Deliverable D2.3. Report on the analysis and comparison of existing noise measurement methods for noise properties of road surfaces

Transcription

1 Collaborative Project FP7-SST-2013-RTD-1 Seventh Framework Programme Theme SST : Innovative, cost-effective construction and maintenance for safer, greener and climate resilient roads Start date: 1 November 2013 Duration: 36 months Deliverable D2.3 Report on the analysis and comparison of existing noise measurement methods for noise properties of road surfaces The research leading to these results has received funding from the European Community s Seventh Framework Programme (FP7/ ) under grant agreement n Main Editor(s) Jørgen Kragh, Danish Road Directorate Due Date 31 December 2014 Delivery Date 30 April 2015 Work Package Dissemination level WP2 Measurement methods for the noise emission properties of road surfaces PU Date: 30/04/2015 Version: Final report 1 (68)

2 Contributor(s) Main Contributor(s) Jørgen Kragh, Danish Road Directorate, Tel , Rasmus Stahlfest Holck Skov, Danish Road Directorate, Tel , Jens Oddershede, Danish Road Directorate, Tel , Contributor(s) (alphabetical order) Fabienne Anfosso-Lèdèe, IFSTTAR Wolfram Bartolomaeus, Marek Zöller, BASt Truls Berge, SINTEF Anneleen Bergiers, BRRC Matthew Muirhead, TRL, Reinhard Wehr, AIT Review Reviewer(s) Phil Morgan, TRL Marco Conter, AIT Date: 30/04/2015 Version: Final report 2 (68)

3 Control Sheet Version History Version Date Editor Summary of Modifications 1 2-Feb-2015 J. Kragh 1 st draft of report structure 2 15-Apr-15 J. Kragh 1 st draft of report 3 30-Apr-15 J. Kragh P. Morgan M. Conter Final version of report Final Version released by Circulated to Name Date Recipient Date Manfred Haider 30/04/2015 Coordinator Consortium European Commission Date: 30/04/2015 Version: Final report 3 (68)

4 Analysis of methods for measuring noise properties of road surface courses Final report Contents 1 Background and aim Measurement methods SPB measurement CPX measurement Other noise measurement methods Measurement of L AE Measurement of L Aeq Method applied and limitations Method Limitations Data received Austrian Institute of Technology (AIT), Austria Federal Highway Research Institute (BASt), Germany Belgian Road Research Centre (BRRC), Belgium Danish Road Directorate (DRD), Denmark French Institute of Science and Technology for Transport, Development and Networks (IFSTTAR), France Transport Research Laboratory (TRL), UK Swedish National Road and Transport Research Institute (VTI), Sweden / Technical University of Gdansk (TUG), Poland SINTEF, Norway Data analysis Initial overview Qualification of data Adaptation of data Temperature correction Speed correction...18 Date: 30/04/2015 Version: Final report 4 (68)

5 5.4 L AE measurement Results obtained on SPB - CPX relation Light vehicles SPB and CPX at same reference speed SPB at 120 km/h and CPX at 80 km/h reference speed SPB at 120 km/h normalised to 80 km/h and CPX at 80 km/h SPB noise levels at different receiver heights - Light vehicles Heavy vehicles SPB relation with CPXH V SPB relation with CPXP v SPB noise levels at different receiver heights - Heavy vehicles SPB noise levels at different heights - Summary Applicability of supplementary indicators Maximum noise level and sound exposure level, L AFmax L AE Background Method applied Results Discussion and conclusions References Appendix Processing of received data General Temperature correction Speed correction Correction of measured L AE Measured differences L AFmax L AE Partners data Processing and identification AIT BASt BRRC DRD IFSTTAR...60 Date: 30/04/2015 Version: Final report 5 (68)

6 9.2.6 TRL SINTEF Dutch data on SPB noise levels at different heights...68 Date: 30/04/2015 Version: Final report 6 (68)

7 Abbreviations Abbreviation AC AIT ASTM BASt BRRC CB CPB Meaning Asphalt concrete Austrian Institute of Technology, AT American Society for Testing Materials Federal Highway Research Institute, D Belgian Road Research Centre, BE Coast-By Method according to UNECE R117 Controlled pass-by method CPX Close-Proximity method according ISO/DIS CPX FR CPXP v CPXH v DRD DGD DoW EACC END CPX noise level from French reference tyre CPX noise level for tyre P1 representing passenger cars at reference speed v CPX noise level for tyre H1 representing heavy vehicles at reference speed v Danish Road Directorate, DK Dutch abbreviation: Thin noise reducing surfacing Description of Work Exposed aggregate cement concrete Environmental Noise Directive (2002/49/EC) END T Estimated noise level difference due to texture changes (ISO 10844) ERNL FR1 Estimated road noisiness level (Descornet/Sandberg model) French CPX test wheel H1 CPX reference test tyre proxy for heavy vehicles according to ISO/TS IFSTTAR L Acpx L Aspb French Institute of Science and Technology for Transport, Development and Networks, FR A-weighted CPX noise level per one-third octave-band A-weighted SPB noise level per one-third octave-band L veh Vehicle sound level measured according to ISO L tx MPD Nord2000 OPA Surface texture level Mean profile depth Nordic prediction method for road traffic noise German abbreviation: Porous asphalt P1 CPX reference test tyre proxy for light vehicles according to ISO/TS PA PMA SINTEF SMA Porous asphalt Porous mastic asphalt, i.e. mastic asphalt EN (Gussasphalt) with an open graded texture at the top to avoid air pumping noise Foundation for scientific and industrial research, NO Stone mastic asphalt Date: 30/04/2015 Version: Final report 7 (68)

8 SPB Statistical Pass-By method according ISO SPL2000 Software developed by DELTA for calculation according to Nord2000 SRTT Standard Reference Test Tyre, P1 TLPA Two-layer porous asphalt TRL Transport Research Laboratory, UK TUG Technical University of Gdansk, PL UTLAC Ultra-thin layer asphalt concrete VTI Swedish National Road and Transport Research Institute, SE WB German abbreviation: Exposed aggregate cement concrete ZOAB Dutch abbreviation: ZOAB(+) = single layer PA; ZOAB-TL = two-layer PA ZWOPA German abbreviation: Two-layer porous asphalt Date: 30/04/2015 Version: Final report 8 (68)

9 Preface The main objective of the ROSANNE project is to advance the harmonization of measurement methods for tyre/road noise emission on various pavements and methods for measuring pavement skid resistance and rolling resistance. Ultimately, the goal is to identify or develop procedures which can be fed into the process of standardization. ROSANNE Work Package 2 (WP2) deals with the characterization of road surface noise properties. Existing measurement methods such as the Statistical Pass-By (SPB) and Close- Proximity (CPX) methods may have different purposes and uses but they are intended to be as consistent as possible and stable over a foreseeable time period. The ultimate aim of WP2 is to provide a draft for the standardization of a procedure for harmonized characterization of road surfaces across Europe. Another objective is to provide methods which enable standardized measurement of road vehicle noise emission data that can be included in European road traffic noise prediction methods. This report deals with Task 2.1 of the ROSANNE project. The objective is to study the relationships between measurements of the acoustic properties of road surfaces made with different methods, particularly those in ISO and ISO/DIS The primary target group for Task 2.1 analyses is those ROSANNE partners working on a harmonised characterization method (in Task 2.3) and input data for traffic noise prediction (in Task 2.4). The analyses in Task 2.1 were conducted by the Danish Road Directorate (DRD) based on DRD data and on data and information received from the ROSANNE partners and others mentioned in Section 4. Executive Summary This report deals with Task 2.1 of the ROSANNE project. The objective is to study the relationships between measurements of the acoustic properties of road surfaces made with different methods, particularly those in ISO and ISO/DIS DRD collected sets of data on CPX noise levels and vehicle pass-by noise levels measured on the same pavements at approximately the same time. These data were pre-processed and grouped into two main groups: those where both types of noise level had been recorded at the same reference speed, and those where CPX noise levels had been recorded at 80 km/h while vehicle pass-by noise levels had been measured at reference speeds km/h. For passenger cars, in the former group of data there was a rather clear connection between CPX noise levels measured with standard P1 reference tyre (Uniroyal Tiger Paw ASTM SRTT F ) and pass-by noise levels, namely L AFmax = 0.95 CPXP db Date: 30/04/2015 Version: Final report 9 (68)

10 according to which the average pass-by noise level is 20.5 db lower than the average CPX noise level. In the latter group, when the pass-by noise levels were normalised to 80 km/h reference speed using speed coefficients derived from an IFSTTAR database, these normalised noise levels were grouped around the average relationship given by the above equation but with a large spread, so that almost half of this latter group of data deviated by ±2 db or more from the average relation. For multi-axle trucks, pass-by noise levels had a clearer relationship with CPX noise levels measured with the standard H1 reference tyre (Avon AV4) than with CPX noise levels measured with the standard P1 reference tyre, given by L AFmax,80 = 0.65 CPXH db The average pass-by noise level from a multi-axle truck on dense pavement was 9.5 db lower than the CPX noise level measured with Avon AV4 tyres. For two-axle trucks the corresponding difference is 12.0 db. The relationships mentioned are based on pass-by noise levels measured at a height of 1.2 m. On average the pass-by noise level decreases by 0.5 db for each metre the measurement height is increased but with rather large variation. On some porous pavements the pass-by noise level was more or less the same at all microphone heights. This underlines that a relative classification of pavement noise properties may depend on the measurement height of the pass-by noise levels and thus a height should be selected which is representative to the noise exposure of people living/working close to the road. At 1.2 m measurement height the relation between the maximum noise level measured with time weighting F and the sound exposure level measured during a vehicle pass-by can be described by the relationship L AFmax - L AE = 2.79 ln(v) db for cars and L AFmax - L AE = 2.23 ln(v) db for trucks. The fact that there is such a connection implies that measuring sound exposure levels rather than maximum pass-by noise levels will not lead to essential improvement in vehicle noise emission values derived from results of vehicle pass-by noise measurements. Date: 30/04/2015 Version: Final report 10 (68)

