Design of diffusive surfaces for improving sound quality of underground stations

Toronto, Canada International Symposium on Room Acoustics 213 June 9-11 ISRA 213 Design of diffusive surfaces for improving sound quality of underground stations Yong Hee Kim (yh.kim@aist.go.jp) Yoshiharu Soeta (y.soeta@aist.go.jp) Living Informatics Research Group, Health Research Institute National Institute of Advanced Industrial Science and Technology (AIST) 1-8-31 Midorigaoka Osaka 563-8577, Japan ABSTRACT This study investigated the effective designs of diffusive surfaces on wall and ceiling in underground station platforms for improving sound quality. For this purpose, two underground stations were constructed with island and side platforms using computer simulation. Scattering characteristics of various diffusive surfaces such as hemisphere, ribbed, box, Schroeder diffuser were surveyed in advance. Diffusers were applied to wall and ceiling surfaces of the simulation model in longitudinal and cross-sectional arrangements. Acoustic parameters such as speech intelligibility, sound pressure level, reverberation time and Interaural cross-correlation coefficient were derived. From the simulation results, contribution of diffusers on the improvement of sound field characteristics will be discussed according to diffuser location and platform style. In addition, the results of scale model tests will be discussed with the simulation results. 1 INTRODUCTION Architectural treatments such as installation of suspended absorbers for reducing reverberation have been attempted to improve sound qualities in public spaces 1-2. Among the public spaces, underground stations have poor acoustic conditions due to its reflective and long enclosure 3-4. Previously, the acoustic effects architectural design elements such as absorptive materials and station dimension were investigated in underground stations 5-8. Especially, Kang reported that diffuser application could be effective for improving speech intelligibility of long enclosures through scale model testing 3. Recently, it was found that more diffused sound fields could contribute to improve speech intelligibility in case of aged people with highly reverberant station by the authors 9. Since effective designed diffuser walls can contribute to improve sound fields in ordinary spaces 1 and diffusive boundary conditions along the length could affect sound fields in long enclosures 3, it is required to find optimum diffusion conditions or suitable diffuser shapes for underground stations. For this purpose, both computer simulation and scale modelling have been widely employed to evaluate architectural elements in underground stations 3,5-8,11. Therefore, this study aimed to find out effective shape and arrangements of diffusers on lateral wall and ceiling surfaces in underground stations for improving sound field characteristics using both computer simulation and scale modelling. Since sound field characteristics can vary according to platform style 4, two typical platform styles of island and side types were considered for the investigation. In addition to the various diffuser applications, effectiveness of tilted lateral walls was also investigated. 1

2 COMPUTER SIMULATIONS 2.1 Models of Underground Stations For the investigation, simplified underground station models were built based on the acoustic fitting process in the previous study 8,12. Stations A and C were based on the actual subway station with island and side platforms, respectively. Additionally, station B was considered to match cross-section area of station C to station A. Cross-section area was 61 m 2 for stations A and B, and 68.2 m 2 for station C. Floor and section plans of the simulation models are shown in Figures 1 and 2. All platform models were connected to rectangular sectional tunnels (length 23 m). Platform length was 15 m with corridor height of 3 m. Width of single track was 3.6 m. Distance between track and platform floors was 1.1 m. Station height was 5.3 m. 2 4 6 8 1 12 14 16 18 2 22 metres 1 metres 8 (a) 6 Figure 1: Floor plans. (a) Stations A (island platform), Station B and C (side platform). 