EXPERIMENTAL AND NUMERICAL MODAL ANALYSIS OF A CONCRETE

HIGH SPEED TRAIN RAILWAY BRIDGE

J. Maeck, G. De Roeck
Department of Civil Engineering, Division of Structural Mechanics,
K.U. Leuven, W. de Croylaan 2, B-3001 Heverlee, Belgium

ABSTRACT

Train induced vibrations are a major environmental concern both in Europe and China. Besides the effect
of vibrations due to passenger and freight trains and subways at relatively low speed, the study of the
vibrational impact of high speed trains is of high interest. In Belgium, for example, new high speed train
lines connect Brussels with Paris and London, while extensions to Amsterdam and Cologne are presently
under construction. In China, this problem will become equally important in the near future, as a high
speed train connection is planned between Beijing and Shanghai.
The partners in this research project are involved in the development of numerical models to predict
traffic induced vibrations. The development, validation and practical use of these models rely on in situ
vibration measurements. A preliminary measurement campaign was undertaken on a high speed train
bridge, with sensors on the bridge as well as on the rails, to be able to get more insight in force transfer
from train to construction.

Keywords: high speed train, dynamic testing, modal analysis

INTRODUCTION

The construction under consideration is a railway bridge for the high speed train between Paris and
Brussels. The bridge is situated near the Belgian village Antoing, close to the Belgian-French border. The
total bridge consists of successively five prestressed concrete bridges of each 50m span, a mixed steel-
concrete bowstring bridge, which is built over a river, and a last 50m span concrete bridge. The bridge
tests and simulations can be situated in a bilateral research program (BIL98/09) between the Belgian
universities K.U. Leuven and V.U.Brussel, and the Northern Jiaotong University, Beijing, China. Aim of
the program is to study the train-structure interaction and the vibrations induced in the environment.

FINITE ELEMENT MODEL

A complete modal analysis of the first 50 span bridge is used as starting point for the experimental
survey.
From symmetry considerations only half a span is modeled in the finite element package Ansys [2]. Eight
noded brick elements are used, three translational degrees of freedom at each node. The concrete is
considered homogeneous, E-modulus 40,000 MPa, density 2500 kg/m3. The bridge section is denoted in
Figure 1. The stone material of the ballast is accounted for as extra mass (1500kg/m3). The supports are
considered fixed in vertical direction.
 Figure 1 Elevation & cross section Antoing bridge

Figure 2 (a) Antoing bridge (b) Maintenance tunnel

Frequencies of all modes under 25 Hz are summarized in Table 1.

Symmetrical modes eigenfrequency (Hz) characterization

B1 3.08 first bending
T1 3.88 first torsional
S1 6.55 first bending of cross section
B3 14.57 second symmatrical bending
S2 14.92 second bending of cross section
Anti-symmetrical modes
B2 9.32 first anti-symmetrical bending
T2 11.17 second torsional
S3 12.67 anti-symmetrical, first bending of cross section
B4 18.44 second anti-symmetrical bending
Table 1 Eigenfrequencies
An overview of the eigenmodes is given in the next plots.
1 ANSYS 5.5.1 1 ANSYS 5.5.1
MAY 30 2000 MAY 30 2000
17:28:52 17:29:18
PLOT NO. 1 PLOT NO. 1
DISPLACEMENT DISPLACEMENT
STEP=1 STEP=1
SUB =2 SUB =3
FREQ=3.083 FREQ=3.878
PowerGraphics PowerGraphics
EFACET=1 EFACET=1
AVRES=Mat AVRES=Mat
DMX =.00108 DMX =.001491

*DSCA=3001 *DSCA=2001
XV =.55007 XV =.55007
YV =-.67695 YV =-.67695
ZV =.48904 ZV =.48904
Z DIST=18.378 Z DIST=18.595
Y XF =13.087 Y XF =13.119
X ZF =2.766 X YF =.291747
A-ZS=-61.527 ZF =2.652
Z-BUFFER A-ZS=-61.527
Z-BUFFER

MODAL ANALYSIS (BRIDGE ANTOING) MODAL ANALYSIS (BRIDGE ANTOING)

Figure 3 T1 & B1
1 ANSYS 5.5.1 1 ANSYS 5.5.1
MAY 30 2000 MAY 30 2000
17:29:37 17:31:37
PLOT NO. 1 PLOT NO. 1
DISPLACEMENT DISPLACEMENT
STEP=1 STEP=1
SUB =4 SUB =4
FREQ=6.553 FREQ=6.553
PowerGraphics PowerGraphics
EFACET=1 EFACET=1
AVRES=Mat AVRES=Mat
DMX =.00166 DMX =.00166

