Low noise surfaces for urban streets

Transcription

1 EUROPEAN COMMISSION DG RESEARCH SIXTH FRAMEWORK PROGRAMME PRIORITY 6 SUSTAINABLE DEVELOPMENT, GLOBAL CHANGE & ECOSYSTEMS INTEGRATED PROJECT CONTRACT No Low-noise surfaces for urban roads and streets Deliverable no. F.D3 Dissemination level Public Work Package WP F.1 New production Technologies for surfaces on urban streets Author(s) Ulf Sandberg, Bent Andersen, Hans Bendtsen, Björn Kalman Co-author(s) Status (F: final, D: draft) F File Name Project Start Date and Duration 01 February 2005, 36 months Ulf Sandberg, Bent Andersen, Hans Bendtsen, Björn Kalman

2 II Preface This report is produced by the Swedish Road and Transport Research Institute (VTI) and the Danish Road Institute/Road Directorate (DRI) as a part of the work carried out by the Forum of European Highway Research Laboratories (FEHRL) in the EU project SILENCE which started in February 2005 and ends in May The report is part of the work carried out in Working Package (WP) F.1 New production technologies for surfaces on urban streets. The following organisations took part in this F.1 Work Package: Swedish National Road and Transport Research Institute (VTI) Danish Road Institute/Road Directorate (DRI) Skanska Contractors, Sweden. The major objective of this part of the SILENCE project was to develop and test concepts for low-noise pavements for urban roads and streets. This report presents the results of this work, with the exception of work reported already in Deliverables F.D1 and F.D2. Unless otherwise stated the pictures and diagrams in this report are made by the authors. The authors Ulf Sandberg, Bent Andersen, Hans Bendtsen, Björn Kalman

3 III TABLE OF CONTENTS SUMMARY V 1 Introduction and background 1 2 Objectives 2 3 Briefly about noise measurement methods 2 4 Thin layers Introduction Danish experiments Thin layer pavement tested in Malmö City Introduction Test section location and construction SPB measurements by DRI Test surfaces Results General remarks Results SPB measurements before repaving Results SPB measurements after repaving Results SPB measurements one year after repaving Results - Noise reduction Results - CPX measurements by M+P/Carl Bro Results - CPX measurements by VTI / TUG within SILENCE Results - Annoyance study Conclusions regarding the Malmö experiment Overall conclusions for thin layers 24 5 Surface dressings Introduction Some earlier data Study of traditional surface dressings used on low-volume highways Development and testing in Denmark of less noisy surface dressings The Fakse experiment The Danish SILENCE experiment Surface dressings tested Noise measurements SPB and CPB noise results CPX noise levels CPX versus SPB spectra Conclusions with regard to the Danish surface dressing experiment Surface dressings with very small aggregates 52 6 Block pavements 54 7 Experiments with a poroelastic cover on a block surface Introduction Problems under investigation Experiments in NR2C and SILENCE - Description 57 Ulf Sandberg, Bent Andersen, Hans Bendtsen, Björn Kalman

4 IV 7.4 Access to suitable material Results - Polishing effect on wet skid resistance Results Stability of block pavement Conclusions More to read 65 8 Conclusions 68 9 Suggestions for implementation References 71 Ulf Sandberg, Bent Andersen, Hans Bendtsen, Björn Kalman

5 V SUMMARY As part of Work Package WP F1 of the SILENCE project, new pavement concepts have been developed and some others have been refined. The work has included the following topics which are dealt with in this report: Noise from different kinds of concrete or stone block pavements has been tested - including an optimized type. This especially considered the development of improved surfacing systems for cultural areas in European cities and towns. This work has primarily been reported in SILENCE Deliverable F.D1 Noise generated at discontinuities like rail crossings has been investigated and design rules have been proposed. This work has primarily been reported in SILENCE Deliverable F.D2 Testing a new type of thin layer surface in the city of Malmö in Sweden. Noise characteristics of different types of surface dressings, with a focus on developing surface dressings giving reduced tyre/road noise emissions Block pavement with a soft cover (so-called poroelastic cover, or "Softbloc") All the above topics have been supported by experimental work in the project. The objectives of WP F1 have been to develop new low-noise surfacing techniques and processes for use on low-speed roads in urban areas; especially areas suffering from frequent intrusions. Earlier Danish results have shown that for newly laid thin layers the noise reduction achieved in relation to a DAC 0/11 (Dense Asphalt Concrete with 11 mm aggregate) or SMA 0/11 reference is around 3 db for passenger cars. After two years this is reduced to 1-2 db. For multi-axle trucks (in the highway case at least) there is a constant noise reduction over two years of around 2 db for the best performing thin layers (AC8o, SMA8+ and UTLAC8). In order to improve the noise reduction, a new experiment was conducted in Denmark. Noise measurements were carried out when the pavements were 1-2 months of age. The best noise reduction, 4.7 db for passenger cars, was measured on an open-graded asphalt concrete with 6 mm aggregate (AC6o); also in this case with a DAC 0/11 as reference surface. Another experiment was conducted on a residential street in the city of Malmö in Sweden, which carries also some through traffic. This location is extra critical since a large proportion of tyres in Sweden in winter-time are equipped with steel studs which are very tough against pavements with small aggregates. Two old wearing courses of dense asphalt concrete with 11 and 16 mm maximum aggregate size were substituted by two new stone mastic asphalt wearing courses - one with 11 mm maximum aggregate size used as a reference surface and the other with 8 mm maximum aggregate size being optimized for noise reduction. The initial noise reduction, measured with the SPB method, of the low-noise-optimized SMA 8 was very good, being almost 7 db compared to the old 16 mm wearing course, and 3.3 db compared to the new reference SMA 11. However, after one year (one winter with traffic using studded tyres) this noise reduction has more or less vanished according to the SPB measurements. The difference between the noise levels at the two new SMA surfaces was reduced to 0.5 db. However, one CPX measurement 8 months after the surface was laid showed that the noise reduction was still 3 db and a second CPX measurement (with other crew and equipment) 14 months after laying showed a remaining noise reduction of 1-2 db. An annoyance study made before and after repaving indicated very positive results in terms of reduced annoyance among the residents along the street with the new noise-optimized SMA surface. Ulf Sandberg, Bent Andersen, Hans Bendtsen, Björn Kalman

6 VI Swedish tests of surface dressings of traditional type indicated that the range of noise for the tested surfaces was 3 db for light traffic and 2 db for heavy traffic. For light traffic the average noise level of the surface dressings was approximately the same as for the reference surfaces (DAC or SMA 0/16), but for heavy traffic the surface dressings are quieter. The surface dressings which are the quietest for light traffic are the noisiest for heavy traffic, and vice versa. For composite traffic of equal (noise) weight this means that the difference between the surfaces is minimal. In general, the quietest surfaces for light traffic are the ones with the smallest chippings, while the contrary is the case for heavy traffic. Danish tests showed that at 80 km/h a small noise reduction was achieved for passenger cars of 1-2 db for two surface dressings with small chippings (2-5 mm, in the top layer). SPB measurement results for multi-axle trucks/busses indicated a noise reduction of up to 1.7 db at 80 km/h. These values for surface dressings are valid only when using a DAC or SMA 0/16 as the reference, for a reference of DAC 0/11 the mentioned noise reductions have to be lowered by 1.0 to 1.5 db in order to reflect the situation in relation to a new DAC 0/11 reference. This is also the case for the noise reductions for heavy vehicles. Generally, the surface dressings with the larger 11 mm aggregate give the highest overall noise level and the surface dressings with the smaller 5 mm aggregate give the lowest overall noise levels. It can thus be concluded that only minor noise reductions are achieved with small-aggregate surface dressings in relation to a DAC 0/11 pavement and that large-aggregate surface dressings increase noise for light traffic but decreases it marginally for heavy traffic. It was not possible to study extra-small-aggregate surface dressings of the type used to increase skid resistance at critical spots in urban areas and often bound with epoxy or other advanced binders. However, it is pointed out that such surface dressings have a potential of providing substantial noise reductions if laid in an optimized way. It is clear that there are substantial differences between existing block pavements. The range between existing ones would be at least db and if the futuristic interlocking block pavement with poroelastic cover is included it will be at least db. These are dramatic differences. In cases where one can accept to surrender some of the historic/cultural values, it would be possible to replace the older type of stone setts with visually rather equal blocks with a flatter surface; such as the flat granite setts tried in Copenhagen and reported here. Depending on the old type of surface, and the orientation of the blocks, one may gain some 4-8 db in this way. This type of surface would not be noisier than an ordinary asphalt surface, and it may then be preferred to use this one rather than to pave the street with asphalt, and in this way save some of the cultural value of the street. In general, the orientation of the blocks should be such that the tyres roll over them at an angle of o rather than hitting the blocks and the joints at right angles. The use of interlocking blocks would generally not generate more traffic noise than an ordinary asphalt surface. With an appropriate design and good workmanship in the laying, as well as a stable sand bedding or other base material, one may even obtain a certain noise reduction compared to a dense asphalt surface; say about 1-2 db. The project included some development and testing for improving poroelastic road surfaces; in particular when laid as a soft cover on interlocking blocks. Three out of five tested poroelastic materials had poor wear resistance with respect to the polishing effect of tyres. But the two remaining materials had very good wear resistance and also showed excellent wet frictional properties throughout the polishing test. One "home-made" poroelastic material with addition of grains of silicon carbide material, together with a soft one-component prepolymerized polyurethane binder, showed a great potential for being a durable and safe road surface. Ulf Sandberg, Bent Andersen, Hans Bendtsen, Björn Kalman

7 VII Stability tests of a block pavement with a cover of poroelastic material, where blocks were laid either in a sand bedding or in a cementitious screed layer, showed that the sand bedding layer had problems with the stability of the blocks, whereas the cementitious screed layer had acceptable performance. It is believed that this type of bedding layer for a SOFTBLOC pavement would be durable. It seems probable that the work in this part of the project has solved two of the major problems connected with poroelastic road surfaces; durable wet friction and a more stable bedding for a block pavement with poroelastic cover. Ulf Sandberg, Bent Andersen, Hans Bendtsen, Björn Kalman

8 1 1 Introduction and background In Work Package WP F1 of the SILENCE project new pavement concepts have been developed and some others have been refined. As far as possible, laboratory work has been supplemented with testing in full scale on urban roads in different European cities. The work has included the following topics: Noise from different kinds of concrete or stone block pavements has been tested - including an optimized type. This work has been reported in SILENCE Deliverable F.D1 [Sandberg & Bendtsen, 2007] Noise generated at discontinuities like rail crossings has also been investigated and design rules have been proposed. This work has been reported in SILENCE Deliverable F.D2 [Bendtsen et al, 2007] Testing a thin layer surface in the city of Malmö. Noise characteristics of different types of surface dressings, with a focus on developing surface dressings giving reduced tyre/road noise emissions Block pavement with a soft cover (so-called poroelastic cover, or "Softbloc") Noise characteristics of road shoulder surfaces with finer textures The effect on noise of grinding-off the peaks of road surface macrotexture It should also be mentioned that apart from the above activities, related work on low-noise road surfaces has been conducted within WP F2 and WP F5 (Testing of novel new road surfacing materials). These are focused on medium and high-speed urban roads. Within F2, DRI has developed improved thin layers, and VTI has developed asphalt rubber surfaces. Testing of these has been conducted within F5. These activities are reported in Deliverables within WP F2 and WP F5.

