CHAPTER 8

Dynamic analysis of structures under high speed train loads: case

studies in Spain

F. Gabaldón Castillo, F. Riquelme, J.M. Goicolea-Ruigómez & J.J. Arribas

Escuela de Ingenieros de Caminos, Technical University of Madrid, Spain

ABSTRACT: A relevant aspect in the design of structures under high speed train loads is the
dynamical effects associated to the moving loads of the train. In this paper some calculation
methods that are usual in the design of high speed train structures are considered. These methods are
based on the integration of the motion equations, with or without considering the vehicle-structure
interaction. Some applications of the vehicle-structure interaction models are shown in order to
analyse the dynamical response of short span bridges, with emphasis in the possible reduction of the
dynamical effects associated to these models. To this end the results obtained using both interaction
models and moving point loads are compared. Finally some dynamic analyses of singular bridges
in the Spanish high speed lines are presented, showing results of interest for the design of these
viaducts. The methodology followed in these calculations corresponds to the Spanish IAPF 2006
code and the Eurocode 1 (EN1991-2).

1 INTRODUCTION

Nowadays in Spain the main part of the investment in transport infrastructure is dedicated to fund-
ing the construction of new railway lines for high speed trains, being this item the most important
too in other countries as France. These new railway lines are a very competitive alternative to other
transport modes for medium distances. At this moment in Spain there are three high speed railways
in operation: Madrid–Sevilla, Córdoba–Málaga and Madrid–Lérida, pertaining the last one to the
railway line Madrid–Barcelona–France. The ADIF authority has in project or construction several
railway lines as Madrid–Valencia–Murcia, Madrid–Segovia–Valladolid, Valladolid–Santiago–
Porto, Madrid–Badajoz–Lisboa, Sevilla–Huelva–Faro, etc. being the three later considered in the
frame of the international high speed railway system Portugal–Spain–Rest of Europe.
These activities have remarked out an important engineering aspect joined specifically to the
design of high speed railway structures: the dynamical effects associated to the train moving loads,
for which basic solutions have been described by Timoshenko and Young [14] and discussed fully
by Fryba [6, 7].
Most engineering design codes for railway bridges have followed the approach of the dynamic
factor proposed by UIC [15], which takes into account the dynamic effect of a single moving load
and yields a maximum dynamic increment of ϕ = 32% for an ideal track without irregularities.
The irregularities have been token into account with another parameter ϕ , leading to the dynamic
factor  = 1 + ϕ + ϕ . This approach cover the dynamic effects associated to a single moving load
but does not include the possibility of resonant effects due to the periodicity of the moving loads, as
this phenomenon does not appear for train speeds below 200 km/h. For velocities upper than 200 or
220 km/h and distances of axles between 13 m. and 20 m. resonant effects may appear. An illustra-
tive example showing experimental resonant measurements and the corresponding computational
results is shown in [2] for the Spanish AVE train crossing at 219 km/h, the Tagus bridge.

113
© 2009 Taylor & Francis Group, London, UK
114 Dynamics of High-Speed Railway Bridges

New European codes include the need for dynamic calculations covering resonant behaviour
[4, 5, 10]. Generally the calculation procedures are based in the direct integration of the structural
response for moving loads modelling the axles of the train. These methods can be applied following
several methodologies: the structural model may be analysed either by the complete discrete system
with N degrees of freedom and a time integrator like the Newmark method, or by a prior modal
analysis and a posteriori time integration of the n significant eigenmodes. Besides, these models
can take into account the vehicle–bridge interaction. If vehicle–bridge interaction is considered
the complexity of the model is increased and a major computational effort is required. Although
these kind of simulation are very interesting from a research point of view they are not useful for
standard design calculations except for unusual situations.

2 MODELS BASED ON DIRECT INTEGRATION WITH MOVING POINT LOADS

These class of methods are based on the direct time integration of the dynamic equations of the
structure, under the actions corresponding to a train of moving loads of fixed values which values are
representative of each axle of the train. The structural model may be analysed through the integration
of the complete discrete system with N degrees of freedom, or through a reduction of the number
of degrees of freedom via a previous modal analysis which reduces substantially the number of
equations. This modal analysis can be performed through an approximate numerical procedure
in order to obtain de eigenfrequencies and eigenmodes of vibration. This kind of procedures are
available in the majority of finite element codes, and for certain cases of simple structures the
spectral analysis can be achieved by analytical methods.
Finite element methods are applicable to arbitrary structures, even when non linear effects must
be considered. A spatial semidiscretisation of the structure is performed into subdomains called
“finite elements” leading a discrete N -d.o.f. system of equations [17]:

