) ( τ (4.4)
Load, W/2
Line of symmetry
Displacement BC
23
where Τ is the frictional torque, f is the coefficient of friction, R
s
is the radius of the shaft
and P
i
is the contact force at the i'th gap element and n is the number of elements in
contact.
4.2 Results and Discussion
The encroachment of the shaft on to the bushing with concomitant reduction in
the clearance continues until the seizure is complete. The process is a complex, non-
linear phenomenon. Analysis shows that TIS is initiated by the ovalization of the bearing
combined with the uniform outward expansion of the shaft yielding contact between the
top of the shaft and the inner bushing surface. This leads to an increase in the contact
forces and the formation of an extra contact area. Increase of contact forces raises the
frictional heat flux and sets up a positive feedback that accelerates the loss of clearance.
The increase in the frictional torque is abrupt once the ovalization of the bearing causes
the shaft to encroach the bushing, as there is further loss in the operating clearance. The
frictional torque increased to exceedingly large values within typically 3 seconds after the
first instance of establishment of new areas of contact. The reasons for such an abrupt
increase in frictional torque are:
(i) As explained previously, the increase in contact forces increases the frictional heat
generated and the increase in frictional heat means that the shaft would expand more
increasing the contact forces and establishing more area of contact. This process leads to
a positive feedback loop and a chain reaction leading to a rapid failure due to TIS.
(ii) The operating clearance of the bearing just before seizure is reduced to a significantly
lower value compared to the steady-state operating clearance. This is due to the thermal
expansion of the journal and the bearing into the operating clearance area. The available
24
clearance just before the extra contact occurs has already reduced to an exceedingly small
value. The following plots, Figures 4.4 - 4.6 illustrate the onset and completion of seizure
for a journal bearing during start-up. The operating parameters are listed below:
W = 4400 N
N = 250 rpm
R
s
= 25.5 x 10
-3
m
R
b
= 51.0 x 10
-3
m
C = 0.0125 x 10
-3
m
L = 51.0 x 10
-3
m
Figure 4.4 – Temperature rise as a function of time in a journal bearing during start up
Plot (a) – Temperature rise, 25 seconds after starting
Plot (b) - Temperature rise, 27 seconds after starting. Note the ovalization and
establishment of extra contact, A, at the top of bearing
Plot (c) - Isothermal plot of temperature rise, 29 seconds after starting – Note the
formation of extra areas of contact and rapid temperature rise
25
Forces after 25 seconds
Forces after 27 seconds
Initial Contact Forces
Forces after 29 seconds
0
1000
2000
3000
4000
5000
6000
7000
8000
C
o
n
t
a
c
t

F
o
r
c
e
,

N
0 50 100 150
Angle from midplane, deg











Figure 4.5 - Variation of contact forces with time during start-up
Figure 4.6 - Variation of frictional torque during start-up – Note that the frictional torque
increased to 50 times the initial frictional torque to indicate seizure

26
Figure 4.4 (a) shows temperature rise on the journal bearing pair just before
ovalization. Figure 4.4 (b) shows the initiation of extra contact at the top of the bearing.
Figure 4.4 (c) shows the temperature increase after 2 seconds after the establishment of
extra areas of contact, which illustrates the substantial rise in temperature and contact
area. Figures 4.5 and 4.6 show the variation of contact forces and frictional torque over
time. It is interesting to note from Figures 4.4-4.6 that, although the frictional torque
increased to 50 times the initial torque, the clearance has not reduced to zero clearance at
all points proving that the frictional torque is a better seizure criteria than zero-clearance.
These simulations reveal that the frictional torque reaches very high values after
approximately 3 seconds after ovalization is first experienced. The ovalization is realized
in the analysis when additional contact is established at the top of the bearing. Therefore,
it will be assumed that the seizure is complete 3 seconds after the first instance of
ovalization. Additional simulations with difficult operating conditions also reveal that
this condition holds.
4.3 Model Parameters
The model used to study the effects of TIS during bearing start-up is given in
Table 4.1. The ranges of the operating variables were arrived by taking the operating
parameters used by Dufrane and Kannel [7] for experimentation and by Kim and
Khonsari [8] and by Hazlett and Khonsari [10] for doing the analytical and finite element
modeling as the basis. The parameters used are:
Table 4.1 – Operating parameters for the finite element model and simulations
Range of shaft radii, R
s
(m) 10 x 10
-3
- 37.5 x 10
-3

Range of bearing lengths, L (m) 51x 10
-3
-100 x 10
-3

ρC
p
(N/m
2
K) 1.23 x 10
6

Conductivity, K (W/mK) 52
27
α (m/mK) 0.3 x 10
-5
– 2 x 10
-5

Young's modulus of shaft, E
s
(GPa) 200
Young's modulus of bush, E
b
(GPa) 110
Poisson's ratio, ν
s
= ν
b
= ν 0.3
Range of loads, W (N) 1000-10000
Range of friction coefficient, f 0.1 – 0.3
Range of speeds N (rpm) 150-2000
Range of radial clearances, C (m) 0.005 x 10
-3
- 0.0325 x 10
-3

Convection on all faces, h (W/m
2
K) 80
Atmospheric Temperature, T

(°C) 25

4.4 Generalization
To derive a relation between the seizure time, t
s
, and the operating parameters, a
statistical procedure followed based on the work done by Hamrock [21].
The seizure time can be written as a function of the various operating parameters.
t
s
= g (N, W, R
s
, R
b
, C, f, L, κ, α).
The variation of the seizure time during the system start-up is studied when the
operating parameters (variables N, W, R
s
, R
b
, C, f, L κ, α) are varied. Then a generalized
equation is derived depending on the individual relationships of the operating parameters
with the seizure time.
The simulations were performed to establish the effects of the operating
parameters on the seizure time during the start-up. The speeds are varied from 150 rpm to
2000 rpm. For speeds lower than 150 rpm, the seizure times are large enough such that
the lubricant flow would have been established during starting. For speeds higher than
2000 rpm and the operating clearance specified, the seizure time is nearly the same, about
2 seconds. This means that the seizure time is almost asymptotic with respect to speed.
All the other parameters were also varied based on a similar rational. The lower limits of
the load and the speed were decided if seizure times exceeded 120 seconds. The
reasoning behind this cut-off point is that the lubricant supply would have been
28
established within 2 minutes of starting the system. The shaft radius was varied from 10
mm to 37.5 mm. The upper limit of the shaft radius was limited as increasing the shaft
radius above 37.5 mm meant that the bearing thickness was less than 10 mm. The length
of the shaft was varied from long bearings (L/D=2) to short bearings (L/D = 0.5). These
are the limits that are generally known as the infinitely long and infinitely short
approximations [22] (ILA and ISA). The clearances were decided based on typical
operating clearances of journal bearings. The friction coefficient was varied from 0.1 to
0.3. These are typical values of friction coefficients for steel on brass sliding [23].
4.4.1 Influence of Operating Parameters
Increasing the speed of the shaft means that the heat generated at the interface of
the journal and the bearing increase in a direct proportionality. Seizure time decreases
with increasing speed of the shaft.
The effect of increase in load is a proportional increase in the frictional heat generated.
This means that the seizure time decreases with increasing load. The influence of load on
the seizure time is determined by varying the load from 1100 to 10000 N keeping the
other factors (N, R
s
, R
b
, C, f, L, κ, α) constant.
The influence of radial clearance on the seizure time is determined by varying the
load from 0.005 to 0.0325 mm keeping the other factors (N, W, R
s
, R
b
, f, L, κ, α) constant.
A decrease in clearance means that the shaft has lesser room to encroach before
contacting the bearing and hence the seizure time decreases with decrease in clearance.
Increase in the shaft radius produces two contrasting effects on the seizure time. Increase
of shaft radius means that the surface velocity of the bearing increases for the same
speed. This means that the frictional heat generated is increased. But an increase in the
radius of the shaft also means that the thermal mass of the shaft is increased. So the net
29
effect of these two modes of heat generated and heat stored decides whether the bearing
seizes faster or slower. It can be inferred from Appendix A that the increase in the shaft
radius decreases the time for seizure up to a particular shaft radius. Beyond a shaft radius
of 25 mm, the effect of the increased thermal mass nullifies the effect of the increase in
the surface velocity. The influence of shaft radius on the seizure time is determined by
varying the load from 10 to 37.5 mm keeping the other factors (N, W, C, R
b
, f, L, α)
constant.
The influence of the outside radius of the bushing on the seizure time is
determined by varying the outside radius from 32.5 to 100 mm keeping the other factors
(N, W, R
s
, C, f, L, α) constant. It was found that varying the outside diameter has
negligible effect on the seizure time. Hazlett [10] varied the convective heat transfer
coefficient on the outside radius of the shaft and did not find any appreciable effect on the
thermoelastic interactions of the bearing. The variation of outside radius may be
significant only for very thin bearings. The simulations were carried out for bearings with
R
s
/t
h
= 2.0 (thin bearings) to R
s
/t
h
= 0.5 (thick bearings) and they seized at around the
same time. Therefore for most applications, the effect of outside radius of the bushing on
the seizure time can be neglected.
The influence of coefficient of friction on bearing seizure time is similar to the
effect of speed. Increasing the coefficient of friction increases the heat generated thus
reducing the seizure time. The influence of the coefficient of friction is determined by
varying the friction coefficient from 0.1 to 0.3 while keeping the other operating
parameters (N, W, R
s
, R
b
, C, L, f, α) constant.
30
The influence of bearing length on the seizure time is determined by varying the
length from 50 to 100 mm keeping the other factors (N, W, R
s
, R
b
, f, C, α) constant. The
effect of the bearing length is two-fold; (i) The increase in bearing length increases the
Hertzian contact width meaning the area of heating on the bushing is more, (ii) The
increase in bearing length means that the bearing area, R
s
*θ*L would also increase.
4.5 Non- Dimensionalization
In order for the analyses and the corresponding statistical procedure to be relevant
to any journal-bearing system irrespective of the units used, use of non-dimensional
parameters is preferable. The non-dimensionalization scheme utilizes only two non-
dimensional operating parameters, while the analyses were carried out for 6 operating
parameters. All the operational variables namely load speed and coefficient of friction
were combined to get a single non-dimensional parameter and all the dimensional
variables like operating clearance, shaft radius and the bearing length were combined into
another non-dimensional parameter as described below.
Non-dimensional seizure time during start-up:
2
s
ss
ss
R
t
t
κ
=
Thermal strain:
s
k
fWωα
ε =
Modified aspect ratio:
L
R
eq
= λ , where
s bi
bi s
eq
R R
R R
R

= =
C
C R R
s s
) ( +

4.5.1 Statistical Treatment of Non-dimensional Operating Parameters
A power series equation is used to find the relation between non-dimensional
thermal strain and non-dimensional seizure time. The equation derived fits the data with a
goodness value of 98.64% as shown in Figure 4.7.
31
ss
t
ss
= 0.0016 ε
-1.2478
0.00
0.50
1.00
1.50
2.00
2.50
3.00
3.50
4.00
ε
0.000 0.005 0.010 0.015 0.020 0.025 0.030
t
The relation between the non-dimensional seizure time and thermal strain is given
by the following equation (4.5)
2478 . 1
0031 . 0

= ε
ss
t
2478 . 1 −
∝ε
ss
t (4.5)











Figure 4.8 – Variation of non-dimensional seizure time with thermal strain
Let us define parameter Γ as

2478 . 1 −
= Γ
ε
ss
t
(4.6)

The relationship between the parameter Γ and the modified aspect ratio was also derived
using power series expressions. The variation of Γ with modified aspect ratio depends on
a particular range of R
s
/C values. Two ranges of R
s
/C values were considered for the
simulations performed and the equation derived fits the data with a goodness value of
97.9% and is given by Equation 4.7.
32
Γ = 90.494 ε
-1.6494
For 500 > R
s
/C > 1000 (4.7)
= 1.825 ε
-1.02718
For 1000 > R
s
/C > 5100

-1.6494 -1.2478
ss
0.494 t λ ε 9 = For 500 > R
s
/C > 1000 (4.8)
-1.02718 -1.2478
ss
1.825 t λ ε = For 1000 > R
s
/C > 5100
4.6 Verification and Analysis
The above empirical relationship (4.8) is verified for its validity using some of the
results published by Hazlett and Khonsari [9,10], Wang, Conry and Cusano [11,12] and
Bishop and Ettles [6].
The first four cases correspond to the simulations reported by Hazlett and
Khonsari [9,10] for journal bearings, the fifth and sixth cases were reported by Hao Wang
[11,12] for an analysis dealing with axle burn-off in railroad roller bearings and cases 7
through 17 were reported by Bishop and Ettles in their study of TIS in journal bearings.
As can be seen from Table 4.2, the empirical results are in agreement with published
results.
In the axle burn-off studied by Hao Wang [12], the nominal operating clearances
ranged from 0.05 mm to 5 mm and the axle bearing was lubricated by grease due to very
heavy loading and the material properties were reported as functions of temperatures. The
material properties used in Table 4.2 for the fifth and sixth cases were averaged values.
Wang et al.
6
reported a number of additional analyses with axle bearings. However, a
number of the clearances were outside the range of simulations presented in this study
and were not considered. Therefore only the 0.05 mm simulations reported was
considered for verification. Applying the empirical relation for the other reported
clearances would lead to erroneous predictions.
33

Table 4.2 - Comparison of Empirical results with published results
No.
Speed
rpm
Load
N
Clearance
m
Shaft
Radius,
m
Bearing
Length
m
Thermal
conductivity
W/mK
Thermal
diffusivity
m
2
/s
Coefficie
nt of
friction
Coefficient of
thermal
expansion,
m/mK
Seizure time
published,
sec
Seizure time
calculated,
sec
1 250 4400 1.25x10
-5
0.0255 0.051 52 4.23x10
-5
0.15 1x10
-5
28 28
2 1800 4400 1.25x10
-5
0.0255 0.051 52 4.23x10
-5
0.15 1x10
-5
2 2.4
3 250 4400 1.25x10
-5
0.0255 0.03825 52 4.23x10
-5
0.15 1x10
-5
21 21
4 250 4400 1.25x10
-5
0.0255 0.0255 52 4.23x10
-5
0.15 1x10
-5
16 14
5 560 160000 5x10
-5
0.0781 0.1562 30 5.47x10
-6
0.3 1.3x10
-5
3 1.3
6 560 160000 50x10
-5
0.0781 0.1562 30 5.47x10
-6
0.3 1.3x10
-5
60 59
7 1000 1 x 10
6
0.5x10
-5
0.5 1 52 1.44x10
-5
0.005 1.1x10
-5
7 7.3
8 1000 1 x 10
6
2.5x10
-5
0.5 1 52 1.44x10
-5
0.005 1.1x10
-5
42 38
9 1000 1 x 10
6
5x10
-5
0.5 1 52 1.44x10
-5
0.005 1.1x10
-5
117 78
10 1000 0.2 x 10
6
5x10
-5
0.5 1 52 1.44x10
-5
0.005 1.1x10
-5
650 583
11 1000 0.3 x 10
6
5x10
-5
0.5 1 52 1.44x10
-5
0.005 1.1x10
-5
400 352
12 1000 0.4 x 10
6
5x10
-5
0.5 1 52 1.44x10
-5
0.005 1.1x10
-5
290 246
13 1000 0.5 x 10
6
5x10
-5
0.5 1 52 1.44x10
-5
0.005 1.1x10
-5
240 186
14 1000 5 x 10
6
5x10
-5
0.5 1 52 1.44x10
-5
0.005 1.1x10
-5
35 11
15 300 1 x 10
6
5x10
-5
0.5 1 52 1.44x10
-5
0.005 1.1x10
-5
510 351
16 500 1 x 10
6
5x10
-5
0.5 1 52 1.44x10
-5
0.005 1.1x10
-5
270 186
17 2000 1 x 10
6
5x10
-5
0.5 1 52 1.44x10
-5
0.005 1.1x10
-5
50 33
34
Bishop and Ettles [6] studied TIS in journal bearings with non-metallic liners
using analytical 1-D and a 2-D analysis. Results from the 2-D analysis of TIS in
lubricated journal bearings were considered for verification in Table 2. Note that
although the simulations done in the present study did not consider fully lubricated cases,
the results derived in Equation (4.8) still holds good. The results in Bishop and Ettles'
paper were reported till the temperature reached 320° above ambient temperatures that
were considered as the burn-off temperature of the linings. The results reported in Table
4.2 were taken when there is complete loss in clearance that was before the limiting
temperature was reached.
To gain further insight into the TIS behavior, we plot the change in the operating
clearance as a function of time based in Dufrane and Kannel's equation (4.1) and the
empirical Equation (4.8) derived in the present study. The result is shown in Figure 4.9
for two heat partitioning factors (n = 0.5 and 1).
ANSYS 5.7 calculates the heat-partitioning factor based on the thermal mass and
material properties at the contact area such that there is continuity of temperature and flux
at the contact interface. A heat partitioning of 1 is unreasonable as it means that all the
frictional heat generated would be transmitted into the shaft. The analyses done by
Dufrane and Kannel [7] did not consider the expansion of the bushing and the thermal
expansion of the shaft was only considered. Also, the bushing was rigidly constrained. As
the present study has considered the bushing constraint the outward expansion of the
bushing, the seizure times were larger than the values obtained using Equation (4.1) used
by Dufrane and Kannel. From Figure 7, it can be seen that the present model compares
close to Dufrane and Kannel's model when a heat-partitioning factor of 0.5 is used in
35
C
l
e
a
r
a
n
c
e

l
o
s
s
,

i
n
Time for seizure, seconds
Dufrane and Kannel's model
n=1
Dufrane and Kannel's model
n=0.5
Present Model
0.0005
0.0010
0.0015
0.0020
0 50 100 150 200
Equation (1). The seizure time formula developed in this study predicts the loss of
clearance with time is not a linear process. Whereas Equation (4.1) implies that TIS
occurs regardless of the size of the clearance, the predicted results here reveal that this is
not true for large clearances. This physically realistic prediction was first discussed by
Khonsari and Kim [24].