9 2 2 Objectives The objectives of WP F1 have been to develop new low-noise surfacing techniques and processes for use on low-speed roads in urban areas; especially areas suffering from frequent intrusions. This WP has also considered the development of improved surfacing systems for cultural areas in European cities and towns. 3 Briefly about noise measurement methods In field experiments carried out in this part of the SILENCE project, basically either the Statistical Pass-By (SPB) or the Close-Proximity (CPX) measurement methods have been used. The SPB measurements were carried out according to ISO :1997. In order to determine the noise levels, L veh, for each vehicle category, regression analyses were carried out on the maximum noise level as a function of the logarithm of the speed (v): L veh = A + B log(v) Prior to these analyses, measured noise levels were normalized for the deviation of the air temperature (T) from 20 C according to Eq. (1), taken from [HARMONOISE, 2004]. L T T corrected = L correction, cars measured = T correction ( T 20) 0.03 ( 20) correction, dual+ mulit axle = T The air temperature was measured with a radiation-shielded electronic thermometer 1 m above the asphalt, while the road temperature was measured in the wheel tracks with an IR thermometer approximately every 30 minutes. The vehicle speed was measured with radar, and the reading was corrected according to Eq. (2) for the angle α between the direction of driving and the direction of the radar beam. vmeasured v corrected = (2) cosα The CPX measurements were carried out according to ISO/DIS :2000. This was and is a non-official draft standard which is still subject of improvement. The measurements were (if nothing else is stated) made by the Technical University of Gdansk (TUG) in Poland using their CPX trailer. Test tyres were the ASTM SRTT tyre (new 16" version) and the BFGoodrich MudTerrain T/A tyre (215/75R15). These tyres have been found by ISO/TC 43/SC 1/WG 33 to be main candidates for new reference tyres since they have showed the best correlation with full tyre/road or vehicle noise measurements (ASTM with light traffic and MudTerrain with heavy traffic), and the ASTM SRTT is already preliminarily decided to be one of the two reference tyres in the future standard. There has been no temperature correction applied to the CPX measurements. Results are always normalized to one of the speeds 30, 50, 80 or 110 km/h. (1)

10 3 4 Thin layers 4.1 Introduction Porous pavements are known to have a noise reducing potential and this family of pavements is already widely used in many, if not most European countries. For example, on the Dutch national highways in 2007 more than 70 % was paved with porous asphalt. In Denmark, porous pavements have been tested but these pavements have so far only been used on an experimental level in a few occasions. The same is valid for Sweden. Porous pavements can be complicated to maintain in the winter period and their durability can be problematic; especially when studded tyres are used. There is therefore a need to develop and test other types of noise-reducing pavements. In the last 5 years or so, thin layers optimized for noise reduction have been in focus in European countries such as France, the U.K, the Netherlands and Denmark. 4.2 Danish experiments In the Danish-Dutch research cooperation designated the DRI-DWW Noise Abatement Programme an important part was to develop and test optimized noise reducing thin layer pavements [DRI 159, 2007]. It was not a part of SILENCE but its results are of interest here. The DRI-DWW Noise Abatement Programme focused specifically on the development of optimised mixes of thin, semi-dense pavements and of thin porous layers. The programme also focused on the monitoring of test sections to achieve as long a time series of measurements as possible within the time framework of the program (which finished by the end of 2007). The main objectives were to analyse the long-term durability of noise reductions as well as the structural performance over time. Full scale testing of thin layers optimized for noise reduction has been carried out in Denmark on a total of four different test sections; two on urban roads and two on highways. They have not been exposed to cars with studded tyres as studded winter tyres are not used in Denmark. In order to reduce both the vibration-generated noise in the frequency region from 500 to 1500 Hz and the noise from air pumping above 1000 Hz the following criteria for the mix design have been used: 1. Maximum aggregate size 6 to 8 mm. 2. Sharp edges of the aggregate. 3. Uniform shape of the aggregate. 4. A steep aggregate size distribution with filler and very small aggregates combined with larger aggregates, and without aggregates of the sizes in-between. 5. Use of a small proportion of oversize aggregate. 6. High built-in air voids. 7. Good compaction of the pavement. In the four Danish experiments related to SILENCE the following pavement types have been optimized [DRI 159, 2007]: Open graded asphalt concrete (ACo). Stone Mastic Asphalt (SMA). A thin layer constructed as an Ultra Thin Layer Asphalt Concrete (UTLAC). In order to achieve further noise reduction, also thin semi-porous pavements (called BBTM) have been included in the latest project on highway M64 near Herning. Testing has also included some reference surfaces of conventional types.

11 Early results have shown that when the thin layers are new on a highway, as well as on urban roads, the noise reduction achieved in relation to a DAC11 reference (Dense Asphalt Concrete with 11 mm aggregate) is around 3 db for passenger cars. After two years this is reduced to only 1 db for the highway experiment and nearly 2 db for urban roads. There is a tendency from the highway experiment that AC8o and SMA8+ have the best noise reduction for passenger cars. For multi-axle trucks in the highway case there is a constant noise reduction over two years of around 2 db for the best performing thin layers (AC8o, SMA8+ and UTLAC8). In order to improve the noise reduction, a new experiment was started on highway M64 near Herning in 2006 [DRI 159, 2007]. Noise measurements were carried out 1-2 months after laying. The best noise reduction, 4.7 db for passenger cars, was measured on a opengraded asphalt concrete with 6 mm aggregate (AC6o). There is a reduction of the vibration noise because of the use of small aggregate and in air pumping noise because of the open structure of the pavements. For multi-axle trucks the best performing pavement, with 4.0 db noise reduction, was an SMA8 with 8 mm maximum aggregate size. Also for multi-axle trucks the vibration noise as well as the air pumping noise was reduced. The reduction of the aggregate size from 8 to 6 mm reduces the vibration noise for passenger cars whereas for multi-axle trucks there is a tendency that pavements with 8 mm aggregate have less vibration noise than pavements with 6 mm aggregate. In one of the SMA6 pavements a certain amount of larger aggregate (size 5-8 mm) has been added, in order to increase the openness of the surface structure (this improved type is designated SMA6+). This has a positive effect that reduces the air pumping noise for passenger cars, without significantly increasing vibration-generated noise. For multi-axle trucks this reduces both the vibration noise as well as the air pumping noise. If a dense asphalt concrete with 16 mm aggregate (DAC16) had been used as a reference on the two highway experiments, as is the case in the Netherlands, the noise reductions would have been more comparable to reductions obtained in the Netherlands. In the M10 experiment the best initial reductions for passenger cars would have been 4.2 db, decreasing to 2.5 db after two years. For the new M64 experiment the initial noise reduction for passenger cars for the best thin layer would have been 6.0 db, which is only marginally less than in the Dutch experiments Thin layer pavement tested in Malmö City Introduction Thin layers are not considered as very interesting by the Swedish Road Administration, due to the expected poor durability against the wear of studded tyres. Such tyres are not used in Denmark and therefore thin layers have been of much greater interest there. Nevertheless, a thin layer was tried in Malmö city in Sweden within this program, since in this part of Sweden the proportion of tyres having studs is relatively low. The location of the trial was a residential street with significant through-traffic having a posted speed of 50 km/h. At such low speed it is expected that thin layers with their small aggregates will be reasonably durable also against some studded tyre traffic. An SMA-type of surface, designated TA-3 and optimized for low noise by Skanska AB, as well as a reference pavement were constructed and laid on a suburban road in Malmö 28 August It resembles an SMA 8 surface. The surfaces and test sections are further described later in this sub-chapter. Statistical Pass-by (SPB) measurements were carried out in June 2006 before the new surfaces were laid, and again in September 2006 when the surfaces were one month old. The SPB measurements were repeated in September 2007

12 after one year of traffic exposure. During the winter months a substantial proportion of the cars on this street were using studded tyres Test section location and construction Ellenborgsvägen is a relatively small municipal street in south-eastern Malmö, Sweden. The annual daily traffic is between 2500 and 3800 vehicles. Approximately 10 % is heavy vehicles (mostly busses) and the speed limit is 50 km/h. The total length of the test section is approximately 950 meters. See Fig. 1. The first 380 meters (starting from Stenkällevägen ) has a noise-reducing pavement and the following reference section, ending at Husie kyrkoväg, is 570 meters long. The street was built in 1964 and the original wearing course was a conventional ABT11 (dense asphalt concrete, DAC 0/11). The pavement was resurfaced at least once since 1964 but no major maintenance had been done, therefore the pavement was severely cracked and had a lot of potholes, repairs and other damages. Figure 1: Aerial photo of the test road with indication of the measurement positions (north is upwards, picture by B. R. Nilsson). Before the final pavement was chosen several laboratory studies and full-scale field test were performed. Extensive discussion with representatives from the city of Malmö was also a part of the decision process. Representatives from The city of Malmö clearly specified that they did not want to spend extra time and money on cleaning and winter maintenance on the noise-reducing pavement compared to a conventional pavement. For these reasons, both single and double layer porous pavements were not used in Malmö. The only realistic option left was to use a thin open graded pavement with optimized surface texture. Several recipes were evaluated in the laboratory before any field tests were performed. The mixes were blended in a laboratory mixer and then slabs were manufactured using a static roller. From the slabs, cores were taken and were later analyzed to determine for example

13 void content, amount of binder, grading, etc. One major challenge during this process was to determine which maximum aggregate size to use. It is widely known that a pavement containing fine aggregate is generally less noisy compared to a pavement with a larger aggregate. However, since many Nordic countries, including Sweden, use studded tyres during the winter it is not possible to use fine mixes because of the wear. After the laboratory tests four mixes were chosen to the preliminary field test in Dalby, a small village outside Malmö near the Skanska asphalt plant. Basic data for the mixes are shown in Table 1. Table 1: Basic data for the four mixes used in the preliminary field test in Dalby 6 Max. aggregate size (mm) Void content (%) Binder content (%) Texture, Sand Patch (mm) Recipe * 1.05 Recipe * 1.15 Recipe * 0.75 Recipe * 0.85 *) Durabit NR, a polymer modified binder provided by Nynäs. After the four initial test sections were laid, an indicative noise measurement was carried out. A microphone was placed 7.5 meters from the centre of the road and 1.2 meters above the surface. This setup is similar to the SPB measurement but in this case only a limited numbers of passing cars were measured at 50 km/h in order to get an indication of the noise reducing potential for the four pavements. Based on these measurements it was concluded that recipe 1 and 2 gave similar noise reduction (>3 db) when data was compared to an old reference section but when recipe 3 and 4 were used noise reduction was close to zero. Since approximately % of the cars in southern Sweden use studded tyres during the winter, and the amount of traffic is not very high on Ellenborgsvägen, it was decided that recipe 1 (max aggregate size 8 mm) should be used. On August 28 and 29, 2006, Skanska laid the noise-reducing pavement and a reference pavement (SMA 0/11) on Ellenborgsvägen. The weather conditions were optimal with sun and relatively warm temperatures. Prior to this the old wearing course had been milled off and at each bus stop the pavement was strengthened to withstand the slow moving traffic (mostly busses) using Skanska s bus-stop concept. In addition to this, manhole covers were replaced with telescopic covers that were easily adjusted when the paver had passed. This saved time and increased the quality and smoothness of the surface. Before the mix was placed, a tack coat was sprayed on the milled surface to ensure good adhesion between layers. To achieve a smooth surface without joints the noise reducing pavement was placed without stops. This required good logistical coordination between the asphalt plant, transport company and the crew operating the paver. The mix thickness was approximately 35 mm and it was compacted using a static roller SPB measurements by DRI DRI has performed SPB measurements at the test site in Malmö [DRI 67, 2008]. Measurement dates, time, and temperatures are given in Table 2 for both test sections. Since the distance between the radar and the measurement cross section are m, and the distance to the middle of the lane are 4-7 m, the speed correction factors according to Eq. (2) are small: Figure 1 shows the test area and the two SPB measurement positions.

14 7 Table 2: General measurement data at both test sections (temperatures in C). Surface Date Time T air, avg (min-max) T road, avg (min-max) 1/cos α SMA 8 (TA3) :55-12: ( ) 25.9 ( ) SMA 11 (ref.) :34-16: ( ) 30.3 ( ) SMA 8 (TA3) :11-12: ( ) 20.8 ( ) SMA 11 (ref.) :57-16: ( ) 25.5 ( ) SMA 8 (TA3) :11-14: ( ) 17.1 ( ) SMA 11 (ref.) :39-16: ( ) 17.6 ( ) The test sections are m long and the measurement positions are chosen near the middle of the test sections. The reference section goes from the east crossing with Hüsie Kyrkoväg 570 m to the bend in the middle of the photo where the optimized test surfacing starts. The length of this section is 380 m and it ends at Stenkällevägen. Position 1 is shown in Figure 2. The microphone is placed on the footpath just before a small byroad Virentoftagatan near Ellenborgsvägen 74. Cars in the near lane (going west) are measured, and the whole transmission path is acoustically hard (paved). The road has a slight increasing gradient towards west as can be seen in the photo in Figure 3. Position 2 is shown in Figures 4 and 5. The microphone is placed on the grass shoulder 0.7 m behind the kerb near Ellenborgsvägen Cars in the opposite lane (going east) are measured Test surfaces The old surfacings on the test road were judged to be dense asphalt concrete with 11 mm maximum aggregate size (AC 11 d 1, equal to DAC 11) on the east test section and 16 mm (AC 16 d or DAC 16) on the western part. Both surfacings (especially the western part, AC 16 d) were severely cracked and had a lot of potholes, repairs, and other damages. The optimized test surface around position 1 is an SMA 8 / TA3 (called TA3 by the manufacturer) 2 with the basic target data: maximum aggregate size 8 mm % air voids 6.5 % polymer modified binder thickness approximately 35 mm The mean texture depth (measured as volumetric patch) should be 1.05 mm. The aggregate grading curve is deviating from standard SMA pavements. The reference surface around position 2 is a standard SMA 0/11 with maximum aggregate size 11 mm. The photos in Figures 6 and 7 show the old surface as seen from the microphone, and close up details of the old surface, the new surface at year 0, and the new surface at year 1, at both test sections. 1 These are Danish road surface designations (used by DRI in this Swedish location). Corresponding international designations would be DAC 0/16 and DAC 0/11, in accordance with the HARMONOISE procedures, and corresponding Swedish designations would be ABT 16 and ABT We therefore use the designation SMA 8 / TA3 here, to emphasize that it is a special variant of SMA 8 which is called TA3 by its constructor.

15 Figure 2: Position 1 seen towards northeast. The radar is in the silver car on the roadside, and the red car in the far lane is measured (Sept. 2007, new surface, year 1). 8 Figure 3: Position 1 seen towards west, (Sept. 2007, new surface, year 1). Figure 4: Position 2 seen towards north-east. The radar is in the silver car on the roadside, and the nearest, right lane is measured (June 2006, old surface).