11 1 Background and aim The most important methods available for characterising road surface noise characteristics are the so-called Statistical Pass-by (SPB) method defined in ISO [1] and the Close- Proximity (CPX) method described in ISO/DIS [2]. The main purpose of Task 2.1, reported here, is to look at results found using these two methods to determine if pavement noise properties are ranked in the same way by the two methods. In that case, a harmonised pavement noise classification system may be established using one of the methods or, for special cases, perhaps a combination of both methods supplemented with measures of proxy parameters such as surface texture or sound absorption. Date: 30/04/2015 Version: Final report 11 (68)

12 2 Measurement methods Advantages, disadvantages and uncertainties involved in using the SPB and CPX methods have been discussed at length earlier, both in the (draft) standards themselves and, for example, in reports like the SILENCE project deliverable F.D12 [3]. The following sections briefly summarise the methods and their characteristics. A few other methods are mentioned in Section SPB measurement The SPB method is based on measuring the maximum A-weighted sound pressure level and the (constant) vehicle speed during the pass-by of a number of individual vehicles in the ordinary traffic. By measuring a sufficient number of pass-by noise levels of different categories of vehicle, representative average vehicle noise levels at specified reference speeds are determined by means of linear regression of noise level against vehicle speed. Such average noise levels from different vehicle categories may then be combined to represent the noise from mixed road traffic. The fact that noise levels are measured at the side of the road assists in having a transparent method, meaning that people living/working close to the road tend to show more confidence in such directly measured noise levels than in predicted noise levels or noise levels at the roadside derived from CPX noise measurements. A major disadvantage of the SPB method is that the measurement result only represents a short section of the road. Another drawback is the sensitivity to sound reflected from structures such as buildings or parked vehicles and to screening provided by guardrails or other objects affecting sound propagation from vehicles to the microphone position. Effects of reflections from objects behind the microphone may be controlled by placing the microphone on a so-called backing board to screen the microphone from reflections from behind, see [4]. 1.2 CPX measurement The CPX method is based on measuring the noise generated by specified reference tyres running on the surface under test. The tyres are mounted on a vehicle, which may be selfpowered or a towed trailer. The vehicle speed is recorded together with the noise level and the measurement results are normalised to a specified reference speed. Noise levels are averaged over segments of the road and may be averaged over different wheel tracks and over several repeated runs. The main advantages of the CPX method are that the results characterise long sections of road and that it is less sensitive than the SPB method to influences from variation in the road environment. Among its disadvantages are that a measurement result only represents tyre/road noise, not the vehicle propulsion system noise, and that results are specific for the applied reference tyre which may not necessarily represent the tyre/road noise from various categories of vehicle. Difficulties may arise when the measuring speed cannot be kept in the Date: 30/04/2015 Version: Final report 12 (68)

13 specified speed range due to slow or stationary traffic. In such cases it may be necessary to compensate by repeating a run several times. 1.3 Other noise measurement methods Measurement of L AE When measuring vehicle pass-by noise levels according to ISO only the maximum A-weighted sound pressure level, L AFmax, with time weighting F is recorded (see Section 7.1). This noise level may occur at different times during the vehicle pass-by, for example when the drive-axle wheels or the engine exhaust opening is near to the microphone, and this noise level may or may not be representative of the energy average noise level during the entire pass-by. The latter may be characterised by the sound exposure level L AE (or SEL), which is the total sound energy received during the pass-by normalised to an event duration of one second. L AE is potentially more representative than L AFmax to the equivalent continuous noise level at the roadside, in particular if the vehicle noise emission is directional in a horizontal plane. When the traffic is dense it is a challenge to record a sufficient part of the noise level time history to get a reliable value of L AE. This is the main reason why the SPB standard requires L AFmax and not L AE measurement. The importance of this is dealt with in Section Measurement of L Aeq Oftentimes the energy-equivalent A-weighted sound pressure level, L Aeq,T, over a time interval T is measured on a building façade to determine before/after changes after having repaved a road. This can either be a tool for classifying and ranking different types of road surfaces according to their influence on noise emission or for evaluating the effect of different road surfaces at a particular site, especially before and after resurfacing. Such measurements are popular with municipal administrations wanting to communicate to citizens that they are cared for, for example in the frame of action plans according to the Environmental Noise Directive (END) 2002/49/EC. Such before/after measurements can also be made by means of the SPB method. If the environment influence (in particular any sound reflections) are kept constant the method requirements may be relaxed and yet a quite good estimate of the difference between pavement acoustic properties may be obtained, although not sufficient for an official characterisation of road surfaces. Date: 30/04/2015 Version: Final report 13 (68)

14 3 Method applied and limitations 3.1 Method ROSANNE data were collected via mail correspondence with project partners and others (see Section 4). This was intended for finding out if CPX and SPB measurements rank road surfaces in the same way and in order to establish relationship(s) between results obtained using the two methods. Other useful information was found in several reports such as a) French reports on measured and calculated differences between CPX and Controlled Pass-by (CPB) noise levels [5] - [7] (see Section 9.2.5). b) A report from a Danish-Dutch project on the acoustic ageing of pavements [8] which comprised data on SPB noise levels measured at different heights (see Section 6.3). As a part of the ROSANNE project DRD also made computations, i.e. Nord2000 simulations, for use in correcting L AE measurement results (see Section 9.1.3). Finally, DRD carried out a series of new measurements explicitly for ROSANNE to determine relations between vehicle pass-by noise levels at different measurement heights and between the maximum A-weighted noise level L AFmax with time weighting F and the sound exposure level L AE for various categories of vehicle. 3.2 Limitations DRD received no data on noise levels measured on tined, ground 1) or polished surfaces, or on surfaces with other directional textures as mentioned in the project Description of Work (DoW) for this Work Package. On non-standard pavements, data are available only from one site with resonators placed at the base of two-layer porous asphalt and three sites with poroelastic pavements. The project DoW indicated that the use of texture indicators to harmonize and/or supplement the methods should be studied. An initial attempt was made as mentioned in Section 6.4 with so little success that no further attempts were made. 1) AIT has a few results of CPXP v measurements on diamond grinding but no such SPB data Date: 30/04/2015 Version: Final report 14 (68)

15 4 Data received This section briefly outlines the data received from project partners and others. The expanded acronyms for all surface types are detailed in the Abbreviations table on page 7. The report uses the terminology given in Table 4-1 concerning reference tyres for CPX measurements, see [9] - [10]. Table 4-1: Reference tyres used for comparison of CPX and SPB resp. CB noise levels Abbreviation Tyre line Tyre size P1 [9] Uniroyal Tiger Paw (ASTM SRTT F ) 225/60 R16 97S H1 [9] Avon AV4 Supervan 195 R14 C 106/104N FR1 [10] French CPX test wheel: Michelin E3A tyre. Load 3.32 kn; Inflation pressure 0.22 MPa 195/65 R Austrian Institute of Technology (AIT), Austria AIT provided data from four roads [11] comprising two SMA and two EACC (see Section 9.2.1) as follows: CPX: CPXP 70, 80, 100 ; SPB: Cars at 120 km/h and multi-axle trucks at 80 km/h. 4.2 Federal Highway Research Institute (BASt), Germany Data were received from BASt [12] for 26 roads sections comprising 11 SMA/PMA, 1 TLPA (ZWOPA), 2 OPA, 11 EACC (WB) 0/8 and 1 EACC (WB) 0/11 (see Section 9.2.2). DRD has omitted incomplete data from two of these roads. Data were as follows: CPX: CPXP 80 and CPXH 80 ; SPB: Cars at 120 km/h (some on SMA at lower speeds) and trucks at average speeds of km/h all measured at 4 microphone heights. 4.3 Belgian Road Research Centre (BRRC), Belgium BRRC sent data from 10 road sections comprising 8 noise reducing thin layers and 2 reference sections (SMA 10 & TLPA) [13] (see Section 9.2.3). Data were as follows: CPX: CPXP 80 and CPXH 80 ; SPB: 3 positions per road section for cars at 80 km/h. SPB results were suspected by DRD to be subject to unwanted propagation effects due to grassland. 4.4 Danish Road Directorate (DRD), Denmark Measurement results collected in 2014 were available from a total of 41 roads (see Section 9.2.4). CPX data comprised CPXP 80 and/or some CPXP 110. SPB noise levels at 1.2 m were available for all roads, and data from supplementary microphone heights at 3 m and 5 m together with measured L AE -values (see Section 5.4) were available for 22 roads. Date: 30/04/2015 Version: Final report 15 (68)