Figure 4.9 – Comparison of seizure time with Dufrane and Kannel's model and the
present model. Note: n = heat partition factor

There are some limitations that restrict the use of these relations depending on the
operating conditions.
1. It is recommended to use the empirical relations for operating conditions close to
the conditions used in the simulations. This is especially true for the operating
clearance.
36
2. The boundary condition for the fixation of the bearing varies with different
applications. The boundary condition used in this analysis allows some flexibility for
the bearing to expand outwards. If the bearing were fully constrained, the seizure
times would be appreciably reduced.
37
CHAPTER 5. TIS OF JOURNAL BEARINGS TRIGGERED BY A TRANSIENT
FLOW DISTURBANCE

The thermoelastic interaction of journal bearing pair in the event of a disturbance
in lubricant flow is a largely neglected area of study. There are a number of situations
whereby a journal bearing system while running steady may suddenly experience a
disturbance in its operating conditions. In this part, we focus our attention to the situation
whereby a system undergoes a transient flow disturbance. A "lubricant supply
perturbation" in an otherwise steady-state operation may be caused by a brief interruption
of lubricant supply due to a clogged filter or air locking, for instance. Transient operating
conditions are also predominant in aircraft during the periods of taking-off, landing and
maneuvering [14].
When the lubrication perturbation occurs for a short period of time, the
lubrication regime can shift from hydrodynamic into the mixed and boundary lubrication
regime. This transition of the bearing operating conditions from the thick-film to the
boundary lubrication regime can significantly affect the operating coefficient of friction
and the convective heat transfer coefficient within the clearance. Ideally once the
lubricant supply is resumed, the bearing reverts back to the thick-film regime, otherwise
TIS becomes imminent. Evaluation of the thermomechanical response following the
resumption of lubricant supply is necessary to understand if the system can survive the
interruption of lubricant flow.
In this chapter, flow interruption is modeled as a transition between various
lubrication regimes. This transition is assumed to take place linearly with time. The
methodology behind the application of the lubricant perturbation can be graphically
represented by Figure 5.1 and 5.2. Lubricant flow disturbance is assumed to occur over a
38
period of 150 seconds from the onset of the disturbance to full resumption of steady
operation. This period consists of an initial rise in the coefficient of friction and
concomitant reduction of convective heat transfer coefficient over a period of 70 seconds,
during which there is a gradual transition from full-film to boundary lubrication. Then, a
duration of 10 seconds in the boundary lubrication followed by a 70 seconds transition to
full-film lubrication is assumed. The coefficient of friction for the fully lubricated bearing
is typically in the order of 0.005. During the transition regime, the coefficient of friction
is ramped linearly with time to 0.15, which represents lubricated metal-to-metal sliding.
A convective heat transfer coefficient of 200 W/m
2
K is assumed for thick-film
lubrication. During the transition period of 70 seconds, the convective heat transfer
coefficient is ramped linearly till the HT coefficient reaches 80 W/m
2
K as shown in
Figure 5.2. The nominal convective coefficient for air at the speeds involved is around 80
W/m
2
K [9,10]. A change in the steady-state thick-film heat transfer coefficient was found
to have minimal effect on the bearing seizure time as the seizure occurs primarily due to
the flow disturbance and not due to the steady-state temperature rise.
39
1000 1100 1200 1300 1400 1500 1600
Time, seconds
0.000
0.050
0.100
0.150
C
o
e
f
f
i
c
i
e
n
t

o
f

f
r
i
c
t
i
o
n
2
80
100
120
140
160
180
200
C
o
n
v
e
c
t
i
v
e

C
o
e
f
f
i
c
i
e
n
t
,

W
/
m


K
1000 1100 1200 1300 1400 1500 1600
Time, seconds














Figure 5.1 – Variation of friction coefficient during lubricant perturbation















Figure 5.2 – Variation of convective HT coefficient during lubricant perturbation
40
5.1 Finite Element Analysis Modeling and Seizure Criteria
The procedure and modeling approach is similar to Section 4, where TIS during
bearing start-up was studied with a few minor differences. The thermal analysis is
performed at thick-film hydrodynamic conditions until steady state is achieved and then
the transient flow disturbance is applied in the transient. The finite element procedure
overview is given below
1. A 2-D static contact analysis was performed to determine the contact forces and
the contact angle.
2. A transient heat transfer analysis was done to determine the time required for the
journal bearing to attain the steady-state conditions. The perturbation conditions
(increase in coefficient of friction and decrease in the convective heat transfer
coefficient) were applied after the time required for the journal-bearing pair to
reach steady state. The methodology of the application of the lubricant
perturbation is done as per the procedure described in Section 5.0 (Figures 5.1 and
5.2).
3. A transient thermoelastic analysis was performed to study the interactions of the
journal-bearing pair under the influence of perturbation. The variation of radial
clearance, contact forces and ovalization of the bearing were studied in this
analysis.
The bearing seizure is assumed to be complete when the frictional torque greatly
exceeds the driving torque. This is similar to the methodology explained in Section 4.1.3




41
5.2 Results and Discussion

The failure of journal bearings undergoing a transient disturbance in flow is
primarily due to the seizure of the shaft in the bearing due to thermal expansion of the
shaft and the bearing and concomitant reduction in clearance. This process is similar to
the TIS of bearings during start up. The time to seize differs from the TIS of bearings
during start up. This is due to the fact that the increase in the coefficient of friction and
the reduction of convective heat transfer is gradual as the bearing moves from a thick-
film regime to boundary lubrication over a period of time. The thermomechanical process
of seizure is explained for a journal bearing with the same operating conditions as
discussed in Section 4 is discussed in this section.






















Figure 5.3a, b – Plots of temperature rise during a transient flow disturbance

Plot (a) – Steady state temperature rise Plot (b) – after 62 seconds
42

Figure 5.3c,d – Plots of temperature rise during a transient flow disturbance
Figure 5.4 - Variation of contact force after the onset of ovalization due to flow
disturbance
Plot (c) – After 64 seconds
Plot (d) – After 65 seconds
After 65 seconds
After 64 seconds
After 62 seconds
0 50 100 150 200
Angle from midplane, deg
0
1000
2000
3000
4000
5000
6000
7000
C
o
n
t
a
c
t

F
o
r
c
e
s
,

N
43
Figure 5.5 – Variation of frictional torque when a transient flow disturbance occurred
Comparing Figures 5.3 through 5.5 with Figures 4.4 through 4.6, it can be seen
that the process of TIS follows a similar procedure, the only difference being the time
taken to seize. It is an interesting fact that the ovalization of the bearing and the
establishment of the extra areas of contact contributed to the rapid seizure of the bearing
in the case of flow interruption also.
5.3 Model Parameters
The model parameters are similar to Table 4.1, except that the operating
parameters are of a wider range than that of the start up analysis.
Table 5.1. Operating parameters for simulating TIS in a journal bearing undergoing a
transient flow disturbance

Shaft Radius, R
s
(m) 10 x 10
-3
-37.5 x 10
-3

Length of Bush, L (m) 51x 10
-3
-100 x 10
-3

ρC
p
(N/mm
2
K)
1.23
44
Conductivity, K (W/mK) 52
α (mm/mmK) 1e-5
Young's Modulus of Shaft, E
s
(GPa) 200
Young's Modulus of Bush, E
b
(GPa) 110
Poisson's Ratio, ν
s
= ν
b
0.3
Load, W (N) 2200-12000
Friction Coefficient, f 0.15
Speed, N (rpm) 500-3000
Radial Clearance, C (m) 0.0125 x 10
-3

Convection on outer face, h (W/m
2
K) 80
Atmospheric Temperature, T

(°C) 25
Transition time, t
ref
(sec) 20-100


5.4 Non-Dimensionalization and Generalization
Similar to Section 4.4, a statistical analysis was performed and an empirical
relationship was derived for arriving at the seizure time when a journal bearing undergoes
an interruption in lubricant supply. As the analysis procedure is different from the
thermoelastic analysis done for predicting seizure time, a different non-
dimensionalization technique was followed. The non-dimensional parameters used in the
analysis are defined below.
Non-dimensional seizure time,
ref
s
s
t
t
t =
Non-dimensional seizure time during flow disturbance:
ref
sp
sp
t
t
t =
Thermal strain:
s
k
fWωα
ε =
Modified aspect ratio:
L
R
eq
= λ , where
s bi
bi s
eq
R R
R R
R

= =
C
C R R
s s
) ( +


45
Using the results of 50 sets of results, the relationship between the non-
dimensional time for seizure and the non-dimensional operating parameters is given by
the following equation (5.1).
( ) ( )
1
5489 0
sp
314 9 0065 0 t

+ = . .
.
λ ε (5.1)

5.5 Analysis of Results
The empirical relation (5.1) provides a useful empirical relation about the seizure
time to be expected when a bearing undergoes a disturbance in lubricant supply. The
physical meaning of the seizure time and restrictions of the empirical relation (5.1) are
explained next.
1. If the empirical relationship predicts a seizure time of, say 48 seconds, it means
that the bearing failed in the transition period before boundary conditions are
achieved. If a bearing seizes between 71 and 80 seconds it means that the seizure
is during the boundary lubrication period. Similarly if seizure time is greater than
80 seconds, it means that the seizure has taken place after the lubrication is
restored.
2. For bearings that seize within 70 seconds, i.e. before the mixed/boundary
lubrication phase occurs, it means that those bearings cannot sustain even a
lubricant cut-off for a few seconds. Most of the cases considered in this analysis
falls within this category.
3. The empirical relationship does not apply to bearings that may be susceptible to
seizure in hydrodynamic regime. Some heavily loaded bearings that operate at
high speeds or with low clearances may seize in fully lubricated conditions. When
the empirical formula (5.1) is applied for these conditions, they give a finite value
46
of seizure time when the bearing had already seized even with no perturbation.
Typically, seizure times of 20 seconds or below fall under this category. If the
empirical relationship yields a result of 20 seconds or lower, it means that the
bearing has already seized or it would seize even for a slight change in the
operating conditions. The TIS failure of lubricated journal bearings has been
studied by Pascovici et.al. [15] where a "no-seizure" condition was arrived based
on the bearing operating parameters. Jang et.al. [16] performed an extensive study
of TIS in thick film journal bearings. The thermoelastic behavior was theoretically
formulated and experimentally verified.
4. The material properties used in the simulations pertain to mild steel shaft and
bronze bearing. So the above empirical relations can be used only for material
combinations that are close to the material properties of the materials that were
used in the simulations. Most of the bearing materials typically have a bronze
base and shaft made of steel. That was the reason behind choosing this material
pair.
5. The Finite Element Analysis neglected the heat transfer in the axial direction.
Only a 2D analysis was performed.
6. The analysis is valid in the transition time ranges of 20 to 80 seconds. The
transition time is the time taken for the bearing to go from fully lubricated mode
to boundary lubrication mode.
7. The boundary conditions for the fixation of the bearing vary with different
applications. The boundary conditions used in this analysis allow some flexibility
47
for the bearing to expand outwards. If the bearing were fully constrained, the
seizure times would be appreciably reduced.
5.6 Verification
A verification is performed for some operating conditions and the validity of the
empirical relation is checked against the simulated results obtained from ANSYS finite
element simulations for the following operating conditions.
W = 8800 N
N = 500 – 2400 rpm
C = 0.0255 x 10
-3
m
R
s
= 0.0255 m
R
b
= 0.051 m
L = 0.051 m

For the above operating conditions, the results of the empirical relation are
compared to the results of the simulations performed using ANSYS 5.7. The results are
tabulated in Table 5.2. The % error was computed and the max error was found to be
7.5%. The comparison is repeated for other simulations also. It was found that the
empirical relation provides a realistic and conservative estimate for seizure time for most
of the cases.
Table 5.2 – Comparison of simulated seizure time and empirical seizure time
Speed Empirical Seizure Time Simulated Seizure time %error
250 99 100 -1.01
500 67 68 -1.49
750 53 58 -1.89
1000 45 50 -4.44
1200 40 46 -5.00
1400 37 44 -5.41
1600 34 38 -5.88
1800 32 34 -3.13
2000 30 30 -6.67
2200 28 30 -7.14
2400 27 26 -7.41

48
An interesting observation can be made from Table 5.2 for the operating speed of
250 rpm. The seizure time predicted by the empirical formula and from the simulations is
about 100 seconds. This means that the bearing seized even after the resumption of
lubricant supply.
49
CHAPTER 6. THERMOMECHANICAL ANALYSIS OF A HEAVILY LOADED
OSCILLATING PIN-BUSHING PAIR

6.1 Introduction and Problem Definition
Heavy-duty earth-moving machinery use linkages that are supported on a pin and
bushing pair. They are subject to very heavy loads generally operating at low speeds.
Typically in the pin-bushing pair, the pin is stationary while the bushing oscillates at low
frequencies. These pin-bushing pairs are typically made of mild steel and surface treated
with a protective coating. Depending on the application, the pin-bushing pair maybe
packed with grease during assembly. Industry is now using automatic grease feeders or
recommends feeding grease periodically. There is no provision for a continuous supply of
lubrication in the system and the packed grease is the only source for lubrication. After
some time of operation in field, the pin-bushing pair exhausts the supply of grease packed
in the assembly by normal operational leak as well as wear and tear. The absence of
proper lubrication leads to the wear of the anti-friction liner on the surfaces of the pin-
bushing pair. Further operation causes intimate metal-to-metal contact between the pin
and the bushing that leads to high heat generation in the contact region. The absence of
proper lubrication, very high loads and improper cooling leads to failure of the bearing
either by galling or by thermally induced seizure (TIS). A typical schematic of the pin-
bushing oscillating pair used in the undercarriage of heavy-duty earth-moving equipment
is shown in Figure 6.1.
Galling is defined as the mode of failure when there is plastic deformation and
localized welding of relative moving parts and subsequent wear of the welded parts due
to the shear loads. Galling can occur due to very high stresses and/or due to very high rise
in temperatures. The combination of heavy loads and high temperatures in the oscillating
50
pin-bushing assembly in earth-moving machinery makes it vulnerable to galling.
Thermally Induced Seizure (TIS), on the other hand, is the mechanism of failure if there
is a complete loss in operating clearance in the pin-bushing assembly as the pin
encroaches into the bushing. The underlying phenomenon is similar to that described in
Chapters 4 and 5, except that in the oscillating pin-bushing assemblies, the temperatures
involved and the time to failure are generally higher. Once failure starts the surface
temperature at the pin-bushing interface shoots up to very high values introducing
possible galling or TIS, depending on the operating conditions.
















Figure 6.1 – Track Retention Assembly with the oscillating pin and bushing

Track link
Stationary pin
Oscillating Bush
51
6.2 Problem Description and Solution Methodology
The objective of the work done in this chapter is to study the thermomechanical
interactions of an oscillating pin-bushing pair undergoing oscillating loading conditions.
The finite element method is used to perform thermal and thermomechanical simulations
to determine the performance of the oscillating pin-bushing pair at various operating
conditions. The objective of this research is to develop an analysis for predicting failure
of pin-bushing under oscillating loads.
The problem is solved in a 3-step routine described briefly below and in detail in
sections 6.4 to 6.6.
(i) Theoretical Hertzian contact analysis
(ii) Transient finite element thermal analysis using ANSYS 5.7
(iii) Transient finite element thermomechanical using ANSYS 5.7
6.3 Theoretical Hertzian Contact Analysis
A theoretical analysis is performed as first steps to determine the contact forces
and ascertain the contact patch using Hertzian theory of elastic contact[21]. The contact
area and the contact pressure are calculated from the Hertzian theory of elastic contact.
For a line contact, the Hertzian half-contact width is given by Equation (6.1) [21] as,
LE
wR
b 598 . 1 = (6.1)
where
b = semi-contact width, m
w = Load acting on the pin, N
R = Equivalent Radius of contact, m
L = Length of contact in the normal direction, m
E = Equivalent Elasticity modulus, N/m
2

52
The contact region is measured using the equivalent elastic modulus and
equivalent radius of contact. For a Hertzian elastic contact, the equivalent radius and the
equivalent modulus of elasticity are given by Equations (6.2) and (6.3)
Equivalent radius:
2 1
2 1
R R
R R
R
±
= (6.2)
Equivalent Young's modulus:
1
2
2
2
1
2
1
1 1
2








+

=
E E
E
ν ν
(6.3)
where the subscripts 1 and 2 refers to the pin and bushing respectively and the
denominator has + for convex contact and – for concave contacts. As a pin and bushing
contact is concave, we use the difference of the radii of the pin and bushing as our
denominator.
6.4 Thermal Analysis
Frictional heat is generated at the interface between the pin and the oscillating
bushing when the two surfaces slide against each other. The heat generated is a function
of the contact forces, coefficient of friction and the sliding velocity. The heat generated is
distributed as per a partition factor between the pin and the bushing. The partition of heat
is dependent on the material properties, thermal mass and the relative velocity of the pin
and the bushing. The finite element analysis program like ANSYS or FlexPDE [25] can
calculate the partition of heat flux automatically taking into account the thermal mass and
material properties of the contacting bodies. The objective of the transient thermal
analysis is to obtain the time-dependent temperature contours of the pin-bushing under
the application of the oscillating frictional heat flux. The results of the transient thermal
analysis are used in doing the thermomechanical analysis to determine the pin-bushing
interactions.
53
6.4.1 Element Types
The pin and bushing are modeled in ANSYS 5.7 using the four-noded solid
thermal elements viz. PLANE55. This element is compatible with the 4-noded structural
solid element used in the thermomechanical analysis. This means that the results of the
thermal analysis can be successfully exported to perform the thermomechanical analysis.
6.4.2 Calculation of Heat Flux
The heat flux applied depends on the contact forces, coefficient of friction and the
linear velocity. The total heat generated is calculated by the following simple equation
fPv Q = (6.4)
where
f = Coefficient of friction
P = Contact force, N
v = Relative velocity, m/s
The heat flux applied is the heat generated applied on the area of contact and is calculated
as
c
p
L R
Q
q
θ * *
= (6.5)
where
R = Radius of the pin, m
L = Length of the bushing in the normal direction, m
θ
c
= Contact Angle, in radians
6.4.3 Implementation of Oscillating Heat Flux
The finite element modeling of the oscillating heat source requires a special
formulation. The modeling is done by applying the heat flux as a function of time and
space.
54
The time taken for 1 complete oscillation (t
osc
) is calculated from the speed of
oscillation, the angle of oscillation and the radius of the shaft. The calculated time for a
single oscillation is broken up into a series of 13 time-steps. The calculated flux is
applied over a set of 6 elements in the mean position, Figure 6.2 (a) and then it is solved
for the 1 time step i.e. t
osc
/13. For the next time sub-step, the flux is moved to the next set
of six elements and then solved again for 2x(t
osc
/13). This solution is appended to the
previous solution. The process is continued until a full oscillation is completed. This
sequence is programmed into a ANSYS macro (sub-routine) and repeated cyclically to
complete the transient analysis.
The application of the flux for 1 cycle is illustrated in Figure 6.3







(a) Mean Position (b) Extreme right - +α (c) Extreme left - -α
Figure 6.2 Schematic of application of oscillating flux, α is the oscillation angle
Sample Calculation of Time for 1 Oscillation
An example of calculation will now be presented for a bushing operating at a
surface velocity of 0.08 m/s with a stationary pin of radius 0.05 m.
⇒ Angular velocity, sec / 6 . 1
05 . 0
08 . 0
radians
R
v
= = = ω = 91.67 degrees/sec
α α
Heat Source
55
The element size used in the finite element model has an angle of 4.5
ο
. The
theoretical Hertzian contact angle, using Equation 6., is found to be 26.12
ο
. This means
that the flux should be applied over 6 elements.
At the angular velocity of 1.6 rad/s, the time taken for application of flux over a
patch of 6 elements before moving to the next set of elements is 0.05 seconds and the flux
is applied and solved for every sub-step as illustrated in Figure 6.3. The time taken to
negotiate one oscillation (t
osc
) is found to be 0.65 seconds.









Figure 6.3 –Application of oscillating heat flux as a function of time and space
6.4.4 Convection Film Transfer Coefficient between Bushing Outer Radius and
Atmosphere
The forced convection coefficient can be theoretically calculated by the empirical
relation, Equation 6.6 suggested by A.F. Mills [26] for a cylinder rotating in a fluid. This
formula is applicable for a cylinder rotating in an infinite quiescent fluid medium.
Nu
D
= 0.133 Re
D
2/3
Pr
1/3
(6.6)
where the dimensionless numbers are,
time = 0.05sec
time = 0.10 sec
time = 0.15 sec
time = 0.20 sec
time = 0.25 sec
time = 0.30 sec
time = 0.35 sec
time = 0.40 sec
time = 0.45 sec
time = 0.50 sec
time = 0.55 sec
time = 0.60 sec
time = 0.65 sec
Total Time for 1 Oscillation = 0.65 seconds
56
Nu
D
= Nusselt Number, hD/k
Re
D
= Reynolds number, ωD
2

k
Pr = Pradtl Number (0.69 for air)
where,
h = Convective heat transfer coefficient, W/m
2
K
D = Diameter of the surface on which convection takes place, m
k = Thermal Conductivity of the convecting fluid, W/mK
ω = Angular velocity of the rotating member, s
-1

ν
k
= Kinematic viscosity of the convecting fluid, m
2
/s
The convection between the bushing outer boundary and the environment was
found by assuming a mild flow of air over the bushing surface at a speed of 5 mph at a
temperature of 300 K. The following parameters are used to calculate the convection HT
coefficient.
Convecting Fluid = Air – Properties are obtained from [18]
Temperature of Fluid = 300 K (Room Temperature)
Radius of Bushing = 0.1 m
Velocity of Air = 5 mph (assumed)
Verification
The validity of the above method was verified by applying the formula to find the
convection coefficient of an axle and cone problem solved by Wang et.al [11]. The
convection coefficient calculated using Equation (6.6) was 46 W/m
2
K (Appendix). This
value was close to the value used by the authors (45 W/m
2
K) in their analysis.

