Tese

Instituto Superior Tcnico Univ.
Tcnica de Lisboa
Escuela de Ingenieros de Caminos Univ. Politcnica de Madrid
Dynamic Behaviour of High Speed Railway
Bridges. Vehicles Lateral Dynamic
Behaviour
Rui Filipe da Silva Dias
Dissertation for the degree of Master of Science in

Civil Engineering
Supervisor: Prof. Jos M. Goicolea

Co-supervisor: Prof. Jorge Proena
November 2007
Dynamic Behaviour of High Speed Railway Bridges. Vehicles Lateral Dynamic Behaviour
Dissertation for the degree of Master of Science in Civil Engineering

Instituto Superior Tcnico
Universidad Politcnica de Madrid
Madrid, November 2007
Rui Filipe da Silva Dias
Supervisor: Jos M. Goicolea

Dr. Ingeniero de Caminos, Canales y Puertos
Escuela Tcnica Superior de Ingenieros de Caminos, Canales y Puertos
Departamento de Mecnica de Medios Continuos y Teor a de Estructuras
Co-Supervisor: Jorge M. Proena

Dr. Engenheiro Civil
Departamento de Engenharia Civil e Arquitectura
Universidad Politcnica de Madrid

Profesor Aranguren s/n
Madrid 28040

Av. Rovisco Pais
1049-001 Lisboa
Phone: +351 962438617

E-mail: rui@mecanica.upm.es
web: http://w3.mecanica.upm.es/ rui
Abstract
This work begins with the study of the assumptions and requirements made by the
new engineering codes, EN19912 and EN1990A2. Additionally, the calculation
models available for the evaluation of dynamic eects due to travelling loads in
railway bridges, were reviewed.
In order to study a particular railway bridge and put in pratice some of the
methods reviewed, the vertical dynamic analysis of the "Arroyo las Piedras" high
speed railway viaduct, was performed and described. This bridge, due to the sin-
gularity of its structure, presents relevant lateral dynamic displacements, leading
to the study on the inuence of these eects, among others, in the lateral dynamic
behaviour of railway vehicles. In order to study this phenomena two other problems
were outlined: track irregularities and eects due to wheel-rail contact problems.
Dierent vehicle models were developed and the dynamic analysis performed, con-
sidering the actions of those eects and the following models: a model with one
degree of freedom, a model with two degrees of freedom and a three-dimensional
model. The methodology was fully developed by the author.
This study led to the conclusion that the bridge does not has signicant vertical
dynamic eects, being the ultimate and serviceability limit states veried. In terms
of vehicle lateral dynamic behaviour, considering this study as a rst approach to
the problem, it was concluded that the safety and the comfort of the passengers are
ensured. However, further studies should be performed.
This work was developed within the scope of the cooperative protocol signed
between Instituto Superior Tcnico and RAVE - Rede Ferroviria de Alta Veloci-
dade, S.A.
Keywords: Lateral vehicles dynamic behaviour, track irregularities, virtual path,

ctitious vehicles, Klingel movement eects
i
Sumrio
Este trabalho inicia-se com o estudo das hipteses e exigncias normativas adop-
tadas pelos novos cdigos de engenharia EN1991-2 e EN1990-A2, para estruturas
ferrovirias. Adicionalmente, foram revistos os modelos de clculo disponiveis e cor-
rentes, necessrios para efectuar uma anlise dinmica de pontes ferrovirias quando
sujeitas a cargas mveis.
Com o objectivo de pr em prtica alguns dos mtodos revistos, ser efectuada
e descrita a anlise dinmica do viaducto "Arroyo las Piedras". Devido singula-
ridade da sua estructura, o viaducto apresenta deslocamentos laterais importantes,
levando assim ao estudo da inuncia destes efeitos, entre outros, no comporta-
mento dinmico lateral de veculos ferrovirios. No sentido de avaliar este fenmeno,
outras duas questes foram estudadas: irregularidades da via e efeitos laterais no
movimento dos veculos devido aos problemas na interface de contacto rodacarril.
Foram desenvolvidos diferentes modelos de veculos ferrovirios, considerando para
a sua anlise a inuncia dos efeitos mencionados: um modelo com um grau de
liberdade, um modelo com dois graus de liberdade e um modelo tridimensional. A
metodologia usada foi totalmente desenvolvida pelo autor.
Com base neste estudo concluuse que o viaducto no possui efeitos dinmicos
signicantes, tendo em conta que os estados limite ltimos e de servio foram veri-
cados. Em relao ao comportamento dinmico lateral dos veculos e considerando
este estudo como uma primeira aproximao ao problema, concluise que a segu-
rana e o conforto dos passageiros no so postos em causa. No entanto, estudos
mais detalhados devero ser realizados.
Este trabalho foi desenvolvido no mbito do protocolo assinado entre o Insti-
tuto Superior Tcnico e a RAVE - Rede Ferroviria de Alta Velocidade, S.A.
Palavras-chave: Comportamento dinmico lateral de veculos, Irregulari-

dades de via, Traado virtual, Veculos ctcios, Movimento de Klingel
iii
Acknowledgments
This work would not be possible without the support and the information provided
by several persons and organizations, along the time it has been developed.
First of all, I want to thank to Professor J.M. Goicolea from the Civil En-
gineering school of Madrid, all the support provided, the availability, the advises,
the rigour, the time spent teaching me all I have learned in the last months and
for making possible the development of this work. I want to thank too, to Profes-
sor Felipe Gabaldn of the same school, for the time spent helping me with FEAP
whenever I needed. However, this work would not have been possible without the
help of Professor Jorge Proena, who proposed a thesis in the eld of high speed
railway bridge behaviour and established the necessary contacts for my integration
in Madrid and for the availability and interest in helping me.
I'd like to thank to Jorge Nazar and Goicochea from Instituto de Caminos
de Hierro de Espaa, how provided important information and invitaded me for
the conference of high speed in Crdoba, to Manuel Cuadrado and CIDI-Calculo,
Investigacion y Desarollo para la Ingeneria, S.L., to J.Vinolas and A. Alonso from
CEIT and to RAVE-Rede Ferroviria de Alta Velocidade.
Thanks is also to Javier, Claudio and Yolanda, from the Mechanic Computa-
tional Group of the Civil Engineering School of Madrid, who helped me with my
integration in the department and for all the help and friendship provided. A special
acknowledgement to my family, friends and Claudia and most importantly, I would
like to thank my parents for the support and guidance throughout my academic
career.
v
Contents
Abstract i
Sumrio iii
Acknowledgments v
Contents x
List of Figures xvii
List of Tables xix
Notations xxi
1 Introduction and objectives 1

1.1 Motivation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.2 State of the Art . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
1.3 Scope of this study . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
2 Dynamic eects 13
2.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
2.2 Phenomena and Concepts . . . . . . . . . . . . . . . . . . . . . . . . 13
2.2.1 The wavelength . . . . . . . . . . . . . . . . . . . . . . . . . 14
2.2.2 Spacing length between axles dk . . . . . . . . . . . . . . . . . 15
2.2.3 Resonance in high-speed railway bridges . . . . . . . . . . . . 15
vii
2.3 Vertical Actions for Design . . . . . . . . . . . . . . . . . . . . . . . . 17
2.3.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
2.3.2 Static Eects . . . . . . . . . . . . . . . . . . . . . . . . . . . 18
2.3.3 Dynamic Eects . . . . . . . . . . . . . . . . . . . . . . . . . . 20
2.3.4 Dynamic Factor (2 , 3 ) . . . . . . . . . . . . . . . . . . . . 21
2.4 Horizontal Actions for Design . . . . . . . . . . . . . . . . . . . . . . 24
2.4.1 Centrifugal Forces . . . . . . . . . . . . . . . . . . . . . . . . . 24
2.4.2 Nosing Forces . . . . . . . . . . . . . . . . . . . . . . . . . . . 25
2.4.3 Traction and Braking Forces . . . . . . . . . . . . . . . . . . . 25
2.5 Verication of the Limit States . . . . . . . . . . . . . . . . . . . . . 25
2.5.1 Ultimate limit states . . . . . . . . . . . . . . . . . . . . . . . 26
2.5.2 Serviceability limit states - trac safety . . . . . . . . . . . . 27
2.5.3 Serviceability limit states - passenger comfort . . . . . . . . . 31
2.6 Methods for Dynamic Analysis . . . . . . . . . . . . . . . . . . . . . 33
2.7 Models Based on Point Loads . . . . . . . . . . . . . . . . . . . . . . 33
2.8 Methods Based on Finite Element Models . . . . . . . . . . . . . . . 34
2.9 Analytical methods based on modal analysis . . . . . . . . . . . . . . 36
2.9.1 Modal Analysis for a Simply Supported Beam . . . . . . . . . 36
2.9.2 Modes of vibration . . . . . . . . . . . . . . . . . . . . . . . . 38
2.9.3 Number of Modes of Vibration to Consider in the Analysis . . 38
2.10 Simplied Methods According to the Train Dynamic Signature . . . . 39
2.10.1 DER method . . . . . . . . . . . . . . . . . . . . . . . . . . . 40
2.10.2 LIR method . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42
2.11 Dynamic Analysis with Vehicle-Structure Interaction . . . . . . . . . 49
3 Vertical Railway Dynamic Analysis - Case Study 51

3.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51
3.2 Methodology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53
3.3 Bridge Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 54
3.3.1 Finite Element Model of the Structure . . . . . . . . . . . . . 54
viii
3.3.2 Modal Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . 55
3.4 Dynamic Analysis for Vertical Loads . . . . . . . . . . . . . . . . . . 56
3.5 Limit States Verication . . . . . . . . . . . . . . . . . . . . . . . . . 62
3.5.1 Dynamic Factor (2 ) . . . . . . . . . . . . . . . . . . . . . . 63
3.5.2 Real Impact Factor real . . . . . . . . . . . . . . . . . . . . . 63
3.5.3 LM71 Model Eects . . . . . . . . . . . . . . . . . . . . . . . 65
3.6 Verication of the Serviceability Limit States . . . . . . . . . . . . . . 65
3.7 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 68
4 Lateral Dynamic Analysis 71

4.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 71
4.2 Bridge Lateral Displacements . . . . . . . . . . . . . . . . . . . . . . 73
4.3 Lateral Stability of Railway Vehicles . . . . . . . . . . . . . . . . . . 85
4.3.1 Track Irregularities . . . . . . . . . . . . . . . . . . . . . . . . 85
4.3.2 Physical Behaviour of the Wheelset on a Straight Track . . . . 92
4.4 Lateral Vehicle Car Characteristics . . . . . . . . . . . . . . . . . . . 97
4.5 Vehicle Model With One Degree of Freedom . . . . . . . . . . . . . . 102
4.5.1 Response due to a prescribed base displacement . . . . . . . . 104
4.5.2 Dynamic Analysis of the Vehicle Model . . . . . . . . . . . . 105
4.6 Model with Two Degree of Freedom . . . . . . . . . . . . . . . . . . . 112
4.6.1 Response due to a prescribed base displacement . . . . . . . . 113
4.7 Three-Dimensional Vehicle Model . . . . . . . . . . . . . . . . . . . . 116
4.7.1 Denition of the model . . . . . . . . . . . . . . . . . . . . . . 116
4.7.2 Dynamic Results of the Model . . . . . . . . . . . . . . . . . . 118
4.8 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 122
5 Conclusions and suggestions for subsequent studies 125
Bibliography 128
Appendix 133
ix
A Train Dynamic Signature - Program (.m) 133
B Fictitious Railway Vehicles - Program (.m) 135
C Two Degree of Freedom Model 139
D Three-Dimensional Model 143
E Piers characteristics 147
x
List of Figures
1.1 High speed lines in Europe - UIC - High speed . . . . . . . . . . . . . 1

1.2 High speed passenger volume growth - UICHigh Speed . . . . . . . . 2
1.3 High speed train Eurostar. . . . . . . . . . . . . . . . . . . . . . . . . 3
1.4 Example of an alignment track irregularities prole in a track length
of 50 m. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
1.5 Wheelset sinusoidal movement in a straight track. Representation of
the movement at v = 400 km/h - Klingel (1883) [23] . . . . . . . . . 5
1.6 Dynamic response of a simply supported beam from the catalogue of
bridges ERRI, with span length, L = 15 m, and frequency of vibration,
f0 = 5 Hz, due to the eects of a single moving load of the train AVE,
P = 172.1 kN, at v = 200 km/h and v = 360 km/h. . . . . . . . . . . 7
1.7 Dynamic behaviour of a simply supported bridge from the catalogue
of bridges of ERRI, due to AVE train eects at v = 200 km/h and
v = 360 km/h. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
1.8 "Arroyo las Piedras" viaduct - Case study [31]. . . . . . . . . . . . . 10
2.1 Amplitude of vibration of a ctitious wave. Geometric interpretation

of the wavelength of excitation. Extracted from Dominguez Barbero [2]. 14
2.2 Spacing length between axles, dk . Extracted from Dominguez Barbero
[2]. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
2.3 Dynamic Displacements of one of the bridges dened in the ERRI
catalogue, with span length L = 15 m, damping = 0.02 and stiness
EI = 7694081 kN/m2 , due to actions of the train AVE at v = 200
km/h and v = 360 km/h. . . . . . . . . . . . . . . . . . . . . . . . . . 17
2.4 Load Model 71 and characteristic values for vertical loads. Extracted
from EN1991-2 [13]. . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
xi
2.5 Load Models SW/0 and SW/2. Extracted from [13] . . . . . . . . . . 19
2.6 Characteristic values of vertical loads for Load Models SW/0 and
SW/2. Extracted from EN1991-2 [13]. . . . . . . . . . . . . . . . . . 19
2.7 Limits for bridge natural frequencies, n0 [Hz], as a function L [m].
Extracted from EN1991-2 [13]. . . . . . . . . . . . . . . . . . . . . . . 21
2.8 Flow chart to evaluate if a dynamic analysis is required. Extracted
from EN1991-2 [13]. . . . . . . . . . . . . . . . . . . . . . . . . . . . 22
2.9 Denition of deck twist and limit values. Extracted from EN1990-A2
[12]. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28
2.10 Denition of angular rotations at the end of decks. Extracted from
EN1990-A2 [12]. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29
2.11 Model of a track/structure system. Extracted from EN1990-A2 [12]. . 30
2.12 Variation of longitudinal shear force with longitudinal track displace-
ment for one track. Extracted from EN1990-A2 [12]. . . . . . . . . . 30
2.13 Recommended levels of comfort. Extracted from EN1990-A2 [12]. . . . 32
2.14 Point Loads dening the actions of a real train (train AVE). . . . . . 34
2.15 Nodal force time history denition for a single moving load F , with
speed v . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36
2.16 Two rsts modes of vibration of a simply supported beam and respec-
tive analytical expression. . . . . . . . . . . . . . . . . . . . . . . . . . 38
2.17 Transfer function of ballast. Extracted from [36]. . . . . . . . . . . . . 39
2.18 Denition of the distances between the axles of the train of loads from
the rst axle of the front car. . . . . . . . . . . . . . . . . . . . . . . . 41
2.19 Denition of Subtrain. Representation of the three rsts subtrains of
the train model. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42
2.20 Train dynamic signatures for the seven real trains, considering a
damping value of =0.00. . . . . . . . . . . . . . . . . . . . . . . . . 44
2.21 Train dynamic signatures for HSLMA model, considering a damping
value of =0.00. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45
2.22 Envelope values for the real trains and for the HSLM model, for
=0.00. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45
2.23 AVE train dynamic signatures for dierent values of damping . . . . 46
xii
2.24 Inuence line, A(r), for a simply supported beam from the catalogue
of bridges dened by the ERRI in [8], with L = 15 m and f0 = 5 Hz. . 46
2.25 Envelope of maximum vertical accelerations as a function of train
speed for a bridge of the catalogue ERRI with L = 15 m, f0 = 5 Hz
and damping = 2% due to the passage of the train AVE, obtained
with the LIR method and with a direct integration in time. . . . . . . 48
2.26 Envelope of maximum vertical accelerations as a function of train
speed for a bridge of the catalogue ERRI with L = 15 m, f0 = 5 Hz
and damping = 2% due to the passage of the train ICE2, obtained
with the LIR method and with a direct integration in time. . . . . . . 48
2.27 Complete vehicle-structure interaction model. . . . . . . . . . . . . . . 49
2.28 Load train with vehicle-bridge interaction: simplied interaction model
and variables denition. . . . . . . . . . . . . . . . . . . . . . . . . . 50
2.29 Reduction of the dynamic response in simply supported bridges with
dierent span lengths and damping = 2%. Bridges ERRI with spans
L = 10, 30 and 40 m (L = 20 m, f0 = 4 Hz, = 20000 kg/m,
U IC71 = 11.79 mm; L = 30 m, f0 = 3 Hz, = 25000 kg/m, U IC71 =
15.07 mm; L = 40 m, f0 = 3 Hz, = 30000 kg/m, U IC71 = 11.81
mm). Performed by J.M. Goicolea [28]. . . . . . . . . . . . . . . . . . 50
3.1 "Arroyo las Piedras" viaduct. . . . . . . . . . . . . . . . . . . . . . . 51

3.2 Cross-section of the "Arroyo las Piedras" viaduct, near a pier. . . . . 52
3.3 Lateral view of the "Arroyo las Piedras" viaduct. Representation of
the twenty spans. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 54
3.4 Torsional moment of inertia dened for the dierent spans in (m4 ). . 55
3.5 Deformed shape of the 10 rst eigenmodes of the "Arroyo las Piedras"
viaduct and respective eigenfrequencies. . . . . . . . . . . . . . . . . . 56
3.6 Representation of the vertical moving loads and respective torsional
moments. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57
3.7 Denition of the Rayleigh curve for the model. . . . . . . . . . . . . . 57
3.8 Dynamic response of the deck in terms of displacements and accele-
rations due to deection eects in the midspan of the span 3 (node
292) at 200 km/h and 400 km/h. . . . . . . . . . . . . . . . . . . . . 58
3.9 Dynamic response of the deck in displacements and accelerations due
to deection eects in the midspan of the span 18 (node 2197) at 200
km/h and 400 km/h. . . . . . . . . . . . . . . . . . . . . . . . . . . . 58
xiii
3.10 Envelope values for the midspan displacement in each span for dier-
ent speed due to deection eects. . . . . . . . . . . . . . . . . . . . . 59
3.11 Envelope values for lateral displacements in the midspan of each span
for dierent speed due to torsional eects. . . . . . . . . . . . . . . . 60
3.12 Envelope values for the midspan displacement in each span for dier-
ent speed due to deection and torsional eects. . . . . . . . . . . . . 61
3.13 Maximum dynamic displacements in the node 292 for the seven Real
Trains, for a speed range dened from v = 120 km/h up to v = 420
km/h. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 64
3.14 Envelope value for the eects produced in node 292. . . . . . . . . . . 66
3.15 Acceleration values produced in node 292. . . . . . . . . . . . . . . . . 67
4.1 Derailment possibly due to vehicle dynamic eects. . . . . . . . . . . . 72

4.2 Lateral displacements of the deck due to torsional eects. . . . . . . 73
4.3 Virtual path measured at a certain axle of the European AVE train at
v = 350 km/h in the 'Arroyo las Piedras' viaduct, which started to
cross the bridge 2.05 seconds after the rst axle of the train. . . . . . 74
4.4 Virtual path measured at a certain axle of the European AVE train at
v = 400 km/h in the 'Arroyo las Piedras' viaduct, which started to
cross the bridge 1.83 seconds after the rst axle of the train. . . . . . 75
4.5 Mode shapes for the rst eigenmodes of lateral vibration. . . . . . . . 77
4.6 Lateral displacements in node 1181. The gure on the left represents
the displacements due to lateral benging of piers and the gure on the
right the displacements due to local torsional eects. . . . . . . . . . . 78
4.7 Total lateral displacements on node 1181. The red mark represents
the instant in which the axle 3 of the AVE train crosses the node. . . 78
4.8 Virtual path for the bogie 3 of R1 train. Performed by CIDI [4]. . . . 79
4.9 Virtual path for bogies 1, 3 and 5 of the R1 train. Performed by CIDI
[4]. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 80
4.10 Virtual path for the bogie 3 of ICE2 train. Performed by CIDI [4]. . . 81
4.11 Virtual path for bogies 1, 3 and 5 of the ICE2 train at v = 350km/h.
Performed by CIDI [4]. . . . . . . . . . . . . . . . . . . . . . . . . . . 82
4.12 Virtual path for bogies 1, 3 and 5 of the AVE train at v = 400km/h.
Performed by CIDI [4]. . . . . . . . . . . . . . . . . . . . . . . . . . . 83
xiv
4.13 Virtual path for the bogie 3 of AVE train. Performed by CIDI [4]. . . 84
4.14 Denitions of track irregularity parameters. (a) Typical track, (b)
gage and alignment, (c) cross level and nominal vertical prole - taken
from Garg and Dukkipati [17]. . . . . . . . . . . . . . . . . . . . . . . 87
4.15 Example of a power spectral density function - from Karl Popp, Holger
Kruse and Ingo Kaiser [22]. . . . . . . . . . . . . . . . . . . . . . . . 89
4.16 Example of a track lateral alignment irregularities prole for a track
with low irregularities in a total length of 1209m. Representation of
the same prole for a length of 100 m. . . . . . . . . . . . . . . . . . 91
4.17 Representation of the Klingel movement on a tangent track. . . . . . . 93
4.18 Flangeway clearance, f wc. . . . . . . . . . . . . . . . . . . . . . . . . 93
4.19 Lateral displacements of the wheelset given by the theory of Klingel.
Example of the movement for a time interval of 2 seconds at 400km/h. 95
4.20 Variation of wheel-rail contact point due to track alignment irregula-
rities. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 96
4.21 Dierence between wet and dry rail for creep forces. Extracted from
Bombardier [26]. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 97
4.22 Representation of yaw and roll motions. . . . . . . . . . . . . . . . . 98
4.23 three-dimensional vehicle model. Extracted from Yean-Seng Wu [35]. . 99
4.24 Vertical dimensions of the train model adopted. Extracted from [35]. . 102
4.25 Reduction of the lateral vibration of railway vehicles to a system with
one degree of freedom. . . . . . . . . . . . . . . . . . . . . . . . . . . 103
4.26 Response of the system with one degree of freedom due to a prescribed
base displacement, y = 0.003 m. Mechanical characteristics adopted
from the ETR-500 train car. . . . . . . . . . . . . . . . . . . . . . . . 104
4.27 Response of the vehicle due to the Virtual Path excitation for a type
of train with a natural frequency of vibration f0 = 0.43 Hz (ETR-500
train car). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 106
4.28 Maximum relative displacement and maximum acceleration responses
for the dierent ctitious vehicles due to virtual path excitation. . . . 107
4.29 Response of the vehicle due to the Virtual Path excitation for a type
of train with a natural frequency of vibration f0 = 0.76 Hz (Fictitious
vehicle). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 107
xv
4.30 Acceleration Response of the vehicles due to the Virtual Path excita-
tion for trains with a natural frequency of vibration f0 = 0.43 Hz (Car
ETR-500) and f0 = 0.76 Hz (Fictitious vehicle). . . . . . . . . . . . . 108
4.31 Vehicles dynamic response in terms of displacements due to the vir-
tual path plus track irregularities excitation. Vehicles with a natural
frequency of vibration f0 = 0.43 Hz (Car ETR-500) and with a natural
frequency of vibration f0 = 0.76 Hz (Fictitious vehicle). . . . . . . . . 109
4.32 Acceleration Response of the vehicles due to the Virtual Path plus
Track Irregularities excitation for trains with a natural frequency of
vibration f0 = 0.43 Hz (Car ETR-500) and f0 = 0.76 Hz (Fictitious
vehicle). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 109
4.33 Maximum relative displacement and maximum acceleration for the
dierent ctitious vehicles due to the combination of virtual path and
track irregularities eects. . . . . . . . . . . . . . . . . . . . . . . . . 110
4.34 Maximum relative displacement and maximum acceleration for the
dierent ctitious vehicles due to the combination of virtual path,
track irregularities and Klingel movements eects. . . . . . . . . . . . 111
4.35 Displacements response of the car ETR-500 due to the combination
of all the three eects considered. . . . . . . . . . . . . . . . . . . . . 111
4.36 Accelerations response of the ETR-500 train car due to the combina-
tion of all the three eects considered. . . . . . . . . . . . . . . . . . . 112
4.37 Representation of the system with two degrees of freedom. . . . . . . . 113
4.38 Response of a system with two degrees of freedom due to a prescribed
constant base displacement, y = 0.003 m. Results given for the degree
of freedom x2 (t). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 114
4.39 Comparation between the response in displacements of the two systems
considered due to the total aects considered in the study. . . . . . . . 115
4.40 Comparation between the response in accelerations of the two systems
considered due to the total aects considered in the study. . . . . . . . 115
4.41 Three-Dimensional vehicle model. . . . . . . . . . . . . . . . . . . . . 116
4.42 Denition of the nodes of the model. The bodies drawn were just
dened for visualization. . . . . . . . . . . . . . . . . . . . . . . . . . 117
4.43 Three-dimensional railway vehicle model with the mechanical proper-
ties of the car ETR-500 and with the geometrical characteristics of
the AVE train. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 117
xvi
4.44 Results obtained for the dierent models developed in this study due
to the eects of the virtual path. . . . . . . . . . . . . . . . . . . . . . 119
to the combination of the virtual path and track irregularities eects. . 119
to consideration of all the eects considered: virtual path, track irre-
gularities and Klingel movement eects. . . . . . . . . . . . . . . . . . 120
4.47 Representation of the maximum relative displacements to the track
(virtual path) for the three models developed and for the three eects
considered in this study: 1Virtual path; 2virtual path plus track
irregularities;3virtual path plus track irregularities and klingel move-
ment eects. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 121
4.48 Representation of the maximum accelerations values for the three mo-
dels developed and for the three eects considered in this study: 1
Virtual path; 2virtual path plus track irregularities;3virtual path plus
track irregularities and Klingel movement eects. . . . . . . . . . . . 121
4.49 Example of lateral dynamic behaviour of the three-dimensional model. 122
C.1 Model with two degrees of freedom modeled in FEAP [30] . . . . . . . 139
E.1 Inertia properties of the "Arroyo las Piedras" viaduct piers, about the
axis of lateral bending. . . . . . . . . . . . . . . . . . . . . . . . . . . 147
E.2 Geometric characteristics of the "Arroyo las Piedras" viaduct piers. . 148
xvii
xviii
List of Tables
2.1 Regular spacing d1 in regular train cars, in meters [m]. . . . . . . . . 15
2.2 Recommended accelerations values to ensure the respective levels of
comfort. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32
3.1 Static and Dynamic eects due to real trains actions. . . . . . . . . . 65
4.1 FRA Track Safety Limits for track alignment irregularities[1]. . . . . 92

