Applicability of ISO 16283-3 for field measurement of sound insulation of partially open windows Lars Sommer Søndergaard 1 Rune Egedal 3

Rasmus Stahlfest Holck Skov 4 FORCE Technology Agro Food Park 13, 8200 Aarhus N, Denmark Birgit Rasmussen 2 BUILD – Department of the Built Environment Aalborg University, Copenhagen, Denmark

ABSTRACT Façade sound insulation regulations are typically focused on closed windows. However, many peo- ple prefer open windows for ventilation purposes, or simply because of the psychological effect of having an open window. As such it is important to be able to correctly quantify the sound insula- tion, also with open windows. The international standard ISO 16283-3 describes a field method for test of sound insulation of facades or façade elements, e.g. a window, which is further explained in the scope: “The element methods aim to estimate the sound reduction index of a façade element, for example, a window. The most accurate element method uses a loudspeaker as an artificial sound source. Other less accurate element methods use available traffic noise”. However, the standard is probably primarily meant for closed windows, and not for open windows. The applicability of ISO 16283-3 for partial open windows is therefore under investigation for such conditions, which are included in an additional Danish environmental noise guideline. Generally, it can be concluded that the traffic noise method is applicable, but care should be taken by using the loudspeaker method for partially open windows, since the results depend highly on the window opening position compared to the loudspeaker position.

1. INTRODUCTION

Typically, regulations for sound insulation for facades and windows are concerned only with closed windows, and accordingly the standards for quantifying the sound insulation are typically designed for closed windows. One might however need to quantify the sound insulation for partially open windows as well. This question is especially important in Denmark, where regulation since 2007 has lined up guidelines for partially open windows, as well if the outdoor noise level exceeds a cer- tain threshold [1, 2, 3].

1 lss@forcetechnology.com 2 bira@build.aau.dk 3 rue @forcetechnology.com 4 rsh s@forcetechnology.com

worm 2022

In these guides a requirement is set that in special situations, where the outdoor noise level from e.g. road traffic or railroads is high, the indoor noise level in e.g. apartments and offices must be be- low a certain level with open windows, when the opening area is 0.35 m 2 for each openable win- dow.

The method to verify compliance with limits consists of a combination of calculation of the fa- çade noise level together with documentation of the sound reduction for the chosen window solu- tion.

The guidelines has naturally led to innovation in this area [4, 5, 6, 7, 8, 9, 10, 11], and the quanti- fication methods (ISO 16283-3 [12] and ISO 10140-2 [13]) have previously been tested for some of the more complex solutions [14], which can be described as a sort of airlock/channel. However, at the moment there is much focus on more simple solutions, and in parallel to the ongoing revision of ISO 16283-3 [15] it has been chosen to study the applicability of this standard for simple partially open windows. In addition, it should be mentioned that open windows and noise also are of strong interest outside Denmark, for example in Hong Kong and Shanghai [16], United Kingdom [17] and Hamburg [18], and where more examples can be found in [4, 5, 9]. 2. ISO 16283-3 ACOUSTICS – FIELD MEASUREMENT OF SOUND INSULATION IN BUILDINGS AND OF BUILDING ELEMENTS – PART 3: FAÇADE SOUND INSULATION

As described in the title the purpose of the standard ISO 16283-3 is to describe field measurements of sound insulation of a façade or elements of the façade. This is further detailed in the scope: “The element methods aim to estimate the sound reduction index of a façade element, for example a win- dow. The most accurate element method uses a loudspeaker as an artificial sound source. Other less accurate element methods use available traffic noise”. However, although not described, the stand- ard is probably only/primarily meant for closed windows, and not for open windows, and for sev- eral reasons one might need to document the sound insulation for open or partially open windows. One of these reasons could be do due to the Danish regulation with partially open windows (open- ing area of 0.35 m 2 ).

Figure 1: Sketch of loudspeaker positions, where P 1 , P 2 and P n mark the loudspeaker positions. The figure is suggested to replace Figure 3 in ISO 16283-3 [15].

