iii

ABSTRACT

ROLE OF RAILWAY VEHICLE-TRACK SYSTEM AND DESIGN
PARAMETERS ON FLAT-INDUCED IMPACT LOAD

Md. Rajib Ul Alam
Wheel flats are known to cause high magnitude impact loads at the wheel-rail
interface, which can induce fatigue damage and failure of various vehicle and track
components. With demands for increased load and speed, the issue of wheel flats and a
strategy for in-time maintenance and replacement of defective wheels has become an
important concern for heavy haul operators. In this study, an analytical model of the
coupled vehicle-track system is developed by integrating a pitch plane model of the
vehicle with a two-dimensional model of the flexible track comprising 3-layers together
with a nonlinear rolling contact model. The track system model is periodically supported
by sleepers and ballasts characterized by their lumped parameters. The commonly used
Hertzian nonlinear contact model is utilized in analysis of the vertical vehicle-track
interactions. Generalized coordinate method is employed to solve for the coupled partial
differential equations of the track and ordinary differential equations of motion for the
lumped-parameter vehicle model. The validity of the coupled vehicle-track system is
demonstrated by comparing the simulation results with the reported measured data and
analytical solutions. The validated model is utilized to investigate the characteristics of
impact forces due to wheel flats and its effect on motions and forces transmitted to
vehicle and track components. The results are analyzed to examine the sequence of
events as the wheel flat enters the contact area. The magnitudes and predominant
frequencies of wheel-rail contact forces are examined in terms of system and operating
iv
parameters. A comprehensive parametric study is performed to study the effects of
selected vehicle-track design and operating parameters on the wheel-rail impact loads,
and forces transmitted to the bearing, railpad and ballast in the presence of single and
multiple wheel flats within either same or adjacent wheels. The study shows that the
magnitudes of cross wheel impact forces are larger when the phase angle between the two
flats is small. The impact force may be higher or lower than that caused by a single flat
depending on the relative positions of the flats. The study further revealed that factors
such as primary suspension stiffness, railpad and ballast stiffness, rail mass, sleeper
spacing, bending stiffness of rail and ballast mass have significant influence on the
impact load, whereas secondary suspension properties and ballast damping show
negligible effects.
























v

TABLE OF CONTENTS

LIST OF FIGURES

ix
LIST OF TABLES

xiv
NOMENCLATURE


xv
CHAPTER 1

INTRODUCTION AND LITERATURE REVIEW

1.1 INTRODUCTION

1
1.2 LITERATURE REVIEW

3
1.2.1 Wheel defects

5
1.2.2 Vehicle models

15
1.2.3 Modeling the track Components

23
1.2.4 Track system models

30
1.2.5 Wheel-rail contact models

35
1.2.6 Simulation methods

37
1.3 THESIS SCOPE AND OBJECTIVES

39
1.4 ORGANIZATION OF THE THESIS


40
CHAPTER 2

VEHICLE-TRACK SYSTEM MODEL AND METHOD OF ANALYSIS

2.1 INTRODUCTION

42
2.2 VEHICLE SYSTEM MODEL

43
2.2.1 Pitch-plane vehicle model

45
2.2.2 Equations of motion 46
vi

2.3 TRACK STRUCTURE

47
2.3.1 Track system model

49
2.3.2 Equation of motion

51
2.4 WHEEL-RAIL INTERFACE

53
2.5 METHOD OF ANALYSIS

58
2.6 SUMMARY


63
CHAPTER 3

MODEL VALIDATION AND VEHICLE-TRACK SYSTEM RESPONSE

3.1 INTRODUCTION

65
3.2 MODEL VALIDATION

66
3.3 RESPONSE ANALYSES OF THE VEHICLE-TRACK SYSTEM
MODEL

72
3.3.1 Wheel-Rail Contact Force Response

74
3.3.2 Force Responses of the Vehicle and Track Components

79
3.3.3 Displacement Responses of the Vehicle-Track Components 84
3.4 SUMMARY


93
CHAPTER 4

PARAMETRIC STUDY

4.1 INTRODUCTION

95
4.2 SELECTION OF IMPORTANT MODEL PARAMETERS

96
4.3 INFLUENCE OF WHEEL FLAT

98
4.3.1 Effect of single wheel flat on wheel-rail impact force 98
vii
4.3.2 Effect of multiple wheel flats on W/R impact loads

103
4.3.3 Effect of flat length and depth

115
4.4 PARAMETRIC STUDY ON VEHICLE PARAMETERS

118
4.4.1 Effect of speed

118

4.4.2 Effect of wheel load

119
4.4.3 Effect of unsprung mass

120
4.4.4 Effect of suspension stiffness and damping on peak W/R
impact load

121
4.4.5 Effect of suspension stiffness and damping on peak bearing
force

122
4.5 PARAMETRIC STUDY ON TRACK MODEL PARAMETERS 124
4.5.1 Effect of Rail Mass per unit Length

124
4.5.2 Effect of railpad stiffness and damping 125
4.5.3 Effect of ballast stiffness and damping

126
4.5.4 Effect of Sleeper mass 133
4.5.5 Effect of Ballast mass 133
4.5.6 Effect of sleeper spacing

135
4.5.7 Effect of bending stiffness of rail

136
4.6 SUMMARY 138
CHAPTER 5

CONCLUSIONS AND RECOMMENDATIONS

5.1 GENERAL

140
5.2 HIGHLIGHTS OF THE PRESENT WORK

141
5.3 CONCLUSIONS 142
viii
5.4 RECOMMENDATIONS FOR FUTURE WORK 144
REFERENCES 146










































ix
LIST OF FIGURES


Figure 1.1 The structure of the vehicle-track interaction models
Figure 1.2 An ideal chord type flat
Figure 1.3 A haversine type flat
Figure 1.4 A non-periodic OOR of wheel
Figure 1.5 Basic compositions of railway vehicle-track system
Figure 1.6 A single DOF one-dimensional model of the vehicle
Figure 1.7 A three-DOF one-dimensional vehicle model
Figure 1.8 A five-DOF pitch-plane vehicle model
Figure 1.9 A typical roll-plane vehicle model with several DOF
Figure 1.10 Two-dimensional 10-DOF pitch-plane vehicle model
Figure 1.11 A three-dimensional 10-DOF vehicle model
Figure 1.12 A double-beam rail model
Figure 1.13 Three-parameter pad models
Figure 1.14 Sleeper modeled as rigid mass resting on elastic ballast
Figure 1.15 A detailed model of ballast considering the stiffness and damping in
shear

Figure 1.16 A lumped-parameter 3-layer track model
Figure 1.17 A single layer track model with continuous support
Figure 1.18 A double layer track model with continuous support
Figure 1.19 Rail beam on discrete supports (only sleeper mass is included)
Figure 1.20 A three-layer model of track system
Figure 2.1 A three-piece freight car truck
x
Figure 2.2 A 5-DOF pitch-plane vehicle model
Figure 2.3 Various layers of the track structure
Figure 2.4 A three-layer railway track system model
Figure 2.5 Wheel/Rail contact model used in the present study
Figure 2.6 A wheel with an idealized flat
Figure 2.7 A wheel with haversine type flat
Figure 3.1 Comparison of wheel-rail impact force response of the present model
with that reported by Zhai et al. [3]

Figure 3.2 Time-history of impact force response predicted by the current model: (a)
single impact; and (b) three-consecutive impacts (v = 27 km/h;
f
L = 52.8
mm;
f
D = 1 mm)

Figure 3.3 Variations in the measured dynamic contact force due to a wheel flat,
reported by Zhai et al. [3]

Figure 3.4 Time history of rear wheel-rail impact force due to a flat on the rear
wheel: (a) single cycle; and (b) three cycles.

Figure 3.5 Time history of the flat free front wheel-rail impact force in the presence
of a rear- wheel flat at v = 70km/h: (a) single cycle; and (b) three cycles

Figure 3.6 Time histories of wheel-rail impact forces developed at front and rear
wheels due to a flat in the front wheel (v = 70 km/h;
f
L = 52 mm;
f
D =
0.4 mm): (a) front wheel-rail impact force; (b) rear wheel-rail impact
force.

Figure 3.7 Time history of front and rear wheel-rail contact force
Figure 3.8 Variations in the bearing force response due to a rear-wheel flat as a
function of static wheel load.

Figure 3.9 Variations in railpad force due to a rear wheel flat as a function of static
wheel load

Figure 3.10 Variations in ballast force due to a rear wheel flat as a function of static
wheel load

xi
Figure 3.11 Time histories of vertical displacements of the rear wheel and rail in the
presence of a flat

Figure 3.12 Time histories of vertical displacements of front wheel (no flat) and rail
at the wheel-rail contact point

Figure 3.13 Comparison of wheel and rail displacement responses of the present
model with those reported by Sun et al. [73]: (a) present model; and (b)
reported results (v = 70 km/h;
f
L = 40 mm;
f
D = 0.35 mm)

Figure 3.14 Displacement responses evaluated at (a) a point on the rail; (b) sleeper;
and (c) ballast beneath the rail point in the presence of a rear wheel flat

Figure 3.15 Time history of car body vertical motion in the presence of a rear wheel
flat (
f
L = 52 mm;
f
D = 0.4 mm)

Figure 3.16 Variation in the bounce and pitch responses of the bogie in the presence
of a rear wheel flat: (a) bounce motion; (b) pitch motion

Figure 4.1
Variations in radius of a wheel with single flat (
f
L = 52 mm
and
f
D = 0.4 mm) as a function of angular position of the contact

Figure 4.2 Influence of size of a front wheel flat on the impact force responses at
the wheel-rail interface: (a) front wheel; and (b) rear wheel

Figure 4.3 Effect of speed on peak front and rear wheel impact loads due to a single
flat on the front wheel (
f
L = 52 mm and
f
D = 0.4 mm)

Figure 4.4 Effect of speed on the peak displacement of the rail at the front wheel-rail
contact point (
f
L = 52 mm and
f
D = 0.4 mm)

Figure 4.5
Variations in radius of a wheel with two same size flats (
f
L = 52 mm
and
f
D = 0.4 mm), which are 90
0
apart

Figure 4.6 Time response of rear wheel-rail impact force with two flats at 45
0
phase
angle

Figure 4.7 Time response of rear wheel-rail impact force with two flats at 90
0
phase
angle

Figure 4.8 Time response of rear wheel-rail impact force with two flats at 135
0

phase angle
xii
Figure 4.9 Time response of rear wheel-rail impact force with two flats at 180
0

phase angle

Figure 4.10 Time history of front and rear wheel impact force responses
Due to a single flat on both wheels in phase
Figure 4.11 Front and rear wheel impact force responses due to a single flat on both
wheels at 45
0
out-of-phase (rear wheel flat ahead 45
0
)

Figure 4.12 Front and rear wheel impact force responses due to a single flat on both
wheels at 90
0
out-of-phase (rear wheel flat ahead 90
0
)

Figure 4.13 Front and rear wheel impact force responses due to a single flat on both
wheels at 135
0
out-of-phase (rear wheel flat ahead 135
0
)

Figure 4.14 Front and rear wheel impact force responses due to a single flat on both
wheels at 180
0
out-of-phase (rear wheel flat ahead 180
0
)

Figure 4.15 Effect of speed on wheel-rail impact load with single and multiple wheels
flats

Figure 4.16 Effect of flat length on W/R impact force for a constant flat depth of
0.4mm

Figure 4.17 Effect of flat length on wheel-rail impact force
Figure 4.18 Effect of flat depth on W/R impact force with a constant flat length of 52
mm

Figure 4.19 Effect of speed on wheel-rail impact force with three loads
Figure 4.20 Effect of static wheel load on peak wheel-rail impact force
Figure 4.21 Effect of unsprung mass on W/R impact load
Figure 4.22 Effect of primary suspension stiffness on peak wheel-rail impact force
Figure 4.23 Effect of primary suspension stiffness on peak bearing force
Figure 4.24 Effect of rail mass on wheel-rail impact force
Figure 4.25 Effect of railpad stiffness and damping on wheel-rail impact load
Figure 4.26 Effect of railpad stiffness and damping on peak pad force
xiii
Figure 4.27 Effect of rail pad stiffness on peak rail displacement
Figure 4.28 Effect of ballast stiffness and damping on peak wheel-rail impact force
Figure 4.29 Effect of ballast stiffness on peak rail displacement
Figure 4.30 Effect of ballast stiffness on peak ballast displacement
Figure 4.31 Effect of ballast stiffness and damping on peak ballast force
Figure 4.32 Effect of sleeper mass on peak wheel-rail impact force
Figure 4.33 Effect of sleeper mass on peak pad force
Figure 4.34 Effect of ballast mass on peak wheel-rail impact force
Figure 4.35 Effect of sleeper spacing on peak wheel-rail impact force
Figure 4.36 Effect of sleeper spacing on peak rail displacement
Figure 4.37 Effect of bending stiffness on peak wheel-rail impact load
Figure 4.38 Effect of bending stiffness on peak rail displacement











xiv

LIST OF TABLES

Table 3.1 Parameters used for examining validity of the model with single wheel flat

Table 3.2 Nominal simulation parameters

Table 4.1 Nominal simulation parameters used for parametric study





































xv


NOMENCLATURE


SYMBOL DESCRIPTION
c
M
Car body mass (kg)
W
Static wheel load (N)

t
M
Bogie mass (kg)
w
M
Wheel mass (kg)
t
J
Bogie mass moment inertia (kg-m
2
)
1 s
K
Primary suspension stiffness (N/m)
1 s
C
Primary suspension damping (N-s/m)
2 s
K
Secondary suspension stiffness (N/m)
2 s
C
f
l
r
l
Secondary suspension damping (N-s/m)
Distance between the front wheel and mass center of bogie (m)
Distance between the rear wheel and mass center of bogie (m)
t
l
Wheelset distance (m)
R Wheel radius (m)
f
L
Flat length (mm)
f
D
Flat depth (mm)
H
C
Non-linear Hertzian spring constant (N/m
3/2
)

r
m
Rail mass per unit length (kg/m)
E
Elastic modulus of rail (N/ m
2
)

I
Rail second moment of area (m
4
)

EI Rail bending stiffness (N-m
2
)
s
M
Sleeper mass (kg)
b
M

Ballast mass (kg)
p
K
Railpad stiffness (N/m)
b
K
Ballast stiffness (N/m)
xvi
w
K Ballast shear stiffness (N/m)
f
K
Subgrade stiffness (N/m)
p
C
Railpad damping (N-s/m)
w
C Ballast shear damping (N-s/m)
f
C
Subgrade damping (N-s/m)
l Length of the rail (m)
s
l
Sleeper distance (m)
( )
c
w t
Car body displacement in vertical direction (m)
( )
t
w t
Vehicle bogie displacement in vertical direction (m)
( )
t
t
Pitch rotation of the car bogie (rad)
1
( )
w
w t
Front wheel vertical displacement (m)
2
( )
w
w t
Rear wheel vertical displacement (m)
( )
c
w t
Car body velocity in vertical direction (m/s)
( )
t
w t
Vehicle bogie velocity in vertical direction (m/s)
( )
t
t
Pitch velocity of the car bogie (rad/s)
1
( )
w
w t
Velocity of front wheel in vertical direction (m/s)
2
( )
w
w t
Velocity of rear wheel in vertical direction (m/s)
( )
c
w t
Car body acceleration in vertical direction (m/s
2
)
( )
t
w t
Vehicle bogie acceleration in vertical direction (m/s
2
)
( )
t
t
Pitch acceleration of the car bogie (rad/s
2
)
1
( )
w
w t
Acceleration of front wheel in vertical direction (m/s
2
)
2
( )
w
w t
Acceleration of rear wheel in vertical direction (m/s
2
)
( )
'
j
P t
Wheel/rail contact force (N) (j =1-2)
1
( ) P t
Front wheel-rail contact force (N)
2
( ) P t
Rear wheel-rail contact force (N)
( ) r t Wheel flat profile function
xvii
( , )
r
w x t
Vertical displacement of the rail (m)
( )
si
w t Vertical displacement of the sleeper (m)( i = 1, 2, 3,N )
( )
bi
w t Vertical displacement of the ballast (m) ( i = 1, 2, 3,N )
( , )
r
w x t
Vertical velocity of the rail (m/s)
( )
si
w t Vertical velocity of the sleeper (m/s) ( i = 1, 2, 3,N )
( )
bi
w t Vertical velocity of the ballast (m/s) ( i = 1, 2, 3,N )
( , )
r
w x t
Vertical acceleration of the rail (m/s
2
)
( )
si
w t Vertical acceleration of the sleeper (m/s
2
) ( i = 1, 2, 3,N )
( )
bi
w t Vertical acceleration of the ballast (m/s
2
) ( i = 1, 2, 3,N )
N Number of the sleepers/ballasts
( )
rsi
F t Rail/sleeper contact force ( i = 1, 2, 3, N )
( )
sbi
F t Sleeper/ballast contact force ( i = 1, 2, 3, N )
i
x Position of the sleeper( i = 1, 2, 3,N )
j Number of wheels considered in the vehicle model (j=1-2)
Gj
x
Position of the wheel (j=1-2)
( ) Y x Mode shape function
K Total number of modes of the rail
k Number of rail mode corresponding to the sleeper position ( k =1, K )
k
Y k th rail mode shape
z A Wheel-rail overlap in vertical direction
( )
k
q t k th mode displacement of rail
( )
k
q t k th mode velocity of rail
( )
k
q t k th mode acceleration of rail
wr
G
Shear modulus of rail
wr
u
Poissons ratio
w

Wheel profile radius (m)
xviii
t
R
Rail profile radius (m)
e Natural frequency of the rail beam


1
CHAPTER 1
INTRODUCTION AND LITERATURE REVIEW

1.1 INTRODUCTION

Railway is the economical, environment friendly, safe and efficient transportation
mode, and it continues to play a very important role in the world commerce and
development. For about 200 years, railroad has been the most popular mode for
transportations of both the passengers and the freight. While much has been achieved, the
European Commission and American Association of Railway have expressed many
concerns that the rail sector has not shown the same progress towards a quieter operation
that the rival modes have achieved. Several research groups, and commissions have
contributed towards better solutions in order to improve safety and reliability, especially
in the freight sector, where such improvements have been least pronounced. A survey of
global market and investment trends has shown that annual investment in rail vehicles
has grown from about $US 18 billion in 1993 to $US 25 billion in 2000 and about $US
31 billion by 2004 [41]. North America accounts for about 20% of the world rail vehicle
market predicts strong growth as cities continue to expand their rail transit systems [41].
In recent years, the operational safety has drawn considerable attention in order to
reduce operating cost as well as unscheduled service interruption that may arise from the
wheel and rail defects. These defects are mainly wheel flats, shelling, spalling,
corrugation of rails and wheels attributed to material fatigue, blocked brakes,
manufacturing flaws, etc. Such defects trigger severe repeated high frequency impact
forces in wheel-rail interface, which may cause failure of various components and lead to
derailments in extreme cases. Formation mechanism of these defects involves complex
2
interactions between the wheel profile and the rail surface, thermo elastic instability and
development of unusual forces in relation to the wheel/rail adhesion [82, 94]. A sudden
change in wheel-rail contact force due to these defects can lead to an accelerated
deterioration of the vehicle and track structure components, such as wheelsets, bearings,
rails, and sleepers. It is estimated that the railway industry in North America is currently
spending nearly $90 millions annually to replace 125,000 wheels due to wheel defects
[37]. Only for one type of wheel defect, namely spalling, North American railroads spend
$15 million annually to replace spalled wheels [64]. The associated unscheduled service
interruption costs are far more significant.
Among all the various wheel defects, a wheel flat is known to be quite common
[3, 38, 43]. This type of defect is generally caused by unintentional sliding of the wheel
on the rail, and occurs mainly when the braking force is too high in relation to the
available wheel/rail friction. The response of the railway vehicle system with a wheel flat
is strongly dependent on the vertical dynamics of wheel-rail interactions. The nonlinear
Hertzian point contact theory is most commonly used in the analysis of impact forces
attributed to the wheel-rail contact. The complexity of the analysis and severity of the
contact forces, however, increases manyfolds when wheel flats are present. A detailed
study on dynamic wheel-rail interactions is thus required in order to characterize the
contact impact forces between the wheel and the rail, which will permit timely detection
and removal of a defective wheel. The demands for high axle loads and operating speeds
in order to enhance the operation efficiency could further amplify the magnitudes of
wheel flat induced impact forces.
3
In this study, an analytical model of the coupled vehicle-track system is
developed by integrating the pitch plane model of the vehicle-track system comprising a
two-dimensional model of the 3-layer flexible track system and a 5-DOF of the vehicle
coupled through a nonlinear rolling contact. Generalized coordinate method is employed
to analyze the vehicle-track interactions. The validity of the coupled vehicle/track system
is demonstrated by comparing the simulation results with the reported experimental data
and analytical solutions. The validated model is utilized to investigate the characteristics
of impact forces in the presence of single or multiple wheel flats. Finally, a detailed
parametric study on the effect of different vehicle-track design and operational
parameters on the wheel-rail impact loads due to wheel flats is presented.
1.2 LITERATURE REVIEW

A study of dynamic wheel-rail interactions in the presence of wheel defects
involves thorough understanding of various contributing factors. These include the
dynamic motion of the vehicle components, deflection of the multiple-layered continuous
track structure, nature of wheel defects, dynamic interactions of the moving wheels with
the flexible track, etc. Relevant reported studies are thus reviewed and briefly
summarized in this chapter in an attempt to build essential background and scope of this
dissertation research. The coupled vehicle-track system dynamics in the presence of
various types of wheel defects have been extensively investigated in the last few decades
to identify the sources of wheel-rail impact loads, and to define the threshold values of
impact loads for timely removal of defective wheels. These defects come into existence
either from the wheel or the rail, or from both. Owing to the complexities associated with
measurements of impact loads caused by a moving load, only a few studies have
4
performed field measurements of impact loads caused by the wheel and rail defects under
limited operating conditions [5, 31, 35, 47, 90]. A vast number of studies have relied on
development of effective simulation models for predicting the dynamic impact loads [3,
17, 18, 32, 45, 53, 92, 95, 110]. An array of vehicle and track system models have been
developed in the past few decades for analyses of wheel-rail vertical interactions for
studies on the ride quality, safety, wear, curving, etc. Knothe and Grassie [42] have
presented a comprehensive review of advancements in the field of railway vehicle
dynamics including the development in dynamic models of study the vehicle-track
interactions due to wheel defects in high frequency range. A state-of-the art review of
rock and roll dynamics of railway vehicles has been reported by Sankar and Samaha [68].
Dahlberg [4, 55] and Taheri et al. [69] have reported review of studies on dynamic
interactions between the rail and track, and vehicle-guideway interactions, respectively.
The studies on dynamic interactions of vehicle and tracks generally involve a vehicle
model, wheel-rail contact model, track model, and irregularities of the wheel or the track.
These component models are integrated, as shown in Fig. 1.1, and are briefly discussed in
the following subsections.



Track Model

Vehicle Model

Wheel Defect Model

W/R Contact Model

Wheel Displacement Input

Contact Force Output

Rail Displacement Input

Contact Force Output

Output Results
Fig. 1.1: The structure of the vehicle-track interaction models
system

5

In order to facilitate the analysis, it is necessary to study in details the modeling of
vehicle and track system combined with their individual components and wheel defects.
It is also required to survey the relevant reported studies to be acquainted with the nature
of effects of these factors on the wheel-rail impact load. The relevant literatures are thus
reviewed and discussed in the following sub-sections to build up the essential background
and to formulate the scope of this dissertation research. A detail review and discussion of
each of the components of the above Fig. 1.1 is also presented in the following sub
sections.

