TRANSPORT

ISSN 1648-4142 / eISSN 1648-3480

2015 Volume 30(2): 217–232
doi:10.3846/16484142.2015.1051108

THE DYNAMIC BEHAVIOUR OF A WHEEL FLAT OF

A RAILWAY VEHICLE AND RAIL IRREGULARITIES
Marijonas Bogdevičius1, Rasa Žygienė2, Stasys Dailydka3, Vilius Bartulis4,
Viktor Skrickij5, Saugirdas Pukalskas6
1, 2, 4, 5Deptof Transport Technological Equipment, Vilnius Gediminas Technical University, Lithuania
3Dept
of Railway Transport, Vilnius Gediminas Technical University, Lithuania
6Dept of Automobile Transport, Vilnius Gediminas Technical University, Lithuania

Submitted 7 July 2014; resubmitted 16 December 2014, 9 May 2015; accepted 10 May 2015

Abstract. The article examines a mathematical model for the system ‘Railway Vehicle Wheel–Track’ that allows exam-
ining the interaction between a wheel flat and a rail in the vertical plane. The dynamics of the railway track is described
using the finite element method while that of the soil and vehicle is expressed applying discrete elements. The model
is used for assessing physical and mechanical properties, the roughness of the wheel, rail surface and their geometry.
The analysis of the dynamic system ‘Railway Vehicle Wheel–Track’ has been conducted. In accordance with the revised
method, forces arising from contact between the wheel flat and the rail are possible to be determined in a more precise
way. The article presents and analyses the results of a mathematical experiment on this system.
Keywords: rail; wheel flat; defects; contact; dynamics; numerical method.

Introduction For dealing with the problems of railway wheel–rail

Along the evolving technology, the number of the means
of transport is increasing, traffic is getting more intense
worldwide and rail traffic is growing. Therefore, spe-
cial attention is paid to railway vehicles, infrastructure
maintenance and traffic safety. In the case of the inter-
action between a moving train wheel and the rail, the
forces resulting from their physical condition, metal
stress (Ahmetzyanov 2003), deformation, noise, etc. are
formed. The article studies the effects of various work-
ing conditions on the surface of the rail (Donzella et al.
2009). Axles can move laterally because of the forces re-
sulting from the interaction between the wheel and the
rail, while a lateral displacement can lead to wheel or
track alterations.
A number of models for wheel–rail interference
have been developed. However, when formulating
wheel–rail issues and setting targets, the scientist pre-
sents (Yazykov 2004; Uzzal 2012; Wu, Thompson 2001;
Torstensson et al. 2011) different solutions to this prob-
lem. A vertical non-linear Hertzian contact model based
on research into the behaviour of two moving cylinders
is widely used for investigation of the wheel–rail interac-
tion. The assumption that a very small contact area is ob-
served inside the contact of two bodies has been made.
interaction, finite element models have been designed
(Recuero et  al. 2011). Numerical models (Wasiwitono
et al. 2007; Zhu et al. 2007; Wu, Thompson 2001; Bezin
et al. 2009) allow predicting the dynamic interaction of
the ‘wheel–rail–soil’ system. Vibrations arising in the
system are divided into finite elements.
Dynamic models allow to predict the state of the
system or the influence of a dynamic system in the event
of having a variety of factors. When performing tests
on dynamic behaviour, the railway track and vehicle
change, and the movements and degrees of freedom are
taken into account. A number of dynamic models hav-
ing various degrees of freedom, taking into account dif-
ferent dynamic characteristics and various parameters,
are presented. The forces arising from the wheel–rail
interaction (Pombo et al. 2007) are calculated referring
to dynamic movement and evaluating tangential forces
under normal forces and velocity between two moving
bodies using the Kalker’s (1979) linear theory, heuristic
non-linear methods and Polach’s (2000) formulation.
Numerical dynamic simulation models are used for
researching trains. Many dynamic models are designed
to monitor and predict the movement of the means of
transportation and contact between the rail and the

Corresponding author: Marijonas Bogdevičius

E-mail: marijonas.bogdevicius@vgtu.lt
Copyright © 2015 Vilnius Gediminas Technical University (VGTU) Press
http://www.tandfonline.com/TRAN
218 M. Bogdevičius et al. The dynamic behaviour of a wheel flat of a railway vehicle and rail irregularities
wheel (Zhu et al. 2007). In order to reduce the number
of the degrees of freedom and decrease simulation time
while creating a dynamic system, the simplification of
elements to springs–dampers and masses is employed.
Sun et al. (2003) is described a dynamic model of
a three-dimensional vertical-horizontal multi-body sys-
tem consisting of a vehicle and the road using Kalker’s
creep and Hertzian contact methods, including the
wheel–rail interaction with a variety of loads.
Railways are systematically inspected during both
production and service processes using various non-
destructive element testing methods. One of the main
indicators characterizing the lifetime of the railway track
is its age. A large focus of scientists goes on defects in
the railway wheel or rail surface, their measurements,
growth monitoring and forecasting (Kasimov 2009;
Markov 2004). One of the main evaluation criteria for a
defected track are the stress intensity factor, as it can be
used for predicting growth in rail defects. Most defects
in the railway track are found in the railhead and mainly
emerge due to contact stress.
For analysing the status of soil, a system for sleep-
ers and rails (Aursudkij 2007; Herron et  al. 2009) at a
critical load and the impact force of equipment wheels
(Kaewunruen et al. 2014), a mathematical model has
been developed.
The load of the wheel plays an important role in
the bending moment of the sleeper; pulse duration is
greater, which explains such phenomena as the fracture
of sleepers from wheel impact.
Wheel and rail response, due to parametric excita-
tion by the varying dynamic stiffness of a periodically
supported rail, has been studied using a spatial quasi-
static method (Bogacz et al. 1993).
The ground properly laid under the sleeper can im-
prove or help with obtaining the desired features of the
upper construction of the track. In order to reduce the
forces affecting sleepers, soil stiffness should be reduced,
and, in order to minimize the bending moment of the
track, stiff soil should be chosen.
The conducted research has concluded that the tie
without support increases vibrations, reduces adhesion
and is likely to lead to an increase in wheel and rail de-
fects and noise, the risk of load stability and even human
life.
Concerning the design of the wheel or rail, there
are forces of different extent generated between the
wheel and the rail in a connection with potential rail
joints and welds, various train speeds, forces of break-
ing, locking of the wheel and the mass of the car. These
forces are able to cause wheel flats of a different size and
produce the latter vibrations of the rails and car. Geo-
metrical defects of the wheel and rail are rarely detected
at the initial stages of operation; however, these faults
increase over time. A number of scientists describe the
contact area of the wheel–rail as a point, or define wheel
geometry by an analytical function in mathematical
models (Barke, Chiu 2005; Bian et al. 2013; Kouroussis
et al. 2011; Nielsen, Abrahamsson 1992; Pieringer et al.
2014; Sackfield et al. 2006, 2007).
Scientists have researched the time and length
history of vehicle mass and displacement acceleration
by examining dynamic processes happening inside the
contact area (Ferrara et al. 2012; Kumar, Rastogi 2009;
Uzzal et  al. 2009; Wang et  al. 2015; Wasiwitono et  al.
2007). The spectra of accelerations of the vehicle mass
are calculated using discrete Fourier transform (Nielsen
et al. 2008; Wang et al. 2015, Wasiwitono et al. 2007).
The analysis of scientific articles showed there was
no single method according to which dynamic processes
occurring in the interaction between vehicle flats and
the track were developed. Each author has preferred dif-
ferent simulation parameters.
The article proposes the revised method for deter-
mining forces arising from the contact of the wheel flat
and the rail and other characteristics of contact. Geo-
metrical irregularities of the wheel and flat, the uneven-
ness of the rail and the micro-unevenness of the bodies
in the contact area are assessed in the revised model.
Particular attention is paid to determining the forces
arising in the contact of the wheel flat and the rail. The
area of contact is divided into a number of intervals at
any given time between the bodies in contact. The in-
terval may extend from 1 to 100 µm. At the points of
contact, possible short-term intervals between the wheel
and the rail are assessed, and therefore, at each time
point between the contacting surfaces, the distribution
of the load varies.
According to the revised method, velocities of slip,
relative longitudinal slip, normal and friction forces and
the moment of friction torque forces acting on the wheel
are established in the contact area between two surface
points in contact at each point in time. Using this infor-
mation allows determining the average values of those
characteristics in the contact area at each point in time.
In regard to the indicated revised method that de-
termines contact force, wheel and rail depreciation prob-
lems, loads on the bearing of the wheel and impact on
soil can be investigated. The theoretical part deals with
the revised method of contact force allowing the deter-
mination of forces acting in the contact and character-
istics of the contact.
The article is aimed at revealing all stages of the
considered issues selected by the scientists and refers to
possible ways of solving them.
1. A Vertical Dynamic Model for the Interaction
between the Railway Track and the Vehicle
The rail profile is considered as a stochastic function
where unevenness parameters of the actual rail are
measured. The mathematical model of the wheel and rail
contact allows us determining the forces resulting from
contact when assessing defects. The carried out investi-
gation is focused on identifying contact forces resulting
from the wheel and rail contact under various defects.
The mathematical model was created in order to analyse
the main defects of the vehicle wheel.
A computational model is composed of a vehicle
and the road divided into finite elements. The article
Transport, 2015, 30(2): 217–232 219