16 9 Figure 5: Position 2 seen towards west (Sept. 2007, new surface, year 1). Figure 6: Photographs of test section 1 (pos. 1, middle of test lane and driving direction indicated) and surface details. The black and white squares are 10 x 10 mm 2. Pos.1: Old surface (middle of lane marked) Old surface: AC 16 d, June 2006 SMA 8 / TA3, Year 0, Sept SMA 8 / TA3, Year 1, Sept. 2007

17 10 Figure 7: Photographs of reference section 2 (pos. 2, middle of test lane and driving direction indicated) and surface details. The black and white squares are 10 x 10 mm 2. Pos.2: Old surface (middle of lane marked) Old surface: AC 11 d, June 2006 SMA 11, Year 0, Sept SMA 11, Year 1, Sept Results General remarks All results are given at the reference conditions 50 km/h (the posted speed limit) and an air temperature of 20 C. Results for heavy vehicles are considered indicative only since they are generally based on the relatively few busses passing by. These results may, however, be used for comparisons provided caution is observed when making conclusions Results SPB measurements before repaving (old conventional surfaces) The results of the SPB measurements on the old pavements in June 2006, before repaving occurred, are presented in Table 3 (traffic statistics) and Table 4 (overview of noise results). The standard uncertainty related to L veh is the standard deviation due to the statistical variance only. The "traffic statistics" show (only) the number of vehicles (passages) used in the measurements and their speed data.

18 Table 3: Traffic statistics during measurements in June 2006, before repaving, at both test sections. "Dual-axle" and "Multi-axle" refer to the sub-category of heavy vehicles. 11 Position Vehicle category Number of passings min. speed [km/h] max. speed [km/h] avg. speed [km/h] stdev. speed [km/h] 1 cars dual-axle multi-axle cars dual-axle multi-axle Table 4: Main results of the SPB noise measurements in June 2006, before repaving, at both test sections. Results within parenthesis are indicative only. Position Vehicle category L veh [db] Std. uncertainty [db] Intercept, A Slope, B 1 cars dual-axle (78.3) multi-axle (80.6) cars dual-axle (86.6) multi-axle For cars, detailed results are given in Figures 8 and 9 showing the scatter plots of the maximum pass-by vehicle noise levels (L pafmax ) of cars as a function of speed. The graphs also show the logarithmic regression lines A + B log(v) and the 95 % confidence limits. These lines are given in a speed range of ± 1.5 standard deviation of the speed around the average speed as stated in [ISO ]. Figure 10 gives the 1/3 octave band spectrum of L veh, cars at 50 km/h at both test sections. The vehicle sound level for cars was 74.1 db at the old surface at pos. 1 (AC 16 d) and 71.4 db at pos. 2 (AC 11 d) 3. 3 These are Danish road surface designations (used by DRI in this Swedish location). Corresponding international designations would be DAC 0/16 and DAC 0/11, in accordance with the HARMONOISE procedures, and corresponding Swedish designations would be ABT 16 and ABT 11.

19 Figure 8: L pafmax as a function of speed for cars at section 1 (pos. 1, AC 16 d, old). Logarithmic regression line and 95 % confidence limits are also shown Pos. 1, old LpAFmax [db] Speed [km/h] Figure 9: L pafmax as a function of speed for cars at section 2 (pos. 2, AC 11 d, old). Logarithmic regression line and 95 % confidence limits are also shown Pos. 2, old LpAFmax [db] Speed [km/h]

20 Figure 10: Third-octave band spectrum of L veh (50 km/h) for cars at sections 1 and 2, before repaving (old surfaces) AC 16d, old, 74.1 db AC 11d, old, 71.4 db 65 L_cars, [db] Centerfrekvens, [Hz] Results SPB measurements after repaving (newly laid surfaces, "year 0") The results of the measurements in September 2006, about one month after repaving, are presented in Table 5 (traffic statistics) and Table 6 (overview of the noise results). Table 5: Traffic statistics during measurements in September 2006 ("year 0") at both test sections. "Dual-axle" and "Multi-axle" refer to the sub-category of heavy vehicles. Position Table 6: Main results of the SPB noise measurements in September 2006 ("year 0") at both test sections. Results in parenthesis are indicative only. Position Vehicle category Vehicle category Number of passings L veh [db] min. speed [km/h] max. speed [km/h] Std. uncertainty [db] avg. speed [km/h] Intercept, A stdev. speed [km/h] 1 cars dual-axle multi-axle cars dual-axle multi-axle Slope, B 1 cars dual-axles (74.8) multi-axles (79.0) cars dual-axles (76.0) multi-axles (78.6)

21 14 Figure 11: L pafmax as a function of speed for cars at test section 1 (pos. 1, SMA 8/TA 3, new, "year 0"). Logarithmic regression line and 95 % confidence limits are also shown Pos. 1, year LpAFmax [db] Speed [km/h] Figure 12: L pafmax as a function of speed for cars at test section 2 (pos. 2, SMA 11, new, "year 0"). Logarithmic regression line and 95 % confidence limits are also shown Pos. 2, year LpAFmax [db] Speed [km/h]

22 15 For cars, detailed results are given in Figures 11 and 12 (above), showing the scatter plots of the maximum pass-by vehicle noise levels (L pafmax ) as a function of speed. The graphs also show the logarithmic regression lines A + B log(v) and the 95 % confidence limits. These lines are given in a speed range of ± 1.5 standard deviation of the speed around the average speed as stated in [ISO ]. Figure 13 shows the 1/3 octave band spectrum of L veh, cars at 50 km/h at both test sections. The vehicle sound level for cars was 67.4 db at the new optimized "low-noise surface" at pos. 1 (SMA 8 / TA3) and 70.7 db at the new reference surface at pos. 2 (SMA 11). Figure 13: Third-octave band spectrum of L veh for cars at 50 km/h at sections 1 and 2, after repaving, "year 0" SMA 8, year 0, 67.4 db SMA 11, year 0, 70.7 db 65 L_cars, [db] Centerfrekvens, [Hz]

23 Results SPB measurements one year after repaving ("year 1") The results of the measurements in September 2007 are given in the following. Table 6 gives the traffic statistics and table 7 the overview of the noise results. For cars, detailed results are given in Figures 14 and 15 showing the scatter plots of the maximum pass-by vehicle noise levels (L pafmax ) of cars as a function of speed. The graphs also show the logarithmic regression lines A + B log(v) and the 95 % confidence limits. These lines are given in a speed range of ± 1.5 standard deviation of the speed around the average speed as stated in [ISO ]. Figure 16 shows the 1/3 octave band spectrum of L veh, cars at 50 km/h at both test sections. The vehicle sound level for cars one year after repaving was 71.4 db at the noise-optimized surface at pos. 1 (SMA 8 / TA3) and 71.8 db at the reference surface at pos. 2 (SMA 11). The appearance of the two surfaces at a close distance is shown in Figure 17. Note the dramatic difference in macrotexture, despite the maximum aggregate size differs only between 11 and 8 mm. The reason is that not only the maximum aggregate size is different; also the grading curves are very different. One may also note that there does not seem to be any porosity (air voids) left in the surface at the age of 8 months. Thus, it is believed that the noise reduction loss over the first year is due mainly to the closing of the porosity and once this has been made, the noise reduction loss will be much slower (if any). Table 7: Traffic statistics during measurements in September 2007 ("year 1") at both test sections. "Dual-axle" and "Multi-axle" refer to the sub-category of heavy vehicles. Position Vehicle category Number of passings min. speed [km/h] max. speed [km/h] avg. speed [km/h] stdev. speed [km/h] 1 cars dual-axle multi-axle cars dual-axle multi-axle Table 8: Main results of the SPB noise measurements in September 2007 ("year 1") at both test sections. Results in parenthesis are indicative only. Position Vehicle category L veh [db] Std. uncertainty [db] Intercept, A Slope, B 1 cars dual-axle multi-axle (79.1) cars dual-axle multi-axle (78.1)

24 Figure 14: L pafmax as a function of speed for cars at test section 1 (pos. 1, SMA 8/TA 3, "year 1"). Logarithmic regression line and 95 % confidence limits are also shown Pos. 1, year LpAFmax [db] Speed [km/h] Figure 15: L pafmax as a function of speed for cars at test section 2 (pos. 2, SMA 11, "year 1"). Logarithmic regression line and 95 % confidence limits are also shown Pos. 2, year LpAFmax [db] Speed [km/h]

25 Figure 16: Third-octave band spectrum of L veh for cars at 50 km/h at sections 1 and 2, "year 1" SMA 8, year 1, 71.4 db SMA 11, year 1, 71.8 db 65 L_cars, [db] Centerfrekvens, [Hz] Figure 17: Appearance of the SMA 11 reference surface (left) and the noise-optimized SMA 8 / TA3 surface (right). Photo shot at an age of 8 months. Coin: 25 mm diameter.

26 Results - Noise reduction The noise reduction at 50 km/h is presented in Table 9, based on the results in Tables 4, 6, and 8. The initial noise reduction in both positions ( L 0, old ) is given relative to the noise from the old surface at the same position, while the noise reduction in year 0 ( L 0 ) and year 1 ( L 1 ) is given relative to the noise at the reference surface (pos. 2) in the same year. The noise reduction for dual- and multi-axle trucks is hardly reliable since the number of passes was low and the speed range was lower than the reference speed (making it necessary to extrapolate to 50 km/h). Table 9: Noise reduction relative to the old surface in positions 1 and 2 (2 nd and 3 rd columns), and noise reduction in position 1 relative to the reference surface in position 2 in year 0 and year 1 (4 th and 5 th columns). Results in parenthesis are indicative only. Vehicle category L 0, old, pos. 1, [db] L 0, old, pos. 2, [db] L 0, [db] L 1, [db] cars dual-axle (3.5) (10.6) (1.2) - multi-axle (1.6) - (-0.4) (-1.0) Figure 18 shows the 1/3 octave band spectra for cars at the noise-optimized surface (SMA 8 / TA3) and at the reference surface (SMA 11). The initial noise reduction was very good - approximately 4 db for frequency bands 800 and 1000 Hz, and 2-3 db for lower and higher frequencies. This indicates that the surface is open-textured and reduces aerodynamic noise ( air-pumping noise ), and that it at the same time is relatively smooth-textured, thus reducing vibration-generated noise. However, in year 1, the noise spectra are almost identical for the two surfaces - the 1/3 octave band levels Hz for the test surface are db lower than the reference. The difference between the overall levels is only 0.5 db. This corresponds well to the rule-of-thumb stating a noise reduction of 0.25 db per 1 mm smaller maximum aggregate size: SMA 11 compared to SMA 8 should then correspond to 0.75 db. The noise at the reference surface (SMA 11) increased approximately 1 db during the first year. Figure 18 shows that this increase particularly took place in frequency bands above 1000 Hz. This indicates that the surface texture after one year is less open than at the new surface; thereby increasing the aerodynamic noise ( air-pumping noise ). It is obvious from Figure 10 (and Figure 19) that the noise from the old surface at position 1 (AC 16 d) is around 3 db higher in the whole frequency range than the noise from the old surface at position 2 (AC 11 d). This indicates that the old surface at position 2 was still in rather good condition: not very uneven and still open at the top surface. Comparison of the 1/3 octave band spectra of L veh at both sections in year 1 to the noise spectra recorded for the old surfaces is shown in Figure 19. The spectra for the new SMA surfaces are similar to the spectrum for the old AC 11 d surface at position 2 in the most important frequency bands Hz. At lower frequencies the 1/3 octave band levels were 3-4 db higher at the old surface indicating a rougher surface perhaps due to ravelling and cracks. However, at frequency bands from Hz the noise from the SMA surfaces are 2-3 db higher than the noise at the old AC 11 d surface. This may be due to a denser surface texture of the SMA wearing courses after one year than of the old dense asphalt concrete surface (cf. Figure 7).

27 Figure 18: Third-octave band spectra of L veh for cars at 50 km/h at sections 1 and 2, for surface ages designated "year 0" (newly laid) and "year 1" (one year old) SMA 8, year 1, 71.4 db SMA 11, year 1, 71.8 db SMA 8, year 0, 67.4 db SMA 11, year 0, 70.7 db L_cars, [db] 4000 Centerfrekvens, [Hz] Figure 19: Third-octave band spectra of L veh for cars at 50 km/h at sections 1 and 2 at the old and the new wearing courses after 1 year of traffic exposure SMA 8, year 1, 71.4 db SMA 11, year 1, 71.8 db AC 16d, old, 74.1 db AC 11d, old, 71.4 db L_cars, [db] Centerfrekvens, [Hz]

28 Results - CPX measurements by M+P/Carl Bro The constructor of the SMA 8 / TA3 surface (Skanska AB) also engaged consultants M+P in the Netherlands via Carl Bro in Denmark to conduct some measurements with the CPX method; using the M+P CPX trailer. It is believed that the tyres used were the four reference tyres CPX A, CPX B, CPX C and CPX D as specified in [ISO/DIS ]. The results from the measurements in "year 1" are shown in Figure 20. These values are probably the socalled CPXL in the draft standard; i.e. meant to represent light vehicles. It appears that in this case the SMA 8 / TA3 surface still after one year of traffic exposure gives a noise reduction of 3 db, while the SMA 11 after the same traffic exposure has approached the noise level of the old DAC 11. There is currently no explanation for the large discrepancies between the results of the SPB and CPX measurements in "year 1", where SPB indicated 0.5 db of noise reduction while CPX indicated 3 db. Normally, one would consider SPB measurements as more relevant and closer to the actual conditions. However, for comparisons such as here, the CPX method is normally very precise. New CPX measurements are planned to be made in Figure 20: Noise level measured with the CPX method (the CPXL index) at 50 km/h 10 months after repaving ("year 1") at section 1 ("SMA 8") and section 2 ("SMA 11"). The value for "DAC 11 old" is a measurement before repaving at position 2.