Md̈ + Cḋ + Kd = f (t) (1)

where M is the mass matrix, C is the damping matrix, K is the stiffness matrix, f (t) is the vector of
external loads, and d is the unknown vector of nodal displacements. In order to integrate in time this
system of equations, a direct integration of the model solves the complete system (1) for each time
step, and because the equations are generally coupled they must be solved simultaneously. This
procedure is valid even when nonlinear effects must be taken into account. In such case the elastic
internal forces and viscous damping should be replaced by a general nonlinear term f int (d, ḋ ).
If as usual the structural behaviour is linear, a modal analysis can be performed [1] resulting in
another system with a remarkable reduction of degrees of freedom. In a first stage, the eigenvalue
problem corresponding to the undamped system is performed solving the generalised eigenproblem
corresponding to the structural discrete system:

(−ω2 M + K)a = 0 (2)

obtaining the more significant n eigenfrequencies ωi i = 1, . . . , n, and the corresponding normal

vibration modes an , being generally n  N . In a second stage, the vibration modes ai oscillating
with their respective frequencies ωi are integrated in time. With this procedure the equations are
decouple, and the modal response of each mode is obtained from the dynamic equation of a system
with a single degree of freedom.
The simplest procedure to define the train loads is applying load histories in each node. For time
step ti and an axle load F, a nodal load FJ is assigned to the node J if the axle is above an element
that contains node J . The magnitude of FJ depends linearly on the distance from the axle to the
node. This procedure is outlined in Figure 1 for a single load. This scheme has been implemented in
the finite element program FEAP [13] both for the ten HSLM–A (High Speed Load Model) trains

© 2009 Taylor & Francis Group, London, UK

 Dynamic analysis of structures under high speed train loads 115

F F F

v v v

J−1 J J+1 J−1 J J+1 J−1 J J+1

ti − ∆t ti ti + ∆t

FJ

ti − ∆t ti ti + ∆t t

Figure 1. Nodal force time history definition for an axle load F moving at velocity v.

y k(t ) y 2(t ) y 1(t )

w (r, t )
maj

kj cj

mBj dk d 1= 0
r
mj = maj + mBj d2

Figure 2. Load train with vehicle–bridge interaction: simplified interaction model and definition of the
variables.

defined in the Eurocode [4, 10] and for the seven real trains defined in the Spanish code IAPF
[10]. All the results for moving point loads described in this paper have been obtained with the
methodology described in this section, using a direct time integration of the significant eigenmodes.

3 MODELS CONSIDERING VEHICLE-BRIDGE INTERACTION

When the structural model considers the relative vertical movement of the vehicles, the dynamic
behaviour of the structure is represented more realistically than using moving point loads. With
these models the train is considered as a set of masses representing the non-suspended masses of
the bogies, the suspended masses of the coaches, springs modelling the suspension of the vehicles,
dampers, etc [16]. In this work the simplified interaction models shown in Figure 2 are considered
[8]. For a train with k axles, each axle j is represented with an interaction element j with a non-
j j
suspended mass ms , a suspended mass ma , and stiffness and dampers constants with values k j and
j
c , respectively.
The equations obtained for the structure and the vehicle are (3) and (4), respectively:
        
MWW MWB ẅ CWW CWB ẇ KWW KWB w
w + +
MBW MBB b̈ CBW CwBB ḃ w
KBW KBB b
  (3)
fW
= f w,ar
B +f B
w,ext

© 2009 Taylor & Francis Group, London, UK

116 Dynamics of High-Speed Railway Bridges

 y    y    y  
MBB MBY b̈ CBB CBY ḃ KBB KBY b
+ +
MYB MYY ÿ CYB CYY ẏ KYB KYY y
  (4)
y,ar y,ext
= fB +fB
fY

In these equations the degrees of freedom have been got into three groups corresponding to the
nodes modelling the contact of the train with the deck (b), the rest of the nodes of the deck (w),
and those corresponding to the suspended masses of the vehicle ( y). In the load vector modelling
the wheel-deck forces, the contribution of the action–reaction term ( f .,ar ) and the external loads
have been distinguished ( f .,ext ).