analyses the task of nonlinear dynamics. It is assumed The potential energy of beam finite element Lp con-
that gaps can occur between two bodies in contact and sists of the potential energies of bending Ep,b, tension
between the rail and the sleeper. The equations of the force Ep,T and elastic foundation Ep,F:
system elements are set employing d’Alembert’s princi-
ple. E p = E p,b + E p,T + E p, F , (3)
The issues of nonlinear dynamics are considered
where:
applying the numerical methods of Newton–Raphson Le 2
1  ∂ 2 w ∂ 2 w0 
and Newmark (1947). E p,b =
2 ∫ ER J yR ( x )  ∂x 2 − ∂x 2  dx =
0
1.1. The System of Equations ‘Railway Vehicle Wheel– 1
Track’ for the Rail
2
{qe }  K1e  {qe } − {qe } {F01e } + E p,b0 ,
T T
(4)
When studying rail interaction with the wheel–rail and where: ER is a modulus of rail elasticity; J yR ( x ) is the
sleepers, it can be assumed that the rail deforms at X in
the Z plane and can be evaluated as follows: axial moment of inertia; {F01e }, E p,b0 ,  K1e  is the
–– variable characteristics of the rail cross-section force vector of finite elements, potential energy evaluat-
(rail unevenness is measured); ing the initial shape of the rail and stiffness matrix re-
–– rail interaction with the roadbed as a tough base; spectively:
–– potential space between the rail and roadbed;
–– the length of the rail and the wheel–rail contact {F01e } = K1e  {w0e } ;
and geometrical imperfections in it;
–– the impact of axial forces in the rail on stiffness E p,b0 =
{w0e }  K1e  {w0e }
T
;
(because of difference in temperature); 2
ER J yR ( x )  ∂2 N w ( ξ ) 
–– the initial deformation of the rail; T
L 2
∂ N w (ξ) 
–– a gap between the sleeper and the rail.  K1e  = ∫  ∂î 2

L3e
 2
 dξ.
The dynamics of the rail has been considered ap- 0   ∂î 
plying the finite element method. Two node beam ele-
The potential energy of the finite element under
ment is used (Fig. 1).
acting tension force is equal to:
Le 2
1  ∂w ∂w 0 
Z
h(x)
E p,T =
2 ∫ Fa  ∂x − ∂x  dx =
0
1
w2
{q }T K  {q } − {qe }T {F02e } + E p,T 0 , (5)
2 e  2e  e
where: Fa is tension force; {F02e }, E p,T 0 ,  K 2e  are the
) X
E Z(x
(x),
w1
AR(x), J yR 2 Y force vector of finite elements, potential energy, the stiff-
E R,
ness matrix of the finite element under operating axial
force respectively,
1 Y
{F02e } = K 2e  {w0e };
Fig. 1. Parameters for assessing rail unevenness
E p,T 0 =
{w0e }  K 2e  {w0e }
T
;
2
Displacement w and initial displacement w0 along T
the Z axes of the beam finite element are equal to: ∂N w ( ξ )  Fa  ∂N w ( ξ ) 
L

 K 2e  = ∫   
∂ξ  Le  ∂ξ 
 dξ .
w =  N w ( ξ ) {qe }; 0
The potential energy of the finite element lying on
w0 =  N w ( ξ ) {w0e } , (1) the foundation is equal to:
L
where: {w0e } is the initial displacement vector; {qe } is 1
( 2
k ( x ) w ( x ) − w0 ( x ) dx = )
2 ∫0 S
E p, F =
a displacement vector;  N w ( ξ ) is a matrix of shape
functions. 1
The matrix of shape functions equals: {q }T K  {q } − {qe }T {F03e } + E p,F 0 , (6)
2 e  3e  e
 N w ( ξ ) = 1 − 3ξ2 + 2ξ3 − Le ξ 1 − 2ξ + ξ2 ,
 ( ) where: kS (x ) is the coefficient of the stiffness of the elas-
tic foundation; {F03e }, E p, F 0 ,  K 3e  are the force vector
3ξ2 − 2ξ3 Le ξ2 (1 − ξ ) , (2) of the finite element, the potential energy of the finite
x element, the stiffness matrix of the finite element under
where: ξ = is a local coordinate; Le is the length of the present elastic basis respectively,
Le
the finite element. {F03e } = K 3e  {w0e };
220 M. Bogdevičius et al. The dynamic behaviour of a wheel flat of a railway vehicle and rail irregularities

1 ξb
E p, F 0 =
2
{w0e }  K 3e  ;
T
{Fcontact } = ∫ N w (ξ) q (ξ) Le dξ ,
T
(13)
L ξa
T
 K 3e  = ∫  N w ( ξ ) kS (x )  N w ( ξ )Le dξ. where: q ( ξ ) is a distributed load in the contact area be-
0 tween the rail and the wheel.
The damping matrix of the beam finite element of The system of equations for the beam finite element
the rail is equal to: of the rail is equal to:

Ce  = α  Me  + β  K e  , (7)  Me  {qe } + Ce  {q e } +  K e  {qe } = {Fe }. (14)

where: a, b are optional coefficients;  Me  is the mass
matrix;  K e  is the stiffness matrix. 1.2. The Equation System ‘Railway Vehicle
Mass matrix of the beam finite element of the rail Wheel–Track’ for the Track
is equal to: In order to more accurately determine the interaction
1
T
 Me  = ∫  N w ( ξ ) ρR AR  N w ( ξ ) Le dξ , (8) between the rail and the wheel–rail, researchers assess
0
the gap properties of the railway sleeper and the sleeper–
where: rR is rail material density; AR is the rail cross- roadbed.
section area. The presented mathematical model for the system
The stiffness matrix is equal to: ‘Railway Vehicle Wheel–Track’ can be used for investi-
gating the interaction between the wheel of rolling stock
 K e  =  K1e  +  K 2e  +  K 3e  . (9) and the rail, when the wheel has or has no flat.
For assessing the impact of the railroad bed on the
The load vector of the finite element is equal to:
system rail–wheel–bogie, the following assumptions and
{Fe } = {Fw } + {FSe } + {Fcontact }, (10) evaluations can be made:
–– the gap of the rail and railroad bed between two
where: {Fw } is a vector of weight force, sleepers i and i+1 at the midpoint of j (Fig. 2);
Le –– the interaction between soil layers under adjacent
T
{Fw } = ∫ N w (ξ) qw Le dξ ; (11) rails.
The dynamic model for the system ‘Railway Vehicle
0
{FSe } is a force vector depending on axial force, bending Wheel–Track’ and its unevenness is presented in Fig. 2.
The potential energy of an elastic element and a
and an elastic basis of the rail,
dissipation function of damping element e pad ,i between
{FSe } = {F01e } + {F02e } + {F03e } ; (12) the rail and the ith sleeper are equal to:

{Fcontact } is a vector of contact force, 1 2

(
E p, pad ,i = k pad ,i wi − qsl ,i H δ pad ,i ;
2
(15) ) ( )
qbg4
mbg4
a) b) c)
ebg34 qj qj
qbg3 qj qj
mbg3
ebg23 ke Fe Ce e ke Ce
e
qbg2
mbg2
qi e qi
ebg34 qi qi
qbg1
mbg1
vtrain
ebg0
Fa i j k Fa

epad,i epad,j epad,k

qsl,i qsl,j qsl,k
msl,i msl,j msl,k
es34,i es34,j es34,k
qs3,i qs3,j qs3,k
ms3,i es33,i,j ms3,i es33,j,k ms3,k
es23,i es23,j es23,k
qs2,i qs2,j qs2,k
ms2,i es22,i,j ms2,j es22,j,k ms2,k
es12,i es12,j es12,k
qs1,i qs1,j qs1,k
ms1,i es11,i,j ms1,j es11,j,k ms1,k
es01,i es01,j es01,k

Fig. 2. Dynamic model for the system ‘Railway Vehicle Wheel–Track’: a – scheme; b – nonlinear elastic-damping discrete
element; c – linear elastic-damping discrete element
Transport, 2015, 30(2): 217–232 221