16 4.5 French Institute of Science and Technology for Transport, Development and Networks (IFSTTAR), France IFSTTAR delivered data comparing the differences between CPX and CPB measurements on 18 dense and 5 porous pavements [14] (see Section 9.2.5). These were measured with the French CPX reference wheel FR1 and not with the reference tyre P1 according to ISO/TS (see Table 4-1) resulting in different CPX levels, i.e. CPX FR rather than CPXP V. IFSTTAR also performed statistical analyses of slope coefficients in a French National database of SPB measurement results [15]. The results are given in Section Transport Research Laboratory (TRL), UK TRL provided data from four types of thin dense asphalt surface courses: 6 mm, 10 mm, and two 14 mm, each measured in two consecutive years, [16] (see Section 9.2.6). Data were as follows: CPX: CPXP 80 ; SPB: Cars at 110 km/h; medium and heavy trucks at 85 km/h. 4.7 Swedish National Road and Transport Research Institute (VTI), Sweden / Technical University of Gdansk (TUG), Poland No data were received from VTI/TUG. SPB and CPX data were available at VTI/TUG from the same surfaces but the CPX data had been collected using the earlier reference tyre Avon ZV1 [17]. 4.8 SINTEF, Norway SINTEF sent data from roads with mostly dense pavements [18] (see Section 9.2.7). DRD extracted the following sets of data: CPX: CPXP 80 and CPXH 80; SPB: Noise levels from cars (14 roads) and trucks (multi-axle and two-axle trucks on 8 roads). Date: 30/04/2015 Version: Final report 16 (68)

17 5 Data analysis 5.1 Initial overview Upon receiving data from partners DRD had a first look at trends in differences between CPX and SPB levels. Various combinations of reference speeds were seen, e.g. in some cases both CPX and SPB noise levels had been recorded at 80 km/h and 110 km/h, in others CPX had been recorded at 80 km/h, SPB at 120 km/h etc. Preliminary corrections were made in each set of data. This first look at the results indicated that there was a surprisingly large range of noise level differences. For example, in seemingly comparable measurement situations a range of differences between CPX and SPB noise levels from below 20 db to more than 25 db was seen. Some partners data set tended to show high differences, others low differences and some had both high and low differences in their data, depending on the pavement type. This was presented at a WP2 meeting 22 October 2014 in Brussels and it was decided to generate overall tables of collected data in order to be able to identify and discard outliers, if any were present. In order to be able to combine data from different partners, their data were qualified and adjusted as described in the following sections. 5.2 Qualification of data Closer inspection of the data revealed that the BRRC SPB data seemed to be affected by an unwanted influence from the terrain between the road and microphone. These data were therefore left out of the data compilation (see further details in Section 9.2.3). 5.3 Adaptation of data The first step in the adaptation of data for the compilation was pre-processing by adjusting for differences in temperature correction and differences in reference speeds. In the next step data were grouped according to pavement type Temperature correction Some of the received data had been temperature corrected using various temperature coefficients. These were un-corrected and all data were normalised to 20ºC air temperature applying the temperature coefficients given in Table 5-1. Table 5-1: Temperature coefficients applied for normalising data to 20 C [19] Pavement type Temperature coefficient [db/ C] Light vehicles Heavy vehicles Dense asphalt -0,10-0,05 Porous asphalt ,025 Cement concrete ,035 Date: 30/04/2015 Version: Final report 17 (68)

18 5.3.2 Speed correction SPB data are determined by means of linear regression of the pass-by noise level on the logarithm of the vehicle speed as in Eq. (6-1). L veh = a + b log 10 (v) (6-1) In cases where CPX and SPB noise levels had been measured at the same reference speed there was no need for speed correction. Data from sites where the measurements had been made at different reference speeds in some cases could be corrected to the same reference speed based on the results themselves. This was the case in some SPB results from BASt which include regression line slopes: for seven SMA pavements the regression line range of validity according to [1] allowed the normalisation of data to 80 km/h for direct comparison with CPXP 80 noise levels. For data where the speed range of validity did not allow such normalisation, relationships are first given between CPXP 80 and SPB(120 km/h). After discussions with ROSANNE partners it was decided to also correct these latter SPB results to 80 km/h. The speed corrections for this normalisation were based on analyses made by IFSTTAR of the speed coefficients b as given in Eq. (6-1) in its database of SPB measurement results, see Section There was no correlation between the speed coefficient and the surface age, the reference speed, the traffic intensity, etc. In all these cases, the analyses resulted in a cloud of uncorrelated data [15]. The following average slopes were found, (the numbers given in parentheses are the number of SPB measurements each coefficient is based on): For light vehicles: b = 28 for porous and semi-porous (N = 211), 30 for dense pavement (N = 258). For heavy trucks: b = 28 for porous (N = 71), 30 for semi-porous (N = 37) and 27 for dense pavement (N = 156). For light vehicles the coefficients given above were applied to normalise the SPB data. The distinction between coefficients, however, is not crucial as illustrated in Section 9.1. Figure 5-1 illustrates the slopes from the BASt data together with the average slope b = 30.3 for dense pavements in the IFSTTAR database. The curved grey line shows 30.3 log 10 (v) relative to the value 0 db at 110 km/h. Each coloured straight line in the figure illustrates the regression line slope from one SPB measurement made by BASt. Each line connects the value of b log 10 (v) at the ends of the range of validity for the regression line, and each line has been positioned so that it has its midpoint on the grey IFSTTAR curve. Within the speed ranges of validity there is little difference. Date: 30/04/2015 Version: Final report 18 (68)

19 Figure 5-1: Illustration of speed slopes in SPB regression lines. Grey line: 30.3 log 10 (v) from IFSTTAR database, relative to its value at 110 km/h. Other lines represent BASt results 5.4 L AE measurement DRD has measured both sound exposure levels and maximum pass-by noise levels for the ROSANNE project on a number of sites during the 2014 measurement season. The measurement method and the results are mentioned further in Section 7. During selected pass-bys the time function (or profile ) of the pass-by noise level for individual vehicles was recorded. In the laboratory the energy sum of recorded signal was calculated and normalised to 1 s event duration. Then corrections were made for any missing tails of the time function, see Section Date: 30/04/2015 Version: Final report 19 (68)

20 6 Results obtained on SPB - CPX relation Section 6.1 deals with the relations between SPB noise levels from light vehicles while Section 6.2 deals with SPB noise levels from heavy vehicles. The first subsection on light vehicles is on results of SPB and CPX measurements made at the same reference speed while the following two subsections deal with the results of such measurements made at different reference speeds. The expanded acronyms used in the Figures for different surface types are detailed in the Abbreviations table on page Light vehicles SPB and CPX at same reference speed This section deals with data for light vehicles measured at nearly the same reference speed. Figure 6-1 shows the pass-by noise levels from light vehicles at 80 km/h as a function of CPXP 80. There is a reasonably straightforward relation with ±0.5 db or ±1 db variation in SPB noise levels for the same CPX noise level, independent of pavement type and the source of measurement results. Around 90 % of the SPB results lie within the grey lines SPB = CPX db and SPB = CPX db shown in Figure 6-1. Suspicions raised by measurement results shown later in Figure 6-3, that German SPB noise levels would be higher than Danish SPB noise levels because Germans have larger/heavier cars than Danes, are not supported by the results in Figure 6-1. Figure 6-1: Relationship between CPXP 80 and L veh measured at 80 km/h. The legend indicates pavement type and data origin Date: 30/04/2015 Version: Final report 20 (68)

21 Figure 6-2 is a compilation of ROSANNE data and earlier DRD data. The figure shows sets of CPX and SPB noise levels measured at the same reference speed, which may have been 50 km/h, 80 km/h or 110 km/h. The ROSANNE data shown in blue, except for three sets of data in orange colour which are considered outliers, were all measured at 80 km/h or 110 km/h reference speed. The (older) Danish data shown in grey were measured at the three reference speeds given in the figure legend. Figure 6-2 shows two trend lines: A blue line for the ROSANNE data (the data denoted ROSANNE in the key only outliers do not contribute to this trend line) A dashed red line for the grey data points from Danish measurements made in for other projects. The blue ROSANNE data points which include the new Danish ROSANNE data show a trend for higher SPB noise levels than the older Danish data. The average relationship between these ROSANNE noise levels is SPB = CPX db. Figures in Sections identify data points from contributing parties. Figure 6-2: Relationship between CPXP 80 and L veh measured at the same reference speed (outliers do not contribute to the blue trend line). Grey data points are from Danish measurements in ; the dashed red line is the trend line for these data SPB at 120 km/h and CPX at 80 km/h reference speed In many cases, the ROSANNE data contained SPB measurement results at reference speeds differing so much from the reference speed used for CPX measurements that the Date: 30/04/2015 Version: Final report 21 (68)

22 validity range for the SPB regression lines did not encompass the CPX reference speed. For these cases it was first decided to normalise the SPB measurement results to 120 km/h and to show their relation with CPX noise levels measured at 80 km/h. The results are illustrated in Figure 6-3. The relationship between SPB and CPX noise levels in this figure is not clear, as follows. SPB data from the pavements denoted as Porous, most data denoted as Dense and some data denoted as SMA are within SPB = CPX ± 1 db Several data denoted as being SMA, EACC, PMA and Dense deviate by 2 db or more from SPB = CPX 18.0 db. The large group of data from EACC with CPXP 80 = db deviate by 4 7 db as does one of the PMA points and one SMA point. Figures in Sections show which data points are from which contributing party. Figure 6-3: Relationship between CPXP 80 noise levels and L veh for light vehicles measured at 120 km/h or measured at 110 km/h and converted to 120 km/h SPB at 120 km/h normalised to 80 km/h and CPX at 80 km/h After discussions at the WP2 meeting on 9 March 2015 it was decided to normalise SPB results measured at 120 km/h reference speed to 80 km/h even though the latter speed is outside the range of validity of the regression lines of noise levels against vehicle speed. This is a deliberate violation of the specifications in the measurement standard [1]. The results are illustrated in Figure 6-4. The figure shows the same blue, grey and orange data points as Date: 30/04/2015 Version: Final report 22 (68)

