CHAPTER 11

Structural analysis of high speed rail bridge substructures.

Application to three Spanish case studies

Juan A. Sobrino
Pedelta & University of Catalonia, Barcelona, Spain

Juan Murcia
University of Catalonia, Barcelona, Spain

ABSTRACT: This paper describes the track-bridge interaction phenomenon in bridges for high
speed railways, proposing a methodology for the structural analysis of the substructure taking
into account the following aspects: horizontal loads and imposed displacements due to the long-
term behaviour of concrete and temperature using a step by step analysis and, if required, the
construction process, behaviour of ballast using a non-linear links between the track and the super-
structure and behaviour of POT bearings, interaction between foundation and ground conditions.
The methodology fulfils the specifications of Eurocode 1 and consist of modelling the whole struc-
ture (track-bridge) with a 2D or 3D model using linear bar elements and non-linear springs and the
analysis of the different short-time and long-time scenarios. To illustrate the methodology, three
case studies of concrete viaducts of the HSR line Madrid-Barcelona-French Border are presented.
The results are compared with those obtained from the simplified design methods.

1 INTRODUCTION

After the conclusion of the first Spanish High Speed Railway Line in 1992, connecting Madrid
and Seville, the strategic railway infrastructure plan developed by the Spanish Ministry of Public
Works as a part of the objective of the European Union (EU) of developing a Trans-European
High-Speed Rail System, includes the construction of more than 4000 km of HSR in a period of
fifteen years [1]. The directives for the rail system interoperability constitute an impelling element
for the railway sector, as new lines, trains and equipment within the EU countries should be either
built or renovated.
As a result of the complexity of the Spanish geography, approximately a 10% of the railway
system consists of bridges and tunnels. Construction of railway bridges for high speed lines repre-
sents a significant cost of the network. Due to the important magnitude of vertical and horizontal
loads, the design of these bridges requires a judicious selection of the structural configurations and
erection procedure.

2 DESIGN CRITERIA OF HSR BRIDGES IN SPAIN

2.1 Design Codes

The HSR lines in Spain are developed by the Ministry of Public Works. The technical specifications
required by the owner for the design of these bridges are as follows:
• Loads should be according to the Spanish Code for Railway Bridges [2] [3] but a check is also
required to fulfil the specifications of Eurocode 1.2 (Traffic Loads on Bridges) and the Annex
2 of Eurocode 1, specifying additional Serviceability Limit States [4] [5].

129

© 2009 Taylor & Francis Group, London, UK

130 Track-Bridge Interaction on High-Speed Railways

• Design of structural elements should be carried out according to the Spanish Codes for concrete
structures or the recommendations for the design of composite and steel road bridges.

2.2 Specific relevant aspects for the design of HSR Bridges

Internal forces due to railway traffic loads are 2 to 2.5 times larger than those induced by road traffic
loads. Dead loads of two ballasted tracks, including all bridge finishes, weights 120 kN/m and the
effect of ballast should be incremented for the design about 30% to take into account possible
increments during the life of the bridge.
Horizontal loads originated by railway traffic (braking and traction, nosing force, wind, track-
structure interaction and centrifugal forces) are also much bigger than similar effects in road bridges.
As an example, maximum braking and traction force in a HSR standard 300 m viaduct is 7000 kN.
The equivalent force in a similar road bridge is 850 kN. Loads induced by centrifugal forces in
railway bridges could also range from 300% to 1500% of the ones caused by the same force in road
bridges.
Apart from the heavy loads considered for the design of HSR bridges, there are some specific
Serviceability Limit States (SLS) to be verified in this type of bridges that could be summarized
as follows:
• Verification of vibrations for traffic safety, limiting the maximum vertical peak deck acceleration
induced by real trains (for instance, the recommended value for a ballasted track is 3.5 m/s2 ).
• Verification of deck twist and vertical deformations of the deck for traffic safety.
• Verification of the maximum vertical deflection for passenger comfort, depending on the train
speed and span length.
• Verification of track, limiting rail stresses due to combined response of the structure and track
to variable actions, limiting the longitudinal displacements induced by traction and braking (for
instance to only 5 mm for welded rails without rail expansion devices or with only one expansion
device at one end of the deck), etc.
Due to these significant loads and the very strict Serviceability Limit States to be fulfilled,
structural elements are much stiffer than in road bridges and for this reason, the optimization of
materials and, in particular, the selection of a judicious structural system and the deck’s slenderness
is basic to obtain economical solutions.

