Traffic noise at rumble strips

Transcription

1 Traffic noise at rumble strips - Inter noise paper 2007 Danish Road Institute Report Ministry of Transport and Energy

2 Road Directorate Guldalderen 12 DK-2640 Hedehusene Denmark Telephone Telefax Roadinstitute.dk Title Traffic noise at rumble strips - Inter noise paper 2007 Dated September 2007 Author Jørgen Kragh, Bent Andersen, Sigurd N. Thomsen Published by Road Directorate, Danish Road Institute Copyright Road Directorate, All rights reserved Photo Bent Andersen Print Electronic ISSN electronic ISBN electronic Reports published by the Danish Road Directorate can be requested from the bookshop: Telephone Telefax schultz@schultz.dk

3 Traffic noise at rumble strips - Inter noise paper 2007 Jørgen Kragh Bent Andersen Sigurd N. Thomsen Danish Road Institute Report Ministry of Transport and Energy

4 xx

5 Contents Preface... 3 Forord... 4 Summary... 5 Resumé Introduction Delimitation Rumble strips Method Selected cars Measurement procedure Road surface Data analysis Results Comments Immediate observations Trends for different cars Correction for the distance from the microphone Modified results Model and interpretation Noise level further from the road Conclusions References

6 2

7 Preface When a vehicle tyre hits a rumble strip of the types applied hitherto the noise level suddenly increases. This applies inside the vehicle as well as in its surroundings. The Danish Road Directorate in co-operation with a Danish county road administration (Frederiksborg Amt) has worked on developing rumble strips giving rise to less noise in the environment while at the same time warning drivers sufficiently. In a pilot study sinusoidal indentations in the pavement were tested along with a recent Swedish proposal applying indentations shaped as a section of a cylinder, as well as some existing rumble strips with rectangular indentations. 3

8 Forord Når en bils dæk rammer en rumlerille af de typer der for tiden anvendes giver det anledning til en pludselig forøgelse af støjniveauet, både inde i bilen og i omgivelserne. Vejdirektoratet og Frederiksborg Amt har arbejdet med at lave rumleriller der giver mindre støj i omgivelserne men alligevel tilstrækkelig varsling af chaufførerne. I et pilotforsøg blev der afprøvet forskellige sinusformede rumlestriber, et nyt svensk forslag samt nogle eksisterende rumleriller. 4

9 Summary A regional Danish road administration (The County of Frederiksborg) has established rumble strips in the middle of a few two-lane rural roads, a total of five different types: 1) maximum 10 mm deep segments of a circle per 0.6 m, 2) maximum 7 mm deep sinusoidal shape of 0.6 m wavelength, 3) maximum 4 mm deep sinusoidal shape of 0.6 m wavelength, 4) 8 mm deep rectangular shape (per 0.33 m), and 5) 4 mm deep rectangular shape (per 0.33 m). Three different passenger cars were driven repeatedly in both lanes at a speed of approximately 80 km/h with their left wheels in the rumble strips. During each pass-by the maximum noise level was recorded by a microphone placed 7.5 m from the middle of the road. The vehicle speed was recorded as well. For comparison, the same vehicles were passing the microphone at rumble strip No. 3 driving on the pavement, 22 years old stone mastic asphalt (SMA), without having their left wheels in the rumble strip. The main results of the measurements and subsequent analyses are given in the table below. The numbers are the estimated increments of the overall A-weighted noise level with time weighting F when driving with one set of wheels in the rumble strip as opposed to driving with both sets of wheels on the reference pavement. The increase is given in relation to the noise level at an old stone mastic asphalt surface and is valid at distances exceeding 25 m or so from the road. Table 1. Increment of noise levels from passenger cars driving with one set of wheels in the rumble strip, relatively to the noise level from a pass-by with all wheels on the SMA pavement, in the nearest lane and in the far lane, respectively. Rumble strip No. [-] Designation Far lane Near lane 1 Segment of circle, max. 10 mm deep Sinus 7 mm Sinus 4 mm Rectangle 4 mm deep Rectangle 8 mm deep 8 5 The sinusoidal strips led to db increase in noise level while the circlesegment strip gave an increase of 2 3 db. The rectangular strips gave rise to significantly higher noise levels (3 7 db) than the sinusoidal strips as well as significantly higher noise levels (2 5 db) than the circle segment strip. 5

