Review of the HDM-111 User Cost Model for Suitability to Canadian Heavy Vehicles

Transcription

1 78 TRANSPORTATION RESEARCH RECORD 1262 Review of the HDM-111 User Cost Model for Sitability to Canadian Heavy Vehicles PETER BEIN Operating expenses of the anadian hea y-vehicle Fleel exceed several times th cxpendilre on the pblic road system, bt sond aid that recog11ie the imp rtance of ser costs in road investment decision are lacking. A state-of-the-art model of ser co ts wa developed after a stdy in Brail by th World Bank. The crrent version of the Highway Design and Maintenance Standards Model, HDM-111, is evalated for relevance to the anadian heavy-vehicle operating condition, and the re 'lts of preliminary calibrations for five typical vehicles are presented. The model i relevant to tdy.ing road infrastrctre isse. h is also easy to operate and calibrate, bt the cost-roghness slope and the effect of road textre, a roadworks decision variable in Canada, remain to be validated. The model will become more sefl once the effects of congested traffic conditions on operating speeds arc incorporated. Maintenance and depreciatio11 costs need closer examination becase they are interrelated and c mplicated by the inclsion of factory warramy co ts in the prchase pri.cc. Maintenance co t formla them elves need reevalation becase they were d velopcd for a low labor-c t environment. Jf the present HDM-Jll " adjsted tilialion metliod" i retained, it shold be modified to accont for the fact t"hat time savings on a hal do not nee sarily translate into additional trips. Relationships for cargo delay and damage de to road condition also need improvement. Trcking makes a major contribtion to the economy. Not conting the smaller operations, Canadian for-hire and private carriers spend well over $2 billion annally in operating costs and engage more than 12, straight trcks, 49, tractors, and 11, trailers (1 ). Trcks move abot twothirds of the commerce between Canada and the United States at a cost of $2.5 to $3 billion, an amont that cold doble in Jess than 1 years with free trade. Dereglation will have growth effects similar to those observed between 1978 and 1987 in the United States (2). Most of the heavy vehicle travel takes place on paved srfaces that accont for 4 percent of the total 84 km of road in Canada. Robst economic argments are needed to spport increased road agency fnding levels and proposed changes in heavy vehicle sie and weight reglation. An accrate assessment of trcking costs is important becase they have significant mltiplier effects in the national economy. Canadian road sers' contribtion to the economy is an order of magnitde higher than the present pblic spending on constrction, maintenance, and reconstrction of all rral and rban roads (3) West 14th Avene, Vancover, British Colmbia, Canada V6R 2X2. Highway ser cost and benefit analyses in Canada rely on otdated data. There is a recognied need to establish ser cost parameters for the Canadian vehicle fleet and develop a methodology for road deterioration relationships (3). Trcking ind try models sch ''s tho e oftrimac (4 or IBl Grop (5), are nable to relate the cost to road srface condition or geometry and are less sefl for st dying road infrastrclre i ses. The applicability of the HDM-UI model f road ser costs to Canadian heavy vehicles is discssed in this paper. HDM-III MODEL Following a large-scale stdy in Brail in 1975 to 1981, the World Bank developed the Highway Design and Maintenance Standards Model. The present version of the model, HDM III, can aid feasibility stdies of individal projects as well as policy stdies for highway networks (6). The model is generic and can be calibrated to sit I cal condition if they djffer from those covered by the model's data base. The research repre ents the largest effort to date to develop a model captring the relationships between costs of conslrclion, maintenance, and tiliation of roads. The model is based on the following premises dedced from earlier stdies: User costs are related to highway constrction and maintenance standards throgh the effect of road geometry and pavement srface condition. Pavement deterioration depends on the original constrction, maintenance, and reconstrction, as well as traffic loading and ncontrollable environmental effects. Srface roghness is the principal road-related factor affecting ser costs in free-flow traffic that can be related to all major pavement performance variables. Interrelationships are strctred to describe known mechanistic and economic phenomena and can be sitably calibrated sing the observed data. User Cost Components User costs depend on a region's economy, vehicle technology, driver behavior, and fleet-operating decisions. In order to facilitate ftre calibrations of the model to different local conditions, the World Bank's goal was to employ generic principles. The goal was reached for vehicle speed and fel consmption models. Tire consmption and vehicle mainte-