4.2 Klingel movement parameters denition. . . . . . . . . . . . . . . . . 94
4.3 Suspension and Mass characteristics of UIC and ETR-500 train car. . 100
4.4 Vehicle inertia properties. . . . . . . . . . . . . . . . . . . . . . . . . 101
4.5 Geometric vertical properties of vehicle model. See gure 4.24. . . . 102
4.6 Vehicle/Track Interaction Safety Limits (FRA). . . . . . . . . . . . . 120
4.7 Vehicle/Track Interaction Safety Limits (SNCF). . . . . . . . . . . . 120
B.1 Mechanical characteristics of cticious railway vehicles dened for

this study with frequencies of lateral vibration dened from 0.2 Hz up
to 1.0 Hz. The lateral stiness, k , is given in (kN/m), the ratio of
damping, c, in (kN/(m/s)) and the mass, M , in (ton). The values
inside boxes correspond to some reference real trains. . . . . . . . . . 137
B.2 Mechanical vehicle characteristics used as reference values in order to
dene the ctitious vehicles . . . . . . . . . . . . . . . . . . . . . . . 138
xix
Notations
C damping matrix
EI deection stiness [kg/m2 ]
Fj axle force assigned to the node j
Fk load of the axle k of the train of loads
L span length
L equivalent span length
Lf inuence length of the loaded part of the curve on the bridge
La,b inuence length for traction and braking forces
M mass matrix
M mass nominal value
Mi generalized mass
K stiness matrix
Qtk characteristic value of centrifugal forces in [kN]
Qvk characteristic values of the vertical loads for LM71, SW/0, SW/2
and "unloaded train" in [kN]
Qsk nosing force
P nominal value of a single moving load
V design speed
T Period of vibration
a lateral acceleration of the vehicle body car
bv vehicle vertical accelerations
c suspension damping nominal value
dk distance from the rst axle to the axle k of a train of loads
d vector of nodal displacements
d vector of nodal speeds
d vector of nodal accelerations
f frequency of vibration
f0 fundamental frequency of vibration
fh0 lower limit for the frequency of bridge lateral vibrations
f (t) vector of external loads of the system
xxi
n0 rst natural bending frequency under permanent loads
nT rst natural torsional frequency under permanent loads
qtk characteristic value of centrifugal forces in [kN/m]
qvk characteristic values of the vertical loads for LM71, SW/0, SW/2
and "unloaded train" in [kN/m]
k spring stiness nominal value
s track gauge
ystat vertical displacement due to static loads
ydyn vertical dynamic displacements
yi vertical displacement for the eigenmode i of vibration
yi vertical speed for the eigenmode i of vibration
yi vertical acceleration for the eigenmode i of vibration
t maximum deck twist
v train speed
ti time step
telem time at which a load takes to cross an element
xk distance from the rst axle to the axle k of a train of loads
f (t) vertical displacement in a certain point of the structure at time t
f (t) vertical speed in a certain point of the structure at time t
f (t) vertical acceleration in a certain point of the structure at time t
mass density
impact factor
i (x) function of the deformed shape for each mode of vibration
0 maximum dynamic increment due to the eects of a single moving
load
00 factor that takes into account track irregularities
2 impact factor for railway lines with good maintenance
3 impact factor for railway lines with normal maintenace
sta,LM 71 Maximum static vertical displacements due to the static actions of
the model LM 71
dyn,real Maximum static vertical displacements due to the actions of the
real trains for a certain speed range
maximum permissible vertical deection
i damping of the structure for the eigenmode i
i frquency of vibration of the eigenmode i
wavelength
D frequency of excitation
xxii
Chapter 1
Introduction and objectives
1.1 Motivation
At the beginning of this millennium, Portugal is facing a new form of European in-
tegration: trans-European transport systems. It is necessary to provide the country
with a better quality and quantity of transport infrastructure, with the same level as
that in more developed countries, in order to achieve new levels of competitiveness
and sustained development.
Figure 1.1: High speed lines in Europe - UIC - High speed

2 Introduction and objectives
In today's global market, connections using transport infrastructures that

favour the ecient mobility of people and goods are an indispensable condition
for economic growth, with particular importance in the country's infrastructure and
social cohesion, and serve as guarantee of balanced development. This is why the
high speed railway project is not only a structural factor of mobility, but will also,
thanks to its multidisciplinary nature, make a substantial contribution to the devel-
opment of a large number of the country's sectors, one of which is engineering.
In recent years the design and construction of high speed railway lines in
Europe has been one of the major activities for civil engineering due to its magnitude
and technical requirements. In countries like Spain and France the investment in
high-speed railway lines began some years ago with the construction of new lines or
the upgrade of others, but in Portugal this is about to begin and nally provide an
ecient transport link between Portuguese cities as well as with Spain and the rest
of the Europe.
High-speed transportation became more and more important in European life,
with a constant increment in the number of passengers along the years, as shown in
gure 1.2, and in Portugal this will certainly happen too.
Figure 1.2: High speed passenger volume growth - UICHigh Speed
High Speed Railway Bridges

Railway bridges, due to high intensity moving loads, are subjected to dynamic eects
that may reach signicant values, which must be considered in the design of these
structures. These dynamic eects are being given greater importance at present, as
a consequence of the increment of the circulation speed in existing or new railway
1.1 Motivation 3
lines, as the case of those intended for high-speed trains.
Figure 1.3: High speed train Eurostar.
In high-speed railway lines, dynamic eects tend to increase even more con-
siderably as a result of resonance eects. This occurs due to the passage of trains
composed by several groups of regularly spaced axles at a certain speed. Thus, the
knowledge of these eects is of major importance for the following reasons:
i vibrations induced by the passage of the trains over the bridge originate, in
general, displacements in structures greater than those produced when the
loading is statically applied;
ii excessive vibrations of the structure may lead to a magnication of the fatigue

phenomena;
iii deformations and accelerations of the bridge should be controlled and kept
within certain limit values, in order to ensure the track stability and constant
wheel-rail contact;
iv the accelerations in vehicles should be limited so that the passengers comfort

can be guaranteed.
Extensive studies were undertaken by the Oce for Research and Experiments
(ORE) of the International Union of Railways (UIC), and subsequently by the Spe-
cialist Subcommittee D214 under the management of the European Rail Research
Institute (ERRI). These recent advances in the understanding of the behaviour of
high speed railway bridges have been reected in new engineering codes, such as the
EN19912 [13], the EN1990A2 [12] and more recently in IAPF [10].
Lateral Dynamic Phenomena of Railway Bridges

Due to railway trac, high-speed bridges are aected by vertical loads and by lon-
gitudinal and transversal horizontal forces. These transversal horizontal forces are
generated by lateral movements of the vehicles from two sources in a straight track:
horizontal track irregularities (Figure 1.4) and sinusoidal motion of conical wheels
along cylindrical rail heads (Figure 1.5). Apart from these two sources, centrifugal
forces also originate, in a curved track, lateral forces, which acts on the bridge in
the outward direction.
8
Track Irregularities
Perfect track
4
Track alignment irregularities (mm)
-2
-4
-6
-8
0 10 20 30 40 50
Distance (m)
Figure 1.4: Example of an alignment track irregularities prole in a track length of

50 m.
Measures and parametric studies made by the committee ERRI D181 and dis-
cussed in the report RP6 [7], have shown that these eects play an important rule
in the phenomena of interaction between bridge-train, in which were established
some project recommendations to cover these lateral eects. Usually the problem
of lateral vibration in railway bridges does not aect the safety of the structure but
can put at risk the comfort or the security of the passengers. On the other hand
can contribute to an accelerated degradation of the railway, like ballast migration
and instability or the track structural damage, which aects the stability of vehicles
movement.
1.1 Motivation 5
0.008
0.006
0.004
Lateral Displacement [m]

0.002
-0.002
-0.004
-0.006
-0.008
0 0.5 1 1.5 2
Time [s]
Figure 1.5: Wheelset sinusoidal movement in a straight track. Representation of the

movement at v = 400 km/h - Klingel (1883) [23]
Additionally, lateral displacements in railway bridges are caused by vertical

loads due to railway trac in bridges with double track, where these loads are ec-
centric. This phenomenon is related with the torsional behaviour of the deck causing
the piers to bend over laterally and displacing the deck at the top of it. The local
torsional eects within spans, induced by the same vertical loads, also has a lateral
component that increase the displacements.
Objectives of this work

The main objectives for this thesis are the following:
1. Study and comprehension of the dynamic eects in railway bridges under high-
speed trac and existing models for analysis and calculation. Review the main
methods and codes for vertical dynamic response analysis of the bridge-train
system. Application of these methods to a real case of high-speed railway
bridge;
2. Application of the lateral dynamic response analysis in a real case of high-speed

railway bridge, completing the dynamic analysis of this bridge. Synthesis of
conclusions;
3. Study in more detail the methodologies and the dierent forms of lateral bridge
instability and vibration, caused by high speed train trac. Proposal of a
simple methodology for the analysis of railway vehicles lateral vibrations;
4. Develop and validate models in nite element programmes (FEAP [30] and
ABAQUS [29]) and in a numerical program (Octave [11]Matlab language) to
simulate the train lateral response when subject to movements in the base,
and to predict the phenomenon of lateral resonance. [3]
1.2 State of the Art

Since the early stages of railways, dynamic eects, due to moving loads from rail-
way trac, have been one of the main design requirements for railway structures.
The basic solution of the dynamic response for a simply supported beam due to a
moving load was obtained in 1905 by Krylo [24] and described by Timoshenko in
1928 [32]. More recently further studies were developed by Fryba [16], among others.
The design codes existing up to now, like the Eurocode [13, 12] and the new
Spanish code IAPF [10], considered for the design of railway bridges the dynamic
response through an impact factor. This factor, currently represented as , repre-
sents the increase in the dynamic response with respect to the static one for a single
moving load, being dened as follows:
= max(1 + 0 + 00 ) (1.1)
where 0 is the maximum dynamic increment due to the eects of a single moving
load at speed v , as shown in gure 1.6, which has a maximum of 0 = 1.32. In turn,
00 is a factor that takes into account the eect of irregularities. The impact factor
so dened is applied to the static eects produced by the real trains.
In order to simplify the consideration of these real trains, in [25] the nominal
train type model LM 71, was dened, which represents the actions of the real trains
in circulation. Finally, an envelope value of the impact factor, , was established,
in order to be applied to the static eects of this model. The new impact factor can
be dened as 2 or 3 , according to the type of track maintenance, as explained in
chapter 2.
However, high speed railway lines pose dynamic problems of higher order, due
to the possibility of resonance eects. This phenomenon appears at speeds above
200 km/h, considering the typical distances between axles in railway coaches and
the main eigenfrequencies of the bridges. Thus, taking into account that the impact
factor does not consider resonance eects, the applicability of this method is subject
to some restrictions, which are discussed in chapter 2.
In gure 1.7 the dynamic response of a simply supported bridge from the
1.2 State of the Art 7
3
v=200Km/h
v=360Km/h
2
Vertical displacements (mm)
-1
sta
-2
sta
-3
0 0.1 0.2 0.3 0.4 0.5 0.6
Time (s)
Figure 1.6: Dynamic response of a simply supported beam from the catalogue of
bridges ERRI, with span length, L = 15 m, and frequency of vibration, f0 = 5 Hz,
due to the eects of a single moving load of the train AVE, P = 172.1 kN, at v = 200
km/h and v = 360 km/h.
catalogue of bridges of ERRI, is presented. This bridge has a span length of L = 15

m and rst frequency of vibration f0 = 5 Hz and it may be seen that the static
eects obtained with the LM 71 model, do not cover the maximum dynamic eects
of the beam due to the passage of the AVE train.
15
18
10
16
Maximum Displacements (mm)
5
14
Displacement (mm)
0
12
-5
10
-10
8
-15
6
360 -20
4 0 1 2 3 4 5
150 200 250 300 350 400
Velocity (km/h)
Velocity (km/h)
v=360 km/h v=200 km/h UICstatic
AVEDynamic UICstatic
Maximum Maximum
Figure 1.7: Dynamic behaviour of a simply supported bridge from the catalogue of
bridges of ERRI, due to AVE train eects at v = 200 km/h and v = 360 km/h.
Within Europe, a joint eort for research and study of dynamic eects in high
speed lines has been carried out within ERRI by the subcommittee D214. These
and other ndings have been included in the recently drafted engineering codes
previously referred.
Thus, along the years several methods were developed in order to analyse the
dynamic behaviour of railway bridges. The methods available in pratice are, in
terms of increasing complexity:
i Impact factor;
ii Dynamic train signature;
iii Moving load dynamic analysis;
iv Vehicle-structure interaction dynamic analysis.
Other subject which must also be considered is the interoperability of railway

infrastructures, enabling the transit of all possible high speed trains in all European
railway lines. Hence, all European high speed train types should be considered for
the design and corresponding dynamic analysis. One of the most valuable results
of the subcommittee ERRI D214 has been the establishment of a High Speed Load
Model (HSLM) that comprises a family of ctitious articulated trains whose dynamic
eects have been proved to be an envelope of the dynamic eects due to all the
current trains, and those foreseen within an agreed set of interoperability criteria
dened in [33]. Thus, dynamic analysis should be performed considering all the
seven real trains available at the present and the ten ctitious trains of the HSLM,
in orther to guarantee the validity of all the existing trains and the new ones that
may appear in the future.
The eects of horizontal forces due to railway trac, in longitudinal and lateral
directions, is other relevant aspect in railway bridges. In order to consider these
eects, in EN1991-2 [13] were established some actions to be considered in the
design of railway bridges, which are discussed in chapter 2 of this work. This code
established that for lateral horizontal forces a concentrated force should be adopted,
named as nosing force, acting at the top of the rails and applicable on both straight
and curved tracks. Aditionally, centrifugal forces must also be considered in curved
tracks.
In order to limit transverse deformations of the deck and ensure the trac
safety, EN1990-A2 [12] - section A2.4.4.2.4 established limits for the lateral deec-
tion, maximum horizontal rotation and change of radius of curvature. In the same
section, in order to limit the transverse vibration of the deck, the rst natural fre-
quency of lateral vibration of a span should not be less than fh0 = 1.2 Hz. This
1.3 Scope of this study 9
value was originally proposed by the subcommittee ERRI D181 in the report RP6
[7], in which the study of lateral forces in railway bridges was performed. This limit
was proposed in order to avoid lateral resonance in railway vehicles, taking into
account that the frequencies of lateral vibration of railway vehicles are, in general,
not greater than 1 Hz.
Vehicle Lateral Dynamic Behaviour

The study of vehicle lateral dynamic behaviour is important in order to evaluate
the security and the comfort of the passengers. Bridge lateral displacements, track
irregularities and rail-wheel contact problems play an important role in these type
of analyses.
In 1883 Klingel [23] developed a model of a free massless wheelset with conical
wheels rolling cinematically on straight rigid rails with a sinusoidal displacement
of constant period (gure 1.5) in order to evaluate the contact wheel-rail problem.
However, the reality is more complex, especially concerning the contact physics.
Taking into account the inertia of the wheelset, the hunting motion requires tan-
gential forces in the contact area, for which Kalker's linearised theory (1967) gives
a more suitable approximation. Other authors like Carter (1926) and Matsudaira
(1952) studied this subject too.
The inuence of track irregularities (gure 1.4) in the case of resonance in
bridges, was studied by the committee ERRI D214 and presented in the report
RP5 [9]. In this document some limit values for the maximum deviations of track
irregularities and conicity were proposed to prevent dynamic problems of higher
order. These track irregularities are usually implemented as random irregularities
based on power spectral density functions, as published by ARGER/F [27], among
others.
The inuence of these eects in vehicle dynamics were considered in analytical
models developed and studied by Garg and Dukkipati [17], Karl Popp [22], Yean-
Seng Wu [35] and Jun Xiang [20], among others.
1.3 Scope of this study

The main subject of this work is the study of the dynamic response of railway
bridges due to railway trac and the evaluation of railway vehicles lateral dynamic
behaviour.
A bridge dynamic analysis of a case study will be performed, where the vertical
dynamic eects will be analysed, as well as the lateral dynamic displacements due
to the same actions. Aditionally, the inuence of these displacements, together with
track irregularities and contact problems eects, in the behaviour of railway vehicles,
will be studied.
Thus, in chapter 2 the dierent methods used in vertical dynamic analysis will
be reviewed, in order to help in the comprehension of the phenomena and concepts
involved in the dynamic response of railway bridges due to high speed railway trac.
The chapter is completed with the assumptions and requirements contemplated in
EN1991-2[13] and EN1990-A2 [12], for the design of railway bridges.
Figure 1.8: "Arroyo las Piedras" viaduct - Case study [31].
In order to exemplify and put in practice some of the methods reviewed in the
previous chapter, in chapter 3 a simple dynamic analysis will be performed for the
case study, "Arroyo las Piedras" viaduct, located in the new high speed railway line
between Mlaga and Crdoba in Spain, shown in gure 1.8.
In chapter 4, the lateral dynamic behaviour of railway vehicles will be studied,
considering the previous viaduct. A signicant importance in the development of
vehicle models will be given, in order to estimate the lateral response of railway
vehicles due to dierent excitations imposed on the wheelsets, such as bridge la-
teral displacements, track irregularities and Klingel movement eects. The lateral
dynamic analysis of the bridge was performed by CIDI - Calculo, Investigacion y
Desarrollo para la Ingenieria, S.L. [4], which provided the most signicant results
essential for this work.
For the study of the vehicle models the following steps were performed:
1.3 Scope of this study 11
1. Evaluation of the bridge lateral dynamic displacements, provided in the report

made by CIDI [4];
2. Generation of a track irregularities prole;
3. Study of the Klingel movement eects;
4. Development of models with one degree of freedom in Matlab code and with
two degrees of freedom in the nite element program (FEAP) [30], in which
the geometrical and mechanical characteristics of some real trains will be con-
sidered. Input of the dierent excitations considered in this study;
5. Development of a three-dimensional model in the nite element program ABAQUS

[29], in order to evaluate, in a more accurate way, the dynamic response of rail-
way vehicles on the viaduct considering the same excitations eects;
6. Analysis of the results.
Finelly, Chapter 5 contains some conclusions and recommendations for further

studies.
Chapter 2
Dynamic Eects in Railway Bridges
due to High Speed Trac
2.1 Introduction
In this chapter the main concepts and phenomena of railway bridges dynamic be-
haviour, will be discussed. Then, the assumptions and requirements dened in the
engineering codes, EN1991-2 [13] and EN1990-A2 [12], that should be considered in
the design of these type of structures, will be reviewed. Finally, the dierent meth-
ods used to perform an accurate dynamic analysis are dened and explained. Some
examples will be performed in order to help in the understanding of the concepts
presented.
2.2 Phenomena and Concepts

As discussed in chapter 1, the main problems of railway bridges dynamic behaviour
are the resonance eects generated on these structures due to high-speed trac. To
dene, understand and predict this phenomena two important concepts are used:
i the denition of wavelength represented by ;
ii the concept of train of loads dened by the nominal loads in each axle of the
train and by the regular spacing between them, dened as dk .
14 Dynamic eects
2.2.1 The wavelength

The excitation wavelength is dened by the following relation:
v
= (2.1)
f0
where f0 is the natural frequency of vibration of the structure, in [Hz], and v the
speed with which the train crosses the bridge, in [m/s]. The denomination of wave-
length of excitation is given by means of the relation between the denition of the
expression (2.1) and the wavelength of a ctitious wave with speed of longitudinal
propagation v and frequency f0 .
Figure 2.1: Amplitude of vibration of a ctitious wave. Geometric interpretation of

the wavelength of excitation. Extracted from Dominguez Barbero [2].
As an example, gure 2.1 shows the amplitude of the displacement of one wave
with longitudinal propagation of the displacement in the midspan of the bridge at
v = 219 km/h.
This graphic is the result of a dynamic analysis performed for a real structure1 ,
to which the resonance occurs at v = 219 km/h for the AVE train.
1 See section E.2. from Dominguez Barbero [2]

2.2 Phenomena and Concepts 15
2.2.2 Spacing length between axles dk

The spacing length between the dierent train axles are normally dened by regular
characteristic spacing values, dk , as show in gure 2.2. Three types of trains are
dened in this gure according to the number of bogies per train car. The rst type
of trains is named as conventional trains, the second type as articulated trains, in
which the bogies are shared by dierent train cars, and nally the last type of trains
indicated in the gure are named as regular trains. In table 2.1 the regular spacings
of some real European high-speed trains are dened.
Table 2.1: Regular spacing d1 in regular train cars, in meters [m].