29 Results - CPX measurements by VTI / TUG within SILENCE CPX measurements were also made within the SILENCE project by VTI. These measurements were made utilizing the equipment and staff of the Technical University of Gdansk (TUG), Poland. The results from the measurements in "year 1" (in October 2007) are shown in Figure 21. Test speed was 50 km/h. The test tyres used were: CPX A, according to [ISO/DIS ] CPX D, according to [ISO/DIS ] ASTM SRTT, standard ref tyre in accordance with ASTM standard F2493 MudTerrain, BF Goodrich MudTerrain T/A The two first tyres are the ones used so far in the CPX method; the latter two tyres are those which are expected to replace the old reference tyres as they are now obsolete. CPX A and ASTM SRTT are intended to simulate the effects of road surfaces on light vehicle noise, while the CPX D and MudTerrain are intended to simulate the effects of road surfaces on heavy vehicle noise. The results in Figure 21 indicate that the SMA 8 / TA3 surface still after one year of traffic exposure gives a noise reduction of 1-2 db compared to the reference SMA 11 surface, depending on tyre type. Figure 21: Noise reduction measured with the CPX method using four test tyres at 50 km/h 14 months after repaving ("year 1"). The noise reduction is the difference between the noise levels measured on section 1 ("SMA 8 / TA3") and on section 2 ("SMA 11"). 3 2,5 Noise reduction [db] 2 1,5 1 0,5 0 CPX A ASTM SRTT CPX D MudTerrain Tyre type

30 Results - Annoyance study In addition to noise measurements, a survey to study how much annoyed people were by the traffic noise before and after repaving was conducted by Lund University, Lund Institute of Technology (LTH) [Turanovic, 2007]. This study was based on prior work done by the Danish Transport Research Institute [Bendtsen et al., 2002]. However, the questionnaire had to be translated and adjusted to Swedish conditions. The major findings from the Swedish study at Ellenborgsvägen were: 89 % of the respondents were very positive or positive to the implemented measures (repaving with the SMA 8 / TA3 surface). The residents along Ellenborgsvägen felt that the noise level had decreased inside their homes (both with opened and closed windows), in the garden, on the terrace or balcony, in the yard and along the street. In comparison with the conventional reference pavement (SMA 11) the noise reducing pavement was perceived as quieter. The number of highly annoyed people was reduced from 27 % to 12 % after repaving with the quieter asphalt. An increased noise level lead to more annoyance among the residents. The source of noise and the time of day were important to the degree of annoyance. The degree of annoyance depended on how long the resident had lived in their current house or apartment. The type of residence played a significant role in the degree of annoyance Conclusions regarding the Malmö experiment On a residential street in the city of Malmö two old wearing courses of dense asphalt concrete with 11 and 16 mm maximum aggregate size were substituted by two new stone mastic asphalt wearing courses - one with 11 mm maximum aggregate size as a reference surface the other with 8 mm maximum aggregate size; the latter being optimized for noise reduction. Statistical pass-by (SPB) noise measurements were carried out before rebuilding the road, one month after repaving, and again one year later. Only results for cars (at 50 km/h and 20 C) are evaluated here. The initial noise reduction of the optimized SMA 8 wearing course was very good, being almost 7 db compared to the old 16 mm wearing course, and 3.3 db compared to the new reference SMA 11. Frequency bands above 1 khz were reduced by 2-3 db, indicating that the surface was open-textured (reducing aerodynamic air-pumping noise). Also frequency bands below 1 khz were reduced, indicating that the surface was smooth-textured (reducing vibration-generated noise). After one year (one winter with traffic using studded tyres) this noise reduction had more or less vanished. The difference between the noise levels at the two new SMA surfaces was reduced to 0.5 db; which essentially occurred around 1 khz, presumably due to the smaller maximum aggregate size. This noise level is almost identical to the noise level from the old dense asphalt concrete with 11 mm maximum aggregate size. Such a reduction, or more, would normally be achieved only by using 8 mm maximum aggregate size instead of 11 mm. However, a measurement with the CPX method by consultants M+P/Carl Bro indicated quite different results, namely that the SMA 8 / TA3 surface also after one year of traffic exposure had a remaining noise reduction of 3 db. Measurements with the CPX method by VTI/TUG indicated an average result of the SPB measurements by DRI and the CPX measurements by M+P/Carl Bro, namely a noise reduction of 1-2 db, depending on tyre type used. It is not known which results that are the most accurate or representative; thus the conclusion is that results after one year are inconsistent. The average of all measurements is a noise

31 reduction of 3 db when the SMA 8 / TA3 surface was new and about half that reduction remaining after one year of exposure to traffic, of which some is by studded tyres. The annoyance study made before and after repaving indicated very positive results in terms of reduced annoyance with the new noise-optimized SMA surface by Skanska Overall conclusions for thin layers Earlier Danish results have shown that for newly laid thin layers the noise reduction achieved in relation to a DAC 0/11 (Dense Asphalt Concrete with 11 mm aggregate) or SMA 0/11 reference is around 3 db for passenger cars. After two years this is reduced to 1-2 db. The results from the experiment in Malmö confirm these observations (after one year). For multiaxle trucks (in the highway case at least) there is a constant noise reduction over two years of around 2 db for the best performing thin layers (AC8o, SMA8+ and UTLAC8). In order to improve the noise reduction, a new experiment was conducted in Denmark. Noise measurements were carried out when the pavements were 1-2 months of age. The best noise reduction, 4.7 db for passenger cars, was measured on an open-graded asphalt concrete with 6 mm aggregate (AC6o); also with a DAC 0/11 as reference surface. The results from the thin layer experiments are useful already when noise reducing pavements are needed for new roads as well as in the pavement maintenance process. But there is a need for further research in order to establish lifetime measurements series of the performance and to investigate the change of the surface structure over the years in order to be able to develop even more acoustically durable thin layers for application on highways as well as urban roads. Obviously, a three-year project such as SILENCE cannot provide such information. There is also a need to further develop and test thin layers adapted to roads where studded tyres are used in the winter period.

32 25 5 Surface dressings 5.1 Introduction Different types of surface dressings are often used on urban and rural roads with limited traffic volume as an inexpensive maintenance measure. This is generally a less expensive pavement maintenance measure in relation to applying a new dense top layer. However, surface dressings are normally considered to result in increased noise emissions from the road traffic. The major reason is that the texture of traditional types of surface dressings is generally very rough. Figure 22 is an illustration of this. Figure 22: Close-up photo of a surface dressing. The rough surface texture caused by the aggregate is evident. Work Package F1 of SILENCE includes a subtask F.1.2 New production technologies for surfaces on urban streets focussing on this subject. As part of this subtask, investigations of traffic noise from streets and roads with surface dressings have been carried out. The objectives were: 1. To produce a ranking of different kinds of surface dressings in relation to a reference pavement. 2. To identify, develop and test less noisy types of surface dressings.

33 Some earlier data Some measurements are presented in various parts of the Tyre/Road Noise Reference Book [Sandberg & Ejsmont, 2002], such as in Figure 23 which is from this book. Note that the relation between noise level and texture (MPD) is very poor. Figure 23: Comparison of A-weighted sound levels (CPXI) measured by TUG with the CPX method on different surfaces. Test tyres in accordance with ISO/DIS In addition, the Mean Profile Depth (MPD) of the macrotexture according to ISO of each surface is indicated by red diamonds connected with red lines. Each bar is filled with a photo of the surface. Surface dressings are shown in the middle. In a part of the EU project SILVIA [SILVIA, 2005], DRI carried out an inventory on the noise emission measured by the SPB method [ISO ] from different types of pavements commonly used in Europe [Delta, 2004]. In this inventory, the aggregate size and the pavement age are not included as active parameters because there were not enough data to do so. The main results in relation to surface dressings can be seen in Figure 24 where Dense Asphalt Concrete (DAC) is used as a reference pavement. Generally the noise level from surface dressings is higher than the noise level from DAC pavements. It can be seen that for passenger cars the difference is 3.5 db at a reference speed of 50 km/h and 5.5 db at 110 km/h. For passenger cars, surface dressings become relatively noisier at high speeds. The difference is less for dual-axle trucks where it is 1.9 db at 50 km/h and 1.5 at 85 km/h. For multi-axle trucks the difference is 2.1 at 50 km/h and 1.6 db at 85 km/h. These results

34 indicate that for heavy vehicles there is a slight tendency that the difference in noise between surface dressings and DAC is reduced as the speed is increased. Figure 24: Typical noise levels for dense asphalt concrete (DAC) and surface dressings (SD) at different speeds for passenger cars and trucks [Delta, 2004]. "Two axle" and "Multi axle" refer to the number of axles for heavy vehicles considered Noise [db] DAC SD km/h 110 km/h 50 km/h 85 km/h 50 km/h 85 km/h 60 Passenger cars Two axle Multi axle Finally, a comprehensive summary of surface dressing measurements in comparison to other road surfaces is shown in Figure 25. The figure shows measurements on one surface per each horizontal line. Note that the average level for the surface dressing group is the highest of all for light vehicles. 5.3 Study of traditional surface dressings used on low-volume highways SILENCE is focused on urban areas. So why study surface dressings on low-volume highways? The answer is: Looking at the traditional use of surface dressings and the variation between such types will constitute a background basis for urban applications It was felt that there was a lack of data measured with presently used methods and distinguishing between various types There was a very good opportunity to link with a national project in Sweden to collect information at a very low cost. Nine surface dressings in middle Sweden were selected for this study. They included the types listed in Table 10. On these, TUG made measurements at 50 and 80 km/h with the CPX method, using the TUG CPX trailer. The same four tyres as described in Section were used. Texture measurements were made by the VTI laser profilometer, from which (among others) MPD values according to ISO were obtained. The tested surfaces were in a range of conditions, mostly in an "average" condition, having been subject to both light and heavy traffic for a number of years. The area is an area where forestry is of prime importance which means that a substantial part of the wear had been caused by heavily loaded 6-8-axle timber trucks. The results are shown in Figures

35 Figure 25: Comparison of A-weighted sound levels measured in France with the SPB method on a great number of different pavements, classified into 11 categories. The figure is based on data collected over many years. The top half contains data relevant to traffic by light vehicles (cars); the bottom half data relevant to traffic by heavy trucks. Note: chps = chippings. Diagrams from [Crocker, 2007], originally obtained from F. Besnard, SETRA, France SPB measurements on 283 pavements - light vehicles (LAmax, temperature 20 C, normalized to speed 90 km/h) A-weighted SPL [db] Porous asphalt concrete max 6 mm chps Porous asphalt concrete max 10 or 14 mm chps Ultra-thin surfacing (<20mm) max 6 mm chps Very thin surfacing (20-25mm) max 6 or 10 mm chps, porous Very thin surfacing (20-25mm) max 6 or 10 mm chps, dense Dense asphalt concrete max 10 or 14 mm chps Surface dressing (chip seal) max 6 or 8 mm chps Surface dressing (chip seal) max 10 or 14 mm chps Ultra-thin surfacing (<20mm) max 10 mm chps Very thin surfacing (20-25mm) max 14 mm chps, dense Cement concrete various surface treatments SPB measurements on 166 pavements - heavy vehicles (LAmax, temperature 20 C, normalized to speed 80 km/h) A-weighted SPL [db] Porous asphalt concrete max 6 mm chps Porous asphalt concrete max 10 or 14 mm chps Ultra-thin surfacing (<20mm) max 6 mm chps Very thin surfacing (20-25mm) max 6 or 10 mm chps, porous Very thin surfacing (20-25mm) max 6 or 10 mm chps, dense Dense asphalt concrete max 10 or 14 mm chps Surface dressing (chip seal) max 6 or 8 mm chps Surface dressing (chip seal) max 10 or 14 mm chps Ultra-thin surfacing (<20mm) max 10 mm chps Very thin surfacing (20-25mm) max 14 mm chps, dense Cement concrete various surface treatments

36 Table 10: Surface dressings tested in Sweden in 2006 Surface Swedish Type of surface No. designation S12 Y1B 8/11 Single surf dressing, 8-11 mm chippings 29 S13a Y1B 11/16 Single surf dressing, mm chippings Eastern direction S13b Y1B 11/16 Single surf dressing, mm chippings Western direction S14 Y2B 11/16+4/8 Double surf dressing, mm chippings in lower layer and 4-8 mm in top layer S15 Y1B 8/16 Single surf dressing, 8-16 mm chippings S16 Y1B 11/16 Single surf dressing, mm chippings S17 - Racked-in surf dressing (smaller chippings on top of single layer) S18 Y1B 4/16 Single surf dressing, 4-16 mm chippings S19 Y1B 8/16 Single surf dressing, 8-16 mm chippings Figure 26 shows noise levels measured by the CPX method, using tyres CPX A and ASTM SRTT (above) and tyres CPX D and Goodrich MudTerrain (below). The results of these have been grouped together two-by two since these pairs of tyres react to surfaces in a similar way, with the two first representing the noise sensitivity to light vehicles and the two last heavy vehicles. Results for 50 km/h are shown in green bars at the left and for 80 km/h in violet bars at the right. Typical reference levels for SMA and DAC surfaces are shown as bands in the diagrams. The following observations are of interest: The range of noise for the tested surfaces is 3 db for light traffic and 2 db for heavy traffic For light traffic the average noise level is approx. the same as for the reference surfaces, but for heavy traffic the surface dressings are quieter The surfaces which are the quietest for light traffic are the noisiest for heavy traffic, and vice versa. For composite traffic of equal (noise) weight this means that the difference between the surfaces is minimal. In general, the quietest surfaces for light traffic are the ones with the smallest chippings, while the contrary is the case for heavy traffic. Further insight into the problem is gained by looking at the relation between noise level and texture (MPD), as shown in Figure 27. Note that in this case the surface type is uniform and the MPD values are generally very high, which explains why the correlation between noise and MPD is significant. It shows that the reason for the results is that the noise levels of the tyres representing light traffic have a positive correlation with texture, whereas the contrary is the case for the tyres representing heavy traffic. This is an effect of the proportions between the texture dimensions and the tyre tread dimensions [Sandberg & Ejsmont, 2002], where the positive correlation generally occurs when the texture dimensions are larger than the tyre tread elements and the negative correlation occurs when the texture dimensions are smaller than the tyre tread elements.

