Rail Vehicle Dynamics

by
Evert Andersson, Mats Berg, Sebastian Stichel

Stockholm 2005

RAlLWAYGROUPKrn~' Div. of Railway Technology

Centre for Research and Educati011 ./ r . / Dep. of Aeronautical & Vehicle Eng.
. R'1
111 al way E ' .
·ngmecnng .:...- /' .//
Kungl Tekniska H6gskolan (KTH)
.,........ ' ".
Preface and acknowledgements

This book is mainly intended for the M.Sc. education in railway engineering at Kungliga
Tekniska Hdgskolan (KTH, Royal Institute of Technology), Stockholm, Sweden. The
compendium has been developed since the 1990s.

The objective is to give fundamental knowledge on rail vehicle dynamics and vehicle-
track dynamic interaction. This subject is even more important today due to trends of
increased vehicle speeds and/or axle loads. The compendium should also be used as a
reference book with its references and index.

Several persons and organisations have been involved in the development of this
compendium. In addition to the authors, significant contributions have been given by
Tech.Lic. Mikael Wrang, Ph.D. Pelle Carlbom, Tech.Lic. Tomas Jendel and Tech.Lic.
Rickard Nilsson. Also Banverket (Swedish National Rail Administration) and Bombardier
Transportation have contributed. Finally several persons have been involved in patiently
editing the texts and figures; in particular M.Sc. Ola Gabrielson, M.Sc. Lena Lukaszewicz,
Mr. Tomas Agren and Ms. Susanne Claesson. We want to cordially thank all these persons
and organisations for their support.

We hope that this book will enhance the knowledge and understanding of rail vehicle
dynamics and vehicle-track dynamic interaction.

Stockholm, 10 January 2005

Evert Andersson Mats Berg Sebastian Stichel

Professor Professor Researcher
Introduction 1
-----------------------------------------------------------
Track components, geometry and flexibility 2
-----------------------------------------------------------
Vehicles 3
-----------------------------------------------------------
Vehicle-track interaction 4
-----------------------------------------------------------
Dynamic analysis techniques 5
-----------------------------------------------------------
Advanced vehicle modelling 6

-----------------------------------------------------------
Wheel-rail guidance mechanisms 7

Creep and creep forces - dynamic stability and curving 8

Track forces and derailment 9

Wheel-rail wear 10

Ride comfort 11

Vehicle gauging 12

Examples on vehicle-track interaction 13

Dynamic calculations in vehicle design 14

Experimental techniques and validation 15

References 16

Symbols 17

Index 18
 Table of contents

1. Introduction ............................................. . 1:1

2. Track components, geometry and flexibility ................... . 2:1
2.1 Track components ............................................. . 2:1
2.1.1 Rails ................................................ . 2:1
2.1.2 Rail fastenings and pads ................................ . 2:3
2.1.3 Sleepers .............................................. . 2:5
2.1.4 Ballast ............................................... . 2:5
2.1.5 Jointed tracks and CWD tracks .......................... . 2:6
2.1.6 Classification of tracks related to their components ......... . 2:6
2.2 Alignment of the track ......................................... . 2:7
2.2.1 Track gauge .......................................... . 2:7
2.2.2 Horizontal circular CUIVes - radius and cant ............... . 2:7
2.2.3 Transition CUIVes and superelevation ramps ................ . 2:8
2.2.4 Longitudinal track gradient and vertical CUIVes ............. . 2:11
2.3 Track irregularities .................... ':'-: ...................... . 2:12
2.3.1 Measuring track irregularities ........................... . 2:12
2.3.2 Classification of track irregularities ....................... . 2:15
2.3.3 Track irregularities for dynamics analysis .................. . 2:17
2.3.4 Rail corrugation ....................................... . 2:19
2.4 Track flexibility ............................................... . 2:20
2.4.1 Vertical flexibility ...................................... . 2:20
2.4.2 Lateral flexibility ...................................... . 2:22
2.4.3 Track models for vehicle dynamics analysis ................ . 2:23

3. Vehicles ................................................. . 3:1

3.1 Classification and design principles .............................. . 3:1
3.2 Running gear, bogies ........................................... . 3:4
3.2.1 General .............................................. . 3:4
3.2.2 Wheelsets ............................................ . 3:4
3.2.3 Bogies ............................................... . 3:6
3.3 Carbodies .................................................... . 3:8
3.3.1 General .. " ..................... " ................... . 3:8
3.3.2 Structural flexibility and ride comfort ..................... . 3:9
3.4 Springs etc.................................................... . 3:13
3.4.1 General .............................................. . 3:13
3.4.2 Rubber springs ........................................ . 3:16
3.4.3 Airsprings ............................................ . 3:18
3.4.4 Coil springs ........................................... . 3:21
3.4.5 Leaf springs .......................................... . 3:21
3.4.6 Traction rods, anti-roll bars etc.......................... . 3:21
3.5 Dampers ..................................................... . 3:21
3.6 Carbody tilt .................................................. . 3:23
3.6.1 Passive and active tilt .................................. . 3:23
3.6.2 Trains with carbody tilt ................................. . 3:25
3.6.3 Three important demands on carbody tilt systems .......... . 3:26

I
 4. Vehicle-track interaction................................... 4:1
4.1 Fundamentals and definitions .................................... 4:1
4.2 Track plane acceleration, cant deficiency etc. ....................... 4:2
4.2.1 Track plane acceleration. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4:2
4.2.2 Equilibrium cant and equilibrium speed. . . . . . . . . . . . . . . . . . . . 4:3
4.2.3 Cant deficiency and cant excess. . . . . . . . . . . . . . . . . . . . . . . . . . . 4:5
4.2.4 Permissible track plane acceleration, cant deficiency and cant excess.. 4:6
4.2.5 Jerk, rate of cant and rate of cant deficiency .. . . . . . . . . . . . . . . 4:8
4.2.6 Permissible jerk etc. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4:9
4.3 Vehicle roll and sway, body plane acceleration ..................... . 4:11
4.3.1 Coefficient of flexibility ................................ . 4:11
4.3.2 Lateral acceleration in vehicle body plane ................. . 4:13
4.3.3 Calculation of vehicle sway .............................. . 4:15

5. Dynamic analysis techniques ............................... . 5:1

5.1 Simple models for analysis ...................................... . 5:1
5.1.1 One-dimensional model with 1 DOF .................... . 5:1
5.1.2 One-dimensional model with 2 DOF .................... . 5:4
5.1.3 Two-dimensional model with two DOF .................. . 5:5
5.2 Calculation of eigenvalues and eigenmodes ........................ . 5:8
5.2.1 One-dimensional model with one DOF .................. . 5:9
5.2.2 One-dimensional model with two DOF '~: ................ . 5:12
5.3 Calculation of frequency response functions ....................... . 5:14
5.4 Power spectral analysis ......................................... . 5:19
5.5 Time step integration .......................................... . 5:22

6. Advanced vehicle modelling ................................. 6:1

6.1 Wheelsets, bogie frames and carbodies as rigid body models .......... 6:1
6.1.1 General formulation. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6:2
6.1.2 Formulation with track-following and body-following reference syst. 6:4
6.2 Wheelsets, bogie frames and carbodies as flexible body models ........ 6:15
6.3 Suspension models ............................................ . 6:19
6.3.1 One-dimensional models for springs ..................... . 6:19
6.3.2 Two-dimensional model for springs ...................... . 6:27
6.3.3 Models for dampers ................................... . 6:28
Appendix 6A General motion of track - following reference system ......... . 6:35
Appendix 6B Airspring modelling ..................................... . 6:36
7. Wheel-rail guidance mechanisms ........................... . 7:1
7.1 Wheel-rail geometry .................. " ...................... . 7:1
7.1.1 Rail profiles and rail inclination ......................... . 7:1
7.1.2 Wheel profiles ........................................ . 7:2
7.1.3 Contact geometry ..................................... . 7:3
7.1.4 Contact geometry, contact point parameters ............... . 7:5
7.2 Guiding mechanisms ........................................... . 7:8
7.2.1 Sinusoidal motion of wheelset .. " ....................... . 7:9
7.2.2 Sinusoidal motion for wheelsets stiffly connected to a frame ... 7:12
7.2.3 Conicity and sinusiodal motions depends on the wheel-rail comb. 7:13
7.2.4 Determination of equivalent conicity in non - linear cases .... . 7:15
7.2.5 Curving perfonnance .................................. . 7:16
Appendix 7A Example of rolling radius difference and equivalent conicity. . . . 7:19
Appendix 7B Calculation of equivalent conicity for nonlinearcontact functions 7:21

II
 Creep and creep forces - dynamic stability and curving ........ . 8:1
8.
8.1 Creep ................................. , ......... , ... , ....... . 8:3
8.1.1 Definitions ........................................... . 8:3
8.1.2 Expressions for creep on left and right wheel .............. . 8:4
8.2 Nonnal contact force and contact patch ........................... . 8:8
8.2.1 Contact geometry and contact pressure according to Hertz theory 8:9
8.2.2 Examples of typical contact areas and contact pressures ..... . 8:12
8.3 Tangential contact - Creep forces ............................... . 8:14
8.3.1 Kalker's linear theory .................................. . 8:17
8.3.2 Approximate nonlinear theory ........................... . 8:21
8.3.3 Kalker's nonlinear creep force theory ..................... . 8:22
8.3.4 Kalker's simplified theory of rolling contact ................ . 8:23
8.3.5 The influence of spin creepage on the lateral creep force .... . 8:25
8.3.6 Some important cases with resulting creep forces ........... . 8:26
8.4 Lateral dynamic stability ....................................... . 8:28
8.4.1 Equations of motion for a wheelset with two degrees of freedom 8:29
8.4.2 Stability analysis by means of eigenvalues and root locii ..... . 8:32
8.4.3 Hunting stability in nonlinear case ....................... . 8:36
8.4.4 Which parameters influence stability? .................... . 8:38
8.5 Curving and flange steering .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8:42
8.5.1 Possible yaw angles of a wheelset in a curve. . . . . . . . . . . . . . . . . 8:42
8.5.2 Typical curving situation and forces for one-point contact. . . . 8:44
8.5.3 Typical contact point positions and larce dir. for two-point contact 8:46
8.5.4 Influence of primary suspension .......................... 8:49

9. Track forces and derailment ................................ . 9:1

9.1 Track forces .................................................. . 9:1
9.1.1 Vertical track forces (O-forces) ......................... . 9:3
9.1.2 Lateral track forces, track shift forces ..................... . 9:12
9.2 Flange climbing - derailment ................................... . 9:15
9.2.1 Y/O for a wheel ....................................... . 9:16
9.2.2 ~ Y/O for a wheelset ................................... . 9:18
9.2.3 Standards and practical approaches to avoid flange climbing ., 9:20
9.3 Vehicle turnover .............................................. . 9:26
9.3.1 Moment method ...................................... . 9:26
9.3.2 Intercept method and turnover process ................... . 9:28
9.3.3 Aerodynamic forces .................................... . 9:30
9.3.4 Train overturning wind velocity and accident frequency ...... . 9:32
9.3.5 Some practical implications ............................. . 9:34

10. Wheel-rail wear. ... . . . . .. . . . . . . . . . .. . . . . .. . . . . . . . . . . . . . . . . 10:1

10.1 Wheel wear ................................................... 10:1
10.1.1 Flange wear ........................................... 10:2
10.1.2 Wheel tread wear. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10:5
10.1.3 Out-of-round wheels. . . .. . ... . .. . .. . . . . . . . . . . . . .. . .. . . 10:5
10.1.4 Example on wheel wear ................................. 10:6
10.2 Rail wear ..................................................... 10:8
10.2.1 Gauge comer wear ..................................... 10:9
10.2.2 Top surface wear ....................................... 10:9
10.2.3 Example on rail wear. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10:9
10.3 Wheel-rail contact environment: friction, wear and lubrication. . . . . . . 10:12

III
 10.4 Prediction of wheel-rail wear. . . . . . . . . .. . . . .. . . . . . . . . . .. . . . . . . . . . 10:15
10.4.1 Simple wear prediction. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10:15
10.4.2 Advanced wear prediction ............................... 10:16

11. Ride comfort 11:1

11.1 Introduction. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11:1
11.2 Criteria for average comfort levels ................................ 11:2
11.2.1 Wertungszahl (Wz) and Ride Index (RI) ................... 11:2
11.2.2 Criteria according to ISO 2631 ........................... 11:4
11.3 Proposed CEN standard. . . . .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11:6
11.3.1 Mean value of ride comfort NMV . . . . . . . . . . . . . . . . . . . . . . . . . 11:6
11.3.2 NVA and NVD . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11:8
11.3.3 Comfort on curve transitions PCT ........... . . . . . . . . . . . . . 11:8
11.3.4 Comfort on discrete events PDE ......................... 11:10
11.4 Appropriate levels of lateral accelerations on passengers ............. 11:11
11.5 Motion sickness. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11:11

12. Vehicle gauging . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12:1

12.1 Definitions and general aspects. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12:1
12.2 Phenomena to be considered. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12:2
12.2.1 Geometric overthrow in curves ........................... 12:2
12.2.2 Lateral displacement of the axles (wheelS1ots) in the track. . . . . 12:5
12.2.3 Lateral displacement in primary and secondary suspensions . . . 12:6
12.2.4 Lateral displacement due to vehicle sway or carbody tilt ...... 12:6
12.2.5 Vertical displacement between wheeiset and track ........... 12:7
12.2.6 Vertical displacement in primary and secondary suspensions .. 12:7
12.2.7 Vertical displacement due to vehicle roll and sway or carbody tilt 12:7
12.2.8 Displacement due to vehicle asymmetry. . . . . . . . . . . . . . . . . . . . 12:8
12.2.9 Space for lateral, vertical and cant track irregularities ........ 12:8
12.2.10 Margins for phenomena not foreseen in the gauging procedures 12:9
12.3 Gauging procedures and standards . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12:9
12.3.1 Gauging principles. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12:9
12.3.2 European standards according to VIC ................. . . . . 12:12
12.3.3 Swedish gauging. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12:16
12.3.4 British gauging . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12:21
12.4 Dynamic gauging development ................................... 12:23

13. Examples on vehicle-track interaction. . . . . . .. . . . . .. . . . . . . . . . . 13:1

13.1 Vehicles and their wheelset guidance .............................. 13:1
13.2 Running on straight track. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13:3
13.3 Running on curved track ........................................ 13:11
Appendix 13A Example of quasistatic curving behaviour for different running gear
designs under different conditions .. . . . . . . . . . . . . . . . . . . . . . . . . 13:18

14. Dynamic calculations in vehicle design ........................ 14:1

14.1 Introduction. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14:1
14.2 Modelling vehicle-track dynamic interaction. . . . . . . . . . . . . . . . . . . . . . . 14:1
14.2.1 Track models .......................................... 14:1
14.2.2 Vehicle models. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14:2
14.2.3 Wheel-rail contact· models .............................. 14:5

IV
 Types of analysis and their application in vehicle design .............. 14:5
14.3.1 Quasistatic analysis. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14:5
14.3.2 Eigenvalue analysis ..................................... 14:6
14.3.3 Analysis in the frequency domain ......................... 14:6
14.3.4 Analysis in the time domain. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14:7
14.4 Design and development process ................................. 14:9
14.5 Reliability of simulation results. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14:11
14.6 Computer softwares ............................................ 14:11

Experimental techniques and validation ..... . . . . . . . . . . . . . . . . . . 15:1

15.1 Vehicle testing and the relevant standards. . . . . . . . . . . . . . . . . . . . . . . . . . 15:1
15.2 Measurement techniques and required equipment. . . . . . . . . . . . . . . . . . . 15:3
15.2.1 Types of measurements in vehicles ........................ 15:3
15.2.2 Wheel-rail forces. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15:3
15.2.3 Wheel wear measurements. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15:10
15.2.4 Other measurements. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15:11
15.3 Evaluation and presentation ..................................... 15:13
15.3.1 General. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15: 13
15.4 Practical considerations and quality assurance ...................... 15:15
15.4.1 Conditions during the tests. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15:15
15.4.2 Errors in measurement and evaluation. ... . . . . . . . . . . . . . . . . . 15:17
15.5 Special test facilities ........................... ::'. . . . . . . . . . . . . . . . 15:18
15.5.1 Vehicle testing laboratories .............................. 15:18
15.5.2 Roller rigs. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15:18
15.5.3 Test tracks. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15:21

16. References ~
.................................................... It .................. • It ...... It ........ ..
16:1
17. Symbols ........................................................
17:1

18. Index. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18:1

v
 Rail vehicle d)'llamics

1. Introduction

The aim of this book is to give an introduction into the dynamics of railway vehicles. It is
not only a reference book, but tries to explain the important phenomena connected with the
topic. In many cases it is also referred to existing standards, mostly UIC and CEN
standards (UIC = Union Internationale des Chemins de fer, CEN = Comite Europeen de
Normalisation).
Dynamics of railway vehicles and vehicle-track interaction covers a wide range of
technical subjects:
• Oscillations of vehicles, including ride stability.
• Track forces, i.e. forces between wheel and rail.
• Derailment due to the above mentioned oscillations and forces.
• Wear of wheels and rails due to the above mentioned oscillations and
forces.
• Passenger ride comfort and comfort for goods.
• Sudden safety risks due to destructioll of track.
• Fatigue of track and vehicle components.
• Air borne and structure borne sound propagation to vehicle and
environment.
Except for the last three all mentioned topics are treated in this book. An example of
consequences of poor running behaviour of a vehicle is shown in Figure I-I.

poor running behaviour

poor
/ high
! large wear of
passenger track forces wheels + rails

~
comfort
+
bad image
l
large track reprofiling of wheels
irr7ularities replacement of wheels
replacement of rails
1 vehicles in workshop
fewer pas- extensive

::;k{;;;;:~"~~
sengers

rll-ow-er-'-r-e-v-en-u-e-s'll higher costs Irl-h-'ig'-'I-le-r-c-o-st-s--'

Figure 1-1 Consequences of poor running behaviour.

As can be seen from Figure 1-1 the dynamic properties of railway vehicles have great
influence also on the economics of a railway system. Below two real examples try to
illustrate the importance of the running behaviour of a vehicle on the costs for the operator.

1:1
Introduction Rail vehicle dynamic.

1. In the middle of the 1970s, the American operator AMTRAK bought express Stockholm - Boden Stockholm-G6teborg
locomotives designed for a maximum speed of 125 miles/h (200 km/h). The (line with many curves) (line with few curves)
locomotives had two three-axle bogies and a maximum axle load of 29 t.
Passenger coach bogies:
The first locomotive delivered derailed during the first test at a speed of 100 miles/h
(about 160 km/h). Investigations showed that the locomotive had run unstably travelling distance (km/year) 300000 240000
including an extreme so-called hunting motion. The lateral forces had become so
Cost savings per coach (SEK/year) by
high that the rails turned over.
reduced wheel flange wear 42000 12000
Consequence: reduced rail wear 14000 6000
After major modifications and tests finally the locomotive was accepted for a reduced running resistance in 4000 2000
maximum speed of 85 mileslh ( about 135 km/h). The price of the locomotives was curve (energy consumption)
reduced with the ratio 85/125, i.e. the ratio between the finally accepted and the Total (SEK/year): 60000 20000
designed maximum speed.
Locomotive bogies:
2. 1984 and 1985 SJ (Swedish State Railways) and ASEA (today part of Bombardier
travelling distance (km/year) 225000 225000
Transportation) carried out several investigations regarding possible cost savings
by introducing so-called radially steering bogies. Such bogies have a soft primary Cost savings per loco (SEK/year) by
suspension in the horizontal plane, so that the wheelsets have the possibility to
negotiate curves with an essentially radial position on the track (down to a certain reduced wheel flange wear 32000 8000
curve radius). reduced rail wear 14000 4000
reduced running resistance in 4000 2000
By radial steering the wheel-rail friction forces can be reduced and thus less wear
curve
in the contact areas between wheel and rail can be achieved.
Total (SEK/year): 50000 14000
The background for the investigations was that wheel wear - especially on
locomotives and passenger coaches - had increased dramatically during the last Table 1-1 Reduced maintenance costs due to better steering in curves for twl
decade. The annual maintenance costs were estimated at that time to between 50 bogies and two lines.
and 100 millions SEK per vehicle.

Wear occured mainly during periods with dry weather, especially in spring and
summer. It could be proved that the maintenance costs were clearly reduced after
the introduction of radial steering bogies. Table 1-1 shows the calculated cost
reductions which could be achieved.
In the table costs for material, labour, energy costs (reduced running resistance in
curves) and costs for having the vehicle in the workshop are included.

Another possibility of reducing the costs due to wear in curves is lubrication of the
gauge face of the outer (high) rail in a curve. If lubrication is applied consequently
the difference in costs between conventional and radial bogies is less.

1:2 1:3
 Rail vehicle dynamic

2. Track components, geometry and flexibility

The presentation of the track in this chapter will concentrate on the topics that 31
important for the vehicle-track interaction. A good introduction to the field of railwa
track can be found in [20].

2.1 Track components

In this section different track components are presented. Figure 2-1 shows a track cro~
section with the most important track components. The track is divided into
superstructure and a substructure. The superstructure includes rails, rail fastenings, pad.
sleepers and ballast. The subgrade is the substructure of a track. The substructure may als
consist of a bridge structure.

Superstructur
Ballast depth Ballast

Sub grade Substructure

Figure 2-1 Track cross section - some important components.

In for instance Germany and Japan, there are also tracks with the ballast replaced b
oblong concrete slabs. Such tracks have higher initial costs but are expected to have Ie,
maintenance, for instance with respect to track irregularities

2.1.1 Rails

The rails have to work as both running surfaces and as load bearing design elements. A
running slllfaces they must be smooth, to ensure that the vehicles run quietly and with
low level of vibration. As design elements the rails have to withstand static and dynami
loads, in all three directions and to transfer them to the ballast with help of the pad;
fastenings and sleepers. Thermal loads have to be endured by the rails, too. Th
requirements on rails can be summarized as follows:

- Tight geometric tolerances.

- High demands on straightness of rails.
- High hardness and tensile strength.
- High toughness.
- High wear resistance.
- High fatigue strength.
- Low brittle transition temperature.
- Good welding properties.

2:1
 Rail vehicle dynamics
Track components, geometry and flexibility

Two widely used rail materials today are a carbon manganese steel with 800-900 MPa 72
706
tensile strength (e.g. UIC900A) and a carbon chrome manganese steel with 1080-1200 52 .[
MPa tensile strength (e.g. UICllOO) which is used on special track sections, like tunnels r-¥---i "j

t
and outer rails in tight curves, or on railway lines with high axle loads.
0 ~Lfl
The so called Vignoles rail is the most common rail type in use today. It was introduced
g ~ :;;
;;;
already in 1837 by the railway engineer Charles B Vignoles. Channel rails are used for
tramways and industry track. Cross sections of these two types of rails are shown in Figure
<1 1S
2-2.

= 65-75 mm
Top surface

16.5
y-~x_is_ __

=140-190mm Gauge face

=120-160 mm
(a) (b)
Figure 2-2 The two most important types of rails: (a) Vignoles rail, (b) Channel rail.

The gauge face points to the centre of the track. The geometry of the rail top surface and z-axis
gauge face is very important for the vehicle-track interaction. In Figure 2-2 also typical
dimensions for Vignoles rails are given. The tendency today is that rail height and width Figure 2-3 m
Rail profile UIC60. Mass per length unit =60.3 kg/m, cross sectional
are increased, because of higher axle loads and higher speeds. The most common rail on areaA = 7.686,10- 3 m2, area moment of inertia Iy = 30.55.10-6 m4, area
main lines in Europe today is the UIC60 rail, where 60 stands for the rail weight 60 kg/m. moment of inertia Iz = 5.129'10- 6 m 4,flexural resistance Wy =335.5,10- 6
The cross section ofthe UIC60 rail is shown in Figure 2-3. In North America and Germany m3 [N2j.
even rails with a weight of 70 kg/m are used on some new lines.
The rails are often mounted on the sleepers with an inclination, as can be seen in Figure 2.1.2 Rail fastenings and pads
2-1. Rail inclination leads to a better fit between wheel and rail profiles, and to a better
transfer of the vehicle loads to the sleepers and the ballast. Typical rail inclinations are 1:20 The main task of rail fastenings is to avoid displacements of the rail on the sleeper, so that
(e.g. France) and 1:40 (e.g. Germany). In some turnouts no inclination exists. As a the track gauge is kept constant. Together with the rail pad also the longitudinal, lateral and
compromise the inclination 1:30 has been chosen in Sweden. vertical loads have to be transferred from rail to sleeper. More requirements are
- Simple mounting and maintenance.
- Electrical isolation.
- Demands on elasticity and acoustic damping (sometimes).
Below some examples of existing rai! fastenings for wooden and concrete sleepers are
shown.

2:3
2:2
 Track components, geometry and flexibility
Rail vehicle dynamics

2.1.3 Sleepers

The sleepers have the following functions:

Heyback - Elastic foundation for rails, i.e. transferring the point loads on the rails to stresses in the
ballast.
Wooden sleeper. Base plate. Rubber - Fixing the rails in lateral direction to keep the track gauge.
pad. Clamping force of 15 kN per fas-
tening. - Provide flexural resistance in the lateral direction in interaction with the rails.
If the distance between sleepers is not constant this can lead to higher bending stresses in
the rails and ballast, which can cause plastic deformations and thus track irregularities.
0.5-0.65 m are typical distances between sleepers on mainlines in Europe.

~ 'l
Today on mainlines in Europe mostly concrete sleepers are used. A concrete sleeper used
in Sweden is shown in Fignre 2-6.

----- -_~_J
K-fastening

(':I_,,,"nl 83

JJ
r--~·;------- Wooden sleeper. Base plate. Usually .&!
8 :!:~
~I
t2~ ~
pad of impregnated wood fibre.
f :: i'"4f§l Often used in Germany.
500 750
Figure 2-4 Examples of so called indirect rail fastenings on wooden sleepers.

~~ ~ O·~
~
[N22J.
0 !!!I §I
2500
Fist
Figure 2-6 Concrete sleeper SJ type S3 [N21J. Weight about 300 kg.
Concrete sleeper. Rubber pad 4.8 mm.
Disadvantage: The strength of the sleeper is
reduced at the fastening bolt. Risk of cracks.
Not used for new lines. 2.1.4 Ballast

The tasks of the ballast are:

Pandrol
Concrete sleeper. Plastics pad 5.5 mm,
today rubber pad about 10 mm. Widely
used in Europe for new lines.
Figure 2-5 Examples of rail fastenings on concrete sleepers [N22J.
- To withstand the static and dynamic loads and distribute them to the subgrade.
- To stabilize the track in lateral and vertical directions.
- Provide some elasticity.
- Provide drainage.
- To be hard and have a good wear resistance.
Ballast stones should have about the same size, be about cubic and have sharp corners tl
provide high friction between the ballast stones themselves and between ballast alli
sleepers. As mentioned above the !:>.allast has to stabilize the track, i.e. the resistance again;
displacements of the sleepers should be high. In Figure 2-7 an example of widths of ballw
shoulders for different curve radii is shown.

)'4
2:.5""
 Track components, geometry and flexibility Rail vehicle dynamics

welded rails: R 2: 500 m, 2.2 Alignment of the track

~~
Alignment of the track means the design geometry or nominal geometry of the track, e.g.
transition curves, circular curves and cant. Deviations from this geometry are called track
irregularities, or misalignment, and are described in the next section. The following
quantities are introduced in this section
- Track gauge
welded rails: 400:s R < 500 welded rails: R < 400 - Curves: curve radius, circular curve, transition curves.
- Cant (superelevation), superelevation ramps.
- Longitudinal gradients, vertical curves.