Train EUROSTAR TGV ICE2 THALYS ETR-Y-500 TALGO AV
d1 18.70 18.70 26.40 18.70 26.10 13.14
When a train is crossing a bridge with a certain speed, the axles cross the
structure with a certain frequency. For example, for the train ETR-Y-500, which
has a regular spacing d1 equal to 26.10 m, at 360 km/h (100 m/s), this frequency
would be dened as f = 100/26.10 = 3.83 Hz, named as frequency of excitation.
2.2.3 Resonance in high-speed railway bridges

Resonance occurs in railway bridges when the frequency of excitation coincides with
the fundamental frequency of vibration of the structure or a multiple of it. This
means that when the characteristic spacing length between train axles, dk , coincides
with the wavelength, or a multiple of it, which is quantied through the expression
(2.2), resonance on the structure occurs.
dk = i, i = 1, 2, ... (2.2)
As an example, a dynamic analysis of a simple supported bridge dened in the

catalogue ERRI [8], with span length L = 15 m, damping = 0.02 and stiness
EI = 7694081 kN/m2 , was performed, using the nite element program FEAP [30].
In gure 2.3 the dynamic response in terms of vertical displacements in the midspan
of the bridge due to the passage of the train AVE at v = 200 km/h and v = 360
km/h, is presented, in which the resonance eects are visible.
The resonance problem was discussed and analysed by Dominguez Barbero in
[2], in which the prediction of this phenomenon, the key factors for the increase of
the dynamic response and the problem of resonance in dierent bridge typologies,
was studied.
16 Dynamic eects
Figure 2.2: Spacing length between axles, dk . Extracted from Dominguez Barbero
[2].
2.3 Vertical Actions for Design 17
30
200km/h
360km/h
20
10
Displacements [mm]
-10
-20
-30
0 2 4 6 8 10
Time [s]
Figure 2.3: Dynamic Displacements of one of the bridges dened in the ERRI ca-
talogue, with span length L = 15 m, damping = 0.02 and stiness EI = 7694081
kN/m2 , due to actions of the train AVE at v = 200 km/h and v = 360 km/h.
To take into account these eects in high speed railway bridges, an accurate
dynamic analysis should be performed using methods that have been developed
along the years. In section 2.6 these methods will be reviewed.
2.3 Vertical Actions for Design

2.3.1 Introduction
The methodology used up to now for the design of railway bridges by the new Euro-
pean engineering codes ([10, 25, 13]) and originally proposed in the UIC Code 776 -
1 R, considered the dynamic response of railway structures through an impact fac-
tor. This factor, generally represented as , represents the increase in the dynamic
response with respect to the static one due to a single moving load. However, this
factor does not include resonance eects.
Along the years the impact factor method has been adopted by dierent na-
tional engineering codes with little variations from the original methodology pro-
posed in the UIC Code 776 -1 R, with respect to some parameters and in how to
use it. In reference [2] these considerations and modications of the original method
were analysed.
18 Dynamic eects
With the increase of trains service speed in railway lines and with the develop-
ment of larger trains than the one considered in the UIC code 2 , the recommendations
proposed proved to be insucient to cover the increase of the dynamic response due
to resonance eects.
Facing this problem, the experts of the European Committee for Normalization
(CEN), responsible for the part 2 of the Eurocode 1, adopted the impact factor, but
imposed some limits for the use of this method. For cases outside those limits, an
accurate dynamic analysis should be carried out, considering the requirements and
assumptions proposed by EN1991-2 [13] and EN1990-A2 [12], as reviewed in this
chapter.
2.3.2 Static Eects

The static eects adopted in the design of railway bridges, should take into account
four models of railway loading, dened in EN1991-2 [13]:
Load Model 71 (and Load Model SW/0 for continuous bridges) to represent
normal rail trac on mainline railways),
Load Model SW/2 to represent heavy loads,
Load Model "unloaded train" to represent the eect of horizontal loads.
Load model 71
The introduction of the Load Model 71 (LM 71), represented in gure 2.4, permitted
the simplication in the design of railway bridges taking into account that this model
can cover the static eects of all the current real trains dened in the code UIC 776-
1R [25].
These characteristic values shall be multiplied by a factor, , on lines carrying
rail trac heavier or lighter than normal railway trac. When multiplied by this
factor, the loads aected are called as "classied vertical loads", being taken as
one of the following values:
0.75 0.83 0.91 1.00 1.10 1.21 1.33 1.46 (2.3)
For checking limits of deection, classied vertical loads and other actions
enhanced by shall be used, except for passenger comfort analyses where should be
taken as 1.
2 The Turbo Train was the train considered in the UIC code with 44.8 m of length
Figure 2.4: Load Model 71 and characteristic values for vertical loads. Extracted
from EN1991-2 [13].
Load Model SW/0 and SW/2

The Load Models SW/0 and SW/2 represent additional load models which must be
considered in some particular situations.
The load model SW/0 represents the static trac eect of vertical loading
due to normal railway trac on continuous beams. In turn, the load model SW/2
represents the static eects of vertical loading due to heavy railway trac. The
load arrangement shall be taken as shown in gure 2.5, adopting the respective
characteristic values, indicated in table 2.6. Note that the load model SW/0 shall
be multiplied by the factor in accordance with 2.3.
Figure 2.5: Load Models SW/0 and SW/2. Extracted from [13]
Figure 2.6: Characteristic values of vertical loads for Load Models SW/0 and SW/2.
Extracted from EN1991-2 [13].
20 Dynamic eects
Load model "unloaded train"

For some specic verications a particular load model is used, called as "unloaded
train", which consists of a vertical uniformly distributed load with a characteristic
value of 10.0 kN/m. It may be used, for example, in the verication of the lateral
stability of a bridge when subjected to lateral actions as the wind action.
2.3.3 Dynamic Eects

The static stresses and deformations induced on bridges are increased and decreased
under the eects of moving loads. These vibrations of the structure result from the
following actions:
the rapid rate of loading due to the speed of trac when crossing the structure
and the inertia response (impact) of the structure;
the passage of successive loads with approximately uniform spacing which can
excite the structure under certain circumstances, creating resonance;
variations in wheel loads resulting from track or vehicle imperfections (inclu-

ding wheel irregularities).
These eects should be considered in the design of the structure through an

impact factor or by means of a dynamic analysis, depending on the situation.
The dynamic behaviour of the structure is inuenced by the speed of trac
in the bridge, the span length, L, of the element, the mass of the structure, the
natural frequencies of vibration of the whole structure and relevant elements, the
associated mode shape along the railway line, the number of train axles, the nominal
loads per axle and the regular spacing of axles, the damping value of the structure,
the vertical irregularities in the track, the unsprung/sprung mass and suspension
characteristics of vehicles, the presence of regularly spaced supports on the deck
slab and/or track, the vehicle imperfections and the dynamic characteristics of the
track (ballast, sleepers, track components, etc).
Requirements for a Static or Dynamic Analysis

The requirements for determining whether a static or a dynamic analysis are re-
quired are shown in gure 2.8, where V [Km/h] is the Maximum Line Speed at the
site, L [m], the span length, n0 [Hz], the rst natural deection frequency under
permanent loads, nT , the rst natural torsional frequency for the same load, v [m/s]
the maximum nominal speed and nally (v/n0 )lim , as given in annex F of EN1992-2
[13]. The frequency rst of vibration, n0 , must be within the limits established in
gure 2.7.
Figure 2.7: Limits for bridge natural frequencies, n0 [Hz], as a function L [m].
Extracted from EN1991-2 [13].
The upper limit (1) is dened as:
n0 = 94, 76L0,748 (2.4)

and the lower limit (2) as:
(
80/L for 4 m L 20 m
n0 =
94, 76L0,748 for 20 m L 100 m
where L is the span length of the simply supported bridge or the equivalent
span length, L , dened for other type of structures. This parameter is quantied
in EN1991-2 [13]section 6.4.5.3.
2.3.4 Dynamic Factor (2, 3)

The dynamic factor , usually named as impact factor, takes into account the dy-
namic magnication of stresses and vibration eects in the structure but does not
22 Dynamic eects
Figure 2.8: Flow chart to evaluate if a dynamic analysis is required. Extracted from
EN1991-2 [13].
consider resonance eects, as previously noted. Whenever the criteria specied in

the previous section is not satised, there is a risk that resonance or excessive vibra-
tion may occur. For such cases a dynamic analysis shall be carried out to calculate
impact and resonance eects. Note that this method is only applicable for speeds up
to 200 km/h, for simply supported bridges with only longitudinal line beam or sim-
ple plate behaviour with negligible skew eects on rigid supports, and for structures
with natural frequencies of vibration within the limits established in gure 2.7.
Hence, the impact factor is taken as an envelope value of the static eects
produced under Load Models 71, SW/0 and SW/2 in order to cover the dynamic
eects due to real trains, for speeds up to 200 km/h:
sta,LM71 dyn,real (2.5)
where:
sta,LM71 Maximum static vertical displacement due to the static actions

of the model LM 71;
dyn,real Maximum static vertical displacement of the bridge due to the

actions of the real train for the range of speeds (20 km/h up to V km/h)3 .
The impact factor, , shall be taken as either 2 or 3 , for railway lines with
good or normal maintenance respectively, dened as following4 :
1.44
2 = + 0.82 1.0 2 1.67 (2.6)
L 0.2
2.16
3 = + 0.73 1.0 3 2.0 (2.7)
L 0.2
where L is the equivalent span length of the element under study, in meters 5 .
3 Being the equation valid for values of V inferior to 200 km/h as considered in reference [9]
4 These values were originally proposed in UIC Code 776 -1 R [25]
5 These impact factors were determined for simply supported beams. The equivalent span, L ,

possible the use of these factors for other structures with dierent supported conditions.
24 Dynamic eects
2.4 Horizontal Actions for Design

In EN1991-2 [13] the models which represent the three types of horizontal forces to
consider in the design of railway bridges, were established and dened as: Centrifugal
forces, nosing forces and longitudinal actions due to traction and braking.
2.4.1 Centrifugal Forces

Where the track on a bridge is curved over the whole or part of the length of the
bridge, the centrifugal force on the track shall be taken into account. They should
be taken to act outwards in a horizontal direction at a height of 1.80 m above the
running surface and shall always be combined with the vertical trac load. These
forces can be dened as following:
V2
Qtk = (f Qvk ) (2.8)
127r
V2
qtk = (f qvk ) (2.9)
127r
where V [km/h] is the specied maximum line speed at the site, r [m] is the radius
of curvature, Qtk [kN] and qtk [kN/m] are the characteristic values of the centrifugal
forces and Qvk [kN] and qvk [kN/m] are the characteristic values of the vertical
loads LM71, SW/0, SW/2 and "unloaded train" models. For the load model HSLM
the characteristic values of centrifugal forces should be determined using the LM71
model. The reduction factor, f , is given by:
s
V 120 814 2.88
f = 1 + 1.75 1 (2.10)
1000 V Lf
subjected to a minimum value of 0.25, in which Lf is the inuence length of the

loaded part of the curved track on the bridge. Thus, the reduction factor can assume
dierent values given as follows:
1. f = 1 for either V 120 km/h or Lf 2.88 m
2. f < 1 for 120 km/h < V 300 km/h and Lf > 2.88 m
3. f(V =300) for V > 300 km/h and Lf > 2.88 m

2.5 Verication of the Limit States 25
2.4.2 Nosing Forces

The nosing force shall be taken as a concentrated force acting horizontally, at the top
of the rails, perpendicular to the centre-line of the track. It shall be applied on both
straight and curved tracks. The characteristic value shall be taken as Qsk = 100 kN.
These forces are considered to cover the forces generated in the wheel-rail contact
interface, which is a very complex problem and review in chapter 4 of this work.
2.4.3 Traction and Braking Forces

Traction and braking forces acts at the top of the rails in the longitudinal direction
of the track. They shall be taken as uniformly distributed over the corresponding
inuence length, La,b , for traction and braking forces in the structural element con-
sidered. The direction of both forces shall take into account the permitted direction
of travel on each track.
The characteristic values for traction and braking forces shall be taken as
follows:
Traction force:
Qlak = 33[kN/m] La,b [m] 1000 kN (for LM 71, SW/0, SW/2 and HSLM)
Braking force:
Qlbk = 20[kN/m] La,b [m] 6000 kN (for LM 71, SW/0, SW/2 and HSLM)
Qlbk = 35[kN/m] La,b [m] (for Load Model SW/2)
2.5 Verication of the Limit States

The engineering code EN1991-2 [13] species the requirements that shall be consi-
dered to ensure trac safety, being dened as follows:
i. The verication of maximum peak deck accelerations shall be regarded as

a trac safety requirement checked at the serviceability limit state for the
prevention of track instability;
26 Dynamic eects
ii. The dynamic increase of load eects shall be allowed for by multiplying the
static loading by the dynamic factor, . If a dynamic analysis is necessary,
the results of the dynamic analysis shall be compared with the results of the
static analysis increased by and the most unfavourable load eects shall be
used for the bridge design;
iii. If a dynamic analysis is necessary, a check shall be carried out in order to es-
tablish whether the additional fatigue loading at high speeds and at resonance
is covered by consideration of the stresses due to load eects from LM 71.
The most adverse fatigue loading shall be used in the design.
2.5.1 Ultimate limit states

For the design of railway bridges and if a dynamic analysis is required, EN1991-2 [13]
species that the stresses and displacements on the structure should be determined
considering the most unfavourable of the following scenarios:
(1 + 0dyn + 00 /2) (HSLM or RT ) (2.11)

or
(LM 71 + SW/0) (2.12)

being 0dyn the dynamic factor of amplication, obtained as follows:
0dyn = max |ydyn /ystat | 1 (2.13)

where ydyn is the maximum dynamic response and ystat the corresponding maximum
static response at any particular point in the structural element due to a real train
or a HSLM model.
On the other hand, if a dynamic analysis is not required, the static load due
to a real train at speed v [m/s] shall be multiplied by:
1 + = 1 + 0 + 00 for track with standard maintenance (2.14)
or
1 + = 1 + 0 + 0.500 for carefully maintained track (2.15)
with:
K
0 = for K < 0, 76 (2.16)
1 K + K4
and
0 = 1, 325 for K 0, 76 (2.17)
where:
v
K= (2.18)
2L n0
The parameter 00 represents the dynamic increment due to track and wheels irre-
gularities and is obtained by the following expression:
( L )2 n 0 L L 2
00 = 56e 10 + 50 1 e( 20 ) (2.19)
100 80
with a = min(V /22, 1), being V the velocity of project in (m/s).
2.5.2 Serviceability limit states - trac safety

In this section the limits of vibration and deformation in railway bridges, which
should be taken into account for the design of new railway bridges, will be reviewed.
Excessive deformations can endanger trac by creating unacceptable changes in
vertical and horizontal track geometry, excessive rail stresses and vibrations in bridge
structures. Excessive vibrations can lead to ballast instability and unacceptable
reduction of wheel-rail contact forces. Excessive deformations can also aect the
loads imposed on the track/bridge system, and create conditions which can cause
passenger discomfort.
In EN1991-2 [13] the criteria for trac safety is dened taking into account
the following dynamic eects on the deck:
i. Vertical accelerations;
ii. Deck twist;
iii. Vertical deformations;
iv. Transverse deformations and vibrations;
v. Longitudinal displacements.
Vertical Accelerations of the Deck

The maximum peak values for bridge deck accelerations, calculated along each track,
shall not exceed the following design values:
28 Dynamic eects
i. 3.5 m/s2 for ballasted track;
ii. 5 m/s2 for direct fastened tracks with track and structural elements designed
for high speed trac.
Also, the range of frequencies to take into account in the determination of the
dynamic response in terms of accelerations, shall not exceed the maximum of the
following values:
i. 30 Hz;
ii. 1.5 n0 ;
iii. the frequency of the third mode of vibration of the member in study.
Deck Twist
The twist of the bridge deck shall be calculated taking into account the characteristic
values of the LM 71 model, as well as SW/0 or SW/2, appropriate multiplied by
and . The Load Model HSML should be considered too, including centrifugal
eects.
The maximum twist t [mm/3 m] of a track gauge s [m] of 1.435 m measured
over a length of 3 m, should not exceed the values given in the table presented in
gure 2.9, where t1 = 4.5, t2 = 3.0 and t3 = 1.5.
Figure 2.9: Denition of deck twist and limit values. Extracted from EN1990-A2
[12].
Vertical Deformations of the Deck

For all structures congurations, loaded with the classied characteristic vertical
LM 71, or with the models SW/0 and SW/2 if required, the maximum total vertical
deection measured along any track due to railway trac actions should not exceed
L/600. The angular rotations at the end of decks, represented in gure 2.10, in the
vicinity of expansion devices, switches and crossings, should be veried.
Figure 2.10: Denition of angular rotations at the end of decks. Extracted from
EN1990-A2 [12].
Transverse Deformation and Vibration of the Deck

EN1990-A2 [12] establishes that the transversal deformation and vibration of the
deck shall be checked for the model LM 71 and SW/0 multiplied by the dynamic
factor , or by the real trains aected by the relevant dynamic factor, if appropri-
ate. Additionally, the designer must consider wind loads, nosing forces, centrifugal
forces in accordance with EN1991-2 [13], and the eect of a transversal dierential
temperature across the bridge.
The transversal deection, h , at the top of the deck also should be limited
according to the values indicated in Table A2.8 - EN1990-A2 [12]. In order to
prevent excessive lateral displacements or vibrations, is recommend that the rst
natural frequency of lateral vibration of a span should be higher than 1.2 Hz.
Longitudinal displacement of the deck

For rails on the bridge and on the adjacent abutment, the permissible additional rail
stresses due to the combined response of the struture and the track, due to variable
actions, should be limited to the following design values:
Compression: 72 KN/mm2
Tension: 92 kN/mm2
30 Dynamic eects
Figure 2.11: Model of a track/structure system. Extracted from EN1990-A2 [12].
Figure 2.12: Variation of longitudinal shear force with longitudinal track displace-
ment for one track. Extracted from EN1990-A2 [12].
When determining the combined response of track and structure to traction

and braking forces, these forces should not be applied on the adjacent embankment
unless a complete analysis is carried out considering the approach, passage over and
departure from the bridge of rail trac on the adjacent embankments to evaluate
the most adverse load eects.
For the determination of load eects in the combined track/structure system a

model based upon gure 2.11 may be used where the longitudinal load/displacement
behaviour of the track or rail supports may be represented by the relationship shown
in gure 2.12. The numbers indicated in gure 2.11 represent the following compo-
nents of the track/structure system model:
1. Track;
2. Superstructure (a single deck comprising two span and a single deck with one
span shown);
3. Embankment;
4. Rail expansion device (if present);
5. Longitudinal non-linear springs reproducing the longitudinal load/ displace-

ment behaviour of the track;
6. Longitudinal springs reproducing the longitudinal stiness K of a xed sup-

port to the deck taking into account the stiness of the foundation, piers and
bearings, etc.
This model is very important to evaluate the security of the track structure
and not the structural security. High track deformations can lead to unfavourable
eects for the structure and for vehicles when these are crossing the bridge.
2.5.3 Serviceability limit states - passenger comfort

For these type of verications EN1990-A2 [12] denes the limiting values for the
maximum vertical deection for passenger comfort, as following:
i. Comfort criteria;
ii. Deection criteria for checking passenger comfort;

32 Dynamic eects
iii. Requirements for a dynamic vehicle/bridge interaction analysis for checking

passenger comfort.
Passenger comfort depends on the vertical accelerations, bv , inside the coach.

These levels of comfort limiting values for the vertical accelerations are presented in
table 2.2.
Table 2.2: Recommended accelerations values to ensure the respective levels of com-
fort.
Very good 1.0 m/s2
Good 1.3 m/s2
Acceptable 2.0 m/s2
Figure 2.13: Recommended levels of comfort. Extracted from EN1990-A2 [12].
In order to limit vertical vehicle acceleration, being the limits dened in table
2.2, vertical displacements should be less than the maximum permissible vertical de-
ection, , obtained from gure 2.13. These values are expressed in function of the
span length L [m], and train speed V [km/h], which is valid only for railway bridges
with three or more successive simply supported spans. Alternatively, these acce-
lerations can be determined considering the vehicle-structure interaction dynamic
analysis.
Additionally, the limiting values of L/ , dened in gure 2.13 are given for
bv = 1.0 m/s2
2.6 Methods for Dynamic Analysis 33
Vertical deections should be determined with the LM 71 model multiplied by

the factor and adopting = 1, being only one track loaded for the case of bridges
with two or more tracks.
2.6 Methods for Dynamic Analysis

Dynamic analysis has been previously referred to, but which type of analysis that
should be performed was not yet specied. As remarked in chapter 1, several meth-
ods have been developed, in an eort to study the dynamic eects in railway bridges
due to high speed trac.
Analytical methods based on direct time integration, or on modal analysis,
have the advantage to make possible the comprehension of the basic principles of
the dynamic response of railway bridges due to high speed trac. These methods
are usually used by nite element programmes and the only disadvantage is related
to the time of the analysis of complex structures, sometimes limiting the analysis of
dierent structural solutions and consequent optimization.
On the other hand, the simplied methods available for dynamic analysis of
railway bridges, are restricted only to simply supported bridges. These methods
are based on the Decomposition of the Resonance Excitation (DER) and on the
Residual Inuence Line (LIR).
Empirical methods are based on real examples, being restricted to several
typologies of structures.
Finally, the methods based on vehicle-structure interaction are applied if a
more rigorous dynamic analysis is required. With this method is possible to study
vehicles vertical dynamic behaviour. These methods will be discussed in the follow-
ing sections.
2.7 Models Based on Point Loads

Considering that the eects of interaction between the structure and vehicles are not
included in these dynamic analysis, in which is considered the interaction between
the bridge and the primary and secondary suspensions, the vehicles are taken as rigid
structures without interaction with the bridge. The loads per axle in the dierent
models are dened by a nominal value, which corresponds to the static load of the
train when stopped. This is why these models are so called as point loads models,
which are used in the methods reviewed in the next sections. Hence, a train model is
characterised by a serie of moving loads with constant values per axle and constant
34 Dynamic eects
spacing between them, as shown in Figure 2.14. The validity for the use of these
Figure 2.14: Point Loads dening the actions of a real train (train AVE).
models is discussed by Dominguez Barbero in [2]section 4.6, in which the principles

for the use of these models are dened, comparing with those that take into account
the vehiclestructure interaction phenomenon. The use of point load models does
not give an exact solution of the problem but most of the time is sucient for
the evaluation of the dynamic behaviour of these structures. On the other hand,
in order to perform a dynamic analysis considering the phenomenon of vehicle
structure interaction, specic train characteristic values (mechanical properties of
suspensions, masses, geometry) are needed, which is some times a dicult task for
the engineers.
2.8 Methods Based on Finite Element Models

Finite element methods are applicable to arbitrary structures if, as usual, the struc-
tural behaviour is linear or even nonlinear. These methods are based on direct time
integration of the dynamic equilibrium equations of the structure, under the actions
of a train of loads. The structural model6 may be analysed through the integration
of the complete model, taking this as a discret system with N degrees of freedom,
or through a reduction of the number of degrees of freedom by a previous modal
analysis.
6 In
this work all the models where modeled in the program of nite elements FEAP [30] or in a
version of this one created by the Mechanical Group of Civil Engineering school of Madrid, named
as FFCC
2.8 Methods Based on Finite Element Models 35
Direct Time Integration Methods

The general dynamic equation for the model of a struture with N degrees of freedom
is given by the following system of equations:
Md + Cd + Kd = f (t) (2.20)
where M is the mass matrix, C the damping matrix, K the stiness matrix, f (t) the
vector of external loads and d the unknown vector of nodal displacements. A direct
time integration of the model solves the complete system (2.20) for each time step,
previously dened. Taking into account that the equations are generally coupled,
they must be solved simultaneously. This methodology was not used in this work.
Modal analysis
Modal analysis is only used for structures with linear behaviour and is based on
the reduction of the number of degrees of freedom, N , of the initial system (2.20).
In a rst stage is made a selection of the n signicant eigenmodes of vibration
(N n), being then the problem integrated in time for each mode shape with the
respective frequency of vibration, i (i = 1, .., n). With this procedure the equations
are decoupled and the modal response of each mode is obtained from the dynamic
equation (2.20) similar to that of a system with a single degree of freedom. A
comparasion between the two methods was performed by Dominguez Barbero in
[2], in which concluded that the results are very similar and acceptable. However,
the number of modes to adopt in the analysis must be representative of the bridge
dynamic response, as commented in section 2.9.3.
Modelling a Train of Moving Loads

The simplest procedure to dene a train of moving loads, in a nite element model,
is applying load histories in each convenient node. For a time step ti and an axle
force, F , a nodal load Fj is assigned to the node j (j = 1, 2, ..) if the axle is above
an element that contains that node. The magnitude of Fj depends linearly on the
distance from the axle to the node. This procedure is outlined in gure 2.15 for a
single moving load, where telem represents the time at which a load, with a certain
velocity, takes to cross the element.
36 Dynamic eects
F
t0 = 0 F 1111
0000
0000
1111
1111
0000
111
000
v 0000
1111
0000
1111
0000
1111
0000
1111
0000
1111
0000
1111
0
1 0
1 0
1 0000
1111
11
0 21
0 31
0
t0 t1 t[s]
F
t1 = telem F 1111
0000
000
111
0000
1111
000
111
111
000 1111
0000
000
111
v 0000
1111
000
111
0000
1111
000
111
0000
1111
000
111
0000
1111
000
111
0000
1111
000
111
0
1 0
1 0
1 0000
1111
000
111
111111111111
000000000000
0000
1111
000
111
11
0 21
0 31
0
t0 t1 t2 t[s]
1
0
0
1
0
1
0
1
0
1
0
1
0
1
F0000000
1111111
000
111
000111
000
F 0
1
0
1 111000
111
t2 = 2 telem 0
1 000
111
000
111000
111
000
111
111
000
v 0
1
0 000
111000
111
1
0
1 000
111000
111
0
1 000
111
000
111000
111
0
1 0
1 0
1
0
1
0
1 000111
111000
000
111
1111111111111
0000000000000
11111
00000
0
1 000111
111000
11
0 21
0 31
0
t0 t1 t2 t3 t[s]
Figure 2.15: Nodal force time history denition for a single moving load F , with
speed v .
2.9 Analytical methods based on modal analysis

These type of methods are based in the modal decomposition of the structure and
in the combination of the dierent modes of vibration due to eects of the train of
moving loads. Additionally, the analysis could be more simplied if the bridge is
modeled as a straight beam, if posible.
In this section the modal analysis for a simply supported beam, considering a
single moving load and a train of moving loads, will be reviewed. The solution of
this problem will be considered as a general methodology and can be used by nite
element programs.
2.9.1 Modal Analysis for a Simply Supported Beam

Consider the motion of the simply supported beam of Bernoulli, with constant
stiness, EI , and density, , due to the actions of a single moving load. From
the analysis of the problem, discussed in reference [2], it is possible to dene the
dynamic equilibrium for the rst mode shape of the beam, by means of the following
2.9 Analytical methods based on modal analysis 37
equation:
F
y1 + 21 1 y1 + 12 y1 = sin(vt/L) (2.21)
M1
with:

x
1 (x) = sin (2.22)
L
1
M1 = L (2.23)
2
s
EI
1 = 2 (2.24)
L4
Studying the case of a train of nF loads, in which each load k is characterised

by its nominal value, Fk , and the corresponding spacing from the rst load in the
head of the train (see gure 2.18), the equation (2.21) can be rewritten for each
mode shape as follows:
nF
X Fk
yi + 2i i yi + i2 yi = i (vt dk ) (2.25)
k=1
Mi
being Mi and i the generalized mass and the damping value for the eigenmode i
(i=1,2,...).
Solving this function in time, the response in terms of displacements, velocity
and accelerations at any point of the beam, (xf ), are dened respectively as:
n
X
f (t) = u(xf , t) = yi (t) i (xf ) (2.26)
i=1
n
X
f (t) = u(xf , t) = yi (t) i (xf ) (2.27)
i=1
n
X
f (t) = u(xf , t) = yi (t) i (xf ) (2.28)
i=1
in which for a single moving load, n = 1.

