Acoustic measurements and psychoacoustic analyses of ventilation di ff users

Lara Stürenburg a , 1

Philipp Ostmann b , 2

Lukas Aspöck a

Dirk Müller b

Janina Fels a

a Institute for Hearing Technology and Acoustics, RWTH Aachen University Kopernikusstraße 5, 52074 Aachen, Germany

b RWTH Aachen University, E.ON Energy Research Center, Institute for Energy E ffi cient Buildings and Indoor Climate Mathieustraße 10, 52074 Aachen, Germany

ABSTRACT Since ventilation systems are not only used in industry but also increasingly in residential buildings, it is important to study its sound radiation and the consequences for our well-being. This research project aims to identify if geometric features of air di ff users a ff ect the rated annoyance of ventilation systems. It is known that the A-weighted sound serves as a criterion to a limited extent. Therefore, psychoacoustic analyses and evaluations need to be included to find out if there are correlations between psychoacoustic features and flow phenomena caused by the geometric features of the air di ff users. Several air di ff users were acoustically measured in a hemi-anechoic room using a developed mobile ventilation unit that supplied the necessary air volume flow. The ventilation unit could be operated in both supply and extract air configuration of the di ff users. With the measured data, the acoustic directivity of the air di ff users can be distinguished, and psychoacoustic features can be evaluated. Future listening experiments will also be extended to audio-visual experiments in virtual reality to increase ecological validity and study the role of the environment.

1. INTRODUCTION

The thermodynamic performance of air di ff users is often limited by the emitted A-weighted sound pressure level. But the A-weighted sound pressure level serves as a criterion to a limited extent when assessing ventilation noise on annoyance and pleasantness. Therefore, psychoacoustic analyses and evaluations need to be included in the assessment process. Susini et al. [1] conducted a psychoacoustic study where unpleasant sound emissions from air-conditioning units were identified with the help of

1 lara.stuerenburg@akustik.rwth-aachen.de

2 philipp.ostmann@eonerc.rwth-aachen.de

21-24 AUGUST SCOTTISH EVENT CAMPUS ? O? ? GLASGOW

listening experiments. However, the results were not linked to flow phenomena, which is an important factor when investigating air di ff users and not rotating machines such as air-conditioning units or fans. During an investigation of the acoustic quality of fans by Töpken and van de Par [2] [3], the three most important noise groups were “unpleasant”, “humming”, and “shrill”. The fan noises could be separated into these adjective groups regarding their perception, although they had the same A-weighted sound pressure level. This shows the importance of psychoacoustic parameters when investing in ventilation noises. Despite the great o ff ered information based on test person studies and psychoacoustic parameters, the results do not provide design recommendations. However, numerical simulation of two exhaust valves by Saarinen and Koskela [4] showed that even minimal changes in the geometrics could result in significant flow noise changes. Therefore, this research project aims to identify if geometric design features of air di ff users enhance specific psychoacoustic features and flow phenomena. In this paper, the acoustic measurements of air di ff users are described, and the first results of the psychoacoustic evaluation are shown. Small deviations in loudness and sharpness between the di ff erent air di ff users can be detected.

2. MEASUREMENT SETUP

Several air di ff users were acoustically measured in a hemi-anechoic room. A developed mobile ventilation unit controllably supplies the necessary air volume flow to the air di ff users.

2.1. Fluid Mechanics The ventilation unit of the investigated system is equipped with a backwards-curved radial fan which is mounted inside an isolated housing to prevent noise radiation. Using a flexible hose, the fan is connected to a calibrated measurement section, which is then connected to the air di ff user. The ventilation unit is placed outside a hemi-anechoic room with only the hose entering it. The hose is guided through a small hole, which was thoroughly insulated to prevent noise from entering the hemi-anechoic room. If necessary, the hose is widened up smoothly to allow the connection to the investigated air di ff user. Depending on the configuration, the di ff users are operated in supply air direction or extract air direction. The volume flow ˙ V is measured with an orifice that was calibrated using a reference orifice that meets the standards of DIN EN ISO 5167-2 . By measuring the pressure drop ∆ p the volume flow ˙ V can be calculated by using the following expression, where ξ is the pressure loss coe ffi cient of the orifice and A the section’s cross-sectional area.

