Applied Acoustics 75 (2014) 52–58

Contents lists available at SciVerse ScienceDirect

Applied Acoustics

journal homepage: www.elsevier .com/locate /apacoust

Modeling chairs and occupants to closely approximate the soundabsorption of occupied full scale theatre chairs

0003-682X/$ - see front matter � 2013 Elsevier Ltd. All rights reserved.http://dx.doi.org/10.1016/j.apacoust.2013.07.003

⇑ Tel.: +82 1029270563.E-mail addresses: youngji@chonbuk.ac.kr, youngji@jbnu.ac.kr

Young-Ji Choi ⇑Department of Architectural Engineering, Chonbuk National University, 664-14 1Ga, Duckjin-Dong, Duckjin-Gu, Jeonju, Jeonbuk 561-156, Republic of Korea

a r t i c l e i n f o a b s t r a c t

Article history:Received 15 May 2013Received in revised form 3 July 2013Accepted 3 July 2013Available online 2 August 2013

Keywords:Sound absorptionModel chairsModel listeners

The present work reports on the process of modeling chairs and occupants to closely approximate thesound absorption of occupied full scale theatre chairs and explains how the best form of model listenerwas determined. Modifying the form of the model listeners to have shorter upper legs and narrowerlower legs, led to improved agreement between model and full scale occupied chairs at all frequenciesincluding at 125 Hz. The measured absorption coefficients of single blocks of model chairs with or with-out model listeners agreed well with the measured values for both full scale types E and G chairs. How-ever, the estimated values for larger sample blocks of model chairs with P/A = 0.5 m�1 showed betteragreement with the measured values for full scale type G chairs than type E chairs due to the differentslopes of the regression lines versus P/A. The present results demonstrate that the model chair and lis-tener accurately simulate the sound absorption characteristics of a particular type of quite absorptive fullscale occupied chairs for all sample sizes of the full scale chairs.

� 2013 Elsevier Ltd. All rights reserved.

1. Introduction

The successful modeling of chairs and occupants in scale mod-els is particularly important because the audience is usually thelargest single component of absorption in an auditorium. Someprevious studies [1–6] have demonstrated the development ofmodel listeners for adding the sound absorption of an audience,seated on chairs, in model lecture rooms or auditoria. The modellisteners have varied in form from egg cartons [1] to the simplifiedhuman forms [2,3,5]. Table 1 summarizes the details of model lis-teners developed in some earlier studies.

Day [2] developed 1/10 scale model listeners with a simplifiedhuman form to investigate the effect of the degree of occupancyon the audience absorption in a model lecture room. The model lis-tener consisted of smoothed softwood body and hardwood head. Asingle layer of surgical gauze was used for simulating human cloth-ing. Hegvold [3] investigated the effects of the amount of clothingworn by a person on the absorption of a group of people and devel-oped 1/8 scale model listeners to simulate the sound absorption ofa typically dressed Sydney audience. The model listeners were con-structed with a rigid polyurethane foam body and a sanded pinehead. Hegvold adopted Day’s simplified human form to modelthe sound absorption of model Sydney listeners.

Cremer et al. [4] used cardboard egg cartons to simulate thesound absorption of the audience in a 1/16 scale model of a

multi-purpose hall and found that the cardboard egg cartons werecapable of simulating the partially diffuse reflections from theaudience. In a recent study, Tahara et al. [5,6] developed 1/16 scalemodel listeners to simulate the audience absorption in a modelauditorium. The model listeners consisted of a wooden head witha single layer of 1 mm felt simulating human hair and a woodentorso. Their model listeners were more similar to a head and torsosimulator rather than the simplified more complete human form ofmodel listeners developed by Day and Hegvold. Similar types ofhead and torso forms of model listeners were used in other re-cently reported scale model predictions [7]. The absorption coeffi-cients of the simple forms of model listener were obtained fromthe measurements of single blocks of model chairs, and thereforethe results were never justified to closely approximate the absorp-tion characteristics of occupied full scale chairs for all sample sizesof the chairs.

Some earlier studies demonstrated the value of using modeltests to better understand the sound absorption of theatre chairs[3,8,9]. Reverberation chamber tests of both model and full scalechairs showed that the sound absorption characteristics of smallerblocks of chairs are not representative of those found for the largerblocks of chairs in an auditorium because the measured absorptioncoefficients vary with the dimensions of the sample and largersamples tend to have lower absorption coefficients due to edge ef-fects [3,8–10]. The edge absorption may be reduced by usingscreens around the seating blocks, but diffraction effects are noteliminated by this method and the absorption coefficients of theseating blocks would still vary with the sample perimeter-to-area

Fig. 1. A photo of high absorption model chairs occupied with model listeners inthe model reverberation chamber.