38 Dynamic eects
2.9.2 Modes of vibration

The function of the deformed shape for each mode of vibration of the beam, i (x),
can be determined analytical or numerically according to the cases that are being
studied. Considering the free vibrations analysis of a simply supported beam, is
possible to dene i (x) as a solution of the equation (2.29), as described in reference
[2].
4 i (x)
4
= i2 i (x) (2.29)
x EI
The solution of this equation is performed taking account of the boundary con-
ditions of the beam and the conditions of deection, rotation, bending moment and
shear forces above the internal and external supports of the beam. As an example,
for a simply supported beam those conditions would be dened by displacements
and bending moments equal to zero on the two supports of the beam. In gure 2.16
the two rsts modes of vibration for this case are presented.
x x
1 (x) = sin(x/L) 2 (x) = sin(2x/L)
Figure 2.16: Two rsts modes of vibration of a simply supported beam and respective
analytical expression.
In general cases, numerical methods solved by computer are necessary to de-

termine the dierent eigenvalues of the struture.
2.9.3 Number of Modes of Vibration to Consider in the Anal-

ysis
Like was referred to before, with the modal analysis method, a bridge model with
N degrees of freedom can be reasonably described by n modal equations, each with
one degree of freedom (N n). This simplication is valid whenever the system is
suciently described by those n modes of vibration considered.
Some studies, performed by the subcommittee ERRI D214 [8], concluded that
vibrations with frequencies fi above 20 Hz do not had sucient energy to contribute
for the bridge dynamic response. The same studies concluded that the response in
2.10 Simplied Methods According to the Train Dynamic Signature 39
terms of displacements, in the midspan of a simply supported beam, is approximately

represented considering only the contribution of the rst mode of vibration.
However, this limit was extended up to 30 Hz due to further studies, in which
was concluded that the transfer function, represented in gure 2.17, has a peak
above 20 Hz for the dynamic acceleration response. This new value was adopted
by EN1990-A2 [12], which proposed the consideration of eigenmodes with respective
mode frequency up to 30 Hz or 1.5 times the frequency of the rst mode of vibration
or the frequency of the third mode of vibration.
Figure 2.17: Transfer function of ballast. Extracted from [36].
Recent studies made by the Federal Institute Research and Testing (BAM) in
Berlin [36], shown that the maximum value of the transfer function is near 60 Hz
(gure 2.17), being proposed for the design of railway bridges a new maximum up
to 60 Hz or the frequency of the third mode of vibration. However, the conclusions
of this study were not still considered in any engineering code.
2.10 Simplied Methods According to the Train Dy-

namic Signature
In following sections simplied methods according to train dynamic signature used in
vertical dynamic analysis of railway bridges, will be reviewed. These methods were
developed by the subcommittee ERRI D214 [8] and are based on the Decomposition
of the Resonance Excitation (DER) and on the Residual Inuence Line (LIR) both
using the concept of train dynamic signature.
40 Dynamic eects
The train dynamic signature is taken as a geometric curve that characterise

the train aggressivity in the dynamic behaviour of the bridge.
These methods determine the response as a combination of harmonic series,
and establish an upper bound of this sum, avoiding a direct dynamic analysis by
time integration. However, the application of these methods is limited to simply
supported bridges, which can be represented dynamically by means of a single har-
monic vibration mode, as discussed before. They have been developed and used for
a number of years within SNCF (French National Railway Company).
2.10.1 DER method

The development used in DER method begins from the analysis of the frequency
of excitation produced by a train of moving loads. This method is based in the
following assumptions:
Applicable only on statistically determinated bridges;
For the analysis of statistically determinated bridges, is considered that the

dynamic response is signicantly represented by the rst mode of vibration,
as shown in Figure 2.16.
Decomposing the dynamic response of the bridge in a Fourier series, the DER
method, centres the study in the components that belong to resonance frequencies,
considering that the main problem in the dynamic behaviour of railway bridges is
the phenomenon of resonance. Thus, the maximum acceleration value is obtained
as a product of two functions: the rst one depends on bridge characteristics and
the second one depends on the train dynamic signature.
Hence, the development of the method is summarised as follows:
1. Reduce the response of a statistically determinate beam to a single degree of

freedom system;
2. Decomposition of the dynamic response of the bridge, in Fourier series;
3. Consideration of the term which corresponds to the condition of resonance

frequencies.
Considering the analytical development of the method, presented in [2], the

maximum accelerations at the midspan of the beam for a certain speed, is given as
111F11111111
000 00000000 F2 F1
000
111
000
111
00000000
11111111
k
00000000
11111111
...
11111
00000
... F4 F3
1111111
0000000 11
00 0000
1111
00000
11111 0000000
1111111
0000000 11
00
00 0000
000
111 1111
00011111111
111 00000000
00000000
11111111
00000000
11111111
00000
11111 1111111 11 0000
1111
0000
1111
x2
x3
xk
Figure 2.18: Denition of the distances between the axles of the train of loads from
the rst axle of the front car.
follows:
y(t) Ct A(L/) G() (2.30)

where the rst factor is a constant that depends from bridge characteristics:
8f02 4
Ct = = (2.31)
K L
The second factor is a function called dynamic inuence line:

cos(L/)
A(L/) =
(2.32)
(2L/)2 1
and the third factor represents the train dynamic signature, dened as follows:
v
u
u XN 2 XN 2
2xk 2xk
G() = t Fk cos + Fk sin
k=1
k=1

x
L
2 N
1e (2.33)
xN
being only dependent on the train characteristics and on the bridge damping coef-
cient. The value xk is dened as shown in gure 2.18.
Subtrain concept
From the expression (2.33) is possible to conclude that the maximum acceleration
occurs only when the last load leaves the bridge. However, real measurements in
42 Dynamic eects
real railway bridges revealed that this is not what happens. In certain situations,
the maximum dynamic response occurs before the train leaves the bridge. To take
into account this fact, the concept of subtrain was introduced, being the value of
G() taken as the maximum of the values obtained for each subtrain, as expressed
in (2.34).
v
uXN 2 XN 2
1 u 2xk 2xk
G() = max t Fk cos + Fk sin
i=1...(N ) xi
k=1 k=1
x

2 i
1e (2.34)
In gure 2.19 the concept of subtrain is represented.
1
0F2 F1
0
1
0
1
0
1
0
1 Sub1
0
1
0
1
0
1
0
1
0
1
0
1
0
1
0
1
F3 F2 F1
1
0
0
1
0
1
0
1
0
1
Sub2
0
1
0
1
0
1
0
1
0
1
0
1
0
1
F4 F3 0
1
F2 F1
0
1
0
1
0
1
0
1 Sub3
0
1
0
1
0
1
0
1
0
1
0
1
0
1
0
1
Figure 2.19: Denition of Subtrain. Representation of the three rsts subtrains of

the train model.
2.10.2 LIR method

This method is based on the residual inuence line and like the DER method is
applicable only on statistically determinated bridges. The mathematic development
of the LIR method begins with the analysis of the vibrations in the Bernoulli beam
due to a single moving load.
This procedure is based on the analysis of the residual free vibrations after
each individual single load crosses the simply supported bridge. The analytical
development of the method is described in [2], as for the DER method, in which the
nal solution in terms of maximum displacements and accelerations, in the midspan
of the simply supported beam, are dened as follows:
ymax = Cdesp A(r) G() (2.35)

ymax = Cacel A(r) G() (2.36)
being Cdesp and Cacel given by:
1
Cdesp = (2.37)
M 02
1
Cacel = (2.38)
M
The factor A(r) is named as the dynamic inuence line and is given as7 :
r
r 2 r

A(r) = e + 1 + 2 cos e r (2.39)
1 r2 r
with r = /2L.
G() is named train dynamic signature but dened in a dierent way from
the DER method 8 , depending too from train characteristics and from the damping
coecient of the structure, given as follows:
v" #2 " #2
u
u X x x
x x
i
X i
e2 e2
i i
G() = t Fi cos 2 + Fi sin 2
i
i

(2.40)
7 Only depends from the bridge parametres.
8 The dynamic train signature have two distinct forms for the DER and LIR methods.
44 Dynamic eects
However, similarly to the DER method, the maximum response may not be
produced only when the last load of the train leaves the bridge, being necessary to
consider the G() as the maximum value of G() for each dierent subtrain:
v" #2 " x #2
u xi i
N t X
u X
G() = max Fi cos (2i ) e2i + Fi sin (2i ) e2i (2.41)
i=1
x1 x1
x1 xi
with i =

Train Signatures for the European Real HS Trains

6000
Talgo
AVE
ETR-Y
Eurostar
5000 ICE2
Thalys
Virgin
4000
G()[kN]
3000
2000
1000
0
5 10 15 20 25 30 35
Wavelength [m]
Figure 2.20: Train dynamic signatures for the seven real trains, considering a damp-
ing value of =0.00.
As an example, a simple program in Matlab code [11] was developed in order

to generate the dierent train dynamic signatures for the seven real trains and for
the ten HSLMA models, presented in gure 2.20 and gure 2.21, for a damping
coecient of = 0.00. The envelope values for the train dynamic signatures of the
real trains and the HSLM model are dened in gure 2.22, and as was expected,
the envelope values for the HSLM model cover the eects of the real trains. In
gure 2.23 the dierent train dynamic signatures for the AVE train are presented
but considering distinct values of damping, .
Train Signatures for HSLM-A

8000
A1
A2
A3
7000 A4
A5
A6
A7
6000 A8
A9
A10
5000
G()[kN]
4000
3000
2000
1000
0
5 10 15 20 25 30 35
Wavelength [m]
Figure 2.21: Train dynamic signatures for HSLMA model, considering a damping
value of =0.00.
Train Signature Envelopes

8000
Real HSML - A
Real HS Trains
7000
6000
5000
G()[kN]
4000
3000
2000
1000
0
5 10 15 20 25 30 35
Wavelength [m]
Figure 2.22: Envelope values for the real trains and for the HSLM model, for =0.00.
46 Dynamic eects
AVE train signatures

4500
=0 %
=1 %
4000 =2 %
=3 %
=4 %
3500
3000
2500
G() [kN]
2000
1500
1000
500
0
5 10 15 20 25 30 35
Wavelength [m]
Figure 2.23: AVE train dynamic signatures for dierent values of damping .
1.8
1.6
1.4
1.2
1
A(r)
0.8
0.6
0.4
0.2
0
0.2 0.3 0.4 0.5 0.6 0.7 0.8
r = /(2L)
Figure 2.24: Inuence line, A(r), for a simply supported beam from the catalogue of
bridges dened by the ERRI in [8], with L = 15 m and f0 = 5 Hz.
For dierent trains, resonance eects are achieved in dierent circumstances.

Taking as an example the Talgo and AVE train signatures and considering a simply
supported bridge with natural frequency of vibration f0 = 5Hz , those resonance
eects occurred for velocities near 235 km/h and 345 km/h, for the Talgo and AVE
trains, respectively.
On the other hand, the inuence line, A(r), is calculated for specic cases,
depending on the length of the bridge, natural frequency, damping and the range of
speeds that it is been considered for the analysis. Figure 2.24 represents the inuence
line of a simply supported bridge dened in the catalogue of bridges studied by ERRI
and presented in [8]. The span length is equal to L = 15 m, with a damping value
of = 2% and with a natural frequency of vertical vibration, n0 = 5Hz .
Being both, C and A(r) independent of any geometrical or mechanical char-
acteristic of the train, it becomes easier to evaluate the critical parameters of the
bridge and the wavelength that maximises resonance eects on the structure (dis-
placements and accelerations). In order to evaluate the total vertical displacements
on the structure, is necessary to add the static eects due to the trains static loads
to those obtained from this method.
Figure 2.25 and 2.26 represents the envelope of maximum vertical accelerations
for the same bridge, due to the actions of AVE and ICE2 trains, obtained with
the LIR method and a direct integration in time of the dynamic response of the
bridge. For this integration, a program developed in Octave [11] (Matlab code) and
developed by Domnguez Barebero, was used.
As may be seen, the LIR method covers the most signicant dynamic eects on
the bridge, due to resonance phenomena. In the case of the ICE2 train, the results
obtained with the LIR method covers the peaks of accelerations, due to resonance,
veried at v = 160 km/h and v = 230 km/h, as well as the results obtained for
speed above v = 200 km/h. There are certain values for which the method does
not give an envelope value, being this related with the denition of the inuence
dynamic line that presents some problems for speed under v = 200 km/h.
For AVE train, the results obtained do not cover the maximum dynamic re-
sponses above v = 200 km/h, but cover the maximum response at v = 355 km/h, as
was expected. However, a dierent result was expected and so the values obtained
should be checked carefully to conclude if, eventually, in this case, for this train in
this bridge these results are in fact correct.
48 Dynamic eects
Train AVE
30
Train signature
Direct integration in time
25
20
Acceleration [m/s2]
15
10
0
100 150 200 250 300 350 400 450
Velocity [km/h]
Figure 2.25: Envelope of maximum vertical accelerations as a function of train speed

for a bridge of the catalogue ERRI with L = 15 m, f0 = 5 Hz and damping = 2%
due to the passage of the train AVE, obtained with the LIR method and with a direct
integration in time.
Train ICE2
10
Train signature
Direct integration in time
9
7
Acceleration [m/s2]
0
100 150 200 250 300 350 400 450
Velocity [km/h]
Figure 2.26: Envelope of maximum vertical accelerations as a function of train speed

for a bridge of the catalogue ERRI with L = 15 m, f0 = 5 Hz and damping = 2%
due to the passage of the train ICE2, obtained with the LIR method and with a direct
integration in time.
2.11 Dynamic Analysis with Vehicle-Structure Interaction 49
2.11 Dynamic Analysis with Vehicle-Structure In-

teraction
Point load models consider that the loads transmitted by the wheels to the rails, are
constant and equal to the nominal load in each axle. However, these actions have
variations due to vehicle suspensions. This phenomenon can be considered through
simplied interaction models as the ones presented in this section.
The consideration of the vertical motion of the vehicles with respect to the
bridge deck, allows for a more accurate representation of the dynamic overall be-
haviour. The train is no longer represented by moving loads of xed value, but
rather by point masses, bodies and springs which represent wheels, bogies and ve-
hicle bodies. In some cases these models may have a non negligible inuence on the
dynamic response of the bridge, like in continuous bridges where point model loads
are appropriate. On the other hand the consideration of vehicle-structure interac-
tion has real signicance in statistically determinate bridges with short spans and
small damping ratios where, according to the results given in [2]section 4.6, the
displacements and accelerations in resonant situations may reach a reduction of 30
%, as shown in gure 2.29.
A general model for a conventional coach on two bogies is shown in gure 2.27,
including the stiness and damping (Kp , cp ) of the primary suspension of each axle,
the secondary suspension of the bogies (Ks , cs ), the unsprung mass of wheels (Mw ),
the bogies (Mb , Jb ), and the vehicle body (M, J ). Similar models may be developed
for articulated or regular trains.
Figure 2.27: Complete vehicle-structure interaction model.
Additionally, a more simplied interaction model can be considered. For a

train with k axles, each axle j can be represented as shown in gure 2.28, with an
interaction element j with a non-suspended mass mjs , a suspended mass mja , and a
50 Dynamic eects
stiness and dampers constants k j and cj , respectively.
Figure 2.28: Load train with vehicle-bridge interaction: simplied interaction model
and variables denition.
The model thus obtained considers a degree of freedom for each mode of the
structure and an extra one for each interaction element. If we consider n modes of
vibration of the structure and k loads, the problem will be described by a system of
n + k dierential equations. This model was implemented by Domnguez Barbero
and the complete description of the methodology is presented in [2].
Figure 2.29: Reduction of the dynamic response in simply supported bridges with
dierent span lengths and damping = 2%. Bridges ERRI with spans L = 10, 30
and 40 m (L = 20 m, f0 = 4 Hz, = 20000 kg/m, U IC71 = 11.79 mm; L = 30 m,
f0 = 3 Hz, = 25000 kg/m, U IC71 = 15.07 mm; L = 40 m, f0 = 3 Hz, = 30000
kg/m, U IC71 = 11.81 mm). Performed by J.M. Goicolea [28].
Chapter 3
Vertical Railway Dynamic Analysis -
Case Study
3.1 Introduction
In order to apply some of the methodologies and normative requirements reviewed
in chapter 2, the dynamic analysis of the railway "Arroyo las Piedras" viaduct [31],
will be performed and reported in this chapter.
Figure 3.1: "Arroyo las Piedras" viaduct.
The viaduct in study is located in the new high-speed line between Crdoba
52 Vertical Railway Dynamic Analysis - Case Study
and Mlaga, in Spain. Is the rst composite steel-concrete high speed railway bridge
in Spain, being an innovative solution in these type of bridges for high speed railway
lines. It brings the strict box girder methods developed for road bridges over the
last few years to railway lines. The new design focuses on typical twin plate girder
solutions, frequently used in Europe, but modied to improve them with strict
box girder capabilities. This characteristic is very important in order to provide the
structure with the torsional stiness required for adequately controlling the dynamic
response when railway trac eccentrically runs along a single track. On the other
hand, the piers are remarkably high, since several of them are higher than 93 m.
This bridge has 20 spans, being the rst one of 50.5 m, the centre ones of
63.5 m and the two nal ones of 44 m and 35 m. The viaduct has a soft curvature
but with a radius grater than 7000 m, being considered as a straight viaduct for
the analysis. The cross section of the deck is made of two 3.85 m deep twin-plate
girders plus a top slab 14 m wide, whose thickness varies from 0.41 m in the deck
longitudinal axis to 0.22 m at the edge of the overhangs. The result is a composite
steel-concrete cross-section with a constant total depth of 4.2 m approximately, as
shown in gure 3.2.
Figure 3.2: Cross-section of the "Arroyo las Piedras" viaduct, near a pier.
In section 2 of this chapter the methodology and the assumptions adopted for
the dynamic analysis, will be described.
In section 3 the nite element model of the viaduct used for the analysis, will
be described. The dierent mode shapes and respective mode frequencies of the
structure obtained from the modal analysis performed, will be reported.
In section 4 the signicant results, obtained from the dynamic analysis, will
be presented, such as the maximum vertical displacements and accelerations in the
midspan of each span of the viaduct, due to the actions of the AVE train. Aditionally,
3.2 Methodology 53
a dynamic analysis will be performed, considering the actions of the seven real trains,
in order to evaluate the dynamic response in the midspan of the span 3.
In sections 5 and 6, verication of the limit states in accordance with the
established in EN1990-A2 [12], will be performed.
Finally in section 7, conclusions about the bridge dynamic behaviour, will be
summarised.
3.2 Methodology
The dynamic analysis was performed taking into account the following methodology:
Development of the model of the viaduct in the nite element program FEAP;
Modal analysis using the nite element model of the structure. Evaluation of
the number of modes to consider in the dynamic analyses;
Denition of the load time histories for each node of the model and for each
velocity considered in the analysis. The speed range was dened from v = 120
km/h up to v = 420 km/h, with an increment of v = 10 km/h;
Direct time integration of the eigenmodes considered for the analysis, due to
the actions of the train AVE and for the speed range dened, in order to obtain
the dynamic response of the viaduct in the midspan of each span;
Determination of the maximum eects in each span of the structure, from the
results of the last stage;
Determination of the static eects due to the actions of the LM 71 model and
of the seven real trains, in the midspan of the span 3 (span with the maximum
dynamic response due to the train AVE);
Verication of the limit states for the span 3.
The choice to perform a more specic dynamic analysis in order to evaluate

the dynamic eects in the midspan of the span 3, and respective verication of the
limit states, is related to the fact that this was the span of the viaduct with the
maximum dynamic eects due to the actions of the train AVE.
3.3 Bridge Model

3.3.1 Finite Element Model of the Structure
In order to perform the dynamic analysis, it was proposed a nite element model
of the viaduct, in which were considered the twenty spans of the structure with the
exact dimensions: one span with 50.5 m, seventeen with 63.5 m and the two nals
with 44 m and 35 m, summing a total length of 1209 m as shown in gure 3.3.
Figure 3.3: Lateral view of the "Arroyo las Piedras" viaduct. Representation of the
twenty spans.
For the vertical dynamic analysis, no piers were considered and the structure
was modeled as a continuous beam with three-dimensional frame elements. Each
frame element has two nodes with 6 degrees of freedom each, and a constant length
of 0.5 m. The model is made of 2418 elements and 21 supports, permitting only the
rotation about the z axis and the translation about x (except in one of the extreme
supports). Thus, the torsion and the vertical deection of the beam are permitted,
being the lateral deection restricted in orther to avoid parasitic eigenmodes of the
structure.
The cross section of the beam was dened with homogenous properties, dened
as following:
Area A = 0.9 m2 ;
Density m = 38333 kg/m3 ;
Modulus of elasticity E = 0.2058 103 GPa;
Moment of inertia for vertical deection If v = 1.8 m4 ;
Moment of inertia for lateral deection If h = 14.0 m4 ;

3.3 Bridge Model 55
Torsional moment of inertia variable along the model, as represented in Figure

3.4 and provide by the designer of the structure:
1.156
0.795
0.331
Span 1:
36.5 m 14 m
1.156 1.156
0.786 0.786
Spans 2 to 18: 0.331
14 m 35.5 m 14 m
1.156
0.877
0.786 0.69
Span 19: 0.331
14 m 16 m 14 m
0.877
0.69
Span 20: 0.331
14 m 21 m
Figure 3.4: Torsional moment of inertia dened for the dierent spans in (m4 ).
3.3.2 Modal Analysis

As referred to in chapter 2, the modal analysis of the structure is essential to sub-
stantially reduce the number of degrees of freedom and consequently the number of
equations that dene the dynamic equilibrium of the viaduct. This analysis provided
the mode shapes and associated frequencies of the viaduct, in order to evaluate the
number of modes of vibration that will be used in the dynamic analysis.
With the nite element model of the viaduct, described in the previous section,
the modal analysis was performed with the nite element program, FEAP1 .
EN1990-A2 [12] requires that the mode shapes considered in the dynamic
analysis should be limited to the number of modes with respective frequencies up
to 30 Hz, in order to perform the evaluation of the maximum deck acceleration due
to railway trac actions.
1 Inthis case was used a dierent version of this program, made by the Dept. of Mechanics and
Structure of the Civil Engineer school of Madrid.
However, in the present study only were consider the mode shapes with respec-
tive eigenfrequencies up to 20 Hz, in order to reduce the time spent in the analysis.
In a real design this number of modes reduction should not be made and the limit of
30 Hz, for the number of modes to be considered in the analysis, should be respected.
The reason for this requirement was dealt with in section (2.9.3).
Hence, it was concluded from the modal analysis that the 102 rst eigenmodes
should be considered in the dynamic analysis, taking into account the fact that the
mode 103 has an eigenfrequency of f103 = 20.31 Hz, above the 20 Hz limit. The
deformed shape of the 8 rst eigenmodes and respective eigenfrequencies are dened
in gure 3.5.
1.28 Hz 1.31 Hz
1.36 Hz 1.43 Hz
1.51 Hz 1.61 Hz
1.71 Hz 1.83 Hz
Figure 3.5: Deformed shape of the 10 rst eigenmodes of the "Arroyo las Piedras"
viaduct and respective eigenfrequencies.
3.4 Dynamic Analysis for Vertical Loads

After the completion of the modal analysis and the denition of the number of modes
of vibration that will be used in the dynamic analysis, a time history of the loads in
each node of the model should be made, as dened in section 2.8. As established in
EN1992-2 [13], only one track was loaded and the one with the most unfavourable
eects for the structure, which in this case is indierent. In order to consider the
eccentric eects of the vertical loads of the train models, it was considered, for
each single load, the action of a torsional moment applicable in the same point and
simultaneously, as shown in gure 3.6. The nominal values of the torsional moments
are dened as:
M i = Pi e i = 1, 2, .., n (3.1)
3.4 Dynamic Analysis for Vertical Loads 57
where Pi is the nominal load per axle, e the eccentricity value and n the number of
axles. The eccentricity was determined according to the established in IAPF [10],
assuming the value e = 2.733 m. Once the vibration modes are known and the
histories of loads for each node of the model dened, the response of each mode of
vibration is integrated on time and the sobreposition of each response, is made. This
procedure is performed numerically by the nite element program, which provide
ecient results. Note that the adoption of smaller time steps, will increase the
time of analysis, being important a previous reection about this parameter. The
F...
M...
F2
1
M2 F1
2 M1
v
3
Figure 3.6: Representation of the vertical moving loads and respective torsional mo-
ments.
time step used for the integration of the dynamic response of the bridge was xed
as T p = 0.02 s, which provide good results, according to the general conclusions
made by Dominguez Barbero in [2] - chapter 4. The damping ratio of the model

0.02
f1 f102 f
Figure 3.7: Denition of the Rayleigh curve for the model.
was dened taking into account a maximum value of = 2%, which was considered
in the model through the Rayleigh curve. This curve is represented in gure 3.7
and was dened through the respectives eigenfrequencies of the rst and the last
eigenmodes, in which were assigned the maximum damping ratio, = 2%. These
conditions ensure that all the eigenmodes have a damping ratio not higher than 0.02.
This method is conservative but ecient and simple to use in this type of analysis.
Note that the adoption of this method is not necessary when a dynamic analysis
based on modal analysis, is used. The damping ratio can be dened exactly for each
mode of vibration. However, the programm used for this analysis does not has this
option and the Rayleigh curve had to be used.
Subsequently, the results of the dynamic analysis in each relevant node, each
speed and each train were postprocessed in order to obtain the maximum displace-
ments and the maximum acceleration in those nodes. Figures from (3.10) to (3.12)
illustrate the maximum dynamic responses in each midspan of the twenty spans of
the structure, for the range of speeds adopted for the analyses. Consider the gure
(3.8) and (3.9), in which the vertical dynamic responses, due to the actions of the
AVE train at 200 km/h and 400 km/h, in the midspan of spans 3 and 18, are repre-
sented. In these gures it is possible to conrm the increase of the dynamic response
with respect of the speed increment, in terms of displacements and accelerations.
Displacement History in Node 292 Acceleration History in Node 292

