TITLE: by: Calibration and standardisation of road roughness measurements using the TRL. profile beam G Morosiuk, C R Jones and M Osman Transport Research Laboratory Crowthorne Berkshire RG45 6AU United Kingdom PAl 276192 Department For International D 1 )~~~Development .MOROSIUK, G, C R JONES and M OSMAN, 1992. Calibration and standardisation of road roughness measurements using the TRL profile beam. In: Proceedings of the Seventh REAAA Conference, Singapore, 22-26 June 1992. CALIBRATION AND STANDARDISATION OF ROAD ROUGHNESS MEA'SUREMENTS USING THE TRRL PROFILE BEAM Dr.G Morosiuk Transport and Road Research Laboratory United Kingdom C R Jones Transport and Road Research Laboratory United Kingdom Mahazan Osman Public Works Department (JKR) Malaysia ABSTRACT This paper describes the development of a new relatidnship between the towed fifth wheel bump integrator (BI) and the TRRL profile beam for use in the calibration of vehicle response instruments used to measure roughness oni asphaltic concrete surfacings. A comparison has also been made with the original BI/profile beam relationship derived from data collected in the United Kingdom and Brazil by the Transport and Road Research Laboratory (TRRL). The results show that the BI/profile beam relationship developed during this study is the more appropriate one to use to calibrate and standardise vehicle response instruments on asphaltic concrete surfaced roads. This study was carried out under a joint cooperative research programme between the TRRL of the United Kingdom and the Jabatan Kerja Raya (JKR) of Malaysia. 1.0 INTRODUCTION Vehicle operating costs are signi ficantly affected by the quality of the road (Morosiuk and Abaynayaka 1982, Hide 1982). As vehicle operati ng costs account for a large proportion of the total transport cost over the life of the road (Parsley -and Robinson 1982), it is important to measure road roughness accurately. In the past the standard instrument used by the TRRL and other road authorities in the UK has been the towed fi fth wheel bump integrator (El) (Keir 1974). Road authorities in other parts of the world use a wide variety of road roughness measuring devices, most being simple response type instruments which need regular calibration. In order to c'ompare different devices, they need to be related to A standardised scale based on the absolute profile of the road itself (Gillespie et al 1980). The profile beam was designed by the Overseas Unit of TRRL to facilitate this calibration and standardisation procedure. 434 The profile beam measures the road profile 'and calculates a single numerical statistic (the root mean square of deviation -RNSD) from the road profile which has been shown to correlate extremely well with most response type instruments. Since roughness measured with the BI has been the UK standard, a relationship between RMSD and the Bl ha's been programmed i-nto the beam's micro-processor so that the profile statistic can be output directly in terms of the El. This relationship was developed from data collected on asphaltic concrete, surface dressed, gravel and earth roads thereby covering a range of roughness between 1000 mm/km and 15000 mmlkm. The equation was not well established at its extremes and therefore it was desirable to collect more data to improve the overall accuracy. The majority of the paved roads in Malaysia are surfaced with asphaltic concrete on which the. roughness is likely to be in the range of 1-000 to 4500 mmlkm. This study was therefore designed to develop a relationship between roughness measured by the Bl and the profile beam on these roads. The relationship was then compared with the original BI/profile beam relationship to establish the most appropriate one for use in calibrating and standardising roughness measurements made on asphaltic concrete surfacings. 2.0 ROAD ROUGHNESS MEASURING SYSTEMS Many systems have been devised and used for measuring road roughness. They can be categorised either as response type instruments or profiling devices. 