23 Figure 6-2. The normalised data are shown in green/orange and they still display a large variation with almost half of the data points being more than 2 db from the average, in both directions. Figure 6-4: Relationship between CPXP V and L veh measured at the same reference speed from Figure 6-2, combined with SPB results measured at 120 km/h but corrected to 80 km/h SPB noise levels at different receiver heights - Light vehicles This section deals with SPB measurement results provided for the ROSANNE project by BASt and DRD. BASt measured simultaneously at 1.2 m, 2.4 m, 3.6 m and 4.8 m height, while DRD measured at 1.2 m, 3 m and 5 m, respectively. All results but those from the German pavement denoted ZWOPA 0/8 + 0/11 are included here. Results from the latter pavement, which is two-layer porous asphalt with Helmholtz resonators below it, showed trends different from the remainder of the data. Figure 6-5 shows the SPB noise level at different heights as a function of the noise level at 1.2 m height. BASt data are from SMA/PMA or EACC pavements while DRD data are from SMA, AC, UTLAC, semi-porous asphalt and PERS. The trend lines shown are essentially parallel with slopes around 1.0, so the general effect of increasing the microphone height is decreasing the measured pass-by noise level. Date: 30/04/2015 Version: Final report 23 (68)

24 Figure 6-5: SPB noise levels from cars measured at 2.4, 3, 3.6, 4.8 and 5 m height as a function of the noise level at 1.2 m height Figure 6-6 shows the same results in a slightly different way. For each measurement height the figure shows the reduction in noise level by increasing the microphone height from the 1.2 m height specified in the SPB measurement standard. There is a clear trend for lower noise levels with increasing microphone height. The spread in differences measured at individual sites is rather large. This variation may be caused by general measurement uncertainty or by frequency spectrum differences due to variations in sound propagation, pavement surface or vehicle speed. Date: 30/04/2015 Version: Final report 24 (68)

25 Figure 6-6: Difference between SPB noise levels from cars measured at 1.2 m and SPB at other measurement heights, for individual surface courses Figure 6-7 shows the average differences from Figure 6-6 and their standard deviations. The trend line for these average differences has a slope of 0.5 db per meter increase in microphone height and the standard deviations of the differences are between 0.4 db and 0.8 db. In Figure 6-8 and Figure 6-9 the differences are shown for the pavement types EACC, SMA, AC and PA, respectively. The trend is for noise levels measured at EACC to decrease more than the others when the microphone height is increased from 1.2 m to 2.4 m. The noise levels at the PA pavements were almost independent of the microphone height. Figure 6-7: Average differences and their standard deviations between SPB noise levels at 1.2 m and SPB at other measurement heights Date: 30/04/2015 Version: Final report 25 (68)

26 Figure 6-8: Differences between SPB noise levels at 1.2 m and SPB at other measurement heights on EACC, SMA, AC and PA pavements Figure 6-9: Average differences and their standard deviations between SPB noise levels at 1.2 m and SPB at other measurement heights on EACC, SMA and AC pavements Figure 6-10 shows pass-by noise levels from cars measured by BASt at three sites with porous asphalt. These noise levels are shown as a function of the microphone height and the results at 1.2 m height are those shown with crosses in Figure 6-3. The measurement results in Figure 6-10 are shown as noise levels relative to the average of 12 BASt measurement results from EACC. The figure illustrates that if pass-by noise levels are measured at 4.8 m the porous pavements are 6.5 db quieter than EACC while measured at 2.4 m height they are 8.0 db quieter. At 1.2 m the difference is 9 11 db. Date: 30/04/2015 Version: Final report 26 (68)

27 Figure 6-10: SPB noise levels from cars measured at different heights on three BAST sites with porous asphalt. The levels are given relative to the average noise level measured at the same heights at 12 sites with EACC 6.2 Heavy vehicles SPB relation with CPXH V BASt and SINTEF contributed CPX noise levels measured with reference tyre H1 (Avon AV4), and SPB noise levels from heavy vehicles. BASt only measured noise levels from multi-axle trucks, while SINTEF also measured noise levels from two-axle heavy vehicles. The results for multi-axle heavy vehicles are illustrated in Figure 6-11, with both CPX and SPB noise levels given at 80 km/h. There is a trend, although not very clear, for increasing SPB noise level with increasing CPX noise level. The results from porous asphalt deviate from those from other types of pavement. In Figure 6-12 the data from SMA, EACC, PMA and AC have been grouped into dense pavements and the trend line for these data is shown. The determination coefficient R 2 is 0.5. Date: 30/04/2015 Version: Final report 27 (68)

28 Figure 6-11: Relationship between CPXH 80 noise levels and SPB noise levels for multi-axle heavy vehicles at 80 km/h Figure 6-12: Relation between CPXH 80 noise levels and SPB noise levels for multi-axle heavy vehicles at 80 km/h. The pavements are the same as in Figure 6-11, and the designation Dense includes SMA, EACC, PMA and AC Date: 30/04/2015 Version: Final report 28 (68)

29 The average difference between pass-by noise levels from multi-axle and two-axle heavy vehicles was 2.5 db. This difference is based on results from four British sites, each measured twice, and from 14 Norwegian sites SPB relation with CPXP v Parties other than BASt and SINTEF only delivered CPX data measured with reference tyre P1 (SRTT). Figure 6-13 shows the relation between CPHP 80 and the pass-by noise levels from multi-axle trucks. No clear overall trend is seen in the data, not even in data for each type of pavement. Figure 6-13: Relation between CPXP 80 noise levels and SPB noise levels for multi-axle heavy vehicles at 80 km/h In order to compare how well the two reference tyres represent pass-by noise levels from multi-axle trucks Figure 6-14 shows data from the same sites as Figure 6-11 but measured with SRTT rather than Avon AV4. Again, no clear trend is seen. In Figure 6-15 data from pavements types SMA, EACC, PMA and AC have been grouped as Dense and the trend line is given. The determination coefficient R 2 = 0.2, and thus the Avon AV4 reference tyre is a better representative of pass-by noise levels from multi-axle trucks than the SRTT, but there is considerable spread in data for individual pavements. Date: 30/04/2015 Version: Final report 29 (68)

30 Figure 6-14: Relation between CPXP 80 noise levels and SPB noise levels for multi-axle heavy vehicles at 80 km/h. The surfaces are the same as in Figure 6-11 Figure 6-15: Relation between CPXP 80 noise levels and SPB noise levels from multi-axle heavy vehicles at 80 km/h. The surfaces are the same as in Figure 6-14, and the designation Dense includes SMA, EACC, PMA and AC Date: 30/04/2015 Version: Final report 30 (68)

31 6.2.3 SPB noise levels at different receiver heights - Heavy vehicles This section, like Section 6.1.4, deals with data from BASt and from DRD. BASt measured pass-by noise levels at 1.2, 2.4, 3.6 and 4.8 m height while DRD measured at 1.2, 3 and 5 m. height. Results from the German pavement denoted ZWOPA 0/8 + 0/11 have been excluded because the data from two-layer porous asphalt with Helmholtz resonators below it showed trends different from the remainder of data. All trend lines in Figure 6-16 are roughly parallel with a slope in the order of 1. The range in noise levels in the data from DRD is small, and data from an individual pavement may have heavy influence on the trend line. Figure 6-16: Relation between SPB noise levels from multi-axle trucks measured at 1.2 m and 2.4, 3, 3.6, 4.8 and 5 m height, respectively Like in Figure 6-6, Figure 6-17 shows the difference between the noise level at 1.2 m and the noise level at other measurement heights. There is a clear trend for lower noise levels with increasing microphone height. The variation in differences measured at individual sites is rather large. This variation may be caused by general measurement uncertainty or by frequency spectrum differences due to different sound propagation condition, pavement surface or vehicle speed. Figure 6-18 shows the average differences from Figure 6-17 and their standard deviations. The trend line for these average differences has a slope of 0.6 db per meter increase in microphone height and the standard deviations of the differences are between 0.3 db and 0.6 db. Date: 30/04/2015 Version: Final report 31 (68)

32 Figure 6-17: Difference between SPB from multi-axle trucks at 1.2 m and SPB at other measurement heights for individual surface courses Figure 6-18: Average difference and the standard deviation of differences between SPB from multi-axle trucks at 1.2 m and SPB at other measurement heights 6.3 SPB noise levels at different heights - Summary The average difference between vehicle pass-by noise levels measured at different heights is about the same for cars and multi-axle trucks. Table 6-1 summarises the average difference between the noise levels at 1.2 m and the noise level at other measurement Date: 30/04/2015 Version: Final report 32 (68)