37 Figure 26: Noise levels measured by the CPX method, using tyres CPX A and ASTM SRTT (above) and tyres CPX D and Goodrich MudTerrain (below). Results for 50 km/h in green bars at the left and for 80 km/h in violet bars at the right. Typical reference levels for SMA and DAC surfaces are shown as bands in the diagrams. Tyre CPXA & ASTM km/h Noise level [db(a)] 102 Approx. noise level of SMA 0/16, SMA 0/11, DAC 0/16 and DAC 0/11 at this position at 80 km/h km/h 94 Approx. noise level of SMA 0/16, SMA 0/11, DAC 0/16 and DAC 0/11 at this position at 50 km/h S12 S13A S13B S14 S15 S16 S17 S18 S19 S12 S13A S13B S14 S15 S16 S17 S18 S19 Surface Tyre CPXD & MudT 104 Noise level [db(a)] km/h Approx. noise level of SMA 0/16, SMA 0/11, DAC 0/16 and DAC 0/11 at this position at 80 km/h km/h Approx. 92 noise level of SMA 0/16, SMA 0/11, DAC 0/16 and DAC 0/11 at this position at 50 km/h 90 S12 S13A S13B S14 S15 S16 S17 S18 S19 S12 S13A S13B S14 S15 S16 S17 S18 S19 Surface

38 Figure 27: Noise levels measured by the CPX method (green symbols and line representing light traffic and violet symbols and line representing heavy traffic) plotted against measured MPD values km/h, CPXA&ASTM 80 km/h, CPXD&MudT 102 Noise level [db] y = -0,7844x 2 + 3,907x + 97,031 R 2 = 0,8932 y = 0,3658x 2-1,8983x + 101,09 R 2 = 0, Mean Profile Depth MPD [mm] 5.4 Development and testing in Denmark of less noisy surface dressings The Fakse experiment In 2003, the NCC contracting company in Denmark conducted an experiment with surface dressings at a highway near Fakse on Zealand. The objective was to investigate the noise emission from different types of surface dressings where the maximum aggregate size was changed. Four test sections with different types of surface dressing types were constructed on a road with a speed limit of 80 km/h. The maximum aggregate size was 2/5 or 8/11 mm. On one test section a double surface dressing was constructed. A bottom layer with 8/11 mm aggregate and on top of this a layer with 5/8 mm aggregate was applied. See Table 11. Table 11: Description of the surface dressings used at the Fakse experiment [Delta, 2004]. Pavement Max. aggregate size [mm] DAC 0/11 (reference) 11 SDS 2/5 type 1 2/5 SDS 8/11 8/11 SDD 8/11+5/8 8/11+5/8 SDS 2/5 type 2 2/5

39 SPB noise measurements were performed by DELTA [Delta, 2004] after the pavements reached an age of 3 months. A dense asphalt concrete with 11 mm aggregate (DAC 0/11) of the same age at the test site was used as a reference pavement. The main results of the noise measurements can be seen in Figure Figure 28: Results of the SPB noise measurements at surface dressings at Fakse when the pavements were 3 months old. db Passenger car 80 km/h Two axle truck 70 km/h Multi axle truck 70 km/h SD 2/5 type 1 SD 2/5 type 2 SD 5/8+8/11 SD 8/11 DAC11 In Figure 29 the noise results are shown relative to the DAC 0/11 pavement at 80 km/h for passenger cars and at 70 km/h for trucks. It appears that the surface dressings generally give higher noise level for passenger cars than the DAC 0/11. For the SDS 8/11 the level is 4.2 db higher. When 5/8 aggregate is applied as a second layer the difference is reduced to 2.6 db. For the SDS 2/5 type 1 the difference is 2.0 db whereas for the type 2 there is no difference (-0.3 db). The situation is much different for two- and multi-axle trucks for which there is no significant difference in noise between the surface dressings and the DAC 0/11. Figure 29: Increased noise level relative to DAC 0/11 for surface dressings at Fakse at 3 months age. Reference speed was 80 km/h for passenger cars; 70 km/h for trucks db Passenger car Tw o axle truck Multi axle truck -1 SD 2/5 type 1 SD 2/5 type 2 SD 5/8+8/11 SD 8/11

40 The frequency spectra in the important frequency range 400 to 4000 Hz can be seen in the following figures. The spectra for passenger cars can be seen in Figure 30. In the frequency range below 1500 Hz the surface dressings give significantly higher levels than the DAC 0/11 reference pavement (up to 7 db). This indicates increased vibration noise caused by a rough surface texture. Only one of the surface dressings with 2/5 mm aggregate (SDS 2/5 type 2) has nearly the same level of vibration-generated noise as the DAC 0/11 pavement. In the frequency range above 1500 Hz, where the air pumping noise is dominant, all the surface dressings had a slightly lower noise levels than the DAC 0/11 pavement (0 to 2 db). This indicates that the pavements have an open structure, which can to some extent reduce "air pumping" noise. Especially the SDS 8/11 gives up to 4 db noise reduction at the high frequencies but in total, this pavement has a noise level 4.2 db higher than the DAC 0/11 because of a very high level of vibration-generated noise. 33 Figure 30: Frequency spectra for passenger cars at 80 km/h [db(a)] DAC11 SD 2/5 type 1 SD 2/5 type 2 SD 5/8+8/11 SD 8/ Passenger cars Frequency [Hz] The spectra for dual-axle trucks can be seen in Figure 31. The same tendencies as for passenger cars are seen but the differences between the pavements are remarkably less significant. At the lower frequencies below 1500 Hz the surface dressings give 0 to 2 db higher noise levels than the DAC 0/11 and at the higher frequencies there is a 0 to 2 db reduction in relation to the DAC 0/11 pavement. The SDS 2/5 type 2 pavement gives 1 to 2 db lower levels over the entire spectrum. Figure 32 shows the spectra for multi-axle trucks/busses. The tendencies are the same as for dual-axle trucks/busses. But the surface dressing with 8/11 mm aggregate gives up to 4 db lower levels at the higher frequencies than the DAC 0/11, indicating that this pavement reduces the "air pumping" noise, because it has an open surface texture.

41 34 Figure 31: Frequency spectra for dual-axle trucks/busses at 70 km/h [db(a)] DAC11 SD 2/5 type 1 SD 2/5 type 2 SD 5/8+8/11 SD 8/ Two axle truck Frequency [Hz] Figure 32: Frequency spectra for multi-axle trucks/busses at 70 km/h [db(a)] DAC11 SD 2/5 type 1 SD 2/5 type 2 SD 5/8+8/11 SD 8/ Multi axle truck Frequency [Hz]

42 5.4.2 The Danish SILENCE experiment In order to investigate the possibilities for developing surface dressings with reduced noise levels compared to the ordinary types, a project was started in Denmark as a part of Task F1 of the SILENCE project [DRI 68, 2008]. During the summer 2006 the former County of Ribe in West Jutland planned a full-scale experiment on Highway 520 with surface dressings that were optimized with the intention to reduce noise. It was planned to construct test sections with six different surface dressings using aggregate sizes of 5, 8 and 11 mm. Statistical Pass-by (SPB) according to ISO and Controlled Pass-by (CPB) measurements were carried out when the surface dressings were about 3 months old. SPB measurements were repeated at two locations one year later including Close-Proximity (CPX) measurements according to [ISO/DIS ]. This part of the report documents the noise levels at different surface dressings with an age of three months and the trend over a one year period at two of the surface dressings. The test site is located in Jutland on Route 469 (highway 520) between the two red crosses, in Figure 33. The annual daily traffic is around 3200 and the posted speed limit is 80 km/h. The surface dressings are located from km 11,900 to km 14,500. Table 1312 defines the location of the surface dressings more precisely. Close-up photographs of the surface dressings are shown in Figure Figure 33: The surface dressings are located between the two crosses Surface dressings tested Six different surface dressings were constructed, see Table 12. Table 12: Data for the tested surface dressings Surface dressing # Aggregate Binder Constructed [dd.mm.yy] Length [m] m_sds 8/11 Vikan Ecobit 15/ m_s_sds 8/11 + 5/8 Vikan Ecobit 15/ ½ m_sds 5/8 + 2/5 Vikan Ecobit 15/ m_sds 2/5 Vikan Ecobit 15/ SDD 8/11 + 5/8 Vikan Ecobit m_sds 5/8 Vikan Ecobit

43 36 The designations of the surfacing dressings used in Table 12 are from the proposed standardized notation in the Harmonoise report "Source modelling of road vehicles" [0], but supplemented with an "m_ " in the beginning, meaning "modified", and "s_" meaning "sandwich". The construction principle of the different surface dressings is shown in Figures and explained below. Figure 38 shows pictures with scales of the tested surfaces. m_sds 8/11 is a modified single surface dressing with aggregate range 8/11 mm g/m 2 of hot binder is sprayed out onto which kg/m 2 8/11 mm aggregate is laid out. The layer is then compressed by a roller and redundant aggregate is removed. m_s_sds 8/11 + 5/8 is a modified sandwich surface dressing with 8/11 mm aggregate range in the bottom layer and 5/8 mm aggregate range in the top layer. First, 8-9 kg/m 2 of 8/11 mm aggregate is laid onto which g/m 2 of hot binder is sprayed. Then, kg/m 2 of 5/8 mm aggregate is laid over it when the binder still is hot. The layer is then compressed so that the 5/8 mm aggregate is wedged by the 8/11 mm aggregate. Redundant aggregate is removed. The type designated "1½ m_sds 5/8 + 2/5" is a special Danish surface dressing where the binder is covered with two layers of aggregate, the second being of smaller size, as shown in Figure 37. The bottom layer aggregate range is 5/8 mm and the top layer aggregate range is 2/5 mm g/m of hot binder is sprayed out onto which 8 9 kg/m 2 of 5/8 mm aggregate is laid when the binder still is hot. After this 5 6 kg/m 2 of 2/5 mm aggregate is laid. The layers are then compressed so that the 2/5 mm aggregate is wedged by the 5/8 mm aggregate. Redundant aggregate is then removed. m_sds 2/5 is a modified single surfacing dressing with aggregate range 2/5 mm g/m 2 hot binder is sprayed out onto which 9 10 kg/m 2 of 2/5 mm aggregate is laid. The layer is compressed and redundant aggregate is removed. SDD 8/11 +5/8 is a double surface dressing with aggregate range 8/11 mm in the bottom layer and aggregate range 5/8 mm in the top layer g/m 2 hot binder is sprayed out onto which kg/m 2 of 8/11 mm aggregate is laid. The layer is then compressed and redundant aggregate is removed. After this g/m 2 of hot binder is sprayed out, in which kg/m 2 of 5/8 mm aggregate is laid. The surface is then compressed a second time and redundant aggregate is removed. Finally, m_sds 5/8 is a modified single surface dressing with aggregate range 5/8 mm g/m 2 of hot binder is sprayed out onto which kg/m 2 of 5/8 mm aggregate is laid. The layer is compressed and redundant aggregate is removed. The test section with surface dressings that was carried out in collaboration with the former County of Ribe, involved an experiment in testing of a new binder without organic solvents, which resulted in use of different doses of bitumen at each of the surface dressings. The use of different doses of bitumen resulted in bleeding at some of the surface dressings during the hot summer days in Because of the bleeding, these surface dressings were sprayed with 2/5 mm aggregate in order to restore the friction. Two of the test sections - the m_sds 2/5 and the m_sds 5/8 - were bleeding but are still a part of the SILENCE project.

