Soil-structure interaction in resonant railway bridges

A. Romero, M. Solı́s, J. Domı́nguez, P. Galvı́n

Escuela Técnica Superior de Ingenierı́a, Universidad de Sevilla, Camino de los Descubrimientos, 41092 Sevilla, Spain

Abstract

This paper explores dynamic soil-bridge interaction in high speed railway lines. The analysis was conducted
using a general and fully three-dimensional multi-body finite element-boundary element model formulated
in the time domain to predict vibrations caused by trains passing over the bridge. The vehicle was modelled
as a multi-body system, the track and the bridge were modelled using finite elements and the soil was
considered as a half-space by the boundary element method. The dynamic response of bridges to vehicle
passage is usually studied using moving force and moving mass models. However, the multi-body system
allows to consider the quasi-static and dynamic excitation mechanisms. Soil-structure interaction was taken
into account by coupling finite elements and boundary elements. The paper presents the results obtained
for a simply-supported short span bridge in a resonant regime under different soil stiffness conditions.
Key words: High speed train (HST), soil-structure interaction, railway bridge, resonant vibration

1. Introduction

In railway bridges, resonance occurs when the load frequency is equal to a multiple of the natural
frequency of the structure. In short span bridges, the actual train operating speed can be close to resonance
velocities. In that case, the high level vibrations reached in the resonance regime can lead to safety, passenger
comfort and train stability problems. Therefore, the dynamic behaviour of railway bridges is an important
design issue. The response of the structure is influenced by many factors such as soil properties, axle
loads, geometry and mass characteristics of the train and track irregularities. These effects are evaluated by
applying dynamic amplification factors to railway bridge standards, which represent the dynamic response
amplification compared to the static response for a single moving load [1]. However, dynamic amplification
factors do not account for resonance effects and their use is limited to train speeds below 220 km/h; in other
cases, further analysis is needed.
References about the dynamic response of railway bridges are quite extensive. Frýba [2] developed a
theoretical model of a bridge using the integral transformation method. This model provided an estimate

Email address: [email protected] (P. Galvı́n)

Preprint submitted to Soil Dynamics and Earthquake Engineering May 17, 2012
of the amplitude of free vibrations. Li and Su [3] studied the influence of vehicle-bridge interaction on
resonant vibrations. They concluded that the maximum bridge response in the resonant regime is reached
at the first resonance speed. Ju and Lin [4] proposed a three-dimensional finite element model to study
resonance effects in multi-span bridges. They concluded that load frequencies and natural frequencies of
bridges should be as different as possible to avoid resonance phenomena. Xia et al. [5] explored the resonance
mechanisms and conditions of the train-bridge system, analysing the resonant regimes according to their
excitation mechanisms.
Studying railway bridge vibrations requires an accurate model of the force induced by the train. Different
vehicle models have been used: the moving load model, the moving mass model and moving oscillator models.
The moving force model is the simplest vehicle model. It can be used if the train speed is low enough to
neglect its inertia and has been widely employed by the engineering community [6, 7, 8, 9, 10]. The moving
mass model is a more sophisticated model. It takes into account the mass of the vehicle but does not consider
the effect of suspension. Finally, comprehensive moving oscillator models have been used by several authors
[3, 4, 5, 11, 12]. Li and Su [3] established that the oscillator model leads to a lower response of the bridge
than the moving force model due to the dynamic train-bridge interaction. Pesterev et al. [11] examined
the asymptotic behaviour of the solution of the moving oscillator problem for large and small suspension
stiffness values. A conclusion of that study was that the moving mass model is not equivalent to the moving
oscillator model with infinite spring stiffness. Liu et al. [12] studied under which conditions the dynamic
train-bridge interaction should be considered for the dynamic analysis of railway bridges. They concluded
that the dynamic vehicle-bridge interaction is relevant for a large train-bridge mass ratio.
Little has been published about the influence of soil-structure interaction (SSI) on railway bridge re-
sponse. Takemiya and Bian [13] and Takemiya [14] studied the soil-foundation-bridge interaction under
moving loads using the substructure method in the frequency domain. They concluded that, in addition to
train load profile, bridge geometry and soil properties, soil-structure interaction is crucial for determining
the bridge and ground response. Recently, Ülker-Kaustell et al. [15] presented a qualitative analysis of the
soil-structure dynamic interaction of a portal frame railway bridge. They concluded that the contribution
of the soil-bridge interaction to the modal damping ratios is quite substantial in soft soils.
This paper presents the development of a three-dimensional numerical model to study the dynamic be-
haviour of a bridge accounting for soil-structure interaction (Fig. 1). The numerical model was based on
a three-dimensional finite element and boundary element formulation in the time domain. The articulated
train configuration was modelled as a multi-body system. The quasi-static and dynamic excitation mech-
anisms were considered. The outline of the paper is as follows: First, the numerical model is presented,
including a brief summary of the finite element and the boundary element time domain formulations as
well as the multi-body model used to represent the vehicle. Second, soil-structure interaction is analysed in
a short span railway bridge. Changes in modal parameters are analysed considering the effect of the soil.
2
Next, induced vibrations are computed for several train speeds. Resonant and non-resonant regimes are
studied. Finally, deck and support abutments are analysed for different soil types.