As described in the scope of ISO 16283-3 the preferred method is to use a loudspeaker, whereas traffic noise is considered less accurate.

However, for a typical Danish context with a partial open window, the sole reason for having this rule is because the outdoor noise level is high, often from traffic noise. Intuitively, it is the win- dows ability to dampen the noise from this exact noise source, which is the most important thing to determine, and as such the outdoor noise source (in this paper traffic noise) which should be the preferred noise source.

Since the geometry, here defined as the position of the noise source versus the opening direction of the partially open window, most likely is of importance, it is probably not unimportant, where the loudspeaker is positioned, if the loudspeaker method is chosen instead of the outdoor noise source.

The comparison of methods and noise sources for a partially open window is therefore carried out in a case study.

Parallelly and as the result of systematic review, a revision of ISO 16283-3 is recently initiated with a number of good inputs. One of the suggested changes is to replace the current figure 3 in ISO 16283-3 with the above shown Figure 1, since it is believed that the current figure in the standard is often misunderstood. We hope that the results of the case study can be useful input to the discus- sions as part of the revision of ISO 16283-3. 3. CASE STUDY

To compare methods and noise sources for a partially open window a suitable test site is needed ful- filling the following:  Periods with sufficient traffic noise to do the traffic noise part of ISO 16283-3. • Periods with no traffic noise to do the loudspeaker part of ISO 16283-3. • A room with a window which can be opened and where the general setup complies with ISO 16283-3. Preferable the window is an outward opening window and preferably an outward side hung window. • Access to the room, the façade, and to open the window (also in the late evening). • A site where no neighbours can be disturbed when using the loudspeaker as a noise source, which will probably be late in the evening to avoid traffic noise.

A meeting room in the main office of FORCE Technology meets most of the requirements. The only requirement not met is to have a side-hung window. The reason that a side hung windows is desired is due to geometry, hence it is easier to position the loudspeaker both shielded from the opened window and unshielded (seen from inside). The solution was to find a way to elevate the loudspeaker to a height of 10 m (in the top position).

Figure 2: Left: Photo from outside showing the partial open window, the microphones on the win- dow and the dodecahedron loudspeaker in top position. Right: Photo from inside the room showing the partial open window, some of the outdoor microphones and the road in the background.

3.1. Setup of room, window and road For the selected room one window, an outward top steered window with the dimensions 1.31 m x 1.21 m, can be opened, see Figure 2 . The middle of the window is approx. 5.4 m above ground. The room is a rectangular meeting room (V = 106 m 3 ) equipped with a large table and several chairs. The window was opened to an opening area of 0.35 m 2 following the Danish regulation.

The ground outside the building is a paved parking lot. The road is parallel to the building, and the middle of the road is located 23 m away from the building. The road has a total width of 11 m. The measurements with traffic noise were performed in the afternoon during rush hour. The meas- urements with the loudspeaker were performed in the evening when there were almost no vehicles. At the time for measuring traffic noise, it took approx. 4 minutes to have 50 motorized vehicles passing, corresponding to 750 vehicles an hour. All measurements were completed during the same day.

N

Figure 3: Aerial photo of the site. The full yellow line shows the path for the moving loudspeaker on the bike path, R’ ml1 , and the dashed yellow line shows the path for the moving loudspeaker on the parking lot, R’ ml2 . The yellow star shows the approx. position of the window.

3.2. Setup of indoor microphones For the indoor setup 5 fixed microphones were distributed over the room positioned with distances to room boundaries complying with ISO 16283-3. The reverberation time (T 20 ) was measured in these positions + one extra position with one corner loudspeaker position used for all microphone positions.

3.3. Setup of outdoor microphones For the outdoor setup 3 microphones were distributed on the actual window and the panels/win- dows next to/below. The microphones were mounted with windshields and secured with duct tape. For the traffic noise measurements 3 microphone positions in total were used. For the loudspeaker measurements 3x3 microphone positions were used i.e., the measurement was repeated 3 times for each setup where the position of the outdoor microphones was changed.