Y.-J. Choi / Applied Acoustics 75 (2014) 52–58 53

(P/A) ratio both in full scale [9–11] and in model tests [3,9]. Thiswould also be true for the approach of putting absorbing samplesin a well in the floor of the reverberation chamber [12]. This lattermethod is also not very practical for objects as large as chairs. Thesound absorption coefficients of blocks of theatre chars have beenshown to be linearly related to the sample P/A ratio [10]. One canpredict the expected absorption coefficients of the larger blocks ofchairs found in auditoria using linear regression fits to the mea-sured values obtained from 5 or more sample blocks of chairs withvaried P/A ratio in a reverberation chamber. This method is re-ferred to as the P/A method. This method can more completelycharacterize particular types of chairs than one sound absorptiontest of a single sample block of chairs in the reverberation cham-ber. The method has been shown to accurately predict the absorp-tion coefficients of chairs measured in an auditorium both in fullscale [10,11] and in model tests [9,13].

Many have previously used scale model chairs and listeners, butprevious efforts to create scale model chairs and occupants havenot always been very thorough. There are a range of absorptioncharacteristics for occupied and unoccupied theatre chairs andthe absorption characteristics vary with P/A and the form of thisvariation varies with chair type [13]. It is not enough to comparesingle samples in reverberation chamber tests. One should validatemodel chairs and occupants to confirm that the variation ofabsorption coefficients with P/A for the model chairs and listenersare similar to those for common types of full scale chairs withoccupants. This would ensure that absorption coefficients of themodel chairs would be similar to the corresponding full scalechairs for all sample sizes of the chairs.

Table 1Summary of the model listeners developed in previous studies. The symbol ‘‘Ø’’ indicates

Researcher (year) Materials

Brebek et al. [1] Egg cartonsCremer et al. [4]Day [2] Head: softwood

Body: hardwoodCloth: surgical gauze adding on the body

Hegvold [3] Head: sanded pineBody: polyurethane foamCloth: no cloth

Tahara et al. [5] Head: woodTorso: woodCloth: felt 1 mm adding on the torso and head

This process could enhance the credibility of future studiesusing these model chairs and occupants such as for investigationof the incremental effects of occupants in chairs or carpet underchairs on the absorption coefficients of each combination. It isimportant to better understand such interactive effects rather thanto depend on measurements of every new situation. The goal ofthis work is to develop model chairs and occupants as closely rep-resentative as possible of common types of full scale theatre chairswith audiences. The present work reports on the process of model-ing chairs and occupants to closely approximate the sound absorp-tion of occupied full scale theatre chairs and describes how thebest form of model listeners was determined. This paper was alsointended to give some useful guidelines on constructing modeloccupants by investigating how the form and dimensions of modellisteners affect to the measured sound absorption characteristics ofoccupied theatre chairs.

2. Measurement procedures

2.1. Model chair construction

The main purpose of developing various types of model theatrechairs and listeners was to accurately simulate their sound absorp-tion characteristics and to better understand how they affect to theacoustical conditions in real auditoria. The model chairs and occu-pants were developed to have similar absorption characteristics, asa function of P/A, to typical more highly absorbing full scale chairsand occupants. The previously reported results for full scale types Eand G chairs [10,11] were used as design goals for the model chairsand occupants.

The type E chairs were considered very highly absorptive chairsand the type G chairs more typically highly absorptive. These highabsorption full scale theatre chairs had thicker absorbing materialand included thick absorbing pads on the chair backs. The type Echairs also had absorptive padding on the rear of the seat backsas well as on the arm rests and sides of the chairs. They also hadquite thick and absorptive seat cushions and perforated seat pansover glass fiber material. Chair type G had cloth covered metal ofthe rears of the seat backs. The slopes and intercept values of theregression lines for both unoccupied and occupied types E and Gchairs were included in Ref. [13]. Beranek and Hidaka’s [14] esti-mates of the average absorption coefficients of the most absorptivegroup of chairs (their Group 1) were less absorptive than the valuescalculated from the measured types E and G full scale occupiedchairs with P/A = 0.5 m�1.