3.1 Time integration based on the modal decomposition

The solution of the equation (3) can be expressed:

  
N  
w φWi (xk (t))
(xk (t), t) = qi (t) (5)
b φBi (xk (t))
i=1

being qi = qi (t) the modal amplitude, N the number of modes used in the analysis, φWi and φBi the
components of the eigenvector i corresponding to the degrees of freedom w and b respectively, and
xk (t) the parameter of position of the axle k.
Normalising the eigenvectors with the mass matrix, the modal equations are:

 T  
φ fW
q̈i + 2ξi ωi q̇i + ωi2 qi = Wi , i = 1, . . . , N (6)
φBi fBW,ar + f W,ext
B

y,ar y,ar
Solving for fB in (4) and substituting fBW,ar = −fB , results:

y,ext y y y
f w,far
B = fB − MBY ÿ − MBB b̈ − CBY ẏ − CBB ḃ − KBY y − KBB b (7)

The equation (6) is re-written via the substitution of the modal expression of b detailed in (5),
in equation (7):

q̈i + 2xi ωi q̇ + ω2 qi = φWi fW + φBi fBext − φBi MBY ÿ − φBi CBY ẏ − φBi KBY y


N
{q̈j φBj + 2vq̇j φB j + v2 qj φB j }
y
− φBi MBB
j=1


N (8)
{q̇j φBj + vqj φB j }
y
− φBi CBB
j=1

y

N
− φBi KBB {qj φBj }
j=1

x,ext y,ext
B = fB
being: f ext +fB y xk = vt (with the hypothesis of constant velocity of the train).

© 2009 Taylor & Francis Group, London, UK

 Dynamic analysis of structures under high speed train loads 117

mv

kv cv
w1 w2
mw
u1 b u2

Figure 3. Bernoulli beam element with simplified vehicle–bridge interaction model.

In order to integrate the movement of the suspended masses the modal expression of b is
substituted in (4), resulting the system of differential equations:


N
MYY ÿ + CYY ẏ + KYY y = fY − MYB {q̈φBj + 2vq̇j φB j + v2 qj φB j }
j=1


N
− CYB {q̇j φBj + vqj φB j } (9)
j=1


N
− KYB {qj φBj }
j=1

The equations (8) and (9) must be integrated in order to obtain the modal amplitudes qj (t) and the
degrees of freedom of the suspended masses y(t). After this, the degrees of freedom corresponding
to the movement of the deck (w and b) are computed from equation (5).

3.2 Direct time integration

In order to carry out the time integration of the whole structure, taking into account all its degrees
of freedom, a model based on the Euler–Bernoulli beam with an additional degree of freedom
corresponding to the axle of the train was developed (see Figure 3). Let w1 , θ1 , w2 , θ2 denote the
degrees of freedom of the beam (displacement and rotation of the nodes), and y and b the degrees of
freedom of the suspended mass (mv ) and the non-suspended one (mw ) respectively. The dynamical
equilibrium equations of the element are:

       
ÿ ẏ y Fy
Minterac + Cinterac + Kinterac = (10)
b̈ ḃ b Fb

being:

     
mv 0 cv −cv kv −kv
Minterac = , Cinterac = , Kinterac = (11)
0 mw −cv cv −kv kv

© 2009 Taylor & Francis Group, London, UK

118 Dynamics of High-Speed Railway Bridges

The displacement of the non-suspended mass b is interpolated from the nodal displacements of
the beam in a standard way via the hermitic shape functions [17]:
⎧ ⎫
⎪ w
⎨ 1⎪ ⎬
θ
b = [N1 N2 N3 N4 ] 1 (12)
⎪
⎩w 2 ⎪
⎭
θ2

Using the transformation matrix T:

 
0 0 1 0 0
T= (13)
N1 N2 0 N3 N4

The motion equations of the element result [9]:

⎧ ⎫ ⎧ ⎫ ⎧ ⎫ ⎧ ⎫
⎪ F1 ⎪ ⎪ẅ1 ⎪
⎪ ⎪ ⎪
⎪ ẇ1 ⎪
⎪ ⎪ w1 ⎪
⎪
⎪ ⎪
⎪ ⎪
⎪ ⎪
⎪ ⎪
⎪ ⎪
⎪ ⎪
⎪ ⎪
⎪
⎪
⎪ ⎪
⎪ ⎪
⎪ ⎪
⎪ ⎪
⎪ ⎪
⎪ ⎪
⎪ ⎪
M
⎨ 1⎬ θ̈
⎨ 1⎬ θ̇
⎨ 1⎬ ⎨ θ1 ⎪
⎬
Fy = M ÿ + C ẏ + K y (14)
⎪
⎪ ⎪
⎪ ⎪
⎪ ⎪
⎪ ⎪
⎪ ⎪
⎪ ⎪
⎪ ⎪
⎪
⎪
⎪ F2 ⎪⎪ ⎪ẅ2 ⎪
⎪ ⎪ ⎪
⎪ẇ2 ⎪⎪ ⎪
⎪w2 ⎪⎪
⎪
⎩ ⎭ ⎪ ⎪
⎪ ⎪
⎪ ⎪
⎪ ⎪
⎪ ⎪
⎩ ⎪
⎭
M2 ⎩ ⎭ ⎩ ⎭ θ2
θ̈2 θ̇2

being:
M = TT MinteracT, C = TT CinteracT, K = TT KinteracT (15)
The matrix (15) must be assembled with the standard matrix of the Bernoulli beam element
in order to obtain the stiffness matrix of the interaction element. The element matrix must be
re-computed in each time step.