Additionally, 1 microphone was positioned at a distance of 2 m from the window, which can be seen in Figure 2 and Figure 4. This microphone can be used to analyse the measurements according to the global method, to be able to compare the results of the open windows with the element and the global methods. This, however, is not the purpose of this paper and will therefore be investi- gated subsequently.

3.4. Setup of stationary loudspeaker The loudspeaker was a dodecahedron loudspeaker positioned in a horizontal distance of 5 m from the building, for 6 different positions in a circle with a radius of 5 m, see Figure 1 and Figure 4.

It was desired to have loudspeaker positions both shielded from the openable window (position 37°, 90° and 127°) and unshielded from the openable part of the window (position 233°, 270° and 323°). Since it was a top steered window, it meant to have both loudspeaker positions in the highest and lowest part of the circle. Pink noise was used.

90 o

127 o

37 o

323 o

233 o

270 o

Figure 4: Photo/sketch of the loudspeaker setup. The blue oval outlines the circle of the 6 loud- speaker positions, see Figure 1, where the loudspeaker at this photo is at the topmost position (90 o ). The orange arrows outline the approximate lines between the 6 loudspeaker positions and the mid- dle of the window.

3.5. Setup of movable loudspeaker It was discussed to use a moving loudspeaker playing pink noise, and this was done as well. To have a battery-operated solution, a loudspeaker of the brand ‘Soundboks’ was used, which was po- sitioned on a sack truck and placed so it pointed towards the window, see Figure 5. It was moved at fast walking speed back and forth in parallel to the building and the road. When direction was changed, the loudspeaker was turned so it always pointed towards the window. Two ‘paths’ were used, one at the empty parking lot at a distance of approx. 10 m to the building over a path of ap- prox. 35 m (2 times both directions) and one at the bike path at the other side of the road, meaning at a distance of approx. 35 m to the building over a path of approx. 80 m (once in each direction). The measurement was paused, when motorized vehicles were passing. The descriptor has here been designated R’ ml (ml: Moving Loudspeaker) and calculated the same way as with road traffic noise (section 3.13 in ISO 16283-3).

Figure 5: Merged photo showing 3 different positions when walking with the movable loudspeaker on the parking lot, R’ ml2 . 4. MEASUREMENT RESULTS

The signals from the microphones were recorded synchronously on a computer using the data ac- quisition software noiseLAB Capture 4.0. For the signals with the dodecahedron loudspeaker occa- sionally a motorized vehicle passed and depending on the number of passing vehicles between 2-4 minutes were recorded for each setting. When analysed in noiseLAB Batch 4.1 a train analysis was performed with a train length of 16 seconds. The 16 seconds with the lowest level for all 8 micro- phones for each setup were used for the further analysis. The 16 seconds are the minimum measure- ment length described in ISO 16283-3 for the relevant frequency range. The train analysis was per- formed to find the period least affected by traffic noise.

For the measurements with traffic noise the full recording (with at least 50 passing motorized ve- hicles) was used for further analysis (5 recordings in total).

For the measurements with the moving loudspeakers, the measurement was paused, when a mo- torized vehicle passed.

All the measurements were corrected for background noise. With traffic noise the signal-to-noise ratio was less than 6 dB for 50 Hz and for frequencies higher than 2.5 kHz. For the 1/3 octave band frequencies with a signal-to-noise ratio less than 6 dB, the values were corrected with 1.3 dB.

The data from the outdoor and indoor microphones was combined as described in ISO 16283-3, and the sound reduction index is shown in Figure 6 and Figure 7. As ‘S’, the area of the test speci- men, the area of the window was used. The relevant single number quantities were calculated ac- cording to ISO 717-1 [19] and are shown in Table 1.

