Acoustic Characteristics of Four Subway Stations in Naples, Italy

Transcription

1 Acoustic Characteristics of Four Subway Stations in Naples, Italy Umberto Berardi a) Department of Architectural Science Ryerson University Toronto, ON, M5B 2K, Canada Gino Iannace b) and Giovanna Giordano Department of Architecture and Industrial Design Luigi Vanvitelli Second University of Naples Aversa, Italy Due to the presence of a large number of travelers and the high traffic of trains, subway stations are noisy environments. This noise prevents listening to safety messages, as well as affects people s conversations and health. Subway stations are also the performance spaces selected by street musicians, who perform in poor acoustic conditions. This study evaluates the acoustic characteristics of four new subway stations in Naples, Italy. Using the impulse response method, monaural parameters were measured. Results indicate average reverberation times longer than three seconds at middle frequency, and values of clarity (both C and D) and sound transmission index (STI) far from optimal for listening perception. As expected, the train tunnels of the subway lines resulted in a high absorption at low frequencies, making subway stations particularly frequency-selective. Acoustic corrections for reducing the noise levels and improving the acoustic characteristics are studied. In particular, this paper investigates the effect of adopting a sound-absorbing and fire-resistant plaster on the ceiling of the subway tunnel. 1 INTRODUCTION An urban subway represents a common way to solve public transit problems providing large capacity, fast speed, and convenient ways to alleviate urban road traffic issues. Unfortunately, trains in small and highly reflective underground environments create serious noise problems which considering the large number of travelers and workers in subway stations, are often particularly problematic. The noise created by the collision and friction of wheels and rails, as a) uberardi@ryerson.ca b) Gino.IANNACE@unina2.it

2 well as the noise from the braking is generally reflected by the hard surfaces (often concrete) that are common in subway stations. Old subway stations in cities such as New York, London or Paris do not have acoustical treatments, and often they were not designed considering acoustical aspects. In fact, they have an abundance of hard surfaces which were preferred for their being resistant, easy to clean, and with minimal risk of the early deterioration that could occur to soft sound absorbing materials. However, new subway stations are often equipped with sound absorbing treatments. An example of this is the Metro Center station in Washington DC, where the waffle exposed concrete roof of this brutalist architecture designed by Harry Weese has sound-absorbing panels in the concrete cavities (Fig. 1). Fig. 1 - Metro Center station in Washington DC with its vault designed by Ben Schumin. Many other measures have been taken into account to decrease the noise in subway stations, among which a commonly adopted strategy is the inclusion of platform screen doors (PSD). For example, the Westminster Station in London designed by Hopkins Architects almost twenty years ago includes a variety of acoustic treatments in it (Fig. 2). This station has automatic PSD that separate the platform from the train tracks, and that only open when the train has stopped. It has perforated metal panels covering mineral sound-absorbing material. A similar design was adopted in another station along the London Jubilee line, the Southwark Station (Fig. 2). Fig. 2 - Westminster Underground Station (left) and Southwark Station (right) in London. Literature data show that the noise level in underground stations is over 6 db higher than that in aboveground stations. Moreover, the noise levels are even higher for stations with island