44 37 Figure 34: Construction of single surface dressings [ EN 12271, 2006]. Figure 35: Construction of double surface dressings [EN 12271, 2006]. Figure 36: Construction of sandwich surface dressings [EN 12271, 2006]. Figure 37: Construction of racked-in surface dressings [EN 12271, 2006].

45 38 Figure 38: Photographs of the different surface dressings when the pavements were roughly 3 months old. The black and white squares are 10 mm x 10 mm. a) m_sds 8/11 b) s_m_sds 8/11 + 5/8 c) 1½ m_sds 5/8 + 2/5 d) m_sds 2/5 e) SDD 8/11 + 5/8 f) m_sds 5/ Noise measurements This section describes the measurements that were carried out by DRI [DRI 68, 2008]. In 2006, Statistical Pass-by (SPB) and controlled pass-by (CPB) measurements were carried out. In 2007, SPB measurements were repeated at two locations. Close-proximity (CPX) measurements were also carried out in SPB and CPB measurements were carried out in conformity with ISO ; CPX measurements were carried out in conformity with ISO/DIS SPB measurements were carried out September Measurements were repeated at two surface dressings 29 August The air temperature was in the range of 16 C to 21 C. The location of the measurement positions and other relevant data are shown in Table 13. See also Figure 39.

46 Table 13: Location of the surface dressings, SPB and CPB measurement positions and meteorological conditions during the measurements. 39 Surface dressing m_sds 8/11 m_sds 8/11 s_m_sds 8/11 + 5/8 1½ m_sds 5/8 + 2/5 m_sds 2/5 m_sds 2/5 SDD 8/11 + 5/8 m_sds 5/8 Location [km] 12,500 to 13,000 12,500 to 13,000 13,300 to 13,600 13,600 to 13,900 14,200 to 14,500 14,200 to 14,500 12,200 to 12,500 11,900 to 12,200 Measurement position [km] Measurement date [yyy.mm.dd] Average air temperature [ C] Average surface temperature [ C] 12, , , , , , , , For the SPB measurements, the following parameters were recorded for each pass-by: The maximum noise level L pafmax Third-octave band frequency spectra The vehicle speed The type of vehicle Whenever the speed was measured at an angle relative to the driving direction, the speed was corrected relative to the angle. The speed correction factor was in the range of to The number of vehicles in each category that has been measured within a reasonable time interval can be seen in Table 14. It appears that the number of measured dual-axle trucks/busses was rather low. Table 14: Number of vehicles for which noise levels were measured during the SPB measurements. Pavement and year of measurement Passenger cars, P Dual-axle trucks/busses, L Multi-axle trucks/busses, F m_sds 8/11, year m_sds 8/11, year s_m_sds 8/11 + 5/8, year ½ m_sds 5/8 + 2/5, year m_sds 2/5, year m_sds 2/5, year SDD 8/11 + 5/8, year m_sds 5/8, year CPB measurements were conducted at 50 km/h and 80 km/h. The speed was measured as in the SPB measurements and corrections due to the angle between radar and driving direction were carried out as in the SPB measurements. Three passenger cars and one dualaxle truck were used as pass-by noise sources, see Table 15. The truck was unloaded. Each vehicle made three pass-bys at 50 km/h and 80 km/h.

47 40 Figure 19: Photographs of the measurement sites. The red arrow points in the northern direction. N N m_sds 5/8 SDD 8/11+5/8 N m_sds 8/11 s_m_sds 8/11+5/8 N 1½ m_sds 5/8+2/5 m_sds 2/5 N

48 Table 15: Data for the vehicles used in the CPB measurements. The symbol "-" means "no data available". Vehicle Type Year Tyre front Tyre rear P1 Skoda Felicia, GLX Goodyear 205/55/R16 91V EAGLE NCT5 Goodyear 205/55/R16 91V EAGLE NCT5 P2 Opel Corsa 1985 Michelin 165/70/R13 79T Pirelli P /70/R13 79T 41 P3 Peguot Airvan 1.6D 2001 Hankook 165/70/R13 79T Centrum k702 L1 Mercedes Benz Michelin X XZE2+ 315/80R22.5 XYZ 12 R22.5 CPX measurements were carried out at 50 km/h and 80 km/h from km 14,600 to 14,000 and from km 13,100 to km 12,500 in the south-eastern direction, using DRI's CPX trailer "decibella"; see Figure 40. Prior to measurements, the tyres were brought to normal operating temperatures by driving at least for 15 minutes. Figure 20: Photograph of DRI's CPX trailer "decibella" and the towing van.

49 For the CPX measurements the following values were recorded: The maximum A-weighted noise level L pafmax Third-octave band frequency spectra The vehicle speed GPS coordinates Air temperatures Tyre temperatures During CPX measurements, the air temperature was 11 C. The temperature of tyre A and D was 29 C and 26 C, respectively. Results are reported at the reference temperature of 20 C. Results deviating from the reference temperature were corrected in conformity with Eq. 1 as follows [HARMONOISE, 2004]: Eq. 1 T T Corr, P Corr, H = 0.05 = 0.03 ( Tmeasured o 20) [ C] ; Passenger cars ( T o 20) [ C] ; Heavy vehicles measured 42 Corrections for speed deviations during CPX measurements were carried out as shown in Eq. 2, where v is the measured speed, v ref is the reference speed and B is a speed constant. Eq. 2 Vref L = L + B corr meas log v SPB and CPB noise results After all corrections are done, a logarithmic regression analysis of the relation between the measured noise levels L pafmax and the speed for every vehicle category has been carried out. Then the regression was calculated as follows: L paf max = B log () v + A ; 1.5 std( vp ) v 1.5 std( vp ) ; std( v ) v std( v ) Heavy vehicles where std is the standard deviation of the speed. H H Passenger cars The data points as well as the regression curves and 95 % confidence intervals can be seen in [DRI 68, 2008] for the three vehicle categories included. The results from the SPB measurements are displayed in Figure 41. There was no reference pavement of the same age as the SD pavements at the test site. Since no or few new Danish measurements have been made at DAC 0/11 at 80 km/h, the NORD 2000 reference has been used for comparisons. This represents a pavement with an average age of 9 years so this reference pavement can be expected to have a noise level 1.0 to 1.5 db higher than a new DAC11. Therefore the noise reductions reported have to be lowered by 1.0 to 1.5 db in order to reflect the situation with a new DAC11 reference. This is also the case for the noise reductions for heavy vehicles.

50 Figure 41: Maximum A-weighted noise level (passenger cars) and 95 % confidence interval. Reference speed was 80 km/h and temperature was 20 C LpAFmax [db] m_sds 8/11 SDD 8/11+5/8 s_m_sds 8/11+5/8 m_sds 5/8 1½ m_sds 5/8+2/5 m_sds 2/5 NORD2000 ref. LpA year 0 80,6 80,1 79,2 79,2 77,4 76,7 78,6 LpA year 1 80,5 77,1 Compared to the NORD 2000 reference DAC 0/11 with an age of 9 years for passenger cars, all surface dressings with 8/11 mm aggregate and one with 5/8 mm aggregate yielded higher noise levels of up to 2.0 db (see Figure 41). The two least noisy were the 1½ m_sds 5/8+2/5 and the m_sds 2/5, both yielding a slight noise reduction of 1.2 db to 1.9 db relative to the 9 year old NORD 2000 reference, respectively. The two noisiest pavements were the m_sds 8/11 and the SDD 8/11+5/8. Measurements were repeated at two surface dressings in year 1 (one year after the surfaces were laid), m_sds 8/11 and m_sds 2/5. At the m_sds 8/11 the noise level was the same. At the m_sds 2/5 the noise level had increased 0.5 db. Noise levels at 80 km/h for dual-axle trucks/busses are shown in Figure 42. All surface dressings yielded lower noise levels than the 9 year old reference, in the range of 0.8 db to 2.8 db. The range between the noisiest and the least noisy surface dressing was 2.1 db. The result for dual-axle trucks/busses only gives an indication of the noise levels, because of the low number of pass-bys during measurements. The same 9 year old NORD 2000 surface was used as a reference for multi-axle trucks/busses. All surface dressings yielded lower noise levels than the reference, in the range of 0.1 db to 1.7 db, see Figure 43. The m_sds 2/5, 1½ m_sds 5/8+2/5 and m_sds 8/11 gave the same noise levels. The SDD 8/11+5/8, s_m_sds 8/11+5/8 and m_sds 5/8 yielded noise levels of 86.8, 86.3 and 85.7, respectively. The increase in noise levels from 2006 to 2007 was 0.7 db at the m_sds 2/5 and 0.1 db at the m_sds 8/11. The results from the CPB measurements are shown in Figures Figure 44 shows the noise levels at 50 km/h and 80 km/h for the passenger cars and the dual-axle truck. The m_sds 2/5 yields the lowest level of the two categories at both speeds. For the passenger cars at 80 km/h, only small differences between the SPB and CPB noise levels are seen; in the range of 0 to 0.3 db with the exception of 0.9 db at the m_sds 5/8. When comparing the CPB noise levels to noise levels of the NORD 2000 reference, as shown for passenger cars in Figure 45, three surface dressings yielded lower noise levels than the reference at 50 km/h, yielding a noise reduction of 0.7 db to 3.6 db. At 80 km/h two surfaces gave lower noise level than the reference; in the range of 0.3 db to 2.2 db.

51 Figure 42: Maximum A-weighted noise level (dual-axle trucks/busses) and 95 % confidence interval. Reference speed was 80 km/h and temperature was 20 C LpAFmax [db] NORD2000 ref. SDD 8/11+5/8 m_sds 8/11 m_sds 2/5 m_sds 5/8 1½ m_sds 5/8+2/5 s_m_sds 8/11+5/8 LpA year 0 85,3 84,5 83,5 83,2 82,6 82,6 82,5 LpA year 1 82,2 Figure 43: Maximum A-weighted noise level (multi-axle trucks/busses) and 95 % confidence interval. Reference speed was 80 km/h and temperature was 20 C LpAFmax [db] NORD2000 ref. m_sds 2/5 1½ m_sds 5/8+2/5 m_sds 8/11 SDD 8/11+5/8 s_m_sds 8/11+5/8 m_sds 5/8 LpA year 0 87,4 87,3 87,2 87,1 86,8 86,3 85,7 LpA year 1 88,0 87,2

52 Figure 44: CPB noise levels for passenger cars and dual-axle truck, normalized to 20 C LpAFmax [db] SDD 8/11+5/8 1½ m_sds 5/8+2/5 m_sds 2/5 m_sds 5/8 m_sds 8/11 s_m_sds 8/11+5/8 P 50 km /h 72,5 69,4 68,2 71,1 73,3 71,8 P 80 km /h 80,0 77,4 76,4 78,3 80,7 79,2 L 50 km/h 76,7 75,2 74,7 76,7 76,6 76,4 L 80 km/h 83,3 81,2 80,9 81,7 82,6 81,7 P 50 km/h P 80 km/h L 50 km/h L 80 km/h Figure 45: CPB noise levels compared to the reference level; for passenger cars, normalized to 20 C. The Ref is the 9 years old NORD 2000 reference. 81 LpAFmax [db] SDD 8/11+5/8 1½ m_sds 5/8+2/5 m_sds 2/5 m_sds 5/8 m_sds 8/11 s_m_sds 8/11+5/8 P 50 km/h 72,5 69,4 68,2 71,1 73,3 71,8 P 80 km/h 80,0 77,4 76,4 78,3 80,7 79,2 Ref. 50 km /h 71,8 71,8 71,8 71,8 71,8 71,8 Ref. 80 km /h 78,6 78,6 78,6 78,6 78,6 78,6 P 50 km/h P 80 km/h Ref. 50 km/h Ref. 80 km/h

53 Figure 46 shows the CPB noise levels and the NORD 2000 reference for dual-axle trucks/busses. All CPB noise levels yielded lower noise levels than the reference at 50 km/h and 80 km/h. At 50 km/h the noise reduction was in the range of 3.1 db to 4.6 db, with m_sds 2/5 as the least noisy and SDD 8/11+5/8, m_sds 5/8 and m_sds 8/11 as the noisiest. At 80 km/h the noise reduction was in the range of 2.0 db to 4.4 db, with SDD 8/11+5/8 as the noisiest and the m_sds 2/5 as the least noisy. 46 Figure 46: CPB noise levels and reference, dual-axle trucks/busses, normalized to 20 C. The Ref is the 9 years old NORD 2000 reference. 86 LpAFmax [db] SDD 8/11+5/8 1½ m_sds 5/8+2/5 m_sds 2/5 m_sds 5/8 m_sds 8/11 s_m_sds 8/11+5/8 L 50 km/h 76,7 75,2 74,7 76,7 76,6 76,4 L 80 km/h 83,3 81,2 80,9 81,7 82,6 81,7 Ref. 50 km /h 79,8 79,8 79,8 79,8 79,8 79,8 Ref. 80 km /h 85,3 85,3 85,3 85,3 85,3 85,3 L 50 km/h L 80 km/h Ref. 50 km/h Ref. 80 km/h CPX noise levels CPX measurements were carried out in "year 1" at two surface dressings, m_sds 8/11 and m_sds 2/5. The Danish CPX index CPX DK is shown in Figure47 and calculated as follows: Eq. 3 CPX ( L ) + 0. ( L ) K DK = 0.85 A 15 D + where K is a constant related to the CPX measurement trailer. For DRI's trailer in 2007 K= 0 db at 50 km/h and K= -0.4 db at 80 km/h. The CPX DK reference values at 50 km/h and 80 km/h are also shown in the figure. At 50 km/h the m_sds 2/5 gives a noise reduction of 1.8 db and 3.5 db at 80 km/h. The m_sds 8/11 yields a noise level just below the reference of 0.6 db. According to the Danish noise classification system of surfaces, the m_sds 2/5 is a class C at 80 km/h [Vejregelrådet, 2006][Vejregelrådet, 2007].

