Subjective evaluation of wind turbine noise using 3-dimensional audio- visual reproduction system Miki Yonemura 1 Institute of Industrial Science, The University of Tokyo 4-6-1, Komaba, Meguro, Tokyo 153-8505, Japan Hyojin Lee 2 Seoul National University #220 Liberal studies, Seoul National University, 1 Gwanak-ro, Gwanak-gu, Seoul, Korea, 08826 Shinichi Sakamoto 3 Institute of Industrial Science, The University of Tokyo 4-6-1, Komaba, Meguro, Tokyo 153-8505, Japan

ABSTRACT In recent years, the installation of new wind turbines has been promoted in Japan. Since wind turbines are often installed in quiet countryside, it is necessary to properly evaluate the noise impact of wind turbine installation. Since wind turbines are thought to have visual as well as auditory effects, we conducted a subjective evaluation experiment on the loudness and annoyance of wind turbine noise (WTN) and road traffic noise (RTN) by magnitude estimation method, using a three-dimensional audio-visual reproduction system consisting of 6 channels of loudspeakers installed around the sound receiving point and a dome screen. As a result, the loudness and annoyance of WTN tended to be evaluated higher than that of RTN. These results were considered to be influenced by the amplitude modulation called “swish sound” and the low-frequency dominant frequency characteristics, which are characteristics of WTN. In addition, when comparing the differences in loudness and annoyance with and without visual information, the two attributes tended to be lower when visual information was present.

1. INTRODUCTION

Wind power is an important renewable energy source and is used in many countries and regions, especially in Europe. There have been a number of evaluation studies on wind turbine noise annoyance, and several guidelines for wind turbine noise assessment have been published based on the results of these studies, such as Environmental Noise Guidelines for the European Region published by the WHO [1]. Recently, the installation of new wind turbines has been promoted in Japan as well, and the need for wind turbine noise assessment is increasing.

1 m-yone@iis.u-tokyo.ac.jp 2 lee.hj.sel@snu.ac.kr 3 sakamo@iis.u-tokyo.ac.jp

inter.noise 21-24 AUGUST SCOTTISH EVENT CAMPUS GLASGOW

In previous studies, it has been pointed out that wind turbine noise is characterized by the predominance of low-frequency components and amplitude modulation due to the blade rotation [2- 3]. Not only these acoustical characteristics, but visual impressions are also reported to affect the noise annoyance [4-5]. Various studies on sound evaluation by presenting audiovisual stimuli have been conducted not only for wind turbines but also for various sound environments such as transportation noise, and the sound environment of urban spaces [6-8]. In addition, the recent development of technologies related to 3-dimensional virtual reality has made it possible to present sound environments with a higher degree of reality.

The purpose of this study was to evaluate the effects of wind turbine noise, which is increasing in Japan. Subjective tests were conducted to evaluate loudness and annoyance of wind turbine noise, using a 3-dimensional audio-visual reproduction system.

2. METHOD

2.1. Recording and Reproduction System for 3-dimensional Audio-visual Stimuli Figure 1 shows the audio-visual recording system. Audio-visual stimuli were recorded using a panoramic camera (Gopro, Fusion) and an ambisonics microphone (Sennheiser, AMBEO VR MIC). The sound field microphone was connected to a multi-channel recorder (TASCAM, DR-680 MK-II). To analyze the original sound field, a sound level meter (RION, NL-52) was also connected to the recorder and recorded simultaneously.

Camera

Microphone

Panoramic Camera

(Gopro, Fusion)

Recorder

Ambisonics microphone (Sennheiser, AMBEO VR MIC)

Sound Level Meter

Figure 1: Recording system.

For the presentation of audio-visual stimuli, 6-channel speaker system [9] with a dome screen were used (Figures 2 and 3).

Audio-visual stimuli were composed of a 6-channel wave file and a MP4 video. The 4-channel ambisonic sound (A-format) was converted into 6-channel (B-format). The color of the recorded video was adjusted using a video editing software (Adobe, Premier Pro) so that the color of the projected video is close to the original color. In the recording described above, two video data were obtained for each of the front and rear cameras of the panoramic camera, but only the video taken by the front camera was used for the projection. The stimuli were played using a personal computer with a projection software (Amateras dome player).

The sound stimuli played by the computer were input to a digital mixer (YAMAHA, DME64N) via an audio interface (RME, Fireface 802). The digital mixer corrected the frequency response with a 1/3 octave band equalizer. This adjustment included correction for individual loudspeaker differences and shielding by the screen. As shown in Figure 2, sound emitted from the loudspeakers

placed in front of and above the subject is interrupted by the screen and its frequency characteristic changes. The corrected sound is passed through a D/A converter (YAMAHA, DA824) and a speaker amplifier (YAMAHA, XM4080), and is reproduced from 6 loudspeakers (TANNOY, T12) installed in an anechoic room.

As for the presentation of visual stimuli, a dome screen with a diameter of 3.6 m was hung from the ceiling. A projector (BenQ, TK800) was located behind the subject to project the visual stimuli through a fisheye lens.

Figure 2: Diagram of 3-dimensional audio-visual reproduction system.