4 2 (a) Figure 2: Section plans. (a) Stations A (island platform), Station B and C (side platform). 2.2 Simulation Settings Ray-tracing software (B&K Odeon 11.23) was employed to derive acoustic parameters and binaural impulse responses. Figure 3 shows source and receiver positions in the simulation models. Sound source (height: 2.8 m) was located 22.5 m away from the rear wall of platform in the longitudinal direction. In total, 14 receivers (height: 1.6 m) were placed on either side from the sound source in the longitudinal direction. Distance between receivers was 2.5 m. In stations A and B, source and receivers were placed as the same manner with station C. Odeon 1985-212 Licensed to: National Inst. Advanced Science & Tech., Japan 6 7 1 8 2 9 3 1 metres4 5 6 7 8 9 1 metres 2 metres 1 Track Track r16 14 r15 13 r 12 14 r 11 13 r 1 12 r 9 11 r1 8 P3 S r8 7 r7 6 r6 5 r5 4 r4 3 r3 2 r2 1 up Tunnel Tunnel (a) r r r r r16 14 r15 13 12 14 11 13 1 12 9 11 r1 8 P3 S 8r 7 7r 6 6r 5 5r 4 4r 3 3r 2 2r 1 Tunnel up Figure 3: Source and receiver positions (S: source, r x : receivers). (a) Stations A (island platform), Station B and C (side platform). Odeon 1985-212 Licensed to: National Inst. Advanced Science & Tech., Japan Acoustic properties of the finishing materials were referred to the previous study 8. There were no differences on assignments of finishing materials between station models. In platform area, floor material was granite tile with low scattering coefficient (sc) of.1. Ceiling was metal sheet with air backing with sc of.3. Wall was covered by glossy metal panel with sc of.1. In tunnel area, floor was concrete track bed with rail hardware. Other surfaces in tunnel were finished as concrete. Tunnel end was treated as high absorption. In the simulation, passengers, background noise and train were not considered. 2 Track Track up

RT [s] In order to quantify sound field characteristics, four acoustic parameters were employed based on IEC 6268-16 13 and ISO 3382-1 14 : speech transmission index (STI), sound pressure level (SPL), reverberation time (RT) and interaural cross-correlation coefficient (IACC). SPL and RT were averaged from.5 to 1 khz. A-weighted filter was applied to calculate IACC. All results derived from all 14 receivers are averaged to compare simulation cases as a single value. In order to emphasize the effects of diffuser, acoustic parameters were also calculated as relative values with reference to the current condition without any diffusers. Thus, for example, ΔSPL means SPL of the current condition without diffusers subtracted from SPL of each simulation case with diffusers. Determination of acoustically-fitted condition was made in terms of RT. Difference of RT between the field measurement 12 and the simulation for station A was less than JND values of RT (Relative 5%) 14 for each frequency band from 125 Hz to 4 khz. 2.3 Sound Fields Distribution according to Platform Style Figure 4 shows the results of acoustic parameters as a function of source-receiver distance. Sound field characteristics of station A with island platform were compared with stations B and C with side platform. As shown in Figure 4(a), side platform showed longer RT than island platform due to reflective track spaces located middle of platform. As shown in Figure 4, the direction toward tunnel showed higher IACC than the direction toward platform center due to strong and reflective reflections from platform rear wall. As results, platform style seems to be more dominant than room volume since station B showed similar characteristics with station C. Therefore, major investigations were made in station A and C. Platform center direction Tunnel Platform center direction Tunnel Island (A) Side (B) Side (C) Island (A) Side (B) Side (C) (a) Source-receiver distance [m] Figure 4: (a) RT and IACC as a function of source-receiver distance in stations A to C. 2.4 Experimental Configurations 2.4.1 Variation of scattering coefficients Effects of diffusive wall and ceiling on sound field characteristics were simply investigated by changing sc of flat surfaces of lateral wall and ceiling. In station A, the treated surface area of lateral wall and ceiling was 9 and 94 m 2, respectively. However, in station C, the treated surface area was only 48 m 2 for lateral wall and 786 m 2 for ceiling because the service facilities such as elevator were located near the lateral walls. In the experiment, sc of.1,.3 and.5 were applied to both wall and ceiling. Thus, nine cases in total were considered including the current situation (sc Wall =.1, sc Ceiling =.3). 