2.2.1 Track gauge

In Figure 2-8 some important quantities concerning the definition of track gauge are
Figure 2-7 Ballast shoulders. Arranged to increase the lateral stability of the track shown. As can be seen the gauge is measured 14 mm below the track plane, which is
in tight curves. SlF 540.1 [N20j.
formed by joining the tops of both rails.
track gauge
top of rail (tor)
2.1.5 Jointed tracks and CWD tracks ------e------
Historically short rail segments were put together to form the rails along the track. The rail
14 mm f 14 mm

segments were only 5-20 m long and connected to each other through joints with a
nominal gap of about 20 mm to allow for some rail expansion during hot summer days and
thereby reduce the risk of track buckling in the horizontal plane.
But as the track stability was improved through the years these jointed tracks could have
longer rails. Today mainline tracks have continuously welded (CWD) rails to allow
smooth vehicle-track interaction and thus reduced wheel and rail damage as well as
improved ride comfort. Due to the very good lateral track stability the track seldom buckle,
even during hot days.
2.1.6 Classification of tracks related to their components
A track subjected to trains of high-speed and/or heavy axle load must be designed to
withstand this traffic so that the track deterioration is limited and the safety is high. Such a
track typically consists of the following high-performance components: CWD rail UIC60,
Pandrol fastenings, concrete sleepers and ballast stones. On the contrary, very unqualified
traffic may only demand a simple track like: jointed rails of 40 kg/m, spike fastenings,
wooden sleepers and sand ballast.
In this way the railways usually classify their tracks into, say, 4-5 different classes.
centre of the track
Figure 2-8 Track gauge etc.
Standard gauge track is a track with a n~minal (design) gauge of 1435 mm. Major parts of
European railways use standard gauge. Also USA, Canada, Mexico, China, Japan (new
high speed lines) and other countries use standard gauge.
Tracks with nominal gauge less than 1435 mm are called Narrow gauge tracks. Typical
gauges are 1000 mm (e.g. South America and Africa) and 1067 mm (Japan, South Alrica,
Queensland in Australia).
Broad gauge tracks are tracks with nominal gauge greater than 1435 mm. Examples are
1524 mm (Finland and Russia), 1600 mm (Ireland), 1668 mm (parts of Spain, Portugal)
and 1676 mm (parts of India).
2.2.2 Horizontal circular curves - radius and cant
In a circular curve the radius R and thus also the curvature is constant. The curve radius is
related to the centre of the track, see Figure 2-9.

~~~"""'''''"'
/" mnerrall"~
s = longitudinal
track position
curve radius
R(s) =R =constant

Figure 2-9 Curve radius R for a horizontal circular curve.

2:6 2:7
 Track components, geometry and flexibility Rail vehicle dynamics

Examples of typical minimum curve radii on standard gauge track are: I/R
- 190 m in turnouts. centre line of track
60-90 m at workshops.
- 30-60 m for metro lines and ~ ____ ____-+________________ ____
~ ~ ~~ ______ ~~ s
- 15-20 m for tramways. tangent Itransition circular curve I transition tangent
track ,curve ,curve track
On mainlines built a century ago or so, the curve radii are typically in the span of 500 to hI
2000 m. For mainlines built during the last 1-2 decades the radii are about 2000-5000 m,
outer (high) rail
sometimes even wider.
In most curves, except on tramways and diverging track of turnouts, the outer rail in a
curve is superelevated to reduce the effect of the centrifugal acceleration for a vehicle
~____-4~____ -+______~i~n~ne=r~(=lo~w~)~r=ai~I__~______~________~~S
negotiating a curve. This is called cant or superelevation. The cant, hI, is defined according
to Figure 2-10. For this definition the base, 2bo , is the distance between the nominal Figure 2-11 Example for alignment of a horizontal curve. In the transition curve the
wheel-rail contact points on the left and right side. For standard gauge track, 2bo is equal to curvature l/R is changed linearly and in the superelevation ramp the cant
1.500 m. For canted curves the outer and inner rails are also called high rails and low rails changes linearly.
respectively.
.'Note that turnouts usually lack cant, and thus the diverging track is directly connected to
tangent track, i.e. no transition curve exists.

As indicated above the curvature and the superelevation in transition curves correspond to
each other. Therefore their geometries below are shown in parallel. In Figure 2-12
transition curves with linearly and parabolically increasing (or decreasing) curvature/
track plane superelevation are shown.
1/R ht
outer rail
-----------'7r'-----

Figure 2 -1 0 Cant lit and cant angle CPt.

The angle CPt in Figure 2-10 is called cant angle and given by inner rail
s

. hI (a)
CPI = arcs1ll 2bo (2-1)
1/R
A typical maximum value for permissible cant is 0.150 m. On some lines with only outer rail
passenger traffic 0.180-0.200 m are permitted. A more detailed introduction into the 1/1<0 hl,o - - - - - - - - - -,;--,------
vehicle-track interaction in curves and a presentation of permissible values for cant, lateral
acceleration, vehicle speed etc., is given in Chapter 4.

2.2.3 Transition curves and superelevation ramps inner rail

~~~ ____-L__________L -______ ~ S
A circular curve with cant can not be combined directly to a tangent track. A transition Lp/2
between these two types of alignment elements is thus necessary. Usually the curve radius (b)
(or the curvature) is changed continuously between tangent track and circular curve, on the
so called transition curve. Also the cant is changed continuously and thus we get a so called Figure 2-12 Transition curves with
superelevation ramp. Usually the transition curve and the superelevation ramp have the (a) linear curvature!cant.
same start and end positions along the track, which is illustrated in Figure 2-11. (b) parabolic curvature!cant.

2:8 2:9
 Track components, geometry and flexibility Rail vehicle dynamics

Mathematically the varying curvature and cant can be described as below: The increased twist in the track of a parabolic ramp increases the risk of derailment,
especially for vehicles with stiff vertical suspension (cf. Chapter 9). Buckling of the track
Linear curvature/cant is also one reason for that the maximum gradient of ramps has to be limited. The linear
ramp gradient is often restricted to 1/400, i.e. a ramp index of 400.
1 (s) =
-R ~ . -L (clothoid) (2-2)
LI Ro
2.2.4 Longitudinal track gradient and vertical curves
and
The longitudinal gradient of a track is limited for two reasons
htCs) =f . hl,o (2-3) - Locomotives and other power units have to be able to pull the train uphill.
I
- The brake systems have to be able to brake a train downhill without thermal problems
where LI is the length of the linear ramp whereas Ro and hl,o are the curve radius and the and too long braking distances.
cant in the circular curve. The ramp transition curve starts at s =O.
As an example; on lines with mixed traffic the permissible gradient in Sweden is about
Parabolic curvature/cant 10-12:5 %0. The high speed lines in France for example, which are only used by TGV
trains, are built with gradients up to 25-35 %0.

~(s) = (lJ2 .~
Ro
for 0<
- s
-< 22
L
(2-4a)
At locations where the gradient changes the connection has to be smoothed with a vertical
curve. Otherwise there is theoretically, an infinitely high vertical acceleration at the
connection positions, which gives rise to very high vertical wheel-rail forces and poor
passenger ride comfort. The radius of such a vertical curve has to be related to the
~(s) -L_
Ro (L pL: sr .~
Ro
for Lp <
2' - s- <
Lp (2-4b) permissible speed on the line. In DIC Code 703 [N30] the quasi static vertical acceleration
has to be lower than a certain recommended limit value

and
(v max,km/h)2
az = 12.96 . Rv m :s: az,recommended (2-8)

hls) = (lpr . 2hl,o for 0<

- s L
-< 22 (2-5a)
Examples of recommended limit accelerations are given in Table 2-1 for different
categories of traffic.

hls) = hl,o - (L T sr .
- 2 hl,o for Lp <
2' - s- <
Lp (2-5b)
Category I Category II Category III Category IV
Vmax 80-120 120-200 '" 250 250-300
where Lp is the length of the linear ramp. [km/h]

For ramps usually also the first derivative of the cant and its inverse value (ramp index) are FS DB SNCF
given. For linear ramps we get Stand Max. Exe. Stand Max. Exe. Stand Max. Stand Max. Stand Max.
az 0.20 0.30 0.40 0.20 0.30 - 0.16 0.24 0.20 - 0.45 0.60
[m/,2]
Ramp gradient h I '(S)_
--
ht,o (2-6)
LI
I: Mixed traffic lines, with passenger train speeds higher than 80 km/h and up to 120 km/h.

II: Mixed traffic lines, with passenger train speeds higher than 120 km/h and up to 200 km/h.
Ramp index (2-7)
III: Mixed traffic lines, designed for passenger train speeds up to 250 lan/h. Values used on
the FS and DB new lines are given.
The advantage with parabolic ramps is that the second derivative does not become
infinitely high at the beginning and end of the transition curve. For parabolic ramps the IV: Dedicated passenger lines with speeds over 250 km/h and capable of reaching 300 km/h.
first derivative varies with the distance s. The maximum gradient 2h"o/Lp is twice as large values used un the SNCF TGV lines are given.
as for linear ramps provided Lp =LI. In practice the advantages with parabolic ramps are
small. The theoretical improvement tends to be negligible compared to the track Table2-l Examples of recommended vertical accelerations in vertical curves
irregularities. according to UIe 703 [N30j.

2:10 2:11
Track components, geometry and flexibility Rail vehicle dynamics

2.3 Track irregularities a ___ real irregularity

Track irregularities are the deviations of the track from the design geometry. They are as
wavelength L
important for the wheel-rail forces and the passenger ride comfort as the alignment of the
track. Therefore it is important to know, at least approximately, the irregularities and to
limit them by tamping the track within certain intervals. ___ recorded irregularity
b

2.3.1 Measuring track irregularities (a)

~ real and recorded irregularity b =a
For measuring of track irregularities special testing cars are usually used. Four quantities
are measured by most of these measuring cars: _____ ,,"'---o=-~_ __ __ a ____ _

Long. level: Geometrical error in the vertical direction of the longitudinal-vertical

plane. L = 2· measuring base
(b)
Line: Geometrical error in the lateral direction of the horizontal plane.
recorded irregu-
Gauge: Track gauge, defined in Section 2.2.1. larity with amp- real irregularity
litude b ' - c::::;;;e;:;2=:::~::S;:e:::1 / b« a
Twist: Twist is the difference of cant between two cross sections of track
divided by the distance. This also gives the cant irregularity.
a
In principle two different measuring systems exist: Inertia based systems, where it is measurin
possible to measure the absolute, or real track irregularities, and relative measuring base
systems with mechanic transducers, where the real track irregularities are found after a wavelen b
L = 6 . measuring base
transformation of the measured signals. (c)
Absolute measuring systems are based, as mentioned, on inertia systems. Often Figure 2 -14 Transfer function b/a for versed sine measurement. Amplitudes a and b,
accelerometers, gyros and as a complement optical systems (laser) are used, as in the wavelength L.
Swedish testing car STRIX [N4]. (a) Definition. (Amplitudes a and b occur at the same time)
(b) Example: L = 2· measuring base gives b=a.
A typical consequence of a relative measuring system is shown in Figure 2-13. The versed (c) Example: L = 6· measuring base gives b« a.
sine which is measured is not equal to the real irregularity, which however has to be known
for correction of the track and as input for dynamic analysis of vehicles.
Another problem with relative measuring systems are irregularities with long
B versed sine wavelengths. As can be seen in Figures 2-15 and 2-16 the transfer functions have
relatively low values for wavelengths larger than five or ten times the measuring base, i.e.
larger than 50-100 m. For high speed trains however the irregularities with long
absolute (real) wavelengths can be very important for the passenger comfort. For lines where trains run
irregularity A B' c with v = 250 km/h (= 70 m/s) wavelengths at least up to 140 m should be recorded.
Figure 2-13 Measuring of versed sine. Measuring base is the distance A-C. The
versed sine is measured relative to a point B', often in the middle between
A and C.

The examples in Figure 2-14 explain the difference between real and measured
irregularity for two different wavelengths of the irregularity. Examples of transfer
functions for longitudinal level and line from the Mauzin vehicle, which has its origin from
France, are shown in Figures 2-15 and 2-16. As can be seen there are singularities in the
transfer function for the line (lateral) irregularity for a wavelength of half the measuring
base, a fourth of the measuring base and so on. The longitudinal level is not measured with
versed sines and the transfer function does not have any singularities.

2:12 2:13
Track components, geometry and flexibility Rail vehicle dynamics

bla 2.3.2 Classification of track irregularities

2 b I-cos IOn
Ii L Most railways classify their tracks with respect to permissible track irregularities. On
mainlines for example less irregularities are permitted. Here we refer to the classification
in the UIC Code 518 concerning test and acceptance of railway vehicles [N28], see also
1.5 [NlO]. Three classes of track qualities are defined with regard to the necessity of
maintenance and to the applicability for acceptance tests of vehicles:
QNl refers to the value which necessitates observing the condition of the track or
taking maintenance measures as part of regularly planned maintenance
operations.
QN2 refers to the value which requires short term maintenance action.

0.5 QN3 refers to the value which, if exceeded, leads to the track section being
excluded from the acceptance analysis because the track quality
encountered is not representative of usual quality standards.
"Value" means in this case both standard deviation and peak value of a measured track
5 10 50 irregularity. In Tables 2-2 and 2-3 the permissible values of longitudinal level and line for
wavelengthL (m) standard deviations and peak values for the classes ONI and ON2 are given. For the
Figure 2-15 Transfer function bfafor line with amplitude a and wavelength Lfor calculation of peak values the permissible values for ON3 are calculated with
Mauzin recording car [55]. Versed sine measuring with base = 5 + 5 =
10m, b = recorded versed sine.
QN3 = 1.3 . QN2

Wavelengths below 3 m and above 25 m of the track irregularities are filtered out.
bla
2
Standard deviation: ON1 [mm] ON2 [mm]
longitudinal level
1.5
o <v<SOkm/h 2.3 2.6
SO < v < 120 kmlh 1.8 2.1
120 < v < 160 km/h 1.4 1.7
160 < v < 200 km/h 1.2 1.5
200 < v < 300 km/h 1.0 1.3

Standard deviation: ONl [mm] ON2 [mm]

0.5 line
o <v<80km/h 1.5 1.8
80 < v < 120 km/h 1.2 1.5
o1L-----~--~~5~~~1~0~----~----5-0~~~~· 120 < v < 160 km/h 1.0 1.3
wavelength L (m) 160 < v < 200 km/h 0.8 1.1
Figure 2-16 Transfer function bfa for longitudinal level with amplitude a and 200 < v < 300 km/h 0.7 1.0
wavelength Lfor Mauzin recording car [55].
b = recorded amplitude. Eight measuring wheels, i.e. no versed sine Table 2-2 Permissible standard deviationsfor longitudinal level and line for track
classes QNl and QN2 according to VIC Code 518 [N28].
measuring system.

2:14 2:15
Track components, geometry and flexibility Rail vehicle dynamics

Maximum isolated error: QNl QN2 Even though the track qualities are defined with standard deviations and peak values the
longitudinal level [mm] [mm] UTC is sceptical whether this actually is sufficient. UTC 518 states:
0<v<80km/h 12.0 16.0
According to current knowledge, it is uneasy to draw a simple relation between
80 < v < 120 km/h 8.0 12.0 track-geometry and vehicle-response. As result, it is not possible either to relate a
120 < v < 160 km/h 6.0 10.0 geometrical value to a track geometric quality or a given track geometrical error to a
160 < v < 200 km/h 5.0 9.0 vehicle-response.
200 < v < 300 km/h 4.0 8.0
2.3.3 Track irregularities for dynamics analysis

Maximum isolated error: QNl [mm] QN2 [mm] As input data for analysis of vehicle dynamics the absolute irregularities as function of
line distance of course have to be known. In principle the unwanted vertical and lateral
o < v < 80 km/h 12.0 14.0 displacement of both rails have to be known. In practice however they are usually
represented similarly to the description in section 2.3.1 as
80 < v < 120 km/h 8.0 10.0
120 < v < 160 km/h 6.0 8.0 • Longitudinal level (mean value of vertical irregularities ofleft and right
rail)
160 < v < 200 km/h 5.0 7.0
200 < v < 300 km/h 4.0 6.0 • Line (mean value of lateral irregularities of left and right rail)
• Cant irregularity (deviation from nominal cant)
Table 2-3 Permissible peak values for isolated track errors for longitudinalleve!
• Gauge irregularity (deviation from nominal track gauge)
and line for track classes QNl and QN2 according to UlC Code 518
[N28]. The four defined track irregularities are illustrated in Figure 2-17.

The values ONI and ON2 relating to standard deviations and peak values for isolated track
errors have been obtained from measurements with the NS measuring vehicle. If the
measuring vehicle of another railway is used for measurements, the values of ON1 and longitudinal direction
ON2 shall be corrected and the coefficients to be applied can be found in Table 2-4.

Coefficient K
Measuring vehicle Longitudinal level Line
BR 1.14 1.20
CFF 0.91 1.24
CFF/long 1.25 -
CFR 1.40 1.95
CSD 1.52 1.77
DB 1.24 1.47
FS 1.33 1.72
NS 1.00 1.00 Figure 2-17 Longitudinal level, line, cant irregularity and gauge.
PKP 0.73 0.71
An example of track irregularities is given in Figure 2-18. As an alternative to the
RENFE 0.91 1.47 representation of track irregularities as functions of distance, they can be represented as
SNCF 0.91 1.47 functions of wavelength or frequency. Power spectral densities (PSD) are widely used for
this purpose. Somewhat simplified it can be said that a PSD is the square of the Fourier
Table 2-4 Correction factors if other measuring vehicles than the NS vehicle are transform of a signal, ct. Chapter 5. The phase shift between the different frequency
used. ao/her = K . aNS vehicle; Peak value a/her = K . peak value NS vehicle components in the signal is lost in a PSD. A PSD can be used either directly as an input for
power spectral analysis, described in Section 5.4, or they can be retransforrned into track
irregularities as function of distance. However, we will not get the original signal in this

2:16 2:17
Track components, geometry and flexibility Rail vehicle dynamics

case but a signal which is statistically equivalent. An example of power spectra of track
irregularities is shown in Figure 2-19.
line long. level

"'.L,_~-_.LA-~-"-J

CYCLES/1'01.1

gauge cant irreg.

""_,':--~-_.LA-~-+.J
CYCLES/.1'14

Figure 2-19 Example of absolute track irregularities as power spectral densities.

Logarithmic axes. On the horizontal axis -1 and -2 mean wavelength
10m and 100 m respectively.

2.3.4 Rail corrugation

In addition to the irregularities above, there are sometimes small irregularities with short
wavelengts on the rail running surface. These irregularities are referred to as rail
corrugation. The wavelengths are typically 20-500 mm and the amplitudes 0.1-1 mm.

These corrugations are due to irregular wear of the rail top surface caused be a rather
o 100 200 300 500 600 700 800
complicated vehicle-interaction at frequencies up to 1000 Hz.
DlSTANCB
The rail corrugations can give extensive vehicle-trade vibration and noise and should be
Figure 2-18 Example of absolute track irregularities. The irregularities are in mm. removed through grinding of the rail head.

2:18 2:19
Track components, geometry and flexibility Rail vehicle dynamics

2.4 Track flexibility 250

Tracks are not rigid but more or less flexible. The pertinent stiffness and damping --0-- concrete
properties of the track are very important for the dynamic forces and the oscillation 200 r------

~V
- x - wood
phenomena occurring between wheel and rail. The loads also influence life of track and
vehicle components. Generally speaking stiffer track gives rise to higher dynamic forces stiffness 150
than softer track. Higher damping leads to less oscillations. [MN/m] ~~
100
Earlier the railways have not looked into these phenomena from a theoretical point of view.
The influence of different track parameters has been investigated by tests and statistical x _ _x _ _ x_x_x_x_x_ _ x _ _x_x_x-x
50 -x-
studies. Today, however, a better understanding of the basic phenomena is desired due to
the necessity to reduce the maintenance costs. In the analysis of vehicle dynamics,
however, very simple track models are used in most cases. Therefore in this section only 0
the most basic phenomena are described. o 10 20 30 40 50
frequency [Hz1

2.4.1 Vertical flexibility

500

,
The vertical flexibility of the track depends on both the load and the frequency of an
450 -0- concret ~

oscillating load. Often, for example, there is a "play" existing between rail and sleeper or
between sleeper and ballast. When the play has been overcome the track becomes stiffer. 400 - x - wood
!I
? ""-
350
An example of the vertical deflection of a rail as function of vertical load for different types 300 /' ~
Jf'
of track is shown in Figure 2-20. It can be seen that the track with wooden sleepers is more
flexible than the track with concrete sleepers. Frozen track is stiffer than non-frozen track.
The dependence of stiffness and damping due to excitation frequency is shown in Figure
damping
[kNs/m]
250
200
150
~
L "
2-21. Again it can be seen that the track with concrete sleepers is stiffer than the track with 100
wooden sleepers. The stiffness has the tendency to increase with increased frequency. 50 ~ -x-x-x___
~-~ r"
Even the damping is higher for the track with concrete sleepers. However, the results for 0
damping are associated with many uncertainties. o 10 20 30 40 50
Vertical displacement frequency [Hz1
(mm)
3 Wooden sleepers, mid October
(non-frozen track) Figure 2-21 Vertical track stiffness and damping for 5-50 Hz excitation and two
different tracks: Wooden sleeper with Heyback fastening and concrete
sleeper (mono lit) with Fist fastening. Static preload 50 leN/rail. Dynamic
load amplitUde 20 leN/rail. Measured on line Ramsj6 - NorrhOg in May
1990. Stiffness and damping values apply to one rail [12}.
2
Wooden sleepers, early March
(frozen track) A special variation in track flexibility arises through the discrete foundation of the rails on
the sleepers. Even though the difference in static deflection is less than a tenth of a
Twin block concrete sleepers,
mid October (non-frozen track)
millimeter, the variations in vertical load between wheel and rail may be high. The
Monolit concrete sleepers, mid frequency which arises is called sleeper passing frequency, see Figure 2-22.
October (non-frozen track)
Twin block concrete sleepers, early
Example: For a speed v = 130 km/Q (36.1 m/s) and a sleeper distance I: 0.65 m the
sleeper passing frequency becomesf: 36.1/0.65: 56 Hz. For v: 200
~~~;;::::::::::::::::====March (frozen track)
Monolit concrete sleepers,
early March (frozen track)
km/h we get f: 85 Hz.
~~~~

.,.."'---------I----------!---I.. vertical force

o 50 100 per rail (kN)

Figure 2-20 Vertical displacement of rail asfunction of vertical static load.

Measured by Banverket.

2:20 2:21
 Track components, geometry and flexibility Rail vehicle dynamics

2.4.3 Track models for vehicle dynamics analysis

As indicated above the vehicle is influenced by the flexibility of the track. Continuous
models of the track based on rails regarded as beams are too complicated to be
implemented in vehicle dynamics simulation models. The main reason is the enormous
calculation time which would be needed, due to the large number of degrees of freedom
which would be needed to represent the track.

(a) Instead a simple model like the one in Figure 2-24 may be used. In this model a piece of
track is situated under each wheelset, hence the track piece is following the vehicle with the
same speed. All different stiffnesses existing between rail head and subgrade (e.g. ballast
and pad stiffness) are concentrated to one lateral and two vertical stiffnesses between track
piece and reference frame. The springs and dampers may be nonlinear. However in most
cases today linear springs and dampers are used.

Sleeper
I

(b)
m,
Figure 2-22 Parametric excitation by discrete foundation of rails. Sleeper passing
[requency:/=v/L (a)t=to (b)t=to+T (T=l//=I/v)

2.4.2 Lateral flexibility

The knowledge about the lateral track flexibility is limited. Little can be found in the
literature. Some tests, however, have been performed. The lateral stiffness may be defined
as in Figure 2-23. Figure 2-24 Track model at each wheelset [12J.
Qo = 70 kN
The sleeper passing effect described above can be taken into account in an approximate
way using such a track model. The vertical track stiffness can be defined as variable with
the longitudinal position x:
stiffness = ;

kztCx) = k zt ,lIomillal . ( 1 + A sin(~X + cp) ) etc., where

A =non-dimensional amplitude for variation of stiffness

L = wavelength of variation of stiffness
cP = phase shift for variation of stiffness

Figure 2-23 Definition of measured lateral stiffness. (example)
Typical stiffnesses, as defined above, are for non-frozen track:
• Track with wooden sleepers with base plate and rail spike: 5-10 MN/m
• Track with concrete sleepers with Fist fastening: 15 - 20 MN/m
Thus the lateral stiffness is much lower than the vertical stiffness.
Another stiffness which has to be regarded is the vertical stiffness between wheel and rail
(d. Chapter 8). Even though both wheel and rail are very stiff a flexibility normal to the
contact plane exists. In simulation mode.ls there are two possibilities of treating this
stiffness (d. Section 5.5). Either the stiffness is taken into account with a spring, or a
constraint in normal direction is introduced, i.e. the stiffness is neglected and the normal
displacements of wheel and rail in the contact plane have to be the same.
An overview of track models for track dynamics analysis can be found in [43].

2:22 2:23
 Rail \'ehic!e dynamics

3. Vehicles

3.1 Classification and design principles

Rail vehicles can be divided into different classes, depending on the field of application.
There are vehicles for different types of rail traffic systems. The most common systems
are:
- Railways. Often with different types of traffic (passenger and freight traffic, suburban
and mainline traffic etc.).
- Subways or metros. Suburban rail system fully or partly under ground. Separated from
other rail traffic and from road traffic.
- Tramways. Suburban rail system with lightweight vehicles fully or partly operating
together with the road traffic.
Rail vehicles can be divided into two main classes:
- Tractive stock. The vehicles in this class are powered. Locomotives take no passengers
whereas motor coaches do. The vehicles can be powered electrically or by diesel fuel.
Rolling stock. The vehicles are non-powered and can be divided into coaches and
(freight) wagons. Some coaches contain a driver's cab (push-pull running).
Rail vehicles consist of two main parts:
- Running gear. The running gear consists of wheels, axle and suspension as well as
components connecting these parts. A wheelset normally consists of two wheels on a
common axle. The running gear should support the carbody as well as guide, brake and,
for a tractive vehicle, drive the vehicle.
- Carbody. This vehicle part carries the payload, i.e. passengers or goods, and/or the
traction equipment.
Depending on the type of running gear, there are two vehicle types:
- Rigid-frame vehicles. The running gear only consist of wheelsets and suspension
components. This principle is shown in Figure 3-l.
- Bogie vehicles. The running gear is a so called bogie. A bogie consists of one or more
wheel sets, a framework and suspension components. See the example of Figure 3-2.
Rigid-frame vehicles are the simplest vehicles. They are also inexpensive and light.
However, these vehicles have limited payload capacity (usually only two wheelsets). The
vehicle length is also restricted by the need for an acceptable curving performance.

From a rail vehicle dynamics point of view, these vehicles also give a rather shaky and
uncomfortable ride since the vehicles have one suspension level only. The wheelsets often
go unstably at only about 100 km/h. A horizontally stiffer wheelset-carbody connection
increases the so called critical speed (see Chapter 7), but gives a worser curving
performance.

In summary, rigid-frame vehicles are most appropriate for rather unqualified transports,
for instance rather light-weight freight traffic with speeds up to 100-120 km/h.

3:1
 Vehicles Rail vehicle dynamics

In rail vehicle dynamics the motions of the vehicle and vehicle parts are important
quantities. For instance displacements, velocities and accelerations; both translational and
Carbody
rotational (angular). Usually motions imply relative motions, i.e. motions relative to the
track design geometry and vehicle speed. In Table 3-1 six important relative motions are
defined. These rigid body motions are shown in Figure 3-3 for a carbody.

Relative motions Symbol Notation

iI
Translation in direction of travel x Longitudinal

Translation in transverse direction, y Lateral

parallel to the track plane

Translation perpendicular to the track plane z Vertical, bounce

Figure 3-1 Rigid-frame vehicle with carbody suspended on two wheelsets. Rotation about a longitudinal axis cp Roll, sway

Rotation about a transverse axis, X Pitch

parallel to the track plane
Carbody
Rotation about an axis perpendicular 1f! Yaw
to the track plane
Secondary suspension -;=:=::...2---::1-__ Bogie frame
Primary suspension Wheelset Table 3-1 Definition of relative motions.

-+--.-x

Figure 3-2 Bogie vehicle with two two-axled bogies.

Bogie vehicles have two levels of suspension, cf. Figure 3-2: primary suspension between
wheelset and bogie frame, and secondary suspension between bogie frame and carbody.
Note that the word "suspension" means not only springs but also dampers and other
elements, cf. Sections 3.4 and 3.5.
Bogie vehicles have a better curving performance and the derailment risk is lower than for
rigid-frame vehicles. The carbody vibrations and wheel-rail forces are also reduced since
there are two levels of suspension.
The bogie frame gives a high-frequency vibration isolation through its inertia, and also a
geometric advantage since disturbances acting on one wheel set are, in principle, halved at
the bogie frame longitudinal mid-point (two-axled bogie, which is the most common
type). However, bogies increase the vehicle weight and costs.
High-speed vehicles demand carefully designed and manufactured bogies. The primary
and secondary suspensions must be optimized to cope with various demands, especially if
the vehicles are to run on lines with many (tight) curves and significant track irregularities.
The subject of rail vehicle dynamics, and vehicle-track interaction, is therefore of utmost
importance in such applications.
z
Figure 3-3 Relative rigid body motions of a carbody. Coordinate system X-Y-Z,
motion components according to Table 3-1.
Three different motion conditions can also be defined:
- Static condition. Normally this means the condition of a vehicle at rest on a perfect,
plane and tangent track. This condition is also referred to as nominal.
- Quasistatic condition. The idealized condition when the vehicle runs at constant
speed through a perfect curve with constant radius and cant as well as constant
wheel-rail contact condition. All forces and relative displacements within the vehicle
and between vehicle and track are constant in time.
- Dynamic condition. The condition when the vehicle has additional time dependent
forces and relative motions, relative to the quasistatic or static condition, mainly due
to track irregularities and transition curves. Motions related to vehicle instability are
also included here.

3:3
Vehicles Rail vehicle dynamics

3.2 Running gear, bogies

3.2.1 General
Wheel inside gauge 2b a
As mentioned above the running gear have the following tasks:
- Support the carbody, at vehicle rest and travel.
- Guide the vehicle, especially curve negotiation. r-----------~-Axle'------~
Wheel
Brake the vehicle and, for tractive vehicles, drive the vehicle.
These tasks should be carried out demanding: Flange

- Acceptable track forces and high safety level.

