Acoustics in wooden buildings – Measurements in the Laboratory and in Single Family Houses : AcuWood Report 1

(1)

Acoustics in wooden buildings –

Measurements in the Laboratory

and in Single Family Houses

Moritz Späh

Andreas Liebl

Philip Leistner

AcuWood Report 1

SP Report 2014:14

(2)

SP Technical Research Institute of Sweden

Box 857, 501 15 Borås, Sweden (headquarters)

SP Rapport 2014:14

ISBN 978-91-87461-64-4

ISSN 0284-5172

(3)

Fraunhofer‐Institut für Bauphysik IBP Forschung, Entwicklung, Demonstration und Beratung auf den Gebieten der Bauphysik Zulassung neuer Baustoffe, Bauteile und Bauarten Bauaufsichtlich anerkannte Stelle für Prüfung, Überwachung und Zertifizierung Institutsleitung Univ.‐Prof. Dr.‐Ing. Gerd Hauser Univ.‐Prof. Dr.‐Ing. Klaus Sedlbauer

Project report No. 1

Measurements in the Laboratory and

in Single Family Houses

WoodWisdom‐Net:

AcuWood – Acoustics in Wooden

Buildings

Development of advanced measurement and rating

procedures for sound insulation in wooden buildings as

basis for product optimisation

Research project 033R056

Term of project 01.10.2010 – 30.09.2013 Moritz Späh, Andreas Liebl, Philip Leistner Stuttgart, 27.06.2013 Project leader Editor Prof. Dr.‐Ing. P. Leistner Dr. M. Späh

(4)

1 Introduction

Wooden multi storey family houses are increasingly build in Europe. Driving forces are better sus‐ tainability, a development towards industrialisation of building elements and related to it, cost re‐ duction in the construction sector. In the past years, legislation has enabled wooden multi storey houses in many countries, including Germany. The main problems of fire protection issues have been solved. However, noise and vibration disturbances experienced by residents tend to increase, even if the building code requirements are fulfilled. Therefore, sound and vibration issues have become the new hindrance for multi storey wooden buildings. The current acoustic requirements in multi storey family houses are based on experience in heavy weight multi storey buildings, as wooden buildings have not been possible previously. The perceived acoustic quality in lightweight buildings is different, compared to heavyweight structures. In particu‐ lar, low frequency sound transmission of airborne and especially impact sound sources lead to com‐ plaints in wooden buildings, and might become very evident and disturbing in lightweight structures [1]. The currently used rating systems for airborne and impact sound transmission in buildings were de‐ veloped in the 1950’s and aimed to rate the building constructions of this time. In the 1990’s the in‐ troduction of spectrum adaption terms in ISO 717 [2, 3] changed the rating system and included (in parts) low frequencies down to 50 Hz. With the introduction of wooden multi‐storey houses with acoustic requirements on the separating elements (floors and walls), it was obvious that the current rating systems did not prevent increased annoyance of living noise, especially impact noise, in wood‐ en buildings. In this project, the aim was to find better technical descriptors of impact noise sources by correlation to subjective ratings of impact noise sources in Buildings. Besides wooden constructions, a concrete floor was also investigated to include the behaviour of common floor design in this study.

1.1 Aim of the project

As problems of noise and vibration disturbances in wooden buildings have been recognised, the aim of this project is to develop sound and impact noise criteria that better correspond to human percep‐ tion in heavy weight and lightweight buildings. The criteria should not only focus on wooden build‐ ings, but also include traditional heavy weight buildings, for example made of brick, concrete etc. The disagreement between the acoustic requirements in national standards and the subjective noise perception of the occupants is a general problem, which applies to wooden and lightweight buildings all over Europe [1, 4, 5].

(9)

Although it has been tried to solve the problems by adding spectrum adaption terms to the conven‐ tional single‐number quantities of the weighted sound reduction index Rw [2, 6] and the weighted impact sound pressure level Ln,w [3, 7], the problems are still not solved [8].The main problem in noise protection in wooden buildings are the impact sound insulation of wooden (lightweight) floors and – to a smaller degree – the airborne sound insulation of the exterior building elements like walls and roofs. Even though there are numerous investigations on propagation and human reception of im‐ pact and airborne sound in wooden buildings, a uniform and consistent approach for adapted rating criteria and requirements is not available yet [9–13].

1.2 Aim of the report

This report documents the conducted measurements in the laboratories of the IBP and in German single family houses in the field. It includes all important information on the constructions of the floors, the laboratories and the room situations in the buildings. It lists the basic measured values for documentation.

2 Literature review

2.1 Impact sound

The objective evaluation of airborne‐ and impact noise is based on measurements in buildings. The measurements are specified in national standards, which are based on ISO 140 [14]. The require‐ ments differ in the European countries in terms of different levels, but also in the descriptors used. A brief historic overview of the development of the criteria for airborne and impact noise is given in [4]. The current rating system in ISO 717 [6, 7] is based on developments in Germany. This is de‐ scribed for example in [15]. As the sound reduction is frequency depending, an arithmetic averaging of the frequency dependent values had previously been used, but it was found that the single aver‐ aged value was not correlating to the subjective impression. Therefore, a rating method was firstly suggested by Cremer [16], where the sound reduction curve measured was compared to a reference curve. This curve was intended to give the airborne sound reduction which was the general sound re‐ duction of standard building elements used at this time and which should be aspired by other ele‐ ments used. This reference curve was shifted so that the curve to be analysed only falls below the reference curve by a certain sum of deviation. A general check for the standardised building construc‐ tions then (homogeneous single leaf walls and floors with floor covering) showed, that this rating method was appropriate and in agreement to the subjective judgements. Some of the rules to gain a single number value by the shifting reference curve have been changed over the years [4], but the reference curves itself has not changed and is still used in ISO 717 for airborne and impact sound.