A variety of 1/10 scale model chairs were systematically testedto develop a model chair with absorption characteristics similar tothose for full scale high absorption chairs. Model chairs having awidth of 0.6 m (full scale) were constructed as connected seatswith arm rests and underpasses as shown in Fig. 1. The height of

the diameter of the model listener’s head measured in mm.

Scale Size (width � length � depth), mm

1/10 –1/161/10 Head: Ø25

Torso: 35 � 75 � 18Hips: 35 � 45 � 18Legs: 35 � 47 � 18

1/8 Head: 22.2 (depth) � 31.7 (length)Torso: 54.6 � 73.6 � 22.2Hips: 54.6 � 57.2 � 22.2Legs: 54.6 � 58.4 � 22.2

1/16 Head: Ø10Torso: 24 � 38 � 12

125 250 500 1000 2000 40000.0

0.2

0.4

0.6

0.8

1.0

Abso

rptio

n co

effic

ient

Frequency, Hz

Felt 1 mm Bubble wrap 3mm Bubble wrap+Felt

Fig. 2. Absorption coefficients of the materials added to model chairs (full scale).

54 Y.-J. Choi / Applied Acoustics 75 (2014) 52–58

the gap under the chairs was 280 mm (full scale). A 1.0 mm singlelayer of felt combined with a 3.0 mm single layer of bubble wrapwas added on the seat back, rear of back, sides and undersides ofthe model chairs and these chairs are referred to as high absorptionmodel chairs. The measured absorption coefficients of the materi-als when added to the model chairs are plotted in Fig. 2. The sam-ples tested had an area of 10.2 m2 (full scale), which corresponds toa P/A ratio of 1.25 m�1. The absorption characteristics of the com-bined bubble wrap and felt materials showed very similar trendsover frequency to those found for unoccupied full scale chairs(see Fig. 5).

The sound absorption coefficients of blocks of theatre chairsvary with sample P/A ratio and the sample blocks of model chairsshould have the same P/A ratio as the full scale blocks of chairsused as design goals. The tests used blocks of 3 rows of 5 modelchairs (R3C5) having a P/A ratio of 1.40 m�1 (full scale) whichwas very similar to P/A of 1.39 m�1 and 1.37 m�1 for the blocksof 3 rows of 6 chairs (R3C6) of both unoccupied full scale types Eand G chairs.

2.2. Model listener construction

One-tenth scale model listeners were constructed using 10 mmthick expanded PVC (poly vinyl chloride) pieces and were scaled tobe representative of average Koreans. The selection of materials forconstructing model listeners similar to the absorption coefficients

125 250 500 1000 2000 40000.0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1.0

Abso

rptio

n co

effic

ient

Frequency, Hz

Felt 0.5 mm Fabric 0.5 mm Pineapple sheet 0.5 mm Expanded PVC 10 mm

Fig. 3. Absorption coefficients of the materials added to model listeners (full scale).

of the human body was carefully considered according to someearlier studies on the measurements of the absorption coefficientsof the human body and hair [15,16]. Fig. 3 plots the measuredabsorption coefficients of the materials used for model listenerconstruction. The samples tested had an area of 10.2 m2 (full scale),

which corresponds to a P/A ratio of 1.25 m�1. The absorption coef-ficients of the PVC pieces are very similar to those measured forreal human body surfaces found in Ref. [15]. The absorption coef-ficients of human hair increased with increasing frequency but de-creased above 2 kHz. The absorption coefficients of the 1 mm feltshowed similar trends as measured for human hair found in Ref.[16]. It was found that there was considerable variation in theabsorption coefficients resulting from different dress types wornby audiences [3]. The audiences wore light to medium clothes,such as a short sleeved shirts or suits, during the measurementsin the full scale reverberation chamber tests and the materialsadded to the model listeners were selected to approximate theabsorption characteristics of the dress types worn by full scaleaudiences.

Prior to adding absorbing materials to the model listeners, thebest form of model listeners was determined by testing 7 typesof model listeners with various forms and dimensions. The absorp-tion coefficients of each set of model listeners were measuredwhile seated on the high absorption model chairs. A total of 7 typesof model listeners with various forms and dimensions are illus-trated in Fig. 4 and they are described in Table 2. ‘Upper legs’ arethe portion of the legs above the knees and ‘lower legs, are the por-tion below the knees. Model listener type A had the widest upperand lower legs and was similar to the model occupant developedby Day [2] and Hegvold [3]. Model listener type B consisted of torsoand upper legs, but had no lower legs. Type C had shorter upperlegs than types A and B model listeners and no lower legs. Modellistener type D had no upper and lower legs and was similar tothe model listeners developed by Tahara et al. [5,6]. Although mod-el listener type D had no upper and lower legs, and hence was notrepresentative of real human body forms, it was intended to inves-tigate the effects of the presence of the upper and lower legs on theoccupied chair absorption characteristics. Model listener types E, Fand G were modified versions of the type A listeners. Type E listen-ers had narrower lower legs than those of the type A model listen-ers. Type F had shorter upper legs and narrower lower legs than theother models. Type G had the two thin separate lower legs makingthem more similar to real humans.

