FIVE SENSITIVITIES OF A COUPLED VOLUME CONCERT HALL: A COMPENDIUM

Transcription

1 FIVE SENSITIVITIES OF A COUPLED VOLUME CONCERT HALL: A COMPENDIUM Michael Ermann 1 1. Virginia Tech School of Architecture + Design; 201 Cowgill Hall (0205); Blacksburg, VA 24061; mermann@vt.edu; Abstract In the execution of a coupled volume concert hall, a dynamic and kinetic architecture defines a space s acoustics. Wrapping a concert hall audience chamber with another room, or coupled volume, and connecting the two with adjustable apertures, creates the opportunity for a double sloped sound decay. The acoustic interaction between the primary and secondary volumes offers a level of variability and control to the sound field that is not possible in traditional, single volume halls. Most distinctly, the double sloped approach promises a measure of simultaneous clarity and reverberance (qualities long thought to be mutually exclusive). While some spaces achieve that promise, others probably do not. Measurements taken in a coupled volume concert hall, simulations of coupled volume concert halls, and listening tests suggest a fickle relationship between the architecture of the hall and its ability to provide both clarity and reverberance simultaneously. The two halls studied suggest five sensitivities of the system: (1) the coupled volume must be exceedingly more reverberant than the audience chamber, (2) the aperture linking the two spaces must be exceedingly small, (3) location within the hall relative to the apertures impacts what is heard, (4) the background noise level must be exceedingly low, and (5) it is unclear whether subjects of a listening test preferred the coupled volume system. 1. Introduction The coupled volume concert hall and its signature sound decay the double sloped acoustic offer a tantalizing promise to designers. By attaching a reverberant coupled volume of space to a room for music making and listening, acousticians and architects hope to create an impulse response that rapidly decays at first, but later decays slower as sound that had been trapped in the coupled volume leaks out. Given that rapid sound decay is the hallmark of acoustic clarity, and given that slower sound decay is the hallmark of acoustic reverberance, and given that clarity and reverberance are each desired yet inversely related, the double slope undertakes to reconcile the two. In this impulse response model, the early rapid slope allows each note to decay and make room for the next (clarity) and the late gradual slope allows each note to linger and blend into the next (reverberance). Author-1

2 Figure 1. Left: Impulse response sound decay, based on the Sabine formula for diffuse sound fields, predicted for a concert hall (main hall only)l. Center: Sabine decay predicted for a single large concert hall of a size equal to a main hall plus a coupled volume. Right: Double sloped sound decay, based on Kuttruff s 1 formula for a non-diffuse, coupled volume condition where the two rooms are only partially connected through apertures. Figure 2. Left: Concert hall massing diagram of a main hall space, saddled by two coupled volume spaces. Right: Interior perspective of the main hall of a coupled volume concert hall, looking toward the stage; the partially open apertures lining the side walls and portions of the upstage wall acoustically connect the main hall space to the coupled volume space. The ray tracing software results presented in this paper utilize this model, which is based on a built U.S. concert hall. Once anathema to good acoustics, the double sloped decay has received attention since the early eighties when a flurry of coupled volume concert halls began construction. 2,3 The process continues today with recent or upcoming coupled volume concert hall openings in Singapore, Miami, and Orange County, California. Cesar Pelli, Raphael Vinoly, I.M. Pei, Jean Nouvel, and the Finnish firm Arkkitehtityöhuone Artto Palo Rossi Tikka Oy have designed coupled volume concert halls, among others. Yet, anecdotal evidence, reputation, and room measurements suggest that of the approximately 20 concert halls built with dedicated coupled volumes designed to provide a double slope decay, perhaps only a handful deliver on that promise. 4,5,6,7,8 Why? This line of research finds five sensitivities inherent in the design of concert halls based on architectural composition, haptic perception, background noise levels, listener sensitivity, and listener preference. Designers of concert halls, note: coupled volume systems are remarkably sensitive to these issues in a way not typical of other spaces designed for unamplified music listening. 2. Methods This line of inquiry, spanning six years, draws on anecdotal evidence, statistical acoustics simulations, geometrical acoustics simulations, in-situ room measurements, and subjective listening tests. Two coupled volume halls have been examined and modeled to derive the Ermann-2

3 results. New Hall is a coupled volume concert hall that has opened since the year 2000; Old Hall is a similar venue (though with fewer apertures connecting the main hall with the coupled volume) that opened in the 1990s. The author gathered anecdotal evidence from interviews with operators and designers of coupled volume concert halls, many from his year working for the architectural acoustics consulting firm that has designed a majority of the world s coupled volume halls. He developed a software program, based on established statistical acoustics to model and compare varying architectural compositions of coupled volume concert halls. 1,9 Further models were created using the geometrical-acoustics-based software, CATT-Acoustic, which both simulates impulse response decays and auralizes the results so that a user may hear the room s simulated impulse response convolved with an anechoic musical recording. In-situ measurements were taken at 16 positions in Old Hall by binaurally recording an organist during rehearsal. All but one of the recording positions were unable to reveal a double sloped decay; one of the positions probably revealed a double slope. That measurement was compared to both the statistically based and geometrically based prediction methods to verify their efficacy as modelers. The results of that comparison neither confirm nor repudiate either model, but the statistically based model appeared to be more accurate. Finally, listeners of varying experience were given paired comparison tests to determine perception of, and preference for, the double sloped decay. Experienced listeners volunteered at acoustics conferences, and generally-less-experienced student listeners volunteered from both architecture and architectural acoustics classes. In each case, subjects were given eleven pairs of CATT-Acoustic auralizations created from models with varying architectural compositions and for each pair were asked (1) if they heard a difference between the two recordings, (2) which one of the two is more double sloped or more likely double sloped, and (3) which one of the two they prefer. Note that this paper is written with architects and designers of coupled volumes in mind as readers. Those wishing to learn more about the technical details involved in the statistical acoustics and geometric acoustics simulations, as well as the in-situ room measurements are directed to the publications referenced in endnote 5. Those wishing to inquire about the anecdotal evidence gathered, or the technical details of the listening tests (which will be published in an upcoming paper) are encouraged to contact the author. 3. Sensitivity I: coupled volume materiality and scale To achieve a double sloped condition, sound must leave the main hall, move into the coupled volume, wait a bit for sound in the main hall to decay rapidly, then slowly leak back into the main hall to form the late gradual decay of the double slope. For this to work, the sound level in the coupled volume created by the impulse in the main hall must, over time, be louder than the sound level in the main hall. Clearly, then, the coupled volume must be reverberant. What has been found here, however, is that the coupled volume must be exceedingly reverberant to effect a double sloped decay. Typically, when we speak of sound reflective materials, we are referring of those with absorption coefficients less than Using the geometrical volume of the coupled volumes found in Old Hall and New Hall as a baseline for size, the world of the double slope then requires an absorption coefficient of less than 0.02 a full order of magnitude more stringent. This limits the materials that can be used for coupled volumes of these sizes to: painted brick, smooth concrete, marble and glazed tile (on concrete). See Figures 3 through 5. Indeed, of the few halls that are Ermann-3

4 reported as successful in achieving a double sloped decay, all have coupled volumes of concrete. Of the many more that are reported as less successful, the author knows of none that are constructed of concrete, but several which are dominated by wood or concrete block. Figure 3. Statistical acoustics predicted decays at 1000 Hz for different material profiles of the coupled volume. Top left: smooth concrete (α=0.02). Top right: 1/2 gypsum board nailed to studs (α=0.04). Bottom left: concrete block (α=0.07). Bottom right: 3/8 plywood over airspace (α=0.09). The green dashed line indicates the predicted decay for the main hall only condition; the blue dashed line indicates the predicted decay for a single large volume equal in size to the main hall and coupled volume combined. For these simulations (in red) the apertures connecting the two halls were set to 0.5% of the total available surface area of the walls and ceiling of the main hall. Ermann-4