Axle box distance 2b l
- Acceptable ride qualities and sound levels.
- Minimal running gear and track maintenance.
- Acceptable weight and costs.
hf = flange height
In the running gear design, aspects like the following must be considered:
If = flange thickness y = cone angle
- The demands listed above.
- Traffic operation type (locomotive, coach, tram, wagon etc.)
- Speed, on tangent and curved track, and axle load.
- Track design geometry and track irregularities. Figure 3-4 Conventional wheelset with definitions. Measures in mm.

- Type of brake system and traction system.

The following numbers for some of the measures in Figure 3-4 are typical for standard
- Loading gauge etc. track gauge unworn wheelsets, and are partly standardized:
- Relevant standards, national and international. Flange thickness It = 32.5 mm
- Flange height hf = 28 mm
3.2.2 Wheelsets
- Wheel inside gauge 2ba = 1360mm
The fundamental component in rail vehicles is the wheelset, i.e. the pair of wheels rolling - Running tread position bo = 750mm
in parallel on the rails. Normally the two wheels are connected by an axle, see Figure 3-4.
- Axle box distance 2b[ = 2000mm
The wheel part in contact with the rail has a tread and a flange. The flange is located
Due to wheel wear during vehicle operation, the flange thickness will decrease whereas the
towards the wheel inside and comes into contact with the gauge face of the rail. The
flange height will increase. Minimnm and maximnm permitted values, respectively,
nominal rolling circle of the wheel tread is called running tread. The nominal wheel
depend on the wheel diameter [N26].
diameter (radius) refers to this circle.
The wheels have conical treads which means that that the rolling circle radius increases as The wheel radius is usually in the range of ro =0.3-0.6 m. A wheelset like above, with axle
the wheel-rail contact point approaches the flange. For linear conical treads the cone angle boxes but without possible brake discs, weighs about 1000-1500 kg.
is constant ( y = Yo). For special vehicles and conditions modified wheelsets can be used:
The axle connects the two wheels, keeping the wheel inside gauge constant and giving the - The wheels have interior suspension (to reduce unsuspended mass and sound level).
wheels the same rotational speed. The axle has two bearings at the axle boxes, usually
located outside the wheels. The weight of the rest of the vehicle is supported on these boxes - The wheels are free to rotate relatively to each other.
via the (primary) suspension. - The axle-boxes are located between the wheels.
The present wheelset, with virtually rigidly connected wheels and conical treads, has For rigid-frame vehicles the carbody is directly suspended onto the wheelsets, as pointed
steering capabilities affecting the vehicle stability and curving performance, cf. Chapter 7. out in Section 3.1.

3:4 3:5
Vehicles Rail vehicle dynamics

3.2.3 Bogies

For heavy freight traffic and most passenger traffic the running gears are designed as
-- . •I
;
. JiIlL
I -'-
bogies, normally two-axled bogies. Figure 3-5 shows such an example.

This two-axled bogie has rubber+coil springs in the primary suspension and airsprings in
the secondary suspension. Hydraulic dampers are used both in the primary and secondary
I. - v

--
''--

I
-1--.-

suspensions. The bogie also has an anti-roll bar to reduce the carbody roll, especially in
curves. The bogie has wheel mounted disc brakes and is also powered. - - .
Ull
a
. .

a
Ull
.
-'-
1
(a)

--- - --
- (b)
--
Figure 3-6 Bogies with stiff and soft wheelset steering. Bogie wheelset base 2a.
(a) Stiffwheelset steering ("stiff bogie").
(b) Soft, or radial, wheelset steering ("soft bogie").
Airspring, support for carbody 7 Wheelset
2 Bogie frame 8 Primary suspension (rubber+coil springs) Note that a soft bogie still has rather high longitudinal, primary stiffness as compared to the
3 Traction rod 9 Axle-box damper
vertical primary stiffness. As Figure 3-6b indicates, a soft bogie also has a significant
4 Vertical damper 10 = Disc brake
11 = Traction motor
lateral, primary suspension. In practice there are "intermediate" bogies, with horizontal
5 Yaw damper
6 Anti-roll bar 12 = Traction gear primary stiffnesses between the ones for stiff and soft bogies.
The most established variant of the conventional bogies above is the so called Jacob's
Figure 3-5 Bogie example. Source: Bombardier Transportation. bogie. In this case the bogie supports two carbodies. The carbody ends are connected by
joints to each other and to the Jacob's bogie, see Figure 3-7.
There are two (extreme) principles of wheel set steering within the bogie frame:
A smaller number of bogies is thus needed, but the axle loads are increased. Rather
- Stiffwheelset steering. The wheelsets have very limited possibility to yaw relative to the light-weight passenger trains without any needs to increase or decrease the number of
bogie frame. carbodies are most suitable for Jacob's bogies. However, carbodies will usually be shorter
with Jacob's bogies because the carbody ends (outside bogie centres) are taken away.
- Soft (radial) whee/set steering. The wheelsets have significant possibility to yaw
relative to the bogie frame and thereby steer to an almost radial position in curves.
The two principles are illustrated in Figure 3-6. The corresponding bogies are often called
stiff and soft bogie, respectively.

A stiff bogie has almost only vertical primary suspension. The horizontal motions of the
wheelsets relative to the bogie frame therefore are very much restricted. A stiff bogie has
consequently good ride stability properties, and gives a rather high critical speed, but poor
performance in curves.

A soft bogie has rather low longitudinal and lateral primary stiffnesses. This results in a
significant possibility for the wheelsets to yaw relative to the bogie frame and good
curving performance. However, instability may occur on tangent track if the longitudinal
stiffness is too low. Figure 3-7 Principle ofJacob's bogie.

3:6 3:7
Vehicles
3.3 Carbodies
3.3.1 General
The main task of the carbody is to carry the payload, i.e. the passengers or goods, and/or
most of the traction equipment of a motor coach or locomotive. The carbody should also
protect the payload and provide comfort for it.
Often carbodies have to fulfill the following demands:
- Safety for passengers and personnel. Aspects like collisions, shunting impacts, fire,
derailment consequences and air pressure shock waves in tunnels.
Comfort for passengers. Easy entrance/exit access, seating and standing comfort,
lighting, temperature, ride comfort, sound and vibrations and attractive interior design.
_ Comfort for goods. Weather protection, ride comfort, vibrations, no displacement of
goods, simple loading and unloading procedure.
Operating and maintenance aspects. Loading gauge, strength, fatigue, surface
treatment, corrosion resistance, maintenance friendliness and lifting possibilities.
- Costs. Manufacturing and purchase costs as well as life cycle costs (LCC).
Aerodynamic design. Applicable to high-speed vehicles.
- Aesthetic design.
These demands should be fulfilled considering the following aspects:
_ Carbody length. For payload vehicles, at least for passenger coaches and freight
wagons with bulky load, it is in general desirable to have a carbody as long as possible
in order to carry much payload. For locomotives required equipment and driver's cab
should be contained.
_ Carbody cross-section. The carbody cross-section, including lateral and vertical
motions, must not interfere with the prescribed loading gauge. Especially in curves, the
carbody ends and middle must not exceed the gauge's outer and inner part respectively.
For a long carbody, the carbody width may therefore have to be reduced. See Chapter
11.
- Carbody stiffness. A too flexible carbody (low stiffness) often leads to significant
structural motions and vibrations, due to the track irregnlarities and vehicle speed, and
as a consequence poor ride comfort. This may restrict the carbody length and other
design parameters and principles. See Section 3.3.2.
_ Carbody-bogie interface. The carbody and running gear (bogies) must match. The
bogies often determine the lowest possible carbody floor level. For motor coaches the
traction equipment imposes further restrictions.
_ Air resistance. For high -speed vehicles a low air resistance is important. Not only must
the vehicle front be properly designed, the carbody walls and roof as well as the bogies
also cause air resistance. The surfaces should be smooth and the pantograph on the roof
must be given special attention.
3:8
Rail vehicle dynamics
3.3.2 Structural flexibility and ride comfort
The carbody structural flexibility (finite stiffness) and related motions and vibrations often
reduce the ride comfort. This is indicated by the vehicle-track simulation example of
Figure 3-8.
If the structural flexibility is considered in the simulations (bottom graph), the maximum
level and dominating frequency of the vertical acceleration history are increased. The
dominating frequency can be estimated to about 10 Hz, a vibration frequency that
significantly influences the human ride comfort, or rather discomfort (see Chapter 10).
XID
Vertical acceleration on carbody floor, middle position (m/s2)
Rigid body motions plus
structural flexibility motions
4 9 10
--
Time (s)
01_. DUE Il-eJ-al
Figure 3-8 Simulated vertical acceleration on carbody floor, middle position.
Influence of structural flexibility. Swedish train X10 at speed 140 km/h.
As an overall measure of the structural flexibility, the lowest eigenfrequency of the
carbody is often used. The eigenfrequency usually refers to the case ofa free carbody, fully
equipped but without payload and can be either calculated or measured.
Figure 3-9 gives an example of the four lowest eigenfrequencies, and corresponding
eigenmodes, of a carbody. Note that the lowest eigenfrequency, 10.9 Hz, is close to the
dominating excitation frequency mentioned above. Although not for the same carbody,
this implies the importance of the lowest eigenfrequency and the corresponding
"banana-like" eigenmode for vertical bending motion.
The second eigenmode of Figure 3-9 is a torsional motion about a longitudinal axis of the
carbody. The third mode is dominated by a lateral bending motion, whereas the fourth
mode is complicated with significant cross-sectional shear. The higher the
eigenfrequency, the more complicated the eigenmode.
3:9
Vehicles Rail vehicle dynamics

A practical example of the negative influence of structural flexibility on the ride comfort is
shown in Figure 3-10. Here a vertical acceleration was measured on the carbody floor of
10.9 Hz an experimental train at high-speed tests (but with an old carbody). After performing a
frequency analysis of the measured acceleration an extreme peak at about 10 Hz became
obvious.

A closer examination showed that not only did the vertical bending mode appear at this
frequency, but also a local mode caused by slender cross beams and attached heavy
equipment. An additional amplification was therefore obtained giving an unacceptable
acceleration level and poor ride comfort. The dashed line in the figure approximately
represents the human sensitivity to different frequencies of vertical vibration.

~f2~~f~~i~~~-!
f2 = 11.3 Hz

I
Floor measurement point

'Vertical acceleration (m/52)

t Bending motion Unacceptable

h = 11.9Hz
. / of carbody&t1oor
.... -.- ----
~-Acceptable
0.8 .,.,., ..... .,.- ... p -

p"-
.-_#.. -
0.4

f4 == 15.0Hz Figure 3 - 10 Frequency analysis of vertical acceleration on carbody floor.

Swedish experimental train, litera X15-1. The carbody has a vertical
bending mode at about 10Hz.

A simple solution to the structural flexibility problem hardly exists. Unfortunately a stiffer
carbody in general also gives a heavier and more expensive carbody, and restrictions on the
possible window and outer door sizes. The carbody might also be too short or give a
reduced cross-section.

The efforts of making carbodies and rail vehicles lighter - to get reduced wheel-rail forces
and wear as well as lower energy consumption etc. - give a potential risk of a too flexible
carbody.

An additional complication is that the demands on ride comfort, sound and vibration levels
have become more and more rigorous during the last 15 -20 years, in spite of the increasing
vehicle speeds.
Figure 3-9 Four carbody eigenmodes and eigenfrequencies. Example for a
passenger coach. Dashed lines indicate the undeformed carbody. Source:
Bombardier Transportation.

3:10 3:11
 Rail vehicle dynamics
Vehicles

Possible actions to mitigate the comfort effects due to structural flexibility must be based 3.4 Springs etc.
on an understanding ofthe full dynamic system of vehicle, track and their interaction. This
also indicates that a certain action may work in most situations, but sometimes it doesn't 3.4.1 General
help and may even worsen the comfort.
Suspension systems are introduced to isolate the carbody from major vibrations and
Still, the list below suggests actions that should be considered to overcome poor comfort shocks, reduce wheel-rail forces and wear as well as to allow for internal vehicle motions
related to structural flexibility: during curve ~egotiation. These systems mainly consist of springs (Section 3.4) and
_ Increase the lowest carbody eigenfrequency, e.g. with proper stiffeners. dampers (SectIOn 3.5). Four types of springs are shown in Figure 3-12 and described in
more detail in Sections 3.4.2-3.4.5.
_ Increase the carbody structural damping, e.g. distributed damping material and
properly designed carbody joints.
_ Avoid connections of springs and dampers etc. at points where the lowest carbody
eigenmodes are dominant. Rubber spring (chevron)
_ Avoid coupling very stiff elements to the carbody, e.g. unmotivated stiff traction rods Cross-section from above
and possible carbody-to-carbody joints. The result may be large, high-frequent forces.
_ Make sure that the lowest carbody eigenfrequencies do not coincide with bogie
eigenfrequencies, e.g. bogie pitch, and eigenfrequencies of underframe mounted
Airspring
1. Carbody interface
equipment.
2. Bogie interface
The last action is illustrated by Fignre 3 -11. Here the eigenfrequencies are separated by 3. Air bag
increasing the frequencies, starting from the track and ending at the carbody components 4. Additional spring
and seats. 5. Surge pipe
6. Orifice
In practise this separation of frequencies cannot always be carried out. In particular, it is
7. Surge reservoir
difficult to make the seats sufficiently stiff. On the other hand, a full separation of
8. Levelling valve
frequencies is not always necessary. 9. Air container
10. Compressor
Component frame~ E o o r with seats 18 HzI
\ I Coil spring
Floor between side sills 14 Hz I
,. ---___t___---,HzI
First vertical bending mode of carbody 10

t
IBOgiT'' " I Leafspring

Track Figure 3 -12 Four different spring types in rail vehicles.

Figure 3-11 Frequency separation for good vibration isolation.
The bogie frequency refers to pure vertical motion of the bogie frame on
the primary suspension. The other eigenfrequencies also refer to similar
idealized single-degree-of-freedom systems.
For a literature study on structural flexibility models for rail vehicles, see [15].
Active spring systems based on sensors, computer control, hydraulics, pneumatics etc. will
play an Important role in future rail vehicles, but such systems are not discussed here.
Spring loads
The .s?rings are subjected to static, quasistatic and dynamic loads, ct. the defined
co~dltlOns at the end of Section 3.1. The static contributions result from carbody dead
weIght, payload and, for primary springs, bogie frame weight etc. The variation in payload

3:13
3:12
Vehicles Rail vehicle dynamics

must be considered and the maximum payload may be almost as high as the carbody dead
weight, for freight vehicles even 10 times that weight.
Quasistatic contributions result from curve negotiation. Constant acceleration or
retardation also give contributions. The dynamic load contributions mainly come from the
track irregularities.
the friction effects are more pronounced which leads to higher stiffness. Also, given a
certain amplitude, increased frequency in general leads to higher stiffness since viscous
effects start to dominate the hysteresis.
ForceF

Force-displacement relations, stiffness

The characteristics of springs can be described by force-displacement graphs. For most
springs, but normally not airsprings, the static load due to dead weight and payload deflects
the spring. Different static loads (preloads) give different deflections. Displacement x
Quasistatic and dynamic spring forces/displacements are of special interest. Figure 3-13
shows some examples of force-displacement relations for quasistatic loading and
unloading of springs.
Force F ForceF
Progressive Figure 3-14 Definition of dynamic stiffness k,}. Harmonic excitation with force
/ amplitude Fo and displacement amplitude .xo.
Llks
1
An example of static and dynamic stiffness is shown in Figure 3-15. Here we assume that a
vehicle runs on tangent track into a curve so that the lateral suspensions are subjected to
increasing lateral spring displacements and forces due to centrifugal effects. Also dynamic
contributions caused by track irregularities are present. From the figure we can see that the
small and fast dynamic motions activate the friction and viscous mechanisms, giving a
higher stiffness (kt). On the contrary, the slow displacement and force increase in the
transition curve gives a lower stiffness (ks).
Displacement

Displacement x Displacement x Circular curve

(a) (b)

Figure 3 -13 Force-displacement graphs for quasistatic loading and unloading.

(a) Elastic behaviour (linear, progressive, degressive).
(b) Hysteresis. Definition of static stiffness ks·

The graphs in Figure 3-13a illustrate elastic behaviour: linear, degressive and progressive.
These types of characteristics can be found for springs in rail vehicles, even though
perfectly linear or elastic materials hardly exist. In practice a more or less pronounced I
Force
hysteresis appears, see the example of Figure 3-13b. For the present slow loading and i
unloading a friction-like behaviour dominates and energy is dissipated (= the area
enclosed by the graph). A certain amount of damping is obtained in this way.
In Figure 3-13b a (quasi-) static stiffness, ks, is also defined. This stiffness is thus the
tangent stiffness when virtually all spring friction has been released.
A dynamic stiffness, kt, can also be defined, see Figure 3-14. Here a harmonic spring
excitation with a "typical" frequency, say 1 and 5 Hz for secondary and primary springs
respectively, and a "small" force and displacement amplitude. As seen in Figure 3-14 the
dynamic stiffness is the quotient between force amplitude and displacement amplitude, i.e.
Displacement
kt =Fo/xo·
Due to the smaller amplitude and the higher frequency, the dynamic stiffness becomes Figure 3 -15 Principal lateral behaviour of a spring during curve entrance.
higher than the static one. This is because for small amplitudes, given a certain frequency, Static and dynamic stiffness (ks and kt).

3:15
Vehicles Rail vehicle dynamics

To simply relate static and dynamic stiffness to each other, a dynamic stiffness factor can be
defined as

(3-1)

This factor depends on the choice of "typical" frequency and "small" amplitude in the
dynamic case, partly also on how complete the friction release is in the static case. Still,
Table 3-2 intends to give some guidelines on values of the dynamic stiffness factor for
different spring types. We can see that the dynamic stiffness factor may even exceed the
value of2.

Spring type Loading type K (a) (b)

Rubber springs mainly shear 1.21)

mainly compression 1.5

as above, rolling/sliding rubber parts 1.5 - 2.5

Airsprings slow motions in curves etc 1.1 - 1.2

fast motions due to track irregularities 1.3 - 1.4 2)

Coil springs 1.0

1) Span of 1.1-1.4 depending on rubber quality, motion amplitude, temperature etc

2) Theoretically about 1.4, but see Section 3.4.3. (c) (d)

Table 3-2 Dynamic stiffnessfactor Kfor different springs. Figure 3 -16 Some different types of rubber springs.
Numbers to be used with care - may not be valid in all situations. (a) Rubber chevrons, (b) Rolling rubber ring,
(c) Rubber pad on coil spring, (d) "Hour-glass" spring.

3.4.2 Rubber springs

Rubber springs are often used in rail vehicles. Some advantages (+) and disadvantages (-)
of rubber springs are:
+ Many possibilities to choose geometry, insert metal plates and provide voids to give
great freedom in choosing mechanical properties of the spring in different directions.
+ Rubber springs are usually light and do not require much space.
+ Rubber springs are rather inexpensive.
- High dynamic stiffness factor (see Table 3-2).
- High stiffness at low temperature, say below -25°C. ..Qm.lli
I'ow"~ lluoerB
- Significant creep. 1\t!«C 9'!!p/c;

Many types of rubber springs exist. Four types are shown in Figure 3-16. The chevron type
in Figure 3-16a consists of rubber reinforced by metal plates, and gives many possibilities
to choose (primary suspension) stiffnesses in different directions.
The rolling rubber ring spring of Figure 3-16b is also shown in the application of Figure
3-17. Figure 3 -17 Example of rolling rubber ring spring.

3:16 3:17
Vehicles Rail vehicle dynamics

Force-displacement graphs for the rubber ring spring subjected to harmonic excitation provided the inertia forces of the airspring is neglected (which is reasonable in ride
with different displacement amplitudes are shown in Figure 3-18. The high dynamic comfort studies with frequencies below 20 Hz). The absolute pressure is now P = Pg + Palm
stiffness depends, in addition to the internal rubber friction, on the friction (sliding) whereas the air bag air volume is V.
between the rubber body and the metal spring housing. Cf. the high dynamic stiffness The airspring vertical stiffness, kz, is given by
factor in Table 3 - 2.

(3-4)

given that Ae is virtually independent of the vertical deformation.

To determine the stiffness kz more conveniently we introduce the gas equation

P . vY = Po . V;; (3-5)

where y is the so called polytropic exponent. For isothermal and adiabatic conditions the
exponent equals y= 1 respectively y = 1.4.
---------Tffi#77'7'~------- Displacement
The stiffness can now be written as

(3-6)

where

Figure 3-18 Force-displacement graphs/or rolling rubber ring spring. (3-7)

Static and dynamic stiffness (ks and ~).
3.4.3 Airsprings
Airsprings are preferably used as secondary suspension in vehicles with high payload,
relative to the dead weight, and high ride comfort demands. The different components of
an airspring have been shown in Figure 3-12. Study this figure once more.
The airspring is pressurized by the compressor through the air pipe and air container.
Normally the static gauge pressure Pgo = 300-700 kPa, i.e. 3-7 times the atmospheric
pressure (Palm =101 kPa). The static absolute pressure is thus Po =400-800 kPa, given the
relation Po =Pgo + Parm·
By means of the gauge pressure the airspring is able to carry a vertical load, mainly due to
the carbody dead weight and payload (passengers). This static vertical load is normally Fzo
= 50-130 kN. The effective cross-sectional area of the air bag, Aeo, is determined by
F zo (3-2)
Pgo = Aeo
A change in load Fzo often gives a corresponding change in gauge pressure Pgo, since the
effective cross-sectional area Aeo is almost independent of the load for most types of
airsprings. Aeo is often 0.1-0.3 m2 (Pgo =500 kPa and Fzo = 100 kN gives Aeo =0.2 m2).
The air bag is made of rubber and typically contains Va = 15 -40 liters of air.
For dynamic vertical loading Equation (3-2) turns into
(3-3)
is obtained from Equation (3-5) and where
dV =-A (3-8)
dz eo
Equation (3 -6) can also be expressed as
(3-9)
by means of Equations (3-6) and (3-2) as well as the relation Po = Pgo + Palm.
The vertical airspring stiffness is thus almost proportional to the load Fzo since Pgo >>Palm
normally applies. The ratio between spring stiffness and carbody mass, or the natural
frequency of the dynamic system for vertical motion, consequently becomes almost
constant. Thus the vertical suspension properties remain almost unchanged and
independent of the load.
In order to soften the airspring, mainly in the vertical direction, an additional air volume is
often introduced in terms of a surge reservoir, see Figure 3-12. This reservoir has an air
volume of about 30-100 liters and is connected to the air bag with a surge pipe. As seen
from Equations (3-6) and (3-9) the stiffness is decreased for an increased air volume.
However, this assumes that the surge pipe has sufficiently large diameter so that the two
connected air volumes essentially work as one. A softer airspring usually gives an
improved ride comfort, but the carbody lateral (and roll) displacements may be siguificant.
The natural frequency mentioned above now becomes about 1 Hz.

3:18 3:19
Vehicles
The airspring vertical stiffness increases in principle with excitation frequency, mainly
because the heat exchange with the surrounding air decreases with faster motions (y
increases from 1). For frequencies above 0.01-3 Hz, depending on type of airspring, there
is hardly any heat exchange. An adiabatic process is thus obtained and the dynamic
stiffness factor becomes about 1.4 (y = 1.4), see Table 3-2 and Equation (3-5). However,
for airspring systems with surge reservoir this additional volume is essentially
disconnected at higher frequencies, say above 15 Hz, so that the stiffness factor becomes in
the range of3-6. Moreover, if the pipe between the two volumes is long and unconstrained
(no orifice) the stiffness factor may reach 6-15 in the frequency range of 5 -10 Hz at small
displacement amplitudes due to a resonance phenomenon [13]. It is the air mass in the pipe
that is subjected to very high accelerations that causes this resonance. For this case the
vibration isolation is limited.
By means of the levelling valve, the airspring top level can be kept independent of the
vertical load, see Figure 3-12. Thus an automatic level control is achieved so that the
carbody floor level is almost independent of the payload. For suburban vehicles this level
coincides with the platform level.
The airspring usually has an emergency rubber spring, giving some suspension if the
airspring is punctured.
Originally the airspring only provided vertical suspension, but since the last 2-3 decades it
also gives horizontal suspension. During curve negotiation, and for significant lateral track
irregularities, the lateral airspring suspension is very important for the lateral ride comfort.
Some different types of airsprings are shown in Figure 3-19.
Unreslrlcled Air Spring
Unrestricted Air Spring Unre.lrlct.d Air Spring
with rubber-tc>matal bonded spring with rubb.r-to-metal bonded spring
Belted Air Spring Belled AIr Spring
with rubbor.lo-tnelal bonded spring
Figure 3-19 Some different types of airsprings.
3:20
Rail vehicle dynamics
In general airsprings also provide a significant sound and vibration isolation.
3.4.4 Coil springs
Coil springs, see Figure 3-12, can be used both as primary and secondary suspension.
Several springs can be used in parallel. The coil (flexicoil) springs can usually transfer
both vertical and horizontal forces.
Coil springs have, for example, the following advantages (+) and disadvantages ( -):
+ Save space in the horizontal plane.
+ Simple to design and manufacture for required stiffnesses and loads.
+ Static and dynamic stiffnesses are almost identical, cf. Table 3-2.
- Need much space vertically.
- Give significant deflection span for significant load span.
- Cannot in general manage large lateral deformations.
- Virtually no damping.
3.4.5 Leaf springs
Leaf springs are common in freight wagons and older coaches and locomotives, see Figure
3-12. Leaf springs typically consist of 8-10 leaves. The following advantages (+) and
disadvantages (-) can be identified for leaf springs:
+ Simple, inexpensive and robust.
+ Limited space in lateral and vertical direction.
+ The friction betweeb the leaves gives significant damping.
- Needs much space in the longitudinal direction.
- Released friction (sliding) at the leaf surfaces often gives sudden vehicle motions.
- Poor vibration and sound isolation.
3.4.6 Traction rods, anti-roll bars etc.
Traction rods transfer longitudinal forces between running gear (bogie) and carbody to
allow for vehicle braking and acceleration (traction), cf. Figure 3-5. They can be
considered as very stiff springs.
Anti-roll bars connect bogie and carbody to reduce the carbody roll, cf. Figure 3-5. An
anti-roll bar can be seen as a torsional spring.
Bumpstops mainly limit the lateral motion between bogie and carbody. The stops first have
to close a gap before they are activated thmugh a small (conical) rubber component and,
later on, a very stiff metallic part.
3.5 Dampers
Specially assigned dampers are usually needed to limit the vehicle motions - kinetic
energy is transferred to heat. Note however, that some of the spring components in the
previous section also give some damping.
3:21
 Rail vehicle dynamics
Vehicles

Dampers can be introduced in both the primary and secondary suspension and are often 3.6 Carbody tilt
hydraulic dampers, cf. Figures 3-5 and 3-20. These dampers normally use OIl to prov~de
If the vehicle speed in track curves exceeds a certain limit, the lateral acceleration in the
viscous, axial forces. In freight vehicles friction dampers are common, based on shdmg
carbody floor plane will be too high and cause unacceptable ride comfort. Both the
between surfaces. quasistatic and dynamic part of the acceleration will be annoying.
For hydraulic dampers, viscous forces are thus transferred. This means that the force
To improve this situation, but still be able to run rather fast in curves on curvy lines, various
depends on the piston velocity. Figure 3-21 shows examples of such characteristics.
carbody tilt systems are developed. These systems tilt the carbody inwards when
negotiating curves and reduce the lateral acceleration in the carbody floor plane.
The carbody tilt system should only affect (improve) the ride comfort and not safety
aspects like wheel-rail forces and derailment. An active tilt system, demanding control
and power supply etc., may of course suddenly stop working but the safety aspects must
not be influenced by this inconvenience. For instance, the tilt system must not move the
carbody centre of gravity significantly inwards in the curve so as to avoid outward vehicle
turnover.
Tilting trains can increase the curve speed by up to 30-35% as compared to conventional
trains. The possible overspeed percentage is, among other things, restricted by:
: The lateral acceleration endurable for passengers.
- The wheel-rail forces and what the track can withstand.
- The tilt angle the tilt system can produce.
The degree of compensation gives a non -dimensional measure of the tilt system capacity.
This degree is defined as the quotient between the carbody tilt angle, and the angle of the
centrifugal and gravitational acceleration resultant (that is rpci'P, see Chapter 4). When the
degree of compensation equals 1, a full compensation is obtained and the quasistatic,
lateral acceleration in the carbody floor plane equals O.
Figure 3-20 Example of how a hydraulic damper works. Compressive and tensile
force, respectively. Source: Bombardier Transportation. 3.6.1 Passive and active tilt
Force The carbody tilt system can be either passive or active, cf. Figure 3-22:
(kN) ,---..,---,----,--,--r-----"1 I
20+----+----r----r--~----+---~ - Passive tilt. The carbody has a high level tilt centre. The centrifugal force through the
carbody centre of gravity then gives the carbody a tilt motion like a pendulum.
- Active tilt. The carbody tilt is enforced by a special tilt system powered by pneumatics,
I
I hydraulics or electricity.
i In case (a) of Figure 3-22 the tilt system is purely passive.
1O-t------'--+::.
! !
I In case (b) the centrifugal force counteracts the active tilt system, which means that a high
~ I !
I power system is required.
/ i i
In case (c), however, the tilt centre is closp. to the centre of gravity and the required system
Y \. i i
: power is limited. Closely located tilt and gravity centres is therefore desirable.
/1 I : l
In Figure 3-23 the tilted carbody cross-section is related to the loading gauge, both for
o~::::.....-J..----,0J-.l----'~--0-+.2---"'----0->'.3 /
0 passive and active tilt.
0.1 0.2
Piston velocity (m/s) Piston velocity In case (a) of Figure 3 - 23 the carbody needs to be rather narrow in its lower part, the part
where a wide carbody is most important (seat level).
Figure 3-21 Common characteristics for hydraulic dampers. To avoid damper
damage, the force cannot exceed 10 kN in the right graph (blow-off). In case (b), the carbody needs to be rather narrow in its upper part.