4 APPLICATION: DYNAMIC ANALYSES OF SHORT SPAN BRIDGES

In these analyses four short span bridges defined in the ERRI 214 catalogue of isostatic bridges
have been considered. The definition of the bridges is summarised in Table 1. The loads considered
was the corresponding to the ICE 350 E high speed train, that is running in Spain with the name
AVE 103, in the range of velocities from 120 km/h to 420 km/h. The same damping ratio ξ = 2%
has been adopted for all the calculations.
The curves in Figures 4 and 5 correspond to the envelopes of accelerations and displacements
in the mid point of the 7.5 m span. It can be observed a left translation of the peaks computed with
the interaction model relative to the point load results. This effect appears because the interaction
models include the mass of the vehicle leading to the reduction of the frequencies of the viaduct
(if both masses have similar values). Besides, a reduction in the highest values of accelerations and
displacement is obtained when using interaction models.

Table 1. Definition of short span bridges from ERRI D214 catalogue.

Length (m) Mass (kg ×103 /m) Frequency (Hz) EI (kNm2 )

5.0 7 16 453919
7.5 9 12 1661921
10.0 10 8 2593823
20.0 20 4 50660592

© 2009 Taylor & Francis Group, London, UK

 Dynamic analysis of structures under high speed train loads 119

With the standard finite element formulation of the Euler–Bernoulli beam element [17] the
expression of the terms in the mass matrix is:



Mij = ρI Ni, x Nj, x dx + ρA Ni Nj dx (16)
L L

The first term in the right hand side of equation (16) corresponds to the rotational part of
the mass matrix (Mrot ), and the second term contributes to the translational part (Mtras ), being
M = Mrot + Mtras . The contribution of Mrot to M is very small for long span bridges. Nevertheless,
although this term is neglected by several computational codes, it is important in short span bridges.
This fact is shown in Table 2 where the values of the first frequency of an isostatic beam are reported
for several lengths.
Figure 6 shows the comparison between the envelopes of displacements for the spans of L = 5.0 m
and L = 20.0 m. It has been considered models with rotational mass and without it, using both point
loads as vehicle-bridge interaction elements.

Figure 4. Acceleration and displacement in the mid span of the 7.5 m beam.

Figure 5. Acceleration and displacement in the mid span of the 7.5 m beam.

© 2009 Taylor & Francis Group, London, UK

120 Dynamics of High-Speed Railway Bridges

Table 2. First frequency of an isostatic beam considering/not

considering the rotational mass matrix.

Frequency (Hz)

Length (m) With R.M. Without R.M.

5.0 13.5478 16.0000

7.5 11.0682 12.0000
10.0 7.6322 8.0000
20.0 3.9515 4.0000

Figure 6. Vertical displacement in the midspan: L = 5 and L = 20 m.

For the 5 m span bridge, comparing the results obtained with rotational mass (black and blue
curves) with those obtained without it (red and yellow curves) an important translation to the left
side is observed for the models with rotational mass. The reason is the remarkable increment of the
structural mass associated to the rotational terms. Comparing, for the same formulation of the mass
matrix, the results obtained using moving point loads with those obtained with interaction models
(the black curve versus the blue one, and the red curve versus the yellow one) another translation
of the resonant peaks is observed, although less relevant than in the former comparison.

© 2009 Taylor & Francis Group, London, UK

 Dynamic analysis of structures under high speed train loads 121

Finally, for the 20 m span can be concluded that the contribution of the rotational is negligible
from a practical point of view.

