2.

LONGITUDINAL DYNAMICS OF RAILWAY BOGIES: A

LITERATURE REVIEW

2.1. INTRODUCTION

Railway bogies are often subjected to longitudinal forces due to a number of train

related dynamics including braking and traction. These longitudinal forces affect the

dynamics of bogies in a complex manner. Severe braking or traction may affect the

safety and stability of bogies adversely. Thus, there is a need to study the bogie

dynamics as a function of the longitudinal forces with a view to minimising the risk to

railway transportation.

For the proper analysis of the dynamic performance of bogies under braking or traction,

basic understanding of the vertical and lateral dynamics of the wagon is required. The

first part of this chapter introduces some important terms that relate to the dynamics of

wagons. The mechanics of wheel-rail contact, which is fundamental to the bogie and

wagon dynamics, is also discussed in this chapter. The wagon braking and traction

systems and their principle of working are reviewed briefly for completeness. In the last

part of this chapter, a review of the current railway wagon simulation software systems

and their limitation to perform wagon longitudinal dynamics simulation is presented.

2.1.1. Basic Axis System and Terminologies

To discuss the dynamics of the wagon a coordinate system containing six degrees of

freedom as shown in Fig 2.1 is normally used. Linear motion along the X, Y and Z

axes are termed as longitudinal, lateral, and vertical translations respectively.

6
 Z

Y Pitch Yaw
X

Roll

Figure 2.1. Six degrees of freedom of wagon movement

The rotations are defined in accordance to the right hand screw rule where the positive

rotation is seen as clockwise if the observer is stationed at the origin and looks at the

axes in the positive direction. Rotary motions about the X, Y and Z axes are termed as

roll, pitch and yaw respectively (see Fig.2.1).

2.1.2. Wagon Assembly and Tracks Construction

Wagon components and assembly

Most wagons that are currently in use consist of a body and two bogies that provide the

necessary suspension. Each bogie basically consists of two wheelsets, a bogie frame

consisting of two side frames and one bolster, as well as spring nests. A few commonly

available bogies are exhibited in Fig.2.2 - 2.4.

Two common types of bogies widely used in freight trains are three-piece bogies and

Y25 bogies. The first is widely used in the United States, Australia, and Asia while the

second in Europe (Harder (2000), Bosso et al. (2000)). Fig.2.2 shows a typical three-

piece bogie and Fig.2.3 shows a typical Y25 bogie.

7
Figure 2.2. Typical design of three-piece bogie (Company Standard Car Truck (2000))

Transversal
beam or bolster

Side beam

End beam
Primary
suspension

Figure 2.3. Typical design of Y25 bogie (Website K. Industrier. AB (2005))

Three-piece bogies consist of a bolster, two side frames and two wheelsets. This type of

bogie does not have primary suspension while the secondary suspension is made from

coil springs that connect the bolster with the side frame. Damping is provided by

friction wedges, which are placed between the side frame and the bolster.

8
In contrast to the three-piece bogies, Y25 bogies do not have secondary suspension.

They have only primary suspensions that connect the wheelsets and the side frame. The

primary suspensions are formed from coil springs and Lenoir links which provide

friction damping. The bogie frame is constructed from two side beams, one transverse

beam or bolster and two end beams.

The three-piece bogies and the Y25 bogies are very popular as they are cheap to

purchase and maintain. However, their simple design leads to low levels of lateral

stability and ride quality, and higher levels of track forces due to vertical and lateral

impacts as well as the angle of attack in curved tracks (Stichel (1999)). Although the

performance of both bogies fulfils most of the requirements for freight wagon

operations, in some cases a higher performance bogie will be needed. In such cases, at

the expense of initial cost, bogies containing both primary and secondary suspensions

are used. An example of this type of bogie is shown in Fig 2.4.

Secondary Primary
suspension suspension

Figure 2.4. Bogie with primary and secondary suspension (Ikamoto (1998))

9
Track construction

Track is one of the most important technical elements required for the railway

operation. Its main function is to provide guidance for the wagons in addition to

supporting the heavy mass of the running train and absorbing the induced vibration.

Track possesses a complex structure with elastic and dissipative properties. Fig. 2.5

(Profilidis (2000)) exhibits the basic construction of a traditional railway track that

consists of a pair of rails, sleepers and track support.

Figure 2.5. A Typical Track Structure (Profilidis (2000))

The requirements for the strength and quality of the track depend to a large extent on

the following load parameters (Esveld (2001)):

− axle load: static vertical load per axle

− tonnage borne: sum of the axle loads

− dynamic and / or impact load

10
The static axle load level, to which the dynamic increment is added, in principle

determines the required strength of the track. The dynamic load components, which

depend on the operational speed and horizontal and vertical track geometry, are also

essential factors in the track structure design.

2.2. DYNAMICS OF THE BOGIES

Train-track dynamics is largely due to the interaction between the complex geometries

of the wheel and the rail within the contact patch that generates much high levels of

contact forces. It also involves many degrees of freedom and forces that change rapidly

and act at the same time at many points in the suspension system and couplers between

wagons. The existence of track irregularities and wheel defects makes the problem

more challenging to mathematically formulate. Hence an in depth understanding of the

subject of the dynamics of multibody systems is essential.

2.2.1 Principles of Wheelset Dynamics

The dynamic characteristics of a railway bogie are defined by the interaction between

the wheel and the rail, the configuration of suspensions and the articulation with

adjacent wagons. Among these, the interaction between the wheel and the rail that

characterises the dynamic behaviour of the railway wheelset running along the track is

the most fundamental factor that affects the bogie dynamics.

The wheelset provides basic guidance of travel to railway bogies. The dynamics of a

bogie is primarily affected by the dynamic behaviour of the wheelset and the track

characteristics. Fig 2.6 depicts a railway wheelset positioned on the track. It could be

seen that the conventional wheelset consisting of two conical wheels separated by a

11
distance compatible to the gauge width of the track fixed to a common axle. This form

has a long history and seems to have evolved by a process of trial and error (Wickens

(1998)).

Figure 2.6. Wheelset on the track (Ikamoto (1998))

The observation of the dynamic behaviour of wheelsets has begun since the early years

of the railway history. Marshal (1938) has reported that, not long after the conical

wheel tread was established in 1821, George Stephenson in his Observation on Edge

and Tram Railway had stated a very clear description of kinematic oscillation as shown

in Fig.2.7. This kinematic oscillation could cause stability problem in the tangent track.

The statement of George Stephenson can be found in Wickens (1998) and Wickens

(2003).