3:23
3:22
Vehicles Rail vehicle dynamics

Advantages (+) and disadvantages (-) of active tilt systems:

+ The carbody can be rather wide at the level where this is most important.
+ The carbody tilt can be applied rather fast at curve entrance.
- The active tilt system is more complicated than the passive one. It is also more
expensive to manufacture and maintain. Active tilt system costs can be estimated to
about 3% of the costs for the entire vehicle.
As a result, active tilt systems can provide carbody tilt angles of about 6-7 degrees relative
to the track plane. The degree of compensation is normally 50-70%.

3.6.2 Trains with carbody tilt

(a) Passive (b) Active 1 (c) Active 2
The four most established tilting trains world-wide are shown in Figure 3-24.
o = Carbody tilt centre
~

Figure 3-22
= Carbody centre of gravity _=____ = Tilt direction

Centrifugal force

Examples of passive and active tilt motion.

_ _ Loading gauge _ _
Carbody cross-section

(a) JR 381, Japan (1973) (b) TALOO, Spain (1980)

Passive, about 4 degrees Passive, about 3.5 degrees

Passive tilt Active tilt

o = Carbody tilt centre
~

Figure 3 -23
= Carbody centre of gravity _= ____ = TIlt direction

Centrifugal force

Gauge clearance for passive and active carbody tilt.

Advantages (+) and disadvantages (-) of passive tilt systems:

+ Simplicity, carbody tilt without active system.
The carbody centre of gravity is displaced 0.1-0.2 m outwards in curves; the risk for
vehicle turnover increases.
- The tilt velocity and timing at curve entrance are rather poor.
- The lower part of the carbody must be narrowed which reduces the seating capacity.. (c) ETR 450 "Pendolino", Italy (1988) (d) X 2000, Sweden (1990)
As a result, passive tilt systems can only provide moderate carbody tilt angles, typically Active, about 7 degrees Active, about 6.5 degrees
3-4 degrees, and limited overspeeds. Figure 3-24 Examples of tilting trains. The years indicate the commercial start. The
angles give the maximum carbody tilt angle relative to the track plane.

3:24 3:25
 Vehicles

Table 3-3 also gives data on some newer tilting trains in commercial operation. For more
information and more tilting trains, see for instance [64] [45].

Train Train type Speed Maximum Power Sen- Country Com-

limit tilt* supply sors mersial
(km/h) (degrees) start
JR381 Multiple unit 120 4 Passive - Japan 1973
Talgo Loco hauled 180 3.5 Passive - Spain 1980
ETR 450 Multiple unit 250 8 Hydraulic (b) Italy 1988
X2000 Loco hauled 200 8 Hydraulic (c) Sweden 1990
VT610 Multiple unit 160 8 Hydraulic (b) Germany 1992
JR383 Multiple unit 120 5? Pneumatic (a) Japan 1995
S220 Multiple unit 220 8 Hydraulic (b) Finland 1995
ETR460 Multiple unit 250 8 Hydraulic (b) Italy 1995
ETR470 Multiple unit 200 8 Hydraulic (b) Italien 1996
VT611 Multiple unit 160 8 Electric (b) Germany 1997
VT612 Multiple unit 160 8 Electric (b) Germany 1998
ICE-T Multiple unit 230 8 Hydraulic (b) Germany 1999
BM73 Multiple unit 210 7.5 Hydraulic (b) Norway 1999
Table 3-3 Tilting trains in commercial operation, see also [34][64][45J.
(*) Tilt relative to the bogie.
(a) In-track sensors.
(b) Gyro and accelerometer in front vehicle.
(c) Accelerometer and tilt sensors in front vehicle.

3.6.3 Three important demands on carbody tilt systems

Required tilt angle
The required tilt angle is mainly determined by the occurrence of (tight) curves and the
requested overspeed in these curves. If the tilt angle requirement is moderate, a passive tilt
system might do. It should be kept in mind that the choice of tilt systems also depends on
the choice of bogie and pertinent wheel-rail forces as well as force limit values.

Tilt timing in transition curves

The carbody tilt history must properly match the vehicle position in the curve, especially at
transition curve entrance and exit. The tilt angle should be increased and decreased
smoothly, but not too slowly.

Reliability
The carbody tilt system must be reliable and work at virtually all times. If it stops to work,
the vehicle curve speed must be reduced not to cause any discomfort. In particular reverse
tilt, i.e. tilt outwards in the curve, must be avoided. The entire system must be able to fulfil
these requirements. Reliability is based on proper choices of system components,
redundancy, monitoring etc. For instance different sensors may be ,:, ,ubled and if a vehicle
computer stops to work, computers of adjacent vehicles may take over.

3:26
 Rail vehicle dynamics

4. VehicIe- track interaction

4.1 Fundamentals and definitions

When describing the dynamic behaviour of a railway vehicle, its motions are often divided
into desired large motions, given by the forward motion and the track alignment, and
smaller undesirable displacements caused by track irregularities. For the mathematical
description different coordinate systems are therefore introduced, shown in Figure 4-l.
Besides the inertial fixed coordinate system called "/", a track-following coordinate
system called "T' is defined. The track-following coordinate system follows the track
alignment, i.e. curves and track cant (superelevation), with the vehicle speed. Vehicle
reactions, as displacements, accelerations and forces, are usually expressed in this
coordinate system. The origin of the track-following coordinate system is usually located
below the centre of gravity of the carbody at the top of rail level. A more detailed
explanation of different coordinate systems used in rail vehicle dynamics can be found in
Chapter 6.

track plane
-------. ------- top of rail

z
T

. -----l> s

Figure 4-1 Definitioll of track-following reference system.

According to above, two definitions are introduced to describe the behaviour of a vehicle
on a track:
Quasistatic behaviour: When a vehicle runs with constant speed on an ideal track (no
irregularities) with a .:onstant curve radius, cant and wheel-rail friction.
In this case all forces and relative displacements in the vehicle and
between vehicle and track are constant with respect to time.
Dynamic behaviour: All additional forces and displacements, mainly due to track
irregularities and self induced vehicle motions. Dynamic forces and
displacements have to be added to the quasi static values.
In the following the basic definitions regarding the quasistatic behaviour of a vehicle in a
curve are introduced.

4:1
 Rail vehicle dynamics
Vehicle-track interaction

For v2/R < 0.3g and ip, < 0.12 rad (= 7°), which almost always is valid, Equations (4-1a)
4.2 Track plane acceleration, cant deficiency etc.
and (4-1b) can be approximated with:

4.2.1 Track plane acceleration v2 . 2 h

ay = R -g' smip, = ~
R
-g._'
2bo
(4-2a)
In case of quasi static curving the vehicle is exposed to two accelerations: the horizontal
centrifugal acceleration which is equal to v 2 /R, and the vertical gravitational acceleration
g. This is illustrated in Figure 4-2a. The resulting acceleration vector can alternatively be az = g (4-2b)
divided into two accelerations parallell and perpendicular to the track plane. The lateral
acceleration parallel to the track plane, ay, is often called track plane acceleration. The with an error for lly and ilz less than 0.003g respectively 0.03g.
acceleration ilz is normal to the track plane, Figure 4-2b.
The side force allgle cP in Figure 4-2(b) can be related to the accelerations fl.. and ilz as
follows ··Y

cP = arctan aay (4-3)

z

It can be .seen that track plane acceleration and side force angle depend on the speed v
curve radllls R and cant h,. '

4.2.2 Equilibrium cant and equilibrium speed

The cant for zero track plan~ ~c~eleration, lly =0, at given curve radius and speed is called
eqUlhbnum cant. Ifthe eqUlhbnum cant is denoted as heq , we get from Equation (4-2a)

.--
~
2b o . ~

~'-
heq = (4-4)
g R
~. . .
~
'PI
...... track plane
. . ... ' . ,
track plane
The equations ab.ove are based?n SI-units. Sometimes for practical use, speed is given in
horizontal plane [km/h] and cant m [mm]. In thIS case Equation (4-4) can be expressed as
horizontal plane
(a) (b)

Figure 4-2 Definition oftrack plane acceleration tly and side force angle CPo The two
subfigures are equivalent to each other. (4-5)
(a) Horizontal and vertical acceleration.
(b) Accelerations in track following coordinate system.
where the di~erent un.its ~re indicated. ~ubstituting even the values 2bo for standard gauge
and the nom mal gravitatIOnal acceleratIOn g, the equation is simplified further to
Regarding railways, accelerations and forces perpendicular and parallell relative to the
track plane are often called vertical/lateral accelerations or forces. In this chapter the
lateral acceleration lly however is called track plane acceleration. Accelerations lly and ilz (4-6)
are given by

In Table 4-1 values for equilibrium cant based on Equation (4-6) are given.
(4-1a)
For. a given cant also an equilibrium or balanced speed, veq , can be defined. We assume
agam zero track plane acceleration, lly =0, and get from Equation (4-2a)
v 2 . (4-1b)
az = R' smipt +g . cosipt

where hi is the cant and 2bo the measuring base for the definition of cant (cf Chapter 2).
Veq = R ·g·h,
2b o
(4-7)

4:3
 Rail vehicle dynamics
Vehicle-track interaction
4.2.3 Cant deficiency and cant excess

For several reasons, fully compensated lateral (track plane) acceleration, ay = 0, can not be
speed [km/h] Radiu achieved in all cases:
Radius 160 170 180 190 200 [nil
40 50 60 7080 90 100 1\0 120 130 140 150
[m] equillbrlUm cant [mm1 - A train sometimes has to stops in a curve, which makes it necessary to limit the
150 maximum cant. Too much cant might in this case cause displacement of goods in freight
150 126 197 175
175 108 169 243 200
trains or discomfort in passenger trains. (In the worst case a vehicle might turn over.) A
200 94 148 212 225 typical limit cant is hl,lim = 150 mm, although some railways dedicated for passenger
225 84 131 189 250 traffic allow up to 180 or 200 mm. For a curve with R = 1000 m, for example, the
250 76 118 170 231 275 equilibrium speed would be as low as 113 km/h for 150 mm cant. Therefore it is
275 69 107 154 2\0 300
300 63 98 142 193 325
desirable to allow a cant deficiency, i.e. a certain amount of uncompensated lateral
.325 58 91 131 178 232 350 acceleration ay that remains in the track plane.
350 54 84 121 165 216 400
400 47 74 106 145 189 239 450 - Not all trains run at the same speed. Therefore it would not be possible to achieve fully
450 42 66 94 128 168 212 ·500 compensated lateral acceleration for all trains anyway.
500 38 59 85 116 151 191 236 550
550 34 54 77 105 137 174 215 600 Cant deficiency is the difference between equilibrium cant and real cant
600 31 49 71 96 126 159 197 238 650
220
650 29 45 65 89 116 147 182 700
700 27 42 61 83 108 137 169 204 243 750 (4-8)
227
750 25 39 57 77 101 127 157 190 800
212 249
800 24 37 53 72 94 119 148 178 850
200 Z35
850 22 35 50 68 89 112 139 168 900 With Equation (4-4) substituted into (4-8) we get
222
900 21 33 47 64 84 106 131 159 189 950
950 20 31 45 61 79 101 124 ISO 179 210 243 1000
143 170 199 231
19 30 42 58 76 96 118
1000
1050 18 28 40 55 n 91 112 116 162 190 220
1050
1100
h
d
= 2b o
g
• .i.
R
- h
/ (4-9)
IJO' 154 181 210 241
1100 17 27 39 53 69 87 107 1150
124 148 173 201 231
1150 16 26 37 50 66 83 103 1200
98 119 142 166 193 221
1200 16. 25 35 48 63 80 1300 Cant deficiency can also be related to the track plane acceleration if we solve Equation
131 153 178 204 232
1300 IS 23 33 44 58 74 91 110 190 216 243 1400 (4-2a) for v2/R and substitute the expression into Equation (4-9). It can then be written as
1400 13 21 30 41 54 68 84 102 121 142 165 1500
95 113 133 154 177 201 227
1500 13 20 28 39 50 64 79 239 1600
74 89 106 125 145 166 189 213
1600 12 18 27 36 47 60 1700
1t7 136 156 178 201 225 hd 2bo
1700 II 17 2S 34 44 56 69 84 100 ZIZ Z37 1800 = g' (4-10)
79 94 111 128 148 168 189 ay
1800 10 16 24 32 42 53 66 2000
38 48 59 71 85 100 116 J3J lSI 171 191 213 236
2000 9 IS 21 29 215 2200
91 105 121 IJ7 155 174 194
2200 9 13 19 26 34 43 S4 65 77 170 189 2500
23 30 38 47 57 68 80 93 106 121 136 153 where ay > 0 is assumed so that hd > 0 (for example ay > 0 in Figure 4-2).
2500 8 12 17 -114 127 142 IS7 3000
48 57 66 77 88 101
3000 6 10 14 .19 25 J2 39 122 135 3500
17 22 27 34 41 49 57 66 76 86 97 109 In Table 4-2 some examples of the relationship between track plane acceleration, side
3500 5 8 12 106 118 4000
30 36 42 SO 58 66 76 85 96
5 7 11 14 19 24 4500 force angle and cant deficiency are given.
4000
21 26 32 38 44 51 59 67 76 85 95 105
4 7 9 13 17 5000
4500
19 24 29 34 40 46 53
60 68 76 85 94
5000 4 6 8 12 IS 71 79 6000
10 13 16 20 24 28 33 39 44 50 57 64 Track plane acceleration Side force angle Cant deficiency
6000 3 5 7 61 67 7000
17 20 24 29 33 38 43 49 55
7000 3 4 6 8 II 14 48 53 59 8000 ay (m/s2) pC) hd (m)
7 9 12 15 18 21 25 29 33 38 43
2 4 5 10000
8000 23 27 30 34 38 43 47
10000 2 J 4 6 8 10 12 14 17 20 35 39 12000 0.654 3.81 0.100
6 8 10 12 14 17 19 22 25 28 32
12000 2 2 4 5 31 15000
3 4 5 6 8 10 II 13 IS 18 20 23 25 28 ZOOOO 0.981 5.71 0.150
15000 1 2 2124
. 2 3 4 5 6 7 8 10 .12 13 IS 17 19
20000 1 1 180 190 200 1.000 5.82 0.153
110 ·)20 130 140 150 160 170
40 50 60 70 80 90 100
speed kmIh] 1.031 6.00 0.158

Table 4-2 Some examples of the relationship between track plane acceleration,
Table 4-1 Equilibrium cant for standard gauge (1435 mm; 2bo =1500 mm) for side force angle and cant deficiency.
different speeds and curve radii. Equations (4-3) and (4-10) have been used and 2bo =1.500 m (standard
gauge) and az = g = 9.81 m/s2.

4:5
 Rail vehicle dynamics
Vehicle-track interaction

A vehicle at the speed v = 50 m/s (180 km/h) is running in a circular curve For conventional vehicles with older runing gear, called category A, according to
Example: Banverket BVF 586.41 [N3] the following is permitted:
withR =1600 m and hI =0.150 m. With Equation (4-8) above we get the cant
deficiency: hd = 0.239 - 0.150 = 0.089 m or the same result, hd = Gy,lim = 0.65 m/s2, equivalent to hd,lim =0.100 m.
1.500/9.81·0.58 = 0.089 m according to (4-10).
For some vehicles with improved running gear, category B,
If hd would become negative, i.e. the real cant is greater than the equilibriu.m ca~t, we relate
to cant excess. A new variable he is then introduced for this excess and IS defmed as Gylim = 0.98 m/s2, equivalent to hd,lim = 0.150 m is permitted.
Finally for SIs tilting trains X 2000, category S, much higher track plane acceleration is
(4-11)
permitted:

so that also cant excess has positive values. Corresponding to Equations (4-9) and (4-10)
Gy,lim =1.60 m/s2, equivalent to hb,lim =0.245 m.
for cant deficiency cant excess can be calculated to This is possible because of a new bogie design which reduces the lateral track shift forces
in curves and the tilt mechanism which keeps the passenger comfort on the same level as
(4-12) for other trains.
On tracks with older spike fastenings the permissible cant deficiency is reduced to hd,lim =
0.080 m for category A and hd,lim =0.100 m for categories Band S.
respectively
, For a given permissible track plane acceleration or cant deficiency the permissible speed
(4-13) . for a given curve radius can be calculated. With Equation (4-2a) we find

Vlim == (4-14)
Thus: he = -hd·
Example: A freight train at a speed v =25 m/s (90 km/h) is running in a circular curve
with R = 1600 m and hi =0.150 m. According to Ekvation (4-2a) we get the where Gy,lim can be related to h d•1im according to Equation (4-10) resulting in
. 25 2 9.81 . 0.150 _ 059 / 2
track plane acceleratIOn ay = 1600 - 1.500 - - . m s
(on standard gauge). From Equation (4-13) a cant excess (4-15)
he =- 1.500 . (- 0 59)
9.81 .
= 0.090 m is calculated.
where the speed is expressed in km/h. This equation is very easy to use.
4.2.4 Permissible track plane acceleration, cant deficiency and cant excess
Cant excess
Cant deficiency Also limit values for cant excess are given in [N9]. The most restrictive limit for cant
The two primary reasons for limiting the permissible track plane acceleration, excess in Sweden is :
alternatively cant deficiency, are Gy,lim = -0.46 m/s2, equivalent to he,lim =0.070 m
_ risk for lateral track displacement, This limit value, however, is not for vehicles at stand still but for ~ehicles at the slowest
_ unacceptable passenger comfort. regular speed of freight trains, normally 90 km/h. (For a vehicle at stand still he =hI which
in turn would limit hI to 0.070 ill).
Also the amonnt of cant excess is limited by the railways. The reason is the risk of goods
displacement in slow freight trains. Besides that, deterioration of the inner rail might In curves with R > 1000 ill and in the diverging track in switches and v > 50 km/h he,lim =
increase if freight trains with high rates of cant excess run slowly through curves. 0.100 ill can be permitted. In the diverging track of switches and v,;; 50 km/h even he,lim
0.130 m is permitted.
In the CEN standard ENV13803-1 [N9] recommended and limit values for ca~t
deficiency are given. For conventional passenger trains .without tilting. under cer~am
conditions 180 mm cant deficiency is allowed. Values for mteroperable high speed hnes
are given in the so called Technical Specifications for Interoperability (TSI) [NIl].
In Sweden like in many other countries permissible values depend on the type of vehicle.

4:7
4:6
Vehicle-track interaction
Rail vehicle dynamics

4.2.5 Jerk, rate of cant and rate of cant deficiency

Lateral jerk is the time derivative of lateral acceleration. Lateral jerk is equal to zero for (4-22)
quasistatic curving. In transition curves, however, the lateral jerk is different from zero,
because the curve radius and cant are changing. It is a matter of ride comfort to limit the Permi~sible values for rate of cant deficiency, track plane jerk and rate of cant can also be
lateral jerk arising in transition curves. For a transition curve with linear change of found III the CEN standard [N9].
curvature and a corresponding (same start point and length) linear (straight)
superelevation ramp, the track plane jerk can be written as 4.2.6 Permissible jerk etc.

As can be ~een in Equations (4-18) to (4-22) the length of the transition curve and the
(4-16)
super~~evatlo~ ramp, Lt. play an important roll regarding limits for jerk and the different
velocities defmed above. This is true especially for high speed vehicles.
where Equations (4-2a), (2-2) and (2-3) are used. The ramp starts at s = 0 and ends at s =
L/. If v still is constant only the track coordinate s depends on time. With Accor~ing to ~anverket BVF 586.41 [N3] the permissible vehicle speed on a given track is
determmed With the following three criteria:
.!L = ds i!:.. = vi!:.. (4-17) Lim
dt dt ds ds 2fij"- ;::: 0.4
l,mm (4-23)
Equation (4-16) can be simplified to

ay = v3(L~J - g2~ov (i';)

v. < 1000 . LI,m
= Lay,o (4-18) lun,km/h - qa • Llh"mm (4-24)

where the track plane jerk becomes a constant in the transition curve.lIy.o is the track plane 1000· Lim
acceleration in the circular curve. V < '
lim,km/h - qb • Llhd,mm (4-25)
Example: A vehicle at a speed of v = 40 mls (144 km/h) is running in a linear transition
curve and superelevation ramp with a lengthL/ =150 m. For Ro =1000 m and wh~r~ qa and % are constant and All, och A41 are the difference of cant and of cant
h"o = 0.150 m according to Equation (4-18) we get the jerk defl.clency bem:een beginning and end of the ramp, i.e. h"o and hd,o for a ramp connecting
a y
= 40 3/(150 . 1000)-9.81 ' 40 . 0.150/(1.500 . 150) = 0.43- stral~ht .track Wlth~ut cant and a circular curve with constant cant h"o . Vlim is the maximum
permissible speed m the actual curve.
0.26 = 0.17 m/s 3 (standard gauge).
(We arrive at the same result with: The constants % and qb shall have the following values:
ay,o = 40 2/1000--9.81 . 0.150/1.500 = 0.62 m/s2; and again Equation Vehicle category A: qa = 6 % =6
(4-18) gives ay = 40/(150 . 1000) . 0.62 =0.17 m/s3). Vehicle category B: qa =5
Also derivatives of the other quantities introduced above are defined, rate of cant, rate of Vehicle category S: qa =4
cant deficiency and rate of cant excess. Still assuming linear change of curvature and linear
ramps we get in a similar way as in Equation (4-18) for the rate of cant If Equa!ion (4-24) is converted to SI -units and inserted into Equations (4-19) and (4-22)
for vehicles of category A with v =Vlim
. v
h ... L-h,p (4-19)
h. < h,,0
I I 1
and I - 6 . h"o . 3.6· = 21.6 = 0.046 m/s (4-26)

hd-r;
- v h d,o (4-20) respectively

respectively . h 1
cP I = 2;0 s 1.5 . 21.6 = 0.031 rad/s = 1.8 o/s (4-27)
. v (4-21)
h ... L-he,o
e I
For vehicles of ca~egory B 20% higher speeds are permitted (6/5=1.2). For trains of
for rate of cant deficiency and rate of cant excess. Even the time derivative of cant angle category S, 50 % higher speds are permitted (6/4=1.5).
might be of interest, especially in combination with tilting trains. The expression can be If Equation (4-25) is converted to SI -units and inserted into Equation (4-20) for vehicles
derived from Equation (4-19) . of category A with v = Vlim

4:9
Vehicle-track interaction Rail vehicle dynamics

4.3 Vehicle roll and sway, body plane acceleration

. V lim • hd,o = _1_ = 0.046 mls (4-28)
hd:S 6. hd,o . 3.6 . vlim 21.6
Up to now it was assumed that the carbody floor always is parallel to the track plane. In
Also here 20% higher speeds are permited for vehicles of categOlY B. For trains category reality of COurse this is not true. Due to the suspension of the vehicle the carbody can sway,
which influences
S, 71% (6/3.5=1.71) higher speed is permitted, i.e. 0.079 m/s.
- ride comfort in the carbody: the roll angle of the carbody increases the experienced
In a similar way we get (category A)
lateral acceleration in curves and on track irregularities with long wavelengths.

a =.-L h < 9.81 _1_ = 0.30 m/s 3 (4-29) - carbody motions relative to fixed installations beside or above the track. Especially
Y 2b o b - 1.50021.6 pantograph motions relative to the catenary have to be regarded.

for jerk. For category Band S we thus get 0.36 m/s3 and 0.52 m/s3 respectively. . - relative displacements between parts of the vehicle, for example between carbody and
bogie.
Example: Do Banverkets rules admit (category A) that the vehicle in the example III
Section 4.2.5 is allowed to run at the desired speed v = 144 km/h ? - distribution of wheel forces between outer and inner rail in curves.

Below the so-called coefficient offlexibility, which describes the vehicle sway, is defined.
L I,m 2: 0.4 . Llhl,mm = 0.4 . 150 = 60 m Thereafter the influence of carbody sway on lateral accelerations in the carbody is
discussed. The problem of vehicle gauging is treated in Chapter 12.
1000 . L I,m _ 1000 . 150 = 167 kmlh
Vlim,kmjh :s qa . Llhl,mm - 6· 150
4.3.1 Coefficient of flexibility

< 1000· Ll,m = 1000 . 150 = 263 km/h The coefficient of flexibility is defined for a stationary vehicle (v = 0) on canted track
Vlim,kmjh - qb . Llhd,mm 6 . 95 without (activated) carbody tilt. Figure 4-3 shows the equilibrium position ofthe carbody
and the important angles. The coefficient offlexibility C'P is defined by the ratio
95 mm cant deficiency are equivalent to 0.62 m/s2 track plane acceleration. All three
criteria are met. The vehicle is allowed to run at v = 144 km/h. C - Cf!c,v=o
'P - cp;- (4-32)
In Sweden the length of superelevation ramps and. transition curves .is relatively. high
compared to many other countries. This makes it easier to Illcrease vehicle speed Without
changing the track geometry. i.e. the relation between carbody sway angle (in relation to track) and track cant angle.

Especially for tilting trains it is important to have long transition curves in order to have
enough time to tilt the carbody.
~
For new or upgraded lines according to Banverkets BVH 586.40 [N6] the ramp length I /i 'Pc
shall be calculated with h. = 0.046 m/s. With Equation (4-19) we get [46] 'Y
~-:--_
l,hm

'Pc

Lim
hl,lim
= -.-Vdim
= hl,lim (4-30)
roll centre
I 'j' . . . . . t-
, '
'hl,lim \Um
g
and rounded

(4-31)

Figure 4-3 Carbody in equilibrium position for a vehicle in a curve with cant.

4:11
Vehicle-track interaction
Rail vehicle dynamics

Typical values for the coefficient of flexibility for different modern vehicles are 4.3.2 Lateral acceleration in vehicle body plane
- Locomotive with pantograph on roof Crp = 0.15-0.25 Whenever the track plane acceleration in a curve is different from zero, the carbody will
- Multiple unit with pantograph on roof Crp =0.15-0.25 lean on its suspension and thus form an angle ({Jc between track plane and carbody plane as
described in the section above. This implies that the lateral acceleration in the carbody
- Multiple unit without pantograph on roof Crp = 0.2-0.4
plane lIye differs from the track plane acceleration, see Figure 4-5b in comparison to Figure
- Passenger coach Crp = 0.2-0.4 4-2b.