2 Bein 79 nance modeling experienced difficlties in gathering controlled experimental data. Generic economic cases of interactions between vehicle maintenance, depreciation, and interest costs were flly appreciated only after HDM-111 was examined for conditions different than in the Brailian stdy. Operating cost relationships were developed for atomobiles, light vehicles, two- and three-axle trcks, an articlated trck, and a bs. Vehicle speed, fel, and tire consmption models were derived from a force-balance eqation. The other ser cost components are: oil; maintenance parts and labor; depreciation as the loss of vehicle market vale; interest on ndepreciated capital tied p in the vehicle assets; and driver, occpant, and cargo delay costs. Administration overhead costs can be calclated either as a lmp sm per vehicle prorated over the annal distance travelled or as a percentage of rnning costs. This cost category cold inclde other fixed costs sch as terminal, registration, insrance, and license costs. A sm of the vehicle operating costs (VOC) and delay costs weighted by each vehicle percentage in the traffic mix is the total cost on a road section or network link. This total cost is sed in the HDM-III highway investment model and in the evalation of alternative individal projects. A stand-alone VOC prediction model is available for an IBM PC or compatible microcompter rnning on an MS DOS 2. or higher version. The program fits a crve to the total ser cost expressed as an exponential fnction of srface roghness. Road geometry featres are fixed parameters in the program. Assmptions and Limitations The HDM-III ser cost sbmode! assmes that roads consist of homogeneos sections of sfficient length for a vehicle to attain a steady-state speed for a given road geometry and srface condition. The speed is a fnction of engine power, road gradient, horiontal crvatre, srface roghness, and driver behavior. The sbmode! in its original form cannot inclde the costs reslting from congested traffic conditions and start-stop operations. Until a sitable revision is made, these costs can be inclded in the HDM-III highway investment model by a generalied procedre. Other ser costs sch as constrctionrelated traffic delays, accidents, and environmental polltion can also be entered from separate estimates. A significant amont of work will be reqired to adapt and calibrate the HDM-111 ser cost sbmode! so as to accrately reflect the heavy-vehicle conditions in all Canadian provinces. Preliminary calibrations have already been accomplished within two Canadian projects (3, 7). The assmptions and limitations of HDM-III ser cost model will now be discssed from the point of view of application to heavy vehicles in Canada. COMPARISON OF BRAZIL STUDY AND CANADIAN HEAVY-VEHICLE OPERATING CONDITIONS Vehicle Technology The Brailian vehicle fleet was spplied by the domestic atomotive indstry, ranked among the world's top ten. Brailian trck designs in the 197s featred engine efficiency and body design for payload maximiation within gro s vehicle weight and axle load limits, bt were not as efficient as their Eropean conterparts. By 198, a range of engines was available that cold rn on a variety of gasoline, gasohol, and diesel fels. Vehicle owners cold match trck specifications to the type of service they wished to provide. Data for typical Canadian and Brailian heavy vehicles are compared in Table 1. The most widely sed interrban trck in Brail was a three-axle rigid vehicle with a nondriven trailing axle, grossing between 18 to 22 tonnes and powered by a 147-hp (SAE) engine. This shold be compared with a 1986 five-axle nit grossing 37.5 (Saskatchewan) to 49 tonnes (Qebec) and eqipped with a 3- to 35-hp engine. The new Canadian heavy-vehicle weight reglations allow a gross combination weight (GCW) of 46.5 tonnes for five-axle combinations and 63.5 tonnes for Roads and Transportation Association of Canada (RT AC) B-trains