HIGH RAILWAY ROLLING NOISE REDUCTION BY COMBINING EXISTING TRACK BASED SOLUTIONS

Transcription

1 HIGH RAILWAY ROLLING NOISE REDUCTION BY COMBINING EXISTING TRACK BASED SOLUTIONS Patrick Vanhonacker 1, Konstantinos Vogiatzis 2 1 I-MOSS, Bertem, Belgium 2 University of Thessaly, Department of Civil Engineers, Laboratory of Transportation Environmental Acoustics, Greece The overall objective of the EU funded Quiet-Track project is to provide step changing track based noise mitigation systems and maintenance schemes for the railway rolling noise. This paper focuses on the combination of existing track based solutions to yield a global performance of at least 6 db(a). The analysis and validation was carried out considering a track section in the network of Attiko Metro line 1 with an existing outside concrete slab track (RHEDA track) where high airborne rolling noise was observed. The procedure for the selection of mitigation measures is based on numerical simulations, combining two software tools for noise prediction, WRNOISE and IMMI with experimental determination of the required track and vehicle parameters. The availability of a detailed rolling noise calculation procedure, which includes the modelling of the wheel/rail source intensity and of the noise propagation with ability to evaluate the effect of modifications at source level (e.g. grinding, rail dampers, wheel dampers, change in resiliency of wheels and/or rail fixation) and of modifications in the propagation path (absorption at the track base, noise barriers, screening) allows for detailed designing of measures and of ranking individual measures. A relevant combination of existing solutions was selected in function of the simulation results. Three distinct existing solutions were designed in detail aiming at a high rolling noise attenuation and not affecting the normal operation of the metro system: Action 1: implementation of sound absorbing precast elements (panel type) horizontally distributed in the track bed; Action 2: Implementation of an absorbing noise barrier with a height of m above rail level and close to the track and vehicle; Action 3: Installation of rail dampers at the selected track. The selected solutions were implemented on-site and the global performance was measured -step by step- for comparison with simulations. A very good agreement was found between simulations and measurements after installation and a global noise attenuation performance of 9 db(a) was obtained at the considered receiver location. 1. Introduction The general objective of Quiet-Track is to provide step changing track based noise mitigation systems and maintenance schemes, to provide reliable improved rolling noise calculation procedures with harmonized monitoring of the required input parameters and to provide track noise management tools, for use in noise mapping and hot spot action plans according to the European Noise Directive, for use as engineering tools and solutions in new railway projects and in refurbishment projects and for use by the track maintenance managers and maintenance industry. The existing rolling noise models will be enhanced with new fundamental features: the integration of the low frequency noise emission and of the actual wheel rail contact conditions for more accurate predictions of the noise emitted by the track. This is very important in an urban environment where low frequency noise penetrates much easier trough nearby windows than the higher frequency noise. Taking into account the actual wheel rail contact conditions is also very important (e.g. multiple point contact in curves) in order to enhance the accuracy of the rolling noise prediction models. The precision of the actual rolling noise models for tangent track, obtained by experienced users, is +/- 3 db(a) on the global noise level and +/- 6 db(a) on the individual octave bands in the averaged equivalent noise spectrum. The rolling noise in the low frequencies bands (below 250 Hz) cannot be calculated accurately with the existing models, based on statistical energy analysis. The aim is to obtain the same precision as above in curves and to obtain a global noise level precision of +/- 3 db(a) when only considering the low frequency bands (up to 250 Hz). The existing models use rail and wheel roughness values which are determined at a couple of selected locations or which come from general databases. On-board monitoring systems will be developed to make it be possible to use the real roughness values and track decay rate values measured directly at the track location where maintenance action is required or where mitigating solutions have to be applied. New track solutions will be developed within Quiet-Track which yield a noise reduction performance of at least 6 db(a) in comparison with the global rolling noise measured on a well maintained standard track in the network of the participating infra managers. A well maintained track is defined as track which is ground (standard rail defect removal grinding) at least once a year for urban networks and which is, for the conventional rail tracks, additionally tamped at least once a year for geometrical defect correction. The solutions have to be applicable to tram, LRT and metro tracks as well as to conventional tracks. 1/10