ΩBEM
ΩFEM

Figure 1: Vehicle-track-soil-structure interaction

2. Soil-structure interaction model

The dynamic response of a structure is considerably affected by the underlying soil. A homogeneous
soil causes damping effects due to wave radiation to infinity [16]. The present study used three-dimensional
finite element [17] and boundary element [18] time domain formulations to explore soil-structure behaviour.
The boundary element method system of equations can be solved step by step to obtain the time variation
of the boundary unknowns (i.e., displacements and tractions). Piecewise constant time interpolation func-
tions are used for tractions and piecewise linear functions are used for displacements. Nine-node rectangular
quadratic elements are used for spatial discretization. Explicit expressions of the fundamental solution of
displacements and tractions corresponding to an impulse point load in a three-dimensional elastic full-space
can be seen in reference [19].
After performing the spatial and temporal discretization, the following equation is obtained for each time
step:
∑
n−1
Hnn un = Gnn pn + (Gnm pm − Hnm um ) exp [−2πα(n − m)∆t] (1)
m=1

where un is the displacement vector and pn is the traction vector at the end of the time interval n, and
Hnn and Gnn are the fully unsymmetrical boundary element system matrices in the time interval n, α is
the soil attenuation coefficient and ∆t is the time step [20].
The equation that results from the finite element method can be expressed symbolically as follows if the
implicit Newmark time integration method is applied [21]:

Dnn un = fn + fn−1 (2)

where Dnn is the dynamic stiffness matrix, un is the displacement vector and fn is the equivalent force
vector at the end of the time interval n.
3
 In this paper, the damping matrix C was considered proportional to the mass matrix M and the stiffness
matrix K:
C = α0 M + α1 K (3)

α0 and α1 are obtained from the i-th (ζi ) and j-th (ζj ) modal damping ratios. The n-th modal damping
ratio is [22]:
α0 α1 ωn
ζn =
+ (4)
2ωn 2
The i-th and j-th modes should be chosen to obtain the damping ratios for all modes that contribute to the
response. If both modes have the same damping ratio ζ, the result is:
2ωi ωj 2
α0 = ζ α1 = ζ (5)
ωi + ωj ωi + ωj
Finally, coupling boundary element and finite element sub-regions provides satisfying equilibrium and com-
patibility conditions at the interface between both regions, as shown in reference [23].

3. Vehicle model

The train type considered in this paper consisted of one front traction car, eight passenger cars and one
rear traction car. Passenger cars adjacent to traction cars shared one bogie with the neighbouring passenger
car, while central passenger cars shared both bogies with the neighbouring cars. Bogie distances and axle
distances of the articulated HST were La = 3 m and Lb = 18.7 m, respectively. The mechanical properties
of the HST are summarised in Table 1.

Description Name Unit Traction cars Passenger cars

Mass of car body Mc kg 55790 24000
Mass of bogie Mb kg 2380 3040
Mass of wheel axle Mw kg 2048 2003
Car body inertia moment Jc kg m 2
1.15 × 10 3
1.48 × 103
Bogie inertia moment Jb kg m2 1.48 × 106 2.68 × 103
Primary suspension stiffness k1 N/m 2.45 × 106 1.4 × 106
Secondary suspension stiffness k2 N/m 2.45 × 106 0.82 × 106
Primary suspension damping c1 N s/m 20 × 103 10 × 103
Secondary suspension damping c2 N s/m 40 × 103 48 × 103

Table 1: Mechanical properties of HST

The multi-body model used to represent the vehicle is shown in Fig. 2.(a). Axles and car bodies were
considered to be rigid parts. Primary and secondary suspensions are represented by spring and damper
elements [24].
4
 xc,1 xc,2 xc,i
ϕc,1 Mc,1 Jc,1 ϕc,2 Mc,2 Jc,2 ϕc,i Mc,i Jc,i

2lc,i

(a)
xbc,1 xbc,2 xbc,3 xbc,4 xbc,j-1 xbc,j

(b)

Figure 2: (a) Multi-body model of an articulated HST. (b) Uncoupled bogies.

