Edinburgh, Scotland EURONOISE 9 October -8 Noise impact of innovative barriers dedicated to freight trains in urban areas Marine Baulac a Jérôme Defrance Philippe Jean Paris Est, CSTB, rue Joseph Fourier, 38 Saint-Martin-d Hères, France Florence Margiocchi Franck Poisson SNCF, rue de Londres, 7379 Paris Cedex 8, France ABSTRACT The development of freight trains is a promising way to sensibly reduce environmental air pollution impact due to lorries in living areas. The modal transfer of goods transportation from road to rail is a sustainable solution endorsed by most of European countries. On the other hand the noise impact of freight trains on residential and recreational areas should be carefully controlled in order not to lose the gain obtained in air quality. In this paper we present a parametric study of theoretical acoustic performance of innovative noise barriers for two types of urban situations: a very dense one for which a set of low height abatement systems are studied, and an open one with a little dwellings density for which a set of multiple-edge barriers are tested. In each of the two cases, the side clearance of the railways as well as train body shape and specific railway noise sources are taken into account in the D Boundary Element Method used here for simulations. The results are given in terms of global acoustic efficiency for different receiver locations. Their analysis shows the potential gain that can be expected compared to classical straight barriers configurations. 1. INTRODUCTION This paper deals with the -years project IMPACT (complex noise protections impact on goods transportation by railway in urban areas) including researchers from CSTB and SNCF. This project is funded by ADEME (French Environment and Energy Management Agency). The main objective of this project is to offer optimized solutions dedicated to reduce noise from rail freight transport. Two kinds of noise protections are considered (as shown in Fig. 1): low-height noise barriers for dense urban areas and multiple edge noise
protections for more rural areas. The first part consists of numerical calculations of the efficiency of the protections and the second one of in-situ measurements on prototypes. Traditional noise barriers are three or four meters high. In order to increase their efficiency, some devices are considered on the top of the barriers. For example, multiple edge barriers are straight barriers with one or several panels on one or the two sides of the central barrier. This kind of barrier has already been studied 1 ; the calculations have shown that a significant efficiency can be obtained. In very dense urban areas, conventional noise protection cannot be considered for many reasons (aesthetic aspects, building constraints ). Low height noise protections as studied previously can be proposed, especially for guided transports since the nearer of the source is the protection, the best is the efficiency. Figure 1: Cross section of: (a) low height noise protections (for dense urban areas), (b) Multiple edge noise barrier (for peri-urban areas).. NUMERICAL APPROACH This part deals with numerical calculations on the two kinds of noise protections. The first step, presented here, is a parametric study of the problematic realized with an advanced numerical code named MICADO, based on the boundary element method (see next paragraph). Another step in progress is a more systematic research 3. The purpose is to use an optimization method (Genetic Algorithms and Nelder Mead algorithm coupled) to directly determine the optimal forms and impedances and thus to build a noise barrier with a maximum efficiency. The function to be optimized is the average pressure field in a zone to be protected and the optimization variables are the shape and the surface impedance of the acoustical protection. A. Description of the numerical code MICADO The Boundary Element Method (BEM) which relies on the Integral Equation theory has been developed in the s and has been since extensively used. Two families of boundary element methods can be distinguished: direct and indirect formulations. The direct formulation relies on the use of the Helmholtz integral equation where the unknown functions are pressure and velocity, while the indirect one is based on an integral formulation assuming that the sound field scattered by a boundary can be represented by a linear combination of a distribution of monopoles (a simple-layer potential) and a distribution of dipoles (a double-layer potential). References -8 are suggested for more details on the method. In this work, the numerical simulations of outdoor sound propagation have been carried out with MICADO. It is a D, D½ and 3D BEM numerical code based on the direct formulation. It has been developed with a variational approach by Jean at CSTB. In the
D version used here, the geometry of the problem is bi-dimensional: the source is an infinite linear coherent source and all the obstacles remain unchanged and infinite along a direction perpendicular to the vertical section plane (as shown in Fig. ). The BEM is a powerful tool in acoustical predictions for complex topologies and geometries in a homogeneous atmosphere. In this work, the meteorological effects can be neglected since the distances considered are shorter than one hundred meters. This method can be very time consuming, a compromise has to be found between accurate results and reasonable calculation times. Preliminary calculations have shown that parameters for a good convergence of results are: frequencies per third octave band and elements per segment and wave length. Figure : Geometrical configuration for D-BEM calculations. B. Details on configurations for numerical simulations Here are given some details about numerical calculations, definition of the efficiency and variable parameters. B1. Acoustical efficiency of noise protections The output of the numerical code MICADO is here the efficiency of the acoustical protection that is to say the difference of noise levels in given receiver points between the configuration with the studied noise protection and a reference configuration. The efficiency is defined by: = (1) where is the noise level, in db(a) in the configuration with the studied noise protection and is the noise level, in db(a) in the reference configuration. For the calculations of noise levels, a typical freight train power spectrum running at 3 km/h has been employed. B. General configuration and variable parameters for the calculations The general configuration is presented in Fig. 3. As the main noise source of freight train is the rolling noise, four point sources are defined: two for the rail noise and two for the wheel noise. Specific noise spectrums are considered for those sources, for several speeds but in this paper, only the 3 km/h case is taken for calculations, see Fig.. The parametric study concerns many variable parameters: Source power level (train speed, addition of dynamic absorbers on the rail, ) Distance between the noise protection and the centre of the line Geometry of the train Consideration of buildings in the cross section or not Geometry of the protections (global shape, global height, absorbing materials, added device size, )
In this paper, only a few of the numerous results obtained with the parametric study are presented in the next section. Hauteur (m) 3 1 - Largeur (m) Figure 3: General configuration for the parametric study. 7 7 3 km/h, rail 3 km/h, w heel Lw (db) 3 1 1 Frquency (Hz) Figure : Noise spectrums. Rail noise: solid line, wheel noise: dashed line. C. Results Here are presented a few results of the parametric study. Results are given as efficiency maps in db(a) and details on the reference configuration and the studied configuration are specified. Fig. shows the efficiency of a low height noise protections compared with a reference configuration with no protection. The low height noise protection brings an efficiency of to 1 db(a) in a large area compared to a configuration with no protection. Fig. shows the impact of the distance between the centre of the line and the protection (low height noise barrier). When the protection is located 1 m nearer the noise sources, the efficiency increases by db(a) in a large area. Fig. 7 shows the impact of the shape of the protection (multiple edge noise protection). The multiple edge noise protection proposed is more than db(a) more efficient than the straight noise protection in a large area.