2 Four different development directions have been identified for solution development each with the objective to yield at least a performance of 6 db(a) noise reduction: Direction 1: combination of existing solutions. The individual performance of many existing solutions at the level of the track is not very good. In most cases (except for grinding) the performance is limited to max. 3 db(a) reduction. In Quiet-Track existing solutions will be combined to yield a global performance of at least 6 db(a). This work is reported in this paper. Direction 2: reduction of rail roughness growth rate. Solutions will be developed which are based on the concept of reduced rail roughness growth rate. These solutions have a far better potential for noise control than the ones based upon increase of the TDR (track damping) since these last ones only reduce the noise by 1-2 db(a) in average. Direction 3: low noise embedded rail. Elastically embedded rails are already in use for vibration reduction. This track type will be optimized for its noise mitigation performance whereby all noise sensitive parameters will be considered and linked to track design parameters such as stiffness and damping of embedment material, use and characteristics of the continuous rail pad. This optimization process will be validated in the network of De Lijn. Direction 4: low noise rail type and hardness selection. Rails are subjected to wear and this wear growth is strongly influenced by the selection of the rail type (and hardness) in function of the type of rolling stock. A procedure will be developed to selected the optimal rail type in terms of minimal rolling noise emission combined with optimal wear characteristics taking into account tangent track and curved track. A procedure for checking the economic viability of the solutions will be developed. This procedure will use recognized noise cost factors (from HEATCO study) to calculate the monetary benefit for comparison with the additional solution related investment cost and/or maintenance cost. This will result in holistic noise management plans for the introduction of the noise abatement solutions and for the noise related track maintenance. This paper reports the work done on combining existing solutions to yield a global noise reduction performance of at least 6 db(a) (direction 1 above). This is done by simulating the combined effect using the WRNoise software. A validation is performed in the network of Attiko Metro line 1, in Athens where an existing outside concrete slab track (RHEDA track) which exhibits high airborne noise is evaluated for noise reduction. The combination of existing solutions is installed and the global noise reduction performance is measured for comparison with simulations. Three types of mitigation measures were investigated, both individually and combined: Absorbing panels on the track: They work on the absorption of sound waves close the source and are thereby able to mitigate both contributions of the rails and of the wheels. Moreover, the application of such panels is confined to the track itself. Since they only influence sound waves that are normally reflected the expected achievable reduction is relatively low. Noise barriers next to the track: in addition to the above absorbing panels, noise barriers next to the track are considered. They prevent the sound propagation by directly reflecting and absorbing the sound waves and therefore have the potential to achieve a higher level of noise reduction. Since they act on the propagation path, they influence both the rail and wheel contribution. Rail dampers: in addition to the absorbing panels and noise barriers, rail dampers are considered. Their application is confined to the track itself. Unlike the other measures, they only influence the contribution of the rail. Their use should only be considered when the rail contribution is at least equally important as the wheel contribution. The procedure for the selection of mitigation measures is based on numerical simulations, combining two software tools for noise prediction, WRNOISE and IMMI. WRNOISE accurately represents the source model for noise generation. The noise propagation influencing measures, such as absorbing panels and noise barriers, are modelled using IM- MI. Since the rail dampers have a direct influence on the track vibration response (the track admittance and TDR), the software WRNOISE is used in combination with IMMI to investigate the efficiency of rail dampers. The noise reduction efficiency is computed by comparing the sound pressure level before and after the installation of each mitigation measure. The insertion loss factor IL in db, shows the difference between the sound pressure levels (SPL in db) in two cases: IL [db] = SPL after SPL before For each mitigation measure, the resulting sound pressure level is calculated in a point at 7.5 m rom the track center and a height of 1.2 m above the top of the rail. Since two distinct simulation software packages are used, it is important to ensure a good interface: all source modelling is performed in WRNOISE, while propagation is handled by IMMI. 2/10