(10)

2.2 Subjective evaluation of impact noise

A recent thorough literature study on the annoyance in dwellings, the perception of impact noise caused by walking and an overview of listening test methods has been conducted as work package 1 within the AkuLite project [17]. Additionally, a procedure for a listening test within the AkuLite pro‐ ject has been proposed here. The following work is based upon this literature study. Nevertheless, the listening test performed did not follow the suggested method described in the report of Thorsson [17]. Instead of recording the vibration of the ceiling during measurement and playback by a loud‐ speaker hanging from the ceiling of the listening room, the recording was made by an artificial dum‐ my head with microphones in the ear channel. This leads to a binaural recorded signal, which was played back by calibrated headphones. This enables the localisation of the source (above the listen‐ er). As mentioned by Thorsson, this method of recording the signals includes the room acoustics of the receiving room. The influence of the localisation of the source (by binaural signals in the listening test) on the subjective judgement was tested in a first listening test. This test proved, that the influ‐ ence cannot be neglected and further listening tests were performed solely with binaural recorded signals. To reduce the spread of the room acoustics conditions in the recordings, all measurements in the la‐ boratory and almost all measurements in the field were conducted in rooms with quite similar sizes. Additionally, the receiving rooms in the laboratory where equipped with additional sound absorbers. This leads to reverberation times close to real situations of about 0.5 s. Some deviations were ac‐ cepted for the building measurements, when the rooms where empty (House E). In this case, addi‐ tional sound absorbers were installed in the receiving room, leading to some longer reverberation times of about 1 s for all frequencies above 50 Hz. All other rooms in the field measurements were normally equipped with furniture and had reverberation times of 0.5 s and slightly above this value at higher frequencies. The measured reverberation times are given in the annexes to this report and re‐ port No.2.Further information on the listening tests are described in AcuWood Report No. 3.

2.3 Objective evaluation of impact noise

The objective evaluation of airborne‐ and impact noise is based on measurements in laboratories and real buildings. The measurements are specified in national standards [6, 7], which are based on ISO 140. The requirements differ in the European countries in terms of different levels, but also in the de‐ scriptors used. An overview of the different descriptors used is given in.[4] . For a long time, the frequency range in 1/3 octave bands from 100 to 3150 Hz was used and became the “traditional” frequency range for requirements in Europe [4]. With the introduction of ISO 717 re‐ vision 1996 the spectrum adaption terms in airborne and impact sound insulation were introduced, extending the possible frequency range for sound insulation descriptors to lower frequencies down to 50 Hz and to higher frequencies up to 5000 Hz. Since 1998, the frequency range in the regulatory minimum requirements in Sweden was extended down to 50 Hz [4]. This is the result of experience in countries with a tradition of light weight building practice, which are mainly the Nordic countries Norway Sweden and Canada. In the criteria for higher sound quality classes, descriptors down to 50 Hz have been introduced in the countries Denmark, Sweden, Norway, Finland, Iceland and in Lith‐ uania in the last decade [4]. From the viewpoint of subjective evaluation, the low frequencies below

(11)

100 Hz play a significant role for walking noise. Studies have shown, that frequencies down to 16 Hz might be necessary to regard for a good correlation between subjective and objective evaluation of walking noise [18]. Unfortunately, especially the measurement of the reverberation time at low fre‐ quencies gets more difficult, the lower the frequency is. In this project the measurements were re‐ stricted to 20 Hz and above, as 20 Hz was the lowest third octave band measureable with the given equipment in practice.

3 Measurements

In the AcuWood project, measurements and recordings of the sounds were conducted, as single number values of measurements were to be correlated with subjective ratings from listening tests. In the receiving room, all signals were recorded, and third octave band measurement values were calcu‐ lated from the recordings. Therefore, measurements and recordings are termed “measurements” in the following.

3.1 Sources

All laboratory and field measurements were performed using the following standardized and non‐ standardized impact noise sources.

3.1.1 Tapping machine

The utilised tapping machines are standardized impact noise sources for building acoustics meas‐ urements according to DIN EN ISO 10140‐5 [19] Annex E. The used tapping machines are listed in section 3.6. In the laboratory measurements, the tapping machine Norsonic type 211 , Sr.‐No. 12958 was utilised, in the filed measurements the tapping machine Norsonic type 211 , Sr.‐No. 706.Both are comparable in the levels they generate, but slightly different in the rhythm they produce (this was the impression in the listening tests). According to the standards DIN EN ISO 10140‐4 [20] and DIN EN ISO 140‐4 [14], measurements were performed with four positions of the tapping machine on the floor, the measurements had a duration of 60 s. Exceptions of the number of positions were neces‐ sary in one of the field measurements (House B, described in detail in section 5.3). A photograph of the tapping machine is shown in figure 1.

(12)

Figure 1: Photograph of the utilised tapping machine.

3.1.2 Modified tapping machine

As modified tapping machine, the above mentioned machines were placed on elastic pads with 12.5 mm thickness and the hammers were falling onto an elastic interlayer of the same thickness. The material below the hammers was Getzner Sylomer (yellow), according to DIN EN ISO 10140‐5 [19] Annex F1, method b. Again, the same four positions were uses as for the tapping machine, and the measurement duration was again 60s. A photograph of the modified tapping machine is shown in figure 2. Figure 2: Photograph of the modified tapping machine.