57
6.4.5. Convective Coefficient in the Gap between Pin and Bushing
The convective heat transfer coefficient in the gap between the pin and bushing
needs to be examined in more detail as the temperature rise in the interface between the
pin and bushing is high. The high temperature difference between the pin and bushing
would contribute to radiation heat transfer between the hot pin and its bushing, which is
at a lower temperature. The radiation heat transfer in the analysis by using the concept of
linearized radiation coefficient, h
rad
. The total heat transfer coefficient in the gap between
pin and bushing thus consists of 2 components.
(i) Forced convective heat transfer coefficient, h
forced
and
(ii) Equivalent convective heat transfer coefficient h
rad

The total convective heat transfer coefficient is a summation of the forced
convection term and the equivalent radiation term.
h
total
= h
forced
+ h
rad
(6.7)
Calculation of Forced Convection Coefficient inside the Pin-Bushing Assembly
The forced convection of air inside the pin-bushing assembly is evaluated using
the empirical relation (6.6) using the operating conditions given below. The fluid
properties of air are varied to include the entire temperature range that can be expected
for various operating conditions.
Convecting Fluid = Air - Properties from Incropera and Dewitt [18]
Temperature of Fluid = 300 K (Room Temperature) - 1100 K (Steady State)
A sensitivity study was done by calculating the forced convection heat transfer
coefficient (h
forced
) at various temperature values and it was found that the increase in
temperature did not have a significant effect in the temperature range of room
58
temperature to steady state temperature (about 1100 K). The average forced convective
heat transfer coefficient of the sensitivity study was found to be 10 W/m
2
K for the
temperature range of 300 K – 1100 K.
Calculation of Equivalent Convective Coefficient due to Radiation
Steady state thermal analyses for various operating conditions reveal that interface
temperatures are in the range of 600

K – 1100

K. These high temperature values can
influence the thermal analysis due to heat transfer by radiation between the hot body and
the colder body. The heat transfer due to radiation is given by Equation (6.8) [18]
) (
4 4
surf rad
T T q − = εσ (6.8)

where, q
rad
= Heat transfer due to radiation, W/m
2

ε = Emissivity of the radiating body
σ = Stefan-Boltzmann's constant
T = Surface Temperature of the hotter body, K
T
surf
= Surface Temperature of the cooler body,
ο
K
The above equation can be rewritten as
) )( )( (
2 2
surf surf surf rad
T T T T T T q − + + = εσ (6.9)
The above equation can be put in the form of a simple convection heat transfer
type equation by summing the terms together to be written as an equivalent convective
coefficient due to radiation.
) (
surf rad rad
T T h q − = (6.10)
where, h
rad
is the equivalent convective coefficient due to radiation expressed as,
) )( (
2 2
surf surf rad
T T T T h + + = εσ
59
The equivalent convective coefficient due to radiation varies with the
temperature. The heat transfer due to radiation is insignificant when the surface
temperatures are very low. Also the heat transfer due to radiation is predominant in the
clearance between the pin and bushing and it is not significant on the outer surface of the
bushing. So, the equivalent convective coefficient due to radiation, h
rad
is used along with
the regular convective coefficient in the pin-bushing clearance. The variation of h
rad
with
time is linear. The value h
rad
is calculated by taking the average temperatures on the
surface of the pin and the bushing.
6.4.6 Boundary conditions for the Thermal Analysis
Heat Flux Applied on the pin, q
p
= Calculated for different load conditions using
Equation 6.5.
Temperatures on the pin and the bushing are coupled in the contact region to
achieve continuity of temperature in the contact zone.
Convective HT coefficient
On the Bushing outer edges = 30 W/m
2
K
In the Pin-Bushing clearance = h
forced
+ h
rad

Average h
forced
= 10 W/m
2
K
Equivalent h
rad
= Linear function of temperature
60
0
5
10
15
20
25
30
35
40
300 800 1300 1800 2300 2800 3300 3800
Time, sec
C
o
n
v
e
c
t
i
v
e

c
o
e
f
f
i
c
i
e
n
t
,

W
/
s
q
m
K
hrad hconv htotal

Figure 6.4 – Variation of effective convective coefficient with time
6.5 Transient Thermomechanical Analysis
6.5.1. Element Types
The solid element PLANE42 is used to model the pin and the bushing. This
element is a 2D bilinear element with the x and y displacements as the degrees of
freedom and the temperatures obtained from the transient thermal analysis can be applied
as nodal loads.
The radial clearance between the pin and the bushing is a problem that has to be
dealt with care. The gap is modeled using 2-noded contact elements CONTAC52.
Contact elements are used to model gap and they come into effect only when the two
nodes that make the element come into contact. The element properties include a normal
stiffness value that governs the resistance to normal load. The finite element programmer
assigns the stiffness value for the contact element. To determine the stiffness value, the
theoretical Hertzian contact width is found. The stiffness value is then fixed at a value by
61
trial and error so that the Hertzian contact width and the contact width found by Finite
Element method are the same. The verification of the stiffness value against the Hertzian
theory is a good way to establish validity.
6.5.2 Meshing
The meshing pattern is scaled such that the element size is reduced in the radial
direction towards the bushing-pin interface. This is done by having coarser elements at
the center of the pin and finer elements at the interface. This means that the elements
towards the center of the pin degenerate from the standard 4 noded rectangular element
into triangles. This methodology is adopted, as our area of interest is the interface of the
pin and bushing











Figure 6.5 – Mesh pattern for the oscillating pin-bushing assembly

6.5.3 Boundary Conditions
The radial load is applied on the pin on the centerline and the bushing is
constrained on the outer diameter. All the degrees of freedom are constrained. The steady
Pin Mesh
Bush Mesh
62
state analysis is performed and the contact element results are analyzed. The contact
element CONTAC52 can display the contact forces and the value of clearance in the gap.
The contact element is activated only when there is a physical contact between the pin
and the bushing. Whenever there is contact established between the pin and bushing, the
contact element results give us the magnitude of the contact force and negative clearance.
The stiffness value of the contact element is adjusted such that the contact angle is in
agreement with the Hertzian contact angle.

Oscillating Bush
Stationary Pin
Restraining Link

Figure 6.6 – Schematic of a Retention track assembly

The transient thermal analysis is followed by a transient thermomechanical
analysis to determine the effects of temperature rise on the operating parameters such as
operating clearance and the frictional torque. The results of the thermal analysis are
applied as thermal nodal loads with respect to time. The radial load acting on the pin is
applied as a point force acting on the centerline.
6.6. Operating Parameters
Table 1 – Operating parameters for the finite element model and simulations
Range of pin radii, R
s
(m) 35 x 10
-3
- 75 x 10
-3

Range of bearing lengths, L (m) 50 x 10
-3
- 167 x 10
-3

Range of radial clearances, C (m) 0.1 x 10
-3
- 0.35 x 10
-3

63
Density of pin-bushing ρ (kg/m
3
) 1.23 x 10
6
Conductivity, k (W/mK) 54
Thermal expansion coefficient, α (m/mK) 1 x 10
-5

Young's modulus of pin-bushing, E (GPa) 200
Poisson's ratio, ν
s
= ν
b
= ν 0.3
Range of loads, W (kN) 66.723- 222.411
Range of friction coefficient, f 0.1 – 0.3
Range of oscillating frequencies, ω
o

(rad/s)
1 - 2
Atmospheric Temperature, T

(°C) 25

6.7 Results and Discussion
Simulations are performed for a wide range of operating parameters, while one of
the test cases is presented in this section for discussion. Simulations of failure of an
oscillating pin-bushing corresponding to the following operating conditions are
presented.
Pin radius, R
s
(m) 0.05
Bearing length, L (m) 0.1
Radial clearance, C (m) 0.25 x 10
-3

Coefficient of friction, f 0.3

Oscillating frequency, ω
o
(rad/s) 2
Load, W (N) 11205

The heat flux as discussed earlier is applied as an oscillating flux with respect to
time on the contact region. The oscillating heat flux produces a temperature field as
shown in Figure 6.7.
Figure 6.7 gives the temperature profile for the first complete oscillation. The
temperature profile for the first four time steps shows the effect of flux moving from the
mean position to the extreme right position. The maximum temperature also can be seen
to move along in the direction of the moving flux. Time steps 4 through 10 give the
temperature profile when the heat flux moves from the extreme right position (+α in
64
Figure 6.2) to the extreme left position (-α in Figure 6.2). When the flux moves from the
extreme right position (Time step #4), it leaves a temperature tail that trails off at the
extreme right position (+α).
Figure 6.7 – Temperature contour during 1 cycle of oscillation
The temperature rise due to the frictional heating causes thermal expansion of the
pin and concomitant reduction in operating clearance. The loss in clearance causes the
pin to ovalize and come into contact with the bushing at the top of the shaft. This leads to
increased area of contact and increased contact forces. As a result, the frictional heat
further contributes to the process of expansion. The temperature profiles are given in
Figures 6.8 through 6.11. From the contour profiles, it can be seen that the temperature
increases to 752
ο
C after about 6 seconds once ovalization is established. This temperature
is sufficient to cause scuffing in steel [2]. Also the frictional torque increased to very high
values 6 seconds after ovalization was established. Thus the failure can be attributed to
both TIS and thermal galling.

65










Figure 6.8 – Temperature contour just before ovalization










Figure 6.9 – Temperature contour 1 second after initialization of ovalization

66










Figure 6.10 – Temperature contour 3 seconds after initialization of ovalization










Figure 6.11 – Temperature contour 6 seconds after initialization of ovalization
Note: The maximum temperature at the contact reached 752
ο
C in Figure 6.11
when galling takes place.
67
The failure pattern is not typically like the example shown in Figures 6.8 to 6.11.
The failure mode has a combined mode of thermally induced seizure (TIS) and thermal
galling only in very heavily loaded pin-bushing assemblies and where the coefficient of
friction and/or the oscillating frequencies are high. In cases were friction coefficients
were low, the maximum temperature reached was about 300-400
ο
C and failure occurred
by thermally induced seizure (TIS). Typically lightly loaded assemblies with low
coefficients of friction failed by TIS mode while the heavily loaded assemblies and when
the coefficients of friction were higher, a combination of thermal galling and TIS was
observed within the range of simulations performed.
6.8 Generalization
A number of simulations (Table 6.2) were performed to study and derive a
generalized equation to determine the time of failure of an oscillating pin-bushing
assembly. Non-dimensional parameters were used to generalize the operating parameters.
The non-dimensional parameters used in this analysis are similar to the non-dimensional
parameters used in Chapter 4 in the study of TIS in journal bearings during start-up and
are given below.
Non-dimensional seizure time:
2
s
fo
fo
R
t
t
κ
=
Thermal strain:
s
o
k
fW α ω
ε =
Modified aspect ratio:
L
R
eq
= λ , where
s bi
bi s
eq
R R
R R
R

= =
C
C R R
s s
) ( +


68
Table 6.2 – Variation of failure time at various operating conditions
Coefficient
of friction, f
Angular
velocity,
ω
o
, (rad/s)
Load, W
(kN)
Pin
Radius, R
p
,
(mm)
Clearance,
C (mm)
Length,L
(mm)
Failure
time, t
fo
(s)
Thermal
Strain,ε
Aspect
Ratio,λ
Non-
dimensional
time for
failure, t
fo

Temperature
at Failure,
(
ο
C)
0.1 1 6053 2.139 x 10
-03
37.19 300
0.1 1.6 2535 3.422 x 10
-03
15.58 322
0.15 1.2 2399 3.849 x 10
-03
14.74 343
0.2 1.4 1488 5.988 x 10
-03
9.143 374
0.2 1.5 1207 6.416 x 10
-03
7.416 380
0.25 1.6 936 8.554 x 10
-03
5.75 417
0.27 1.8 749 10.97 x 10
-03
4.602 442
0.3 2
111.21
582 12.83 x 10
-03
3.58 492
66.72 3018 3.849 x 10
-03
18.544 329
0.2 1.5
222.42
50 0.25 100
564 12.83 x 10
-03
100.5
3.47 461

0.1 322 250.5 1.98 205
0.15 604 167.17 3.71 265
0.2 1006 125.5 6.18 314
0.25 1207 100.5 7.42 380
0.3 1408 83.83 8.65 440
50
0.35 2532 71.93 15.56 465
35 2380 49.35 29.84 405
75
100
805 225.75 2.20 446
50 564 201 3.47 461
100 1207 100.5 7.42 380
0.2 1.5 111205
50
0.25
166.67 2938
6.416 x 10
-03
60.3 18.05 325
69
t = 0.006 ε
-1.4064
fo
f
o
0.002 0.004 0.006 0.008 0.010 0.012 0.014
ε
0.0
5.0
10.0
15.0
20.0
25.0
30.0
35.0
40.0
t
The number of simulations is reduced by just varying the non-dimensional
parameters instead of varying all the operating parameters making sure that the physical
significance of the operating parameters is not lost. The effect of various operating
parameters like load (W), coefficient of friction (f) and frequency of oscillation (ω
o
) on
the failure time is found by varying the non-dimensional parameter ε. The number of
simulations required is thus greatly reduced. The results of varying the various operating
parameters are given in Table 6.1.










Figure 6.12 – Thermal Strain vs non-dimensional failure time
The results presented in Table 6.1 were curve fitted to a goodness value within
98.9% as shown in Figures 6.12 and 6.13. The expressions relating the thermal strain,
aspect ratio and non-dimensional failure time were combined to give the following
equation 6.11 for failure time for pin-bushing that undergo failure under oscillating
boundary conditions.
70
71 70 255 2
t
4064 1
fo
. .
.

=

λ
ε
(6.11)












Figure 6.13 – Variation of aspect ratio with non-dimensional seizure time
6.9 Verification
The primary reason for failure in the analysis of the pin-bushing assembly is due
to the increase in temperature due to frictional heating. The application of proper
boundary conditions for the thermal analysis has to be verified to ensure accuracy of the
model. A comparison was done between the heat conduction finite element problem done
using ANSYS for a one domain problem with the oscillating heat source applied on the
pin. Krishnamurthy and Khonsari [27] verified the accuracy of the finite element model
by applying Duhamel's theorem to problems involving oscillating heat source. They also
performed dimensional and non-dimensional analysis using another FE solver, FlexPDE
[24]. The results of the finite element problems solved using ANSYS and FlexPDE
f
o
t
fo
= (2.255λ − 70.71)
ε
0 50 100 150 200 250 300
λ
0.000
0.005
0.010
0.025
0.015
t
0.020
−1.4064
ε

1
.
4
0
6
4
71
matched closely with the analytical solution. The convergence of the finite element
model was also checked using three different mesh sizes.
6.10 Restrictions and Applications
(i) The finite element was solved by approximating it as a 2-dimensional
problem, assuming that the contact pressure is uniform across the bearing
length and there is no misalignment of the shaft in the bearing. The material
properties were assumed to be constant and the analysis did not consider
plastic deformation.
(ii) The time for failure as given in Equation (6.11) gives the approximate failure
time (t
fo
) after there is complete loss of initial lubrication in the pin-bush
assembly and loss of anti-friction coating.
(iii) Equation (6.11) gives the failure time for pin-bush assembly when there is
continuous oscillating motion. But in practice, there is only intermittent loads
acting on the pin-bushing assembly. The time to failure predicted is for the
worst-case scenarios. To obtain a practical failure time, the loading cycle of
the oscillating pin-bushing has to be incorporated with Equation (6.11).
(iv) The Hertzian contact pressure in the contact region is elliptic and hence the
frictional heat flux is also elliptic. But a constant heat flux is applied in the
thermal analysis.
(v) The oscillating frequency, ω
o
, is assumed to be constant for small intervals of
time.
72
CHAPTER 7. CONCLUSIONS

7.1 TIS in Journal Bearings During Start Up
When rotating machinery that is supported on fully lubricated bearings are started
up from rest, the lubrication flow may not have been established and there would be
metal-to-metal contact. The effect of the dry sliding during start-up was analyzed by
studying the effect of start-up friction on the bearing operating parameters such as
clearance loss and frictional torque by a thermoelastic finite element model.
A series of simulations were performed by varying the operating parameters. A statistical
analysis was performed using the simulated results to determine the relationship of these
operating parameters on the seizure time. An empirical relationship was developed to
estimate the seizure time using these operating parameters. The procedure for applying
the empirical relation and the interpretation of results that can be deduced from the
empirical relation is described. The restrictions involved in using the empirical relation
are also specified. The accuracy of the empirical relation was verified using published
results. The final expression for the seizure time is given by the following equations
1.6494 - -1.2478
s
0.494 t λ ε 9 = For 500 < R
s
/C < 1000

-1.02718 -1.2478
s
1.825 t λ ε = For 1000 < R
s
/C < 5100
7.2 TIS in Journal Bearings During Transient Flow Disturbance
The sudden stoppage in the lubricant flow, in a fully flooded journal bearing was
analyzed. The interruption in the lubricant flow was modeled in finite element analysis by
an increase in the coefficient of friction and a decrease in the convective cooling. The
effect of the lubricant cut-off was analyzed by studying its effect on the bearing operating
parameters such as clearance loss and frictional torque by a thermoelastic finite element
73
model. An expression relating the seizure time to the operating parameters for a journal
bearing undergoing a transient flow disturbance was derived in the same procedure as
explained for the case of TIS during start up. The empirical relationships relating seizure
time to operating parameters when a flow disturbance is encountered are given by the
following equation.

( ) ( )
1
5489 . 0
314 . 9 0065 . 0

+ = λ ε
sp
t

7.3 TIS and Thermal Galling in Oscillating Pin-Bushing Analysis
The analysis of a heavily loaded pin-bushing assembly subject to oscillating
loading was analyzed using finite element thermal analysis and thermomechanical
analysis. The reasons for bearing failure were analyzed by performing simulations for
different types of operating conditions. It was found that failure of these pin-bushings
occurred by thermally induced seizure for lighter boundary conditions and by a
combination of TIS and thermal galling for more severe boundary conditions. The time
for failure in the non-dimensional form was derived from a series of simulations and is
given by the following equation.
71 70 255 2
t
4064 1
fo
. .
.

=

λ
ε

74
BIBLIOGRAPHY
1. Ling F. F. and Saibel E. Thermal aspects of galling of dry metallic surfaces in
sliding contact. Wear, 1958, 1, 80-91

2. Dyson A. Scuffing. Treatise on Material Science and Technology, 1979, 13, 175-
215

3. Gecim B. and Winer W. O. Steady State Temperature in a Rolling Cylinder
Subject to Surface Heating and Convective Cooling. ASME Journal of Tribology,
1984, 106, 120-127

4. Patula E. H. Steady State temperature Distribution in a Rotating Roll Subject to
Surface Heat Fluxes and Convective Cooling. ASME Journal of Tribology, 1981,
103, 36-41

5. Ulysee P. and Khonsari M. M. Thermal Response of Rolling Components Under
Mixed Boundary Conditions: An Analytical Approach, ASME Journal of Heat
Transfer, 1993, 115, 857-865

6. Bishop J.L. and McC. Ettles C.M. The seizure of journal bearings by
thermoelastic mechanisms. Wear, 1982, 79, 37-52

7. Dufrane K. and Kannel J. Thermally induced seizures of journal bearings.
ASME Journal of Tribology, 1989, 111, 288-92

8. Khonsari M.M. and Kim H.J. On thermally induced seizure in journal bearings.
ASME Journal of Tribology, 1989, 111, 661-7

9. Hazlett T.L. and Khonsari M.M. Finite element model of journal bearing
undergoing rapid thermally induced seizure. Tribology International, 1992a, 25,
177-82

10. Hazlett T.L. Thermoelastic behavior of journal bearing undergoing seizure – a
finite element study. M.S. Thesis, University of Pittsburgh, Pittsburgh,
Pennsylvania, 1990

11. Wang H., Conry T.F. and Cusano C. Effects of Cone/Axle Rubbing Due to
Roller Bearing Seizure on the Thermomechanical Behavior of a Railroad Axle.
ASME Journal of Tribology, 1996, Vol.118, pp.311-319

12. Wang H., Axle Burn-off and Stack-up Force Analyses of a Railroad Roller
Bearing using the Finite Element Method. Ph.D. Thesis, Department of
Mechanical and Industrial Engineering, University of Illinois at Urbana-
Champaign, 1996, p 35, pp.140-147.
75
13. Wang Q., Seizure Failure of journal-bearing conformal contacts. Wear, 1997,
210, pp.8-16.