3 db db 2. Pass-By Noise Measurements 2.1 Reference case without mitigation measures A full program of acoustic reference noise measurements with normal train operation, according to the standard EN ISO 3095:2005, was completed. All airborne acoustic measurements were conducted according to the above standard with the 1 st set of 2 microphones placed at 7.5 m distance from the center of the track and at a height of 1.2 m as well as the 2 nd set of 2 microphones placed at 25 m distance from the center of the track and at a height of 4 m as per the provisions of the recent 2002/49/CE directive (figure 1). The initial measurement campaign yields a 1 st series of acoustic measurements of the existing reference airborne noise with the RHEDA system. Figure 1: Measurements location & equipment according to EN ISO 3095:2005(E) 2.2 Absorbing panels on the track After the implementation of the sound absorbing precast elements ( SoundSorb panels) in the track bed over a length of approx. 110 m based (see figure 2), a 2 nd series of acoustic measurements was executed. It is important to underline that for both 1 st and 2 nd series, the same exact locations of measurements were used and same vehicles operating were measured in order to ensure similar and comparable conditions. Figure 2: Installation of horizontal precast SoundSorb panels The measurement results for both the initial scenario and action 1 implementation at both locations (ch1 at 7.5 m and ch2 at 25 m) are presented hereafter in figure 3. The results of the implementation of the "SoundSorb" absorbing material in precast panels indicates a significant noise attenuation of approximately 2.1±1.4 db(a) and 3.0±1.8 db(a) (as per the following figure) in the position of the 1 st (7.5 m) and the 2 nd (25 m) measurement locations respectively. The implementation of absorbing material between an adjacent to the rails in precast panels has proven to ensure a significant attenuation. Figure 3: Noise levels for 2 nd series (track panels) & 1 st series (reference) - Comparative results 3/10

4 db db db db 2.3 Noise Barriers In a second phase, an absorbing concrete barrier of an approx. length 110 m and a height of m, in the boundaries of the railway track (using also sound absorbing material "SoundSorb") was implemented (see figure 4). Figure 4: No noise barrier (left), noise barrier detail (middle) and noise barrier completed (right) at Quiet Track test site in ISAP line 1 With the implementation of this mitigation measure, a distinct 3 rd series of acoustic measurements were executed. The results of the implementation of absorbing barrier proven to be especially efficient with a significant attenuation compared with action 1 conditions (3 rd series vs 2 nd series of measurements) of approximately 5.8±-2 db(a) and 4.1±2.4 db(a) (as per figure 5) in the position of the 1 st (7.5 m) and the 2 nd (25 m) measurement locations respectively. Figure 5: Noise levels for 2 nd series & 3 th series (implementation of noise barriers) - Comparative results 2.4 Rail Dampers The installation of rail dampers in the selected 110m track was completed (see figure 6) and with the implementation of this 3 rd action, a specific 4 th series of acoustic measurements was executed. Figure 6: Rail dampers installed at test site The results of the implementation of rail dampers proven to be marginally and rather not conclusive positive efficient with an attenuation (4 th series vs 3 rd series of measurements) of approximately 1.7±-2.5 db(a) and 1.0±2.6 db(a) (as per figure 7) in the position of the 1 st (7.5 m) and the 2 nd (25 m) measurement locations respectively. Figure 7: Noise levels for 3 rd series & 4 th series (rail dampers) - Comparative results 4/10