5 DYNAMIC ANALYSIS OF AN ARCH BRIDGE

5.1 Description of the structure and modelling

In this section the results obtained from the dynamic analysis of an arch bridge, designed for the
Ourense–Santiago high speed line, are shown. This viaduct is formed by a concrete arch with a
span of 170 m and a deck with length 630 m. The deck and the arch are connected with five piles.
The material used for the arch and deck is concrete H40 (characteristic resistance fck = 40 kN/mm2 )
and for the piles concrete H30 ( fck = 30 kN/mm2 ). The area and moment of inertia of the arch are
7.76 m2 and 14.87 m4 , respectively. For the deck these values are A = 11.25 m2 and I = 25.91 m4 .
The dynamic analysis was carried out with a 2D finite element model. The model have 1639
beam elements, leading to a system with 4872 degrees of freedom. All the computations have been
carried out with the finite element code iris [12]. The mesh of the model is showed in Figure 7.
The joins of the piles with the deck are encastred, having the pile in this point the same rotation
that the deck. In the left end support the horizontal displacement and bending rotation are allowed,
and the right end support is a fixed point.
For the time integration the first 114 eigenmodes have been considered in order to take into
account all the frequencies lower than 30 Hz, and a modal damping ξ = 2% has been assigned to
all the modes. Figure 8 shows the four lower eigenmodes computed.

Figure 7. Arch bridge. Finite element model.

y y
z x z x
Deformation(x500000): modes of number of mode, step 1. Deformation(x200000): modes of number of mode, step 2.

f1 = 0.954 Hz f2 = 1.396 Hz

y y
z x z x
Deformation(x50000): modes of number of mode, step 3. Deformation(x50000): modes of number of mode, step 4.

f3 = 1.477 Hz f4 = 1.577 Hz

Figure 8. Arch bridge. First to fourth eigenmodes.

© 2009 Taylor & Francis Group, London, UK

122 Dynamics of High-Speed Railway Bridges

Figure 9. Arch bridge. Displacements in the 3rd midspan.

Figure 10. Arch bridge. Accelerations in the 3rd midspan.

5.2 Results
The calculations were performed for a range of velocities from 120 to 420 km/h every v = 5 km/h,
considering the following high speed trains:

1. The ten trains corresponding to the HSLM-A (High Speed Load Model) IAPF 2006 [10] and
Eurocode [4].
2. The seven European high speed trains defined in the Spanish IAPF code [10]: AVE, EUROSTAR
373/1, ETR-Y, ICE-2, TALGO-AV, THALYS and VIRGIN, modelled with moving point loads.
3. The AVE 103 Spanish high speed train modelled both with moving point loads as with simplified
interaction elements, in order to compare the results obtained with these methodologies.

Figures 9 and 10 show the displacements and acceleration in the centre of the 3rd span obtained
with the AVE-103 train (with and without interaction). Figures 9 and 10 show these variables in
the keystone of the arch.
It can be observed from these figures that the dynamical effects are more relevant in the keystone
of the arch. Besides there isn’t reduction of the dynamical effects with the interaction models
because the spans of the structure have moderately long lengths.
Figures 13 and 14 show the bending moments in the left support and in the keystone of the arch
respectively, computed both for the seven real trains defined in the Spanish code IAPF [10] and
for the ten HSLM-A (High Speed Load Model) trains defined in the Eurocode [4, 10]. By a direct

© 2009 Taylor & Francis Group, London, UK

 Dynamic analysis of structures under high speed train loads 123

Figure 11. Arch bridge. Displacements in the keystone of the arch.

Figure 12. Arch bridge. Accelerations in the keystone of the arch.

comparison of the highest dynamical values with the static envelopes in each point the dynamic
amplification factor obtained is ≈2.5.

5.3 Conclusions
The two principals eigenmodes mainly contain the deformation of the arch. Their frequencies are
lower than those usual in standard bridges with shorter lengths. The first mode ( f1 = 0.954 Hz) leads
to non symmetric movements and the second mode ( f2 = 1.396 Hz) is symmetric. The following
modes correspond to the movement of the deck and the piles, and in consequence is convenient to
take into account the dynamical effects in these elements.
Resonant effects could appear for low velocities due to the low values of the main frequencies.
Nevertheless, because the wave length of the first modes (≈170 m) is one order of magnitude
higher than the length of the train coaches (between 13 and 27 m) there are simultaneously several
axle loads in the bridge and therefore their effects are mutually cancelled.
Finally, in this structure the consideration of interaction model is not relevant because the results
obtained are very similar with those computed with moving point loads.

© 2009 Taylor & Francis Group, London, UK

124 Dynamics of High-Speed Railway Bridges

Figure 13. Arch bridge. Bending moments in the left support of the arch.