The equations of motion for the uncoupled multi-body systems shown in Fig. 2.(b) can be written as
follows:
M̃ẍ + C̃ẋ + K̃x = F̃ (6)

where M̃, C̃ and K̃ are the mass, damping and stiffness matrices, respectively. Vehicle response is described
by the displacement xc and rotation φc of the body, the displacement xb and rotation φb of the bogies and
the displacement of the wheels xwr and xwf (Fig. 3).

xbc

k2, c2
xb
Mb, Jb x ϕb
k1, c1

xw,r xw,f
Mw Mw
2lw

Figure 3: Multi-body model of a bogie

The mass matrix of each bogie (Eq. 7) is composed of the bogie mass Mb , the bogie inertia moment Jb
and the wheel masses Mw :

Mb = diag (0 Mb Jb Mw Mw ) (7)

5
The stiffness and damping matrices of a bogie can be written as:
   
k2 −k2 0 0 0 c −c2 0 0 0
   2 
   
 −k2 2k1 + k2 0 −k1 −k1   −c2 2c1 + c2 0 −c1 −c1 
   
   
Kb =  0 0 2
2k1 lw −k1 lw k1 lw  Cb =  0 0 2
2c1 lw −c1 lw c1 lw 
   
   
 0 −k1 −k1 lw k1 0   0 −c1 −c1 lw c1 0 
   
0 −k1 k1 lw 0 k1 0 −c1 c1 lw 0 c1
(8)
where k1 and c1 are the stiffness and damping of the primary suspension, k2 and c2 are the stiffness and
damping of the secondary suspension, and 2lw is the distance between the axles of a bogie.
The equation of motion of the whole train can be obtained from the displacement constraint equations
between car bodies and bogies. The constraint equations used to obtain the equation of motion of the front
traction car were [24]:

xbc,1 = xc,1 − φc,1 lc

(9)
xbc,2 = xc,1 + φc,1 lc

where xc,1 and φc,1 represent vertical displacement and rotation of the car body, respectively, and 2lc
represents the bogie distance in a vehicle. A similar expression can be drawn for the first passenger car (Fig.
2). The vertical displacement xbc,n of the m-th vehicle can be written as follows:

∑
m
( )
xbc,n = 2 (−1)n+i xc,i + (−1)n (xc,2 + lc,2 φc,2 ) (10)
i=3

The constraint equation for the whole train can be expressed as:

xbc = Lxc (11)

Introducing Eq. (11) into Eq. (6) leads to the following equation:

Mẍ + Cẋ + Kx = F (12)

where M, C and K are the mass, damping and stiffness matrices of the articulated HST (Fig. 2.(a)). The
mass matrix is obtained by assembling the car body mass matrix:

Mc = diag (Mc Jc ) (13)

where Mc is the mass of the car body and Jc is the inertia moment of the car body. The degree of freedom
of rotation of the vehicle makes it possible to consider the pitch moment of inertia of the car directly.
Finally, the multi-body system was coupled with the soil-structure interaction model, imposing equilib-
rium and compatibility conditions at each wheel-rail contact point. A Hertzian contact spring was considered
6
between wheels and rails [24, 25]. As the vehicle moves along the track, contact points between wheels and
rails change with time according to its speed. A moving node is created at each wheel-rail contact point
in the rail to couple the vehicles and the track [23, 26]. Thus, the track mesh changes at each time step.
Therefore, the mass, damping and stiffness matrices vary at each time step and the finite element system
of equations obtained becomes non-linear. Nevertheless, the time domain formulation makes it possible to
solve the non-linear system of equations [27].
Induced vibrations due to HST passage are generated by several excitation mechanisms: a quasi-static
contribution, a parametric excitation due to the discrete support of the rails and a dynamic contribution.
Usually, the quasi-static contribution is modelled as constant moving forces, neglecting the inertia effects
of the vehicle. The multi-body system proposed in this paper allows considering the sprung and unsprung
masses and the vehicle’s suspension. The dynamic contribution accounts for track and wheel irregularities.
Displacement uc at the wheel-rail contact point is equal to the sum of rail displacement ur due to the quasi-
static contribution and the parametric excitation, and rail unevenness uw/r perceived by an axle [28, 29] in
order to consider the unevenness:
uc = ur + uw/r (14)

Random track unevenness uw/r (y) was modelled as a stationary Gaussian random process characterized by
its one-sided PSD function S̃uw/r (ky ). The spectral representation theorem was used to generate samples of
track unevenness uw/r (y) as a superposition of harmonic functions with random phase angles [28, 29]:
n √
∑
uw/r (y) = 2S̃uw/r (kym )∆ky cos(kym y − θm ) (15)
m=1

where kym = m∆ky is the wavenumber sampling used only to compute the artificial profile, ∆ky is the
wavenumber step and θm are the independent random phase angles uniformly distributed in the interval
[0, 2π]. The artificial track profile was generated from the PSD function following the ISO 8608 standard
[30]:
( )−w
ky
S̃uw/r (ky ) = S̃uw/r (ky0 ) (16)
ky0
An artificial profile was obtained from the PSD function with ky0 = 1 rad/m and S̃uw/r (ky0 ) = 2π ×10−8 m3 .
Wheel-rail unevenness in current high speed lines is commonly assumed to be w = 3.5 .