5 2.5 Conclusion After completion of the installation of the final action, the results of all actions 1, 2 & 3 were compared with the initial RHEDA track noise behavior. The comparative results are presented in figure 8. The result of the implementation of all actions combined proves to be especially efficient with a noise attenuation compared with the initial RHEDA track of approx. 9.6±-2.1 db(a) at the 1st location (7.5 m from the track ). Figure 8: all actions vs initial RHEDA track measured noise levels in 1/3 octave bands 3. Rolling noise simulation With the experimental input data described above, it is possible to perform calculations using WRNOISE and IMMI to evaluate the efficiency of the mitigation measures. The procedure to interface WRNOISE and IMMI is the following: A simulation is performed in WRNOISE, based on a full experimental characterization of the track and the vehicle (the wheels). The agreement between simulation results and experimental pass-by measurements is verified. The comparison is based on a receiver point at 7.5 m from the track center and a height of 1.2 m above the railhead. Using IMMI, the acoustic source power is calculated that corresponds to the sound pressure level in the response point, calculated by WRNOISE. Using IMMI with the source power calculated in step 2, the influence of absorbing panels and noise barriers on the sound pressure level in the response point is calculated. Finally, a second calculation is performed in WRNOISE to determine the effect of the rail dampers, based on a second experimental characterization of the track, after installation of the rail dampers. 3.1 Reference case without mitigation measures The first step is the calculation of the rolling noise level with WRNOISE and the comparison with the measured noise level during pass-by. For the calculation, a vehicle speed of 60 km/h was assumed, which corresponds to the actual average pass-by speed. The necessary transformations have been performed to take into account the effect of track loading due to the train during pass-by, while the measurements are performed in unloaded condition. Figure 9 shows the comparison of the noise level predicted by WRNOISE and the measured noise level, obtained from the measurements campaign above, measured in a receiver point at 7.5 m from the track center and a height of 1.2 m above the railhead. Since the effect of absorbing panels and the noise barrier is predicted using IMMI, in the second step, the sound power is calculated in IMMI that corresponds to the sound pressure level in the response point, calculated by WRNOISE. In the figure 10, the result (sound pressure level at 7.5 m from the track center) obtained by WRNOISE and IMMI, after calibration of the source power in IMMI, is presented indicating a nearly perfect match between both results and consequently, a good interface between WRNOISE and IMMI was created. Figure 9: Comparison of the measured and predicted noise level in a receiver point at 7.5 m from the track center Figure 10: Comparison between the predicted sound pressure level in WRNOISE and IMMI, after calibration of the source power in IMMI 5/10

6 3.2 Absorbing panels on the track The absorbing panels are installed between and beside the rails along the track, as shown in figure 11 hereafter. The absorbing panels are made of SoundSorb material, which is a cement base, open-cellular material that can be added to the concrete. It has an acoustical value (NRC) of 0.7 to 1.0 according to ASTM C-423 test. The Noise Reduction Coefficient (NRC) is a scalar representation of the amount of sound energy absorbed upon striking a particular surface such that the NRC of 0 indicates a perfect reflection (rigid ground) and the NRC of 1 denotes to a perfect absorption. Figure 11: Scheme of the track and the absorbing panel configuration According to the technical datasheet of SoundSorb, a 3 thickness of this material can result in a NRC of 0.95 that shows this material acts as an almost perfect absorbing material. In the software IMMI, the track is considered as a line source and the absorbing panels are modelled by replacing the ground surface (along the track) with a new material with an absorbing coefficient of 0.95 instead of the ground, which is by default assumed as a rigid surface with an absorbing coefficient of zero. Figure 12 shows the global sound level in db(a) (between 100 and 8000 Hz) computed by IMMI in the field points around the track, before and after installation of the absorbing panels. Figure 13 shows a top view, at a height of 1.2 m. Figure 12: Global level in db(a), predicted by IMMI in the field points around the track; above, before installation of the absorbing panels and below, after installation of the absorbing panels Figure 13: Global level in db(a), predicted by IMMI in the field around the track before (above) the installation of the absorbing panels and after (below) the installation of the absorbing panels. The black point shows the location of microphone at a distance of 7.5 m from the track at 1.2 m above the ground. Figure 14 shows the sound pressure level in db(a) computed by IMMI in the field at a point at 7.5 m from the track center and at a height of 1.2 m above the top of the rail. An insertion loss of 1-2 db in a frequency range from 200 to 600 Hz is obtained. The overall level before installation is 78.1 db(a), while the overall level after installation is 76.0 db(a), resulting in an overall gain of 2.1 db(a). Figure 14: Sound pressure level (SPL) in db(a) predicted by IMMI in the field at a distance of 7.5 m from the track at 1.2 m above the top of rail. Results are presented before (reference) and after the installation of the absorbing panels. 6/10