that are eight-axle configrations. Canadian two- and three-axle trcks are sed mainly in rban service. The three-axle tractor body design in Brail was cab-over-engine, bt conventional design was favored in North America for some ti!!je becase of its lower maintenance cost and better fel economy. Five-axle combinations were lighter and not as nmeros as in Canada, and larger combinations were nsal. bses were also lighter in Brail. An extensive se of lighter materials in the body and mechanical components of both the trck and trailers has led to increased payload capacity per axle. In a qest for lighter eqipment, North American trck operators have considered sacrificing vehicle drability and trck life. Advances in trck combinations technology have prodced nits of seven and more axles that did not exist in Brail. Recent reglatory changes allowing higher gross vehicle weights are certain to increase trck prodctivity. Modern tractors are powered by energy-efficient power plants with semiatomatic transmissions. Aerodynamic design is considered an important featre of eqipment operated on long hals. The vehicles are increasingly being eqipped with radial tires that agment fel efficiency by p to 3 percent compared with bias-ply tires. "Electronic" engines that can be programmed for a specified horsepower, torqe, and speed for a given hal are already available. The bs spply in Brail ranged from integral and platform, rear-engine vehicle with air sspen ion and air-conditioning for long-distance operations on paved roads to traditional front-engine, ladder-type chassis versions for se on npaved rotes. All bses crossing state bondaries or traveling rotes longer than 3 km within a state had to be fitted with a tachograph. Rles governing speed limits and driving hors were strictly enforced for both bses and trcks. Fleet Operatinn Compter applications in trcking management have created an nprecedented potential for prodctivity improvements that were technically impossible for operators participating in the Brailian stdy. Compteried dispatching and roting systems can help the dispatcher's jdgment in optimiing the company's pick p and delivery operations. Maintenance, parts inventory, fel, and tire control software, when sed intelligently, redce fleet operating costs and increase service reli-

3 TRANSPORTATION RESEARCH RECORD 1262 TABLE I DIFFERENCES IN VEHICLE TYPE AND UTILIZATION CANADIAN PRAIRIE VEHICLE PROVINCES, 1986 BRAZIL, Axle Trck<l) Crb weight, kg Max. rated engme power, hp (SAE) Typical annal km '!Jpical annal hors driven 3-Axle Tmck(l) Crb weight, kg Max. rated engine power, hp (SAE) Typical annal km Typical annal hors driven 5cAxle Trck(l) Crb weight, kg Max. rated engine power, hp (SAE) Typical annal km Typical annal hors driven 7-Axle Combi:nation(l) Crb weight, kg Max. rated engine power, hp (SAE) Typical annal km Typical annal hors driven.hys Crb weight, kg Max. rated engine power, hp (SAE) Typical annal km Typical annal hors driven IntemationaJ(2) 14,6 5, ,4 32,(3) /5,ooo(4) 1,/26 Urban/Farms InternationaJ(2) 21,5 11, 18 2,7 32, 1, Urban Kenworth W9oo(2) 37,5 14,5 35 2,15 14,Qo(4) 1,6(4) Western Star(2) 53,5 16, 4 2,1 14,)(4) 1,6(4' M.c.1.(7) 17,2 12,9 3 2,1 >1,? 1,6? Mercedes Ben , 5, ,8 11, 2,4 Mercedes Ben ,5 6, ,8 11, 2,4 Scania 11/39 22, 14, ,2?? (6) Mercedes Ben 362(8) 11,5 8, ,8? 2,4 ~~~-= Brail data from reference (8) Trimac (1986) data for $eneral merchandise cargo; Vehicle dealer atomollve data on most poplar makes; Estimate sing 33 km/h average rnning speed in rban service; Lea Associates 1987 srvey and "1985 Travel on Saskatchewan Highways"; Trimac (1986) data for blk cargo; Eqivalents to 7-axle combinations did not ex.isl in the Brail stdy; Motor Coach Indstries (M.C.I.), 3-axle bs; 2-axle bs. ability. Less-than-trckload freight can be consolidated and empty trck rns eliminated by sing compteried matching service for the interchange of information between shippers and carriers. Electronic data interchange between shippers, carriers, and consignees eliminates paperwork, speeds p deliveries, and enables manfactrers and distribtors to synchronie the flow of goods. This is possible owing to high-technology developments sch as on-board compters in trcks, driver-dispatcher commnication systems, and satellite tracking of vehicles. These systems are now becoming a reality among larger motor carriers, who will be able to fill the transport fnction in the jst-in-time, cost-redcing method of materials and goods management in manfactre and distribtion. Rolling Resistance The largest contribtion lo vehide rolling resistance arises from the hysteresis of tire materials cased by deflection on road srface macrotextre. In the Brailian stdy, rolling resistance was regressed throgh controlled experiments on longitdinal road roghness, which was considered the primary variable. Srface textre, tire, and ssp n ion characteri tics