1
)2 H (δ pad,i ) ; {q }
T
Φ pad ,i = c pad ,i w i − q sl ,i
2
( (16) bj = qsl , j qs1, j qs 2, j qs 3, j  ; (23)

{qRij }
T
δ pad ,i = qsl ,i − wi − ∆ sl ,i , (17) = wi w j  .
( )
where: H δ sl ,i is the Heaviside function; ∆ sl ,i is gaps
1.3. Mathematical Model for the Wheel–Rail Contact
between the rail and the ith sleeper; k pad ,i , c pad ,i are the
stiffness and damping coefficients of the pads; qsl ,i , q sl ,i For analysing the interaction between the rail and the
are the displacement and velocity of sleepers respective- wheel, it is assumed that:
ly; wi, w i are the displacement and velocity of the rail. –– the wheel–rail profile has potential defects
The potential energy function of elastic ballast ele- (Fig. 3b);
ments between ith and jth sleepers is equal to: –– the unevenness of the rail surface is possible.
1 As regards the mathematical model for the wheel
E p,b,i , j =  ks 01,i q2 + ks12,i qs 2,i − qs1,i +
2  s 1, i
2
( ) and rail, micro-unevenness, localized slip, normal and
tangential forces and the moments of forces along the
(
ks 23,i qs 3,i − qs 2,i )2 + ks34,i (qs 4,i − qs3,i )2 + length of the wheel–rail contact are assessed.
Considering the above introduced mathematical
( ) +k (q )
2 2
ks11,i , j qs1,i , − qs1, j s 22,i , j s 2,i − qs 2, j + model, contact length is divided into small sections
where force is set in contact using Hertzian contact
(q )  ,
2
ks 33,i , j − qs 3, j (18) theory.
s 3,i
To determine unevenness in the contact zone of the
where: ks 01,i , ks12,i , ks 23,i , ks 34,i , ks11,i , j , ks 22,i , j , wheel and the rail and to use Hertzian contact theory,
ks 33,i , j are stiffness coefficients of ballast elements;  a curvature radius of the rail must be known. Thus, rail
qs1,i , qs 2,i , qs 3,i are displacements of ballast and sub unevenness is described using the second degree Hermit
 
ballast at the ith point. polynomials.
A dissipation function of damping ballast elements In order to assess rail unevenness, the rail is divided
between ith and jth sleepers is equal to: into a number of sections NR. This number may not co-
1 incide with a finite number of rail elements. The vertical
Φ s 01,i =  cs 01,i q 2 + cs12,i q s 2,i − q s1,i
2 s1,i
( )2 + unevenness of the rail in each section is described as
follows:
(
cs 23,i q s 3,i − q s 2,i )2 + cs34,i (q s 4,i − q s3,i )2 + (2 ) (2) d ∆Ζ i +
∆Ζ ( ξ ) = H 01 ( ξ ) ∆Ζ i + LRe H11 ( ξ )
dx
( ) + c (q )
2 2
cs11,i , j q s1,i , − q s1, j s 22,i , j s 2,i − q s 2, j + 2
(2) d ∆Ζ i + H (2) ξ ∆Ζ +
L2Re H 21 ( ξ ) 02 ( ) i +1
dx 2
(q )  ,
2
cs 33,i , j − q s 3, j (19)
s 3,i
(2) d ∆Ζ i +1 + L2 H (2) ξ d 2 ∆Ζ i +1 , (24)
LRe H12 ( ξ ) Re 22 ( )
where: cs 01,i , cs12,i , cs 23,i , cs 34,i , cs11,i , j , cs 22,i , j , cs 33,i , j dx dx 2
are damping coefficients of ballast elements. where: LRe is the length of the finite element of the rail;
The system of equations for sleepers, ballast and DZi, DZi+1 are vertical unevenness at the ith and (i+1)
sub-ballast blocks is established from the second order (2 )
th point; H ik is the Hermit polynomial of the second-
Lagrange equation: order:

  { }   { }
  { } { }
 Mi , j  qi , j + Ci , j  qi , j +  K i , j  qi , j = Fi , j , (20) (2 )
H 01 ( ξ ) = 1 − 10ξ3 + 15ξ 4 − 6ξ5 ;

{ }
where:  Mi , j  , Ci , j  ,  K i , j  , Fi , j are mass, damp- (2 )
H11 ( ξ ) = −6LRe ξ3 + LRe ξ + 8LRe ξ 4 − 3LRe ξ5 ;
ing and stiffness matrices of the ballast and sub-ballast
block and a part of the rail shown in the Appendix 1; (2 ) 1 3 3 1
H 21 ( ξ ) = L2Re ξ2 − L2Re ξ3 + L2Re ξ 4 − L2Re ξ5 ;
{ } { }{ }{ }
Fi , j is a vector of forces; qi , j , qi , j , qi , j are the
(2 )
2 2 2 2
vectors of displacement, velocity and the acceleration of H 02 ( ξ ) = 10ξ3 + 15ξ 4 + 6ξ5 ;
ballast, the sleeper and the rail.
{ }
The vector qi , j is equal to: (2 )
H12 ( ξ ) = −4 LRe ξ3 + 7 LRe ξ 4 − 3LRe ξ5 ;

{ } 
{ } { }
T T T
qi , j = {qbi }
T
qbj qRij  , (21) (2 ) 1 1
  H 22 ( ξ ) = L2Re ξ3 − L2Re ξ 4 + L2Re ξ5 . (25)
2 2
{ } { }
where: {qbi } , qbj , qRij are the vectors of the dis- The derivative of the vertical unevenness of the rail
placement of the ballast and sub ballast block and a part is equal to:
of the rail: d ∆Ζ ( ξ ) d ∆Ζ ( ξ )
= =
{qbi } T
= qsl ,i qs1,i qs 2,i qs 3,i  ; (22) dx LRe dξ
222 M. Bogdevičius et al. The dynamic behaviour of a wheel flat of a railway vehicle and rail irregularities

 dH (2) (2 ) The irregularities of the rail are described as func-

1 01 dH11 d ∆Ζ i tion ∆Z R ( x ) .
 ∆Ζ i + LRe +
LRe  dξ dξ dx Fourier series are applied widely in engineering to
(2 ) (2 ) solve different problems. In order to describe the geom-
dH 21 d 2 ∆Ζ i H 02 etry of the wheel flat mathematically, Fourier series are
L2Re + ∆Ζ i +1 +
dξ dx 2 dξ the most suitable. However, determining the number of
harmonics is very important as it affects the accuracy of
(2 )
dH12 d∆Ζ i +1
(2 )
dH 22 d 2 ∆Ζ i +1  the obtained results.
LRe + L2Re . (26) A profile of the vehicle wheel is defined as a func-
dξ dx dξ dx 2 
tion of radius variable, depending on the polar angle.
A track geometry car was used for measuring 1 kil- The radius RW (θ ) of the wheel profile is described us-
ometre of railway tracks, and the received results were ing Fourier series:
recorded at the increments of 0.25 m. This stretch has NH
been broken down into sections each of which consists RW (θ ) = ∑ ack cos ( kθ ) + ask sin ( kθ ) , (30)
of 5 points. The vertical unevenness of the rail as well k =1
as the first and second derivatives, with respect to the x
where: ack, ask are Fourier coefficients; NH is the number
coordinate, are calculated as:
of harmonics.
 5
A real vehicle’s wheel flat and a calculation scheme
 ∆Ζ R (ξ R ) = ∑ N i (ξ R ) ∆Ζ R,i ; for the profile of the vehicle’s wheel flat are shown in
 i =1
Fig. 3.
i( R)
 d ∆Ζ 5 dN ξ
 1 The depth of the wheel flat is described as follows:

R
= ∑
LRe i =1 dξ R
∆Ζ R,i ; (27)
 dx  2
 LF  
 d 2 ∆Ζ 1 5 d 2 N i (ξ R ) 
∆ F = RW 0 1 − 1 −   , (31)

 dx 2
R
= ∑
L2Re i =1 dξ2R
∆Ζ R,i , 

 2RW 0  

where: RW0 is a nominal wheel radius; LF is the length
where: shape functions N i (ξ R ) , i = 1,...,5 are equal to: of the wheel flat.
(
N1 (ξ R ) = ξ R (ξ R − 1) 4ξ2R − 1 / 6 ; ) a)
N 2 (ξ R ) = −4ξ R (2ξ R − 1) ( ξ2R )
−1 / 3 ;

( )(
N 3 (ξ R ) = ξ2R − 1 4ξ2R − 1 ; )
(
N 4 (ξ R ) = −4ξ R (2ξ R + 1) ξ2R − 1 / 3 ; )
(
N 5 (ξ R ) = ξ R (ξ R + 1) 4ξ2R − 1 / 6 , ) (28)

where: xR is local coordinate,

 ( x − xi ) 
ξR =  2 − xi  − 1 , ξ R ∈  −1,1 .
 xi +1 
The relation between the x coordinate and local co-
ordinate xR is equal to: b) Z