14. Lacey S. and H. Kawamura H., Bearings for Aircraft Gas Turbine Engines (Part
1), NSK Technical Journal – Motion and Control, 1998, 5, pp. 1-8

15. Pascovici M.D., Khonsari M.M. and Jang J.Y., On the Modeling of
Thermomechanical Seizure. ASME Journal of Tribology, 1995, 117, 744-7

16. Jang J.Y., Khonsari M.M. and Pascovici M.D., Thermohydrodynamic Seizure:
Experimental and Theoretical Analysis. ASME Journal of Tribology, 1998, 120,
8-15

17. Cook R.D., Malkus D.S. and Plesha M. E., Concepts and Applications of Finite
Element Analysis, John Wiley and Sons, 1989, Third Edition

18. Incropera F.P. and Dewitt D. P. Fundamentals of Heat and Mass Transfer, John
Wiley and Sons, 1996, Fourth Edition

19. ANSYS 5.7 Online Users Manual, 2001, ANSYS Inc

20. Peterson M.B. and Winer W.O. Wear Control Handbook, ASME, 1980, 115

21. Hamrock B.J. Elastohydrodynamic lubrication of point contacts. Ph.D Thesis,
Institute of Tribology, Department of Mechanical Engineering, The University of
Leeds, 1976, 44-66, pp. 93-102

22. Khonsari M.M., Booser E.R., Applied Tribology, John Wiley and Sons Inc.,
2001

23. Bolz R. E., Tuve T.L., CRC Handbook of Tables for Applied Engineering
Science, CRC Press, Second Edition, 1970, p 621, pp 117-19, pp 153-155

24. Khonsari M.M. and Kim H.J., Discussion on paper titled "Thermally induced
seizures of journal bearings" by Dufrane K. and Kannel J., ASME Journal of
Tribology, 1989, 111, 292

25. FlexPDE: A Flexible Solution for Partial Differential Equations Version 2.20e 3D
Online Reference Manuals, 1996-2001, PDE Solutions Inc

26. Mills A.F., Heat and Mass Transfer, Richard D Irwin Inc., 1995

27. Krishnamurthy H., "Application of Duhamel's Theorem to Problems Involving
Oscillating Heat Source" M. S. Thesis, Department of Mech. Engr., LSU, 2002
76
APPENDIX - MAPLE PROGRAM TO CALCULATE CONVECTION
COEFFICIENT

The formula to find the convection coefficient of an axle and cone problem solved by
Wang et.al [11]. The convection coefficient calculated using Equation (6.6) was 46
W/m
2
K while Wang et. al. reported the convection coefficient to be 45 W/m
2
K for the
operating conditions reported below.
Train speed = 55 mph;
Axle rotating speed = 560 rpm
Ambient air temperature = 25
o
C
Shaft radius = 78.632 mm

Maple Program:

> restart;
> with(linalg):
Warning, the protected names norm and trace have been redefined and
unprotected
> Nusselt[forced] := 0.133*Reynolds^(2/3)*Pr^(1/3);
:= Nusselt
forced
.133 Reynolds
( ) / 2 3
Pr
( ) / 1 3

> Reynolds := omega*Dia^2/nu;
:= Reynolds
ω Dia
2
ν

>
> omega := V/R;
:= ω
V
R



Free Convection coefficient from Incropera and Dewitt

> Nusselt[free] := (0.60 +
0.387*Ra^(1/6)/(1+(0.559/Pr)^(9/16))^(8/27))^2;
:= Nusselt
free


















+ .60
.387 Ra
( ) / 1 6










+ 1 .7209732340








1
Pr
( ) / 9 16
( ) / 8 27
2

77
> Nusselt[total] := Nusselt[forced] + Nusselt[free];
:= Nusselt
total
+ .133








V Dia
2
R ν
( ) / 2 3
Pr
( ) / 1 3


















+ .60
.387 Ra
( ) / 1 6










+ 1 .7209732340








1
Pr
( ) / 9 16
( ) / 8 27
2

> Ra := g*beta*(T[s]-T[inf])*Dia^3/(nu*alpha);
:= Ra
g β ( ) − T
s
T
inf
Dia
3
ν α

> beta := 1/T[inf];
:= β
1
T
inf

> Dia := 2*R;
:= Dia 2 R

Enter the Kinematic viscosity of the fluid - air
> nu := 15.89e-6;
:= ν .00001589

Enter the velocity of the shaft in m/s
> V := 4.6112;
:= V 4.6112

Enter the radius of the shaft, in meter
> R := 0.078632;
:= R .078632

Enter the thermal conductivity of the fluid, in SI units
> k := 26.3e-3;
:= k .0263

Enter the Prandtl Number for the fluid
> Pr := 0.69;
:= Pr .69

Enter the Acceleration due to gravity value in the proper units;
> g := 9.8;

:= g 9.8

Enter the Diffusivity of the fluid
> alpha := 22.5e-6;
:= α .0000225

Enter the Fluid Bulk Temperature (in degree K)
> T[inf] := 300;
:= T
inf
300
78
Enter the Mean Temperatur rise on the surface of the shaft;
> T[s] := 373;
:= T
s
373

> h := Nusselt[total]*k/Dia;
:= h 46.09811930

79
VITA
Mr. Rajesh Krithivasan was born on February 4, 1976, in Madurai, India. He earned his
bachelor’s degree in mechanical engineering from Bharathiyar University, India. After
his undergraduate studies, he worked as a design engineer in Ashok Leyland Limited,
India, for 2 years. He joined Louisiana State University in August 1999 to pursue a
master’s degree in mechanical engineering. His area of interests include tribology,
computer aided designing/manufacturing and finite element methods. He successfully
defended his thesis work on October 7, 2002, and will receive the degree of Master of
Science in Mechanical Engineering in December 2002.

ACKNOWLEDGEMENTS I would like to express my gratitude to my advisor, Dr. Khonsari, for his invaluable guidance and encouragement throughout the course of this thesis work. I am sincerely thankful to my committee members, Dr. Yitshak Ram and Dr. Wenjin Meng, for their support and spending their valuable time in evaluating my thesis work. My special thanks goes to Mr. Haribabu Krishnamurthy for all the fruitful discussions we had that were instrumental for the successful completion of this thesis. I also thank all my friends in the tribology group (CeRoM), for their moral and technical support, without which I would not have succeeded in finishing my thesis. This work was supported in part by a grant from The Louisiana Board of Regents' Industrial Ties Research Subprogram with Caterpillar Inc (Grant # ITRS-007B-2002).

ii

TABLE OF CONTENTS ACKNOWLEDGEMENTS………………………………………………………………..ii NOMENCLATURE………………………………………………………………………..v ABSTRACT………………………………………………………………………………vii CHAPTER 1. INTRODUCTION…………………………………………………………1 CHAPTER 2. LITERATURE SURVEY …………………………………………………2 CHAPTER 3. FINITE ELEMENT ANALYSIS: PROCEDURE AND OVERVIEW…...9 3.1 Thermal Analysis – Theory and Finite Element Formulation…………………9 3.2 Thermoelastic Analysis – Theory and Finite Element Formulation …………12 CHAPTER 4. THERMALLY INDUCED SEIZURE IN JOURNAL BEARINGS DURING START UP ……………………………………………………15 4.1 Finite Element Modeling Procedure………………………………………….17 4.1.1 Transient Thermal Finite Element Analysis ……………………….18 4.1.2 Non-linear Transient Elastic Finite Element Model ……………….20 4.1.3 Seizure Criterion …………..……………………………………….22 4.2 Results and Discussion……………………………………………………….23 4.3 Model Parameters…………………………………………………………….26 4.4 Generalization………………………………………………………………...27 4.4.1 Influence of Operating Parameters…………………………………28 4.5 Non- Dimensionalization …………………………………………………….30 4.5.1 Statistical Treatment of Non-dimensional Operating Parameters….30 4.6 Verification and Analysis…………………………………………………….32 CHAPTER 5. TIS OF JOURNAL BEARINGS TRIGGERED BY A TRANSIENT FLOW DISTURBANCE…………………………………………………37 5.1 Finite Element Analysis Modeling and Seizure Criteria……………………..40 5.2 Results and Discussion……………………………………………………….41 5.3 Model Parameters…………………………………………………………….43 5.4 Non-Dimensionalization and Generalization………………………………...44 5.5 Analysis of Results…………………………………………………………...45 5.6 Verification…………………………………………………………………...47 CHAPTER 6. THERMOMECHANICAL ANALYSIS OF A HEAVILY LOADED OSCILLATING PIN-BUSHING PAIR …………………………………49 6.1 Introduction and Problem Definition ………………………………………..49 6.2 Problem Description and Solution Methodology .…………………………..51 6.3 Theoretical Hertzian Contact Analysis ………………………………………51 6.4 Thermal Analysis …………………………………………………………….52 6.4.1 Element Types ……………………………………………………..53

iii

6.4.2 Calculation of Heat Flux …………………………………………...53 6.4.3 Implementation of Oscillating Heat Flux…………………………..53 6.4.4 Convection Film Transfer Coefficient between Bushing Outer Radius and Atmosphere ……………………………………………55 6.4.5. Convective Coefficient in the Gap between Pin and Bushing …….57 6.4.6 Boundary Conditions for the Thermal Analysis …………………...59 6.5 Transient Thermomechanical Analysis………………………………………60 6.5.1 Element Types ……………………………………………………..60 6.5.2 Meshing…………………………………………………………….61 6.5.3. Boundary Conditions ……………………………………………..61 6.6 Operating Parameters…………………………………………………………………….62 6.7 Results and Discussion……………………………………………………….63 6.8 Generalization………………………………………………………………...67 6.9 Verification…………………………………………………………………...70 6.10 Restrictions and Applications……………………………………………….71 CHAPTER 7. CONCLUSIONS………………………………………………………….72 7.1 TIS in Journal Bearings During Start Up …………………………………….72 7.2 TIS in Journal Bearings During Lubricant Flow Disturbance ……………….72 7.3 TIS and Thermal Galling in Oscillating Pin-Bushing Analysis……………...73 BIBLIOGRAPHY ………………………………………………………………………..74 APPENDIX - MAPLE PROGRAM TO CALCULATE CONVECTION COEFFICIENT …………………………………………………………..76 VITA ……………………………………………………………………………………..79

iv

seconds v . W/m2 Heat generated. m Equivalent radius of contact. J/kgK Young's modulus. m2 Radial clearance. m Speed of the shaft. °C Seizure time during start-up. m Radius of the journal. seconds Failure time for oscillating pin-bushing. W/mK Length of the bearing in the normal direction. rpm Heat partition factor Contact force between the journal and the bearing. m Ambient temperature. N Heat flux entering the shaft. m Specific heat capacity. W Inside radius of the bearing. seconds Seizure time when lubricant flow disturbance occurs. N/m2 Coefficient of friction Convective heat transfer coefficient. m Outer radius of the bearing.NOMENCLATURE As C Cp E f h k L N n P qs Q Rbi Rbo Rs Req T∝ tss tsp tfo = = = = = = = = = = = = = = = = = = = = = Area of the shaft that is in contact with the bushing. W/m2K Thermal conductivity.

seconds u W = = = = = = = = = = = = Surface velocity of the journal. m2/s Non-dimensional modified aspect ratio Poisson's ratio Density of the material. radians Frictional torque. kg/m3 Contact angle. Nm Angular velocity. N Coefficient of thermal expansion.t ss t sp t fo tref = = = = Non-dimensional seizure time during start-up Non-dimensional seizure time during lubricant flow disturbance Non-dimensional failure time for oscillating pin-bushing Transition time for the bearing to go from fully lubricated conditions to boundary lubrication conditions. radians/seconds α ε κ λ ν ρ θc τ ω ωo Subscripts s b = = Shaft or journal or pin Bearing vi . radians/seconds Oscillating frequency. m/s Load acting on the shaft. m/mK Non-dimensional thermal strain Thermal diffusivity.

ABSTRACT This thesis presents a finite element parametric study of the transient thermomechanical interactions of shaft-bushing pairs in bearings to gain insight into the nature of two categories of failures encountered in practice. speeds. operating clearance. Good agreement between the empirical and published results attests to the capability of the model and its potential for predicting bearing failure. friction coefficients and thermal expansion coefficients are performed to gain insight into the phenomenon of TIS and thermal galling. vii . An extensive set of parametric simulations covering a wide range of loads. bearing dimensions. The first type of failure deals with occurrence of the so-called thermally induced seizure (TIS) during the start-up period followed by an investigation of TIS due to a transient flow-disturbance. The second part deals with the thermomechanical interactions of pin-bushing assembly under heavy oscillating loads where the failure is by TIS and/or thermal galling. A statistical procedure is applied to the simulated results and an appropriate empirical relationship is derived that predicts the time to failure for each category.

INTRODUCTION The study of thermal effect on bearing performance has been considered to be an important subject since the evolution of tribology as a field of study. The thesis concentrates on the transient thermomechanical interactions of journal and bearings under varying operating conditions encountered in practice. 1 . Galling failure. on the other hand refers to the condition whereby there is gross damage on the sliding surfaces characterized by the formation of local welding of surface. Thermally induced seizure and galling are examples of such failures relevant to this work.CHAPTER 1. The driving force behind this is the frequent failure of tribological components due to metal-to-metal contact and the associated rise in the frictional heating. The research focused on the failure of journal bearings during start-up. Thermally induced seizure occurs when the rise in operating temperature causes a partial or complete loss in operating clearance leading to the seizure of the journal in the bearing. and the failure of a heavily loaded pin-bushing pair subject to oscillating heat fluxes. failure of journal bearings during a transient flow disturbance.

load and time. white. LITERATURE REVIEW Ling and Saibel [1] performed a study of failure of bearings due to thermal galling of sliding surfaces in contact. This criterion would be more appropriate for high energy sliding. etch-resistant phase on the sliding surfaces. The presence of this transformed layer. The governing partial differential equation and boundary conditions were solved using the finite Fourier transform. scuffing and scoring. The identity of the phase is commonly thought to be a mixture of austenite. The analogy between the analysis done by Gecim and Winer and a journal bearing undergoing 2 . like disc brakes and clutches. The surface temperatures were calculated from Blok's criterion for predicting the flash temperature due to the asperity contact. "the metallurgical reaction on a scuffed or a galled surface is characterized by the formation of a hard. This work provided the author with the idea to use a similar failure criterion for bearings. This type of failure mode is sometimes referred to by technical terms like galling. often separated from the original untransformed bulk of the material by a tempered layer suggests that the scuffed surfaces have been subject to high temperatures and rapid cooling". According to Dyson [2]. Dyson [2] presented a review of a type of failure and where gross damage characterized by the formation of local welding spots occurs between the sliding surfaces. Gecim and Winer [3] performed a steady-state thermal analysis of a rotating cylinder subjected to heating in a particular region on its surface while the rest of the surface was subjected to convective cooling. martensite and carbide. Galling criteria was expressed as a function of the surface velocity. Galling was predicted to occur when the sliding surface reached the recrystallization temperature of the metal.CHAPTER 2.

Bearing seizure was considered to have occurred when the temperature rise on the surface of the shaft exceeded 360οC. The cause of this type of failure was identified to be dry metal-to-metal contact during the bearing start-up and associated rise in the contact temperature. Bishop and Ettles considered the temperature rise to be dependent primarily on the load. Bishop and Ettles [6] analyzed the thermoelastic interaction of a journal in a plastic bushing that was interference-fit with the shaft.frictional heating is striking. contact pressure and the sliding velocity. A similar analysis was performed by Patula [4] to determine the steady state temperature of a rotating roller used in rolling of hot ingots in metal forming industry. speed and clearance and arrived at a critical PV/C number as a bearing seizure number. The work is of particular importance as the encroachment of the shaft to the bushing and the concomitant reduction 3 . In that study. Ulysee and Khonsari [5] generalized the conditions of a cylinder undergoing heating and cooling by including non-uniform heating and cooling. as the temperature rise is not appreciable as the thermal conductivity of plastic was low. The frictional heat generated was calculated as the product of the coefficient of friction. Bearings that had been out of service for a relatively long time are particularly susceptible because of the lack of adequate supply of the lubricant in the contact area. They obtained an analytical expression for the steady state solution using the Fourier transform method. Dufrane and Kannel [7] performed a study of the thermoelastic interactions of a journal bearing undergoing a catastrophic seizure leading to a complete loss in operating clearance. the shaft was considered to expand radially outwards due to the rise in temperature. The expansion of the bushing was ignored.

Although it gave good results that matched experiments. This linear equation holds good only for very short transient times. They developed two models. The 2-D axial model included the effect of heat transfer in the axial direction. Dufrane and Kannel performed a series of experiments to determine the effect of dry friction that led to failure by seizure. A onedimensional thermal and thermoelastic analysis was performed to estimate the seizure time as a function of the operating parameters. Khonsari and Kim [8]. performed a 2-D finite element analysis to study the thermoelastic behavior of journal bearings undergoing seizure during start-up. The results proved that the seizure time tended to "flatten out" with increasing time. A heat-partitioning factor was calculated based on the ratio of the areas of contact of the journal and the bearing. The bearing thermal expansion was not considered in the theoretical analysis. This non-linear expression closely captures the physics of the reduction in clearance with time. The shaft expanded radially outward and seizure was assumed to be complete when the total operating clearance vanished. it is somewhat restrictive and does not accurately represent the behavior of the system. The experimental results showed that typically seizure occurred within 30 seconds in most journal bearings operating in dry conditions. If a misalignment were present in the axial direction. The linear variation of seizure time with clearance implies that all bearings irrespective of the operating clearance would eventually seize. one with an aligned journal and bearing and the other with an axial misalignment. A linear analytical expression was derived relating the bearing operating conditions and the seizure time. inspired by the work done by Dufrane and Kannel.in the operating clearance (on the order of microns) was found to occur rapidly. the shaft comes into contact with 4 .

Based on the results of the thermomechanical analysis.the bearing on a very small area. further aggravating the thermomechanical interaction. They developed a thermomechanical finite element model using the finite element package ANSYS. the bearing is assumed to be seized. The thermal analysis was first done to study the effects of the frictional heating on the contact area of the bushing and the entire area of the shaft. The results established that TIS is triggered by the ovalization of the bushing and formation of new contact patches at the top of the bearing. 10] continued the research to gain further understanding of the behavior of a journal bearing undergoing seizure. Hazlett and Khonsari found that the encroachment of the shaft into the bearing was non-linear with time. The results of the thermal analysis were used as thermal loads for solving the thermoelastic model. Hazlett and Khonsari [9. When the frictional exceeded a limiting value. These elements come into effect only when there is a contact between the journal and the bearing. Frictional torque was introduced as the basis for seizure. Khonsari and Kim approximated this as a point heat source and presented results reveled that the misalignment of the journal in the axial direction led to faster seizure times. The thermal contact resistance at the contact area between the shaft and the bushing was modeled using convection link elements. The status of these elements was used to find the variation of clearance and contact forces over time. The establishment of new contact patches accelerated the seizure process by increasing the contact forces and thus 5 . The operating clearance was modeled by contact elements available in the ANSYS element library. The thermal analysis incorporated the partition of frictional heat generated at the contact area based on the ratio of the areas of contact of the journal and the bearing.

Conry and Cusano [11. Radiation was also considered in the thermal analysis. Temperature dependent material properties and friction coefficients were used because of the high temperatures involved. More recently.0. The seizure patterns were similar to those reported by Hazlett and Khonsari. An axially symmetric finite element and a radially symmetric finite element analyses were performed simultaneously to simulate the 3D analysis.12] performed a finite element study of the burn-off and force stack-up in a railroad axle. It revealed that seizure in unlubricated conformal contacts was 6 . This model of heat partitioning was found to match closely with the results of the partitioning method used by Hazlett and Khonsari.increasing the frictional torque. When the clearances are low (0. Failure of the railroad followed two patterns depending on the clearances involved. When the clearances are sufficiently large (2 mm – 5 mm). This may be visualized as a positive feedback of thermal energy into the system accelerating the process of seizure. the failure was due to seizure of the axle in the bearing cone.05 mm . This is of importance as the temperature rise in the journal-bearing system was very large. New contact areas were established that triggered seizure. Wang. The bearings operated for a long time subjected to frictional heating continuously and the temperatures reached recrystallization temperatures when the failure happened due to axle burn-off at high temperatures. The partition of heat between the journal and the bearing was done by iteratively comparing the surface temperatures of the radial and axisymmetric models. Wang [13] performed a review of published results on TIS in conformal contacts. The thermal analysis was similar to the analysis procedure performed by Hazlett and Khonsari.5 mm). the failure was due to axle burn-off.