- Non -tilted power unit for tilting trains Crp =ca 0.1
- Freight wagons Crp = 0.05-0.1 !-;t ( carbody

I.
In a vehicle negotiating a curve with a non-compensated track plane acceleration, the .j
~~~
carbody normally is rolling towards the outer side of the curve like in Figure 4-4a and
4-4c. For zero track plane acceleration the carbody is not rolling at all relative to the track
plane, see Figure 4-4b.
j

/ rpc = 0
m
rl
2 .

~ ~
.---.
:I /
~....
track plane . '. . track plane
' P , ' . '.. .
~.
(a) (b)
Figure 4-5 Definition of track plane acceleration /lye. The two subfigures are
equivalent to each other.
(a) Horizontal and vertical acceleration relative to ground plane.
(b) Accelerations in the carbody reference system.
(a) (b) (c)
Similar to Equations (4-1a,b), the accelerations lIyc and a.c are given by
Figure 4-4 Rolling of carbody for quasistatic curving. Centrifugal force mv 2/R,
gravitational force mg, roll angle ({Jc and cant angle ({Jt.
(a) No cant leads to rolling towards the outer side of the curve. (4-33a)
(b) Full compensation of track plane acceleration, lIy = 0, gives no
rolling at all.
(c) Cant is not sufficient for compensating the track plane (4-33b)
acceleration. The carbody rolls towards the outer side of the
curve, as in (a). Note that the angle CPc relative to the track plane is negative in Figure 4-5, i.e. in case of
cant deficiency and lack of tilt mechlL.".ism. For v2/R < 0.3 g and 0 :0; CPt + ({Jc < 0.2 rad
("" 12°) Equation (4-33) can be approximated with:
In vehicles with primary and secondary suspension often both contribut~ equaly to Crp ,i.~.
the bogie sway is about Crp/2. As can be seen in Figure 4-3 the carbod~ IS dls?laced also III
lateral direction. The combined lateral and sway motion can be descnbed With help of the (4-34a)
position of the roll centre.

For moderate angles and non-linearities the roll coefficient and the roll centre are more or (4-34b)
less inependent of the track cant. with an error in aye and a.c of less than 0.004g respectively 0.04g.

4:12
4:13
 Vehicle-track interaction
Rail vehicle dynamics

The ratio between carbody plane and track plane acceleration is of particular interest 4.3.3 Calculation of vehicle sway

a yc ]; - g(cp/ + fPc) Vehicle and suspension are assumed to be symmetric with respect to the centre line.
1 - (4-35)
a; = ~ - gcpt v2
If - gCPI

where Equations (4-2a) and (4-34a) have been used and the linearisation sin cP = cp. For a
stationary vehicle we get from Equation (4-35)

(4-36)

where Cop is the coefficient of flexibility which has been defined in Section 4.3.1. It can be
seen that

(4-37)

For small angles and linear vehicle properties this can be extended to the quasistatic case
Rzgure4-6 Quantities for calculation of sway.
(4-38)
The following quantities are introduced:

Substituting (4-38) into (4-35) we get the important relation m mass for suspendled part of vehicle
g gravitational acceleration
ay track plane acceleration
(4-39) /1cp' = /1cp = sway angle for the suspellded mass
b" HI: see figure above
This means that the track plane acceleration has to be multiplied by the factor 1+Cop to get k.pc =spring rate for anti roll bar
the lateral acceleration in the carbody plane. The greater Cop the greater ay,c' In modern
vehicles usually torsional rods (anti roll bars) are installed in a lateral direction under the
carbody to limit the vehicle sway.
A tilt system can be interpreted as a "negative coefficient of flexibility" in Equation (4-39)
(CPc becomes positive in Figure 4-5), and hence the carbody plane acceleration becomes
smaller than the track plane acceleration. Tilting can also be interpreted as increased cant,
(cpt + CPc > CPt), to be able to increase speed without worsening passenger comfort.

Figure 4-7 Forces, moments and deformations in a spring.

Springs 1 and 2 are described in the yz-plane with the matrix equation

where kyrp = krpy < 0 (4-40)

(4-41)
For springs which are symmetric around the y-axis and loaded with the vertical force F
the off -diagonal stiffness can be written as Z>

4:14
4:15
 Rail vehicle dynamics
Vehicle-track interaction

Since kyrp < 0, we can state that

(4-42)

kycp increases the sway angle Llcp' }

We get the following equations: . . => destabilizin
mcreases lateral dIsplacement LI y' g
(4-43)
FyI = kyydYI + kyrpdcp'
* HJ and mg have the same influence
(4-44) * kq,c (roll stabilizer) is stabilizing
McpI = kyrpdYI + kcprpdcp'
* bs is strongly stabilizing
(4-45)
LlYI = LlYa - HI . Llcp'
Note: For vehicles with primary and secondary suspension the calculation becomes more
(4-46) complicated. After reduction an equation system with four unknowns remains;
LlY2 = LlYI = LlYa - HI . Llcp'
sway angles and lateral displacements for bogie respectively carbody.
Llzl = za(mg) + bs . Llcp' (4-47)
Note: DIC 505 includes equations for calculation of sway. The resnlts however are often
between 20-30% too low!
LlZ2 = za(mg) - bs . Llcp' (4-48)

McpI = Mcp2 = kcpydYI + kcprpdcp' = kyrpdYI + kcprpdcp' (4-49)

(4-50)
Fy2 = FyI = kyyLIYI + kyrpdcp'
FzI = F za + kzz . bs . Llcp' (4-51)

Fz2 = F za - kzz . bs . Llcp' (4-52)

Equilibrium:
(4-53)
may - FyI - Fy2 = 0

may' HI + mg' HI . Llcp' - FzI . bs +

+ Fz2 . b s - McpI - Mcp2 - kcpc . Llcp' = 0 (4-54)

Equations (4-53) and (4-50) give

2kyydYI + 2kycp . Llcp' = may (4-55)

Equations (4-54), (4-51) and (4-52) give

2kyrpdYI + (2b;k zz + kcpc-mgHI) . Llcp' = mayfil (4-56)

In the last equation it is assumed that for normal springs 2kcpcp < < 2b; kzz·
With Equations (4-55) and (4-56) we can write
H _ kycp
I kyy
Llcp' = may--------------~------~k2~ycp-
(4-57)
2b; kzz + kcpc mgHI 2-
ky-y

~,[ k~j
kycp
HI -
kyy (4-58)
Lly' = - 1 - 2kycp
2kyy
2b; kzz + kcpc mgHI - 2-
kyy

4:17
4:16
 Rail vehicle dynamics

5. Dynamic analysis techniques

Even though dynamic analyses of railway vehicles today usually are performed by means
of computer simulations, all calculations are based on mathematical models of vehicles.
Modelling of carbody, bogie, wheelset and some suspension elements has been discussed
in Chapter 3, modelling of track irregularities and track flexibility has been discussed in
Chapter 2. Geometrical and mechanical properties of the wheel-rail contact are described
in the following chapters.

In this chapter simple models describing the vertical dynamics of a typical railway vehicle
are introduced and the most common types of dynamic analyses are discussed with help of
these models. Simple models are also very useful to check simulation results. It can be
difficult to decide whether results of simulations with a model with 50 or 100 degrees of
freedom are reasonable or not. More detailed descriptions of simulation of railway vehicle
dynamics can be found in [23), [44) and [62]. See also Chapter 14.

5.1 Simple models for analysis

Defining a model it is important to think about the phenomena which shall be studied. In
studying the dynamic behaviour of railway vehicles, it is often sufficient to represent
reality in a frequency range between 0 and 150 Hz. Studying only carbody dynamics
normally frequencies up to 20 Hz have to be considered. Higher frequencies are filtered
out in the primary and secondary suspension.

5.1.1 One-dimensional model with 1 nOF

The most simple model to investigate the vertical dynamics of a railway vehicle is shown
in Figure 5-1. The model is one-dimensional. It only consists of an unsuspended
(unsprung) mass and a suspended (sprung) mass.

p<"n,
C(
-----L-
Figure 5-1 One-dimensional model with one DOF (z). Masses m and m~, viscolls
damping c, stifness k, speed v, displacements z(t) and zw(t) and track
irregularities zr(s). Damping force Fo dynamic spring force Fled and
dynamic contact force Qd.

5:1
 Rail vehicle dynamics
Dynamic analysis techniques

With the assumption of linear characteristics for spring and damper we get
The model is associated with the following assumptions:
_ The vehicle is running on tangent track (without superelevation or gradient). (5-5a)
_ The vehicle is symmetric with respect to a longitudinal vertical plane.
_ Only vertical track irregularities act on the vehicle. (5-5b)

_ Wheel and rail are totally stiff (rigid).

_ Springs and dampers have linear characteristics. Substituting Equation (5-5) into Equation (5-4) yields
_ The length of the vehicle is much shorter than the dominating wavelength in the
irregularities. This means that pitch motion can be neglected. mz + c(z - zw) + k(z - zJ = 0 (5-6a)

_ The vehicle runs at constant speed.

_ Wheel and rail do not loose contact.
(5-6b)

_ The vehicle has perfectly circular wheels.

Finally by substituting Eq. (5-2a,b) into Eq. (5-6a) we get the equation of motion for the
The assumption that wheel and rail do not loose contact with each other results in that the
vertical displacement of the wheel set is the same as for the track irregularity. Therefore suspended mass
ZW(t) is not an independent degree of freedom, thus the model consists of only one degree
of freedom. The position set) of the vehicle at each time I can be described as mz + cz + kz = cz w + kzw = cz; v + kz, (5-7a)

(5-1) while substituting f21 from Eq. (5-6b) into Equation (5-3b) gives the contact force
s(l) = So + vi
The displacement, velocity and acceleration of the wheel can be expressed by Q = Qo + Qd = (m + mw)g - mz - mwzw =

Zw = Z,[S(t)]
(5-2a) (m + mw)g - m z- mw-i,' v 2 (5-7b)

In conclusion the model consists of one DOF, and thus one equation of motion. The
equation of motion is an ordinary second order differential equation. The right hand side
can be interpreted as force representing the track excitation. The Q- force which often is of
interest (cf. Chapter 9) can be calculated with help of Equation (5-7b) if z(t) is known. As
Z = d22Zt[s(t)] = liz = liCit v) = it' v2 (5-2c) can be seen in (5-Th) the dynamic part of Q is directly related to the inertia forces of the
dl dt dt
w W
two masses. For a well suspended mass the acceleration z is much lower than the
As well the spring force as the contact force Q have a static and a dynamic part: acceleration zwaf the unsuspended wheelset. Even though m is significantly higher than
l1lw the wheelset mass usually is responsible for the largest part of the inertia force and in
turn the largest part of the dynamic contact force.

(5-3b)
For simplicity in Figure 5-1 only the dynamic parts are taken into acount. The
displacement of the suspended mass is therefore only the dynamic displacement relative to
the static equilibrium. Accordingly the equilibrium equations write as
Therefore it is important to keep the unsuspended mass as low as possible, especially for
high-speed train. From Equation (5-Th) it can be seen as well that the acceleration Zw is
equal to v 2 divided by the curvature of the track irregularity z~' . This can be compared
with the centrifugal aceleration in a curve v 2/R (cf. Chapter 4).

(5-4a)
mz = -Fk,d-Fc

(5-4b)

5:3
 Rail vehicle dynamics
Dynamic analysis techniques

and an expression for the dynamic wheel-rail contact force

5.1.2 One-dimensional model with 2 DOF
Almost all railway vehicles consist of carbody, bogie frames and wheelsets. Therefore an (5-11)
improvement of the model above is to introduce as second suspended mass, i.e. to
introduce a second vertical degree of freedom. Such a model is shown in Figure 5-2. The total contact force can be calculated to

(5-12)

5.1.3 Two-dimensional model with two DOF

~ow w.e go back to. a ~odel with only one suspension level but include the length
dimenSIOn of the vehicle mstead. The model then looks like in Figure 5-3. This model
sh.ould be able to represent the vertical dynamics for a two-axle freight wagon or a bogie
with two wheelsets but without carbody.
==> v

s
zc(s)
Figure 5-2 One-dimensional model with two degrees of freedom (Zc, 7<1). Carbody
mass me> bogieframe mass mb and wheelset mass mw- Primary suspension
with stiffness k1 and damping Cj. Secondary suspension with stiffness k2
and damping C]. Speed v. Displacements Zc(t), zt,(t), zw(t) and track
irregularity zds).

The three force equations of the system with respect to the static equilibrium can be written
as
(5-8a) Figure 5-3 Two-~imensional model with two DOF (z,x). Masses m, mwl and mw 2.
Mass In.ertla momentJ. Damping CI and Cz; stiffnesses kl and kz. Longitu-
dinal dIstance from centre of gravity to wheelset L loch L z. Speed v. Dis-
placements z(t), Zwl(t) och zwz(t) and track irregularities Zt(s). Rotation,
or pUch, X(t). Damping forces Fel and Fez; dynamic suspension forces
(5-8c) Fkl •d and Fk2,d; dynamic contakt forces QI,d och QZ,d.

The force! moment equations for the suspended mass write as

Using Equation (5-2) one gets the two equations of motion in matrix form as
(5-13a)

[~< ~"]{::} +[-'~, ,,-+'~,1{::} +[~t, k,-+\](::] (5-13b)

= {CIzw~kIZW} {Clz'tVO+klZt}
(5-9)
while the force equations for the two wheelsets become

or in short form
(5-10)
M"i + Cx + Kx = F

5:5
Dynamic analysis techniques Rail vehicle dynamics

(5-13c) where the static parts of the contact forces can be calculated with a static equilibrium.
In matrix form the equations of motion (5-16a,b) can be written as
mw2 Zwz = FkZ,d + FeZ - QZ,d (5-13d)

M'i:+Cx+Kx=F (5-17)
With linear charasteristics for springs and dampers we can write

(5-14a,b) where

(5-14c,d) M = [~ J] (5-17a)

Substituting Equation (5-14) into Equation (5-13) gives

(5-17b)
mz + CJ(Z - jeLJ - ZWJ) + CZ(Z + XL z - ZW2) +

+ kJ(z - XLJ - zwJ) + kz(z + XL2 - Zw2) = 0 (5-J5a)

(5-17c)
IX - cJLJ(Z - iLJ - ZWJ) + CzL2(Z + XL 2 - ZW2)-

- kJLJ(z - XL J - ZWJ) + kzLz(z + XL z - zW2) = 0 (5-J5b)

(5-17d)
(5-15c)

mwZ Zw2 - CZ(Z + iL z - Zwz) - kz(z + XLz - zwz) = - QZ,d (5-J5d)

(5 -17e,J,g)

The equations of motion (5-15a,b) can be written as

Like before ZwJ and Zwz, and their derivatives, in Equations (5-16) and (5-17d) can be
replaced by the expresions in Eq. (5-2).

For the special case CJ =Cz =C, kl =k2 =k and Ll =L2 =L, i.e. a symmetric vehicle, the
matrices C, K and F are simplified to

2C
C= [ 02cL 2
0] (5-JSa)

(5-16a,b)

while the expressions for the contact forces become, cf. Eqs. (5-3) and (5-15c,d),
2k
K= [ 02kL 2
0] (5-1Sb)

(5-1Sc)

All matrices become diagonal matrices and thus the two equations of motion get
uncoupled.

5:6 5:7
Dynamic analysis techniques
Rail vehicle dynamics

5.2 Calculation of eigenvalues and eigenmodes

(5-21)
Calculation of eigenvalues (eigenfrequencies, relative damping) and eigenmodes
(eigenvectors) gives valuable information about the principle dynamic properties of a
vehicle. It is important to know the eigenfrequencies of the vehicle in order to avoid or
resonances. For example there is the risk that the frequency of the hunting motion of the
wheelsets or bogies coincides with the yaw eigenfrequency of the carbody. Eigenmodes it = Au + B
(5-22)
give a good indication about the type of the dominant motions in the vehicle.
where
A disadvantage of this method of analysis is that the model has to be linearized before the
calculation of eigenvalues and eigenmodes. In strongly nonlinear models this might lead to
major errors in the results. It is important to be aware of the displacement amplitudes
around the working points for suspensions and wheel-rail contact used for the
(5-22a)
Iinearisation.

The solution for the free vibration of the system in Figure 5-2 is calculated from the
homogeneous part of the equations of motion (5 -10) (5-22b)

Mx+Cx+Kx=O (5-19a)

The solution of this equation system is assumed to be

it = .E...u
dt
= .E...{x}
dt x
= {x}x (5-22c,d)

x(t) = yel' (5-19b) T~e ~o called system matrix A has the dimension 2n x 2n. There are several algorithms
eXlstmg to solve Eq. (5-22). With Eq. (5-22) and B=O we get
where A. and yare the eigenvalues and eigenvectors, respectively, of the system. For a
system with n degrees of freedom one gets 2n eigenvalues and 2n eigenvectors. it = Au
(5-23a)
Substituting Equation (5-19b) into (5-19a) gives With the same as above, i.e.

(5-19c)
u(t) = {XX(t)} {y}
(t) = ly el' = ael' (5-23b)
because of el' > O. For a non-trivial solution the following condition has to be satisfied:
we get
(5-19d)
(A -lI)a =0 (5-23c)
The eigenvalue A. can be calculated from this equation. Substituting the eigenvalues into
with a non-trivial solution that has to fulfill
(5-19c) gives the eigenvectors. (5-19d) is quadratic in A.. For a system with n DOF we get
2n eigenvalues and thus 2n eigenvectors.
det(A - 1I) =0 (5-23d)
For systems with many DOF there are in principle no numerical algorithms which solve
Eqs. (5-19c) and (5-19d). Therefore often the second order differential equations are
rewritten into twice as many first order differential equations. Eq. (5-19a) can then be S.2.1 One-dimensional model with one DOF
written as If t~e right hand side (track excitation) is set to zero we get the following equation of
motion

(5-20) mz+cz+kz=O
(5-24a)

Beca~se the model on~y h~s one D~F Eq. (5-24a) can be solved easily without neglecting
where I is a n x n unit matrix. Inverting the matrix on the left hand side of the equation dampmg?r tr.ansformmg mto two first order equations (cf. below). With the same guessed
gives solutIOn lIke m (5-19b) we get

'i:R
5:9
Dynamic analysis techniques
Rail vehicle dynamics

z(t) = yell (5-24b)

For relative damping ~ '" 0 the eigenvalues are on the left side of the imaginary axis. The
where y is a constant. Substitution into Eq. (5 - 24a) yields linear system can in this case be regarded as stable. Note that the distance from the
"crosses" to the origin in Figure 5-4 is equal to 010 ,
(ml 2 +d+k)Y=0 (5-24c) The eigenvector y is for this 1-DOF system an arbitrary number, because with eigenvalues
according to Eq. (5-26) all values for y fulfill Equation (5-24c).
which in turn gives for the non-trivial solution
Example: A carbody with mass m = 38200 kg, damping c == 160 kNslm and
det( ml 2 + d + k) = mll + cII. + k = 0 (5-24d) suspension stiffness k == 2160 kNlm has with Eq. (5-26a) the undamped
eigenfrequency 01 0 = )2160000/38200 = 7.52 rad/s, and thus in Hz
This second order polynomial in A has two eigenvalues fo = 7.52/2JT = 1.20 Hz. With Eq. (5-26b) we get for the relative
damping ~ = c/(2mw o ) = 160000/(2 . 38200 . 7.52) = 0.278, i.e.
~ < 1. According to (5-26c) the damped eigenfrequency becomes
(5-25)
wd = 7.52 II - 0.278 2 = 7.22 rad/s, and in Hz
fd = 7.22/2JT = 1.15 Hz. With Eq. (5-26) finally the eigenvalues are
r al) damping which is undercritical, i.e.
Moreorless all railway vehicles ha.v~ a (ver l~ This means that a single excitation is AU = - 2.09 ± 7.22i.
c < cc = 2.;mk where Cc is the cn~lcal dampmg. m in c > c usually leads to a
----------------------------------------------------------------------
damped out with ~n oscillat.ing motl~n. O(5~~~)r~:;!r~:iti~al ~~mpin~ means that (cI2m)2
rather bumpy runnmg behavIOUr. For q. In the following it is shown that transformation into two first order equations like in Eq.
< kim and thus we can write (5-23d) of Course yields the same eigenvalues like Eq. (5-26). The system matrix is here
given by, cf. Eq. (5-22a),
(5-26)

where i 2 == -1 and where

(5-27)

(5-26a) Substituting into Eq. (5-23d) yields

!-
" =
C
Co =
C ___
2.;mk - 2mwo
c_ (5-26b)
det(A -AI) = det[_~
-A 1]
-ffz-A = A2 + ffzA + ~ =0
(5-28)

wd=Woj1-~2 (5-26c) which after multiplying with m gives Equation (5-24d).

. relative damping (~ < 1) and the damped ----------------------------------------------------------------------

are the undamped elgenfrequency, . I . Eq (5-26) are thus complex
The two elgenva ues m .
eigenfrequency of the. system. . f uency of the system (only one DOF). The We will now look into the solution of (5-24a) in more detail. With (5-24b) and the
conjugated and resuhlt m t~e s:~~pe:~;~~~~inate system like in Figure 5-4. eigenvlaues in (5-26) we have to extend the solution to
eigenvalues can be s own III a

1m
z(t) = Yle(-~(U"+i(UdJt + Y2e(-~w"-i(UdJt = e-~(Uot(YJei(Udt + Yze-iWdt)
(5-29)
where Yl and Yz are constants. With Euler's formula
X· Old

Re
eiy = cosy + isiny
-~wo. (5-30)
Eq. (5-29) can be rewritten to
X -Old

(5-31)

Figure 5-4 Eigenvaluesfor model with one DOF. (See Figure 5-1). that the solution is
z
where Cj and Cz are constants given by the initial conditions z(O) and (0). It can be shown

5:11
 Rail vehicle dynamics
Dynamic analysis techniques

and we get the characteristic equation

z(t) = e-t;WJlZ(O) +wz;O)~Wo sinwdt + z(O) COS Wd t1 (5-32)

Figure 5-5 illustrates how the solution could look like. (5-35)

z(t) Since it is almost impossible to calculate the symbolic solution of this equation by hand,
only the solution of the undamped case, i e CI = C2 =0, will be given

(5-36)

This equation is called the characteristic equation for the calculation of A. This is a
quadratic expression in AZ and the roots can be found as

kl
[- k2
- ( 1+-
mb+me
me
mb
)]2 -mbme
4klkz]
--

(5-37)

Example: A carbody with mass me = 38200 kg is isolated against vibrations with

Free vibration for system with undercritical damping. a secondary suspension stiffness of k2 =2160 leN/m, a bogieframe mass
Figure 5-5 mb = 6000 kg and a primary stiffness kl = 11200 leN/m (CI =C2 = 0). With
Equation (5-37) the quadratic eigenvalues become
5.2.2 One-dimensional model with two DOF Ai,2 = - 1141.6 ± 1094.4, which leads to
For the model from Figure 5-2 with 2 DOF we get the following equation of motion for
Ai = - 47.2 and A~ = - 2236.
The four eigenvalues are calculated to
free vibrations .11,2 = ± 6.87i and .13,4 = ± 47.3i,
giving the undamped natural frequencies
Wj = 6.87 rad/s respectively UlJ. = 47.3 rad/s, i.e. h = 1.09 Hz respectively
h = 7.53 Hz.
The eigenvectors of the example above can be calculated to

With the solution YI Yell =

= {Ybl {I} .
0.165 respeclively Yz = {Ye2}
Yb2 = {- 0.026}
1 (5-38a,b)

(S-33b)
•. From the eigenvectors we get information about the amplitudes of the displacements and
'. the phase shift between the oscillations of the different masses. In an undamped
only OOand 180° phase shift exist. In the first eigenmode mainly the carbody
we get while in the second eigenmode the bogie frame mass oscillates. In the first
the two masses oscillate in phase and in the second mode they oscillate with a
shift of 180 The two modes are shown in Figure 5-6.
0
•

(5

5:13
Dynamic analysis techniques
Rail vehicle dynamics

1
-0.026 (5-42a)

~o ire~~igat~ vibration is?la:ion in the carbody it is more interesting to evaluate its

s~~:;:: IOns ue to the excitatIOn. The corresponding frequency response function can be
0.165
1

(5-42b)

One-dimensional model with 1 DOF

Mode 1 (11=1.09 Hz) Mode 2 (12=7.53 Hz)
The equation of motion is
Figure 5-6 Eigenvectors for example with two degrees of freedom.

An example for a system with two degrees of freedom containing damping is discussed in m z + cZ + kz = c Zw + kzw
(5-43)
Chapter 8.
With what is said above in this section we get

5.3 Calculation of frequency response functions

(5-44)
In frequency response analysis the particular solution of the equations of motion is and
obtained that is due to the excitation of the system irrespective of the homogeneous
solution. Frequency response functions describe the relationship between a harmonic
z(t) = 2 Re (Zeiw~
excitation at different points in the system and the response of the system, also at different (5-45)
points of the system. The functions are in general complex-valued and contain
Substituting Eqs. (5-44) and (5-45) into (5-43) yields
information about amplitudes and phases with respect to the exciting force.
Also frequency response analysis is a method which demands linear or linearized ( - mw 2 + ciw + k)z = (ciw + k)zw
equations of motion. Frequency response functions can for example give valuable (5-46)
information about how well the carbody is isolated against different excitation frequencies and thus
or wavelengths in the track irregularities. Peaks in the frequency response function
indicate that the pertinent excitation frequencies can be critical. Often peaks arise near HL(iw) = ciw + k
eigenfrequencies (natural frequencies) of the system. - mw 2 + ciw +k (5-47a)
Introduce a periodic force into the equation system (5-19a)
H; (iw) = - w2HUiw) = _ w2 ciw + k
F(t) = 2Re(Fes~ (5-39) - mw 2 + ciw +k (5-47b)

In this case a linear system has to respond with a periodic oscillation

With kim = w~ and clm = 2~wo from Eq. (5-26a,b), Eq. (5-47) can be rewritten as
x(t) = 2 Re(:xes~ (5-40)
HUJ3, 1;) = 1 + i21;/3
Especially an undamped harmonic input is interesting to study. This means that s = iw (1 - /3 2) + i21;(3 (5-48a)
where w is the circular frequency of the excitation.
If this is introduced into Eq. (5-40) we get
(5-48b)
(Mw 2 + Ciw + ]():x = F
If the system has n degrees of freedom n2 frequency response functions can be "a,"u""'W the frequency response function for accelerations has been normalised with 2
where -wo
because every input Ilk (k=l,l1) gives an output '![. (;=1,11). The frequency
J
functions are therefore given as
(5-49)

,\·14
';:1 'i
 Rail vehicle dynamics
Dynamic analysis techniques

Amplification of displacement respectively acceleration is given by the absolute value of

. . xcitation and undamped system frequency. The frequency response the frequency response functions, i.e.
I.e. the ratIO between e h i d ending on the dimesionless variables fJ and~.
functions in Eq. (5-48) are t us on y ep

(5-50a)

,
5

(5-50b)
4
1; = 0.1
Figure 5-7 illustrates these two expressions.
3
If 1; = 0.2 From the figure the following can be concluded:
\1
2 - Low damping yields high amplification at resonance (fJ = 1).
~1=0.5~
1
~ 1;=0.\ - High relative damping gives large accelerations, also for e.g. fJ > 3.

~
An important task in vehicle dynamics is therefore to find adequate damping.

One-dimensional model with 2 DOF

2 3 4 5
1
(a) If we introduce a bogie mass this acts like another filter for high frequency excitations ..
The frequency response functions for the model with 2 DOF in Figure 5-2 illustrates this.
he equations of motion for this model are

With the solution

(5-52)

and

(5-53)

after substitution into Eq. (5-51) we get (cf. Eqation (5-33c)),

(b)

Frequency respnse junction jor displacement and acceleration of track

Figure 5-7
excited 1 DOF system.
(a) Displacement. .
(b) Acceleration (normaizsed).

We are looking for the frequency response function between Zw and zc. With (5-42) it can
be written

5:17
 Rail vehicle dynamics
Dynamic analysis techniques

(5-55)

With Cramer's rule it can be shown that

(a) ~

~
~ (b)
100

Zc f
= Zw = V
10 /
/
r/
(5-56)

W
which yields 1 10 100
FigureS-8 F~equency. response function for acceleration in a system with and
without primary suspension.
(a) W~thoutprimary suspension (one degree offreedom).
(b) With primary suspension (two degrees offreedom).
(5-57)

Frequency response functions between Zw and :i c are like before 5.4 Power spectral analysis
In S~c.tion 2.3 it is mentioned that track irregularities can be represented bower s ectral
(5-58) densItIes (PSD). In power spectral analysis - which also is a linear met!o~ an~
of sis
ipower
e Ispectra
.f of
. hthe vehicle response are calculated by muItI·pIYI·ng power spectra ofytrack-
rr. gtu. an ~es WIt the frequency response functions between the point of excitation and the
In principle frequency response functions like in Eqs. (5-48a,b) can be derived. The
~Olll. III t e vehicle to be analyzed. Power spectra are often calculated with help of the
expressions, however, become rather complicated. Instead an example with specific
a~~n~:a~sfo:m of a sI~al. If CK is the fourier coefficient of a discrete fourier spectrum
values is given. calculated a: [;~fuency Interval between two coefficients, power spectral density is
Example: A carbody with mass me = 38200 kg is isolated against vibrations with a
secondary stiffness k2 = 2160 kN/m, a secondary damper Q = 160 kNs/m, a
bogieframe mass mb = 6000 kg, a primary stiffness k] = 11200 kN/m and a C2
primary damper c] =240 kNs/m. Figure 5-8 shows the frequency response S(wK) = 2d: ' K = 1, ... , N. (5-59)
function for acceleration in the carbody with and without primary
suspension. In th.is way thfe spectral ~ensity gets independent of the frequency interval in contrast to the
f
ouner trans orm, cf. FIgure 5-9.
The frequency response function for the system with two degrees of freedom has two
peaks, almost at the natural frequencies of the system (cf. Section 5.2). Comparing the
frequency response function for the two degrees and one degree of freedom systems, it can
be observed that the high frequency part of the acceleration is efficiently filtered out by the
primary suspension.

<j·le)

5:18
Dynamic analysis techniques Rail vehicle dynamics

:~:r; I'M: !' I

S(w)

I ! ®,' i:
S(w)
o 2 4 6 8 10 12 14 16 18 20

~:~Ji±±=:::'·· Ii·]
lC z
2 K-

~~~~~--~~w
Llw
Figure 5-9 Calculation of spectral density from a discrete fourier transform (linear o 2 4 6 8 10 12 14 16 18 20

l]ft!······~
scale on horizontal axis).

