Truck behaviour on highway facilities

Transcription

1 Trck behavior on highway facilities J. J. HAJEK, PhD,PEng, Senior Research Engineer, anda. J. UGGE, PEng, Research Engineer, Pavements and Roadway Section, Research and Development Branch, Ontario Ministry of Transportation, Canada The recent availability of weight-in-motion (WIM) traffic monitoring data is exploited for the prpose of investigating safety-related trck behavior on highway facilities. This paper briefly describes WIM technology and its capabilities to generate traffic monitoring data which are important for traffic safety analysis. The body of the paper provides examples of data analysis that demonstrate the seflness of WIM data for investigating safety-related trck behavior. The examples inclde determination of trck exposre rates, and evalation of vehicle speed and headway distribtions as a fnction of trck type and trck load. 1. INTRODUCTION With the advent of the Strategic Highway Research Programme (SHRP) and its national satellite programs, sch as UK-SHRP and Canadian SHRP, weight-in-motion (WIM) scales have become commonplace. For example, the Ontario Ministry of Transportation is crrently operating eight in-highway WIM scales and, to comply with the SHRP and C-SHRP gidelines, is planning to install one additional WIM scale in The WIM scales are providing, and will contine to provide, a large amont of detailed traffic monitoring data for individal highway vehicles, sch as axle spacing and weights, vehicle length and speed. This is in addition to the traditional aggregated traffic characteristics sch as daily and annal vehicle volmes and Eqivalent Single Axle Loads. Considering the effort associated with the installation and operation of WIM scales, and with sbseqent data retrieval and processing, we shold ensre that the wealth of traffic monitoring data generated by WIM scales is properly tilized. Becase of its original association with the SHRP-related pavement research effort, it is often assmed that WIM data are only applicable to pavement damage research. Many potential sers of WIM-type traffic monitoring data do not know (a) (b) (c) the data monitoring capabilities of WIM technology, type of data available, and how the data can be tilized within their area of interest [1]. The obj ective of this paper is to show, by practical examples, that WIM data are also sefl for safety-related analyses and indeed indispensable for stdying trck behavior on highway facilities. The data sed in this stdy were obtained from two WIM scales, both located on Hwy 42 in Ontario, Canada. Highway 42 is a for-lane rral freeway with a speed limit of 1 km/h. One WIM scale is located in an eastbond trck lane near Sarnia and is referred to in this paper as Location 1; the second scale spans all for lanes near London and represents Location 2. Both scales se piezoelectric cable technology [2]. Low traffic volmes (abot 3 vehicles per hor dring daytime and 75 vehicles per hor at night in the trck lane) enable a large degree of traffic operational freedom. However, Location 2 is abot 2 km downstream from a freeway entrance ramp. De to highway alignment constraints, traffic may not yet reach its free-flow eqilibrim at this location. The highway grade for two or more kilometers before either location is at a level. 2. DESCRIPTION OF TRAFFIC MONITORING DATA PROVED BY WIM SCALES A typical WIM scale consists of magnetic loops and axle sensors embedded in the pavement, and a microcompter hosed in a roadside cabinet. Magnetic loops and axle sensors respond to axles passing over the pavement by generating electric signals. The signals are processed by the compter and are transformed into engineering parameters for each individal vehicle. These parameters inclde a time stamp, instantaneos vehicle speed, vehicle length, distances between consective axles, and axle weights. The knowledge of engineering parameters for individal vehicles, combined with a knowledge of the impact of vehicle weight and dimension reglations on trck design, can be tilized to determine highway vehicle types for classification prposes. A jdicios classification scheme can pinpoint and select specific trck types of interest. For example, Figre 1 shows one sch scheme designed for identification of flly loaded 6-ax1e tractor-semitrailers with a liftable ("belly") axle. It is possible to narrow the field even frther by specifying also the period of occrrence and speed or headway parameters. 418 Heavy vehicles and roads technology, safety and policy. Thomas Telford, London, 1992.