2.4.2 Diffusers on lateral walls Source-receiver distance [m] Table 1 described four diffuser profiles used: hemisphere diffuser, ribbed diffuser, Quadratic residue diffusor (QRD) and box diffuser. Scattering coefficients were estimated based on the measurement data 15. Diffusers were installed on the limited region of lateral walls to compare simulation results between the stations. The treated surface area was 216 m 2 in total. In cases of ribbed diffuser and QRD, the effects of diffuser direction were also investigated: horizontal and vertical directions, which are denoted by -H or -V. 3

Table 1: Details of the diffuser profiles with various shapes installed on lateral walls. Structural Coverage Case No. Diffuser shape sc height density HS Hemisphere diffuser 2 mm 14%.25 RD1 25 mm.45 Ribbed diffuser 5% RD2 125 mm.34 QRD QRD N=7 25 mm 1%.57 Box1A 25%.64 25 mm Box1B 5% - Box diffuser 25%.54 125 mm 5% - Examples of interior view Additionally, the effects of diffuser arrangements on lateral walls were investigated by comparing five simulation cases (full, 2/3 and 1/3 coverage in both horizontal and vertical arrangements) using ribbed diffuser with its structural height of 25 mm. Table 2 shows the details of simulation configurations for investigating diffuser arrangements on lateral walls. Table 2: Details of diffuser arrangements on lateral walls (Grey parts: diffusers). Case No. Diffuser area Arrangement Diagram Full 216 m 2 None 2/3-V 144 m 2 Vertical 1/3-V 72 m 2 2/3-H 144 m 2 Horizontal 1/3H 72 m 2 2.4.3 Diffusers on ceilings Among the diffusers in Table 1, four diffusers (R2-H, R2-V, and ) with its structural height of 125 mm were selected to investigate effects of diffuser type and direction on ceilings as shown in Figure 5. Diffusers were installed on the limited region of ceilings to compare simulation results between the stations. The treated surface area was 252 m 2 in total for both stations A and C. (a) (c) (d) Figure 5: Interior view of the simulation models with ceiling diffusers. (a),, (c) and (d) 2.4.4 Tilted wall profiles Five cases according to the inclined angles of -4.8, -2.4 (upward), (the current condition), +2.4 and +4.8 (downward) were prepared by tilting wall profiles facing upward and downward. The treated wall area was 216 m 2 for both station A and C. 4

STI STI 2.5 Results 2.5.1 Variation of scattering coefficients Figure 6 shows the simulation results in terms of STI and IACC for stations A and C. As shown in Figures 6(a) and, wall and ceiling surfaces with higher scattering coefficients showed decrease of STI in both stations. Especially, increase of ceiling scattering coefficient by.4 yielded larger variation of STI than increase of wall scattering coefficient by.4. As shown in Figures 6(c) and (d), higher scattering coefficients on ceiling showed decrease of IACC in both stations. However, there was no clear tendency of IACC by variation of wall scattering coefficients in island platform as shown in Figure 6(c), whereas increase of wall scattering coefficients yielded higher IACC in side platform as shown in Figure 6(d)..56.55 Island platform (A) W.1 W.3 W.5.56.55 Side platform (C) W.1 W.3 W.5.3 W.1 W.3 W.5.3 W.1 W.3 W.5.54.54.2.2.53.52 C.1 C.3 C.5.53.52 C.1 C.3 C.5 (a) Scattering coefficient Scattering coefficient (c) Scattering coefficient (d) Figure 6: STI in stations A (a) and C, and IACC in stations A (c) and C (d) according to variation of scattering coefficients (x) of wall (W x) and ceiling (C x). Figure 7 shows IACC distribution in both stations according to variation of wall scattering coefficients when the ceiling scattering coefficient was.3. As shown in Figure 7(a), it seems that unclear tendency of IACC