4.1. Comparison of measurement results for the three different noise sources Figure 6 shows the sound reduction index for the measured data, which for this figure is shown pri- marily as areas to make it easy to distinguish the results between the three different noise sources. Each area is coloured between the minimum and maximum sound reduction index of each group of data, and additionally the arithmetic mean has been calculated, which is shown with a full line.

Figure 6: The sound reduction index in the frequency area 50-5000 Hz is shown as areas for each of the three types of noise source, blue shows measurement with the dodecahedron positioned in 6 po- sitions according to ISO 16283-3, yellow shows measurement with traffic noise (measured for 5 pe- riods) and green shows measurement with the moving loudspeaker. For each of the three types of noise source the arithmetic mean is shown as well with a full line.

The figure shows that measurements with the dodecahedron loudspeaker, R’ 45 o , for nearly all 1/3 octave bands gives the highest sound insulation. When comparing the results with traffic noise and the moving loudspeaker, it can be seen that the moving loudspeaker clearly gives the highest sound insulation of the two for 1/3 octave band fre- quencies above 400 Hz, where the sound insulation for 1/3 octave band frequencies below 400 Hz is approximately comparable for the two types of noise source.

‘Sound reduction index, R' [4B] 15 10 ° 63 125 250 500 1000 2000 4000 41 octave band center frequency [Hz]

4.2. Comparison of measurement results with the dodecahedron loudspeaker Figure 7 shows the sound reduction index for the measured data with the dodecahedron loudspeaker in the used 6 different positions, and additionally the sound reduction index of the arithmetic aver- age of the sound reduction measured with traffic noise has been added for comparison.

Figure 7: The sound reduction index in the frequency area 50-5000 Hz is shown for the 6 dodecahe- dron loudspeaker positions according to ISO 16283-3, together with the arithmetic mean of the measurements with traffic noise as noise source.

When comparing the curves, it can be seen there are 1/3 octave band frequency ranges, where the sound reduction index is approx. the same (50 Hz-160 Hz and 500 Hz-1000 Hz) and 1/3 octave band frequency ranges where there is a huge difference in sound reduction (160 Hz-500 Hz and 1000 Hz-5000 Hz). For the frequency range with the huge difference, the measurement with the loudspeaker in the top position (90°) clearly stands out as having the highest sound reduction, where the loudspeaker in the bottom position (270°) often has the lowest sound reduction. This is also as expected, since the top position is also the most shielded by the open window, and the bot- tom position is the least shielded.

° H 63 125 250 500 1000 2000 4000, 41 octave band center frequency [Hz] \, | [Ra 270" =e rly 223" —eo rly. Rl, 90°

4.3. Comparison of single number quantities for the measurement results In Table 1 the single number quantities (SNQ) of selected measurements are shown. Comparing the SNQ for the arithmetic average of the traffic noise with the different positions of the dodecahedron loudspeaker a difference in the range of approx. 2-6 dB is seen indicating that one should be very careful substituting the two methods.

When comparing the SNQ for the different positions of the dodecahedron loudspeaker differ- ences in the range 0-3 dB can be seen (4 dB in a single case), with largest difference between the ground positioned loudspeaker (270 o ) and the topmost positioned loudspeaker (90 o ). It is therefore clear that one should be very careful in choice of loudspeaker position especially with outward opening windows (which can somewhat shield the loudspeaker). When comparing the SNQ for traffic noise and for the moving loudspeaker, the difference is in the range 0-2 dB indicating that this method might be worth investigating further. Table 1: The calculated single quantities (SNQ) values related to Figure 6 and Figure 7. The values are calculated according to ISO 717-1. For traffic noise (R’ tr,s ) the SNQ including 5000 Hz should not be used, since the signal-to-noise ratio for traffic noise was lower than 6 dB for 1/3 octave band frequencies above 2.5 kHz.