 ξ 
x =xi +  1 + R  ( xi +1 − xi ) , x ∈  xi , xi +1  .
2
LF

 2 
After using the functions N i (ξ R ) , i = 1,...,5 of the
F

1
sections of the vertical unevenness of the rail, the first
2
and second derivatives, with respect to the x coordinate,
are calculated as: 1

 5 X

 ∆Ζ R = ∑ N i (ξ R ) ∆Ζ R,i ;

)
w (

i =1
1 5 dN i (ξ R )

R

 d ∆Ζ R
 = ∑
LRe i =1 dξ R
∆Ζ R,i ; (29)
 dx
 d 2 ∆Ζ 1 5 d 2 N i (ξ R )

 dx 2
R
= ∑
L2Re i =1 dξ2R
∆Ζ R,i . Fig. 3. Vehicle’s wheel flat (a) and calculation scheme for the
profile of the vehicle’s wheel flat (b)
Transport, 2015, 30(2): 217–232 223

While operating, the profile of the railway wheel– 

rail is changing and affects wheel–rail interaction. CW
Wheel flats (Fig. 3a) are most commonly caused by
an uneven surface, temperature differences, etc. Wheel
flats of the axle occur on the tread surface. Wheel dam- RW(i–1) RW(i+2)
age is usually determined visually or by template sup-
port, depending on staff qualification. To control the
rolling surface, other diagnostic methods are used. For
example, the application of Lenz and vibrant force sen- i–1 i i+1 i+2
whe
el
sors assist in measuring the critical values of the wheel

surface with the help of comparison.
Le–1 Le Le+1
For determining wheel–rail intersection points, de-

ra
il
Lcontact
fining the X and Z coordinates of the wheel–rail and the
rail is necessary.
Penetration rate at point k with local coordinate xk ()
of contact is:
δ k =  N w (ξ k ) {qe } + ∆Ζ R cos (φk ) − =0
1  2  3  4 ...  i  i+1  N–1 N = 1 

(q − ( R
bg 1 W0 )
− RW ( Ψk ) sin ( Ψk ) , ) (32)
Le

if 0 ≤ φnW ≤ π : q()

0 ≤ Ψ k ≤ φnW , θk = 2π − φnW + Ψ k ;

 =0
φnW < Ψ k ≤ 2π, θk = Ψ k − φnW ; 1  2  3  4 ...  i  i+1  N–1 N = 1 
Le
if π < φnW ≤ 2π :
Fig. 4. The contact of the wheel and rail: a – changes
0 ≤ Ψk ≤ φnW , θk = 2π − φnW + Ψk ; in penetration in the contact area; b – chart of speed and
 forces operating at point k of the vehicle’s wheel and rail
φnW ≤ Ψk ≤ 2π, θk = Ψk − φnW ;
Ωt
φnW = 2π (a − nW ) , when a = ; Force Fcontact ( x ) appearing at the contact point is
2π
determined using Hertzian contact theory. In addition,
 Ωt  the hysteresis of contact forces is evaluated:
nW = integer  ;
 2π 
x k = xc − RW (θk ) cos ( Ψk ) ,
( ) ( )
Fcontact ( x ) = kRW ( x ) δ ( x ) D δ ( x ) H δ ( x ) , (36)
n

where:
where: RW (θk ) is the radius of the wheel–rail profile; 4
qbg1 is the vertical displacement of the wheel–rail; {qe } kRW ( x ) = E R(x) ; (37)
3 ekv
is a displacement vector for the finite element of the rail;
1 1 − ν2R 1 − νW 2
Yk is the angle (Fig. 4). = + ; (38)
The deviation angle: Eekv ER EW
R ( x ) RW
dw R ( xk ) 1  ∂N w ( ξ k )  R(x) = R ; (39)
tg (φk ) = − =−   {qe }. (33) RR ( x ) + RW
dx LRe  ∂ξ  δ
Penetration speed at the contact point is equal to: () 3
D δ = 1 + 1 − e 2
4  (
δ max
, ) (40)

V  ∂N w ( ξ k ) 
δ k =  N w (ξ k ) {q e } +   {q } + where: kRW is contact stiffness, e is speed restitution co-
LRe  ∂ξ  e efficient; ER, EW are the modulus of elasticity of the rail
V d ∆Ζ and wheel; nR, nW are Poisson’s coefficients of the rail
cos (φk ) − ∆Ζφ k sin (φk ) − and wheel; RW estimated by (Eq. 30); rail radius RR is
LRe dξ equal to: 3
∂Rk (θk )   2 
Ω sin ( Ψk ) , d ∆Ζ ( x k ) 
2
qbg 1 − (34) dw 
∂θ 1 +   +
where: W is the angular velocity of the wheel set,   dx  dx 
  d 2w
Ω = V RW 0 ; RR ( x k ) = + 2 , (41)
 d 2 ∆Ζ ( x k )  dx
 1  ∂N w ( ξk )  V  ∂ 2 N w ( ξk )    2 
φ k = − cos 2 ( φk )   dx
 L  ∂ξ  { e } L2  ∂ξ2  {qe }  .
q + 
 Re   Re    where the derivatives of the vertical unevenness of the
(35) rail equal:
224 M. Bogdevičius et al. The dynamic behaviour of a wheel flat of a railway vehicle and rail irregularities

 dH (2) ξ d i ∆Ζ 
d ∆Z ( xk ) ij ( k )
2 2 a)
1 j k
∑∑  dξ  ; (42) CW
= i
Z
dx LRe i = 0 j =1  dx 
 RW()
RW(1)
 d 2 H (2) ξ d i ∆Ζ 
d 2 ∆Ζ ( xk ) ij ( k )
2 2
1 j
dx 2 ∑∑
= 2
LRe

 d ξ 2 dx i
 . (43)

RW(k) RW(NP)
i = 0 j =1   1 N
(xk)
The alteration of penetration in the contact of the
wheel and rail evaluating the unevenness of the wheel
k
Z(x1) Z(xk) ZR(xNP)
and rail, forces operating in the kth point of the contact
X
of the wheel and rail and velocity are shown in Fig. 4.
Load distribution in the interval x ∈  x1 ...x NP  in RR(xNP)

the contact area is equal to:

dFcontact ( x ) dFcontact ( x ) dkRW ( x ) CR
q(x) = = +
dx dkRW dx
dFcontact ( x ) dR ( x ) dFcontact ( x ) dδ ( x )
b)
Z
+ ; (44)
dR dx dδ dx
CW qbg1
dkRW ( x ) dk  dR dRR 
= RW  ; (45)
dR  dRR dx 
X
dx RW()
F2K
k

2
RW
dR
= ; (46) k RW(k)
dRR ( R + R )2
R W FTK(xk) k V
dkRW 2 1
= Eekv . (47)
dR 3 R(x) Fig. 5. Wheel and rail geometric parameters
By using expressions (32–47), load distribution in for the contact area
the local coordinate system is equal to:
dFcontact ( ξ ) dFcontact ( ξ ) dkRW ( ξ ) where: NP is the number of points in the contact in-
q (ξ) = = + terval.
Le dξ dkRW Le dξ
The coordinate of resultant penetration at the finite
dFcontact ( ξ ) dR ( ξ ) dFcontact ( ξ ) dδ ( ξ ) element of the rail in contact zone xaver ,δ is equal to:
+ . (48)
dR Le dξ dδ Le dξ x NP

∫ xδ ( x ) dx
The load vector of the eth finite element of the rail
x1
in contact length (Fig. 5) is equal to: xaver ,δ = , (52)
x NP
ne ξi +1
{Fe } = ∑ ∫
T
 N w ( ξ ) q ( ξ )Le dξ , (49) ∫ δ ( x )dx
x1
i =1 ξi
where: x1 and xNP are rail coordinates x.
where: ne is the number of contact intervals in the fi-
Contact load and displacement vectors of the finite
nite element of the rail, when penetration is positive
elements of the rail and wheel as well as force acting in
(δ ≥ 0 ) in all contact interval ξ ∈ ξi , ξi+1  . the contact are equal to:
Vertical force affects the mass of the wheel in the
contact interval and is equal to:
ξ
{FR,BW }T = {{FRe }T , FBW }; (53)
ne i +1
FBW = ∑∑ ∫ qe ( ξ )Le dξ .
e i =1 ξ
(50)
{qR,BW }T = {qRe }T , qbg1  . (54)
i
The coordinate xcont ,aver ,q ∈  x1 ,..., x NP  of load Vector {FR, BW } is expanded in the surroundings
distribution operating in the contact interval is equal to:
x NP of point {qR, BW ,k } using Taylor series and is equal to:
∫ xq ( x ) dx
x1 {FR,BW ,k +1 } = {FR,BW ,k } + KT ,R,BW ,k  {∆qR,BW ,k }, (55)
xcont ,aver ,q = x NP
, (51)
where:  KT , RBC ,k  is the Jacobi matrix,
∫ q ( x )dx
x1
Transport, 2015, 30(2): 217–232 225