in island platform caused by the outliers from the receivers at 12.5 to 15 m from the sound source near tunnel area. However, in side platform, IACC variation was found at the near receivers within 1 m from the sound source as shown in Figure 7..1 Island platform (A) C.1 C.3 C.5.1 Side platform (C) C.1 C.3 C.5 Scattering coefficient.6.4.2 Island platform (A) W.1 W1C3 C.3 W3C3 W.3 C.3 W5C3 W.5 C.3.6.4.2 Side platform (C) W.1 W1C3 C.3 W.3 W3C3 C.3 W.5 W5C3 C.3-2 -15-1 -5 5 1 15 2 (a) Source-receiver Source-to-receiver distance [m] [m] Source-receiver Source-to-receiver distance [m] [m] Figure 7: IACC as a function of source-receiver distance according to scattering coefficients of lateral walls (.1 to.5) when ceiling scattering coefficient was.3. (a) Station A and station C. 2.5.2 Diffusers on lateral walls Figure 8 shows differences of acoustic parameters between the current and other cases with various lateral wall diffusers. Results of RT were not included because RT showed similar tendency with SPL. As shown in Figure 8(a), most diffuser shapes contributed to increase STI. Especially, QRD profiles were effective to increase STI in both stations. In island platform, box diffusers with its coverage density of 5% were also effective to increase STI. In case of, STI was slightly decreased after installing diffusers in both stations. According to diffuser directions, there was no significant change of STI. As shown in Figure 8, large amount of sound energy attenuation in terms of SPL was observed according to installation of diffusers. Diffuser profile with higher scattering coefficient tended to show larger reduction of SPL. As shown in Figure 8(c), IACC was mainly decreased only in case of in both stations. The diffusers of QRD-V and In island platform 5-2 -15-1 -5 5 1 15 2

Diffuser types Diffuser types Diffuser types and Box1B in side platform showed decrease of IACC, but the amount of varied IACC was relatively small. Except for these diffusers, IACC was increased by most diffusers. It is similar results from the above investigation using variation of scattering coefficients in the simulation model as shown in Figure 6(d). Amount of varied SPL and IACC by installing diffusers in side platform was larger than that in island platform. This indicates that it is important to treat lateral walls in side platform since lateral walls are located close to the passenger area. According to diffuser direction, vertical diffusers showed larger variation of SPL than horizontal diffusers Box1B diffuser diffuser Box1A QRD-V QRD-H HD -.2-.1.1.2 (a) Δ(STI) w/diffuser-w/o diffuser Δ(SPL) w/diffuser-w/o diffuser (c) Figure 8: (a) Δ(STI), Δ(SPL) and (c) Δ(IACC) according to wall diffuser types. Figure 9 describes the results of acoustic parameters according to diffuser arrangements on lateral walls using ribbed diffusers (RD1). As shown in Figures 9(a) and, fully-arranged diffusers in both stations showed increase of STI regardless of diffuser direction. 2/3-covered horizontal arrangements for and in island and side platforms, respectively, showed slight increase of STI. Meanwhile, all vertical arrangements for showed decrease of STI in both stations. It reveals a possibility to avoid vertical arrangement in case of vertical diffusers. As shown in Figures 9(c) and (d), IACC was increased by all diffuser arrangements in both stations. Similar to Figure 8(c), increased amount of IACC in side platform was larger than that in island platform. Δ(STI) w/diffuser-w/o diffuser.2.1 -.1 -.2 Full 2/3-V 1/3-V 2/3-H 1/3-H Δ(STI) w/diffuser-w/o diffuser Diffuser arrangements Diffuser arrangements Diffuser arrangements Diffuser arrangements (a) (c) (d) Figure 9: Δ(STI) in stations A (a) and C, and Δ(IACC) in stations A (c) and C (d) according to wall diffuser arrangements. 2.5.3 Diffusers on ceilings.2.1 -.1 -.2 Box1B Box1A QRD-V QRD-H Full HD 2/3-V 1/3-V 2/3-H 1/3-H -1 -.5.5 1 -.1.1 Figure 1 shows differences of acoustic parameters between the current and other cases with various ceiling diffusers. As shown in Figure 1(a), in side platform showed increase of STI. In case of, STI was decreased by installing diffusers in both stations. SPL was decreased according to installation of diffusers as shown in Figure 1. IACC was increased by installing diffusers in both stations as shown in Figure 1(c). In particular, IACC in island platform was largely increased by, of which direction is orthogonal to the platform length. Thus, was effective to increase STI, but it accompanied with increase of IACC. 6 Δ() w/diffuser-w/o.1 -.1 Full 2/3-V 1/3-V 2/3-H 1/3-H Box1B Box1A QRD-V QRD-H HD Δ() w/diffuser-w/o Δ() w/diffuser-w/o diffuser.1 -.1 Full 2/3-V 1/3-V 2/3-H 1/3-H