270 o 233 o 323 o 37 o 127 o 90 o R’ tr,s R’ ml1 R’ ml2

[dB] [dB] [dB] [dB] [dB] [dB] [dB] [dB] [dB]

R’ w (1 dec) 8.9 11.1 10.0 9.6 10.6 11.9 6.4 8.8 7.7

R’ w 8 11 10 9 10 11 6 8 7

R’ w + C , 100-3150 Hz 8 10 9 9 10 11 6 8 7

R’ w + C tr, 100-3150 Hz 8 10 9 8 9 10 6 7 6

R’ w + C , 50-3150 Hz 8 10 9 9 10 11 6 8 7

R’ w + C tr, 50-3150 Hz 8 9 9 8 9 10 6 7 6

R’ w + C , 50-5000 Hz 8 10 9 9 10 12 3* 7 6

R’ w + C tr, 50-5000 Hz 8 9 9 8 9 10 5* 7 6

Assuming that ISO 12999-1 [20] is applicable also for partially open windows the expanded uncer- tainty for a 95 % two-sided confidence level with a type C situation (repeatability) can be estimated to be in the range of 0.8 – 2.0 dB, with 0.8 dB for R’ w and 1.4 dB for R’ w + C tr,100-3140 Hz.

5. DISCUSSION, CONCLUSION AND SUGGESTIONS

Typically, regulations for sound insulation of facades and windows are concerned only with closed windows, and accordingly the standards for quantifying the sound insulation are typically designed for closed windows. However, one might need to quantify the sound insulation for partially open windows as well. This question is especially important in Denmark, where regulation since 2007 has lined up guidelines for partially open windows if the outdoor noise level exceeds a certain threshold. The method to verify compliance with limits consists of a combination of calculation of the façade noise level together with documentation of the sound reduction for the chosen window solution.

It can however be discussed what the best method is to document the sound reduction of the cho- sen window solution. One possible method is the ISO 16283-3, which however most likely is aimed for closed windows, even though it is not explicitly described in the standard. In parallel to the initi- ated revision of ISO 16283-3, it has been chosen to study the applicability of this standard for sim- ple partially open windows by a case study.

In the case study, primarily the use of traffic noise and loudspeaker positions has been tested (all positions complying with ISO 16283-3) together with the use of a moving loudspeaker.

The results show large differences (up to 6 dB) for single number quantities (SNQ) when com- paring traffic noise and loudspeaker positions, and additionally large differences (up to 3 dB) are found when comparing loudspeaker positions, where the largest differences (and highest sound in- sulation) are observed with the loudspeaker position, which is most shielded by the partial open window (the top position). The ‘classification’ of a large difference should here be seen in the con- text of that it is exactly the same window tested in an unchanged open position for measurements on the same day (with very low wind speed).

Great care is therefore recommended with the use of loudspeaker as noise source for measure- ments of sound insulation of partially open windows, and if the purpose of the measurement is to document the sound insulation ability towards a specific noise source (e.g. traffic noise) that this noise source is used as noise source for the measurement. Suggestions for future work:  It would be beneficial with more similar studies comparing the methods for partially opened windows to have a larger base of data.  In the case study a moving loudspeaker method was supplementary investigated which showed promising results. It is suggested to test the method for more situations.  For this study, only the standard for field measurements of sound insulation of simple partial open windows has been investigated. Previously the lab and field standards has been compared for a complex construction [14]. The applicability of ISO 10140 for simple open windows should be investigated in the laboratory as well.  Use the findings in this paper as recommendations for the revision of ISO 16283-3.  Compare measurement results from the element method with the global method. 6. ACKNOWLEDGEMENTS

FORCE Technology wishes to thank the Danish Agency for Institutions and Educational Grants for financial support of the work. The authors would like to thank FORCE Technology for giving ac- cess to rooms, to thank Jens Oddershede for helping with the measurements, to thank Inge Lis Kjær for proof reading and to thank Martin Leerbæk for inspiration to the moving loudspeaker method.