1.2.1 Wheel defects
The term wheel defects is used to describe various types of wheel imperfections
that may develop either during operation or in manufacturing, and wheel reprofiling
stages. There are different types of defects present in the railway wheels, which range
from short wavelength to large wavelength defects. These include flats, shelling, spalling,
corrugation, eccentricity, etc. Nielsen et al. [82] described the formation of these defects
and their effects on wheel-rail impact loads through a comprehensive review on
published studies. It was concluded that timely detection and replacement of a defective
wheel offers large economic returns by reducing the maintenance and repair cost. Barke
and Chiu [38] also presented a comprehensive review on published studies on the effects
of wheel defects and track design parameters on the dynamic forces transmitted to the
track and vehicle components. The reported studies have facilitated the developments in
wheel removal criteria and detection of defective wheels in order to prevent the failure or
6
service interruptions. Despite these developments, continued efforts are being made to
develop more reliable wheel defect detection algorithms and systems, and wheel removal
criteria.
Wheel flat is the most common type of wheel defect encountered by railway
industry, which develops due to sliding of the wheel on the rail under braking, when the
barking force is larger than the available wheel/rail friction or when the brake system
poorly adjusted or defective [8, 73]. During sliding, the part of the wheel tread could be
removed causing a flat surface to form on the wheel profile. This surface irregularity
causes impact loads on the rail and track structure as the wheel rolls. This impact load
induces high frequency vibration of the track and the vehicle components [89]. It has
been suggested that excessive magnitude of the impact load may even shear the rail [30].
A number of studies have attributed the wheel sliding to poorly adjusted, frozen, or
locked wheels or excessive braking forces in relation to the wheel/rail adhesion [7, 90].
Contaminations on the rail surface, such as leaves, grease, frost, and snow could also
aggravate the sliding at the wheel-rail contact. In [90], it is reported that the area of actual
wheel/rail contact changes during the formation of a flat and thus the shape of the
railhead also has a great influence on the growth of the wheel flat.
With continuing increases in the in the axle loads and operating speed, the wheel
flats are becoming increasingly common. The high magnitude of impacts due to wheel
flats, whether single or multiple, not only induce high magnitude impact force and stress
on vehicle components but also to the rails and the sleepers [3, 53]. Wheel flats thus
affect track maintenance and the reliability of the vehicles rolling elements [88]. In
7
addition to safety and economic considerations, these defects reduce passenger comfort
and significantly increase the intensity of noise [43].
Various railroad organizations have set for the criteria for removal of wheels with
flats. The proposed threshold values of the wheel flats are primarily based on flat size and
the impact load produced by the flat. Different organizations, however, have defined
different threshold values for the wheel defects. The AAR [91] criteria for removal of
wheel from service imposed that a railway wheel with 50.8 mm long single flat or
38.1mm long two adjoining flats cannot be placed or continue to be in service. The AAR
also states the threshold value of wheel-rail impact loads due to a single flat. According
to AAR rule [101], a wheel should be replaced if the peak impact forces due to single flat
approaches in the 222.41 to 266.89 kN range.
According to Swedish Railway, the condemning limit for a wheel flat is based on
a flat length of 40 mm and flat depth of 0.35 mm [10]. Transport Canada safety
regulations [108] require that a railway company may not continue a car in service if
wheel has a slid flat spot that is more than 63.50 mm in length or two adjoining flat spots
each of which is more than 50.80 mm. According to UK Rail safety and standard board
[109], freight vehicle with axle load equal to or over 17.5 tones a wheel with flat length
exceeding 70 mm must be taken out of service. These removal criteria for defective
wheels are mostly based on the magnitude of the impact force induced by the flat, while
considerable discrepancies among the different criteria exist.
The assessment of potential damages caused by a wheel flats and development of
reliable criterion necessitate development of effective impact load prediction tools. A
wide range of mathematical descriptions have thus evolved to characterize the geometry
8
of wheel flats in order to investigate the impact loads [3, 5, 7, 12, 14, 30, 37]. On the
basis of the reported flat geometries, the wheel flats have been classified as chord type
flat, cosine type flat and combined flat. A chord type flat model considers the flat as a
newly formed fresh flat with relatively sharp edges, where the interacting force between
the wheel and the rail is estimated based on the assumption that one of two edges of the
flat is always in contact with the rail.
The chord flat models have been used widely in various studies on wheel-rail
impact load, rail acceleration, and noise [7, 14, 30]. A chord type wheel flat model is
shown in Fig. 1.2. The geometry of this type of flat is described by its length (
f
L ) and
depth (
f
D )



Mathematically, the wheel profile with chord type flat can be expressed as [14]

(1 )
0
R cos
r
u
=
`
)

0 u
u
< <
>
(1.1)

u

f
D
o
f
L
R

x
( ) t r
A
Fig. 1.2: An ideal chord type flat
[14]

9
Where, is the angle subtended by the flat at the wheel centre, R is the radius of the
wheel, r is the variations in the wheel radius due to the flat, u is the angle of an arbitrary
point within the flat zone or the cord, and expressed as:

1
1
sin ( / )
sin [( ) / ]
f
x R
L x R
u

=
`

)

0 / 2
/ 2
f
f f
x L
L x L
s <
s <
(1.2)

Where,
f
L is the length of the flat, x is the longitudinal coordinate of an arbitrary point
within the flat curve.
The edges of the chord type flat tend to become more rounded as the wheel
continues to be service due to wear and/or deformation under repeated impact loads.
Subsequently, the chord type flat can be modeled as a cosine flat, which is also known as
a haversine flat or rounded flat. Haversine flat model is widely used for analysis of
dynamic behavior of rail vehicles and tracks together with the wheel-rail impact load due
to flat as used in [3, 5, 12, 37]. A haversine flat model with its mathematical expression is
shown in Fig. 1.3. A haversine flat is expressed as [37]:
1
( ) [1 cos(2 / )]
2
f f
r t D x L t = (1.3)
Where
f
D is the flat depth that may be calculated as:
2
/(16 )
f f
D L R = (1.4)
10


Unlike the chord type flat, a haversine flat yield more uniform and continuous
contact between the wheel and the rail. The longitudinal position of wheel-rail contact
center is thus generally considered as the projection of the wheel center [7, 13, 37]. The
impact loads predicted by the haversine flat have generally shown reasonably good
agreements with the measured data [37].
It has been suggested that a wheel flat may not be truly represented by a chord or
the haversine function. A combined wheel flat model was introduced by Ishida and Ban
[30], where the characteristics of both chord type and cosine type flats are combined
together to analyze the dynamic behavior of the wheel and the rail. The model results,
however, did not show substantial advantage in enhancing the dynamic wheel-rail impact
load prediction ability when compared to the haversine flat. A comparative study on rail
acceleration response due to three types of flats was also presented in this study. From the
results, it was concluded that a chord flat model is more sensitive and cosine type flat

f
D
o
R
x
f
L
( ) t r
Fig. 1.3: A haversine type flat
11
model is less sensitive to measured rail acceleration, while the combined flat model yield
better agreement with the measured data.
The operating speeds of railway vehicles have continued to increase to enhance
the operational efficiency. The potential effects of high-speed operation on the wheel-rail
interactions caused by the wheel flats have thus been emphasized in many studies [5, 14,
23, 35, 37]. The majority of these studies have investigated the wheel-rail impact loads
due to single wheel flat using a non-linear Hertzian contact spring [3, 5, 23, 35]. Dong
[37] and Hou et al. [14] used finite element method to study the increase in impact force
due to a wheel flat with increasing speed. Thompson and Wu [8] also performed non-
linear analysis of wheel-rail impact loads under different operating speeds. These studies
have invariably concluded that the magnitudes of the contact force increase with
increasing speed. These studies have also identified the ranges of speed, where the
variations in wheel-rail impact load remain relatively insensitive to operating speed [3,
37]. Furthermore, the magnitudes of the wheelrail impact forces have a small peak in the
low speed range (3040 km/h), and after that speed the magnitudes increase with increase
in speed. A few studies have shown that the maximum contact force can decrease at
higher speeds, which are beyond the practical speed limit.
Sun et al. [10, 73] compared contact force due to wheel flats derived from the
model and measured data reported by Fermer and Nielsen [31]. The study concluded that
the magnitude and frequency of the impact forces derived from the model agree
reasonably with the measured data. The magnitude of wheel-rail impact forces derived
from both models and measured data were approximately 50 percent greater than the
static wheel load.
12
Experimental and theoretical studies on impact loads due to wheel flats have
been described by Johansson and Nielsen [5, 31], and Newton and Clark [35]. These
studies have evolved into a number of significant findings, namely: (i) wheel defects
often lead to large impact loads but they are not always easily detected by visual
inspection of the wheel; (ii) for a constant velocity, the impact load increases with
increase in the flat length; and (iii) a nonlinear track model yields more accurate
prediction of the wheel-rail impact force due to wheel defects than the linear track model.
A freshly formed flat is known to grow as the wheel continues in service. Only minimal
efforts, however, have been made to study the rate of growth of a wheel flat and the
major contributing factors. Jergeus et al. [90] performed experiments to study the flat
growth and concluded that the rate of growth is very high at the beginning. It is thus
essential to take the wheelset out of service as quickly as possible when a wheel flat is
observed.
In general, wheels carrying any type of defects are referred to as out-of-round
(OOR) wheels. Nevertheless, many of the individual studies concerned with wheel-rail
impact load have considered a polygon shape of the wheel only as out-of-roundness, as
shown in Fig. 1.4 [5, 21, 89, 94]. Non-uniformity of the wheel profile is very common in
practice. When a perfectly round wheel is in use, the irregularities may develop in the
wheel trade to cause out-of-roundness (OOR) condition. The OOR may be characterized
by both periodic and non-periodic defect. The clamping of wheel during reprofiling may
cause periodic OOR, while a non-periodic OOR may be caused by unbalance in the
wheelset or by inhomogeneous material properties of the wheel [82]. Both of these types
OOR are usually found in disc-braked wheelsets. The vibration caused by OOR wheels
13
are transmitted from the wheels via the bogies to the passenger compartments, which
often perceived as annoying coupled with humming noise. In addition to the deterioration
of comfort, it has been reported that maintenance costs associated with OOR wheels tend
to be considerable [21].
A comprehensive review of studies involving classifications of OOR, their
formations, wheel-rail impact loads attributed to OOR has been presented by Nielsen and
Johansson [82]. The study emphasized the need for development of improved wheel
removal criterion that will not only be based upon the geometry of the OOR but also on
the computed and/or measured maximum wheel/rail impact loads. The study further
concluded that a complete model of the vehicle-track system is required to study the
long-term wear behavior due to OOR, and more investigations on the formation and
control of OOR are needed.



Johansson and Nielsen [5], and Johansson and Andersson [94] have reported that
periodic wheel OOR leads to increase in ride vibration levels, especially at certain
speeds. The magnitude of vertical wheelrail contact force increases due to the wheel
Fig. 1.4: A non-periodic OOR of wheel [94]
14
OOR, which contributes to reduce vehicle-track system fatigue life. Barke and Chiu [38]
have recently reported a review of studies on the effects of OOR on the wheel-rail impact
loads. The study concluded that a deeper understanding of the effect of impact loads on
fatigue lives of vehicle-track components is required using the comprehensive vehicle-
track models.
Apart from the wheel flat defects, a number of studies have investigated the
impact loads caused by various other types of surface defects and their propagation.
These include wheel spalling, wheel shelling, and wheel and rail corrugation that are
characterized by wavelength. Railway wheel spalling is assumed to occur as the result of
fine thermal cracks joining to produce the loss of a small piece of tread material. The
thermal cracks are developed owing to heating and rapid cooling of the wheel tread
during and after block braking. Stone [65] have presented an interpretive review on wheel
spalling and shelling. Some highlights of this review can be summarized as: (i) spalling
and shelling are dominated by rolling contact resistance; (ii) some factors like loss of
material strength, quasi-static thermal stresses must be considered in model to analyze
wheel spalling.
Shelling of wheel is another type of wheel defect that is assumed as the result of
rolling contact fatigue. It is manifested by loss of flakes of material from the wheel tread.
Eric et al. [96] have shown diagrammatically the relationship between shelling formation
and the related various operating and environmental parameters. A study on formation of
railway wheel shelling due to thermal effect is conducted by Moyar and Stone [66]. The
study concluded that periodic rail chill in the case of hot-braked treads has a strong effect
15
on shelling and it can be minimized by uniformly distributed braking force, and
maintaining tread temperatures as low as possible.
Corrugation is characterized by an almost regular sequence of shiny peaks and
dark troughs generally spaced about 30-50 mm apart. A recent review on corrugation
characteristics treatments is presented by Sato et al. [67]. This study concluded that in
order to understand the formation of corrugation, a systematic study of the corrugation
phenomena including experimental as well as theoretical investigations is essential.
The effect of rail corrugation on wheel-rail dynamic impact load is investigated
by Jin et al. [17, 44, 85], Nielsen and Igeland [23], and Sun and Simson [74].
Experimental study of corrugation formation on the rail top surface is carried out by Suda
et al. [93] in order to investigate the entire process of corrugation and the phase relation
between corrugation profile, contact load, and slip. Several analytical models have been
developed to study the corrugation formation process as reported in [32, 45, 92, 95]. All
these studies have shown that once the corrugation has formed, it will lead to an
accelerated deterioration of the track structure and the vehicle and to the generation of
high frequency noise.
1.2.2 Vehicle models
The main components of a rail vehicle system are car body, bogie/side frame,
wheel, primary suspension, and secondary suspension etc. The car body rests on two
bogies each containing two wheelsets. The spring and damping elements connecting the
wheelset bearings and the bogie frame are referred to as the primary suspension. The
secondary suspension connects the bogie frame to the car body, as illustrated in Fig. 1.5.
16



The type of vehicle model employed in a dynamic study mainly depends on the
objective of the formulation. According to Knothe et al. [42], in the low frequency range
the major issues in the vehicle system concern the curving, stability, and ride quality
performance. In the lower frequency range, up to 50 Hz, the track essentially behaves as a
relatively stiff spring and its effect on the vehicles behavior is small, especially in the
vertical direction [37]. The vehicle can thus be modeled as a lumped-parameter system.
A wide range of linear and nonlinear lumped-parameter models have been developed for
studies on lateral stability, curving, and ride comfort [13, 21, 35, 62, 63]. These includes
pitch-plane, roll-plane and three-dimensional models of the vehicle with several layers of
track.
When the vertical dynamic forces due to wheel and rail irregularities such as
wheel flats and rail joints are of concern, the wheel-rail interactions may be investigated
Fig. 1.5: Basic compositions of railway vehicle-track system [37]
17
using a simple model of an effective wheel mass with a constant force acting on the
wheel. Such a simplified model is illustrated in Fig. 1.6, and has been employed in
numerous studies, where the contribution due to vehicle dynamics are assumed negligible
[5, 7, 8, 15, 22, 23, 31, 35, 39]. This simplifying assumption is justified by the very high
frequency components of the wheel-rail interactions, where the vehicle wheels remain the
most active component. Such model, however, do not permit the analyses of effects of
one wheels flat on the forces imparted by an adjacent wheel. Furthermore, analysis of
dynamic forces imposed on the bearings would require appropriate considerations of
dynamic motions of the bogie (with or without car body) with proper suspension
parameters [2, 6, 17, 19, 23, 37]. If the interaction between the two wheels on different
wheelsets is of concern, a half car model with several degrees-of-freedom would be most
appropriate. Such a model has been employed by Zhai et al. [3, 53], Schwab et al. [72],
Sun et al. [73], and Cai et al. [86] for analyses of wheel-rail impact loads under wheel
flat. The dynamic behaviors of railway wheelsets in the medium frequency range in both
vertical and lateral planes have also been investigated in [10, 12, 13, 14, 74]. A
comprehensive three-dimensional model, however, would be essential for investigating
both pitch and roll effect of the bogie and car body.
The analyses of very high frequency components of deflections and forces up to
20 kHz, for noise emissions studies, elastic wheel, and/or wheelsets models need to be
considered. Grassie et al. [36, 57] have investigated the behavior of railway wheelset
under high frequency excitation in vertical plane. A number of simulation models for an
elastic wheel have been developed using different approaches are developed and applied
in studies on vehicle-track interactions. Szolc [18] developed an elastic wheelset model to
18
identify the sources of polygonalization of wheels. Similar models have also been used
by Clause and Schiehlen [78] to study vibration behavior of railway bogie, and by
Schneider et al. [79] to study wheel noise generation. Finite element methods have been
widely used to study the noise generation by elastic wheelsets with OOR defects [75, 76,
77, 78]. The elastic wheelset models, however, are meaningful only for stress, fatigue,
and failure analysis, and may not be significant for analyses of wheel-rail interaction
forces.
Apart from the vehicle components modeling, vehicle system models can be
primarily categorized into three groups. These are one-dimensional, two-dimensional,
and three-dimensional models. One-dimensional model is the simplest model that
considered a single wheel with static force representing the static load due to the car and
bogie, as shown in Fig. 1.6. The contact between the wheel and rail is maintained by
either linear or non-linear spring. Such a simple model has been widely used in many
studies on wheel-rail interaction forces due to wheel and rail defects [1, 5, 15, 20, 21, 22,
28]. Such a model would also be adequate for high frequency vibration analysis when the
interaction between the wheel and rail with irregularities is of concern. However, this
type of model is insufficient for analyzes of effects of impact forces due to wheel and rail
defects on various vehicle components. Moreover, the contributions due to pitch and roll
motions of the vehicle, and the presence of multiple defects in different wheelsets cannot
be evaluated by this model.
One-dimensional vehicle model with two- or three- DOF involving motions of
either the bogie or car or both are also quite common in the analysis of vehicle-track
system [2, 17, 30, 35]. A three-DOF one-dimensional model, as shown in Fig. 1.7,
19
generally considers car, bogie, wheel, and primary and secondary suspensions. Such
models can effectively predict the dynamic forces between the bogie and wheel, i.e.
bearing force. The effect of car and bogie pitch and roll dynamic responses, and that of
one wheel flat to other wheels, however, cannot be evaluated.



Alternatively, two-dimensional models with several degrees-of-freedom have also been
developed for analyses of vehicle-track interactions. A two-dimensional model can be
either a pitch-plane or a roll-plane model depending on the type of analysis. A simplest
pitch-plane vehicle model consists of two wheels coupled to the sideframe through the
primary suspension and a static force acting at the center of the side frame due to car load
[6, 10, 23, 37, 86]. Some models also consider the car body coupled to the side frame
through secondary suspension, as shown in Fig. 1.8 [25, 27]. These models are sufficient
to study the vertical and pitch dynamic responses of the vehicle and dynamic coupling
between the two adjacent wheels in the presence of wheel and rail defects. Furthermore,
inclusion of the secondary suspension into the model allows investigation of dynamic
response of the car body. However, contributions due to roll dynamic response of the
W
H
C

Wheel
Rail

Vehicle Load
Hertzian contact spring
Fig. 1.6: A single DOF one-dimensional model of
the vehicle
20
wheelsets are ignored and effect of one wheels irregularity to the other wheel within the
same wheelset could not be obtained.






Bogie
Car body
Secondary
suspension
Primary
Suspension
Wheel
Fig. 1.7: A three-DOF one-dimensional vehicle model [2, 35]
Fig. 1.8: A five-DOF pitch-plane vehicle model [37]
21
A two-dimensional vehicle model in the roll plane consists of a wheelset, side
frame and car body connected together through the primary and secondary suspension, as
shown in Fig. 1.9 [25, 26]. Such models are useful for analyses of dynamic vertical and
lateral response of the vehicle system. The effect of one-wheel defect to the adjacent
wheel within the same wheelset can be effectively investigated. However, contributions
due to pitch dynamics response of the car body and wheelsets and effects of one wheel
defect to the other wheels within the same bogie or the car could not be investigated.
A few studies have developed a pitch-plane vehicle model of half of the vehicle
including half the car body and two bogies and four wheelsets [3, 13, 17, 53]. This model
includes not only the pitch motion of the car body but also that of the bogie. Such models
exhibit 10 to 12 DOF, as shown in Fig. 1.10. Such a model can permit analyses of
dynamic interaction between the leading and trailing bogie and wheels. Furthermore,


Wheel
Bogie
Car body
Rail
Fig. 1.9: A typical roll-plane vehicle model with several DOF
22
impact force due to a wheel defect present in the front bogie wheels could also be
determined at the rear bogie wheels. Considering that the vehicle pitch mostly occurs at
low frequencies, a quarter-vehicle pitch-plane model would most likely be sufficient for
analyses of dynamic wheel-rail interactions in the presence of a wheel defects.
Comprehensive three dimensional vehicle models with relatively large number of
DOF have been increasingly used in recent years. The detailed representations of vehicle
components used in these models make them attractive and more realistic nature towards
the new researchers. A typical three-dimensional vehicle model incorporates half of the
car body, two bogie sideframes, and two wheelsets, while the total number of DOF may
range from 10 to 37. The primary and secondary suspensions are modeled by linear
spring-damper elements along the lateral and vertical directions. Such a model provides


all the advantages of roll, pitch plane models, and would be sufficient to investigate the
influences of coupled vertical, pitch, and lateral dynamics of the vehicle. Furthermore,
Fig. 1.10: Two-dimensional 10-DOF pitch-plane vehicle model [3]
23
the cross effects of the four wheels of a bogie and the leading and trailing wheelsets can
be effectively investigated. Fig. 1.11 illustrates a three-dimensional model of the vehicle
that can be run over a flexible track.


1.2.3 Modeling the track Components
The railway track system comprises a periodically supported rail of infinite
length. The track system is a combination of multiple layers of the rail, sleepers, ballasts,
and sub-ballasts. A number of simulation models of the track of varying number of layers
and complexity have been reported in literature. The treatment of each layer is described
below together with the modeling considerations.
Rail
The flexible rail forms the most important components in studies involving
analysis of wheel-rail interactions. The simplest way to model a rail is to consider it as a
lumped mass, spring, and viscous damper [37]. The model parameters are generally
derived from theory of beams on elastic foundations. Li and Selig [110] developed a
Fig. 1.11: A three-dimensional 10-DOF vehicle model [14]
24
lumped parameter rail model to study the wheel-rail impact load due to rail joints. The
study also presented a comparison between the lumped-parameter rail model and a finite
element model, and concluded that both models could provide consistent prediction of
peak magnitude of the impact force. The lumped-parameter rail model offers
considerable advantages, which include its simplicity for efficient computation and its
ability to deal with low frequency vehicle vibration. Furthermore, the non-linear
properties of the track components can be easily incorporated in the model. Such models,
however, cannot be considered sufficiently accurate for prediction of high frequency
wheel/ rail impacts, and contribution due to higher modes of the continuous elastic rail.
The rail is mostly modeled as a continuous beam, either as an Euler-Bernoulli
beam or a Timoshenko beam. The Euler beam rail model was used in earlier days for
static and stability analysis, which was considered to provide an accurate model for
representation of the rails response to vertical excitations at frequencies below 500 Hz
[42]. The Euler beam model of the rail has thus been used in many studies on dynamic
analysis of the vehicle-track system. Zhai et al. [3, 53], Jin et al. [13, 17], Morys [21], Cai
et al. [54] have employed Euler beam model to study the effects of rail dipped joints,
wheel flats, rail corrugation, and OOR defects under a moving force on the track. The
Euler beam model, however, assumes negligible shear deformation and rotational inertia
of the rail. The rail thus exhibits relatively higher stiffness, which may yields
overestimation of the dynamic force in the high frequency range. This model is thus not
suited for lateral dynamic analysis that involves the lateral flexibility of the rail web.
Alternatively, the rail has been widely modeled as a Timoshenko beam for predicting
dynamic loads due to the wheel and rail imperfections [5, 6, 7, 8, 9, 10, 12, 19, 22, 23, 28,
25
31, 37, 39]. Timoshenko beam model of the rail has been used to study the effect of
nonlinearity and railpad stiffness on the wheel-rail impact and wheel-rail noise
generation. It has been reported that Timoshenko beam model is adequate for frequencies
up to 2.5 kHz for vertical responses [42]. However, for lateral and torsional modes,
railhead and foot need to be modeled as independent Timoshenko beams interconnected
by continuous springs.
Wu and Thompson [46] have developed a double Timoshenko beam rail model
suitable for very high frequency analysis (up to 5 kHz). In this model, the rail is divided
into two sections: the upper part represents the head and web, while the lower part
represents the foot, as shown in Fig. 1.12. These two parts are connected by springs, such
that the two beams vibrate together as a single Timoshenko beam at low frequencies. A
relative motion between the two beams, however, occurs at high frequencies, which
represents the cross-sectional deformation between the railhead and the foot. This model
has shown good agreement with results obtained from a FEM model and measured data
in terms of frequency-wavenumber relation. The model, however, poses considerable
difficulties in identifying cross-sectional and coupling parameters.


Fig. 1.12: A double-beam rail model [46]
26

Railpad and fastener
Railpads, made of rubber, plastic or composite materials, are used to support the
rail, to protect sleepers from wear and damage, and provide electrical insulation of the
rails. The railpads and fasteners are usually modeled as a parallel combination of spring
and damping elements for analysis of vertical dynamics. A few studies have considered
structural damping with a constant loss factor [111]. The railpad is generally modeled as
an spring elements [9, 15, 16] or combined spring and damping elements to form either
continuous [2] or discrete [3, 13, 17, 21] rail support. The discrete models consider the
visco-elastic pad model at a point on the rail foot at the center of sleeper support to
investigate the wheel-rail interactions [3, 5, 6, 9, 19, 22, 23, 28, 31]. Such models have
also been used for analyses of noise emissions [7], and effect of rail corrugations [13, 17,
39, 74]. The deflection behavior of the rail pad at higher frequencies has been
experimentally measured by Thompson et al. [47, 50], and Fermer and Nielsen [31, 48].
The data was used to examine the validity of the analytical model developed by Fenander
[51]. Oscarsson [1] proposed a train-track interaction model to encompass non-linear
railpad stiffness to investigate its influence on rail corrugations.
Andersson and Oscarsson [49] have developed a state-dependent three-parameter
viscoelastic railpad model, as shown in Fig. 1.13, to study the dynamic behavior of pad
under high and low frequency excitations. The model provided higher pad stiffness at
higher frequencies. It was concluded that both low and high frequency behavior of
railpads could be accurately modeled with the inclusion of three parameters. The
influence of the state-dependent properties of the railpad on the wheel-rail contact force
27
was, however, observed to be relatively small, while the model parameter identification
was considered to be tedious [49].





Sleeper
The sleepers support the rail and transmit vertical, lateral, and longitudinal forces
from the rail to the ballast. A sleeper can be considered either as a transverse beam with
either uniform or variable cross section or as discrete mass on an elastic foundation
representing the ballast, as shown in Fig. 1.14. This type of simple sleeper model has


been most widely used in analysis of vehicle-track interactions in the presence of wheel
and rail defects [1, 6, 8, 13, 17, 22, 23, 28, 31, 39]. These studies have invariably
concluded that sleeper modeled as a rigid mass is adequate for prediction of dynamic
vehicle-track interaction force, while the bending stiffness is neglected that may yield an
overestimation of impact force at higher speeds.
Sleeper
Ballast

Stiffness
Damping
(a) (b)

Fig. 1.13: Three-parameter pad models [49]

Fig. 1.14: Sleeper modeled as rigid mass resting on elastic ballast

28
Alternatively, sleepers have been characterized by either an Euler-Bernoulli or
Rayleigh-Timoshenko beam model [56, 57]. These studies showed good agreements
between the model and measured responses of the rail and the sleeper at frequencies
below 700 Hz. Grassie and Cox [57, 58] have developed flexible sleepers as uniform
Timoshenko beam of prescribed flexural rigidity and mass per unit length, which
provided satisfactory correlation of computed sleeper strains and deflections with the
measured data acquired using a test train running over a sinusoidally corrugated rail. An
effective methodology for determining the sleeper parameters, however, has not yet been
formulated; the measured data have been used to identify a set of model coefficients [59].
A rigid mass representation of the sleeper has been considered adequate for
wheel-rail dynamic interactions. Knothe and Grassie [42] concluded that sleeper response
to contact forces at the railhead up to 1 kHz could be effectively predicted using the rigid
body sleeper model. Dahlberg [55] reported that a concrete sleeper could be adequately
modeled as a rigid mass under excitation below 100 Hz. the Euler-Bernoulli beam theory
would suffice under excitation up to 400 Hz, while Rayleigh-Timoshenko beam theory
should be used for accurate description of sleeper vibration at higher frequencies.
Ballast and subgrade
Ballasts are coarse stones placed under and around the sleepers to form a bed. It
limits sleeper movement by resisting vertical, transverse, and longitudinal forces
transmitted to the track. From a physical point of view, modeling of ballast materials and
its interactions with the sleepers is a very complicated task. The ballasts and subgrades,
however, do not greatly affect the wheel-rail contact, primarily due to their distant
placement from the rail. The specifications of various ballast materials have been
29
reviewed by Peplow et al. [60], while Jacobsson [61] presented a review of ballast
materials with particular emphasis on constitutive and mathematical modeling. The
ballast and subgrade are generally grouped together as the foundation for the track, and
modeled as parallel spring-damper element [2, 5, 6, 7, 8, 9, 19, 21, 22, 23, 28, 29, 31, 35,
37, 39]. Zhai et al. [3, 53] and Jin et al. [13, 17] have considered the ballast as lumped
masses below ties, which are interconnected by the springs and dashpots in shear, as
shown in Fig. 1.15. This model permits for analysis of distributed ballast deflections
under excitations at the wheel-rail interface. Such models, however, require a large
number of system parameters and could lead to greater uncertainty of the simulation
results.





Although coupling the ballast and subgrade together as a single mass is very
common in practice, some researchers have modeled them as two individual masses
either connected in shear [10, 12], or through vertical springs and dampers without any
shear coupling [81] to investigate the influence of different parameters on dynamic wheel
loads. This type of model provided better correlation with the measured response, while
Sleeper
Ballast

Damper

Spring
Fig. 1.15: A detailed model of ballast considering
the stiffness and damping in shear [3, 17]

30
the difficulties associated with identification of adequate model parameter values have
been acknowledged.