3 platforms (one platform at the center and railway tracks on both sides) than for side platforms, although the earlier are often preferred for economic reasons. The trend in designing modern subway stations is to include sound absorption panels for less reverberation, to adopt PSD, and to select the finishing in order to guarantee internal scattering. This last aspect is particularly important to improve the sound diffusion and to increase the sound-to-noise ratio. The research in the field of the acoustic of subway stations is generally limited. Many studies about the lack of a diffuse sound field were conducted by Jan Kang twenty years ago using scale model measurements, numerical simulations, and theoretical explanations 1. Those studies allowed to understand the characteristics of long spaces respect to the typical room acoustic theory and to investigate the nonlinear decay characteristics and non-diffusivity of these spaces. Later, Yang and Shield 2 used ray-tracing simulations to investigate subway spaces and to predict the speech intelligibility in underground stations with rectangular cross sections. More recently, Lam and Li 3 have developed a new theoretical model for rectangular long enclosures. On the practical side, Su and Caliskan 4 presented the acoustical design of three new subway stations in Ankara, Turkey. Through computer simulations, the authors showed that sound absorbing materials were necessary on the ceiling, the sidewalls, and beyond perforated metal panels suspended from the ceiling. The lay-in material used behind 15% perforated aluminum panels was SoundTex by Freudenberg which has an absorption coefficient of 0.65 at 0 Hz. Tang et al. 5 recently used a finite-element software named Actran to analyze the acoustical performance of an underground subway station in Shanghai. This analysis allowed figuring out the best position for the sound insulation material, a 3cm-thick polyurethane foam, in order to reduce the noise levels through the station. The Japan Health Research Institute has recently promoted several studies on subway stations investigating different topics, including the effect of interior materials 6, the benefits of PSD 7, and the public address systems 8. In particular, the PSD, especially mobile full-height doors, showed to reduce noise at the low frequency by making the sound field more diffuse 7. As a result, PSD have been encouraged since they also help to prevent accidents from service trains passing through the station or falls from the platform onto the track area, and they guarantee a more efficient climatic control within the station. The study by Shimokura and Soeta has also helped understanding the effect of the train position (as a sound source) in subway station 6 : in fact, the sound field for a sound source near or in the tunnel presented a higher strength (G) by 5.1 db and longer reverberation time by 0.7 s compared to the sound source in the station, mainly due to the convergence effect of the tunnel 6. This was also confirmed by the fact that sound sources located at the center of the platform present strong early reflections, while sound sources in the tunnel present long reverberation and a sound transmission mainly in the direction parallel to the track 6. The attention for the speech intelligibility in subway stations for hazard warning or other information has often led to adopt advanced public address systems 2,9. However, minimum acoustic standards have rarely been defined for subway stations. In this paper, the values of the STI index will be considered together with other more common parameters in room acoustics to describe the acoustic characteristics of four subway stations in Naples, Italy. 2 THE SUBWAY STATIONS IN NAPLES The origins of the subway system in the city of Naples date back to 1963, but only a partial opening has occurred in the last two decades. When it will be completed, the subway line will design a closed loop through the city center. Trains have a maximum speed of km/h, a length of 35 meters, and are composed of four carriages each hosting seats and 372 standing places.

4 The construction of the various subway stations of Naples has been promoted with the idea to realize a series of art gallery stations. Famous architects have collaborated with artists to create stations that are functional and comfortable, and that should be able to rehabilitate city neighbors. This means that the typical "non-places" of public transit areas in subway stations have been planned as community spaces where art pieces are exhibited to both passengers and visitors. As a result, more than 1 art pieces have already been located in these stations. The four subway stations considered in this paper (and shown in Fig. 3) are: Garibaldi station: it was designed by the French architect Dominique Perrault in The station is structured as a single, bright environment through opened suspended escalators. Natural light penetrates from the transparent ceiling and connects the station to the street level; Università station: the interior of this station is characterized by fluent volumes and strong colors. The design architect Karim Rashid was inspired by digital languages and global communication networks to explore innovative media, especially in the continuously changing colors surfaces; Toledo station: in this station designed by the Catalan architect Oscar Tusquets Blanca, the blue structure of skylight provides a strong link between the outside and the inside, while the walls are decorated by waves in relief with blue mosaics, which increase in intensity with the depth; Dante station: in this Gae Aulenti s project, the gates are made of clear glass. Several contemporary art works are displaced along the platforms which also have mosaics. Fig. 3 Photos of the four investigated subway stations: Garibaldi station (top left), Università station (top right), Toledo station (bottom left), and Dante station (bottom right).