  KT , R, R,k   KT , R, BW ,k   When using expressions (49) and (63), resultant re-

 KT , RBC ,k  =    = sistance force and torque about the Y axis of the wheel
  KT , BW , R,k   KT , BW , BW ,k   depend on friction and normal forces in the contact and
   
are equal to:
  ∂ {F }   ∂ {FR }  
ne
 R
    ( ) ( )
FX , Friction = − ∑∑ FBW ,e ,i µ εe ,i cos α e ,i ; (65)
  ∂ {qRe }   ∂qbq1   e i =1
ne
  , (56)
  ∂ {FBW }   ∂ {FBW }   ( ) ( )
MY = ∑∑ FBW ,e ,i µ εe ,i RW Ψe ,i +
    e i =1
  { Re }   ∂qbq1  
∂ q
FBW ,e ,i RW ( Ψe ,i ) sin ( α e ,i ) ; (66)
{ }
where: ∆qRB ,k is an increment of displacement vector
where: e is the number of the finite element of the rail;
{ }
qR, BW ,k .
i is the point of the wheel–rail contact; ae,i is the angle
The system of nonlinear equations for the finite ele-
between the vertical axis of the wheel and radius.
ments of rails in contact with wheel mass is equal to:
1.4. System for Equations for a Railway Vehicle
{ } { } {
 M R, BW  qR, BW + CR, BW  q R, BW +  K R, BW  qR, BW − }
1/8 of the vehicle body with cargo mass, 1/4 of bogie
{ } { } {
 KT , R, BW ,k  ∆qR, BW ,k = FR, BW (t ) + FR, BW ,k , (57) } mass and 1/2 of wheel set mass have been assessed in the
mathematical model for the railway vehicle.
{
where: FR, BW (t ) } {
is force varying in time; FR, BW ,k } As for the revised method of contact force, the
is nonlinear force dependant on point k;  M R, BW  , wheel is divided into two masses, where mass mbg1 is in
  direct contact with the rail while mass mbg2 is the main
CR, BW  ,  K R, BW  are mass, damping and stiffness
mass of the wheel set. The use of the additional mass
matrices respectively,
mbg1 of the wheel enables a more accurate assessment
  M Re  0  of forces acting in contact and the calculation of kinetic
 M R, BW  =   ; (58) parameters for individual parts of the wheel.
 0 mbg 1   A system for the nonlinear equation for railway ve-
 
hicle movement out of contact is equal to:
 C  0 
CR, BW  =   Re 
 0
;
0 
(59)
  { }    { }
 Mbg  qbg + Cbg  qbg +  K bg  qbg =
 { }
 K 
 K R, BW  =   Re 
0  {F (q
BW bg qRe qbg q )} + {F (t )},
Re bg (67)
. (60)
{ }
T
 0 0  where: qbg = qbg 1 qbg 2 qbg 3 qbg 4  ; qbgi is the dis-
Knowing the contact points of wheel–rail length placement of the mass of the ith vehicle;  Mbg  , Cbg  ,
(Fig.  5), at each of these places, a longitudinal relative  
wheel slip can be determined in the following way: { ( )} {
 K bg  , FBW qbg qRe qbg q Re , Fbg (t ) are mass,
  }
∆V ( ξ ) damping and stiffness matrices, nonlinear load and
ε (ξ) = , (61) weight force vectors respectively. The introduced matri-
V
ces and vectors are shown in Appendix 2.
where:
The nonlinear load vector is equal to:
∆V = V − ΩRW ( Ψk ) cos ( Ψk ) − u k ; (62)

 hk + ∆Ζ (ξ k )   dN w (ξ k )  {F (q
BW bg qRe qbg q Re )} = F contact (q
qRe qbg q Re .
bg
 )
u k =    {q e } + The total nonlinear equation system for the whole
 Le   dξ 
movement system is equal to:
(hk + ∆Ζ (ξk ))  d2 Nw (ξk )  {q }, { } { }
V e (63)  M  {q} + C  {q } +  K  {q} + FNL (q, q ) = F (t ) , (68)
L2e  dξ
2

where: hk is the distance from the geometric centre to where:   M  , C  ,  K  , {FNL (q, q )} , {F (t )}
are
the surface of the rail head. mass, damping and stiffness matrices and nonlin-
Friction force in the contact interval is equal: ear load and external force vectors respectively;

{ }
{q}T = {qR }T , {qB }T , qbg  is the total displacement
T
ne
FTK = ∑∑ FBW ,e ,i µ εe ,i , ( ) (64)  
e i =1 vector; {qR } is a displacement vector of the rail; {qS } is
where: εe,i is a wheel slip in the ith point of the eth finite
∆V (ξi )
{ }
a displacement vector of ballast; qbg is a displacement
element of the rail, εe ,i =
V
( )
; µ εe ,i is the friction
vector of the vehicle.
The nonlinear total system of equations for mo-
coefficient between the rail and wheel. tion is solved applying Newmark (1947) and Newton–
226 M. Bogdevičius et al. The dynamic behaviour of a wheel flat of a railway vehicle and rail irregularities

Raphson method. The accuracy of the solution to the Table 1. Data calculation. Track and vehicle system
nonlinear problem is increased by combining the above Notation
Definition
methods.
{
Non-linear force FNL (q, q ) is extracted in the } Static load Fx = 100 kN
Taylor series at the surroundings of point {qk }: The second moment of the area of
the rail about Y axis J yR = 3.54 ⋅10 −5 m 4
{ } { }
FNL (q, q ) = FNL,k +  KT ,k  {∆qk } + CT ,k  {∆q k } , Cross-sectional area of the rail AR = 82.65 ⋅10 −4 m2
Poisson’s coefficient of the rail νR = 0.30
(69)
Elastic modulus of the rail ER = 206 GPa
where:
ρR = 7850 kg/m3
{
 ∂ FNL (qk , q k )
 KT ,k  = 
}  ; Rail density
Rail mass per meter mR = 65 kg/m
 ∂ {q}  Pad damping coefficient C pad = 45 kNs/m
  Pad stiffness k pad = 140 MN/m

CT ,k  = 
{
 ∂ FNL (qk , q k ) 
.
} Sleeper spacing
Sleeper mass
L p = 0.5435 m
msl = 265 kg
 ∂ {q } 
  Damping coefficient of ballast Cs11,i , j = 10 kNs/m
Then, the total system of equations (33), at the mo- Cs 22,i , j = 13 kNs/m
ment of time t + Dt, is equal to: Cs 33,i , j = 15 kNs/m
Cs 01 = 90 kNs/m
 M  {qt +∆t } + C  {qt +∆t } +  K  {qt +∆t } + Cs12 = 70 kNs/m

{ } { }
CT  ∆qt +∆t ,k +  KT  ∆qt +∆t ,k =
Cs 23 = 60 kNs/m
Cs 34 = 50 kNs/m

− {FNL (qt +∆t , qt +∆t )} + {F (t )}. Stiffness coefficient of ballast ks11,i , j = 15 MN/m
(70)
ks 22,i , j = 16 MN/m
When applying Newmark (1947) and Newton– ks 33,i , j = 17 MN/m
Raphson methods, the total system for linear algebraic ks 01 = 180 MN/m
equations is solved in each of kth iterations: ks12 = 170 MN/m
ks 23 = 160 MN/m
{
 At +∆t ,k  {∆qk } = − Pt +∆t ,k , } (71) ks 34 = 150 MN/m
where: Mass of ballast ms1 = 500 kg
 1 ms 2 = 300 kg
 At +∆t ,k  =   M  + ms 3 = 200 kg
 β∆t 2
1/8 of the car body mass mbg 4 = 8743 kg
γ
β∆t
( ) ( 
C  + CT ,k  +  K  +  KT,k   ;

) 1/4 bogie mass mbg 3 = 700 kg
mbg 2 = 640 kg
1/2 of the wheel set mass
{Pt +∆t ,k } =  M  {qt +∆t ,k } + C  {qt +∆t ,k } + Mass of the wheel in contact mbg 1 = 110 kg

Car body stiffness kbg 34 = 2,55 MN/m

 K  {qt +∆t ,k } + {FNL,k } − {F (t + ∆t )} , Bogie stiffness kbg 23 = 6.5 MN/m
Wheel set stiffness kbg 12 = 56 MN/m
where: b, g are the Newmark’s coefficients: γ = 1/ 2 ,
 
