Scale model measurements of the acoustic performance of vented and absorptive road covers Isabelle Schmich, Jerôme Defrance, Gabriel Kirie, Pierre Bruder Centre Scientifique et Technique du Bâtiment (CSTB), 24, rue Joseph Fourier, F-38400 St Martin d Hères, France, {isabelle.schmich, j.defrance, bruder}@cstb.fr, gabriel_kirie@yahoo.fr An efficient solution for road traffic noise abatement is the use of vented (or louvered) absorptive covers. These devices are quite complex road noise protectors composed of rows of acoustic absorbent baffles located above the platform which is enclosed by two barriers, or at the top of a trench. In previous works 2D Boundary Element Method (BEM) calculations have been carried out on a series of parallel baffle covers with a constant geometry in the direction of the road lanes. In order to study 3D geometries of such covers, scale model measurements are done since the cost of 3D BEM simulations would be too high considering the range of studied frequencies. Measurements are performed using an MLS method at a scale of 1 to 20 in the frequency range 1000-20000 Hz (50-1000 Hz at full scale). Different types of vented road covers varying in baffle spacing and design have been tested. Results are presented in terms of acoustic performance due to the addition of each studied cover as a function of frequency and receiver position. 1 Introduction The present work on louvered and absorptive road covers as an efficient solution for road traffic noise abatement is complementary to a previous study [T1] which aims predict the acoustical performance of different types of louvered road covers. They consist of a range of acoustic absorbent baffles located above the platform which is enclosed by two barriers, or at the top of a trench. In our case (figure 1) they are placed parallel to the traffic. The top view (figure 2) shows a range of parallel lines. This case has also been treated in 1:20 scale model measurements. Figure 1: Louvered road cover. Cross-section Baffles Baffles Barrier Road sources Barrier Road sources Figure 2: Louvered road cover. Top view 2 2D Boundary Element Method (BEM) calculations The calculation code MICADO is a powerful model, developed by CSTB, and based on the Boundary Element Method [2,3]. It is currently the only theoretical approach that is able to give exact calculation results that can be used as a reference. As shown in figure 3 and 4, the following parameters and range of variation have been used: The number of lanes defines the total width of the protection L. The lanes have a width of 3.5 m, the hard shoulder for emergency use has a width of 2.5 m. The width of the central median has been adjusted so that L corresponds to an integer value of e. The total height of the protection H of the barriers that vary from 7 to 10 m. The impedance condition Ze of the lining of the barriers: the barriers are reflective with a width of 10 cm. It is possible to leave the internal side blank or to cover it on the upper ¾ with 10 cm of wood fibre concrete or 5 or 10 cm of glass-wool. The space e between two successive baffles: 0.5, 1 or 2 m The height h of the baffles: 0.5, 1 or 2 m The inclination of the baffles θ: from 0 to 20 in increments of 5 The orientation of the baffles: the baffles can be tilted parallel on each side of the road axis or symmetrically with respect to the road axis. In 1147
the last case, if there is an odd number of baffles the central baffle is left vertical. The impedance d of the lining of the baffles: the baffles are reflective and have a thickness of 5 cm. The cases where the baffles have their lateral sides covered with 5, 10 and 20 cm of glass-wool have also been treated Figure 3: Geometric parameters for the 2D calculations The sources that represent the vehicles are positioned on the ground. The calculations have been done for a flat and reflecting site which corresponds to a rectangle with a vertical height of 40 m and a width of 100 m. The map begins at 1 m from the exterior side of the barrier. The horizontal and vertical steps are in increments of 2 m. h R e Noise map θ Ze e L h S 100m Zd 40m H L(f i ) : relative noise level defined in NF EN 1793-3 [4] used for the road noise at the frequency f i Lw c (S j, f i ) : noise level in the covered case for the source S j at the frequency f i Lw n (S j, f i ) : noise level in the case without louvered cover for the source S j at the frequency f i To simplify the comparison, the calculations have been done for a reference case and each studied case varies only one parameter from this reference case. The reference case corresponds to a 6-lane road enclosed by two 7 m high barriers. The barriers are covered on the upper ¾ with 10 cm of porous wood cement. The baffles of the cover are vertical, spaced by 1 m and are 1 m high. They are covered on each side with 5 cm of glass-wool. 3 Some results of 2D Boundary Element Method (BEM) calculations Figure 5 shows the EA map for the