1.2.4 Track system models

The reported track system models for analysis of dynamic train-track interactions
can be grouped into three categories, namely: (i) lumped parameter models; (ii) rail beam
on continuous supports; and (iii) rail beam on discrete supports. Dahlberg [55] presented
a detailed description of different types of track system models with their relative merits.
The simplest lumped parameter track model can be described by a single effective rigid
mass supported on track bed by linear spring and damping elements. The simple track
model can also be extended to include two or more layers models of the sleepers and
ballasts, as shown in Fig. 1.16. Such a model has been used by Ahlbeck and Daniels [32]
to study the formation of rail corrugations and wheel/rail impact forces due to wheel
OOR defects. Despite the computational simplicity, the lumped parameter models are
adequate for low frequency vibration analysis of the train-track system. The higher

Fig. 1.16: A lumped-parameter 3-layer track model [14]
31

deflection modes and stresses in the track, however, can be evaluated using beam models
of track on continuous elastic support. A number of beam models based on either Euler
or Timoshenko beam theory have been developed. The most basic form of the beam
model consists of a rail beam seated on a bed of continuous layer of visco-elastic
elements or springs only [2, 16, 35]. The mass of the sleeper is often lumped with that of
the rail beam, assuming uniform distribution of sleepers over the track length. Newton
and Clark [35] developed Timoshenko beam model of the rail on continuous foundation
to investigate the wheel-rail impact load due to a wheel flat, as shown in Fig. 1.17.
Arnold and Joel [16] and Wen et al. [24] applied similar models to study the wheel-rail
impact load due to rail joints. Sing and Deepak [33] performed experiments to estimate
the track stiffness and damping properties.


Euler beam on elastic foundation with random stiffness and damping has been
recently used to investigate deformation under a moving load [34]. Such a model is
considered to represent the track system fairly well, and can provide a close form
analytical solution. Furthermore, there is a possibility to replace the subsoil foundation by
the frequency and wave number dependent stiffness. The model, however, presents
certain limitations, mainly: (i) the sleeper mass could not be adequately distributed over
Fig. 1.17: A single layer track model with continuous support [35]
32
the rail; (ii) sleeper bending effect are neglected and dynamic behavior of sleeper can not
be investigated; (iii) the contribution of elastic rail pads are ignored; (iv) the contributions
due to discrete rail supports from individual sleeper are ignored; and (v) for too many
simplifications used in the model could overestimate the wheel-rail impact loads.
Grassie et al. [36] developed a two-layer track model, which consists of two
Timoshenko beams supported on continuous spring-damper elements. Similar models
have also been applied by Tassilly and Vincent [95] to study rail corrugations, Thompson
[75] to study wheel-rail noise generations, and Bitzenbauer and Dinkel [2] to investigate
the dynamic interactions problems between a moving vehicle and substructure. Wu and
Thompson [15] have also developed a similar model (Fig. 1.18) with inclusion of rigid
sleeper mass to study the wheel-rail impact loads due to rail joints. The model permits for
analysis of dynamic behavior of the sleeper considering both symmetric and asymmetric
bending modes, while only limited information could be derived for dynamic behavior of
the total track system.


The rail beam models on discrete supports have been most widely used for
analysis of impact forces caused by wheel flaws [5, 7, 9, 19, 21, 29]. The supports are
usually characterized by either only sleeper masses or both sleeper and ballast masses
connected by spring-damping elements periodically. These models are also referred to as
two-layer, three-layer, and four-layer models depending on the number of layers
Fig. 1.18: A double layer track model with continuous support [2]
33
considered. The sleepers and ballasts can be modeled as either beams or rigid masses. A
two-layer model generally employs the sleeper masses alone, while the ballast and
subgrades are modeled as spring-damping elements, as shown in Fig. 1.19 [6, 8, 22, 23,
28, 31, 39]. Similar track models have been employed to study impact load and noise
emissions caused by wheel and rail defects [5, 7, 8, 22, 23].
Similar models have also been used by Morys [21] to study the effect of OOR
wheel profiles. Lei [19], and Andersson and Abrahamsson [28] investigated the dynamic
response of vehicle and track under high-speed conditions. The primary advantage of
this type of track models lies in the fact that responses corresponding to three resonance
frequencies of the track structure, the rail and the sleeper can be adequately captured [55].
Furthermore, it includes the effect of discrete sleeper support into the impact analysis.


Euler-beam sleeper models have also been used to study the impact load due to
wheel flat and the track behavior due to corrugated rail [35, 56]. Both rail and sleepers
have been considered as Timoshenko beams by Nielsen and Igeland [23] to study the
impact load due wheel flat and rail corrugation, by Dong [37] to investigate the impact
load due to wheel flats and rail joints, and by Morys [21] to study the growth of out-of-
roundness of wheels.
Fig. 1.19: Rail beam on discrete supports (only sleeper mass is included) [55]
34
Alternatively, three-layer track models have been developed where the ballast is
modeled as a mass in addition to the sleeper (Fig. 1.20). The figure shows additional
ballast masses interconnected by means of shear spring-damper for analysis of distributed
ballast deflections. Zhai and Cai [3, 53] developed a three-layer model to study the
wheel-rail impact loads due wheel flats and rail joints and the influence of the ballast
density on the wheel-rail contact force. Oscarsson [1] has used a similar model to
simulate the train-track interactions with stochastic track properties. Jin et al. [13, 17]
used the same model to study the effect of rail corrugations on vertical dynamics of the
rail and track. Sun et al. [10, 12] proposed a four-layer track model incorporating two
different masses for ballast and subgrade to investigate the wheel-rail impact load due to
flat. A five-layer model was also developed by Ishida and Ban [30] to analysis the effect
of different types of wheel flats in terms of impact loads and rail acceleration.


Newton and Clark [35] studied the effect of speed on wheel-rail impact force
considering three different track models, namely, Euler beam on elastic foundation
(EBOEF), Rayleigh-Timoshenko beam on elastic foundation (TBOEF) and Discrete
support model (DSM). The study concluded that at low speeds (up to 50 km/h), the
Fig. 1.20: A three-layer model of track system [53]
35
EBOEF yields an underestimate of the impact force, while the TBOEF track model yields
similar to the experimental result. The EBOEF model, however, resulted in
overestimation of the impact force at higher speeds. The DSM model resulted in
considerably lower impact force than the EBOEF and TBOEF models in the entire speed
range.
1.2.5 Wheel-rail contact models

The accurate description of contact between the wheel and rail is a necessary
condition to obtain reliable prediction of not only the impact forces in the presence of
wheel and rail profile defects, but also for curving and lateral stability analysis. The
rolling contact problem in railway vehicles has been extensively investigated over past
several decades. These studies have evolved in numerous methods for analysis of the
vehicle-track contact problem. Elkins [100] presented a state-of-the-art review of various
wheel-rail dynamic contact models, which were divided in two categories: Hertzian and
non-Hertzian contact models. The relative merits and demerits of these two types of
contact models together with those of other contact models are also described in [100].
Hertzian contact model is perhaps the simplest and most widely used to
characterize the rolling contact in railway vehicle. The model uses a single point contact
that lies under the wheel geometric center, where the contact region is described by a
very small ellipse. A Hertzian contact spring is also widely used to represent the stiffness
between the wheel and rail at the point of contact. The contact force is described as a
nonlinear function of the wheel-rail overlap or relative deflection, such that:

H
P C z
|
= A (1.5)
36
Where P is the contact force, z A is the wheel-rail overlap in the vertical
direction, exponent | is a constant ( | =1.5) and
H
C is the Hertzian wheel/rail contact
coefficient that depends on wheel-rail geometry and materials.
Yan and Fischer [103] have analyzed applicability of the Hertz contact theory to
wheel-rail contact problems and concluded that the wheel-rail contact dynamics can be
described with reasonable accuracy for purely elastic contact when the surface curvature
of the rail within the contact area remains unchanged. A major disadvantage of non-linear
Hertzian contact model is that it underestimates the impact force at low speed and
overestimates the impact force at a higher speed [22]. It has been suggested that a
linearized contact spring could adequately represent the wheel-rail contact when
variations in the overlap are very small [70]. The linearized contact spring has been
widely used to study the rail corrugations, vibration due to high frequency irregularities
on wheel-rail tread, and noise generations [45, 70, 75, 97]. The linearization, however,
yields overestimation of the contact stiffness in the vicinity of discontinuity and thereby
the impact loads [37].
Non-Hertzian contact models based on predicted contact area and shape have
been proposed by Kalker [99]. In the presence of a wheel flat, this approach requires
extensive computation as the contact area must be established as a function of wheel
angular position relative to the rail. A study of vehicle-track interaction by Baeza et al.
[88] has used the non-Hertzian contact model for a wheel with flat. The study stated that
it is not viable to solve this contact model simultaneously with the integration of
differential equation of motion. The study therefore, proposes pre-calculation of the
contact model for a set of relative wheel flat positions with respect to the track for
37
different normal loads. Interpolation of the contact force model is then used for the
dynamic simulation. Pascal and Sauvage [102] have shown alternate methods to calculate
the wheel-rail contact forces using nonHertzian contact patch. The solutions of contact
problems for an elliptical contact zone are available from the works by Kalker [98] and
Shen et al. [107], and later mostly used by others for analysis of wheel-rail squealing
noise [80], corrugation studies [92, 95] etc.
For analysis of vehicle system dynamics, instead of using single-point contact,
multiple point wheel-rail contact models were used in [104, 105, 106]. Several multipoint
contact models based on elliptic and non-elliptic profile are cited in [106]. A multiple
point contact model based on Hertzian static contact theory has been developed by Dong
[37] to study the wheel-rail impact load due to wheel flat. This model has also been
employed later by Sun [10] for the same purpose. Both of these studies reported that
multiple contact model shows good correlation between the predicted and experimental
data. The study, nevertheless, assumes that contact region is symmetric about the vertical
axis, and the results obtained are very similar to those predicted by Hertzian point contact
model. A very recent study by Zhu [52] developed a multipoint adaptive contact model to
account for the asymmetric contact as the as the flat enters the rail. Further study is,
however, required to establish the spring stiffness for the model.

1.2.6 Simulation methods

There are two different techniques widely used in the analysis of wheel-rail
interaction forces. These are frequency domain technique and time domain technique. In
frequency domain technique, an imaginary strip containing the wheel/rail irregularities is
pulled at a steady speed between the vehicle and the track, while the vehicle model is
38
held in a fixed position on the rail. Frequency domain technique has been used to study
the impact load due to wheel defects and the formation of corrugations and noise
generations [5, 71, 83, 92 ]. This technique takes less time to analyze and is effective for
the prediction of the frequency response related to excitation. However, this technique is
limited to investigation of the linear models only. When nonlinearity is present either in
vehicle-track system model or in contact model, it is necessary to adopt the solution
process in time domain. Two methods are generally employed in the dynamic analysis of
vehicle-track interaction in time domain, namely, modal analysis method and finite
element method.
In modal analysis method, a mode shape function is assumed depending on the
type of beam and type of support considered in the track model. All the partial
differential equations those depict the dynamic motion of continuous beam are converted
into ordinary differential equations. Rayleigh-Ritz approximation is employed in this
method to approximate the finite number of the lowest natural frequencies of a
continuous system. This method has been used to study the wheel-rail impact load due to
wheel flats and rail joints, and rail corrugations [3, 13, 17, 53, 73]. Sun et al. [10] carried
out an extensive study with three-dimensional coupled vehicle-track model represented
by 2077 equations and the simulation took 95 min on a Pentium 4 machine. Generalized
coordinate method has also been employed to study wheel-rail impact loads due to flat,
rail joints, rail corrugations, and noise generations [23, 35, 44, 80, 85, 86].
With the recent progress of computer capacity, use of finite element method for
solution of train-track interaction has become more attractive. Finite element method has
been applied to study the impact load due to wheel and rail defects and formation of OOR
39
in wheel profiles [5, 7, 19, 22, 23, 28]. The extreme adaptability and flexibility of finite
element method have made it a powerful tool to solve PDE over complex domain.
However, the accuracy of the obtained solution is usually a function of the mesh
resolution and the solution often requires substantial amounts of computer and user time,
particularly when extensive parametric study is required.
1.3 THESIS SCOPE AND OBJECTIVES

From the review of relevant literature, it is evident that considerable research
efforts have been made to predict the impact loads caused by wheel flats. Although the
presence of multiple flats in one wheel or different wheels within the same bogie is quite
common in practice, vast majority of the efforts focus on the interactions due to a single
flat. The presence of multiple flats may create larger impact forces at the wheel-rail
interface depending on the size and position of the flats. Increase in demand for higher
speeds and haul loads may further increase magnitude of wheel-rail impact loads due to
multiple wheel flats. The development of an adequate model of coupled vehicle and track
system is thus desirable for simulation of impact loads caused by wide range of wheel-
rail irregularities and to predict the dynamic responses of individual components of both
the vehicle and the track systems. The track model must be in sufficient details as it plays
a strong role in development of wheel-rail interaction forces, especially at higher speeds.
The primary objective of this dissertation research is thus formulated to develop a
comprehensive general-purpose dynamic model of the railway vehicle/track system
capable of predicting dynamic responses of each individual components of the coupled
vehicle-track system in the presence of single as well as multiple flats. The model needs
to be validated based on available results, which can then be applied in the analysis of
40
vertical dynamic forces due to wheel defects and parametric excitation of the track in the
high frequency range. The specific objectives of the present work are as follows:
- Develop a two-dimensional pitch-plane model of the vehicle to study the
interactions of two wheels, taking into account the contribution of vehicle pitch
motion.

- Develop a two-dimensional three-layer model of the track using Euler beam
formulation of the rail, supported on discrete elastic supports, where the sleepers
and ballasts are considered as rigid bodies, and rail pad and subgrade as vsico-
elastic elements.

- Formulate a haversine model for the single as well as multiple wheel flats to study
the wheel-rail impact forces.

- Formulate a simulation methodology for simultaneous solution of partial
differential equations representing the motion of the continuous rail and ordinary
differential equations for motion of the vehicle.

- Evaluate the impact forces arising from single as well as multiple wheel flats and
influences of one wheel flat on the forces imparted at the interface of the adjacent
wheel.

- Investigate the influences of variations in various design and operating parameters
on the magnitudes of the impact force, as well as forces and responses of vehicle
and track system components.

1.4 ORGANIZATION OF THE THESIS
In chapter 2, a two-dimensional pitch-plane vehicle model is developed together
with a two-dimensional three-layer track model. The two models are coupled by the non-
linear Hertzian wheel-rail contact model. Rayleigh-Ritz method is used to analyze the
coupled continuous rail track and lumped-parameter vehicle system models.
In chapter 3, the vehicle-track system model is validated under wheel flat
conditions using both theoretical and experimental results obtained from the literature.
The responses of individual components of the vehicle and track system are also derived
41
in terms of displacement together with the wheel-rail impact force as functions of
operating speed and flat geometry.
In chapter 4, the impact forces due to a single wheel flat as well as multiple flats
are investigated with both in phase and out of phase conditions. Effects of a flat in one
wheel on the forces developed at the interface of the other wheel within the same bogie
are also investigated. A comprehensive parametric study is conducted for better
understanding of the roles of various design and operational factors affecting the impact
loads induced by wheel flats.
In chapter 5, important conclusions drawn from this study and a list of
recommendation for further studies in this area are presented.













42
CHAPTER 2
VEHICLE-TRACK SYSTEM MODEL AND METHOD OF ANALYSIS

2.1 INTRODUCTION

This research is primarily concerned with the vertical wheel-rail impact load
associated with vertical dynamics of the coupled railway vehicle and track structure
system. In general, a vehicle-track system model for simulation of vertical dynamic
interactions is composed of vehicle model, track model and the contact model with rail
and wheel irregularities. It has been suggested that the vertical dynamics of the vehicle
alone contributes only little to the wheel-rail impact force [37, 42]. The vehicle model
may thus be greatly simplified to ensure representative wheel loads and their variations.
In this study, the vehicle system is modeled as a 5-DOF lumped mass model comprising a
quarter of the car body and half of the bogie coupled to two wheels through the primary
suspension. The analysis of dynamic impact loads, however, necessitates a most
comprehensive modeling of the continuous track system [4, 37]. A number of track
models of varying complexities have been developed, as described in chapter 1 [10, 12,
13, 17, 53]. In this study, a multiple-layer track system model comprising the rail pads,
the ballasts, and the subgrade is considered to study the coupled rail-vehicle system
dynamics in the presence of wheel defects.
Analyses of impact loads caused by a defective wheel in the vicinity of the
primary contact under study can be effectively investigated by considering the roll plane
vehicle model [25, 26]. The impact load caused by a defective wheel within the adjacent
bogie, however, requires a pitch plane model of the vehicle moving along the track. The
modal coordinates method is used to analyze of vertical dynamic interactions between
43
railway vehicle and the track system. The rail is modeled as an Euler Bernoulli beam that
considers transverse deflections only; this method is considered to be sufficiently
accurate for such analysis [3, 13, 17].
Considering that, the dynamic wheel-rail interactions concerned with the wheel
flaws are mostly dependent upon vertical dynamics of the vehicle-track system of
interest, the vehicle is assumed to be traveling on a straight track. This assumption is well
justified since the lateral and longitudinal relative motions between the wheel and rail are
small, and creep forces at the wheel-rail contact interface have little effect on the
dynamic vertical forces. The forward speed of the vehicle is also assumed to be constant,
while the contribution due to the track roughness is considered small in relation to forces
that may be caused by wheel defects.
In the present study, the dynamic wheel-rail impact load together with individual
system responses are investigated in the time domain. A pitch plane model of the vehicle-
track system is formulated to evaluate the effects of the wheel flats on the dynamic forces
developed at the wheel-rail interface. The generalized coordinates method is used to
convert the partial differential equations (PDE) describing the deflections of the
continuous track to ordinary differential equations (ODE). The deflection response of the
rail is determined from the theory of Euler simply supported beam.

2.2 VEHICLE SYSTEM MODEL
In typical North American freight car system, shown in Fig. 2.1, the wheelset to
side frame connection consists of only a bearing and bearing adapter with associated
friction. The lateral and yaw motions of the wheelsets relative to the side frames are thus
generally very small. The elastomeric pads between the bearing adapter and the
44
sideframe, however, form the primary suspension, whose stiffness is usually considered
in the modeling process. The bolster is connected to the sideframes by a combination of
vertical springs, as shown, in parallel with the friction plates, which constitute the
secondary suspension. The lateral motion is restricted by the bolster gibs. The dry friction
at the centerplate together with the stiffness of continuous contact with the side bearings
resists the truck rotation relative to the centerplate.



In the present study, the vehicle system is considered as a half truck, as shown in
Fig. 2.1, which supports quarter of the car body. Although a large numbers of vehicle
system models have been developed to study the wheel-rail impact loads in the presence
of wheel defects [3, 5, 7, 8, 10, 23, 30, 35, 37, 40], the majority of these consider the
dynamic analyses with a single wheel coupled with primary suspension [5, 7, 8, 30, 35,
40]. Other studies limited their investigations of the impact loads at the wheel-rail contact
point only. A two-dimensional roll plane vehicle model offers to investigate the effect of
one-wheel defect to the adjacent wheel within the same wheelset [25, 26]. However,
Fig. 2.1: A three-piece freight car truck

45
contributions due to pitch dynamics response of the bogie and effects of one wheel defect
to the other wheels within the same bogie could not be investigated. A pitch-plane
vehicle model is employed in this study to account for pitch oscillations of the bogie and
to investigate the effects of a defect on one wheel on the contact points of the adjacent
wheels.
2.2.1 Pitch-plane vehicle model
The vehicle system model used in this study consists of a quarter car supported on
a bogie, while the side frame is supported on two wheelsets. The primary suspension
connecting the wheels and the bogie frame is modeled as a parallel combination of a
linear spring and a viscous damping element. The secondary suspension connecting the
bogie frame and the car body is also modeled by parallel spring and damping elements.
The mass of the car body
c
M , bogie mass
t
M , wheel mass
w
M are coupled through the
suspension elements, as shown in Fig. 2.2. The total vehicle system model is represented
by a 5-DOF dynamic system that includes the car body vertical motion, ( )
c
w t , the bogie
vertical and pitch motions, ( )
t
w t and ( )
t
t , respectively, and vertical motions of the
wheels,
1
( )
w
w t and
2
( )
w
w t . The primary suspension stiffness and damping elements are
represented by
1 s
K and
1 s
C respectively, while
2 s
K and
2 s
C represent the stiffness and
viscous damping coefficient due to secondary suspension. The contact force between the
leading wheel and rail is denoted as
1
( ) P t and that between the trailing wheel and the rail
as
2
( ) P t .
t
J , and ( ) r t are mass moment of inertia of the bogie, wheel defect profile,
respectively.
f
l and
r
l are the distance from the mass center of the bogie to the front and
rear wheel centers, respectively.
46


2.2.2 Equations of motion
The equations of motion of the vehicle model are derived upon neglecting the
contribution due to track roughness, while the contact forces developed at the wheel-rail
interface are represented by
1
( ) P t and
2
( ) P t . It is further assumed that the resultant
primary suspension force acts at the bogie mass center. The equations of motion of the
vehicle system are summarized below.

Car body bounce motion:

2 2 2 2
0
c c s c s c s t s t
M w C w K w C w K w + + = (2.1)

Bogie bounce motion:

1 1 1 2 2
( ) ( ) ( )
t t s t f t w s t r t w s t c
M w K w l w K w l w K w w + + + + (2.2)
1 1 1 2 2
( ) ( ) ( ) 0
s t f t w s t r t w s t c
C w l w C w l w C w w + + + + =

Bogie pitch motion:
c
M
2 s
C
2 s
K
( )
c
w t
v
t
M ,
t
J ( )
t
t
( )
t
w t
2
( )
w
w t
1
( )
w
w t
w
M
w
M
1 s
K
1 s
K
1 s
C
1 s
C

( ) r t ( ) r t

2
( ) P t
1
( ) P t
f
l
r
l
Car body
Side frame
Front Wheel
Rear Wheel
Fig. 2.2: A 5-DOF pitch-plane vehicle model
47

1 1 1 2 1 1
( ) ( ) ( )
t t s f t f t w s r t r t w s f t f t w
J K l w l w K l w l w C l w l w + + + +
1 2
( ) 0
s r t r t w
C l w l w = (2.3)

Front wheel vertical motion:

1 1 1 1 1 1 1 1
( ) ( ) ( ) 0
w w s w t s w t s f t s f t
M w C w w K w w C l K l P t + + + = (2.4)

Rear wheel vertical motion:

2 1 2 1 2 1 1 2
( ) ( ) ( ) 0
w w s w t s w t s r t s r t
M w C w w K w w C l K l P t + + + + + = (2.5)


2.3 TRACK STRUCTURE

The dynamic forces developed at the wheel-rail interface,
1
( ) P t and
2
( ) P t , are
derived upon development and analysis of the track model. The conventional railway
track structure can be considered to consist of various discrete subsystem layers
representing the rails, sleepers, railpads, fasteners, ballast, sub-ballast, and the sub-grade.
Rails are connected to the sleepers through rail-pads and fasteners, which are supported
by the ballast. The ballast bed rests on a sub-ballast layer, which forms the transition
layer to the subgrade. The different layers of the tack structure are shown in Fig. 2.3.
The modern rail is made of steel and its cross section is derived from an I-profile
that serves as a carrier of the vertical load of the train that is distributed over the sleepers.
Railpads are usually synthetic materials to provide some cushioning effect between the
rail and the sleeper. The properties of the pads affect the overall track stiffness, while soft
railpads attenuate the high frequency vibration and permit larger deflection due to load.
The modeling of the track structure thus necessitates identification of appropriate
parameters of the railpad, apart from other structural layers.
48


Sleepers are made of either wood or concrete. The concrete sleepers are
increasingly being employed in railway transportation systems for their low maintenance
cost and longer durability. However, the wooden sleepers are still being used in railway
transportation system due to their low cost. Irrespective of the type used, the primary
function of the sleeper is to support the rails and maintain a good level of the track.
Ballast and sub-ballast are beds of coarse crushed hard stones. The standard depth of the
ballast is in the order of 0.3 m, but it is packed to 0.5 m around the sleeper ends to ensure
lateral stability [55].
From the physical point of view, the modeling of ballast materials and their
interactions with the adjacent layers is quite complex. In reference [37], it has been
shown that ballast properties have little effects on the dynamic wheel-rail contact forces,
mostly due to distant location of the ballast from the wheel-rail interface. The ballast,
sub-ballast, and sub-grade layers are thus generally lumped together and modeled by
equivalent spring and damping coefficients [21, 23, 28, 29]. A few studies have also
Fig. 2.3: Various layers of the track structure

49
considered the ballast as lumped masses below the ties, which are interconnected by
springs and dashpots in shear [3, 13, 17]. In this study, the latter model of the ballast is
used in order to achieve adequate response of the ballast, sleeper, and rail layers
individually. The effect of the three main masses in the track structure, namely, rail,
sleeper, and ballast masses on the wheel-rail impact load can also be effectively
investigated using this model. Furthermore, the shear coupling between the ballast
masses allows for considerations of influences of the adjacent ballast masses to the
specific ballast deflection.

2.3.1 Track system model

In this study, two-dimensional model of the track system is developed upon
consideration of three subsequent layers. These include the rail, and the lumped
representation of the sleeper and the ballast. A finite length of the track structure is
considered by including a total of 100 sleepers/ballasts. The track is assumed to be
symmetric with respect to its centerline. The rail is discretely supported on the sleepers,
ballasts/subballasts and subgrades elements, as shown in Fig. 2.4, where two layers of
discrete masses below the rail represent the sleeper and the ballast, respectively. The rail
beam is modeled as an Euler-Bernoulli beam and the conventional beam theory is used to
analyze its deflection response under a moving load. The rotatory inertia of the beam
cross section and beam deformations due to the shear force are considered negligible for
the Euler beam.

50



Railpads are placed between the steel rails and the sleepers to protect the sleepers
from wear and impact damage, and they provide electrical insulation of the rails. The
continuous rail beam is supported on the discrete spring-damper elements representing
the rail pads and fasteners. The rail is coupled to discrete sleeper masses, which provide
support of the rails, and preserve level and alignment of the track through the rail pads.
The ballast blocks are interconnected elastically so that a vertical deflection of one ballast
block will be distributed to others via the shear springs and dampers.
It has been suggested that a total of 50 to 60 sleepers/ballast elements would be
sufficient for analysis of dynamic of wheel-rail impact load due to a wheel flat [53, 94].
In this study, 100 sleepers/ballasts are considered in order to investigate the contribution
due to pitch dynamic of the bogie and various vehicle design parameters. The motion of
the rail beam coupled with the sleeper and ballast is expressed as ( , )
r
w x t , while
( )
si
w t and ( )
bi
w t describe the motions of the sleeper and ballast masses.
p
K ,
p
C ,
b
K and
b
C are the railpad and ballast stiffness and damping coefficients, respectively.
w
K andC
w

s
M
p
K
p
C
b
K
b
C
w
K
w
C
f
K
f
C
b
M
Rail

Sleeper
Ballast

r
m
( , )
r
w x t
( )
bi
w t
( )
si
w t
x
Fig.2.4: 3 Layer Railway track system model.

Fig. 2.4: A three-layer railway track system model.

51
are the shear stiffness and damping coefficients of the ballast, respectively. The subgrade
stiffness and damping are denoted as
f
K and
f
C , respectively. The rail mass per unit
length is represented by
r
m ,
s
M is mass of half of the sleeper, and
b
M is mass of each
ballast block.
A major assumption is made during the modeling of the track system model is
that the track system only vibrates in the vertical plane. Since the longitudinal and lateral
motions have little effect on the vertical wheel-rail interactions, the longitudinal and
lateral displacements of all the masses are neglected. Furthermore, the length of the track
considered in simulation is assumed to be sufficiently large to minimize the contributions
of assumed boundary conditions. The infinite long track is modeled as a track of finite
length, whose two ends are simply supported.