4 Bein 81 were ignored. Road-srface textre characteristics are definitely a location-specific variable, depending on materials, methods of constrction, and the weathering effects :-if climate (9). Tire properties have changed considerably relative to the bias-ply technology prevailing in Brail. Neither load conditions nor pavement textre was fond to be a significant determinant of test trck rolling resistance, bt tire pressre was fond to be highly significant (1). Between a flsh seal, hot mixes, and a chip seal, the macrotextre varies by almost eight times, whereas microtextre varies by almost two times (11). Rolling resistance is 25 percent higher on deeply textred asphalt than on medim textre asphalt, and almost 7 percent higher than on smooth concrete (12). The HDM-Ill relationships relating to rolling resistance wold need revision in Canada. Coast-down experiments are recommended sing vehicles of contemporary sspension and tire designs on a range of typical Canadian road srfaces in order to verify the Brailian rolling resistance relationships. This verification is important, considering that the forcebalance eqation nderlies fel and tire consmption and time-related ser cost estimation. It is also important to qantify the effect of the pavement deflection bowl. It has been observed in Canada that loaded trcks need to gear down in order to climb a gradient created in front of the wheels on thin pavements. Vehicle Speeds The HDM-Ill VOC model has been developed for steadystate speeds encontered nder free-flow traffic conditions. Speed change and stop-and-go cycles along the roadway are not modeled in the present version, bt at least one compatible model for these traffic conditions exists (12) and cold be inclded in a ftre version of HDM. The new version wold be mch more sefl in Canada for both highways and rban roads. In the Brailian stdy, vehicle loads were estimated visally and payloads recorded in broad weight classes. Trck payload has an important inflence on speeds, and visal observations are not accrate estimates of loads carried. One stdy (7) encontered some complications in calibrating the speed relationships, possibly for this reason. For lack of other measres, the HDM-Ill roghness-related speed fnction was adopted from the standard Mays-metereqipped Chevrolet Opala car. Its average rectified velocity of sspension motion was sed as a srrogate for that of the heavier vehicles. This assmption may not be sitable for trcks, especially those haling high-vale cargo ssceptible to damage by vibrations and shocks. Fel Consmption The Brailian stdy fond that fel consmption cold be predicted by sing a constant nominal engine speed rather than the actal one. Adjstment factors have been introdced to allow corrections for changes in vehicle technology and for experimental driving conditions. The HDM-llI defalt factors are specific to the Brailian stdy. They were obtained by correlating experimental data with actal road-ser fel consmption. Specific vehicle and engine technology imposed by the trcks in Brail does not permit extrapolation to modern Canadian trcks, and corrections will be reqired. HDM-llI fel consmption relationships are based on an inadeqate approach to representing pavement textre in the rolling resistance eqations. Pavement textre cases energy losses in the tires that are responsible for an additional resistance compared with driving on a srface withot textre. A Swedish model (13) predicts a 4 percent increase in fel consmption for heavy vehicles traveling on deeply textred srfaces compared with smooth roads. For free-flow traffic, an Astralian model (12) fond a reasonable agreement with HDM-llI fel consmption predictions for straight trcks and combinations. A Canadian stdy (7) fond calibration of HDMllI fel consmption to be one of the easiest adaptation tasks. The omission of textre, an important decision variable in highway management in developed contries, has cased some controversy. Zaniewski (14) cold not find any dependence of fel consmption on