(13)

3.1.3 Japanese rubber ball

The Japanese rubber ball is a standardized source, developed in Japan for impact noise generation and measurement. It is described in DIN EN ISO 10140‐5 [19] Annex F2. In the measurements, the Japanese rubber ball of the Fachhochschule Stuttgart – University of Applied Sciences was em‐ ployed. The rubber ball was dropped from a height of 1 m and caught after each drop. The height was set approximately by the operator. Tests showed that the repeatability of the ball drops was very high, giving a standard deviation of the ball drops at the same position in general below 1 dB. The measurements were performed on the same four positions (exception: house B, the same two posi‐ tions) as the tapping and modified tapping machine positions. In the laboratory, the ball drop was re‐ peated 10 times on each floor position, giving a total of 40 measurements, which were arithmetically averaged. The signals on the different microphone positions were energetically averaged. In the field measurements, the number of ball drops measured was reduced to 5 on each floor position, giving a total of 20 ball drops on the floor. Each ball drop was recorded within a time period between 3 and 10 s, and the L,F,max value was taken in third octave band as measured value, analysed with third octave

band filters by the acoustic software Artemis by Head Acoustics. A photograph of the Japanese rub‐ ber ball is shown in figure 3. Figure 3: Photograph of the Japanese rubber ball.

3.1.4 Real sources: walking persons

As real sources, walking persons were also measured in the laboratory and in the field. Here, different persons with different footwear were employed during the tests. The footwear was normal male shoes with rubber sole, male shoes with leather sole, semi‐high‐heeled shoes for the female walkers and socks for male and female walkers. In a study on the wooden beam floor with suspended ceiling, the differences of a greater number of walkers were studied. This is described in [21]. During the pro‐ ject, the number of different walkers was kept low, and usually medium walkers (in terms of levels produced) were mostly employed. In the field measurements, always the same male walker was us‐ ing the same shoes and was walking with socks; a female walker was not employed in the field. Gen‐

(14)

erally it was tried to engage the same walkers on all floor coverings. Unfortunately, this was not al‐ ways possible. Therefore, the different walkers are indicated by their first name. On each floor in the laboratory and in the field, the walking persons were walking in a circle across the four above mentioned excitation positions. The speed of walking was close to two steps per sec‐ ond, the measurement was done for a time of 60 s for each walking person. (In some of the field measurements, the background levels were relatively high. As the signals were recorded, times of high background noise in the recordings were not included in the generation of third octave band levels and also not included in the listening test signals. Therefore, in some cases the averaging was shorter than 60 s. A photograph of one walking person is shown in figure 4. Figure 4:Photograph of the walking person.

3.1.5 Real sources: drawing of chair across the floor

As another real source, a standard four leg chair was used. To generate normal chair moving sounds on the floor, it was drawn by a rope for a distance of about 1 m across the floor. The speed was about 20 cm/s, so the signals were about 5 seconds long. The signal was recorded for 10 s. The drawing of the chair was performed on the similar four positions as the operation of the tapping and modified tapping machine and the ball. In the laboratory, the drawing of the chair was repeated 10 times on each position, giving in total 40 signals. The signals were averaged arithmetically. The averaged sig‐ nals of the different microphone positions were energetically averaged. In the case of carpet as floor covering, the procedure of the measurements was the same. On carpet, the source acted differently, as the main excitation mechanism was the slip‐stick‐effect of the feet of the chair on the floor. On carpet, a stick‐slip‐effect did not occur, and the chair gave a very different excitation of the floor it‐ self. This should always be kept in mind when analysing the measurement results of the drawing of the chair. A photograph of the drawing of the chair is shown in figure 5.

(15)

Figure 5: Photograph of the drawing of the chair.

3.2 Sound pressure level

3.2.1 Tapping machine, modified tapping machine, walking persons, drawing of chair

The sound pressure levels in the receiving room of the different sources are calculated by energetic averaging of all microphone positions. The sound pressure level is calculated by:  





n Li

n

L

10 log

1

10

/10

(

1

)

with:

L = energetic averaged sound pressure level dB Li = sound pressure level of each microphone in the same room dB

3.2.2 Japanese rubber ball

As the Japanese rubber ball is an impulse sound source, the max values of the signals with time weighting fast ( = 125 ms) was used. The averaged sound pressure level of the ball is calculated by:









n L F F i

n

L

_,_max , ,max/10

10

1 log

10 (

2

)

with:

(16)

LF,max = energetic averaged maximum sound pressure level in dB

Li,F,max = sound pressure level of each microphone in the same room in dB

3.3 A‐weighted sound pressure level

To compare the different impact sound sources on the basis of a single number value, the A‐ weighted standardized sound pressure level Ln,T,A was calculated from the measurements.

3.3.1 Tapping machine, modified tapping machine, walking persons, drawing of chair

For all sources, the sound pressure level L in the receiving room (Equation 1) was standardized to a reverberation time of 0.5 s and A‐weighted, giving:















n L L A T n i A i T n

L

_, _,

₁₀

_log

₁₀

, , , /10

₍

3

)

with: Ln,T,A = the A‐weighted standardized sound pressure level in dB LA,i = the A‐weighting values for the third octave bands i in dB Ln,T,i = the standardized sound pressure level for the third octave bands i in dB, given by

















0 ,

10 log

T

L

_n_T

(

4

)

where: L = sound pressure level in the receiving room (Equation 1) in dB T = measured reverberation time in the receiving room in s T0 = reference reverberation time of 0.5 s

3.3.2 Japanese rubber ball

For the ball, the maximum sound pressure level L in the receiving room (Equation2) was standardized to a reverberation time of 0.5 s and A‐weighted, giving:

(17)















n L L A T n F i A i T n F

L

_,_max, _, _,

₁₀

_log

₁₀

,max,, , , /10

₍

5

)

with: LF,max,n,T,A = the A‐weighted standardized maximum sound pressure level in dB LA,i = the A‐weighting values for the third octave bands i in dB LF,max,n,T,i = the standardized maximum sound pressure level for the third octave bands i in dB, given by

