reference case. The negative values show a diminution of the noise due to the road cover. The positive values show the area for which the noise level with road cover is superior to the noise level without road cover. Sources 2m 6m 4m 6m 2m Figure 4: Site description The Excess Attenuation (EA) in db(a) of the louvered cover for a road traffic noise has been calculated. It is written as: L( fi ) + Lwc ( S j, fi ) 10 10 i j EA = 10 log L( f ) + Lw ( S f ) (1) i n j, i 10 10 i j with: f i : set of frequencies from 50 to 2000 Hz in third octave bands S j : set of lanes (incoherent sources) Figure 5: EA map in db(a) for the reference case All calculated EA maps can be separated into two areas: An area where the height is lower than the height of the barrier. This area corresponds approximately to the ground floor and first floor of a house. The EA does not vary much except for the 20 m next to the infrastructure. It is the area where the influence of the louvered road cover is the least intense because the most effective attenuation is due to the barrier. So the absolute value of the EA is the smallest. 1148
An area where the height is larger than the height of the barrier. This area corresponds to all the upper floor levels of a house (beginning with the second floor). The EA has a radial distribution using the upper edge of the barrier closest to the calculation area as the axis. Generally there will be a zone between 20 and 45 relative to the horizontal line where the EA is the lowest which means that the louvered road cover has its maximum efficiency in this zone. The influence of the number of lanes can be summarised as: the larger the road, the more effective the louvered road cover. The calculation of the influence of the height of the barriers shows that the higher the barrier, the larger is the area of efficiency of the louvered road cover. The tests on the influence of the barrier coverage show that the more they are covered by absorbing material, the more they are effective. It should be emphasised that the influence of the absorbing material will only be effective for frequencies higher than 500 Hz. Because the louvered covers are efficient at high frequencies, it remains to investigate how to increase the excess attenuation at the lowest frequencies. The efficiency of a louvered cover increases with the size of the total surface of the baffles. For the same total surface the louvered cover is more effective with a larger space between the baffles and higher baffles as shown in figure 6 and 7 or 8 and 9. The more baffles there are, the more efficient it will be. The larger the inclination of the baffles the better is the performance. Figure 8: EA map in db(a) for 0.5 m high baffles with a spacing of 1 m between them Figure 9: EA map in db(a) for 1 m high baffles with a spacing of 2 m between them The study of the influence of the type of baffle cover shows that it is an important parameter for the optimisation of louvered road covers. It applies to all calculation areas, but it is only effective for high frequencies. Figure 6: EA map in db(a) for 2 m high baffles with a spacing of 1 m between them Figure 10: EA map in db(a) for reflecting 1 m high baffles with a spacing of 1 m Figure 7: EA map in db(a) for 1 m high baffles with a spacing of 0.5 m between them 1149
EA (db) Figure 11: EA map in db(a) for absorbing (10 cm of glass wool), 1 m high baffles with a spacing of 1 m 4 Measurement preliminary work 1:20 scale model measurements of the louvered road covers shown in figure 1 have been realized and compared to the 2D BEM calculations results. Figure 14: Experimental results compared with BEM 2D calculations For the absorbing material it was necessary to find a material that best represents mineral wool at a scale of 1:20. It should have the same acoustic impedance for the two different scales. Tests have been made for Ardennes baize (felt). The validation experiments have been done in the two configurations shown in figure 15 and 16. Figure 15 shows the perpendicular incidence of the source and figure 16 the grazing incidence of the source. Source Tweeter Axis Figure 12: Acoustic source: tweeter mounted on a rotating axis For the material used in the model we have chosen high-impact polystyrene as the reflecting material so that the mass law of the real size reflecting material is respected. The experimental set-up for validation of our choice is shown in figure 13. Figure 14 shows the experimental results compared with BEM 2D calculations for validation. 4.5cm Source 1 m 1 m 2 mm 15 cm Figure 13: Validation experimental set-up 7.5 cm 2 m 10 cm Tested material Figure 15: Validation set-up for the absorbing material with perpendicular incidence of the source 3 cm Source Tested material 1 m Figure 16: Validation set-up for the absorbing material with grazing incidence of the source 2 cm 2.5 cm Figures 17 to 20 show the comparison between the 2D BEM calculations results for 5 and 10 cm of mineral wool and the 1:20 scale model measurements results with 2 and 4 mm of Ardennes baize. The variations are similar between the 2D BEM calculations and the measurements. The interference positions are reasonably reproduced. 1150