6 DYNAMIC ANALYSIS OF A PERGOLA BRIDGE

6.1 Description of the structure and modelling

This viaduct will be in the Madrid–Valencia high speed line. It is a pergola bridge composed of
64 attached longitudinal beams with box section typology. The length of the precasted beams varies
from 35.35 m of the shortest one to 40.62 m. The separation between the axles of the beams is 4.5
or 6 m, leading to 326.75 m of total width.
In order to verify the requirements in the Spanish IAPF code [10] and Eurocode 1 [4], the
dynamic analysis comprises the following points:

1. Determination of the impact coefficient for velocities lower than 220 km/h, via simplified
methods.
2. Determination of the impact coefficient for velocities higher than 220 km/h, via dynamic
calculations. These computations consider the following aspects:
– Time integration of all the eigenmodes with frequencies lower than 30 Hz.
– The actions applied are the corresponding to the ten trains of the HSLM-A model (High Speed
Load Model A) defined in [4, 10], which are dynamic envelopes of the effects of the possible
real trains.
– The calculations are performed for a range of velocities from 120 to 300 km/h every
v = 5 km/h. The highest velocity is 20% higher than the design velocity vdesign = 250 km/h.

© 2009 Taylor & Francis Group, London, UK

 Dynamic analysis of structures under high speed train loads 125

Figure 14. Arch bridge. Bending moments in the keystone of the arch.

3. Verification of the ELS (Usage Limit State) requirements for the maximum values of acceleration,
vertical displacements and rotations
Taking into account the dimensions and typology of the viaduct, and because the tracks are
mainly in transversal direction (see Fig. 15), it has been considered convenient to perform a three
dimensional finite element model.
This bridge has been modelled with 7220 four nodes shell elements and 6342 nodes leading
to a model with 37730 degrees of freedom. Figures 16 and 17 show a general view of the mesh.
The boundary conditions adopted correspond to a point support of the ends of each beam as is
showed in the detail of Figure 18. For the evaluation of the vibrating mass of the viaduct it has
been considered the mass of concrete and the added masses corresponding to other non-structural
members (ballast, sleepers, rails, etc). The values adopted for the mechanical properties of concrete
are Young modulus E = 3.5 GPa and Poison ratio ν = 0.2.

6.2 Dynamic analysis of the structure

The dynamic analysis of the structure has been performed in accord to the requirements in the
appendix B of IAPF 2006 code [10], via the time integration of the modes of oscillation of the
structure. In order to take into account all the modes with associated eigenfrequencies lower than
30 Hz, it is necessary to consider 552 eigenmodes in the dynamic computations. The deformed
shapes corresponding to the fourth lower modes of oscillation are showed in figure 19. The damping

© 2009 Taylor & Francis Group, London, UK

126 Dynamics of High-Speed Railway Bridges

Figure 15. Pergola bridge. General plant view.

Figure 16. Finite element model of the pergola bridge. Lower view.

Figure 17. Finite element model of the pergola bridge. Upper view.

© 2009 Taylor & Francis Group, London, UK

 Dynamic analysis of structures under high speed train loads 127

Figure 18. Finite element model of the pergola bridge. Detail of the model.

is defined as Rayleigh damping [1] considering a fraction of critical damping rate ξ equal to 2% (for
concrete structures). The applied loads correspond to the 10 trains defined in the HSLM-A model
for a range of velocities from 120 km/h to 300 km/h every 5 km/h, leading to 370 analyses with 552
eigenmodes each one. To perform this computations in a reasonable lapse of time they have been
used 24 PENTIUM machines, with 2.6 GHz and 512 Mb RAM, running in parallel processes.
The results obtained from each one of the 370 analyses have been post-processed in order to
obtain the maximum values of vertical displacement, vertical acceleration and “in plane” rotations.
These maximum values have been considered for reporting the results obtained in the following
sections.

6.3 Simplified calculation of the impact coefficient

The simplified methodology for the evaluation of the impact coefficient 2 described in the IAPF
2006 code [10] is only valid for velocities lower than 220 km/h, and has the following requirements:
1. The typology of the viaduct must be conventional.
2. The first eigenfrequency must be valued between the two bounds defined in the expression (2.5)
of the IAPF 2006 code [10].
For the particular case of this viaduct the lower bound is fmin = 2.73 Hz and the upper one
fmax = 6.23 Hz. Therefore the first frequency of the viaduct ( f1 = 4.31 Hz) verifies this requirement.
For structures with good maintenance the expression of the simplified impact coefficient is:

1.44
2 = √ + 0.82 = 1.06 (17)
L − 0.2

being L defined in the IAPF 2006 [10] as the “characteristic length” of the viaduct (for this bridge
L = 38 m).

6.4 Calculation of the impact coefficient from dynamic analysis

The Spanish IAPF 2006 code requires a complete and specific dynamic analysis of structures in
high speed lines with design velocity upper than 220 km/h. From the results of this analysis using
the high speed trains, an impact coefficient  must be obtained. This impact coefficient  will be
compared with the impact coefficient 2 from expression (17), and the coefficient with the highest
value will be considered for the design of the structure.