10 Resumé Frederiksborg Amt har fået fræset rumleriller midt på vejen, i alt 5 forskellige typer: 1) maks. 10 mm dybt cirkelafsnit pr. 0,6 m ( svensk ), 2) max. 7 mm dyb sinus med bølgelængde 0,6 m, 3) maks. 4 mm dyb sinus med bølgelængde 0,6 m, 4) 8 mm dybt rektangel (pr. 0,33 m) ( gammel ) og 5) 4 mm dybt rektangel pr. 0,33 m ( gammel ). Tre forskellige personbiler kørte forbi gentagne gange i begge retninger med ca. 80 km/t med det venstre sæt hjul i rumlerillen. Ved hver forbikørsel blev støjniveauet målt i et punkt ved vejsiden, 7,5 m fra midten af kørebanen, og farten blev registreret. Til sammenligning kørte de samme biler forbi på belægningen ved rumlerille nummer 3 (en 80 kg SMA fra 1984) uden at have det venstre sæt hjul i rillen. Hovedresultaterne af støjmålinger og analyser er vist i Tabel 1. Tallene angiver den beregnede forøgelse af det A-vægtede støjniveau med tidsvægtning F i en afstand på mere end 25 m fra vejen ved at køre med det ene sæt hjul i rumlerillen i stedet for at køre med begge sæt hjul på referencebelægningen. Forøgelsen gælder relativt til støjniveauet ved en gammel SMA-belægning. Tabel 1. Forøgelsen af støjniveauet fra personbiler, der kører forbi med det ene sæt hjul i rumlerillen, dels i den fjerneste, dels i den nærmeste vognbane, relativt til kørsel med begge sæt hjul på referencebelægningen. Rille nr. [-] Betegnelse Fjern Nær 1 Cirkelafsnit, 10 mm dybt( Svensk ) Sinus 7 mm Sinus 4 mm Rektangel 4 mm dybt ( Gammel ) Rektangel 8 mm dybt ( Gammel ) 8 5 De sinusformede rumleriller gav en forøgelse i støjniveau på 0,5 1 db og de svenske en forøgelse på 2 3 db. De gamle rumleriller på Isterødvejen gav væsentligt højere støjniveauer (3-7 db) end de sinusformede rumleriller og væsentligt højere støjniveauer (2-5 db) end de svenske rumleriller udformet som cirkelafsnit. 6

11 1. Introduction Rumble strips in the middle of roads are in some cases established to improve traffic safety. Such strips assist in preventing vehicle drivers from crossing the road centre line without noticing it. They create noise/vibration in the vehicle to warn drivers, but at the same time the noise level outside increases and this may cause annoyance to road neighbours. The Danish Road Institute (DRI) conducted a pilot study in 2006 to test rumble strips giving rise to low noise levels in the environment. 7

12 2. Delimitation Only noise levels from passenger cars at 80 km/h have been investigated. The effect of the rumble strips on heavy vehicle noise levels is not known. The warning effect on drivers, including noise and vibration levels inside the car, has not been investigated by DRI but we know that vibration inside a car has been measured by DanCrash, a Danish consultant. Drivers involved in the pilot study agreed that noise/vibration in their vehicle when driving on the tested rumble strips would give sufficient warning. 8

13 3. Rumble strips Table 1 shows some characteristics of the rumble strips tested in the study and Figure 1 Figure 3 illustrate the profiles of rumble strips No. 1, No. 2 3, and No. 4 5, respectively. In 2004 Danish road authorities had rumble strips milled on both sides of the centre line of some two-lane rural roads. Rectangular indentations were milled, 4 mm and 8 mm deep, respectively, per 0.33 m. These rumble strips (named No. 4-5 in the following) gave rise to noise complaints. Swedish road authorities have decided to mill rumble strips with cylinder segment shaped indentations per 0.6 m. Such a rumble strip was milled, called No. 1 in the following. Table 2. Summary of rumble strip characteristics. Rumble strip No. Indentation [-] 1 Segment of cylinder, max. 10 mm deep 2 Sinus 7 mm top to bottom 3 Sinus 4 mm top to bottom 4 Rectangle 4 mm deep 5 Rectangle 8 mm deep In a British investigation three measurement series have been carried out [1] on sinusoidal rumble fields with various wavelengths and amplitudes. The conclusion was that the best warning effect would be obtained by applying a waveform generating an excitation frequency of 37 Hz. The British experiments were made at 30 mph 48 km/h and the corresponding wavelength was 0.36 m. In the final measurement series rumble fields with 35 cm wavelength and amplitudes of 4.14 mm and 6.62 mm, respectively, were used [1]. At a speed v [m/s] and with a wavelength [m] the car is excited by a frequency f [Hz]: v f or v (1) f At v = 80 km/h = 22 m/s frequency f = 37 Hz is obtained at a wavelength = 22/37 = 0.6 m. Sinusoidal rumble strips were milled with this wavelength and with amplitudes ±3.5 mm and ±2 mm, respectively. 9