6 0.5
v = 200km/h v = 200km/h
v = 400km/h v = 400km/h
4 0.4
0.3
2
0.2
Accelerations (m/s2)
Displacement (mm)
0
0.1
-2
0
-4
-0.1
-6
-0.2
-8 -0.3
-10 -0.4
0 2 4 6 8 10 12 14 0 2 4 6 8 10 12 14
Time [s] Time [s]
Figure 3.8: Dynamic response of the deck in terms of displacements and accelerations
due to deection eects in the midspan of the span 3 (node 292) at 200 km/h and
400 km/h.
Displacement History in Node 2197 Accelerations History in Node 2197

4 0.25
v = 200km/h v = 200km/h
v = 400km/h 0.2 v = 400km/h
2 0.15
0.1
0
Accelerations (m/s2)
Displacement (mm)
0.05
0
-2
-0.05
-0.1
-4
-0.15
-6 -0.2
-0.25
-8 -0.3
0 5 10 15 20 25 30 0 5 10 15 20 25 30
Time [s] Time [s]
Figure 3.9: Dynamic response of the deck in displacements and accelerations due to
deection eects in the midspan of the span 18 (node 2197) at 200 km/h and 400
km/h.
10
Maximum Displacement (mm)

8
4
150 200 250 300 350 400
Train Velocity (km/h)
span 1 span 2 span 3 span 4 span 5

8.5
8
7.5
6.5
5.5
150 200 250 300 350 400

8.5
8
7.5
6.5
5.5
5
150 200 250 300 350 400
9
7
1
150 200 250 300 350 400
Figure 3.10: Envelope values for the midspan displacement in each span for dierent
speed due to deection eects.
3.4
3.2
3
2.8
2.6
2.4
2.2
1.8
1.6
1.4
150 200 250 300 350 400

1.9
1.85
1.8
1.75
1.7
1.65
1.6
1.55
1.5
150 200 250 300 350 400

1.9
1.85
1.8
1.75
1.7
1.65
1.6
1.55
1.5
150 200 250 300 350 400
2.6
2.4
2.2
1.8
1.6
1.4
1.2
0.8
150 200 250 300 350 400
Figure 3.11: Envelope values for lateral displacements in the midspan of each span
for dierent speed due to torsional eects.
12
11
Maximum Displacement (mm) 10
5
150 200 250 300 350 400

10
9.5
8.5
7.5
6.5
150 200 250 300 350 400

10
9.5
8.5
7.5
6.5
150 200 250 300 350 400
10
8
2
150 200 250 300 350 400
Figure 3.12: Envelope values for the midspan displacement in each span for dierent
speed due to deection and torsional eects.
From the analysis of the results presented in gure 3.12, is concluded that the
span 3 has the maximum displacements and that, in general, there are no signicant
dynamic eects on the structure, with important peaks due to resonance. The same
conclusion can be taken observing the dynamic response due to deection eects, in
gure 3.10. On the other hand, the dynamic response due to torsional eects has
relevant peaks for speeds around 200 km/h, 250 km/h and 350 km/h. However, in
these specic results a strange phenomenon is visible. It would be expected similar
responses in equal spans, taking into account that the transmission of torsional
eects, from one span to another, is restricted at every supports. In spans 3, 4
and 5, this is not what happens. The reason may be related with the number of
modes of vibration considered for the analysis. Probably there were not consider the
sucient modes of vibration that could represent the correct torsional behaviour of
the structure. If in an accurate dynamic analysis this situation occurs, precaution
must be taken and the model should be checked.
3.5 Limit States Verication

Through the values obtained from the dynamic analysis, according to EN1991-2 [13]
in section 6.4.6.5, and already reviewed in chapter 2, the dynamic amplications
veried in each relevant point for each train model and speed is determined by the
following dynamic enhancement:
0dyn = max |ydyn /ystat | 1 (3.2)

where ydyn is the maximum dynamic response and ystat the corresponding maximum
static response at any particular point in the structure, both due to a Real Train or
a HSML model.
This dynamic factor is further increased due to the consideration of track
defects and vehicle imperfections, by the value 0.500 , for carefully maintained tracks,
or by 00 , for standard maintenance tracks.
Thus, for the design of the structure, the most unfavourable value from the
following enhancements, should be considered:
(1 + 0dyn + 0.500 ) Sstat (3.3)

or
(LM 71stat ) (3.4)
where Sstat represents the static eects due to the real trains or the HSLM models,
in a relevant point, and (LM 71stat ) the static eects due to the LM 71 model, in
3.5 Limit States Verication 63
the same relevant point.
3.5.1 Dynamic Factor (2)

For speeds under 200 km/h the impact factor, 2 , is calculated as dened in chapter
2 - section 2.3.4, for a carefully maintained track:
1.44
2 = + 0.82 (3.5)
L 0.2
with: 1.00 2 1.67
For this structure, the determinant length, L , is dened in EN1991-2 [13]

- table (6.2), by the case 5.2 (Girders and spans over n spans), and is dened as
follows:
P20
Li
L = 1.5 i=1
= 90.675m (3.6)
20
Thus, the impact factor assumed the value 2 = 0, 97447 1, 00.
3.5.2 Real Impact Factor real

In order to evaluate the real impact factor, dened as:
real = 0dyn + 1 (3.7)

is necessary to obtain the maximum static displacements of the structure, for each
train model (RT or HSML), in order to obtain 0dyn , which is dened through the
equation 3.2.
For the determination of the real impact factor was considered the span 3,
taking into account that this was the span with maximum vertical displacements on
the bridge, ydyn , due to the actions of the AVE train.
In order to obtain the maximum relation, ydyn /ystat , all the seven real trains
were considered. Thus, it was performed the dynamic analysis to determine the
maximum dynamic eects in the midspan of the span 3 (node 292), due to the
actions of the real trains and for the same speed range of the previous analysis. In
gure 3.13 is presented the maximum vertical displacements in the node 292 due to
the trac actions of the real trains.
Node 292
12.5
12
11.5
Maximum displacements (mm)
11
10.5
10
9.5
8.5
7.5
7
150 200 250 300 350 400
Train velocity (km/h)
AVE Eurostar Talgo Virgin

ETR-Y ICE2 Thalys
Figure 3.13: Maximum dynamic displacements in the node 292 for the seven Real
Trains, for a speed range dened from v = 120 km/h up to v = 420 km/h.
The evaluation of the maximum static displacement in the node 292, was
performed through a dynamic analysis where the train crossed the bridge with a very
low speed (10 km/h). Is considered that the results given by this simplication are
acceptable and approximated to the ones obtained from an accurate static analysis.
However, faster than a real one. In this quasi-static analysis, the eccentricity of
the vertical loads were considered too, combining both eects, torsional and vertical
deection.
Thus, considering the results presented in table 3.1, the real impact factor is
dened as:
0dyn + 1 = |ydyn /ystat | (3.8)
0dyn + 1 = 0.0121/0.0076 = 1.5853 (3.9)
This factor will be increased by means of the value 00 , calculated through the
expression 2.19 dened in chapter 2. Thus, considering v 22 m/s, the value of 00
obtained was 00 = 0.0024, which is a negligible value, being the real impact factor
nally dened as:
1 + 0dyn + 0.500 = 1.5853 (3.10)
Hence, the static eects of the real trains should be multiplied by this value
3.6 Verication of the Serviceability Limit States 65
Table 3.1: Static and Dynamic eects due to real trains actions.
Train model ystat (m) ydyn (m) ydyn /ystat
AVE 0.0073 0.0113 1.5431
ETR 0.0074 0.0116 1.5806
Eurostar 0.0076 0.0120 1.5785
ICE2 0.0076 0.0121 1.5853
Talgo 0.0085 0.0119 1.3979
Thalys 0.0074 0.0117 1.5823
Virgin 0.0077 0.0099 1.2892
in order to take into account the real dynamic eects produced by these trains on
the viaduct.
3.5.3 LM71 Model Eects

The static displacements due to the application of the LM 71 model, were calculated
in order to determine the enhancement (3.4), dened in section 5 of this chapter. It
was applied on the viaduct in order to obtain the maximum vertical displacements
in the node 292. Taking into account the eccentric eects of this vertical loads, the
maximum static displacement in the node 292 is dened as:
LM 71 = 32.1 + e 1.876 = 37.2 mm (3.11)
Thus, the two enhancements values, dened in section 5, are dened as follows:
(1 + 0dyn + 0.500 ) Sstat = 1.5853 8.5 = 13.48 mm (3.12)

and
(LM 71stat ) = 37.2 mm (3.13)
being the static eects obtained with the LM 71 model, almost three times larger
than the dynamic eects produced by the real trains in the node 292, as shown in
gure 3.14.
3.6 Verication of the Serviceability Limit States

In the evaluation of the serviceability limit states it was only checked the vertical
accelerations and the limiting values for the maximum vertical deection.
Node 292
15
14
(1 + dyn + 0.5 ) Sstat
13
Displacements (mm)
12
11
10
7
150 200 250 300 350 400
AVE Talgo
ETR-Y Thalys
Eurostar Virgin
ICE2 Envelope value - 13.48 mm
Node 292
40
(LM 71stat )
35
Displacements (mm)
30
25
20
15
(1 + dyn + 0.5 ) Sstat
10
150 200 250 300 350 400

AVE Thalys
ETR-Y Virgin
Eurostar Envelope value - 13.48 mm
ICE2 Envelope value - 37.2 mm
Talgo
Figure 3.14: Envelope value for the eects produced in node 292.
3.6 Verication of the Serviceability Limit States 67
To ensure trac safety, in which a dynamic analysis is required, the verication

of the maximum accelerations in the deck of the viaduct, due to the vertical actions
of the railway trac, shall be regarded as a trac safety requirement in order to
prevent track instability. The maximum peak value for bridge deck accelerations,
is specied in chapter 2 and is dened as 1.0 m/s2 for a good level of passenger
comfort. This criteria is veried since the maximum acceleration, obtained from the
dynamic analysis for vertical loads, are less than 0.6 m/s2 in the node 292, as shown
in gure 3.15.
Node 292
0.6
0.5
Maximum accelerations (m/s2)
0.4
0.3
0.2
0.1
0
150 200 250 300 350 400
AVE Eurostar Talgo Virgin

ETR-Y ICE2 Thalys
Figure 3.15: Acceleration values produced in node 292.
In terms of the deection criteria, the results from the dynamic analysis are
acceptable, taking into account the limits established by the EN1990-A2, in section
A2.4.4.3.2.(5) [12] and reviewed in chapter 2. Considering the span length of the
span 3, equal to 63.5 m, and the design speed V = 350 km/h, the upper limit for
the vertical deection is dened as:
(L/) 0.9 = 2600 0.9 (3.14)
being
= 63.5/2600 0.9 = 27.14 mm (3.15)
From the results presented in table (3.1), it was concluded that the deection
criteria for the span 3 is veried, being the maximum dynamic displacement due to
the ICE2 train equal to 12.1 mm, lower than the limit value presented in (3.15).
3.7 Conclusions
From the dynamic analysis performed in this chapter several conclusions can be
drawn and summarised as follows:
i The viaduct has an acceptable vertical dynamic behaviour, considering the

actions of the AVE train, and particularly considering the seven real trains
in the midspan of the span 3 (node 292), which is the point with the most
unfavourable dynamic eects. The dynamic displacements and accelerations
comply with the requirements for trac safety and passenger comfort;
ii The results obtained from the dynamic analysis in the midspan of the span 3
(node 292), performed with the seven real trains, are not signicantly dierent
from the ones obtained with the AVE train. However, the maximum dynamic
displacements occur due the action of the ETR train, and for the ICE2 train
in terms of accelerations;
iii Evaluating the accelerations in the node 292, is important to remark that
there are some resonance eects for velocities near 250 km/h, specially for the
VIRGIN train;
iv The maximum dynamic eects on the viaduct, and particularly for the node
292, are satised for the maximum speed considered in the analysis, v = 420
km/h;
v The statical eects on the structure due to model LM 71, dened in section
2.3.2, cover the dynamic eects due to the actions of the seven real trains,
being almost three times higher. This result was inferred in the analysis of the
dynamic behaviour of the node 292 but the same conclusions should probably
extend to the whole structure;
vi A more specic dynamic analysis should be performed, in order to evaluate

the dynamic behaviour in all the relevant points of the structure, due to the
actions of the seven real trains and the HSLM model;
vii Taking into account that the limit dened in EN1990-A2 [12], for the number
of modes to be considered in the dynamic analysis was not respected, in order
to reduce the time of the analysis, this should be considered in a more specic
study;
viii In spite of these simplications taken into account in the analysis, it is con-
sidered that the results obtained are representative of the general dynamic
behaviour of the structure. The possible increase and decrease of the dynamic
3.7 Conclusions 69
eects due to a more specic dynamic analysis would not be signicant. For
this reason it is concluded that the "Arroyo las Piedras" viaduct has an ac-
ceptable vertical dynamic behaviour.
Chapter 4
Lateral Dynamic Analysis
4.1 Introduction
Railway vehicles aect bridges not only by means of vertical loads, but also through
longitudinal and transversal horizontal loads. These horizontal transversal forces
are generated by lateral motions of railway vehicles from two sources in a straight
track: Horizontal track irregularities and sinusoidal motion of conical wheels along
cylindrical rail heads. Apart from these two sources, centrifugal forces are also orig-
inated in curved tracks, which act on the bridge in the outward direction. However
this case will be not discussed in this work.
Measures and parametric studies made by the committee ERRI D181 [7] have
shown that these eects play an important role in the bridgetrain dynamic interac-
tion, being established some design recommendations in order to cover these lateral
eects and avoid resonance phenomena in vehicle lateral vibrations. These recom-
mendations were adopted by the new European engineering codes ([13, 10]). One of
these recommendations was the limitation of the natural frequency of spans lateral
vibration, flateral , as follows:
flateral 1.2 Hz
The problem of lateral vibration in railway bridges does not aect the safety
of the structure but instead can put in risk the comfort or the security of the passen-
gers. On the other hand, this problem can contribute to an accelerated degradation
of the railway, like ballast migration and instability or the track structural damage,
which aects the stability of vehicles movement.
72 Lateral Dynamic Analysis
Figure 4.1: Derailment possibly due to vehicle dynamic eects.
In order to study the lateral dynamic behaviour of railway vehicles when cross-
ing the "Arroyo las Piedras" viaduct, several models were developed and will be
described in this chapter. The reason for this analysis is related with the important
lateral dynamic eects generated on the viaduct due to the special characteristics
of the structure. For this reason it is interesting to evaluate the lateral dynamic
behaviour of railway vehicles in this situation.
Thus, section 2 focuses on the study of the bridge lateral dynamic behaviour
due to the passage of railway vehicles.
In section 3, the eects of track irregularities and contact forces between the
interface wheel-rail, considered in this study, will be reviewed.
In section 4 the general, geometric and mechanicall train characteristics adopted
for this analysis, will be specied.
In section 5 the one-dimensional models developed for the present study will
be described, as well as the results obtained from the dynamic analysis performed,
considering the eects presented in section 3. These models consist in a model with
one degree of freedom and another with two degrees of freedom.
In section 6 a three-dimensional model will be described and the dynamic
results compared with the other models.
Finally, in section 7, conclusions of the study will be summarised.
4.2 Bridge Lateral Displacements 73
4.2 Bridge Lateral Displacements

Lateral displacements in railway bridges result, fundamentally, from vertical loads
due railway trac and will be studied in this section.
1 1 2
Support Midspan
Figure 4.2: Lateral displacements of the deck due to torsional eects.
In railway bridges with double track, the eccentric vertical loads of railway
trac, presented in section 3.4, are responsible for torsional eects on the deck.
This leads to lateral displacements at the top of the piers with consequent lateral
displacement of the deck, as represented in gure 4.2. In this gure, 1 , represents the
lateral displacement of the deck due to the lateral bending of piers, which depends
on their exibility. However, also the torsion of the deck within spans is responsible
for lateral displacements, being represented in gure 4.2 as 2 . Resonance occurs
when the frequency of excitation, due to lateral impacts and vertical loads, coincides
with the natural frequencies of both eects, torsion and lateral deection of the deck.
In the following, the concept called here as virtual path, will be used.
The virtual path denes the displacement of the track (horizontal or vertical,
depending on the case) at a moving point which follows the train on its motion
along the bridge. Therefore it will be represented as a displacementtime curve.
Note that the virtual path represents deformation of bridge due to train loads only.
The bridge-vehicle interaction is not considered.
It may also be represented as a displacementlongitudinal track coordinate
curve, by a simple change of variables. This change represents the equivalence
between time and train position, v = x t, with appropriate choice of zero value of
coordinates.
In order to perform the study of lateral dynamic behaviour of railway vehicles,
the virtual path of the horizontal lateral displacements of the bridge shall be used.
A virtual path is given for each axle of the train, being dierent from one to
another. In gure 4.3 a virtual path for a certain axle of the AVE train at v = 350
km/h, is presented, in which the red line represents the lateral displacements, 1 ,
and the green line the lateral displacements, 2 . The combination of both eects is
represented by the blue line. From the analysis of this gure, is possible to take the
following conclusions:
lateral displacements of the deck due to bending eects on piers, represented

by the red line, without signicant dynamic vibrations but instead a quasi-
static displacement with a semi-wave length coincident with the total bridge
length;
lateral displacements due to local dynamic torsional eects,within spans, rep-

resented by the green line, with distinct wave lengths along the time, in which
the maximum values are obtained in the midspan of each span of the bridge.
0m 1209 m
8
Lateral displacement (mm)
0
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15
Time(s)
Deck lateral displacement without torsion vibration

Lateral displacement due to torsion vibration
Virtual Path
Figure 4.3: Virtual path measured at a certain axle of the European AVE train at
v = 350 km/h in the 'Arroyo las Piedras' viaduct, which started to cross the bridge
2.05 seconds after the rst axle of the train.
In gure 4.4 the virtual path for the same axle of the AVE train, but for
a speed of v = 400 km/h, is represented. In this case, the lateral displacements
are higher than the ones obtained for v = 350 km/h due to an increment of the
dynamic response. The displacements 1 , still have a quasi-static behaviour but
0m 1209 m
8
Lateral displacement (mm)
0
0 1 2 3 4 5 6 7 8 9 10 11 12 13
Time(s)
Deck lateral displacement without torsion vibration
Lateral displacement due to torsion vibration
Virtual Path
Figure 4.4: Virtual path measured at a certain axle of the European AVE train at
v = 400 km/h in the 'Arroyo las Piedras' viaduct, which started to cross the bridge
1.83 seconds after the rst axle of the train.
the displacements due to local torsional eects, 2 , are much more aected by this
increment in the dynamic response, with higher amplitudes and shorter wave lengths.
Thus, it is concluded that the virtual path depends on the axle considered and on
the respective speed.
For the present work, some virtual paths were performed for some represen-
tative vehicles and for certain velocities when crossing the "Arroyo las Piedras"
viaduct. This study was developed by CIDI (Calculo, Investigacion y Desarollo
para la Ingenieria, S.L.) [4], which was responsible for the development of this part
of the work. The trains considered for the analysis were the ICE2, the AVE and the
Load model for freight UIC 71 (R1), being considered for each train the following
speeds:
- ICE2 and AVE: 50, 100, 150, 200, 250, 300, 350 and 400 (km/h)
- Train R1: 10, 54, 75, 100, 125, 150 (km/h)
Due to the long time needed for computation, only ve axles for each train
were considered. These are uniformly distributed along the train and dened by the
relative distance to the rst axle, as follows:
Train ICE2:
- xaxle[1] = 0 m
- xaxle[2] = 98.51 m
- xaxle[3] = 177.71 m
- xaxle[4] = 256.91 m
- xaxle[5] = 336.06 m
Train AVE:
- xaxle[1] = 0 m
- xaxle[2] = 95.08 m
- xaxle[3] = 200.15 m
- xaxle[4] = 295.23 m
- xaxle[5] = 332.63 m
Train R1:
- xaxle[1] = 0 m
- xaxle[2] = 175 m
- xaxle[3] = 350 m
- xaxle[4] = 425 m
- xaxle[5] = 725 m
For the performance of each virtual path the deformed shape in every 10 cm
along the deck, was considered. t = 0 s was the instant considered in which the
rst axle enters on the bridge. Some of the virtual path performed are presented in
gures 4.8, 4.9, 4.10, 4.11, 4.12 and 4.13.
It is concluded from the analysis of these results that the local torsional eects
of the deck represents the most unfavourable dynamic eects on the bridge, in all the
cases studied. The virtual paths with the maximum displacements correspond to
the R1 train, being the dynamic responses almost equal for the same axle at distinct
speeds. However, the virtual paths for the axles of the AVE train at v = 400 km/h,
are the ones with the most signicant dynamic vibrations, being the maximum
values veried for the axle 3, as represented in gure 4.12. In order to evaluate the
inuence of these eects in the lateral dynamic behaviour of railway vehicles, the
virtual path which corresponds to this axle, will be considered.
Pretending to validate these results, a lateral dynamic analysis of the viaduct
was performed. For the analysis the model developed and described in chapter
3, was used, but considering this time the piers of the structure. The geometric
and mechanical characteristics of these elements are presented in Annex E, being
modeled with frame elements of 1 m and with variable characteristics. The model
was developed in FEAP [30] and the loads dened as described in chapter 3 - section
3.4. The lateral displacements in the midspan of the span 10 (node 1181) were
evaluated and presented in gures 4.6 and 4.7, considering the eects of the AVE
train at 250 km/h, 300 km/h, 350 km/h and 400 km/h. The number of modes of
vibration was limited to 85, due to high computation time and some limitations
of the programme used, being considered only the eigenmodes with frequencies of
vibration up to 10 Hz. In gure 4.5 the rsts four modes of vibration are represented.
0.29 Hz 0.37 Hz
0.45 Hz 0.51 Hz
Figure 4.5: Mode shapes for the rst eigenmodes of lateral vibration.
In the same gure, the time in which the axle 3 of the train crossed the node
1181, is indicated, in which the lateral displacement at this instant, t = 7.12 s, is
approximately equal to 9 mm. This value is in accordance with the results performed
by CIDI and presented in gure 4.13, concluding that this ones are valid.
8 8
250 km/h 250 km/h
300 km/h 300 km/h
7 350 km/h 7 350 km/h
400 km/h 400 km/h
6 6
Displacements [mm]
Displacements [mm]
5 5
4 4
3 3
2 2
1 1
0 0
0 5 10 15 20 25 30 0 5 10 15 20 25 30
Time [s] Time [s]
Figure 4.6: Lateral displacements in node 1181. The gure on the left represents
the displacements due to lateral benging of piers and the gure on the right the
displacements due to local torsional eects.
axle 3 crosses the node

10
250 km/h
9 300 km/h
350 km/h
8 400 km/h
Displacements [mm]
-2
0 5 10 15 20 25 30
Time [s]
Figure 4.7: Total lateral displacements on node 1181. The red mark represents the
instant in which the axle 3 of the AVE train crosses the node.
79
Figure 4.8: Virtual path for the bogie 3 of R1 train. Performed by CIDI [4].
80 Figure 4.9: Virtual path for bogies 1, 3 and 5 of the R1 train. Performed by CIDI [4].
81
Figure 4.10: Virtual path for the bogie 3 of ICE2 train. Performed by CIDI [4].
Figure 4.11: Virtual path for bogies 1, 3 and 5 of the ICE2 train at v = 350km/h. Performed by CIDI [4].
82
83
Figure 4.12: Virtual path for bogies 1, 3 and 5 of the AVE train at v = 400km/h. Performed by CIDI [4].
Figure 4.13: Virtual path for the bogie 3 of AVE train. Performed by CIDI [4].
84
4.3 Lateral Stability of Railway Vehicles 85
4.3 Lateral Stability of Railway Vehicles

Additionally to the excitation provided from the lateral deformation of the structure,
two sources of horizontal lateral forces in a straight track will be considered for the
study of the vehicles lateral dynamic behaviour1 :
Horizontal track irregularities;
Sinusoidal motion of conical wheels along cylindrical rail heads.
These lateral forces can aect the stability of railway vehicles and are responsi-
ble for the generation of lateral instability phenomena such as the commonly known
hunting. This phenomenon is associated with conically shaped or otherwise proles
steel wheels running on steel rails to negotiate gentle curves without slipping. When
a wheelset is disturbed from the central position on tangent track (e.g., due to track
irregularities and bridge lateral displacements) or when the curve is too tight, large
horizontal forces, called creep forces, are generated at the wheel-rail interface. These
horizontal forces are responsible not only for the steering and centring capability but
also, unfortunately, these restoring forces due to coned or proled wheels can result
in the vehicle following a sinusoidal path on tangent track. This is considered as a
self-excited instability inherent in the vehicle's design. Lateral stability models of
railway vehicles are presented in [17] - chapter 7.
4.3.1 Track Irregularities

Description of Track Geometry
Track geometry is dened in terms of four irregularities consisting of gage, cross level,
alignment and vertical surface prole. Gage is dened as the horizontal distance
between two rails. Cross level is the dierence between the elevations of the two
rails. Alignment is the average of the lateral positions of two rails (often referred to
as the centre line). Finally, Vertical surface prole is the average elevation of the
two rails. These parameters are illustrated schematically in gure 4.14.
Alignment and cross-level variations are the major causes of lateral vibration
in railway vehicles, whereas vertical prole has little inuence on lateral vehicle
dynamics. However, gage plays an important role in the lateral stability of railway
vehicles.
1A more occurate description of these eects may be found in [17] - chapters 3 and 4.
These parameters can be dened mathematically through dierent methods, as

described by Fryba [16] or Garg and Dukkipati [17], within others, being one specif-
ically reviewed in the following section. However, the irregularities so dened have
limitations and must be within security values established by several engineerings
codes as the FRA track safety standards [1].
Analytical Representation of Track Geometry