2.3.2 Equation of motion

The equations of motion of the entire track system are derived upon integrating
the equation of motion for the rail as an Euler beam with the differential equations of
motions for the discrete sleeper and ballast supports. The deflection of the continuous rail
can be derived from the partial differential equation for the Euler beam as [53]:
( ) ( )
( ) ( )
4 2
2
'
4 2
1 1
, ,
( ) ( )
N
r r
r rsi i j j
i j
w x t w x t
EI m F t x x P t x x
x t
o o
= =
c c
+ = +
c c

(2.6)

Where N is total number of sleepers considered in the model, k is the number of
deflection modes considered for the rail beam and j is the number of wheelsets
incorporated in the vehicle model, which represent the number of moving point loads
acting on the beam. E is the elastic modulus of rail beam materials and I is the second
moment of area. The coordinate x represents the longitudinal position of the beam with
52
respect to the left end support of the rail beam.
i
x defines the position of the i th sleeper
and ( ) x o is the Dirac delta function. ( )
rsi
F t is force developed at the i th rail/sleeper
interface, given by
( ) ( ) ( ) ( ) ( ) , ,
rsi pi r i si pi r i si
F t K w x t w t C w x t w t ( ( = +

(2.7)
The ( )
'
j
P t defines the total vertical force acting at the j th wheel and rail interface. It
comprises both the static vehicle load and the contact force ( )
j
P t , j =1, 2, such that:
( ) ( )
'
[0.5( ) ]
j j c t w
P t P t M M M g = + + + ; j =1, 2
The contact force ( )
j
P t is derived using the Hertzian contact model described in Eq.
(2.11), such that:

3/ 2
( ) [ ( ) ( , ) ( )] ;
j H wj r j j
P t C w t w x t r t = j =1, 2
Where
H
C is Hertzian wheel/rail contact coefficient and ( )
j
r t is the wheel flat function.
The contact force diminishes when a loss of contact of the wheel with the rail is
encountered, when [ ( ) ( , ) ( )] 0;
wj r j j
w t w x t r t s j =1, 2
The rail beam deflection ( ) ,
r
w x t is evaluated through solution of Eq. (2.6) using the
generalized coordinates method described in section 2.5. The deflection ( ) ,
r
w x t is
expressed by the product of the k th mode deflection mode ( )
k
Y x and the time
coordinate ( )
k
q t . The equation of motion for the discrete sleeper and ballast masses are
derived as follows:

( ) ( ) ( ) ( ) ( ) ( ) ( )
si si p b si p b si b bi b si
M w t C C w t K K w t C w t K w t + + + +
53
( ) ( ) ( ) ( )
1 1
0
K K
p k i k p k i k
k k
C Y x q t K Y x q t
= =
=

; 1, 2,..... i N = (2.8)


( )
( )
( )
( )
( ) ( ) ( ) 2 2
bi bi b f w bi b f w bi b si b si
M w t C C C w t K K K w t C w t K w t + + + + + +

( ) ( ) ( ) ( )
( 1) ( 1) ( 1) ( 1)
0
w b i w b i w b i w b i
C w t K w t C w t K w t
+ +
= ; 1, 2,..... i N = (2.9)

The shear coupling between the first ballast mass and the beam support, and the
last ballast mass and the beam support, however, are not considered in Eq. (2.9), i.e.
0 0
( ) 0;
b b
w t w = = and
( 1) ( 1)
( ) ( ) 0.
b N b N
w t w t
+ +
= =
The ballast force developed at the sleeper-ballast interface can be expressed as:
( ) ( ) ( ) ( ) ( ) ( ) ( )
sbi bi si bi bi si bi wi bi b i wi bi b i
F t K w t w t C w t w t K w w C w w
+ +
( ( = + + +


( ) ( )
wi bi b i wi bi b i
K w w C w w

+ + (2.10)

2.4 WHEEL-RAIL INTERFACE

Wheel-rail interface distinguishes railways from other forms of land transports. In
order to simulate the wheel-rail interaction force, the contact between the wheel and rail
must be established by a well-defined method, since an accurate solution of the rolling
contact problem is quite complex. The essential characteristic of the wheel-rail contact is
its extreme stiffness in vertical direction. The area of contact is generally very small,
while the interface supports the traction, braking, and curving forces apart from the
vertical forces.
The interaction between the vehicle and the track system is achieved at the
wheel/rail interfaces through wheel/rail force compatibility. The wheel-rail contact can be
54
represented as shown in Fig. 2.5, where the load due to the car-body is represented by W,
while assuming negligible vertical dynamics of the vehicle. The wheel-rail contact has
been widely described by the nonlinear Hertzian contact theory commonly used in the
wheel/rail interaction problems [3, 8, 13, 17, 35, 53, 85].
According to the Hertzian contact theory, the wheel-rail contact force is related to
the rail deflection in a nonlinear manner, such as:
3/ 2
( ) ( )
H
P t C z t = A (2.11)
Where ( ) z t A is the wheel-rail overlap in the vertical direction. In the absence of a
wheel defect, the overlap is defined by the relative motion of the wheel with respect to
the rail:
( ) ( ) ( , )
w r
z t w t w x t A = (2.12)
In the above equation, ( )
w
w t and ( , )
r
w x t are the wheel and rail deflections in
vertical direction, respectively.


The magnitude of the wheel-rail overlap is dependent upon many design and
operating factors. These include the speed, car load, material properties, wheel, and track
surface roughness. The wheels in service, however, may exhibit surface non-uniformity
W
H
C

Wheel
Rail

Vehicle Load
Hertzian contact spring
( )
w
w t
( , )
r
w x t
Fig. 2.5: Wheel/Rail contact model used in the present study.
55
attributed to brake-induced wheel flats, shelling, etc. In such situations, the overlap and
thus the contact force is predominantly influenced by the nature of the wheel defects.
Although, a wide-range of wheel defects have been documented in the literature, the
wheel flats are shown to be among the commonly observed defects and cause excessive
impact loads at the wheel-rail interface. The wheel flat can be divided into two
categories: (i) chord type flat; and (ii) haversine type flat. The former one represents a
freshly formed flat, while the latter is used to describe the geometry of the flat on the
wheel in service that yields rounded edges [5, 7 12, 37].
A chorded or newly formed flat is also referred to as the ideal flat, which is
geometrically shown in Fig. 2.6. When a rigid wheel with an ideal flat rolls over a rigid
rail, its motion can be directly related to the wheel and flat geometry described by the flat
length
f
L and the flat depth
f
D .




u

f
D
o
f
L
R

x
( ) t r A
Fig. 2.6: A wheel with an idealized flat [7]

56
An ideal flat is expressed as [37]:

(1 )
0
R cos
r
u
=
`
)

0
,
u
u u
< <
> s
(2.13)

Where is the angle subtended by the flat at the wheel centre, R is radius of the wheel,
r is variations in the wheel radius due to the flat, and u is the angular coordinate of
contact point within the flat zone or the cord, and expressed as:

1
1
sin ( / )
sin [( ) / ]
f
x R
L x R
u

=
`

)

0 / 2
/ 2
f
f f
x L
L x L
s <
s <
(2.14)

Where x is the longitudinal coordinate of the contact point within the flat with respect to
point A.
After formation of a wheel flat, the sharp edges at the end of the profile become
rounded, while continued in service. This type of flat is commonly known as haversine
flat, which is shown in Fig. 2.7.



f
D
o
R
x
f
L
( ) t r

Fig. 2.7: A wheel with haversine type flat
57

The variation in the radii of a contact point ( ) r t within a haversine flat is expressed as:
1
( ) [1 cos(2 / )]
2
f f
r t D x L t = (2.15)
Where
f
D is the flat depth that has been related to wheel radius R , in the
following manner [37]:
2
/(16 )
f f
D L R = (2.16)
The presence of a wheel defect obviously greatly influences the instantaneous overlap
( ) z t A between the wheel and the rail, which can be expressed as:
( ) ( ) ( , ) ( )
w r
z t w t w x t r t A = (2.17)
Where ( ) r t is the function of the wheel flat geometry, as described in Eqs. (2.13) and
(2.15).
The contact force developed at the wheel rail interface can thus be estimated from the
Hertzian contact model, defined in Eq. (2.11). The Hertzian contact coefficient
H
C has
been related to material properties of the rail and the wheel geometry, such that [73]:
4
3(1 )
wr e
H
wr
G R
C
u
=

(2.18)
where
wr
G and
wr
u are shear modulus and Poissons ratio of the rail, respectively.
e r w
R R R = , where,
r
R is the rolling radius of the wheel, and
( )
w t
w
w r
R
R
R

where
w
is the wheel profile radius and
t
R is the rail profile radius.



58
2.5 METHOD OF ANALYSIS

The equations of motion of the coupled vehicle-track system together with the
nonlinear contact are solved in the time domain. As the vehicle moves along the track,
the contact force and deflections of the rail and those of the lumped masses of vehicle and
the track are calculated through numerical integration of the differential equations. In this
method, the unknown wheel-rail contact force is determined from the individual
responses of the vehicle and track components. The time-domain approach also permits
for consideration of loss of contact between the wheel and the rail.
The wheel-rail interaction models have been widely solved using two different
approaches. These include the Finite Element Method and Generalized Coordinate
Method. The Finite Element Method (FEM) is employed for finding approximate
solutions of partial differential equations (PDE) describing deflection of the continuous
track under moving contact forces. The solution approach is generally based on the
transferring the PDE to Ordinary differential equations (ODE), which are subsequently
solved by various well-established techniques, such as finite difference method. Finite
element method is extremely versatile and powerful tool to solve PDEs with different
boundary conditions and has been used by many researchers for analysis of wheel-rail
interactions [19, 20, 24, 37]. The accuracy of the obtained solution, however, is usually a
function of the mesh resolution, while the solutions are generally demand more
computational time.
On the other hand, the generalized coordinate method utilizes assumed mode
shape to convert the PDE into ODEs. The assumed mode shape are required to satisfy the
boundary condition of the continuous rail [17, 53]. The RayleighRitz approximation is a
59
very good choice for approximating a finite number of the lower mode of a continuous
system. This method has been widely used in various studies on the railway vehicle
dynamics [3, 13, 17, 53]. The modal analysis of the rail requires that the track be modeled
over a finite length depending on the number of modes considered. The rail is than
describe through its modal parameters and the physical deflection of the rail is
determined through modal superposition. The primary advantage of the modal method is
its

computational efficiency, while most

important limitation is the difficulty in capturing
strong nonlinear effects of system parameters.
The track system model formulated in this study comprises both ODEs and PDE
describing the deflection of the lumped sleeper and ballast masses, and the continuous
rail, respectively. The PDE is expressed as ODEs by assuming a mode shape function.
The Rayleigh-Ritz method is used to express the fourth order PDE describing the motion
of the continuous rail by a series of second order ordinary differential equations in terms
of the time coordinates. The resulting ODEs of the track and vehicle systems are then
solved in time domain to obtained responses of individual components of the vehicle-
track system model. The relative responses between the components are used to derive
the dynamic interaction forces. The closed form of the solutions by using this method
largely depends on the accurate assumption of the mode shape and the number of modes
considered. Zhai and Cai [53] have suggested that good convergence of solution can be
obtained if the number of modes is equal to or more than 60. In this study, a total of 100
modes of the rail beam are considered for analysis of the coupled vehicle-track system
model.
60
The deflection mode of continuous beam with simply supported boundary
conditions can be derived from the Euler-Bernoulli equation of the beam. The deflection
modes and the natural frequencies of an Euler beam in the absence of sleeper supports
and external loads have been well documented and can be expressed as [84]:
( ) sin
k
k x
Y x
l
t | |
=
|
\ .
; and

2
k
r
k EI
l m
t
e
| |
=
|
\ .
, k =1, 2, 3K (2.19)

Where ( )
k
Y x is the deflection mode,
k
e is the corresponding natural frequency
and l is the beam length.
The above equations were obtained for the following boundary conditions:
(0) ( ) 0 Y Y l = = ; and
(0) ( ) 0 Y Y l '' '' = = (2.20)
The deflection response of the rail is then derived from :
( )
1
, ( ) ( )
K
r k k
k
w x t Y x q t
=
=

(2.21)
Where K is the number of modes considered.
The rail deflections in the vicinity of the contact points with the wheels are
derived from:
1
1
( , ) sin ( )
K
r k k
k
k
w x t Y vt q t
l
t
=
| |
=
|
\ .

; and
2
1
( , ) sin ( ) ( )
K
r k t k
k
k
w x t Y vt l q t
l
t
=
(
=
(

(2.22)
Where
t f r
l l l = + is the wheel base of the bogie.
61
The substitution of the rail deflection response from Eq. (2.21) together with the
mode shape
k
Y in the PDE, Eq. (2.6) yields a set of ODEs in ( ) q t , expressed as:
( ) ( ) ( ) ( ) ( ) ( ) ( ) ( )
4
1 1 1 1
N K N K
k pi k i k i k k pi k i k i k
i k i k
r
EI k
q t C Y x Y x q t q t K Y x Y x q t
m l
t
o o
= = = =
| |
+ + +
|
\ .



( ) ( ) ( ) ( ) ( )
2
'
1 1 1
( )
N N
pi k i si pi k i si j k Gj
i i j
C Y x w t K Y x w t P t Y x o o o
= = =
=

(2.23)

Where (2/ )
r
m l o = and k =1, 2, 3K

The above equations incorporate the effects of sleeper supports through the restoring and
dissipative forces developed by the rail pad in the vicinity of the sleeper support. The
total force of the rail pad, attributed to rail deflections alone, can be further simplified as:

( ) ( ) ( ) ( ) ( ) ( )
1 1 1 1
N K N K
pi k i k i k pi k i k i k
i k i k
K Y x Y x q t C Y x Y x q t
= = = =
+

=

( ) ( ) ( ) ( ) ( ) ( )
1 1 1 1
N K N K
pi k i k i k pi k i k i k
i k i k
K Y x Y x q t C Y x Y x q t
= = = =
+

(2.24)

As examples, these forces associated with the first mode ( 1 k = ) can be derived from:
( ) ( ) ( ) ( ) ( ) ( )
1 1 1 1 1 1
1 1
N N
pi i i pi i i
i i
K Y x Y x q t C Y x Y x q t
= =
+

=
( ) ( ) ( ) ( ) ( ) ( ) ( ) ( )
1 1 1 1 1 1 2 1 2 1 3 1 3 1 1
...............
p N N
K q Y x Y x Y x Y x Y x Y x Y x Y x ( + + + +

+

( ) ( ) ( ) ( ) ( ) ( ) ( ) ( )
1 1 1 1 1 1 2 1 2 1 3 1 3 1 1
...............
p N N
C q Y x Y x Y x Y x Y x Y x Y x Y x ( + + + +

(2.25)

The forces associated with the second deflection mode ( k =2) are derived in a similar
manner as:
( ) ( ) ( ) ( ) ( ) ( )
2 2 2 2 2 2
1 1
N N
pi i i pi i i
i i
K Y x Y x q t C Y x Y x q t
= =
+

=

62
( ) ( ) ( ) ( ) ( ) ( ) ( ) ( )
2 2 1 2 1 2 2 2 2 2 3 2 3 2 2
...............
p N N
K q Y x Y x Y x Y x Y x Y x Y x Y x ( + + + +

+

( ) ( ) ( ) ( ) ( ) ( ) ( ) ( )
2 2 1 2 1 2 2 2 2 2 3 2 3 2 2
...............
p N N
C q Y x Y x Y x Y x Y x Y x Y x Y x ( + + + +

(2.26)

The forces associated with the K th mode considered in the analysis is also expressed as:
( ) ( ) ( ) ( ) ( ) ( )
1 1
N N
pi k i k i k pi k i k i k
i i
K Y x Y x q t C Y x Y x q t
= =
+

=

( ) ( ) ( ) ( ) ( ) ( ) ( ) ( )
1 1 2 2 3 3
.........
p K K K K K K K K N K N
K q Y x Y x Y x Y x Y x Y x Y x Y x ( + + + +

+

( ) ( ) ( ) ( ) ( ) ( ) ( ) ( )
1 1 2 2 3 3
........
p K K K K K K K K N K N
C q Y x Y x Y x Y x Y x Y x Y x Y x ( + + + +

(2.27)

The equation of motion of the rail supported on discrete sleepers and railpads are
formulated upon substituting for component of the railpad force, described above, in Eq.
(2.23). These equations are summarized below, as examples,
1 k =
( ) ( ) ( ) ( ) ( ) ( ) ( ) ( )
4
1 1 1 1 1 1 1 1
1 1
1
N N
pi i i pi i i
i i
r
EI
q t C Y x Y x q t q t K Y x Y x q t
m l
t
o o
= =
| |
+ + +
|
\ .


( ) ( ) ( ) ( ) ( )
2
'
1 1
1 1 1
( )
N N
pi i si pi i si j k Gj
i i j
C Y x w t K Y x w t P t Y x o o o
= = =
=

(2.28)

2 k =

( ) ( ) ( ) ( ) ( ) ( ) ( ) ( )
4
2 2 2 2 2 2 2 2
1 1
2
N N
pi i i pi i i
i i
r
EI
q t C Y x Y x q t q t K Y x Y x q t
m l
t
o o
= =
| |
+ + +
|
\ .


( ) ( ) ( ) ( ) ( )
2
'
2 2
1 1 1
( )
N N
pi i si pi i si j k Gj
i i j
C Y x w t K Y x w t P t Y x o o o
= = =
=

(2.29)

3 k =
( ) ( ) ( ) ( ) ( ) ( ) ( ) ( )
4
3 3 3 3 3 3 3 3
1 1
3
N N
pi i i pi i i
i i
r
EI
q t C Y x Y x q t q t K Y x Y x q t
m l
t
o o
= =
| |
+ + +
|
\ .


( ) ( ) ( ) ( ) ( )
2
'
3 3
1 1 1
( )
N N
pi i si pi i si j k Gj
i i j
C Y x w t K Y x w t P t Y x o o o
= = =
=

(2.30)

100 k =
63

( ) ( ) ( ) ( ) ( ) ( ) ( ) ( )
4
100 100 100 100 100 100 100 100
1 1
100
N N
pi i i pi i i
i i
r
EI
q t C Y x Y x q t q t K Y x Y x q t
m l
t
o o
= =
| |
+ + +
|
\ .


( ) ( ) ( ) ( ) ( )
2
'
100 100
1 1 1
( )
N N
pi i si pi i si j k Gj
i i j
C Y x w t K Y x w t P t Y x o o o
= = =
=

(2.31)

The deflection response of the rail is finally derived from summation of a total of 100
deflection modes, as:

1 1 2 2 3 3 100 100
( , ) ( ) ( ) ( ) ( ) ( ) ( ) ................................ ( ) ( )
r
w x t Y x q t Y x q t Y x q t Y x q t = + + + + (2.32)

The equations of motion of the vehicle system described by Eqs. (2.1) to (2.5),
and of the track system derived in Eqs. (2.6) to (2.9) and (2.23) together with the Hertzian
nonlinear contact model in Eq. (2.17) describe the vertical dynamics of the coupled
vehicle-track system. These equations can be solved simultaneously to obtain the wheel-
rail interaction force and the responses of the vehicle-track system components.

2.6 SUMMARY
A two-dimensional pitch plane model of the coupled vehicle-track system is
developed to investigate the effects of the wheel flats on the vehicle-track interaction
force and the dynamic responses of vehicle and track components. The model is
formulated assuming constant forward speed of the vehicle, while moving on a straight
track. The track is considered to be symmetric with respect to its center line. The rail
sleepers are assumed to be uniformly spaced, while a nonlinear Hertzian contact model is
considered for deriving the rolling wheel contact force. The railpad, primary and
secondary suspensions are assumed to possess linear stiffness and damping properties.
64
The vehicle is modeled as lumped parameter system, which consists of quarter of
a car body, half bogie, primary, and secondary suspension, and rigid wheels. The rail is
modeled as an Euler beam simply supported at the extreme ends, which ignores shear
deformation and rotational inertia of the rail. The track is modeled as a three-layer system
with rigid sleeper and ballast masses. The mass moment inertia of the sleeper and ballast
are ignored. The rail pad and subgrade are modeled as combinations of discrete spring
and damping elements. The continuous rail is supported on uniformly spaced sleeper
masses via railpads that are modeled as discretely spaced spring and damping elements.
The shear coupling of the interlocking ballast blocks is incorporated through shear
springs and damping elements. The vehicle is modeled in pitch plane that can be used to
simulate the effect of wheel defect on one wheelset to other. The PDE representing the
motion of the continuous rail beam is converted to a set of ODEs by using generalized
coordinate method. The proposed vehicle-track system is validated using the reported
data on dynamic impact force in the presence of a wheel flat in the subsequent chapter.
The validated model is applied to study the wheel-rail impact force and vehicle system
response to a single wheel flat.







65
CHAPTER 3

MODEL VALIDATION AND VEHICLE-TRACK SYSTEM RESPONSE

3.1 INTRODUCTION

The coupled vehicle-track model together with the Hertzian contact model
developed in chapter 2 is used to obtain the vehicle-track interaction responses in terms
of contact force and displacements of various vehicle and track components. The primary
objective in developing the model was to examine the dynamic wheel-rail impact force in
the presence of wheel flats. For this purpose, a multi-layer track system coupled with the
pitch-plane vehicle model with two wheelsets is utilized. The validity of the model
should, however, be examined first using the available test and reported results that deal
with wheel flat and its impact. One such study by Zhai et al. [3] presents both
experimental and analytical results for a well-defined flat introduced at one wheel. The
present model parameters are adjusted to represent the vehicle-track system presented in
the purpose of examining the model validity [3]. Due to the lack of available data for
wide range of operating conditions and vehicle-track system design parameters in
addition to wheel defects, only a limited validation is possible for a study of this
magnitude. Since the contact force is very sensitive to the wheel and rail responses
especially when irregularities are present either in the wheel or the rail, the proposed
model is validated in time domain.
The vast majority of the reported studies focus on response analyses due to a flat
in a single wheel [3, 7, 35, 42, 73]. The proposed pitch plane model permits for analysis
of the influence of a flat in one wheel on the contact force response of the other wheel. In
this chapter, the validated model is utilized for a baseline Canadian fright car system to
66
investigate the responses in the presence of wheel flats either in front or rear wheels of a
bogie. The analyses are performed with parameters representative of a bogie of a C.P. rail
car with 102 kN load. The response characteristics are obtained for a fixed size flat in one
of the wheels as a function of constant forward speed.
3.2 MODEL VALIDATION

In order to validate the present model in the presence of a wheel flat, the
equations of motion of vehicle and track system described in Eqs. (2.1) to (2.9) are
combined with wheel defect and wheel-rail interface models presented in Eqs. (2.15) and
(2.17), respectively. The validity of the resulting model is then evaluated using the data
reported by Zhai et al. [3]. In this reported study, a 10-DOF pitch plane model of the full
car was considered with a three-layer model. The rail was modeled as an Euler-Bernoulli
beam. All the parameters of the vehicle, the track, and defect size considered in this
validation are identical to those reported by Zhai et al. [3], except for the primary
suspension properties and Hertzian contact stiffness, which were not reported. These
parameters were obtained for a typical wagon track system used in North America [37].
The parameters employed in this simulation are listed in Table 3.1.
The listed parameters resulted in the total track length of 54.5 m and a total of 305
coupled ordinary differential equation for the model. The dynamic response of the entire
vehicle-track system is evaluated under a constant speed of 27 km/h as reported by Zhai
et al. [3] in the presence of a 52.8 mm long and 1 mm deep flat in the leading wheel. The
variation in the dynamic contact force obtained at the interface of the defective wheel and
rail is evaluated and compared with those reported in [3], as shown in Fig. 3.1. The
simulation results were obtained under steady-state condition.
67
Table 3.1: Parameters used for examining validity of the model with single wheel flat
[3, 37]





Notation Parameter Value
c
M Car body mass (half) 38500 kg

t
M Bogie mass 1100 kg
w
M Wheel mass 1200 kg
t
J Bogie mass moment of inertia 760 kg-m
2

1 s
K Primary suspension stiffness 7.8810
5
kN/m

1 s
C Primary suspension damping 3.5 kN-s/m

2 s
K Secondary suspension stiffness 5.3210
3
kN/m
2 s
C Secondary suspension damping 70

kN-s/m

t
l Wheelset distance 0.875 m
R Wheel radius 0.42 m
f
L Flat length 52.8 mm
f
D Flat depth 1 mm
H
C Hertz spring constant

8710
9
N/m
3/2
r
m Rail mass per unit length 60.64 kg/m
EI Rail bending stiffness 6.62 MN-m
2

s
M Sleeper mass 118.5 kg
b
M
Ballast mass 739 kg
p
K Railpad stiffness 120 MN/m
b
K Ballast stiffness 182 MN/m
w
K Ballast shear stiffness 147 MN/m
f
K Subgrade stiffness 78.4 MN/m
p
C Railpad damping 75 kN-s/m
b
C Ballast damping 58.8 kN-s/m
w
C Ballast shear damping 80 kN-s/m
f
C Subgrade damping 31.15 kN-s/m
l Length of the rail 100 sleepers long
l
s
Sleeper spacing 0.545 m
68
It can be seen that the contact force response predicted by the current model
agrees reasonably well with the response reported by Zhai et al. [3]. It should be noted
that the reported study employed a 10 DOF pitch plane model of the entire car, while the
present study considers only a 5-DOF model of the quarter car. Despite this
simplification, the current model yields nearly similar impact force response, which
suggests negligible contributions due to the vehicle pitch. A reduced model would thus be
sufficient for accurately predicting the dynamic contact force due to a wheel flat. The
amplitude of impact force and the period of vibrations predicted by both models are in
very good agreement. The peak impact force predicted by both models is in the order of
245 kN. The dominant period of oscillation of the contact force is approximately 0.01 s
for both models. The results attained from both the models suggest that this period of
oscillation rapidly approaches to nearly 0.02 s after the excitation due to wheel flat has
passed the contact region, although some differences in the responses become evident.

Fig. 3.1: Comparison of wheel-rail impact force response of the
present model with that reported by Zhai et al. [3].