54 47 Figure 47: CPX DK levels and reference at 50 km/h and 80 km/h, normalized to 20 C LpAFmax [db] CPXdk 50 km/h CPXdk 80 km/h CPXdk ref. 50 km /h CPXdk ref. 80 km /h m_sds 8/11 m_sds 2/5 CPXdk 50 km/h CPXdk 80 km/h CPXdk ref. 50 km/h CPXdk ref. 80 km/h CPX versus SPB spectra In Figure48, spectra for passenger cars measured with the SPB method are shown. At the test site there was no reference pavement. Therefore, for comparison, the spectra from the new dense asphalt concrete (DAC 11) at Fakse (see Section 5.4.1) has been included as a kind of reference pavement for passenger cars. Figure 48: Third-octave band spectra from 400 to 4000 Hz for passenger cars at 80 km/h. Normalized to 20 C [db] m_sds 8/11 SDD 8/11+5/8 SDS 8/11+5/8 SDS 5/8+2/5 m_sds 2/5 m_sds 5/8 FAKSE DAC Frequency [Hz]

55 48 In the frequency range below 1500 Hz, for passenger cars the surface dressings give significantly higher levels than the DAC 0/11 reference pavement (up to 7 db). This indicates increased vibration-generated noise caused by a rough surface texture. The surface dressings with 2/5 mm aggregate (m_sds 2/5) have nearly the same level of vibrationgenerated noise as the DAC 0/11 pavement. The pavements with the larger aggregate have the highest vibration-generated noise. If a second layer of smaller aggregate are applied, the vibration-generated noise is reduced by 1 to 3 db. In the frequency range above 1500 Hz, where the air pumping noise is dominant, the surface dressings with 11 mm aggregate give 1 to 3 db lower noise levels than the DAC 0/11 pavement. This indicates that the pavements have an open texture, which can to some extend reduce air pumping noise. The surface dressings with 5 mm aggregate give 1 to 2 db higher noise levels above 1500 Hz, indicating that the surface texture is not very open and therefore cannot reduce the air pumping noise. Generally, the surface dressings with the larger 11 mm aggregate give the highest overall noise levels and this is caused by the high vibration-generated noise. The surface dressings with the smaller 5 mm aggregate give the lowest overall noise levels. The same tendencies were found in the Fakse measurements (see Section 2.1). Figure 49 shows the spectra for multi-axle trucks and busses. Because of the low number of dual-axle trucks/busses spectra for the latter are not shown. The picture is different for the multi-axle vehicles than for cars. There is no big variation in the frequency spectra for the heavy vehicles. The surface dressings with 11 mm aggregate give the highest level of vibration-generated noise; 1 to 3 db higher than the pavements with 5 mm aggregate. For the air pumping noise (over 1500 Hz) the surface dressings with the smaller 5 mm aggregate give 1 to 3 db higher noise levels than the pavements with the larger aggregate. Figure 49: Third-octave band spectra from 400 to 4000 Hz for multi-axle trucks//busses at 80 km/h. Normalized to 20 C [db] m_sds 8/11 SDD 8/11+5/8 SDS 8/11+5/8 m_sds 5/8 SDS 5/8+2/5 m_sds 2/ Frequency [Hz]

56 In Figure 50 the CPX DK and SPB and CPB spectra for passenger cars at 80 km/h from m_sds 8/11 and m_sds 2/5 are compared. The SPB and CPB spectra are displayed on the axis on the right side. The dynamic range in both plots is 25 db. At the m_sds 8/11 the resemblance of the SPB and CPX spectral shape is good below 1 khz, while above 1 khz the SPB spectrum yields lower levels. At the m_sds 2/5 the resemblance of the shape is good from 630 Hz and upwards. Spectra from the CPB measurements are also shown. At 50 km/h only the CPX and CPB spectra are compared. At m_sds 2/5 the CPX spectrum seems to be more irregular than at 80 km/h. The resemblance of the shape is good in some parts of the frequency range. See Figure 51. The last comparisons between spectra are shown in Figure 52 where CPX DK spectra at 50 km/h and 80 km/h are compared. CPX DK at 50 km/h is displayed on the right axis. The plots show that vibration-generated noise is more pronounced at lower than at higher speeds Conclusions with regard to the Danish surface dressing experiment SPB measurements showed that at 80 km/h a small noise reduction was achieved for passenger cars of 1.2 db and 1.9 db for the pavements designated 1½ m_sds 5/8+2/5 and m_sds 2/5, respectively. SPB measurement results for multi-axle trucks/busses indicated a noise reduction of up to 1.7 db at 80 km/h. All surface dressings yielded lower or equal noise levels compared to the NORD 2000 reference. This represents a pavement with an average age of 9 years so this reference pavement can be expected to have a noise level 1.0 to 1.5 db higher than a new DAC11. Therefore the noise reductions mentioned have to be lowered by 1.0 to 1.5 db in order to reflect the situation with a new DAC11 reference. This is also the case for the noise reductions for heavy vehicles. Consequently, in the frequency range below 1500 Hz for passenger cars the surface dressings give significantly higher levels than a new DAC 0/11 reference pavement (up to 7 db). This indicates increased vibration-generated noise caused by a rough surface texture. The surface dressings with 2/5 mm aggregate (m_sds 2/5) give nearly the same level of vibration-generated noise as the DAC 0/11 pavement. The pavements with the larger aggregate give the highest vibration-generated noise. If a second layer of smaller aggregate are applied the vibration-generated noise is reduced by 1 to 3 db. In the frequency range above 1500 Hz where the air pumping noise is dominant the surface dressings with 11 mm aggregate give 1 to 3 db lower noise levels than a new DAC 0/11 pavement. This indicates that the pavements have an open texture, which can to some extent reduce air pumping noise. The surface dressings with 5 mm aggregate give 1 to 2 db higher noise level over 1500 Hz indicating that the surface texture is not very open and therefore cannot reduce the air pumping noise. Generally the surface dressings with the larger 11 mm aggregate give the highest overall noise level and this is caused by the high vibration-generated noise. The surface dressings with the smaller 5 mm aggregate give the lowest overall noise levels. For multi-axle trucks and busses there is no big variation in the frequency spectra. The surface dressings with 11 mm aggregate give the highest levels of vibration-generated noise; 1 to 3 db higher than the pavements with 5 mm aggregate. For the air pumping noise (over 1500 Hz) the surface dressings with the smaller 5 mm aggregate give 1 to 3 db higher noise levels than the pavements with the larger aggregate. To get an indication of the acoustical behaviour at 50 km/h, CPB measurements were carried out, indicating that the m_sds 2/5 yields a better noise reduction at 50 km/h than at 80 km/h when compared to NORD 2000 reference, thus a noise reduction of 3.6 db at 50 km/h. At 80 km/h CPB measurement results for dual-axle trucks showed that all surface dressings gave lower noise levels compared to the NORD 2000 reference surface; in the range of 3.1 db to 4.6 db at 50 km/h and 2.0 db to 4.4 db at 80 km/h.

57 Figure 50: CPX DK versus SPB and CPB spectra for passenger cars at 80 km/h, normalized to 20 C. The right half shows data for m_sds 8/11, while the left half shows data for m_sds 2/5. Measurements in "Year 0" LpAFmax [db] CPX SPB CPB LpAFmax [db] CPX SPB CPB Frequency [Hz] Frequency [Hz] Figure 51: CPX DK versus CPB spectra for passenger cars at 50 km/h, normalized to 20 C. The right half shows data for m_sds 8/11, while the left half shows data for m_sds 2/5. Measurements in "Year 0" LpAFmax [db] CPX CPB LpAFmax [db] CPX CPB Frequency [Hz] Frequency [Hz]

58 51 Figure 52: CPX DK at 50 km/h versus CPX DK at 80 km/h, normalized to 20 C. The right half shows data for m_sds 8/11, while the left half shows data for m_sds 2/5. Measurements in "Year 1" LpAFmax [db] CPX 80 km/h CPX 50 km/h LpAFmax [db] CPX 80 km/h CPX 50 km/h Frequency [Hz] Frequency [Hz]

59 Surface dressings with very small aggregates There is another class of surface dressings which may offer low-noise properties, but which could not be tested in this project due to limited resources and too low interest from road authorities. This family of surface dressings can be described as resin bonded aggregate systems based on very small aggregates having very high Polished Stone Value (PSV). They are often glued onto the underlying very smooth pavement by epoxy. These are designed to produce a pavement surface with a high skid resistance which will be maintained throughout the service life of the treatment and mainly used at spots on streets and roads where exceptional skid resistance is needed, such as at and near intersections, or at the end of downhill slopes where vehicles need to stop. These surfaces are generally proprietary. In the U.K. they are often approved in accordance with the type approval system called HAPAS/BBA. Commercial products include, for example: Tyregrip (used in a number of countries, e.g. USA, U.K.) Truegrip Shellgrip EP-grip Epoxy-Durop ShoBond (used extensively in Hong Kong) Italgrip (used in Italy and the USA) The surface typically comprises either a polyurethane binder or a two-component epoxy binder and a graded (1 mm to 3 or 4 mm) calcined bauxite aggregate. In some cases (Italgrip) metal slag is used as aggregate. It is common that some colour, frequently red, is added to make drivers observe the critical spot or lane on the road better. A typical appearance of such a surface is shown in Figure 53. From the surface texture, it can be expected that these surfaces in general should provide a low-noise surface, since the texture has a potential to be near an optimum. However, to reach this goal one must try to achieve the following features: The aggregate shall be glued as close together as possible, with little spacing between chippings The base surface must be as smooth and even as possible in order to provide a firm base onto which the aggregate is bonded, but without being pressed down into the base surface There must not be significant ravelling Strange enough, very few noise measurements have been conducted on such surfaces. In the SILVIA project it was intended to include noise measurements on Italgrip, but this was never realized. The first noise measurement known on such a surface was conducted by the Belgian Road Research Centre and resulted in one of the quietest surfaces ever measured by the Centre [Descornet, 1979]. It was hoped that SILENCE could test such a surface, but with budget cuts this appeared impossible. The Italgrip surface is described in [Sandberg & Ejsmont, 2002]. In later years such a surface has been in operation on a motorway in Wisconsin where it gave remarkable decreases in traffic accidents while providing a small noise reduction. The noise reduction could have been much higher had the surface been laid on a smooth, polished cement concrete rather than on a diamond-ground concrete.

60 53 Figure 53: Typical high-friction surface dressing with very small aggregate. Photo from a street in Okinawa, Japan. Similar surface is common in Tokyo and other Japanese cities. The coin is 22 mm in diameter. The insert at the lower right is a photo of the m_sds 2/5 surface in Figure 38d. The surface tested by DRI in SILENCE designated m_sds 2/5 was an inexpensive attempt to obtain such a surface dressing. However, it failed to give a substantial noise reduction (compared to a DAC 0/11), due to failure to meet the two first features listed above; see Figure 38d and the insert in Figure 53 above. The aggregate was partly pressed down into the basecourse and the binder was pressed up between the chippings. To achieve a more ideal texture, such as in Figure 53, one must have a better basecourse and a more advanced binder, such as epoxy. While the surface dressings tested in SILENCE were not specifically designed for urban areas, the surface dressings described in this section are targeted for "hot spots" on urban streets; albeit not for noise-reducing purposes but for exceptionally high and durable skid resistance. This makes it important to test in some future project such surfaces for noise properties.