s

2 · ∆ p

ρ · ξ , with A = π

˙ V = A ·

4 · D 2

The ventilation unit can be equipped with di ff erent orifices to allow a wider volume flow range since a certain minimum pressure drop ∆ p is required for accurate measurements. For the targeted volume flow range, two orifices are used, with both using the same section diameter D = 139 . 5 mm. The bigger orifice was calibrated to ξ high = 15 . 45 and the smaller one to ξ low = 56 . 13. The air density ρ is calculated from the measured air temperature, relative humidity and absolute pressure by using the Magnus formula [5] and the ideal gas law. Due to the operating limits of the ventilation unit, the maximum achievable volume flow is at 600 m 3 / h. In the following, the investigated air di ff users are briefly described. The di ff users cover a broad range of rated volume flows (200 m 3 / h - 800 m 3 / h) and are therefore suited for a wide range of possible applications or di ff erent room sizes. An overview of all di ff users is shown in Figure 1.

2.1.1. Swirl di ff users The di ff user with the highest rated volume flow is a TROX TDF 600 (see Figure 1a). It is rated at a volume flow of 800 m 3 / h and its swirl vanes are manufactured by sheet punching. The vanes have a

swept shape. The other swirl di ff users are rated at 700 m 3 / h and are mainly distinguished by their vanes. The Wildeboer DTQ (see Figure 1b) is also manufactured by sheet punching, with small spacing between the vanes. The Wildeboer DXQ (see Figure 1c) has di ff erently sized plastic vanes that are inserted into the main plate alternatingly.

2.1.2. Slot di ff users As a rather simple slot di ff user, the TROX TSD (see Figure 1d) was selected with two slots. It is rated at 350 m 3 / h at the selected length of 1 m. Its vanes can be rotated, allowing for either a ceiling- attached or ceiling-normal airflow. A more complex slot di ff user with four slots is the LTG LWmodule 12clean (see Figure 1e). The flow is guided by rotatable cylinders. In addition to the main flow through the cylinders, an air curtain is produced by thin bypass channels over the whole length. The di ff user is rated at 400 m 3 / h at the selected length of 1 m and fully opened cylinder position. Since only one of the two connection ducts is used, the e ff ective rated volume flow reduces to 200 m 3 / h. This setup was recommended for measurements by the manufacturer.

Figure 1: Selected air di ff users: a) swirl di ff user from TROX, b) swirl di ff user DTQ from Wildeboer, c) swirl di ff user DXQ from Wildeboer, d) slot di ff user from TROX, and e) slot di ff user from LTG.

2.2. Acoustics The measurements took place in a hemi-anechoic room (see Figure 2) with a lower limiting frequency of 100 Hz. To reconstruct the environment, the air di ff users were integrated into a ceiling tile construction of 2.4 x 2.4 m 2 . A microphone arc with a radius of 195 cm measured the acoustic signals for each air di ff user at four di ff erent room positions, and three volume flows (200 m 3 / h, 400 m 3 / h and 600 m 3 / h). The microphone arc contains 19 microphone capsules (Sennheiser KE 4-211-2) at a 5° distance from each other. The first microphone (No. 1) was placed parallel to the air di ff user, and the last microphone (No. 19) was centred directly above the air di ff user. To investigate the directional radiation patterns of the air di ff users without increasing the measurement e ff ort considerably, four microphone arc positions were chosen at 45° intervals. The room positions were: 180 - as in Figure 2, opposite of the hose, at 180° to the hose; 135 - in the Figure at the left corner, at 135° to the hose; 90 - in the Figure at the front site, at 90° to the hose; and 45 - in the Figure at the front corner, at 45° to the hose.

Figure 2: Measurement setup with the microphone arc in a hemi-anechoic room. The air di ff users were integrated into a ceiling tile construction.

Further measurements were taken with a low noise microphone (G.R.A.S. 40HL ½”) positioned 195 cm above the air di ff user as microphone No. 19 of the microphone arc. For all measurement setups and configurations, three recordings were conducted, each with a length of 11 seconds.