road roghness, possibly becase the Serviceability Index does not captre pavement textre characteristics, according to Claffey (15). The other factor important in increasing fel consmption is a higher wheel slippage in Canada becase of the presence of slsh, snow, and ice on road srfaces in winter. Winter fels were not sed in the Brail tests. The Scania test tractor in Brail had a high fel consmption when rnning loaded on grades compared with a similar Eropean version. Investigations have reveal~d that the vehicle was fitted with a low-speed differential that may be partly responsible for the differences. This was not realied when the fel data were collected and analyed. The fel stdy reslts for this vehicle type shold be sed with cation (T. Watanatada, World Bank, npblished data, 1987). Tires Tire data are difficlt to collect becase tires are moved between vehicles and axles. Tire life varies even nder identical operating conditions according to load carried, position on the vehicle, speed of operation, and driver behavior. Tire tiliation also depends on a company's standards of maintenance for tires and vehicles, and on recapping policy. Tire technology and manfactre contine to change rapidly. Research into tire costs for road vehicles is reqired to measre the benefits of new tread and carcass materials, recapping techniqes, new tire types, and central tire inflation eqipment on large articlated vehicles. Steel-braced radial tires were not available in Brail dring the periods in which ser srveys were performed, and these tend to give different tire costs per kilometer. The limitation of the rolling resistance coefficient with regard to captring pavement textre properties also applies to the tire consmption relationships. A srvey of bs companies has fond that anglar stones in the pavement cased accelerated tire wear (1). The HDM-llI relationship for tire consmption of trcks and bses is intended for se with roads of moderate horiontal alignment and a well-designed sperelevation. These conditions are met on Canadian arterials and collectors, where most of the trck transportation takes place.

5 82 Maintenance Costs HDM-III maintenance cost relationships will most likely be difficlt to transfer to Canada. Maintenance expenditres are sensitive to price and wage levels and the trade-offs of depreciation and interest charges, all of which are linked to the sie, strength, and strctre of the local economy and to the type of transport service offered by the operator. For example, an operator providing service of high reliability at peak periods in a competitive economy will set high levels of inspection and condct preventive maintenance. These costs will be balanced by keeping reserve vehicles to a minimm. Despite these transferability problems, close agreement was fond between the cost-roghness slope in Brailian and Soth African data on bs parts consmption (JO). Maintenance parts and labor costs, and their trade-off with depreciation and interest charges, are likely to prove highly resistant to a mechanistic approach. Economics, not technology, is the key factor in predicting these components. Calibration of the maintenance cost relationships of HDM-III for Saskatchewan (7) met some difficlty becase of their mathematical strctre. Depreciation and Interest In the Brailian stdy, no data were collected to estimate the relationships between depreciation and interest costs and road characteristics. HDM-III considers speed, tiliation, and service life to be interdependent. A nmber of relationships are provided for calclating tiliation and service life, bt only "the adjsted tiliation method" is recommended for Canada. It assmes that each vehicle operates on a fixed rote throghot a given year and that the annal hors available for driving are constant and independent of vehicle speed and rote characteristics. Vehicle operators maximie vehicle prodctivity by making as many trips as possible within the availability constraints. HDM-III has two methods for calclating the average annal depreciation and interest. The constant-vehicle-life method ses a straight-line depreciation for a specified vehicle life, which is assmed to be a constant regardless of any calclated vehicle speed. The average annal interest is calclated on the average vehicle price over its lifetime. In the varyingvehicle-life method, vehicle life decreases (or increases) somewhat as vehicle speed increases (or decreases). This means that the