Psychoacoustic properties of tire-road noise and the relation to noise annoyance André Fiebig 1 Technische Universität Berlin Einsteinufer 25, 10587 Berlin, Germany Christoph Jakobs 2 Technische Universität Berlin Einsteinufer 25, 10587 Berlin, Germany

ABSTRACT Rolling noise as the interaction between tire and road contributes strongly to the total road noise in particular as engines become quieter due to more strict noise regulations and the emergence of electric motors. Thus, road traffic noise can be effectively reduced by optimizing vehicle tires and the pavement. Over the last decades, different pavements were developed to reduce tire/road noise as for example a porous asphalt surface or stone mastic asphalt reduce significantly rolling noise over a period of about 10 years. In the past the noise reduction due to different pavement types and layers were validly determined, several measurement methods established, and level reduction values incorporated in noise directives like the German guidelines for noise protection at roads. However, little attention was paid to assess the annoyance reduction effect of different pavements in detail. As the road tire noise is not only reduced in level but is also affected in spectrum, road traffic noise annoyance might not be accurately predicted by considering the level reduction only. 1. INTRODUCTION

To estimate the traffic noise in terms of sound pressure level at different receiver points (immission) resulting from road traffic (passenger cars, motorcycles, light and heavy trucks) multiple factors must be appropriately considered. Regarding the emission of vehicles suitable emission models must be applied to estimate reliably the noise contribution of a vehicle. In the current German guidelines for “Noise Protection at Roads RLS-19” [1] published in 2019 as part of the “Regulation for Traffic Noise Protection” (16. BImSchV [2]) among others, corrections for the road surface 𝐷 𝑆𝐷,𝑆𝐷𝑇 (𝑣) in dependence of vehicle type and speed are defined. The corrections for the road surface are based on the “Statistical Pass-by Method” (SPB) and the “Close Proximity Method” (CPX) [3]. The corrections of diverse road surfaces are related to a reference road surface ( 0 𝑑𝐵 ) and range from −5.5 𝑑𝐵 (open porous asphalt) to +7.0 𝑑𝐵 (cobblestone). Several investigations analyzed the acoustic effects of different types of road surfaces in detail with special emphasis on low noise pavements (see exemplarily [4]). In this context, most of the investigations mainly focused on level changes due to the road surface and its relation to annoyance [5]. However, it is assumed that the change of annoyance resulting from different pavements cannot

1 andre.fiebig@tu-berlin.de 2 christoph.jakobs@campus.tu-berlin.de

be solely described in terms of the A-weighted sound pressure level. It seems that the level reduction alone does not represent completely the change of noise annoyance due to the psychoacoustic properties. In general, it was shown decades ago that psychoacoustic parameters can improve the estimation of noise annoyance, for example based on the parameter loudness [6]. The benefit of psychoacoustics to describe noise annoyance reactions towards road traffic noise in particular, was shown in different studies already, for example quantifying the annoyance increase close to intersections [7] or related to the specific annoyance reactions towards powered two-wheelers [8]. In the context of pavements and their impact on the resulting tire-road noise, similar perceptual effects are expected and were demonstrated in a few studies. For example, Wühler and Guski demonstrated that the annoyance reduction due to low noise pavements cannot be sufficiently explained based on 𝐿 𝐴𝑒𝑞 [9]. Fastl and Zwicker illustrated the benefit of the parameter loudness to describe the acoustic effects of low noise pavements with respect to noise annoyance [10]. Later further psychoacoustic parameters were discussed to account for the changed noise character at low noise pavements [11]. However, it appears that the perceptual effects of different road surface types changing the road-tire noise were not systematically investigated and comprehensively understood. Thus, the following listening experiment investigates the perceptual effect of different road surfaces based on short-term measurements which were judged with respect to different evaluation criteria in an online experiment. The results of the online experiment guide the design and execution of a subsequent listening experiment in laboratory under controlled conditions.