Time (s)
F
o
r
c
e


(
k
N
)
Time (s)
F
o
r
c
e


(
k
N
)
Time (s)
F
o
r
c
e


(
k
N
)
Current model
Zhai et al. [3]
69
Figure 3.2 (a) further illustrates the wheel-rail contact force response of the
present model over a longer duration for examining the possible response when the
excitation due to the flat is absent. The results clearly illustrate frequency of oscillation in
the order of 56 Hz, which agrees well with that reported in studies [73, 87]. Figure 3.2 (b)
shows the contact force response over three consecutive impacts of the wheel flat with
the rail. The results clearly show the periodic nature of the impact force with a period of
0.3519 s that corresponds to one revolution of the wheel at the selected speed of 27 km/h.
The results further show that the contact force response diminishes as approaches its
steady state value at t= 0.2 s after the impact.
The results in Fig. 3.2 further show the sequence of events in terms of contact
force and their frequency of oscillation, which can be explained in the following manner.
As the wheel flat enters the contact area, there is a sudden drop in the contact force
followed by a large peak due to the wheel-rail impact. This referred to as P1 force in
literature [3, 23, 53], lasts for a duration of 0.008 s, which correspond to the duration of
the flat in contact with the rail. The frequency of P1 can therefore be referred to as the
excitation frequency due to the flat, and is a function of flat size and forward speed. The
following sequence of peak force known as P2 force oscillates at a frequency of 125 Hz.
This is primarily due to oscillation of the rail on the support pads. From various
simulation runs, it is noticed that this frequency is unaffected by the flat size and speed,
and is a function of pad properties, which is also reported in [55]. As shown in Fig. 3.2,
the final sets of oscillation in wheel rail contact force takes place at a frequency of 56 Hz,
which is attributed to the coupled vehicle-track natural frequency.
70
1.04 1.05 1.06 1.07 1.08 1.09 1.1 1.11 1.12 1.13
0.5
1
1.5
2
x 10
5
Time (s)
W
h
e
e
l
/
R
a
i
l

c
o
n
t
a
c
t

f
o
r
c
e

(
N
)
(a)
0.7 0.8 0.9 1 1.1 1.2 1.3 1.4
0
0.5
1
1.5
2
2.5
x 10
5
Time (s)
W
h
e
e
l
/
R
a
i
l

c
o
n
t
a
c
t

f
o
r
c
e

(
N
)

(b)
Fig. 3.2: Time-history of impact force response predicted by the current model:
(a) single impact; and (b) three-consecutive impacts (v = 27 km/h;
f
L = 52.8 mm;
f
D = 1 mm)

71
Time (s)
F
o
r
c
e


(
k
N
)
Time (s)
F
o
r
c
e


(
k
N
)
Time (s)
F
o
r
c
e


(
k
N
)

Fig. 3.3: Variations in the measured dynamic contact force due to
a wheel flat, reported by Zhai et al. [3]

Zhai et al. [3] also performed measurements of the dynamic contact force caused
by 52.8 mm long and 1mm deep wheel flat. The wheel-rail impact force was
continuously measured by a Wheel Impact Load Detector (WILD), while the vehicle was
running over the track at a speed of 27 km/h. WILD is a wheel flat dynamic detection
system that is widely used in the rail industry for detecting defective wheels. Figure 3.3
illustrates the variations in measured dynamic contact. The comparison of the contact
force predicted by the current model with the measured data also shows reasonably good
agreement, particularly with respect to the trend and the corresponding wavelength of the
contact force. The peak contact force predicted by the model, however, is slightly lower
than the measured peak force, while the difference is in the order of 10 kN. This
difference could be attributed to consideration of linear properties of the vehicle and track
system. A few studies have been shown that consideration of nonlinear track properties
72
would yield relatively higher peak magnitude of the contact force [8, 22, 94]. The results
presented in Figs. 3.1 to 3.3, suggest that the proposed pitch plane model of the vehicle-
track system can effectively predict the wheel-rail contact force in the presence of a
wheel flat.
3.3 RESPONSE ANALYSES OF THE VEHICLE-TRACK SYSTEM MODEL

Wheel flats are known to cause high magnitudes of impact forces at the wheel-rail
interface, which may cause fatigue failure of the railway infrastructure. The magnitudes
of the contact forces are strongly dependent upon a number of design and operating
parameters, such as wheel load, speed, track and vehicle properties and geometry of the
wheel flat. Furthermore, the impact forces would strongly influence the dynamic
responses and fatigue of the vehicle and track system components. The validated vehicle-
track system model is applied to investigate the dynamic contact force as well as dynamic
responses of the components at a constant forward speed. The analyses are performed
under a haversine wheel flat (
f
L = 52 mm and
f
D = 0.4 mm). The type of flat considered
closely meets the wheel removal criterion recommended by a number of American and
European railroad organizations, such as AAR [91], Transport Canada [108], Swedish
Railway [10], and UK Rail Safety and Standard Board [109].
The equations of motion for the vehicle-track model are solved for the track
length involving 100 sleepers, and a forward speed 70 km/h. The simulations were
performed using a time step size of 0.000078 s, which is significantly smaller than the
time required for the flat to overcome the contact with the rail. The simulation parameters
used for analysis are given in the Table 3.2, which represent those of a typical North
American railroad car.
73
Table 3.2: Nominal simulation parameters [3, 37]
Notation Parameter Value
c
M Car body mass (quarter) 19400 kg

t
M Bogie mass 500 kg
w
M Wheel mass 500 kg
t
J Bogie mass moment of inertia 176 kg-m
2

1 s
K Primary suspension stiffness 7.8810
5
kN/m

1 s
C Primary suspension damping 3.5 kN-s/m

2 s
K Secondary suspension stiffness 6.1110
3
kN/m
2 s
C Secondary suspension damping 158 kN-s/m

t
l Wheelset distance 1.25 m
R Wheel radius 0.42 m
f
L Flat length 52 mm
f
D Flat depth 0.4 mm
H
C Hertz spring constant

8710
9
N/m
3/2
r
m Rail mass per unit length 60.64 kg/m
EI Rail bending stiffness 6.62 MN-m
2

s
M Sleeper mass 118.5 kg
b
M
Ballast mass 739 kg
p
K Railpad stiffness 120 MN/m
b
K Ballast stiffness 182 MN/m
w
K Ballast shear stiffness 147 MN/m
f
K Subgrade stiffness 78.4 MN/m
p
C Railpad damping 75 kN-s/m
b
C Ballast damping 58.8 kN-s/m
w
C Ballast shear damping 80 kN-s/m
f
C Subgrade damping 31.15 kN-s/m
l Length of the rail 100 sleepers long
l
s
Sleeper spacing 0.6 m

74
3.3.1 Wheel-Rail Contact Force Response
The time history of the rear wheel-rail impact force in the presence of a rear
wheel flat is shown in Fig. 3.4. Figure 3.4 (a) shows the variation in contact force over a
single cycle, while Fig. 3.4 (b) illustrates the impact force response over three
consecutive cycles. The results clearly show high magnitude impact force occurring at
the rear wheel-rail interface due to the flat. The high magnitude impact force occurs for a
very short duration, while the force response approaches a low magnitude steady value in
approximately 0.05 s. The results further show that as the flat comes into contact with the
rail, the contact force decreases sharply from its static equilibrium. The contact force then
increases rapidly to its peak value. The first peak in the impact force is commonly
referred to as P1 force, which is primarily attributed to the wheel defect. The magnitude
of this peak impact force approaches as high as 180 kN for 52 mm long and 0.4 mm deep
flat considered in the study. The ratio of peak force to the static load is nearby 1.77. The
second peak in contact force with relatively lower magnitude is commonly known as P2
force, which is primarily attributed to the stiffness and damping properties of the track
[53]. The magnitude of this secondary peak is significantly smaller than the primary
peak. The impact force responses attained over three consecutive cycles show the time
interval between impacts, which equals the period of one wheel revolution.
A freight car system with two bogies consists of four wheelsets and eight wheels.
There may be one or more wheels with defects. The dynamic interactions of a defective
wheel with the track may also impose considerable magnitudes of contact force to
another wheel within the same bogie, which is free of defects. The proposed pitch plane
vehicle model permits for analysis of effects of a flat on one of the bogie wheels on the
75
1.06 1.08 1.1 1.12 1.14 1.16
2
4
6
8
10
12
14
16
18
x 10
4
Time (s)
R
e
a
r

W
/
R

I
m
p
a
c
t

f
o
r
c
e

(
N
)


(a)
1.05 1.1 1.15 1.2 1.25 1.3 1.35 1.4 1.45
0
0.5
1
1.5
2
x 10
5
Time (s)
R
e
a
r

W
/
R

I
m
p
a
c
t

f
o
r
c
e

(
N
)
(b)
Fig. 3.4: Time history of rear wheel-rail impact force due to a flat on
the rear wheel: (a) single cycle; and (b) three cycles.


contact force imposed on the other wheel. Since the distance between the two bogies of a
car is significantly larger than the distance between the wheels within a bogie, it is safe to
76
assume that defective wheel on one of the bogie wheels would impose only negligible
magnitudes of contact force on the other bogie wheels.
Figure 3.5 illustrates the variations in the contact force at the interface of the flat-
free front wheel and the rail. The results clearly show significantly high contact force at
the front wheel-rail interface, even though the front wheel is considered to be free of
defects. The peak impact force caused by the rear wheel flat approaches 115.5 kN. The
ratio of peak force to the static load is nearby 1.14. Unlike the impact force response at
the rear wheel contact, the impact force of the front wheel exhibits a smaller magnitude
peak prior to the sharp drop in contact force, which can be attributed to out-of-phase
motion of the two wheel-rail contact points. The oscillations in the contact force of the
front wheel tend to diminish over a short duration of approximately 0.04 s. The wheel-rail
contact force, however, continues to oscillate at a frequency near 56 Hz until an impact of
the rear wheel flat occurs during the subsequent cycle as shown in Fig. 3.5 (b). Similar
trends in the force response of a perfect wheel have also been reported by Cai [86]. The
impact force developed at the perfect wheel-rail contact due to a flat in another wheel is
also referred to as the cross-wheel force, and has been attributed to side frame pitch [37,
86].
The influence of location of a wheel flat in a bogie is further investigated by
considering a flat in the front wheel alone coupled with a flat free rear wheel. Figure 3.6
illustrates the variations in the contact forces developed at the front and rear wheels
interfaces with the rail. Owing to the symmetry of the bogie and the vehicle model about
the central lateral axis, a front wheel defect yields similar variations in the direct and
cross-wheel impact forces, as observed in Figs. 3.4 and 3.5. The impact load developed at
77
the flat free rear wheel contact is comparable with that of the flat-free front wheel caused
by rear flat, as shown in Fig. 3.5. This response is attributed to the oscillations of the rail
caused by front-wheel flat.
1.07 1.08 1.09 1.1 1.11 1.12 1.13 1.14 1.15 1.16
8
8.5
9
9.5
10
10.5
11
11.5
12
x 10
4
Time (s)
F
r
o
n
t

W
/
R

i
m
p
a
c
t

f
o
r
c
e

(
N
)

(a)
1.05 1.1 1.15 1.2 1.25 1.3 1.35 1.4
0.8
0.85
0.9
0.95
1
1.05
1.1
1.15
1.2
1.25
x 10
5
Time (s)
F
r
o
n
t

W
/
R

i
m
p
a
c
t

f
o
r
c
e

(
N
)

Fig. 3.5: Time history of the flat free front wheel-rail impact force in the presence of a
rear- wheel flat at v = 70 km/h: (a) single cycle; and (b) three cycles.



78
0.94 0.95 0.96 0.97 0.98 0.99 1 1.01
2
4
6
8
10
12
14
16
18
x 10
4
Time (s)
F
r
o
n
t

w
h
e
e
l

i
m
p
a
c
t

f
o
r
c
e

(
N
)

(a)

0.95 0.96 0.97 0.98 0.99 1 1.01
7.5
8
8.5
9
9.5
10
10.5
11
11.5
12
x 10
4
Time (s)
R
e
a
r

w
h
e
e
l

i
m
p
a
c
t

f
o
r
c
e

(
N
)

(b)

Fig. 3.6: Time histories of wheel-rail impact forces developed at front and rear wheels
due to a flat in the front wheel (v = 70 km/h;
f
L = 52 mm;
f
D = 0.4 mm):
(a) front wheel-rail impact force; (b) rear wheel-rail impact force.

The variations in front and rear wheel contact force responses over the same time
interval are compared in Fig. 3.7. The results clearly show a considerable phase
79
difference between the peak contact forces developed at the leading wheel with a flat and
the flat-free trailing wheel, which is mainly due to the transmission delay of the contact
force associated with the deflection of the continuous rail. The peak magnitude of the
cross-wheel force due to the front wheel flat occurs at a delay of 1.6 ms with reference to
the large magnitude peak at the front wheel-rail interface. For a wheelset axle spacing of
2.5 m, the transmission speed of peak contact force is about 1563 m/s. Such a high speed
could only be achieved by the wave propagation of the rail that mostly transmits the
impact load [37].
1.084 1.086 1.088 1.09 1.092 1.094 1.096 1.098 1.1 1.102
2
4
6
8
10
12
14
16
18
x 10
4
Time (s)
I
m
p
a
c
t

F
o
r
c
e

(
N
)


Rear wheel contact force (without flat)
Front wheel contact force (with flat)

Fig. 3.7: Time history of front and rear wheel-rail contact force


3.3.2 Force Responses of the Vehicle and Track Components

The high magnitude contact forces developed at the wheel-rail interface are
transmitted to the vehicle components and track layers, which may cause components
fatigue or failure. The transmission of contact force to the sideframe occurs through the
bearing adapters. The deflections of wheel and side frame give rise to the primary
80
suspension force, also referred to as bearing force, which is eventually transmitted to the
car body. The magnitudes of bearing forces strongly depend upon the nature of flat,
speed, wheel load and suspension properties. The variations in bearing force are
investigated using the model parameters listed in Table 3.2 and a flat within the rear
wheel (
f
L = 52 mm;
f
D = 0.4 mm). The responses, evaluated under three different load
conditions (W = 63, 82, and 102 kN), are compared in Fig. 3.8. The bearing force tends
to rapidly decrease from its static level of 100 kN as the wheel flat enters the wheel-rail
contact region, as it was observed for the wheel-rail contact force in Fig. 3.6. Similar to
the contact force, the bearing force increases to a high value of nearly 160 kN, although
the rate of rise is considerably smaller than that of the contact force. The bearing force
decays in an exponential manner, thereafter, which is attributed to the primary suspension
damping.
The results further show that variations in the wheel load do not influence the
oscillation frequency of the bearing force. The ratio of peak bearing force to static
bearing force varies nonlinearly with the wheel load. These ratios for wheel loads of 102,
82 and 63 kN are obtained as 1.59, 1.73, 1.93, respectively. The nonlinear variations in
the peak bearing force with the wheel load are mostly attributed to the nonlinear
variations in the peak wheel-rail impact force with the static wheel load, which is further
attributed to the nonlinear wheel-rail contact model. The ratios of peak wheel-rail impact
load to static wheel load were obtained as 1.77, 1.96, and 2.45 for static loads of 102, 82,
and 63 kN, respectively.
81
0.81 0.815 0.82 0.825 0.83 0.835 0.84 0.845 0.85 0.855 0.86
2
4
6
8
10
12
14
16
x 10
4
Time (s)
B
e
a
r
i
n
g

f
o
r
c
e

(
N
)


W=102 kN W=82 kN W=63 kN
Fig. 3.8: Variations in the bearing force response due to a rear-wheel flat
as a function of static wheel load.

The dynamic force developed by the discrete rail pads, also referred to as the
reaction force at each rail-sleeper interface, is further evaluated to study the nature of
forces transmitted to different track layers. The dynamic rail pad force is computed from
relative deflection of the rail and sleeper, using Eq. (2.7). As an example, the variations in
the rail pad force developed at sleeper no. 22, due to a flat within the rear wheel, are
illustrated in Fig. 3.9. The variations in the railpad force are investigated using the model
parameters listed in Table 3.2 and a flat within the rear wheel (
f
L = 52 mm;
f
D = 0.4
mm), and three different load conditions (W=63, 82, 102 kN) at a constant forward
vehicle speed of 70 km/h. Figure 3.9 shows that as the wheel flat enters the wheel-rail
contact region, the pad force rapidly decreases from its static level of 82.23, 66.88, and
51.57 kN under static wheel loads of 102, 82, and 63 kN, respectively. The pad force then
increases to peak value of 120.4, 104.9, and 91.64 kN, respectively. The ratios of the
peak pad force to static pad force are obtained as 1.46, 1.57, and 1.77 for static wheel
82
load of 102, 82, and 63 kN, respectively. The figure further shows that the pad force
gradually decreases after the impact, as the wheel moves away from the location of the
sleeper. It can be seen that the variations in the wheel load do not influence the oscillation
frequency of the pad force. The results also show that the magnitudes of pad forces are
significantly smaller than those of the contact forces, which can be attributed to inertia
force of the rail.
0.676 0.678 0.68 0.682 0.684 0.686 0.688 0.69 0.692 0.694 0.696
2
3
4
5
6
7
8
9
10
11
12
x 10
4
Time (s)
P
a
d

f
o
r
c
e

(
k
N
)


W=102 kN W=82 kN W=63 kN

Fig. 3.9: Variations in railpad force due to a rear wheel flat as a function of
static wheel load.

The impact force developed at wheel-rail interface due to a wheel flat is also
transmitted to the ballast blocks, which is ultimately transmitted to the ground. The force
developed at the sleeper-ballast interface is known as ballast force. The magnitudes of
ballast forces also depend upon the wheel load, ballast properties and the nature of the
flat. The effect of wheel flat on dynamic ballast force is investigated in the presence of a
single flat in the rear wheel. The resulting variations in the ballast force are shown in Fig.
83
3.10. The ballast force is computed by summing up the reaction forces obtained from a
particular sleeper and the shear forces arising from couplings with the adjacent ballast
blocks, as shown in Eq. (2.10). The time histories of the ballast forces for ballast no. 22
under three different load conditions (W= 63, 82 and 102 kN) are shown in Fig. 3.10. The
results suggest that variations in ballast forces are similar to those observed for the pad
forces shown in Fig. 3.9. The peak magnitudes of the ballast forces, however, are
considerably smaller than those of the pad forces. The peak values of forces are attained
as 108.4, 92.49 and 75.62 kN under static wheel loads of 102, 82 and 63 kN, respectively.
The ratios of peak to static ballast forces for wheel loads of 102, 82, and 63 kN are
obtained as 1.32, 1.38, and 1.46, respectively. The normalized ballast forces relatively
less sensitive to variations in the static wheel load in the range of variations considered.
The figure further shows that the ballast force gradually decreases after the wheel impact,
as the wheel moves away from the location of the ballast. The variations in the wheel
loads do not show significant influence on the oscillation frequency of the ballast force,
as observed for the wheel-rail contact and bearing forces.
84
0.68 0.685 0.69 0.695 0.7
1
2
3
4
5
6
7
8
9
10
11
x 10
4
Time (s)
B
a
l
l
a
s
t

F
o
r
c
e

(
N
)


W=82 kN W=102 kN W=63 kN
Fig. 3.10: Variations in ballast force due to a rear wheel flat as a function of
static wheel load.


3.3.3 Displacement Responses of the Vehicle-Track Components
The deflection responses of different vehicle and track system components are
further investigated to study the transmission of wheel flat-induced rail deflections to
various vehicle components and track layers, as a vehicle moves over the track with a
particular load and speed. Excessive deflections may cause fatigue damage of the
vehicle-track components, especially the track layers leading to increase maintenance
cost or track failure. The deflections of the vehicle and track components are evaluated
under constant static wheel load of 102 kN in the presence of a rear wheel flat (
f
L = 52
mm;
f
D = 0.4 mm).
Figure 3.11 and 3.12 illustrate the variations in displacements responses of wheel-
rail contact points at front and rear wheels, respectively. The results show deflections of
the wheel masses and deflections of rail at wheel-rail contact points. The application of
the static wheel load causes the rail to deform downwards at the contact point. However,
85
a relaxation of rail compression prior to the impact at the contact point is evident as the
flat approaches the contact region, while the wheel moves downwards due to its flat
geometry. This is followed by further compression of the rail due to impact force
developed at the wheel-rail interface, as seen in Fig. 3.11. Subsequently, the rail profile
tends to recover its steady value, while the deflection response exhibits oscillation near
56 Hz, which has been referred to as the coupled vehicle-track system resonance [10, 43,
73, 87]. Both the wheel and rail displacement responses exhibit oscillation at the same
frequency. The peak deflection of the rail under interaction with the rear wheel flat
approaches 1.715 mm. The steady deflection of the rail under the static wheel load is in
the order of 1.582 mm.
The impact forces caused by the rear wheel flat also yields deflections of flat-free
front wheel and the corresponding wheel-rail contact point, as shown in Fig. 3.12,
although the peak deflections are relatively small. This is attributed to the influence of the
cross wheel impact force. The results further show that peaks in deflections responses of
the wheel and the rail occur at same instants. Furthermore, in the absence of flat, the
difference between the displacements of the wheel and the rail remains nearly constant
over the entire revolution of the wheel.

86
0.85 0.9 0.95 1 1.05 1.1
-1.8
-1.7
-1.6
-1.5
-1.4
-1.3
x 10
-3
D
i
s
p
l
a
c
e
m
e
n
t

(
m
)
Time (s)


Wheel Displacement
Rail Displacement
Fig. 3.11: Time histories of vertical displacements of the rear wheel and rail
in the presence of a flat
0.85 0.9 0.95 1 1.05
-1.75
-1.7
-1.65
-1.6
-1.55
-1.5
-1.45
x 10
-3
Time (s)
D
i
s
p
l
a
c
e
m
e
n
t

(
m
)


Wheel Displacement Rail Displacement
Effect of other wheel flat
Effect of other wheel flat

Fig. 3.12: Time histories of vertical displacements of front wheel (no flat) and rail at the
wheel-rail contact point.

87
The wheel and rail displacement responses at the contact point are compared with
the responses reported by Sun et al. [73], as shown in Fig. 3.13. In the reported study, the
displacement responses of the wheel and rail were evaluated under a constant speed of 70
km/h and static wheel load of 106 kN, using a pitch plane model of the vehicle coupled
with 4-layer track model. The responses were evaluated in the presence of a 40 mm long
and 0.35 mm deep flat. The present model was also evaluated under identical flat
geometry, forward speed and static wheel load. The computed wheel and rail
displacement responses are shown in Fig. 3.13 (a), while the reported results are shown in
Fig. 3.13 (b). The results presented for three consecutive impacts, generally show good
agreements with the responses reported in [73]. The static deflection of the wheel mass
predicted by both models is in the order of 1.78 mm. The initial displacement peaks in the
rail displacement response, prior to the impact, predicted by both models are in the order
of 1.4 mm. Although the displacement responses of both the rail and wheel during the
entire period of simulation compare very well with those reported by Sun et al. [73], the
magnitudes of peak displacements encountered during impact differ. Such differences are
mostly likely attributed to differences in the track parameters and the vehicle models used
for the analyses.

88
0.75 0.8 0.85 0.9 0.95 1 1.05 1.1 1.15
-2
-1.9
-1.8
-1.7
-1.6
-1.5
-1.4
-1.3
x 10
-3
Time (s)
W
h
e
e
l

a
n
d

r
a
i
l

d
i
s
p
l
a
c
e
m
e
n
t
s

(
m
)


Wheel displacement Rail displacement

(a)


(b)
Fig. 3.13: Comparison of wheel and rail displacement responses of the present
model with those reported by Sun et al. [73]: (a) present model; and (b) reported
results (v = 70 km/h;
f
L = 40 mm;
f
D = 0.35 mm).

Fig. 3.14 (a) illustrates the variations in displacement response at a single point on
the rail, while flat (
f
L = 52 mm;
f
D = 0.4 mm) is assumed to exist in the rear wheel.
Figures 3.14 (b) and (c) illustrate the resulting variations in displacement responses of the
89
sleeper and ballast block (i= 22) located beneath the point considered on the rail. The
results clearly show the differences between peak deflections of the point on the rail,
sleeper and ballast under the front and rear wheel contacts. The point on the rail
undergoes a peak deflection of 1.58 mm under the front wheel contact, which increases to
1.715 mm under the rear wheel. Similarly, the peak deflections of the sleeper under front
and rear wheel contacts are obtained as 0.88 mm and 0.99 mm, respectively, while those
of the ballast are 0.46 mm and 0.5 mm. While the deflection response at the front wheel-
rail contact point is affected by the rear-wheel flat, the effect is not evident in Fig. 3.14
(a) due to the relatively large ordinate scale. This effect, however, is evident in the ballast
response as shown in Fig. 3.14 (c). The rail, sleeper, and ballast deflections exhibit small
fluctuations every time as the deflections approach their steady values. These are
attributed to impact forces occurring at adjacent locations of the rail. The results also
show slight rail lift-off from the sleeper, sleeper lift-off from the ballast and ballast lift-
off from the sub-grade occur in the order of 0.052 mm, 0.0229 mm, and 0.0073 mm,
respectively. These lift-offs can be attributed to the dynamics of the moving load that
separate the track components from each other.
The displacement responses of different vehicle system components due to a
wheel flat are also evaluated, as shown in Fig. 3.15 for the car body bounce, and Fig. 3.16
for the bogie bounce and pitch. The car body bounce motion, shown in Fig. 3.15, reveals
that a wheel flat does not affect the car body mass deflection response. This is attributable
to very low natural frequency the car body, near 2.75 Hz. Consequently, the car body
mass has been considered as a static load acting on the bogie in many studies [5, 6, 22,
23, 28, 37].
90
0.165 0.294 0.423 0.552 0.681 0.81 0.939
-5
-3
-1
0
Time (s)
w
b
0.162 0.291 0.42 0.549 0.678 0.807 0.936 1.04
-10
-8
-6
-4
-2
0
w
s
R
a
i
l

(
m
m
)
W
r

(
1
3
.
2
,
t
)
(
m
m
)
(
m
m
)
(
m
m
)
0.16 0.2895 0.419 0.5485 0.678 0.8075 0.937 1.04
-15
-10
-5
0
w
r

(
1
3
.
2
,
t
)
Wheelset
distance
Front wheel without def ect
Rear wheel with def ect
Rail lif t of f
Sleeper lif t of f
Rear wheel with def ect
Front wheel without def ect
Ballast lif t of f
Front wheel without def ect
Rear wheel with def ect
(a)
(b)
(c)

Fig. 3.14: Displacement responses evaluated at (a) a point on the rail; (b) sleeper;
and (c) ballast beneath the rail point in the presence of a rear wheel flat



91
0 0.5 1 1.5 2 2.5 3
-2.5
-2
-1.5
-1
-0.5
0
x 10
-3
Time (s)
D
i
s
p
l
a
c
e
m
e
n
t

(
m
)

Fig. 3.15: Time history of car body vertical motion in the presence of
a rear wheel flat (
f
L = 52 mm;
f
D = 0.4 mm)


The presence of a wheel flat, however, significantly influences both the bounce
and pitch deflection responses of the bogie, as shown in Fig. 3.16. Both the vertical and
pitch deflections of the bogie increase rapidly as the flat interacts with the rail. This is
caused by relatively high magnitude bearing forces transmitted to the side frame, as
observed in Fig. 3.8. As the flat moves away, the deflection responses oscillate around
the static equilibrium near the frequency of 56 Hz. In the presence of a flat, the peak
vertical deflection of the bogie increases from the static value of 1.816 mm to 1.904 mm.
A rapid change in the pitch response of the bogie also occurs at the time when the
defective wheel strikes the rail. The peak pitch angle, however, is relatively small, in the
order of 0.810
-6
rad. The results suggest important effects of wheel flat on the vertical
deflection of the bogie mass, which would be further influenced by an array of factors
related to operating variables and flat geometry.
92
0.65 0.7 0.75 0.8 0.85 0.9 0.95 1
-1.92
-1.9
-1.88
-1.86
-1.84
-1.82
-1.8
-1.78
-1.76
x 10
-3
Time (s)
B
o
g
i
e

D
i
s
p
l
a
c
e
m
e
n
t

(
m
)
(a)
0.65 0.7 0.75 0.8 0.85 0.9 0.95 1
-10
-8
-6
-4
-2
0
2
4
6
8
x 10
-5
Time (s)
B
o
g
i
e

p
i
t
c
h

r
o
t
a
t
i
o
n

(
r
a
d
)


(b)
Fig. 3.16: Variation in the bounce and pitch responses of the bogie in the presence of a
rear wheel flat (
f
L = 52 mm;
f
D = 0.4 mm): (a) bounce motion; (b) pitch motion.