12
 Figure 2.7. Wheelset Kinematic Oscillation (Wickens (2003))

Klingel (1883) formulated the first mathematical relationship for this kinematic

oscillation and derived the relationship between the wavelength LH and the wheelset

conicity λw , nominal wheel radius rw and lateral distance between the wheel-rail

contact points 2l as shown in Eq.2.1.

rw l
L H = 2π ( )1 / 2 (2.1)
λw

This simple formula is derived purely from the geometry analysis.

Although the conical wheel causes stability problems in the tangent track, it helps the

wheelset to negotiate the curves with ease. A wheelset with conical wheels can

maintain a pure rolling motion while running on a curve if it moves outward and takes

the radial position. In his book, Wickens (2003) reported that Redtenbacher (1885) had

performed a theoretical analysis to improve the understanding of the conical wheelset

negotiating curves as illustrated in Fig.2.8.

13
 Figure 2.8. Conical wheelset on a curve (Wickens (2003))

From the geometry in Fig. 2.8 it can be seen that there is a simple relationship between

the lateral movement of the wheelset y , the radius of curve R , the wheel radius rw , the

lateral distance between the points of contact of the wheels with the rails 2l , and the

conicity λw of the wheels in order to sustain pure rolling that is shown in Eq.2.2.

OAB = OCD
(rw − λw y ) /( R − l ) = (rw + λw y ) /( R + l ) (2.2)
y = rwl / Rλw

As can be seen, the above two analyses of kinematic oscillation and curving are purely

based on geometries of the wheel and the track. More detailed analysis of the dynamic

behaviour of railway wheelset must be performed if the forces acting at the wheel-rail

contact patch are desired.

14
2.2.2. State-of-the-art of the Study of Wagon and Bogie Dynamics

History and development

The revolution in the analysis of railway wagon dynamics started when the theory of

creep (further discussed in Section 2.3) was first introduced by Carter (1926). With this

theory, the wheel-rail contact forces can be determined and the equations of motion that

describe the wheelset dynamics can also be derived.

Following the development of the creep theory that defines the interaction between the

wheel and the rail, extensive studies of wagon dynamics were undertaken. In general

the problem of wagon dynamics can be divided into two parts: the hunting or the

stability problem which deals with the tangent track, and the curving behaviour which

deals with curved track. Unfortunately there is a conflict between stability on the

straight track and curving behaviour on the curved track. For example a softer and more

flexible primary spring will give a better curving performance but it will reduce the

lateral stability in the tangent track. Similarly harder springs will improve the lateral

stability but lead to poor curving performance (defined by increase in angle of attack).

Gilchrist (1998) has presented a good paper that reviews the history and the

development of the research on the optimisation of the hunting and curving problems.

Matsudaira (1952) was the first to solve the complexity of the wheelset equation of

motion and concluded that through proper design, a spring-restrained wheelset could be

made stable up to any required critical speed as demonstrated by his theoretical work

and experiments on a roller rig, where the stability was observed directly. Using his

theory, he specified the suspension design parameters for the bogies of the first

Shinkansen high speed train that was successfully introduced into service in 1964.

15
Following Matsudaira’s work, from the mid fifties until the mid sixties most of the

research on railway wagon dynamics were focused on the subject of stability. For

example, de Pater (1956, 1961) investigated the non-linearity of the wheel-rail profile

and its effect to wagon stability, and Wickens (1965-6) proved the importance of lateral

damping of the primary suspension for the dynamic stability of the wheelset.

However, the more extensive study on wagon dynamics came after Kalker (1967), who

studied the rolling contact between two elastic bodies, provided a theory that could be

used to calculate forces generated in the wheel-rail rolling contact patch. The work of

Kalker and the development in numerical techniques have widely opened the

possibility of studying wagon dynamics through computer modelling to obtain better

results. Since the pioneering work of Kalker, a large amount of research on wagon

dynamics has been reported. It is impossible to present all of them in this review; a

brief summary, however, is provided. The research covers the study on stability,

curving performance, wagon-track dynamic interaction and control.

Hunting and stability

Rinehart (1978) has examined the hunting stability of three-axle locomotive bogies.

Due to symmetry, he has simplified the system using an eleven DOF model

representing a “half locomotive” (one bogie only). A set of laboratory test data of

natural frequency and mode shapes was used to validate the model. The results showed

that a hunting frequency of more than 4 Hz was obtained and the predicted hunting

frequency from the mathematical model had good agreement with the measurement.

Tuten et al. (1979) investigated how various wheel profiles and asymmetric loading

affected the stability of wagons. The investigation used a nine DOF wagon model. The

16
results showed that the wagon stability strongly depended on the location of axles

having different values of effective conicity and contact angles. The wheel profile mix

of a particular bogie was shown to be of much greater importance than whether the

bogie was put in the leading or trailing location. The asymmetric loading was also

found to affect the lateral stability.

Renger (1984) modelled a railway vehicle with two-axle bogies to examine its lateral

stability and ride quality. The vehicle response against the lateral centre line and the

cross-level deviations of the track were evaluated. A few important results provided by

the simulation are as follow:

(i) The stability and the riding quality could be improved by optimising the

secondary lateral damping.

(ii) Instead of using the primary damping, the stability and lateral ride quality could

be better improved through the optimisation of the primary stiffness.

De Pater (1989) studied the lateral stability of wagons containing two axle bogies. The

study was purely analytical using a mathematical model. He showed that an appropriate

choice of the lateral stiffness connecting the two wheelsets could increase the critical

speed.

Ahmadian (1998) investigated the non-linear oscillation of the wheelset with flange

contact. Bifurcation theory was applied to analyse the instability due to hunting. The

studied case was a rail wheelset containing nonlinear primary yaw dampers. The result

of the study showed that hunting could occur at speeds below the critical speed

17
computed through a linear analysis due to nonlinearities caused by flange contact,

gauge clearance and yaw dampers.

Yabuno et al. (2001) studied the stabilisation control for the hunting motion of the

wheelsets. They proposed a control strategy to limit the hunting motion of the

wheelsets. The control method focused on the asymmetry in the stiffness matrix that

was the principal cause of hunting. In order to reduce the effect of the asymmetrical

component of the matrix, a lateral force proportional to yaw motion was applied. An

experimental study was also conducted to validate the theoretical result. It was

concluded that the control strategy significantly increased the critical speed.

Mohan (2003) reported an investigation on the nonlinear analysis of the controllable

primary suspensions to improve hunting stability of wagons through the use of various

primary and secondary stiffness and damping parameters. It was concluded that the

critical velocity of wagons was more sensitive to the primary longitudinal stiffness

compared to other parameters. He also proposed a method to control hunting stability

using semi-active control of the primary longitudinal stiffness.