The best form of model listener was selected and tested withvaried added absorbing material to develop a model listener withabsorption characteristics most similar to those found for the fullscale occupied theatre chairs. Table 3 describes various absorbingmaterials added to the best form of model. Four different typesof absorbing materials were added to the head, torso and upperlegs of the model listeners. No absorbing materials were addedto the lower legs of model listeners. A total of 6 model listenerswith varied added absorbing materials were constructed. Theabsorbing material named ‘pineapple sheet’ is a 0.5 mm thick pa-per having a pineapple pattern on it. Three absorbing materials, felt0.5 mm, fabric 0.5 mm and pineapple sheet 0.5 mm were added tothe torso and upper legs of the model listeners. Felt, 1.0 mm thick,was added to the head of model listeners to simulate the soundabsorption by human hair (see Table 3).

All tests were of sample blocks of 5 rows of 3 chairs in each rowwith a row-to-row spacing of 0.91 m (full scale). This particularsample block of model chairs had a P/A ratio of 1.55 m�1 that issimilar to P/A of 1.53 m�1 for both unoccupied full scale types Eand G chairs. After selecting the most successful model listener,the absorption coefficients of larger sample blocks of occupiedmodel chairs with P/A = 0.5 m�1 were estimated from the reverber-ation chamber measurements of seven different sample blocks of

125 250 500 1000 2000 40000.0

0.2

0.4

0.6

0.8

1.0

1.2

1.4

1.6

1.8

Abso

rptio

n co

effic

ient

Frequency, Hz

Full scale type E chairs (unoccupied, P/A=1.39m-1) Full scale type G chairs (unoccupied, P/A=1.37m-1) Average, full scale types E and G chairs Model chairs (unoccupied, P/A=1.40 m-1)

Fig. 5. Measured absorption coefficients of an R3C6 block of unoccupied modelchairs compared with those measured for R3C6 blocks of both unoccupied full scaletypes E and G chairs as well as the average values of all two types of full scale chairs[10,13].

Fig. 4. Sketch of front and side views of 7 types of model listeners (unit: mm, full scale).

Table 27 Types of model listeners with various forms and dimensions. The symbol ‘‘Ø’’indicates the diameter of the model listener’s head, measured in mm.

Name Size (width � length � depth), mm

Head Torso Upper legs Lower legs

Type A Ø25 36 � 60 � 10 36 � 40 � 10 36 � 30 � 10Type B Ø25 36 � 60 � 10 36 � 40 � 10 –Type C Ø25 36 � 60 � 10 36 � 25 � 10 –Type D Ø25 36 � 60 � 10 – –Type E Ø25 36 � 60 � 10 36 � 40 � 10 20 � 30 � 10Type F Ø25 36 � 60 � 10 36 � 30 � 10 15 � 30 � 5Type G Ø25 36 � 60 � 10 36 � 30 � 10 5 � 30 � 5

Y.-J. Choi / Applied Acoustics 75 (2014) 52–58 55

chairs having a range of P/A values between 1.29 and 2.40 m�1 andcompared to those estimated for full scale occupied chairs. Themeasured absorption coefficients of sample blocks of chairs varywith sample P/A and comparing the estimated absorption coeffi-cients of larger sample blocks of chairs typically found in auditoriafor both model and full scale chairs would be more appropriate.