2. MEASUREMENTS OF PASS-BY NOISES

Measurements were performed with a calibrated handheld sound level meter (Norsonic 140) and a binaural headset (SQuadriga I) in a distance to center of the road of 7.5 m without any reflecting surfaces close to the measurement point. The measurement height of the level meter was 1.2 m, the height of the binaural system around 1.70 m. All sites considered built-up areas in Berlin and are roads with relatively new pavements (see table 1) in order to investigate potential perceptual effect in case of pavements with large acoustic effects. It needs mentioned that for specification purposes with respect to [1], the pavements need to be of older age. Different kinds of stone mastic asphalts (SMA) were considered in the survey. Due to an optimized grain composition and binder usage of the asphalts, low-noise stone mastic asphalts achieve a void content that is up to 15 % higher than that of the standard variant. LO stands for noise optimized and LA for low noise (see table 1). SMA is the most frequently used surface layer in Berlin according to the senate of Berlin. Moreover, pass-by noises on a cobblestone pavement were measured for the purpose of comparison. Table 1 illustrates the selected roads in Berlin with the respective properties. The measurements were performed following the Statistical Pass-by Method (SPB) during times of low traffic and background noise. More than 120 single pass-by events per site were measured and analyzed to determine the respective correction value 𝐷 𝑆𝐷,𝑆𝐷𝑇 (50𝑘𝑚/ℎ) following the German specification defined in TP KoSD-19 [8]. The cobblestone road was chosen because of several noise complaints noted by the Berlin authority in this area. Moreover, the cobblestone road shows a relatively high daily traffic volume (see table 1). The correction values 𝐷 𝑆𝐷,𝑆𝐷𝑇 (50𝑘𝑚/ℎ) including the resulting uncertainties related to the reference road surface defined in [1] based on the measured data are shown in table 2.

Table 1: Selected measurement sites

Site (street in Berlin) Road surface Year of installation

DTV

Karl-Marx-Allee MA 8 S 2019 31.700

Veitstraße SMA 5 S LO 2019 7.100

Lichtenberger Straße SMA 8 LA 2020 23.400

Wiltbergstraße SMA 8 S 2020 8.400

Hauptstraße cobblestone no data 7.100

Table 2: Determined correction values 𝐷 𝑆𝐷,𝑆𝐷𝑇 (50𝑘𝑚/ℎ) and uncertainty values using the Statistical Pass-by Method and the German specification defined in TP KoSD-19 [12]

Site (street in Berlin) Road surface Level difference Uncertainty

Karl-Marx-Allee MA 8 S -3.7 dB(A) 0.51

Veitstraße SMA 5 S LO -3.0 dB(A) 0.27

Lichtenberger Straße SMA 8 LA -3.8 dB(A) 0.32

Wiltbergstraße SMA 8 S -0.2 dB(A) 0.32

Hauptstraße cobblestone +6.2 dB(A) 0.24

3. LISTENING EXPERIMENTS

In order to determine the perceptual differences of road traffic noise resulting from different road surfaces an online listening experiment was designed and performed. This experiment was carried out to investigate systematic differences between varying tire-road noises and to test the general feasibility for a subsequent laboratory experiment performed under controlled conditions. The online listening experiment was realized with the portal SoSci Survey (https://www.soscisurvey.de), which is free of charge for scientific, non-commercial surveys. This study was considered to prove the general feasibility of such a experimental approach and to study the potential relevance of a perceptual related correction value beyond the correction value 𝐷 𝑆𝐷,𝑆𝐷𝑇 (50𝑘𝑚/ℎ) following the German specification defined in TP KoSD-19 [12]. A detailed determination of a psychoacoustic correction value as a dB-equivalent needs to be based on a laboratory experiment performed under controlled conditions as described in section 3.3.

3.1. Method Subjects 29 participants (24 male, 4 female, 1 non-binary) took part in the online experiment. 11 participants belonged to the age group ranging from 20 to 29 years and 11 persons to the 30 to 39 years group. The rest of participants indicated an age equal to 40 years or older. Stimuli In total, 18 sounds with a duration of 12 s measured on 5 different road surfaces were presented. Each sound contained three single pass-by events. The level categories ranged from 60 dB(A) to 71 dB(A) ( 𝐿 𝐴𝑒𝑞 ) related to the single pass-by events. For the listening experiment, three different single pass-