Curving and derailment

Sweet et al. (1984) studied the running safety of wagons against wheel-climb

derailment. The study employed theoretical modelling and experiments. The results

showed that the derailment quotient or the ratio of the lateral to the vertical force (L/V

ratio) at the wheel-rail interface alone was not sufficient to predict the safety against

derailment. Under dynamic conditions, a significantly larger derailment quotient could

occur without causing actual derailment if it only occurred for a very short time. Hence

18
the criterion for derailment was accounted for by both the derailment quotient and the

corresponding duration.

Effects of wheel-rail contact geometry to the wheelset steering forces was studied by

Mace et al. (1996). The study involved both field experiments and theoretical analysis.

It was reported that the hollow worn wheel adversely affected the wheelset steering

during negotiating curves, leading to the generation of large negative steering moments.

These negative steering moments caused a number of undesirable effects such as the

track gauge widening, the rail roll over, the extensive wear of the wheel and the rail,

and the increased train rolling resistance.

Haque et al. (1996) reported a non-linear wheelset model for derailment prediction. The

modelling provided special emphasis on safety-related behaviour of the wheelset

negotiating both the tangent and curved tracks. The wheelset models accounted for

non-linearities due to wheel-rail profile geometry and creep force and the longitudinal

translation of the contact patch as a function of wheelset yaw angle. The wheelset

model was claimed to have lateral, yaw, and spin degrees of freedom and considered

single-point and two-point contact as well as transition from one to the other. The

authors exhibited that the model was capable of predicting the wheelset dynamic

behaviour during wheel climb, wheel lift, steering characteristics during curve

negotiation and also limit cycle behaviour on the tangent track.

Nagase et al. (2002) reported experimental results of the wheel climb derailment. The

experiment was performed using a model bogie and a model track (1:5 scale). The risk

of the derailment was evaluated using Nadal formula (ratio of the lateral force to the

vertical force applied to the wheel) as well as by measuring the wheel vertical

19
displacement using a high-precision laser displacement sensor. As a result, it was

found that the adhesion coefficient had a major influence on the occurrence of wheel

climb derailment.

Optimisation of hunting and curving behaviour

Wickens (1991) provided a detailed review on optimisation of the hunting and curving

behaviour through suspension design using sophisticated mathematical modelling.

Current suspension technology has opened possibilities to develop innovative bogie

designs with optimal performance in both tangent and curved tracks.

Matsumoto et al. (1999) proposed some methods to optimise the curving behaviour

without reducing the hunting speed of the wagon. These methods included optimised

worn tread profiles of wheels, independently rotating wheels in the rear axle, and

asymmetric arrangement of the longitudinal primary suspension. The third method had

earlier been proposed by Suda and Anderson (1994).

The improvement of the compatibility of the lateral stability and the curving

performance of a railway passenger bogie was recently reported by Dukkipati and

Narayanaswamy (2004). The authors proposed that the stiffness of the primary

suspension of the leading axle be set different to the trailing axle.

Wagon and track interaction

All of the above reported research works have been focusing on the study of wagon

and/or bogie dynamics without providing much attention to the track structure by

assuming the track as a rigid or a simple elastic support. There are several studies, on

the other hand, that focus only on the railway track dynamics by simplification of input

20
disturbance from the measurement or pre-calculated wagon dynamic characteristics

(Grassie (1992), Luo et al. (1996), and Kerr (2000)). Some other studies on track

dynamics have considered the track and the wagon as a multi-body system but

simplified the case to only the vertical interaction between the wagon and the track

(Zhai and Sun (1993)and Ripke and Knothe (1995)).

However, sometimes it is necessary to fully describe both the wagon and the track

structures in 3D to examine their interaction with each other in all directions. This is

because the track structure can affect the wheel rail interaction forces that play an

important role in determining the dynamic behaviour of wagons. On the other hand,

wagon suspension design can also affect the forces imposed on the track.

Sun and Dhanasekar (2001) introduced a three dimensional wagon track system

dynamics (3D-WTSD) model, which fully describes the dynamic behaviour of the

wagon and the track when the wagon runs under constant speed on tangent tracks. The

3D-WTSD model can be used, for example, to investigate the effect of track design

parameters both on the wagon and the track dynamics or vice versa to investigate the

effect of suspension design parameters on both the wagon and the track dynamics. It

also can be an effective tool to investigate the effect of track on wagon dynamics and

the effect of lateral and vertical impact applied by the wagon to track dynamics (Sun

(2002)). From the results of simulations using 3D-WTSD model, Sun and Dhanasekar

(2004) also reported the importance of track modelling for the determination of the

critical speed of wagons. A similar example of the simulation that takes into account

the 3D dynamic behaviour of the wagon and the track was presented by Anderson and

Abrahamsson (2002).

21
In the case of wagon dynamics under traction/braking condition, the vertical and the

lateral impact forces applied to the track may not be an issue. However, when the

traction/braking force is applied to the wheels a large longitudinal creep force arises in

the wheel-rail contact area. This force will dissipate through the structure below the rail

such as the sleepers and the ballast. If this longitudinal force is big enough, the sleepers

can be displaced from their position. If a train is braked or accelerated on the

construction such as the railway bridge, the longitudinal force generated will be also

passed through the construction. Therefore, in designing the track structure and the

railway bridge this longitudinal force is required to be accounted for.

All of the above described simulation models deal with only constant speed without

any reference to deceleration or acceleration. The dynamics of wagons under

braking/traction is reviewed in Section 2.5.

2.3. THEORY OF WHEEL-RAIL ROLLING CONTACT

This section briefly reviews the development of wheel-rail rolling contact theory, which

forms the foundation for determining the wheel-rail interaction forces. The accurate

calculation of the wheel-rail interaction forces is very important in the modelling of the

railway wagon and bogie dynamics. Kalker (1991) has provided a very good

presentation on this subject. Another good reference is a book by Garg and Dukkipati

(1984). Details of the mathematical analysis on rolling contact phenomena can be

found in Jacobson and Kalker (2000) and Kalker (1990).

22
2.3.1. The Concept of Creep

Consider two rigid bodies that are in contact at a point. If any one or both of these

bodies are rotated and/or moved relative to each other, the contact point will also shift

its original position; the resulting velocities of the contact point over each body might

or might not be equal to each other. When the velocities are equal, the bodies are said

to be undergoing pure rolling (with no creep); under unequal velocities, they are said to

be undergoing rolling coupled with sliding (i.e., with creep). Creep or creepage is a

dimensionless term (except for spin creepage) defining the deviation of the actual

rolling condition between two rigid bodies from pure rolling to rolling coupled with

sliding.

