Professional Documents
Culture Documents
Tcnica de Lisboa
Escuela de Ingenieros de Caminos Univ. Politcnica de Madrid
Behaviour
November 2007
Dynamic Behaviour of High Speed Railway Bridges. Vehicles Lateral Dynamic Behaviour
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.
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
Notations xxi
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
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
Bibliography 128
Appendix 133
ix
A Train Dynamic Signature - Program (.m) 133
x
List of Figures
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
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
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
4.45 Results obtained for the dierent models developed in this study due
to the combination of the virtual path and track irregularities eects. . 119
4.46 Results obtained for the dierent models developed in this study due
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
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.
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.
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;
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;
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
4 Introduction and objectives
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].
8
Track Irregularities
Perfect track
4
Track alignment irregularities (mm)
-2
-4
-6
-8
0 10 20 30 40 50
Distance (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
-0.002
-0.004
-0.006
-0.008
0 0.5 1 1.5 2
Time [s]
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;
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;
6 Introduction and objectives
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]
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.
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.
8 Introduction and objectives
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;
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.
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.
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
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;
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.
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
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 .
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.
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.
dk = i, i = 1, 2, ... (2.2)
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.
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.
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:
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
2.3 Vertical Actions for Design 19
Figure 2.4: Load Model 71 and characteristic values for vertical loads. Extracted
from EN1991-2 [13].
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
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;
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].
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.
Figure 2.8: Flow chart to evaluate if a dynamic analysis is required. Extracted from
EN1991-2 [13].
2.3 Vertical Actions for Design 23
where:
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
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
2. f < 1 for 120 km/h < V 300 km/h and Lf > 2.88 m
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)
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.
or
1 + = 1 + 0 + 0.500 for carefully maintained track (2.15)
with:
K
0 = for K < 0, 76 (2.16)
1 K + K4
2.5 Verication of the Limit States 27
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).
i. Vertical accelerations;
v. Longitudinal displacements.
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].
2.5 Verication of the Limit States 29
Figure 2.10: Denition of angular rotations at the end of decks. Extracted from
EN1990-A2 [12].
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.
Compression: 72 KN/mm2
Tension: 92 kN/mm2
30 Dynamic eects
Figure 2.12: Variation of longitudinal shear force with longitudinal track displace-
ment for one track. Extracted from EN1990-A2 [12].
2.5 Verication of the Limit States 31
1. Track;
2. Superstructure (a single deck comprising two span and a single deck with one
span shown);
3. Embankment;
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.
i. Comfort criteria;
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
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
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).
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
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.
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 .
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
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
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
Figure 2.16: Two rsts modes of vibration of a simply supported beam and respective
analytical expression.
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.
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:
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:
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)
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
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:
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 =
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.
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.
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
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.
2.10 Simplied Methods According to the Train Dynamic Signature 47
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]
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 2.28: 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.
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);
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 ;
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 ).
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.
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
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.
58 Vertical Railway Dynamic Analysis - Case Study
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.
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.
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.
3.4 Dynamic Analysis for Vertical Loads 59
10
4
150 200 250 300 350 400
Train Velocity (km/h)
8
Maximum Displacement (mm)
7.5
6.5
5.5
150 200 250 300 350 400
Train Velocity (km/h)
8
Maximum Displacement (mm)
7.5
6.5
5.5
5
150 200 250 300 350 400
Train Velocity (km/h)
span 11 span 12 span 13 span 14 span 15
9
7
Maximum Displacement (mm)
1
150 200 250 300 350 400
Train Velocity (km/h)
span 16 span 17 span 18 span 19 span 20
Figure 3.10: Envelope values for the midspan displacement in each span for dierent
speed due to deection eects.
60 Vertical Railway Dynamic Analysis - Case Study
3.4
3.2
3
Maximum Displacement (mm)
2.8
2.6
2.4
2.2
1.8
1.6
1.4
150 200 250 300 350 400
Train Velocity (km/h)
1.85
1.8
Maximum Displacement (mm)
1.75
1.7
1.65
1.6
1.55
1.5
150 200 250 300 350 400
Train Velocity (km/h)
1.85
1.8
Maximum Displacement (mm)
1.75
1.7
1.65
1.6
1.55
1.5
150 200 250 300 350 400
Train Velocity (km/h)
span 11 span 12 span 13 span 14 span 15
2.6
2.4
2.2
Maximum Displacement (mm)
1.8
1.6
1.4
1.2
0.8
150 200 250 300 350 400
Train Velocity (km/h)
span 16 span 17 span 18 span 19 span 20
Figure 3.11: Envelope values for lateral displacements in the midspan of each span
for dierent speed due to torsional eects.
3.4 Dynamic Analysis for Vertical Loads 61
12
11
5
150 200 250 300 350 400
Train Velocity (km/h)
9.5
Maximum Displacement (mm)
8.5
7.5
6.5
150 200 250 300 350 400
Train Velocity (km/h)
9.5
Maximum Displacement (mm)
8.5
7.5
6.5
150 200 250 300 350 400
Train Velocity (km/h)
span 11 span 12 span 13 span 14 span 15
10
8
Maximum Displacement (mm)
2
150 200 250 300 350 400
Train Velocity (km/h)
span 16 span 17 span 18 span 19 span 20
Figure 3.12: Envelope values for the midspan displacement in each span for dierent
speed due to deection and torsional eects.
62 Vertical Railway Dynamic Analysis - Case Study
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.
1.44
2 = + 0.82 (3.5)
L 0.2
with: 1.00 2 1.67
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)
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:
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:
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.