(2) Oil flow interruption ⇒ Surface damage (Wear) ⇒ High heat generation ⇒ Reduction in bearing clearance ⇒ Seizure. starvation of the oil was found to be a direct cause for seizure. surface roughness and microstructure of the shaft and the bushing. 7 . Lacey and Kawamura [14] performed a study on the effects of lubricant flow disturbance in aircraft gas turbine engine bearings. The oil flow interruption typically lasts for 15 to 30 seconds. landing and sudden maneuvering.primarily due to a thermal ratcheting effect in a positive feedback of increases of interfacial pressure and heat. The effects of the oil flow interruption in lubricant supply took place in either of the two mechanisms described below: (1) Oil flow interruption ⇒ Adverse ∆T ⇒ Reduction in bearing clearance ⇒ Excessive Hertzian stresses ⇒ High heat generation ⇒ Bearing seizure. In lubricated journal bearings. the inspiration to perform a similar study on journal bearings was developed. Thermally induced Seizure (TIS) can also occur in circumstances where there is a disturbance in the steady-state operation of the journal bearing. The starvation led to wear and temperature rise in the contact area that caused the onset of seizure. These bearings are required to survive with little or no oil under severe operating conditions such as those encountered during flight take-off. Other factors that affect the seizure process are the surface quality and chemical composition. Although this study was done to improve the performance of ball bearings in aircraft engines. The aircraft engine bearings are not only required to survive these operating conditions but also to resume normal operation once the lubricant flow is re-established.

Jang. the thermal response of the journal bearing is obtained. A theoretical THD analysis was done to determine the thermal response of the lubricating oil. It can also occur in bearings running in fully lubricated condition. Performing an energy balance on a lumped system. The viscous dissipation caused the oil temperature to rise. Khonsari and Pascovici [16] performed a comprehensive analytical and experimental study on the thermohydrodynamic (THD) seizure in journal bearings. This phenomenon was discussed by Pascovici. 8 . The journal was made of polyamide and the bearing was made of glass-reinforced epoxy resin. contaminant blockage. Khonsari and Jang [15] and an analytical model was formulated for this transient analysis. airlock etc. A limiting temperature is taken as the condition for onset of seizure. The thermal expansion of the shaft and its encroachment into the bushing was determined using a standard 2-D heat conduction equation. Thermally induced seizure is not a phenomenon restricted to bearings operating under dry or boundary lubricated conditions. The seizure time was determined when the driving torque requirement exceeded a certain range. This limiting temperature is when the clearance between the journal and the bearing is completely lost. A "no-seizure" condition is also derived based on the limiting temperature. The boundary conditions at the interface of the shaft and the fluid film were determined from the THD analysis. Good agreement of theoretical and experimental seizure time was observed.This study provided the author the inspiration to study the effects of flow disturbance in journal bearings that may occur due to supply line blockage. The experimental set-up studied THD seizure in an unloaded bearing that is operating in fully lubricated conditions.

For an isotropic material with no internal heat generation. Fortunately. the finite element analysis procedure.CHAPTER 3.1 and 3.1) 9 . compute the temperatures at the nodes. The finite element analysis is a simple and handy tool that is used with good accuracy in engineering. In this chapter. Equation 3.2 give the governing equation for heat conduction under transient condition in rectangular and polar coordinates respectively [18]. The commercial FEM software package ANSYS 5. it is possible to devise a single mesh layout for both problems: a computer program can read a single data file.1 Thermal Analysis – Theory and Finite Element Formulation Heat conduction analysis must be performed to determine the material temperatures and the heat flow rates.7 was utilized to perform a detailed analysis of the thermoelastic interactions of journal and bearing. stresses etc. The temperature distribution is also needed in order to perform an analysis for thermally induced stresses. and 19]. then use these temperatures in a thermomechanical analysis to compute the displacement. the type of elements used and the boundary conditions involved are presented with the theoretical background [17. ∂ 2θ ∂ 2θ ρ C p ∂θ k 2 +k 2 = k ∂t ∂x ∂y (3. 18. 3. FINITE ELEMENT ANALYSIS: PROCEDURE AND OVERVIEW The objective of this thesis is to study the thermomechanical interactions of journal-nearing systems subject to different types of boundary conditions like unlubricated bearing start-up. a fully-lubricated bearing undergoing a disturbance in the lubricant oil supply and a journal and bearing subject to oscillating heating.

3 gives the equivalent finite element formulation for the transient thermal analysis.∂ 2θ 1 ∂θ ∂ 2θ ρ C p ∂θ + + = k ∂t ∂r 2 r ∂r ∂ϕ 2 where θ = Rise in temperature. The elements are arranged in a global matrix formulation to arrive at the complete solution. where the surface boundary conditions like heat flux or convection are specified. ([k ] + [H ]){T }+ [C ] ∂T  = {Rq }+ {Rh }    ∂t  (3.3) where [k] = element thermal stiffness matrix = T ∫ [N ] [h][N ]dS Ve T ∫ [B] [Κ ][B]dV [H] = Se 10 .2 can be solved analytically if we know the geometry of the solid and its material properties and four boundary conditions (2 in the x/r-direction and 2 in the y/φ-direction) and one initial condition. There are normally two types of boundary conditions to solve a thermal analysis: (i) the Dirichlet boundary condition. J/kgK t = time. Note that Equation 3. The finite element formulation does not depend on the coordinate system.2) k = thermal conductivity of the material. seconds Equation 3. Equation 3. kg/m3 Cp= Specific heat capacity of the material.3 is the formulation for a single element [17]. where the temperature boundary conditions are specified and (ii) the Neumann boundary conditions. W/mK ρ = Density of the material. οC (3.1 and 3.

Four-noded elements as shown in Figure 3. Boundary condition matrix on surface. . The shape function of a finite element is defined as the interpolation function that describes the distribution of the degree of freedom (temperature. -b) 2 (a. The temperature field across the element is modeled as a linear function of the nodal temperatures.4 and 3.1 were used in the analyses in this thesis for modeling thermal solids. [B] is the strain displacement matrix. 4 (-a. Se [Κ] = Thermal conductivity matrix of the material [N] is the shape function of the element used in the finite element analysis. Se {Rq} = Se ∫ [N ] T {Rh} = Se ∫ [N ] T hT f dS . Boundary condition matrix on surface.1 – 4-noded bilinear element  (a − x)(b − y ) (a + x)(b − y ) (a + x)(b + y ) (a − x)(b + y )  [N] =  . which is a spatial derivative of the shape function matrix. displacements etc. [N]. b) x 4-noded Bilinear Element 1 (-a.[C] = Ve ∫ [N ] T ρ C P [N ] dV qb dS .4) 11 .  4ab 4ab 4ab 4ab   (3.) across the element. the shape function matrix and the strain displacement matrix are given by the following equations 3. b) y 3 (a. -b) Figure 3. For such a 4-noded element.5. .

6) has been solved [17]. [K ]{D}= {R} where [K] (3. stresses.s are available after the structural governing equation (3.2 Thermoelastic Analysis – Theory and Finite Element Formulation The thermoelastic analysis is done to determine the displacements at nodes.s) are known. strains. etc. T2.f. If T1. These nodal d.1 =∫ b T ∫ [B] [E ][B]t dxdy a −b − a {D} {R} = Nodal displacement matrix. The results of the thermal analysis are input as nodal loads to determine the temperature effects on the elastic behavior. The shape functions are modified accordingly but the bilinearity is maintained.6) = Structural stiffness matrix of the bilinear element. Stress on an element can be calculated when its nodal degrees of freedom (d. as shown in Fig 3. T3 and T4 are the nodal temperatures of the element shown in Figure 3.o.f.o. A parametric formulation effectively solves for complex geometries. y) = (a − x)(b − y ) (a + x)(b − y ) (a + x)(b + y ) (a − x)(b + y ) T1 + T2 + T3 + T4 4ab 4ab 4ab 4ab 3.1.∂  ∂x [B] =  0  ∂  ∂y   0  ∂ [N ] ∂y  ∂  ∂x   (3.5) The shape functions in finite element packages like ANSYS have a robust solver that can solve for elements that are not necessarily a regular rectangular shape. = Load matrix. 12 . then the temperature field across the element is given by: T(x.

T4]T Equation 3. {Te} = [T1. T4]T then the thermal nodal loads for solving the thermomechanical analysis are given by the following matrix [17]. T3. If the nodal temperatures of an element is given by the following matrix. The temperature at nodes is found by performing a thermal analysis that is solved before running the thermomechanical analysis. for isotropic materials  E  The stresses in a finite element can be solved from the solution of Equation 3.8 [17]. The stress field for plane stress problems is given by Equation 3.7) The stress field for plane strain problems is given by Equation 3. 13 . the generalized Hooke’s law is given by the following equation: {σ }= [E ]({ε }− {ε 0 }) where {ε0} represents the strains due to the thermal expansion caused by the nodal loads exported from the thermal analysis. T2.[E] E = Young's modulus matrix =  0  0  0 E 0 0 0  . T2.7 and 3. σ x  E   σ y  = 2 τ  1 − ν  yx    0  1 ν ν 1 0   1 −ν  0 0  2    αT       [B ]{D}− αT     0     (3.8 are used to solve for the stresses in plane stress conditions and plane strain conditions respectively. For a general case.7 [17]. T3.6 and imposing the temperatures as nodal loads acting at the nodes. Thermal Load Matrix: {Re} = εAeα [T1.

8) The specific boundary conditions. 5 and 6 as per the specific problem. 14 . σ x  ν 1 − ν E    ν 1 −ν σ y  = τ  (1 − 2ν )(1 + ν )  0 0   yx     0  1 − 2ν   2  0  αT       [B ]{D}− αT     0     (3. element types and loading for the various analyses are described in brief in Sections 4.

Dufrane and Kannel [7] analyzed the catastrophic seizure of bearings due to dry friction by a simple 1D equation relating the seizure time to the bearing operating parameters and material properties. A series of experiments were also conducted to determine the seizure time t ss = CρC p 2(1 + ν )αq s • 1 −1  Rs  + n  Rbo − (n − 1)  Rs Rbo      (4. When the surface temperature of the shaft or the bushing reaches 320ο above the ambient temperature. The seizure criterion was based on a cut-off temperature.1) Equation (4. The analysis included axial length of the 15 . Khonsari and Kim [8] performed a comprehensive 2D numerical analysis of a bearing undergoing seizure during start-up. A significant amount of work has been reported that analyzed the thermomechanical interactions in stationary loaded bearings susceptible to TIS. These conditions occur during start-up or in an event of lubricant supply blockage. A critical PV/C number was proposed to be an influential parameter for assessing the seizure time. the shaft is assumed to seize in the bearing. Although TIS can take place in lubricated bearings. it is predominant when a hydrodynamic bearing happens to operate in the boundary or mixed lubrication regimes. Bishop and Ettles [6] analyzed the thermoelastic interaction of a journal in a plastic bushing that is interference-fit with the shaft. This means that the bearing will seize even if the clearance is very large. THERMALLY INDUCED SEIZURE IN JOURNAL BEARINGS DURING START UP Thermally induced seizure (TIS) in journal bearings is a mode of failure that can occur quite suddenly and end up with a catastrophic damage to the system.CHAPTER 4.1) predicts a linear relation between the seizure time and the operating clearance.

Their analysis revealed that the thermoelastic deformation between the journal and the bearing led to a reduction in clearance in a non-linear fashion. Other factors that affect the seizure process are the surface quality and chemical composition. The aim of the work is to provide a simple empirical relation for practicing engineers and lubrication system designers to predict the seizure time.al. It revealed that seizure in unlubricated conformal contacts was primarily due to a thermal ratcheting effect in a positive feedback of increases of interfacial pressure and heat. In lubricated journal bearings.12] analyzed the axle burn-off and stack-up force of a railroad roller bearing using the finite element method and independently verified the results of Hazlett and Khonsari's findings. 16 . This study established a non-linear nature for the clearance loss as a function of time. [11. Hazlett and Khonsari [9. More recently. The objective of this work is to perform a comprehensive study of seizure in bearings during start-up and arrive at a seizure time evaluation formula that is a function of the various operating parameters.10] performed a detailed finite element analysis to gain insight into the nature of the contact forces and encroachment of the mating pair leading to TIS of a dry bearing during start up. surface roughness and microstructure of the shaft and the bushing. which can help them to design instrumentation systems and warning devices to take necessary precautions. starvation of the oil was found to be a direct cause for seizure. Wang [13] performed a review of published results on TIS in conformal contacts. The effect of bearing parameters on the seizure time was also performed as a sensitivity study.bearing as well as the effects of a misaligned shaft on the thermoelastic interactions. More recently. The starvation led to wear and temperature rise in the contact area that caused the onset of seizure. Wang et.

7. Analysis Model The model consists of a shaft rubbing on the inner surface of the bushing as shown in Figure 4.1 Finite Element Modeling Procedure The finite element modeling is done using ANSYS 5. 2. The variation of radial clearance. The finite element model of the present work employs a finer mesh than the mesh used by Hazlett and Khonsari to evaluate the contact forces with more accuracy.10] was recreated.1. The contact forces results in the generation of frictional heat on the entire surface of the shaft and in the area where it contacts the bushing inner radius. A transient heat transfer analysis was done to model thermal effects of dry frictional heating on the journal and the bearing. Due to the rise in temperature. A simplified 2-dimensional analysis is performed. A 2-D static contact analysis was performed to determine the contact forces and the contact angle. A transient thermoelastic analysis was performed to study the interactions of the journal-bearing pair during bearing start-up. The analysis of a bearing undergoing TIS during start up is done by the following steps: 1.4. 3. The analysis assumes that the contact pressure is uniform in the axial direction and no crowning or misalignment is present in the system. [19] First. the analysis done by Hazlett [9. The effect of bearing length is analyzed in the 2-D analysis by taking into account the change in the contact width and change in the heat flux generated with change in bearing length. contact forces and ovalization of the bearing were studied in this analysis. the shaft expands and its encroachment to the bushing leads to 17 .

a loss of clearance. At some point in time, the bearing clearance reduces to a minimum and shaft starts to encroach the bearing. Analyses show that typically during TIS, the following three phenomena occur: (i) The contact forces increase, increasing the heat generated. (ii) The contact angle increases causing a higher percentage of heat entering the bush. (iii) New areas of contacts are established resulting in a chain reaction of events leading to a rapid loss in the operating clearance. In the simulations presented in this paper, these processes were implemented by performing a thermal analysis and a thermoelastic analysis in a stepwise linear fashion. The model utilized a one-half symmetry and neglected the heat conduction in the axial direction.

Rbi L Load, W Rs Journal

2C Pillow Block

Figure 4.1 – Schematic of a journal supported on a pillow block
4.1.1 Transient Thermal Finite Element Analysis Thermal Elements - The thermal analysis is done to determine the temperature

distribution in the journal-bushing pair. The journal and bushing were modeled as 4noded solid thermal elements viz. PLANE55. PLANE55 element has a single degree of

18

freedom, namely temperature. This element is compatible with the 4-noded structural solid element used in the thermomechanical analysis. This means that the results of the thermal analysis can be successfully exported to perform the thermoelastic analysis
Loading - The loading applied in the thermal analysis consists of the heat

generated by the frictional contact at the shaft-bushing interface. The total heat generated in the contact zone is:
Q = fu ∑ Pi
i =1 n

(4.2)

where f is the coefficient of friction of the rubbing surfaces, ∑Pi is the summation of the contact forces; n the total number of elements in contact and u is the surface velocity of the shaft. The summation of the contact forces was equal to the total load acting on the system, W. The frictional heat generated heats the entire surface area of the shaft in an on-off mode. Thus the surface of the shaft is intermittently heated in the contact area and cooled in the clearance area. The frictional heat generated heats only the surface of the bushing that is in contact. Temperature and flux continuity exists in the contact patch of contact of the journal and the bearing. The heat flux due to the frictional heating is computed using the following formulae.
qs = f .u. ∑ Pi f .u. ∑ Pi = As 2πR s L

(4.3)

Boundary Conditions - The rotating shaft is heated periodically when it makes

contact with the bushing. This can be thought of as an on-off type heating. It was shown by Hazlett [10] that the on-off heating could be modeled as an average heat flux on the entire surface. Also, there is dissipation of heat by convective cooling by the air within the clearance of the journal and the bushing

19

Convection on inner surface, h Bearing inner radius, Rbi Shaft radius, Rs Convective Cooling on outer surface h

Heat Flux at interface, q

Figure 4.2 - Finite Element Model for Thermal Analysis, with loads and boundary conditions. (Not to scale)

To represent the periodic heat dissipation in the finite element model, the nodes on the surface of the shaft are coupled. The temperature on the surface of the journal and the bushing at the interface is constant and is modeled by coupling the temperatures at the nodes on the interface. The outer surface of the bushing is subject to natural convection. The finite element mesh, thermal boundary conditions and the thermal loads are schematically represented in Figure 4.2. For clarity, the operating clearance in Figure 4.2 is scaled by a factor of approximately 300.
4.1.2 Non-linear Transient Elastic Finite Element Model Elastic/Thermoelastic Elements - The steady state analysis to find the contact

forces and the transient thermoelastic analysis use two types of elements in the Finite Element Program ANSYS 5.7. The solid element PLANE42 is used to model the journal

20

and the bush. This element is a 2D bilinear element with the displacements in the x and y directions as the degrees of freedom. The radial clearance between the journal and the bush is modeled using two-noded contact elements, namely CONTAC52. Contact elements are used to model gap and they come into effect only when the two nodes that make the element come into contact. The element properties include a normal stiffness value that governs the resistance to normal load. The finite element programmer assigns the stiffness value for the contact element. To find the stiffness value, the theoretical Hertzian contact width [20] must be determined. The stiffness value is then fixed at a value by trial and error so that the Hertzian contact width and the contact width found by Finite Element method are the same. The ANSYS 5.7 manual suggests that the normal stiffness value for the contact element CONTAC52 can vary in the range of 0.01 to 100 times the stiffness of the underlying solid element material. The manual also suggests the programmer to use his judgment in the trial and error method used to finalize the stiffness value. The verification of the stiffness value against the Hertzian theory [20] was followed to establish validity.
Loading - The loading for the non-linear thermoelastic analysis consists of the

thermal loads applied as nodal temperatures and the radial force acting on the journal. The time dependent thermal load is obtained from the results of the transient thermal analysis. The static load, W is applied to act in the negative y-direction on the shaft. As the model utilizes half-symmetry, a load of W/2 is applied.
Boundary Conditions - Symmetry boundary conditions are used to model the

one-half symmetry as shown in Figure 4.3. The constraint of the bearing on its outer surface is modeled by fixing the bearing at the node under the shaft on the outer edge of

21

∑ Pi (t ) i =1 m (4. The present model assumes that TIS is complete when the frictional torque reaches at least 50 times the driving torque. This constraint approximates the boundary conditions that could be expected from a pillow block type of bearing as shown in Figures 4. W/2 Displacement BC Figure 4. When the frictional torque increases beyond the extent of the driving torque capability.1 and 4.1.3.3 Seizure Criterion Frictional torque is the torque resisting the driving torque exerted by the motor. f .ANSYS Model for Elastic Analysis.the bearing on the symmetry plane.3 .Rs . with loads and boundary conditions 4.4) 22 . it can be concluded that the journal has seized in the bearing. Line of symmetry Load. The contact forces acting on the gap elements at any instant of time determine the frictional torque at any time. τ (t )= 2.

This process leads to a positive feedback loop and a chain reaction leading to a rapid failure due to TIS. This leads to an increase in the contact forces and the formation of an extra contact area. The reasons for such an abrupt increase in frictional torque are: (i) As explained previously. (ii) The operating clearance of the bearing just before seizure is reduced to a significantly lower value compared to the steady-state operating clearance. f is the coefficient of friction. 4. Increase of contact forces raises the frictional heat flux and sets up a positive feedback that accelerates the loss of clearance. The available 23 .2 Results and Discussion The encroachment of the shaft on to the bushing with concomitant reduction in the clearance continues until the seizure is complete. The increase in the frictional torque is abrupt once the ovalization of the bearing causes the shaft to encroach the bushing. Rs is the radius of the shaft and Pi is the contact force at the i'th gap element and n is the number of elements in contact. The process is a complex. as there is further loss in the operating clearance. The frictional torque increased to exceedingly large values within typically 3 seconds after the first instance of establishment of new areas of contact. Analysis shows that TIS is initiated by the ovalization of the bearing combined with the uniform outward expansion of the shaft yielding contact between the top of the shaft and the inner bushing surface. This is due to the thermal expansion of the journal and the bearing into the operating clearance area.where Τ is the frictional torque. the increase in contact forces increases the frictional heat generated and the increase in frictional heat means that the shaft would expand more increasing the contact forces and establishing more area of contact. nonlinear phenomenon.

at the top of bearing Plot (c) .clearance just before the extra contact occurs has already reduced to an exceedingly small value.5 x 10-3 m Rb = 51.6 illustrate the onset and completion of seizure for a journal bearing during start-up. The operating parameters are listed below: W = 4400 N N = 250 rpm Rs = 25. Note the ovalization and establishment of extra contact. A. The following plots.4 .0 x 10-3 m C = 0.0125 x 10-3 m L = 51. 27 seconds after starting.4. Figures 4.Isothermal plot of temperature rise.4 – Temperature rise as a function of time in a journal bearing during start up Plot (a) – Temperature rise. 29 seconds after starting – Note the formation of extra areas of contact and rapid temperature rise 24 .Temperature rise.0 x 10-3 m Figure 4. 25 seconds after starting Plot (b) .