2.1 Response Type Instruments Most roughness measuring systems use instruments either mounted inside a vehicle or towed by a vehicle at a constant speed. Measurements made with this type of system are the result of the interaction of three basic components - the road, the vehicle and the instrument. They measure and record the cumulative displacement of an axle relative to the body of a vehicle or trailer caused by the unevenness of the road. The majority of these instruments are time, vehicle and speed dependent, and therefore suitable calibration procedures need to be employed to provide meaningful roughness data. The BI is a single-wheeled trailer, manufactured to rigid specifications, which enables it to be used as a standard instrument for measuring road roughness. It consists essentially of a single axle and a rectangular chassis within which a pneumatic tyred wheel is mounted, the load and tyre size being standardised.. An integrator unitisone on one side of the chassis and detects the total downward movement of the wheel axle relative to the chassis. W~hen properly maintained, the measurements are stable over time and do not depend on the characteristics of the towing vehicle. Measurements are generally taken at a steady speed of 32 km/hr (20 mph). BI's are not readily available in most countries but integrator units, similar to the one mounted on the BI, are commonly fitted inside a vehicle. In this case the integrator unit is fitted to the chassis of a vehicle directly over the rear axle to which it is connected. 435 Vehicle mounted instruments are dependent on speed and the suspension characteristics of the vehicle. The speed can be standardised, but the suspension characteristics differ between vehicles and may change for any particular vehicle over a period of time. Therefore it is necessary to calibrate these vehicles at regular intervals. The profile beam has been designed for the calibration of this type of instrument and to calculate the standard equivalent BI roughness. 2.2 Profiling Devices These devices record the longitudinal profile of the road and then analyse this profile using standard mathematical procedures. There are several elaborate and expensive profilometers developed by a number of research bodies such as the TRRL high speed profilometer in the UK (Dickerson and Mace, 1976), the APL 125 in France (Lucas and Viano, 1979) and the GMR Profilometer in the USA (Walker et al, 1970). 3.0 THE TRRL PROFILE BEAM The profiling instrument used for this project was the TRRL profile beam. This equipment consists of an aluminum beam, which is used as the datum, supported at, each end by a tri po d. The measuring unit, containing a linear transducer and a battery powered micro-processor, is suspended from the beam with its sensor wheel resting on the road surface. The wheel traverses the road along the beam's length and the transducer measures the distance from the beam to the wheel at 100mm intervals. Earlier research work by TRRL in the UK and Brazil showed that sampling the road profile at 300mm intervals over a 1.8 metre baselength produced the most statistically significant correlations between profile numerics and response type instruments operated at 32 km/hr. The length of the profile beam therefore was fixed at 3.6 metres enabling two complete baselengths to be measured in one placement. The beam's micro-processor calculates a single numerical statistic (the Root Mean Square of Deviation - RMSD) from the road profile. The RMSD is derived by determining the profile deviation at 300mm intervals from a simple linear regression line, analogous to the ideal road surface profile, for a 1.8 metre baselength and then calculating the root mean square of these deviations. For a 1.8 metre baselength, with 7 profile points spaced at 300mm intervals, the regression line y = a + bx is calculated and the deviations d, determined. The RMSD for a 1.8 metre baselength is then computed as: RMSD 1.8 LA 12 [] n The weighted mean RMSD for a section of road containing N baselengths of length 1.8 metres is given by: RMSD = IYRMSD, 2 [2] YN 436 Measurements of the road profile are taken in both wheelpaths ie RMSD (left) and RMSD (right). The micro-processor then calculates the mean RMSD for a section of road using the formula: Mean RMSD / (RIISDI 2 * PL) + (RMSDR 2 * PR) [3] PL + PR where PL and PR are the number of placements in the left and right wheelpaths respectively; normally these are the same. The Reference Roughness (RR), which is the equivalent BI roughness estimated from the profile beam's numeric, is then calculated from the formula: RR = 2045(RMSD) + 68(RMSD)2 (4] This equation, developed from research work in the UK and Brazil, covers a range of roughness between 1000 mmlkm and 15,000 mm/kin, as measured with the BI, on asphaltic concrete, surface dressed, gravel and earth roads. Roughness measurements from a vehicle response device can be standardised by running the vehicle over the same lengths of road profiled with the TRRL beam. For each section of road the vehicle response reading (VR) is input to the micr6-processo~r which then -solves the second order polynomial: y = A + Bx + CX2 [5] where y = values of RR calculated from RMSD x = values of VR The y value is referred to as the 'Calibration Roughness' and is the estimated roughness in terms of the BI. 4.0 SITE MIEASUJREMENTS In this study the BI and the profile beam were used on various asphaltic concrete surfaced roads to develop a new relationship between the two instruments for use in calibrating vehicle response devices. The vehicle used in this calibration exercise was a Toyota stationwagon fitted with an integrator unit which. was run over the same lengths of road as the other two instruments. 