Personal Computer with Projection Software J Projector ‘Ausdlo Interaco oro — Digital Mixer COOH DIA converter oro ‘Speaker Amp, oro {6ch. Loudspeakers

Figure 3: 3-dimensional audio-visual reproduction system.

2.2. Test Stimuli 12 audio-visual stimuli listed in Table 1 were used. The types of sound source are divided into 2 groups, wind turbine noise (WTN) and road traffic noise (RTN). The power of wind turbines recorded in this study ranged between 1250 and 1670 kW. The A-weighted sound pressure level ( L A,30s ) are measured using omnidirectional microphone (RION, NL-52) in the position of the listener’s head. The presentation level of the sound was adjusted to be equal to the A-weighted sound pressure level at the measurement site. However, in order to adjust the range of presentation levels to be the same for the WTN and the RTN, the R-1 stimulus shown in Table 1 was presented with a reduction of approximately 5 dB from the original level. The A-weighted sound pressure levels were approximately 40 to 55 dB for both WTN and RTN.

Screen shots of visual stimuli are presented in Figures 4 and 5. To study the effect of visual stimuli, two visual stimuli were set up for the same audio stimulus: one was a video recorded simultaneously with the audio stimulus, and the other was an interior view of an anechoic room, which is shown in

Figure 5. The measurement points of W-5 and W-6 were the same and the audio stimuli were similar, but the camera was set up in a different direction to show different views.

The duration of the stimuli was 30 seconds for each. The frequency characteristics of the audio stimuli are shown in Figure 6. Comparing the difference between RTN and WTN, the frequency characteristics of the RTN had a peak in the 500-1000 Hz band, whereas the frequency characteristics of the WTN did not have significant peaks and was prominent in the low frequency range. Figure 7 shows the temporal variability of audio stimuli. As shown in Figure 7(a), R-1~3 were recorded at a distance of 100 m away from the road and showed steady state, while R-4 and R-5, which were recorded near the road (20~50 m), showed fluctuation due to passing vehicles. As for the WTN, W5- 7 had periodical fluctuation in levels with a period of about one second (swish sound).

Table 1: A-weighted sound pressure levels of test stimuli.

Sound Source Type ID L A, 30s (dB)

W-1 W-2 W-3 W-4 W-5 W-6 W-7

37.1 42.6 47.0 49.4 51.7 54.1 55.1

Wind turbine noise

R-1 R-2 R-3 R-4 R-5

40.0 44.4 49.9 52.7 57.7

Road traffic noise

Figure 4: Screen shots of visual stimuli (WTN).

Figure 5: Screen shots of visual stimuli (RTN and anechoic room).

(a) Road Traffic Noise 1-5 (b) Wind Turbine Noise 1-4 (c) Wind Turbine Noise 5-7

60

60

60

1/3 oct. band S.P.L.(dB)

50

50

50

40

40

40

30

30

30

R-1, 40.0 dB R-2, 44.4 dB R-3, 49.9 dB R-4, 52.7 dB R-5, 57.7 dB

W-1, 37.1 dB W-2, 42.6 dB W-3, 47.0 dB W-4, 49.4 dB

W-5, 51.7 dB W-6, 54.1 dB W-7, 55.1 dB

20

20

20

10

10

10

63 125 250 500 1k 2k 4k 8k

63 125 250 500 1k 2k 4k 8k

63 125 250 500 1k 2k 4k 8k Center Frequency (Hz)

Figure 6: Frequency characteristics of audio stimuli.

Center Frequency (Hz)

Center Frequency (Hz)

(a) Road Traffic Noise 1-5 (b) Wind Turbine Noise 1-4 (c) Wind Turbine Noise 5-7

70

70

70

W-1, 37.1 dB W-2, 42.6 dB

W-3, 47.0 dB W-4, 49.4 dB

W-5, 51.7 dB W-6, 54.1 dB

W-7, 55.1 dB

60

60

60

L A,Fast (dB)

50

50

50

40

40

40

R-1, 40.0 dB R-2, 44.4 dB

R-3, 49.9 dB R-4, 52.7 dB

R-5, 57.7 dB

30

30

30

0 5 10 15 20 25 30

0 5 10 15 20 25 30

0 5 10 15 20 25 30

Figure 7: Temporal variability of audio stimuli.

Time (sec)

Time (sec)

Time (sec)

2.3. Procedure The loudness and the annoyance of test stimuli were evaluated by ME method. The participants were asked to listen to the test stimuli and evaluate the loudness/annoyance of the stimulus. The experiment was conducted in two separate sessions, and participants evaluated loudness in one session and annoyance in the other. After several rehearsals, the main experiment was conducted: the 24 test stimuli were presented twice each in random order.

12 subjects with normal hearing ability were attended to the evaluation experiments. 6 of the 12 participants had the loudness evaluation session first, while the others had the annoyance evaluation session first.

3. Results and Discussions

A total of 24 responses were obtained for each test stimulus, two times each from 12 participants. All the ME values assigned to the loudness and the annoyance were geometrically averaged for each test stimulus.