5 Figure 4. Statistical acoustics predicted decays at 125 Hz. Note the limited material palette available to effect a double sloped decay. Figure 5. Geometric acoustics predicted decays at 1000 Hz. Top left: smooth concrete. Top right: 1/2 gypsum board nailed to studs. Bottom left: concrete block. Bottom right: 3/8 plywood over airspace. Multiple red lines indicate simulations at different receiver positions. It is not enough, it appears, to simply specify an exceedingly sound reflective material in which to clad the coupled volume. The designer must also be wary of the typical residue of construction Ermann-5

6 processes, fire and egress requirements, building systems, and end-user patterns if any of these elements are to be exposed to the sound energy that enters and later leaves the coupled volume. Spray-on fireproofing, metal ducts, air cavities behind plaster finishes, doors, miscellaneous items held in storage, grilles and diffusers, raceways, conduit, and lighting may affect performance. In a typical space, even one designed for unamplified music listening, many of these ancillary elements may not have a significant impact, but in the sensitive world of the coupled volume, small changes in materiality leverage large changes in the behavior of sound. Figure 6. Residues of building processes make an impact on impulse responses in double sloped decay. Left: 1000Hz comparison of concrete coupled volume versus the same condition with 5% of the surfaces in sheet metal duct and 5% of the surfaces covered in spray-on fireproofing. Right: 125 Hz comparison of plaster on brick versus typical plaster condition. Figure Hz comparison of Old hall as it is drawn and published 11 vs how it was observed on a visit. The visit revealed additional concrete structure, not apparent in the drawings (approximately 10% of total surface area), steel raceways, catwalks, pipes, roof truss structure and music stands in storage (2.5%), partially exposed velour drapes (1.1%), wood doors, organ stop cases, crates in storage (2.3%), sheet metal ducts, electrical panels (0.9%), cardboard tubes (0.1%), and rubber conduit (0.5%). 4. Sensitivity II: aperture size In these systems, adjustable apertures control the sonic transparency between main hall and coupled volume. When the apertures are fully closed, the system operates as a single, smaller, room. When the apertures are partially open, the system may operate as a coupled volume, producing a double sloped impulse response sound decay. When the apertures are fully open, Ermann-6

7 the system approaches that which would be found in a single, larger volume comprising the main hall plus the coupled volume. Where then, in this continuum from closed to open, lies the threshold when single smaller room becomes a coupled volume system, and, as we continue to open the apertures exposing the main hall to the coupled volume, where does the coupled volume system approximate a single, large room? On this question, again, the anecdotal evidence, the statistical simulations and the geometric simulations generally agree. And again, the system is remarkably sensitive. Interviews with concert hall designers occurring before the quantitative portion of this research commenced suggested that a double sloped decay may be audible immediately, upon opening the doors a very small amount. The double sloped effect gives way to the single-large-room-decay when the doors connecting the two approach 3% to 5% of the total available wall and ceiling space. (For clarity, descriptions of apertures degree of openness will be expressed as the percentage of New Hall s available main hall surface area, not including the floor.) The interviews indicated that the double sloped phenomenon peaks at approximately 0.5%. The statistical and geometric simulations bear this out: at 0.1% a discernable double slope emerges, at 0.4% it peaks, and at 3% to 5% it settles back to approach a classic, single (larger) volume Sabine decay. Again, we see the sensitivity of the system to the architectural composition of the spaces. When both aperture size and coupled volume reverberance are allowed to interact, the inherent fickleness of the double sloped decay emerges. To effect such a decay, the designer must work with the hall operator and target a limited sweet spot, where both of the following conditions are met: the coupled volume is much more reverberant than the main hall and the apertures linking the two spaces are relatively small. Note that coupling constant is used in the figures that follow. This term refers to the ratio of the time required for an impulse to decay by 60db relative to the early decay, extrapolated out to approximate the time it would take to decay by 60db, if the early decay was allowed to continue linearly (on a decibel scale). A linear, Sabine, decay will have a coupling constant of 1.0; a higher coupling constant is suggestive of a double sloped decay. For more on the coupling constant, refer to endnote [5]. Ermann-7

8 Figure Hz statistical acoustics comparison of different aperture sizes. Based on New Hall with a concrete coupled volume. Ermann-8

9 Figure Hz geometric acoustics comparison of different aperture sizes. Each decay line is associated with a different location within the hall. Based on New Hall with a concrete coupled volume. Ermann-9

10 Figure Hz statistical acoustics comparison of reverberation ratios (coupled volume/main hall), aperture sizes, and coupling constant. The higher the coupling constant, the more likely the impulse response will be heard as having a double slope. Figure Hz geometric acoustics comparison of reverberation ratios (coupled volume/main hall), aperture sizes, and coupling constant. Note that Figs. 10 and 11 identify a narrow sweet spot, with very reverberant coupled volumes and very small apertures, in which double sloped decays can be created. Ermann-10

11 5. Sensitivity III: aural haptic perception Haptic perception involves one s sense of where one is in space. Anecdotal evidence, geometric acoustics, in-situ measurements, and listening tests suggest that a listener s proximity and view to the apertures connecting the main hall to the coupled volume space may impact the character of what is heard. When visiting Old Hall, the staff alerted the author to the location the front of a balcony with excellent acoustic lines-of-sight to the apertures from which the double sloped decay is most likely to be reported. This was the only location of the 16 measured where the author heard a double slope, and the impulse response taken there seems to indicate that a mild one may be present. In contrast, impulse responses taken from organ stop chords at seats very near the apertures themselves appear closer to the predicted linear Sabine decay for the single-large-room condition than they do to the double sloped condition hardly surprising given that the listener is in, or close to, both spaces simultaneously. Geometric acoustics predictions suggest that while there is no discernable difference in the coupling constant between locations that vary along the height axis or length axis, something different may be heard along the width axis when listeners move within two meters of an aperture. Not only do the graphically plotted impulse responses look different, but listeners to a paired comparison of two recordings auralized from different locations had a relatively easy time identifying the difference between them. Both simulations were created in what is believed to be the easiest to identify as a double sloped condition (1% open aperture and smooth concrete coupled volume). One, taken at orchestra level, house center, was by far preferred to the other, taken at orchestra level, house left, near the apertures. Figure Hz in situ room measurements, derived from recordings of organ stop chords. The house center measurement impulse response, with clear acoustic lines-of-sight appears to be closer to the double slope prediction, while the impulse response associated with the measurement location immediately adjacent to an aperture appears to be closer to the prediction for a single-largevolume linear Sabine decay. Ermann-11

12 Figure Hz geometric acoustics New Hall simulations (0.5% open apertures, smooth concrete coupled volume condition). The receiver positions near the apertures, as a group, are notably different from the group of receiver positions farther from the apertures. This split was evidenced only in the simulated hall conditions that created the most dramatic double sloped decays (the ones with the highest coupling constant). Figure Hz geometric acoustics New Hall simulations. Coupling constant does not vary between locations along length of room. 6. Sensitivity IV: background noise Maintaining low levels of background noise is paramount in any space for unamplified listening, but in the realm of the coupled volume, double sloped perception requires exceptionally low background noise. This is because if the crossover point the elbow in the impulse response Ermann-12

13 where the initial rapid decay gives way to the later-arriving slower decay falls below the noise floor, the double slope can not be heard. To account for this, the designers of coupled volume concert halls interviewed for this line of research, target the threshold of hearing as a maximum acceptable background noise level. Anecdotal evidence, in situ noise measurements, and both statistical and geometric acoustic simulations suggest that a low signal-to-noise ratio will likely eliminate perception of the double slope. However, absolute levels of background noise are not solely responsible for determining if a cross-over point falls below the noise floor. Obviously, the sound level of the music being played impacts the location of the crossover point relative to the noise floor. Less obvious is the impact of aperture size. With increasing aperture size comes a crossover point that is sooner or higher in the impulse response and, therefore, less likely to fall below the noise floor. See Figure 8 and note the location of the crossover point relative to the size of the aperture. Figure Hz in-situ room measurement at orchestra level of Old Hall. Note the elevated level of the background noise relative to the predicted double slope of the curve. In this situation, even if the hall s architectural composition were favorable to the creation of a double sloped decay, it likely would be drowned by the background noise, which was clearly audible to the researcher. 7. Sensitivity V: perception and preference The larger question, of course, is Who Cares? In other words, are listeners able to perceive the double slope decay? And if they are, are they able to perceive it during music, or only at stop chords? Finally, if they can perceive it, do they prefer it? For this study, listeners were given auralizations of pairs of recordings and asked about perception and preference. Through analysis of the data, trends appeared. First, many of the listeners were liberal responders. That is, even when each member of the pair of recordings were identical, the subjects often identified them as different from one another. For this reason, a group of proficient listeners were identified who recognized the control groups of pairs that were identical and were teased out as a separate category. The other two categories of respondents were (1) student volunteers and (2) professionals attending acoustics conferences. Generally the proficient listeners were more successful at correctly identifying those recording pairs that were indeed different relative to the non-proficient professionals, and the nonproficient professionals were generally more successful than the students. Second, listeners generally were not able to correctly identify the more double sloped recording with any consistency when the impulse responses associated with the two recordings Ermann-13