© 2009 Taylor & Francis Group, London, UK

128 Dynamics of High-Speed Railway Bridges

f1 = 4.31 Hz f2 = 4.43 Hz

f3 = 4.51 Hz f4 = 4.52 Hz

f5 = 4.57 Hz f6 = 4.63 Hz

Figure 19. Pergola bridge. First to fourth eigenmodes.

From the dynamic analysis results, the real impact coefficient of the train i in some point of the
structures is:
δidin,rel
i = (1 + 0.5ϕ ) (18)
δest,tipo
being δidin,rel the maximum movement obtained for the train i, δest,tipo the movement obtained stati-
cally using the train defined in the IAPF 2006 (“tren tipo”) from the UIC train, and ϕ a coefficient
related to the tracking irregularities [3] (in this bridge ϕ = 0.0141).

© 2009 Taylor & Francis Group, London, UK

 Dynamic analysis of structures under high speed train loads 129

0.6

0.55

0.5

Impact coefficient
0.45

0.4

0.35

0.3
120 140 160 180 200 220 240 260 280 300
v (km/h)

Figure 20. Pergola bridge. Impact coefficient  in the midpoint of the track.

Figure 20 shows the impact coefficient  obtained applying the expression (18) for each velocity
and each train.
The impact coefficient is the highest value obtained considering all the trains, and in this structure
is the one obtained with the HSLM-A9 train running with a velocity v = 290 km/h:

 = maxi = 0.517 (19)

i

Comparing this value of  with the one obtained in (17) using the simplified methodology
results the impact coefficient to consider in the design of the viaduct:

 = 1.06 (20)

6.5 Verification of the usage limit state

The deformations, displacements and accelerations in the deck of railway bridges are related to the
Usage Limit States for the structure, and to the Ultimate Limit State of the track and vehicles [11].
For this reason in the IAPF 2006 and Eurocode [10, 4] are included the Usage Limit States that
must be considered in order to avoid security problems.

6.5.1 Accelerations
In accord to the section 4.2.1.1.1 of IAPF 2006 and section A.2.4.4.2.1 of Eurocode 1, the maximum
vertical acceleration of the deck must verify:

amax ≤ 0.35 g (21)

for bridges with ballast. In accord to [3], the values obtained from the dynamic computations (for
an ideal track) must be incremented by a factor (1 + 0.5ϕ ) in order to take into account the track
irregularities. The maximum values of the acceleration are showed in Table 3. From these results
can be concluded that the structure verifies the requirement (21).

6.5.2 Rotations of the deck

For bridges with double track and ballast the values of the rotation in the transition tie bar-deck are
bounded with the value [10, 4]:
θ ≤ 3.5 · 10−3 rad (22)

© 2009 Taylor & Francis Group, London, UK

130 Dynamics of High-Speed Railway Bridges

Table 3. Pergola bridge. Maximum values of the vertical acceleration.

Train a (m/s2 ) a (m/s2 )

HSLM A01 1.89 (v = 250 km/h) 2.01 (v = 260 km/h)

HSLM A02 2.41 (v = 265 km/h) 3.00 (v = 285 km/h)
HSLM A03 1.89 (v = 290 km/h) 2.19 (v = 300 km/h)
HSLM A04 2.97 (v = 295 km/h) 2.98 (v = 300 km/h)
HSLM A05 1.92 (v = 290 km/h) 1.79 (v = 295 km/h)
HSLM A06 1.95 (v = 285 km/h) 1.79 (v = 290 km/h)
HSLM A07 2.16 (v = 290 km/h) 1.87 (v = 295 km/h)
HSLM A08 1.52 (v = 265 km/h) 2.57 (v = 295 km/h)
HSLM A09 1.95 (v = 290 km/h) 1.98 (v = 295 km/h)
HSLM A10 2.06 (v = 285 km/h) 2.22 (v = 295 km/h)

Rotation in the transition deck-tie bar Incoming end

0.08

0.07
$\theta ( \ rm {rad}\codot 10{-3})$

0.06

0.05

0.04

0.03

0.02
120 140 160 180 200 220 240 260 280 300
$v$ (km/h)
Rotation in the transition deck-tie bar Outcoming end
0.08

0.07
$\theta ( \ rm {rad}\codot 10{-3})$

0.06

0.05

0.04

0.03

0.02
120 140 160 180 200 220 240 260 280 300
$v$ (km/h)

Figure 21. Pergola bridge. Envelopes of rotation in the deck-tie bar transition.