5 .Variation of frictional torque during start-up – Note that the frictional torque increased to 50 times the initial frictional torque to indicate seizure 25 . N 6000 5000 Forces after 27 seconds 4000 Forces after 25 seconds 3000 Initial Contact Forces 2000 1000 0 0 50 100 150 Angle from midplane. deg Figure 4.8000 7000 Forces after 29 seconds Contact Force.6 .Variation of contact forces with time during start-up Figure 4.

it will be assumed that the seizure is complete 3 seconds after the first instance of ovalization. Therefore.4 (a) shows temperature rise on the journal bearing pair just before ovalization.5 and 4. Additional simulations with difficult operating conditions also reveal that this condition holds. Figures 4.Figure 4.6 show the variation of contact forces and frictional torque over time. which illustrates the substantial rise in temperature and contact area. the clearance has not reduced to zero clearance at all points proving that the frictional torque is a better seizure criteria than zero-clearance. K (W/mK) 10 x 10-3 .4-4. Figure 4.1 – Operating parameters for the finite element model and simulations Range of shaft radii.23 x 106 52 26 . Figure 4.6 that. These simulations reveal that the frictional torque reaches very high values after approximately 3 seconds after ovalization is first experienced.5 x 10-3 51x 10-3 -100 x 10-3 1. although the frictional torque increased to 50 times the initial torque.1. The ranges of the operating variables were arrived by taking the operating parameters used by Dufrane and Kannel [7] for experimentation and by Kim and Khonsari [8] and by Hazlett and Khonsari [10] for doing the analytical and finite element modeling as the basis. The parameters used are: Table 4. 4.4 (b) shows the initiation of extra contact at the top of the bearing.37. L (m) ρCp (N/m2K) Conductivity.4 (c) shows the temperature increase after 2 seconds after the establishment of extra areas of contact.3 Model Parameters The model used to study the effects of TIS during bearing start-up is given in Table 4. The ovalization is realized in the analysis when additional contact is established at the top of the bearing. Rs (m) Range of bearing lengths. It is interesting to note from Figures 4.

3 1000-10000 0. C. f. The reasoning behind this cut-off point is that the lubricant supply would have been 27 . For speeds higher than 2000 rpm and the operating clearance specified. α). Eb (GPa) Poisson's ratio. W. The seizure time can be written as a function of the various operating parameters.3 x 10-5 – 2 x 10-5 200 110 0. ts. νs = νb = ν Range of loads.1 – 0.4 Generalization 0. L κ. W (N) Range of friction coefficient. The simulations were performed to establish the effects of the operating parameters on the seizure time during the start-up. the seizure time is nearly the same. f. f Range of speeds N (rpm) Range of radial clearances. The variation of the seizure time during the system start-up is studied when the operating parameters (variables N. C (m) Convection on all faces. the seizure times are large enough such that the lubricant flow would have been established during starting. For speeds lower than 150 rpm. ts = g (N. The lower limits of the load and the speed were decided if seizure times exceeded 120 seconds. Rs.α (m/mK) Young's modulus of shaft. Then a generalized equation is derived depending on the individual relationships of the operating parameters with the seizure time. L. Rb.005 x 10-3 . α) are varied.0. about 2 seconds. Rb.0325 x 10-3 80 25 To derive a relation between the seizure time. κ. All the other parameters were also varied based on a similar rational. a statistical procedure followed based on the work done by Hamrock [21]. T∝ (°C) 4. W.3 150-2000 0. Rs. h (W/m2K) Atmospheric Temperature. and the operating parameters. C. Es (GPa) Young's modulus of bush. The speeds are varied from 150 rpm to 2000 rpm. This means that the seizure time is almost asymptotic with respect to speed.

Rs.0325 mm keeping the other factors (N. The influence of load on the seizure time is determined by varying the load from 1100 to 10000 N keeping the other factors (N. Rb. 4. κ. Rb. The friction coefficient was varied from 0. f. Increase in the shaft radius produces two contrasting effects on the seizure time. L. The shaft radius was varied from 10 mm to 37. A decrease in clearance means that the shaft has lesser room to encroach before contacting the bearing and hence the seizure time decreases with decrease in clearance. Seizure time decreases with increasing speed of the shaft. C. The influence of radial clearance on the seizure time is determined by varying the load from 0. W.1 Influence of Operating Parameters Increasing the speed of the shaft means that the heat generated at the interface of the journal and the bearing increase in a direct proportionality. f.3. κ. This means that the frictional heat generated is increased. α) constant. Increase of shaft radius means that the surface velocity of the bearing increases for the same speed. L. This means that the seizure time decreases with increasing load. The length of the shaft was varied from long bearings (L/D=2) to short bearings (L/D = 0. The clearances were decided based on typical operating clearances of journal bearings.005 to 0.1 to 0.5 mm. But an increase in the radius of the shaft also means that the thermal mass of the shaft is increased.4. The effect of increase in load is a proportional increase in the frictional heat generated. These are typical values of friction coefficients for steel on brass sliding [23].established within 2 minutes of starting the system. So the net 28 .5 mm meant that the bearing thickness was less than 10 mm. The upper limit of the shaft radius was limited as increasing the shaft radius above 37.5). Rs. These are the limits that are generally known as the infinitely long and infinitely short approximations [22] (ILA and ISA). α) constant.

Rb. It can be inferred from Appendix A that the increase in the shaft radius decreases the time for seizure up to a particular shaft radius. W. α) constant. W. L. The influence of shaft radius on the seizure time is determined by varying the load from 10 to 37. C. Rb. f. α) constant.1 to 0. The influence of coefficient of friction on bearing seizure time is similar to the effect of speed.5 to 100 mm keeping the other factors (N. f. Rs. Beyond a shaft radius of 25 mm. L. C. Rs. Hazlett [10] varied the convective heat transfer coefficient on the outside radius of the shaft and did not find any appreciable effect on the thermoelastic interactions of the bearing. The influence of the coefficient of friction is determined by varying the friction coefficient from 0. The variation of outside radius may be significant only for very thin bearings. It was found that varying the outside diameter has negligible effect on the seizure time. the effect of outside radius of the bushing on the seizure time can be neglected.effect of these two modes of heat generated and heat stored decides whether the bearing seizes faster or slower. the effect of the increased thermal mass nullifies the effect of the increase in the surface velocity. The simulations were carried out for bearings with Rs/th = 2. f. Increasing the coefficient of friction increases the heat generated thus reducing the seizure time. W. C.5 (thick bearings) and they seized at around the same time. Therefore for most applications. α) constant.0 (thin bearings) to Rs/th = 0. 29 .5 mm keeping the other factors (N.3 while keeping the other operating parameters (N. The influence of the outside radius of the bushing on the seizure time is determined by varying the outside radius from 32. L.

where Req = Rs Rbi R (R + C) = s s Rbi − Rs C 4.7.5 Non.Dimensionalization In order for the analyses and the corresponding statistical procedure to be relevant to any journal-bearing system irrespective of the units used. 30 .5. Non-dimensional seizure time during start-up: t ss = t ssκ Rs 2 Thermal strain: Req L ε= fWωα ks Modified aspect ratio: λ= . Rs*θ*L would also increase. The equation derived fits the data with a goodness value of 98. Rs. (i) The increase in bearing length increases the Hertzian contact width meaning the area of heating on the bushing is more.The influence of bearing length on the seizure time is determined by varying the length from 50 to 100 mm keeping the other factors (N. use of non-dimensional parameters is preferable. W. All the operational variables namely load speed and coefficient of friction were combined to get a single non-dimensional parameter and all the dimensional variables like operating clearance. f. The effect of the bearing length is two-fold. (ii) The increase in bearing length means that the bearing area. 4. shaft radius and the bearing length were combined into another non-dimensional parameter as described below. Rb.64% as shown in Figure 4. α) constant. The non-dimensionalization scheme utilizes only two nondimensional operating parameters. while the analyses were carried out for 6 operating parameters.1 Statistical Treatment of Non-dimensional Operating Parameters A power series equation is used to find the relation between non-dimensional thermal strain and non-dimensional seizure time. C.

00 1.9% and is given by Equation 4.000 0.00 0.0031ε t ss ∝ ε −1.00 3.5) ε tss = 0.2478 (4.2478 t ε (4.6) The relationship between the parameter Γ and the modified aspect ratio was also derived using power series expressions.025 0.2478 4.00 2.8 – Variation of non-dimensional seizure time with thermal strain Let us define parameter Γ as Γ= ss −1. Two ranges of Rs/C values were considered for the simulations performed and the equation derived fits the data with a goodness value of 97. 31 .50 3.030 tss Figure 4.00 0.005 0.0016 ε -1.5) t ss = 0.50 0.50 −1.50 1.010 0.The relation between the non-dimensional seizure time and thermal strain is given by the following equation (4.015 0.7.2478 2. The variation of Γ with modified aspect ratio depends on a particular range of Rs/C values.020 0.

the empirical results are in agreement with published results.825 ε -1.10] for journal bearings.494 ε -1.2478 λ t ss = 1.12] for an analysis dealing with axle burn-off in railroad roller bearings and cases 7 through 17 were reported by Bishop and Ettles in their study of TIS in journal bearings.2478 λ -1. Wang et al.2 for the fifth and sixth cases were averaged values. Conry and Cusano [11.8) -1. the nominal operating clearances ranged from 0. 32 . Applying the empirical relation for the other reported clearances would lead to erroneous predictions.05 mm simulations reported was considered for verification. The first four cases correspond to the simulations reported by Hazlett and Khonsari [9. As can be seen from Table 4.8) is verified for its validity using some of the results published by Hazlett and Khonsari [9.825 ε -1.12] and Bishop and Ettles [6].7) (4.10]. Wang.02718 4. the fifth and sixth cases were reported by Hao Wang [11. In the axle burn-off studied by Hao Wang [12].494 ε -1.6 reported a number of additional analyses with axle bearings. Therefore only the 0. a number of the clearances were outside the range of simulations presented in this study and were not considered.6494 For 500 > Rs/C > 1000 For 1000 > Rs/C > 5100 For 500 > Rs/C > 1000 For 1000 > Rs/C > 5100 (4.Γ = 90.6 Verification and Analysis The above empirical relationship (4. The material properties used in Table 4. However.6494 = 1.2.02718 ⇒ t ss = 90.05 mm to 5 mm and the axle bearing was lubricated by grease due to very heavy loading and the material properties were reported as functions of temperatures.

1x10 42 38 1.44x10-5 0.1x10 35 11 -5 1.44x10 0.23x10 0.1x10 650 583 1.5x10 0.44x10 0.5 1000 0.2 .44x10 0. published.5 6 -5 2000 1 x 10 5x10 0.5 1000 1 x 106 5x10-5 0.005 1 52 1.5 500 1 x 106 5x10-5 0.1x10-5 240 186 -5 1.005 33 .051 52 4.005 1 52 1.15 0.44x10 0.47x10-6 0.5 6 -5 300 1 x 10 5x10 0.Comparison of Empirical results with published results Coefficient of thermal Seizure time Seizure time expansion.005 -5 1 52 1.3 -5 1 52 1.44x10 0.25x10-5 0.23x10-5 0.5 1000 0.5 Bearing Thermal Thermal Coefficie Length conductivity diffusivity nt of m W/mK m2/s friction 0.0255 250 4400 1.3 0.3x10-5 60 59 -5 1.1x10 290 246 1. m/mK sec sec -5 1x10 28 28 1x10-5 2 2.15 0. rpm N m m -5 250 4400 1.44x10-5 0.3x10 3 1.44x10 0.0255 52 4.15 -6 0.Table 4.0255 1800 4400 1.005 1 52 1.44x10-5 0.005 -5 1 52 1.1x10 510 351 1.051 52 4.25x10-5 0.005 -5 1 52 1.1x10-5 117 78 -5 1.005 -5 1 52 1.3 1.3 -5 1.23x10-5 0.1x10 7 7.44x10-5 0.2 x 10 5x10 0.1562 30 5.1x10-5 270 186 -5 1.0255 250 4400 1.4 1x10-5 21 21 1x10-5 16 14 -5 1.1x10-5 400 352 -5 1.005 -5 1 52 1.5 6 -5 1000 0. calculated.005 -5 1 52 1.03825 52 4.3 x 106 5x10-5 0.5 6 -5 1000 5 x 10 5x10 0.15 -5 0.5 6 -5 1000 0.5 x 106 5x10-5 0.0781 560 160000 50x10-5 0.5 6 -5 1000 1 x 10 2.005 1 52 1.1562 30 5.44x10 0.4 x 10 5x10 0.0255 560 160000 5x10-5 0.47x10 0.5x10-5 0.23x10-5 0.25x10 0.1x10 50 33 No.25x10-5 0. 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 Shaft Speed Load Clearance Radius.0781 1000 1 x 106 0.

the bushing was rigidly constrained. Note that although the simulations done in the present study did not consider fully lubricated cases.1) and the empirical Equation (4. To gain further insight into the TIS behavior.Bishop and Ettles [6] studied TIS in journal bearings with non-metallic liners using analytical 1-D and a 2-D analysis.1) used by Dufrane and Kannel. From Figure 7. Results from the 2-D analysis of TIS in lubricated journal bearings were considered for verification in Table 2. ANSYS 5.5 is used in 34 . The results reported in Table 4. we plot the change in the operating clearance as a function of time based in Dufrane and Kannel's equation (4. the results derived in Equation (4.7 calculates the heat-partitioning factor based on the thermal mass and material properties at the contact area such that there is continuity of temperature and flux at the contact interface. A heat partitioning of 1 is unreasonable as it means that all the frictional heat generated would be transmitted into the shaft.9 for two heat partitioning factors (n = 0.8) derived in the present study. Also. The results in Bishop and Ettles' paper were reported till the temperature reached 320° above ambient temperatures that were considered as the burn-off temperature of the linings. the seizure times were larger than the values obtained using Equation (4. As the present study has considered the bushing constraint the outward expansion of the bushing.5 and 1). it can be seen that the present model compares close to Dufrane and Kannel's model when a heat-partitioning factor of 0. The analyses done by Dufrane and Kannel [7] did not consider the expansion of the bushing and the thermal expansion of the shaft was only considered. The result is shown in Figure 4.8) still holds good.2 were taken when there is complete loss in clearance that was before the limiting temperature was reached.

Whereas Equation (4. seconds Figure 4.Equation (1).0010 Dufrane and Kannel's model n=0. 35 . It is recommended to use the empirical relations for operating conditions close to the conditions used in the simulations. Note: n = heat partition factor There are some limitations that restrict the use of these relations depending on the operating conditions. This physically realistic prediction was first discussed by Khonsari and Kim [24].0005 0 50 100 150 200 Time for seizure. The seizure time formula developed in this study predicts the loss of clearance with time is not a linear process. in 0.5 Dufrane and Kannel's model n=1 0. the predicted results here reveal that this is not true for large clearances. 1.9 – Comparison of seizure time with Dufrane and Kannel's model and the present model.0020 Clearance loss.1) implies that TIS occurs regardless of the size of the clearance. 0. This is especially true for the operating clearance.0015 Present Model 0.

36 .2. The boundary condition used in this analysis allows some flexibility for the bearing to expand outwards. If the bearing were fully constrained. the seizure times would be appreciably reduced. The boundary condition for the fixation of the bearing varies with different applications.

the lubrication regime can shift from hydrodynamic into the mixed and boundary lubrication regime. There are a number of situations whereby a journal bearing system while running steady may suddenly experience a disturbance in its operating conditions.2. flow interruption is modeled as a transition between various lubrication regimes. we focus our attention to the situation whereby a system undergoes a transient flow disturbance. In this part. landing and maneuvering [14]. for instance. A "lubricant supply perturbation" in an otherwise steady-state operation may be caused by a brief interruption of lubricant supply due to a clogged filter or air locking. In this chapter.CHAPTER 5. otherwise TIS becomes imminent. TIS OF JOURNAL BEARINGS TRIGGERED BY A TRANSIENT FLOW DISTURBANCE The thermoelastic interaction of journal bearing pair in the event of a disturbance in lubricant flow is a largely neglected area of study. When the lubrication perturbation occurs for a short period of time. Transient operating conditions are also predominant in aircraft during the periods of taking-off.1 and 5. Ideally once the lubricant supply is resumed. The methodology behind the application of the lubricant perturbation can be graphically represented by Figure 5. Lubricant flow disturbance is assumed to occur over a 37 . Evaluation of the thermomechanical response following the resumption of lubricant supply is necessary to understand if the system can survive the interruption of lubricant flow. This transition is assumed to take place linearly with time. the bearing reverts back to the thick-film regime. This transition of the bearing operating conditions from the thick-film to the boundary lubrication regime can significantly affect the operating coefficient of friction and the convective heat transfer coefficient within the clearance.

During the transition period of 70 seconds. during which there is a gradual transition from full-film to boundary lubrication.2.10]. This period consists of an initial rise in the coefficient of friction and concomitant reduction of convective heat transfer coefficient over a period of 70 seconds.period of 150 seconds from the onset of the disturbance to full resumption of steady operation. During the transition regime.005. The nominal convective coefficient for air at the speeds involved is around 80 W/m2K [9. A change in the steady-state thick-film heat transfer coefficient was found to have minimal effect on the bearing seizure time as the seizure occurs primarily due to the flow disturbance and not due to the steady-state temperature rise. which represents lubricated metal-to-metal sliding. A convective heat transfer coefficient of 200 W/m2K is assumed for thick-film lubrication. a duration of 10 seconds in the boundary lubrication followed by a 70 seconds transition to full-film lubrication is assumed.15. the coefficient of friction is ramped linearly with time to 0. Then. 38 . the convective heat transfer coefficient is ramped linearly till the HT coefficient reaches 80 W/m2K as shown in Figure 5. The coefficient of friction for the fully lubricated bearing is typically in the order of 0.

1 – Variation of friction coefficient during lubricant perturbation 200 Convective Coefficient.2 – Variation of convective HT coefficient during lubricant perturbation 39 .100 0. W/m K 180 2 160 140 120 100 80 1000 1100 1200 1300 1400 1500 1600 Time.150 Coefficient of friction 0. seconds Figure 5.0. seconds Figure 5.000 1000 1100 1200 1300 1400 1500 1600 Time.050 0.

3. 2.3 40 . The thermal analysis is performed at thick-film hydrodynamic conditions until steady state is achieved and then the transient flow disturbance is applied in the transient. A transient heat transfer analysis was done to determine the time required for the journal bearing to attain the steady-state conditions.1 Finite Element Analysis Modeling and Seizure Criteria The procedure and modeling approach is similar to Section 4. The bearing seizure is assumed to be complete when the frictional torque greatly exceeds the driving torque. contact forces and ovalization of the bearing were studied in this analysis.0 (Figures 5. A transient thermoelastic analysis was performed to study the interactions of the journal-bearing pair under the influence of perturbation. where TIS during bearing start-up was studied with a few minor differences. The perturbation conditions (increase in coefficient of friction and decrease in the convective heat transfer coefficient) were applied after the time required for the journal-bearing pair to reach steady state.1 and 5.2). The methodology of the application of the lubricant perturbation is done as per the procedure described in Section 5. The variation of radial clearance.1.5. A 2-D static contact analysis was performed to determine the contact forces and the contact angle. The finite element procedure overview is given below 1. This is similar to the methodology explained in Section 4.