93
3.4 SUMMARY
The coupled vehicle-track model formulated in this dissertation is validated in the
presence of a wheel flat using the reported analytical and measured data on the resulting
impact force. The comparison of wheel-rail impact force responses of the model with the
available data revealed good agreements. The effects of a single wheel flat on the
responses of vehicle and track components in terms of impact loads and displacements
are further investigated for both defective wheel and the flat-free wheel. The results
suggest that a simpler Euler beam rail model can provide reasonably good prediction of
the contact force and displacement responses of the vehicle-track system in the presence
of a flat. The characteristics of the wheel-rail impact loads, and bearing, railpad, and
ballast forces due to a haversine wheel flat were also investigated as a function of the
static wheel load. The results clearly show that the magnitudes of these transmitted forces
increase considerably under a flat-induced excitation, which may cause rapid fatigue of
the vehicle and track components. The magnitudes of these impact forces could be
approaches twice the static load under medium wheel loads. The results also suggested
significant influences of the wheel flat on displacement responses of the vehicle and track
system components except for the low natural frequency car body. The study also
revealed that a flat in one of the bogie wheels would also influence the forces developed
at the adjacent wheel-rail contact point. A defective wheel may thus cause damage to
wheels of the adjacent wheelset. The magnitude of this cross-wheel force depends on the
vehicle, and track properties, and the flat geometry. It is shown that a flat in a wheel not
only affects the wheel-rail deflections but also the deflections of the sleepers and the
ballast. A comprehensive parametric study is undertaken in the subsequent chapter in
94
order to investigate the effect of vehicle-track model design and operating parameters on
the impact force response.









































95
CHAPTER 4

PARAMETRIC STUDY
4.1 INTRODUCTION

The presence of defects in the wheel can cause high magnitude impact forces on
the wheel-rail interface. These high impact forces are transmitted to vehicle-track
subsystems, and may lead to rapid component fatigue. Among the several types of wheel
defects, only wheel flat is considered in the present study. As mentioned in chapter 1,
wheel flat not only causes the high frequency vibration in the vehicle track system but
also affects the track maintenance and the reliability of the vehicle's rolling elements. The
vast majority of studies dealing with the wheel flat utilize simple vehicle models and
consider single flat on a wheel. The presence of multiple flats within a wheel or axle of a
car is not very uncommon in practice. In case of multiple flats, the sizes and locations of
the flat would play an important role on the peak wheel-rail impact loads. Quantitative
analyses of dynamic impact loads in both wheel and rail for single and multiple flats are
therefore important in predicting the safe operating limits.
The magnitude of impact forces imparted on the vehicle and track components,
however, depend upon various design and operating parameters. In this chapter, the
proposed vehicle-track system model is applied to study the effects of selected design and
operating parameters on the vehicle-track responses due to single as well as multiple
flats. Responses in terms of wheel-rail impact load and forces transmitted to bearings,
pad and ballast are evaluated in an attempt to identify desirable design and operating
factors. The parametric study includes variations in selected vehicle, track, operational as
well as flat parameters. The two-dimensional coupled vehicle-track model, developed in
96
chapter 2 and validated in chapter 3, is considered as the baseline model. The parameter
effects are evaluated by varying a single parameter at a time, while the others are held
constant and equal to their nominal values.
4.2 SELECTION OF IMPORTANT MODEL PARAMETERS

In reality, the vehicle and track system parameters may vary widely depending on
the type, size and its application. A few studies have suggested that some of the vehicle
and track design parameters could strongly influence the peak impact loads, imparted on
the rail by a wheel flat [10, 37, 86], although the influences of various vehicle and track
parameters have not been characterized over representative ranges of operating
conditions. The reported studies have mostly emphasized the effect of speed, which has
significant influence on the peak wheel-rail impact force. Apart from speed, the flat-
induced impact forces strongly depend upon the wheel load, as it was shown in section
3.3.2. The effects of various design parameters therefore must be evaluated over
representative ranges of speeds and operating loads. The maximum allowable speed of a
freight car in Canada and USA is 108 km/h and 190 km/h, respectively [86]. The
influences of selected parameters on the nature of impact loads are thus evaluated in 10 to
120 km/h speed range, while nominal speed of 70 km/h is considered. The load per wheel
for a freight train may also vary from a fully loaded condition to an empty car. This
variation is in the range of 34.325-117.68 kN in Sweden [5], and 3.11-146.8 kN in
Canada [43]. The parametric study is performed under static wheel load in the range of
63-122 kN.
Apart from the load and speed variations, the geometry of the flat may also vary
considerably. The range of flat is chosen in accordance with the threshold values for
97
wheel removal by different railway organizations. Variations may occur not only in the
length but also in the depth of the flat. The AAR [91] criteria for removal of wheel from
service requires that a railway car with a wheel with a 50.8 mm long single flat or
38.1mm long each of the two adjoining flats cannot continue to be in service. According
to Swedish Railway, the condemning limit for single wheel flat is 40 mm long and 0.35
mm deep [10]. Transport Canada safety regulations [108] require that a railway company
may not continue a car in service if a wheel has a slid flat spot that is more than 63.50
mm in length or two adjoining flat spots each of which is more than 50.80 mm. In order
to examine the stipulated guidelines, the study considers the flat length range of 30-80
mm, while the nominal length is chosen as 52 mm. The flat depth range is covered from
0.2-0.7 mm, where the nominal value is considered as 0.4 mm.
The analysis of wheel-rail contact forces greatly relies upon the number of modes
considered in modeling of the continuous rail beam. The length of the rail should also be
sufficiently large in order to minimize the contribution due to end conditions. Zhai et el.
[53] suggested that a good convergence of the solution can be obtained for modes equal
to or exceeding 60. Many studies have considered 100-120 modes of the rail beam to
reduce the effect of boundary conditions [13, 17, 73]. In the present study, the solutions
are obtained by considering a total of 100 rail modes. The distance between two
consecutive sleepers also has a strong influence on the overall vehicle-track system
response, which may vary from 0.54 to 0.79 m [3, 35], although the vast majority of
studies consider sleeper spacing in 0.6-0.7 m range [13, 37, 45, 85].
The Hertzian contact coefficient also differs widely in different reported studies,
which can greatly influence the magnitude of wheel-rail impact force. The contact
98
coefficient value in the present study is estimated from the Eq. (2.18). Other parameters
of track system model, such as the pad and ballast stiffness, damping coefficient, and rail
mass per unit length are selected from [3] based on three-layer track model. While some
of these parameters of the vehicle-track system model are varied, others are held constant
(Table 4.1).
4.3 INFLUENCE OF WHEEL FLAT

Depending on the size of the flat, the peak dynamic load created by the wheel flat
can be as high as 2-3 times the static load. Multiple wheel flats within a wheel or axle of
a freight car are also quite common in practice. A single wheel with multiple flats can
impart very high frequency impact loads, while the magnitude and frequency of the
impact load would depend upon the number of flats and their relative positions.
Furthermore, for a given phase difference between two flats, the forward speed may also
play an important role. The impact forces arising from different possible locations of the
flats may not be same. With these considerations, a detailed study on wheel flat
parameters is conducted, using the baseline model parameters listed in Table 4.1.
4.3.1 Effect of single wheel flat on wheel-rail impact force
In the presence of a flat in a wheel, a sharp change occurs in the profile of the
wheel, as described earlier in section 2.4. Figure 4.1 illustrates the variations in radius of
a wheel with a 52 mm long and 0.4 mm deep flat, as a function of wheel angular position.
The influence of flat size on the impact force is investigated by considering three
different flat sizes for the front wheel alone: (i) Flat 1-
f
L = 45 mm;
f
D = 0.3 mm; (ii)
Flat 2-
f
L = 52 mm;
f
D = 0.4 mm; (iii) Flat 3-
f
L = 58 mm;
f
D = 0.5 mm.

99
Table 4.1: Nominal simulation parameters used for parametric study [3, 37]


Notation

Parameter Value Type Range
c
M Car body mass (quarter), kg 19400

Variable

11350-
23500
t
M Bogie mass (half), kg 500 Fixed
w
M Wheel mass, kg 500 Variable 300-900
t
J Bogie mass moment inertia, kg-m
2
176 Fixed
r
m Rail mass per unit length, kg/m 60.64 Variable 30-100
EI Rail bending stiffness, MN-m
2
6.62 Variable 4-8
s
M Sleeper mass, kg 118.5 Variable 80-140
b
M
Ballast mass, kg 739 Variable 600-900
1 s
K Primary suspension stiffness, MN/m 788

Variable
500-
1000
1 s
C Primary suspension damping, kN-s/m 3.5

Variable 1-6
2 s
K Secondary suspension stiffness, MN/m 6.11

Variable 4-9
2 s
C Secondary suspension damping, kN-s/m 158

Variable 130-180
p
K Railpad stiffness, MN/m 120 Variable 60-180
b
K Ballast stiffness, MN/m 182 Variable 120-240
w
K Ballast shear stiffness, MN/m 147 Fixed
f
K Subgrade stiffness, MN/m 78.4 Fixed
p
C Railpad damping, kN-s/m 75 Variable 50-90
b
C Ballast damping, kN-s/m 58.8 Variable 40-90
w
C Ballast shear damping, kN-s/m 80 Fixed
f
C Subgrade damping, kN-s/m 31.15 Fixed
H
C Hertz spring constant, N/m
3/2
8710
9
Fixed

t
l Wheelset distance, m 1.25 Fixed
R Wheel radius, m 0.42 Fixed
f
L Flat length, mm 52 Variable 30-80
f
D Flat depth, mm 0.4 Variable 0.2-0.7
l Length of the rail, m 60.00 Variable 40-80
l
s
Sleeper spacing, m 0.6 Variable 0.4-0.8
100
0 40 80 120 160 200 240 280 320 360
0.4195
0.4196
0.4197
0.4198
0.4199
0.42
Angular position (Degree)
R
a
d
i
u
s

(
m
)

Fig. 4.1: Variations in radius of a wheel with single flat (
f
L = 52 mm
and
f
D = 0.4 mm) as a function of angular position of the contact.

The impact force responses at front and rear wheel-rail (W/R) contacts obtained
for the vehicle running at a speed of 70 km/h are shown in Fig. 4.2. The figures illustrate
that the peak W/R impact force increases as the size of the flat increases. The ratios of
this peak impact load to the static wheel load are 1.67, 1.77, and 1.87 for 45 (flat 1), 52
(flat 2), and 60 mm (flat 3) long flats, respectively. The magnitudes of the secondary peak
also increase with flat sizes. The presence of the flat in the front wheel also influences the
W/R impact load at the rear wheel, while the variations are relatively small, as shown in
Fig. 4.2 (b). The ratios of the peak impact load to the static load are 1.134, 1.149, and
1.154 for the three flats considered, respectively. The small differences in these ratios can
be attributed to the small variations among the size of the flats.
101
1.495 1.5 1.505 1.51 1.515 1.52
0
2
4
6
8
10
12
14
16
18
20
x 10
4
Time (s)
F
r
o
n
t

W
/
R

I
m
p
a
c
t

F
o
r
c
e

(
N
)


Flat1 Flat2 Flat3
(a)
1.495 1.5 1.505 1.51 1.515
0.8
0.85
0.9
0.95
1
1.05
1.1
1.15
1.2
1.25
x 10
5
R
e
a
r

W
/
R

I
m
p
a
c
t

F
o
r
c
e

(
N
)
Time (s)


Flat1 Flat2 Flat3
(b)
Fig. 4.2: Influence of size of a front wheel flat on the impact force responses at
the wheel-rail interface: (a) front wheel; and (b) rear wheel.



102
The magnitude of the impact force, however, is strongly dependent upon the
vehicle forward speed. Figure 4.3 illustrates the variations in the peak W/R impact force
with speed in the range of 10-120 km/h. The results are presented for both front and rear
wheel impact forces under the influence of a 52 mm (flat 2) flat on the front wheel, while
static wheel load of 102 kN is considered. The figure illustrates that wheel-rail impact
load increases with inceasing the speed of the vehicle. The rate of increase, however, is
considerably larger in the defective front wheel conact force. The front wheelrail impact
force exhibits higher magnitude in the lower speed range (2030 km/h) and tends to
increase rapidly at speeds above 80 km/h. The relatively higher magnitude of peak impact
force near 25 km/h has been attributed to elasticity of the ballast [35]. The results shown
in Fig. 3.4 are consistent with the results reported in [3, 23, 42, 43, 86]. The cross-wheel
impact force developed at rear wheel-rail contact point increases only slightly with
vehicle speed, while the effect of front wheel flat becomes evident at speeds above 90
km/h.
0
50
100
150
200
250
300
350
400
10 20 25 30 40 50 60 70 80 90 100 110 120
Speed (km/h)
P
e
a
k

W
/
R

I
m
p
a
c
t

f
o
r
c
e

(
k
N
)
Front wheel impact force Rear wheel impact force
Fig. 4.3: Effect of speed on peak front and rear wheel impact loads due to a single flat on
the front wheel (
f
L = 52 mm and
f
D = 0.4 mm)
103
Figure 4.4 illustrates the peak displacement response of the rail at the contact
point between the defective wheel and the rail as a function of forward speed. The results
show peak rail displacement near the low speed of 25 km/h, which causes higher
magnitude of contact force, as observed in Fig. 4.3. The peak rail deflection decrease
with increase in speed at speeds above 25 km/h. The peak rail displacement reaches a
peak in the order of 1.877 mm, where the static deflection of the rail is 1.58 mm.
1.55
1.6
1.65
1.7
1.75
1.8
1.85
1.9
1.95
10 20 25 30 40 50 60 70 80 90 100 110 120
Speed (km/h)
P
e
a
k

r
a
i
l

d
i
s
p
l
a
c
e
m
e
n
t

(
m
m
)
W=102 kN

Fig. 4.4: Effect of speed on the peak displacement of the rail at the front wheel-rail
contact point (
f
L = 52 mm and
f
D = 0.4 mm)

4.3.2 Effect of multiple wheel flats on W/R impact loads
Although multiple flats in a single wheel are widely observed, the resulting
impact forces have not been reported. In this section, the pitch-plane vehicle-track system
model is applied to study the influences of multiple flats. The analyses are performed for
two different cases: (i) two identical size flats on the same wheel located at different
angular positions (phase angle) (ii) two same size flats on the different wheels either in
phase or out-of-phase. The flats are defined such that the leading flat comes into contact
with the rail first followed by trailing flat, which enters the contact with a defined phase
104
angle. The variations in radius of a wheel with two flats that are 90
0
apart are illustrated
in Fig. 4.5, as a function of the wheel position.
0 50 100 150 200 250 300 350
0.4195
0.4196
0.4197
0.4198
0.4199
0.42
Angular position (Degree)
R
a
d
i
u
s

(
m
)
Leading flat
Trailing flat
Fig. 4.5: Variations in radius of a wheel with two same size flats (
f
L = 52 mm
and
f
D = 0.4 mm), which are 90
0
apart.

Wheel-rail impact load due to multiple flats on single wheel
Two same size flats with specified phase angles are considered to study
the effect of location of multiple flats on the wheel-rail impact load. Both flats are 52 mm
long and 0.4 mm deep. The analyses are performed under flats within the rear wheel only,
while the front wheel is considered to be perfect. The variations in the W/R impact loads
are evaluated by considering different phase angles between the two flats, namely, 45
0
,

90
0
, 135
0
and 180
0
. Figures 4.6 to 4.9 illustrate the variations in W/R impact forces at the
rear wheel-rail contact for different phase angles, respectively, over two consecutive
cycles. The results clearly show two distinct peaks in the impact force response during
each cycle occur as the two flats strike the rail. Although two flats are identical in size,
the magnitudes of the impact forces induced by the two flats differ. The peak impact load
induced by the first flat is in the order of 180.6 kN irrespective of the position of the
second flat. The magnitude of the peak impact force induced by the second flat tends to
105
be smaller, when the flats are 45
0
apart. This difference is in the order of 4.8 kN, while
the time lapse between the two peak forces directly related to the vehicle speed and the
phase difference between the two flats. The difference in the magnitudes of the impact
forces due to two flats decreases as the phase angle increases. For 90
0
phase angle, this
difference is observed to be in the order of 3.5 kN. This difference diminishes, when the
flats are placed more than 90
0
apart, as it is evident in Fig. 4.8 and 4.9, for 135
0
and 180
0
phase angles, respectively. The results suggest that the magnitudes of impact forces
induced by two flats are similar to that caused by a single flat. The frequency of the
impact force, however, increases with increasing the number of flats, which could yield
rapid fatigue of the vehicle and track components.
0.65 0.7 0.75 0.8 0.85 0.9 0.95 1 1.05
0.2
0.4
0.6
0.8
1
1.2
1.4
1.6
1.8
2
x 10
5
Time (s)
W
/
R

I
m
p
a
c
t

F
o
r
c
e

(
N
)
Trailing flat
Leading flat
Fig. 4.6: Time response of rear wheel-rail impact force with two flats at 45
0
phase angle
106
0.65 0.7 0.75 0.8 0.85 0.9 0.95 1 1.05
0.2
0.4
0.6
0.8
1
1.2
1.4
1.6
1.8
2
x 10
5
Time (s)
W
/
R

I
m
p
a
c
t

F
o
r
c
e

(
N
)
Leading flat
Trailing flat
Fig. 4.7: Time response of rear wheel-rail impact force with two flats at 90
0
phase angle
0.65 0.7 0.75 0.8 0.85 0.9 0.95 1 1.05
0.2
0.4
0.6
0.8
1
1.2
1.4
1.6
1.8
2
x 10
5
Time (s)
W
/
R

I
m
p
a
c
t

F
o
r
c
e

(
N
)
Leading flat
Trailing flat
Fig. 4.8: Time response of rear wheel-rail impact force with two flats at 135
0
phase angle
107
0.65 0.7 0.75 0.8 0.85 0.9 0.95 1 1.05
0.2
0.4
0.6
0.8
1
1.2
1.4
1.6
1.8
2
x 10
5
Time (s)
W
/
R

I
m
p
a
c
t

F
o
r
c
e

(
N
)
Leading flat
Trailing flat
Fig. 4.9: Time response of rear wheel-rail impact force with two flats at 180
0
phase angle
The variations in the impact forces induced by multiple flats on a wheel would
strongly depend upon the speed of the vehicle. The time lapse between the peak forces
caused by two flats will particularly decrease with an increase in vehicle speed. The
effect of speed on the wheel-rail impact load due to multiple wheel flats is therefore
examined in the speed range of 0-120 km/h. The peak magnitude of impact forces caused
by two flats with phase angles of 45
0
, 90
0
and 135
0
were investigated. The peak forces
caused by the two flats were very similar as observed in Figs. 4.6-4.9, irrespective of the
vehicle speed. The variations in the peak impact forces with speed were thus very similar
to those observed for a single wheel flat in Fig. 4.3.

Wheel-rail impact load due to flats in both wheels
A single flat in one wheel may be followed by another flat of same or different
size in the adjacent wheel. Their relative positions can also be different. The positions of
a flat on two different wheels may be same i.e. in-phase or different i.e. out-of-phase. The
108
effect of multiple wheel flats on the wheel-rail impact loads for both in-phase and out-of-
phase conditions are presented in this subsection.

Figure 4.10 illustrates the time history of the front and rear wheel-rail impact load
due to a 52 mm long and 0.4 mm deep single flat on both front and rear wheels in same
location i.e. flats are in phase. It can be seen that the magnitude of peak front and rear
wheel-rail impact forces are in the order of 163.9 kN. This peak is approximately 16 kN
less than the peak magnitude of wheel-rail impact load in the presence of single wheel
flat of same size. This difference can be attributed to the symmetry of the bogie and the
vehicle model about the central lateral axis due to the presence of flats on both wheels at
the same position. In the presence of flats on both wheels, both front and rear wheels
move together while the pitch motion of the bogie is not excited. On the other hand, in
case of single wheel flat, owing to the pitch motion of the bogie, the deflections of wheel
is more, while the rail deflection remains same. These motions of the wheels increase the
wheel-rail overlap in case of a single wheel flat that ultimately increases the wheel-rail
impact load. A further analysis on impact load due to flats with phase angles is carried
out to examine the loads at the impacting wheel as well as the adjacent wheel. In the
presence of a phase, the pitch dynamics of the bogie is expected to increase the impact
load in comparison to those of in-phase presented in Fig. 4.10.
109
0.4 0.45 0.5 0.55 0.6
0.2
0.4
0.6
0.8
1
1.2
1.4
1.6
1.8
2
x 10
5
Time (s)
Time (s)
F
r
o
n
t

W
/
R

I
m
p
a
c
t

F
o
r
c
e

(
N
)
0.4 0.45 0.5 0.55 0.6
0.2
0.4
0.6
0.8
1
1.2
1.4
1.6
1.8
2
x 10
5
Time (s)
R
e
a
r

W
/
R

I
m
p
a
c
t

F
o
r
c
e

(
N
)
Fig. 4.10: Time history of front and rear wheel impact force responses
due to a single flat on both wheels in phase.




110
The time histories of contact forces at front and rear wheel-rail contact points due
to a single flat on both the front and rear wheels at 45
0
, 90
0
, 135
0
and 180
0
out-of-phase
are shown in Figs. 4.11 to 4.14, respectively. The rear wheel flat is assigned ahead of the
front wheel flat with a specific phase angle. The impact forces are evaluated at a constant
speed of 70 km/h. As shown in Fig. 4.11 for 45
0
phase angle, the difference between the
peak impact force due to the direct and cross wheel flat is pronounced. The figures
illustrate two distinct peak impact forces corresponding to the positions of the respective
flats. The results demonstrate a very little effect of one wheel flat on the impact force

0.4 0.45 0.5 0.55 0.6
0.2
0.4
0.6
0.8
1
1.2
1.4
1.6
1.8
2
x 10
5
Time (s)
F
r
o
n
t

w
h
e
e
l

c
o
n
t
a
c
t

f
o
r
c
e

(
N
)

Impact load due to rear wheel flat

0.4 0.45 0.5 0.55 0.6
0.2
0.4
0.6
0.8
1
1.2
1.4
1.6
1.8
2
x 10
5
Time (s)
R
e
a
r

w
h
e
e
l

i
m
p
a
c
t

f
o
r
c
e

(
N
)
Impact load due to front wheel flat

Fig. 4.11: Front and rear wheel impact force responses due to a single flat on both wheels
at 45
0
out-of-phase (rear wheel flat ahead 45
0
)

111
0.4 0.45 0.5 0.55 0.6
0.2
0.4
0.6
0.8
1
1.2
1.4
1.6
1.8
2
x 10
5
Time (s)
F
r
o
n
t

w
h
e
e
l

C
o
n
t
a
c
t

F
o
r
c
e

(
N
)
Impact load due to Rear wheel flat

0.4 0.45 0.5 0.55 0.6
0.2
0.4
0.6
0.8
1
1.2
1.4
1.6
1.8
2
x 10
5
Time (s)
R
e
a
r

w
h
e
e
l

C
o
n
t
a
c
t

F
o
r
c
e

(
N
)
Impact load due to Front wheel flat

Fig. 4.12: Front and rear wheel impact force responses due to a single flat on both wheels
at 90
0
out-of-phase (rear wheel flat ahead 90
0
)

generated at the other wheel-rail contact point. The magnitudes corresponding to flat at
each wheel are same as those obtained for a single wheel flat earlier. The results indicate
that the flat at the other wheel has little or no effect on the peak force generated at the
contact when flat is encountered. The results, on the other hand show that the cross wheel
effect is more pronounced primarily for low phase angles. For the sequence of flat
considered, the cross wheel flat effect at the rear wheel is apparent for phase angles 45
0
and 90
0
as shown in Figs. 4.11 and 4.12, respectively. The extent of the effect is,
however, dependent on speed, phase angle between the flats as well as the phase between
112
the direct and cross wheel responses. For the given speed, the results show that cross
wheel effect on the rear wheel is more pronounced at 90
0
(Fig. 4.12) than that at 45
0
(Fig.
4.11). The results shown in Fig. 4.13 and Fig. 4.14 for phase angles of 135
0
and 180
0
,
respectively, illustrate that there is no cross wheel effect at either wheel-rail contact when
the phase angle is large.