7 3.3 Noise barriers IMMI software is used for modelling the effect of the noise barrier. The noise barrier is installed at a distance of 2.6 m from the track center between the track and the existing buildings nearby. The barrier is a concrete wall covered by the same absorbing material as used in the track. Figure 15 schematically shows the configuration. First, the noise reduction efficiency of the individual concrete wall is examined and then the effect of this measure together with the absorbing panel on the track is investigated. Figure 15: Scheme of the track, the absorbing panel and the noise barrier configuration Figure 16 shows the global sound level in db(a) (between 100 and 8000 Hz) computed by IMMI in the field points around the track. Results clearly display a significant improvement at points located at right side of the noise barrier. Figure 3.9 shows the top view at a height of 1.2 m; this figure clearly shows the influence of the length of the noise barrier. Figure 16: Global level in db(a) predicted by IMMI in the field around the track (above) before the installation of the noise barriers and (below) after the installation of the noise barriers Figure 17: Global level in db(a) predicted by IMMI in the field around the track before (above) the installation of the noise barriers and after (below) the installation of the noise barriers Figure 18 shows the sound pressure level in db(a) computed by IMMI in the field at a point at 7.5 m from the track centre and at a height of 1.2 m above the top of the rail. An insertion loss of up to 7 db is obtained. The overall level before installation is 78.1 db(a), while the overall level after installation is 70.7 db(a), resulting in an overall gain of 7.4 db(a). Figure 18: Sound pressure level (SPL) in db(a) predicted by IMMI in the field at a distance of 7.5 m from the track at 1.2 m above the top of rail. Results are presented before and after the installation of the noise barrier. 7/10

8 3.4 Rail dampers As a mitigation measure implemented at the source, rail dampers are designed to reduce the contribution of the rail in the overall rolling noise. Since the rolling noise is generated due to the dynamic interaction of the rail and the wheel, increased vibration attenuation with distance along the rail (the Track Decay Rate-TDR) reduces the average vibration level of the rail and as such reduces the rolling noise radiated by the rail. This section investigates the effect of a damping treatment on the rail vibration response and investigates how this may control and reduce the rolling noise. Therefore, the focus will be on the variation of TDR in the presence of the rail dampers. The TDR measured before and after the rail damper installation are compared and the performance of the rail dampers is investigated by comparing the sound pressure level computed before and after of the damper installation. The sound pressure level for each case is computed using the software WRNOISE Field measurements Two measurement campaigns were performed at the validation site. A first measurement campaign was performed before installation of the rail dampers, a second was performed after installation of the rail dampers. Each time, the track admittance as well as the track decay rate (TDR) function were measured. In the following, the rail without the rail damper is called an undamped rail and the rail with a rail damper (see figure 19) is called a damped rail. The track admittance was measured by exciting the railhead using an instrumented impact hammer and measuring the rail vibration response at the same point in vertical and horizontal direction using two accelerometers mounted on the railhead close to the excitation position as shown in figure 20. According to the direction of the excitation and the direction of the vibration measurement, vertical, horizontal and cross admittances are determined. Figure 19: Track with rail dampers Figure 20: Determination of the track admittance by the hammer impact test Figure 21 shows the track (rail) admittances in vertical (a) and horizontal (b) direction measured by means of a hammer impact test, for the undamped and for the damped rail. Figure 22 shows the vertical and the horizontal track decay rate in 1/3 octave band frequency obtained for the undamped and for the damped rail. (a) (a) (b) Figure 21: (a) Vertical and (b) horizontal track (rail) admittance functions measured by means of the hammer impact test for an undamped and a damped rail (b) Figure 22: Vertical (a) and horizontal (b) track decay rate in 1/3 octave band frequency measured for an undamped and a damped rail 8/10