EA (db) EA (db) Figure 17: Comparison of the EA between the 2D geometry of figure 15. The calculation parameter was 5 cm of mineral wool and the measurement parameter was 2 mm of Ardennes baize Figure 18: Comparison of the EA between the 2D geometry of figure 16. The calculation parameter was 5 cm of mineral wool and the measurement parameter was 2 mm of Ardennes baize EA (db) Figure 20: Comparison of the EA between the 2D geometry of figure 16. The calculation parameter was 10 cm of mineral wool and the measurement parameter was 4 mm of Ardennes baize 5 Scale model measurements The louvered road cover shown in figure 1 has been constructed and measured. The baffles are of rigid framing with 5 cm of mineral wool on each side. They are represented by baffles of high-impact polystyrene with 2 mm of Ardennes baize on each side. The barriers are represented by 5 mm of high-impact polystyrene covered with 4 mm of Ardennes baize corresponding to 10 cm of mineral wool (figure 21). The source is placed in the middle of the road at a height of 45 mm. The receiver is placed 1 m from the barrier at a height of 30 cm. EA (db) Figure 19: Comparison of the EA between the 2D geometry of figure 15. The calculation parameter was 10 cm of mineral wool and the measurement parameter was 4 mm of Ardennes baize Figure 21: 1:20 Scale model for the louvered road cover The comparison of the EA in narrow bands obtained from 2D BEM calculations and the measurements is shown in figure 22. It is difficult to quantify the correlation, so it has been presented in 1/3 rd octave 1151
bands in figure 23. The correlation is quite good up to 630 Hz where the gap between both approaches increases. EA Figure 22: Comparison of the EA in narrow bands between the 2D BEM calculations and measurements in the case of the geometry of figure 15 The geometry of the model is not perfect. For example the baffles are sometimes warped and are no longer parallel. It could be possible to use thicker baffles or some other material such as metal which still have the same necessary properties for the scale model. The choice of material for modelling mineral wool at the scale 1:20 is quite sensitive and there already exists some differences between calculations and measurements for only one reflection. These differences can be added in the case of multireflections. A better choice of the baize or the division of the measurements in different parts (a different material for each frequency range) could improve it. A last point is that the measurement environment is not infinite in the 3 rd dimension, so that in our case all acoustic energy that reaches the receiver 15.9 ms after the first acoustic wavefront detected by the microphone is not taken into account. This can lower the quality of the measurements. Acknowledgements The authors are grateful to Emmanuel Thibier and to the French Institution ADEME (Agence de l Environnement et de la Maîtrise de l Energie) for its financial support. References Figure 23: Comparison of the EA in 1/3 octave bands between the 2D BEM calculations and measurements in the case of the geometry of figure 15 6 Conclusion The results are quiet satisfactory even if the correlation between calculations and measurements increases after 630 Hz. This discrepancy can be explained by the following reasons: The source is theoretically an omnidirectional point source which is not the case of the used tweeter. It can also be considered that the tweeter is an obstruction for the reflecting rays on the ground, which can bring some errors in the measurement results. To solve the problem the use of an electrostatic sparker which gives the possibility to do free field measurements seems appropriate [5,6]. [1] J. Defrance, J. Roland, M. Perraudeau, M. Baulac, Characterisation, performance and conception of vented and absortive road covers. Internoise 2004, Prague. [2] P. Jean, A variational approach for the study of outdoor sound propagation and application to railway noise. Journal of Sound and Vibration; 212(2):275-294 (1998) [3] P. Jean, J. Defrance, Y. Gabillet. The importance of source type on the assessment of noise barriers. Journal of Sound and Vibration; 226(2):201-16 (1999) [4] Europeen Norm EN 1793-3: Road traffic noise reducing devices Test method for determining the acoustic performance part 3 : Normalized traffic noise spectrum, 1997. [5] M. Wayne, W. Wright, N. Menendrop. Acoustic radiation from a finite line source with N wave excitation. Journal of the Acoustic Society of America, 43 966-971, (1968). [6] R.E. Klinkowstein. A study of acoustic radiation from an electrical spark discharge in air. Report of Acoustics and Vibration Laboratory Massachusetts Institute of technology, July 1974. 1152