61 54 6 Block pavements Deliverable F.D1 reported the WP F1 study made on block surfaces. For completeness, the summary from this deliverable is reproduced below. The reader who has a deeper interest in this particular issue should read the Deliverable [Sandberg et al, 2007]. Block surfaces, made of stones or cement concrete, have a widespread use in today's cities and towns. One major use is to preserve the ancient cultural style of an old city, with its historical cobblestones or stone setts on the streets. Another major use is, as an integrated part of traffic management schemes, to alert vehicle drivers that they drive on an intermodal street; i.e. a street shared by vehicles of various kinds (e.g. cars, busses, bicycles) and pedestrians, where one would like to keep them apart as much as possible. The message conveyed by the paving stones or blocks increases traffic safety and improves the urban environment by making drivers more alert, reducing speeds, and maybe also the traffic volume. A third use is to enhance the aesthetical impression of the street and its environment. In both the latter cases, the type of block pavement used is mostly modern interlocking blocks made of cement concrete. There are many examples where the use of block surfaces has led to complaints from people living or working close to streets with such pavements; complaints that mostly focus on the extra noise generated by the traffic on such surfaces. The authorities then will have to balance the needs for a quieter environment against the needs of preserving the historic and cultural values that these pavements generally represent. The EU-project SILENCE has recognized this sensitive balance and therefore included a subtask F1.1 with the title Noise reduction for paving stone surfaces for streets of high cultural or historic importance. In this subtask investigations of noise from streets with block pavements ("paving stones") were carried out. The objectives were stated as: 1. To produce a ranking in relation to noise of commonly used types of paving stones 2. To develop and test quieter types of paving stones Deliverable F.D1 constitutes the result of this SILENCE subtask as it stood at the time of writing. Experiments had been conducted in Denmark, Slovenia and Sweden to study the mentioned topics; results of which were reported in the Deliverable. There is also a section providing a review of results of similar studies presented in the open literature. To achieve the first objective, experiments were carried out in Denmark. Five streets, each with different types of block pavements in the Copenhagen area, were selected for noise measurements according to ISO "Measurement of the influence of road surfaces on traffic noise - Part 1: Statistical Pass-By method" (popularly referred to as "the SPB method"). The study included one type of cement concrete blocks and four types of granite setts, with different levels of surface texture and roughness. To achieve the second goal a special type of granite blocks, developed by the municipality of Copenhagen (one of the setts mentioned above), was applied on a street in Copenhagen in the summer/autumn 2005 and tested within SILENCE. From the Danish study, the following simple, practical guidelines for design of paving stone and other block surfaces having as favourable acoustic properties as possible are offered: The individual blocks should have as even (plane) a surface as possible The joints between the blocks should be as narrow as possible The blocks should have a uniform size in order to ensure the same width of the joints and by this making it easier to minimize the width It is better to have an angle of 45 than 90 between the direction of the joints and the driving direction of the vehicles on the road

62 In general, the stone setts were noisier than the flat concrete or granite blocks. The best noise performance in this investigation was achieved by using flat concrete blocks or flat granite blocks. The difference between the noisiest and the quietest block surface was found to be 10 db. Compared to the NORD 2000 reference, as practised in Denmark and later to be used in the five Nordic countries, all pavements showed a reduction in high frequency (air pumping) noise with the exception of the large and old stone setts. When using traditional stone setts as compared to modern concrete or flat granite blocks, the low-frequency vibration-excited noise is substantially increased. Considering the Danish results and measurements regarding the acoustical properties of block surfaces reported in the literature; i.e. the influence on traffic noise emission, it is clear that there are substantial differences between existing block surfaces. The range between existing ones would be at least db and if the futuristic interlocking block pavement with poroelastic cover is included it will be at least db. These are dramatic differences. In cases where one can accept to surrender some of the historic/cultural values, it would be possible to replace the older type of stone setts with visually rather equal blocks with a flatter surface; such as the flat granite setts tried in Copenhagen and reported here. Depending on the old type of surface, and the orientation of the blocks, one may gain some 4-8 db in this way. This type of surface would not be noisier than an ordinary asphalt surface, and it may then be preferred to use this one rather than to pave the street with asphalt, and in this way save some of the cultural value of the street. In general, the orientation of the blocks should be such that the tyres roll over them at an angle of o rather than hitting the blocks and the joints at right angles. The use of interlocking blocks would generally not generate more traffic noise than an ordinary asphalt surface. With an appropriate design and good workmanship in the laying, as well as a stable sand bedding or other base material, one may even obtain a certain noise reduction compared to a dense asphalt surface; say about 1-2 db. The use of a porous layer on the top of the blocks would (tentatively) improve the acoustic qualities a little extra, with a potential noise reduction in comparison to dense asphalt surfaces (DAC 0/11 and SMA 0/11) of at least 2-3 db; although this benefit is likely to diminish by time when clogging occurs. The ultimate choice would be an interlocking block surface with a cover of poroelastic material of the type described in the Deliverable F.D1. With such a surface one would obtain a traffic noise reduction corresponding to the best that can be offered by the most advanced (double-layer) porous asphalt surfaces. However, compared to the porous asphalt surfaces, the one with blocks and poroelastic cover would most probably be able to keep its low-noise properties for a longer time due to the elasticity of the surface which would prevent dirt from getting stuck in the voids. It is important, however, to provide for a very stable base material to lay the blocks into. It should be observed that tear and wear, stability problems, as well as poor maintenance, might increase the noise emission from stone setts and other block pavements over the years. The Deliverable F.D1 could therefore only provide some "snapshots" of the overall situation. Nevertheless, it is hoped that the Deliverable F.D1 will advance the knowledge about this family of road surfaces. 55

63 56 7 Experiments with a poroelastic cover on a block surface 7.1 Introduction A poroelastic road surface is a surface which combines a high degree of elasticity (by using a high proportion of rubber or similar material) and a high degree of porosity; i.e. air voids. It was proposed in [Sandberg & Ejsmont, 2002] that the term poroelastic road surface (PERS) shall only be used for surfaces with a design air voids content of no less than 20 % and a proportion of rubber of at least 20 % by weight. In the Q-CITY project, the term has been used for surfaces with as low air voids content as % (which are not really porous surfaces) and rubber content of around 10 % by weight (which does not feel elastic when walking on it). The authors think that this wider use of the term only makes it confusing, since such surfaces are not very different from conventional surfaces. The PERS provides an efficient combination of noise-reducing measures: (1) sound absorption and elimination of air pressure gradients by its high air voids content, (2) a smooth surface giving low vibration excitation to the tyre, (3) low contact angles in the leading and trailing edges of the tyre/road interface, and (4) a soft, elastic impact between tyre tread and road surface elements. If the hysteresis losses in the deflection of the system can be made low (by material selection and design) the rolling resistance losses of such a system may be low or at least not higher than for a conventional tyre/road contact. The EU project SILVIA included a large part in which trials with PERS were conducted; see [Sandberg et al, 2006]. In February 2008, a consortium led by FEHRL institutes proposed a new project to the EU Commission named PERSUADE, which would focus on advanced development of PERS. One of the tasks in SILENCE WP F1 was to construct a block pavement with a soft and thus quiet cover. For this purpose, a poroelastic road surface has been selected as the soft cover. The projected use of such a surface would be in urban or suburban streets where noise emission is a major problem while at the same time one would prefer to have a surfacing which stands out as "unusual" in order to notify vehicle drivers of the need to drive softly. Since in such a location, homes and businesses are likely to be close to the street, also aesthetical properties would be important. Normally, in such a location one would choose to use interlocking blocks to meet the two latter purposes (unusual surface with pleasing visual appearance), but the noise problem would still be there. The design described in the following is intended to take care of also the first problem, i.e. the noise emission. During the course of this project, it was found that the EU project NR2C ("New Road Construction Concepts"; see had partly similar aims as SILENCE WP F1 and it was decided to have a cooperation between the two projects, aiming at producing and testing a couple of interlocking block pavements with PERS cover. 7.2 Problems under investigation Two particular and major problems of PERS and block pavements were selected for study in SILENCE, namely: The wet skid resistance must be maintained at an acceptable level for a reasonable operating time for a typical urban street condition The stability of the system of blocks and stabilizing layer must be sufficient for a reasonable operating time and for a mix of light and heavy vehicles typical of an urban street Both of these problems must be solved before a block pavement with PERS cover can be used in actual traffic. These were focused on in the experiments conducted by VTI and ZAG within this common programme, where VTI focused on the first problem and ZAG focused on the second one. The block pavement with soft cover which was studied was named "SOFTBLOC" by VTI and ZAG.

64 7.3 Experiments in NR2C and SILENCE - Description As mentioned in Deliverable F.D1 all field tests with poroelastic road surfaces has either been prematurely terminated due to lack of adhesion between the underlying surfaces or due to wet friction properties being below regulatory limits at some point of the tests. The one exception is the trial in Japan with a block pavement covered with poroelastic material. In that case the test was interrupted due to severe rutting and pavement blocks becoming displaced. Insufficient drainage of the stabilizing sand layer immediately below the paving blocks was a plausible cause for the rutting. However, even if this surface had not been forced to be removed due to the rutting, it would most probably not maintain a sufficient wet friction after some months of operation, as the problem with durability of the frictional properties had not been solved at that occasion. A laboratory test scheme was set up to study means for having an interlocking block surface covered with a poroelastic material that (together with the blocks) is both resistant to rutting and has a durable wet friction. For improving the frictional properties, a test programme was conducted to study the influence on the durability of friction of binder hardness, addition of silicon carbide to the mix and pre-treatment of the rubber crumb. For simulation of the polishing effect of rolling tyres on a road surface, the VTI Pavement Testing Machine was used. This pavement testing machine is a circular track with a diameter of approximately 5 metres. Four or six wheels are rolling on this surface at a speed up to 70 km/h. See Figure 54. The power to the traction of the wheels is delivered by electrical engines attached on each of the wheel axles. The load on each wheel is adjustable but is usually fixed at 450 kg. The wheels do not follow a circular track; rather there is a latter movement of the wheels to simulate a realistic distribution of wheel paths. The test can be made in dry or wet conditions. When wet conditions are used, fresh water is constantly sprinkled over the surface. The evolution of the friction was tested with the British Pendulum method and by a manually pushed equipment developed at VTI which measures the friction on a partially slipping wheel pushed at a speed of approximately 5 km/h. Figure 54: The Pavement Testing Machine ("PVM") at VTI. 57

65 For the test, the test track in the pavement testing machine was divided in five different sections. In each section, a different type of mix was produced. After curing and initial check of friction, stiffness and permeability, the traffic polishing commenced in wet conditions. At regular intervals the wet friction and rutting were measured. The test continued until approx tyre passages had occurred, corresponding to some years of traffic polishing. Figure 55 shows manufacturing of one of the test samples. 58 Figure 55: Manufacturing of test samples of PERS at VTI. The mixes listed in Table 16 were produced. Two plates, A and B, were produced of each mix type and mounted in the PVM for polishing and wear tests. In many of the previous experiments the lack of adhesion between the poroelastic layer and base-layers was the reason for failures. This is also the reason for choosing a pavement structure where poroelastic layer is laid on concrete (paving) blocks. In this way the critical gluing can be made in a more controlled and proper manner and environment. Taking 4 into account problems and reasons for failure of previous test fields in Japan, the project participants identified and selected a keypoint topic for further research work at the ZAG institute 5 ; namely the foundation for the blocks (bedding layer) has to be improved, intending to avoid the Japanese failure. 4 The section about the experiments at ZAG in Slovenia has been written by Mr Darko Kokot at ZAG 5 ZAG Ljubljana (Zavod za gradbenistvo Slovenije) is Slovenia's national building and civil engineering institute

66 59 Table 16: Mixes produced for testing in the PVM at VTI. Number & Name Rubber Binder Rubber content % (w/w) Binder content % (w/w) Silicon carbide content % (w/w) 1 Rosehill, abrasive Not available Not available Not available Not available Not available 2 Hard, reference Un-aged Hard, two component Hard, abrasive Un-aged Hard, two component Hard, abrasive and aged Aged Hard, two component Soft, abrasive Un-aged Soft, one component To solve problems with displacement of the intermediate layer of asphalt-coated sand on which the interlocking blocks in the Japanese tests were placed, ZAG focused on the bedding layer onto which the cement concrete blocks are placed. The entire pavement structure will be as strong as possible to avoid bearing capacity failures, but it will also be as traditional as possible. Except for the poroelastic layer, the pavement structure tested in Slovenia was a traditional pavement structure for stone/cement concrete blocks paved on roads in Slovenia, and for preparing it only local materials are used. The testing structure was prepared in a wooden mould, where each pavement layer was placed and compacted. This procedure can be seen in Figures Four different systems were tested, as illustrated in Figure 60. First, ZAG performed tests on a structure where cement concrete blocks with a glued poroelastic overlay were placed onto a sand bedding layer. The second test system was similar but included watering of the structure. The third and fourth systems were similar to the first two ones, except that the cement concrete blocks were placed onto a cementitious screed bedding layer. In total there were four test cycles. One cycle consisted of a piston-induced dynamic loading of the test structure, as can be seen in Figure 61. There were 100,000 vertical loadings applied through a heavy vehicle tyre, each time of 3 tons. The loading equals 100,000 passes of 12 ton axle load. The vertical displacement of the structure was monitored and followed by three LVDTs, mounted on a framework and placed along the tyre. Field testing could not be conducted within the programme due to time and budget constraints.

67 60 Figure 56: The unbound layer is placed in the mould. Figure 57: Compacting the asphalt layers.

68 61 Figure 58: Placing cement concrete blocks with poroelastic overlay. Figure 59: The test construction is finished.

69 Figure 60: The four different systems tested. 62 poroelastic overlay cement concrete blocks sand, bedding course 5 cm asphalt layers cm unbound layer 30 cm A) Construction with sand bedding course B) Sand bedding course, adding water poroelastic overlay cement concrete blocks cementitious screed, bedding course 5 cm asphalt layers cm unbound layer 30 cm C) Construction with screed bedding course D) Screed bedding course, adding water Figure 61: Test assembly with a sketch showing the testing principle.