STI Diffuser type Diffuser type Diffuser type -.2 -.1.1.2 (a) Δ(STI) w/ diffuser-w/o diffuser Δ(SPL) w/ diffuser-w/o diffuser (c) Figure 1: (a) Δ(STI), Δ(SPL) and (c) Δ(IACC) according to ceiling diffuser types. 2.5.4 Tilted wall profiles -2-1.5-1 -.5.5 1 -.1.1.2.3.4 Δ() w/ diffuser-w/o diffuser Figure 11 shows STI and IACC values according to the tilted angles of the lateral walls. As shown in Figure 11(a), there were no significant correlations of STI with the tilted angles of the lateral walls in both stations although STI followed the second-order regression model of the tilted angles in case of side platform (R 2 =.87). However, IACC showed significant correlations with the tilted angles of the lateral walls as shown in Figure 11. In side platform, IACC also followed the second-order regression model of the tilted angles. The lowest IACC was found at the current condition without tilting walls in side platform. Meanwhile, IACC in island platform was decreased according to increase tilted angle of lateral walls. As inclining lateral walls facing downward, lower IACC can be expected in island platform..54.25.53 (a) Tilted angle [ ] Tilted angle [ ] Figure 11: (a) STI and IACC variations according to the tilted angles of the lateral walls. 3 SCALE MODEL 3.1 1/25 Scale Models y = 8E-5x 2 + 3E-5x +.5349 R² =.8655 y = -.1x +.5261 R² =.3472.52-5 -2.5 2.5 5 y =.17x 2 -.18x +.168 R² =.9983 y = -.53x +.1716 R² =.9892.1-5 -2.5 2.5 5 In order to validate simulation results with diffusers, scale model experiments were planned. Scale model methods were widely employed to evaluate acoustic effects of diffusers in various spaces 3,16-17. In consideration of scale model size for underground stations, a scale factor of 1/25 was selected. From the 1/25 scale models, frequency range up to 3,84 Hz can be measured through an AD/DA converter with a sampling rate of 192 khz. Figure 12 shows scale model modules with its width of 91 mm. For scale model construction of station A, 6 pieces of module A, 2 pieces of module Y, and 2 pieces of module I are required. For station C, 6 pieces of module C and 4 pieces of module I are required..2.15 (a) (c) (d) Figure 12: Modules for making scale models of stations A and C in 1:25 scale. (a) Module A (Platform area of station A), Module C (Platform area of station C), (c) Module Y (Tunnel area of station A) and (d) Module I (Tunnel area of both stations) 7

3.2 Model materials Main body of scale model is made of MDF (medium density fiberboard). Absorption coefficients of model surfaces will be acoustically fitted by lacquering various painting and varnishing materials through the measurements of test samples in 1/25 reverberation chamber. In addition, diffuser profiles and model passengers will be reproduced based on its dimension and acoustic characteristics as shown in Figure 13. (a) 3.3 Remarks Figure 13: (a) Model diffuser (QRD) and model passengers in 1:25 scale. Currently, scale model is under construction. In the scale model testing, effects of lateral wall and ceiling diffusers on the sound field characteristics will be investigated with diffuser profiles of QRD and ribbed diffusers (RD1 and RD2). In addition, effects of passengers on the sound field distribution will be investigated. 