β = 1/ 4 ; Dt is integration time step; t is time. Damping coefficient of the car body Cbg 4 = 10 kNs/m
At the initial moment of time t = 0, displacement Damping coefficient of the bogie Cbg 3 = 100 kNs/m
{ }
vector qt +∆t ,k is determined from the equation sys- Damping coefficient of the wheel set
Damping coefficient of mass in
Cbg 2 = 50 kNs/m
tem:
(
  K  +  K  ∆q
    T ,k 

){ } {(
t +∆t ,k = − P qt +∆t ,k ;
(72)
)} contact Cbg 1 = 44.2 kNs/m

{ } {
 qt +∆t ,k +1 = qt +∆t ,k + ∆qt +∆t ,k . } { }

Wheel radius
Elastic modulus of the wheel
RW = 0.495 m
EW = 210 GPa
Poisson’s coefficient of the wheel νW = 0.30
2. Results of Mathematical Modelling Exponent n = 3/2
Maximal penetration velocity δ max = 10 m/s
2.1. Parameters for the System ‘Railway
Vehicle Wheel–Track’ Restitution coefficient e = 0.65
Friction coefficient µ = 0.3
In order to compile the mathematical model for the sys-
tem ‘Railway Vehicle Wheel–Track’, data on the four-axle
freight vehicle 12–9780 were used. It is assumed that the The distance between sleepers is 0.5435 m, the
vehicle moves at a speed of 100 km per hour. The fat of number of sleepers is equal to 41 and is divided into 10
the vehicle wheel is LF = 100 mm. A static load on the beam finite elements.
rail is 100 kN. It has been calculated that the rail has no The number points in the contact interval are equal
unevenness. to NP = 1001. The total time of vehicle motion is 0.560 s,
Transport, 2015, 30(2): 217–232 227

integration time step is Dt = 10–6 s. The total number of a)

unknowns is 970. 0.4955
The parameters for the developed system are pre- 0.495
sented in Table 1.
0.4945

2.2. Research and Analysis of the Dynamic Process 0.494

RW [m]
of the System ‘Railway Vehicle Wheel–Track’ 0.4935
The carried out research is aimed at determining the 0.493
forces operating under the interaction between the
0.4925
wheel flat and the rail and at investigating the impact
of forces on the dynamics of the rail. The wheel loses 0.492
0 0.1 0.2 0.3 0.4 0.5 0.6
contact with the railway track due to the rail, sleepers, t [s]
high stiffness of the rail bed and low damping, dynamic b)
properties of a vehicle, a high speed of the vehicle and a 0.4955
significant wheel flat.
0.495
The initial centre of the vehicle wheel coincides
with the geometrical centre of the sleeper. The initial 0.4945
position of the flat is in the middle of the top of the 0.494

RW [m]
wheel (Fig. 6). 0.4935
The initial displacements of the rail under a static
load on the rail of 100 kN are shown in Fig. 7. The initial 0.493

position of the wheel (t = 0) is on the eleventh sleeper 0.4925

(x = 5.47 m). 0.492
The geometrical parameters of the wheel flat in- 0.277 0.278 0.279 0.28 0.281 0.282
t [s]
clude RW0  = 0.495 m and LF  = 100  mm. A profile of
the wheel flat of the railway vehicle is described using a Fig. 8. Changes in the effective rolling radius of the rotating
four hundred and one harmonics (NH = 401) (Eq. 30) wheel flat: a – defined under the time from 0 to 0.6 s;
b – defined under the time from 0.277 to 0.283 s
(Figs 8–9).
LF
×10–7

0.5
v
RW–RW0 [m]

CW 0

–0.5

–1

0 20 40 60 80 100
x [m]

Csl Fig. 9. A deviation of wheel radius from the nominal radius

under the flat length of LF = 100 mm

Fig. 6. The initial position of the wheel

A deviation of wheel radius from the nominal ra-
–0.2 dius when the length of the flat is equal to LF = 100 mm
is shown in Fig. 9.
–0.4 Changes in vertical force FZ operating upon the
–0.6
wheel over time are shown in Fig. 10. Due to the wheel
w [mm]

flat (LF  = 100 mm, DF = 2.53 mm), load (100 kN) and
–0.8 vehicle velocity is 100 km/h, in time gaps from 0.388 to
0.398 s, wheel loses its contact with the rail (Fig. 10b).
–1
In order to determine whether the vehicle wheel
–1.2 has flats, forces operating on sleepers when the wheel
0 5 10 15 20 passes a certain section of railway rails have to be iden-
x [m] tified.
Fig. 7. The initial displacements of the rail under Forces operating from the rail (through the pad)
a static load equal to 100 kN are shown in Fig. 11.
228 M. Bogdevičius et al. The dynamic behaviour of a wheel flat of a railway vehicle and rail irregularities

600 300
a) b)
500 250
400
200
300
Fz [kN]

Fz [kN]
150
200
100
100

0 50

–100 0
0 0.1 0.2 0.3 0.4 0.5 0.39 0.392 0.394 0.396
t [s] t [s]

Fig. 10. Changes in vertical force FZ operating upon the wheel over time: a – to 0.55s; b – from 0.388 to 0.398 s

20 a) 1500

0
1000
–20

a [m/s ]
2
500
Fpad [kN]

–40

–60
0
–80 27 sleeper 28 sleeper
29 sleeper 30 sleeper
31 sleeper 32 sleeper –500
–100 0 0.1 0.2 0.3 0.4 0.5
33 sleeper 34 sleeper
35 sleeper 36 sleeper t [s]
–120
0.26 0.31 0.36 0.41 0.46 0.51 b) 1000
t [s]

Fig. 11. Forces operating on sleepers from the rail

(through the pad) 500
a [m/s2]

Forces on operating sleepers from the rail (through

pad) are shown in Table 2. 0

Fig. 11 and Table 2 show that an impact on the rail

occurs between the 31 and the 32, id est., at a distance of
0.2065 m from the centre of the 31th sleeper. –500
0.39 0.392 0.394 0.396 0.398 0.4
Impact force on the rail affects four sleepers, and t [s]

both sides obtain the mostly affected sleeper (31 sleeper). c) 30

When the wheel with no flat passes through the 20
rail, the load (from 31 sleeper) to sleepers distributes 10
almost evenly to both sides. However, when the wheel
a [m/s2]

0
has a flat, 31 and 32 sleepers are mostly loaded, and the
–10
load to other adjacent sleepers is almost evenly distrib-
uted to both sides. –20

The acceleration of the wheel is one of the most –30 mbg3

mbg4
characteristic parameters evaluating a dynamic load on –40
0 0.1 0.2 0.3 0.4 0.5
the rail and vehicle. t [s]
Changes of accelerations of masses mbg2, mbg3, mbg4 Fig. 12. Acceleration changes over time: a – mbg2;
due to vehicle wheel flat are shown in Fig. 12 b – mbg2 at time 0.39–0.40 s; c – mbg3 and mbg4

Table 2. Peak forces operating on sleepers from the rail (through pad) and their percentages when the wheel has or has no flat
27 28 29 30 31 32 33 34 35 36
Measurement units
Number of sleepers
Value of force when the wheel has no flat at the time of 0.3931 s
[kN] 0.9 –0.7 –6 –15 –26 –21 –9 –2 0.4 0.7
[%] –3 3 21 58 100 81 34 8 –1 –3
Value of force when the wheel has a flat at the time of 0.39504 s
[kN] 3.1 5.0 –7 –55 –111 –68 –14 4 3.5 0.5
[%] –3 –4 7 50 100 62 12 –4 –3 0
Transport, 2015, 30(2): 217–232 229

The maximum value of the acceleration of mass a) 103

mbg2 are equal to 1200 m/s2 and the acceleration of the 102
vehicle body mass mbg4 – 6.7 m/s2. However, (Fig. 12a),
101
the pending system shows that, up to the time of 0.30 s,

log(Fz) [kN]
a transitional process can be observed; therefore, the 100

maximum acceleration of the wheel is equal to 900 m/s2 10–1

when time t = 0.5 s and the acceleration of the mass of 10–2
the car body mbg4 is 0.7 m/s2.
10–3
Sudden changes in vehicle accelerations are seen
in the acceleration curve (Fig. 12a) (at time: 0.06, 0.17, 10–4
0 1000 2000 3000 4000 5000
0.28, 0.39, 0.50 s) when wheel impact occurs on the rail. f [Hz]
Acceleration spectra of the wheel set and verti-
102
cal forces occurring in contact between the wheel and b)
the rail (Fig. 13) are determined using discrete Fourier 101
transform.

