In-situ measurements at gymnasiums for sound absorption characteristics of building materials using the ensemble averaging technique

Toru Otsuru 1 , Reiji Tomiku 2 , Noriko Okamoto 3

Faculty of Science and Technology, Oita University 700 Dannoharu, Oita 870-1192, Japan

ABSTRACT The applicability of the in-situ measurement method for sound absorption characteristics of materials using the ensemble averaging technique, i.e. EA method, is examined. A series of in-situ measurement is conducted at two gymnasiums located in Oita-city, Japan. Two kinds of building materials are measured: one is a small sized glass-wool panel which is brought around from one gymnasium to another; and the others are building materials installed on the floors / walls of each gymnasium. The glass-wool panel has the dimensions of 0.5 m by 0.5 m by 0.05 m and with the density of 32 kg m − 3 . Several measurements are conducted during badminton plays are undergoing. Measured sound absorption coe ffi cients of the glass-wool panel revealed that most results agree well with those measured in reverberation rooms and that measured values of floors and walls show acceptable consistencies.

1. INTRODUCTION

The authors presented an in-situ measurement method for sound absorption characteristics of materials using the ensemble averaging technique, namely EA method [1, 2]. The method consists of two types: one using two microphones; and another utilizes a pressure–velocity sensor (Microflown [3]), PU-sensor. In one of successive studies [4], the mathematical–physical model of EA method measurement was given and the uncertainty of absorption coe ffi cient measured by the EA method was proved to satisfy the tentative requirement for the computational room acoustics simulations [5]. Comparisons with other methods and the method’s geometrical configuration were given [6]; and the method’s practicability as well as the reproducibility were revealed, too [7–9]. Recently, to show the applicability of the EA method in field environments, trial measurements were conducted at two sports arenas of gymnasiums; and almost satisfactory results were exhibited [10]. One of the results, however, showed certain discrepancy from others. Then, we conducted additional EA method measurements at the same sports arena.

1 otsuru@oita-u.ac.jp

2 tomiku-reiji@oita-u.ac.jp

3 n-okamoto@oita-u.ac.jp

a slaty. inter.noise 21-24 AUGUST SCOTTISH EVENT CAMPUS O ¥, ? GLASGOW

2. EA METHOD, IN BRIEF

Figure 1 illustrates measurement settings of the EA method to measure a specimen’s ensemble averaged surface normal impedance ⟨ Z n ⟩ which is defined by:

⟨ Z n ⟩ = ⟨ P surf ⟩

⟨ U n , surf ⟩ . (1)

Here, ⟨·⟩ denotes ensemble averaging; and P surf and U n , surf relatively express sound pressure and particle velocity at specimen surface. Both P surf and U n , surf are values in the frequency domain. When a PU-sensor is employed, both values are measured directly. While, a two-microphone PP-sensor is employed, U n , surf is obtained from the di ff erence of the two sound pressure values [1]. Then, corresponding absorption coe ffi cient ⟨ α ⟩ is defined by:

⟨ α ⟩ = 1 −  ⟨ Z n ⟩− 1 ⟨ Z n ⟩ + 1

2 (2)

In practical measurements with a standard two channel fast Fourier transform analyzer (2ch FFT), ⟨·⟩ is performed automatically using linear averaging function. So, calculation of Equation 1 is easy to perform as a transfer function between ⟨ U n , surf ⟩ and ⟨ P surf ⟩ . Note that EA method with PU-sensor is expected to be more accurate than EA method with PP-sensor because of its geometrical straightforwardness that both particle velocity and sound pressure are measured at almost a single point close to specimen surface. On the other hand, EA method with PP-sensor is easier to be employed because of the stability and cost–e ff ectiveness of precise microphones manufactured for scientific measurements. Both sensor types require appropriate calibrations to ensure preciseness.

Figure 1: Measurement settings of EA method.

3. ADDITIONAL MEASUREMENT OUTLINE

In this manuscript, outline of measurements by EA method with PU-sensor is discribed.

3.1. Measurement settings As illustrated in Figure 1, a PU-sensor (Microflown Co., PU regular) is placed at 1 cm above the specimen; and it is plagued to a 2ch FFT(B&K Co., PULSE). From experience, the distance from specimen to sensor is determined safer side so as not to destroy the sensitive sensors at various field environments. The direction of the PU-sensor is set normal to the material surface. Transfer function H U , P between particle velocity and sound pressure sensors is calculated on the 2ch FFT in the frequency domain. Linear averaging in the time domain is performed 150 times applying a Hanning window with 0.33 s time length. Therefore, net measurement time for a single EA method finishes within 49.5 s.