3. RESULTS

With the sound pressure levels of the 19 signals from the microphone arc at four room positions, the sound power levels were calculated according to the standard DIN EN ISO 3745. The results for all air di ff users and the three di ff erent volume flows are listed in Table 1. 3 The “B.” represent the second vane position of the slot di ff user from TROX, which emits ceiling-attached airflow, while the first vane position emits ceiling-normal airflow. The table shows that the sound power level increases for all air di ff users when the volume flow increases. The slot di ff user from LTG has the lowest sound

Table 1: The sound power level in dB of the air di ff users for di ff erent volume flows. 3

Di ff user Volume Flow (m 3 / h)

200 400 600

Slot TROX 59.4 68.6 75.3

Slot TROX B. 58.6 67.9 74.4

LTG Extract * 53.5 62.3 68.8

LTG Supply * 53.3 60.8 68.1

Swirl TROX 60.9 70.1 75.8

WB DTQ 60.4 69.2 75.7

WB DXQ 60.0 68.1 74.4

3 Only one half of the slot di ff user from LTG was measured. The volume flow was halved to compare it with the other air di ff users.

65

65

Slot TROX Slot TROX B. LTG Extract * LTG Supply *

TROX WB DTQ WB DXQ

Slot TROX Slot TROX B. LTG Extract * LTG Supply *

TROX WB DTQ WB DXQ

Sound Pressure Level (dB SPL)

Sound Pressure Level (dB SPL)

60

60

55

55

50

50

45

45

40

40

35

35

30

30

200 400 600 Volume Flow (m³/h)

200 400 600 Volume Flow (m³/h)

(a) Microphone No. 19

(b) Low noise microphone

Figure 3: Comparison of the band-pass filtered sound pressure level in dB SPL between (a) the microphone No. 19 of the microphone arc and (b) the low noise microphone. They measured at the same position: centred 195 cm above the air di ff users.

power levels of all air di ff users for each volume flow. 3 The sound power levels of the slot di ff user from TROX in both vane positions are similar to the sound power levels of the swirl di ff users from TROX and Wildeboer.

In Figure 3 the sound pressure level in dB SPL is shown over the volume flow for the microphone No. 19 of the microphone arc (Figure 3a) and the low noise microphone (Figure 3b). These two microphones were placed at the same position: centred 195 cm above the air di ff users. The signals were band-pass filtered between 100 Hz and 4 kHz. It can be seen that the sound pressure levels compared between these two microphones are very similar, and di ff er by a maximum of 3 dB for the measurements of the slot di ff user from TROX in the first vane position at a volume flow of 600 m 3 / h.

In Figure 4a the sound pressure level in dB SPL is shown over the frequency in Hz for all 19 microphones of the microphone arc. The spectrum is shown for the swirl di ff user of TROX at a volume flow of 600 m 3 / h and room position 90 . The highest sound pressure levels are between 75 Hz and 150 Hz. For frequencies above 150 Hz, the sound pressure levels decrease almost logarithmically with the frequency. In Figure 4b the directivity of the sound pressure level in dB SPL measured with the microphone arc can be seen. Shown is the measured deviation of the sound pressure level in reference to microphone No. 19 (at 0°) of the swirl di ff user from TROX at a volume flow of 600 m 3 / h and room position 90 . The one-third octave bands 250 Hz, 500 Hz, 1 kHz, 2 kHz, 4 kHz, and 8 kHz are plotted for all microphones of the microphone arc. Microphone No. 1 is marked at 90°, going down in 5°-steps to microphone No. 19 at 0°. It can be seen that the di ff erences between the microphones are smaller than ± 10 dB SPL. Parallel to the air di ff user (75°-90°; microphones No. 4-1) the frequency bands of 1 kHz and 2 kHz have higher sound pressure levels than at the more vertical microphone positions (0°-30°; microphones No. 19-13). For the frequency band 250 Hz the sound pressure level decrease in the parallel position (60°-90°), and also the sound pressure level of the frequency band 500 Hz is low from 45°-90° (microphones No. 10-1).