14 were more similar to one another (0% open apertures versus 0.1% open apertures, for instance). Listeners were, however, more able to correctly identify the recording that was more double sloped when the impulse responses differed greatly from one another (0% open apertures versus 1.0% open apertures, for instance). Again, here, proficient listeners performed the best. Finally, on the question of whether listeners prefer the more double sloped recording, the evidence is not as conclusive. It does appear that proficient listeners prefer the recording that is less double sloped, while non-proficient professionals and students were split or unsure of which they prefer. In other words, it may be that, when given a choice, listeners do not prefer the double sloped decay! Figure 16. Pair of auralizations used to detect perception and preference of the double sloped decay. Click on the icon to hear each decay. Subjects were asked (1)if they could detect a difference between the two recordings, (2) if a difference exists, which one is more likely to be double sloped or is more dramatically double sloped? and (3) which one is preferred? It should be noted that the listening test portion of this inquiry is ongoing and only preliminary results are reported here. The latest listening test administered involve those with truncated decays so that in identifying perception and preference, listeners must rely on running music and are denied the stop chords at the end of musical passages. It should be noted that the perception and preference results published here are consistent with others reported in separate studies.12,13,14 8. Conclusions Simply designing a coupled volume concert hall is not enough to ensure that the space will produce a double slope decay; producing a double slope decay will not ensure that it will be heard everywhere in the space, nor will it ensure that it will be heard above the background noise; and hearing a double slope decay will not ensure that it will be perceived or even preferred. As a rule of thumb for purposes of schematic design, to achieve a double sloped decay in the concert halls modeled here, a designer should (1) create a coupled volume that is at a minimum, four to seven times as reverberant as the main hall, and possibly much more, (2) maintain small aperture sizes, less than 1.5% of the available surface area of the main hall, (3) keep background noise to the level of the threshold of hearing, and (4) inform the client that the double sloped effect, sought after as a way to reconcile the competing qualities of reverberance and clarity, may not be able to be perceived and may not be preferred. Ermann-14

15 The idea of a kinetic architecture, where spatial composition, materiality, and haptic perception may profoundly impact the aural environment is an exciting one to many acousticians and designers. This natural excitement is credited with the genesis of the studies outlined here. However, the fickleness of the system, where small changes in architecture leverage large changes in acoustics, and the unconvincing value of the double slope itself, assuming it is successfully achieved, should give pause. This is not to condemn the entire arena of music-piece -specific adjustable acoustics, nor does it dismiss the other possible uses of a coupled volume (absorptive chamber when velour banners or drapes are deployed, place for lighting effects, auxiliary stage for performance, giver of added reverberance when doors are opened). This line of research simply frames the double slope decay itself as fickle-at-best and unwanted-at-worst. 8. Acknowledgements As this is a compendium, many of the images are based on those found in references 6, 7, and 8. Thanks to Marty Johnson, Bill Yoder, Rebecca Stuecker, and Chiss Yoder for their help researching this paper. 9. Endnotes [1]Kuttruff, H., Room Acoustics, Elsevier, New York, [2] Johnson, R., Kahle, E., Essert, R., Variable coupled volume for music performance, Conference proceedings from Music and Concert Hall Acoustics, 1995, p [3] Xiang, N., Goggans, P., Li, D., Measurement of decay times in coupled spaces, 141 st Meeting of the Acoustical Society of America, Chicago, Journal of the Acoustical Society of America, Vol. 109, 2001, p [4] Based on dozens of interviews and presentations, both formal and informal, with acoustic consultants involved in the design of coupled volume concert halls, [5] Please see the following three references for extensive literature reviews on the concepts presented in this paper. [6] Ermann, M., Coupled volumes: aperture size and the double sloped decay of concert halls, Building Acoustics, Vol. 12, No. 1, 2005, pp [7] Ermann, M., Coupled volumes: secondary room reverberance and the double sloped decay of concert halls, Building Acoustics, Vol. 12, No. 3, 2005, pp [8] Ermann, M., Johnson, M., Exposure and materiality of the secondary room and its impact on the impulse response of coupled volume concert halls, Journal of Sound and Vib., Vol. 284, 2005, pp [9] Cremer, L. and Müller, H., Principles and Applications of Room Acoustics, Applied Science, London, [10] Egan, D, Architectural Acoustics, McGraw-Hill, New York, [11] Beranek, L., Concert Halls and Opera Houses, Springer-Verlag, New York, [12] Atal, B.S., Schroeder, M.R., and Sessler, G.M., "Subjective Reverberation Time and its Relation to Sound Decay," Proc. Fifth Intl. Congress on Acoust., [13] Picard, D. Audibility of Non-Exponential Reverberation Decays, A thesis submitted to Rensselaer Polytechnic Institute (Greene Building, th St., Troy, NY 12180). [14] Bradley, D., Analysis of parameter effects on sound energy decay in coupled volume systems, A thesis submitted to the University of Nebraska, Ermann-15

16 PFbook_outline_text.indd 3 5/21/06 4:38:27 PM

17 PFbook_outline_text.indd 4 5/21/06 4:38:27 PM

18 PFbook_outline_text.indd 5 5/21/06 4:38:28 PM

19 PFbook_outline_text.indd 11 5/21/06 4:38:44 PM

20 PFbook_outline_text.indd 30 5/21/06 4:39:06 PM

21 PFbook_outline_text.indd 31 5/21/06 4:39:08 PM

22 PFbook_outline_text.indd 32 5/21/06 4:39:11 PM

23 PFbook_outline_text.indd 33 5/21/06 4:39:14 PM

24 PFbook_outline_text.indd 34 5/21/06 4:39:16 PM

25 PFbook_outline_text.indd 35 5/21/06 4:39:20 PM

26 PFbook_outline_text.indd 36 5/21/06 4:39:21 PM

27 PFbook_outline_text.indd 37 5/21/06 4:39:24 PM

28 BUILDING ACOUSTICS Volume 13 Number Pages Mapping the Sound Field of a 400 Seat Theater Michael. Ermann a, M. R. F. Kidner b, and D. Mennitt c a School of Architecture and Design, Virginia Tech (0205), Blacksburg, VA USA b,c Vibration and Acoustics Laboratory, Virginia Tech, Blacksburg, VA USA b Now at Department of Mechanical Engineering, University of Adelaide, Adelaide SA, Australia 5005 a mermann@vt.edu; b mkidner@mecheng.adelaide.edu.au; c dmennitt@vt.edu (Received 19 September 2005 and accepted 12 March 2006) ABSTRACT To what extent does a receiver s location in a room impact what is heard and can that impact be modeled? For this line of study, acoustic parameters of a 400 seat theater were mapped as a function of position at every-seat resolution to quantify the spatial disparity of a sound field. A 1:20 scale model and a ray-tracing software model were then constructed. This study seeks to (1) relate the sound decay rate and pressure level measured at one theater seat to that measured at others, (2) compare the inter-seat sound field maps measured across the frequency spectrum with an eye toward the Schroeder cut-off frequency, and (3) determine the models reliability in replicating the measured inter-seat variation. Keywords: Room Acoustics, Spatial Variation, Computer Modeling, Scale Modeling 1. INTRODUCTION The individual location of each listener in a room can have a profound impact on what is heard. Prediction of this variation is important, as the consistency of the acoustic response within a space is indicative of its perceived quality. 1 Indeed Cremer and Müller s 2 observation that a study of the different impulse responses in the same hall makes it astonishing that one would dare to speak of the acoustics of a hall at all without at least indicating where the observation was made, prompted this study. While numerous authors have studied room acoustics parameters of performance spaces, the variation within a single space has been largely ignored, with few exceptions. Siebein et al. 3 studied a wide variety of rooms and found the spatial disparity of acoustical parameters dependant on size, shape, and other architectural features. From the sound level distribution in auditoria, Barron and Lee 4 found that the total reflected sound level is related to source-receiver distance and can be predicted in reasonably diffuse halls with certain characteristics. This paper reports on experiments conducted to chart the spatial distribution of measured acoustic parameters of the Lyric Theater (and compare them with those