4.1 Selection of Test Sites A total of 21.test sites with asphaltic concrete surfacings were selected covering a range of roughness between 1200 mm/km and 3600 mm/km. The sites were flat and straight so that the speed of the vehicles could be easily and safely maintained at 32 km/hr. Each site was 500.4 metres in length (139 beam placements) and had a uniform roughness throughout its length. 4.2 Road Roughness Measurements The profile of each section of road was measured using the profile beam in both wheelpaths to obtain the RMSD statistic for each wheelpath and thus a mean RMSD for a site. 437 The El and the stationwagon were then run over each site at a steady speed of 32 km/hr. The BI, ~being a single-wheeled trailer, was run in each wheelpath separately whereas the stationwagon responds to the profile of both wheelpaths. The number of passes on each site was dependent on the consistency of the readings with the minimum number of passes being three. The roughness readings from both instruments recorded on the 21 sites, together with the RMSD values, are given in Table 1. TABLE 1 ROUGHNESS MEASUREMENTS ON SITE (Note: *BI not measured) 5.0 ANALYSIS OF DATA 5.1 Relationship.Between the BIland-,Profile Beam The relationship between the reference roughness (RR) and the RMSD statistic (equation [4] ) incorporated in the profile beam's micro-processor was used to estimate the roughness of a site in terms of the Bl. The reference roughness values estimated from this relationship were then compared with the measured Bl roughness values. Both the El and the profile beam measure roughness in a single wheelpath therefore it was possible to compare 40 pairs of data. These data points are listed in Table 2 and plotted in Figure 1. The comparison shows that the observed BI roughness 438 Toyota Bump Integrator (mm/kin) Profile Beam (RMSD) Site station- No wagon (mm/kmn) v/s o/s Mean v/s { o/s [Mean 1 470 1705 1780 1743 0.459 0.484 0.472 2 510 1700 * 1700 0.495 0.463 0.479 3 540' 1850 * 1850 0.589 0.603 0.596 4 1460 3225 3055 3140. 1.365 1.285 1.326 5 905 2385 2550 2468 0.999 0.858 0.931 6 735 2285 2020 2153 0.988 0.807 0.902 7 905 2295 2380 2338 1.018 0.879 0.951 8 560 1535 1945 1740 0.587 0.683 0.637 9 1375 '2985 3305 3145 1.260 1.319 1.290 10 1575 3020 3630 3325 1.237 1.531 1.392 11 1030 2750 2635 2698 1.250 1.047 1.153 12 1260 2850 2960 2905 1.204 1.177 1.191 13 670 2080 2055 2068 0.806 0.685 0.748 14 725 2300 1965 2133 1.098 0.717 0.927 15 940 2430 2370 2400 1.083 0.864 0.980 16 315 1330 1420 1375 0.383 0.407 0.395 17 305 1285 1340 1313 0.365 0.390 0.378 18 295 1200 1320 1260 0.340 0.361 0.350 19 315 1300 1310 1305 0.362 0.365 0.364 20 405 1690 1725 1708 0.478 0.540 0.510 21 320 1365 1300 1333 0.330 0.371 0.351 figures were generally substantially higher than the. predicted roughness values from equation [41. TABLE 2 COMPARISON BETWJEEN OBSERVED AND PREDICTED BI ROUGHNESS VALUES * Using equation [4] * BI not measured The BI roughness reading on an asphaltic concrete road in good condition is known~ to be approximately 1200 mm/km (Gillespie 1980). The predicted roughness from equation [4] was below 1000 mm/km on eight sites and fell as low as 700 mm/km on a number of these sites. This suggests that equation [4] underestimates the roughness of relatively smooth asphaltic concrete surfacings. Using the data from each wheelpath for all sites, a more appropriate relationship for estimating BI roughness values from RMSD numerics was derived. The new reference roughness relationship for asphaltic. concrete surfacings (RRac) is given below and plotted in Figure 2. RRac = 691 + 1811(RMSD) [6 ] R2 = 0. 954 This new relationship is based on 40 points whereas the original relationship (equation [4]) was based on only 8 points at this end of the range of roughness (ie < 3600 mm/kmn). 