4. High speed railway bridges

This section analyses the effect of soil-structure interaction on railway bridges. A simply supported 12 m
long railway bridge was studied. The deck (Fig. 4.(a)) was composed of a 0.25 m thick concrete slab. The
slab rested over five pre-stressed concrete beams with a 0.75 × 0.3 m rectangular cross-section. The distance
3
between beams was 1.39 m. The concrete properties were the following: density ρ = 2500 kg/m , Poisson’s
2
ratio ν = 0.2 and Young’s modulus E = 31 × 109 N/m .
7
 UIC-60
1.46
Rail-pad
1.30
Concrete sleeper 6.00
0.72

Ballast layer 0.70

0.25

1.39 1.39 1.39 1.39 1.20

0.75 7.30 1.20
2.40
0.3 0.3 0.3 0.3 0.3
5.84 1.20

(a) (b)

Figure 4: (a) Deck cross-section. (b) Abutment geometry.

The deck leaned over two concrete abutments (Fig. 4.(b)) with the following properties: density ρ =
3 2
2500 kg/m , Poisson’s ratio ν = 0.3 and Young’s modulus E = 20 × 109 N/m . The beams rested on
laminated rubber bearings. The thickness of the bearings was 20 mm and their stiffness and damping values
were kb = 560 × 106 N/m and cb = 50.4 × 103 Ns/m, respectively.
A single ballast track was located over the deck. The track was composed of two UIC60 rails with a
bending stiffness EI = 6.45×106 Nm2 and a mass per unit length m = 60.3 kg/m for each rail. The rail pads
had a thickness of 10 mm and stiffness and damping values of krp = 150×106 N/m and crp = 13.5×103 Ns/m,
respectively. The pre-stressed concrete mono-block sleepers had the following characteristics: length l =
2.50 m, width w = 0.235 m, height h = 0.205 m (under the rail) and mass m = 300 kg. A distance d = 0.6 m
3
between the sleepers was considered. The ballast had a density ρ = 1800 kg/m , Poisson’s ratio ν = 0.2, and
2
Young’s modulus E = 209 × 106 N/m . The width of the ballast was 2.92 m and the height was h = 0.7 m.
The structure was assumed to be located on top of a half-space that represented the soil. The soil was
a homogeneous viscoelastic soil with Poisson’s ratio ν = 0.35 and mass density ρ = 1800 kg/m3 . Four
different S–wave velocities were analysed, corresponding to a soil with infinite stiffness (Cs = ∞ m/s), a stiff
soil (Cs = 400 m/s), a medium soil (Cs = 250 m/s) and a soft soil (Cs = 150 m/s).

4.1. Soil-structure dynamic behaviour

Resonance in railway bridges occurs when the load frequency of the train becomes close to the natural
frequency of the structure. For a resonant regime to begin, the structure also needs to be lightly damped.
Fig. 6 shows four natural frequencies and mode shapes of the structure: the first bending mode(symmetric),
the first torsional mode, the first bending of cross-section mode (symmetric) and the first antisymmetric
deck bending mode shape, respectively. At least, the contribution of the first symmetric and antisymmetric
bending modes to the response of simply supported railway bridges should be considered [6]. Therefore,
8
 ΩFEM

ΩBEM

Figure 5: Soil-structure discretization

the finite element damping matrix (Eq. 2) is obtained by assigning the same damping ratio η = 2 % to
these modes considering ωi = ω1 and ωj = ω4 . Rayleigh damping parameter values are α0 = 2.3 s−1 and
α1 = 1.24 × 10−4 s.

(a) f1 = 11.80 Hz (b) f2 = 21.90 Hz (c) f3 = 29.99 Hz (d) f4 = 47.82 Hz

Figure 6: Four mode shapes of the structure

The influence of the soil-structure interaction on the dynamic behaviour was evaluated by means of
the deck receptance. The deck receptance was computed by the frequency content of the track response.
Fig. 7 shows the vertical displacement at the centre of the mid-span deck due to an impulsive load P (t) =
−1 N (H(t) − H(t − 0.045 s)) acting on both rails. Fig. 7.(a) shows how the maximum displacement reached
during the acting load increased as the soil stiffness decreased. The receptance (Fig. 7.(b)) shows that the
response was governed by the first deck bending (symmetric) mode. The first resonant frequencies of the
structure ranged from f1 = 11.80 Hz, when the soil was not taken into account, to f1 = 9.89 Hz for the
soft soil. The damping of the system, obtained from the free vibration response, reached a maximum value
ζ = 6.4 % in the soft soil and a minimum value ζ = 2 % when the interaction was not considered. Table 2
summarizes the results for the different soil types considered. The soil-structure interaction led to a decrease
in frequencies and an increase in damping ratios.