Figure 21 shows the envelopes of the rotations both in the incoming as the out-coming tie bars
of the bridge. It can be verified that the computed values are two orders of magnitude lower than
the required one.
6.5.3 Comfort
In section 4.2.1.2 of the IAPF 2006 code [10] the values of the vertical displacements are limited in
order to ensure the comfort of the passengers. Besides the comfort level is classified in accord to the

© 2009 Taylor & Francis Group, London, UK

 Dynamic analysis of structures under high speed train loads 131

maximum values obtained for the vertical acceleration. In this bridge the maximum acceleration
has a value upper than 2.0 m/s2 , and therefore the comfort level is considered as “allowable”.
The requirements for comfort are established in terms of the ratio L/δ, being L the length and
δ the maximum vertical displacement. The value of this ratio must be corrected with a factor that
depends on the maximum acceleration value and the number of spans. For the Pergola bridge
described in this paper the comfort requirements are satisfied.

ACKNOWLEDGEMENTS

The authors acknowledge the financial support of Ministerio de Fomento of Spain to the project
“Análisis dinámico de estructuras sometidas a acciones de trenes de alta velocidad” through the
research program “Acciones Estratégicas del Área Sectorial de Construcción Civil y Conservación
del Patrimonio Histórico Cultural” of the “Plan Nacional de Investigación Científica, Desarrollo e
Innovación Tecnológica 2002–2003”.

REFERENCES

[1] Clough, R.W. & Penzien, J. 1994. Dynamics of Structures. Second Edition, McGraw-Hill.
[2] Domínguez Barbero, J. 2001. Dinámica de puentes de ferrocarril para alta velocidad: métodos de
cálculo y estudio de la resonancia. Ph. D. thesis (in Spanish). E.T.S. Ingenieros de Caminos, Universidad
Politécnica de Madrid.
[3] ERRI D214 committee. 1999. Ponts-rails pour vitesses >200 km/h; Etude Numérique de l’influence
des irrégularités de voie dans les cas de résonance des ponts. Rapport technique RP 5. European Rail
Research Institute (ERRI). March.
[4] Eurocode 1. 2003. Actions on structures – Part 2: Traffic loads on bridges. CEN.
[5] Ferrovie dello Stato, Italy. 1997. Sovraccarichi per il calcolo dei ponti ferroviari. Istruzioni per la
progettazione, l’esecuzione e il collaudo. Code I/SC/PS-OM/2298.
[6] Fryba, L. 1972. Vibration of solids and structures under moving loads. Academia, Noordhoff.
[7] Fryba, L. 1996. Dynamics of railway bridges. Thomas Telford.
[8] Gabaldón, F., Riquelme, F.Y. & Goicolea, J.M. 2005. Análisis dinámico de estructuras sometidas
a acciones de trenes de alta velocidad, considerando la interacción vehículo-estructura. Métodos
Numéricos en Ingeniería. SEMNI/APMTAC.
[9] Ju, S. & Lin, H. 2003. Numerical Investigation of a steel arch bridge and interaction with high-speed
trains. Engineering Structures. Vol. 25. pp. 241–250.
[10] Ministerio de Fomento, Spain. 2006. Instrucción de acciones a considerar en el proyecto de puentes de
ferrocarril. Draft J.
[11] Nasarre & de Goicochea, J. 2002. Estados límites de servicio en relación con la vía en puentes de ferro-
carril. Puentes de Ferrocarril. Proyecto, Construcción y Mantenimiento. Congreso del Grupo Español
de IABSE. Vol. 2. Madrid, Junio.
[12] Romero, I. iris: User Manual. Universidad Politécnica de Madrid. http://w3.mecanica.upm.es/∼ignacio/
iris.html.
[13] Taylor, R.L. FEAP. A Finite Element Analysis Program. University of California, Berkeley.
http://www.ce.berkeley.edu/∼rlt/feap.
[14] Timoshenko, S. & Young, D. 1995. Vibration problems in engineering. Van Nostrand, 3rd ed..
[15] Union Internationale des Chemins de Fer. 1979. Charges a prendre en considerations dans le calcul des
ponts-rail. Code UIC 776-1R.
[16] Yang, Y.B., Yau, J.D. & Wu, Y.S. 2004. Vehicle-Bridge Interaction Dynamics With Applications to
High-Speed Railways, World Scientific.
[17] Zienkiewicz, O.C. & Taylor, R.L. 2000. The finite element method, Butterworth-Heinemann.

© 2009 Taylor & Francis Group, London, UK