3 TRACK-DECK INTERACTION

The design basis of the track-bridge interaction phenomenon is established in the Eurocode 1
[4] [5] and in the UIC 774-3 leaflet [6]. The interaction is taken into account in the numerical
model by means of non linear springs that reproduce the horizontal interaction of the system
rail + fastenings + sleepers and the deck (Fig. 1). Non linearity is modelled by a bilinear horizontal
force per unit rail length – relative displacement law, as shown in Figure 2. In the structural analysis,
the existence of rail expansion devices, the type and the loading and maintenance state of track
(affecting the value of the k parameter of bilinear law) and the substructure stiffness have also to
be considered. It is necessary to include the substructure in the track-bridge interaction model as
long as it can affect significantly the global behaviour of the bridge [7].
The situations in which the interaction appears are those where a relative horizontal displacement
between the track and the deck occurs. The UIC 774-3 leaflet specifies the main actions to be taken
into account: the changes in temperature in the rail and in the deck, the horizontal forces due to
braking and acceleration and the bending of the deck caused by vertical traffic loads (bending
generates horizontal movements at the end of the deck due to its rotation). The values of the above
mentioned actions are defined in the EC-1 [4] [5] and in the UIC 774-3. The linear combination of the
results obtained from the independent calculation of each action (temperature, braking/acceleration,
bending) is not valid since the problem is not linear. A rigorous study demands to carry out a step-
by-step non linear analysis (called complete calculation in the UIC 774-3), including all the actions

© 2009 Taylor & Francis Group, London, UK

 Structural analysis of high speed rail bridge substructures 131

Figure 1. Elements to be considered in the analysis of track-bridge interaction.

Longitudinal stiffness of the ballasted bed

F (KN/ml)
k=0 Loaded track
60

k=60 KN/m2
40

k=0 Unloaded track

20 2
k=20 KN/m

0
0 0,002 u (m)

Figure 2. Longitudinal resistance of the track respect to longitudinal displacement.

and the loading conditions of the track (affecting the non-linear springs representing the effect of
the track-deck interaction). However, the UIC 774-3 admits the linear combination of the results to
check the additional stresses in the rail. For some specific cases, a simplified methodology using
diagrams and tables is provided to obtain the stresses in the rail and the reactions at the supports.
On the other hand, the deformation of the concrete deck due to creep and shrinkage can be as
important as for the temperature changes and should also be considered [8]. In [9] a method to
combine this action with the rest of actions is proposed in order to obtain the stress envelope for
the rail. The temperature changes and the creep and shrinkage are considered together in the same
load case, as an equivalent temperature change of the deck. The worst situation for the rail would
happen in one of these cases: a) maximum increase of temperature in the early ages (without creep
and shrinkage deformation), and b) maximum concomitant decrease of temperature and maximum
creep and shrinkage deformation. The braking/acceleration and bending results would be added
afterwards.

4 METHODOLOGY FOR THE STRUCTURAL ANALYSIS OF HSR BRIDGES

4.1 Analysis model

The analysis model used to obtain the longitudinal forces distribution among the viaduct includes
the track, the deck, the substructure and their interaction non-linear effects (fastenings/ballasted

© 2009 Taylor & Francis Group, London, UK

132 Track-Bridge Interaction on High-Speed Railways

Figure 3. Force-displacement cycle for a sliding POT bearing.

bed and bearings). The rails and the deck are modelled with beam elements in their centre of
gravity. The supports are modelled in the same way, taking also into account the loss of stiffness
due to the long-term shrinkage. Geometric and material non linearities have not been considered;
which would turn out to be suitable for very high supports. The foundations are represented as
equivalent stiffness for horizontal movement and rotation in the support base. The track system
(rail + fastenings + sleppers + ballast) and deck interaction is represented by non-linear springs
with a bilinear law according to Figure 2.
POT bearings, fixed, guided or sliding, are the most commonly used type of bearings in the Span-
ish high-speed railway bridges. They have two different friction coefficients (static and dynamic)
should be considered. This friction coefficient µ varies from 0.5% to 5%. The behaviour of the
POT for horizontal movements can be represented by a spring with a Coulomb’s friction law as
shown in Figure 3.
The model of the whole set in 2 dimensions is shown in Figure 4. In case there is no rail expansion
device in the end of the viaduct, it is necessary to include the adjacent platform (300 m behind the
abutment according to the UIC 774-3 leaflet [6]).

4.2 Analysis type

The type of analysis that is going to be carried out is the so called complete calculation [7] [8]. It
consists in a step-by-step analysis for each different action and static arrangement.

© 2009 Taylor & Francis Group, London, UK

 Structural analysis of high speed rail bridge substructures 133

Figure 4. Continuous-deck bridge model.