In order to study vehicle-track interactions and to predict vehicle response, some
form of analytical description of track geometry and track irregularities is required as
described in [17] - section 3.8. Because is impossible to establish a complete detailed
analytical representation of the entire track, one must be satised with statistical
representation. Numerous measurements have shown that these irregularities repre-
sent a stationary stochastic process and they may be described by spectral density
functions.
Track geometry models based on PSDs are very useful in determining the
sustained type of vehicle response. They can be used to calculate mean square
values of rail deviations, vibration levels in the vehicle and forces at the wheel-rail
interface. However, this is a limited tool because isolated geometric variations are
obscured by the averaging property of the PSDs. These variations are often the
cause of undesirable responses. This problem is dealt with in the reference [17].
PSD track denition discussed in references as [14, 6, 17, 16] will be reviewed.
Power Spectral Densities of Track Irregularities

The track alignment for a certain point of the railway line, A (x), is dened by the
lateral proles of the right and left rail dened as yr and yl , respectively, as follows:
A (x) = (yr yl )/2 (4.1)
The independent excitation proles can be understood as stochastic Gaussian

ergodic processes. Each of the above mentioned ergodic processes = (x), for
0 x , are characterised by the mean value:
Z L
1
= lim (x)dx (4.2)
L L 0
Figure 4.14: Denitions of track irregularity parameters. (a) Typical track, (b) gage
and alignment, (c) cross level and nominal vertical prole - taken from Garg and
Dukkipati [17].
and the correlation function

Z L
1
R () = lim (x)(x )dx (4.3)
L L 0
which includes for the = 0 the mean square value

Z L
1
2 = lim 2 (x)dx (4.4)
L L 0
The Fourier transformation of the correlation function results in the power

spectral density S which is dened as:
Z
S() = R ()ej d (4.5)

where is the distance frequency and, therefore, the coordinate in the fre-
quency domain. Then, the mean square value 2 may also be obtained from the
spectral density S :
Z
1
2 = S()d (4.6)
2
being sometimes written as
() = 2S() (4.7)
where the so-called one-sided spectral density is dened in the range 0

only. Therefore, the mean square may be obtained as follows:
Z
1
2 = S()d (4.8)
2 0
Measurements of track irregularities made in German railway lines (DB),

shown that the power spectral densities can be standardised by functions as the
one proposed by ARGER/F [27], as V an A , for vertical or lateral alignment
irregularities:
2c
V,A () = A (4.9)
(2r + 2 )(2c + 2 )
The values of the constant factors r and c , dened in [27], are:

r = 0,0206 rad/m,
c = 0,8246 rad/m.
The value A denes the grade of track irregularities, being Alow and Ahigh used
to dene a track with low or high grade of track irregularities, given as in [27]:
Alow = 0, 59233 106 radm,

Ahigh = 1, 58610 106 radm.
Generation of track irregularities

Figure 4.15 shows a power spectrum density of the measured irregularity proles
and the analytical ones. It has to be noted that the measured data represent track
irregularities only for wavelengths between 0.8 m and 84 m. The limitation is due to
the measurement equipment and, therefore, valid measured track irregularities are
restricted to a certain range of frequencies , as may be seen in this gure.
Comparisons of measured data with the analytical power spectral density show
good agreement in the valid range of wavelength or distance frequencies dened
in Figure 4.15. The mentioned distance frequencies range corresponds to a range
between 1.3 Hz and 138.9 Hz at a vehicle speed of v = 400 km/h, which were
determined by the following relation:
v(m/s)
l(m) = (4.10)
f0 (Hz)
Figure 4.15: Example of a power spectral density function - from Karl Popp, Holger
Kruse and Ingo Kaiser [22].
As described by Goicolea Ruigomez [18] and by H. Claus and W. Schiehlen

[5], the numerical process to generate track histories for geometrical irregularities
over a certain distance is given by the following expression:
1
NX
r(x) = 2 An cos(n x + n ) (4.11)
n=0
which represents the spectral density of N discrete frequencies n .

The independent random phase angles n are uniformly distributed in the
range between 0 and 2 . The expression (4.11) is dened within the interval of
frequencies [0 , f ] with an increment set to:
f 0
= (4.12)
N
In turn, An is dened as:
r
1
An = S(n )n for n = 0, 1, 2, ...N 1 (4.13)
2
Figure 4.16 represents an example of a lateral track alignment irregularities

prole, generated for the viaduct in study and adopting a low level of irregularities,
being A = Alow . Note that for a high-speed railway track, this value is reasonable
taking into account that the degree of maintenance of these type of lines has an
important signicance.
The wavelengths adopted to generate this prole are within 0.8 m and 84 m,
which is the range available for the use of the expression (4.11).
Taking into account that the frequencies of lateral vibration of railway vehicles
are within 0.2 Hz and 1.0 Hz, no resonance eects are expected due to the consider-
ation of this track proles. As discussed by Esveld in [14], the waveband associated
with car body accelerations at high speed is from 70 m to 120 m at v = 300 km/h,
which means that for v = 400 km/h it would be higher, being outside the limits
acceptable for the use of the method proposed here.
However, taking into account that this is a high-speed railway line and that the
quality and maintenance of the track is very important, the use of higher wavelengths
may not be a realistic assumption. Thus, the values assumed are reasonable.
The limits established by the Track Safety Standards of FRA [1] are indicated
in table 4.1, being A, B and C dened as follows:
10
Track Irregularities example
6
-2
-4
-6
-8
-10
0 200 400 600 800 1000 1200
Distance (m)
8
Track Irregularities
Perfect track
6
-2
-4
-6
-8
0 20 40 60 80 100
Distance (m)
Figure 4.16: Example of a track lateral alignment irregularities prole for a track
with low irregularities in a total length of 1209m. Representation of the same prole
for a length of 100 m.
Table 4.1: FRA Track Safety Limits for track alignment irregularities[1].
Class of track A (inches) B (inches) C (inches)
6.......... 1/2 3/4 1 21
7.......... 1/2 1/2 1 41
8.......... 1/2 1/2 3/4
9.......... 1/2 1/2 3/4
AThe deviation from uniformity of the mid-chord oset for a 31-foot chord
may not be more than the limits indicated;
BThe deviation from uniformity of the mid-chord oset for a 62-foot chord
may not be more than the limits indicated;
CThe deviation from uniformity of the mid-chord oset for a 124-foot chord
may not be more than the limits indicated.
In this table is specied a maximum deviation for a wavelength of 37.8 m (124-

foot) equal to 19.05 mm (3/4-inches), for tracks with the higher quality - class of
track 9, which proves that the track alignment irregularities prole, generated for
this study, is reasonable and within the security limits.
4.3.2 Physical Behaviour of the Wheelset on a Straight Track

Theory according to Klingel
If a wheelset with conical tire proles is laterally displaced from the central position,
this displacement is counteracted due to dierent rolling radius of the wheels.
This results in a periodical movement of the wheelset which was described
theoretically by Klingel in 1883 [23] and is therefore, often referred to as the Klingel
movement, shown in gure 4.17. This case is also dealt with in [14] where is shown
that the lateral displacements of the wheelset in a perfect rolling movement is given
by the expression:
x
y = y0 sin(2 ) (4.14)
LK
in which y0 and LK are, respectively, the amplitude and the wavelength of the
Figure 4.17: Representation of the Klingel movement on a tangent track.
sinusoidal lateral displacement. In turn, the wavelength is dened as follows:
r
rs
LK = 2 (4.15)
2
where r is the wheel radius in central position of the wheelset, s the track width
and the conicity of the wheel tread (inclination).
The Klingel movement is therefore purely a kinematic movement in which
forces are not included. The lateral displacement, y , is an harmonic, undamped
function of the distance coordinate x as long as the amplitude moves within the
angeway clearance f wc, which is dened as shown in gure 4.18.
Figure 4.18: Flangeway clearance, f wc.

Introducing the speed, the time domain frequency of the Klingel movement is:
v
f= [Hz] (4.16)
LK
If this frequency coincides with the natural frequencies of the rolling stock,
the vehicle ride becomes unstable. The progressively increasing conicity in the case
of worn proles due to increasing lateral axle movement, therefore, has an adverse
eect in this respect.
However the reality is more complex, especially concerning the contact physics
and taking into account the inertia of the wheelset, the hunting motion requires
tangential forces in the contact area. In fact these kinematic oscillations exist but
could have stability problems in the tangent track due to speed increase.
For the present study the Klingel movement will be dened by the parameters
indicated in table 4.2.
Table 4.2: Klingel movement parameters denition.

P arameters V alues
r 0,455 m
s 1,435 m
1:20
Lk 16,055 m
f 6,21
These parameters were assumed according to several cases studied, including

the one specied in [14], in order to obtain the expression (4.14) and thus, to trace
the lateral displacements of the wheelset in the straight track of the bridge in study.
The amplitude of the sinusoidal movement is dened as y0 = 0.007 m, accord-
ing to [19]. Hence, the expression (4.14) results in:
x
y = 0.007 sin(2 ) (4.17)
16.055
and can be re-written as a function of time and represented in gure 4.19, as
follows:
vt
y = 0.007 sin(2 ) (4.18)
16.055
0.008
0.006
0.004
Lateral Displacement [m]

0.002
-0.002
-0.004
-0.006
-0.008
0 0.5 1 1.5 2
Time [s]
Figure 4.19: Lateral displacements of the wheelset given by the theory of Klingel.
Example of the movement for a time interval of 2 seconds at 400km/h.
According to the expression (4.16), this movement has a frequency of excitation

f = 6.21 Hz at v = 400 km/h.
Hunting movement
As is remarked in [14] and already commented here, the Klingel theory is simple and
instructive but does not include the eect of couple axles, mass forces, and adhesion
forces. In fact, the amplitude y0 of the Klingel movement is dependent on alignment
track irregularities, in the dynamic vehicle behaviour, and on the circulation speed.
Generally speaking, the value of y0 due to slip will increase with speed until it
is equal to half the angeway clearance. Flanging then occurs as a result of which
the axle will rebound.
This means that the lateral movement takes on a completely dierent be-
haviour to that described above and known as hunting. The movement changes
from harmonic to a zig-zag shape, in which the wavelength becomes shorter and the
frequency increases quickly until it is in the critical range for the rolling stocks and
resonance occurs.
Wheel-Rail Interaction Forces

The calculation of wheel-rail contact forces has always been important for the dy-
namic analysis of railway vehicles, their running qualities and forces and stresses
acting on the track.
A review of the various wheel-rail rolling contact theories, that have been
developed in the last two decades, was made in [17] chapter 4, in which the initial
portion of the chapter is concerned with the phenomenon of creep. Creep forces
result from the slip phenomena that takes place at the points of contact between
the wheels and rails as the wheelset moves along the rails, which experience lateral
and yaw displacements.
Figure 4.20: Variation of wheel-rail contact point due to track alignment irregulari-
ties.
The dynamic behaviour of a railway vehicle is signicantly aected by these

interaction creep forces in the wheelrail interface of contact. These forces depend
on the adhesion, creep and wear characteristics. The geometry of the wheel and
rail proles aects the adhesion-creep-wear characteristics, as well as the dynamic
behaviour of the vehicle, because the creep forces are signicantly inuenced by the
area of contact and contact stresses between wheel and rail. The surface roughness,
as well as environmental conditions such as contamination due to water, oil, dirt,
snow, and other factors, usually inuence the adhesion-creep-wear characteristics,
as shown in gure 4.21. The materials of the wheel and rail also play an important
role in this regard.
Wheel-rail forces are functions of at least four independent variables (multi-
dimensional problem), particularly creep forces:
Flateral = f (sx , sy , , a/b) (4.19)
where sx , sy and represent the creepages and a/b the shape of the contact
area.
In a computer simulation, the computation of these forces is repeated many
times for each wheel in each integration step. Therefore a short calculation time is
very important. Among many theories developed around this problem, the simplied
4.4 Lateral Vehicle Car Characteristics 97
Figure 4.21: Dierence between wet and dry rail for creep forces. Extracted from
Bombardier [26].
theory used in Kalker's programme FASTSIM [21] appears to be one of the must
ecient.
Taking these facts into account, in the present work, only the kinematic move-
ment proposed by Klingel will be adopted, leaving the analysis of vehicle dynamics
considering the creep forces, for further studies. It is considered that this assump-
tion is reasonable for the current work and will provide acceptable results.
4.4 Lateral Vehicle Car Characteristics

The weight of a vehicles car body is transmitted to rails by components that are
called trucks (USA) or bogies (Europe). Generally, two bogies are used for each vehi-
cle or, like the AVE train, there is a bogie in each joint of two vehicles. In passenger
and freight vehicles, each bogie usually consists of two wheel-axle sets that are con-
nected through some type of primary suspension to the bogie frame. This frame
supports the weight of the car body through a secondary suspension system located
between the car body and the bogie frame. There are dierent kind of bogies -
passenger bogies and freight bogies. Between them there are several dierences that
are studied in [17]2 .
Figure 4.23 represents a vehicle model extracted from [35]. For this work,
in order to study the vehicles lateral dynamic behaviour, only certain degrees of
2 Inchapter 5 of this reference the rail vehicle bogie congurations and suspensions are discussed
as well as the equations of motion of a wheel-axle set on tangent and curved track. Is this same
chapter a bogie model is reproduced being originally proposed by Murray [34]
freedom from those represented in this gure, will be considered. The locomotive
car body is assumed to be rigid and is assigned degrees of freedom with respect
to lateral displacement We and roll, e . For each bogie, the lateral displacement,
wt , and roll, er , motions are allowed. The yaw motion is not considered in this
study at any vehicle component. For wheelsets only lateral displacements, ww , are
permitted. In the same gure, lc and ls represents the longitudinal and the lateral
distances between the suspensions and the centre mass point of the vehicle car body,
being dened as lc = 6.12 m and ls = 1.23 m. These values were adopted from the
geometry of the AVE train locomotive.
3 1
2 v 2
1
1
0 0
1
0
1
0 1
3 0 3 1
1 0
1
yaw roll
Figure 4.22: Representation of yaw and roll motions.
Linear springs and dampers are combined in primary and secondary suspen-
sions. The values of mass, moment of inertia and the roll moment of inertia are
taken into account for the dierent locomotive components.
In table 4.3 lateral and vertical characteristics are indicated for some repre-
sentative trains.
These values are the result of combinations between the particular values of
the dierent spring and damper elements of those locomotives. The total values for
lateral and vertical stiness and damping, indicated in this table, results from the
sum of inverses:
1
kt = 1 (4.20)
kyw
+ k1yt
and
1
ct = 1 1 (4.21)
cyw
+ cyt
Figure 4.23: three-dimensional vehicle model. Extracted from Yean-Seng Wu [35].

Table 4.3: Suspension and Mass characteristics of UIC and ETR-500 train car.
Car UIC Car ETR-500
Primary 12800 8700
Lateral Stiness (kN/m) Secondary 320 256
Total 312 259
Primary - -
Lateral Damping (kN/m((m/s)) Secondary 59 40
Total 59 40
Primary 2920 3220
Vertical Stiness (kN/m) Secondary 860 722
Total 664 590
Primary 5 15
Vertical Damping (kN/m((m/s)) Secondary 74 65
Total 5 12
Suspended Mass (ton) 32 34.23
Frequency of lateral vibration (Hz) 0.5 0.43
where kyw and kyt are the stiness values for primary and secondary suspensions, and
cyw and cyt the damping values for primary and secondary suspensions, respectively.
In some locomotives the primary suspension does not have the primary damping
system, being just a single spring. These values will be considered in the dierent
models dened in this work.
In railway vehicles, both vertical and lateral stiness of the primary suspen-
sion are always high, which is necessary for a stable running of the wheelsets. In
comparation, the secondary suspension is much soft. Additionally, the mass of the
car body is always high (about 32 ton). The result is a distinct dynamic behaviour
between car body and bogies, decoupled by a frequency ratio of about 1:10.
It is known that the frequency of lateral vibration of the trains is low, be-
tween 0.2 Hz to 1 Hz [7], being the excitations with frequencies multiple of these
values responsible for resonance eects in the dynamic lateral response of vehicles.
The frequency of vibration of the vehicles presented in table 4.3, results from the
expression: r
1 K
f0 [Hz] = (4.22)
2 M
where M is the suspended mass and K the total lateral stiness of the train.
In order to evaluate the inuence of the lateral spring stiness, which is often
unknown, on the dynamic behaviour of railway vehicles that may use the "Arroyo
las Piedras" viaduct, a calculation based on ctitious vehicles was performed. These
vehicles had as reference some of the characteristic values of several real vehicles,
provided for this study and presented in table B.2 of annex B. Taking these values
into account the characteristics of the ctitious vehicles resulted from a decrement of
the mass, stiness and damping values of these ones in order to obtain the frequency
of vibration needed, between 0.2 Hz and 1.0 Hz, as presented in table B.1 of annex
B. As an example, for the denition of ctitious vehicles with frequencies of lateral
vibration between 0.43 Hz and 0.5 Hz, the characteristics of ETR-500 and UIC train
car (table 4.3), were taken into account.
The inertia mechanical properties adopted for the dierent components of the
train models, car body and bogies, are dened in table 4.4. This information were
provided by CEIT - Centro de estudios e Investigaciones Tcnicas de Guipzcoa [15]
and were adopted from the characteristic values of the head car of the Bilbao metro
unit. These values were used in [15] in order to develop a vehicle model of this
unit and study the rail corrugation evolution. These are not the characteristics of
a high-speed railway vehicle but, taking into account that no more information was
provided due to condentiality issues, they were adopted. However, it is considered
that characteristics are reasonable and that the use of more accurate values would
not bring important dierences.
Table 4.4: Vehicle inertia properties.

V ehicle Components Inertia(kg.m2 )
Carbody (x) 80346
Carbody (y) 1050380
Carbody (z) 1043910
Bogie (x) 983
Bogie (y) 1799
Bogie (z) 2909
The properties of the wheelsets are not dened in this table, taking account
of the fact that for the present study the wheels are considered as xed to the rails
and massless. If the interaction forces between wheel-rail interface were considered,
the mechanical properties of wheelset should had been considered. However, as
discussed in section 4.3.2, it will be not considered in this work. The co-ordinate
system is presented in gure 4.23.
In gure 4.24 the vertical distances between the centre of mass of the dierent
vehicle components are presented. hcs represents the vertical distance between centre
of gravity of car body and lateral secondary suspension system, hts the vertical
distance between the lateral secondary suspension system and the bogie centre of
gravity, htp the vertical distance between the centre of bogie gravity and the lateral
Figure 4.24: Vertical dimensions of the train model adopted. Extracted from [35].
primary suspension system and r0 the nominal wheel radius. These values were
adopted from Yean-Seng Wu [35] and are dened in table 4.5.
Table 4.5: Geometric vertical properties of vehicle model. See gure 4.24.
Item V alue(m)
hcs 0.75
hts 0.42
htp 0.20
r0 0.455
In the following sections, the dierent train models developed for the analysis
of the vehicles lateral dynamic behaviour when crossing the viaduct under analysis,
will be described and analysed.
4.5 Vehicle Model With One Degree of Freedom

This model was developed in order to analyse in the most simplied way the dynamic
behaviour of the car body when submitted to a lateral base motion, due to the
dierent phenomena described in section 4.3. In gure 4.25 the reduction of the car
body lateral vibrations into a system with a single degree of freedom, is represented.
This model is formed by a spring with stiness k , a dashpot of damping, c, and a
point mass M . The motion of this system is described, according to Newton's second
4.5 Vehicle Model With One Degree of Freedom 103
x(t)
M v
M
x(t)
k c
y(t)
F( t) = ky(t) + cy(t)
Figure 4.25: Reduction of the lateral vibration of railway vehicles to a system with
one degree of freedom.
law and D'Alembert's principle, by the dierential equation (4.23) as follows:
M x(t) + cx(t) + kx(t) = F (t) (4.23)
where x(t) is the displacement of the mass M at time t and F (t) dened as F (t) =
cy + ky . This expression can be rewritten as:
F (t)
x(t) + 2w0 x(t) + w02 x(t) = (4.24)
M
with: r
k
w0 = (4.25)
M
p
wD = w0 1 2 (4.26)
c
= (4.27)
w0 2M
The value of k represents the total lateral stiness of the vehicle, c the total
lateral damping and M the suspended mass. This solution is described in references
[6, 17].
4.5.1 Response due to a prescribed base displacement

In order to validate the model, the response of the system due to a constant pres-
cribed base motion of y = 0.003 m, will be evaluated. The exact solution of this
system due to a prescribed base motion, is given as follows:
c
t F (t)
x(t) = Ae 2M sin(wD t + ) + (4.28)
k
The characteristics of the ETR-500 train car, described in table 4.3, will be
considered, being the stiness, equal to k = 249 kN, the damping value, equal to
c = 40 kN and the mass of the car body dened as M = 34 ton.
0.005
Dynamic Response
Prescribed base displacement
0.0045
0.004
0.0035
Displacement [mm]
0.003
0.0025
0.002
0.0015
0.001
0.0005
0
0 2 4 6 8 10
Time [s]
Figure 4.26: Response of the system with one degree of freedom due to a prescribed
base displacement, y = 0.003 m. Mechanical characteristics adopted from the ETR-
500 train car.
Considering the equation (4.28) in order to estimate the values of A and , it

is necessary to take into the initial conditions x(0) = 0 x(0) = 0, being:
r
1
= (tan) 1
1 (4.29)
2
and
x0
A= (4.30)
sin
Just as a static system may be stable or unstable, so also may be the dynamic
system. The criterion for stability of a static system is that it should return to
its initial conguration after a small disturbance, and the same criterion holds for a
dynamic system. For the damped spring-mass system of the example, it will oscillate
with diminishing amplitude until it approaches its initial rest state.
As may be seen in gure 4.26 the system oscillates with damped vibration
around the value of the prescribed base motion, ry = 0.003 m, until it stops. Note
k
that the frequency of vibration is equal to f0 = = 0.43 Hz with a period of
M
T = 2.32 s.
4.5.2 Dynamic Analysis of the Vehicle Model

Having dened the model, the dynamic behaviour of the system will be studied
considering the following displacements, yt , as an input on the base of the model:
Lateral displacements of the bridge represented by the virtual path dened for
the axle 3 of the AVE train at v = 400 km/h;
Track lateral alignment irregularities prole;
Displacements of the wheelset due to klingel movement eects in a straight

track.
All these eects were already studied and dened in previous sections as well
as the parameters and correspondly justications.
In order to perform the dynamic analysis of this model, a programme in Octave
(Matlab code) was developed, as presented in annex A. This program integrates
the response in time using the Newmark-Beta method, or the so called trapezoidal
method.
For the analysis the following combination of those eects, were considered:
1. Displacements due to virtual path;
2. Displacements due to virtual + track irregularities;
3. Displacements due to virtual + track irregularities + klingel movement.
Dynamic response due to virtual path eects

The dynamic response of the model in terms of displacements, due to virtual path
eects, is represented in gure 4.27, in which the response of the vehicle is quasi-
static without signicant vibrations. The relative lateral displacements between car
10
6
Displacement [mm]
-2
0 2 4 6 8 10 12 14
Time [s]
Virtual Path Relative Displacements
Vehicle Response
Figure 4.27: Response of the vehicle due to the Virtual Path excitation for a type of
train with a natural frequency of vibration f0 = 0.43 Hz (ETR-500 train car).
body and railway track are presented, in the same gure, where is posible to see a
maximum value less than 2 mm.
In order to understand how the natural frequencies of vibration of the vehicles
inuence the dynamic behaviour of the models due to these eects, the study of the
ctitious vehicles dened in section 4.4, was performed. The maximum relative late-
ral displacements and the maximum lateral accelerations obtained for the dierent
ctitious vehicles due to virtual path eects, are presented in gure 4.28.
As was remarked in section 4.4, in order to vary the lateral frequency of ve-
hicles vibration, the stiness and mass values of the reference vehicles presented
in table 4.3 and B.2, were changed and presented in table B.1. Considering that
the damping value does not has inuence in the frequency of vibration, they were
adopted according to the reference values. In order to cover the possibility of exist-
ing vehicles with less damping than the adopted, the same dynamic analysis were
performed but considering half of the initial damping values dened in table B.1.
The results for this case are presented too in gure 4.28.
From the analysis of these results it may be seen that the maximum value
in terms of lateral displacements is given for the ctitious vehicle with a natural
frequency of lateral vibration equal to f0 = 0.76 Hz. The dynamic behaviour of
this vehicle with respect to time is presented in gure 4.29, where is possible to see
relevant vibrations comparing with the response of the ETR-500 train car, with a
frequency of lateral vibration equal to f0 = 0.43 Hz, presented in gure 4.27. The
3.2 0.09
3
0.08
2.8
0.07
2.6
Displacement [mm]
Acceleration [m/s2]
2.4 0.06
2.2 0.05
2
0.04
1.8
0.03
1.6
1.4 0.02
0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
Frequency [Hz] Frequency [Hz]
V. Path V. Path - 50% nominal damping V. Path V. Path - 50% nominal damping
Figure 4.28: Maximum relative displacement and maximum acceleration responses

for the dierent ctitious vehicles due to virtual path excitation.
lateral accelerations of both vehicle models with respect to time, are presented in
gure 4.30.
10
6
Displacement [mm]
-2
0 2 4 6 8 10 12 14
Time [s]
Virtual Path Relative Displacements