The psychoacoustic values in Figure 5 and 6 were evaluated with the software ArtemiS Suite by HEAD acoustics. The given values are the mean values of three conducted measurements in each condition. In Figure 5a the A-weighted sound pressure level in dB is shown over the volume flow in m 3 / h for all

60

TROX-90-600

TROX-90-600 Ch01 Ch02 Ch03 Ch04 Ch05 Ch06 Ch07

Ch08 Ch09 Ch10 Ch11 Ch12 Ch13 Ch14

Ch15 Ch16 Ch17 Ch18 Ch19 Ch24 (ref)

0 30

Sound Pressure Level (dB SPL)

10

330

40

5

60

300

0

20

250 Hz 500 Hz 1 kHz 2 kHz 4 kHz 8 kHz

-5

90

270

-10

0

-20

120

240

-40

150 180 210

100 1k 10k Frequency (Hz)

(b) Directivity

(a) Spectrum

Figure 4: (a) Spectrum of the swirl di ff user from TROX measured with the 19 microphones of the microphone arc at a volume flow of 600 m 3 / h and position 90 . (b) Directivity of the sound pressure level in dB SPL of the same air di ff user at the same room position as in (a), plotted for six one-third octave bands in reference to microphone No. 19 (at 0°).

measured air di ff users. For all air di ff users, except the slot di ff user from LTG, the A-weighted sound pressure level increases approximately linear with increasing volume flow. In Figure 5b the loudness in sone is shown over the volume flow in m 3 / h for all air di ff users with a logarithmic ordinate. It was calculated according to the DIN 45631. The loudness increases approximately logarithmically when increasing the volume flow for all air di ff users, except the slot di ff user from LTG. It can be determined that the loudness for all air di ff users is approximately three times higher when doubling the volume flow from 200 m 3 / h to 400 m 3 / h. When increasing the volume flow from 400 m 3 / h to 600 m 3 / h, the loudness approximately doubles.

In Figure 6a the sharpness in acum is shown over the volume flow for all measured air di ff users. The sharpness is calculated according to the DIN 45692. With the logarithmic ordinate, a logarithmic

55

8

Slot TROX Slot TROX B. LTG Extract * LTG Supply *

TROX WB DTQ WB DXQ

Slot TROX Slot TROX B. LTG Extract * LTG Supply *

TROX WB DTQ WB DXQ

Sound Pressure Level (dB (A))

50

4

45

Loudness (sone)

2

40

35

0.8 1

30

25

0.4

20

200 400 600 Volume Flow (m³/h)

200 400 600 Volume Flow (m³/h)

(b) Loudness

(a) A-weighted sound pressure level

Figure 5: (a) A-weighted sound pressure level in dB and (b) loudness in sone over the volume flow for all measured air di ff users.

1.4 1.6

0.2

Slot TROX Slot TROX B. LTG Extract * LTG Supply *

TROX WB DTQ WB DXQ

Slot TROX Slot TROX B. LTG Extract * LTG Supply *

TROX WB DTQ WB DXQ

1.2

0.16

1

Roughness (asper)

Sharpness (acum)

0.8

0.12

0.6

0.08

0.4

0.04

200 400 600 Volume Flow (m³/h)

200 400 600 Volume Flow (m³/h)

(b) Roughness

(a) Sharpness

Figure 6: (a) Sharpness in acum and (b) roughness in asper over the volume flow for all measured air di ff users.

sharpness increase can be seen for an increasing volume flow. The maximum sharpness is calculated for the slot di ff user from LTG in the supply air setting with 1.27 acum at 600 m 3 / h, and the lowest sharpness is calculated for the swirl di ff user from TROX with 0.35 acum at 200 m 3 / h. In Figure 6b the roughness in asper is shown over the volume flow for all measured air di ff users and is calculated according to the ECMA-418-2 [6]. The roughness is minimal for the slot di ff user from LTG in the supply air setting with 0.03 asper at 200 m 3 / h and is maximal for the slot di ff user from TROX in both vane positions with 0.18 asper at 600 m 3 / h. It is noticeable that with an increasing volume flow from 400 m 3 / h to 600 m 3 / h, the roughness does not increase strongly for any air di ff users.