5 Figure 4 shows the plans of the four station, whereas Fig. 5 reports their cross section. The figures clarifies that that the four stations comparable dimensions and a circular cross section. Fig. 4 Plans of the different stations with indication of the source (S) and receiver (x) positions, as well as of the microphone location for the train noise measurements ( ). Garibaldi station Università station Toledo station Dante station Fig. 5 Cross section of the different stations with the microphone position.

6 T30 (s) 3 ROOM ACOUSTIC MEASUREMENTS Room acoustic measurements were performed in the different stations using an ambisonics microphone BRAHMA. The acoustic measurements were done during late evening hours without passengers, so the impulse response was recorded in empty conditions. Seven receivers at the height of 1.6 m from the floor were located along the platform (Fig. 4). Using the software DIRAC, monaural parameters were then calculated. First, an analysis of the variability of the results in the different receiver position was done. Figure 6 shows the reverberation time with the maximum and minimum values in two stations. The graphs show that the values had similar trends, longer reverberation time at middle frequency, while at high frequency, the higher absorption of the air results into a decrease of the reverberation time. The graphs show a high variability of results at low frequency: the standard deviation in Garibaldi station was in the range of 0.5 s for the frequency bands of 63 Hz and 125 Hz and around 0.1 s above these frequency bands, whereas, in the Università station, the standard deviation was 0.4 s for low frequency bands and reduced to 0.3 s at high frequency bands. Figure 6 also confirms that subway stations are particularly frequency selective 6 : the reverberation time at 63 Hz was 1.16 s in the Garibaldi station and 0.74 s in the Università station, whereas at middle frequency, the reverberation was much longer, being 3.4 s in the Garibaldi station and 2.8 s in the Università station at 0 Hz. Although the results in Fig. 6 have a similar pattern to those reported in other studies, the absolute values of the reverberation time was generally longer. Referring to the Turkish Noise Control Act, Su and Caliskan 4 state that the optimum reverberation time for an unoccupied metro stations at 0 Hz should be between 1.2 s and 1.4 s. In that paper, the authors discuss about the values found in some stations in Ankara where the T30 resulted around 1.5 s, a value well below those shown in Fig Garibaldi min Max Frequency (Hz) T30 (s) Universita' min Max Frequency (Hz) Fig. 6 Reverberation time in the stations of Garibaldi and Università together with minimum and maximum values recorded among all the combinations of sound-receiver. Figure 7 reports the results of the room acoustic measurements in the four stations according to the ISO The figure shows the reverberation time (T30), clarity (C) and the definition (D). Average values for different source-receiver combinations are reported in the frequency octave bands from 63 Hz to 4 khz. Looking at this figure, it emerges that the frequency selective behavior is typical in every station and it is generally common to have high sound absorption in the 63 Hz band, which depends on the cross section and depth of the train tunnel. The differences among the stations are generally

7 limited, although the Università station consistently shows shorter reverberation time at low frequency and a longer one at high frequency, as a consequence of its smaller cross section and volume. However, it should be mentioned that attached to the walls of the different stations, there are paintings realized on metal panels in the Garibaldi station) or plastic panels (in the Università station), and the different absorption of these panels influence the overall absorption. 3.5 T30 (s) 3.5 T30 (s) - modified Garibaldi Università Toledo C (db) Dante Garibaldi Università Toledo D (-) Dante Garibaldi Università Toledo C (db) - modifieddante Garibaldi Università Toledo D (-) - modified Dante Garibaldi Università Toledo Dante Garibaldi Università Toledo Dante Fig. 7 Acoustical data for the reverberation time (T30), the clarity (C) and the definition (D) for the investigated subway stations in the current situation (left) and after the acoustical correction intervention (right).