0 max , , max, ,

10 log

T

L

_F _n_T _F

(

6

)

where: LF,max = maximum sound pressure level in the receiving room (Equation2) in dB T = measured reverberation time in the receiving room in s T0 = reference reverberation time of 0.5 s

3.4 Airborne sound reduction

3.4.1 Measurements in the laboratory

All measurements in the laboratories were conducted on the basis of DIN EN ISO 10140‐4 [20]. All la‐ boratories were equipped with linings, reducing the flanking transmission to a great extent above 100 Hz. The weighted sound reduction index Rw , the weighted standardized sound pressure level dif‐ ference DnT,w and the spectrum adaption terms were calculated according to DIN EN ISO 717‐1:2006 [6]. Differing from DIN EN ISO 10140‐4, the receiving rooms in the laboratories were treated to have a reverberation time of close to 0.5 s. The reason was that simultaneously with the measurements, recordings for the subjective listening tests were performed. Therefore, similar reverberation condi‐ tions to normal in living rooms were realised. This was considered more important than a reverbera‐ tion time between 1 and 2 s. The measurements were performed with stationary microphones. The number of microphone positions in the sending and receiving rooms were 6, the number of loud‐ speaker positions in the sending room was 2. This leads to 12 independent measurements in sending and receiving room. The averaging time was 60 s. The reverberation time was measured by the method of stationary signal suddenly turned off. In the sending room, the measurement of the re‐ verberation time was performed at 6 independent microphone positions and one loudspeaker posi‐ tion. In the receiving room, the measurement was executed at 6 independent microphone positions and two different loudspeaker positions, giving a total of 12 independent measurements. The signal was pink noise. The sound reduction index was calculated by:

(18)



















A

S

L

R

₁ ₂

10 log

(

7

)

with:

R = sound reduction index in dB L1 = Sound pressure level in the sending room in dB L2 = Sound pressure level in the receiving room in dB S = Area of the separating element in m² A = equivalent sound absorption area in m² with:















T

V

A

0 .

16 (

8

)

where:

V

= volume of the receiving room in m³

T

= reverberation time of the receiving room in s

3.4.2 Measurements in the field

All measurements in the field were conducted on the basis of DIN EN ISO 140‐4 [14]. The weighted sound reduction index R´w , the weighted standardized sound pressure level difference D´nT,w and the

spectrum adaption terms were calculated according to DIN EN ISO 717‐1:2006 [6]. In all field meas‐ urements, flanking transmission was included. All the measurements were performed with stationary microphones. The signal was pink noise. Further details are given at the description of the specific measurements. The sound reduction index in the field was calculated by:



















A

S

L

R

´

₁ ₂

10 log

(

9

)

with:

R´ = sound reduction index in dB, including flanking transmission

(19)

L1 = Sound pressure level in the sending room in dB L2 = Sound pressure level in the receiving room in dB S = Area of the separating element in m² A = equivalent sound absorption area in m²

3.5 Impact sound pressure level of the tapping machine

3.5.1 Measurements in the laboratory

All measurements in the laboratories were conducted on the basis of DIN EN ISO 10140‐4 [20]. The weighted normalized impact sound pressure level Ln,w , the weighted standardized impact sound pressure level LnT,w and the spectrum adaption terms were calculated according to DIN EN ISO 717‐ 2:2006 [7]. Differing from DIN EN ISO 10140‐4, the receiving rooms in the laboratories were treated to have a reverberation time of close to 0.5 s. The measurements of the impact noise sources were performed with stationary microphones. The number of microphone positions in the sending room was 2, in the receiving rooms the number was 6. The number of tapping machine positions in the sending room was 4. This leads to 8 independent measurements in sending room and to 24 meas‐ urements in the receiving room. The averaging time was 60 s. The reverberation time was measured by the method of stationary signal suddenly turned off. In the sending room, the measurement of the reverberation time was performed at 6 independent microphone positions and one loudspeaker posi‐ tion. In the receiving room, the measurement was executed at 6 independent microphone positions and two different loudspeaker positions, giving a total of 12 independent measurements. The meas‐ urement signal was pink noise. The normalized impact sound pressure level was calculated by:

_















0 2

10 log

A

L

_n

(

10

)

with:

Ln = normalized impact sound pressure level in dB L2 = sound pressure level in the receiving room in dB A = equivalent sound absorption area in m² A0 = reference sound absorption area of 10 m² Additionally, the standardized impact sound pressure level was calculated by:

_















0 2 ,

10 log

T

L

_n_T

(

11

)

(20)

with:

Ln,T = standardized impact sound pressure level in dB L2 = sound pressure level in the receiving room in dB T = measured reverberation time in s T0 = reference reverberation time of 0.5 s A correction for the airborne sound transmission to the impact noise measurements was applied for L´n and L´nT. For the laboratory measurements, this correction was very small (≤ 0,1 dB). As the focus of the investigation was real living situations, the analysis of the signals within the AcuWood‐Project was based on standardized impact sound levels with reference to 0.5 s.

3.5.2 Measurements in the field

All measurements in the field were conducted on the basis of DIN EN ISO 140‐7 [22]. The weighted normalized impact sound pressure level L´n,w , the weighted standardized impact sound pressure lev‐ el L´nT,w and the spectrum adaption terms were calculated according to DIN EN ISO 717‐2:2006 [7]. In all field measurements, flanking transmission was included. All the measurements were performed with stationary microphones. Further details are given at the description of the specific measure‐ ments. . The normalized impact sound pressure level was calculated by:

_















0 2

10 log

´

A

L

_n

(

12

)

with:

L´n = normalized impact sound pressure level in dB, including flanking transmission L2 = sound pressure level in the receiving room in dB A = equivalent sound absorption area in m² A0 = reference sound absorption area of 10 m² The standardized impact sound pressure level was calculated by:

_















0 2 ,

10 log

´

T

L

_n_T

(

13

)

with:

(21)

Lń,T = standardized impact sound pressure level in dB, including flanking transmission L2 = sound pressure level in the receiving room in dB T = measured reverberation time in s T0 = reference reverberation time of 0.5 s A correction for the airborne sound transmission to the impact noise measurements was applied for Lń and LńT. This correction was small (≤ 0,2 dB) As the focus of the investigation were real living situations, the analysis of the signals within the AcuWood‐Project was based on standardized impact sound levels with reference to 0.5 s.