0.4 0.45 0.5 0.55 0.6
0.2
0.4
0.6
0.8
1
1.2
1.4
1.6
1.8
2
x 10
5
F
r
o
n
t

W
h
e
e
l

C
o
n
t
a
c
t

F
o
r
c
e

(
N
)
Time (s)
Impact load due to Rear wheel flat

0.4 0.45 0.5 0.55 0.6
0.2
0.4
0.6
0.8
1
1.2
1.4
1.6
1.8
2
x 10
5
Time (s)
R
e
a
r

W
h
e
e
l

C
o
n
t
a
c
t

F
o
r
c
e

(
N
)

Impact load due to Front wheel flat

Fig. 4.13: Front and rear wheel impact force responses due to a single flat on both wheels
at 135
0
out-of-phase (rear wheel flat ahead 135
0
)


113
0.4 0.45 0.5 0.55 0.6
0.2
0.4
0.6
0.8
1
1.2
1.4
1.6
1.8
2
x 10
5
Time (s)
F
r
o
n
t

W
h
e
e
l

C
o
n
t
a
c
t

F
o
r
c
e

(
N
)

Impact load due to Rear wheel flat

0.4 0.45 0.5 0.55 0.6
0.2
0.4
0.6
0.8
1
1.2
1.4
1.6
1.8
2
x 10
5
R
e
a
r

W
h
e
e
l

C
o
n
t
a
c
t

F
o
r
c
e

(
N
)

Time (s)
Impact load due to Front wheel flat

Fig. 4.14: Front and rear wheel impact force responses due to a single flat on both wheels
at 180
0
out-of-phase (rear wheel flat ahead 180
0
)
The effect of speed on wheel-rail impact force in the presence of flat on both front
and rear wheels at particular phase difference is further examined in the speed range of 0-
120 km/h. Two identical flats of 52 mm length and 0.4 mm depth are assumed to be
present on the front and rear wheels at 45
0
, 90
0
, 135
0
and 180
0
phase difference. The peak
magnitude of impact force developed at front wheel due to the direct and cross wheel flat
as a function of speed were investigated. The variations in the front wheel peak impact
force due to the direct and cross wheel flat were very similar to that observed in Fig. 4.3
in the presence of a single wheel flat. Similar to the results presented in Figs. 4.11-4.14
114
for 70 km/h, the cross wheel effect at rear wheel can be observed at high speeds and for
low phase angle between the two flats. These results are not reported here in order to
avoid repetition of the result trends seen in Figs. 4.11-4.14.
Figure 4.15 illustrates the impact load behavior as a function of speed due to
identical flat on each wheel located at the same position or the phase angle between the
flats is zero. The results are compared with those obtained for only one wheel flat while
the other is perfect. As the results show, the impact force produced by each flat at each
wheel is dependent on the flat at that wheel and not whether one or both wheels have a
flat. Only deviation from this trend is seen around 50-90 km/h, where both wheels with
flat in phase lead to slightly lower impact load than a single flat cause. This reconfirms
the observations made earlier regarding the excitation of the bogie pitch motion when
only one wheel has a flat. From the results, it is apparent that the pitch effect is most
dominant around 65 km/h, for a single flat and vehicle parameters used.
0
50
100
150
200
250
300
350
400
10 20 25 30 40 50 60 70 80 90 100 110 120
Speed (km/h)
P
e
a
k

W
/
R

i
m
p
a
c
t

f
o
r
c
e

(
k
N
)
Rear wheel flat only Flats in both rear and front wheels in phase
Fig. 4.15: Effect of speed on wheel-rail impact load with
single and multiple wheels flats
115
4.3.3 Effect of flat length and depth
Wheel flats are usually defined by their length and depth. There is no human
control over the formation of the flats when the wheel is in service. The railway
authorities have set the allowable limits for the flat length and depth in order to continue
in service. It is thus of interest to examine the effect of the flat size on wheel-rail impact
load. In examining the effect of flat size in this investigation, three different forward
speeds are considered, while the static load acting per wheel is taken as 102 kN, which
represent a fully loaded vehicle. In practice, when a wheel flat is freshly developed, the
length and depth of the flats are geometrically related. With use, however, the flat edges
become rounded effectively changing the length of the flat without affecting its depth.
For the given depth, therefore, there can be a range of flat length. In this study, one
parameter is varied at a time, therefore, the flat length is varied for a fixed depth and vice
versa. Figure 4.16 shows the effect of flat length on wheel-rail peak impact loads for a
constant depth of 0.4 mm. It can be seen that the impact force decreases with increase in
the flat length irrespective of speed. For the given parameters as listed in Table 4.1, when
the flat length is larger than 60 mm, the rate of decrement of impact load is insignificant.
For a closer examination of significantly large impact forces corresponding to
shorter flat lengths, further simulations are carried out for flat lengths in the range of 5-40
mm. The peak wheel-rail impact load responses for three different forward velocities (v=
50, 70, 90 km/h) are shown in Fig. 4.17. The results clearly show that there exists a
critical length for the flat, which is also reported in [43]. For the selected depth of flat
(
f
D = 0.4 mm), the critical flat length is sensitive to forward speed and is found to
increase with the speed. Further simulations show that the critical flat length is sensitive
116
to the depth size, and increases with an increase in depth. The effect of flat depth on
wheel-rail impact load for a constant flat length of 52 mm is shown in Fig. 4.18. The
figure shows that an increase in the flat depth also increases the peak impact force, while
the rate also increases when the flat depth is larger. Comparison of Fig 4. 16 with Fig.
4.18 shows that the effect of flat depth is exactly opposite to that of flat length. For a
given flat depth, an increase in flat length essentially smoothens the effect of flat on the
dynamic impact load.
0
50
100
150
200
250
300
350
400
450
500
30 40 50 60 70 80
Flat length (mm), Nominal value: 52 mm
P
e
a
k

W
/
R

i
m
p
a
c
t

f
o
r
c
e

(
k
N
)
v=50 km/h v=70 km/h v=90 km/h

Fig. 4.16: Effect of flat length on W/R impact force for a constant flat depth of 0.4mm

117
120
170
220
270
320
370
420
470
520
5 10 15 20 30 40 50
Length (mm)
P
e
a
k

I
m
p
a
c
t

l
o
a
d

(
k
N
)
v= 50 km/h v= 70 km/h v= 90 km/h
Fig. 4.17: Effect of flat length on wheel-rail impact force
0
50
100
150
200
250
300
350
400
450
0.2 0.3 0.4 0.5 0.6 0.7
Flat depth (mm)
P
e
a
k

W
/
R

c
o
n
t
a
c
t

f
o
r
c
e

(
k
N
)
v=50 km/h v=70 km/h v=90 km/h

Fig. 4.18: Effect of flat depth on W/R impact force with a constant flat length of 52 mm



118
4.4 PARAMETRIC STUDY ON VEHICLE PARAMETERS

The previous section examined the effect of various flat configurations on the
wheel-rail impact load. The presence of a flat in a wheel causes large dynamic force on
vehicle system. Several vehicle system and operating parameters can have further
influence on the impact force generated due to a flat. It has been mentioned that some of
the vehicle parameters such as speed, wheel/axle load, and suspension properties have
significant influence on wheel-rail impact loads [3, 35, 43, 86]. In this subsection, effects
of different operating and vehicle parameters on peak wheel-rail impact force and peak
bearing force are investigated. A flat of 52 mm length and 0.4mm depth is considered to
be present on either a single wheel or both wheels at same location. Identification of the
important parameters those mostly affect the wheel-rail impact loads and their trends are
essentially the focus of this investigation.
4.4.1 Effect of speed
Speed has large influence on the wheel-rail impact load, especially in the presence
of the wheel flat. Figure 4.19 shows the relationship between the vehicle speed and the
wheelrail impact load due to a single flat on both front and rear wheel in phase. The
peak impact loads are evaluated under three different load conditions (W=63, 82, and 102
kN). It can be seen that the wheel-rail impact forces attain a peak in the lower speed
range (2035 km/h) and rise gradually with increase in speed. This result is consistent
with the outcomes reported in several studies [3, 35, 43, 86]. In low speed range, the rate
of increment in peak impact load due to flat is lower which can be justified by the
combined motions of wheel, rail, and tie without separation during the impact process.
However, when the speed reaches over 80 km/h, the rate of increment of impact force is
119
very significant. At these higher speeds, the separation occurs and the peak impact force
increases more rapidly with speed. At very high speed, the peak impact force may reach a
maximum value and reduce afterward since the wheel would tend to fly over the flat at
very high speed. Nevertheless, this speed is beyond the practical operating speed of
conventional freight car.
4.4.2 Effect of wheel load
During the operation of a freight car, its load can fluctuate from empty car to the
fully loaded car. Figure 4.20 illustrates the simulated peak impact force for different
wheel loads in the presence of single wheel flat for two different speeds (v= 50 and 70
km/h). The peak wheel-rail impact load is evaluated in the presence of a front wheel flat
of 52 mm length and 0.4mm in depth. The figure shows that for a constant speed,
increasing the axle load linearly increases the peak wheel-rail impact force. For the given
parameters as listed in Table 4.1 and for higher speeds, the maximum peak impact load
can be greater than twice the static load for partially loaded car, and close to twice the
static load for a fully loaded car.
120
0
50
100
150
200
250
300
350
0 10 20 25 30 35 40 50 60 70 80 90 100 110 120
Speed (km/h)
P
e
a
k

W
/
R

c
o
n
t
a
c
t

f
o
r
c
e

(
k
N
)
W=102 kN W=82 kN W=63kN
Fig. 4.19: Effect of speed on wheel-rail impact force with three loads
80
100
120
140
160
180
200
220
50 60 70 80 90 100 110 120 130
Static load per wheel (kN)
P
e
a
k

W
/
R

I
m
p
a
c
t

F
o
r
c
e

(
k
N
)
v=50 km/h v=70 km/h

Fig. 4.20: Effect of static wheel load on peak wheel-rail impact force
4.4.3 Effect of unsprung mass
Among vehicle parameters, the wheel represented as unsprung mass is directly
exposed to the impact load. Figure 4.21 shows the influence of wheel mass on the wheel-
121
rail impact load in the presence of a single flat on both wheels in phase. Since the impact
load depends on the static wheel load also, three different loading conditions are
considered here. Figure 4.21 shows that increasing the wheel mass increases the wheel-
rail impact load up to 600 kg. Thereafter, there is no significant increment in peak impact
load due to the increase in wheel mass. The trend is similar for all three static loads.
Since the unsprung mass is relatively low compare to the vehicle mass and static load, the
effect of a change in unsprung mass is also not very significant. The largest influence is
observed for a combination of low unsprung mass with low static load.
120
125
130
135
140
145
150
155
160
165
170
300 400 500 600 700 800 900
Unsprung mass (kg)
P
e
a
k

I
m
p
a
c
t

f
o
r
c
e

(
k
N
)
W=102 kN W=82 kN W=63 kN
Fig. 4.21: Effect of unsprung mass on W/R impact load

4.4.4 Effect of suspension stiffness and damping on peak W/R impact load
There are no resilient materials to form primary suspension in the conventional
three-piece freight car system in North America. However, the linear spring and damper
elements in between the wheel and the side frame are used in modeling of primary
suspension. The effects of primary suspension stiffness and damping on the peak wheel-
122
rail impact loads are examined in the presence of a single flat on both wheels in phase.
The simulations were carried out for a constant forward speed of 70 km/h and three static
loads (W= 63, 82 and 102 kN). Figure 4.22 illustrates that increase in the primary
suspension stiffness slightly decreases the peak wheel-rail impact load. For the given
parameters, the maximum variation in peak wheel-rail impact load is around 4 kN
irrespective of static load. However, the peak impact force was observed as insensitive to
the primary suspension damping coefficient. This behavior can be attributed to the large
nominal value of primary suspension stiffness. Effects of secondary suspension stiffness
and damping on the peak wheel-rail impact force were also investigated under similar
conditions. The study revealed that secondary suspension stiffness and damping
coefficients have no significant effect on peak wheel-rail impact force.
4.4.5 Effect of suspension stiffness and damping on peak bearing force
The primary suspension of a freight vehicle effectively represents the bearings
that support the wheelsets on the side frame. Effect of primary and secondary suspension
stiffness and damping on the peak bearing force is further evaluated in the presence of a
single flat of 52mm long and 0.4mm deep. It is clear from Fig. 4.23 that increasing the
primary suspension increases the peak bearing force. Thus, making the primary
suspension softer can reduce the bearing force. This will, however, lead to a slight
increase in the wheel-rail impact load, as shown in Fig. 4.22. Furthermore, reducing
primary suspension stiffness significantly may compromise the design of a typical three-
piece track system as it may introduce warp motion of the side frame. Effects of
secondary suspension stiffness and damping on peak bearing forces were further
investigated in the presence of identical size of flat and three static wheel loads (W= 63,
123
82 and 102 kN). The results demonstrated that the secondary suspension stiffness and
damping has negligible influence on peak bearing force.

120
130
140
150
160
170
180
500 600 700 800 900 1000
Primary stiffness (MN/m)
P
e
a
k

W
/
R

c
o
n
t
a
c
t

f
o
r
c
e

(
k
N
)
W=63 kN W=82 kN W=102 kN

Fig. 4.22: Effect of primary suspension stiffness on peak wheel-rail impact force
80
100
120
140
160
180
200
500 600 700 800 900 1000
Primary suspension stiffness (MN/m)
P
e
a
k

b
e
a
r
i
n
g

f
o
r
c
e

(
k
N
)
W=63 kN W=82 kN W=102 kN

Fig. 4.23: Effect of primary suspension stiffness on peak bearing force
124

4.5 PARAMETRIC STUDY ON TRACK MODEL PARAMETERS
As shown in chapter 3, wheel flat has significant influence on track system
components. It causes large dynamic impact force at the interfaces of track components.
Excessive deflections in the components of the track also arise due to the presence of a
wheel flat. In this section, the effects of different track model parameters on wheel-rail
impact force, pad force, ballast force, as well as rail, sleeper and ballast displacements are
investigated. The objective of this subsection is to identify the important track model
parameters those have considerable influence on the peak impact forces and
displacements. The investigation is carried out at a speed of 70 km/h in the presence of
two same size flats on front and rear wheels at same position. The size of the flat is
considered as 52 mm long and 0.4mm deep and the static loads acting on the wheel are
63, 82, and 102 kN. The basic model parameters are same as those listed in Table 4.1,
unless specified.
4.5.1 Effect of Rail Mass per unit Length
The effect of rail mass per unit length on peak wheel-rail impact load is shown in
Fig. 4.24. The figure shows that an increase in rail mass increases the peak wheel-rail
impact load under all three static loads. Increasing the rail mass increases the rail
effective mass and rail stiffness, this in turn increases the wheel-rail impact force. The
effect of different axle loads is more prominent at lower value of rail mass. For a higher
value of rail mass, the peak impact load due to the flat becomes independent of the static
wheel load.
125
100
120
140
160
180
200
220
30 40 50 60 70 80 90 100
Rail mass (kg/m)
P
e
a
k

W
/
R

c
o
n
t
a
c
t

f
o
r
c
e

(
k
N
)W=102 kN W=63 kN W=82 kN
Fig. 4.24: Effect of rail mass on wheel-rail impact force

4.5.2 Effect of railpad stiffness and damping
Railpad is known to be an influential parameter in the study of dynamic wheel-
rail impact load. As discussed in section 3.2, it affects the dynamic response of the track
system both in terms of second peak and its frequency. Usually, a softer pad leads to
reduce dynamic wheel-rail contact forces. For a constant speed of 70 km/h and three
different static load conditions, Fig. 4.25 shows that an increase in the pad stiffness and
damping increases the peak wheel-rail contact force to some extent. Increasing the rail-
pad stiffness enhances the tie-rail connection and increases the track effective mass. This
may contribute to an increase in the peak impact load. These results are consistent with
the results reported in [31, 40, 43].
The effects of rail pad stiffness and damping on the peak pad force at 22
nd
sleeper
are examined and shown in Fig. 4.26. The results show that the pad force increases
linearly with an increase of stiffness and damping of the pad. The difference in peak pad
126
force due to higher static wheel load is also significant. Increasing the pad stiffness and
damping coefficient increases both the relative deflection and velocity of the rail and
sleeper. The pad being sandwiched between them, thus also experience larger dynamic
force.
The peak rail displacement, however increases as the pad stiffness is increased.
Effect of railpad stiffness on peak rail displacement at the 22
nd
sleeper is shown in Fig.
4.27. It can be seen that increase in the rail pad stiffness reduces the peak rail
displacement in a significant manner. An increase in pad damping coefficients, however,
showed no influence on the peak rail displacement.
4.5.3 Effect of ballast stiffness and damping
The stiffness of the ballast mainly forms the stiffness of the overall track system,
which may influence the dynamics of vehicle and track significantly. The effect of ballast
stiffness and damping on the wheel-rail impact force, and the rail and ballast deflections
are shown in Figs. 4.28 to 4.30 under three different wheel loads. The peak values of
impact force as well as the rail and ballast peak displacements were computed when the
wheel flat strikes the rail above the 22
nd
ballast. For the range of ballast stiffness and
damping values, Fig. 4.28 shows that an increase in the ballast stiffness and damping
slightly reduces the peak wheel-rail impact load. Figures 4.29 and 4.30 show that an
increase in ballast stiffness reduces peak rail displacement while increases the peak
ballast deflection. This behavior can be attributed to the increase in total track stiffness
due to the increase in ballast stiffness that in turn reduces the rail deflection. The results
obtained from this study further revealed that ballast damping has negligible influence
both on the peak rail and ballast displacements.
127
130
140
150
160
170
180
190
60 80 100 120 140 160 180
Pad stiffness (MN/m)
P
e
a
k

W
/
R

i
m
p
a
c
t

f
o
r
c
e

(
k
N
)
W=63 kN W=82 kN W=102 kN

90
110
130
150
170
190
4 5 6 7 8 9 10
Pad damping coefficient (kN-s/m)
P
e
a
k

i
m
p
a
c
t

f
o
r
c
e

(
k
N
)
W=63 kN W=82 kN W=102kN

Fig. 4.25: Effect of railpad stiffness and damping on wheel-rail impact load

128
60
70
80
90
100
110
120
130
140
150
60 80 100 120 140 160 180
Pad stiffness (MN/m)
P
e
a
k

p
a
d

f
o
r
c
e

(
k
N
)
W=63 kN W=82 kN W=102 kN

70
80
90
100
110
120
130
140
5 6 7 8 9
Pad damping coefficient (kN-s/m)
P
e
a
k

p
a
d

f
o
r
c
e

(
k
N
)
W=63 kN W=82 kN W=102kN

Fig. 4.26: Effect of railpad stiffness and damping on peak pad force


129
0.5
0.7
0.9
1.1
1.3
1.5
1.7
1.9
2.1
2.3
60 80 100 120 140 160 180
Pad stiffness (MN/m)
P
e
a
k

r
a
i
l

d
i
s
p
l
a
c
e
m
e
n
t

(
m
m
)
W=63 kN W=82 kN W=102 kN
Fig. 4.27: Effect of rail pad stiffness on peak rail displacement

The effect of rail ballast stiffness and damping coefficient on the peak ballast
force are further investigated for the 22
nd
number ballast and shown in Fig. 4.31. The
figures clearly show that increase in ballast stiffness linearly increases the peak ballast
force. However, increase in the ballast damping decreases the peak ballast force to some
extent. These results are consistent with the results reported in [37].

130
125
130
135
140
145
150
155
160
165
170
120 140 160 180 200 220 240
Ballast stiffness (kN/m)
P
e
a
k

W
/
R

c
o
n
t
a
c
t

f
o
r
c
e

(
k
N
)
W=63 kN W=82 kN W=102 kN

125
130
135
140
145
150
155
160
165
170
175
40 50 60 70 80 90
Ballast damping coefficient (kN-s/m)
P
e
a
k

W
/
R

c
o
n
t
a
c
t

f
o
r
c
e

(
k
N
)
W=63 kN W=82 kN W=102 kN

Fig. 4.28: Effect of ballast stiffness and damping on peak wheel-rail impact force


131
1
1.2
1.4
1.6
1.8
2
120 140 160 180 200 220 240
Ballast stiffness (kN/m)
P
e
a
k

r
a
i
l

d
i
s
p
l
a
c
e
m
e
n
t

(
m
m
)
W=63 kN W=82 kN W=102 kN
Fig. 4.29: Effect of ballast stiffness on peak rail displacement
0.3
0.35
0.4
0.45
0.5
0.55
0.6
120 140 160 180 200 220 240
Ballast stiffness (kN/m)
P
e
a
k

b
a
l
l
a
s
t

d
i
s
p
l
a
c
e
m
e
n
t

(
m
m
)
W=63kN W=82 kN W=102 kN
Fig. 4.30: Effect of ballast stiffness on peak ballast displacement

132
60
70
80
90
100
110
120
130
120 140 160 180 200 220 240
Ballast stiffness (kN/m)
P
e
a
k

b
a
l
l
a
s
t

f
o
r
c
e

(
k
N
)
W=63 kN W=82 kN W=102 kN
60
70
80
90
100
110
120
40 50 60 70 80 90
Ballast damping coefficient (kN-s/m)
P
e
a
k

r
a
i
l

b
a
l
l
a
s
t

f
o
r
c
e

(
k
N
)
W=63 kN W=82 kN W=102 kN
Fig. 4.31: Effect of ballast stiffness and damping on peak ballast force


133
4.5.4 Effect of Sleeper mass
In practice, the sleeper mass can be varied depending on the type of materials
used in sleepers. The effect of a rigid sleeper mass on the peak wheel-rail impact force
and pad force is shown in Figs. 4.32 and 4.33, respectively. The responses computed for
the 22
nd
sleeper show that both peak wheel-rail impact force and pad force increase due to
the increase in sleeper mass. The increase in W/R impact force is however, highly
insignificant. The effect of sleeper mass on peak rail and sleeper displacements were also
investigated in the presence of flat and at a forward speed of 70 km/h. However, the
results revealed negligible influence of sleeper mass on both peak rail and sleeper
displacements.
4.5.5 Effect of Ballast mass
Although ballast mass is the largest mass in the track system, it is located farthest
from the wheel-rail impact point. The effect of ballast mass on the peak wheel-rail impact
force, ballast force, and peak rail deflection and ballast deflections are investigated at a
vehicle speed of 70 km/h. Figure 4.34 shows the effect of ballast mass on the peak wheel-
rail impact force. It can be seen that a very slight increase in the impact force results from
an increase in the ballast mass. Furthermore, insignificant effects of ballast mass on peak
ballast force, rail and ballast displacement are observed.



134
125
130
135
140
145
150
155
160
165
170
175
80 90 100 110 120 130 140
Sleeper mass (kg)
P
e
a
k

W
/
R

C
o
n
t
a
c
t

f
o
r
c
e

(
k
N
)
W=63 kN W=82 kN W=102 kN
Fig. 4.32: Effect of sleeper mass on peak wheel-rail impact force
70
80
90
100
110
120
130
70 80 90 100 110 120 130 140 150
Sleeper mass (kg)
P
e
a
k

p
a
d

f
o
r
c
e

(
k
N
)
W=63 kN W=82 kN W=102 kN
Fig. 4.33: Effect of sleeper mass on peak pad force



135
125
130
135
140
145
150
155
160
165
170
175
600 650 700 750 800 850 900
Ballast mass (kg)
P
e
a
k

W
/
R

c
o
n
t
a
c
t

f
o
r
c
e

(
k
N
)
W=63 kN W=82 kN W=102 kN
Fig. 4.34: Effect of ballast mass on peak wheel-rail impact force
4.5.6 Effect of sleeper spacing
Distance between the sleepers has a direct influence on the deflection of the rail.
An increase in sleeper distance is thus known to cause larger wheel-rail dynamic force
even if there is no defect in wheel-rail contact interface. The effect of sleeper spacing on
the peak wheel-rail impact loads and deflection of the rail due to the nominal flat size
under three wheel loads are shown in Figs. 4.35 and 4.36. The figures show that increase
in the sleeper spacing linearly increases the peak rail deflection while decreases the peak
wheel-rail impact forces. Higher sleeper spacing reduces the overall track stiffness that
makes the track softer, which ultimately results more deflection of the track and less
wheel-rail impact force. Although the peak wheel-rail impact force reduces due to higher
sleeper spacing, it will adversely affect the load carrying capacity of the rail.
136
100
110
120
130
140
150
160
170
180
190
0.4 0.6 0.8
Sleeper distance (m)
P
e
a
k

W
/
R

i
m
p
a
c
t

f
o
r
c
e

(
k
N
)
W=63 kN W=82 kN W=102 kN
Fig. 4.35: Effect of sleeper spacing on peak wheel-rail impact force
0
0.5
1
1.5
2
2.5
0.4 0.6 0.8
Sleeper spacing (m)
P
e
a
k

r
a
i
l

d
i
s
p
l
a
c
e
m
e
n
t

(
m
m
)
W=63 kN W=82 kN W=102 kN
Fig. 4.36: Effect of sleeper spacing on peak rail displacement
4.5.7 Effect of bending stiffness of rail
Bending stiffness of the rail is the product of the elastic modulus of rail and rail
second moment of area. The effects of bending stiffness on peak wheel-rail impact force
and rail displacement are shown in Figs. 4.37 and 4.38. Figure 4.37 shows that increase in
the bending stiffness increases the peak wheel-rail impact force. This increment can be
137
attributed to the dynamics of the moving vehicles. When the vertical bending stiffness of
the rail is low, the passing wheels generate large deflections at the wheel-rail contact
point, which ultimately leads to larger wheel forces. The deflection of the rail decreases
with increase in rail bending stiffness as shown in Fig. 4.38.
125
135
145
155
165
175
4 5 6 7 8
Bending stiffness (N-m
2
)
P
e
a
k

W
/
R

i
m
p
a
c
t

f
o
r
c
e

(
k
N
)
W=63 kN W=82 kN W=102 kN
Fig. 4.37: Effect of bending stiffness on peak wheel-rail impact load
0.7
0.9
1.1
1.3
1.5
1.7
1.9
2.1
2.3
4 5 6 7 8
Bending stiffness (N-m
2
)
P
e
a
k

r
a
i
l

d
i
s
p
l
a
c
e
m
e
n
t

(
m
m
)
W=63 kN W=82 kN W=102 kN

Fig. 4.38: Effect of bending stiffness on peak rail displacement
138
4.6 SUMMARY
In this chapter, the developed coupled vehicle-track model is employed to study
the wheel-rail impact load caused by the single and multiple flats. A detailed parametric
study on the impact loads is carried out in presence of single and multiple flats with
different sizes and their relative positions. The effects of design and operational
parameters of vehicle and track on wheel-rail impact load, bearing force, railpad force
and ballast force are investigated.
This study shows that the presence of wheel flat in either front or rear wheel has
considerable effect on the cross wheel impact load. Although this cross wheel impact is
considerably less than that of the direct wheel impact force, its effect can be significant in
high-speed condition. The cross wheel impact force also depends on the size of the wheel
flat and track system parameters.
In investigations of impact loads due to multiple flats with the pitch plane
model, it is observed that the presence of two flats on different wheels on same location
has considerable effect on peak wheel-rail impact load. The magnitude of the peak wheel-
rail impact load produced by two in-phase flats for certain speeds is lower than the
impact load produced by a single flat of same size on one wheel. The study further
illustrates that the magnitudes of the wheel-rail impact forces generated due to multiple
flats either in the same or different wheels are same to that generated by single wheel flat
when the flats are far away from each other.
The study shows that wheel-rail impact load is significantly influenced by some
operating parameters such as wheel load and vehicle speed. The flat size in terms of
length and depth of flat is shown to have opposite effect on the generated peak impact
139
force. However, some factors such as primary and secondary suspension damping and
secondary suspension stiffness have negligible influence on the wheel-rail impact force
and bearing force, while primary stiffness has considerable influence on the wheel-rail
impact force as well as bearing force. Among the track parameters, the rail mass, railpad
stiffness, and damping, bending stiffness of rail and ballast mass show considerable effect
on the wheel-rail impact loads while the other factors show negligible effect. The rail
pad, and ballast stiffness and damping also generate considerable influence on the rail
pad and ballast force, respectively. Rail parameters such as sleeper spacing has the most
significant influences on both track displacement and impact force. In general, the
parameter that increases rail displacement tends to reduce the magnitude of peak impact
load.













140
CHAPTER 5
CONCLUSIONS AND RECOMMENDATIONS
5.1 GENERAL
As set out in chapter 1, the overall objective of this work is to study the
characteristics of vertical dynamic of wheel-rail impact loads in the presence of wheel
flats through developing a comprehensive vehicle-track system model. The specific
objectives include: formulation of mathematical model of 2-D pitch-plane vehicle and
three-layer track model based on Euler beam theory, coupling of the vehicle and track
interaction model through nonlinear Hertzian contact spring, analysis of the responses of
each components of the vehicle-track system, and investigation of the wheel-rail impact
loads in the presence of multiple wheel flats. Modal analysis method incorporating the
MATLAB predefined function was applied to solve the coupled partial and ordinary
differential equations. The results from the developed mathematical model were
compared with the reported analytical and measured data in the presence of a flat. A
comprehensive parametric study covering a wide range of parameters from the vehicle
and track system model was carried out to identify the major factors those affect the
wheel-rail impact loads. A thorough investigation of predicted wheel-rail impact loads
due to a single as well as multiple wheel flats was accomplished and the impact loads due
to multiple flats were compared with the impact loads due to same size single flat. In the
present chapter, highlights of the present work are summarized. Based on the results
some general and specific conclusions are drawn, and a direction for future investigations
is established.