In the case of the wheelset running over the rails, creepage is defined in both the

longitudinal and the lateral directions and also about the common normal of the contact

patch (spin) as shown in Fig.2.9 (Dukkipati (2000)). The formulation is provided in

Garg and Dukkipati (1984) and Dukkipati (2000) as in Eq.(2.3.a-c).

(a) Longitudinal creepage (b) Lateral creepage (c) spin creepage

Figure 2.9. Creepage (Dukkipati (2000))

23
 (longitudinal velocity of wheel – longitudinal velocity of rail) at the point of contact
ξx = (2.3.a)
Nominal Velocity

(lateral velocity of wheel – lateral velocity of rail) at the point of contact

ξy = (2.3.b)
Nominal Velocity

(angular velocity of wheel – angular velocity of rail) about normal axis at the point of contact
ξ sp = (2.3.c)
Nominal Velocity

where ξ x , ξ y , ξ sp are the longitudinal, the lateral and the spin creepages respectively.

It is important to note that the longitudinal and the lateral creepages are dimensionless,

whereas the spin creepage has the dimension of L-1.

2.3.2. The Development of Wheel-Rail Rolling Contact Theory

The wheel-rail rolling contact theory explains the relationship between the creep forces

and the creepage. In general wheel-rail rolling contact theory states that there exists a

unique relationship between the creepage and the forces generated at the wheel-rail

contact patch. These forces are called creep forces as they are generated due to the

existence of the creepage.

The relationship between creepage and creep forces was first defined by Carter (1926)

who was concerned with the action of locomotive wheels when large tangential forces

were transmitted during acceleration and braking. Carter has shown that the difference

between the circumferential velocity of a driven wheel and the translational velocity of

the wheel over the rail has a non-zero value as soon as a braking or traction couple is

24
applied to the wheel. The difference increases if the braking or traction couple

increases, which means that there exists a relationship between the couple and the

velocity leading to saturation when the Coulomb friction maximal value is reached.

However, the formulation given by Carter was based on solving the integral equation of

the two dimensional analysis for a cylinder rolling on a plane which only considers the

force on the rolling direction. It is clearly insufficient for the purpose of rail wagon/

bogie simulation due to the complex geometries of the wheel and the railhead.

Vermeulen and Johnson (1964) proposed a creep-force law, which included the

longitudinal and the lateral creepages. However, the spin creep was left out. To

calculate the shape and the size of wheel-rail contact, Hertz theory is used. The Hertz

theory defines the contact area between the wheel and the rail as elliptical and the ratio

of the semi axes of the ellipse as a function of the curvature of the wheel and the

railhead. The treatment of the Hertz theory in detail can be seen in Johnson (1985).

The most successful method of calculating the creep force is presented by Kalker

(1967) who then wrote the computer program CONTACT, a universal program for all

contact problems of bodies that can be described by half-space. He also has written a

program called DUVOROL, which efficiently handles all possible rolling contact

problems of bodies with identical elastic constants that touch each other according to

the Hertz theory. DUVOROL was used by British Rail to construct a book of tables in

support of rail vehicle simulation.

The computational time of both CONTACT and DUVOROL, which are based on

Kalker’s exact theory, is high and hence they are not suitable for real time applications

in vehicle simulation (this is the reason why British Rail constructed a book of tables).

25
Concerned with this, in 1973 Kalker introduced the simplified theory of rolling contact

and then used the theory to build a fast algorithm and computer program FASTSIM

(Kalker (1982)).

Shen et al. (1983) improved the Vermeulen and Johnson law by including the effect of

spin using Kalker’s exact linear theory. The model used by Shen et al. is also called the

heuristic model and is well known as Shen-Hedrick-Elkins or SHE theory. Fig. 2.10

presents the comparison of the creepage – creep force curve produced using FASTSIM,

DUVOROL, and the heuristic model of Shen-Hedrick-Elkins for small spin. The figure

shows that the three methods agree very closely.

FR
µr N

normalised creepage

Figure 2.10. Creepage - Creep Force Curves (Shen et al. (1983))

In summary, Kalker (1991) defined the rolling contact theories and their interrelation as

shown in Fig 2.11. He also gave a suggestion that the contact mechanics aspect of the

wheel and the rail can be treated with the following routines:

26
Driving/braking wheel action Shen-Hendrick-Elkins (SHE)

Non flanging vehicle dynamics BR table book, linear theory, SHE, FASTSIM

Flanging vehicle dynamics BR table book, FASTSIM

Two-point contact BR table book, FASTSIM

Wear (profiles) FASTSIM, CONTACT

Hertz CONTACT
B.R. table book
Non-Hertzian
Unrestricted creepage
Carter
2D validation

Johnson/Vermeulen Linear theory

ξ sp = 0 Small Creepage

Unrestricted ξx ,ξ y

Shen-Hedrick-Elkins Simplified theory validation

Small ξ sp Unrestricted ξ x , ξ y , ξ sp
Unrestricted ξx ,ξ y FASTSIM

validation

Figure 2.11. Wheel-rail rolling contact theories and their interrelation (Kalker (1991))

To complete our discussion on the development of the wheel-rail rolling contact theory,

we should review another method to determine the creep forces which has been

recently proposed by Polach (1999). This method is claimed to perform better under

high creepage, although it is based on Kalker’s work with simplification of the

distribution of normal and tangential stresses in the wheel-rail contact patch. According

27
to this theory the creep forces can be computed efficiently with significant saving in

computational effort. Application of this method to wagon dynamics simulation was

also reported by the author (Polach (2001, 2005)). Polach also extended his creep force

model for large creep application by introducing reduction factors for the Kalker

coefficient to differentiate the areas of adhesion and slip.

2.4. FREIGHT WAGON BRAKING AND TRACTION SYSTEM

Traction and braking may be regarded as a process of conversion of energy. If a rail

wagon is at rest the kinetic energy remains zero, whilst a moving rail wagon possesses

significant kinetic energy. Braking reduces the speed of the wagon which means

reducing the wagon kinetic energy, whilst traction does the opposite. However, wagons

usually do not have their own traction system but are pulled or pushed by locomotives.