by events from the same level category were selected and combined resulting in a natural sounding road traffic noise sequence. The assigned level of a single pass-by was determined by means of the monaural microphone (Norsonic 140). For the listening experiment the respective corresponding time interval of the binaural recording was used. Thus, small level variations can be observed between the single pass-by events (see figure 1). Although the absolute level is not relevant, the level differences between the measurements remain constant independent from the playback level. The following surfaces were considered SMA 8 S (Berlin, Wiltbergstraße), MA 8 S (Berlin, Karl Marx Allee), SMA 8 LA (Berlin, Lichtenberger Straße), SMA 5 S LO (Berlin, Veitstraße), and cobblestone (Berlin, Hauptstraße). Moreover, additional measurements on Lichtenberger Straße (SMA 8 LA) were performed for studying the perceptual effect of a wet surface. Figure 1 shows 5 road traffic noise stimuli with three single pass-by events analyzed with respect to A-weighted sound pressure level, loudness according to ISO 532-1 [13] and sharpness determined based on DIN 45692 [14]. Procedure All sounds were judged on 11-point rating scales with respect to annoyance , loudness and sharpness . The categories were labelled with numbers from 1 to 11 and in addition the uneven categories were verbally labelled with ’not at all’ ( gar nicht ), ’slightly’ ( wenig ), ’fairly’ ( mittelmäßig ), ’quite’ ( ziemlich ), and ’very’ ( sehr stark ), cf. [15]. Before requesting judgments, several sounds were randomly presented for illustrating road traffic noises used during the experiment. This included road traffic noise examples from all measured road surfaces. During the presentation of sounds the subjects should adjust the volume to a volume similar to typical environmental noises containing road traffic noises. After choosing an appropriate playback level the participants were requested to strictly keep the volume constant. In the second part of the experiment, a paired comparison test was carried out with a subset of sounds. The participants had to indicate the more annoying, louder and sharper sound out of two presented stimuli. The experiment lasted approximately 20 minutes.

Figure 1: Comparison of stimuli used in the listening experiment from the same level category with

respect to the A-weighted sound pressure level (top), loudness according to ISO 532-1 [13] (middle) and sharpness according to DIN 45692 [14] (bottom). From left to right the road traffic

noises related to the road surface: SMA 8 S, MA 8 S, SMA 8 LA, SMA 5 S LO, cobblestone

3.2. Results As expected, the noise sequences consisting of three single pass-by noises resulting from different road surfaces were differently judged in a statistically significant way with respect to all evaluation criteria, i.e., annoyance, loudness, sharpness. Table 3 presents the results from one-way ANOVAs for the level category of 66 dB(A). This means that beyond the sound pressure level other acoustic properties play a role with respect to the assessment of the road traffic noises. This is in line with the current knowledge about the effect of pavements on annoyance [16]. Table 3: Results of one-way ANOVA for the evaluation criteria ‘sharpness’, ‘loudness’ and ‘annoyance’ for the level category of 66 dB(A)

evaluation criteria F p-value

sharpness 3.311 0.0127*

loudness 10.13 <0.01*

annoyance 7.097 <0.01*

Figure 2: Annoyance (average values) of road traffic noises measured on different road surfaces all

with the same level category (level category of 66 dB(A)) The effect of a wet surface turned out to change the noise annoyance ratings statistically significant way as shown in figure 3 for the same level category. Due to the wet pavement the loudness and level respectively increases due to the increase of the tire-road noise contribution. At the same time the resulting road traffic noise contains much more high frequent components resulting in a sharper road traffic noise. In case of comparing pass-by noise events with the same sound pressure level, the changed spectral content (from 1.3 𝑎𝑐𝑢𝑚 to 1.6 𝑎𝑐𝑢𝑚 ) accounts for the higher annoyance which result in an annoyance penalty of approximately 4 𝑑𝐵 in addition to the level increase (of about 1dB). The results of the paired comparison test supported the observations based on the rating scales presented here, which speaks for the clearness of the observed annoyance differences due to the different pavement types.