2.3. Reverberation chamber measurements

The volume of the 1/10 scale model reverberation chamber was300 m3 (full scale) and it was built using 20 mm thick acrylic pan-els. In the model measurements, a 1.37-s logarithmic sine sweepfrom 1 kHz to 100 kHz was used, which corresponds to full-scale

Table 3Absorbing materials added to the best form of model listener. The meaning of letter symb

Name Materials, thickness mm

Head Torso

Best type-Fe – Felt, 0.5Best type-F – Fabric, 0.5Best type-PF – Pineapple sheBest type-PP – Pineapple sheBest type-33FePP Ø33Felt, 1 Pineapple sheBest type-27FePP Ø27Felt, 1 Pineapple she

Symbol

FPFe

frequencies from 100 Hz to 10 kHz. To eliminate the unwanted ef-fects of air absorption, the model chamber was filled with nitrogenduring each test. The reverberation chamber was kept at a constanttemperature of 22 �C and a relative humidity of 4%. Six combina-tions of two source positions and three receiver positions were se-lected for measuring the absorption coefficients of the unoccupiedchairs. A 20 dB range of each decay, from �5 dB to �25 dB, wasused to calculate reverberation times according to the proceduresdescribed in ISO 354 [12].

Prior to the measurements, the diffusivity of the sound field inthe reverberation room was examined according to ISO 354 [12].There was no evidence of non-linear decays over this range whenthe absorption of the chairs was measured. The measurementswere made in 1/3 octave bands, but the absorption coefficientswere presented as octave band values derived by averaging thethree individual 1/3 octave sound absorption coefficients in eachoctave band. A repeatability test of the measurements for theabsorption coefficients of the chairs was carried out to checkwhether the results were consistent for each measurement. Themeasurements were repeated three times, and the results werepresented as the mean absorption coefficients.

ols is explained in the table below the main table.

Upper legs Lower legs

Felt, 0.5 –Fabric, 0.5 –

et, 0.5 Fabric, 0.5 –et, 0.5 Pineapple sheet, 0.5 –et, 0.5 Pineapple sheet, 0.5 –et, 0.5 Pineapple sheet, 0.5 –

Meaning

FabricPineapple sheetFelt

56 Y.-J. Choi / Applied Acoustics 75 (2014) 52–58

By comparison the full scale measurements were performed inthe 254 m3 reverberation chamber, at the National Research Coun-cil in Ottawa, having fixed diffuser panels as well as a large rotatingvane. More details are included in Refs. [10] and [13].

3. Results and discussion

3.1. Developing model chairs and listeners representative of the soundabsorption of occupied theatre chairs

The measured absorption coefficients of the unoccupied highabsorption model chairs were compared in Fig. 5 with those valuesmeasured for unoccupied full scale types E and G chairs as well asthe average of both types of full scale chairs. The measured absorp-tion coefficients of unoccupied model chairs in Fig. 5 show moresimilar characteristics in absorption coefficient over the frequencyto those measured for type G chairs than with the type E chairs re-sults. Chair type G had slightly higher absorption coefficients at250 Hz than those measured for model chairs and the type E chairs.This is most likely due to the resonant sound absorption of the thinmetal seat pans of type G chairs. The average absorption coeffi-cients of the more absorptive two types of theatre chairs showedgood agreement with the measured results for the model chairs.

The materials of highly absorbing chairs can influence theabsorption characteristics of the occupied chairs, but the occupantsare the most important factor determining their absorption charac-teristics [13]. Various types of model audiences were developed inearlier studies, but guidelines on how to construct more accuratemodel listeners were generally not available. Therefore, it was veryuseful to provide measurement results showing how the form anddimensions of model listeners affect to the measured absorptioncharacteristics of occupied theatre chairs. Fig. 6 compares the mea-sured absorption coefficients of the types A to D model listenersseated on the high absorption model chairs to illustrate the effectsof the form of the model listener. Adding model listeners to thechairs generally led to increases in the absorption coefficientswhich were largest at 125 Hz. The increases in 125 Hz absorptioncoefficients varied from 0.06 to 0.27. The largest increases in125 Hz absorption coefficients occurred for the types A and B lis-teners. The type D model listener having no upper and lower legshad the smallest increase in absorption coefficient at 125 Hz. Sincethe absorption of the expanded PVC pieces at 125 Hz (Fig. 3) wasquite small, the increases in the absorption coefficients at this fre-quency with model listeners added to the chairs, were likely due to

125 250 500 1000 2000 40000.0

0.2

0.4

0.6

0.8

1.0

1.2

1.4

1.6

Abso

rptio

n co

effic

ient

Frequency, Hz

Unoccupied model chairs Type A+chairs Type B+chairs Type C+chairs Type D+chairs

Fig. 6. Measured absorption coefficients of occupied chairs with varied forms ofmodel listeners compared with the absorption coefficients of the unoccupied chairs.

changes to the form of the model listener. The inclusion of upperand lower legs in the model listeners led to larger increases inabsorption coefficients at 125 Hz. At 125 Hz the cavities betweenand under the chairs may act as resonant cavities and putting legsinto them might change the cavities enough to change the reso-nant absorption of the cavities and to increase the low frequencyabsorption of the model theatre chairs.