14 The actually measured surface profiles of rumble strips No. 1 3 are shown in Figure 4. Due to dense traffic, the surface profiles of rumble strips No. 4 5 could not be measured. " " v h k Figure 1. Cylinder segment ; strip No. 1; h 0.01 m; k 0.15 m; strip width 0.30 m; = 0.6 m ,0 0,5 1,0 [m] 1, [mm] 7 mm / 0.60 m [mm] 4 mm / 0.60 m -8 Figure 2. Sinusoidal; strip No. 2 (top) and No. 3 (bottom). " " k Figure 3. Rectangular ; strip No. 4 and No. 5; = 0.33 m; k = 0,1 m. 10

15 4. Method The measurements were Controlled Pass-By measurements (CPB). 4.1 Selected cars Three passenger cars were selected based on advice from DanCrash. Information on the cars is summarised in Table 2. Table 3. Data on the selected cars and their tyres. Car No Make Volkswagen Skoda Toyota Model Golf 1,8 Octavia 1,9 TDI Combivan (Corolla Verso) Year km Tyres Gislaved Speed 516 Front Tyres Rear 185/60 R14 82T Michelin Energy 185/60 R14 XT2 Continental ContiEcontact 3 195/65 R15 91H Goodyear Ultragrip 6 (M+S) 195/65 R15 35 [mm] [m] , , , , , [mm] [m] , , , , , [mm] , , , , ,5 15 Figure 4. Extracts of surface profiles of rumble strip No. 1 (top), No. 2 (middle) and No. 3 (bottom) measured using DRI laser equipment. [m] 11

16 4.2 Measurement procedure A microphone was placed on one side of the two-lane road at a distance of 7.5 m from the road centre line and at a height of 1.2 m above the road surface, cf. Figure 6. The drivers attempted to pass the measurement positions at a constant speed of 80 km/h. No vehicles passed at the same time in the opposite lane. During each pass-by the maximum noise level was recorded and the vehicle speed was measured with radar. Noise levels were recorded during 3 5 undisturbed pass-bys of each car in each direction with their left wheels on the rumble strips. When driving in the near lane the left wheels rolling on the rumble strip were screened from the microphone by the car body while these wheels were directly visible during pass-by in the opposite direction in the far lane. For comparison, the same vehicles passed the microphone at rumble strip No. 3 driving on the pavement, a 22 years old stone mastic asphalt (SMA) without their left wheels touching the rumble strip. 4.3 Road surface The road surface at rumble strips No. 1 3 was a 22 years old 80 kg stone mastic asphalt, probably SMA 11. Rumble strip No. 4 was milled in a 6 years old thin layer surface BBTM 8 (Danish designation TB 8k) while No. 5 was milled in a 6 years old asphalt concrete AC 11o. The reference section chosen at rumble strip No. 3 was 22 years old SMA Data analysis Measured pass-by noise levels L pafmax were corrected from the actual speed v to 80 km/h by adding -37 log 10 (v/80) which is the basic relation between maximum noise level and vehicle speed in Nord2000, the new Nordic prediction method for road traffic noise [2]. These corrected noise levels were averaged and a statistical uncertainty u of the average was calculated: u s n, s the standard deviation of the mean; n = the number of pass-bys. 12