3a. b – Plots of temperature rise during a transient flow disturbance 41 . Plot (a) – Steady state temperature rise Plot (b) – after 62 seconds Figure 5. This is due to the fact that the increase in the coefficient of friction and the reduction of convective heat transfer is gradual as the bearing moves from a thickfilm regime to boundary lubrication over a period of time.5. The thermomechanical process of seizure is explained for a journal bearing with the same operating conditions as discussed in Section 4 is discussed in this section.2 Results and Discussion The failure of journal bearings undergoing a transient disturbance in flow is primarily due to the seizure of the shaft in the bearing due to thermal expansion of the shaft and the bearing and concomitant reduction in clearance. The time to seize differs from the TIS of bearings during start up. This process is similar to the TIS of bearings during start up.

deg Figure 5.3c. N 5000 After 65 seconds 4000 After 64 seconds 3000 2000 After 62 seconds 1000 0 0 50 100 150 200 Angle from midplane.d – Plots of temperature rise during a transient flow disturbance 7000 6000 Contact Forces.Variation of contact force after the onset of ovalization due to flow disturbance 42 .4 .Plot (c) – After 64 seconds Plot (d) – After 65 seconds Figure 5.

except that the operating parameters are of a wider range than that of the start up analysis.4 through 4.3 Model Parameters The model parameters are similar to Table 4.5 with Figures 4.5 – Variation of frictional torque when a transient flow disturbance occurred Comparing Figures 5. Operating parameters for simulating TIS in a journal bearing undergoing a transient flow disturbance Shaft Radius.23 43 .5 x 10-3 51x 10-3 -100 x 10-3 1.1.3 through 5. Rs (m) Length of Bush.6. the only difference being the time taken to seize. L (m) ρCp (N/mm2K) 10 x 10-3 -37. Table 5. 5. It is an interesting fact that the ovalization of the bearing and the establishment of the extra areas of contact contributed to the rapid seizure of the bearing in the case of flow interruption also. it can be seen that the process of TIS follows a similar procedure.1.Figure 5.

0125 x 10-3 80 25 20-100 5. As the analysis procedure is different from the thermoelastic analysis done for predicting seizure time.4. νs = νb Load. Non-dimensional seizure time.15 500-3000 0. where Req = Rs Rbi R (R + C) = s s Rbi − Rs C 44 . Es (GPa) Young's Modulus of Bush. W (N) Friction Coefficient. C (m) Convection on outer face. K (W/mK) α (mm/mmK) Young's Modulus of Shaft.4 Non-Dimensionalization and Generalization Similar to Section 4. The non-dimensional parameters used in the analysis are defined below.Conductivity. ts = ts t ref t sp t ref Non-dimensional seizure time during flow disturbance: t sp = Thermal strain: ε= Req L fWωα ks Modified aspect ratio: λ= . Eb (GPa) Poisson's Ratio. N (rpm) Radial Clearance. a statistical analysis was performed and an empirical relationship was derived for arriving at the seizure time when a journal bearing undergoes an interruption in lubricant supply. tref (sec) 52 1e-5 200 110 0.3 2200-12000 0. a different non- dimensionalization technique was followed. f Speed. T∝ (°C) Transition time. h (W/m2K) Atmospheric Temperature.

it means that those bearings cannot sustain even a lubricant cut-off for a few seconds. If a bearing seizes between 71 and 80 seconds it means that the seizure is during the boundary lubrication period. Similarly if seizure time is greater than 80 seconds. The empirical relationship does not apply to bearings that may be susceptible to seizure in hydrodynamic regime.1) are explained next. they give a finite value 45 . When the empirical formula (5. The physical meaning of the seizure time and restrictions of the empirical relation (5. Most of the cases considered in this analysis falls within this category. Some heavily loaded bearings that operate at high speeds or with low clearances may seize in fully lubricated conditions.1) is applied for these conditions. it means that the bearing failed in the transition period before boundary conditions are achieved. 2.5489 (0.e. i.Using the results of 50 sets of results. For bearings that seize within 70 seconds. If the empirical relationship predicts a seizure time of.1) provides a useful empirical relation about the seizure time to be expected when a bearing undergoes a disturbance in lubricant supply.1). before the mixed/boundary lubrication phase occurs.5 Analysis of Results ( 0.0065λ + 9. t sp = ε 5. the relationship between the nondimensional time for seizure and the non-dimensional operating parameters is given by the following equation (5.1) The empirical relation (5. 3.314 ))−1 (5. 1. it means that the seizure has taken place after the lubrication is restored. say 48 seconds.

[15] where a "no-seizure" condition was arrived based on the bearing operating parameters. Only a 2D analysis was performed. Typically. [16] performed an extensive study of TIS in thick film journal bearings. The analysis is valid in the transition time ranges of 20 to 80 seconds. The transition time is the time taken for the bearing to go from fully lubricated mode to boundary lubrication mode. The Finite Element Analysis neglected the heat transfer in the axial direction. Most of the bearing materials typically have a bronze base and shaft made of steel. The boundary conditions used in this analysis allow some flexibility 46 . The TIS failure of lubricated journal bearings has been studied by Pascovici et. That was the reason behind choosing this material pair. Jang et. 5. So the above empirical relations can be used only for material combinations that are close to the material properties of the materials that were used in the simulations. seizure times of 20 seconds or below fall under this category. The material properties used in the simulations pertain to mild steel shaft and bronze bearing. If the empirical relationship yields a result of 20 seconds or lower. 6. 4. it means that the bearing has already seized or it would seize even for a slight change in the operating conditions.of seizure time when the bearing had already seized even with no perturbation. The thermoelastic behavior was theoretically formulated and experimentally verified. 7.al. The boundary conditions for the fixation of the bearing vary with different applications.al.

051 m L = 0.2 – Comparison of simulated seizure time and empirical seizure time Speed 250 500 750 1000 1200 1400 1600 1800 2000 2200 2400 Empirical Seizure Time 99 67 53 45 40 37 34 32 30 28 27 Simulated Seizure time 100 68 58 50 46 44 38 34 30 30 26 %error -1.5%. 5.49 -1.for the bearing to expand outwards.13 -6.89 -4. The results are tabulated in Table 5. W = 8800 N N = 500 – 2400 rpm C = 0.88 -3. Table 5.14 -7.0255 x 10-3 m Rs = 0.00 -5. the seizure times would be appreciably reduced.44 -5.2.41 -5.7.0255 m Rb = 0.051 m For the above operating conditions. the results of the empirical relation are compared to the results of the simulations performed using ANSYS 5. If the bearing were fully constrained.6 Verification A verification is performed for some operating conditions and the validity of the empirical relation is checked against the simulated results obtained from ANSYS finite element simulations for the following operating conditions. The % error was computed and the max error was found to be 7.41 47 .01 -1.67 -7. The comparison is repeated for other simulations also. It was found that the empirical relation provides a realistic and conservative estimate for seizure time for most of the cases.

An interesting observation can be made from Table 5. This means that the bearing seized even after the resumption of lubricant supply.2 for the operating speed of 250 rpm. The seizure time predicted by the empirical formula and from the simulations is about 100 seconds. 48 .

Typically in the pin-bushing pair. A typical schematic of the pinbushing oscillating pair used in the undercarriage of heavy-duty earth-moving equipment is shown in Figure 6. Galling is defined as the mode of failure when there is plastic deformation and localized welding of relative moving parts and subsequent wear of the welded parts due to the shear loads. the pin-bushing pair maybe packed with grease during assembly. These pin-bushing pairs are typically made of mild steel and surface treated with a protective coating. There is no provision for a continuous supply of lubrication in the system and the packed grease is the only source for lubrication. After some time of operation in field. the pin-bushing pair exhausts the supply of grease packed in the assembly by normal operational leak as well as wear and tear.CHAPTER 6. Depending on the application. Galling can occur due to very high stresses and/or due to very high rise in temperatures. THERMOMECHANICAL ANALYSIS OF A HEAVILY LOADED OSCILLATING PIN-BUSHING PAIR 6. They are subject to very heavy loads generally operating at low speeds.1 Introduction and Problem Definition Heavy-duty earth-moving machinery use linkages that are supported on a pin and bushing pair. The absence of proper lubrication leads to the wear of the anti-friction liner on the surfaces of the pinbushing pair. the pin is stationary while the bushing oscillates at low frequencies. The combination of heavy loads and high temperatures in the oscillating 49 . Industry is now using automatic grease feeders or recommends feeding grease periodically. Further operation causes intimate metal-to-metal contact between the pin and the bushing that leads to high heat generation in the contact region.1. very high loads and improper cooling leads to failure of the bearing either by galling or by thermally induced seizure (TIS). The absence of proper lubrication.

Thermally Induced Seizure (TIS). the temperatures involved and the time to failure are generally higher. except that in the oscillating pin-bushing assemblies. depending on the operating conditions. Oscillating Bush Track link Stationary pin Figure 6. Once failure starts the surface temperature at the pin-bushing interface shoots up to very high values introducing possible galling or TIS. The underlying phenomenon is similar to that described in Chapters 4 and 5. on the other hand. is the mechanism of failure if there is a complete loss in operating clearance in the pin-bushing assembly as the pin encroaches into the bushing.pin-bushing assembly in earth-moving machinery makes it vulnerable to galling.1 – Track Retention Assembly with the oscillating pin and bushing 50 .

The finite element method is used to perform thermal and thermomechanical simulations to determine the performance of the oscillating pin-bushing pair at various operating conditions.3 Theoretical Hertzian Contact Analysis A theoretical analysis is performed as first steps to determine the contact forces and ascertain the contact patch using Hertzian theory of elastic contact[21]. The contact area and the contact pressure are calculated from the Hertzian theory of elastic contact. For a line contact.6. The objective of this research is to develop an analysis for predicting failure of pin-bushing under oscillating loads. the Hertzian half-contact width is given by Equation (6.7 Transient finite element thermomechanical using ANSYS 5.7 6. b= 1. N/m2 51 . (i) (ii) (iii) Theoretical Hertzian contact analysis Transient finite element thermal analysis using ANSYS 5. The problem is solved in a 3-step routine described briefly below and in detail in sections 6. m Load acting on the pin.1) b w R L E = = = = = semi-contact width. m Length of contact in the normal direction. N Equivalent Radius of contact.598 where wR LE (6.1) [21] as.4 to 6.6.2 Problem Description and Solution Methodology The objective of the work done in this chapter is to study the thermomechanical interactions of an oscillating pin-bushing pair undergoing oscillating loading conditions. m Equivalent Elasticity modulus.

The finite element analysis program like ANSYS or FlexPDE [25] can calculate the partition of heat flux automatically taking into account the thermal mass and material properties of the contacting bodies. The results of the transient thermal analysis are used in doing the thermomechanical analysis to determine the pin-bushing interactions. 6. The objective of the transient thermal analysis is to obtain the time-dependent temperature contours of the pin-bushing under the application of the oscillating frictional heat flux.2) Equivalent Young's modulus: 1 − ν 1 2 1 − ν 2 2  E = 2 +  E2   E1 (6. The heat generated is distributed as per a partition factor between the pin and the bushing.3) where the subscripts 1 and 2 refers to the pin and bushing respectively and the denominator has + for convex contact and – for concave contacts.2) and (6. The heat generated is a function of the contact forces. The partition of heat is dependent on the material properties. thermal mass and the relative velocity of the pin and the bushing.3) Equivalent radius: R= R1 R2 R1 ± R2 −1 (6.4 Thermal Analysis Frictional heat is generated at the interface between the pin and the oscillating bushing when the two surfaces slide against each other. For a Hertzian elastic contact. As a pin and bushing contact is concave. we use the difference of the radii of the pin and bushing as our denominator. 52 . coefficient of friction and the sliding velocity.The contact region is measured using the equivalent elastic modulus and equivalent radius of contact. the equivalent radius and the equivalent modulus of elasticity are given by Equations (6.

The total heat generated is calculated by the following simple equation Q = fPv (6. coefficient of friction and the linear velocity. The modeling is done by applying the heat flux as a function of time and space.3 Implementation of Oscillating Heat Flux The finite element modeling of the oscillating heat source requires a special formulation. m Length of the bushing in the normal direction.6. in radians θc 6. PLANE55.2 Calculation of Heat Flux The heat flux applied depends on the contact forces.4.5) qp = R * L *θ c where R L = = = Radius of the pin.4. This element is compatible with the 4-noded structural solid element used in the thermomechanical analysis.1 Element Types The pin and bushing are modeled in ANSYS 5.4. This means that the results of the thermal analysis can be successfully exported to perform the thermomechanical analysis.7 using the four-noded solid thermal elements viz. m Contact Angle. 6. m/s The heat flux applied is the heat generated applied on the area of contact and is calculated as Q (6. N Relative velocity. 53 .4) where f P v = = = Coefficient of friction Contact force.

The process is continued until a full oscillation is completed. ⇒ Angular velocity. tosc/13.05 54 .2 Schematic of application of oscillating flux. α is the oscillation angle Sample Calculation of Time for 1 Oscillation An example of calculation will now be presented for a bushing operating at a surface velocity of 0.08 = = 1. Figure 6.08 m/s with a stationary pin of radius 0. This sequence is programmed into a ANSYS macro (sub-routine) and repeated cyclically to complete the transient analysis.6radians / sec = 91. the flux is moved to the next set of six elements and then solved again for 2x(tosc/13).2 (a) and then it is solved for the 1 time step i.3 Heat Source α α (b) Extreme right . For the next time sub-step.05 m.+α (c) Extreme left . The calculated flux is applied over a set of 6 elements in the mean position. The calculated time for a single oscillation is broken up into a series of 13 time-steps.-α (a) Mean Position Figure 6. This solution is appended to the previous solution.The time taken for 1 complete oscillation (tosc) is calculated from the speed of oscillation.e.67 degrees/sec R 0. ω = v 0. The application of the flux for 1 cycle is illustrated in Figure 6. the angle of oscillation and the radius of the shaft.

6) 55 . time = 0.05sec time = 0.35 sec time = 0.12ο. the time taken for application of flux over a patch of 6 elements before moving to the next set of elements is 0.6 rad/s.45 sec time = 0.25 sec time = 0.50 sec time = 0.05 seconds and the flux is applied and solved for every sub-step as illustrated in Figure 6. using Equation 6.3. is found to be 26..10 sec time = 0.55 sec time = 0.30 sec time = 0.133 ReD2/3 Pr1/3 where the dimensionless numbers are. Equation 6.15 sec time = 0.65 seconds Figure 6.65 sec Total Time for 1 Oscillation = 0.5ο. At the angular velocity of 1.60 sec time = 0.F.40 sec time = 0. The time taken to negotiate one oscillation (tosc) is found to be 0. This means that the flux should be applied over 6 elements. This formula is applicable for a cylinder rotating in an infinite quiescent fluid medium.4 Convection Film Transfer Coefficient between Bushing Outer Radius and Atmosphere The forced convection coefficient can be theoretically calculated by the empirical relation.3 –Application of oscillating heat flux as a function of time and space 6.20 sec time = 0. Mills [26] for a cylinder rotating in a fluid.The element size used in the finite element model has an angle of 4.65 seconds.4. The theoretical Hertzian contact angle. (6.6 suggested by A. NuD = 0.

= = = Nusselt Number. m2/s ω νk The convection between the bushing outer boundary and the environment was found by assuming a mild flow of air over the bushing surface at a speed of 5 mph at a temperature of 300 K. hD/k Reynolds number. 56 .NuD ReD Pr where. This value was close to the value used by the authors (45 W/m2K) in their analysis.69 for air) h D k = = = = = Convective heat transfer coefficient. ωD2/νk Pradtl Number (0.al [11]. s-1 Kinematic viscosity of the convecting fluid. The convection coefficient calculated using Equation (6. W/m2K Diameter of the surface on which convection takes place. Convecting Fluid = Air – Properties are obtained from [18] 300 K (Room Temperature) 0. m Thermal Conductivity of the convecting fluid. The following parameters are used to calculate the convection HT coefficient. W/mK Angular velocity of the rotating member.6) was 46 W/m2K (Appendix).1 m 5 mph (assumed) Temperature of Fluid = Radius of Bushing Velocity of Air Verification = = The validity of the above method was verified by applying the formula to find the convection coefficient of an axle and cone problem solved by Wang et.

Properties from Incropera and Dewitt [18] 300 K (Room Temperature) . hrad.5. Convecting Fluid = Air . The total heat transfer coefficient in the gap between pin and bushing thus consists of 2 components. hforced and Equivalent convective heat transfer coefficient hrad The total convective heat transfer coefficient is a summation of the forced convection term and the equivalent radiation term. The fluid properties of air are varied to include the entire temperature range that can be expected for various operating conditions. Convective Coefficient in the Gap between Pin and Bushing The convective heat transfer coefficient in the gap between the pin and bushing needs to be examined in more detail as the temperature rise in the interface between the pin and bushing is high.1100 K (Steady State) Temperature of Fluid = A sensitivity study was done by calculating the forced convection heat transfer coefficient (hforced) at various temperature values and it was found that the increase in temperature did not have a significant effect in the temperature range of room 57 . which is at a lower temperature. htotal = hforced + hrad Calculation of Forced Convection Coefficient inside the Pin-Bushing Assembly (6.7) The forced convection of air inside the pin-bushing assembly is evaluated using the empirical relation (6. (i) (ii) Forced convective heat transfer coefficient.6) using the operating conditions given below. The radiation heat transfer in the analysis by using the concept of linearized radiation coefficient.6.4. The high temperature difference between the pin and bushing would contribute to radiation heat transfer between the hot pin and its bushing.

W/m2 Emissivity of the radiating body Stefan-Boltzmann's constant Surface Temperature of the hotter body. qrad = = = = = Heat transfer due to radiation.10) where.8) where. The average forced convective heat transfer coefficient of the sensitivity study was found to be 10 W/m2K for the temperature range of 300 K – 1100 K.8) [18] 4 q rad = εσ (T 4 − Tsurf ) (6. hrad is the equivalent convective coefficient due to radiation expressed as. The heat transfer due to radiation is given by Equation (6. οK ε σ T Tsurf The above equation can be rewritten as 2 q rad = εσ (T 2 + Tsurf )(T + Tsurf )(T − Tsurf ) (6.temperature to steady state temperature (about 1100 K). q rad = hrad (T − Tsurf ) (6. Calculation of Equivalent Convective Coefficient due to Radiation Steady state thermal analyses for various operating conditions reveal that interface temperatures are in the range of 600 K – 1100 K. These high temperature values can influence the thermal analysis due to heat transfer by radiation between the hot body and the colder body.9) The above equation can be put in the form of a simple convection heat transfer type equation by summing the terms together to be written as an equivalent convective coefficient due to radiation. 2 hrad = εσ (T 2 + Tsurf )(T + Tsurf ) 58 . K Surface Temperature of the cooler body.

Also the heat transfer due to radiation is predominant in the clearance between the pin and bushing and it is not significant on the outer surface of the bushing.6 Boundary conditions for the Thermal Analysis Heat Flux Applied on the pin. hrad is used along with the regular convective coefficient in the pin-bushing clearance. 6. The value hrad is calculated by taking the average temperatures on the surface of the pin and the bushing. The variation of hrad with time is linear. Convective HT coefficient On the Bushing outer edges In the Pin-Bushing clearance Average hforced Equivalent hrad = = = = 30 W/m2K hforced + hrad 10 W/m2K Linear function of temperature 59 .4. The heat transfer due to radiation is insignificant when the surface temperatures are very low.The equivalent convective coefficient due to radiation varies with the temperature. So.5. the equivalent convective coefficient due to radiation. qp= Calculated for different load conditions using Equation 6. Temperatures on the pin and the bushing are coupled in the contact region to achieve continuity of temperature in the contact zone.

The stiffness value is then fixed at a value by 60 .40 hrad hconv htotal Convective coefficient. W/sqmK 35 30 25 20 15 10 5 0 300 800 1300 1800 2300 2800 3300 3800 Time. The finite element programmer assigns the stiffness value for the contact element. The element properties include a normal stiffness value that governs the resistance to normal load.1. To determine the stiffness value.5 Transient Thermomechanical Analysis 6. The gap is modeled using 2-noded contact elements CONTAC52. This element is a 2D bilinear element with the x and y displacements as the degrees of freedom and the temperatures obtained from the transient thermal analysis can be applied as nodal loads. The radial clearance between the pin and the bushing is a problem that has to be dealt with care. sec Figure 6.5. the theoretical Hertzian contact width is found. Element Types The solid element PLANE42 is used to model the pin and the bushing.4 – Variation of effective convective coefficient with time 6. Contact elements are used to model gap and they come into effect only when the two nodes that make the element come into contact.