33 heights. The table also gives average differences found in Dutch measurement results [8], see also Figure Table 6-1: Average difference between pass-by noise levels at 1.2 m and other measurement heights. ROSANNE data and Dutch data from 2010 (see Figure 9-34) ROSANNE Dutch data 2010 Height Cars Heavy Cars Heavy [m] [db] [db] [db] [db] Applicability of supplementary indicators An attempt was made to find out if surface texture or sound absorption data would be useful parameters for pavement classification, but it seems as if their effects are too small or too random compared to the spread in data for such parameters to contribute to an improved classification. Texture data were included in the data received from BASt, i.e. for 11 of their 23 pavements 2). However, DRD data collected in other projects than ROSANNE on many sites with different types of pavement showed no correlation between Mean Profile Depth (MPD) and CPXP 80, see Figure Another discouraging example is illustrated in Figure Figure The first of these figures show the CPX noise level measured on two consecutive sections of road paved seven weeks earlier with a surface layer of SMA 8. These sections of road had been paved with an interval of four days during the second half of June 2014 by the same contractor using asphalt from the same plant mixed after the same recipe. The noise level was 1.5 db higher on the second section (N2_2: km 2.86 km 4.34) than on the first section (N2_1: km 1.00 km 2.80). These noise level data points are marked by labels in Figure ) plus BRRC data, where noise levels were influence by unwanted propagation effects, see Sec Date: 30/04/2015 Version: Final report 33 (68)

34 Figure 6-19: Measured CPX noise levels as a function of MPD on Danish pavements Figure 6-20: Example of CPX noise level recording on a motorway newly paved with SMA 8 Date: 30/04/2015 Version: Final report 34 (68)

35 Figure 6-21: Frequency spectra measured on two consecutive sections of new Danish SMA 8 built with 4 days interval using the same mix recipe etc. Left: CPXP 80 ; Right: SPB 80 from cars Figure 6-22: Surface profile texture spectra measured the same day as the CPX measurements on two consecutive sections of SMA built with a 4 days interval The left part of Figure 6-21 shows the A-weighted frequency spectra of the CPX noise levels from Figure The difference in noise levels occurs at frequencies above 800 Hz. Ten weeks after the CPX measurements, supplementary SPB measurements were taken at a position at the side of each road section. The SPB frequency spectra in the right part of Figure 6-21 display very much the same differences as the CPX spectra. Figure 6-22 shows the surface profile texture spectra from each of the two road sections. The first part (represented by km ) has the highest texture levels at all but the smallest wavelengths. Table 6-2 summarises the noise levels and gives the MPD values. The table also gives the void contents (± their standard deviation) measured on 12 drill cores taken from each of the two sections of road. It is seen, that on the road section having the lowest MPD value the Date: 30/04/2015 Version: Final report 35 (68)

36 measured CPX and SPB noise levels were higher than the measured noise levels on the section having the highest MPD value. The noise levels ERNL and END T calculated according to the Descornet/Sandberg model and with the ISO 10844:2011 procedure, respectively, show the opposite trend. Table 6-2: Measured CPX noise levels and other pavement characteristics on two consecutive road sections with new SMA 8 Section SMA 8 CPXP 80 SPB cars,110 MPD Void ERNL END t [db] [db] [mm] [%] [db] [db] N2_1 km ± ± N2-2 km ± ± Difference Date: 30/04/2015 Version: Final report 36 (68)

37 7 Maximum noise level and sound exposure level, L AFmax L AE 7.1 Background It has been speculated that the maximum noise level with time weighting F recorded during a vehicle pass-by might not be representative of the total sound power emitted by the vehicle, particularly for long vehicles such as multi-axle trucks. During the 2014 measurement season DRD therefore supplemented some of its normal noise monitoring measurements by measuring both maximum noise levels L AFmax according to ISO and sound exposure levels L AE. Figure 7-1 shows an example of the noise level measured with time weighting F as a function of time during a vehicle pass-by. Figure 7-1: Noise level time history during vehicle pass-by and measured noise levels The figure also illustrates the maximum A-weighted noise level L AFmax and the energyequivalent noise level L Aeq,T measured over T = 6.8 s around the maximum. This equivalent noise level normalised to 1 s duration is the sound exposure level L AE. The sound exposure level, in principle, is a better descriptor of the vehicle noise emission than L AFmax in terms of the contribution of the vehicle to the equivalent noise level at road neighbours because it represents an integration of the variation of vehicle noise emission in a horizontal plane. 7.2 Method applied Noise level time histories were recorded during the pass-by of a number of vehicles. Every 25 ms the measuring system stores the time-weighted total A-weighted noise level and onethird-octave band frequency spectrum. DRD selected vehicle pass-bys with time level histories little affected by background noise, including noise from other vehicles. The sound exposure level for each pass-by was calculated based on part of the noise level time history as illustrated in Figure 7-1, and a correction was made for the missing part of the time level history as described in Section 9.1.3, and the difference between L AFmax and L AE was determined for each pass-by. Date: 30/04/2015 Version: Final report 37 (68)

38 7.3 Results Figure 7-2 illustrates the results of measurements mad at a microphone height of 1.2 m. Measurement results from 3 m and 5 m height are shown in Section The relation in Figure 7-2 between L AFmax and L AE appears to be approximately the same for cars and heavy vehicles. The average relation is L AFmax - L AE = 2.79 ln (v) db for cars and L AFmax - L AE = 2.23 ln (v) db for trucks. These relations are so similar that the relation for trucks should be valid for both two-axle and multi-axle trucks. Figure 7-2: L AFmax L AE at 1.2 m measurement height on a variety of pavements as a function of the vehicle speed. Square data points: Multi-axle vehicles; Diamond shaped data points: Cars. Data points show the average difference and error bars show 95 % confidence intervals Date: 30/04/2015 Version: Final report 38 (68)

39 8 Discussion and conclusions The objective of the study is to investigate the relationships between measurements of the acoustic properties of road surfaces made with different methods, particularly those in ISO and ISO/DIS There seems to be a reasonably 1:1 relationship between CPX noise levels measured with tyre P1 (the standard reference test tyre, SRTT), and SPB noise levels from passenger cars, as long as both types of noise level are measured at the same reference speed. The average relationship found in the data collected for ROSANNE at a microphone height of 1.2 m is Cars: L AFmax = 0.95 CPXP db (9-1) This equation yields a 20.5 db average difference between the two measured quantities, and almost 90 % of all data are within ±1 db around this trend line. Sets of data where the CPX noise levels had been measured at 80 km/h reference speed while the vehicle pass-by noise levels had been measured at km/h reference speeds did not show the same clear pattern. The reasons for this are not known. When corrected for the speed difference using the speed coefficients given in Section almost half of these data deviated by ±2 db or more from the average relation given in Eq On dense pavements an average 9.5 db difference was found between the CPX noise level measured with reference tyre H1 (Avon AV4) and the pass-by noise level from multi-axle heavy trucks measured at 1.2 m height. This implies a 12.0 db average difference between CPXH 80 and L AFmax from two-axle trucks, based on average observed differences between pass-by levels for two-axle and multi-axle trucks. The agreement between the CPX noise levels measured with reference tyre P1 (SRTT) and the pass-by noise levels was not as good. The average relationship between multi-axle truck pass-by noise levels and CPXH 80 is: Multi-axle trucks: L AFmax,80 = 0.65 CPXH db (9-2) When higher SPB microphone positions are used the average noise level decreases by approximately 0.5 db per meter the microphone height is increased. There was a trend for a larger change in noise level when increasing the height from 1.2 m to 2.4 m on EACC than on SMA. The reasons for this are not known. Dutch SPB data collected at different heights indicated little need to distinguish between different types of pavement (thin noise reducing asphalt layers, single layer or two-layer porous asphalt), see Figure 9-34 in Section 9.3. At 1.2 m microphone height very similar relationships were found for cars and heavy vehicles between the maximum vehicle pass-by noise level L AFmax and the sound exposure level L AE : Cars L AFmax - L AE = 2.79 ln (v) db (9-3) Trucks: L AFmax - L AE = 2.23 ln (v) db (9-4) where v is the vehicle speed in km/h. A comparison was made with noise levels predicted according to the Nordic prediction method for road traffic noise. Nord2000 predicted maximum noise levels slightly lower (0.2 db) than measured and at the same time predicted sound exposure levels slightly higher (0.9 db) than measured in the speed range around 110 Date: 30/04/2015 Version: Final report 39 (68)

40 km/h. Therefore the predicted difference L AFmax - L AE deviated by approximately 1 db from the measured differences in this speed range. This deviation, however, is not important to the ROSANNE project. The measurements carried out in the ROSANNE project show good correlation between maximum noise levels and sound exposure levels. This does not support a notion that more reliable noise emission values will be obtained by measuring sound exposure rather than maximum noise levels. The simulations mentioned in Section confirm that the sound exposure level is more sensitive than the maximum noise level to variations in the ground surface between vehicle and measuring microphone and that it is important to correct for any missing tails of the noise level time history. Such corrections are given in Section An attempt made to find out if surface texture or sound absorption data would be useful parameters in pavement classification gave discouraging results, perhaps because the effects of such parameters are too small or random compared to the large spread in data from different sources. The comparison of pavements, and hence their classification, may depend on the microphone height used for pass-by noise measurements. It is therefore essential that such measurements for pavement noise classification purposes are taken at a height which is representative of the noise exposure of people living/working close to the road. Date: 30/04/2015 Version: Final report 40 (68)