Vehicle Response
Figure 4.29: Response of the vehicle due to the Virtual Path excitation for a type of
train with a natural frequency of vibration f0 = 0.76 Hz (Fictitious vehicle).
0.08
0.06
0.04
Acceleration [m/s2]
0.02
-0.02
-0.04
-0.06
0 2 4 6 8 10 12 14
Time [s]
Vehicle Response with natural frequency of vibration f=0.43Hz
Figure 4.30: Acceleration Response of the vehicles due to the Virtual Path excitation
for trains with a natural frequency of vibration f0 = 0.43 Hz (Car ETR-500) and
f0 = 0.76 Hz (Fictitious vehicle).
Dynamic response due to the combination of virtual path and track irre-
gularities eects
Considering now the combination of virtual path and track irregularities eects,
presented in section 4.3.1, the response of the previous models are presented in gure
4.31. From these results is posible to see that the maximum relative displacements
of the ctitious vehicle are higher than the displacements of the vehicle ETR-500
trai car. However, with the accelerations the same phenomenon does not occur, as
shown in gure 4.32.
The results of the dynamic analysis performed for the ctitious vehicles, due
to the combination of these two eects, are presented in gure 4.33. The maximum
relative displacement was obtained for the ctitious vehicle with a frequency of
lateral vibration equal to f0 = 0.89 Hz. In terms of accelerations, the vehicle with
the maximum value is the same as the one of the previous case. However, the gap
between this vehicle response and the others, decreased. The reason for this change
was expected considering that the new excitation has a wider range of frequencies.
Another relevant aspect is the gap between the dynamic responses due to the eects
of the virtual path and due to the combination of both eects. This proves that the
consideration of track irregularities is very important in the dynamic behaviour of
the vehicles.
10
6
Displacement [mm]
-2
0 2 4 6 8 10 12 14
Frequency [Hz]
Vehicle response with natural frequency of vibration f=0.43 Hz

Vehicle response with natural frequency of vibration f=0.76 Hz
Figure 4.31: Vehicles dynamic response in terms of displacements due to the virtual
path plus track irregularities excitation. Vehicles with a natural frequency of vibration
f0 = 0.43 Hz (Car ETR-500) and with a natural frequency of vibration f0 = 0.76 Hz
(Fictitious vehicle).
0.5
0.4
0.3
0.2
Acceleration [m/s2]
0.1
-0.1
-0.2
-0.3
-0.4
-0.5
0 2 4 6 8 10 12 14
Time [s]

Figure 4.32: Acceleration Response of the vehicles due to the Virtual Path plus Track
Irregularities excitation for trains with a natural frequency of vibration f0 = 0.43 Hz
(Car ETR-500) and f0 = 0.76 Hz (Fictitious vehicle).
5 0.6
4.5
0.5
4
0.4
Displacement [mm]
Acceleration [m/s2]
3.5
0.3
3
0.2
2.5
0.1
2
1.5 0
0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
V. Path V. Path
V. Path - 50% nominal damping V. Path - 50% nominal damping
V. Path + Track Irreg. V. Path + Track Irreg.
V. Path + Track Irreg.- 50% nominal damping V. Path + Track Irreg.- 50% nominal damping
Figure 4.33: Maximum relative displacement and maximum acceleration for the dif-
ferent ctitious vehicles due to the combination of virtual path and track irregularities
eects.
Dynamic response due to the combination of virtual path, track irregu-

larities and Klingel movements eects
Finally, the inuence of klingel movement eects was considered in the analysis.
Because it was made for the previous cases, the maximum dynamic responses of the
ctitious vehicles were determined and presented in gure 4.34. From the analysis
of these results is possible to see that the consideration of Klingel movement eects
is so important as the consideration of track irregularities. Note that the increment
in the dynamic response is more signicant for lateral accelerations than for lateral
displacements.
The maximum acceleration corresponds to the vehicle model with a frequency
of vibration equal to f0 = 0.43 Hz, which corresponds to ETR-500 vehicle train car.
The reason for this eect is related with the fact that the combination of all the
eects considered in this study results in an excitation with a predominant frequency
multiple of f0 = 0.43 Hz. The dynamic response of this vehicle is represented in
gure 4.35 in terms of relative lateral displacements and in gure 4.36 in terms of
accelerations. Comparing the results presented in gure 4.35 with the ones presented
in gure 4.27, due to the consideration of the virtual path, is possible to see that
the levels of vibration are more signicant due to higher levels of excitation.
5.5 0.8
5 0.7
4.5 0.6
Displacement [mm]
Acceleration [m/s2]
4 0.5
3.5 0.4
3 0.3
2.5 0.2
2 0.1
1.5 0
0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
V. Path V. Path
V. Path - 50% nominal damping V. Path - 50% nominal damping
V. Path + Track Irreg. V. Path + Track Irreg.
V. Path + Track Irreg.- 50% nominal damping V. Path + Track Irreg.- 50% nominal damping
V. Path + Track Irreg. + contact wheel/rail V. Path + Track Irreg. + contact wheel/rail
V. Path + Track Irreg. + contact wheel/rail - 50% nominal damping V. Path + Track Irreg. + contact wheel/rail - 50% nominal damping
Figure 4.34: Maximum relative displacement and maximum acceleration for the dif-
ferent ctitious vehicles due to the combination of virtual path, track irregularities
and Klingel movements eects.
10
6
Displacement [mm]
-2
0 2 4 6 8 10 12 14
Time [s]
Virtual Path Relative Displacments
Vehicle Response
Figure 4.35: Displacements response of the car ETR-500 due to the combination of
all the three eects considered.
0.8
0.6
0.4
0.2
Aceleration [m/s2]
-0.2
-0.4
-0.6
-0.8
0 2 4 6 8 10 12 14
Time [s]
Acceleration response
Figure 4.36: Accelerations response of the ETR-500 train car due to the combination
of all the three eects considered.
4.6 Model with Two Degree of Freedom
The model presented in the last section considered just one suspended mass, which
represented the car body mass. With a model with two degrees of freedom, the
inuence of bogies mass in the lateral dynamic response of the car body is considered.
This model was developed in the nite element program FEAP [30], as descri-
bed in annex C. The model contains two rigid bodies with mass (M1 , M2 ) connected
by two type of lateral and vertical suspensions (primary and secondary), in which
only the lateral suspensions are considered. Each suspension is represented by one
spring with stiness (k1 , k2 ) and one dashpot of damping (c1 , c2 ). These values
are taken as the sum of the nominal stiness and damping of all the suspensions of
the same level (primary and secondary). The system is represented in gure 4.37.
Taking into account that the problem is expressed in terms of incremental displace-
ments or accelerations, the Newmark-Beta method is used in order to integrate the
dynamic response.
The consideration of a second rigid body in the model, the bogie mass, will
modify the total response of the system in comparation with the response of the
system with a single degree of freedom, producing an expected increment in the
total lateral dynamic response of the car body.
4.6 Model with Two Degree of Freedom 113
M2 x2(t)
k2 c2
x1(t)
M1
k1 c1
y(t)
Figure 4.37: Representation of the system with two degrees of freedom.
4.6.1 Response due to a prescribed base displacement

In order to analyse the response of the model with two degrees of freedom a prescri-
bed constant base displacement of y = 0.003 m was imposed to the model, as for the
previous model. The characteristic values of the model correspond to the ETR-500
Car and are given as follows:
M1 = 2 3.4 ton
M2 = 34 ton
k1 = 8700 kN/m
k2 = 256 kN/m
c1 = 0.0 kN/(m/s)
c2 = 40.0 kN/(m/s)
Note that the bogie mass M1 , was assumed as 10% of the car body mass M2 .
In order to simplify, the model with a single degree of freedom will be named
as model 1 and the model with two degree of freedom as model 2. The dynamic
responses of these two models are represented in gure 4.38, in which the displace-
ments x2 (t) of the car body of the model 2 are larger than the the displacements of
the model 1 and the frequency of vibration is approximately the same, as expected.
Note that this example was useful in order to validate the model with two degrees
5
4.5
4
Displacement (mm) 3.5

3
2.5
2
1.5
1
0.5
0
0 2 4 6 8 10
Time [s]
Response for the model with one degree of freedom

Response for the model with two degree of freedom
Prescribed base displacement
Figure 4.38: Response of a system with two degrees of freedom due to a prescribed
constant base displacement, y = 0.003 m. Results given for the degree of freedom
x2 (t).
of freedom.
Dynamic Analysis of the Vehicle Model

For the analysis of this system no ctitious vehicles will be used, being only con-
sidered the train model with the relevant dynamic behaviour, obtained from the
analysis of the model 1, in order to evaluate the increase in the dynamic response
due to the consideration of a second mass. This train model has the same mechani-
cal characteristics as ETR-500 train car and the response in terms of displacements
is represented in gure 4.39. These values results from the combination of three
eects considered in this work: virtual path, track irregularities and klingel move-
ment. In the same gure the response to this eects of the model 1 is represented
too. Comparing both responses the maximum dierences are not signicant, being
less than 1 mm.
However, there was an important increment in the acceleration response ob-
tained with the model 2. The dierence between the two models is represented in
gure 4.40, in which the accelerations increased almost twice, taking into account
that the maximum acceleration obtained with the model with a single degree of
freedom is a = 0.8 m/s2 and with the 2DOF model, a = 1.568 m/s2 .
The reason for this dierence may be related with the damping value considered
in each model. The model 1 considered a single suspended damper, where the model
4.6 Model with Two Degree of Freedom 115
10
6
Displacements [mm]
Body Car response - 2 degree of freedom model

Body Car response - ODF Model
Difference between the model responses
-2
0 2 4 6 8 10 12
Time [s]
Figure 4.39: Comparation between the response in displacements of the two systems
considered due to the total aects considered in the study.
2 adopted two suspensions being one considered as a spring with no damping. Thus,
the excitation on the base of the system is amplied by this spring that connects
the wheelset and the bogie (M1 ), and this amplied excitation will be transmitted
to the car body, increasing its acceleration. In fact the damping is not zero and in
future studies a damping dierent from zero should be considered in the analysis.
1.5
max(2df )
1
max(sodf )
0.5
Acelerations [m/s2]
-0.5
-1
-1.5
Body Car response - 2 degree of freedom model

Body Car response - ODF Model
-2
0 2 4 6 8 10 12
Time [s]
Figure 4.40: Comparation between the response in accelerations of the two systems
considered due to the total aects considered in the study.
As was remarked, the maximum acceleration response of the model with two
degree of freedom is a = 1.568 m/s2 . This value is outside the limit established by
the FRA [1] in order to comply the criterion of passengers comfort, as reviewed in
the following section.
4.7 Three-Dimensional Vehicle Model

4.7.1 Denition of the model
The three-dimensional model, represented in gure 4.41 was dened in the nite
element program ABAQUS [29], being necessary the knowledgement of geometrical
and mechanical characteristics of a typical high speed train car, to dene it. Ho-
wever, part of the data used in the development of the model does not correspond
to the characteristics of a high-speed train car. In section 4.4 was indicated all the
data needed for the denition of this model as well as the assumptions considered.
Nevertheless, the values assumed are representative. Annex D contains the input
le that denes it.
Figure 4.41: Three-Dimensional vehicle model.
The components of the model were considered as rigid bodies, which are formed
by the car body, bogies and wheelsets. These rigid components were, in turn, dened
through nodes specically located in space, which are the following points:
1. The reference node of the rigid component;

2. The points of connection with the other components.
The points that are assigned to a certain reference point have the same be-
haviour as this one. In Figure 4.42 the nodes of the model are represented.
The connection between components was made through elements that simulate
the dierent suspensions of the vehicle. Each dened node has six degrees of freedom.
4.7 Three-Dimensional Vehicle Model 117
Figure 4.42: Denition of the nodes of the model. The bodies drawn were just dened
for visualization.
The only points with applied boundary conditions were the reference points of the
wheelsets in which only the lateral displacements are allowed.
In order to understand and visualise the behaviour of the model, the compo-
nents of the train were drawn and assigned to the respective reference nodes of each
part. These bodies are named as display bodies and have the same motion that the
respective reference node. In Figure 4.43 the complete model is represented.
Figure 4.43: Three-dimensional railway vehicle model with the mechanical properties
of the car ETR-500 and with the geometrical characteristics of the AVE train.
The suspensions were considered by means of connectors with vertical and

lateral properties, in order to simulate vertical and lateral suspensions, whose values
are dened in table 4.3.
4.7.2 Dynamic Results of the Model

With the model dened the lateral dynamic analysis were performed taking into
account the combination of the following eects:
1. Virtual path;
2. Virtual path + Track Irregularities;
3. Virtual path + Track Irregularities + Klingel movement.
In order to integrate the dynamic response in time, ABAQUS [29] uses a

general direct-integration method, called the Hilber-Hughes-Taylor (HHT) operator
that is an extension of the trapezoidal rule. This operator is implicit: the integration
operator matrix must be inverted, and a set of simultaneous nonlinear dynamic equi-
librium equations must be solved at each time increment. This solution is achieved
iteratively using Newton's method.
The results obtained in terms of displacements are represented in gure 4.44,
4.45 and 4.46 for each combination of eects considered. Additionally, the results
obtained with the other vehicle models due to the same combination of eects, are
presented in the same gures, in which it may be seen that the results are in good
agreement. The dynamic responses do not have signicant dierences, being these
increased with the increment of the excitation.
Figure 4.47 and 4.48 presents the results in terms of maximum relative dis-
placements and accelerations of the three models considered in this study due to each
combination of eects. It may be seen that for each combination, the dierences
between responses are insignicant, with a maximum variation of 0.35 mm. In terms
of accelerations these dierences are higher, as was expected, taking into account
that the acceleration values result from a second time derivative of displacements.
In order to check if these results are within security limits, the safety limits
established by the Federal Railroad Administration (FRA) [1] and by the French
National Railway Company (SNCF), were considered. In table 4.6 the limit for
the carbody lateral peak-to-peak acceleration proposed by FRA, is indicated. This
value must be measured as the algebric dierence between the two extreme values
of measured accelerations in a one second time period and shall not exceed 0.5 9.8
m/s2 . The lter indicated in this table was not used but, taking into account the
accelerations obtained with the model with two degree of freedom, represented in
gure 4.40, the maximum value peak-to-peak obtained was 3 m/s2 .
Combination 1
9
6
Displacement [mm]
-1
0 2 4 6 8 10 12
Time [s]
model with 1 degree of freedom
model with 2 degrees of freedom
Three-dimensional model
Figure 4.44: Results obtained for the dierent models developed in this study due to
the eects of the virtual path.
Combination 2
10
8
Displacement [mm]
-2
0 2 4 6 8 10 12
Time [s]

Figure 4.45: Results obtained for the dierent models developed in this study due to
the combination of the virtual path and track irregularities eects.
Combination 3
10
8
Displacement [mm]
-2
0 2 4 6 8 10 12
Time [s]

Figure 4.46: Results obtained for the dierent models developed in this study due
to consideration of all the eects considered: virtual path, track irregularities and
Klingel movement eects.
Table 4.6: Vehicle/Track Interaction Safety Limits (FRA).

Accelerations Safety limit Filter/window
Carbody Lateral 0.5 g peak-to-peak 10 Hz 1 sec window
On the other hand, the SNCF established that for lateral car body accelerations
the absolute safety criteria is veried for a 2.5 m/s2 , as indicated in table 4.7,
taken as maximum a value and not as a maximum dierence peak-to-peak.
Table 4.7: Vehicle/Track Interaction Safety Limits (SNCF).

Parameter Safety limit Comfort limit
Carbody Lateral Acceleration 2.5 m/s2 1.5 m/s2
However, under normal conditions the acceleration values should not exceed
the limit of comfort, a = 1.5 m/s2 . This limit is indicated in gure 4.48 by the
red line, in which may be seen that the results obtained with the model with two
degrees of freedom violates the comfort limit. The same do not happen with the
three-dimensional model, which will be taken as the nal results of this study.
2.6
2.5
2.4
Relative displacements (mm)

2.3
2.2
2.1
1.9
1.8
1.7
1.6
11
00 11
00 00
11 1
0
1.5
00
11 00
11 00
11 0
1
0.5
00
11
1 1.5
00
11
2 2.5
Combination of effects 11
00
3
0
1
3.5
Model with 1 degree of freedom

Figure 4.47: Representation of the maximum relative displacements to the track

(virtual path) for the three models developed and for the three eects considered in
this study: 1Virtual path; 2virtual path plus track irregularities;3virtual path plus
track irregularities and klingel movement eects.
1.6
Comfort limit
1.4
100%
1.2
Accelerations (m/s )
2
0.8
0.6 42%
0.4
0.2 11%
11
00 11
00 00
11 11
00
0
00
11 00
11 00
11 00
11
0.5
00
11
1
00
11
1.5 2
Combination of effects
00
11
2.5 3
00
11
3.5

Figure 4.48: Representation of the maximum accelerations values for the three mo-
dels developed and for the three eects considered in this study: 1Virtual path;
2virtual path plus track irregularities;3virtual path plus track irregularities and
Klingel movement eects.
In the same gure the percentage of inuence of each combination of eects

in the total lateral dynamic response of the vehicles that cross the "Arroyo las
Piedras" Viaduct, are dened. These values were obtained taking into account the
response of the three-dimensional model. Thus, was concluded that the inuence
of the lateral displacements of the viaduct represents 11% of the total response of
railway vehicles. In turn the combination of bridge lateral displacements with track
irregularities contributed 42% to the total dynamic response, being approximately
31% the contribution of track irregularities. Finally, the consideration of Klingel
movement eects contributed 58% for the total dynamic response, being the more
signicant eect.
Figure 4.49: Example of lateral dynamic behaviour of the three-dimensional model.
4.8 Conclusions
From the dynamic analysis performed with the dierent models developed, several
conclusions can be drawn, as sucently described:
i. The results were obtained considering several eects on railway bridges and on
railway vehicles when moving along a straight double track. In the study of
some of these eects, such as the track irregularities proles and the interaction
between wheelset-rail, simplications were taken and explained. The inuence
of these simplications on the total dynamic response of the vehicles is not
clear but it is considered that the results obtained are relevant and as a rst
approach of the problem can be taken as reference values;
4.8 Conclusions 123
ii. Through the dynamic analysis of the rst model with a single degree of free-
dom, was concluded that the vehicle with the maximum dynamic responses in
terms of accelerations was the ETR-500 car, whose characteristics are dened
in table 4.3. For this reason this was the vehicle taken into account for the
analysis of the other models. Thus, the results obtained are considered as
envelope values of the lateral dynamic response of the trains that may cross
the viaduct;
iii. From the analysis of the maximum relative dynamic displacements of the three
vehicle models, it may be seen that the values are very similar but unexpected
are almost equal for the model with a single degree of freedom and the three-
dimensional model;
iv. From the analysis of the maximum acceleration values obtained, the model
with a single degree of freedom gives the lowest values of all. On the other
hand the results obtained with the model with two degree of freedom are
the most unfavourable. The three-dimensional model gives an average of the
results obtained with the simplest models;
v. Finally, taking into account the values obtained with the three-dimensional
model, the bridge lateral displacements considered through the virtual path,
represents 11% of the total response. Thus, is concluded that the lateral
displacements of the bridge do not have a signicant inuence on the total
dynamic lateral behaviour of vehicles;
vi. There are no risk of excessive vibration in the dynamic behaviour of railway
vehicles when crossing the "Arroyo las Piedras" viaduct. The accelerations
values obtained are within the limits established by the FRA [1] and by SNFC,
for the criterion of passenger comfort;
vii. The use of three-dimensional vehicle models is recommended in order to con-

sider the dierent components of the trains in a more complete way and to
evaluate the response of the model including the phenomenon of yaw, that was
not considered in this study in order to simplify the analysis.
Chapter 5
Conclusions and suggestions for
subsequent studies
Some of the main problems in railway bridges due to the actions of high-speed
railway trac were discussed in this work. The causes of these problems and the
methods of analysis developed for the study of such phenomenon, were reviewed.
Along the work the "Arroyo las Piedras" viaduct, located in the new high-
speed railway line between Crdoba and Mlaga, in Spain, was object of case study
due to the singularity of the structure. This singularity is a result of the innova-
tive solution adopted, in order to provide the structure with the torsional stiness
required for adequately controlling the dynamic response when railway vehicles ec-
centrically run along a single track. Aditionally, the piers are remarkably high, since
several of them are higher than 93 m, which provides a signicant lateral exibility
to the structure.
From the study developed in this work the following conclusions and assump-
tions can be taken, being summarised as follows:
According to the dynamic analysis performed for the "Arroyo las Piedras"
viaduct, and described in chapter 3, it was concluded that the vertical dy-
namic eects on the structure are not signicant. The dynamic eects were
determined for the midspan of each span of the structure due to the actions
of the AVE train, being the maximum dynamic displacements veried at the
span 3. Taking this into account, the same analysis was performed in order to
evaluate the dynamic eects in the midspan of span 3 (node 292 of the nite
element model of the viaduct), considering the actions of the seven real trains.
126 Conclusions and suggestions for subsequent studies
In this case the maximum displacements were obtained due to the actions of
the ETR train and in terms of accelerations due to the ICE2 train. These
maximum dynamic eects were considered as envelope values of the dynamic
response of the structure due to the actions of the seven real trains. The ser-
viceability limit states were evaluated and it was concluded that the structure
is within the normative limits. The statical vertical displacement on the node
292 due to the actions of the LM 71 model were determined and was concluded
that it covers the vertical dynamic displacements obtained in this node for the
seven real high-speed trains.
In an accurate dynamic analysis of this viaduct, a more completed analysis
should be performed, considering the eects of the real trains and the HSLM
model in the relevant points of the structure. However, the results obtained
from this study are representative of the general dynamic behaviour of the
viaduct.
The eects considered and described in chapter 4, in order to study railway

vehicles lateral dynamic behaviour were: bridge lateral displacements, conside-
red by the denition of virtual path, track alignment irregularities and Klingel
movement eects. The assumptions and parameters assumed to dened these
eects were discussed in section 4.3. The values adopted are reasonable for
a rst approach of the problem but in further studies these eects must be
reviewed and more realistic values should be considered.
The mechanical and geometrical characteristics adopted in the vehicle models

do not correspond exacly to a high-speed train vehicle. The reason is related
with condential problems from the organization that were contacted and that
could provide these informations. Thus, several values were adopted according
to the information that was posible to obtain, which were described in section
4.4. The values adopted are reasonable but for a more realistic analysis the
exact characteristics of a high-speed vehicle should be considered;
In order to perform the evaluation of railway vehicles lateral dynamic be-

haviour, three vehicle models were developed: a model with one degree of
freedom, a model with two degrees of freedom and a three-dimensional model.
From the lateral dynamic analysis of these models, considering the excitation
eects indicated before, the following main conclusions were reached:
1. from the analysis of the ctitious vehicles with a single degree of free-
dom, the ETR-500 train car was the vehicle with the most unfavourable
dynamic accelerations. For this reason the mechanical characteristics of
this vehicle were considered in the following two models;
127
2. criteria of passenger comfort veried considering the dynamic results ob-

tained with the three-dimensional model, taking into account the limits
proposed by FRA [1] and SNFC (see gure 4.48);
3. increment of the dynamic response with the increment of the excitation;
4. the maximum relative dynamic displacements obtained from the model
with a single degree of freedom and the three-dimensional model are
almost equal, considering the combination 3 of eects (see gure 4.47);
5. the dierence between the maximum dynamic responses in terms of acce-
lerations increases with the increase of the excitation;
6. in general, the model with a single degree of freedom gives the lowest
results and the model with two degrees of freedom gives the most un-
favourable results;
7. the inuence of bridge lateral displacements represents 11% of the total
lateral dynamic response of vehicle models, being the consideration of
track irregularities responsable for 31% and the Klingel movement eects
responsable for 58%. With this results it is concluded that, although
the signicant lateral exibility of the structure, the lateral dynamic be-
haviour of the viaduct would not be responsible for relevant lateral vi-
brations on vehicles.
Although the simplifying assumptions considered in this study, the results ob-
tained are reasonable and signicant as a rst approach to the problem. Ho-
wever, further studies should be performed, in which more realistic parameters
should be considered. A special attention should be given to the parameters
used to generate track irregularities proles and to the eects produced by the
forces generated in the wheel-rail interface.
128 Conclusions and suggestions for subsequent studies
Bibliography
[1] FRA Federal Railroad Administration. Track safety standards; nal rule. 49
CFR Part 213, Part II, June 22, 1998.
[2] Jaime Domnguez Barbero. Dinamica de puentes de ferrocarril para alta veloci-
dad: metodos de clculo y estudio de la resonancia. Tesis Doctoral, ANCI:3.1
3.69, 2001.
[3] Klaus-Jurgen Bathe. Finite Element Procedures in Engineering Analysis.