7. REFERENCES

1. Danish Ministry of the Environment, 2007, Vejledning 4/2007, Støj fra veje (Guidance 4/2007

Noise from roads). 2. Danish Ministry of the Environment, 2007, Tillæg til vejledning nr. 5/1984: Ekstern støj fra

virksomheder (Addition to guidance 5/1984: Environmental noise from plants). 3. Danish Ministry of the Environment, 2007, Tillæg til vejledning nr. 1/1997: Støj og vibrationer

fra jernbaner (Addition to guidance 1/1997: Noise and vibration from railroads). 4. Søndergaard LS, Olesen HS. Lydmæssig optimering af "Russervinduer" - Miljøprojekt nr. 1417,

2012 (Acoustical optimization of Supply Air Windows - Environmental project no. 1417, 2012). DELTA/Danish Ministry of the Environment, 2012. 5. Søndergaard L. S, Egedal R, Hansen MB. Open windows with good sound insulation – Envi-

ronmental project no. 2047, 2019). DELTA/Danish Ministry of the Environment, 2019. 6. Søndergaard L. S, Olesen HS. Investigation of sound insulation for a Supply Air Window. Proc.

43rd Forum Acusticum; 27 June - 1 July 2011; Aalborg, Denmark. 7. Søndergaard L. S, Legarth SV. Investigation of sound insulation for a Supply Air Window –

field measurements and occupant response. Proc. 43rd International Congress on Noise Control Engineering, Inter-noise; 16 - 19 November 2014; Melbourne, Australia. 8. Søndergaard L. S, Egedal R. Open windows with better sound insulation. Proc. 45nd Interna-

tional Congress on Noise Control Engineering, Inter-noise; 21 - 24 August 2016; Hamburg, Ger- many. 9. Rasmussen B, Erfaringer med lydisolerende åbne vinduer i trafikstøjbelastede boligområder

(Experiences with sound insulating open windows in traffic noise exposed residential areas), Danish Building Research Institute, (2015). www.sbi.dk/SBi2015:08. 10. Rasmussen B, Experiences with sound insulating open windows in traffic noise exposed

housing. Proc. 44th International Congress on Noise Control Engineering, Inter-noise; 9 - 12 August 2014; San Francisco, USA. 11. Hansen MB, Investigating the impact of noise incidence angle on the sound insulation of a

supply air window, Proc. 44th International Congress on Noise Control Engineering, Inter- noise; 9 - 12 August 2014; San Francisco, USA. 12. ISO 16283-3:2016, Acoustics - Field measurement of sound insulation in buildings and of

building elements -- Part 3: Façade sound insulation. 13. ISO 10140-2:2021, Acoustics — Laboratory measurement of sound insulation of building

elements — Part 2: Measurement of airborne sound insulation. 14. Søndergaard L. S, Egedal R, Bolberg M, Hansen MB. Sound insulation of open Supply Air

Windows, comparing laboratory and field measurements. 46nd International Congress on Noise Control Engineering, Inter-noise; 27 - 30 August 2017; Hong Kong. 15. Result of voting systematic review ISO 16283-3_2016, document number ISO/TC 43/SC

2/WG 18 N 824, dated 29. June 2021. 16. East and South East Asia Regional Chapter of the Acoustical Society of America, International

noise seminar – Innovative acoustic windows – the latest trendy noise mitigation designs, Hong Kong (23 Oct 2016) & Shanghai (25 Oct 2019). 17. Institute of Acoustics, ACOUSTICS VENTILATION AND OVERHEATING Residential De-

sign Guide, Version 1.1, January 2020. 18. HafenCity Hamburg, Schallschutz bei teilgeöffneten Fenstern (2011) 19. ISO 717-1:2020, Acoustics – Rating of sound insulation in buildings and of building elements –

Part 1: Airborne sound insulation. 20. ISO 12999-1:2020, Acoustics — Determination and application of measurement uncertainties in

building acoustics — Part 1: Sound insulation