Adding model listener type D to the chairs caused a larger in-crease in absorption coefficients at 2000 and 4000 Hz than whenadding the other model listener types to the chairs. This was be-cause the model type D listeners, consisting of a torso only, didnot cover some absorptive areas of the seat of the chairs that othermodel listener types, with upper and lower legs, did cover. Exceptfor model listener type D, adding model listeners to the chairscaused a decrease in absorption coefficients at 4000 Hz. The great-est decrease was 0.14 which occurred when the type A model lis-teners were in the chairs. The decreased absorption at 4000 Hz wasdue to the model listeners covering some absorptive areas of theseat and back of the chairs and reducing the total absorption ofthe occupied chairs at high frequencies.

Fig. 7 shows the measured absorption coefficients of highabsorption model chairs occupied with model listener types A, E,F and G. Model listener types A, E, F and G have the same formof torso, but the dimensions of upper and lower legs varied (see Ta-ble 1). Types F and G listeners have the shortest upper legs and nar-rowest lower legs among the four model listener types. Addingmodel listener type E, having narrower lower legs, to the chairsslightly decreased the absorption coefficients at 1000 and2000 Hz compared to chairs occupied with model listener type A.The differences were 0.03 and 0.05 at 1000 and 2000 Hz. Addingmodel listener type F having the shortest upper legs and the nar-rowest lower legs to the chairs led to a smaller increase in theabsorption coefficients at 125 Hz than occurred for chairs occupiedwith types A and E model listeners. Type G model listeners, havingtwo thin lower legs, had a slightly increased absorption coefficientat 125 Hz compared to chairs occupied with types A and E modellisteners. However, the difference in absorption coefficients at125 Hz between chairs occupied with types A and G listenerswas only 0.04. Chairs with type G listeners were less absorptiveat 1000 and 2000 Hz than chairs occupied with type A listeners.Adding type G model listeners to the chairs decreased the absorp-tion coefficients at 1000 and 2000 Hz more than those found forchairs occupied with type A model listeners. The differences were0.09 and 0.12 at 1000 and 2000 Hz respectively. Type F listeners, a

125 250 500 1000 2000 40000.0

0.2

0.4

0.6

0.8

1.0

1.2

1.4

1.6

Abso

rptio

n co

effic

ient

Frequency, Hz

Unoccupied model chairs Type A+chairs Type E+chairs Type F+chairs Type G+chairs

Fig. 7. Absorption coefficients of model listeners with varied dimensions of the legsof the model listeners.

125 250 500 1000 2000 40000.4

0.6

0.8

1.0

1.2

1.4

1.6

1.8

Abso

rptio

n co

effic

ient

Frequency, Hz

Type F-Fe+chairs Type F-F+chairs Type F-PF+chairs Type F-PP+chairs Type F-33FePP+chairs Type F-27FePP+chairs Full scale type E chairs, occupied Full scale type G chairs, occupied Average, full scale types E and G chairs

Fig. 8. Measured absorption coefficients of type F form model listeners with variedadded absorbing materials.

0.0

0.2

0.4

0.6

Slop

e

Frequency, Hz

Type E chairs Type G chairs Model chairs

(a)

125 250 500 1000 2000 4000

125 250 500 1000 2000 40000.0

0.2

0.4

0.6

0.8

1.0

1.2

1.4

1.6

1.8

(b)

Inte

rcep

t

Frequency, Hz

Type E chairs Type G chairs Model chairs

Fig. 9. (a) Slopes and (b) intercepts of regression lines of absorption coefficientsversus P/A for occupied full scale theatre chairs and model chairs.

Y.-J. Choi / Applied Acoustics 75 (2014) 52–58 57

modified human form of model listener having shorter upper legsand narrower lower legs, reduced the increase in the absorptioncoefficients at 125 Hz to 0.07. Therefore, model listener type Fwas selected as the best form of model listeners.