29 200 Mapping the Sound Field of a 400 Seat Theater predicted by a scale model and a CATT-Acoustic 5 software model). The Lyric Theater in Blacksburg, Virginia seats 400 in a volume of approximately 3,000 cubic meters. Its shoebox shape measures approximately 20 meters long, 15 meters wide and 10 meters high with a balcony along the rear wall. It has been recently restored to its original 1929 condition and now hosts cinema and live music. To study this spatial variation in parameters, the impulse response of the Lyric Theater was measured in each seat of the half-plan. The data was post-processed to obtain early decay time, EDT, and a modified strength, G m, at a one-seat resolution. The results depict substantial decay time discrepancies in low frequencies, inter-seat decay time convergence above the Schroeder cut-off frequency, the spatial extent of the acoustic near-field of the sound source used, an observable increase in sound pressure level in the aisles relative to that found in the seating areas, an acoustic shadow of decreased sound pressure level below the balcony, and a mysterious region of decreased sound pressure level, one seat deep running the half-width of the room. 2. SOUND FIELD MAPPING OF THE THEATER 2.1. Mapping method Because the room is nearly symmetric about its longitudinal axis, the impulse response was measured in the house-left seats only. A simple mechanical impulsive source, made from two boards of medium density fiberboard (0.5 m 0.5 m 0.02 m) and spring loaded to clap together when remotely triggered, excited the room. The source was measured in an anechoic chamber and found to be sufficiently repeatable, omnidirectional, and broadband (see Figure 1). A 16 meter linear microphone array quickly measured the response at multiple locations within the main room. Fifteen omnidirectional Acousticel TMS-130A and TMS-130B microphones were spaced at just over 1 meter. This spacing corresponds to the distance between rows of seats. National Instruments hardware acquired the microphone signals along with a trigger channel for the mechanical source. The source was set on vibration isolation, which was in turn set on the proscenium stage. Two people suspended the microphone array along the length of the theater so that each microphone was above one seat (see Figure 2). In this way, a data set could be collected for an entire column of seats. Data were also collected along the aisles. Each measurement simultaneously recorded four seconds of data at a sampling rate of 16 khz and three measurements were taken at each position for a total of 675 measurements at 225 positions. The impulse responses were filtered into full-octave bands from 63 Hz to 4000 Hz to obtain 4,725 data points for mapping. A nonlinear model of an exponential decay with a stationary noise floor 6 estimated the early decay time. A simple linear fit was also performed. 7 The time histories of data sets that yielded outlier reverberation times, far from the mean value, were plotted and compared visually to their calculated decay curves to ensure an accurate fit. Modified relative strength, G m 4, was evaluated for each position. Note this parameter is labeled modified as the source reference level is measured in an anechoic chamber at a distance of 1 meter, rather than the standard 10 meters. While this method deviates from

30 BUILDING ACOUSTICS Volume 13 Number Figure 1. Directivity plots (sound pressure level at one meter, referenced to 20 µpa) of the mechanical source as measured in an anechoic chamber. (a) 125 Hz octave band. (b) 250 Hz octave band. (c) 500 Hz octave band. (d) 1000 Hz octave band. (e) 2000 Hz octave band. (f) 4000 Hz octave band.

31 202 Mapping the Sound Field of a 400 Seat Theater Figure 2. A microphone snake was stretched along a column of seats to obtain 15 simultaneous measurements. ISO standard 3382, the authors believe the results presented here remain valid because they focus on inter-seat variances rather than absolute parameters or comparisons across multiple spaces Mapping results Early Decay Time, EDT, and modified strength, G m, were obtained for each position at octave bands from 63 Hz to 4000 Hz. Figure 3 (a through d) maps the decay times to the seats on the orchestra level. The general trend marks an average EDT decrease with increasing frequency from 3.4s in the 63 Hz band to 2.2s in the 250 Hz. Note that the average reverberation time is high for cinema; indeed, the researchers observed that speech can be unintelligible when listening to a movie. More important for this line of inquiry, the map reveals significant inter-seat decay time variation at low frequencies in some places, and similar inter-seat decay times in others. This is particularly notable at the 63 Hz octave band, but also plainly visible at the 125 Hz octave band. Figure 4 charts the 63 Hz impulse response of two sets of adjacent seats: the top image corresponds to two adjacent seats with similar decay profiles, while the bottom image corresponds to another pair in the same row, having strikingly less similar profiles. In reviewing individual impulse responses, the researchers found the room littered with these kinds of pairings.

32 BUILDING ACOUSTICS Volume 13 Number a e b c d f g h (s) (s) Figure 3. i j k l (s) Reverberation time half-plan maps for theater, scale model and geometric model. The four columns correspond to 63 Hz, 125 Hz, 250 Hz and 500 Hz bands respectively. The top row (a d) depicts the theater measurements, the middle row (e h) depicts the scale model simulations, and the bottom row (i l) depicts the CATT-Acoustic simulations. Solid white holes (e,f) indicate no data due to poor signal to noise ratio and grey maps indicate lack of available data as explained in the text.

33 204 Mapping the Sound Field of a 400 Seat Theater db Time (s) db Figure Time (s) 63 Hz octave band impulse responses measured at two pairs of adjacent seats in the same row. Similar impulse responses (top) and dissimilar impulse responses (bottom). Figure 5 describes three very different 63 Hz impulse responses at three nonadjacent theater seats and questions the appropriateness of attempting to describe the reverberation time of a room as a single number if the octave band in question belongs to the modal world found below the room s Schroeder cut-off frequency 2. The plots in Figure 5 (a,b) reveal smooth but vastly unequal decay rates in two seats. Figure 5 (c) describes a fast initial decay, followed by late-arriving reflections that yield a short EDT a single number that does not capture the effect of the distinct echoes that arrive after 0.5 seconds. Higher than the Schroeder frequency, at octave bands above 125 Hz,

34 BUILDING ACOUSTICS Volume 13 Number a 10 db EDT = 3.5s Time (s) 10 0 b 10 db EDT = 6.5s Time (s) 10 0 c 10 db EDT = 0.91s Figure Time (s) Impulse responses, (solid line) in the 63 Hz third octave band measured at distinct seats in the theater. Non-linear, (dash-dot line), and linear, (dashed line), fits to the response decay. Estimation of point at which decay reaches the noise floor (dotted line).

35 206 Mapping the Sound Field of a 400 Seat Theater a b (db) (db) c d (db) (db) Figure 6. Modified strength half-plan maps (G m ). (a) Theater measurement at 1000 Hz. (b) Theater measurement at 4000 Hz. (c) CATT-Acoustic computer model simulation at 1000 Hz. (d) CATT-Acoustic computer model simulation at 4000 Hz.