log(Acceleration)
The maximum amplitude of vertical wheel force 100
(40 kN) can be observed at frequency f = 503 Hz that
is approximately obtained examining the system: body– 10–1
springs, i.e:
10–2
1 kRW 1 1011
f = = = 503 Hz.
( π)
2 ∑ mbgi (2π ) 104 10–3
0 1000 2000 3000 4000 5000
f [Hz]
The maximum value of the acceleration amplitude
is achieved at 1006 Hz, i.e., the frequency of acceleration Fig. 13. Parameters occurring in contact between the vehicle’s
is equal to 2f. wheel flat and the rail: a – amplitude of the vertical force;
The history of time for the displacements of vehicle b – amplitude of wheel set acceleration
mass in the dynamic system ‘Railway Vehicle Wheel–
Track’ is shown in Fig. 14. a) 0.00
The alterations of displacements, under an impact mbg2
between the wheel flat and the rail (Fig.  14), demon-
strate that the vertical displacements of the wheel set mbg3
are altered most, but the displacements of the bogie and –0.02
body vary to a small extent (Fig. 14a). The loss of contact
Displacement [m]

is observed by comparing the displacements of the wheel

flat (Fig. 14b), after impact forces (0.06, 0.17, 0.28, 0.39,
0.50 s) occur in the contact area due to the wheel flat.
–0.04

Conclusions
The article has proposed the revised contact force meth- mbg4

od, which, unlike the model of two or more interacting –0.06

0.00 0.07 0.14 0.21 0.28 0.35 0.42 0.49 0.56
points, is based on the split of the contact zone to a set t [s]
number of intervals. It allows determining forces and
other contact characteristics occurring in the contact b) –1.3
zone of the wheel and rail at any moment of time. –1.4
Describing the unevenness of the system ‘Railway –1.5
Vehicle Wheel–Track’ possibly appearing in contact be- –1.6
tween the wheel flat and the rail, short-term gaps be-
Displacement [m]

–1.7
tween the wheel and rail, and possible gaps between the
rail and sleepers have been evaluated. –1.8

The system ‘Railway Vehicle Wheel–Track’ has –1.9

been examined under a wheel flat of 100 mm, a velocity –2
of 100 km/h and a static load of 100kN. The computa- –2.1
tional model consists of the vehicle and the track divided –2.2 Rail Wheel
into finite elements. It is assumed, that the rail has no –2.3
irregularities in calculations. 0.00 0.07 0.14 0.21 0.28 0.35 0.42 0.49 0.56

The obtained results have shown that the wheel t [s]

loses its contact with the rail when a flat (LF  = 100 mm Fig. 14. The history of time for displacements: a – wheel flat
and DF  = 2.53 mm) appears in the wheel rail contact (L = 100 mm); b – wheel and rail
zone under a velocity of v = 100 km/h.
230 M. Bogdevičius et al. The dynamic behaviour of a wheel flat of a railway vehicle and rail irregularities
Comparative research has shown that the vertical
contact force of the wheel occurring from the wheel to
the rail is distributed into four sleepers to both sides.
The study has shown that, by using the revised con-
tact force method, the problems of wheel and rail wear
can be examined, and the forces occurring in contact
between the wheel flat and the rail can be identified. Its
size and the points where wheel flats occur can be de-
termined more accurately.
Further studies suggest examining the forces oc-
curring within wheel bearings, in contact between the
wheel flat and the rail, using the revised contact force
method.
Acknowledgements
This work has been supported by the European Social
Fund within the project ‘Development and application
of innovative research methods and solutions for traffic
structures, vehicles and their flows’, project code VP1-
3.1-ŠMM-08-K-01-020.
References
Ahmetzyanov, M. H. 2003. O mehanizme razvitiya kontakt-
no-ustalostnyh povrezhdenij v rel’sah, Vestnik Nauchno-
Issledovatel’skogo Instituta Zheleznodorozhnogo Transporta
(2): 41–45. (in Russian).
Aursudkij, B. 2007. A Laboratory Study of Railway Ballast Be-
haviour Under Traffic Loading and Tamping Maintenance:
Thesis submitted to the University of Nottingham for the
degree of Doctor of Philosophy. 234 p. Available from In-
ternet: http://eprints.nottingham.ac.uk/10321
Barke, D. W.; Chiu, W. K. 2005. A review of the effects of out-
of-round wheels on track and vehicle components, Pro-
ceedings of the Institution of Mechanical Engineers, Part F:
Journal of Rail and Rapid Transit 219(3): 151–175.
http://dx.doi.org/10.1243/095440905X8853
Bezin, Y.; Iwnicki, S. D.; Cavalletti, M.; De Vries, E.; Shahzad, F.;
Evans, G. 2009. An investigation of sleeper voids using a
flexible track model integrated with railway multi-body
dynamics, Proceedings of the Institution of Mechanical En-
gineers, Part F: Journal of Rail and Rapid Transit 223(6):
597–607. http://dx.doi.org/10.1243/09544097JRRT276
Bian,  J.; Gu,  Y.; Murray, M. H. 2013. A dynamic wheel–rail
impact analysis of railway track under wheel flat by finite
element analysis, Vehicle System Dynamics 51(6): 784–797.
http://dx.doi.org/10.1080/00423114.2013.774031
Bogacz,  R.; Krzyżynski,  T.; Popp, K. 1993. On vertical and
lateral dynamics of periodic guideways for maglev vehi-
cle, in R. Bogacz, K. Popp (Eds.). Dynamical Problems in
Mechanical Systems: Proceedings of the 3rd Polish-German
Workshop, 26–31 July 1993, Polish Academy of Sciences,
Warszawa, Poland, 219–233.
Donzella, G.; Mazzù, A.; Petrogalli, C. 2009. Competition be-
tween wear and rolling contact fatigue at the wheel–rail
interface: some experimental evidence on rail steel, Pro-
ceedings of the Institution of Mechanical Engineers, Part F:
Journal of Rail and Rapid Transit 223(1): 31–44.
http://dx.doi.org/10.1243/09544097JRRT161
Ferrara, R.; Leonardi, G.; Jourdan, F. 2012. Numerical model-
ling of train induced vibrations, Procedia – Social and Be-
havioral Sciences 53: 155–165.
http://dx.doi.org/10.1016/j.sbspro.2012.09.869
Herron, D.; Jones, C.; Thompson, D.; Rhodes, D. 2009. Char-
acterising the high-frequency dynamic stiffness of railway
ballast, in M. Pawelczyk, D. Bismor (Eds.). Proceedings of
the 16th International Congress on Sound and Vibration:
Recent Developments in Acoustics, Noise and Vibration
(ICSV16), 5–9 July 2009, Kraków, Poland, 1–8. Available
from Internet: http://resource.isvr.soton.ac.uk/staff/pubs/
PubPDFs/Pub11199.pdf
Kaewunruen, S.; Remennikov, A.; Aikawa, A.; Sakai, H. 2014.
Free vibrations of interspersed railway track systems in
three-dimensional space, Acoustics Australia 42(1): 20–26.
Kalker, J. J. 1979. The computation of three-dimensional roll-
ing contact with dry friction, International Journal for Nu-
merical Methods in Engineering 14(9): 1293–1307.
http://dx.doi.org/10.1002/nme.1620140904
Kasimov, B. R. 2009. Issledovanie rosta vnutrennih prodol‘nyh
treshhin v rel‘se v zone vykolov, in Aktual’nye Problemy So-
vremennyh Nauk – 2009: Internet-Konferenciya. Available
from Internet: http://www.rusnauka.com/17_APSN_2009/
Tecnic/48342.doc.htm (in Russian).
Kouroussis, G.; Gazetas, G.; Anastasopoulos, I.; Conti, C.; Ver-
linden, O. 2011. Discrete modelling of vertical track–soil
coupling for vehicle–track dynamics, Soil Dynamics and
Earthquake Engineering 31(12): 1711–1723.
http://dx.doi.org/10.1016/j.soildyn.2011.07.007
Kumar, V.; Rastogi, V. 2009. Investigation of vertical dynamic
behaviour and modelling of a typical Indian rail road ve-
hicle through bond graph, World Journal of Modelling and
Simulation 5(2): 130–138.
Markov, D. P. 2004. Zadir bokovyh poverhnostej rel’sov i
grebnej koles, Vestnik Nauchno-Issledovatel’skogo Instituta
Zheleznodorozhnogo Transporta (4): 40–43. (in Russian).
Newmark, N. M. 1947. Influence charts for computation of ver-
tical displacements in elastic foundations, University of Illi-
nois Bulletin 44(45): 1–14. Available from Internet: https://
www.ideals.illinois.edu/bitstream/handle/2142/4573/engi-
neeringexperv00000i00367.pdf
Nielsen, J. C. O. 2008. High-frequency vertical wheel–rail con-
tact forces: validation of a prediction model by field testing,
Wear 265(9–10): 1465–1471.
http://dx.doi.org/10.1016/j.wear.2008.02.038
Nielsen, J. C. O.; Abrahamsson, T. J. S. 1992. Coupling of physi-
cal and modal components for analysis of moving non-lin-
ear dynamic systems on general beam structures, Interna-
tional Journal for Numerical Methods in Engineering 33(9):
1843–1859. http://dx.doi.org/10.1002/nme.1620330906
Pieringer, A.; Kropp, W.; Nielsen, J. C. O. 2014. The influence
of contact modelling on simulated wheel/rail interaction
due to wheel flats, Wear 314(1–2): 273–281. http://dx.doi.
org/10.1016/j.wear.2013.12.005
Polach, O. 2000. A fast wheel–rail forces calculation computer
code, in The Dynamics of Vehicles on Roads and on Tracks:
Proceedings of the 16th IAVSD Symposium, 30 August – 3
September 1999, Pretoria, South Africa, 728–739.
Pombo, J.; Ambrósio, J.; Silva, M. 2007. A new wheel–rail contact
model for railway dynamics, Vehicle System Dynamics 45(2):
165–189. http://dx.doi.org/10.1080/00423110600996017
Recuero, A. M.; Escalona, J. L.; Shabana, A. A. 2011. Finite-
element analysis of unsupported sleepers using three-di-
mensional wheel–rail contact formulation, Proceedings of
the Institution of Mechanical Engineers, Part K: Journal of
Multi-body Dynamics 225(2): 153–165.
http://dx.doi.org/10.1177/2041306810394971
Transport, 2015, 30(2): 217–232 231

