Reduction of noise of railway steel bridges with tuned absorbers on

CFA/DAGA'04, Strasbourg, 22-25/03/2004
Reduction of noise of railway steel bridges
with tuned absorbers on rail
Franck POISSON1, Olivier COSTE2
SNCF - Direction de la Recherche, 45 rue de Londres, 75379 PARIS cedex 08, email : [email protected]
2
SIGNAL DEVELOPPEMENT, 17 rue Albin HALLER, 86000 POITIERS, email : [email protected]
25
INTRODUCTION
VB2N & 17000 - 75 km/h
VB2N & 17000 - 65 km/h
20
Most of railway steel bridges produce much more noise
than a ballasted track. Located in populated area, they may
generate annoyance for the population. Specific noise
reduction solutions must be developed. A research
program has been started by the French railway company
SNCF for Réseau Ferré de France to deal with this
problem.
The aim of this project consists in develop a methodology
based on measurements and/or simulations to choose the
best solution to reduce the noise of a given steel bridge.
Z6100 - 65 km/h
Z20500 - 80 km/h
15
dB
Z20500 - 50 km/h
10
5
0
10k
4k
6.3k
2.5k
1k
1.6k
630
400
250
160
63
100
25
-5
40
1
Third octave bands (Hertz)
Figure 3: spectrum of the sound pressure level increase
due to the train pass by on the test bridge GAVIGNOT
A GLOBAL APPROACH
The first step of the study concerns the selection of a test
bridge. The Gavignot bridge in Enghien les Bains (see
figure 1) is representative of the bridges of the French
railway network and its structure is one of the noisiest
[1,2].
The SPL rises up around 40 Hz and 400 Hz. The sound
pressure level increase in dB(A) is mainly due to the high
frequency energy but the lower part generates annoyance
too, by causing the windows to vibrate.
Then, the simulation of the vibroacoustic behavior of the
bridge must be conducted on the whole frequency range.
The finite element method is used below 200 Hz and the
statistical energy analysis (SEA) in the higher frequency
band.
The low frequency problem is presented in [3]. The deck
plate radiates the main part of the acoustic energy around
40 Hz. Solutions exist and will be tested on the Gavignot
bridge in 2004.
Figure 1: steel bridge GAVIGNOT selected for the study
THE HIGH FREQUENCY
PROBLEM
The rail fastening system is representative: rail is
supported by wood sleepers (or longitudinal sleepers)
directly fastened on the steel deck plate (see figure 2).
Up to 200 Hz, the SEA is used to predict the noise radiated
by the bridge and identify parts of the structure which
radiate the much more noise. The model is presented
figure 4.
Bras verticaux
d’extrémité
Bras verticaux
intermédiaires
Tôles de
Platelage
Figure 2: rail supported by sleepers or longitudinal
sleepers
Partie haute de la
Petite Poutre
Partie basse de la
Grande Poutre
tôle basse de la
Petite Poutre
As a first step of the study, a measurement campaign is
carried out to localize on the whole frequency band the
sound pressure level (SPL) increase due to the bridge.
Then, the noise radiated by several passengers trains is
measured in front of the bridge and few hundred meters far
from the bridge on a ballasted track. Differences between
SPL are presented on figure 3.
Partie haute de la
Grande Poutre
Croisillons
Console
Tôle haute de
Grande Poutre
Rambarde
Entretoises
Longerons
Passerelle
Figure 4: SEA model of the GAVIGNOT bridge
The bridge is built with plates and L corner beams. It is 18
meters long and 4 meters wide. Vibration and acoustic
967
CFA/DAGA'04, Strasbourg, 22-25/03/2004
In this case, the objective is to maintain on the whole
frequency band a decay rates around 3 dB/m, mainly
around 1KHz to reduce the SPL in dB(A).
Examples of tuned absorbers for ballasted track are
presented figure 7. Figure 8 shows how the decay rates are
modified.
measurements have been used to optimise sub-systems
parameters and simplify the model. The main parts of the
structure used in the simplified model to simulate the passby noise radiation are the deck plate, the two longerons
(longitudinal beams) and the cross beam.
The noise spectrum computed by the model is presented
figure 5.
110
Mesure
Bruit du pont seul
Bruit du pont avec une réflexion sur la route
Bruit du pont avec une réflexion sur la route et bruit du rail
100
Bruit du rail seul
90
dB (A)
80
70
Figure 7: CORUS (left) and SOCITEC (right) tuned
absorbers
60
50
40
DECAY RATES (Vertical)
30
10000
100
DR (dB/m)
Figure 5: sound pressure level prediction for a Corail pass
by (high frequency)
The noise predicted by the SEA approach (green line) is
lower than the measured one (red line). Some
measurements have been carried out to estimate the
contribution of the site amplification (orange line). The
rolling stock noise is much more difficult to estimate but
the measurement of vibratory level of the rail during a
train pass-by shows that the rail behavior is specific on the
bridge. The decay rates of the energy in the rail on the
bridge confirm this result. They are much lower than on a
ballasted track (see figure 6).
10
63
Frequency (Hz)
Figure 8: decay rates measured on a ballasted track
without (blue)/with(red) tuned absorbers.
Using the Track Wheel Interaction Noise Software
(TWINS), the impact of the decay rates modification on
the noise radiated by the system is estimated. Due to the
poor decay rates on the steel bridge, the expected
efficiency of tuned absorbers is around 6 dB(A), that is to
say, 1 or 2 dB(A) much more than on a ballasted track.
dB/m
9
hors pont
7
pont
CONCLUSION
6
5
To confirm simulation results, tuned absorbers will be
added to the rail of the Gavignot bridge in april 2004. SPL
reduction and decay rates will be measured.
If its efficiency is validated, this solution may offer a good
alternative to the noise barrier solution wich is around 2,5
times much more expensive.
In the same time, the SNCF homologates two tuned
absorbers for ballasted track to complete its noise
reduction solutions panel.
4
3
2
1
0
100
1
0,1
Pente de régression
8
10
63
0
80
0
10
00
12
50
16
00
20
00
25
00
31
50
40
00
50
00
1000Frequency (Hz)
80
10
0
12
5
16
0
20
0
25
0
31
5
40
0
50
0
100
1000
10000 Hertz
Figure 6: decay rates on the test bridge and on the track
If the rail is taken into account in the SEA model (figure 5,
blue lines), it becomes the major source between 600 Hz
and 1000 Hz which is responsible of the dB(A) sound
pressure level. This result leads to dedicated noise
reduction solutions like rail dampers.
REFERENCES
[1] P. DINGS, “Measures for noise reduction on steel
railway bridges”, Inter-Noise Budapest, 1997.
[2] J. G. WALKER, N. S. FERGUSON, M. G. SMITH,
“An investigation of noise from trains on bridges”, JSV,
193(1), 1996.
[3] F. POISSON, L. DIELEMAN, “Noise of railway steel
bridges: a complete study”, WCRR, 2003.
[4] L. GIRARDI, F. FOY, M. FUMEY, “Mise en voie
d’absorbeurs dynamiques sur le rail pour la réduction du
bruit de roulement”, Revue Générale des Chemins de Fer,
octobre 2003.
THE TUNED ABSORBERS
SOLUTION
Decay rates of the vibratory energy in the rail which are
directly linked to the radiated noise are very low on the
whole frequency band. Tuned absorbers which consist in
adding a mass spring system to the rail offers the
possibility to enhance the decay rates [4].
968