8 The clarity parameter assumes desirable values only at very low frequency, and for some stations at higher frequency, whereas it has negative values at middle frequency. The exception is the Università station, where the parameter C has positive values also in the frequency bands of 0 Hz and 1 khz. The definition (evaluated using the parameter D, so limiting to a ms time interval) follows pretty well the behavior described for the clarity. Values of D are above 0.4 for the Università station through almost all the frequency bands. Both Garibaldi and Toledo stations stand out for the really low values of D. In particular, in the Toledo station, the D is generally well below 0.2 in any frequency band except for the 63 Hz band, whereas in the Garibaldi station, the D assumes really low values in the frequency bands from 2 Hz to 1 khz. 4 TRAIN NOISE MEASUREMENTS Some measurements of the noise levels in the different stations with the transit trains were done. This allowed to evaluate the noise to which the passengers and people working in the stations are exposed during their waiting in the station or their daily activity respectively. Figure 8 reports the spectrum and two-hour time history in the different stations. The results show that: in the Garibaldi station, the Leq was 75.4 dba and the maximum SEL was dba; in the Università station, the Leq was 75.1 dba and the maximum SEL was 98.5 dba; in the Toledo station, the Leq was 81.7 dba and the maximum SEL was dba; in the Dante station, the Leq was 75.8 dba and the maximum SEL was 98.7 dba. #1 [Average] 0Hz 67.9dB A* 75.5 #1 [Average] 0Hz 69.4dB A* k 2 k 4 k 8 k 16 k Lin* A* WED 04/03/15 15h38m27s dB 0h05m18s dB #1 Leq 0ms A SEL k 2 k 4 k 8 k 16 k Lin* A* WED 04/03/15 16h03m35s0 75.1dB 0h03m41s dB #1 Leq 0ms A SEL 100 Spectrum 15h30 15h 15h 16h00 16h10 16h20 16h30 16h 16h 17h00 17h10 Spectrum 15h30 15h 15h 16h00 16h10 16h20 16h30 16h 16h 17h00 17h10 #1 [Average] 0Hz 74.0dB A* 81.8 #1 [Average] 0Hz 69.5dB A* k 2 k 4 k 8 k 16 k Lin* A* WED 04/03/15 16h44m15s dB 0h05m45s dB #1 Leq 0ms A SEL k 2 k 4 k 8 k 16 k Lin* A* WED 04/03/15 16h57m16s dB 0h03m14s0 98.7dB #1 Leq 0ms A SEL 100 Spectrum 15h30 15h 15h 16h00 16h10 16h20 16h30 16h 16h 17h00 17h10 Spectrum 15h30 15h 15h 16h00 16h10 16h20 16h30 16h 16h 17h00 17h10 Fig. 8 Noise recorded in the Garibaldi station (top left), Università station (top right), Toledo station (bottom left), and Dante station (bottom right) during some train transits. Looking at the sound levels in different frequency bands, it emerges that in any station the recorded noise levels drops consistently above 2 khz. Noise in the low frequency bands