Pseud random noises YR PU-sensor or? Mic. 2 channel FF Analyzer Specimen Hard Floor x: measurement point

3.2. Sensor calibration We recommend on site calibration when a PU-sensor is utilized because atmospheric conditions, especially humidity, a ff ects sensor sensitivity to some extent [7]. If one needs accurate values, onsite calibration is preferable. It is often the case that atmospheric conditions at field environments fluctuate rather widely, e.g. relative humidity di ff erence rises greater than 5 %. Therefore, at all measurements presented here, the PU-sensor was calibrated just before starting the EA method measurement using an acoustic tube [6].

3.3. Sound source During the previous measurements at Arena-X, a group of six persons were playing badminton in the other end of the arena. To avoid COVID-19 infection, all machinery ventilators were kept turned on making background noise and all doors were kept open. However, during the previous measurements at Arena-Univ, all doors were kept open but no machinery ventilator is equipped. Moreover, Arena- Univ locates in the woods, and environmental noise level is usually very low. When the additional measurements are carried out at Arena-X, all the ventilators are kept turned on and all the doors are kept open like previous measurements. While, there are no extra people except measurement sta ff s. Then, to realize a pseudo random incidence condition, six portable loudspeakers (JBL Co., Micro Wireless) that respectively radiate incoherent pink noises are moved around by three persons’ hands on a virtual sphere with about 1.5 m radius from the centre of the sample. The loudspeakers might be too tiny to supply enough acoustic power in the frequency region especially less than 200 Hz.

3.4. Sports arenas Ensemble averaged surface normal impedance ⟨ Z n ⟩ and corresponding absorption coe ffi cient ⟨ α ⟩ of materials are measured by EA_pu method at a sports arena "Arena_X" (about 3,500 m 2 , about 4,000 audience seats) in a public gymnasium where some of previous measurements were conducted. The previous measurements were conducted at two arenas: one is at Arena_X; and another is at "Arena_Univ" (about 500 m 2 , no audience seat) in Oita University Second Gymnasium.

Figure 2: Snapshot of EA method measurement of vertical concrete wall CN_wl at Arena_Univ..

3.5. Materials Measured materials are classified into two kinds: one is portable materials suit for comparisons between measurement places; and another is building onsite materials installed to walls and floors. As for the portable material, sound absorption characteristics of a glass-wool panel "GW_pl" with the dimensions of 0.6 m × 0.6 m × 0.05 m and density of 32 kg m − 3 are measured. It is laid on

the wooden floor and measurement point is set at the centre of the panel. Note that the same panel was brought around and its absorption characteristics were measured in the previous measurements at both Arena_X and Arena_Univ.. Based on the results and discussion given in literature [10], GW_pl is placed carefully to avoid unnecessary vibrations. As for the onsite materials, the arenas’ wooden walls "WD_wl" and concrete walls "CN_wl" are measured at Arena-Univ. Here, though both walls stand vertically, there is no distinct di ffi culty for the EA method to measure absorption characteristics of vertical specimens (Figure 2). For both WD_wl and CN_wl, measurement points are selected where PU-sensor is easy to set and pseudo-random incidence condition is expected. Measurements of GW_pl and CN_wl are repeated three times, while, those of WD_wl are two times.

4. RESULTS AND DISCUSSION

4.1. Portable specimen: GW_pl Figure 3 shows comparisons of measured ⟨ α ⟩ values of GW_pl between times and places. As for places, results measured at two arenas, Arena_X and Arena_Univ., are compared to those at Oita University reverberation room, Rev. Room. While, as for times, results of additional measurements (Add.) and those of previous measurements (Prev. and Reference) are plotted together. Here, mean values of three times measurements in one-third octave band are plotted to compare. On the whole, excellent agreements are observable except ⟨ α ⟩ values of Arena-X (Prev.) in the frequency region between 100 Hz and 400 Hz. The discrepancies are considered to be caused by unnecessary vibration of glass-wool panel which diminishes in additional measurements with careful material installation.

Figure 3: Comparisons of ⟨ α ⟩ values of GW_pl between times (Add. and Prev.) and places (Arena-X, Arena-Univ and Reverberation Room).

4.2. Onsite materials on vertical walls: WD_wl and CN_wl Provided that a specimen is surrounded by su ffi ciently large and hard floor / wall, we can expect EA method to provide appropriate sound absorption characteristics of the specimen regardless of the specimen’s location. To prove it, two onsite materials on vertical walls are measured: WD_wl and CN_wl. Figure 4 shows measured ⟨ α ⟩ values of WD_wl and CN_wl by the EA method. There, reference data [11] are plotted to compare. Note that the reference data are one-octave-band reverberation room absorption coe ffi cient. Taking account of the measurement situations, considerably good measurement repeatability of the EA method is confirmed for both materials. Moreover, overall tendency of ⟨ α ⟩ values of CN_wl show excellent agreement with that of the reference data.