17 5. Results The detailed results can be found in [3] while a summary is given in Figure 5. The main results of the measurements and subsequent analyses including the corrections and model calculations mentioned in Section 6 and 7 - are given in Table 4. The numbers in the table are the increments of the overall A-weighted noise level when driving with one set of wheels on the rumble strip as opposed to driving with both sets of wheels on the reference pavement. These increments are given in relation to the noise level at an old stone mastic asphalt surface and they are valid at distances exceeding 25 m or so from the road. Some observations are given in the following section. Table 4. Increment of noise levels from cars driving with one set of wheels on the rumble strip, relatively to the noise level from a pass-by with all wheels on the SMA pavement, in the near and far lane, respectively. Rumble strip No. [-] Indentation Far lane Near lane 1 Segment of cylinder, 10 mm deep Sinus 7 mm top to bottom Sinus 4 mm top to bottom Rectangle 4 mm deep Rectangle 8 mm deep

18 6. Comments 6.1 Immediate observations A detailed analysis of the data summarized in Figure showed that when driving in the far lane the noise levels at the sinusoidal rumble strips No. 2 3 were less than 1 db higher than the noise levels at the reference section, while driving on rumble strip No. 1 the noise levels were almost 3 db higher and at rumble strip No. 4-5 the noise levels were 5 6 db higher than at the reference. Based on these immediate observations noise levels seemed to decrease with 5 db or so by using a sinusoidal shape rather than a rectangular indentation. For pass-bys in the near lane the differences were smaller and the noise levels from pass-bys on the sinusoidal rumble strips were lower than the levels from pass-bys on the reference surface. This is discussed below. Further analyses led to the final results shown in Table Trends for different cars The results in [3], when looked at in detail, indicate that complicated interactions between road surface, rumble strip, tyre and car model determine the resulting noise levels. To deduce these relationships would require more measurements and more thorough analyses than has been possible here. 6.3 Correction for the distance from the microphone When driving in the near lane the cars were nearer to the microphone when passing on the reference pavement than when passing with their left wheels on the rumble strip. When driving in the far lane the cars were further away from the microphone when passing on the reference pavement than when passing with their left wheels on the rumble strip. These differences in distance were corrected for by subtracting 0.8 db from the reference noise level from the near lane and by adding 0.5 db to the reference noise level from the far lane [3]. 6.4 Modified results The measurement results seemed at first to be inconsistent. At rumble strip No. 3, for example, the pass-by noise in the far lane rumble strip was higher than the reference noise level while the pass-by noise level in the near lane was lower than the reference noise level. The reference measurements were made with exactly the same microphone position and imme-diately subsequent to the rumble strip measurement. The trend is present in the results both with and without correction for the difference in distance from the microphone to the vehicle: the noise level on the rumble strip was lower than the reference in the near lane and higher than the reference in the far lane. The difference is significant and this led to further analyses. 14

19 86 84 Far lane R.1 R.2 R.3 Near lane R.1 R.2 R.3 Figure 5. Mean pass-by noise levels ± uncertainty. X-axis label: a.b = rumble strip No. a and car No. b. R = reference (i.e. R.3 = Car No. 3 on the reference pavement). Table 5 shows the average of the pass-by noise level for the three cars on each of the rumble strips relative to the (corrected) average pass-by noise level on the reference surface. The table also shows the difference between these characteristic numbers which is used for the model considerations in Section 7. 15

20 Table 5. Average increase in noise level from the three cars on each rumble strip, and difference in increments. Rumble strip No. [-] Far lane Near lane Difference Near - Far 1 2,8 0,5-2,3 2 0,8-0,2-1,0 3 0,9-0,8-1,7 4 5,8 1,4-4,4 5 6,4 1,9-4,5 16

21 7. Model and interpretation A simple model was used to account 1) for the difference in distance between pass-bys on the rumble strips and pass-bys on the reference next to the rumble strips, and 2) for the seemingly inconsistent results mentioned in Section 6.4. The model is illustrated in Figure 6 which shows the positions of the wheels during pass-by. The wheels nearest to the road centre line are at a distance x from this line and each wheel is represented by a noise source at the road surface. The wheels on the rumble strip are assumed to emit z db more sound power than the wheels on the reference surface. The source at the wheels on the far side of the car as seen from the microphone is assumed to be y db screened by the car body and the nearest wheels. Using this model, the theoretical difference in noise level when driving in the near and far lane was calculated as illustrated in Figure 7. The figure presupposes x = 0.5 m and y = 3 db. These numbers were found to fit the results from the reference measurements where z 0 db by the model definition. The model is not suited, however, for estimating y, because y is a characteristic of the lower of two noise levels which are added. Figure 7 shows that the pass-by noise level in the far lane becomes higher than the pass-by noise level in the near lane when z is large ( Near Far is negative). In this situation, the noise from the wheels on the rumble strip dominates over the noise from the other set of wheels. The noise source on the rumble strip is screened when the vehicle is in the near lane and un-screened when the vehicle is in the far lane. 2,4 2,0 1,6 1,2 0,8 Microphone CL Near lane Far lane 0 db +z-y db +z db -y db 0,4 0,0 0 5 x x 10 Figure 6. Vertical section sketching the wheel positions when driving in the two directions, cf. text. 17