3.6 Equipment used

For the measurements of the sound reduction index and the reverberation time following equipment was used: ‐ Real Time Analyser Norsonic type 840 S.‐No.: 1607 (Laboratory measurements) ‐ Real Time Analyser Norsonic type 840 S.‐No.: 18736 (Field measurements) ‐ Power Amplifier Klein und Hummel, type AK 120 (Laboratory measurements) ‐ Power Amplifier Norsonic 235, S.‐No. 22595 (Field measurements) ‐ Dodecahedron loudspeaker Norsonic type 229, , S.‐No. 22568 ‐ Microphones B&K type 4165, S.‐No.: 674849 and S.‐Mo.: 1604478 (Laboratory measurements) ‐ Preamplifier Norsonic 1201, S.‐No. 22062 and S.‐No.22063 (Field measurements) ‐ Mikrophones B&K type 4165, S.‐No. 1158476 and S.‐No 1330519 (Field measurements) ‐ Calibrator Bruel & Kjaer 4230 S.‐No. 1472576 For the recording of the calibrated signals, the following equipment was used: ‐ Head Acoustics Frontend SQLab III, S.‐No.: 35020102 ‐ Dummy heads Head Acoustics type HDM I.Q. S.‐No.: 13001362 and 13001363 ‐ Microphones G.R.A.S. type 46 AE, S.‐No.: 88711, 88712, 88713, 88717, 88719, 88720, 88727, 88730

(22)

‐ Tapping machine Norsonic type 211 , Sr.‐No. 706 ‐ Tapping machine Norsonic type 211 , Sr.‐No. 12958

3.7 Listening tests and questionnaires

With the recorded signals of the dummy head in the receiving rooms, listening tests were performed. The listening tests are a main and crucial part of the of the AcuWood study. The listening tests per‐ formed are described in AcuWood‐report No. 3. Additional questionnaires were conducted within the project in Germany and Switzerland, also described in AcuWood report No. 3.

4 Laboratory measurements

4.1 Floor coverings of the laboratory measurements

As the measurements were planned to be as representative for real building situations as possible, different floor coverings were applied in the laboratory. At the bare wooden floor, floor coverings were not applied, as this is rarely found in buildings. Nevertheless, for the floors with floating floor (wooden beam floor with dry floating floor, the same floor with additional suspended ceiling and the concrete floor with concrete floating floor), measurements on the bare floating floors and with addi‐ tional different floor coverings were performed. Four different typical floor coverings were measured: laminate, parquet, tiles and carpet. The floor coverings laminate, parquet and tiles were combined with an intermediate foam layer between the cover material and the floor. The foam layers are often used to reduce the impact noise of the floor cover, but also to compensate unevenness of the floor surface and in the case of the tiles, to decouple the tiles from the floor. In many cases and for similar reasons, these interlayers are also used for the installation of laminate and parquet on concrete floors. The choice of the material was based on previous measurements of the reduction of the nor‐ malized impact sound pressure level on a bare homogeneous concrete floor of 140 mm thickness (P9 of IBP), according to DIN EN ISO 10140‐5 [19]. The floor coverings are listed in Table 1. Table 1: Floor coverings with interlayer and measured reduction of the normalized impact noise level.

Number Floor covering Interlayer Reduction of the normalized impact sound pressure level 1 Laminate, 7 mm (Meister Classic LC 100, Buche Stab 3) Ribbed foam interlayer (WPT SRL 160 + XPS‐foam ribbed) 20 [dB] (measure‐ ment from 21.05.2012)

(23)

2 Parquet, 13 mm (Meister Diele PD 400 cotta‐ ge, naturmatt lackiert) Foam interlayer (WPT SRL 140 s) 15 [dB] (measure‐ ment from 21.05.2012) 3 Standard tiles, 8 mm, size 30 x 30 cm, glued with 2 mm tile adhesive Decoupling layer (WPT E 210) 16 [dB] (measure‐ ment from 16.05.2012) 4 Standard carpet (Feinschlingenware mit Tex‐ tilrücken, 4 mm thickness, Polhöhe 2 mm, Polgewicht 360 g/m², OBI Rambo) None 23 [dB] (measure‐ ment from 16.05.2012) The above described floor coverings were used on all three different floors. Therefore, none of the floor coverings were glued to the floor, but laid out evenly. Besides the carpet floor cover, all other floor covers did not cover the whole floor area in the laboratories. Nevertheless, the area of the floors covered by the coverings was big enough to use different excitation positions for the impact sources. The influence by the additional floor covers on the airborne sound reduction was considered low, and therefore it was not investigated in detail. As the measurements were conducted in Laboratories with homogeneous heavy weight flanking walls and linings, a correction of the impact noise levels by airborne sound transmission was not necessary.