Fig. 8 shows the measured absorption coefficients of type F formmodel listeners with varied added absorbing material. The mea-sured absorption coefficients of full scale occupied theatre chairsare also plotted for comparison in Fig. 8. Adding 0.5 mm feltabsorbing material to the model listeners reduced the absorptioncoefficients of the occupied chairs at most frequencies. That is,the felt did not increase the absorption coefficients of the occupiedchairs as desired. Because the expanded PVC pieces and the felt hadsimilar absorption coefficients (see Fig. 4), adding the 0.5 mm feltabsorbing material to the model listeners covered the expandedPVC pieces and reduced the total absorption of the occupied chairsat most frequencies. Adding 0.5 mm fabric absorbing material tothe model listeners led to increases in absorption coefficients atmost frequencies and the increase in absorption coefficients at125 Hz was the largest among the six model listeners with variedadded absorbing materials.

Adding 0.5 mm pineapple sheet absorbing material to the mod-el listeners slightly reduced the absorption coefficients at 125 Hzand increased the absorption coefficients at other frequencies.Adding 1 mm felt absorbing material to the Ø33 mm head of themodel listeners increased the absorption coefficients at most fre-quencies and was found to be the most successful model listenerwith similar absorption coefficients to the full scale occupied the-atre chairs. At 125 Hz, model listener type F-33FePP had a slightlyhigher absorption coefficient than the full scale occupied type Gchairs with a difference of 0.06. The occupied model chairs withmodel listeners type F-33Fe-PP showed good agreement of theirabsorption coefficients with the occupied full scale type G chairs.The difference between the absorption coefficients, averaged over

Table 4Summary of: the standard errors of the estimate of the octave-band data about the regressiThe values of slopes (b), and intercepts (a1) for high absorption model chairs (ns = not sig

Chair type N Frequency, Hz

125

Model chairs, occupied 7 r 0.052p �b 0.106a¥ 0.664

frequency, for model chairs occupied with the most successfulmodel listeners, and those measured for the full scale occupiedtype G chairs, were no more than 0.1.

3.2. Comparison of estimated absorption coefficients for samples ofoccupied model chairs with P/A = 0.5 m�1 with those for full scaleoccupied chairs

Linear regression lines were fitted to plots of absorption coeffi-cient versus P/A for various blocks of chairs measured in the modelreverberation chamber. Theses analyses were performed using theOrigin plotting and analysis software [17] to perform standardleast squares linear regression fits to the data. The slopes and inter-cept values of the regression lines were included in Table 4 as wellas the statistical significance of the results. Fig. 9(a) plots the slopes

on lines (r), the statistical significance of the octave-band regression results (p-value),nificant, �p < 0.2, ��p < 0.05, ���p < 0.01).

250 500 1000 2000 4000

0.040 0.049 0.061 0.047 0.043��� ��� ��� ��� ���0.307 0.264 0.317 0.394 0.4700.686 0.953 0.949 0.868 0.786

125 250 500 1000 2000 40000.0

0.2

0.4

0.6

0.8

1.0

1.2

1.4

1.6

1.8

Abso

rptio

n co

effic

ient

Frequency, Hz

Full scale type E chairs ,occupied Full scale type G chairs ,occupied Average, full scale types E and G chairs Model chairs ,occupied

Fig. 10. Comparison of absorption coefficients versus frequency for samples ofoccupied model chairs with P/A = 0.5 m�1 with Bradley et al.’s [13] estimates of theabsorption coefficients of two types of full scale occupied chairs as well as theaverage of all two types of full scale chairs.

58 Y.-J. Choi / Applied Acoustics 75 (2014) 52–58

of the regression lines versus frequency for occupied theatre chairsand model chairs. The intercepts of the regression lines are plottedversus octave band frequency in Fig. 9(b). The slopes and interceptsof the regression lines for the type G and model chairs in Fig. 9(a)and (b) are very similar at most frequencies. The type E chairs havehigher absorption coefficient values at most frequencies. The mod-el chairs are more representative of the more intermediate charac-teristics of the type G chairs than the more highly absorptive type Echairs.

The absorption coefficients of model chairs for larger sampleblocks of chairs with P/A = 0.5 m�1 were calculated and comparedwith those estimated for full scale types E and G occupied chairsshown in Fig. 10. This gives more reliable comparisons by usingthe mean trend over several measurements of blocks of chairs.The estimated absorption coefficients of full scale type G occupiedchairs were good agreement with those calculated for model chairsover frequency. Although the measured absorption coefficientsfrom one single sample blocks of chairs for full scale chair E typewere very similar to those measured for model chairs (see Fig. 5),the estimated values for larger sample blocks of chairs were quitedifferent at most frequencies. This is likely due to the differentslope values over frequency for both full scale type E and modelchairs shown in Fig. 9(a). It is needed to understand the differencesin characteristics of different types of chairs.