22 As mentioned earlier the pass-by noise at rumble strip No. 3 was higher than the reference noise level in one direction while it was lower than the reference noise level in the opposite direction. The above model cannot reflect this. One possible explanation could be different noise emission (different values of z) in the two directions. This does not seem probable, however, based on measured rumble strip texture spectra in the two directions. These spectra are summarized in Figure 8 on the last page of the present report. Three graphs compare the spectrum from the right (H) side of the road (= the far lane) and from the left side (V) (= the near lane) while the fourth graph in the figure shows the average spectrum from the near and far lane of each of the three rumble strips. For wavelengths smaller than 100 mm or so all spectra are rather uniform with a texture level of db re. 1 m. At 630 mm wavelength there is a peak corresponding to the 0.6 m wavelength of the sinusoidal indentations and the distance from cylinder segment to cylinder segment. At rumble strip No. 1 lower peaks were found at 315 mm and 200 mm. Although there is some uncertainty as to the absolute texture levels [3], the authors believe the comparison between the surface texture in the near and far lane to be useful. Another possibility could be that tyre/road noise levels were different for wheels rolling in the wheel track as opposed to rolling on the surface between the wheel tracks. This latter was assumed in the calculation made as a basis of Table 5. Based on trialand-error calculation, the emitted tyre/road noise has been assumed to be 1.2 db lower between the wheel tracks than in the wheel tracks. Later it became possible to carry out a CPX measurement which revealed for the near lane a 0.7 db lower noise level and in the far lane a 0.4 db lower noise level from tyre A when rolling between the wheel tracks than when rolling in the right wheel tack. 18

23 6 5 4 Near - Far Increment z Figure 7. Calculated difference between the pass-by noise level in the near and far lane as a function of the increment z in tyre/road noise level from the rumble strip. x = 0.5 m; y = 3 db. In Table 6, columns 2 3 show for each rumble strip the numerically smallest and largest difference ( Min, Max ) between the measured noise levels from each of the three cars when passing in the near and far lane, respectively. Columns 4 5 show the corresponding minimum and maximum value of z determined by means of a modified version of Figure 7 assuming the above mentioned 1.2 db lower noise levels from wheels between the wheel tracks than from wheels in the wheel tracks. The range of z-values is largest for rumble strips No. 1, 4 and 5 where z is largest. Based on the ranges given in Columns 4-5, a trial-and-error calculation was performed by varying z to obtain the best fit between on one hand 1) the measured and calculated difference between the pass-by levels on the rumble strips in the near and far lane, and on the other hand 2) the calculated and measured increment in pass-by noise levels on the rumble strip as compared to the reference pass-by noise levels in the near and far lane, respectively. 19

24 Table 5 shows in Column 6 the finally chosen representative value of z per rumble strip. A detailed inspection of the table reveals that the correspondence between measured and calculated differences is not perfect. This is hardly to be expected either, since the presupposed identical tyre/road noise emission in both directions, on the rumble strip, in the wheel tracks and between the wheel tracks, respectively, is probably too simple. 20

25 8. Noise level further from the road The finally chosen representative z-values from Table 6 have been used in the modified model from Section 7 to calculate the noise levels at further distance from the road, i.e. above 25 m or so where the distance from all vehicle noise sources are approximately equal. The results are given in Table 7. These results have been expressed as the increment in pass-by noise level compared to the noise level from a pass-by on the reference surface. Table 6. Measured and calculated differences between 1) near and far lane pass-by noise levels and 2) rumble strip and reference pass-by noise levels, cf. text and [3]. Strip No. Measured Near Far z Calculated Rumble re. Ref. Measured Rumble re. Ref. Calc. Meas. Col Min Max Min Max Repr. Far Near Far Near Far Near Infinite Table 7. Increment in pass-by noise level in the far lane and near lane, respectively, relatively to the pass-by noise level on the reference pavement. Rumble strip No. [-] Far lane Near lane