Thus, the two enhancements values, dened in section 5, are dened as follows:
Node 292
15
14
(1 + dyn + 0.5 ) Sstat
13
Displacements (mm)
12
11
10
7
150 200 250 300 350 400
Train velocity (km/h)
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
Figure 3.14: Envelope value for the eects produced in node 292.
3.6 Verication of the Serviceability Limit States 67
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
Train velocity (km/h)
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:
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).
68 Vertical Railway Dynamic Analysis - Case Study
3.7 Conclusions
From the dynamic analysis performed in this chapter several conclusions can be
drawn and summarised as follows:
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;
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.
70 Vertical Railway Dynamic Analysis - Case Study
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
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
Support Midspan
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
74 Lateral Dynamic Analysis
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:
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)
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
4.2 Bridge Lateral Displacements 75
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
76 Lateral Dynamic Analysis
- 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
4.2 Bridge Lateral Displacements 77
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.
78 Lateral Dynamic Analysis
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.
-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
4.2 Bridge Lateral Displacements
Figure 4.8: Virtual path for the bogie 3 of R1 train. Performed by CIDI [4].
Lateral Dynamic Analysis
80 Figure 4.9: Virtual path for bogies 1, 3 and 5 of the R1 train. Performed by CIDI [4].
81
4.2 Bridge Lateral Displacements
Figure 4.10: Virtual path for the bogie 3 of ICE2 train. Performed by CIDI [4].
Lateral Dynamic Analysis
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
4.2 Bridge Lateral Displacements
Figure 4.12: Virtual path for bogies 1, 3 and 5 of the AVE train at v = 400km/h. Performed by CIDI [4].
Lateral Dynamic Analysis
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
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.
Z L
1
= lim (x)dx (4.2)
L L 0
4.3 Lateral Stability of Railway Vehicles 87
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].
88 Lateral Dynamic Analysis
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
() = 2S() (4.7)
2c
V,A () = A (4.9)
(2r + 2 )(2c + 2 )
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]:
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].
90 Lateral Dynamic Analysis
1
NX
r(x) = 2 An cos(n x + n ) (4.11)
n=0
f 0
= (4.12)
N
r
1
An = S(n )n for n = 0, 1, 2, ...N 1 (4.13)
2
10
Track Irregularities example
6
Track alignment irregularities (mm)
-2
-4
-6
-8
-10
0 200 400 600 800 1000 1200
Distance (m)
8
Track Irregularities
Perfect track
6
Track alignment irregularities (mm)
-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.
92 Lateral Dynamic Analysis
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.
x
y = y0 sin(2 ) (4.14)
LK
in which y0 and LK are, respectively, the amplitude and the wavelength of the
4.3 Lateral Stability of Railway Vehicles 93
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.
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.
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
4.3 Lateral Stability of Railway Vehicles 95
0.008
0.006
0.004
-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.
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.
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.
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.
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
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
4.4 Lateral Vehicle Car Characteristics 99
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
4.4 Lateral Vehicle Car Characteristics 101
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.
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
102 Lateral Dynamic Analysis
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.
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.
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].
104 Lateral Dynamic Analysis
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.
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.
Lateral displacements of the bridge represented by the virtual path dened for
the axle 3 of the AVE train at v = 400 km/h;
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:
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
4.5 Vehicle Model With One Degree of Freedom 107
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
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]
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).
108 Lateral Dynamic Analysis
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
Vehicle Response with natural frequency of vibration f=0.76Hz
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.
4.5 Vehicle Model With One Degree of Freedom 109
10
6
Displacement [mm]
-2
0 2 4 6 8 10 12 14
Frequency [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).
110 Lateral Dynamic Analysis
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
Frequency [Hz] Frequency [Hz]
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.
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.
4.5 Vehicle Model With One Degree of Freedom 111
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
Frequency [Hz] Frequency [Hz]
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.
112 Lateral Dynamic Analysis
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.
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)
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
114 Lateral Dynamic Analysis
5
4.5
4
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.
10
6
Displacements [mm]
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
Figure 4.40: Comparation between the response in accelerations of the two systems
considered due to the total aects considered in the study.
116 Lateral Dynamic Analysis
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.
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:
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.
1. Virtual path;
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.
120 Lateral Dynamic Analysis
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.
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.
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.
4.7 Three-Dimensional Vehicle Model 121
2.6
2.5
2.4
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
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.
122 Lateral Dynamic Analysis
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;
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.
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
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.
[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.
[12] EN1990-A2. Basis of structural design - annex a2: Application for bridges
(normative). European Committee for Standardization (Cen), Brussels, 2005.
[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.
[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.
[25] UIC leaets. Uic leaets. Internacional Union of Railways, Paris, 2001.
[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.
[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.
[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:
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);
fichero = fopen('freqtrain.txt','rt');
[k,npu] = fscanf(fichero,'%*g %g %*g %*g', Inf);
fichero = fopen('freqtrain.txt','rt');
[c,npu] = fscanf(fichero,'%*g %*g %g %*g', Inf);
fichero = fopen('freqtrain.txt','rt');
[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));
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:
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
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,
*CONNECTOR ELASTICITY, COMPONENT=3
64,
*CONNECTOR DAMPING, COMPONENT=2
16.25,
*CONNECTOR DAMPING, COMPONENT=3
10,
*CONNECTOR BEHAVIOR, NAME=Muelle2
*CONNECTOR ELASTICITY, COMPONENT=2
805,
*CONNECTOR ELASTICITY, COMPONENT=3
2175,
*CONNECTOR DAMPING, COMPONENT=2
3.75,
*BOUNDARY
Referencia-Wheelset1, 1, 2
Referencia-Wheelset1, 4, 6
Referencia-Wheelset2, 1, 2
Referencia-Wheelset2, 4, 6
*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