9 generally stands a few decibels lower than the maximum values which were often measured in the middle frequency bands (from 2 Hz to 1 khz). The high sound levels at low frequency, also considering the absorption of the metro tunnel, was due to the spectrum characteristics of the noise emitted by the train. These spectra reflect the room acoustics measured in section 3. Moreover, looking at the time history, it is also noticeable to see that values above dba were recorded repeatedly within the stations at the train arrival or more often as the trains enter into the tunnel (outbound). The high SPL values recorded in the different stations are well above the results obtained in other modern subway stations. For example, Tang et al. 5 measured SPL values around 65 dba in the seat areas and below db at the entrance or exit of the train. It is also worthy to mention that the station investigated by Tang et al. had a PSD system. The values recorded in the stations in Naples are also above those found on the platform edge of the Ankara stations 4 which resulted in sound levels around -82 dba, generally below the Turkish limit of 85 dba. Finally, the values found in Naples stations are more similar to those reported by Soeta and Shimokura for old underground stations without PSD 7. This comparison seems to suggest that the lack of a proper Italian standard for maximum noise levels in subway stations has discouraged the design of solutions which would have helped to reduce the noise levels in the new subway stations of Naples. However, it is worthy to note that since the stations are a working environment too, their noise levels should still respect the Italian law n.81/2008 ( Testo unico sulla salute e sicurezza sul lavoro ) which, considering the noise levels recorded in the stations, would prescribe several attenuation noise control actions. 5 IMPROVING THE ACOUSTICS OF THE STATIONS The study then focused on improving the acoustic conditions in the investigated subway stations. The software Odeon was used for the simulation. Ray-tracing simulation software have been widely used to predict acoustic environments in train stations as well as in auditoria 8,12. Air absorption was included considering a temperature of 20 C and a relative humidity of %. The simulated sound source was omnidirectional with a sound emitting power of db. The binaural impulse responses were derived from the simulation. Acoustic fitting was carried out by changing the absorption and scattering coefficients of the finishing materials in order to match the simulated values with the measured ones reported in Fig. 7. Acoustical agreement between field measurements and simulations was determined in terms of just noticeable difference (JND). As seen in section 1, sound absorbing treatments in subway stations have often considered the application of mineral fibers. The main advantages of this material is its resistance to high humidity with limited mold growth issues, and its being class 0 fiver, hygienic and anti-bacterial. Alternative materials to mineral fibers for subway stations could be: Thermatex face pattern by AFM, which has a high ISO classification of Class 1 in accordance ISO 14644, and it is cleanable, pressure washable, and moisture resistant; this material has an absorption coefficient of 0.45 at 125 Hz, 0. at 1 khz, and 0.8 at 2 khz; Starsilent (Acoustimed) by Pyrok is a spray sound absorbing plaster which increases both the fire retarding performance and the sound absorption one, although with this material is hard to have a completely smooth surface, and as a result more dust would accumulate. Finally, the sound absorbing material that was selected for the Naples stations consisted of a fire proof plaster with cork granules (diathonite acoustic by Diasen ). This material was intended to be applied to the roof of the stations for a surface of almost 0 square meters in each station. This material has a sound absorption of 0.16 at 63 Hz and 125 Hz, 0.28 at 2 Hz, 0.63 at 0

10 Hz, 0.74 at 1 khz, and 0.66 at 2 khz. Moreover, it is environmentally friendly and recyclable since it is done of natural materials such as cork, clay, diathomeic powders, and hydraulic binder. The product also presents a high breathability, good thermal insulation characteristics, dehumidifying capacity, excellent fire resistance, and it is anti-mold and bacteriostatic. The simulation with acoustic corrections of the four stations using this material improved significantly the acoustics of the stations, as shown in Figs. 7 and 9. The reverberation time reduced significantly in all the stations, becoming always shorter than 2 s at any frequency band (Fig. 7). In particular, the T30 assumed values below 1 s at frequency above 0 Hz in Garibaldi, Università, and Dante stations, and only in the Toledo station it was estimated to be around 1.5 s after the acoustic correction. The obtained results after the acoustic corrections are similar to those showed by Kim and Soeta for some Japanese stations 8. Fig. 9 STI values in the current situation and after the acoustic corrections.