4.3 Step-by-step load combinations

The steps considered in the analysis are listed below:
1. Construction of the structure (under self-weight -D1- and prestressing if existing).
2. Phase between the conclusion of the bridge and the welding of the rails, taking into account the
short-term creep and shrinkage deformation of the deck (evaluated after 3 months).
3. Introduction of the imposed dead load (D2) of the superstructure (in steps 1 and 2 the loads act
in the deck and the substructure. Only after welding the rails the track-bridge interaction starts to
act. Rails and the non-linear springs representing the track-deck interaction have to be included.
4. Final creep and shrinkage deformation (considering two cases: just after welding the rails and,
eventually, the long-term situation).
5. Change of temperature (positive or negative) of the deck. Change of temperature of the rails is
only considered when an expansion device exists at the end of the viaduct.
6. Vertical and horizontal (braking/acceleration) loads due to traffic according to EC-1.
7. Beginning/final of the process.
8. Steps 1, 2 and 3 are always considered in the same way. However, in each one of the steps 4, 5
and 6 a pair of hypotheses exist with respect to the operating action. It means having 8 different
load combinations for the substructure, if the rails are continuous. If an expansion device is
placed near the bridge it would also be necessary to consider the change of temperature of the
track (step 5), which represents16 different load combinations.
The results will enable to obtain the force envelopes of the supports. In the next section, this
methodology is used to analyse three Spanish high-speed railway viaducts. The results are compared
with the obtained by means of simplified calculation procedures.

5 CASE STUDIES

The proposed methodology has been applied in three viaducts of the HSR Madrid-Barcelona, with
commonly used static arrangements.
As a first estimation, longitudinal forces can be evaluated to be equal to the friction force
generated when the bearing slides under vertical permanent loads: µND1 +D2 . On the other hand, at
the fixed point (in general placed in one of the abutments) the longitudinal force is evaluated as the
sum of the external horizontal forces due to braking, acceleration,
 etc. (Fext ) and the compensation
of the forces of friction generated in the rest of the supports ( µND1 +D2 ). Therefore the bearing are

© 2009 Taylor & Francis Group, London, UK

134 Track-Bridge Interaction on High-Speed Railways

usually expected to slide independently from the support stiffness. It is also assumed that moving
supports are unable to generate additional friction under vertical traffic load. These assumptions
omit the real effect of track-bridge interaction and substructure stiffness. In the following lines,
the results obtained with the methodology proposed in this paper are compared to these reference
values.

5.1 Selles Viaduct

Selles Viaduct is a continuous post-tensioned concrete bridge with an overall length of 167 m
(Fig. 5). The decks has 5 spans of 28 + 3 × 37 + 28 m. The typical cross-section (Fig. 6) is a box
girder 3.3 m depth (L/11.2).
It has a fixed support in one of the abutments and longitudinally sliding (guided or fiexed in the
transverse direction) POT bearings at the rest of the supports. The pier height varies from 10 to
12 m (Fig. 7).
A numerical model, following the criteria described in paragraph 4, has been carried out. Accord-
ing to the obtained results (Table 1), the piers take a significant part of the external longitudinal
forces (braking/acceleration). It is due to the fact that the bearing slides under traffic loading,
reaching the value µND1 +D2 +q . However, achieving this maximum depends on the stiffness of the

Figure 5. View of the bridge.

Figure 6. Typical cross-section of Selles Viaduct.

Figure 7. Elevation of Selles viaduct.

© 2009 Taylor & Francis Group, London, UK

 Structural analysis of high speed rail bridge substructures 135

substructure. For high pier stiffness and/or low sliding friction coefficient, sliding is needed to make
movements compatible (case µ = 0.01). For very slender piers or high sliding friction coefficient
(case µ = 0.05) , the compatibility can be achieved by column deformation without bearing sliding.
Therefore the value is below µND1 +D2 +q as happens for the second scenario.
The force at the fixed point is lower than expected (Table 2) because piers take braking and
acceleration forces and also a part of these external forces are transferred to the embankments
through the rails by means of the track-bridge interaction. On the other hand, the continuity of the
rails can introduce over-stresses in the bridge system as bridge deformations are constrained.

5.2 Avernó Viaduct

This viaduct consists on 14 spans having a total length of 810 m (maximum span length 60 m)
(Fig. 8). The distribution of spans is 45 + 12 × 60 + 45 m. Having small curvature it has been
nevertheless analyzed as a straight bridge for our purpose. The deck is a post-tensioned concrete
box girder (Fig. 9) with constant depth of 4 m (H = L/15). The fixed support is in one of the
abutments. Some piers reach a maximum height of 40 m (Fig. 10).
Similar conclusions can be drawn from the analysis of the Avernó viaduct (Tables 1 and 2). Piers
can take a significant part of the horizontal traffic load. Even though the piers are higher than in

Table 1. Longitudinal force at the top of piers 1 and 4 in Selles

River and Averno viaducts respectively.