Figure 3: Annoyance ratings of road traffic noises with the same 𝐿 𝐴𝑒𝑞 measured on a SMA 8 LA on

a dry and wet condition

3.3. Discussion

The data can be used to explore potential penalties or bonus values as dB-equivalents to account for the perceptual effects of different pavements beyond the sound pressure level. In the current German Guidelines for Noise Protection at Roads RLS 19 [1] different correction values 𝐷 𝑆𝐷,𝑆𝐷𝑇 (𝑣) are defined to address the effect of the road surface on the resulting traffic noise.

Due to the grain size and no further treatment with special binders the Wiltbergstraße has a very smooth surface. This construction method of the asphalt has a correction value of 0 dB according to [13] and represents the standard surface layer used in Berlin.

Figure 4: Annoyance ratings (arithmetic means and standard errors) over level categories in dB(A)

for different road surfaces

As also the determined correction value at the Wiltbergstraße based on acoustic measurements was close to zero (see table 2), the measured pass-by noises can be considered as a kind of reference and compared to the annoyance responses of other tire-road noise conditions. Figure 4 illustrates the shifts between the annoyance functions over level depending on the pavement type. The shifts of the annoyance functions due to the pavement type can be interpreted as the relevance of psychoacoustic properties beyond the level.

Table 4 shows the results of the determination of correction values due to psychoacoustic effects, which were computed on the basis of the comparison of the ratings of varying pavement types related

67.7 dB(A) to the “reference” (Wiltbergstraße). For the evaluation of the psychoacoustic effects the annoyance of the “reference” (Wiltbergstraße) at the level of 67,7 dB(A) was determined. This sound pressure level represents the sound pressure level according to [1] considering the corrections for measurement conditions for the measurement of Wiltbergstraße according to [8]. Therefore, the annoyance at the given level can be considered as the reference annoyance.

The determination of the correction values beyond 𝐷 𝑆𝐷,𝑆𝐷𝑇 (50𝑘𝑚/ℎ) was carried out by using the annoyance function over level based on a linear regression of the Wiltbergstraße data. Any rating differences between the road traffic noises were then converted to dB-equivalent values as shown in table 4.

Table 4: Determined psychoacoustic correction values (in addition to level) as a dB-equivalent for annoyance difference related to the reference value of 67,7 dB(A) related to SMA 8 S as the reference with a 𝐷 𝑆𝐷,𝑆𝐷𝑇 (50𝑘𝑚/ℎ) of −0.2 𝑑𝐵

Road surface psychoacoustic correction value as a dB-equivalent

(related to the reference value 67.7 dB(A))

MA 8 S -5.9 (bonus)

SMA 8 LA -3.9 (bonus)

SMA 5 S LO - 3.0 (bonus)

Cobblestone +2.6 (penalty)

Due to annoyance differences related to the reference value of 67.7 dB(A) as shown in fig. 2 psychoacoustic correction values in the range of few dB can be determined. This observation means that potentially low noise pavements are even more effective in terms of noise annoyance reduction than assumed based on level changes alone. Of course, as the acoustic effectiveness of low noise pavements decreases with time, the observed psychoacoustic correction values can be considered as a best-case estimation. Moreover, due to the general limitations of online experiments the observed results may only give some first indications about the general magnitude of potential psychoacoustic correction values. Thus, a detailed laboratory study is planned to investigate further the perceptual effect of low noise pavements under controlled conditions.

4. CONCLUSIONS

Different traffic road noises were subject to noise annoyance judgments in an online listening experiment. It turned out that the traffic road noise stimuli which were recorded at roads with different road surfaces were differently judged beyond the A-weighted sound pressure level 𝐿 𝐴𝑒𝑞 in an online listening experiment. This indicated the potential need of additional perception related correction values in addition to the level correction 𝐷 𝑆𝐷,𝑆𝐷𝑇 (𝑣) as defined in the German guidelines for “Noise Protection at Roads RLS-19” [1]. It appears likely that the spectral changes of the tire-road noise due to the respective (low noise) pavement type play an additional role with respect to noise annoyance. In case of the cobblestone road traffic noise, an increase in annoyance was observed which cannot be solely explained with an increase of loudness. In this case temporal noise patterns seem to be important beyond any spectral changes attracting additionally attention. Although the online experiment provides meaningful data as the level differences between all measurements remain constant, due the missing control of the experimental conditions including the control of the absolute playback level the results need to be interpreted with caution.