studied configuration - 1 1 studied configuration - reference - 1 1-1 1 3 3 1 1 Figure : Efficiency of a low height noise protection, in db(a). reference -! " #! $ "% &&'!! " " (" 1 1 1 1 "! " #!! )!* 1 1 1-1 1 3 3 1 1 Figure : Influence of the distance between the line and the protection. studied configuration 1 reference 1 1 +"! " '! " #!%,- &&'! +,&!! " 1 1 1-1 1 3 3 Figure 7: Influence of the shape of the multiple edge protection. 1
3. EXPERIMENTS After studying and simulating the efficiency of the protections, the second part consists of in-situ measurements on prototypes. Two kinds of noise protections are considered: low-height noise barriers for dense urban areas and multiple edge noise protections for more rural areas. SNCF with the support of Réseau Ferré de France (the Infrastructure owner) takes in charge this part of in-situ applications. A. Railway constraints of exploitation One of the main objectives of the project is to propose solutions dedicated to freight transport. That is why, we have chosen a dense urban area (near Paris) with freight traffic for low-height noise barriers and a more rural area (Rhône Alpes area) with freight traffic too for multiple edge noise barrier. In each of the two cases, the side clearance of the railways, as well as train body shape and railway constraints of exploitation and of security have to be taken into account. Thus, before measuring in-situ, the site selection is quite difficult. Due to these railways operating constraints, the implementation of low-height noise barriers solutions would be particularly difficult and sensitive. To be as most efficient as possible, the low-height noise barriers have to be installed as close as possible to the rail. In order to take into account the danger area it is quite important to establish the best compromise between this danger area and the relevant efficiency of the noise barrier. A danger area is the area in which personnel, tooling or the equipment being handled may be struck by rail traffic or may be endangered by the blast effect. This area extends laterally, in relation to each track, over a distance of a few metres measured from the outside edge of the nearest rail (limit easily identified by the exposed part of the catenary s supports on the track side). This limited distance depends on circulating conditions of the line (type of traffic, speed, urban constraints ). An engineering study will be performed for each urban areas identified, in this part of the project. B. Measurements After numerical simulations performed in the first part, the measurements on prototypes in test site are planed in order to compare the measured efficiency with the calculated efficiency and to validate the numerical approach. At the end of the project, it is expected that these prototypes will be installed in a durable way and thus bring solutions to two black noise spots sites identified. The two optimised prototypes will be characterized in experiments in situ, in two different ways: SNCF will perform acoustic measurements with commercial trains: behind barriers and in free field, with various heights, The objective is to determine the average attenuation brought by the device and to analyze for each waveband the gain obtained compared to a conventional straight and reflective barrier. Moreover CSTB will measure the efficiency of reflexion and transmission according to standards EN 1793- and -. The main aim of this measurement campaign is to validate the acoustic performances of the barriers and to obtain an acoustic characterization of these new products.
These experimental studies will complete the numerical approach and will lay the emphasis on the expected potential gain in comparison with more classical noise barriers configurations.. CONCLUSION As a conclusion, the parametric study has shown a significant efficiency of both kinds of protections. Work is still in progress for the numerical part and the experimental part. Numerical optimization calculations are still running in order to obtain optimal barriers for given configurations. For the experimental part, work is also in progress to find test sites to build prototypes with a design that would be determined by the optimization part. REFERENCES 1. M. Baulac, J. Defrance, P. Jean, Optimization of multiple edge barriers with genetic algorithms coupled with a Nelder-Mead local search, Journal of Sound and Vibration 3(1), 7, pp.71-87.. M. Baulac, J. Defrance, P. Jean, F. Minard, Efficiency of Noise Protections in Urban Areas: Predictions and Scale Model Measurements, Acta Acustica 9(),, pp.3-39. 3. M. Baulac, Optimisation des protections anti-bruit routières de forme complexe, PhD Thesis, Université du Maine, Le Mans, France,. R.D. Ciskowski, C.A. Brebbia, Boundary Element Methods in Acoustics, London, Elsevier Applied Science, 1991.. S.N. Chandler-Wilde, D.C. Hothersall, Efficient calculation of the green function for acoustic propagation above a homogeneous impedance plane, Journal of Sound and Vibration 18(), 199, pp.7-7.. P. Jean, A variational approach for the study of outdoor sound propagation and application to railway noise, Journal of Sound and Vibration 1(), 1998, pp.7-9. 7. P. Jean and Y. Gabillet, A boundary element method program to study D noise barriers with ground effects, in Proceedings of Euronoise 199, Lyon, France, 199.. R. Seznec. Diffraction of sound around barriers: Use of the boundary elements technique, Journal of Sound and Vibration 73(), 198, pp.19-9.