4 SUMMARY In this study, effectiveness of diffusive lateral wall and ceiling was investigated for improving sound fields characteristics in underground stations with island and side platforms through computer simulations. From the simulation results, contribution of specific diffusers such as QRD and ribbed diffusers with very high scattering performances on the improvement of sound field characteristics was found in terms of STI and IACC. According to the platform style, different effects of diffusers on sound field distributions were found. Based on the simulation results, application guidelines of diffusers for improving sound fields in underground stations can be summarized according to platform style as followings: In an island platform, it is recommended to apply lower scattering coefficients on both lateral walls and ceilings, or to use QRD and Box (5%) diffusers on lateral walls for increase STI. In order to decrease IACC, it is recommended to use ribbed (small and horizontal) diffuser on lateral walls, or to tilt lateral walls facing upward. In a side platform, it is recommended to apply lower scattering coefficients on both lateral walls and ceilings, to use QRD on lateral walls, or to use ribbed (small and horizontal) diffuser on ceilings for increase STI. In order to decrease IACC, it is recommended to use of ribbed (small and horizontal) diffuser on lateral walls. As a further study, more investigations on finding effective lateral wall and ceiling diffusers will be carried out using 1/25 scale models. ACKNOWLEDGEMENTS This work was supported by a Grant-in-Aid for Young Scientists (A) from the Japan Society for the Promotion of Science (2368686). 8

REFERENCES 1 Hideki Tachibana, Desirable conditions for acoustic environment of public spaces, Noise Control (Sōon seigyo) (in Japanese) 23(4), 25 21 (1999). 2 Hiroki Kobayashi, Tasuku Kitamura, Minoru Okashita, Yasuhiko Izumi, Kojiro Fujii and Naoya Toyama, Development of the sound absorber for station concourse, Showa Group Technical Review, 56(1), 42 46 (26). 3Jian Kang, Acoustics of Long Spaces Theory and design practice (Thomas Telford, London, 22). 4 Ryota Shimokura and Yoshiharu Soeta, Characteristics of train noise in above-ground and underground stations with side and island platforms, J. Sound Vib., 33(8), 1621 1633 (211). 5 Jian Kang, Improvement of the STI of multiple loudspeakers in long enclosures by architectural treatments, Appl. Acoust. 47(2), 129 148 (1996). 6 Chye Heng Chew, Integrated bus/rail station, Appl. Acoust. 56(1), 57-66 (1999). 7 Zühre Sü and Mehmet Çalışkan, Acoustical design and noise control in metro stations: case studies of the Ankara metro system, Build. Acoust. 14(3), 23 221 (27). 8 Yong Hee Kim and Yoshiharu Soeta, Temporal and spectral characteristics of public address announcements in subway platforms with various finishing materials using computer simulation, INTER-NOISE Proc., New York, 212(7), 4797 488 (212). 9 Yong Hee Kim and Yoshiharu Soeta, Effects of reverberation and spatial diffuseness on the speech intelligibility of public address sounds in subway platform for young and aged people, To be published, ICA213, Montreal (213). 1 Trevor J. Cox and Peter D Antonio, Acoustic Absorbers and Diffusers Theory, design and application (Taylor & Francis, New York, 29). 11 Chan Hoon Haan, Predicted effect of station design changes on high speed train noise, Build. Acoust. 9(4), 311 323 (22). 12 Ryota Shimokura and Yoshiharu Soeta, Evaluation of speech intelligibility of sound fields in underground stations, Acoust. Sci. & Tech., 32(2), 73 75 (211). 13 IEC 6268-16, Sound system equipment Part 16: Objective rating of speech intelligibility by speech transmission index (International Electrotechnical Commission, Geneva, 23). 14 ISO 3382-1, Acoustics Measurement of room acoustic parameters Part 1: Performance spaces (International Organization for Standardization, Geneva, 29). 15 Michael Vorländer, Auralization Fundamentals of Acoustics, Modelling, Simulation, Algorithms and Acoustic Virtual Reality (Springer, Aachen, 28). 16 Jin Yong Jeon, Jong Kwan Ryu, Yong Hee Kim and Shin-ichi Sato, Influence of absorption properties of materials on the accuracy of simulated acoustical measures in 1:1 scale model test, Appl. Acoust., 7, 615 625 (29). 17 Yong Hee KIM, Jae Ho KIM and Jin Yong JEON, Scale Model Investigations of Diffuser Application Strategies for Acoustical Design of Performance Venues, Acta Acustica united with Acustica, 97(5), 791 799 (211). 9