36 BUILDING ACOUSTICS Volume 13 Number the modal density is sufficient to approximate a uniform sound field with comparable decay times measured across the audience plane and it is only here, comfortably above the Schroeder Cut-off frequency, where one can confidently attribute a single reverberation time to the entire room (see Figure 3 (c,d)). Maps of modified strength in the 1000 Hz and 4000 Hz octave bands measured in the theater are shown in Figure 6 (a,b). The high strength region near the stage (top of the figures) clearly indicates the source position and defines the spatial extent of the near-field. As expected, the balcony creates an acoustic shadow on the orchestra level that can be seen toward the bottom of the same figures. Note the observable G m level contrast between the aisle and seating areas in the 1000 Hz octave band, where unoccupied upholstered seats typically have the highest absorption coefficient. 7 The researchers are wholly unable to explain the narrow band of depressed G m seats mapped at as much as 5 db below the seats directly in front of and behind them which stretch the entire width of the room near the midway point between the front of the stage and the rear of the room. (Perhaps both the low-strength band and high-strength band are part of the acoustic shadow created by the balcony, only the high strength band has been layered by second order sound reflections from the balcony face.) This phenomenon was found at all frequencies, but is especially sharp at the higher octave bands (see Figure 6 (b)). Note that the geometric acoustics employed in the software CATT-Acoustics model predicted the existence, relative magnitude, and location of the loudness depression (see Figure 6 (c,d)). Readers who have measured this phenomenon before, or who wish to speculate on its cause, are encouraged to contact the authors. 3. MODELING THE THEATER 3.1. Modeling method A simple model of the theater was created in CATT-Acoustic, which uses geometric acoustics to predict octave-band echograms based on a 3D CAD model of a room. 5 Absorption coefficients from reference texts were used 2 and the source location was the same as that used for the full scale tests. A 1/20 scale model of the theater was built 8,9 with medium density fiberboard representing the plaster surfaces, felt representing the audience seating areas, and golf tees representing the seats themselves (see Figure 7). The measurements were made using air as the propagation medium and the effect of the sound absorption of air on the decay rate was compensated for during post processing of the data. 10,11 An AcoPacific /4 microphone and type 4016 preamp, measured the impulse response at each seat in the half-plan. An array of three fabric dome tweeters was used as the source and was placed on the stage. WinMLS software 12 measured impulses at 221 receiver locations over a scaled bandwidth from 40 Hz to 250 Hz. The modeling fidelity in this set of experiments should be noted. Both the computer CATT-Acoustic model and the scale model rely on approximations of absorption coefficients based on interviews with the theater staff, the limited 1920s era construction drawings, the 1990s era renovation drawings available, and observations

37 208 Mapping the Sound Field of a 400 Seat Theater Figure 7. Photograph of the theater scale model viewed toward the stage. The rear balcony can be seen protruding in the image foreground. made in the room. The researchers believe, nonetheless, that results from these models are (1) valid, because the results are specifically measuring the models ability to accurately distinguish decay times and loudness between seats rather than predict the absolute sound field at a given point in space, (2) relevant, because designers and researchers must know if they can rely on their models to identify and locate sourcereceiver configurations that produce outlier sound fields, and (3) accurate, because the types of models reported on here are similar to those that would be tested during the room design phase of a building project. Moreover, while air absorption renders the scale model unstable at predicting room sound fields above the 125 Hz octave band, it is precisely the stable lower frequencies that are of interest because it is there that the inter-seat in situ room measurements differ wildly in decay time Modeling results These studies suggest that for this room, the models created are limited in their usefulness and accuracy in predicting the measured inter-seat variations. However, each of the two modeling techniques may be well-suited to predict at least one of the sound field parameters explored. At 63 Hz, the scale model truthfully depicts the kind of inter-seat decay time variations measured in situ; at 125 Hz, it accurately identifies

38 BUILDING ACOUSTICS Volume 13 Number the Schroeder frequency where the modal sound field begins to cross over to a diffuse and consistent sound field, see Figure 3(a, b, e, and f). Because of the inherent scaling limits created by air above 5000 Hz, the model cannot simulate the scaled frequencies above the 125 Hz octave band where room measurements suggest a truly diffuse sound field, nor is the scale model always able to predict the location of outlier sound fields. The CATT-Acoustic model, with its reliance on geometric acoustics, is not able to predict the effect of low frequency modes measured in the room, see Figure 3 (a, b, i, and j). This is consistent with acknowledged limitations of ray-tracing based numerical model validity at low frequencies and is the reason the software does not attempt to simulate below 125 Hz. The computer model also predicts shorter reverberation times than are measured in the room itself. The CATT-Acoustic model is, however, adroit at predicting the spatial differences in loudness measured in the theater, see Figure 6. Not only does it identify and locate the acoustic shadow cast at the house-rear portion of the main floor, it successfully predicts the mysterious horizontal band of lower sound strength located midway between the stage and the rear of the room. 4. CONCLUSION The haptic 13 impact of spatial location plays a significant role in the sound field measured in a theater, but its role as a differentiator is limited by considerations of frequency and architectural composition. Measurements made at 225 locations in octave bands below the Schroeder cut-off frequency expose a wildly modal world with little worth attributed to single-number room decay time averages. By contrast, measurements made above the cut-off frequency reveal rigidly uniform impulse responses, and therefore, identical decay times and are good candidates for single-number room decay time averages. Where inter-seat decay time measurements vary largely as a function of the frequency filter applied, strength measurements vary more as a function of location. The every-seat measurements establish clear boundaries between the acoustic near- and far-fields, define higher sound level regions in aisles and lower sound level regions beneath balconies, and describe acoustical-spatial sound level peculiarities, likely influenced by the particular architecture of a room. The results also suggest that that computer and scale models are limited in their ability to accurately predict the response of this room in fine detail with two exceptions. First, scale modeling effectively represents the variability of low-frequency decay times across receiver locations, and second, computer ray-tracing modeling effectively locates areas of increased and decreased strength within a room. ACKNOWLEDGMENTS The authors would like to thank Dr. Marty Johnson, Julie Redenshek, Adam Tawney, and Joe McCoy for their contributions.

39 210 Mapping the Sound Field of a 400 Seat Theater REFERENCES [1] Beranek, L. (1996). Concert and opera halls; How they sound, Acoustical Society of America, Woodbury, NY. [2] Cremer, L., and Müller, H. (1982). Principles and applications of room acoustics, Applied Science Publishers, London. [3] Siebein, G. W., Chiang, W., Cervone, R. P., Doddington, H. W., and Schwab, W. K., (1992). Acoustical measurements in lecture halls, theaters, and multi-use rooms, J. Acoust. Soc. Am. Vol. 92, pp [4] Barron, M. and Lee, J., (1988) Energy relations in concert auditoriums, J. Acoust. Soc. Am. Vol. 84, pp [5] anon., CATT Acoustics Software for room acoustic consulting and virtual reality. [6] Karjalainen, M., Antsalo, P., Mäkivirta, A., Peltonen, T., and Välimäki, V., (2001). Estimation of modal decay parameters from noisy response measurements, AES 110 th Convention, Amsterdam, The Netherlands, May. [7] Bies, D. A. and Hansen, C. H., (2002). Engineering Noise Control, Spon Press, NY. [8] Spandöck, F., (1934). Akustische modellversuche, Ann. Der Phsyik Vol. 20, pp [9] Xiang, N. and Blauert, J., (1993). Binaural scale modeling for auralisation and prediction in auditoria, Applied Acoustics, Vol. 38, pp [10] Doddington, H., (1995) An investigation of possible approaches to correction of acoustic impulse responses for excess air absorption at full scale and model conditions, University of Florida Department of Aerospace Engineering Mechanics and Engineering Science Report on NSF Sponsored Project MSS [11] Evans, L. B. and Bass H. E., (1972). Tables of absorption and velocity of sound in still air at 68 F (20 C), Wyle Laboratories report wr72-2. [12] anon., Win MLS, Measurement tool for audio, acoustics and vibrations. [13] O Neill, M., (2001) Corporeal Experience: A Haptic Way of Knowing. Journal of Architectural Education. Association of Collegiate Schools of Architecture/MIT Press, September.