Glass-wool panel (r= SOmm) Absorption Coefficient, | ‘Arena-X(Add.) Frequency (Hz)

Whereas, the agreement of ⟨ α ⟩ values between WD_wl and the reference data seems not very good. It is probably because that materials of WD_wl and of reference book are not very similar. Besides, it is not easy for employed small loudspeakers to radiate enough sound energy in the frequency region below 400 Hz. In addition, Arena-Univ locates in the woods and environmental background noise is very low. Then, we consider that such zigzag fluctuations are observed in the low frequency region of ⟨ α ⟩ values of WD_wl.

Figure 4: Measured ⟨ α ⟩ values of WD_wl (left, blue and red lines) and CN_wl (right, blue, red and orange lines) at Arena-Univ Reference data [11], reverberation room absorption coe ffi cient, are plotted in black lines.

5. CONCLUSIONS

To confirm the applicability of the EA method, a series of sound absorption measurement is conducted. Results of portable material, glass-wool panel, measured at two sports arenas in field environments show good agreements with those at a reverberation room. In former studies, we already confirmed that the results of the reverberation room agree well with those measured at di ff erent two reverberation rooms [8]. Moreover, absorption characteristics of two materials installed to vertical walls are measured and results show satisfactory tendencies and agreements with literature data.

ACKNOWLEDGEMENTS

The authors are grateful Mr. Tajima for supporting this work continuously as technical sta ff . This work was partly supported by JSPS KAKENHI Grant Number 19H02298.

REFERENCES

[1] Yasuo Takahashi, Toru Otsuru, and Reiji Tomiku. In situ measurements of surface impedance and absorption coe ffi cients of porous materials using two microphones and ambient noise. Applied Acoustics , 66(7):845–865, 2005. [2] Toru Otsuru, Reiji Tomiku, Nazli Bin Che Din, Noriko Okamoto, and Masahiko Murakami. Ensemble averaged surface normal impedance of material using an in-situ technique: Preliminary study using boundary element method. The Journal of the Acoustical Society of America , 125(6):3784–3791, 2009. [3] Hans-Elias de Bree, Peter Leussink, Twan Korthorst, Henri Jansen, Theo SJ Lammerink, and Miko Elwenspoek. The µ -flown: a novel device for measuring acoustic flows. Sensors and actuators A: Physical , 54(1-3):552–557, 1996.

_Abwrpton Cota (=) Wooden wall (WD_w)) ‘Avonpion Cot {+1 Concrete wall (CN_l) Refrece Da

[4] Toru Otsuru, Reiji Tomiku, Noriko Okamoto, Emi Ueda, and Asami Nakamura. Uncertainty examination of ensemble averaging method for sound absorption of materials in a reverberation room. Acoustical Science and Technology , 37(3):129–132, 2016. [5] Michael Vorländer. Computer simulations in room acoustics: Concepts and uncertainties. The Journal of the Acoustical Society of America , 133(3):1203–1213, 2013. [6] Nazli Bin Che Din, Toru Otsuru, Reiji Tomiku, Noriko Okamoto, and Kusno Asniawaty. Measurement method with a pressure-velocity sensor for measuring surface normal impedance of materials using ensemble averaging: Comparison with other methods and its geometrical configuration. Acoustical Science and Technology , 33(2):86–95, 2012. [7] Toru Otsuru, Reiji Tomiku, Noriko Okamoto, and Saki Yamauchi. Sound absorption deviation measured using pressure–velocity sensors by ensemble averaging technique under di ff erent relative humidity conditions. Applied Acoustics , 122:121–127, 2017. [8] Noriaki Sakamoto, Toru Otsuru, Reiji Tomiku, and Saki Yamauchi. Reproducibility of sound absorption and surface impedance of materials measured in a reverberation room using ensemble averaging technique with a pressure-velocity sensor and improved calibration. Applied Acoustics , 142:87–94, 2018. [9] Noriaki Sakamoto, Toru Otsuru, Reiji Tomiku, Ippei Eto, Daiki Matsuoka, and Ryota Yoshimoto. Investigating the practicability of an ensemble averaging method for measuring sound absorption and surface normal impedance. Applied Acoustics , 156:166–175, 2019. [10] Toru Otsuru, Reiji Tomiku, Noriko Okamoto, and Siwat Lawanwadeekul. Trial applications at gymnasiums of in-situ sound absorption measurement method by ensemble averaging technique. In INTER-NOISE and NOISE-CON Congress and Conference Proceedings , volume 263, pages 4532–4537. Institute of Noise Control Engineering, 2021. [11] Zyun-iti Maekawa, Jens Rindel, and Peter Lord. Environmental and architectural acoustics . CRC Press, 2010.