The power spectrum is the square of the fourier .transform,. this .means that it is a
real-valued expression. However the phase informatIOn of the signal IS lost. The spectral
density of the vehicle response is found as follows

(5-60)
o 2 4 6 8 10 12 14 16 18 20
Frequency [Hz 1
One has to be aware of the fact that measured power spectra of track irregularities usually
are given with spatial circular frequency. Before multiplying them with the frequency Figure 5 -1 0 Power spectra of carbody acceleration for a system with two degrees
response function they have to be divided by the speed offreedom respectively one degree offreedom. Vehicle data as in the
examples above.
1
Szw(w) = Ii . Szw(Wspatial)
[s m. m
2 .
. rad
m] (5-61)
Usually a Gaussian distribution of track irregularities is assumed. In a linear system in this
case also the vehicle response has a Gaussian distribution, and can be described by mean
The spectral density of accelerations can be calculated similar to the section before value and standard deviation. The standard deviation, bz, is evaluated by integrating the
power spectrum
(5-62)
00

In Figure 5-10 power spectra of carbody accelerations for the example i~ the sections
before are shown for a vertical excitation with the following spectral density
b~ = JSz<w) dw (5-64)

o
4.028 . 10- 7 (5-63)
Szw(Wspatial) = 0288 . 10 -3 + 0.68w 2 . + w 4spaLLal
. In practice however it is not integrated between 0 and 00 but as
. spallal

W2
The power spectrum is valid for a standard track in the network of the German railways
[7]. For poor track the nominator has to be set to 1.0785' 10- 6 .
bi = J Siw)dw (5-65)

wI

Standard deviation is a quantity which is also used when analyzing measurements. Mean
value +3 . standard deviation is sometimes assumed to be a statistical maximum value of a
signal.

5:20 5:21
 Rail vehicle dynamics
Dynamic analysis techniques

With the power spectrum of an acceleration like in Equation (5-62), it is easy to evaluate
comfort values. If it is assumed that even a human being's sensibility for vibrations can be ~ .E
j
on ;,
expressed by a frequency response function, the "subjective standard deviation" of for '"
0 '':
'"
~ ,...;-
example vertical carbody acceleration can be calculated as
~
""
~

00 .c
;;:
(5-66) oJ
2
b~
z,carbody,subj
f IHhumall(w)12 . S.. z,car
b0 dy (w) dw $l ~
.~
o
More about evaluation of ride comfort can be found in Chapter 11.
";;:
Q.)
.c
.~

u
.§
'""
>,
'0 ,..............j.._.....,~. .,._ ... _...J~
5.5 Time step integration '"
.~

Time step integration of the differential equations of motions is the most frequently used ~
method of analysis today. The reason for this is that all non-linearities in the real vehicle -
especially in the wheel-rail contact - can be taken into account in the model. Of course
time step integration needs the longest computing time of all methods presented in this
chapter. Usually 10-20 seconds of vehicle running, which means a travelled distance
between 500-1000 m, depending on vehicle speed are simulated.
The equations of motion are often transformed to first order differential equations

(5-67)
U 'T f(u,t), U = {x(t),x(t)}T

where functionf describes spring and damper forces, contact forces, mass forces and track
irregularities. A wide range of solvers exists for the numerical integration of the 1.. ·················-·,.. ~~···-:········-···· ..J~
differential equations. It shall be mentioned here that the wheel-rail contact can cause
problems for the integrator. One approximation is to regard the normal contact as rigid. In
this case one gets an algebraic equation for the calculation of the contact force, which
needs special algorithms.
Another possibility is to model the normal contact with a spring stiffness (Hertzian
,....... -........ i-.·-..·~~········:.···-··-··· .. ~~
stiffness) of about 1.5 . 109 N/m. This introduces high eigenfrequencies in the system, one
gets a so-called stiff equation system. Not all integrators are suited for stiff systems.
Besides the time step has to be very short, at least shorter than lIfmax. With half the mass of
a wheelset of about 800 kg the maximum time step becomes

(5-68)
Llt max = _1_ = ~ = be = 0.005 s 1h2'.~--74-.--~-_~4.--~-.J3°
fmax Wmax 11.5 . 10 9 /800 '-3 160_35_".rt.: k~bpll1h.Fh

Even 5 ms is often not small enough, but the lateral wheel-rail contact with the overcritical Figure 5-11 Results from time step integration.
damped eigenvalues (compare Chapter 8) limits the maximum time step to about 0.1 - 1
ms.
Figure 5-11 shows some results from a time step integration with GENSYS [60]. Thefilter
effect of the suspension can clearly be observed. The higher frequencies are filtered out
gradually.
5:23
 Rail vehicle dynamics

6. Advan.ced vehicle modelling

The models in Chapter 5 only describe the vehicle dynamics in the longitudinal-vertical
plane for very simple vehicles. Three-dimensional models of arbitary vehicles during
curve negotiation call for more advanced vehicle models, in particular if models of flexible
bodies and/or advanced suspension systems are to be considered.

In this chapter three-dimensional models of the vehicles' main mechanical components -

wheelsets, bogie frames and carbodies - are described. Models for suspension components
are also described.

6.1 Wheelsets, bogie frames and carbodies as rigid body models

Wheelsets, bogie frames and carbodies together represent a clear Iy dominating part of the
total vehicle mass. If these components are modelled as rigid bodies it is, in addition to
motions and forces, only the bodies' mass properties that are involved in the pertinent
equations of motion.

However, in particular wheel sets and carbodies often have a structural flexibility that
cannot be neglected in the modelling. These components then have to be modelled as
flexible bodies, see Section 6.2.

For the model "rigid body", which is treated in the present section, the distance between
two arbitrary points of the body is by definition constant and independent of the body
motion. The body motion can therefore be described by only six motions, three
translations and three rotations. Thus a rigid body has six degrees of freedom, cf. Figure
3-3.

If the body performs large rotations the three angles cannot be superposed arbitrarily
though: a certain order of the angles has to be defined and the body rotation cannot be
described by a vector. However, the body angular velocity and acceleration can always be
represented by vectors.

The six equations of motion of the rigid body achieve their simplest form if we let the three
unknown translations, and their time derivatives, refer to the body centre of gravity. Then
the mass moments of inertia also refer to axes through the centre of gravity.

In this way the six scalar equations of motion can be formulated as two equations of motion
in vector form: one force equation and one moment equation.

Below a general formulation of the force and moment equations of a rigid body is
described (Section 6.1.1). Then a formulation often used in rail vehicle dynamics is
introduced (Section 6.1.2).

6:1
Advanced vehicle modelling
Rail vehicle dynamics
6.1.1 General formulation
The acceleration and force vectors of E uation 6
the base vectors of the inertial system q ( -1) can be expressed as components of
According to Newton the force equation for a rigid body C can be written as
J C'
a = ale] + a2 e2 + a3 e3
mfF =F (6-1) (6-4a)
F = F]e] + Fze z + F 3 e3
Wh . (6-4b)
where m is the body mass and where laC' is a vector that describes the acceleration of the en EquatIOn (6-4) is introduced i E .
body centre of gravity C* relative to an inertial (fixed) reference system I. F is the force equations nto quatlOn (6-1), we get the well-known scalar
vector due to external loads, dead weight, suspension forces etc .. Thus the acceleration
vector and the force vector must be parallel and have the same direction.
(6-5a)
According to Euler the moment equation of a rigid body can be expressed as
(6-5b)

(6-2) (6-5c)
The Euler marne t '.
n equatIOn IS derived in a bod .
vectors for this system (C) d Y- fixed reference system. If the base
. are enoted e' , d'
where] is the body's inertia tensor, with respect to its centre of gravity, and where JwC is a EquatIOn (6-2) can be written' I ,ez an e3 the vectors and inertia tensor in
III components as
vector that describes the body angular velocity relative to the inertial system 1. (The letter
w has thus nothing to do with frequency here, cf. Chapter 5). Moreover, M is the moment
vector whereas the body angular acceleration is described by the vector
(6-6a)
+ wi e~- + w3' e'3
(6-3) (6-6b)

(6-6c)
Thus time differentiations of the angular velocity vector lw C in an inertial system I
respectively a body-fixed system C, give the same result. The introduced quantities are
illustrated in Figure 6-1.
(6-6d)

J I ] etc. are the body's mass moments of inertia .

These moments of inertia are time ind wl~h respect to the body centre of
(base vectors e' e' d' ) ependent Since they refer to a body-fixed
] , 2 an e 3 •

I ofE f
qua lOn (6-6) into Equation (6-2) gives three scalar moment equations

J I1 W'] - (J'Z2 - J33

' )w
" w3 + J I2 (w:! - w' w' )
2
+ J ?' (w' Z _ ' 2 I· f
3 ]
_3 2 W3 )+JI3(W3 +W;WZ)=MI
(6-7a)
J' " (
22 W2 - J~3 -Ji] )W3 W{ +J.J.3(W3 -W'W')
+J' ( 2 1 2
13 W) - W? ) + J{2 (W{ + Wz w3 ) = M;
(6-7b)
J' " (
33 W3 - J{] -J22 )w; Wz +J{3 (W; _ w' w,)
+ J' (w'2 _ '2) '( . , 2 3
12 I W2 + J 23 W2 + W3 W; ) = Mi,
(6-7c)
Figure 6-1 Quantities in Newton and Euler equations of motion. thus have q d' .
ua ratlC (nonlinear) terms of angular velocities.

6:3
 Advanced vehicle modelling Rail vehicle dynamics

For the special case when the deviatoric moments 112 = 113 = 123 = 0, Equation Track-following reference system
(6-7) becomes
We star: by extending the gene~al formula~i?n to a track-following system for the rigid
wi - (122 - l '33 )W 2
, w3, =Mi (6-8a) ~~~:. FIgure 6-2 shows the pertment quantities for a traCk-following system T, cf. Figure
III

122 w2 - (133 - Jf11 )

w3' wI, =M2 (6-8b)

Track plane c
133 w) - (Ill - I'22 )wI, w 2, = M3 (6-8c)
--- ---Top of rail
The quadratic terms in angular velocities above are often referred to as gyroscopic terms.

If the body also is axisymmetric, say with respect to e'z ' the relation 133 = I II holds and
Equation (6-8) yields

I II WI - (122 - I'11 )w 2, w3, = Ml (6-9a)

122 Wz = M2 (6-9b)

l '22 )WI,
III w) - (Ill - W 2, = M3 (6-9c)

For instance, wheelsets should be considered as axisymmetric. If e'z is the wheel set axis of
symmetry the angular velocity W is much higher than wi and w3 due to the wheelset
z
nominal rotational speed vlro (v is the vehicle speed and ro is the radius of the wheelset's
running circles).

6.1.2 Formulation with track-following and body-following reference systems

All vehicles have a large desirable motion and a small undesirable motion, cf. Sections 3.1
and 4.1. In particular for rail vehicles, the large motion is known since the nominal track
geometry (curves etc.) and the vehicle speed are known. For the large motions of the
different vehicle bodies, the nominal positions of the bodies within the vehicle are also
e3
known.
Figure 6-2 Quantities of track-following reference system T.
When the dynamics of a rail vehicle and its different components is modelled, it is in
principle necessary to include the known motion above since the important parameters
The ~rigi.n of the ~ystem T, denoted To, is moving along the nominal track centre line
vehicle speed and nominal track geometry somehow need to be incorporated in the
(longItudmal coordmate s) with the vehicle speed. The motion of To, relative to an inertial
models. In addition, the small relative motions of the vehicle bodies need to be determined
system I, is described by the position vector IpTo, the velocity vector Iv To and the
since they usually should be minimized.
acceleration vector laTo.
A track-following reference system (coordinate system) is therefore assigried to each
vehicle body. Knowing the body's nominal position within the vehicle, e.g. height of e
The orient.ation ofthe system Tis given byits base vectors iiI' 2 and ii 3. The vector iiI is
centre of gravity above top of rail, the formulation can be extended to a body-following parallel wIth the track centre line and in the direction of travel
I . e e
. 2 is perpendicular to I'
reference system. The body's track-following and body-following systems have the same para lel.to the track plane and directed to the right with respect to the direct of travel. The
orientation, i.e. they are parallel, but different positions of their respective origin. vector e 3 completes a right-handed Cartesian coordinate system.
The aim is to express the body motions as unknown motions relative to the known motions
The angular velocity vector lro T and the angular acceleration vector I WT define the time
of its body-following system. The relative translations are longitudinal, lateral and
vertical motions (x, y respectively z) whereas the small relative rotations are roll, pitch and dependence of the T -system's orientation, with respect to the inertial system.
yaw (cp, X respectively '1f!), cf. Section 3.1. The following relations apply

6:4 6:5
 Rail vehicle dynamics
Advanced vehicle modelling

(6-10 a)
________-=~-r~~--------L.M

(6-10b)

dscosXI ~
ds cos X,
linT = !4(lwT) = Td(lw T) + Iw T X Iw T = ~(I(OT) (6-10 c)
dt dt

where thus time differentiations of the vector IwT in an inertial system 1 respectively a
tan 1jj,~
track-following system T give the same result.
IT' hi" lved in Equation (6-lOb,c), play an
The angular velocity ve.ct?r w , whl~ a s: t~ 1~;a~k-fOllOwing reference system. The
important role in descnbmg the motion 0 e . k t Figure 6-4 Relation between the angles CPt, Xt and l/Jt on the one hand and
vector lw T is a function of nominal track ge~metry (horizontal curve radIUS, trac can, horizontal curve radius R, vertical curve radius Rv and the angle (j5t on
track gradient, vertical curve radius) and vehicle speed. the other hand.
k number of steps First we must find the
To determine this dependence we must ~a e.\ t and thos~ of the track-following
relation between the base vectors of the ~nertIa .sys em As seen in Figure 6-3 a system with base vectors e1
(i=1,2,3) is first created through a
system. Figures 6-3 and 6-4 illustrate this relatIOn. rotation 1/J, about the vertical base vectors e3 = ~. This rotation corresponds to track yaw
and thus a horizontal curve radius, see Figure 6-4.

Then a system of base vectors ef is created through a rotation X, about base vectors
e1 = e~. This rotation corresponds to track gradient and vertical curve radius.
With these two rotations the nominal track centre line is defined. The remaining rotation
represents a cant (superelevation) and is carried out by a rotation cp, about the base vectors
ef e
= l · In this way the track-following reference system has been created.

The final rotation is thus realized through a rotation about the track centre line. This is
normally not the way track cant is designed. Instead the inner rail keeps it vertical level in a
horizontal curve whereas the outer rail is elevated to create the cant alone. However, such a
cant can be represented by a combination of rotations X, och cp,. From Figure 6-4 it is also
seen that the cant angle cp, is related to a similar angle i'/fl in a vertical-lateral plane.

Among the applied rotations 1/J" XI and cp I the yaw roation 1/J, definitely is large, e.g. a 90
degree tum of the track. The rotations XI and CPI are typically up to 4 respectively 7 degrees.
In order to relate the three rotations to common quantities like curve radii, track gradient
and track cant they have to be applied in the order presented above. Note that the track
gauge is not present in the equations above.
The relations between the different sets of base vectors in Figure 6-3 are given by

e~
Transformation from inertial to track-following system.
Figure 6-3
The base vector e3 is vertical and dIrected downwards.

6:7
6:6
Advanced vehicle modelling Rail vehicle dynamics

where the components are achieved by means of Equations (6-11b,c) and (6-13) as
e1} [CO~1jJ, sinl/J, o]{e l}
{~ = - Slll1jJ, COS1jJ, 0 ez (6-11 a)
(6-14a)
~ 0 0 1 e3
3

(6-14b)

{e!}
ez =
elf
X
[COS I 0 - sin
0 1 0
SinXI 0 COSX,
X'l{e1}
~
~
(6-11 b)
(6-14c)
3 3

For small angles x,(track gradient) and tpt(cant) - and the angular velocities (P" X, and 1fJ,

{e'} = ° -co~tp, e~
el 1 0 0 el
= (Pt' w2= Xt and w3 = 1fJ,.
e2
3
[0 sintp,]
Sllltp, costp,
{B}
eB
3
(6-11 c)
in the same order of magnitude - we get WI

The angular velocities (P" Xt and 1fJ, in Equations (6-14a,b,c) can now be related to the
nominal track geometry and the vehicle speed as
Insertion of Equations (6-11a) into (6-11b), which in turn is introduced in Equation
(6-11c), gives the relation between the base vectors of the inertial system and those of the
track - following system as (6-15a)

cosX,sin 1jJ,
(6-15b)
sintpt sinx,sin1jJt + costp,cos1jJ,
costp,sinxlsin1jJ, - sintp,cos1jJ,
(6-11 d)
. d1jJ, d1jJ, ds cos X,
1jJt = dt = dS dt = ~v (6-15c)

For small angles the matrix in Equation (6-11d) tends towars a unity matrix. According to where Equations (6-12a,b) have been used. Compare Equation (6-15a) with Equation
Figure 6-4 the following relations hold (4-22) for a linear (straight) superelevation ramp.

dscosX,
d1jJ, =-R-- (6-12a) Insertion of Equation (6-15) into (6-14) finally gives andwI' w2 w3 -
and thus fw in T
the track-following reference system - expressed as functions of vehicle speed and
nominal track geometry
(6-12b)
(6-16a)

(6-12c)
(6-16b)
In Equations (6-12a) and (6-12c) the term cos Xs is present. But with a maximu~ track
gradient X, of, say, 4 degrees this term is very close to unity. For completeness we shll keep
this term though. (6-16c)

The angular velocity vector f(J)T can now be expressed as a superposition of three angular
velocities corresponding to the rotations of Figure 6-3:
(6-13)
General expressions in component form Of the velocity and acceleration vectors of the
track-following system (fv To , fa To, l(J)T and f Ii?) yield lengthy equations, especially for
the angular acceleration vector. These expressions are therefore put into an appendix, see
Appendix 6A.

If f(J)T is expressed in only the base vectors of the track-following system we arrive at Here we restrict ourselves to the special case of horizontal curve with constant radius and
cant - with a cant realized through rotation about the track centre line and without track
(6-14) gradient - giving the simpler expressions

6:8 6:9
 Rail vehicle dynamics
Advanced vehicle modelling

(6-17a) The. t.ranslations x, y and z in Equation (6-19) represent the time dependent and unknown
positIOn of the centre of gravity. This time dependent position is thus the position of the
body's centre of gravity relative to the body-following system, or alternatively, relative to
(6-17b) the track-following system if the nominal measures I, band 11 are considered.

(6-17c)
(6-17d)
where a = dv/dt is the vehicle acceleration.
Body-following reference system
After determining the motion of the track-following reference system above, we can
proceed to determine the motion of the body-following reference system. As mentioned
earlier, the latter system has the same orientation as the track-following system but another
origin position. The aim is that the body motions can be expressed as (small) unknown
motions relative to the motions of the pertinent body-following reference system.
If we start with the position vectors in Figure 6-2 and then time differentiate them once and
twice, with respect to the inertial system, the motions of the body centre of gravity can be
expressed as
(6-18a)
Figur~ ?-5 demonstrates the position vector TopC' in a longitudinal-vertical plane, i.e. the
qua?~ltJes band yare not present. As seen from Equation (6-19), 11 and z are defined
positive downwards. In the example of Figure 6-5 the (positive) distance vertically from
top of rail to the centre of gravity equals -(I1+z). For instance a wheel set with radius 0.5 m
gives 11 = -0.5 m whereas a carbody often has a centre of gravity between h =-1.5 and h =
-2.0m.
The lo~gitu.dinal measure I should be chosen as small as possible to avoid unnecessary
approximatIOns III the equations of motion below. The lateral measure b gives the nominal
lateral position of the centre of gravity relative to the nominal track centre line. Since
~ehicles ~re designed to be almost symmetric with respect to a longitudinal-vertical plane
I~tersectmg the track centre line, the measure b is small. (It is hard to require passengers to
Sit or stand symmetrically in the carbody though!).
I +x
I C*
-+-
I
- (h + z)

ej
- - - - ' ' - - - - - - - - $ - - l . . : . . L - - - - - - - - - - - - Top of rail

e3

Figure 6-5 Position vector T.pC* in a longitudinal-vertical plane.

In Equation (6-18c) the last four acceleration terms (the last line) originate from the Force equations
second derivative of the position vector Tope'. The second last term is the so called
If no,:", Equation (6-19), as well as Equations (6-14) and (6-77a), are inserted into
Coriolis acceleration which thus is a cross product of two velocity vectors.
EquatIOn (6-18) we get the motion components of the centre of gravity as
The position vector TopC' for the body centre of gravity C* can be expressed in the
track-following, and body-following, system's base vectors e; (i = 1,2,3): (6-20a)

(6-19) I C· •
v = vel + [.xe• l + Ye2
. • • ]
+ ze3 +
In Equation (6-19) the measures I, band h are constant (time independent) and d.escrib~ ~he + [wle j + w2e Z + w3e 31 x [(I + x)e l + (b + y)e 2 + (h + z)e 31 (6-20b)
nominal position of the centre of gravity. They also define the di~erence m positIOn
between the track-following and body-following systems. We can Imagme thiS by an
undeformable rod of length Jt2 + b2 + h2 attached at its ends to the origins of the
track-following and body-following systems.

6:11
6:10
 Rail vehicle dynamics
Advanced vehicle modelling

Moment equations
laC' = ae l + [Wle l + W2eZ + W3e3] X vel + [xej + ye 2 + 2ii3] +
Fi~allY, we need to express the moment equation in the track-following (body-following)
+ [d,je j + d,ze z + d,3 e3] X[(I + x)e l + (b + y)e2 + (h + z)e3] + rse etrence sYhstem. For the .body's angular velocity and acceleration, relative to the inertial
ys em, we ave the relatIOns
+ 2[w l e + W2e2 + W3e3] X [xel
j + yez + ze3] +
(6-23a)
+ [Wle l + wze z + W3 e3] X ([wje j + wiz + W3e3] X [(I + x)ej + (b + y)e2 + (h + z)e 3]l
(6-20c)
I{JJ. C = Id (I{JJ T + T{JJC )
di = Id
di(I{JJT) + Td
dt (T{JJC) + I{JJT X T{JJC = linT + TinC
Insertion of the expression for laC' according to Equation (6-20c) and the force vector (6-23b)
(6-21)
where Tinc = !A(T(J)C)
dt .
in the force equation (6-1) gives three scalar force equations for the body related to the F::he bOdy;s rotation relative to the body-following system we now introduce the
base vectors of the track-following (and body-following) system: u t town ang es cp, X and 1jJ for the body's roll, pitch respectively yaw motions These
ro af IOns are often small
. ' say bel ow 2 d R
egrees. easonable ·for the
approximations
m[x + wz(2z + wly - wzX) - W3(2Y + W3 x - wjz) + d,zz - d,3Y] per ment angular velOCity and acceleration vectors are therefore
TC ., ., .,
= Fx - m[a + wz(wjb - wzl) - W3(W31- wlh) + d,zh - d,3 b ] (6-22a) {JJ = cp e j + Xez + 1jJ e3 (6-24a)

T·C .. , .. , .. ,
{JJ = cpe j + Xe2 + 1jJe3 + (wje j + W2e2 + w3e3) X (tiJel + xe 2 + 1je 3 )
m[y + W3(2x + wzZ - W3Y) - Wj(2z + WjY - wzx) + d,3 x - d,jZ] = (6-24b)
= Fy - m[ w3 v + w3(w zh - W3b) - wj(wjb - wz/) + d,3 / - d,jh }6-22b) ~~~: thu~ no spe~ific order of application of the rotations are implied, cf. Equation
( a) With EquatIOns (6-13) and (6-14). Inserting Equation (6-24) into (6-23) gives

m[ 2 + wj(2Y + W3x - wjz) - w2(2x + wzz - W3Y) + d,jY - d,zx] = I{JJC = [wle l + wze z + w3e3 ] + [tiJel + Xez + 1je3 ] (6-25a)

= Fz - m[ - W2v + Wj(w31 - wjh) - wz(wzh - W3 b) + d,j b - d, zl}6-22C)

linC = [d,1 ej + d, z e2 + d,3e3] + [qie l + Xez + i'ie3] +
where the unknown displacements x, y, Z and their time derivatives are collected at the + [Wj el + wze z + w3 e3 ] X [tiJ el + Xe2 + 1j e3 ] (6-25b)
left-hand sides. The mass m and the measures I, band h are known. For given nominal
track geometry and vehicle speed, v and a as well as all wi and d, i terms are also known. (b d
~ II vector
The moment . M can aIso b e expresse d m
. the base vectors of the track - following
a Y-I0 owmg) system
Expressions for these angular motions are found in Equations (6-77c,d) in Appendix 6A.
The force components Fx, Fy and Fz include extemalloads, dead weight, suspension M = Mcpe l + Mxe z + M",e 3 (6-26)
forces and for wheelsets also wheel-rail contact forces.
whereas the corresponding inertia tensor car: be approximated by
The force equations (6-22) are thus linear in the unknown x, y, Z and their time derivatives.
Equation (6-22) can be compared with Equation (6-5) of the general formulation. The
formulation in a track-followinglbody-following system thus gives significantly longer
expressions but this formulation is in principle necessary for general rail vehicle dynamics (6-27)
analyses.
Note that no approximations have been done in the derivation of Equation (6-22). In
mass moments of inertia are thus still assumed to be time independent even though
practice there are of course certain terms that are more important to consider than other.
Terms that are "small" for the railway application at hand may therefore be neglected. now not are related to the body-fixed system.

6:13
6:12
 Rail vehicle dynamics
Advanced vehicle modelling

Inserting Equations (6-25) to (6-27) into Equation (6-2) gives thre~ moment equations
, ' d' I I form the equaHons become
that refer to the base vectors e l' e 2 an e 3 · n sca ar
1 (~ + ip + w21jJ - W3X) - (lx - 1l/J)(w2 + X)(w 3 +~) +
: J~(~' + X - wi', - 2W, ~ - +.) + ],,[(w, + i)' - (0" + ,p)'} +
(6-30)
where - vlro is known and where now the unknown relative motion Xcan be assumed
small. Still, the wheelset's nominal rotation is assumed above to refer to the base vector
e
(axis) z and not to the body-fixed e 2 giving an approximation.

6.2 Wheelsets, bogie frames and carbodies as flexible body models

+l<f!1l!(~3 + iP + W1w2 + 2W 1 X + ip1.) "" Mcp (6-28a)
Structural flexibility

lX(~2 +X + w3ip - Wl~) - (lV) -lcp)(w3 + 1jJ)(WI + ip) + In Section 6.1 we have assumed the bodies to be rigid. However, in many railway
applications it is necessary to consider the structural flexibility of the bodies in vehicle-
. " .. ) [(' + ,j,)2 _ (w + ip)21 + track interaction simulations, see e.g. [15]. By structural flexibility we here imply that the
+lXl/J ( w3 + iP - W1(;:'2 - 2w2if! - if!X + lcpl/J w3 'r 1 body deformations, due to forces acting on the bodies, are considered in the analyses.

+1IJlX( ~1 + if; + WZW3 + 2W z1jJ + 1.1jJ) "" Mx (6-28b)

Whee/sets have a significant structural flexibility, in particular in axle bending and torsion.
Some types of bogies, e.g. torsionally flexible bogies, have a flexibility of the bogie frames
that influences the ride stability and therefore could not be neglected. Probably the

1l/J(~3 + iP + WjX - w2ip) - (lcp - lx)(w 1 + ip)(W2 + x) + structural flexibility of carbodies is paid most attention since it often results in poorer ride
comfort. Figure 6-6 illustrates the importance of the carbody structural flexibility.

+lcpl/J( ~j + if; - W2w3- 2W31. - 1.~) + lCPx[ (WI + ip)2 - (W2 + 1.)21 + Power spectral density of carbody acceleration
Logarithmic scale

+lXl/J(~Z + X + W3Wj + 2W 3 ip + 1jJip) "" Ml/J

(m/sZ)2/Hz
(6-28c) -2
.~

These moment equations include a number of quadratic terms in the unkno:v n an~lar -3 /

velocities ip, X and 1jJ. The angular velocities wi and angular accelerations Wi are given -4 .i '.
. A d' 6A The moment components
by Equations (6-77c) respectively (6-77d) m ppen IX . / .'..
' \

T
and T
include moments due to external loads, suspension forces and for -5 / ,
'T
x
- /
<P' l/J
wheelsets also moments due wheel-rail contact forces. , /
-6 "

For the special case of 1IJlX = 1<pljJ = lXl/J = 0, Equation (6-28) is turned into
-7

l<p(~j + if; + w2~ - W3 X) - (lx - 1l/J)(w2+1.)(W3 +~) "" M<p (6-29a) -8 __L -________~________- L________~__________L -_ _ _ _ _ _~
L-'_-_._'~.~.~

o 5 10 15 20 25 30
Frequency (Hz)
lx(~z + X + w3ip - w1 1jJ) - (il/J- lcp)(W 3 + ~)(Wl + ip) = Mx (6-29b) Figure 6-6 Power spectra density of carbody acceleration (example).
Dotted line: Measured lateral acceleration on carbody floor centre.
Dash-dotted line: Resultafter filtering out the rigid body contribution to
: + 1/J.. + WjX
ll/J (w3 " - w2if!
,.) - (1 cp -l)(W
x 1 +ip)(W Z +X)=Ml/J (6-29c)
the acceleration.