Prentice-Hall Civil Engineering and Engineering Mechanics Series, 1982.
[4] S.L. CIDI-Calculo, Investigacion y Desarollo para la Ingeneria. Analsis de las

deformaciones laterales del viaducto del arroyo de las piedras - primer informe.
Documento n1 - Indice A, Noviembre 2006.
[5] H. Claus and W. Schiehlen. Modeling and simulation of railway bogie structural
vibrations. Dynamic of Vehicles on roads and tracks, August 22, 1997.
[6] Ray W. Clough and Joseph Penzien. Dynamics of strustures. Mc Graw Hill,
1993.
[7] ERRI D181. Forces laterales sur les ponts ferroviaires. ERRI D181/RP6, 1996.
[8] ERRI D214. Design of railway bridges for speed up to 350km/h; dynamic load-
ing eects including resonance. Technical report, European Railway Research
Institute (ERRI), 1998.
[9] ERRI D214. Etude numerique de l'inuence des irregularites de voie dans les
cas de resonance des ponts. ERRI D214/RP5, 1999.
[10] Ministerio de Fomento. Instruccin de acciones a considerar en el proyecto de

puentes de ferrocarril. Borrador, pendiente de publicacion, 2002.
[11] John W. Eaton. Octave v.2.0.16. OCTAVE, A high-level interactive language

for numerical computations., 2000.
130 BIBLIOGRAPHY
[12] EN1990-A2. Basis of structural design - annex a2: Application for bridges
(normative). European Committee for Standardization (Cen), Brussels, 2005.
[13] EN1991-2. Actions on structures - part2: General actions - trac loads on

bridges. European Committee for Standardization (Cen), Brussels, 2003.
[14] Coenraad Esveld. Modern railway track. Second Edition, 2001.
[15] J.Vinolas A.Alonso F.Gonzalez, J.Perez. Use of active steering in railway bogies
to reduce rail corrugation on curves. CEIT-Centro de estudios e Investigaciones
Tcnicas de Gipuzkoa, 2007.
[16] L. Fryba. Dynamic of railway bridges. Thomas Telford, Prague, 1996.
[17] Vijay K. Garg and Rao V. Dukkipati. Dynamics of Railway Vehicle Systems.
academic Press, 1984.
[18] Felipe Gabaldn Castillo J.M. Goicolea and Juan Jos Montejo. Estudio
dinmico del sistema de va en placa - calculo de las acciones dinamicas pro-
ducidas el trco ferroviario. Grupo de Mecnica computacional - Escuela de
Caminos de Madrid, mayo 2007.
[19] Jorge Ambrsio Joo Pombo and Miguel Silva. A new wheel-rail contact model
for railway dynamics. IDMEC - Instituto Superior Tcnico.
[20] Qingyuan Zeng Jun Xiang and Ping Lou. Transverse vibration of train-bridge
and train-track time varying system and the theory of random energy analysis
for train derailment. Vehicle System Dynamics, 41:129155, 2004.
[21] J.J. Kalker. Fast algorithm for the simplied theory of rolling contact. Vehicle
System Dynamics, pages 113, 1982.
[22] Holger Kruse Karl Popp and Ingo Kaiser. Vehicle-track dynamics in the mid-
frequency range. Vehicle System Dynamics, pages 423464, 1999.
[23] Klingel. Uber den lauf der eisenbahnwagen auf gerader bahn. Organ Fortschr.
des Eisenbahnwesens, 20:113123, 1883.
[24] A. Krylo. Uber die erzwungenen schwingungen von gleichformingen elastichen

staben. Mathematische Annalen, 1905.
[25] UIC leaets. Uic leaets. Internacional Union of Railways, Paris, 2001.
[26] Oldrich Polach. Creep forces in simulations of traction vehicles running on

adhesion limit. Bombardier Transportation Winterthur, Switzerland., 2003.
BIBLIOGRAPHY 131
[27] Arbeitsgemeinschaft Rheine-Freren. Rad/schiene-versuchs- und demonstra-

tionsfahrzeug, denitionsphase r/s-vd. Ergebnisbericht der Arbeitsgruppe
Lauftechnik, MAN, Mnchen, 1980.
[28] J.M. Goicolea Ruigomez. La consideraco de los fenomenos dinmicos en el

proyecto de puentes ferroviarios. Jornadas Tcnicas: Ingenieria para Alta Ve-
locidad. Veinte aos de Experiencia en Espaa, pages 161183, 2007.
[29] SIMULIA. Abaqus version 6.5. ABAQUS, nite element analysis, 2004.
[30] R.L. Taylor and J.c. Simo. Feap v.7.1f. FEAP, Finite Element Analysis Pro-
gram, 1999.
[31] Ache Asociacin Cientco tcnica del Hormigon Estructrural. Viaduct de "ar-
royo las piedras" primer viaducto mixto de las lneas de alta velocidade espao-
las. Hormign y acero, 243, Primer Trimestre de 2007.
[32] S.P. Timoshenko. Vibration problems in engineering. Van Nostrand, 1928.
[33] TSI. Technical specication for interoperability. Ocial Journal of the Euro-
pean Communities, 2002.
[34] J.K. Hedrick D.N. Wormley W. Murray, W. Baum and A.K.Kar. Perfor-
mance limits of rail passenger vehicles: Evaluation and optimization. Re-
port DOT/RSPA/DPB-50, Department of Transportation, Research and Spe-
cial Programs Administration, Oce of University Research, Washington, D.C.,
1979.
[35] Yeong-Bin Yang Yean-Seng Wu and Jong-Dar Yau. Three-dimensional analysis

of train-rail-bridge interaction problems. 2001.
[36] Manfred Zacher and Matthias Baebler. Dynamic behaviour of ballast on railway
bridges. Dynamic of High-Speed Railway Bridges - Porto, pages 125142, 2005.
Appendix A
Train Dynamic Signature - Program
(.m)
In order to performe de train dynamic signature of a certain train, for a certain
range of velocities and for a certain ratio damping, , a program in Octave/Matlab
code ([11]) was developed. This program was dene as follows:
impronta(train, m, vi, vf), where the variables are dened, with:

train: the train of which the train dynamic signature is required;
m: ratio damping of the structure;
vi: inicial velocity of the range of velocities considered;
vf: nal velocity of the range of velocities considered.
sub.m: this is a sub-function called by the previous one, in order to calculate

de train dynamic signeture and performe the graph of the results.
This two functions are dened as follows:
Function impronta.m
As an example, the train dynamic signature for the train AVE, with a ratio of damp-
ing, = 0.04, and a range of velocity between v = 120 km/h and v = 420 km/h,
when its been study a simply supported bridge with the rst eigenfrequency, f0 = 5
Hz.
134 Train Dynamic Signature - Program (.m)
#!/usr/bin/octave
sub('ave', 0.04, 120, 420);
Function sub.m
function sub(train,m,vi,vf)
traindef = [train,'.def'];
fichero = fopen(traindef,"rt");
[f,npun] = fscanf(fichero,'%*g %g',Inf);
fichero = fopen(traindef,"rt");
[x,npun] = fscanf(fichero,'%g %*g',Inf);
trainsig = [train,'-sig.txt'];
fich1 = fopen(trainsig,'wt');
f0=5;
wavelengthi=vi/f0;
wavelengthf=vf/f0;
for lambda = wavelengthi:0.01:wavelengthf;
delta=x/lambda;
u=(cumsum(F.*cos(2*pi*delta).*exp(-2*pi*m*delta))).**2;
s=(cumsum(F.*sin(2*pi*delta).*exp(-2*pi*m*delta))).**2;
w=max((u.+s).**(0.5));
fprintf(fich1,'%g %g \ n', lambda, w);
endfor
fclose(fich1);
fichero = fopen(trainsig,"rt");
[x,npun] = fscanf(fichero,'%g %*g',Inf);
fichero = fopen(trainsig,"rt");
[y,npun] = fscanf(fichero,'%*g %g',Inf);
Gmax=max(y)
plot(x,y);
title('impronta dinamica ICE2');
xlabel('longitude de onda [m]');
ylabel('G(alpha) [KN]');
traineps = [train,'.eps'];
print(traineps, '-deps');
replot
endfunction
Appendix B
Fictitious Railway Vehicles -
Program (.m)
For the analysis of the ctitious railway vehicles a program in Matlab/Octave was
perfomed. In this program the dynamic equation (4.28) for the system with one de-
gree of freedom is integrated with the methos of NewmarkBeta. The characteristic
of the ctitious vehicles are dened in table B.1. The mechanical characteristics here
dened were obtained taking into account the characteristics of several real railway
vehicles provided for this study, which are indicated on the same table inside boxes.
Some of this vehicles are indicated in table B.2. As an example of this program,
the code dened in order o calculate the maximum accelerations of the vehicles, is
dened as follows:
#!/usr/bin/octave
fichero = fopen('freqtrain.txt','rt');
[freq,npu] = fscanf(fichero,'%g %*g %*g %*g', Inf);
[k,npu] = fscanf(fichero,'%*g %g %*g %*g', Inf);
[c,npu] = fscanf(fichero,'%*g %*g %g %*g', Inf);
[m,npu] = fscanf(fichero,'%*g %*g %*g %g', Inf);
fichero = fopen('path-irreg-klingel.txt','rt');
[tiempo,npun] = fscanf(fichero,'%g %*g %*g %*g %*g',Inf);
fichero = fopen('path-irreg-klingel.txt','rt');
[disp,npun] = fscanf(fichero,'%*g %*g %*g %*g %g',Inf);
d=length(disp);
for i=1:npu
136 Fictitious Railway Vehicles - Program (.m)
Dt = tiempo(2)-tiempo(1);
Dt4=4/Dt;
Dt42=4/(Dt**2);
Dt2=2/Dt;
kh=m*Dt42+Dt2*c+k;
x(1)=0;
xd(1)=0;
xdd(1)=0;
fich = fopen('final.txt','wt');
fprintf(fich,'0 0 0\n');
for n=2:npun
if(n==d)
dispd(n)=(disp(n)-disp(n-1))/(tiempo(n)-tiempo(n-1));
else
dispd(n)=(disp(n+1)-disp(n-1))/((tiempo(n)-tiempo(n-1))*2);
endif
x(n)=(1/kh)*(k*disp(n)+c*dispd(n)+m*(Dt42*x(n-1)+
Dt4*xd(n-1)+xdd(n-1))+c*(xd(n-1)+Dt2*x(n-1)));
xdd(n)=Dt42*(x(n)-x(n-1))-Dt4*xd(n-1)-xdd(n-1);
xd(n)=xd(n-1)+(Dt/2)*(xdd(n-1)+xdd(n));
fprintf(fich,'%g %g %g\n', tiempo(n), x(n), xdd(n));
endfor
fclose(fich);
fich = fopen('ETRacce-total.txt','wt');
fichero = fopen('path-irreg.txt','rt');
[disp2,npun] = fscanf(fichero,'%*g %g %*g %*g',Inf);
fichero = fopen('final.txt','rt');
[resp,npun] = fscanf(fichero,'%*g %*g %g',Inf);
acc=max((resp.**2).**0.5);
for n=1:npun
fprintf(fich,'%g %g \ n',tiempo(n), acc(n))
endfor
fclose(fich);
endfor
137
Table B.1: Mechanical characteristics of cticious railway vehicles dened for this
study with frequencies of lateral vibration dened from 0.2 Hz up to 1.0 Hz. The
lateral stiness, k , is given in (kN/m), the ratio of damping, c, in (kN/(m/s)) and
the mass, M , in (ton). The values inside boxes correspond to some reference real
trains.
Freq.(Hz) k c M Freq.(Hz) k c M
0.2 142.86 85 90 0.6 913.83 37 65
0.21 142.86 80 85 0.61 960 37 65
0.22 166.67 80 85 0.62 960 37 63
0.23 166.67 75 80 0.63 960 37 61
0.24 187.5 75 80 0.64 959.18 37 60
0.25 210 75 82 0.65 959.18 37 58
0.26 215.89 75 81 0.66 958.33 37 56
0.27 214.85 70 77 0.67 958.33 37 54
0.28 222.48 65 70 0.68 967.53 37 53
0.29 229.17 60 67 0.69 967.32 36 51
0.3 235.86 60 65 0.7 967.11 36 50
0.31 240 60 65 0.71 966.89 36 49
0.32 257.14 60 65 0.72 966.67 36 47
0.33 272.73 60 65 0.73 980 17.5 47
0.34 288 60 65 0.74 1020 17.5 47
0.35 300 55 63 0.75 1050 17.5 47
0.36 290.24 55 58 0.76 1061.09 17.5 47
0.37 318.73 50 58 0.77 1110 17.5 47
0.38 319.75 50 55 0.78 1140 17.5 47
0.39 281.25 40 47 0.79 1140 17.5 46
0.4 275.96 40 43 0.8 1170 17.5 46
0.41 246.58 40 37 0.81 1200 17.5 46
0.42 247.76 35 35 0.82 1200 17.5 45
0.43 248.68 40 34 0.83 1210 17.5 44
0.44 271.46 40 36 0.84 1230 17.5 44
0.45 309.57 40 38 0.85 1260 17.5 44
0.46 338.16 40 40 0.86 1280 17 44
0.47 367.31 40 42 0.87 1290 17 43
0.48 396.45 40 44 0.88 1300 17 43
0.49 434.94 39 46 0.89 1320 17 42
0.5 482.76 39 48 0.9 1350 17 42
0.51 511.91 39 50 0.91 1380 17 42
0.52 559.71 39 52 0.92 1380 17 41
0.53 607.49 39 54 0.93 1400 17 41
0.54 645.96 38 56 0.94 1430 17 41
0.55 684.42 38 58 0.95 1450 17 41
0.56 732.18 38 60 0.96 1500 17 41
0.57 789.18 38 62 0.97 1520 16.5 41
0.58 864.63 37 64 0.98 1550 16.5 41
0.59 893.86 37 65 0.99 1560 16.5 40
0.6 913.83 37 65 1 1590 16.5 40
Fictitious Railway Vehicles - Program (.m)
Table B.2: Mechanical vehicle characteristics used as reference values in order to dene the ctitious vehicles
Car UIC Car ETR 500 Car ERRI B176 Wagon UIC Locomotive BR56
Primary 12800 8700 2468 1500 300
Lateral Stiness (kN/m) Secondary 320 256 320 770
Total 312 259 283 1500 1216
Primary - - - 34 -
Lateral Damping (kN/m((m/s)) Secondary 59 40 13.1 34 84
Total 59 40 13.1 17 84
Primary 2920 3220 2928 2600(+3kN ) 7890
Vertical Stiness (kN/m) Secondary 860 722 860 2720
Total 664 590 665 2600(+3kN ) 2023
Primary 5 15 3.36 - 50
Vertical Damping (kN/m((m/s)) Secondary 74 65 11.26 - 150
Total 5 12 3 - 38
Suspended Mass (ton) 32 34.23 32 41 81.2
Frequency of lateral vibration (Hz) 0.5 0.43 0.47 0.96 0.26
138
Appendix C
Model with two Degrees of Freedom
- Developed in FEAP
For the analysis of the model with two degrees of freedom a nite element program
was used. The system was modelised with two mass points and two elements with
the mechanical characteristics of the primary and secondary suspensions, (k1 , c1 )
and (k1 , c1 ), respectively.
x2
M2
k2 , c2
M1
k1 , c1
y(t)
Figure C.1: Model with two degrees of freedom modeled in FEAP [30]
Thus, the system represented in gure C.1 was dened in FEAP [30] as follows:
feap * Model with two degrees of freedom. Railway vehicle. Integrador

directo *
,2,2,2
param
k1=8700
k2=256
140 Two Degree of Freedom Model
m1=3.423*2
m2=34.23
u0=1.0
t1=0.0009
tt=13
c0=0
c1=0
a0=0
a1=40/k2
coor
10.0,0.0
20.0,1.0
30.0,2.0
elem
11,1,2
22,2,3
mass
20.0,m1
30.0,m2
boun
11,1
21,0
31,0
disp
10.0,u0
mate,1
truss
elask1
cros1.0
damp,rayl,c0,c1
mate,2
truss
elask2
cros1.0
141
damp,rayl,a0,a1
end
batch
tplo
prop1
dtt1
tran,newm
loop,time,5000000
timett
tang1
next,time
end
disp,3,2
acce,3,2
2,1
T ime, Displacement in time
stop
142 Two Degree of Freedom Model
Appendix D
Three-Dimensional Model -
Developed in ABAQUS
In order to perform the three-dimensional model in the nite element program
ABAQUS [29]
*HEADING
Simplified Model for a Railway Vehicle
*ASSEMBLY, NAME=Vagon
*NODE, NSET=Referencia-Body
1, 0., 0., 0.
*NODE, NSET=Vagon
2, -6.12, -0.75, 1.23
3, 6.12, -0.75, 1.23
4, 6.12, -0.75, -1.23
5, -6.12, -0.75, -1.23
*NODE, NSET=Bogie1
6, -6.12, -0.75, 1.23
7, -6.12, -1.37, 1.23
8, -6.12, -1.37, -1.23
9, -6.12, -0.75, -1.23
*NODE, NSET=Bogie2
10, 6.12, -0.75, 1.23
11, 6.12, -1.37, 1.23
12, 6.12, -1.37, -1.23
13, 6.12, -0.75, -1.23
*NODE, NSET=Referencia-Bogie1
14, -6.12, -1.17, 0.
*NODE, NSET=Referencia-Bogie2
144 Three-Dimensional Model
15, 6.12, -1.17, 0.

*NODE, NSET=Wheelset1
16, -6.12, -1.37, 1.23
17, -6.12, -1.37, -1.23
*NODE, NSET=Wheelset2
18, 6.12, -1.37, 1.23
19, 6.12, -1.37, -1.23
*NODE, NSET=Referencia-Wheelset1
20, -6.12, -1.37, 0.
*NODE, NSET=Referencia-Wheelset2
21, 6.12, -1.37, 0.
*RIGID BODY, REF NODE=Referencia-Body, TIE NSET=Vagon
*RIGID BODY, REF NODE=Referencia-Bogie1, TIE NSET=Bogie1
*RIGID BODY, REF NODE=Referencia-Bogie2, TIE NSET=Bogie2
*RIGID BODY, REF NODE=Referencia-Wheelset1, TIE NSET=Wheelset1
*RIGID BODY, REF NODE=Referencia-Wheelset2, TIE NSET=Wheelset2
*ELEMENT, TYPE=MASS, ELSET=Masa
1, 1
*ELEMENT, TYPE=MASS, ELSET=MasaB
2, 14
*ELEMENT, TYPE=MASS, ELSET=MasaB
3, 15
*ELEMENT, TYPE=ROTARYI, ELSET=Inercia
4, 1
*ELEMENT, TYPE=ROTARYI, ELSET=InerciaB
5, 14
*ELEMENT, TYPE=ROTARYI, ELSET=InerciaB
6, 15
*ELEMENT, TYPE=CONN3D2, ELSET=Secondary
7, 2, 6
8, 3, 10
9, 4, 13
10, 5, 9
*ELEMENT, TYPE=CONN3D2, ELSET=Primary
11, 7, 16
12, 11, 18
13, 12, 19
14, 8, 17
*CONNECTOR SECTION, ELSET=Secondary, BEHAVIOR=Muelle1
CARTESIAN,
*CONNECTOR SECTION, ELSET=Primary, BEHAVIOR=Muelle2
145
CARTESIAN,
*END ASSEMBLY
*MASS, ELSET=Masa
34.23,
*ROTARY INERTIA, ELSET=Inercia
80.346,
*MASS, ELSET=MasaB
3.423,
*ROTARY INERTIA, ELSET=InerciaB
0.983,
*CONNECTOR BEHAVIOR, NAME=Muelle1
*CONNECTOR ELASTICITY, COMPONENT=2
108.5,
64,
*CONNECTOR DAMPING, COMPONENT=2
16.25,
10,
*CONNECTOR BEHAVIOR, NAME=Muelle2
805,
2175,
3.75,
*BOUNDARY
Referencia-Wheelset1, 1, 2
*STEP
*FREQUENCY
10,
*BOUNDARY
Referencia-Wheelset1, 3,
Referencia-Wheelset2, 3,
*END STEP
*STEP, PERTURBATION
*STATIC
*BOUNDARY
146 Three-Dimensional Model
Referencia-Wheelset1, 3
Referencia-Wheelset2, 3
*DLOAD
Masa, GRAV, 9.8, 0., -1., 0.
*END STEP
*STEP, INC=25000
*DYNAMIC, DIRECT
9E-4, 14
*AMPLITUDE, NAME=Historia, INPUT=pik.txt
*BOUNDARY, AMPLITUDE=Historia
Referencia-Wheelset1, 3, , 1.
Referencia-Wheelset2, 3, , 1.
*END STEP
Appendix E
Piers characteristics
In this appendix the geometrical characteristics of the "Arroyo las Piedras" viaduct
piers are dened, as well as the inertia properties about the axis of lateral bending
(I11 ). These properties were used in the model developed in FEAP [30] in order to
evaluate the lateral dynamic behaviour of the viaduct in the midspan of the span
10 (node 1181), described in chapter 4.
26.72 m4
3 1
69.067 m4
I11
Figure E.1: Inertia properties of the "Arroyo las Piedras" viaduct piers, about the
axis of lateral bending.
Piers characteristics
Figure E.2: Geometric characteristics of the "Arroyo las Piedras" viaduct piers.
148

Tese

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Tese

Uploaded by

Copyright:

Available Formats

Instituto Superior Tcnico  Univ.

Dynamic Behaviour of High Speed Railway

Bridges. Vehicles Lateral Dynamic

Rui Filipe da Silva Dias

Dissertation for the degree of Master of Science in

Supervisor: Prof. Jos M. Goicolea

Dissertation for the degree of Master of Science in Civil Engineering

Madrid, November 2007

Rui Filipe da Silva Dias

Supervisor: Jos M. Goicolea

Co-Supervisor: Jorge M. Proena

Universidad Politcnica de Madrid

Instituto Superior Tcnico

Phone: +351 962438617

Keywords: Lateral vehicles dynamic behaviour, track irregularities, virtual path,

Palavras-chave: Comportamento dinmico lateral de veculos, Irregulari-

List of Figures xvii

List of Tables xix

1 Introduction and objectives 1

3 Vertical Railway Dynamic Analysis - Case Study 51

4 Lateral Dynamic Analysis 71

5 Conclusions and suggestions for subsequent studies 125

B Fictitious Railway Vehicles - Program (.m) 135

C Two Degree of Freedom Model 139

D Three-Dimensional Model 143

E Piers characteristics 147

1.1 High speed lines in Europe - UIC - High speed . . . . . . . . . . . . . 1

2.1 Amplitude of vibration of a ctitious wave. Geometric interpretation

3.1 "Arroyo las Piedras" viaduct. . . . . . . . . . . . . . . . . . . . . . . 51

4.1 Derailment possibly due to vehicle dynamic eects. . . . . . . . . . . . 72

3.1 Static and Dynamic eects due to real trains actions. . . . . . . . . . 65

4.1 FRA Track Safety Limits for track alignment irregularities[1]. . . . . 92

B.1 Mechanical characteristics of cticious railway vehicles dened for

Figure 1.1: High speed lines in Europe - UIC - High speed

In today's global market, connections using transport infrastructures that

Figure 1.2: High speed passenger volume growth - UICHigh Speed

High Speed Railway Bridges

lines, as the case of those intended for high-speed trains.

Figure 1.3: High speed train Eurostar.

ii excessive vibrations of the structure may lead to a magnication of the fatigue

iv the accelerations in vehicles should be limited so that the passengers comfort

Lateral Dynamic Phenomena of Railway Bridges

Figure 1.4: Example of an alignment track irregularities prole in a track length of

Lateral Displacement [m]

Figure 1.5: Wheelset sinusoidal movement in a straight track. Representation of the

Additionally, lateral displacements in railway bridges are caused by vertical

Objectives of this work

2. Application of the lateral dynamic response analysis in a real case of high-speed

1.2 State of the Art

catalogue of bridges of ERRI, is presented. This bridge has a span length of L = 15

ii Dynamic train signature;

iii Moving load dynamic analysis;

iv Vehicle-structure interaction dynamic analysis.

Other subject which must also be considered is the interoperability of railway

Vehicle Lateral Dynamic Behaviour

1.3 Scope of this study

Figure 1.8: "Arroyo las Piedras" viaduct - Case study [31].

1. Evaluation of the bridge lateral dynamic displacements, provided in the report

2. Generation of a track irregularities prole;

3. Study of the Klingel movement eects;

5. Development of a three-dimensional model in the nite element program ABAQUS

6. Analysis of the results.

Finelly, Chapter 5 contains some conclusions and recommendations for further

Instituto Superior Tcnico Univ.

2.1 Amplitude of vibration of a ctitious wave. Geometric interpretation

4.1 Derailment possibly due to vehicle dynamic eects. . . . . . . . . . . . 72

3.1 Static and Dynamic eects due to real trains actions. . . . . . . . . . 65

B.1 Mechanical characteristics of cticious railway vehicles dened for

Figure 1.2: High speed passenger volume growth - UICHigh Speed

ii excessive vibrations of the structure may lead to a magnication of the fatigue

Figure 1.4: Example of an alignment track irregularities prole in a track length of

2. Generation of a track irregularities prole;

3. Study of the Klingel movement eects;

5. Development of a three-dimensional model in the nite element program ABAQUS

i the denition of wavelength represented by ;

Figure 2.1: Amplitude of vibration of a ctitious wave. Geometric interpretation of

As an example, a dynamic analysis of a simple supported bridge dened in the

2.3.2 Static Eects

2.3.3 Dynamic Eects

These eects should be considered in the design of the structure through an

The upper limit (1) is dened as:

consider resonance eects, as previously noted. Whenever the criteria specied in

subjected to a minimum value of 0.25, in which Lf is the inuence length of the

2.5 Verication of the Limit States

i. The verication of maximum peak deck accelerations shall be regarded as

2.5.2 Serviceability limit states - trac safety

For the determination of load eects in the combined track/structure system a

6. Longitudinal springs reproducing the longitudinal stiness K of a xed sup-

ii. Deection criteria for checking passenger comfort;

Vertical deections should be determined with the LM 71 model multiplied by

models is discussed by Dominguez Barbero in [2]section 4.6, in which the principles