141
5.2 HIGHLIGHTS OF THE PRESENT WORK
From the review of relevant literature, it was concluded that accurate prediction of
wheel-rail dynamic loads is mainly dependent on the wheel-rail contact model and the
track system model. The most common contact model used in such studies is the Hertzian
nonlinear point contact model. In this study, the Hertzian contact model is adopted to
calculate the vehicle-track impact forces in vertical direction. A five-DOF pitch plane
vehicle model is developed together with 3-layer track system model. Care is taken in the
modeling of the track system to include railpads, sleepers, ballasts, and subgrades. The
developed model can simulate the wheel lift-off from the rail, the rail lift-off from the
sleeper and sleeper lift-off from the ballast.
Considerable time and effort is devoted in order to validate the developed model
in the presence of a flat. The validation is achieved by comparing the results with
theoretical and measured data available in literature. The validated model is employed to
investigate the vertical dynamic response characteristics of the vehicle-track system. The
dynamic forces induced by the single and multiple wheel flats are addressed and
discussed in this study. This study also investigated the effect of one wheel flat on the
other wheel within the same bogie. The influences of the flat size as well as various
design and operating parameters on wheel-rail impact loads are analyzed and discussed in
this investigation.




142
5.3 CONCLUSIONS
The major conclusions drawn from the present research work are summarized in
the followings:
- The modal analysis method is a very convenient tool in the dynamic analysis of
coupled lumped parameter vehicle and continuous track system. The simulation
results showed that the simplified five DOF pitch-plane model is adequate to
predict the vertical wheel-rail impact load due to a wheel flat. In the simulation,
the total length of the rail equal to 100 sleeper-spans is found to be sufficient to
represent an infinite long track at all practical speeds. A total of 100 rail modes
considered in the study is also found to be sufficient enough for accurate
prediction of wheel-rail impact load.
- In presence of a wheel flat, wheel-rail impact load is significantly affected by the
vehicle speed, axle loading, and size of the flats. This study showed that vehicle
secondary suspension properties have little effect on the impact load, while
primary suspension stiffness has obvious influence on the bearing force.
- The pitch-plane vehicle model shows that the dynamic load at one wheel is also
affected by the adjacent wheel flat within the same bogie. The results also
revealed that the deflection of rail under a wheel of a bogie is also affected by the
other wheel flat. Although the impact force at the defective wheel is significantly
higher than the impact force arises in flat-free wheel, the repetition of this impact
force after a certain time interval can cause fatigue damage and fracture in the rail
or vehicle components.
143
- In case of multiple flats, when flats are present in both front and rear wheel at
same position, wheel-rail impact load induced by the two flats are less than that of
the same size single wheel flat. This is more prominent at certain speeds where
pitch dynamics of the bogie is excited. However, when multiple flats are present
on different wheels in out-of-phase conditions, both wheel flats have effect on the
response of wheel-rail contact force depending on the relative positions of the
flats and speed of the vehicle. The effect of cross wheel flat on wheel-rail impact
loads is more when the phase difference between the flats is less and vice versa.
- When multiple flats are present within one wheel, wheel-rail impact load that
arises at the position of second flat is mostly affected by the location of the first
flat within the same wheel. The effects are significant when the locations of the
flats are close to each other.
- The magnitude of the peak impact loads largely depends on both length and depth
of the wheel flats. For a given set of vehicle-track system parameters and vehicle
speed, there exists a critical length of the flat depending on the depth where the
impact load is the largest. Thus, the present AAR wheel removal criteria due to
flat, which mainly specify the flat length, may not be sufficient without the
addition of flat depth.
- Although the location of the sleepers and ballasts masses are far from the wheel-
rail contact area, some of these parameters such as sleeper and ballast mass,
ballast stiffness and damping have some effect on the wheel-rail impact load in
the presence of wheel flat. Railpad stiffness and damping are found to have
significant effect on the wheel-rail impact force. Furthermore, bending stiffness
144
and mass of the rail have also strong effect on the wheel-rail impact loads and rail
deflections.
- Most sensitive rail parameter for impact load and rail deflection is found to be the
sleeper spacing. In general, the parameters that lead to increase in rail deflection
tend to generate lower peak impact force due to flat.

5.4 RECOMMENDATIONS FOR FUTURE WORK
The present research work yields significant insights on the problems associated
with the wheel-rail impact loads in the presence of single and multiple wheel flats. In
this study, more attentions are directed towards the track system modeling with
simplified approach in the vehicle model. Although this study clearly demonstrates
reasonably accurate and useful results with simplified analytical model, the potential
usefulness and accuracy can be further enhanced upon some other more considerations. A
list of further studies that can be carried out with the developed model along with
recommendations for model improvement is presented in the followings:
- Although the Hertzian nonlinear point contact model is widely used in the study
of vertical vehicle-track interaction, this model assumes that contact occurs at
wheel centre point that is not always accurate especially in case of defective
wheel. Thus, a multipoint contact model that can accommodate the partial contact
in the presence of wheel defect would be a better alternative to predict the wheel-
rail interaction forces.
- The developed two-dimensional pitch plane vehicle model can be extended to
three-dimensional full vehicle model with two bogies and four wheelsets in order
145
to incorporate the roll effect of the car and bogie. The two-dimensional track
model can be extended to three-dimensional track system to accommodate the
three-dimensional vehicle model in order to investigate the interaction between
left and right wheels and rails.
- The rail can be modeled as Timoshenko beam on discrete support that includes
the shear deformation and rotatory inertia of the beam. The track model can be
further improved by including the nonlinear effects of the railpads and ballasts
stiffness and damping. The lumped parameter model of the sleepers can also be
extended to continuous beam model in order to couple with the 3-D track system
model.
- The present model is only validated with the analytical and measured data
available from literature. An experimental effort to validate the present work will
enhance the study, especially for multiple wheel flats. For better understanding of
the multiple wheel flats effect on wheel-rail impact loads, further studies with
incremental phase angles among the flats either in one wheel or different wheels
should be carried out. There must be a critical phase angle for multiple flats that
may help with wheel removal criteria in the presence of multiple flats.
- The developed model can be employed to study the vehicle-track interaction
problems due to other type of wheel and rail defects such as rail joints,
corrugation, shelling etc. The model can be further refined by including the
bending vibration of the wheelsets to study the vertical and lateral interaction of
the vehicle-track system.


146
REFERENCES
1. Oscarsson, J., Simulation of Train-Track Interaction with Stochastic Track
Properties Vehicle System Dynamics 2002, Vol. 37, No. 6, pp. 449-469.
2. Bitzenbauer, J., and Dinkel, J., Dynamic interaction between a moving vehicle and
an infinite structure excited by irregularities Fourier transforms solution Archive
of Applied Mechanics 72 (2002) 199-211.
3. Zhai, W. M., Cai, C. B., Wang, Q. C., Lu, Z.W., and Wu., X. S., Dynamic effects
of Vehicles on Tracks in the case of raising train speed Proceedings of the
Institution of Mechanical Engineers, Part F, v 215, 2001, p125-135.
4. Dahlberg, T., Some railroad settlement modelsa critical review Proceedings of
the Institution of Mechanical Engineers, Vol 215 Part F, p289-300.
5. Johansson, A., and Nielsen, J. C. O., Out of round railway wheels, wheel rail
contact forces and track response derived from the field tests and numerical
simulation Journal of Rail and Rapid transit volume 217 part F, 1995, p135-146.
6. Dahlberg, T., Vertical dynamics Train /Track interaction verifying a theoretical
model by full scale experiments Vehicle system dynamics supplement 24 (1995)
pp 45-57.
7. Wu, T.X., and Thompson, D.J., A hybrid model for the noise generation due to
railway wheel flats. Journal of Sound and Vibration (2002) 251(1) 591-604.
8. Thompson, D., and Wu, T.X., The effect of non-linearity on wheel/rail impact
Journal of Rail and Rapid transit, volume 218, part F, 2003, p1-15.
9. Wu, T.X., and Thompson, D.J., Behavior of the normal contact force under
multiple Wheel/Rail interaction. Vehicle System Dynamics 37 (3) (2002), pp.
157174.
10. Bizindavyi, L., Fang, Z., Green, M.F., Campbell, T.I., Anderson, R. J., and
Moucessian A, Parametric Study of Dynamic Response of Resilient Track for
Transit System, 4th Structural Specialty Conference of the Canadian Society for
Civil Engineering, Montreal, Quebec, Canada, June 5-8, 2002.
11. Sun, Y. Q., Dhanasekar, M., and Roach, D., A three-dimensional model for the
lateral and Vertical Dynamics of wagon-track system. Proceedings of the
institution of Mechanical Engineers, Part F: Rail and Rapid Transit. Vol. 217 (1),
2003, p 31-45.
12. Sun, Y. Q., and Dhanasekar, M., Importance of the track modeling on the
determination of the critical speed of wagons Vehicle system Dynamics
Supplement 41 (2004), p232-241.
147
13. Jin, X., Wen, Z., Wang, K., and Xiao, X., Effect of passenger car curving on rail
corrugation at a curved track Volume 260, Issue 6, 10 March 2006, p619-633.
14. Hou, K., Kalousek, J., Dong, R., A dynamic model for an asymmetrical
vehicle/track system Journal of Sound and Vibration 267 (2003) 591604.
15. Wu, T.X., and D.J. Thompson, On the impact noise generation due to a wheel
passing over rail joints Journal of Sound and Vibration 267 (2003) 485496.
16. Arnold, D. K., and Joel, E.C., Analysis and tests of bonded insulated rail joints
subjected to vertical wheel loads International Journal of Mechanical Sciences 41
(1999) 1253 -1272.
17. Jin, X., Wang, K., Wen, Z., and Zhang, W., Effect of rail corrugation on vertical
dynamics of railway vehicle coupled with a track Acta Mech Sinica (2005) 21, 95
102.
18. Szolc, T., Medium frequency dynamic investigation of the railway wheelset-track
system using a discrete-continuous model, Archive of Applied Mechanics 68
1998, pp. 30 45.
19. Lei, X., Dynamic analysis of the track structure of a high speed railway using
finite elements Proceedings of Institution of Mechanical Engineers, Vol. 215 Part
F, 2001, p-301-309.
20. Koro, K., Abe, K., Ishida, M., and Suzuki, T., Timoshenko beam finite element for
vehicletrack vibration analysis and its application to jointed railway track
Proceedings of Institution of Mechanical Engineers, Vol. 218, Part F, 2004, p-159-
172.
21. Morys, B., Enlargement of out-of-round wheel Profiles on high speed trains
Journal of Sound and Vibration (1999) 227(5), p965-978.
22. Nielsen, J.C.O., and Oscarsson, J., Simulation of dynamic traintrack interaction
with state-dependent track properties Journal of Sound and Vibration, vol. 275
(2004) 515532.
23. Nielsen, J.C.O., and Igeland, A., Vertical dynamic interaction between train and
track-influence of wheel and track imperfections Journal of Sound and Vibration,
187(5), p825-839, 1995.
24. Wen, Z., Jin, X., and Zhang, W., Contact-impact stress analysis of rail joint region
using the dynamic finite element method Wear 258 (2005) p13011309.
25. Jaschinski, A., Multibody simulation of flexible vehicles in interaction with
flexible guidways Journal of Vehicle system dynamics, supplement, 24, 1995, pp
31-44.
148
26. Shen, G., and Pratt, I., The development of a railway dynamics modeling and
simulation package to cater for current industrial trends Proceedings of the
Institution of Mechanical Engineers, Part F, v. 215, 2001, p167-178.
27. Chudzikiewicz, A., Simulation of rail vehicle dynamics in MATLAB
environment Vehicle System Dynamics, 33, 2000, p107119.
28. Andersson, C., and Abrahamsson, T., Simulation of interaction between a train in
general motion and a track Vehicle System Dynamics, 2002, Vol. 38, No. 6, pp.
433455.
29. Wickens, A.H., Static and dynamic stability of asymmetric two-axle railway
vehicles possessing perfect steering, Vehicle System Dynamics 11(1982), pp.898
1006.
30. Ishida, M., and Ban, T., Modelling of the wheel flats for track dynamics XXX
Convegno Nazionale AIAS- Alghero (SS), 12-15 September 2001.
31. Fermer, M., and Neilson, J. C. O., Vertical interaction between train and track with
soft and stiff rail pads Proceedings of the Institution of Mechanical Engineers, Part
F, v 209, 1995, p 39-47.
32. Ahlbeck, D.R., and Daniels, L.E., Investigation of rail corrugations on the
Baltimore Metro Wear, 144 (1991) 197-210.
33. Sing, V.V., and Deepak, D., Evaluation of track stiffness and track damping,
Journal of Sound and Vibration (1984), 97 (2), p 129-135.
34. Fryba, L., Nakagiri, S., and Yoshikawa, N., Stochastic finite elements for a beam
on a random foundation with uncertain damping under a moving force, Journal of
Sound and Vibration (1993), 163 (1), p 9-27.
35. Newton, S.G., and Clark, R.A., An investigation into the dynamic effects on the
track of wheel flats on railway vehicles Journal of Mechanical Engineering
Science, 1979, 21(4), 287297.
36. Grassie, S.L., Gregory, R. W., Harrison, D., and Johnson, K. L., The dynamic
response of railway track to high frequency vertical excitation Journal of
Mechanical Engineering Science, 1982, 24(2), 7790.
37. Dong, R.G., Vertical dynamics of railway vehicle-track system, Ph.D. Thesis,
Dept. of Mechanical and Industrial Engineering, Concordia University, Montreal,
Canada, 1994.
38. Barke, D.W., and Chiu, W. K., A review of the effects of out-of-round wheels on
track and vehicle components Proceedings of the Institution of Mechanical
Engineers, part F, 2005, 219, pp 151-175.
149
39. Ilias, H., The influence of railpad stiffness on wheelset/track interaction and
corrugation growth journal of sound and vibration (1999) 227(5), 935-948.
40. Ripke, B., and Knothe, K., Simulation of high frequency vehicle-track
interactions Journal of Vehicle system dynamics, supplement, 1995, pp 72-85.
41. Briginshaw, D., Bright outlook for Rail Investment Introduction International
Railway Journal, March 15, 2001.
42. Knothe, K., and Grassie, S.L., Modeling of railway track and vehicle/track
interaction at high frequencies, Vehicle System Dynamics 22 1993, pp. 209 262.
43. Dukkipati, R.V., and Dong, R., Impact loads due to wheel flats and shells Journal
of Vehicle System Dynamics, 31 1999, pp. 122.
44. Jin, X., Wen, Z., Zefeng, W., Weihua, Z., and Zhiyun, S., Numerical simulation of
rail corrugation on a curved track Computers and Structures 83 (2005) 20522065.
45. Matsumoto, A., Sato, Y., Tanimoto, M., and Kang, Q., Study on the formation
mechanism of rail corrugation on curved track, Vehicle System Dynamics,
supplement 25 (1996).p450-465.
46. Wu, T.X., and Thompson, D.J., A double Timoshenko beam model for vertical
vibration analysis of railway track at high frequencies Journal of Sound and
Vibration (1999) 224(2), 329-348.
47. Thompson, D. J., and Verheij, J.W., The dynamic behavior of rail fasteners at high
frequencies Applied Acoustics, Vol. 52, No. 1, pp. l-17, 1997.
48. Wu, T.X., and Thompson, D.J. The effects of local preload on the foundation
stiffness and vertical vibration of railway track Journal of Sound and Vibration
(1999), 219 (5), 881-904.
49. Andersson, C., and Oscarsson, J., Dynamic train/track interaction including state-
dependent track properties and flexible vehicle components, Vehicle System
Dynamics Supplement 33 (1999) 4758.
50. Thompson, D.J., Vliet, W.J., and Verheij, J.W., Development of the indirect
method for measuring the high frequency dynamic stiffness of resilient elements.
Journal of Sound and Vibration, Vol. 213, 169-188, 1998.
51. Fenander, A., Frequency dependent stiffness and damping of railpads
Proceedings of the Institution of Mechanical Engineers, Part F: Rail and Rapid
Transit. Vol. 211, 1997 p51-62.
52. Zhu, J. J., Development of an adaptive contact model for analysis of wheel-rail
impact load due to wheel flats, M.A Sc. thesis, Concordia University, Montreal,
Canada, 2006 .
150
53. Zhai, W., and Cai, Z., Dynamic interaction between a lumped mass vehicle and a
discretely supported continuous rail track Computers and Structures Vol. 63, No.
5, pp. 987-997, 1997.
54. Cai, .W., Choung, Y.K., and Chan, H.C., Dynamic response of infinite continuous
beams subjected to a moving force: an exact method J. Sound Vibration, 1988,
123(3), p461-472.
55. Dahlberg, T., Railway track dynamics - a survey, Division of Solid Mechanics,
Linkping University SE-581 83 Linkping, Sweden.
56. Clark, R. A., Dean, P. A., Elkins, J. A., and Newton, S. G., An investigation into
the dynamic effects of railway vehicles running on corrugated rails, Journal of
Mechanical Engineering Science 24, p65-76, 1982.
57. Grassie, S.L., and Cox, S.J., The dynamic response of railway track with flexible
sleepers to high frequency vertical excitation Proceedings of the Institution of
Mechanical Engineers, part D, 198, pp 117-124.
58. Grassie, S.L., and Cox, S.J., The dynamic response of railway tracks with
unsupported sleepers Proceedings of the Institution of Mechanical Engineers, part
D, 199, pp 123-135.
59. Grassie, S. L., Dynamic modelling of concrete railway sleepers Journal of Sound
and Vibration (1995), 187 (5), p 799-813.
60. Peplow, A. T., Oscarsson, J., and Dahlberg, T., Review of research on ballast as
track substructure, Report F189, Solid Mechanics, Chalmers University of
Technology, Gothenburg, Sweden, 1996.
61. Jacobsson, L., Review of research on railway ballast behavior-experimental
findings and constitutive models Report F208, Solid Mechanics, Chalmers
University of Technology, Gothenburg, Sweden, 1998.
62. Garg, V.K., and Dukkipati, R.V., Dynamic of Railway Vehicle Systems
Academic Press, 1984.
63. Diana, G., Cheli, F., Bruni, S., and Collina, A., Dynamic Interaction between Rail
Vehicles and Track for High Speed Train, Supplement to Vehicle System
Dynamics 24. Swets and Zeitlinger 1995.
64. Stone, D. H., Carlson, F. G., and Bachhuber, C. G., Effect of brake-system
components on wheel spalling Proceedings of the 1999 ASME/IEEE Joint
Railroad Conference, 1999, p 177-183.
65. Stone, D. H., "An Interpretive Review of Railway Wheel Shelling and Spalling",
Rail Transportation-1992, RTD - Volume 5, ASME, pp. 97 - 104.
151
66. Moyar, G. J., and Stone, D. H., An analysis of the thermal contributions to railway
wheel Shelling Wear, 144 (1991), pp117-138.
67. Sato, Y., Matsumoto, A., and Knothe, K., Review on rail corrugation studies,
Wear 253 (2002) 130139.
68. Sankar, T.S., and Samaha, M., Research in rail vehicle dynamics-state of the art,
Shock and Vibration digest, 1986, 18(2), 9-18.
69. Taheri, M.R., Ting, E.C., and Kukreti, A. R., Vehicle-guideway interactions: a
literature review, Shock and Vibration digest, 1990, 22(6), 3-9.
70. Thompson, D., Wu, T.X., and Armstrong, T., Wheel/rail rolling noise - the effects
of nonlinearities in the contact zone Tenth international congress on sound and
vibration, 7-10 July 2003, Stockholm, Sweden.
71. Thompson, D.J., Wheel-Rail Noise Generation, Parts I, Journal of Sound
Vibration 161 (3) (1993), pp. 387482.
72. Schwab, C.A., and Mauler, L., An interactive track/train dynamics model for
investigation system limits in high speed track Proceedings of the 13th IAVSD
Symposium ,Kingston, Canada, 1994, 502-514.
73. Sun, Y.Q., and Dhanasekar, M., A dynamic model for the vertical interaction of
the rail track and wagon system International Journal of Solids and Structures 39
(2002) 13371359.
74. Sun, Y. Q., and Simson. S., Vehicle Track modeling for rail corrugation
investigation Proceedings of the Joint Rail Conference (JRC) , March 16-
18,2005,Pueblo, Colorado, USA, p65-71.
75. Thompson, D., Theoretical modelling of wheel rail noise generation Journal of
Rail and Rapid transit, volume 205, part F, 1991, 137-149.
76. Schneider, E., and Popp, K., Noise Generation in railway wheels due to rail-wheel
contact forces. Journal of Sound and Vibration (1988), 120 (2), p 227-244.
77. Ganesan, N., and Ramesh, T.C., Free vibration analysis of composite railway
wheels, Journal of Sound and Vibration (1992), 153 (1), p 113-124.
78. Claus, H., and Schiehlen, W., Modeling and Simulation of Railway Bogie
Structural Vibration, Supplement to Vehicle Systems Dynamics 28(1998), p.538
552.
79. Schneider, E., Irretier, H., and Popp, K., Noise Generation in Railway Wheels due
to Rail-Wheel Contact Forces, Journal of Sound Vibration, 120(1988), pp. 227
244.
152
80. Fingberg, U., A Model of Wheel-Rail Squealing Noise, Journal of Sound
Vibration. 143(1990), p.365 377.
81. Ishida, M., and Miura, S., Dynamic Wheel load on long/short wavelength of track
irregularities, Proceedings of An IHHA conference on Freight Car Trucks/Bogies,
Montreal, Quebec, Canada, June 9-12, 1996 p-7.5-7.8.
82. Nielsen, J.C.O, and Johansson, A., Out of round railway wheels a literature
survey Journal of Rail and Rapid transit, volume 214, part F, 1995, p79-91.
83. Hempelmann, K., Hiss, F., Knothe, K., and Ripke, B., The Formation of Wear
Patterns on Rail Tread. Wear 144 (1991), pp.179195.
84. Clarence, W.S., Vibration fundamental and practice CRC press, USA, 2004.
85. Jin, X., Wen, Z., Kaiyun, W., Zefeng, W., and Weihua, Z., Effect of a scratch on
curved rail on initiation and evolution of rail corrugation Tribology International
37 (2004) 385394.
86. Cai, Z., Modeling of Rail track dynamics and wheel rail interaction Ph D thesis,
Dept. of Civil Engineering, Queens University, Kingston, Ontario, Canada 1992.
87. Dukkipati, R.V., and Dong, R., Idealized steady state interaction between railway
vehicle and track Proceedings of Institution of Mechanical Engineers Volume 213
Part F, 1999, p15-29.
88. Baeza, L., Roda, A., and Nielsen, J.C.O., Railway vehicle/track interaction
analysis using a modal sub-structuring approach Journal of Sound and Vibration
293 (2006) 112124.
89. Snyder, T., Stone, D.H., and Kristan, J., Wheel Flat and Out-of round formation
and growth Proceedings of the 2003 IEEE/ASME Joint Rail Conference April 22-
24, 2003 Chicago, Illinois.
90. Jergeus, J., Odenmarck, C., Lunden, R., Sotkovszki, P., Karlsson, B., and Gullers,
P., Full-scale railway wheel flat experiments Proceedings of Institution of
Mechanical Engineers Volume 213 Part F, 1998,pp1-13.
91. Lonsdale, C., Dedmon, S., and Pilch, J., Effects of increased gross rail load on 36-
inch diameter freight car wheels available at
www.standardsteel.com/rdpapers/2001
92. Valdivia, A., A linear dynamic wear model to explain the initiating mechanism of
corrugation In: Proceedings of the 10th IAVSD-Symposium on the Dynamics of
Vehicles on Roads and on Tracks. Amsterdam, 1988, pp. 493496.
153
93. Suda, Y., Komine, H., Iwasa, T., and Terumichi, Y., Experimental study on
mechanism of rail corrugation using corrugation simulator Wear 253 (2002) p162
171.
94. Johansson, A., and Andersson, C., Out-of-round railway wheelsa study of wheel
polygonalization through simulation of three-dimensional wheelrail interaction
and wear Vehicle System Dynamics Vol. 43, No. 8, August 2005, 539559.
95. Tassilly, E., and Vincent, N., A linear model for the corrugation of rail, Journal of
Sound and Vibration (1991), 150, p 25-45.
96. Eric, M., and Kalousek, J., Identifying and interpreting railway wheel defects
Proceedings of An IHHA conference on Freight Car Trucks/Bogies, Montreal,
Quebec, Canada, June 9-12, 1996 p-5.8-5.21.
97. Grassie, S. L., A contribution to dynamics design of railway track Proceedings of
12
th
IAVSD symposium held at Lyon, France, 1991, pp 195-209.
98. Kalker, J. J., Survey of wheel rail rolling contact theory vehicle system dynamics,
1979, volume 8, p317-358.
99. Kalker, J.J., A simplified theory for non-Hertzian contact Proceedings of 8
th

IAVSD symposium, Cambridge, Massachusetts, USA, 1983 pp.295-302.
100. Elkins, J.A., Prediction of wheel-rail interaction: the state-of-the-art Proceedings
of 12
th
IAVSD symposium, Lyon, France, 1991, pp1-27.
101. Tajaddini, A., and Kalay, S.F., Time to revise wheel-removal rules? Association of
American Railroads regulations, Railway Age, September 1995.
102. Pascal, J.P., and Sauvage, G., The available methods to calculate the wheel/rail
forces in non Hertzian contact patches and rail damaging Vehicle System
Dynamics, 22, p1993, 263275.
103. Yan, W., and Fischer, F.D., Applicability of the Hertz contact theory to rail-wheel
contact problems Archive of Applied Mechanics 70 (2000) 255-268.
104. Piotrowski, J., Contact loading of a high rail in curves: Physical simulation method
to investigate shelling. Vehicle System Dynamics, 17, p5779, 1998.
105. Kalker, J.J., and Piotrowski, J., The elastic cross-influence between two quasi-
Hertzian contact zones. Vehicle System Dynamics, 17, p337355, 1988.
106. Piotrowski, J., A Theory of Wheelset forces for two point contact between wheel
and rail. Vehicle System Dynamics, 11(2), 1982, p6387, 1982.
107. Shen, Z.Y., Hedrick, J.K. and Elkins, J.A.,A Comparison of Alternative Creep
Force Models for Rail Vehicle Dynamic Analysis Proceedings of the 8th IAVSD-
154
Symposium on the Dynamics of Vehicles on Roads and on Tracks. MIT,
Cambridge, MA, 1983, pp. 591605.
108. Transport Canada Railway Freight Car Inspection and Safety Rules available at
http://www.tc.gc.ca/railway/Rules/TC_0-06-1.htm.
109. Reilly, K., Railway Wheelsets, Rail Safety and Standards Board, London, UK,
June 2003.
110. Li, D., and Selig, E.T., Wheel/track dynamic interaction: track substructure
perspective, Vehicle System Dynamics, volume 24, Supplement, 1995, p183-196.
111. Yaua, J. D., and Yang, Y.B., Vertical accelerations of simple beams due to
successive loads traveling at resonant speeds Journal of Sound and Vibration 289
(2006)210228