It is important to notice that the ~oment ~quations (6-28) and (6-29~.area~~~:~~:~~~:: Figure 6-6 is based on measurements for a vehicle running along a certain track. The peak
due to the simplifications done m Equatl~n. (6-24). For s~me t~er~t I~nangle is not small at 2-3 Hz disappears when the lateral rigid body acceleration is filtered out. The resulting
approximations may be too coarse. In additIOn, for a whee set e pi c . spectra shows the significant contribution from carbody structural flexibility with a peak at
. l i a r speed X· = - vir o' However, the absolute pitch moHon can, .12-15 Hz. Note that humans have the highest sensitivity to vibrations in the frequency
h
due to t e nomma angu 0 . ' •
range of 0-20 Hz. See also Chapter 11.
neglecting track gradients, be expressed as the sum of nominal and relaHve pitch velOCity

6:15
Advanced vehicle modelling Rail vehicle dynamics

Eigenmode representation (6-33)

In rail vehicle dynamics simulations it is desirable to repres~nt. t~e body structural
where mi (mass), Ci (damping), ki (stiffness) andfi (force) constitute the modal parameters
flexibility by only a few degrees of freedom (in addition to the. SIX ~Igld body degrees ~f of mode number i.
freedom). One common way of finding such a representatIOn IS to use th.e .body s
eigenmodes found from eigenvalue analysis of the (free) body. Then we get SIX ngld body If the eigenmodes, or eigenvectors, Yi are mass-normalized the modal masses are m; == 1
modes and a number of structural modes, preferrably those with the lowest giving
eigenfrequencies (cf. Figure 3-9). The principle of adding equations of motion due to the
body structural flexibility is described below. (6-34)
Consider a (material) pointAo of a body C. Its nominal position relative to the body centre where ~; and OJ; are the relative damping and (undamped) eigenfrequency respectively for
of gravity is given by the vector C'p'\ see Figure 6-7. The displacement of the point Ao mode i, cf. Section 5.2.1.
due to body structural flexibility can be defined by the vector thus reaching a new A'r, Example: Often the bogie pitch motion can excite the fundamental vertical carbody
pointA. bending mode through stiff longitudinal traction rods between carbody and
,_ __~4
bogies. See Figure 6-8.
-------------- ,,//

--- -- ---
.-"
\
'- ..... .....
I
I --- I
I
\
\
\
I
I \
\
\ ,
I \
.-"
.-" -""'
\

I
/
/
c'r' \
\
/
/

---
C

Figure 6-7 New position A of a point Ao due to deformation of the body C.

The position vector for point A relativte to the origin of the inertial system I can therefore
be written as
Figure 6-8 Coupling between bogie pitch motion andfirst vertical bending mode of
(6-31)
the carbody.
where fpC' is the position vector from the origin of the inertial system to the body ~e~~re ~f
The modal data above are often based on a finite element model of the body in question.
gravity, cf. Equation (6-20a). The motion of the point Ao due to structural fleXibility IS
Such a model may range from a simple beam model with a limited number of beam
assumed as elements, to a three-dimensional model with many thousands of solid elements. Once the
mass matrix M and stiffness matrix J( of the flexible body have been set up the eigenvalue
(6-32) problem can be solved, cf. Equation (5-19a) with C= O. With the resulting eigenvectors Yb
;=1 the modal masses and stiffnesses can be determined as

where YXl·e'1 + yyl·e'2 + yzt.e 3' == y.(A

I
o) is a vector function for point Ao and
. ' • (6-35a)
deformation mode number i (i == 1, 2 , 3 ... n). The deformatIOn modes I are time
independent but associated with time dependent modal degrees of freedom f ;(t). (6-35b)
In order to describe the dynamics of a flexible body we therefore need 6 + n degrees of . The (relative) damping needs in principle to be measured through experimental modal
freedom: x, y, z, cp, X, 1jJ and the modal degrees of freedom f 1 to f n' In this way verified values of body relative damping, ~b can be used in the
If we now choose the deformation modes to be eigenmodes of the free body, the following ICle-lI'"CK dynamics simulations. Typically rail vehicle bodies have relative damping

equation of motion can be added for each modal degree of freedom f i of 1-4 %.

6:16 6:17
 Rail vehicle dynamics
Advanced vehicle modelling

6.3 Suspension models

The modal loads are determined by
(6-35c) In Chapter 5 only very simple models have been used for suspension springs and dampers.
fi = yTF In this section we will also introduce more advanced suspension models. We will give
examples on one- and two-dimensional models for springs as well as models for dampers.
. the body These forces are introduced to the body
where F is a vector of forces actmg on .
The dynamic behaviour of the linear models is illustrated through transfer functions,
through suspension components etc.
especially stiffness and damping. The stiffness S is defined as the ratio between force
amplitude and displacement amplitude. The damping Z is the ratio between force
Summary amplitude and velocity amplitude. Also force-displacement and force-velocity diagrams
... t wheel-rail forces etc the structural flexibility are shown.
For a reasonable predictIon of nde comfor, .d d d modelled. Data is often obtained
of the vehicle bodies often needs to be CO~SI ereI andal analysis The representation of
.. I t I sis and expenmenta mo .
through fmlte e emen ana y .. f b d . nmodes is the most common 6.3.1 One-dimensional models for springs
structural flexibility through a limited set 0 0 y elge

approach. Linear spring model

k

F F

c::::> Xl c::::> Xz
Figure 6-9 Linear spring model.

Stiffness k (frequency independent), no damping, no mass.

Harmonic excitation with frequency OJ and amplitude xo:

(6-36)
gives

F = k(x z - Xl) = kxoe iwt (6-37)

The stiffness S is found as:

(6-38)

The following force-displacement graph is obtained:

6:19
 Advanced vehicle modelling
Rail vehicle dynamics

F
F

Figure 6-10 Force-displacement graph for linear spring model at harmonic

excitation. Figure 6-12 Fo~ce-.displacement graph for nonlinear spring model at harmonic
excItatIOn (example).

Nonlinear spring model Model with linear spring in parallel with linear viscous damper

F F
F

C>Xl
k
Figure 6-11 Nonlinear spring model.
~Xl
Variable stiffness (but frequency independent), no damping, no mass. Model with linear spring in parallel with linear viscous damper.
Harmonic excitation with frequency wand amplitude Xo:
k (frequency independent), damping c (frequency independent), no mass.
.narmClllIC excitation with frequency wand amplitude.lQ:
(6-39)

gives (6-41)
F = f(xl - xl) = arbitrary function of xl - Xl (6-40)
F = k(x2 - Xl) + c d(x2 - x l )/dt = (k + iwc)xo eiw1 (6-42)
Example of force-displacement graph (piecewise linear, symmetric):
functions as stiffness S and damping Z:

6:20
6:21
 Rail vehicle dynamics
Advanced vehicle modelling

(6-43) Force-displacement graphs:

S(iw) = F/(x2 - Xl) = k + iwc
(6-44) F F

IS(iw)1 /
/

/
/ k
/
/
1
/
/
+--------;1-"--------+---- Xz-X 1
/-'1 Xo
// I c
...::::::.. __ --.l
// 1
/
/
/ -Fo
/ ----- ./
k I----~- - -- - -/"/"-(::::. --- -- - - - -- -- (a) (b)
// 1
// 1 F F
/// 1

~/~/~/~ _________~I------------------------------~w Fo
/5
= ""2kxo ./ Fo = fikxo
./
klc
.h r ·scous
Stiffness for model of linea: sl!ring in parallel WIt mear VI
Figure 6-14
damper, as a function of eXCItatIOn frequency.
-+----+-l.L..~-----+--_ XZ-X 1
(6-45)
Z(iw) = F /[d(x2 - xl)/dt] =c - ik/w
(6-46)

(c) (d)
Force-displacement graphs for model of linear spring in parallel with
Iz iw)1 linear viscous damper, as a function of excitation frequency.
(a) General (b) w = 0 (c) W = 0.5 k/c (d) W = k/c

The area of the ellipse equals

------------
c ----------T--
(6-47)
I
L-______--~I------------------~w 'and corresponds to the energy dissipated during one cycle of excitation. For a given
klc c and amplitude -'0, the energy loss per cycle thus increases linearly with the
Damping for model of line~r s?ring in parallel with linear viscous frequency w.
Figure 6-15
damper, as a function of eXCItatIOn frequency.

6:23
 Rail vehicle dynamics
Advanced vehicle modelling
Linear model of spring and damper in series, in parallel with a spring
Linear model of spring and damper in parallel, in series with a spring

k
c

F
F

~X3
~Xl ~Xl ~X2
Figure 6.17 Linear model of spring and damper in parallel, in series with a spring. Figure 6 ·18 Linear model of spring and damper in series, in parallel with a spring.
. (all frequency independent), no mass.
Parallel stiffness kp , series stiffness k, and dampmg c Stiffness k, series stiffness k, and damping c (aU frequency independent), no mass.
Harmonic excitation with frequency wand amplitude x,,: Harmonic excitation with frequency wand amplitude Xo:
(6-48)
(6-53)

Stiffness S:
S(iw) = FI(x3 - xl) = 1/[I/(kp + iwc) + Ilks] = S(iw) = F l(x3 - xl) = k + [l/(l/ks + l/iwc)] =
(6-49)
= [(kp + iwc)(ks)2 + «kp)2 + w 2 c2 )ksl I [(k p + k s)2 + w 2c21 (6-54)

2 2 0] (6-50)
IS(iw)F = [«kp)2 + w 2c2)(ks)2] I [(kp + k s) + w c- (6-55)

(6-51)
whenw = 0 /S(iw)/ =k whenw = 0 (6-56)

(6-52)
\S(iw)\ = ks
whenw = 00 /S(iw)/ = k + ks whenw = 00 (6-57)

6:25
 Advanced vehicle modelling
Rail vehicle dynamics

Model of spring and friction damper in series, in parallel with a spring 6.3.2
Two-dimensional model for springs

F-----(")!

L
k;j (i, j == 1, 2, 3, 4, 5, 6)
N
~XI
Figure 6-19 Model ofspring andfriction damper in series, in paraUel with a spring. xs, Fs

Stiffness k, series stiffness ks, friction coefficient fl and normal force N (all frequency
independent), no mass.

Harmonic excitation with frequency wand amplitude Xo:

(6-58)

Force-displacement graph (PN < ks.xo):

Two-dimensional linear model for springs.

F The force F; is related to displacements Xj according to

(i,j '= 1,2,3,4,5,6)

(6-59)

F'= lex
(6-60)
k is the a 6x6 stiffness matrix.

stiffness matrix is symmetric and the de

degrees of freedom it can be sh gr~~S o~freedom 1 and 4 are uncoupled from
components.' own at t e model only has four independent

Figure 6-22 below tw d £ .

o e ormatIOn casesof the model above are illustrated.

Figure 6- 20 Force displacement graph for model of spring and friction damper in
series, in parallel with a spring, as a function of excitation frequency.

The energy loss per cycle is independent of the frequency w.

6:27
 Advanced vehicle modelling
Rail vehicle dynamics
DampingZ:
Fz r---~ Mcp
-- -- ....--. Z(iw) = Fj[d(xz - Xj)jdt] = C

,-- -
z Fy
I -- -- if!
Fy
FOIce-velocity and force-displacement graphs:
(6-64)

/
/ J /
/
I
I
I
I F
I Mcp {
\
I
{ (
I
\ \ WCXo
I I \ \
I J \
I I

I c
(a) (b) _...I
1
Figure 6-22 Two deformation cases of two-dimensional model for springs. The
bottom support of the spring model is clamped.
(a) y '" 0, Z '" 0, rp = ° - wXo

(b) y = z = 0, rp '" ° d(X2 - xj}/dt

The relation between forces/moment and displacements/rotation can be expressed as

FYj_ [kYY
[F z _
Mcp krpy
° °° krprp
kyrpj[y]
°
k zz Z
if!
(6-61)

where kycp = kcpy in the stiffness matrix has a negative value.

Force-velocity gra h fi l'
See also Appendix 6B for airspring modelling. excitation. p or mear viscous damper model at harmonic

6.3.3 Models for dampers F

Linear viscous damper model

WCXo--:=_-l-__
c

c::::.:> XI

Figure 6 -23 Linear viscous damper model.

- WCXo -"''=---1-_ _- -
Damping c (frequency independent), no stiffness, no mass.
Harmonic excitation with frequency wand amplitude;\fl:

(6-62) Force-d~splacement graph fior linear

harmOnic excitation. viscous damper model at
gives
(6-63) force F and the displacement Xz - Xl are 90 d
X2 - Xl = 0, zero force F = 0 whe egrees out of phase (maximum force F
nxz -Xl :::xo).

6:29
 Rail vehicle dynamics
Advanced vehicle modelling
Model with linear viscous damper in series with a linear spring

The area of the ellipse (6-65)

c k
A = :n;WC(XO)2
. dissipation per cycle.
corresponds to the energl Ie thus increases linearly F F
For a given damping c and amplitude Xa , the energy loss per eye
with the frequency w.

Nonlinear viscous damper model c:>- Xl

Figure 6-28 Model with linear viscous damper in series with a linear spring.
-o-,...--II.....
~ F

Damping c, stiffness k (both frequency independent), no mass.

Harmonic excitation with frequency wand amplitude Xv:
~Xl
(6-68)
Figure 6-26 No nlinear viscous damper model.
d ) 0 stiffness no mass. Transfer functions as stiffness S and damping Z:
Variable damping (but frequency indepen ent , n ,
. wand amplitude Xv :
Harmonic excitation with frequency S(iw) = F /(x3 - Xl) = l/(l/k + l/iwc) = kwc (wc + ik)/(k 2 + w 2c 2)
(6-66)
(6-69)

gives (6-67) (6-70)

F = f(d(x2-xNdtl = arbitrary function of d(X2 - xl)/dt

Example of force-velocity graph (piecewise linear, symmetric): IS(iw) 1

d(X2 - xl)/dt
~----------------+-------------------------------~w
k/c
6-29 Stiffness for model with linear viscous damper in series with a linear
spring, as a function of excitation frequency.

Z(iw) = F/[d(x3 - xl)/dtj = 1 /(l/c + iw/k) = kc(k - iwc)/(k2 + W2c 2)

(6-71)

. d per model at
Force-velocity graph jar nonlinear VISCOUS am (6-72)
Figure 6-27
excitation (example).
6:31
 Rail vehicle dynamics
Advanced vehicle modelling
Model with nonlinear viscous damper in series with a linear spring
Iz iw)1
k
------------------------
F F

c::> Xl
L -________________ -4---------------------------. W
k/c
Figure 6-32 M odeI WIt. h nonlinear viscous damper in series with a linear spring.
Figure 6-30 Damping for model with linear viscous damper in series with a linear
spring, as a function of excitation frequency.
I ness k (b oth frequency independent), no mass.
Variable damping , sfff
For low frequencies the model works as a pure viscous damper model, whereas for high
Harmonic excitation with frequency wand amplitude xo:
frequencies it acts like a pure spring model.
Force-displacement graphs: (6-74)
F F
/ Example of force-displacement graph (simulated):

Force F (kN)
:;ir-----r----
Xo

(b)
F
F

-h1--__------~-----~t..I--- XrXj
/
/
/
/
/ /
/ (d)
(c)
Force-displacement graphs for model with linear viscous damper
Figure 6-31 series with a linear spring, as a function of excitation frequency. g~--~~~~~~--~--~--L-~~~---L--~--~
.1.S .S ·2.5 .() 7.S :Lj
(a) General (b) W = 0 (c) W =0.5 k/c (d) W = k/c
Displacement X3 - Xl (mm)

The area of the ellipse

series with a linear SPri~g (::a:;l:). WIth nonlinear viscous damper in
6-33 Force-displacement gra hfi d I .
A = nwc(x o)2/(1 + (wclk)21
corresponds to the energy loss per cycle.

6:33
Advanced vehicle modelling
Rail vehicle dynamics

Model with friction damper and linear spring in series Appendix 6A G J

enera motion of track- following reference system
N The velocity and acceleration vectors for th .
6-2 and Equation (6-17) can be w 'tt . e track-followmg reference system, cf. Figure
, n en m component form as
IvT, _ Id(l T) ,
- di p a = vel
F (6-77a)

laTa = !.!l.(lvTa) _ Td(l T) I T

dt - dt V a + W X Iv Ta = ae I
+ v(w 3 e2 _ ") _
{J)2e3-
N , ( V2 1 2
ael + -cos- COS _ L . ,v2 1
R X, ((i, R sm((i,)e2 - (-cos 2x sinm
R'
+ Rv
v- cos((i, )'e
3
~Xl ~X3
V 't"

(6-77b)
Figure 6-34 Model with friction damper and linear spring in series.

Friction coefficient,u, normal forceN and stiffness k (all frequency independent), no mass.

Harmonic excitation with frequency (J) and amplitude Xo:

(6-75)

Force-displacement graph (,uN < kx,,):

Fo =,uN

+ Rv + v2(1)'
(.2:. R ) cos((i, - v2
Rv ((i; sin((i,] 2 +e

[(!!:.
+ R + v2(1.)')
R
2 v2 (
COS X,COS((i, - R cosx, ((i; sin((i,cosx, + 1v COS((i,Sin x ,) _
-Xo

- (.2:.
Rv + V2(Rv
1 )') sm((i,
. v2 ((i; cos((i, ]e3
- Rv
(6-77d)

where (1/R)' = d(I/R)/ds etc.

Figure 6-35 Force-displacement graph for model with friction damper in series
with a linear spring.

The enclosed area of the parallelogram

A = 4,uN(xo -,uN/ k) (6-76)

corresponds to the energy loss per cycle. The energy loss is thus independent of the
frequency (J).

6:34
6:35
Advanced vehicle modelling Rail vehicle dynamics

Appendix 6B Ail'spring modelling (6-78a)

(6-78b)
Basic modelling
. . mechanical component and a wide range of airspri?g (6-78c)
The airspnng IS a rather complex . d . . ulations of vehicle dynamics
. I h text of time omam slm I
models may be applicable. n t e co~ . h d th s there is a need to restrict the mode
behaviour the computational effort IS hlg an . u st not be too complicated. (6-78d)
camp IeXI't'y. Therefore , an airspring model for this purpose mu
. dimensional airspring model to 12 basic degrees (6-78e)
A reasonable start is to restrict the three- . t th top (subscript t) and bottom
. d three rotatIOns a e .'
of freedom: three translatlOn~ an I .ons and rotations, along With theu
(subscript b) surface respectIVely. These tran~ ~l. 6 36 (6-78j)
corresponding forces and moments, are shown m 19ure - .
where it is assumed that the X-, Y- and Z-axes are principal axes of the airspring. The
</>t,Mq,t
moment equations (6-78d) and (6-78e) refer to the bottom of the spring. Note that the
Xt,Fxt 7 deformed configuration is considered in these equations since the translations of the
airspring ends are present. The relative vertical displacement -z( + z b (change of airspring
height) is often less than, say, O.OSh and may be neglected in the equations. However, the
influence of the relative displacements X t - xb and Yt - Yb' in combination with the large
compressive force F zt , is significant and gives important moment contributions.

h In an airspring model the forces and moments of Equation (6-78) should now be related to
airspring end displacements and pertinent velocities as well as to the preload P. It is
sufficient, and appropriate, to formulate these relations for the forces and moments of one
of the airspring ends, and then use Equation (6-78) to determine the forces and moments of
,the other end. Here the airspring top forces and moments are selected for the formulation.
l::::::=¢S;;L-----:Y::-b-,F~Yb --:;;MXb the simple case of linear elastic behaviour, it can be shown that the forces and
zb,Fzb,P b of the airspring top are given by:

(6-79a)

Figure 6-36 . rno dIN

Basic quantities of an airsprmg e. 0 minal configuration. (6-79b)

. F'19ure 6 - 36'.
The following airspring quantities can be identified III (6-79c)

• Preload,P (=PI=-P b) (6-79d)

• Nominal height, h
• End displacements/rotations, Xt to VJb (6-7ge)

• End forces/moments, Fxt to M1J!b

(6-79j)
. . lane and straiaht track, the displacements are
For static equilibrium, With the vehicle on a pF _ P _ P ~nd F =P b =-p.
set to zero and the only non -zero forces are zl - 1- zb (6- 79) eight different stiffnesses are introduced. If for instance Y( ;<' 0 and the
displacements are zero, the stiffness ky relates Yt to FYI by the simple equation
The inertial effects of the airspring are assume d negrIgi'ble given that frequency range '= ky . Yt· The stiffnesses kx, kz, krp, kx and k", can be explained analogously. The
main interest is, say, 0-10 Hz. . two stiffnesses, kxx and ky.p, couple the longitudinal force/translation to the
The followmg . eqUi'I'b
. SIX . m conditions should therefore hold during general
I nu rotation/moment, respectively the lateral force/translation to the roll
motions:

6:37
 Rail vehicle dynamics
Advanced vehicle modelling

. ;'" 0 and the other displacements equal to zero, (6-80e)

For instance, the case above wIth Yt . t' oment M = k A. • Yt. At the
F - k . Y but also a resls mg m tlJI Y'Y
requires not only a force yt - y t . . F _ k . Yt and a resisting M1jJ =0 0 (6-80f)
airspring bottom this case also requires a reslstmg force yb -:-. - Y tions 6-78b
_ (k + k h + P) . Y according to the eqUIlIbrIum equa with the three stiffness parameters ky, ky</> and k z.
moment M",b - - y¢ y "
and 6-78d respectively. See Figure 6-37.
0
M ¢t
Advanced modelling, discussion

_ Fyt
! P
The main "problem" is now, however, that airsprings do not have a purely linear elastic
behaviour and the stiffnesses are not independent of the preload. Still it is fruitful to let a
more advanced airspring model turn into the basic model of Equation (6-80) as a special
case.
1>,=0 -r--
I H
I
\
,
I
From the description in Section 3.4.3 the most obvious disadvantage with the basic
I I Yt I
airspring model above is probably the preload independence of the stiffnesses. In Equation
\ / (3-9) the vertical stiffness strongly depends on the preload P = F zoo During a rail vehicle
\ I
I ~\ / h dynamics simulation the preload is seldom changed (unchanged number of passengers in
\ I I '- / each simulation) and the preload dependence must thus not be time dependent. The
'-~~ I -~
I following linear preload dependence seems reasonable
I I
I I
I I (6-81)

F":f where stiffness values are specified for two preloads: stiffness ka for P = P a and stiffness
kb for P = P b' For other preloads Equation (6-81) can be used. The preload dependence
is usually somewhat more pronounced for k z than for ky and ky</>o
M¢b V .' ;'" 0 other
Still the limitation of the linear elastic behaviour of the basic model remains. Qne
Forces and moments at pure lateral shear of an alrsp~tng.'y t ' limitation is the lack of friction representation. From Figures 3-13b and 3-18 we saw that
Figure 6-37 1 0 Actual force and moment duectlOns. , hysteresis occurs even for quasistatic excitation. This also applies to airsprings and for
displacements equa zer .
botb vertical and horizontal motion. This hysteresis cannot be captured by introducing
F d F since F is included in viscous elements in an airspring model since the velocities in question are very small and
Note that the force F zt has to be evaluated before xt an be ~t roxim::ed by F zt = p
bardly give any force contributions. The friction introdnces non-linear spring behaviour.
~quatio~s (6-(769a7)9an)d (~-(~_bi9~)~~hgee~~~~::i:~~;:~~t~~ns inP~quation (6-79).
111 EquatIOns - a an . A two-parameter friction model has recently been suggested by [11] to represent the

For the common Situ~fftion ofaxis~:r::~~~d ~:~;:~~~t

number of dIfferent stI nesses can
. .'
:I/
.th respect to the Z-axIs, the
ive since kx = ky, kx = k", and
"ff an often be
':friction behaviour in rubber springs. The model is also applicable to other spring types like
,"l1"JlIII~'_ which partly can been seen as rubber springs. The two parameters describe the
.rna,xim.um friction force and how "fast" this force is reached, cf Figure 3-13b in which the
h h) I addition the fIve stI nesses c
kxx = - ky¢ (note the minus sign t oug . n _ ~ and k can often be neglected as can be seen as a linear elastic force superposed by a friction force.
reduced to only three since the moments related to k", - x . 1jJ . •r (left and right)
limitation with the basic model above is that it does not have any time
com ared to the roll and yaw mo~ent~ transferred by a~ auspnng pal
der1en(ience. For instance, for a harmonic excitation the model responds independently on
thrO~gh vertical and longitudinal auspnng forces respectIvely. frequency. From Equation (3-9) we see that this is a limitation since the parameter n
.., E I" (6- 79) can be written as: on the frequency. Also, it is obvious from Table 3-2 that most springs have a
With the simplIfIcatIOnS above qua IOn
(6-S0a) dynamic stiffness than static one. Viscous elements can be introduced to cope with
Fxt = ky . (Xt-Xb + hXb) - ky</> . (xt-Xb) +P . Xb
(6-S0b) tbe rubber spring model of [11], viscous effects are approximately considered by a
viscous damper with a series linear spring. For lateral motion the model can be
(6-S0c) rem·o.'o.ntorl as in Figure 6-38.

dynamics of an airspring mainly depends on the pressure and volume of the air
(6-S0d) also on the air flow resistance in the surge pipe, cf Section 3.4.3. A well-known

6:39
6:38
Advanced vehicle modelling

attempt to model this behaviour is shown in the leftmost part of Figure 6-39, where also
the friction model referred to above has been introduced.
In Figure 6-39 the vertical stiffness is represented by a (linear) air bag spring in series with
a (linear) reservoir stiffness. The viscous damper may be set linear provided there is no
significant air flow resistance. For high resistance the damper is non-linear, preferably the
damper force is set proportional to the square of the velocity over the element.
ky

2 friction parameters

cy kyv

Figure 6-38 One-dimensional model of a physical spring considering elastic,

friction and viscous behaviour [11}.

In Figure 6-39 the possible effect of "change of area stiffness", cf Equation 3-4, has been
included. This stiffness may be negative though. Also an emergency spring has been
introduced.
Aspects not considered in the figure are cross-piping, i e connected airsprings in a pair (left
and right) and possible dynamic effects of the levelling valve.

Air bag
stiffness
Friction Change
of area
stiffness &
damping stiffness

Reservoir Reservoir
stiffness damping

Figure 6-39 Example of model for vertical airspring motion.

h:40
 Rail vehicle dynamics

7. Wheel-rail guidance mechanisms

In this chapter lateral guiding mechanisms of a wheelset are discussed. In Chapter 8
equations for so-called creep and creep forces are introduced, and the dynamic behaviour
of a wheelset considering tangential forces between wheel and rail are shown. The two
following chapters are thus describing many of the phenomena which are unique for
dynamics of railway vehicles. They are crucial for the understanding of the dynamic
vehicle-track interaction. Sometimes this special part of railway vehicle dynamics is
called lateral dynamics.

In Section 7.1 we concentrate on the geometric interaction between wheel and rail. Section
7.2 describes kinematic guiding mechanisms for the system wheelset-track on straight and
curved track.

7.1 Wheel-rail geometry

Wheel and rail geometry should always be regarded in combination with each other,
because it is the interaction of the wheel and the rail profiles that determines the guiding
performance of the system. Wheel and rail profile means the geometric shape of the wheel
and rail surface.

7.1.1 Rail profiles and rail inclination

Today's rail profiles are shaped to fit together with the shape of the wheels, especially
when they are worn in. A widely used rail profile in Europe - the UIC 60 profile - is shown
in Figure 2-3. In most cases the rails are mounted with an inclination inwards because also
the wheel profiles are inclined (coned), see Figure 7 -1. The amount of inclination however
varies between 1:40 (for example Germany) and 1:20 (for example France). Sweden is an
exception with inclination 1:30. In some turnouts rails are mounted without inclination.

rail inclination y,-

~- ~
I
Figure 7·1 Rails are llsually mounted with an inclination to fit to the COiled wheel
profiles. Rail inclination is defined either with the inclination angle y, or
with inclination ratio arctan y,.

7:1
 Wheel-rail guidance mechanisms
Rail vehicle dynamics
7.1.2 Wheel profiles
r----------1351---___~__ ;..I
~-----70--______~~~1
The wheelset is the basic design element in a railway vehicle. Each wheel has a wheel _____
~--------55---- ~~
tread, where the wheel usually is in contact with the rail, and a flange. The task of the
flange is to improve the lateral guiding when this becomes necessary. The two wheels are
normally connected to each other with a rigid axle, so that both wheels always have
8, 1:15
essentially the same rotational velocity. In reality, of course, there exists a wheelset
torsional stiffness, thus a certain torsional deformation between left and right wheel can B - --- - --- -
A
---...Y
occur. The typical frequency of the torsional eigerunode of a wheelset is 50 - 80 Hz.
Usually this can be neglected for low frequency dynamics.

As mentioned above, wheel profiles are usually more are less conical. The reason for this is
that conicity provides steering performance of a wheel set, which will be described in
Section 7.2. Wheels with a straight cone from the start, will soon get a worn profile as
shown in Figure 7-2. Aj : y = + 60 Ej :
z = - 2.636 y = - 38,427 GM : y = - 55
2 = + 12
z = + 16
Bj : y = + 32,158 Fj : y = - 39.764
z = - 0,780 HI: y = - 62.765
z = + 15.675
wheel tread wear z = + 25.149
worn wheel profile Cj : y = - 26 FM : y = - 58.558

'"
2=+2,741 Hz: y = - 70
z =+ 8.835
Lly z = + 9.519
Dj : y = - 35
new profile /
z = + 6,867
Gj : y = - 49.663 HM : y = - 49.5
straight cone 2 =+ 26.748
z = + 9.519
DM : y = - 26,211
z = + 16,446
flange wear
Figure 7-3
Standardized VIC/ORE S1002 wheel profile.
Figure 7-2 After a certain running distance, wheels get a typical "worn shape".