Viaduct µ %µND1 +D2 % µND1 +D2 +q

Selles 0,05 125% 85%

River 0,01 148% 100%
Avernó 0,05 125% 84%
0,01 148% 100%

Table 2. Longitudinal force in top of fixed abutment in Selles River and Averno viaducts.

% FH UIC
Viaduct µ %Fext + µND1 +D2 %Fext + µND1 +D2 +q simplified

Río 0,05 67% 55% 59%

Selles 0,01 108% 102% 98%
Avernó 0,05 59% 53% –
0,01 53% 50% –

Figure 8. View of Avernó bridge. Figure 9. Typical cross-section of Selles viaduct.

© 2009 Taylor & Francis Group, London, UK

136 Track-Bridge Interaction on High-Speed Railways

Figure 10. Elevation of Avernó viaduct.

Figure 12. Typical cross-section of Anguera

Figure 11. View of Anguera bridge. viaduct.

Selles bridge, they are not slender enough to prevent sliding of the bearings for low sliding friction
coefficient. The force in the abutment is lower than the maximum friction admissible load due to
the capacity of the rest of the supports to take some horizontal load.

5.3 Anguera Viaduct

The long viaduct over Anguera River consists of 28 simply supported spans of 34 m (Figs 11 and
12). The deck is formed by pre-cast prestressed concrete U girders with 2.45 m depth and a 30 cm
in situ slab. Piers do not exceed 10 m of height, being very stiff. Every span has a fixed support
and the movable one in each side, therefore each pier has a fixed bearing and a sliding one.
Longitudinal forces for this structural arrangement are usually evaluated similarly as for con-
tinuous viaducts, but considering each span independent from the other. This criterion obviously
doesn’t consider accurately the track-bridge interaction. On the other hand, for similar span lengths,
the friction forces generated in the sliding bearings of contiguous spans are equivalent and are
compensated at the pier.
According to the analysis the simplified criteria (par. 5) is clearly conservative since it does not
consider the redistribution of forces through the continuity of the rails (Tables 3 and 4). Moreover,
this continuity of the track allows to transferring part of these forces to the adjacent platform.

6 CONCLUSIONS

Track-bridge interaction has a main role in the global behaviour of the HSR bridges in the distribu-
tion of the longitudinal forces and therefore has to be taken into account in the structural analysis
of the substructure. A methodology to obtain longitudinal forces in supports has been presented in
this paper. It has been applied to three Spanish HSR viaducts and compared with the commonly

© 2009 Taylor & Francis Group, London, UK

 Structural analysis of high speed rail bridge substructures 137

Table 3. Longitudinal force in top of column no 7

in Anguera viaduct.

µ %FH simplified

0,05 44%
0,01 50%

Table 4. Longitudinal force in top of fixed

abutment in Anguera viaduct.

µ % FH , simplified UIC

0,05 71%
0,01 58%

simplified calculation procedures. Differences are shown between both methods. From the com-
parison it is shown the relevance of considering track-bridge interaction, sliding support behaviour
and support stiffness in the structural analysis. Further developments should include geometric
nonlinearity of high supports and variable ballast stiffness and sliding friction coefficient.

REFERENCES

[1] “Strategic infrastructures and transport plan”, Ministry of public works, 2005.
[2] “Code IAP – Actions on railway bridges”, Ministry of public works, 1972 (in Spanish).
[3] “Code RPX – Composite bridge Code”, Ministry of public works, 1995 (in Spanish).
[4] EN 1991-2 “Eurocode 1: Actions on structures. Part 2 Traffic loads on bridges”.
[5] EN 1990 PrAnnex A2 “Eurocode: Basis of design. Annex 2: Application for bridges”, 2002.
[6] UIC, Leaflet-774-3 Track/bridge interaction Reccomendations for calculations, 2nd edition. 2003.
[7] Manterola, J.; Astiz, M.A.; Martínez, A. Puentes de Ferrocarril de Alta Velocidad, Revista de Obras
Públicas No 3386, abril 2000 (in Spanish).
[8] González Requejo, P; et al. Alta velocidad: El fenómeno de interacción vía- tablero en puentes. Revista
de Obras Públicas No 3418., Febrero 2002 (in Spanish).
[9] Cuadrado Sanguino, M.; González Requejo, P. Consideración de las deformaciones por retracción y
fluencia en el estudio del fenómeno de interacción vía-tablero en el proyecto de puentes ferroviarios.
Revista de Obras Públicas, No 3446, Julio-Agosto 2004 (in Spanish).

© 2009 Taylor & Francis Group, London, UK