*DSCA=1001 *DSCA=1001
XV =.55007 XV =-1
Z YV =-.67695 DIST=9.772
ZV =.48904 XF =13.261
Y DIST=17.572 ZF =1.737
X XF =13.261 A-ZS=88.473
ZF =1.737 Z-BUFFER
A-ZS=-61.527
Z-BUFFER
Z

Y X

MODAL ANALYSIS (BRIDGE ANTOING) MODAL ANALYSIS (BRIDGE ANTOING)

Figure 4 S1 (axonometric and front view)

1 ANSYS 5.5.1 1 ANSYS 5.5.1
MAY 30 2000 MAY 30 2000
17:29:54 17:30:06
PLOT NO. 1 PLOT NO. 1
DISPLACEMENT DISPLACEMENT
STEP=1 STEP=1
SUB =5 SUB =6
FREQ=14.568 FREQ=14.924
PowerGraphics PowerGraphics
EFACET=1 EFACET=1
AVRES=Mat AVRES=Mat
DMX =.002076 DMX =.001531

*DSCA=1001 *DSCA=1001
XV =.55007 XV =.55007
Z YV =-.67695 YV =-.67695
Y ZV =.48904 Z ZV =.48904
DIST=18.45 Y DIST=18.414
X XF =13.225 XF =13.221
ZF =1.107 X YF =.728566
A-ZS=-61.527 ZF =1.689
Z-BUFFER A-ZS=-61.527
Z-BUFFER

MODAL ANALYSIS (BRIDGE ANTOING) MODAL ANALYSIS (BRIDGE ANTOING)

Figure 5 B3 & S2
1 ANSYS 5.5.1 1 ANSYS 5.5.1
MAY 30 2000 MAY 30 2000
15:00:41 15:01:22
PLOT NO. 1 PLOT NO. 1
DISPLACEMENT DISPLACEMENT
STEP=1 STEP=1
SUB =3 SUB =4
FREQ=9.322 FREQ=11.171
PowerGraphics PowerGraphics
EFACET=1 EFACET=1
AVRES=Mat AVRES=Mat
DMX =.001496 DMX =.001408

*DSCA=1501 *DSCA=1501
XV =.550073 XV =.550073
Z YV =-.676947 YV =-.676947
ZV =.489043 Z ZV =.489043
Y DIST=18.932 DIST=18.819
Y
X XF =13.316 XF =13.283
ZF =1.525 X YF =.242912
A-ZS=-61.527 ZF =2.104
Z-BUFFER A-ZS=-61.527
Z-BUFFER

MODAL ANALYSIS (BRIDGE ANTOING) MODAL ANALYSIS (BRIDGE ANTOING)

Figure 6 B2 & T2
1 ANSYS 5.5.1 1 ANSYS 5.5.1
MAY 30 2000 MAY 30 2000
15:01:43 15:02:07
PLOT NO. 1 PLOT NO. 1
DISPLACEMENT DISPLACEMENT
STEP=1 STEP=1
SUB =5 SUB =6
FREQ=12.668 FREQ=18.44
PowerGraphics PowerGraphics
EFACET=1 EFACET=1
AVRES=Mat AVRES=Mat
DMX =.001757 DMX =.002355

*DSCA=1001 *DSCA=1001
XV =.550073 XV =.550073
YV =-.676947 Z YV =-.676947
Z ZV =.489043 Y ZV =.489043
Y DIST=18.177 DIST=18.378
X
X XF =13.327 XF =13.328
ZF =1.733 YF =.140E-03
A-ZS=-61.527 ZF =1.008
Z-BUFFER A-ZS=-61.527
Z-BUFFER

MODAL ANALYSIS (BRIDGE ANTOING) MODAL ANALYSIS (BRIDGE ANTOING)

Figure 7 S3 & B4

EXPERIMENTAL TEST PROCEDURE

The measurement campaign is focused on the bridge response to ambient vibration. Excitation of the
bridge is due to train passages primarily and to wind load, road traffic (underneath bridge) and micro-
tremors.

The experimental data to be measured on the bridge are accelerations, strains and a deflection.
The bridge span is divided in 12 equidistant zones. A total of 29 vertical, 4 transversal and 2 longitudinal
accelerations, 18 strains and 1 vertical deflection are measured.
At the lower side of the girder, at the mid-section, two rosettes of resistance strain gauges are glued , at
the center and at the location under one of the railway tracks, in order to investigate the strain changes
during train passages. At the same location of one of the strain rosettes, the vertical deflection at one point
of the bridge girder is measured with a LVDT. Also at the same location, the acceleration is measured.