Contact geometry
This "worn shape" is rather typical and differs usually only with some tenths of a
millimeter for different wheels, at least from the middle of the wheel tread to the steep part A contact area arises between wheel and r a i l . . .
of the flange. The "worn shape" is also rather independent from the initial shape of the the contact area, and not too small th . If radl.1 of ,:heel and rail are constant along
profile. After about 100 000 - 200 000 km running distance the wheel shape gets fairly about 1-2 cm 2. The calculation ft'h e contact area IS elhptical. The total contact area is
stable. the present chapter however w 0 e contact area IS describ d· th "
It . e III e >ollowing chapter. In
middle of the area. e re a e to a contact pOint which is thought to be in the
To minimize wear most European countries today use wheels which already from the start
have so called "worn profiles". Such wheel profiles have been standardized by different The location of the contact point at each moment
railways and mc. A typical example of such a wheel, the UIC/ORE S1002 can be seen in on the relative position between wh I d .1 depends on the shape of the profiles and
ee an ral. In particular reI t· I .
Figure 7-3. 1011 and yaw are of interest. ' a Ive ateral displacement,

Similar to wheels also rails have profiles adapted to wear. With worn profiles on wheels On wheels and rails with "worn profiles"'he contact .
and rails the inclination of the cone will vary depending on where wheel and rail get into the flange when the wheel is d·"1 d I pomt usually moves continuously
displacement of the same con~sp tace. aterally towards the rail. With such a
contact. When the wheel flange is near the gauge corner of the rail, the inclination becomes
the right side of Figure 7-4. ac pomt we always get one-point contact, like
higher than if the contact is between wheel tread and rail head, see Figure 7-4.
other wheel-rail geometries the contact oint c .
Shape and wear of wheel profiles influence to a high degree a vehicles steering capacity For certain relative dis lacement p an su?denl~ Jump from one location to
and dynamic behaviour. It is thus one of the most important parameters for lateral same time. In this case we a~ talk· sbwheel and rail get Illto contact at two points at
dynamics of railway vehicles. of Figure 7-4. mg a out two-pOint contact, see example on the left

7:2
7:3

Wheel-rail guidance mechanisms
wheel profile not adapted
. for wear, e.g. with straight
cone
Rail vehicle dynamics
7.1.4 Contact geometry, contact point parameters
wheel profile with "worn shape" (e.g.
UlC!ORE S 1002), on rail adapted to
When a wheel is displaced in the lateral direction relative to the rail, rolling radius, contact
wear (e.g. UIC 60, inclination 1:30 or
angle and the vertical coordinate of the contact point will change, as in Figure 7 -6.
1:40)

wheel
wheel
no flange contact running ;
circle "I

flange contact ~ \ Figure 7-6 Definition of contact point quantities: Change of rolling radius LI r,
contact angle y, and wheel rise LIz. The three quantities depend on lateral
~
../
one-point contact displacement Lly. Rolling radius in running circle rD'
two-point cont::-/ -
(d)
·For a given wheel-rail combination three so-called contact geometry functions, or simpler
(b) 'f
h tw n wheel and ral . dontact functions, can be defined:
Figure 7-4 Different types of contact e ee
(7-1a)
Lateral clearance .
. dis lacement of a wheel until flange contact IS y = fz(Lly) (7-1b)
Lateral clearance means ~he posslb~e Ph definition when the nominal flange co~tact
reached. Flange contact eXists acC~rdl~g ~f. t e dius has increased with 10 mm, see Figure
oint is reached, i.e. when the nomlUa ro mg. ra. . LIz = NLly) (7-1 c)
~ -S. Also the flange thickness, ft' is defined III thiS pomt.
These functions are in general strongly nonlinear. If the contact point suddenly jumps as
before, also the contact point functions get discontinuous.
. I \, lateral clearance
running C\fC~

is defined as the quotient of rolling radius change and relative lateral

'----:=-.J...........o::::l::--t----r-110 mm

noLnal point of flange contact I (I (7-2)

hi d' I cement of a whee un I
"Lateral clearance" means the possi e ISP a
Figure 7-5
flange contact is reached. . h ht to be in a centered position, steering mechanism of a wheelset however is not due to conicity or change in rolling
As starting point the wheelset IS t oug . for one wheel, but due to the difference in rolling radii between right and left wheel.
i.e. the lateral clearance for both wheels IS the same. wheelset is moved laterally from a centered position, a difference in rolling radii
see also Figure 7 - 7,

(7-3)

(7-3a)

(7-3b)

7:5
 Wheel-rail guidance mechanisms
Rail vehicle dynamics

Aeq
0.50
0.45
0.40
0.35
0.30
0.25 UIC601:40
0.20
0.15 Swedish rail, BV 50, 1:30
UIC601:30
0.10
0.05 UIC601:20

O.OO~---':=========;~=;;;::===~;;:::;;:=::::::::==~~~~;;~
1430
1435 1440 track gauge [mm] 1445
Figure 7-8 E' I .
Figure 7-7 Lateral displacement to the right leads to increased rolling radius on the qUlva ent comcity for VIC/ORE SI002 wheel on different rails.
right wheel, while rolling radius for the left wheel decreases.
Cpefticient of roll angle

Equivalent conicity The coefficient of roll angle, r, for a wheelset is defined as

Dividing Equation (7-3) by My, we get a dimensionless measure which is called r= _ Cfiw = Llzr - Llz[
equivalent conicity Lly Lly . 2b o (7-5)

(7-4) ~~::~:wr~~~~~~~~I~n~~ ~~::theelset, and where &r and &1 are the vertical elevations
Like the three contact functions defined above, equivalent conicity usually is a nonlinear Gravitational stiffness and coefficient of contact angle difference
function, especially for track gauges smaller than nominal gauge. In Appendix 7A the
difference of rolling radii and equivalent conicity for different gauges for an VIC/ORE The wheelset in Figure 7 9 h .. .
. d - as a pOSItive lateral dIsplacement. Also lateral displacement
S1002 wheel and a VIC 60 rail is shown. For wheels with straight cones Aeq =y. .• :'c~~~~~to~~gl~S a~e assumed t~ be small so that the vertical components of the normal
w ee s are approxImately equal to the static wheel force Qo.
In general Aeq is a function of wheel and rail profiles, wheel inside gauge, flange thickness,
rail inclination, track gauge and relative lateral displacement. Large wheel inside gauge (> This leads to N.r = Q /cos Y d N. Q /
'orces b I l o r an I = 0 cos YI. The lateral components of the normal

1360 mm on normal track gauge) and large flange thickness ( > 32 mm ) have the same l' can e ca cu ated to
influence as smaller track gauge, i.e. the lateral clearance decreases. The contact point
YNr = NrsinYr = QotanYr
moves to the flange for smaller lateral displacements. (7-6a)
In Figure 7-8 the calculated equivalent conicity for a lateral displacement of ±2 mm for a
VIC/ORE S1002 wheelan different rail profiles is shown. As it can be seen rail inclination (7-6b)
has a significant influence on the conicity. The large difference in conicity comparing the gives a resulting appro~imate lateral force on the wheelset because of the different
new and the worn Swedish rail profile is also interesting. For the worn profile the conicity angles on the left and nght wheel. The resulting lateral force on the wheelset can be
is very high. It is rather typical that the conicity gets higher on worn rails. This has to be as
taken into account when designing vehicles, for example in stability analysis (cf. Chapter
8). Of course even worn wheel profiles can change conicity significantly. LYN = -Qo(tanYr- tanyJ (7-7)
In VIC 518 [N28] another definition of equivalent conicity is given. The wavelength of the
differenlce ?f contact angles is often written with help of the so-called coefficient OF
kinematic yaw of a given wheelset on a given track at an amplitude of lateral movement of ang e difference Y
± 3 mm is measured or calculated. With the Klingel Equation (7-10) the conicity of a
tapered profile wheelset whose transverse movement has the same wavelength of K = tanYr - tanYI
kinematic yaw as the wheel set under consideration is determined. My (7-8)

7:7
 Rail vehicle dynamics
Wheel-rail guidance mechanisms

7.2.1 Sinusoidal motion of wheel set

Substituting this into Equation (7 - 7) leads to
I Y N = -2Qo . K • Lly
(7-9)
Note that the resulting lateral force always acts in the o~posite . direction as the
displacement t1y. This means that a wheelset w~th successively lllcre~sl~g wheel contact
angles has the tendency to centre itself on the raIl because of the gravitatIOnal load (2Qo)'
Therefore this effect is called gravitational stiffness.
Imagine a wheelset running on tangent track with an initial lateral displacement, as in
Figure 7-10. In this case the rolling radii of the left and right wheel differ from each other,
because of the coned wheels. As said above both wheels have the same angular velocity,
which means that the wheel with the greater rolling radius will run a longer distance than
the other wheel. This forces the wheelset to yaw and to roll back to the centered position.
Because of the yaw angle the wheelset will pass the centered position and the rolling radius
of the other wheel becomes greater. This process continues.

------- --- y

/
YN1
Figure 7.9 Different directions of normal forces on right and left wheel for a
lateral displacement.

7.2 Guiding mechanisms

Figure 7·10 A wheelset's steering capability on tangent track.
In this section we are going to discuss the basic steering mechanisms of a wheelset on the Sinusoidal or hunting motion for wheelset with coned wheels.
track, both on tangent track and in curves. To study the principal behaviour we make some
simplifying assumptions: Following the centre of gravity of the wheelset we would observe an almost sinusoidal
only one non-suspended wheelset moves along the track with constant speed )J lateral motion of this point around the centreline of the track. Klingel and Boedecker
derived in the 1880's an equation which describes the sinusoidal motion - often called
_ no sliding between wheel and rail exists. hunting motion - of a non-suspended wheelset [42]. The wavelength Lw is described by
_ the mass of the wheelset is neglected.
_ the wheelsets have straight cones and no flanges. L -?
w -Dr V[b;"-:-;:;,
---;:---..1.- (7-10)
_ there is a stiff connection between the two wheels, which implies that they roll with the
same angular velocity. where
Such a wheelset with stiffly connected wheels has a selfsteering capability both on tangent bo = half the lateral distance between contact points
track and in curves.
ro nominal wheel radius
. 1. conicity or equivalent conicity, see Section 7.1.4.

It can be seen that an increasing conicity leads to a smaller wavelength. With a constant
forward speed v of the wheelset the higher conicity leads to higher frequency f of the
sinusoidal motion, because of f = v/Lw .

7:9
7:8
 Wheel-rail guidance mechanisms
Rail vehicle dynamics
Understanding Klingel's equation We now assume
A single wheel set is rolling along a straight track at constant speed Vo. The wheels are
assumed rigidly connected through the axle, i.e. both wheels have the same rotational y == Yo sinwt
(angular) speed (7-14)
Equations (7-13) and (7-14) then give
(7-11) . _ A
1jJ - - b7voYosinwt
We also assume that: o 0
(7-]5)
- The conicity does not vary with lateral wheelset displacement, i.e. we have straight
coned wheels.
(7-]6)
- The wheelset is rolling without sliding at the contacts with the two rails.
- Inertia (mass) forces are neglected. if the integration constant is set to zero. At pure rOiling we also have

The assumptions on pure rolling without sliding and negligible inertia forces, are Y == vosin1jJ = vo 1jJ
reasonable for a free whee/set at low speed. (For higher speeds the oscillatory frequencies (7-17)
become higher giving larger inertia forces, which in turn usually cause sliding at the From (7-14) we get
wheel-rail interfaces).
Y == Yo W cOS(Ot
(7-]8)
Equations (7-17) and (7-18) give

1jJ == Yo ~coswt
o
(7-19)
I \' . The (7-16) and (7-19) result in

,i ___________ _ ,I\_~_v_o (7-20)

Vo \

/', \
(7-21)
= To + LIT = To + A . y I' wavelength of the oscillatory motion then becomes

$<---'--+-------\- - J-
(7-22)=(7-10)
form~la is thus called Klingel's formula I '.
to reallZe that (7-13) and (7-17) are valid. n the denvatlOn above the important steps
Figure 7-11 Wheelset with lateral displacement y respectively yaw angle 'IjJ.

For pure rolling without sliding the speed of the hub of the two wheels are, cf. Figure 7-11,'

V2 = [J(r 0 - A . Y)

7:11
Wheel-rail guidance mechanisms Rail vehicle dynamics

Frequency of sinusoidal
7.2.2 Sinusoidal motion for wheelsets stiffly connected to a frame
motion (Hz)
The Klingel equation above can be extended to two wheelsets in a stiff bogie or vehicle Aeq = 0.50
10
frame with a totally stiff suspension in the horizontal plane, as in Figure 7-12 Free wheel set
bo =0.75m
8
ro = 0.5 m
;'eq = 0.25
6

O~--~--r--'--~--_ speed (km/h)

----- o 150 200
Frequency of sinusoidal
motion (Hz)
Wheelsets with totally stiff suspension in a bogie or vehicle frame. The
Figure 7-12 10
wavelength of the sinusoidal motion increases compared to a free Stiff suspension
wheelset. bo = 0.75 m
8
ro = 0.5 m

Lo,sliff = 2n jbo;.ro j 1 + (:}

(7-23)
4
6

~ Aeq =0.50
~Aeq=0.25
Thus, a bogie or vehicle frame with two stiffly connected wheelsets will show a longer 2
wavelength and lower frequency of the sinusoidal motion. ~Aeq=0.10
Aeq - 0.025
In Figure 7-13 the frequency of the sinusoidal motion according to Kingel's equations o speed (km/h)
with respect to vehicle speed and conicity, is shown. In reality the frequency often is o 50 150 200
somewhat lower because of the additional mass forces, which lead to increased sliding Frequency of sinusoidal motion accord' .
between wheel and rail with increased speed, which is indicated with the dashed lines in " free wheelset andfor stiff . mg to Klmgel's equations for a
Figure 7 -13. However, a real wheelset suspension is never totally stiff. Therefore, the true speed and conicity. suspensIOn respectively, asfunction of vehicle
frequency of the sinusoidal motion will be found between the frequency for a free wheelset" In reality the frequency often is somewh I
and the frequency for a totally stiff suspension. mass forces which lead t ' d a~ ower because of the additional
. 0 mcrease shdmg b tw h
mcreased speed which . . d' e een w eel and rail with
Klingel's equations are often a rather good approximation of the real wavelength, in diagram. ' IS m Icated with dashed lines in the upper
of all simplifying assumptions. However, they can not explain the phenomenon
initial amplitude of the sinusoidal motion decreases for speeds under a so-called
speed, and increases for speeds above. To explain this, mass forces and contact
(especially tangential forces) in the contact point have to be regarded, which implies "Conicity andsmuslo
' . d al motions depends on the wheel-rail combination

one has to look on the differential equations of a wheelset. This will be done in mdlcated the sinusoidal motions are fun
lateral dynamic behaviour on the trac~a%ental .for th~ ~heelse~ steering and
of lateral dynamic stabil"ty . . particular It IS very Important for
84 A . I , I.e. whether a motion is d d
" quite significant part of the w k ' . ampe or not; see further
equivalent conicity is here a' or on raIl vehicle dynamics deals with these
. . n Important and sometimes determining parameter
com city and sinusoidal motions are i . .
~urves. In sharper curves some of t~p;rtant o.n strat.ght track and on wide
bogies) are more or less p d ' heels (In partl.cular the leading outer
. . resse qUite hardly t d h
smusOldal motions is very limited. owar s t e outer rail; thus the

7:13
 Rail vehicle dynamics
Wheel-rail guidance mechanisms

, d d' on geometrical Excessively high conicities may be produced by the following circumstances:
, , within a wide range epen Illg
The equivalent COlllClty ~ay vary he hi hest conicity in normal operation is often - Small track gauge
conditions in the wheel-rail Illterface, T 'big icity This is clearly seen from the
- Wide wheel gauge
10-20 times higher than the lowest POS~I e ~nh I ~nd rail profiles, rail inclination,
examples shown in Figure 7 -8, The exa~t ~h~~~~s; a:: determinining the value of the - Low rail inclination (including zero inclination)
track gauge, wheel gauge and [lang - Thick flanges
equivalent conicity in each actual case, ,
" ro erl (for example maintaining dynamic Hollow worn wheel profiles
Rail vehicles must be designed III order t~ ru~ p ~ I/range of possible conditions in the - Flat top surfaces on rails (built in or due to wear)
stability) in the whole spe~d range,wlthlll ~,~:, ~ to design running gear that is able to
wheel-rail interface, PractIcally It IS ?ften I, ~~u t h'gh speed Sometimes it is also Very low conicities may be produced by the following circumstances:
, b'I' t t 0 high COlllCltIeS a I , "
maintain dynamiC sta I Ity a 0 , ' , Th's wI'll be further descnbed III - Wide track gauge
"
difficult to mamtalll stab'l't
I I Y at v ery low COlllcltJes, I
- Small wheel gauge
Section 8A,
, , a nonnal praxis to maximise the equivalent conicity due to the - High rail inclination
In Europe It IS now
difficulties mentioned above [NlO][N2S]: - Thin flanges
It is, of course, important to make a proper selection of all these parameters, A close
Aeq ;$ 0.50 for speeds up to 140 km/h
co-operation between track and vehicle engineers is necessary, To a certain extent, but far
for speeds above 140 km/h and up to 200 km/h fipm completely, these parameters are regulated in national or European standards,
Aeq ;$ OAO
for speeds above 200 km/h and up to 230 km/h Normally a straight coned wheel profile (see Figure 7-2) will worn in to a typical worn
Aeq ;$ 0,35
hollow profile within 100 - 200 000 km of running distance, However, the worn profile
Aeq ;$ 0,30 for speeds above 230 km/h and up to 250 km/h may in some cases be at least slightly dependent on the initial profile, A difference of (say)
for speeds above 250 km/h and up to 280 km/h D,S mm in the wheel (or rail) profile coordinates may produce abnormal conditions. This is
Aeq ;$ 0,25
important to know, for example, when reprofiling wheels in maintenance workshops or
Ae ;$ 0,15
for speeds above 280 km/h and up to 350 km/h grinding operations on rails.
q " 't mmon if the track
T d above occur ThiS IS qUi e co
Sometimes higher values than speci I: 1430 mm fo; standard gauge track) or the, wheel
Determination of equivalent conicity in non-linear cases
gauge is very small (for example belo ' I thO 's the case if the wheel profiles are
h' k 's too large In partlcu ar IS I 2 I
gauge or flange t IC ness I " fl t shape' see for example Figure 7 - , n A smaller track gauge produces less lateral wheel-rail clearance and a faster increasing
worn to a hollow shape ~r the raIl he~~ wo;~ ~o \h~ut any inclination inwards, which may rolling radius difference if the wheelset is moved laterally within in the track. This is
some European track sWltche~ t~e ra! are ai, w~ 'tches, Examples of conditions . clearly seen in the example of Appendix 7A.
Iso roduce excessive COlli city locally III t ese S":'I ,
a P, I ' 't ill be further discussed III SectIOn SA, Up till now we have been dealing with conicity and eqivalent conicity as a single value,
producmg abnorma COlllCI Yw
" ' ' , d in the EuropeaIl regulations the extremely for all given by the wheel-rail geometrical interface,
Although equivalent COlllClty IS now max:m~~ 7 -8 very low coni cities are shown at
most cases the contact functions, in particular the rolling radius difference !J.rr-I, is
low coni cities m:e not r~gu~ated ,at all., n la:~~:r e track gauges, The same may happen non-linear when the wheel flanges approaches the gauging faces of the rails. The
certain wheel-rail combmatlOns, III partIcu ' Ig lateral wheel-rail clearances, The
ea,U/vlliP"t conicity is therefore normally dependent on the amplitude of displacement, as
with worn wheel flanges which also producefqUlt~ a~g:igned and tested also for very low
vehicles and their running gear should there ore e e the second page of Appendix 7A, This is at least the case when hollow worn wheel
conicities, although this is not yet regulated: are used (or the corresponding "worn-type" profiles as UIC/ORE 51002). It is
not correct to assign a single value of the equivalent conicity because it is
Aeq 0,01 to 0,02,
2: . «lU,~llLl.U" dependent. This complicates the matter quite a lot.

, ' h I-rail interface have already been briefly

The influence of certam parameters ,m the Ws e: t 'mportaIlt influence will be further the equivalent conicity is used for characterisation of the wheel - rail geometrical
indicated above, With reference to Figure 7 - t e mos I conUlllOllS with regard to sinusoidal motion and dynamic stability on straight or slightly
listed and described below, track. This means that the wheelset is oscillating quite closely to the centre of the
with plus-minus some millimeters. Therefore equivalent conicity should be
determined for some millimeters of oscillation amplitude, for example ± 3 or ±4 mm,
whatever is anticipated to be the typical amplitude of the sinusoidal motion in relation to
the track, Below is described some more or less complicated methods to determine the
equivalent conicity for defined lateral oscillation amplitudes:

7:15
7:14
 Wheel-rail guidance mechanisms
Rail vehicle dynamics
1. In the simplest case the conicity is determined and expressed according to Equation
7 -4, simply by taking the rolling radius difference at the defined desired amplitude, for
example L'l.YO =±4 mm. lfthe nonlinearites within this range is considerable this method
will not produce representative values of conicity.
2. A more accurate method is to take the average of conicities (determined as in 1) over a
=
number of wheel-rail displacemnents, for example L'l.y 1, 2, 3, 4mm, up to the desired
maximum wheel-rail amplitude L'l.yo. This method is still quite simple but produces
quite reasonably a representative value of the conicity for the desired amplitude in most
cases. A further refinement of this method would be to make a linear regression
R
applying the method of least squares.
3. A third and more complicated method is proposed by the UIe [N29]. The basic
principle is to determine a single value of equivalent conicity producing the same
wavelength of the sinusoidal motion for a single free wheelset. Calculation of the
wavelength is based on the same fundamental Equations (7 -12) and (7 -13) as Klingel's
equation. However, in this case the calculated rolling radius differences (non-linear
with respect to L'l.y) are applied. This calculation is made by numerical integration.
When a wavelength has been calculated, Equation (7-10) is used to determine the
equivalent conicity. The calculation procedure is shown in detail in Appendix 7B.
All simple or advanced calculation methods above produce a single value of the equivalent
conicity which is dependent on the desired and assumed oscillation amplitude. Therefore,
a value of the equivalent conicity must be followed by the assumed amplitude L'l.yo.
It is here necesary to emphasize that non-linear relations can not be correctly represented
by linear approximations and single scalar values. This is an attempt to approximately
Free wheelset with straight cones in a
describe complex relations in the wheel-rail interface. A higher degree of precision in Rollmg r d' fi h curve.
calculations requires the correct contact functions to be calculated and used in simulations. a !Us or igh (outer) and low (inner) wheel.
This is normally made in time step simulations.
With the assumpt.i~ns made in the beginnin of Secti .
However, the equivalent conicity is fundamental and is very often used, both in the basic 5 CL'IIlt:m,cal condItIOns which must exist t g t o~ 7.2, we are now lookmg into the
" written or per ect radIal steering (cf. Figure 7-14). ltcan
theory as well as in practical work, standards and specifications.

7.2.5 Curving performance

(7-24)
Since a free wheelset is steering into the centre of the track after it has been displaced from
the equilibrium position, it has also the capability to steer in a curve. In principle it will be
displaced towards the outer rail (high rail) as much as needed so that the difference in
rolling radius corresponds to the difference in travelled length on high and low rail. (7-25a)
We assume that the centrifugal forces, projected into the track plane, can be neglected.
This can be achieved either by low speed or by running at balanced speed. r;fl = ro - Llr

If the wheelset is allowed to steer freely around the curve, it will run radially with respect' (7-25b)
into Equation (7-24) we get
to the curve radius. Radial steering means that the wheelset axle always points to the centre
of the curve. Such a wheelset is said to have a radial self-steering capability.
(ro + Llr) (R - bo) = (ro - Llr) (R + bo)
(7-26)
forward calculation gives

Llr bo
r;; = R
(7-27)

7:17
 Rail vehicle dynamics

Wheel-rail guidance mechanisms

Appendix 7A Exa.~ple of rolling radius difference and equivalent
(7-28) comclty
2 bo
rout - rin:= rolf

Example: For wheels with straight cones the conicity A becomes the same as the cone Difference in rolling radii
inclination tan Yo = Yo. and Llr =yaL!y. We get Ilrr-l =rr - rl (mm)

Lly '" 1'0 . bo

Yo R
With the following data: 1'0 = 0.5 m, Yo = 0.05 (inclination 1:20) ,
bo = 0.75 m and R = 500 m it is found that

Lly '" .
g~5 ~'6~ =0.015 m =15 mm
which means that 15 mm lateral displacement is needed to achieve perfect
radial steering in this case. lfthe cone inclination is Yo'" 0.25 instead, a lateral
displacement of 3 mm would have been sufficient for perfect steering.

The example illustrates that a greater conicity or cone inclination improves the curving
performance of the wheelset. Therefore a wheel profile with continuously increasing
contact angle towards the wheel flange is preferable. In this case the difference in rolling
radius can be great and the wheelset can steer radially in relatively tight curves without full

flange contact.
In the example above a difference in rolling radius between outer and inner wheel of 1.5
mm is needed for perfect steering. With the wheel-rail combination in Appendix 7A at a
gauge of 1440 mm (which is rather typical in curves) a lateral displacement of about 7 mID
would be needed to achieve this difference in rolling radius.
The steering performance of a wheelset in a curve depends not only on the geometrical
conditions, but also on the friction coefficient. For low friction (I-t = 0.2-0.3) the wheel has
the tendency to "slide down" from the gauge corner of the outer rail so that a sufficieut ;
difference in rolling radius cannot be obtained. On the other hand for high friction ( !-l ='
0.5 -0.6, dry and clean railS) a relatively great contact point angle and a high difference in
rolling radius can be achieved.
This section is concluded with the following statement:
On dry rails and with wheels and rails in "worn shapes", free or softly
suspended wheelsets on normal gauge usually achieve sufficient difference in
rolling radius for radial steering in curves down to 140-200 m radius, without
Rail: VIC60 (1:40)
flange contact.
In real vehicles the (longitudinal) stiffness of the wheelset suspension often makes Track gauge =1428 - 1444 mm
Wheel: VIC/ORE S1002
impossible.

7:19

7·1R
 Rail vehicle dynamics
Wheel-rail guidance mechanisms

Appendix 7B Calculation of equivalent conicity for nonlinear

3. 4. 6. 8. (mm) contact functions
2.
Track gauge = 1428. 0.76 0.93 1.77 1.50 1.26
(mm) 1430.
1432. - -0.~5 Q.,-~8_.,
- - ___ a 0.73
28 ' 0.4 6
11'2410
. .'
i·g ~1eq
___ Jl
Starting from Section 7.2.1, Equations (7-12) and (7-13), which among other things
assume no sliding at the wheel-rail interfaces. Instead of assuming straight conical wheel
0.23 . --6.-23--; 0.50 1.05 treads and thus a constant conicity A, the wavelength of the sinusoidal wheel set motion is
1434. 0.19 0.19 '0.32 0.91
1436. 0.16 0.15 0 .15 Lo-.-f6-LO_.~~ __ now calculated based on tm contact functions, i.e. nonlinear functionsLl rr-/ as function of
1438. 0.11 0.12 0.12 0.10 0.24 lateral displacementomLly between wheel set and track. The calculation of the wavelength
1440. 0.07 0.08 0.10 0.08 0.12 uses numerical integration. Finally the equivalent conicity Aeif is calculated from the
1442. 0.05 0.05 g:g~ 0.07 0.08 Klingel's formula (7 -10) for a single wheelset.
1444. 0.03 0.04 . 'ty)
Aeq (equivalent COnlCI First we substitute y = Ll y in order to simplify the expressions below.
on
In analogy with Equations (7-12) and (7-13) we get
I"

(1)

Also from Equation (7 -17) we have

. dy
y = dt = vo 1/J
(2)

. dy dy dx dy
y = dt = dx . dt = dx . Vo
(3)

... last two equations give

dy
Track gauge dx = 1/J (4)

L--l----~-+-I--"'rT( 1428, rum

LJ ~ 1\ (5)

Equations (1) and (5) now result in

d2y Llrr _ 1
dx 2 = 2horo (6)

(6) is integrated twice with respect tox. The initial values are set to: y =Llyo and dy/dx
O. As a result we get a periodic function, in principle as in the figure below, with
4·

-18 -1.' Lly (rum)

Rail: UIC60 (1:40)

Track gauge =1428 - 1444 mm x
Wheel: VIC/ORE S1002

Calculated oscillatory motion based on nonlinear contact functions.

7:21
Wheel-rail guidance mechanisms

Finally the conicity A =Aeq is determined from Klingel's formula (7-10), thus

and consequently

(7)

The calculated equivalent conicity depends on the assumed displacement amplitude Llyo,
exemplified below.

, ,
0.8 r - - - - - - r - - - - - - r - -

0.6

0.4
0.2
OL---~----~--~--~ Llyo (mm)
o 2 4 6 8

Figure 7-16 Two examples on calculated equivalent conicity as a function of

displacement amplitude ~Yo·

7')).