Reducing the speed of the wagon requires significant reduction to kinetic energy. The

simplest way of reducing the energy is to convert it into heat by contacting material to

the rotating wheels or to discs attached to the axles. The material creates friction and

converts the kinetic energy into heat energy. With the reduction in kinetic energy, the

wagon slows down and when the kinetic energy is fully nullified, the wagon comes to

static equilibrium. The vast majority of freight trains are equipped with braking systems

that use compressed air as the force to push blocks on to wheels or pads on to discs.

These systems are known as "air brakes" or "pneumatic brakes". There are several

types of air brake systems that are currently used in trains, which differ in aspects of

their control systems, main control equipment, auxiliary equipments and pressure level.

This is due to the different standard operating requirements of various railway

networks.

28
Fig 2.12 shows a typical layout of air brake components of a freight wagon (Bureau

(2002)) in North America. A similar system is used in the freight wagons of Australia.

In this system the compressed air is transmitted along the train through pipes. A control

valve, that is the AB type control valve in the figure, controls the pressure level of

compressed air used to produce the braking force.

Figure 2.12. Freight wagon air brake system (Bureau (2002))

The braking force produced by the brake cylinder, usually mounted at the wagon

underframe, is transmitted through a set of levers and rods to the wagon bogie. The

force is then distributed to the wheels through the bogie brake rigging, which consist of

levers and brake beams, fitted in each bogie. A slack adjuster is fitted within the

mechanical link arrangement. The slack adjuster both takes up and lets out slack in the

rods and lever system in order to keep the clearance between the brake shoes and

wheels to a specified level.

29
Fig 2.13 exhibits the typical bogie brake rigging diagram of the three piece bogies that

are equipped with one-side push brake shoe arrangement (Handoko et al. (2004)). The

link consists of rods and levers suspended from the underframe and bogies, and linked

with pins and bushes. The brake rigging requires careful setting up and regular

adjustment to ensure the forces are evenly distributed to all wheels.

F
fixed end pin
B in bolster

guiding slot
in side frame

Figure 2.13. Typical bogie brake rigging

It can be seen from the rigging diagram that any bad adjustment of the brake rigging

could lead to uneven distribution of braking forces to each wheel. Such a situation can

occur when either the centre-pin on rod AB is slightly off-centred or if the fixed-end

pin in the bolster is disorientated. The uneven distribution of the braking force to

wheels may also occur during curve negotiation if the bogie deforms in shear (warping)

mode.

2.4.1. Calculation of Brake Shoe Force

The brake shoe force applied to wheels is calculated from the brake cylinder piston

thrust, the total brake rigging ratio (effectively the multiplication factor by which the

brake piston force is leveraged by the brake rigging geometry), and the counter-forces

30
exerted by the brake storage spring of the slack adjuster as shown mathematically in

Eq. (2.4).

FB = FCT .it − ( FR .ib ) (2.4)

where FB is the brake shoe force, FCT is brake cylinder piston thrust, it is total brake

rigging ratio, FR is counter force exerted by slack adjuster, and ib is the bogie brake

rigging ratio. Brake cylinder piston thrust is determined by the cylinder piston area and

the pressure in the brake cylinder as shown in Eq.2.5.

π
FCT = D 2 . pc (2.5)
4

where D is effective diameter of the piston and p c is the air pressure in the brake

cylinder. The effective brake shoe force that would be actually acting on the wheel is

normally less than the calculated brake shoe force above. This is because of power

losses in the rigging system due to friction in the pin joints of the brake rigging levers.

If we introduce the brake rigging efficiency η to represent these power losses, then the

effective brake shoe force can be written as Eq.2.6.

FBeff = FB .η (2.6)

2.4.2. Brake Application and Release Timing

Brake application time is the time required to build the pressure in the brake cylinder

and is opposite to the brake release time, which is the time required to empty the

cylinder. Ideally, the brake is applied simultaneously at the same time on every wagon

of a train. This condition is easy to achieve if the electric brake control system is used.

31
However it will be difficult for the pure pneumatic brake system where the brake

command is conveyed through the brake pipe along the train. This is because the

response time is limited by the wave propagation in the compressed air system.

For modern freight traffic, it is required to shorten the application and release timing

with greater braking force for efficiency of operation. However for the pure pneumatic

brake system, problems will arise if the application time is too short because the front

end of the train may reach the maximum brake pressure while the rear end would not

have reached the full brake pressure yet. Derailments have occurred in which the rear

portion of the train "ran into" the front portion in such circumstances. So there should

be an optimum brake application time for a given train.

A similar situation (in reverse) happens with the brake release time. The brake will be

released if the brake pipe reaches a certain pressure and all the air in the brake cylinder

will be exhausted to the atmosphere. Because of the inertia of the air in the pipe system,

the brake pipe pressure in the wagon near to the locomotive, from where the

compressed air is supplied, may have reached the required pressure while the rear end

would not have yet.

Due to the reasons explained above and also to assure the compatibility among brake

systems in use, the UIC, BS, AAR, and other main railway standards have limited the

brake application and release times to certain values. This limitation is prescribed for

safety reasoning and decided based on experience. For example, the UIC standard

prescribes the brake application and release timing for freight cars fitted with single

pipe gradual release of brake as follows:

32
Application time : 0 to 95 % max brake cylinder pressure is 18 sec to 30 sec

Release time : 0 to 95 % max brake cylinder pressure is 45sec to 60 sec

From the application time and release time we can clearly see that, during the braking

process, the force applied to the wheel will not be constant; rather it will build up

slowly from zero to maximum. This also means that the wagon deceleration will

change with time.

2.4.3. Traction

Traction, in general, can be viewed as the reverse process of braking. From this point of

view, all parameters that influence the dynamics of wagons during braking can also

affect the dynamics of wagons during traction. However, most of the freight wagons

are not self-propelled. To accelerate they get the pulling or pushing force from a

locomotive. The traction forces in a locomotive are usually generated by diesel engine

or electric traction motors that produce torque transmitted to the wheelset using a gear

box, whilst the wagons receive the traction force through pulling or pushing action of

the mechanical couplers.

2.5. LONGITUDINAL DYNAMICS OF BOGIES AND WAGONS

2.5.1. Basic Principle of Braking Dynamics

During braking, forces or torques are applied to the wheelsets in order to decelerate the

wagon. This process also produces reaction force in couplers and pitch torque to the

wagon body and bogie. These forces affect the running stability and curving

performance of bogies and wagons. On the other hand the dynamic response of the

33
wagon such as the change of the load distribution to the wheel can also affect the

braking performance. The braking performance is usually measured through the

stopping distance and also from the occurrence of skidding or wheelslide. With the

reduction of the wheel load, the chance of the skid occurrence increases. Braking also

involves friction, a complex phenomenon. During the process of braking, friction

occurs between the brake shoe and the wheel and between the wheel and the rail as

shown in Fig. 2.14.