26 9. Conclusions The rumble strips with sinusoidal shape led to only db increase in noise level while the rumble strip with cylinder-segment indentations gave an increase of 2 3 db. Rectangular indentations gave rise to significantly higher noise levels (3 7 db higher) than the rumble strips with a sinusoidal profile as well as significantly higher noise levels (2 5 db higher) than the cylinder segment strip. These increments in pass-by noise level are relatively to the noise level from pass-bys on old stone mastic asphalt and they are valid at distances exceeding 25 m or so from the road. 22

27 10. References [1] G. Watts et al., Optimisation of traffic calming surfaces, Proceedings Internoise 2001, The Hague, [2] J. Kragh et al., Nord2000 Users Guide, DELTA Technical Report AV 1171/06, Hørsholm, [3] J. Kragh, B. Andersen, Traffic noise at rumble strips a pilot study (in Danish), Danish Road Institute / Road Directorate Note 51/

28 70 [db re. 1 m] 70 [db re. 1 m] 65 H_ V_ H_ V_ [mm] [mm] , , , , [db re. 1 m] 70 [db re. 1 m] 65 H_0-50 V_ No.1 No.2 No , ,5 10 [mm] , ,5 10 [mm] Figure 8. Texture spectra for the rumble strips. Abscissa = texture wavelength in mm. Ordinate = texture level in db re. 1 m. H = Near lane; V = Far lane; Rumble strip No. 1 = Top left. No. 2 = Top right; No. 3 = Bottom left. Bottom right: Average texture spectra from the near and far lane for rumble strips No. 1, No. 2 and No. 3, resp. 24

29 Rapport / Report Nr. No. Titel/Title/Shortcut Forfatter/Author 139 Holdbarhed af Drænasfalt Carsten Bredahl Nielsen Jørn Raaberg Erik Nielsen 140 Indbygning af skrivekridt - Et fuldskalaforsøg 141 Noise reducing pavements - State of the art in Denmark 142 Economic assessment of traffic noise in planning Danish experiences 143 Organising urban noise abatement - New ideas 144 Two-layer porous asphalt - for urban roads 145 Thin noise reducing pavements - Experiences Sten Thorsen Poul Panduro Knud A. Pihl Hans Bendtsen Bent Andersen Lars Ellebjerg Hans Bendtsen Hans Bendtsen Lene Nøhr Michelsen Brian Kristensen Hans Bendtsen Bent Andersen Jørn Råberg Lars Ellebjerg Jørgen Kragh Hans Bendtsen Bent Andersen 146 Cost-benefit analysis on noise-reducing pavements Lars Ellebjerg 147 Traffic management and noise - INTER-NOISE 2006 Hans Bendtsen Lars Ellebjerg 148 Noise reducing thin layers for highways - INTER-NOISE 2006 Hans Bendtsen Sigurd N. Thomsen 149 Noise reducing thin pavements urban roads Sigurd N. Thomsen Hans Bendtsen Bent Andersen 150 Integration of noise in PM Systems - Pavement Management and noise 151 Noise Control through Traffic Flow Measures - Effects and Benefits Hans Bendtsen Bjarne Shmidt Lars Ellebjerg 152 Noise from Railway Crossings Inter Noise Paper 2007 Hans Bendtsen Sigurd Thomsen 153 Optimized thin layers for highways - Inter noise paper Noise from streets with paving stones - paper for Inter-Noise 2007 in Istanbul 155 Traffic management and noise - Paper for Inter Noise 2007 in Istanbul Hans Bendtsen Bent Andersen Sigurd Thomsen Bent Andersen Hans Bendtsen Jørgen Kragh Sigurd Thomsen Lars Ellebjerg Hans Bendtsen 156 Traffic noise at rumble strips Inter noise paper 2007 Jørgen Kragh Bent Andersen Sigurd Thomsen

30 Road Directorate Niels Juels Gade 13 P.O. Box 9018 DK-1022 Copenhagen K Denmark Telephone Telefax Road Directorate Guldalderen 12 DK-2640 Hedehusene Denmark Telephone Telefax Road Directorate Thomas Helsteds Vej 11 P.O. Box 529 DK-8660 Skanderborg Denmark Telephone Telefax vd@vd.dk Roadinstitute.dk