3.1. Loudness Figure 8 shows the relationships between perceived loudness and L A,30s . The circles show the WTN stimuli and triangles show the RTN stimuli. The filled symbols represent the stimuli with original video (i.e., with visual information) and open symbols represent the stimuli with the picture of the anechoic room (i.e., without visual information). Statistical tests (Wilcoxon signed-rank test) were used to examine the effect of the visual stimuli, and pairs of stimuli with significant differences (p<0.1) are marked with asterisks.

Comparing the WTN and the RTN, the perceived loudness of the WTN stimuli were higher than that of the RTN at the same L A,30s , showing the same trend as in previous studies [6, 8]. In the interviews conducted after the experiment, several participants commented that steady sounds such as the WTN seemed louder than time-varying sounds such as the RTN.

Focusing on the difference between the stimuli with and without visual information, the stimuli with visual information tended to be perceived as smaller than those without visual information. Significant differences were found in all five conditions for the RTN and in three of the seven conditions for the WTN in which the L A,30s was moderate (47 – 52 dB).

35 40 45 50 55

RTN, video RTN, anechoic room WTN, video WTN, anechoic room

*

** **

*

Loudness (-)

30

*

**

25

20

**

N = 12 **: p <0.05 * : p <0.1

15

*

10

30 35 40 45 50 55 60

L A,30 s (dB)

Figure 8: Relationships between perceived loudness and A-weighted sound pressure levels.

3.2. Annoyance Figure 9 shows the relationships between perceived annoyance of the stimuli and L A,30s . The plots are the similar to those in Figure 8. Comparing the WTN and the RTN, the annoyance of WTN stimuli were higher than that of the RTN at the same L A,30s , which is similar to the results of the loudness evaluation ! described in section 3.1. In the interviews conducted after the experiment, several participants commented that the fluctuating sound and the low, steady sound were annoying. This can be explained by the physical characteristics of audio stimuli (time variability and frequency characteristics) shown in Figures 6 and 7.

Focusing on the difference due to visual information, regarding the RTN stimuli, only two pairs of stimuli showed significant differences in annoyance, while significant differences were found in all stimuli in the loudness experiment. On the other hand, regarding the WTN stimuli, significant differences were found in four of the seven pairs, and the influence of visual information was similar to that in the loudness evaluation experiment.

35 40 45 50 55

** *

**

RTN, video RTN, anechoic room WTN, video WTN, anechoic room

**

Annoyance (-)

30

** **

25

20

N = 12 **: p <0.05 * : p <0.1

15

10

30 35 40 45 50 55 60

L A,30 s (dB)

Figure 9: Relationships between perceived annoyance and A-weighted sound pressure levels. 4. CONCLUSIONS

In order to evaluate the perception of WTN, a 3-dimensional audio-visual reproduction system was constructed and loudness and annoyance of WTN were evaluated, in comparison with RTN. As a result, both loudness and annoyance were rated higher for WTN than for RTN, and it was considered that this was partly due to the temporal variability (swish) and frequency characteristics (low frequency band dominance) of WTN. As for the effect of visual information, both loudness and annoyance tended to be lower when the visual stimuli were presented simultaneously with audio stimuli. The effect of visual information depended on the type of sound source, and further study is needed to clarify the cause of this effect. 5. ACKNOWLEDGEMENTS

This work was supported by JSPS KAKENHI Grant Number JP17H03351. 6. REFERENCES

1. World Health Organization, Environmental Noise Guidelines for the European Region; World

Health Organization Regional Office for Europe: Geneva, Switzerland (2018), 2. Tachibana, H., Yano, H., Fukushima, A., Sueoka, S. Nationwide field measurements of wind

turbine noise in Japan. Noise Control Engineering Journal , 62(2) , 90–101 (2014). 3. Cooper, S. Wind Farm Noise—Modulation of the Amplitude. Acoustics , 3 (2) , 364–390 (2021). 4. Pedersen, E., Larsman, P. The impact of visual factors on noise annoyance among people living

in the vicinity of wind turbines. Journal of Environmental Psychology , 28 (4) , 379–389 (2008). 5. Szychowska, M., Hafke-Dys, H., Preis, A., Kociński, J., Kleka, P. The influence of audio-visual

interactions on the annoyance ratings for wind turbines. Applied Acoustics , 129 , 190–203 (2018). 6. Fastl, H. Audio-visual interactions in loudness evaluation. Proc. of ICA 2004, 1-6 (2004). 7. Preis, A., Kociński, J., Hafke-Dys, H., Wrzosek, M. Audio-visual interactions in environment

assessment. Science of The Total Environment , 523 , 191–200 (2015). 8. Asakura, T., Tsujimura, S., Yonemura, M., Hyojin, L., Sakamoto, S. Effect of immersive visual

stimuli on the subjective evaluation of the loudness and annoyance of sound environments in urban cities. Applied Acoustics , 143 , 141–150 (2019). 9. Yokoyama, S., Ueno, K., Sakamoto, S., Tachibana, H., 6-channel recording/reproduction system

for 3-dimensional auralization of sound fields. Acoustical Science and Technology , 23(2) , 97– 103 (2002).