Sackfield, A.; Dini, D.; Hills, D. A. 2007. Contact of a rotating APPENDIX 1

wheel with a flat, International Journal of Solids and Struc-
tures 44(10): 3304–3316. Mass matrices of the ballast and sub-ballast block and a
http://dx.doi.org/10.1016/j.ijsolstr.2006.09.025 part of the rail:
Sackfield, A.; Dini, D.; Hills, D. A. 2006. The contact problem
for a wheel having a ‘flat’, Wear 261(11–12): 1265–1270.
http://dx.doi.org/10.1016/j.wear.2006.03.009   (
 Mi , j  = diag  Mbi ,bi   Mbj ,bj   M Rij , Rij  , (A.1.1)
      )
Sun, Y. Q.; Dhanasekar, M.; Roach, D. 2003. A three-dimen-
where:  Mbi ,bi  ,  Mbj ,bj  ,  M Rij , Rij  are the blocks of
sional model for the lateral and vertical dynamics of wagon-
track systems, Proceedings of the Institution of Mechanical the mass matrix of sub-ballast and a part of the rail:
Engineers, Part F: Journal of Rail and Rapid Transit 217(1):
31–45. http://dx.doi.org/10.1243/095440903762727339 (
 Mbi ,bi  = diag msl ,i ms1,i ms 2,i ms 3,i ; )
Torstensson, P. T; Nielsen, J. C. O.; Baeza, L. 2011. Dynamic
train–track interaction at high vehicle speeds: modelling
of wheelset dynamics and wheel rotation, Journal of Sound
  (
 Mbj ,bj  = diag msl , j ms1, j ms 2, j ms 3, j ; )
and Vibration 330(22): 5309–5321.
http://dx.doi.org/10.1016/j.jsv.2011.05.030
  (
 M Rij , Rij  = diag mR,i mR, j . )
Uzzal, R. U. A. 2012. Analysis of a Three-Dimensional Railway Damping coefficients of sub-ballast and a part of
Vehicle-Track System and Development of a Smart Wheel- the rail:
set: A Thesis in the Department of Mechanical and Indus-  Cbi ,bi Cbi ,bj Cbi , Rij 
trial Engineering Presented in Partial Fulfillment of the  
Requirements for the Degree of Doctor of Philosophy at Ci , j  =  Cbj ,bi Cbj ,bj Cbj , Rij  , (A.1.2)
 
Concordia University. Montreal, Quebec, Canada. 293  p.  
Available from Internet: http://spectrum.library.concordia. CRij ,bi CRij ,bj CRij , Rij 
ca/973753
Uzzal, R. U. A.; Stiharu, I.; Ahmed, W. 2009. Design and Anal- where: Cbi ,bi  , Cbj ,bi  , CRij ,bi  , Cbi ,bj  , Cbj ,bj  ,
ysis of MEMS based accelerometer for automatic detection CRij ,bj  , Cbi , Rij  , Cbj , Rij  , CRij , Rij  are the blocks
of railway wheel flat, World Academy of Science, Engineering        
and Technology, International Science Index 3(5): 785–793. of the damping matrix of sub-ballast and a part of the
Wang, K. Y.; Liu, P. F.; Zhai, W. M.; Huang, C.; Chen, Z. G.; rail:
Gao, J. M. 2015. Wheel/rail dynamic interaction due to ex- T
citation of rail corrugation in high-speed railway, Science Cbj ,bj  = Cbi ,bj  ;
   
China Technological Sciences 58(2): 226–235.
T
http://dx.doi.org/10.1007/s11431-014-5633-y CRij ,bi  = Cbi , Rij  ;
Wasiwitono,  U.; Zheng,  D.; Chiu, W. K. 2007. How useful is    
track acceleration for monitoring impact loads generated T
CRij ,bj  = Cbj , Rij  .
by wheel defects?, in Proceedings of the 5th Australasian    
Congress on Applied Mechanics (ACAM 2007), 10–12 De- Stiffness coefficients of sub-ballast and a part of the
cember 2007, Brisbane, Australia, 502–507. rail:
Wu, T. X.; Thompson, D. J. 2001. A Hybrid Model for Wheel/  K bi ,bi K bi ,bj K bi , Rij 
Track Dynamic Interaction and Noise Generation due to  
Wheel Flats. ISVR Technical Memorandum No. 859. 47 p.  K i , j  =  K bj ,bi K bj ,bj K bj , Rij  , (A.1.3)
 
Available from Internet: http://resource.isvr.soton.ac.uk/  
staff/pubs/PubPDFs/Pub1127.pdf  K Rij ,bi K Rij ,bj K Rij , Rij 
Yazykov, V. N. 2004. Primenenie modeli negercevskogo kontakta
where:  K bi ,bi  ,  K bj ,bi  ,  K Rij ,bi  ,  K bi ,bj  ,  K bj ,bj  ,
kolesa s rel’som dlya ocenki dinamicheskih kachestv gruzo-
vogo teplovoza: Dissertaciya na soiskanie uchenoj stepeni  K Rij ,bj  ,  K bi , Rij  ,  K bj , Rij  ,  K Rij , Rij  are the blocks
kandidata tehnicheskih nauk. Bryansk: Bryanskij gosudars-        
tvennyj tehnicheskij universitet, 151 s. (in Russian). of the stiffness matrix of sub-ballast and a part of the
Zhu, J. J.; Ahmed, A. K. W.; Rakheja, S. 2007. Adaptive contact rail:
model for simulation of wheel–rail impact load due to a T
 K bj ,bj  =  K bi ,bj  ;
wheel flat, in Proceedings of the 13th National Conference on    
Mechanisms and Machines (NaCoMM07), 12–13 December T
2007, Bangalore, India, 157–164.  K Rij ,bi  =  K bi , Rij  ;
   
T
 K Rij ,bj  =  K bj , Rij  .
   
Load force vectors:

{F } {F } {F }  ,

T T T
= {Fbi }
T
b,i , j bj Rij

where: {F } , {F } , {F } are vectors in the points i
bi bj Rij
and j of load forces.
232 M. Bogdevičius et al. The dynamic behaviour of a wheel flat of a railway vehicle and rail irregularities

APPENDIX 2
The mass of the railway vehicle:

( )
 M RB  = diag mbg 1 mbg 2 mbg 3 mbg 4 , (A.2.1)

where: mbg 1 , mbg 2 , mbg 3 , mbg 4 are the masses of

the vehicle.
Damping and stiffness matrices of the vehicle:

 cbg 12 −cbg 12 0 0 
 
 −cbg 12 cbg 12 + cbg 23 −cbg 23 0 
 
Cbg  =  0 −cbg 23 cbg 23 + cbg 34 −cbg 34 
;
 
 0 0 −cbg 34 cbg 34 
  
(A.2.2)

 kbg 12 − kbg 12 0 0 
 
−k k +k −kbg 23 0 
 K bg  =  bg 12 bg 12 bg 23 ,
   0 − kbg 23 kbg 23 + kbg 34 −kbg 34 
 
 0 0 − kbg 34 kbg 34 
  
(A.2.3)
where: kbg 12 , kbg 23, kbg 34 and cbg 12 , cbg 23, cbg 34 are
damping and stiffness coefficients of vehicle elements.
Weight force vector of the railway vehicle:

{F (t )} =  −m
bg bg 1 g −mbg 2 g −mbg 3 g −mbg 4 g  . (A.2.4)