41 References [1] ISO :1997, Acoustics Measurement of the influence of road surfaces on traffic noise Part 1: Statistical Pass-By method [2] ISO/2 nd DIS : 2015, Acoustics Measurement of the influence of road surfaces on traffic noise Part 2: The close-proximity method [3] Manfred Haider, Ulf Sandberg, Noise classification methods for urban road surfaces. User manual: Measurement methods. SILENCE Deliverable F.D12 [4] ISO/PAS :2013, Acoustics Measurement of the influence of road surfaces on traffic noise Part 4: The Statistical Pass-By method using a backing board [5] Fabienne Anfosso-Lédée, Yves Pichaud, Tyre-road noise measurements. Correlation between CPB and CPX measurements, LCPC Report MENRT 03K228, May 2006 [6] Fabienne Anfosso-Lédée, F. Besnard, Simulations of propagation at various receiver positions. Influence of road surface type and discontinuity, Document N 173 presented to ISO/TC 43/SC 1/WG 33, 18 March 2005 [7] Fabienne Anfosso-Lédée, Modeling the local propagation effects of tire-road noise : propagation filter between CPX and CPB measurements, Proc. Internoise 2004 [8] SUPSIL = DRD R460, 2013 [9] ISO/WD for ISO/TS :2014, Acoustics Measurement of the influence of road surfaces on traffic noise Part 3: Reference tyres [10] Fabienne Anfosso-Lédée, personal communication to J. Kragh 13-Feb-14 [11] Reinhard Wehr, personal communication to J. Kragh 24-Sep-14 [12] Wolfram Bartolomaeus, personal communication to J. Kragh 2-Oct-14 [13] Anneleen Bergiers, personal communication to J. Kragh 1-Oct-14 [14] Fabienne Anfosso-Lédée, personal communication to J. Kragh 1-Oct-14 [15] Fabienne Anfosso-Lédée, personal communication to J. Kragh 4-Nov-14 [16] Matthew Muirhead, personal communication to J. Kragh 22-Sep-14 [17] Piotr Mioduszewski, personal communication to J. Kragh 4-Sep-14 [18] Truls Berge, personal communication to J. Kragh 12-Sep-14 [19] Ulf Sandberg et al. ROSANNE Deliverable D2.2: Report on temperature influence and possible corrections for measurement of noise properties of road surfaces (to be published) [20] Anneleen Bergiers, personal communication to J. Kragh 8-Dec-14 Date: 30/04/2015 Version: Final report 41 (68)

42 9 Appendix Processing of received data 9.1 General Data arrived in different formats and DRD adapted them to a common format. In this connection, data were normalised to the same air temperature and to the same reference speeds Temperature correction Some of the received data had been corrected for temperature influence. Such data were un-corrected applying the temperature coefficients originally used to correct them. These uncorrected data were then normalised to 20 C air temperature using the coefficients given in Table Speed correction Some data were speed corrected by DRD using the speed coefficient determined by IFSTTAR as described in sections 0 and Figure 9-1 illustrates the influence of the value of the speed coefficients found for dense and semi-porous asphalt, respectively. Within the ranges of reference speeds in the received data it makes little difference which speed coefficient is used. Figure 9-1: Speed dependence of pass-by noise levels from cars. Relative noise levels calculated with average speed coefficients found in IFSTTAR data base. The speed 110 km/h is used as a reference in the illustration Date: 30/04/2015 Version: Final report 42 (68)

43 Figure 9-2 and Figure 9-3 show the same grey line curve as Figure 9-1. Each coloured straight line illustrates the regression line slope from one SPB measurement made by AIT (Figure 9-2) or DRD (Figure 9-3). Each line connects the value of L = b log 10 (v) at the ends of the range of validity for the regression line, and each line has been positioned so that it has its midpoint on the grey IFSTTAR curve. Figure 9-2: Illustration of slopes found in AIT linear regression analyses vs. IFSTTAR average Figure 9-3: Illustration of slopes found in DRD linear regression analyses vs. IFSTTAR average Date: 30/04/2015 Version: Final report 43 (68)

44 9.1.3 Correction of measured L AE Figure 9-4 shows a screen dump from the DRD analyser as an example of the noise level time history during a car pass-by. The operator can select the time interval T to be analysed for determining the sound exposure level, L AE, see Figure 7-1, and the analyser delivers L AE and the number of db down from the maximum noise level at the cursor positions denoted Lower limit and Upper limit, resp. These numbers were used to correct for the missing tails of the noise level time history as described below. Figure 9-4: Screen dump from DRD analyser showing the noise level time history during a car pass-by at 101 km/h on SMA 11 The missing tails were simulated by means of software denoted SPL2000 and developed by Birger Plovsing, DELTA, to calculate according to the Nordic prediction method for road traffic noise, Nord2000. Calculations were made for various scenarios representing different sound propagation conditions: All hard ground (dense asphalt) between the centre line of the lane and the measurement position. Hard ground on the first part and grassland on the last 2.4 m of the propagation path, representing the situation at a typical Danish regional highway. Hard ground on the first part and grassland on the last 3.75 m of the propagation path, representing the maximum amount of the terrain surface allowed in the SPB standard [1] to be different to the pavement under test. Date: 30/04/2015 Version: Final report 44 (68)

45 SPL2000 delivers time histories of the instantaneous noise level, calculated as if the total sound power of the vehicle were concentrated at the same horizontal position on the road (but with sources at different heights). Figure 9-5: Time histories of instantaneous noise levels during a car pass-by calculated according to Nord2000 for different sound propagation situations, a) hard ground (dense asphalt); b) hard/2.4 m grassland; and c) hard/3,75 m grassland. Top: 5 m measurement height; Bottom: 1.2 m measurement height Date: 30/04/2015 Version: Final report 45 (68)

46 Figure 9-5 shows examples of calculated noise level time histories during a car pass-by. The top part of the figure shows the time history at a height of 5 m and the bottom part at a height of 1.2 m. At 5 m the time history is the same, no matter the terrain, while at 1.2 m the tails are attenuated more the more grassland there is along the propagation path. Such simulations were made for various speeds, both for cars and heavy trucks, and for measurement heights 1.2, 3.0, and 5.0 m. The noise levels time histories for trucks were slightly less sensitive to terrain differences than those for cars, reflecting the fact that truck noise sources tend to be higher up than the sources of car noise. Differences in speed did not have a significant effect. Figure 9-6 and Figure 9-7 show how much of the total sound exposure level is missing if the noise level time history is integrated over other angles than ± 90 around the maximum. If, for example, integration is made over ± 75 and the terrain is all hard, then 0.9 db of the sound exposure level is missed. If the terrain is partly grassland, less of the sound exposure is missing when integrating over less than ± 90. The results also emphasize the need for taking sound propagation effects into account when deriving vehicle sound power levels from measured sound pressure levels. Figure 9-6: Missing contributions to L AE when integrated over less than ± 90 as a function of the integration angle Figure 9-7 shows the same computation results as Figure 9-6 but in this case the missing contributions to the sound exposure level are shown as a function of the number of db down from the maximum instantaneous noise level. The figure also shows trend line equations used by DRD for correcting the measurement results. Date: 30/04/2015 Version: Final report 46 (68)

47 Figure 9-7: Missing contributions to L AE when integrated over less than ± 90 as a function of the number of db down from the maximum instantaneous noise level SPL2000 delivers an instantaneous noise level calculated using the Nord2000 source model, so supplementary corrections were needed to account for the difference between this maximum instantaneous noise level and the measured maximum noise level with time weighting F. Figure 9-8 shows this difference as a function of the vehicle speed, calculated by means of the expression in Eq which is valid for an omnidirectional point source. L AFmax = maximum noise level with time weighting F, [db] L inst = maximum instantaneous (peak) noise level, [db] a = (slant) distance perpendicularly to the road from source to microphone position = 6.6 m v = source speed [m/s] 3) T = integration time = 0.25 s Table 9-1 shows the resulting corrections. ΔL is the average db-down value before and after the maximum noise level and v is the vehicle speed in km/h. 3) Elsewhere in this report v is in km/h Date: 30/04/2015 Version: Final report 47 (68)

48 Figure 9-8: Calculated difference between instantaneous and time weighted maximum noise level as a function of the vehicle speed. Full line = calculation result. Dashed line = trend line Table 9-1: Corrections applied for the missing contributions from noise level time history tails Height [m] Hard Hard/2.4 m grass Hard/3.75 m grass ln( L) ( v ) e 0.171ΔL - ( v ) e 0.164ΔL - ( v ) e 0.159ΔL - ( v ) Measured differences L AFmax L AE Figure 9-9 shows the differences between the measured maximum noise level and sound exposure level at 1.2 m microphone height from 19 sites for cars and from 11 of these 19 sites for multi-axle vehicles. The noise levels from each individual vehicle were normalised to the average vehicle speed using the trend line in data from that site. The figure shows the average difference between L AFmax and L AE from each site and the error bars show 95 % confidence intervals of these differences for individual vehicle pass-bys. Results have been included from sites where data from 3 or more vehicle pass-bys were available. The difference is about the same for trucks and cars, and at 50 km/h L AFmax is approximately 3 db lower than L AE. This difference decreases with increasing speed to approximately zero at 120 km/h. Figure 9-9 also shows the difference as predicted according to Nord2000 which predicts equal values of L AE and L AFmax.at km/h. These differences have been corrected according to Equation 10-1 and Figure 9-8. A comparison of absolute noise levels from cars revealed a trend for Nord2000 to predict 0.2 db higher L AFmax on the average and 0.9 db lower L AE on the average than the measured noise levels. The latter differences may partly be due to the fact that the original measurement results behind the Nord2000 source model were not corrected for missing tails of the noise level time histories. Date: 30/04/2015 Version: Final report 48 (68)