9 3.4.2 Rolling noise simulation Since the installation of the rail dampers directly influence the source of the rolling noise by influencing the rail vibration, a simulation is performed in WRNOISE using the measured input parameters. Figure 23 shows the resulting sound pressure level in the response point at 7.5 m from the track center and 1.2 m above the railhead. Between 250 and 2000 Hz, where the rail contribution dominates the overall sound pressure level, a reduction up to 7 db is obtained. The overall level before installation is 78.1 db(a), while the overall level after installation is 76.3 db(a), resulting in an overall gain of 1.8 db(a). Using the insertion loss obtained using WRNOISE, a new simulation can be performed (1) to assess the effect at different locations, using the more advanced propagation model in IMMI and (2) to assess the combined effect of all mitigation measures. Figure 24 shows the global sound level in db(a) (between 200 and 4000 Hz) computed by IMMI in the field points around the track; an overall reduction of the sound pressure level is observed after installation of the rail dampers. Figure 23: Sound pressure level (SPL) in db(a) predicted by WRNOISE in the field at a distance of 7.5 m from the track at 1.2 m above the top of rail. Results are presented before (reference) and after the installation of the rail dampers. Figure 24: Global sound pressure level in db(a) predicted by IMMI in the field around the track (above) before the installation of the rail dampers and (below) after the installation of the rail dampers 3.5 Combined effect of mitigation measures Figure 25 shows the global sound level in db(a) (between 200 and 4000 Hz) computed by IMMI in the field points around the track; the effect of the absorbing panels can be observed in the vicinity of the track, while the effect of the rail dampers in an overall reduction of the sound pressure level. Finally, the effect of the noise barrier is observed at the right of the figure. Figure 25: Global level in db(a) predicted by IMMI in the field around the track (above) before installation of the mitigation measures and (below) after installation of the mitigation measures. 9/10

10 Figure 26 shows the sound pressure level in db(a) computed by IMMI in the field at a point at 7.5 m from the track centre and at a height of 1.2 m above the top of the rail. The effect of adding first the absorbing panels, afterwards the sound barrier and finally the rail dampers is illustrated. The overall level before installation is 78.1 db(a), while the overall level after installation of all three mitigation measures is 68.9 db(a), resulting in a predicted overall gain of 9.2 db(a). The figure also shows the measured sound pressure levels for all 4 cases, showing overall a good agreement between the simulated and the measured result. Table.1 summarizes the results for all cases. The efficiency of the noise barrier is slightly underestimated in the simulation, especially in the frequencies below 400 Hz. Figure 26: Sound pressure level (SPL) in db(a) predicted by WRNOISE in the field at a distance of 7.5 m from the track at 1.2 m above the top of rail. Results are presented before (reference) and after the installation of the mitigation measures. Table 1. Comparison of measured and simulated overall sound pressure levels MITIGATION MEASURES ACTIONS Overall sound pressure level [db(a)] Measured Simulated Reference Absorbing panels Absorbing panels + noise barrier Absorbing panels + noise barrier + rail dampers Conclusion This paper describes the use of numerical tools for the optimal combination of existing noise mitigation solutions. As a prototype test site the Athens Metro line 1 (former ISAP Line no 1) in Athens Metro network, where the system RHEDA is installed, has been selected. The procedure for the selection of distinct mitigation measures is based on numerical simulations, combining two software tools for noise prediction, WRNOISE and IMMI. The software IMMI is a field-proven software solution for environmental pollution mapping, integrating outdoors sound propagation (road traffic, railway, industrial and recreational noise), air dispersion modelling (gases, dust, ) and interfaces to CAD and GIS packages. The software WRNOISE is very efficient on the modelling the sound intensity at the source and has been validated by numerous experimental measurements. The selected noise mitigation measures, absorbing track panels- noise barriers- rail dampers, have been installed successively and their effect has been measured and compared with the numerical results. The reduction in overall sound pressure level after installation of all three mitigation measures is about 9 db(a). This significantly exceeds the reduction of 6 db(a) that was aimed. The predicted noise levels show a good agreement with the measured noise levels. ACKNOWLEDGMENT This work has been carried out as part of the QUIET-TRACK FP7 research project, funded by the European Union. 10/10