Thus, a detailed laboratory experiment was already performed as a follow-up to the presented study, and its results will be published soon. For it, additional data from the Bundesanstalt für Straßenwesen (BASt) is used to cover outside built-up areas with higher speeds including the noise from trucks. The benefit of a deeper understanding of the annoyance reduction effect of low noise pavements is to improve the effectiveness of chosen actions to manage environmental noise beyond level considerations. 5. ACKNOWLEDGEMENTS

We gratefully acknowledge the support from the Bundesanstalt für Straßenwesen (BASt) providing numerous pass-by noise recordings. 6. REFERENCES

1. RLS-19. Richtlinien für den Lärmschutz an Straßen (German Guidelines for Noise Protection at

Roads), Forschungsgesellschaft für Straßen- und Verkehrswesen , No. 052, ISBN: 978-3-86446- 256-6, 2019 2. 16. BImSchV. Sechzehnte Verordnung zur Durchführung des Bundes- Immissionsschutz-

gesetzes, 1990 3. Bartolomaeus, W. Guidelines for Noise Protection at Roads – RLS19, Lärmbekämpfung , 14 (6),

Springer Verlag, 203-207 4. Sandberg, U., Ejsmont, J.A. Tyre/road noise: reference book, Kisa, Informex, ISBN 9163126109

9789163126109, 2002 5. Gołebiewski, R., Makarewicz, R., Nowak, M., Preis, A. Traffic noise reduction due to the porous

road surface, Applied Acoustics , 64 (5), 481-494, 2003 6. Weber, R., Mellert, V. Vergleichende Beurteilung von Verkehrsgeräuschen. Korrelation mit

Lautstärkeparametern. In: Fortschritte der Akustik, DAGA 78, Proceedings, 1978 7. Fiebig, A., Papenfus, T. Untersuchungen zur akustischen Störwirkung von Kreuzungspunkten im

Rahmen der Überarbeitung der RLS-90, DAGA 2010, Proceedings, Berlin, 2010 8. Fiebig, A., Marla, P., Sottek, R. Noise of electric and combustion-powered scooters and resulting

annoyance, Euronoise 2012, Prague, Czech Republic, 2012 9. Wühler, K., Guski, R. Untersuchungen der subjektiven Wirkung von lärmmindernden Straßen-

decken. In: Zeitschrift für Lärmbekämpfung, 39 (2), 1992 10. Fastl, H., Zwicker, E. Beurteilung lärmarmer Fahrbahnbeläge mit einem Lautheitsmesser. In:

Fortschritte der Akustik, DAGA 86. Proceedings, 1986 11. Ritterstaedt, U., Vogt-Sädler D. Psychoakustische Bewertung eines offenporigen Asphalts, In:

Fortschritte der Akustik, DAGA 2014, Proceedings, 2014 12. TP KoSD-19. Technischen Prüfvorschriften zur Korrekturwertbestimmung der Geräusch-

emission von Straßendeckschichten“ (TP KoSD-19), Forschungsgesellschaft für Straßen- und Verkehrswesen , 2019 13. ISO 532-1. Acoustics. Methods for calculating loudness. Part 1: Zwicker method, International

Organization for Standardization Geneva, Switzerland, 2017 14. DIN 45692:2009-08. Measurement technique for the simulation of the auditory sensation of

sharpness, German Institute for Standardization (DIN), Beuth, 2009 15. Rohrmann, B. Designing verbalized rating scales: Sociolinguistic concepts and psychometric

findings from three cross-cultural projects. Reports, Roman Research Road, Melbourne, 2015 16. Freitas, E., Mendonça, C., Santos, J.A., Murteira, C., Ferreira, J.P. Traffic noise abatement: How

different pavements, vehicle speeds and traffic densities affect annoyance levels, Transportation Research Part D: Transport and Environment , 17 (4), 321-326, doi.org/10.1016/j.trd.2012. 02.001, 2012 17. Dudenhöfer, B., Nolte, K. Leitfaden lärmtechnisch optimierte Asphaltdeckschichten in Berlin,

2018