40

41 The Acoustics of Multifamily Housing For Architects, Builders, Developers, and Occupants 2010 Michael Ermann and the Acoustical Society of America Table of Contents IMPACT NOISE 2 COMMUNITY NOISE 56 DIFFERENT THAN AIRBORNE NOISE 3 POSSIBLE IN WOOD? 3 IMPACT INSLATION CLASS (IIC) 4 ACHIEVING HIGHER IMPACT NOISE PERFORMANCE 5 SHORTCOMINGS OF THE IIC RATING 10 IMPACT NOISE CHECKLIST 14 ACOUSTICAL PRIVACY 19 AIRBORNE NOISE 19 LOW FREQUENCY SOUND 20 FLANKING 20 TRANSMISSION LOSS (TL) 21 SOUND TRANSMISSION CLASS (STC) 22 ACHIEVING HIGHER ACOUSTICAL PRIVACY 23 PRIVACY CHECKLIST 29 ANNOYANCE 56 SOURCES 56 IT S THE WINDOWS AND DOORS, NOT THE WALLS 57 DISTANCE 59 COMMUNITY NOISE CHECKLIST 60 MECHANICAL EQUIPMENT NOISE 67 SOURCES 67 LOCATION.LOCATION.LOCATION 67 CENTRAL SYSTEMS 68 DUCTED RETURNS TOO 68 MECHANICAL SYSTEM NOISE CHECKLIST 70 PLUMBING NOISE 77 NUISANCE 77 AMPLIFICATION 77 TURBULENT FLOW AND CAVITATION 77 WATER HAMMER 78 DEFECTIVE PARTS 78 EXPANSION AND CONTRACTION 78 DRAINING WATER 78 RUNNING WATER 78 PLUMBING NOISE CHECKLIST 82

42 Noise and Privacy Quiz From an acoustical point of view, how could this apartment be improved (answer in back of booklet)? Adjacent Apartment Trash Shoot D W A/C Compressor Units AHU Elevator Television Adjacent Apartment 1

43 ACOUSTICAL PRIVACY Airborne noise. Airborne sound transmission between rooms is generated by people talking or shouting, equipment running, and the sound amplification associated with stereos and television sets. Sound energy moves through the air to the wall assembly and floor-ceiling assembly, where it is radiated through the structure to the other side. Generally, occupants find louder noises and noises that start and stop or fluctuate to be particularly annoying but, as in the case of a dripping faucet, occupants may even be annoyed by mere audibility. Because people generally are annoyed by sounds that are (1) created by sources they are not involved with, (2) unpredictable, (3) perceived as unnecessary, and (4) generated by people toward whom they don t have a favorable attitude, airborne sound isolation between residential units can be particularly vexing. 19

44 Privacy Low frequency sound. One can sometimes hear the bass beat of a car stereo for what seems like a two block radius, but can t make out the lyrics of the song on that car stereo until the car door is opened in close proximity. In this way low frequency sound energy travels far and easily moves through some building assemblies, particularly those that are light weight. In the context of multi-family housing, the low-pitched hum of mechanical equipment and the amplified bass notes associated with stereos and TVs played too loudly pass through many wall and floorceiling assemblies barely attenuated. Designers beware: Sound Transmission Class (STC) is an easy method of comparing the airborne acoustical isolation of building assemblies, and is effective at summarizing performance related to speech privacy, but is inadequate at summarizing performance associated with low frequency amplified music and mechanical equipment hum. Flanking. Keeping sound out is like keeping water out. The overall performance of an assembly is more a function of its performance at the weakest point, not the average, therefore a small leak can render an assembly ineffective. This can only really be combated by thorough detailing and construction supervision, particularly where the floor meets the wall in wood construction. 20

45 Privacy Electrical outlets facing opposite units should not occupy the same inter-stud wall cavity; niches for bookshelves or fire extinguishers should be well detailed; cabinets and medicine cabinets should not be designed back-to-back; conduit, pipes, ducts, and other penetrations should not move through sensitive assemblies, and when they do, the wall should be sealed at the penetration. Generous quantities of caulk should be used, particularly where drywall meets the subfloor. Designers beware: Sound Transmission Class (STC), while helpful in making comparisons, is only a description of the performance of the wall or floor ceiling assembly and does not account for sound flanking or installation quirks, better considered with whole-system-thinking. Transmission Loss (TL). The airborne sound-insulating properties of a building element can be quantified by measuring its Transmission Loss (TL). The higher the TL values, the more robust the assembly is at attenuating the penetration of sound. Tested building elements will have Transmission Loss values at each of several frequency bands, from low-pitched tones to higher-pitched tones, and because airborne sound attenuation is only as good as the weakest link in the chain, a high value in one octave band will not necessarily make up for a low value in another. 21

46 Privacy When accounting for low frequency noises associated with amplified music, transportation noise, and mechanical equipment rumble, special attention should be paid to select a building assembly with high 63Hz, 125Hz, and 250Hz octave band TL values. Sound Transmission Class (STC). For easy comparison of building elements, Sound Transmission Class (STC) provides a single number rating. Like Transmission Loss, the higher the building assembly s STC rating, the more effective the assembly is at preventing the transmission of sound. But unlike Transmission Loss which links multiple values to a single assembly to account for performance variation between octave bands, Sound Transmission Class combines multiple values from across the frequency spectrum, weights them, and compiles one number to address all the octave bands. The casualty of this simplification is low frequency performance, which STC does not sufficiently relate. 22

47 Privacy Achieving higher acoustical privacy in design. Building elements that are massive, airtight, and structurally discontinuous perform the best. 1. Mass. In general, the more dense the material, the more noise it will attenuate for a given thickness. For example, solid concrete is a better sound insulator than solid wood (of equal thickness), and a thicker concrete wall will attenuate sound more effectively than a thinner concrete wall. Multiple layers of thicker gypsum board on the outside of a wall outperform a single thinner layer. Doubling the weight of the wall by adding a layer to both outer surfaces can increase STC by more than 5 points. 2. Airtight. The best assemblies for maintaining acoustical privacy have surfaces with few or no interruptions and are sealed. A 1/16 inch (2mm) crack 16 inches (400mm) long will reduce a 9 foot long STC 50 wall to an STC 40 level. Try not to interrupt party walls or other acoustically sensitive walls with doors and systems such as electrical outlets, doorbells, fire alarms, intercoms, cabinets, phone jacks, conduit, ducts, grilles, and pipes. 23

48 Achieving Privacy 3. Structurally discontinuous. A cavity wall outperforms a solid wall of equal weight. Maintaining a resilient connection between the wall structure and the panel(s) on one side of an assembly renders the assembly limp, increasing performance relative to rigid-mounted panels. A small room, like a closet, can be designed as a buffer zone, provided the small room extends the full length of the wall in question. 4. Sound absorbing materials in the cavity. In lightweight walls especially, fuzzy material such as fiberglass can improve the performance of a wall significantly, but sound absorbing insulation is not a substitute for mass and air-tightness. Indeed, the true benefit of a wall or ceiling cavity filled with sound absorbing materials can only be realized with structural discontinuity (item 3, above). 24

49 Principles of Sound Transmission Performance Baseline Better Single Layer Gypsum Board Multilayer Gyp. Bd. with Staggered Panel Joints Standard Block Wall Increased Mass Grout Filled Block Wall Single Surface Use of Airspace Two Surfaces with Cavity 25

50 Principles of Sound Transmission Performance Baseline Better 16 o.c. 16 o.c. 24 o.c. Standard Stud Wall (studs 16 o.c.) Limp (Wide Spacing Between Studs) Stud Wall (studs 24 o.c.) Standard Stud Wall Staggered Studs Standard Stud Wall Structural Discontinuity Double Stud Wall 26

51 Principles of Sound Transmission Performance Baseline Better Sound-absorbing blanket (to reduce coupling between block layers) One Wythe of CMU Two Wythes of CMU with Cavity Standard Stud Wall With Resilient Brackets Sound-absorbing blanket (friction fitted between vertical wood furring) CMU Wall Metal channel (to resiliently support gyp. bd.) Structural Discontinuity Non-hardening caulking (to seal perimeter of gyp. bd.) CMU Wall with Furring, Resilient Clips and Gyp. Bd. 27