4.2 Laboratory with wooden beam floor

The floor is a wooden beam floor according to DIN EN ISO 10140‐5 appendix C, floor C1 [19]. In a first measurement series, the bare floor was measured. Then the floor was modified to represent normal floor conditions, by using a dry floating floor and additionally applying different standard floor co‐ vers. The floor and the configuration of sending and receiving room is described in section 5.2, the audio recording and measurement setup, the equipment used and the measurement objects are de‐ scribed in section 5.3. Section 5.4 deals with the listening tests, in section 5.5 results of the measure‐ ments and of the listening tests are presented. Photographs of the laboratory and basic data of the measurements are given in appendix A.

4.2.1 Description of the laboratory

The described measurements and recordings were conducted in the laboratory p8 of the IBP in Stuttgart. The laboratory is made to test wooden floor constructions. It consists of concrete walls and floors and offers a frame, where a lightweight floor can be installed. All walls are equipped with lightweight linings with resonance frequency of approximately 60 to 80 Hz, reducing the flanking transmission in the frequency bands for standard testing from 100 to 5000 Hz. A sectional drawing of

(24)

the laboratory is shown in figure 6. The room sizes are 4.78 m x 3.78 m x 3.82 m for the sending room and 4.78 m x 3.78 m x 2.67. Figure 6: Sectional view of the laboratory p8 of IBP. The wooden floor construction was installed on the console, separating the laboratory into two rooms.

4.2.2 Basic floor construction

The laboratory was equipped with a standardized floor according to DIN EN ISO 10140‐5 Appendix C, floor C1 [19], which is a lightweight wooden beam floor. This kind of floor represents approximately standard floors of (prefabricated) wooden single family houses in Germany, where no regulations on sound insulation and impact noise are given. The floor is shown in figure 7. Figure 7: Sectional view of the wooden beam floor according to DIN EN ISO 10140‐5. .(1: floor plate wooden chip board with 22±2 mm thickness, screwed into beams every 300 ±50 mm; 2: wooden beams with 120 mm width and 180 mm height; 3: mineral wool with 100 mm thickness and flow re‐ sistance between 5 and 10 kPa s/m² according to ISO 9053; 4: wooden battens with 24 mm width and 48 mm height and with 625 mm distance screwed into the beams; 5: gypsum cardboard with 12,5 mm thickness and density of 800 ±50 kg/m³, screwed directly into the battens every 300 ±50 mm) The weighted sound reduction index of the bare floor shown in figure 2 is Rw = 46 dB, the weighted normalized impact sound pressure level of the floor is Ln,w = 74 dB. The graph of the sound reduction index is shown in appendix A2.

(25)

In its initial state, the bare floor produced cracking noises when walkers were walking across it. These were mainly due to the weight of the walker, leading to vertical movement of the top plate edges. Therefore the top plate edges were connected to each other by screwed lashes, reducing the edge movement considerable and reducing the cracking noises to a minimum.

4.2.3 Modified floor construction with floating floor

For the measurements and recordings, the intention was to use a floor construction which is common in Germany. The above described bare floor according to DIN EN ISO 10140‐5 is nowadays rarely found in Germany, as the acoustic performance is too low. Very common is the use of a floating floor to improve the acoustic properties of floors in new single family houses as well as for refurbishment of old buildings. Therefore, a dry floating floor system was applied to the bare floor.. It consists of a 18 mm thick gypsum fibre board, laminated on 10 mm thick wood fibre (KNAUF BRIO 18WF). The wood fibre acts as a resilient layer between the bare floor and the gypsum fibre board. The floor con‐ struction is shown in figure 8. Figure 8: Sectional view of the wooden beam floor with floating floor .(1‐5 floor according to DIN EN ISO 10140‐5, with floating floor of 10 mm wood fibre and 18 mm gypsum fibre board KNAUF BRIO 18 WF)) The weighted sound reduction index of the floor with floating floor in figure 3 is Rw = 54 dB, the weighted normalized impact sound pressure level of the floor is Ln,w = 68 dB. The graph of the sound reduction index is shown in appendix A2, the normalized impact sound pressure level is shown in ap‐ pendix A2. With the dry floating floor, the cracking noises were again reduced, but not totally abandoned. The remaining cracking noises are caused by the construction of the floor and can be considered to be typical for this kind of floors.

4.2.4 Configuration of sending and receiving room

Recordings of the impact noise were performed in both sending and receiving room. Therefore both rooms were adjusted in their absorption to normal living conditions. The goal of the adjustments was to set the reverberation time near 0.5 s. As both rooms were equipped with linings, reverberation at frequencies between 50 and 100 Hz was already short. The sending room was additionally equipped with 8 absorbers of different types, of which one was installed on the ceiling of the room. Pictures of

(26)

the sending room are shown in Appendix A 1. The reverberation times of the sending room is given in appendix A2. Note: below 50 Hz the reverberation time was longer than 1 s and was not much changed by the absorbers. Similar to the sending room, the linings in the receiving room gave low reverberation time at fre‐ quencies between 50 and 100 Hz. The receiving room was additionally equipped with 5 Absorb‐ ers and some thin foam linings. Pictures of the receiving room with the absorbers are shown in . The basic data of the measurements of the wooden beam floor are given in Appendix A2: Basic data of the laboratory with wooden beam floor.

4.2.5 Measurements

The following recordings / measurements were performed: ‐ Measurement of the sound reduction index Rw for the bare floor and the floor with floating floor ‐ Measurement of the normalized impact sound pressure level Ln,w of the bare floor and the floor with floating floor ‐ Calibrated recording of the sound field in the sending and receiving room of different impact noise sources on the floor with floating floor and additionally with different floor coverings on top of the floating floor. The recordings can be used either to generate sound files for listen‐ ing tests and for generating “measurements” by analysing the sound files with the Head Acoustics software Artemis and its different tools.