2 HIGHWAY SAFETY 2 3 Axle # L > 2.4 I. ~I LI > all dimensions are in m Criteria for Trck Definition (a) (b) (c) (d) (e) (f) (g) (h) Figre 1/ 6-axle trcks (co ideneify ehe proper nmber of axles) single (steering) - dal (tractor) - single (liftable) - dal (trailer) axle arrangement axle 2-3 spacing from 1.7 to 1.83 m (co cover known range of drive axle spreads) axle 3-4 spacing greater than 4. m (co ensre ehae ehe semierailer is ae lease 1 m long) axle 4-5 spacing greater than 2.4 m (co ensre independence of axle 4) axle 4-5 spacing greater than axle 5-6 spacing (co ensre ehae axles 4,5,6 do noe form a eriple axle) axle 5-6 spacing from 1.7 to 3.5 m (co cover known range of erck axle spreads) gross weight within 1 kg of allowable load (specifies a flly loaded erck) Criteria identifying 6-axle tractorsemi trailers with a liftable axle Considering the variety of vehicles on Ontario's highways and large samples of vehicles analyzed, there mst be a certain level of speclation associated with any vehicle classification scheme based on WIM data. There is no reason to believe that these ncertainties have a significant effect on the observations presented herein. safety. The accident rate is defined as the nmber of accidents divided by the nmber of kilometers travelled (exposre rate). Becase trcks are often registered simltaneosly in several jrisdictions, or are registered somewhere other than the jrisdiction in qestion, it is difficlt to estimate vehicle kilometers travelled by different trck classes and ths to obtain accident rates for different trck types. WIM data can help in establishing trck exposre measres particlarly for facilities where WIM scales have been installed. For example, referring to Figre 2, 5-axle trcks (defined in this case as a 3-axle tractor with one dal axle semitrailer, or 3S2) comprise abot 55% of the total trck volme on Hwy 42. The volme percentage of 3S2s can be directly related to the percentage of accidents involving the 3S2s on this facility. It shold be stressed that while WIM-type data can be sed to enmerate the poplation of trcks of any particlar type (e.g., flly loaded 6-axle trcks with one liftable axle, as otlined above), the WIM data cannot discern the body style, commodity/load, owner or other information that might identify the specific natre of the trck and its owner. However, the identification of nmbers and magnitdes is still very important becase it does allow infreqent or lowprobability events to be examined, which is not practical with the small samples that can be garnered by manal srvey methods (4] Vehicle Behavior on Highway Facilities Althogh highways are designed to serve a mix of vehicle types, the effects of varios types of vehicles on traffic operations and safety are not niform. In this context, a nmber of factors inflencing traffic safety have been identified, inclding speed, passing, merging and other lane changing manoevres, splash and spray, aerodynamic bffeting, blockage of view, and lateral placement. Other important factors related to highway safety are vehicle handling, stability and braking capabilities, load and load distribtion, and vehicle headway (5]. 3. SAFETY ANALYSIS Safety is a maj or isse in all debates or polemics abot changes in highway traffic reglations and in vehicle weights and dimensions. Invariably, it is conclded that adeqate information abot safety implications of the proposed changes is lacking (3]. WIM data can contribte to analysis of safety isses by providing niqe and detailed information on a) freqency of different vehicle types sing highway facilities, b) vehicle behavior on highway facilities, and c) trck payload and its distribtion Freqency of Vehicle Types Using Highway Facilities The knowledge of accident rates for different trck types is instrmental in identifying the inflence of vehicle design parameters on highway ill El... >.l«h '-' "-' \ or or more Nmber of trck axles Figre 2/ Occrrence of different trck types, March

3 HEA VYVEHICLES AND ROADS WIM type data alone can be sed to evalate trck driving behavior in terms of speed and headway distribtions as a fnction of axle (vehicle) weight, and time of day (e.g., day timenighttime visibility). In addition, weather conditions (precipitation) were also taken into accont in this paper. For a more comprehensive assessment of trck behavior, WIM data need to be spplemented from other sorces sch as video recordings. a) Vehicle speed distribtion Excessive vehicle speed, and particlarly speed differentials between different vehicles, is considered to be one of the main cases of accidents [6]. When different vehicle types exhibi t different speeds (e. g., loaded trcks moving more slowly than the prevailing traffic, particlarly on pgrades), the speed variance of the traffic