9
 −8 −6
x 10 x 10
1.5 1.2

1 1

Receptance [m/N/Hz]
Receptance [m/N]

0.5
0.8
0
0.6
−0.5
0.4
−1

−1.5 0.2

−2 0
0 0.2 0.4 0.6 0.8 1 0 10 20 30 40 50 60 70 80
Time [s] Frequency [Hz]
(a) (b)

Figure 7: (a) Time history and (b) frequency content of deck receptance for Cs = ∞ m/s (grey dashed line), Cs = 400 m/s
(black dashed line), Cs = 250 m/s (grey solid line) and Cs = 150 m/s (black solid line)

4.2. Soil-structure vibrations

The resonant condition of a bridge excited by a row of moving forces can be expressed as follows [2, 5]:

fn d
vn,i = (n = 1, 2, ..., i = 1, 2, ...) (17)
i

where vn,i is the resonant train speed, fn is the n-th resonant frequency of the bridge and d is a characteristic
distance between moving loads. Fig. 8 shows the maximum vertical acceleration at the centre of the
deck for a range of train speeds between 30 m/s and 130 m/s (108 km/h and 468 km/h, respectively). The
deck acceleration was found to increase with train speed. Local maxima were reached at resonant speeds
corresponding to the first bending mode shape, considering the distance between bogies d = 18.7 m. Fig. 8
shows maximum vibration levels at speed v1,2 = 110.4 m/s when soil-bridge interaction was not considered.
The response of the structure changed substantially when soil-structure interaction was considered. The
second resonant speed of the first mode shape decreased to v1,2 = 103 m/s and v1,2 = 95 m/s for stiff and
medium soil, respectively, due to the change in the dynamic behaviour of the system (Table 2). In addition,
the maximum level of acceleration reached in the resonant regime was significantly lower when the soil-
structure interaction was considered. No resonant effects occurred with soft soil. In all cases, the maximum
acceleration at the centre of the mid-span deck was below amax = 3.5 m/s2 in the range of operating speeds
on current high speed lines, that is, the limit set by the European Committee for Standardisation (CEN)
[1].
Soil-structure vibrations induced by train passage at the second resonant speed of the first mode of
vibration v1,2 are analysed below for each soil type. Although the maximum response occurred at the first
resonant speed, this speed was much higher than the current operation speed, for example v1,1 = 220.7 m/s
(794 km/h) when soil-structure interaction was not taken into account.
Fig. 9 shows the dynamic amplification factors (DAF ) of the soil and the structure due to a HST

10
 20

15

Acceleration [m/s2]
10

0
40 60 80 100 120
Train speed [m/s]

Figure 8: Maximum vertical acceleration at the centre of the mid-span deck for Cs = ∞ m/s (grey dashed line), Cs = 400 m/s
(black dashed line), Cs = 250 m/s (grey solid line) and Cs = 150 m/s (black solid line). The second resonant speeds of the
first mode are marked with vertical lines.

Soil type cS [m/s] f1 [Hz] ζ1 v1,2 [m/s] DAF

Infinitely stiff - 11.80 0.020 110.4 4.4
Stiff soil 400 11.01 0.034 103 2.3
Medium soil 250 10.16 0.046 95 1.9
Soft soil 150 9.89 0.064 92.5 1.8

Table 2: Summary of soil-bridge interaction according to modal properties and resonant speeds

travelling at the speed v1,2 shown in Table 2. The DAF is defined as:

ud
DAF = (18)
us

where ud are the displacements for the different soil types and train speeds at the time step when the
maximum deck response is achieved, and us is the maximum static deck displacement without considering
soil-bridge interaction. As expected, maximum dynamic amplification was reached when soil-bridge inter-
action was not considered, and bridge displacements decreased with soil stiffness. As the soil became softer,
its influence on bridge response increased. This effect is further described in section 4.3.
Fig. 10 shows the time histories of vertical acceleration at the centre of the mid-span deck for speed v1,2
2
and the acceleration limit for a ballasted track a = 3.5 m/s [1]. Figs. 10.(a-c) show a gradual increase of
the response of the structure with successive bogie passage. The amplification of the response with each
bogie passage was greater when soil interaction was not considered and decreased when soil stiffness was
lower. There was no evidence of amplification in soft soil (Fig. 10.(d)).
Fig. 11 shows the frequency content of vertical acceleration at the centre of the mid-span deck for train
speeds from 30 m/s to 130 m/s. The frequency content was normalized to the acceleration level at the first
resonance frequency for a train travelling at v1,2 when soil-structure interaction was not considered. Fig.
11.(a) shows a global maximum at the first resonant vibration f1 = 11.80 Hz for the second resonant speed
11
 (a) 4 3 2 1 0 (b) 4 3 2 1 0