4. Conclusions

The process of modeling chairs and occupants resulted inabsorption characteristics that closely approximated the soundabsorption of full scale occupied theatre chairs and the details ofthe successful model listeners were reported. A modified form ofmodel listener (type F), having shorter upper legs and narrowerlower legs than a typical human form of model listener (type A), re-duced the increase in the absorption coefficient at 125 Hz andachieved better agreement with the absorption coefficients ofoccupied full scale chairs. The differences between the absorptioncoefficients, averaged over frequency for model chairs, occupied

with the most successful model listeners (type F-33FePP), andthose measured for the full scale occupied theatre chairs were nomore than 0.1. The measured absorption coefficients from one sin-gle sample block of chairs for model chairs with or without modellisteners showed good agreement with the measured values forboth full scale types E and G chairs. However, the estimated valuesfor larger sample blocks of model chairs representative of chairs inan auditorium (P/A = 0.5 m�1) were closer to the measured valuesfor full scale type G chairs than to the type E chair results. Thatis, the model chairs and occupants were a better model of the P/A effect for the type G chairs.

The present results demonstrate that the model chair and lis-tener that were developed accurately simulated the sound absorp-tion characteristics of the type G full scale chairs. As the type Gchairs are representative of many more absorptive full scale chairs,the model chairs and listeners should be more widely useful.

The effects of P/A on the incremental effects of adding occu-pants have not been fully explored and the incremental effects ofcarpet under chairs have also not been considered. These are issuesthat also need to be better understood so that we can better esti-mate their effects on the total sound absorption in an auditorium.They will be discussed in a forthcoming companion paper.

Acknowledgements

I would like to thank Prof. Dae-Up Jeong at Chonbuk NationalUniversity for his help with the development of the model listen-ers. I also would like to acknowledge Dr. John S. Bradley at NationalResearch Council in Canada for his advice and discussions with thepreparation of the manuscript. This work was partly supported byNational Research Foundation of Korea Grant funded by the KoreanGovernment (2012R1A1B5000491).

References

[1] Brebeck D, Bucklein R, Krauth E, Spandock R. Acoustically similar models as anaid in space acoustics. Acustica 1967;18:213–26.

[2] Day BF. A tenth-scale model audience. Appl Acoust 1968;1:121–35.[3] Hegvold LW. A 1:8 scale model auditor. Appl Acoust 1971;4:237–56.[4] Cremer L, Muller HA, Schultz TJ. Principles and applications of room acoustics,

vol. 1. Applied Science Publishers LTD.; 1982; p. 179.[5] Tahara Y, Shimoda H. 1/16 Scale model experiment for room acoustics. J

Environ Eng AIJ 2006;608:1–7.[6] Tahara Y, Shimoda H. 1/16 Scale model experiment for room acoustics physical

properties and auralized sound quality. Proceedings of 19th ICA; 2007.[7] Nagata Acoustics. News 12-04 (No.292); 2012.[8] Barron M, Coleman S. Measurements of the absorption by auditorium seating –

a model study. J Sound Vib 2001;239(4):573–87.[9] Choi YJ, Bradley JS, Jeong DU. Effects of edge screens on the absorption of

blocks of theatre chairs. Appl Acoust 2012;73:470–7.[10] Bradley JS. Predicting theatre chair absorption from reverberation chamber

measurements. J Acoust Soc Am 1992;91:1514–24.[11] Martellota F, Cirillo E. Experimental studies of sound absorption by church

pews. Appl Acoust 2009;70(3):441–9.[12] ISO 354-Acoustics. Measurement of sound absorption in a reverberation room;

2003.[13] Bradley JS, Choi YJ, Jeong DU. Understanding chair absorption characteristics

using the perimeter-to-area method. Appl Acoust 2013;74:1060–8.[14] Beranek LL, Hidaka T. Sound absorption in concert halls by seats, occupied and

unoccupied, and by the hall’s interior surfaces. J Acoust Soc Am1998;104(6):3169–77.

[15] von Gierke HE. Sound absorption at the surface of the body of man andanimals. J Acoust Soc Am 1949;21(1):55.

[16] Treeby Bradley E, Pan Jie, Paurobally Roshun M. An experimental study ofacoustics impedance characteristics of human hair. J Acoust Soc Am2007;122(4):2107–17.

[17] OriginLab data analysis and graphing software. <http://www.originlab.com/>.