439 BI ROUGHNESS (mm/kmn) Site No. Vergeside Offside Observed Predicted* Observed Predicted' 1 1705 953 1780 1006 2 1700 1029 *961 3 1850 1228 *1258 4 3225 2918 3055 2740 5 2385 2111 2550 1805 6 2285 2087 2020 1695 7 2295 2152 2380 1850 8 1535 1224 1945 1428 9 2985 2685 3305 2816 10 3020 2634 3630 3290 11 2750 2663 2635 2216 12 2850 2561 2960 25,01 13 2080 1692 2055 1433 14 2300 2377 1965 1501 15 2430 2294 2370 1818 16 1330 793 1420 843 17 1285 755 1340 807 18 1200 703 1320 747 19 1300 749 1310 755 20 1690 993 1725 1124 21 1365 682 1300 768 5. 2 Calibration of the Vehicle Mounted Integrator Unit The mean -RMSD values given in Table 1 were used to calibrate the stationwagon using the new reference roughness relationship (equation [6]). These reference roughness values and the corresponding vehicle response readings were then used to derive the calibration equation for the vehicle as shown in equation [7]. CR = 567 + 2.67(VR) - 0.00064(VR)2 [7] R2 = 0.976 where CR is the Calibrated Roughness in mmlkm. VR is the vehicle roughness reading in mmnlkm. 6.0 CONCLUSIONS AND RECOMMENDATIONS Vehicle response type systems are commonly used throughout the world as they offer a rapid and relatively inexpensive method of measuring the roughness of long lengths of road. A method of standardising these -roughness measurements has been established using the TRRL profile beam. This study has developed a more appropriate relationship-between the BI and road profile for roads with asphaltic concrete surfacings. It is recommended that this relationship, shown in equation [6), should be used in place of the original relationship (equation [4]) when using the procedure outlined in Section 3.0 to standardise roughness measurements from vehicle response devices used on asphaltic concrete surfaced roads. 7.0 ACKNOWLEDGEMENTS This study was carried out under a joint cooperative research programme between the Transport and Road Research Laboratory (TRRL) of the United Kingdom and the Jabatan Kerja Raya (JKR) of Malaysia. The authors would like to thank the Director Gene~ral of the JKR for permission to publish the paper. The work described in this paper forms part of the research programme of the Overseas Unit (Unit Head: Mr 3 5 Yerrell) of the TRRL carried out on behalf of the Overseas Development Administration. Any views expressed are not necessarily those of the Administration, the Department of Transport or the JKR. The paper is published by permission of the Chief Executive of the TRRL. 8.0 REFERENCES DICKERSON, R S and D G W MACE (1976). A high-speed profilometer - preliminary description. Department of the Environment, TRRL Report SR182UC. Crowthorne, (Transport and Road Research Laboratory) . GILLESPIE, T D. M W SAYERS and L SEGEL (1980). Calibration of Response-Type Road Roughness Measuring System. Transportation Research Board Report No. 228. 440 HIDE, H (1982). Vehicle operating costs in the Caribbean: results of a survey of vehicle operators. Department of Environment Department of Transport. TRR.L Report LR 1031. Crowthorne, (Transport and Road Research Laboratory). KEIR, W G (1974). Bump-integrator measurements in routine assessment of highway maintenance needs. Department of Environment Department of Transport, TRRL Report 'SR26UC. -Crowthorne, (Transport and Road Research Laboratory). LUCAS J and A VIANO (1979). Mesure syst~matique de 1'uni sur le r~seau routier - APL A grand rendement, version 1978. Bull. liason Laboratoire Central Des Ponts et Chauss~es - 1.01. MOROSIUK, G and S W ABAYNAYAKA (1982). Vehicle operating costs in the Caribbean: an experimental study of vehicle performance. Department of Environment Department of Transport, TRRL Report LR 1057. Crowthorne, (Transport and Road Research Laboratory). PARSLEY, L L and R ROBINSON (1982). The TRRL road investment model for developing countries (RTIM2). Department of Environment Department of transport. TRRL Report LR 1057. Crowthorne, (Transport and Road Research Laboratory). WALKER, R S. F L ROBERTS and W R HUDSON (1970). A profile measuring, recording and processing system. Research Report No 73-2, Center for Transportation Research, The University of Texas at Austin. Crown Copyright 1991. The v~iews expressed in this paper are not necessarily those of the Department of Transport or the Overseas Development Administration. Extracts from the text may be reproduced, except for commercial purposes, provided the source is acknowledged. 441 4 _ 3.5 Line of equality O 3 2.5 E .r 2 V0 1.5 0. 50.5 1.5 2.5 3.5 Observed B11 roughness (mmlkin) (thousands) Fig.1 Comparison of observed and predicted BI roughness 4 0 3.5 0.2.5 E 2 E E 1.5 81691 + 1811 (RMSD)S 0) 0 .2 0R 06 =01 .29541. ~~~~~~~~ 1~~Poil em(MD Fig. Reltionhip etwen 81andprofile beam forMaSphlD oceerod)nMlyi 442