µ b FB

+ FB
Nw

µr Nw

Figure 2.14. Forces at the braked wheel

The brake shoe force applied to the wheel FB produces tangential force µ b FB . If the

wheel has radius rw , a braking torque TB is generated as shown in Eq.2.7.

TB = µb FB rw (2.7)

The free body diagram of the braked wheel shown in Fig. 2.15 represents a wheel

moving longitudinally in the x -direction at speed V and with angular velocity ω .

J y , rw and TB denote the polar moment of inertia, wheel radius and brake torque

respectively. At the contact point between the wheel and the rail, longitudinal and

vertical forces Fx and FZ respectively arise as the reaction to the brake torque and the

static weight mg .

34
 J yω&
=
TB rw
+ + mV&
mg
Fx
Fz

Figure 2.15. Free body diagram of braked wheel

By balancing the forces in the x - and z - directions and moments about the centre of

mass of the wheel, three scalar equations of a braked wheel are established:

mV& = − Fx ⎫
⎪
Fz = mg ⎬ (2.8)
J yω& = Fx rw − TB ⎪⎭

where m is the mass of the wheel and wagon supported by the wheel and g is the

gravity constant, while over dots denote differentiation with respect to time.

2.5.2. Skid and Friction Coefficient

Severe braking force can cause the wheelset to get locked and slide on the rail. This

phenomenon is called skidding. This could lead to geometry damage to the wheels

(wheel flat) and the railhead. Skidding also makes the stopping distance longer that

would be dangerous to the train operation. Hence skidding should be avoided during

braking as a matter of priority.

Skidding occurs when the braking force exceeds the adherence offered by the wheel-

rail contact patch. Thus, to avoid skidding of the wheels during brake application the

35
brake force at the brake shoe must invariably be kept lower than the adhesion at the

rail. Skidding does not occur when the relationship in Eq. (2.9) is fulfilled.

µ b FB < µ r N W (2.9)

Currently many rail wagons are equipped with equipment to prevent skidding.

However, there are still many wagons that do not possess this equipment, especially

freight wagons, due to costs.

The difficulty in controlling skidding is related to the friction characteristic between

materials that vary in nature. For example, friction between the wheel and the rail

varies with rail surface condition (Macfarlane (2000)) such as rail corrugations, rail

head contamination from oily deposits, leaves, water, ice, sand, etc, and the wheel tread

surface condition as the application of the different type of brake shoe on the car (iron

block, composition block, disc). Friction between the wheel and the rail also varies with

the position of the wheelset along the track. The leading wheelset usually encounters

the dirtiest rail and worst adhesion condition, and then cleans it for the wheels that

follow. For design purpose, the friction coefficient between the wheel and the rail is

usually assumed to be between 0.10 - 0.30.

In Section 2.3, the relation between creepage and the generated creep force has been

explained. However, many experiments reported in the literature show that the relation

has a peak followed by decay with increase in creepage. To explain this phenomenon,

Nielsen and Theiler (1996) proposed the modelling of the friction coefficient as a

function of slip velocity. Following the theory of Nielsen and Theiler, the relation

between the slip percentage and the friction between wheel and rail was proposed by

Ohishi et al. (2000), as exhibited in the adhesion force versus slip velocity curve shown

36
in Fig.2.16. Because the adhesion forces depend only on the friction coefficient and

normal load, it is clear that the friction coefficient is also affected by the slip. The

relation between them is not linear and depends on the wheel/rail condition (wet or

dry).

Adhesion adhesion slip adhesion slip

area area area area
force
DRY
WET

Maximum
adhesion force

Slip velocity

Figure 2.16. Adhesion force against slip velocity (Ohishi et al. (2000))

Coefficient of friction between the brake shoes and the wheel also does not remain

constant. It varies with the sliding speed between the brake shoes and the wheal tread

that depends on the velocity of the wagons. This also means that it changes

continuously during the braking process. Fig 2.17 exhibits a typical plot of kinetic

friction coefficient between two surfaces as a function of sliding speed and the

Barwel’s formula that shows the relationship between the friction coefficient µk and

sliding speed vs where the constant c f depends on the material (Rabinowicz (1995)).

37
 µk

µk = c f vs −1/10

Sliding speed, vs

Figure 2.17. Friction coefficient against sliding speed

2.5.3. Dynamics Due to Traction

Because traction can be viewed as the reverse process of braking, the three scalar

equations in the Eq. (2.8) are still applicable with only a change of sign (+/-). As a

consequence, in the reverse of the skid phenomenon during braking, locomotives

exerting excessive traction torque can make the wheelset rotate without any

longitudinal motion. This condition is referred to as ‘roll-slip’ and becomes a subject of

interest in the locomotive drive simulation as reported by Muller and Kogel (2000).

Roll-slip can lead to railhead damage (engine burn or wheel burn). As for braking,

traction can also cause large longitudinal forces in the wheel-rail contact patch.

2.5.4. State-of-the-art of Braking and Traction Dynamics Research

Balas (2001) developed a model for the sliding wheel of a railway car during braking.

This work is mainly to assist the study and design of the braking equipment, including

the Anti-lock Braking System (ABS). In this model the friction coefficient between the

wheel and the rail is considered as a function of the slip of the wheel as shown in Fig

2.18. The slip of the wheel is defined as shown in Eq.2.10.

s = (vcar − v w ) / vcar (2.10)

38
where: s is the slip (always between 0 (no braking) and 1 (locking)), vcar is the

velocity of the car and v w is the velocity of the wheel. This concept has been used by

Ohishi et al. (2000) to design a control system that would prevent slip during traction of

an electric motor coach. The difference is only in the shape of the friction coefficient

against slip curve where Balas considers the pseudo sliding due to elasticity of the

wheel. If this pseudo sliding is ignored, then the curve in Fig.2.16 used by Ohishi will

be obtained.

uncontrollable
sliding

Figure 2.18. Wheel-rail friction coefficient against slip (Balas (2001))

Independently Cocci et al. (2001) presented a railway wagon model with an anti-slip

braking system. The model is set up in the ADAMS/Rail and Simulink platform.

Similar to Balas, Cocci et al. also considered the friction coefficient between wheel and

rail as a function of wheel slip. The bogie suspension is modelled in three dimensions

where longitudinal, vertical, and lateral stiffness are considered in the model. However

longitudinal dynamics due to the effect of anti-slip control for optimum deceleration

was particularly attended. Modelling the complete bogie is required to take into account

the effect of load distribution to the wheel due to track geometry and/or track

irregularities.