This is done by having coarser elements at the center of the pin and finer elements at the interface.trial and error so that the Hertzian contact width and the contact width found by Finite Element method are the same.3 Boundary Conditions The radial load is applied on the pin on the centerline and the bushing is constrained on the outer diameter.5 – Mesh pattern for the oscillating pin-bushing assembly 6. This methodology is adopted.5. as our area of interest is the interface of the pin and bushing Pin Mesh Bush Mesh Figure 6.2 Meshing The meshing pattern is scaled such that the element size is reduced in the radial direction towards the bushing-pin interface. The steady 61 . The verification of the stiffness value against the Hertzian theory is a good way to establish validity. This means that the elements towards the center of the pin degenerate from the standard 4 noded rectangular element into triangles.5. All the degrees of freedom are constrained. 6.

Restraining Link Oscillating Bush Stationary Pin Figure 6.75 x 10-3 50 x 10-3 . The results of the thermal analysis are applied as thermal nodal loads with respect to time. Whenever there is contact established between the pin and bushing.35 x 10-3 62 . The contact element is activated only when there is a physical contact between the pin and the bushing. 6.state analysis is performed and the contact element results are analyzed.0. The radial load acting on the pin is applied as a point force acting on the centerline. Rs (m) Range of bearing lengths. C (m) 35 x 10-3 . The stiffness value of the contact element is adjusted such that the contact angle is in agreement with the Hertzian contact angle.6 – Schematic of a Retention track assembly The transient thermal analysis is followed by a transient thermomechanical analysis to determine the effects of temperature rise on the operating parameters such as operating clearance and the frictional torque. Operating Parameters Table 1 – Operating parameters for the finite element model and simulations Range of pin radii.167 x 10-3 0. The contact element CONTAC52 can display the contact forces and the value of clearance in the gap.1 x 10-3 .6. the contact element results give us the magnitude of the contact force and negative clearance. L (m) Range of radial clearances.

The oscillating heat flux produces a temperature field as shown in Figure 6. f Range of oscillating frequencies. α (m/mK) Young's modulus of pin-bushing. Figure 6. Pin radius. The maximum temperature also can be seen to move along in the direction of the moving flux. T∝ (°C) 6.3 2 11205 The heat flux as discussed earlier is applied as an oscillating flux with respect to time on the contact region.1 – 0.23 x 106 54 1 x 10-5 200 0. Time steps 4 through 10 give the temperature profile when the heat flux moves from the extreme right position (+α in 63 .05 0. W (kN) Range of friction coefficient. while one of the test cases is presented in this section for discussion.411 0.723.1 0. W (N) 0. ωo (rad/s) Atmospheric Temperature. f Oscillating frequency. C (m) Coefficient of friction.7 Results and Discussion 1. E (GPa) Poisson's ratio. Rs (m) Bearing length.7 gives the temperature profile for the first complete oscillation.7. νs = νb = ν Range of loads.25 x 10-3 0.Density of pin-bushing ρ (kg/m3) Conductivity.3 1-2 25 Simulations are performed for a wide range of operating parameters.222. Simulations of failure of an oscillating pin-bushing corresponding to the following operating conditions are presented. The temperature profile for the first four time steps shows the effect of flux moving from the mean position to the extreme right position. L (m) Radial clearance. k (W/mK) Thermal expansion coefficient.3 66. ωo (rad/s) Load.

2) to the extreme left position (-α in Figure 6. From the contour profiles. Thus the failure can be attributed to both TIS and thermal galling. The loss in clearance causes the pin to ovalize and come into contact with the bushing at the top of the shaft. This leads to increased area of contact and increased contact forces. When the flux moves from the extreme right position (Time step #4). Also the frictional torque increased to very high values 6 seconds after ovalization was established.2). it can be seen that the temperature increases to 752οC after about 6 seconds once ovalization is established. Figure 6.8 through 6. This temperature is sufficient to cause scuffing in steel [2]. it leaves a temperature tail that trails off at the extreme right position (+α). 64 .Figure 6. As a result. the frictional heat further contributes to the process of expansion.7 – Temperature contour during 1 cycle of oscillation The temperature rise due to the frictional heating causes thermal expansion of the pin and concomitant reduction in operating clearance. The temperature profiles are given in Figures 6.11.

Figure 6.8 – Temperature contour just before ovalization Figure 6.9 – Temperature contour 1 second after initialization of ovalization 65 .

11 – Temperature contour 6 seconds after initialization of ovalization Note: The maximum temperature at the contact reached 752 οC in Figure 6.Figure 6.11 when galling takes place. 66 .10 – Temperature contour 3 seconds after initialization of ovalization Figure 6.

where Req = Rs Rbi R (R + C) = s s Rbi − Rs C 67 . Non-dimensional parameters were used to generalize the operating parameters.8 Generalization A number of simulations (Table 6. the maximum temperature reached was about 300-400 οC and failure occurred by thermally induced seizure (TIS). The non-dimensional parameters used in this analysis are similar to the non-dimensional parameters used in Chapter 4 in the study of TIS in journal bearings during start-up and are given below. 6.2) were performed to study and derive a generalized equation to determine the time of failure of an oscillating pin-bushing assembly.The failure pattern is not typically like the example shown in Figures 6. a combination of thermal galling and TIS was observed within the range of simulations performed. Non-dimensional seizure time: t fo = t foκ Rs 2 Thermal strain: ε= Req L fWω oα ks Modified aspect ratio: λ= . The failure mode has a combined mode of thermally induced seizure (TIS) and thermal galling only in very heavily loaded pin-bushing assemblies and where the coefficient of friction and/or the oscillating frequencies are high. In cases were friction coefficients were low.11. Typically lightly loaded assemblies with low coefficients of friction failed by TIS mode while the heavily loaded assemblies and when the coefficients of friction were higher.8 to 6.

3 0.2 0.416 x 10-03 0.5 100.65 15.2 0.5 83.58 492 18.5 60.25 0.42 8.8 2 1.5 167. (rad/s) 0.849 x 10-03 12.71 6.27 0.416 x 10-03 8.56 29.75 417 4.15 0.Table 6. time for Ratio. Rp.139 x 10-03 3.1 0.602 442 3.05 205 265 314 380 440 465 405 446 461 380 325 50 0.988 x 10-03 6.15 0.83 x 10-03 111.42 0.5 111205 35 75 100 6.416 380 100.74 343 9. Length.2 1 1.3 1.42 18.5 6053 2535 2399 1488 1207 936 749 582 3018 564 322 604 1006 1207 1408 2532 2380 805 564 1207 2938 Thermal Strain.1 0. f C (mm) (kN) (mm) time.554 x 10-03 10.21 50 0.18 7.143 374 7. of friction.58 322 14.2 0.75 201 100.47 7.4 1.19 300 15.93 49.5 5.84 2.1 0. Radius.98 3.422 x 10-03 3.83 x 10-03 3.5 1. tfo (s) (mm) ωo.25 0.6 1.17 125.2 – Variation of failure time at various operating conditions Angular Pin Coefficient Load.97 x 10-03 12.L Failure velocity.2 1. tfo 37.20 3.35 225.72 222.6 1.47 461 250.ε 2.35 NonTemperature Aspect dimensional at Failure.2 1. W Clearance.3 0.83 71.849 x 10-03 5.25 100 66.25 50 50 100 166.67 68 .544 329 3.λ (οC) failure.

9% as shown in Figures 6. The number of simulations required is thus greatly reduced.The number of simulations is reduced by just varying the non-dimensional parameters instead of varying all the operating parameters making sure that the physical significance of the operating parameters is not lost.12 and 6. The effect of various operating parameters like load (W).008 0.0 0.006 ε -1.0 tfo = 0.0 25.002 0.0 5. 69 . 40.0 35.0 15.014 ε Figure 6. The results of varying the various operating parameters are given in Table 6.010 0.1 were curve fitted to a goodness value within 98.012 0. aspect ratio and non-dimensional failure time were combined to give the following equation 6.1. The expressions relating the thermal strain.0 10.006 0.4064 t fo 20.0 30. coefficient of friction (f) and frequency of oscillation (ωo) on the failure time is found by varying the non-dimensional parameter ε.12 – Thermal Strain vs non-dimensional failure time The results presented in Table 6.11 for failure time for pin-bushing that undergo failure under oscillating boundary conditions.13.0 0.004 0.

Krishnamurthy and Khonsari [27] verified the accuracy of the finite element model by applying Duhamel's theorem to problems involving oscillating heat source.4064 0. The application of proper boundary conditions for the thermal analysis has to be verified to ensure accuracy of the model.9 Verification λ The primary reason for failure in the analysis of the pin-bushing assembly is due to the increase in temperature due to frictional heating.11) 0.71) t ε 0.t fo = ε −1.255λ − 70.025 0.015 ε tfo = (2.4064 − 70.005 0. The results of the finite element problems solved using ANSYS and FlexPDE 70 .71 (6.020 fo −1.010 0.000 0 50 100 150 200 250 300 Figure 6.255λ −1.13 – Variation of aspect ratio with non-dimensional seizure time 6. A comparison was done between the heat conduction finite element problem done using ANSYS for a one domain problem with the oscillating heat source applied on the pin. FlexPDE [24]. They also performed dimensional and non-dimensional analysis using another FE solver.4064 2.

assuming that the contact pressure is uniform across the bearing length and there is no misalignment of the shaft in the bearing. 6. To obtain a practical failure time.matched closely with the analytical solution.11) gives the failure time for pin-bush assembly when there is continuous oscillating motion. (iii) Equation (6.10 Restrictions and Applications (i) The finite element was solved by approximating it as a 2-dimensional problem.11). ωo. (iv) The Hertzian contact pressure in the contact region is elliptic and hence the frictional heat flux is also elliptic. The time to failure predicted is for the worst-case scenarios. (ii) The time for failure as given in Equation (6. But a constant heat flux is applied in the thermal analysis. there is only intermittent loads acting on the pin-bushing assembly. But in practice. the loading cycle of the oscillating pin-bushing has to be incorporated with Equation (6. (v) The oscillating frequency. The material properties were assumed to be constant and the analysis did not consider plastic deformation.11) gives the approximate failure time (tfo) after there is complete loss of initial lubrication in the pin-bush assembly and loss of anti-friction coating. is assumed to be constant for small intervals of time. 71 . The convergence of the finite element model was also checked using three different mesh sizes.

An empirical relationship was developed to estimate the seizure time using these operating parameters. CONCLUSIONS 7.2 TIS in Journal Bearings During Transient Flow Disturbance The sudden stoppage in the lubricant flow.02718 For 500 < Rs/C < 1000 For 1000 < Rs/C < 5100 7.1 TIS in Journal Bearings During Start Up When rotating machinery that is supported on fully lubricated bearings are started up from rest. A statistical analysis was performed using the simulated results to determine the relationship of these operating parameters on the seizure time.2478 λ -1. The procedure for applying the empirical relation and the interpretation of results that can be deduced from the empirical relation is described. The final expression for the seizure time is given by the following equations t s = 90. in a fully flooded journal bearing was analyzed. The effect of the lubricant cut-off was analyzed by studying its effect on the bearing operating parameters such as clearance loss and frictional torque by a thermoelastic finite element 72 . The effect of the dry sliding during start-up was analyzed by studying the effect of start-up friction on the bearing operating parameters such as clearance loss and frictional torque by a thermoelastic finite element model.494 ε -1. the lubrication flow may not have been established and there would be metal-to-metal contact.2478 λ -1.CHAPTER 7. A series of simulations were performed by varying the operating parameters. The interruption in the lubricant flow was modeled in finite element analysis by an increase in the coefficient of friction and a decrease in the convective cooling. The restrictions involved in using the empirical relation are also specified.6494 t s = 1. The accuracy of the empirical relation was verified using published results.825 ε -1.

5489 (0. The empirical relationships relating seizure time to operating parameters when a flow disturbance is encountered are given by the following equation.255λ − 70. The time for failure in the non-dimensional form was derived from a series of simulations and is given by the following equation.0065λ + 9.71 73 . t fo = ε −1. It was found that failure of these pin-bushings occurred by thermally induced seizure for lighter boundary conditions and by a combination of TIS and thermal galling for more severe boundary conditions.3 TIS and Thermal Galling in Oscillating Pin-Bushing Analysis The analysis of a heavily loaded pin-bushing assembly subject to oscillating loading was analyzed using finite element thermal analysis and thermomechanical analysis. t sp = ε ( 0.model.4064 2. The reasons for bearing failure were analyzed by performing simulations for different types of operating conditions.314))−1 7. An expression relating the seizure time to the operating parameters for a journal bearing undergoing a transient flow disturbance was derived in the same procedure as explained for the case of TIS during start up.

Dyson A. 1996. ASME Journal of Heat Transfer. 1979. 1989. ASME Journal of Tribology. 37-52 7. Hazlett T. O. Pennsylvania.D.S. 1. 106. Scuffing.311-319 12. 79. Wear. and Kim H. Thesis. ASME Journal of Tribology. 1989. 1981. The seizure of journal bearings by thermoelastic mechanisms. 1984. and Kannel J. 1993.L. Ph.M. Ettles C. Ling F. Patula E. On thermally induced seizure in journal bearings. Steady State Temperature in a Rolling Cylinder Subject to Surface Heating and Convective Cooling. 25. Steady State temperature Distribution in a Rotating Roll Subject to Surface Heat Fluxes and Convective Cooling. ASME Journal of Tribology. Wang H. Thermoelastic behavior of journal bearing undergoing seizure – a finite element study. M. 13. Conry T. Vol. ASME Journal of Tribology. 80-91 2. M.J. 175215 3. Dufrane K. 1996. Wear.M. Finite element model of journal bearing undergoing rapid thermally induced seizure. Hazlett T. 111. 74 . Tribology International. 1992a. and Khonsari M. 857-865 6. University of Pittsburgh. Pittsburgh. Khonsari M. F. Effects of Cone/Axle Rubbing Due to Roller Bearing Seizure on the Thermomechanical Behavior of a Railroad Axle. and Khonsari M. 661-7 9. Ulysee P. p 35.BIBLIOGRAPHY 1.. and McC.F. Thermal aspects of galling of dry metallic surfaces in sliding contact. University of Illinois at UrbanaChampaign. Axle Burn-off and Stack-up Force Analyses of a Railroad Roller Bearing using the Finite Element Method. 115. and Winer W. Thermal Response of Rolling Components Under Mixed Boundary Conditions: An Analytical Approach. 1982. Wang H. 111. 1958.M. Gecim B. 1990 11. 36-41 5.140-147.118. H. 288-92 8. pp. Treatise on Material Science and Technology. 103.L. 120-127 4. Bishop J. and Cusano C. Department of Mechanical and Industrial Engineering. ASME Journal of Tribology. and Saibel E. pp.L.. Thermally induced seizures of journal bearings. Thesis. 177-82 10.

44-66. CRC Press. ANSYS 5.P. 1989. Thermohydrodynamic Seizure: Experimental and Theoretical Analysis. 117.F. 2002 75 . ASME Journal of Tribology. and Winer W.M. 292 25. Department of Mech.O. 14.. 1997. 1996.. and H. Peterson M. Pascovici M. Third Edition 18. Thesis.20e 3D Online Reference Manuals.R.Y. "Application of Duhamel's Theorem to Problems Involving Oscillating Heat Source" M. 1995 27.. Lacey S. John Wiley and Sons. Mills A. E. 2001. and Dewitt D..D.. and Jang J..D.Y..S. Booser E. John Wiley and Sons.. Bearings for Aircraft Gas Turbine Engines (Part 1).. ASME Journal of Tribology.. LSU. p 621. Heat and Mass Transfer. John Wiley and Sons Inc.7 Online Users Manual.13. On the Modeling of Thermomechanical Seizure. Seizure Failure of journal-bearing conformal contacts. 1998. Wang Q..L..D.M. 8-15 17. Jang J. Khonsari M. pp.M. pp 153-155 24. Discussion on paper titled "Thermally induced seizures of journal bearings" by Dufrane K. Fourth Edition 19. PDE Solutions Inc 26. Kawamura H. Institute of Tribology. Khonsari M. pp 117-19. 120. Cook R. Concepts and Applications of Finite Element Analysis. Malkus D. ASME Journal of Tribology. Ph. 1989.. 1976. Wear Control Handbook. Fundamentals of Heat and Mass Transfer. ASME. E. 744-7 16. pp. Richard D Irwin Inc.8-16.. Khonsari M. 2001 23. Khonsari M. 210. CRC Handbook of Tables for Applied Engineering Science. ANSYS Inc 20. 1998. Department of Mechanical Engineering. and Pascovici M. 115 21. Krishnamurthy H.M. and Plesha M.. 93-102 22.J.. and Kim H. Applied Tribology. Hamrock B.. 1996-2001. Elastohydrodynamic lubrication of point contacts. 1995. pp. The University of Leeds. Engr. 111.D Thesis. Second Edition. 1970. 1980. 1-8 15. Incropera F.J. Wear. and Kannel J. S. NSK Technical Journal – Motion and Control. FlexPDE: A Flexible Solution for Partial Differential Equations Version 2. P..B.. 5. Bolz R. Tuve T.

Train speed = 55 mph.6) was 46 W/m2K while Wang et.60 + 0. The convection coefficient calculated using Equation (6. Axle rotating speed = 560 rpm Ambient air temperature = 25 oC Shaft radius = 78.387 Ra ( 1/6 ) 2 76 . al.387*Ra^(1/6)/(1+(0.632 mm Maple Program: > restart. ( 2/3 ) ( 1/3 ) Nusselt forced := . the protected names norm and trace have been redefined and unprotected > Nusselt[forced] := 0. reported the convection coefficient to be 45 W/m2K for the operating conditions reported below.APPENDIX .MAPLE PROGRAM TO CALCULATE CONVECTION COEFFICIENT The formula to find the convection coefficient of an axle and cone problem solved by Wang et.60 +          ( 8/27 )   ( 9/16 )    1    1 + .  Nusselt free :=  .133*Reynolds^(2/3)*Pr^(1/3). > with(linalg): Warning.7209732340          Pr       .559/Pr)^(9/16))^(8/27))^2. Reynolds := ω Dia 2 ν > > omega := V/R.133 Reynolds Pr > Reynolds := omega*Dia^2/nu. ω := V R Free Convection coefficient from Incropera and Dewitt > Nusselt[free] := (0.al [11].

Dia := 2 R Enter the Kinematic viscosity of the fluid .6112 Enter the radius of the shaft.5e-6.8. > g := 9.0000225 Enter the Fluid Bulk Temperature (in degree K) > T[inf] := 300. in meter > R := 0. g := 9. g β ( Ts − Tinf ) Dia 3 Ra := να > beta := 1/T[inf].7209732340  1         Pr        .69.00001589 Enter the velocity of the shaft in m/s > V := 4.6112. Tinf := 300 77 . Pr := .60 +          ( 8/27 )   ( 9/16 )      1 + .133   Rν  2     ( 2/3 ) Pr ( 1/3 )  +  . R := . β := 1 Tinf > Dia := 2*R. in SI units > k := 26. ν := .89e-6.078632 Enter the thermal conductivity of the fluid.8 Enter the Diffusivity of the fluid > alpha := 22.air > nu := 15.078632.0263 Enter the Prandtl Number for the fluid > Pr := 0.387 Ra ( 1/6 ) 2 > Ra := g*beta*(T[s]-T[inf])*Dia^3/(nu*alpha).3e-3. α := . k := .69 Enter the Acceleration due to gravity value in the proper units.> Nusselt[total] := Nusselt[forced] + Nusselt[free]. V := 4.  V Dia Nusselt total := .

h := 46. > T[s] := 373.09811930 78 .Enter the Mean Temperatur rise on the surface of the shaft. Ts := 373 > h := Nusselt[total]*k/Dia.

He successfully defended his thesis work on October 7. After his undergraduate studies. computer aided designing/manufacturing and finite element methods. 2002. India. he worked as a design engineer in Ashok Leyland Limited. He joined Louisiana State University in August 1999 to pursue a master’s degree in mechanical engineering. India. 1976. His area of interests include tribology. in Madurai. and will receive the degree of Master of Science in Mechanical Engineering in December 2002. He earned his bachelor’s degree in mechanical engineering from Bharathiyar University. Rajesh Krithivasan was born on February 4. for 2 years. 79 .VITA Mr. India.