(c) 4 3 2 1 0 (d) 4 3 2 1 0

Figure 9: Dynamic amplification factor of ground-structure displacements for a HST travelling at (a) v = 110.4 m/s, (b)
v = 103 m/s, (c) v = 95 m/s and (d) v = 92.5 m/s on (a) infinitely stiff soil, (b) hard soil, (c) medium soil and (d) soft soil

12
 Acceleration [m/s2] 20 10

Acceleration [m/s2]
10 5

0 0

−10 −5

−20 −10
0 0.5 1 1.5 2 0 0.5 1 1.5 2
(a) Time [s] (b) Time [s]

5 5
Acceleration [m/s2]

Acceleration [m/s2]
0 0

−5 −5
0 0.5 1 1.5 2 0 0.5 1 1.5 2 2.5
(c) Time [s] (d) Time [s]

Figure 10: Time histories of vertical acceleration at the centre of the mid-span deck for a HST travelling at (a) v = 110.4 m/s,
(b) v = 103 m/s, (c) v = 95 m/s and (d) v = 92.5 m/s on (a) infinitely stiff soil, (b) hard soil, (c) medium soil and (d) soft soil

v1,2 = 110.4 m/s and local maxima for v1,3 , v1,4 , v1,5 and v1,6 . As the soil-bridge interaction increased
(Figs. 11.(b-d)), the acceleration level at the bogie passage frequency fb = v/lb became dominant and the
frequency content of the response of the bridge at the first resonance frequency decreased.
Fig. 12 shows the frequency content of vertical acceleration at a train speed v1,2 for the different soil
types studied. Maximum vibration level was reached at the first resonance frequency when soil interaction
was not considered. The amplification of the resonant response decreased with lower soil stiffness, as well
as the relationship between frequency content at the first resonance frequency and bogie passage frequency.
Both levels were similar for soft soil. This matched the severity of the response in the resonant regime.

4.3. Abutment support response

Fig. 13 shows the time histories and one-third band spectra of vertical acceleration at the bottom of
the first and second support abutments at v1,2 for each soil type. The vertical acceleration of the deck is
superimposed on the one-third band spectra. Resonant behaviour was reached in the abutment in stiffer and
medium soil. Acceleration levels increased with lower soil stiffness and the maximum response was obtained
in soft soil, although resonant behaviour was not observed. The first abutment reached higher acceleration
vibration levels than the second abutment due to a sudden stiffness change between the track-soil and the
track-bridge-soil systems. In addition, the difference between the abutments and deck responses decreased
in softer soil, since soil-bridge interaction became dominant.
Fig. 14 shows the time histories and one-third band spectra of the vertical reaction forces at the bottom

13
 (a) (b)

(c) (d)

Figure 11: Normalized frequency content of vertical acceleration at the centre of the mid-span deck for a HST travelling from
30 m/s to 130 m/s on (a) infinitely stiff soil, (b) hard soil, (c) medium soil and (d) soft soil

10

8
Acceleration [m/s2/Hz]

0
0 10 20 30 40 50 60 70 80
Frequency [Hz]

Figure 12: Frequency content of vertical acceleration at the centre of the mid-span deck for a HST travelling at v = 110.4 m/s
(infinitely stiff soil, grey dashed line), v = 103 m/s (hard soil, black dashed line), v = 95 m/s (medium soil, grey solid line) and
v = 92.5 m/s (soft soil, black solid line).

14
 0.5 1
Acceleration [m/s2]

Acceleration [m/s2]

Acceleration [m/s2]
0.2
0.1 0.5

0 0 0
−0.1 −0.5
−0.2
−0.5 −1
0 0.5 1 1.5 2 0 0.5 1 1.5 2 0 0.5 1 1.5 2
(a) Time [s] (b) Time [s] (c) Time [s]

150 150 150

[dB, ref 10−6 m/s2]

[dB, ref 10−6 m/s2]

[dB, ref 10−6 m/s2]

Acceleration

Acceleration

Acceleration
100 100 100

50 50 50

0 0 0
1 1.6 2.5 4 6.3 10 16 25 40 63 100 1 1.6 2.5 4 6.3 10 16 25 40 63 100 1 1.6 2.5 4 6.3 10 16 25 40 63 100
(d) One−third octave band spectra [Hz] (e) One−third octave band spectra [Hz] (f) One−third octave band spectra [Hz]