52 Principles of Sound Transmission Performance Baseline Better Standard Stud Wall Absorption in the Airspace Standard Stud Wall with Insulation in the Cavity Back-to-back outlets > 2 ft. separation so at least one stud will be between outlets Fibrous insulation (to deaden cavity airspace) Sound leak Non-hardening caulk Caulked outlet box openings and perimeter joint (to prevent sound leaks) Stud Wall with Outlets in the Same Cavity Airtight Stud Wall with Outlets in Separate Cavites 28

53 Privacy Checklist Early Design 1. Program and space-plan with acoustics in mind. Keep the quiet spaces and noisy spaces far away from one another, not only in plan, but in section as well. This is by far the most effective, least costly, and most architectural of the solutions available. 2. Recognize that some rooms are simply too noisy to be adjacent to noise-sensitive spaces, period. 3. Design rooms that are not noise sensitive as buffer zones between noisy spaces and quiet spaces. For instance, one might place a row of closets, utility rooms, vestibules, and bicycle storage rooms between units. Experience suggests that the room two-doors-down is much quieter than the adjacent room. 4. Recognize that within a dwelling, an open plan will not afford acoustic privacy. For instance, if the watching-television space and dining space are in plain sight of one another, no acoustical treatment will provide for meaningful aural separation between the two. 29

54 Privacy Checklist Assembly Performance 1. Do not confuse sound absorption with sound transmission loss. A material s sound absorption or an assembly s impact noise performance has little and often no effect on its sound transmission properties. Noise Reduction Coefficient (NRC), and Impact Isolation Class (IIC) are independent of sound Transmission Loss (TL) and Sound Transmission Class (STC). Acoustical ceiling tile typically has no meaningful effect on the transmission of sound between occupied rooms. 2. Be conservative and specify an assembly that wellexceeds the minimum required. Sound Transmission Class (STC) regularly varies +/- 2 points from measurement to measurement. Some vary more. Know that some manufacturers, when publishing results from acoustic tests, will put forth the highest score ever achieved rather than a typical score. 30

55 Privacy Checklist Assembly Performance 3. If measuring as-built assembly performance in the field, know that field test values usually come in below those measured in the laboratory. This is because, in situ, construction irregularities and flanking paths compromise the robustness of the more controlled samples tested as panels in the lab. Nominally, one may assess a penalty of five points when translating to field measurements if there is the clear recognition that in some cases, the penalty may be more than ten points. 4. Know that sound more easily passes between units if exterior windows of adjacent units are located near one another. 5. Specify massive, airtight, and structurally discontinuous assemblies for walls and floorceilings. 6. Two layers of gypsum board on one or both sides of a wall outperforms one layer. Three layers outperforms two. When layering, stagger the gypsum board seams so they do not align. Attach layers of gypsum board together with visco-elastic adhesives rather than rigid curing adhesives or screws. 31

56 Privacy Checklist Assembly Performance 7. In concrete block construction, know that the densities of available products vary, and that heavier block outperforms lighter block. To increase performance, fill cells with sand or mortar. Staggering wood studs so that each stud only makes contact with one wall surface is more effective than normative stud wall construction where each stud makes contact with both surfaces. A double stud wall with separate sole plates separated by one inch (25mm) is better still, and recommended practice for party walls in multifamily light wood frame construction. In double stud construction, only affix gypsum board to the outside of the stud assembly, keeping the cavity between the exposed surfaces free of sheathing surfaces. 8. If structural considerations allow it, increase the spacing between studs from the normative 16 inches (400mm) to 24 inches (600mm). The same is true for the spacing of resilient channel. Note that this approach is not recommended for joists in floor-ceiling assemblies as it is associated with an increase in impact noise. 32

57 Privacy Checklist Assembly Performance 9. Use resilient connections to attach a ceiling, or one surface of a wall, to the structure. Flanking issues can arise when improperly long screws are used that short circuit the resilient connection by biting directly into the stud or joist. Unless proper supervision will be present during installation to ensure that either short screws are used or that panel attachments are only made to the resilient channel between the structural members, specify a resilient channel system with clips. Cabinets or baseboard trim attached directly to the studs can short circuit the isolation provided by the resilient connections. 10. Know that lighter-weight steel studs are limper, and therefore more effective, than heavier-weight steel studs, and more effective than wood studs. 11. Coupling a gypsum board wall to a concrete block wall with furring increases performance of the block wall. Insulation between the furring and resilient clips to attach the gypsum wall surface is better. A double block wall with a fiberglass insulation filled cavity between withes is better, still. 33

58 Privacy Checklist Avoid Flanking 1. Use generous quantities of non-hardening caulk to ensure a tight seal (1) where wall board meets the floor, (2) where ceilings meet walls, and (3) at penetrations from ducts, electrical outlets, pipes, etc.. To seal larger holes, use firestop putty. 2. Electrical outlets, phone jacks, cable wall jacks, pipe and duct penetrations, etc. should not occupy the same inter-stud wall cavity. When possible, do not locate these flanking opportunities in party walls. 3. Use plastic vapor-barrier electrical outlet boxes: they outperform metal electrical outlet boxes in acoustic tests. 34

59 Privacy Checklist Avoid Flanking 4. Niches for bookshelves or fire extinguishers should be well considered, kitchen cabinets and medicine cabinets should not be designed on party walls. 5. Conduit, pipes, ducts, and other penetrations should not move through sensitive assemblies, and when they do the wall should be sealed at the penetrations. This may require use of sleeves, grout, caulk, and packing. 6. Run walls all the way to the underside of the slab do not terminate the wall short of the slab simply because it has gone through a hung ceiling and will appear proper when viewed from within the room. 35

60 Recommended: Party Wall Block Construction STC 56 5/8 (16mm) gypsum board, screwed to channels Resilient clips with hat channel 8 x 8 x 16 3-cell lightweight concrete masonry units (34 lbs/block) Resilient clips with hat channel 5/8 gypsum board, screwed to channel Concrete Construction STC 58 8 (200mm) thick flat concrete panel (95 psf) Wood Construction STC 62 5/8 (16mm) gypsum board, screwed 12 (300mm) o.c. (2 Layers) Double row of 2 x 4 studs, spaced 16 (400mm) o.c. on separate plates spaced 1 (25mm) apart 1 1/2 (40 mm) thick sound attenuation blanket 5/8 gypsum board, screwed 12 o.c. (2 layers) 36

61 Privacy Checklist Avoid Flanking 7. Conduct preliminary tests of the sound insulating effectiveness of a wall or floor-ceiling prior to painting and final completion. Visually inspect for cracks or gaps in surfaces. Your ears are excellent acoustic instruments: run a noisy device such as a vacuum cleaner or power tool in a closed room and listen in the adjacent room for locations where the noise is leaking through. A physician s stethoscope can help with this too. 8. Locate building control joints where needed. The proper use of control joints to account for differential expansion and contraction will minimize the future cracking of walls and therefore minimize the potential for sound flanking through cracks. Because control joints offer vibration isolation as well, locate rotating and reciprocal-motion equipment such as pumps, compressors, chillers, cooling towers, exhaust fans, air handlers, washers, and dryers on a separate building segment separated from dwelling unit segments with a building control joint. 37

62 Privacy Checklist Avoid Flanking 9. Where critical adjacencies exist between noisy and quiet spaces and there must be a door connecting the two, design a vestibule between two doors instead. Solid wood doors and hollow metal doors outperform hollow wood doors. Gasketed doors outperform those without gaskets. Louvered doors provide almost no acoustical separation. Highperforming, proprietary, acoustical doors are available, but generally very expensive. 10. Ensure that sound attenuating blanket insulation at least two inches (50mm) thick is specified for cavities in assemblies designed to keep out noise. Supervise installation to ensure the insulation doesn t sag and isn t compressed. 11. Know that firestops and gusset plates can short circuit the separation intended for double wall constructions if they rigidly connect the two walls intended to be structurally separated. 38