4. DISCUSSION

The sound pressure level and the sound power level increase for all air di ff users with increasing volume flow. The di ff erences in the sound pressure level between the air di ff users become smaller with increasing volume flow. It was shown that the measurements between microphone No. 19 of the microphone arc and the low noise microphone show little di ff erence in the sound pressure levels. Therefore it can be inferred that the data of both measurement setups are suitable, and the recordings are reproducible. The directivity of the sound pressure level showed an angle di ff erence divided into di ff erent frequencies. The angle di ff erence can be explained due to the geometric structure of the air di ff users, which is often such that the airflow is ceiling-attached. In the shown Figure 4b, the one-third octave bands of 1 kHz and 2 kHz have an increased sound pressure level for the angles near the ceiling (75°-90°). Similar directivity patterns are given when looking at the other measured air di ff users. For the measurements without any air di ff user, only with the hose and the ceiling tile construction, the directivity pattern is similar. It even shows increased sound pressure levels for the 1 kHz and 2 kHz one-third octave bands. This would imply that the geometric design of the air di ff users reduces the sound pressure levels of these frequencies and that there is further potential to decrease these sound pressure levels by geometric modifications to the di ff users. The results of the A-weighted sound pressure level and the loudness are similar. It is noticeable that the loudness increases logarithmically. This could be an important factor when assessing the noises according to their pleasantness and annoyance. This must be further investigated through listening experiments with subjects. It is noteworthy to look at the results for the slot di ff user from LTG in the extract air setting and the supply air setting. Especially at a volume flow of 200 m 3 / h, the sound power levels, loudness, and roughness di ff er significantly from the results of the other air di ff users. This can be

due to the fact that only one half of the air di ff user was measured. The levels will add up in full operation, resulting in a minimum 3 dB increase in the sound power level. However, this air di ff user has increased sharpness compared to the other air di ff users. Nevertheless, the sharpness values are not high. This is due to the fact that the signals contain mainly frequencies below 1 kHz, and the calculation of sharpness includes a weighting that only takes e ff ect above 16 Bark ( ≈ 3 kHz). Also, the perceived sharpness is less when there is a broad spectrum of low frequencies in signals [7]. The roughness values are relatively small, which can be expected for unmodulated broadband noise [8]. The threshold for roughness is at approximately 0.1 asper [7]. Therefore, the sounds of all air di ff users generated with a volume flow of 200 m 3 / h might not be perceived as rough. For the sounds generated with higher volume flows of 400 m 3 / h and 600 m 3 / h, the roughness might be perceived but with values between 0.11 and 0.18 not as strongly.

5. CONCLUSION

In this study, acoustic signals from five di ff erent air di ff users were measured, two of them in di ff erent settings. The results of this study seem to indicate that there are di ff erences between the manufacturers and types of air di ff users regarding the psychoacoustic parameters loudness, sharpness and roughness. These findings now need to be confirmed with the help of listening experiments with subjects. Based on the results of the psychoacoustic analyses, numerical simulations and upcoming listening experiments, the air di ff users will be modified in their geometric design to avoid unpleasant sound features.

ACKNOWLEDGEMENTS

This project was funded by the AiF (Arbeitsgemeinschaft industrieller Forschungsvereinigungen) and the German Federal Ministry for Economic A ff airs and Climate Action (BMWK) based on a resolution of the German Bundestag (IGF No.: 21611 N / 1).

REFERENCES

[1] P. Susini, S. McAdams, S. Winsberg, I. Perry, S. Vieillard, and X. Rodet. Characterizing the sound quality of air-conditioning noise. Applied Acoustics , 65:763–790, 2004. [2] S. Töpken and S. van de Par. Perceptual dimensions of fan noise and their relationship to indexes based on the specific loudness. Acta Acoustica United with Acoustica , 105:195–209, 2019. [3] S. Töpken and S. van de Par. Determination of preference-equivalent levels for fan noise and their prediction by indices based on specific loudness patterns. J. Acoust. Soc. Am. , 145:3399–3409, 2019. [4] P. Saarinen and H. Koskela. Flow noise from an exhaust valve - prediction by simulations as compared with measurements. In ROOMVENT 2011 . [5] G. Magnus. Versuche über die Spannkräfte des Wasserdampfs. Annalen der Physik , 137(2):225– 247, 1844. [6] Psychoacoustic metrics for itt equipment — part 2 (models based on human perception). Standard, Ecma International, Geneva, CH, December 2020. [7] H. Fastl and E. Zwicker. Psychoacoustics: Facts and Models . Springer, Berlin, Heidelberg, 3 edition, 2007. [8] D. Havelock, S. Kuwano, and M. Vorländer. Handbook of Signal Processing in Acoustics . Springer, New York, NY, 1 edition, 2008.