Figure 13: (a-c) Time history and (d-f) one-third band spectra of vertical acceleration at the the bottom of the first abutment
support (black solid line) and the second abutment support (grey line) for a HST travelling at (a,d) v = 103 m/s (Cs = 400 m/s),
(b,e) v = 95 m/s (Cs = 250 m/s) and (c,f) v = 92.5 m/s (Cs = 150 m/s). The dotted line represents the vertical acceleration of
the deck superimposed on the one-third band spectra

of the first and second support abutments for the resonant train speeds. Reaction forces were higher when
soil-interaction was not considered. These forces were superimposed on the one-third band spectra. The
highest force transmitted to the soil was reached when the resonant regime occurred on stiff soil. In this
case, the contribution of high frequencies was noticeable. Reaction forces decreased with soil stiffness and
minimum forces were computed on soft soil.

5. Conclusions

This paper presents a numerical model to predict vibrations on railway bridges. The numerical model
was based on three-dimensional finite element and boundary element formulations in the time domain.
Articulated HSTs were modelled as a multi-body system. It was possible to consider the different excitation
mechanisms accurately. The following conclusions can be drawn from the results obtained:

1. Soil-structure interaction leads to changes in dynamic behaviour. The fundamental periods and damp-
ing ratios of the response were higher when soil-structure interaction was considered than when it was
not.
2. The resonance condition in railway bridges depends on resonance frequencies. Resonant train speeds
were lower when soil-bridge interaction was considered. Amplification in the resonant regime was also
lower.
3. Therefore, dynamic effects on railway bridges considering soil-structure interaction are an important
issue in structure design. Computed bridge deck acceleration is more realistic and the limits set
by railway bridge standards could be satisfied with a more slender structure. Moreover, resonance
15
 5 5 5
x 10 x 10 x 10
6 6 6
Reaction force [N]

Reaction force [N]

Reaction force [N]

4 4 4

2 2 2

0 0 0

−2 −2 −2
0 0.5 1 1.5 2 0 0.5 1 1.5 2 0 0.5 1 1.5 2
(a) Time [s] (b) Time [s] (c) Time [s]
6 6 6
10 10 10
Reaction force [N]

Reaction force [N]

Reaction force [N]

4 4 4
10 10 10

2 2 2
10 10 10
1 1.6 2.5 4 6.3 10 16 25 40 63 100 1 1.6 2.5 4 6.3 10 16 25 40 63 100 1 1.6 2.5 4 6.3 10 16 25 40 63 100
(d) One−third octave band spectra [Hz] (e) One−third octave band spectra [Hz] (f) One−third octave band spectra [Hz]

Figure 14: (a-c) Time history and (d-f) one-third band spectra of the vertical reaction forces at the bottom of the entry
abutment support (black solid line) and the exit abutment support (grey line) for a HST travelling at (a,d) v = 103 m/s
(Cs = 400 m/s), (b,e) v = 95 m/s (Cs = 250 m/s) and (c,f) v = 92.5 m/s (Cs = 150 m/s). The dotted line represents the
reaction forces without considering soil interaction superimposed on the one-third band spectra.

effects may occur at lower operation speeds than those predicted when soil-bridge interaction is not
considered.
4. The response of the abutment support depends on soil stiffness. This response was higher in soils
with lower stiffness. However, the maximum force transmitted to the soil and resonant response were
reached in the stiffest soil.

Acknowledgments

This research was financed by the Spanish Ministry of Science and Innovation (Ministerio de Ciencia
e Innovación) under research project BIA2010-14843. This financial support is gratefully acknowledged.
The authors also wish to acknowledge the support provided by the Andalusian Scientific Computing Centre
(CICA).