63 Flanking Transmission of Airborne Noise A I E B J K C L D M O N F H P R Q G A. Open Plenums Over Walls, False Ceilings B. Unbaffled Duct Runs w/o Min 2 Elbows C. Outdoor Path, Window to Window D. Continuous Unbaffled Inductor Units E. Hall Path, Open Vents F. Hall Path, Louvered Doors G. Hall Path, Openings Under Doors H. Open Troughs in Floor-Ceiling Structure I. Poor Seal at Ceiling Edges J. Poor Seal around Duct Penetrations K. Poor Mortar Joints, Porous Masonry Block L. Poor Seal at Sidewall, Filler Panel, etc. M. Back-to-Back Cabinets, Poor Workmanship N. Holes, Gaps at Wall Penetrations O. Poor Seal at Floor Edges P. Back-to-Back Electrical Outlets Q. Holes, Gaps at Floor Penetrations R. Back-to-Back Phone/Data Outlets 39

64 Floor Path Flanking Sound Leak Flanking Path Fair Wall Continuous OSB Floor Between Units Floor With Joists Parallel to Wall - Sound Transmission Class (STC) With second layer Plywood, No Mechanical Contact with Wall + With Gypsum Concrete on Resilient Mat No Mechanical Contact with Wall Better REFERENCE: J.D.Quirt and T.R.T. Nightingale, Airborne Sound Insulation in Multifamily Buildings, National Research Council Canada Construction Technology Update No. 66 March

65 Concrete Sound Transmission Loss 125Hz 250Hz 500Hz 1000Hz 2000Hz 4000Hz SOUND TRAVELS THROUGH (STC) STC 10 4 x 8 x 16 Solid lightweight Conc. block (23 lbs/block) (100x200x400mm) thick (100mm) Flat concrete panel (54 psf) 44 8 x 8 x 16 3-cell lightweight Conc. Block (28 lbs/block) (200x200x400mm) 6 thick (150mm) Flat concrete panel (75 psf) STC 40+ RECOMMENDED BETWEEN CORRIDOR AND FAMILY ROOM STC 50+ BETWEEN CLASSROOMS STC 55+ BETWEEN KITCHEN AND BED ROOM thick (200mm) (95 psf) ROBUST LOW FREQUENCY VALUES - USEFUL FOR TRANSP. NOISE, AMPLIFIED MUSIC, AND MECHANICAL NOISE SOUND LESS LIKELY TO TRAVEL THROUGH WEAK LOW FREQUENCY VALUES - NOT USEFUL FOR TRANSP. NOISE, AMPLIFIED MUSIC, AND MECHANICAL NOISE 41

66 125Hz 250Hz 500Hz 1000Hz 2000Hz 4000Hz 5/8 gypsum bd (16mm) 2x4 studs (38x89mm) 16 on center (400mm) 5/8 gypsum bd Stud Walls Sound Transmission Loss STC 34 SOUND TRAVELS THROUGH (STC) With 2 thick sound attenuation blanket 38 Without insulation One side of gypsum wall mounted on resilient channel STC 40+ RECOMMENDED BETWEEN CORRIDOR AND FAMILY ROOM With 2 thick sound attenuation blanket and one side mounted on resilient channel STC 50+ BETWEEN CLASSROOMS STC 55+ BETWEEN KITCHEN AND BED ROOM One side with three layers gypsum board The other side with two layers mounted on resilient channel SOUND LESS LIKELY TO TRAVEL THROUGH 51

67 125Hz 250Hz 500Hz 1000Hz 2000Hz 4000Hz 5/8 gypsum board (16mm) 2x4 studs (38x89mm) 16 on center (400mm) 5/8 gypsum board Stud Walls Sound Transmission Loss STC 34 SOUND TRAVELS THROUGH (STC) With 2 thick sound attenuation blanket With double row of 2x4 studs, spaced 16 o. c. on separate plates spaced 1 apart Without insulation 45 STC 40+ RECOMMENDED BETWEEN CORRIDOR AND FAMILY ROOM With 2 layers of gypsum board on each side STC 50+ BETWEEN CLASSROOMS STC 55+ BETWEEN KITCHEN AND BED ROOM With 1 1/2 thick sound attenuation blanket (40mm) SOUND LESS LIKELY TO TRAVEL THROUGH 53

68 1/8 monolithic float glass (3mm) (1.4 lb / sq. ft.) Glass Sound Transmission Loss 125Hz 250Hz 500Hz 1000Hz 2000Hz 4000Hz STC 26 SOUND TRAVELS THROUGH (STC) 10 1/2 insulated glass (13mm): 1/8 + 1/8 double glass With 1/4 air space /4 monolithic float glass (6mm) (2.9 lb / sq. ft.) Double glass: 1/4 laminated + 3/16 monolithic glass With 2 air space (50mm) 35 STC 40+ RECOMMENDED BETWEEN CORRIDOR AND FAMILY ROOM /4 + 1/8 double glass With 2 air space 39 STC 50+ BETWEEN CLASSROOMS Double glass: 1/4 laminated + 1/4 laminated With 1/2 air space STC 55+ BETWEEN KITCHEN AND BED ROOM Double glass: 1/4 laminated + 3/16 monolithic glass With 4 air space (5.9 lb / sq. ft.) TO MAINTAIN ADEQUATE SOUND ISOLATION, ENSURE THAT WINDOWS AND DOORS HAVE STC VALUES NO LOWER THAN 5 POINTS BELOW THE WALL ROBUST LOW FREQUENCY VALUES - USEFUL FOR TRANSP. NOISE, AMPLIFIED MUSIC, AND MECHANICAL NOISE SOUND LESS LIKELY TO TRAVEL THROUGH WEAK LOW FREQUENCY VALUES - NOT USEFUL FOR TRANSP. NOISE, AMPLIFIED MUSIC, AND MECHANICAL NOISE 54

69 Louvered door, 25 to 30% open Doors Sound Transmission Loss 125Hz 250Hz 500Hz 1000Hz 2000Hz 4000Hz STC 12 SOUND TRAVELS THROUGH (STC) /4 hollow-core (45mm) wood door, no gaskets, 1/4 air gap at sill (6mm) (1.5 lb / sq. ft.) 19 With gaskets, and drop seal STC 40+ RECOMMENDED BETWEEN CORRIDOR AND FAMILY ROOM 1 3/4 hollow-core 16 gauge steel door, glass fiber filled with gaskets, and drop seal (7 lb / sq. ft.) STC 50+ BETWEEN CLASSROOMS STC 55+ BETWEEN KITCHEN AND BED ROOM TO MAINTAIN ADEQUATE SOUND ISOLATION, ENSURE THAT WINDOWS AND DOORS HAVE STC VALUES NO LOWER THAN 5 POINTS BELOW THE WALL 75 SOUND LESS LIKELY TO TRAVEL THROUGH 55

70 Not-Recommended Dwelling Layout Plumbing Fixture on Party Wall Plumbing Fixture on Bedroom Wall Trash Shoot Adjacent Exposed to Living Area A/C Compressor Units Adjacent to Windows Single Stud Wall Construction Partial Length Wall not a Buffer D W AHU A/C with Louvered Door & Unducted Return No Vestibule, One Door Elevator Adjacent to Dwelling Partial Length Wall Not a Sound Barrier Cabinets, Dishwasher, Disposal & Plumbing on Party Wall Single Stud Wall Construction TV Mounted on Party Wall 88

71 Recommended Dwelling Layout A/C Remotely Located with Ducted Supply and Return Noisy Appliances in Separate Room with Door Noisy cabinets, Dishwasher, Disposal, Refrig.& Plumbing Away from Noise-Critical Walls Plumbing Fixtures Not on Party Wall Closet as Buffer Zone for Plumbing & Airborne Noise Area Adjacent to Windows Free of Mech. Equip., Dumpsters & Other Noisy Outdoor Equipment Double Stud Wall for Critical Adjacencies AHU W D Full Length & Full Height Wall. Door to Separate Living & Sleeping Areas Double Stud Party Wall Double Stud Wall Construction and Closet Buffer Zone TV & Stereo Not Mounted on Party Wall Vestibule with Door on Each Side Between Corridor & Dwelling Elevator & Trash Shoot Separated from Dwelling with a Utility Closet Buffer Zone 89

72