49 Figure 9-9: Differences between measured maximum noise level and sound exposure level at 1.2 m microphone height. Data points show average differences and error bars show 95 % confidence intervals of the differences Figure 9-10 and Figure 9-11 shows the differences between maximum noise level and sound exposure level measured at 5 m and 3 m height, respectively. These results are based on fewer vehicle pass-bys than Figure 9-8. Figure 9-10: Differences between measured maximum noise level and sound exposure level at 5 m microphone height. Data points show average differences and error bars show 95 % confidence intervals of the differences. N is the number of pass-bys Date: 30/04/2015 Version: Final report 49 (68)

50 Figure 9-11: Differences between measured maximum noise level and sound exposure level at 3 m microphone height. Data points show average differences and error bars show 95 % confidence intervals of the differences. N is the number of pass-bys 9.2 Partners data Processing and identification In this section the received data are described in a little more detail than in Section AIT AIT data were delivered by Reinhard Wehr [1]. Processing AIT data CPX data had not been corrected for temperature effects and DRD corrected them according to Section SPB data had been corrected using db/ C for both SMA and EACC. DRD un-corrected and re-corrected them according to Section DRD decided at first to relate AIT CPXP 80 with SPB(120) because the average vehicle speeds were close to 120 km/h and thus data would be easy to compare with similar data from BASt. The pass-by noise levels were calculated as L veh = a + b log 10 (v/v ref ), where intercept a and slope b were given in the data and v ref = 50 km/h. Identification of AIT data in total data Figure 9-12 shows where AIT data points are in the total data set for cars from Figure 6-3, and Figure 9-13 shows the AIT data points for heavy vehicles in the data set from Figure Date: 30/04/2015 Version: Final report 50 (68)

51 Figure 9-12: Position of data points from AIT in Figure 6-3 Figure 9-13: Position of data points from AIT in Figure 6-13 Date: 30/04/2015 Version: Final report 51 (68)

52 9.2.2 BASt BASt data were delivered by Wolfram Bartolomaeus [12]. Processing BASt data CPX data from BASt had not been corrected for temperature effects and DRD corrected them according to Section SPB data for dense surfaces (asphalt and cement concrete) had been corrected using db/ C for cars and db/ C for trucks. For open porous asphalt data had been corrected by db/ C for cars and db/ C for trucks. For cars DRD used SPB data from EACC directly as delivered. SPB data from SMA and porous asphalt were un-corrected and re-corrected by DRD according to Section All SPB data for trucks were un-corrected and re-corrected by DRD according to Section Data from 7 SMA pavements had been measured by BASt at reference speeds between 70 km/h and 100 km/h. These data were corrected to reference speed v ref = 80 km/h as L veh = a + b log 10 (v ref ), using intercept a and slope b given in the data. Identification of BASt data in total data Figure 9-14 shows the BAST data points in Figure 6-1. Figure 9-15 shows the BAST data points in Figure 6-3. Figure 9-16 shows the BAST data points for heavy trucks in Figure 6-11, and Figure 9-17 shows the BAST data points for heavy trucks in Figure Figure 9-14: Position of data points from BASt in Figure 6-1 Date: 30/04/2015 Version: Final report 52 (68)

53 Figure 9-15: Position of data points from BASt in Figure 6-3 Figure 9-16: Position of data points from BASt in Figure 6-11 Date: 30/04/2015 Version: Final report 53 (68)

54 Figure 9-17: Position of data points from BAST in Figure BRRC BRRC data were delivered by Anneleen Bergiers [13]. Ten sections of road having different types of pavement had been monitored for two years, and measurements had been repeated several times each year. At each test section of road BRRC used three positions for SPB measurements. CPX noise levels were given for the 20 m segment nearest to the SPB measurement position or in some cases the average was given from two consecutive 20 m segments. Processing of BRRC data BRRC results had not been corrected for temperature and DRD normalised the results 20 C air temperature using db/ C for all types of pavement. An initial data analysis was made and the results from three sections are illustrated in Figure Figure The figures show the measurement results as a function of time. The left y-axes show the values of the SPB noise levels while the right y-axes show the values of the CPX noise levels. Figure 9-18 shows measurement results from a reference section of road with SMA 10. The noise levels were fairly constant over time. Figure 9-19 shows results from a section of road with a wearing course having 0/8 mm aggregates; no other information on the pavement is available. At this section of road the CPX noise levels gradually increased while the trend for SPB results was for more constant noise levels over time but with large differences between results from different positions measured at approximately the same time. Figure 9-20 shows results from a section or road having a 30 mm wearing course with 0/2 mm and 0/4 mm Date: 30/04/2015 Version: Final report 54 (68)

55 aggregates. At this section of road the CPX noise levels also increased gradually with time but with larger spread than in Figure The trend for SPB results was for rather constant noise levels over time but with large differences between noise levels measured in different positions at approximately the same time, particularly in month No. 26. DRD determined trend lines in BRRC data and trend line values at pavement ages two months and 26 months, respectively. These values are shown in Figure 9-21 together with the remainder of ROSANNE data measured at the same reference speed (see Figure 6-2). Many, though not all, BRRC data points for the same CPX noise level are at lower SPB noise levels than the remainder of data, and DRD considers them outliers as explained below. A BRRC site is shown in Figure There is a partly grass-covered central area between the road and the SPB microphone which DRD suspects to have influenced the sound propagation by adding extra ground attenuation. Calculations made using SPL2000 for the geometry in Figure 9-23 yielded the results in Table 9-2. The calculated extra attenuation of L AFmax provided by the grass surface in excess of that provided by asphalt is almost 2 db and the sound exposure level is affected even more. On the shown measurement site and on other sites, the ground requirements in ISO [1] are violated since there is another surface than pavement along the first part of the propagation path [20]. Therefore DRD omitted BRRC data from the analyses of the relationship between CPX and SPB noise levels. Figure 9-18: SPB noise levels from cars and CPXP 80 as a function of time on BRRC road section No. 1, SMA 10. Data point labels x.yz indicate x = section No.; y = position No.; z = measurement organisation and/or personnel Date: 30/04/2015 Version: Final report 55 (68)

56 Figure 9-19: SPB noise levels from cars and CPXP 80 as a function of time. BRRC road section No. 10 with 0/8 mm aggregates. Data point labels x.yz indicate x = section No.; y = position No.; z = measurement organisation and/or personnel Figure 9-20: SPB noise levels from cars and CPXP 80 as a function of time. BRRC road section No. 3 with a 30 mm surface layer having 0/2 mm and 0/4 mm aggregates. Data point labels x.yz indicate x = section No.; y = position No.; z = measurement organisation and/or personnel Date: 30/04/2015 Version: Final report 56 (68)

57 Figure 9-21: BRRC data points from road sections #1 - #10 together with the remainder of ROSANNE data from Figure 6-2. Note: Axes differ from Figure 6-2 Figure 9-22: Typical site in BRRC measurement series having a partly grass covered central area Date: 30/04/2015 Version: Final report 57 (68)

58 Figure 9-23: Cross section showing the geometry used in SPL2000 calculations Table 9-2: Results of Nord2000 calculations without and with a grass covered central area Terrain SPB 80 [db] AE Amax All hard With 3 m grass in centre Difference DRD Processing of DRD data DRD CPX data had not been corrected for temperature effects and DRD corrected them according to Section DRD SPB data for both cars and trucks had been corrected using db/ C. SPB data for cars were un-corrected and re-corrected according to Section SPB data for trucks had already been corrected according to Section and were used directly. At six sites SPB noise levels from cars had been measured at 110 km/h reference speed. These SPB data from five sections were corrected to 120 km/h as L veh = a + b log 10 (v ref ), using intercept a and slope b given in the data. For one site the reference speed 120 was outside the range of validity and the correction to 120 was made by means of the expression L veh,120 = L veh, log 10 (120/110) using the speed coefficient found by IFSTTAR [15]. Identification of DRD data in total data Figure 9-24 shows the DRD data points in Figure 6-2, Figure 9-25 shows the DRD data points in Figure 6-3 and Figure 9-26 shows the DRD data points in Figure Date: 30/04/2015 Version: Final report 58 (68)

59 Figure 9-24: Position of data points from DRD in Figure 6-2. Note: axes differ from Figure 6-2 Figure 9-25: Position of data points from DRD in Figure 6-3 Date: 30/04/2015 Version: Final report 59 (68)

60 Figure 9-26: Position of data points from DRD in Figure IFSTTAR CPX-SPB relation IFSTTAR uses Michelin tyres for CPX noise measurements and no information was available on the relation between these CPX FR noise levels and CPXP v so it was decided not to include IFSTTAR data in the analyses of CPX-SPB relations. Table 9-3 shows CPX FR and controlled pass-by (CPB) noise levels at 90 km/h measured by IFSTTAR in two projects, PREDIT and DEUFRAKO, each comprising 12 road surfaces [14]. In PREDIT the overall average difference CPX-CPB was 22.2 db (std. dev. 1.4 db) while for three porous pavements the average 23.5 db difference (std. dev. 0.3 db) was slightly larger than the average 21.8 db for the remainder of pavements. In DEUFRAKO, however, the average 22.1 db difference at two porous pavements did not deviate from the average 21.9 db (std. dev. 0.5 db) for all pavements. These differences are between CPX and vehicle pass-by noise levels for the same tyre line and may not necessarily be the same as the difference between CPXP v and the SPB noise level from cars in normal traffic. Date: 30/04/2015 Version: Final report 60 (68)