4.2.6 Measurement setup

The measurement of the airborne sound insulation was performed according to DIN EN ISO 10140‐4 [20]. The number of loudspeaker positions was two, the sound pressure levels were measured by continuously moving microphones on two paths; the results were averaged. The reverberation time was measured according to DIN EN ISO 10140‐4 [20]. Two loudspeaker posi‐ tions and for each 6 microphone positions were used. The number of independent measurements was 12. For the impact noise sources, calibrated recordings were performed. The noise produced in the send‐ ing room was recorded with two microphones and a dummy head. The positions of the microphones in the sending room are shown in figure 9.

(27)

Figure 9: Floor plan of the sending room with positioning of the microphones and the dummy head. The noise in the receiving room was recorded by 6 microphones. Additionally, a dummy head was set up in the receiving room, recording the noise of the impact sources binaurally. At the time of the measurements it was not clear, if it was necessary to have both recordings of microphone and of the dummy head. Especially for the listening tests the recordings of the dummy head could have an in‐ fluence. The influence was determined by listening tests, described in section 4.3. The positioning of the microphones and the dummy head in the receiving room is shown on figure 10. Figure 10: Floor plan of the receiving room with positioning of the microphones and the dummy head

(28)

The different impact noise sources were placed at similar positions on the floor, to make the record‐ ings and measurements as comparable as possible. For the tapping machine, the modified tapping machine and the Japanese rubber ball, four excitation positions were defined. One of the excitation position was placed directly over a beam of the floor (position 3), one other was placed over a bay of the floor (position 4). The other two positions were partly on a beam and on the adjacent bay. For the tapping machine and the modified tapping machine at all four positions, recordings were made with a length of 60 s. The rubber ball was dropped at each position from a height of 1 m. This was repeated 10 times by a person at each excitation position. For each drop, a 10 s recording was made, including the signal of one ball drop. The chair was drawn over a path of about 1 m length across the same four excitation positions men‐ tioned before. The speed was about 20 cm/s, giving a signal of about 5 s length. As well as for the ball, the recording of the chair was repeated 10 times at each position. The original recordings of each signal were 10 s long. For analysis, the recordings were cut to include only the drawing noise of the chair. Other impacts like bringing the chair back to the starting point were excluded from the analysed file. The walking noise of the different walkers was recorded for 60 s. In this time, the walkers were walk‐ ing at a speed of approximately two steps per second (2Hz) on a circle. The circle position was so that the walkers would walk approximately across the excitation positions of the other sources. The circle was big enough for the walkers to walk in a normal manner and without stopping. The walking noise measurements were extracted by averaging the 60 s long walking signals of the walkers. The posi‐ tions of excitation on the floor in the sending room are shown in figure 11. Figure 11: Floor plan of the sending room with excitation positions of the impact sound sources.

(29)

4.3 Listening tests

With the recorded signals from the laboratory, listening tests were conducted to judge the annoy‐ ance and the loudness of the impact noise signals in the receiving room. Additionally it was tested if there is an influence of the different signals on the rating by the listeners. Therefore mono‐signals of one microphone (microphone 1) and the binaural signal of the dummy head were used.

4.3.1 Aim of the listening tests

The main focus of the listening tests was to get subjective ratings of the different impact noise sources on different floor coverings. Within the AcuWood project, the ratings were planned to be cor‐ related with objective single number ratings. Therefore, this first listening test aimed to prepare for further listening tests in terms of organisation, hardware used etc. and to evaluate the influence of the recordings by microphone and dummy head on the subjective rating. This was important to de‐ termine the appropriate further measurement procedure within the AcuWood project.

4.3.2 Procedure of the listening tests

The listening tests were performed with representative recordings cut to an appropriate length. For the comparison of microphone and dummy head, recordings of microphone 1 and the dummy head were chosen, see figure 10. The recordings of the tapping machine, the modified tapping machine and the walkers were cut to a length of 20 s. The chair signals were cut to a length of 7 s, the ball drop recordings were cut to a length of 1 s. All recordings were aurally checked to be free of background noise or other not relevant artefacts and that they were representative for the source. For the walk‐ ers, cracking noises of the floor in the recordings could not always be avoided. They were typical of such floor constructions and thus also part of the signals to be judged. The signals were adjusted to a calibrated level by playing the signals via headphones to a dummy head. For the listening test, a sample of 23 test persons (9 female and 14 male) was available. The age of the subjects was from 20 to 32 years with a median of 24 years. As material, dummy head recordings from 6 ceiling construc‐ tions (bare floor, floor with floating floor and additionally with 4 different floor covers) and 7 different impact noise sources gave 42 signals to be rated. Additionally the microphone recordings of the 7 sources were tested against the dummy head recordings. The sound files were played randomly to the listeners over headphones. The answers were given by indication on a computer screen. Judgements were asked for the individual noise sensitivity, the an‐ noyance and the loudness of the recorded signals. The scales were for the individual noise sensitivity a 11‐point rating scale (from “not at all” to “extremely”), the annoyance of the signals on a 11‐point rating scale according to ISO/TS 15666 (from “0” to “10”) and the loudness on a 51‐point rating scale according to ISO 16832 (from “0” to “50”).