In order to measure the train induced interaction forces, strain gauges are glued to both vertical sides of
two rails of one track, again in rosette formation.
The sensor locations are indicated in Figure 8. Each sensor location has a unique number, A denotes
acceleration measurement, S strain gauge measurement and D deflection measurement. Measurement
direction is indicated by x, y or z (x longitudinal, y transverse, z vertical). Three reference accelerometers
are used (A3z, A4z, A4y).

Figure 8 Sensor locations

For each set-up, three train passages are measured at sampling frequency 5000 Hz, 240k samples (49sec);
reference A3z is used as trigger, with pretrigger of 6sec.
Measurement acquisition hardware equipment (portable PC, KEMO anti aliasing filter and amplifier,
DAT recorder) is installed at the transverse maintenance tunnel between the 1st and 2nd span.
 Figure 9 (a) Acquisition equipment (b) Strain gauge rosette on rail
The LVDT at the girder is installed on a stiff cable stayed antenna construction.

EXPERIMENTAL RESULTS

Figure 10 shows the time signal and frequency contents of a typical acceleration data series of one of the
accelerometers (A3z) placed in the maintenance tunnel at the France to Brussels side. Two trains are
passing, a long train in direction Brussels , a short train in direction France. High frequency contents is
noticed during train passage.

Figure 10 Time signal & spectrum (1st passage) A3z

The vertical displacement measured by the LVDT on the stiff antenna, at the center of the span is shown
in Figure 11. Maximum deflection is 1500µm. From the free vibration after passage the first
eigenfrequency at 3.12Hz is clearly detectable. Higher eigenfrequencies are more difficult to detect as
displacement spectrum is decreasing quickly and many of the higher modes have zero displacement at the
sensor location.
 Figure 11 LVDT signal
The strain gauge signal (S17x) is denoted in Figure 12. Although the strain level reaches a low peak value
of 6µS the 13 boogie passages are clearly detectable. Boogie 2&3 and 11&12 are close resulting in
double peak in the time signal. From this timesignal the train speed is estimated at 265km/h.

Figure 12 (a) Strain gauge on bridge (b) High speed train composition

DATA PROCESSING

The stochastic subspace identification technique is used to extract the modal characteristics from the data
[3]. The technique has been implemented (SPICE [4]) in the MATLAB environment [5]. Data is
decimated with a factor 100 (low pass filtering at 50Hz) and limited to free vibration after train passage,
to result in frequencies and damping ratios of Table 2.
mode eigenfrequency (Hz) damping ratio (%) characterization
B1 3.12 0.55 first bending
T1 3.85 0.95 first torsional
S1 6.59 1.2
B2 8.54 1.3 first anti-symmetrical bending
T2 10.34 2.0 second torsional
S3 11.99 0.98
B3 18.56 1.1 second symmetrical bending
B4 19.28 0.56
Table 2 Experimental modal characteristics

Figure 13 Experimental modes 1,2,3

Figure 14 Experimental modes 4,5,7

CONCLUSIONS

A full modal analysis was carried out on a prestressed concrete bridge for the high speed train, using the
acceleration data captured in the maintenance tunnels. The experimentally determined eigenfrequencies
showed a good correspondence with the numerical results from a finite element model, as far as the
lowest eigenfrequencies are concerned. For further agreement between experiments and FE model,
updating of E-modulus and boundary condition characteristics is now being carried out.
The measured strains on the concrete and the deflection gave good data, provided that resistance strain
gauges and LVDT have their limitations in maximum frequency range and dynamic resolution.
However strain measurement on the rails was disturbed by the presence of high current short circuit
between rails and during high speed train passage. Strain signals denoted a lot of electrical noise (50Hz
and multiples). Solutions for this phenomenon could be the use of special guarded gauges or optical fiber
technology, to avoid electromagnetic interference.
It was suggested also to use more LVDT in the next measurement campaign, also in transversal direction.
Focus will be then on the quantification of transferred forces from high speed train to structure.

ACKNOWLEDGEMENT

The research has been carried out in the framework of the bilateral research project ’Traffic induced
vibrations in the built environment’ BIL98/09 with a financial contribution by the Ministry of the Flemish
Community.

REFERENCES

[1] Traffic induced vibrations in the built environment, Project proposal, internal document, May 1999
[2] ANSYS revision 5.5.1, Swanson Analysis System, 1999
[3] B. Peeters , G. De Roeck, Reference based stochastic subspace identification in civil engineering,
Inverse Problems in Engineering, V.8(1), pp.47-74, 2000
[4] B. Peeters, B. Van den Branden, G. De Roeck, Output-only modal analysis: a GUI for MATLAB,
2nd Benelux Matlab Usersconference, Brussels, Belgium, March 1999
[5] MATLAB revision 5.3, The Mathworks Inc., 1999