61 Table 9-3: Overview of differences ΔL between CPX FR and CPB noise levels in IFSTTAR measurement series [14] SIMULTANEOUS CPB - CPX MEASUREMENTS (CPB with CPX test vehicle) Vref = 90 km/h Pourous pavements indicated in grey cases MEASUREMENTS "PREDIT" reml romds + 2 PesP PrMcks Test tyre Michelin Energy XI1 size R15 Type pavement CPX ISO CPB DL (db(a)) ECF 0/6 102,8 79,5 23,2 BBTM 0/10 cl1 99,6 77,1 22,4 BBSD 0/10 100,6 78,5 22,1 ES MCSD 6/10 105,3 81,7 23,6 BB5r 0/10 99,5 76,2 23,3 BB5R 6/10 0/14 98,5 74,7 23,8 ES MC5D 10/14 4/6 104,7 85,4 19,3 ES MCSD 6/10 103,4 82,2 21,2 ECF MC 0/2 4/6 100,8 78,9 21,9 BBSD 0/10 96,8 76,9 19,9 BBSD 0/10 (test track) 100,0 77,5 22,4 BB5R 0/10 (test track) 97,2 73,9 23,3 MEASUREMENTS "5EUFRAKO" LCPC Test track Test Tyre Michelin Energy E3A size R15 Pavement Section T air ( C) T road surface ( C) T tyre surface ( C) tyre hardness (shore A) CPB L Amax (90 km/h) (db(a)) CPX L 20m (90 km/h) (db(a)) DL (db(a)) PAF 0C6 A' 1E,5 36,3 3E, ,4 E5,5 22,1 VTAF 0C6 M2 20,5 36,5 38,E 63 73,4 E5,4 22 DAF 0C10 (ner) E (1) 23,5 53B4 44,E 61 73,4 E5,5 22,1 FlexiNle AspOMlP G 17 2E,2 34, ,6 E6,7 22,1 SBDB 0B8C1B5 F 21 40,8 38, E6,E 21,9 SMnd AspOMlP 0C4 I ,8 43, ,1 E6,6 21,5 DAF 0C10 (old) E (2) 17 28B7 34, ,8 EE 22,2 Folgrip F 16,5 25,5 33,5 61,5 77,1 EE,E 22,8 VTAF 0C10 M , ,5 EE 21,5 FemenP FoncrePe ,1 32,5 61,5 77,5 E8,3 20,8 Resin Epoxy I1 1E 37,E 38, ,6 EE,8 22,2 SBDB 8C10 A 23 45, ,2 101,5 21,3 Table 9-4 displays the result of a statistical analysis of slope coefficients in the IFSTTAR database [15], see also Section For each type of pavement the table gives the average slope of the regression line of pass-by noise level against the logarithm of the vehicle speed for light and heavy vehicles, respectively. The number of measurements on which each average is based is given in italics. Date: 30/04/2015 Version: Final report 61 (68)

62 Table 9-4: Slopes of regression lines. Pass-by noise level against vehicle speed in IFSTTAR data base [15] POROUS SEMI-POROUS THIN LAYERS Light Vehicles (LV) Heavy Trucks (TR) type nb_meas SlopeLV nb_meas Slope TR BBDr 0/ , ,1 BBDr 0/ ,5 BBDr 0/ , ,3 AVG POROUS , ,5 BBTM 0/10-t ,9 6 32,3 BBTM 0/6-t , ,0 BBTM 0/4 2 20,7 AVG SEMI POROUS 83 28, ,4 NON POROUS THIN LAYERS AND DENSE ASPHALT SURF. DR. BBM 0/ ,4 6 24,2 BBUM 25 29,4 9 26,2 BBTM 0/8-t1 7 28,4 6 34,0 BBTM 0/10-t , ,6 BBTM 0/ , ,8 BBTM 0/6-t , ,5 BBSG 0/ , ,4 BBSG 0/ , ,4 ECF 20 32,7 6 29,1 ES 55 34,2 7 29,2 ES 10/ ,5 CEMENT C. BC 11 32,9 7 27,7 AVG DENSE , ,0 Light Vehicles (LV) Heavy Trucks (TR) type nb_meas SlopeLV (dba nb_meas Slope TR AVG POROUS , ,5 AVG SEMI POROUS 83 28, ,4 AVG DENSE , ,0 Influence of terrain between vehicle and microphone position IFSTTAR has investigated the influence of the terrain surface between vehicle and measurement position on the measured pass-by noise level by modelling and in field Date: 30/04/2015 Version: Final report 62 (68)

63 experiments [5] - [7] for various combinations of ground surface such as those illustrated in Figure 9-27: dense pavement hard ground dense pavement porous ground porous pavement hard ground porous pavement porous ground Figure 9-27: Examples of combinations of ground surfaces in IFSTTAR simulations and field measurements [7] Based on these investigations IFSTTAR concluded that when the first 2.5 m of ground below the propagation path is pavement then the remainder of the terrain may be grass-covered without detrimental effect on the pass-by noise level. This would imply a. o. that the present 3.75 m requirement in ISO is unnecessarily strict. However, the BRRC measurement results shown in Figure 9-21, from sites where the distance from the vehicle centre line to the edge of a strip of partly grass-covered ground was m, indicate that already at this distance a ground impedance shift may cause substantial extra attenuation. This is supported by Figure 9-28 which shows the pass-by noise level at 7.5 m distance from a car driving 80 km/h calculated by means of SPL2000 with the geometry illustrated in the figure. With first 2.5 m pavement and then 5 m grass-covered ground between the vehicle centre line and the microphone position, the calculated maximum noise level with time weighting F is 1.3 db lower than if all of the terrain had been pavement. Date: 30/04/2015 Version: Final report 63 (68)

64 Figure 9-28: Cross section illustrating the geometry used in SPL2000 calculations and calculated pass-by noise level from a car driving 80 km/h as a function of the amount of grasscovered terrain between the microphone position and the vehicle centre line Another conclusion worth mentioning from IFSTTAR investigations is that pavement classification depends on the selected microphone height and thus a measurement position representative of the conditions at road neighbours should be selected [6]. This is in line with findings in the ROSANNE data, see Section TRL TRL data on CPXP 80 and pass-by noise levels from cars at 110 km/h and medium and heavy trucks at 85 km/h on thin asphalt layers were supplied by Matthew Muirhead [16]. Processing TRL data Pass-by noise levels from cars and trucks had been corrected by TRL to 20 C using the correction 0.03 ((0.7 T s +T a )/2-20) where T s is the surface temperature and T a the air temperature. These data were un-corrected and re-corrected by DRD according to Section CPX data were presumably un-corrected but no information was available on the temperature conditions during the measurements. DRD used the data as received. SPB noise levels from cars had been measured at 110 km/h reference speed. These SPB data were corrected to 120 km/h as L veh_120 = L veh_ log 10 (120/110) using the speed coefficient found in IFSTTAR data. Pass-by noise levels from trucks measured at reference speed 85 km/h were converted to 80 km/h L veh_80 = L veh_ log 10 (80/85) using the speed coefficient found in IFSTTAR data. Identification of TRL data in total data Figure 9-29 and Figure 9-30 show the position of TRL data in in Figure 6-3 and Figure Date: 30/04/2015 Version: Final report 64 (68)

65 Figure 9-29: Position of data points from TRL in Figure 6-3 Figure 9-30: Position of data points from TRL in Figure 6-13 Date: 30/04/2015 Version: Final report 65 (68)

66 9.2.7 SINTEF SINTEF data were delivered by Truls Berge [17], mostly from roads with dense asphalt: SRTT and Avon AV4 noise levels measured in the right wheel track at 80 km/h and SPB noise levels from cars (on 14 roads) and trucks (multi-axle two-axle trucks on 8 roads). Processing SINTEF data SPB data from SINTEF had not been corrected for temperature and DRD corrected them according to Section CPX data had been corrected by db/ C on dense and db/ C on porous pavements. In three cases data contained information on the air temperature and DRD uncorrected and re-corrected them according to Section Data from the remaining sites did not include temperatures, and these CPX data were used as received. All data could be derived at 80 km/reference speed and no speed corrections were made. Identification of SINTEF data in total data Figure Figure 9-33 show the position of SINTEF data in Figure 6-2, Figure 6-11 and Figure Figure 9-31: Position of data points from SINTEF in Figure 6-2 Date: 30/04/2015 Version: Final report 66 (68)

67 Figure 9-32: Position of data points from SINTEF in Figure 6-11 Figure 9-33: Position of data points from SINTEF in Figure 6-13 Date: 30/04/2015 Version: Final report 67 (68)

68 9.3 Dutch data on SPB noise levels at different heights Figure 9-34 shows passby noise levels measured at 1.2 and 5 m height as a function of the pass-by noise level measured at a height of 3 m on a number of Dutch roads having different types of noise reducing pavements [8]. The upper charts show noise levels from multiaxle heavy vehicles, lower charts from cars. The data indicate little need for distinguishing between single-layer porous (ZOAB(+) and two-layer porous asphalt (ZOAB-TL) on one hand, and noise reducing thin asphalt layers (DGD) on the other. Figure 9-34: Pass-by noise levels at 5 m and 1.2 m as a function of the noise level measured at 3 m. Top: Multi-axle trucks; Bottom: Cars Date: 30/04/2015 Version: Final report 68 (68)