(30)

4.4 Results of measurements in the laboratory with wooden beam floor

4.4.1 Repeatability of the source excitation

The repeatability of the rubber ball was tested by comparing 10 single ball drops at the same posi‐ tion, comparing results of ball drops at four different floor positions and comparing the measurement of the signals at 6 microphone positions. The results were that:  The 10 ball drops at the same position gave quite low standard deviation, from 20 to 630 Hz about 1 dB, above the standard deviation increased to about 2 dB at 2000 Hz. Therefore, the repeatability of the source itself is comparable to other impact sources and relatively high  Adding the different microphone positions to the analysis shows, that at low frequencies between 25 and 315 Hz the standard deviation is mainly influenced by the microphone posi‐ tion, the values of the standard deviation reach up to 6,5 dB at 31.5 Hz.  Analysing the whole dataset of 10 drops per excitation position, 6 different microphone po‐ sitions and 4 excitation positions shows, that the excitation positions have an influence on the standard deviation at very low frequencies below 40 Hz, and also from 315 Hz on up‐ wards to 5000 Hz, where the standard deviation reaches values between 2.5 and 5 dB. The repeatability of the chair, drawn across the floor showed very similar results. The standard devia‐ tion of the source itself was between 0.5 and 1.8 dB. Again, at the low frequencies the different mi‐ crophone positions where the reason of a standard deviation reaching 5.5 dB and at high frequencies between 2000 and 5000 Hz, the excitation positions of the chair resulted in a standard deviation of up to 4 dB. The repeatability of the walkers was not tested. It was assumed, that the recording of walking for 60 s gives a good average of the walking noise of one person. The spread of different walkers was not tested in this measurement series. It was assumed, that the same person could be walking on all dif‐ ferent floors. Unfortunately, during the measurements it showed that it was not possible to rely on the same walker for all situations. (Walkers were not always available because of holidays, illness, termination of the work contract etc.). Therefore at the wooden beam floor with floating floor and suspended ceiling, section 4.5, a study of the spread of walking signals of different walkers was per‐ formed. This is reported by Spinner [21].

4.4.2 A‐weighted standardized sound pressure level

For the floor with dry floating floor, the summed level Ln,T,A,50‐2500 was between 71.8 dB(A) for the

tapping machine and 22.5 dB(A) for the male walker with socks (modified tapping machine 47.1 dB(A), chair 62.8 dB(A), Ball LF,max,n,T,A,50‐2500 = 62.1 dB(A), female walker with hard footwear Ln,T,A,50‐ 2500 = 36.4 dB(A), male walker with hard footwear Ln,T,A,50‐2500 = 29.9 dB(A)). The spectra showed that

the tapping machine produced more high frequency excitation then all other sources, and the rubber ball had a max spectrum quite similar to the spectrum of real walkers, but about 30 dB higher.

(31)

4.4.3 Weighted normalized impact sound pressure levels of the different floors

The weighted normalized impact sound pressure levels of the different floors, measured with the tapping machine are given in Table 2.

Table 2: Weighted normalized impact sound pressure level Ln,w an spectrum adaption term CI,50‐2500 of

the different floors.

Floor Ln,w CI,50‐2500 Ln,w + CI, 50‐2500

Bare floor 73.1 1.6 74.7 Floor with dry floating floor 67.2 0.4 67.6 Floor with dry floating floor and laminate 64.4 2.6 67.0 Floor with dry floating floor and parquet 64.9 2.0 66.9 Floor with dry floating floor and tiles 62.6 2.1 64.7 Floor with dry floating floor and carpet 60.1 4.9 65.0 The results show, that for the wooden floor with floating floor, which represents a typical construc‐ tion in wooden single family houses, the different floor coverings give not much difference in the weighted normalized impact sound pressure level. The carpet reaches the lowest value of the nor‐ malized impact sound pressure level, but considering the spectrum adaption term CI,50‐2500, it is com‐

parable to tiles and 2 dB lower then parquet and laminate. The high reduction of the normalized im‐ pact sound pressure level for laminate cannot be found when the floor is installed on the wooden floor (with floating floor), and the ranking of the floor coverings in the real situation is very different then given by the reduction of the normalized impact sound pressure level according to the standard DIN EN ISO 10140. Tests of the dependency of the max impact sound produced by the rubber ball falling from different heights showed that the levels and spectra were quite similar for the height of 1 m and of 0.8 m. This again shows that the spectra are quite independent of little changes (of a few cm) of the falling height. Therefore, the applied procedure to let a person drop the ball can be regarded with a high re‐ producibility, as already shown by the low standard deviation. This is the case for the floating floor, but also for the case of floating floor with carpet.

(32)

4.4.4 Listening tests

The detailed results of the listening test are available in the file {AcuWood 120308 AnL.pptx}. The main results are:  The annoyance and the loudness are rated very similar for all sources and all floors  The tapping machine gives the highest annoyance and loudness rating, the male walker with socks the lowest.  The modified tapping machine gives similar ratings then the male walker with hard foot‐ wear.  The ratings of the dummy head recording were similar to the ratings of the microphone re‐ cordings for all sources with fixed excitation position.  All moving sources (walkers) were rated more annoying and with higher loudness when recorded with the dummy head. Here the possibility to localise the source with the dummy head recordings seems to lead to a greater annoyance and higher loudness judgement. Therefore, recordings in further objects will be performed with both the dummy head and microphones.  The wooden beam floor was rated most annoying, the floor with floating floor and carpet least annoying.  The three other floor coverings of laminate, parquet and tiles were judged equal annoying and with equal loudness, the floor with floating floor and no floor covering was rated slight‐ ly more annoying and slightly higher in loudness then the cases with floor coverings.

4.5 Laboratory with wooden beam floor and suspended ceiling

The floor is the same wooden beam floor of the laboratory p8 on the IBP, according to DIN EN ISO 10140‐5 Appendix C, floor C1, described in section 4.2.2. The floor was modified to represent normal floor conditions by using a dry floating floor, described in section 4.2.3 and additionally a suspended ceiling in the receiving room. Again, the same different standard floor covers were applied on this floor.

4.5.1 Description of the floor construction

The bare floor according to DIN EN ISO 10140‐5 is nowadays rarely found in Germany. Therefore, a dry floating floor and a suspended ceiling were applied to the bare floor. The suspended ceiling was

Acoustics in wooden buildings – Measurements in the Laboratory and in Single Family Houses : AcuWood Report 1