11 It is important to note that the frequency selective behavior which was measured in all the stations and was confirmed in the simulation of the current status, would be strongly attenuated by the sound absorbing additional intervention. This means that the quality of the perception in any subway station would increase significantly. For example, the C will become generally positive, with the only exception of the Toledo station in the frequency bands of 125 Hz and 2 Hz. Finally, the result of the simulations with the sound absorption correction showed increased clarity values especially above 0 Hz, where the C values increased in all the four stations. Figure 9 reports the results of the STI simulated in the four stations before and after the application of the sound-absorbing plaster. The color maps in Fig. 9 indicate a distinctive improvement of the STI values, especially in the Università and Toledo stations. The average value of the index STI in the Garibaldi, Università, Toledo and Dante stations after the introduction of the sound absorbing materials became 0.66, 0.68, 0.57, and 0.66 respectively. These values correspond to the good labelling class of STI, and indicate values more in line with the desired standards in modern metro stations 11. In fact, some previous studies have found that in underground stations and road tunnels, good values of the STI should be above CONCLUSIONS The present paper has investigated four subway stations recently completed in Naples, Italy. Together with being functional to the new subway line in this major Italian city, these stations have been thought as a sort of art gallery where visitors and passengers could appreciate artworks. In fact, many travelers visit these stations even when they are not interested in travelling in the subway. As a result, it is evident that the comfort conditions in these stations represent a fundamental element of these new stations. However, the stations have been designed without acoustic treatments, and the reverberation time is always particularly long. Both clarity and definition have shown particularly low values, while the STI index in the current condition resulted lower than 0. in all the subway stations. Moreover, the noise measurements reported in this paper have also indicated that serious concerns for the staff working in these stations should be addressed. This study has hence investigated a strategy to improve the acoustic conditions by applying a cork granule sound-absorbing plaster on the roof of the stations. Thanks to the sound absorption behavior at medium and high frequency of this material, the stations would result in a more balances acoustic spectrum, where the low frequency absorption of the gallery would be balanced by the high absorption of the new plaster. The simulated results show that the average STI in the Garibaldi, Università, Toledo and Dante stations after the introduction of the sound absorbing material would be equal to 0.66, 0.68, 0.57, and 0.66 respectively. The STI improvements would guarantee a better acoustical perception of the public address systems and more comfortable environments. Finally, the values obtained after the corrections would align with the standards often set for new subway stations. Further studies will focus on simulating the effects on the propagation of the sound from the specific public address system currently installed in the stations with and the without the acoustic corrections presented in this paper. This ongoing study will also take into consideration the importance of the directivity and spectrum characteristics of the actual installed loudspeakers in order to better evaluate the current and future room acoustic conditions in these new stations.

12 7 ACKNOWELDGEMENT The first author is grateful to the useful discussion with Raj Patel, Global Leader for Acoustics, Audio Visual, and Theatre Consulting at ARUP. The authors wish to thank the station managers for allowing access to their facilities and for their valuable cooperation during the measurement sessions. 8 REFERENCES 1. J. Kang, A Method for Predicting Acoustic Indices in Long Enclosures, Applied Acoustics, 51(2), 169-1, (1997). 2. L. Yang and B. M. Shield, The prediction of speech intelligibility in underground stations of rectangular cross section, J. Acoust. Soc. of Am., 109(1), , (2001). 3. P.M. Lam and K.M. Li, The predicted reverberation time in a rectangular long enclosure, Proceedings of Inter-noise, (2004). 4. Z. Sü and M. Çaliskan, Acoustical Design and Noise Control in Metro Stations: Case Studies of the Ankara Metro System, Building Acoustics, 14(3), , (2007). 5. C.H. Tang, Y.S. Wang and H. Guo, Sound Field Simulation and Optimization at an Underground Subway Station, Building Acoustics, 20(3), , (2013). 6. R. Shimokura and Y. Soeta, Sound field characteristics of underground railway stations Effect of interior materials and noise source positions, Applied Acoustics, 73, , (2012). 7. Y. Soeta and R. Shimokura, Change of acoustic characteristics caused by platform screen doors in train stations, Applied Acoustics, 73, , (2012). 8. Y.H. Kim and Y. Soeta, Architectural treatments for improving sound fields for public address announcements in underground station platforms, Applied Acoustics, 74, , (2013). 9. M.F. Harrison, Calculating speech intelligibility for the design of public address system at railway stations, Mechanical Engineers, 215, , (2001). 10.ISO , Acoustics - Measurement of room acoustic parameters - Part 1: performance spaces, (2009). 11.B.U. Rubin, Audible Information Design in the New York City Subway System: A Case Study, Proceedings of the 5th Int. Conf. on Auditory Display - ICAD, (1998). 12.U. Berardi, Simulation of acoustical parameters in rectangular churches, J. of Building Performance Simulation, 7(1), 1 16, (2014).