39
Olson (2001) has studied the longitudinal dynamics of ground vehicles that include

non-linear wheel braking and acceleration models. Although his work is focused on

road vehicles, it is still appropriate to the study of railway wagon braking. In

formulating the equations of motion of wheel under braking condition, Olson

considered the slip as a dynamic state variable, replacing the absolute rotational rate of

the wheel speed.

Lixin and Haitao (2001) studied the dynamic response of wagons in a heavy haul train

during braking mode. For such purpose they have set up a model using ADAMS/Rail

software that could predict the three dimensional dynamic response of the wagon under

braking conditions. An open-top freight wagon used in China was the case study. From

their investigation it was concluded that the application of braking has adversely

affected the lateral and vertical dynamic performance of the wagon. However, the

investigation was limited to constant brake shoe force that was distributed evenly to the

wheels. In actual conditions, the brake force is a time function and may not be

symmetric and could be distributed unevenly to the wheels. Lixin and Haitao (2001)

did not investigate the effect of wheel-lock or skid phenomena. Both of these

conditions (asymmetric brake forces and wheel skid) can lead to a more serious

situation. This thesis (Chapter 9) describes these phenomena in detail.

Berghuvud (2002) investigated the effect of brake application to wagon curving

performance using parameters that define the wagon curving performance such as the

wheelset angle of attack, the track forces, and the wear in the contact patch between the

wheels and the rails. He found that the wagons with different types of bogies respond in

different ways to curving as a function of the applied braking force. However he did not

consider the effect of variation in speed or deceleration as he examined the wagon

40
running on downhill slopes while braking was continuously applied to keep its speed

constant. The braking force applied also remained constant and symmetric with no

wheel skid.

Suda and Grencik (1996) explained the mechanism of deterioration of curving

performance under braking conditions where the braking torque reduced the steering

torque of the wheelset. This mechanism is explained in Fig.2.19 that shows a free body

diagram of a wheelset running along a curved track.

FyL
FxL

FBL

FxR
FyR
FBR

Figure 2.19. Curving diagram of a wheelset (Suda and Grencik (1996))

While negotiating a curve, the rolling radius of the outer wheel becomes larger than

that of the inner wheel leading to the generation of longitudinal creep forces FxR and

FxL , where in general FxR ≠ FxL . These creep forces produce a steering torque that

guides the wheelset to follow the curve appropriately. Application of brake produces

additional longitudinal forces FBR and FBL on the contact points which produce a

steering moment as shown in Eq.(2.11).

M = a [( FxL − FBL )+( FxR + FBR )] , (2.11)

41
where a is the semi distance between the contact points. However, the resultant of total

longitudinal creep force and lateral creep force at the contact points cannot exceed the

maximum frictional force between the wheel and the rail. The resultant creep forces

are calculated as shown in Eq.(2.12).

FCR = ( FxR + FBR )2 + FyR 2 ; FCL = ( FxL − FBL )2 + FyL 2 (2.12)

where FCR and FCL are resultant creep forces on the right and left rail respectively. If

FCR and FCL exceed the creep condition, they becomes saturated leading to reduction

in the longitudinal creep force. The directions of longitudinal creep force and additional

force due to braking are the same on the inner wheel, so the reduction in longitudinal

force on the inner rail is larger than that on the outer wheel. Because of this reduction

of longitudinal forces, the steering moment on the wheelset also reduces.

Malvezzi et al. (2003) carried out an investigation of the braking in trains. They

performed probability analysis of train deceleration during braking. The aim of the

work was to determine the probability that the real deceleration is lower than the

nominal value multiplied by a safety margin.

The analysis of braking in a train was also carried out by Durali and Shadmehri (2003).

The authors reported the analysis of train derailment due to severe braking with various

wagon weight configurations. From the results, the optimum configuration of wagons

and the critical derailment velocity can be determined. The authors also claimed that

the results were in excellent agreement with the field experience although they did not

present any comparison with the field data in the paper.

42
Excessive traction torque applied to the locomotive wheels could cause roll-slip.

Because this roll-slip reduces the traction power and can damage the locomotive’s

wheels, it is very important to avoid it during the operation of the motive power. For

this reason, studies on the subject of slip controllers have been extensively performed.

Among them was the work recently reported by Frylmark and Johnsson (2003). In their

thesis Frylmark and Johnsson studied several methods of slip controller such as

adhesion observer based controller, fuzzy logic slip controller and hybrid slip control

method. As a summary, the authors presented the advantages and disadvantages of each

method. They also proposed a few improvements and present ideas that may be

interesting in future research.

2.6. WAGON SIMULATION SOFTWARE PACKAGES

The computational simulation of the dynamic behaviour of railway wagons has been a

standard design task in the railway industry during recent years (Schupp (2003)).

Software packages such as VAMPIRE and NUCARS (Iwnicky (1999)) have been

specifically developed for this purpose, while general multibody dynamics software

tools such as and ADAMS, SIMPACK and UNIVERSAL MECHANISM have a

module which is intended to simulate the railway wagon dynamics. Multibody

dynamics computational method has been used as a tool to develop these software

packages.

Using VAMPIRE as a tool, McClanachan et al. (2004) have shown that it is possible to

adequately model freight wagons containing three piece bogies during constant speed

operation. In their paper, the data from field tests were compared with the simulation of

roll, bounce and pitch.

43
Shabana and Sany (2001) reported a survey of rail vehicle track simulations which

include flexible multibody dynamics. In their paper they have pointed out that, with the

recent development in computational mechanics, it is possible to develop a tool to

comprehensively analyse the complex dynamics of railway vehicles and tracks.

Shen and Pratt (2001) developed a railway dynamics modelling and simulation package

for current industrial trends. They suggested that the future software tool should adopt

the object oriented and knowledge based techniques that aid the human thought

process. According to them the new simulation packages should be adaptable to design

changes.

2.6.1. Reference Coordinate System and Formulation of Equation of Motion

This section discusses how the multibody system approach is currently used in the

formulation of railway vehicle dynamics simulation tools. The review includes the

equation of motion and the formulation of the wheel-rail contact problem. These are the

specific modelling features that make the railway vehicle dynamic simulation unique.

However the review presented here also shows that the current approach is not ideal to

simulate railway wagon/ bogie dynamics during braking or traction.