flow increases. The difference in speed variance has been directly linked to the increase in overall accident rates [7]. The primary vehicle characteristic affecting acceleration and speed performance of trcks is the weight/power ratio. Speed is also one of the main parameters governing energy consmption. Overloaded and speeding trcks may also constitte an additional safety hazard. An example of vehicle speed distribtion for cars and trcks, obtained by the WIM scale at Location 1 dring daylight hors and at night, is sl1wn in Figre 3. Data were obtained for for consective weekdays withot any precipitation in March of Overall, data in Figre 3 indicate that trck drivers are more disciplined than car drivers. Some specific observations * The majority of the cars were speeding. Dring 4aytime, abot 53% of all car drivers exceeded the speed of 11 km per hor, while at night 42% of all car drivers exceeded this 6, , 5 ~4 c Q) ~ ,'p ",., ",, ", -"""';i"""l.' " \\ \\ \.~, Vehicle speed (km/h) Cars Daytime 8 to 18 Night time - 22 to 4 Trcks a Daytime 8 to 18 Night time - 22 to 4 >13 Figre 3/ Vehicle speed distribtion, Hwy 42, trck lane, March 19 to 22, 1991 speed. The corresponding nmbers for trck drivers were 16% and 1% respectively. * Compared to cars, trck speed distribtion is more niform. Looking at the extremes, dring daytime, 1.3% of cars had speeds less than 8 km/h compared to only.3% of trcks. At the high end, 1. 3% of cars (in the trck lane) exceeded the speed of 13 kmjh compared to.1% of trcks. * The more niform speed distribtion observed for trcks is reflected in their lower speed variance. Using the empirical relationship between speed variance and accident rates (Eqation 4 in Ref. 7), the data in Figre 3 sggest that the accident rate of cars only, wold be abot 7 (accidents per 1 million vehicle miles of travel); the accident rate of trcks only wold be 55, while the corresponding combined accident rate of cars and trcks wold be abot 82. b) Headway distribtion According to Ontario's Highway Traffic Act [8], mai.ntaining "reasonable and prdent" headway (Le. the time between sccessive vehicles) is mandatory for all drivers. There is an extra stiplation for drivers of commercial vehicles which, while driving at speeds exceeding 6 kmjh, "shall not follow within 6 meters of another motor vehicle". Vehicles travelling at 1 kmjh (27.8 m/s) wold have a "front bmper-to-front bmper" spacing of only 28 metres with one second headway. This redces the actal space between vehicles to only abot 24 metres for an average car and, of corse, considerably more for even small trcks. One or two second headways are clearly nsafe at highway speeds. Figres 4 and 5 compare the difference in headway distribtions of cars and trcks. The figres se the same data set as that sed for Q) 8 c Q) 7 ;J I Trck following trck m Car following trck m Car following car ~ Trck following car Figre 4/ Daytime vehicle headway distribtion, Hwy 42, trck lane, March 19 to 22,

4 HIGHWAY SAFETY. '"' '"' 4-! 1" I Trck following trck m Car following trck m Car following car ~ Trck following car Figre 5/ Nighttime vehicle headway distribtion, Hwy 42, trck lane, March 19 to 22, 1991 Figre 3. The greater discipline of trck drivers, indicated by the speed distribtion, is also indicated by the headway distribtion. Some observations * Dring daytime, 7% of all cars followed other cars with a one second headway, while only 2.5% of trcks did so. Nevertheless, considering an average trck speed of 1 kmjh, more than 2.5% of all trcks appear to be in violation of the Ontario Highway Traffic Act headway reqirement. * Also dring daytime, 3.5% of all trcks were following other trcks with a headway of 1 sec while only 2.5% of trcks were following cars with this headway. The difference in the headway distribtion for these two cases was fond to be statistically significant. * Trcks tend to travel in convoys. This is particlarly evident at night when 7.5% of all trcks had headway of less than 6 sec compared to only 1.1% for cars. 3.3 Trck Payload and its Distribtion Another descriptive parameter sefl in accident stdies (and provided by WIM data) is the gross vehicle weight and its distribtion - among axles and, in the case of mltiple trck nits, between trck nits. Braking performance of trcks is not as good as that for cars - braking distances are significantly larger. A recent investigation [9] ana1yzed braking capabilities of different trck types for nloaded, partially loaded and flly loaded vehicles. It appears that braking distances increase sbstantially when trcks are partially loaded or nloaded, where the magnitde of this degradation in braking capabilities depends on the trck type. A comprehensive analysis of trck accidents in Ontario [1] indicated that except