References

[1] European Committee for Standardisation (CEN), Eurocode 1: actions on structures-part 2: traffic loads on bridges. (2008).
[2] L. Frýba, A rough assessment of railway bridges for high-speed trains, Engineering Structures (23) (2001) 548–556.
[3] J. Li, M. Su, The resonant vibration for a simply supported girder bridge under high-speed trains, Journal of Sound and
Vibration 224 (5) (1999) 897–915.
[4] S. H. Ju, H. T. Lin, Resonance characteristics of high-speed trains passing simply supported bridges, Journal of Sound
and Vibration 267 (5) (2003) 1127–1141.
[5] H. Xia, N. Zhang, W. W. Guo, Analysis of resonance mechanism and conditions of train-bridge system, Journal of Sound
and Vibration 297 (3-5) (2006) 810–822.
16
 [6] P. Museros, E. Alarcón, Influence of the second bending mode on the response of high-speed bridges at resonance, Journal
of Structural Engineering 131 (3) (2005) 405–415.
[7] P. Museros, M. D. Martı́nez-Rodrigo, Vibration control of simply supported beams under moving loads using fluid viscous
dampers, Journal of Sound and Vibration 300 (1-2) (2007) 292–315.
[8] M. D. Martı́nez-Rodrigo, J. Lavado, P. Museros, Transverse vibrations in existing railway bridges under resonant condi-
tions: Single-track versus double-track configurations, Engineering Structures 32 (7) (2010) 1861–1875.
[9] M. D. Martı́nez-Rodrigo, J. Lavado, P. Museros, Dynamic performance of existing high-speed railway bridges under
resonant conditions retrofitted with fluid viscous dampers, Engineering Structures 32 (3) (2010) 808–828.
[10] M. D. Martı́nez-Rodrigo, P. Museros, Optimal design of passive viscous dampers for controlling the resonant response of
orthotropic plates under high-speed moving loads, Journal of Sound and Vibration 330 (7) (2011) 1328–1351.
[11] A. V. Pesterev, L. A. Bergman, C. A. Tan, T. C. Tsao, B. Yang, On asymptotics of the solution of the moving oscillator
problem, Journal of Sound and Vibration 260 (3) (2003) 519–536.
[12] K. Liu, G. D. Roeck, G. Lombaert, The effect of dynamic train-bridge interaction on the bridge response during a train
passage, Journal of Sound and Vibration 325 (1-2) (2009) 240–251.
[13] H. Takemiya, X. C. Bian, Shinkansen high-speed train induced ground vibrations in view of viaduct-ground interaction,
Soil Dynamics and Earthquake Engineering 27 (6) (2007) 506–520.
[14] H. Takemiya, Analyses of wave field from high-speed train on viaduct at shallow/deep soft grounds, Journal of Sound and
Vibration 310 (3) (2008) 631–649.
[15] M. Ülker-Kaustell, R. Karoumi, C. Pacoste, Simplified analysis of the dynamic soil-structure interaction of a portal frame
railway bridge, Engineering Structures 32 (11) (2010) 3692–3698.
[16] O. von Estorff, Dynamic response of elastic blocks by time domain bem and fem, Computers and Structures 38 (3) (1991)
289–300.
[17] O. C. Zienkiewicz, The finite element method, 3rd Edition, McGraw-Hill, 1986.
[18] J. Domı́nguez, Boundary elements in dynamics, Computational Mechanics Publications and Elsevir Aplied Science,
Southampton, 1993.
[19] P. Galvı́n, J. Domı́nguez, Analysis of ground motion due to moving surface loads induced by high-speed trains, Engineering
Analysis with Boundary Elements 31 (11) (2007) 931–941.
[20] P. Galvı́n, J. Domı́nguez, High-speed train-induced ground motion and interaction with structures, Journal of Sound and
Vibration 307 (3-5) (2007) 755–777.
[21] N. M. Newmark, A method of computation for structural dynamics, ASCE Journal of the Engineering Mechanics Division
85 (1) (1959) 67–94.
[22] R. W. Clough, J. Penzien, Dynamic of Structures, McGraw-Hill, New York, 1975.
[23] P. Galvı́n, A. Romero, J. Domı́nguez, Fully three-dimensional analysis of high-speed traintracksoil-structure dynamic
interaction, Journal of Sound and Vibration 329 (24) (2010) 5147–5163.
[24] X. Sheng, C. J. C. Jones, D. J. Thompson, A theoretical model for ground vibration from trains generated by vertical
track irregularities, Journal of Sound and Vibration 272 (3-5) (2004) 937–965.
[25] C. Esveld, Modern Railway Track, MRT Productions, Zaltbommel, 2001.
[26] A. Romero, P. Galvı́n, J. Domı́nguez, Short span bridges dynamic behaviour account for the vehicle-track-structure-soil
dynamic interaction [comportamiento dinámico de viaductos cortos considerando la interacción vehı́culo-vı́a-estructura-
suelo], Revista Internacional de Metodos Numéricos para Cálculo y Diseño en Ingenierı́a 28 (1) (2012) 55–63.
[27] S. Y. Chang, Nonlinear error propagation analysis for explicit pseudodynamics algorithm, Journal of Engineering Mechan-
ics ASCE 123 (2003) 841–850.
[28] G. Lombaert, G. Degrande, Ground-borne vibration due to static and dynamic axle loads of intercity and high-speed

17
 trains, Journal of Sound and Vibration 319 (3-5) (2009) 1036–1066, cited By (since 1996) 16.
[29] G. Lombaert, G. Degrande, J. Kogut, S. Franc.ois, The experimental validation of a numerical model for the prediction of
railway induced vibrations, Journal of Sound and Vibration 297 (3-5) (2006) 512–535.
[30] International Organization for Standardization, ISO 8608:1995 Mechanical vibration road surface profiles-reporting of
measured data (1995).

18

View publication stats