The common method of deriving the equations of motion of the railway wagon is the

transfer coordinate (or moving coordinate) system that moves along the track at the

speed of the wagon. This coordinate system is referred to as track-based or track-

following moving reference system (Zboinski (1999)). With this method it is very

convenient to describe the position of the contact point between the wheel and the rail

and the direction of wheel-rail contact force. By assuming steady state motion of

wagons, a simple form of the equations of motion can be obtained. However, applying

44
this method to the braking and/ or traction condition is rather cumbersome if not

impossible.

Fig 2.20 illustrated this problem where a body i , with body coordinate system

( oi , xi , y i , z i ) attached to it, is moving with respect to a non-inertial transfer coordinate

reference system R ( or , x r , y r , z r ). As presented by Schiehlen (1984) and Xia (2002),

the Newton-Euler equations of motion for the moving body i with respect to the

transfer coordinate system and the inertial reference frame O ( oo , xo , y o , z o ) are as per

detail shown in Eq.(2.13) and Eq.(2.14).

xr
x0 rR or xi
rRi oi
o0 yr yi
y0 ω Ri
zr ωR zi
z0

Figure 2.20. The description of the motion of a body in a moving reference frame

m i ⎡⎣&&
rR + ω (
&% + ω
% R r&R + ω R )
% 2R rRi + 2ω rRi ⎤⎦ = Fi
% R r&Ri + && (2.13)

& R +ω
Iiω % R Iiω R + ω
% R ω Ri (I xi + I yi + I zi ) + 2ω
% Ri I i ω R + I i ω
& Ri + ω
% Ri I i ω Ri = M i (2.14)

where mi and I i are the mass matrix and the inertia tensor of the body respectively, Fi

and M i are the external force and moment applied to the body written in the transfer

coordinate frame respectively, rR and ω R are the vector position and the angular

velocity of the transfer coordinate system respectively, and rRi and ω Ri are the vector

position and angular velocity of the body reference in respect to the transfer coordinate

45
system respectively. Over dots ( ⋅ ) represent differentiation with respect to time and

tildes (~) represent skew symmetric matrix of the vector.

If the transfer coordinate system moves with constant speed along a tangent track, then

the Newton-Euler equations of motion reduce to a very simple form as shown in

Eq.2.15 and Eq.2.16.

rRi = Fi
mi&& (2.15)

& Ri + ω
Iiω % Ri I i ω Ri = M i (2.16)

The equations will be slightly more complex if the system moves in curving, although

it can be handled without much difficulty as long as the speed remains constant because

we could pre-define the angular velocity of the transfer coordinate system for the

known geometry of the curve.

The problem becomes very complex when we deal with the braking condition where

the speed does not remain constant. Currently, the solution to this problem is obtained

by pre-guessing the speed profile based on the initial velocity and deceleration and

assuming the transfer coordinate system to move with this speed profile. This approach

will not provide an exact solution because the deceleration actually comes from the

force applied to the body in the system, or more specifically it is caused by the force

generated at the contact points due to brake forces applied to the wheels. To obtain an

exact solution, the contact kinematics should be calculated as a function of the variation

of the velocity of the moving body and it must be calculated in real time (on line)

during the simulation.

46
Another consequence of using a moving reference frame is that the wheelset rotation

with respect to its lateral axis, which is referred to as wheelset pitch, is not explicitly

included in the formulation. Because the speed of the wagon has been pre-defined and

assumed to be the speed of the transfer coordinate system along the track, the nominal

wheelset pitch velocity has also to be pre-defined as a function of the speed. Thus, the

effect of the severe application of heavy braking and traction such as skid and slip

cannot be accounted for in the formulation.

Due to the reasons explained above, the best way to treat the dynamics during braking

and traction is to describe the absolute body motion in the fixed inertia coordinate

system. This thesis presents such a formulation which is discussed in Chapter 3.

Another matter that should receive careful attention is the interaction between the

wheel and the rail itself. Under braking torque, the locked region in the wheel-rail

contact area is moved backward and generates areas of compression and tension as

shown in Fig 2.21 (Dukkipati (2000)). In parallel, the forward motion of the wagon is

decelerated by the longitudinal creep force developed in the contact area. The larger the

longitudinal creep force, the further the zero line of stress is displaced from a line

drawn perpendicular to the rail and through the centre of the wheel. If the longitudinal

creep force is large enough, all the contact area becomes a slip area or in other words

the wheel will purely slide (skid) on the rail. Skidding affects the safety of the wagon

so it is important to consider it in the wheelset dynamic model. The problem will be

more complicated if the friction coefficients between the wheel and the rail and

between the brake shoe and the wheel are treated as dynamic variables.

47
 Figure 2.21. Stress distribution in the contact area during braking (Dukkipati (2000))

We should also note that the magnitude of the total creep force is limited by the

saturation adhesive force between the wheel and the rail. As the total creep force is the

resultant of the longitudinal and lateral creep forces, we can deduce that at saturation

the longitudinal creep force could effectively modify the lateral and roll motions

between the wheel and the rail. From this fact we can also state that the application of

the braking force affects the lateral as well as the longitudinal and the vertical dynamics

of the wagon, especially at the onset of and during skidding.

2.7. SUMMARY

Railway wagon, bogie and wheelset dynamics cover the subjects of linear and non-

linear stability, curving performance, ride quality and comfort analysis, wagon track

interaction, and dynamic control systems. This extensive research is greatly supported

by the emergence of wheel-rail rolling contact theory and the improvement of

numerical analysis algorithms and computational strategies over the last three decades.

48
Only limited information on the dynamic behaviour of wagons and bogies under

longitudinal forces due to braking and traction are found in the published literature. As

wagons are mostly not self-propelled, the interest in the research of the effect of

longitudinal forces due to braking is much larger. Much of the research on wagon

braking is concerned only with the optimisation of braking distance and to avoid wheel

skid. At the onset of and during skidding, the longitudinal creep force affects the lateral

creep force at the wheel-rail contact patch as the magnitude of the resultant of the

lateral and longitudinal creep force is limited by the saturation adhesive force between

the wheel and the rail. The effect of the braking force to the lateral dynamics of the

wagon bogie especially at the onset of and during skidding is addressed in this thesis.

The complexity of modelling the dynamics of bogies during traction/braking emerges

as we cannot use a transfer coordinate system that moves along the track at the speed of

the wagon. The position and orientation of the contact patch, the creepages and creep

forces, which are usually defined in the track reference frame, need to be transferred to

the absolute reference frame to formulate the equations of motions. A comprehensive

explanation of the dynamic behaviour of the bogie under traction/braking condition is

undoubtedly required to maximise the efficiency and minimise the risk of railway

rollingstock operation. This thesis provides a contribution to this important area.

49