for twin trailers, the highest accident rates were for nloaded straight trcks, semi-trailers, and bobtail tractors. This may sggest a link between braking capabilities of nloaded, partially loaded and flly loaded trcks and accident freqencies. Payload distribtion can also affect trck operating characteristics and ths safety. Reference 4 shows how WIM data can be sed to evalate payload distribtion of 6-ax1e trcks with one liftable axle. Systematic evalation of WIM data for loaded and nloaded trcks can provide valable insights into their operating characteristics and behavior. Figre 6 provides an example of speed distribtion for 6 or more axle trcks dring daytime obtained on a passing lane at Location 2. While the difference in mean speed between nloaded and loaded trcks is only 1.5 kmjh, the corresponding speed variance, with its safety implications, differs by 1 (kmjh)2. An example of headway distribtion for loaded and nloaded trcks with 6 or more axles, obtained for a trck lane at location 2, is given in Figre 7. It appears that the loaded trcks have a more prononced proclivity to follow cars with nsafe headways than the nloaded trcks. '(The difference in the headway distribtion was fond to be statistically significant.) The shorter headways observed for the loaded trcks can be attribted to the lack of spare power of these trcks to pass slower moving cars. 4. DISCUSSION The above discssion on extracting specific information of interest in the highway safety area from WIM-type data is not exhastive. It simply illstrates possible sage in the traditional application areas. For example, it is also possible to stdy more complex fnctions of traffic flow sch as the relationship between I' I-l I-l 4-! ' i i EtJ w,",,! \,,,'\ \ I \ \ \ _loaded n = H---nloaded n = Vehicle speed (km/h) Figre 6/ Vehicle speed distribtion, location 2, passing lane, March

5 HEAVY VEHICLES AND ROADS (l) ~ Cl o 2 4-< o , n = 313 o I Loaded 6-or-more-axle trcks following car m Unloaded 6-or-more-axle trcks following car Figre 7/ Daytime trck headway distribtion for 6-or-more-axle trcks,hwy 42, lane 1 vehicle speed, headway, payload, and environmental conditions for different vehicle categories, and to provide data to develop and manage police enforcement strategies. Sprred by Intelligent Highway Vehicle System needs, work is nderway on a range of new sensor technologies, that shold soon be able to spplement IHM-type data with visal and other vehicle characteristics to obtain a comprehensive pictre of traffic flow. 2. "IRD Weight- in-motion Piezoelectric 16P", User's Manal, International Road Dynamics, Inc., Saskatoon, Saskatchewan, McGee, H. W., "Accident Data Needs for Trck Safety Isses", Transportation Research Record 152, TRB, National Research Concil, Washington, D.C., 1986, pp Billing, J.R., and Hajek, J.J., "Weigh-in Motion Scale Data as a Basis for Geometric Design Standards Development, Weight Compliance Assessment, and Reglatory Policy Development", Ontario Ministry of Transportation, September "Twin Trailer Trcks", TRB Special Report 211, TRB/NRC, Washington, D.C., Graber, N.J., and Gadiraj, R., "Impact of Differential Speed Limits on the Speed of Traffic and the Rate of Accidents", Paper No. 9261, Presented at the TRB annal meeting, Washington, D.C., Janary Garber, J.N., and Gadiraj, R., "Speed Variance and its Inflence on Accidents", AAA Fondation for Highway Safety, Washington, D.C. 236, Jly The Highway Traffic Act (as amended to May 1989), Government of Ontario, 88 Bay Street, Toronto, Ontario, M7A ln8, May Parker, D. J., and Htchinson, B. G., "Large Trck Braking at Signalized Intersections", Report TDS-88-l, Ontario Ministry of Transportation, Downsview, Ontario, M3M lj8, December Byco, C., Saccomanno, F.F., and Stewart, A., "Factors Affecting Trck Accident Rates in Ontario", Dept. of Civil Engineering, University of Waterloo, Janary CONCrS IONS 1. This paper has demonstrated by examples that WIM data can provide insights that have previosly not been available to a range of traffic safety related isses. 2. WIM scales, becase of their nobtrsiveness and continos operation, can provide trly nbiased, statistically reliable data yielding a realistic long- term pictre of exposre rates for specific trck types and behavior of the drivers. Both parameters are important for identification of relative inflence of trck design parameters on driver performance and safety. 6. ACKNOWLEDGMENTS The athors grateflly acknowledge the contribtion of Mr. John Billing of the Ontario Ministry of Transportation, to many aspects of this paper. 7. REFERENCES 1. Hajek, J.J., Billing, J.R., and Kennepohl, G., "Application of Weigh-In-Motion Data in Transportation Planning", Paper No , Presented at the TRB annal meeting, Washington, D.C., Janary