service life, lifetime kilometerage, and depreciation charges change less proportionately than speed. This method was fond nsitable for low vehicle tiliation levels in New Zealand (16). Occpant and Cargo Delay Costs In the HDM model, the cost of crew labor and cargo-holding cost are considered to be variable rather than fixed costs. This means that the time spent on loading, nloading, and layovers is not charged against this cost category. The cargo-holding cost per 1, vehicle-km is defined as the prodct of vehicle hors spent by cargo in transit and the ser-specified cargo-holding cost per vehicle-hor delayed. In TRANSPORTATION RESEA RCH RECORD 1262 general, the cargo-holding cost is small, bt for cargo of high vale the cost will be significant. For example, a $1 million cargo delayed 2.5 hr becase of a redction of trck rnning speed from 1 km/hr to 8 km/hr will give rise to a delay cost of 2.9 cents/km. CALIBRATION OF HOM-III FOR CANADIAN HEAVY VEHICLES Uncalibrated HDM-111 Trimac Conslting Services Ltd. carries ot a srvey of Canadian trck operating costs every 2 years ( 4). Tri mac's assmptions on trck characteristics and nit costs are representative of the indstry, and were fed into the Highway User Benefit Assessment Model (HUBAM) sed by Transport Canada and into the ncalibrated HDM-III. Figre 1 compares the reslts of Trimac, HUBAM, and ncalibrated HDM-III analyses for three trcks. HUBAM does not analye vehicles larger than five axles. Costs are fairly close for the two-axle trck, provided that driver costs are adjsted and interest charges in HUBAM are taken into accont. Trimac's driver cost is high becase it reflects an rban pick-p and delivery service at an average 32 km/hr, whereas HDM-III and HUBAM assme intercity hal at free-flow speeds. Fel and oil costs in HUBAM are overestimated, as is maintenance cost in HDM-III. Total costs for a five-axle Trimac trck are similar to HUBAM, bt HDM-III is mch too high becase maintenance and tire costs are abot for times higher than the corresponding figres. The HDM-III reslts for a seven-axle trck sffer from a similar deficiency, althogh the other cost components seem comparable to Trimac's indstry rates. Figre 2 compares the VOC changes on rogher roads relative to a smooth road with a Riding Comfort Index (RCI) vale of 8. HDM-III provides mch larger benefits than HUBAM for any improvement project, particlarly for trcks. This mst reslt from the overprediction of trck maintenance costs and the strong inflence of road roghness on these costs. RT AC and Saskatchewan data shown in Figre 2 indicate disagreement in Canada on this crcial relationship nderlying all highway feasibility, maintenance, and rehabilitation decisions. The most likely relationships wold be between the HDM-111 and HUBAM crves. Figre 3 emphasies the need for an accrate calibration of vehicle speed in HDM-III. The HUBAM speeds are highest becase HDM-III speeds are determined by an 8-km/hr limit in force dring the stdy in Brail. Neither model is realistic abot the gap between atomobile and trck free-flow speeds on present highways. Trimac speeds of five- and seven-axle trcks are not comparable becase they are trip averages inclding stops and rban driving at each end of an intercity hal. Calibrated Relationships Typical Canadian trcks and a bs operating in the prairie provinces were chosen for trial calibrations of the HDM-III model. Vehicle operating characteristics, tiliation, ato-

6 Bein 83 E 6.:J_... Cf) f-- lj.j AXLE E "'- ' Cf) f-- lj.j TRI MAC HDM3 HU BAM Figre 4 shows the total VOC in cents per kilometer as a fnction of roghness expressed in RCI nits. The seven-axle trck has the steepest slope and the three-axle bs the gentlest slope. The two-axle trck has the lowest VOC and the sevenaxle trck the highest voe, bt the three-axle trck is more expensive to operate in cents/km than the five-axle. The breakdown into VOC components is shown in Figre 5 for the seven-axle trck. Crew cost is significant, bt depreciation and interest are greater. The largest cost increase with roghness occrs in the maintenance and depreciation components, while fel cost does not change mch with roghness. Fel cost drops at lower RCI vales becase of redced speeds on rogh roads, bt crew time increases then. For a three-axle bs, tires cost less becase there are fewer of them and the load is lighter. Maintenance is also less becase there are fewer components to break down compared to a seven-axle trck. Figre 6 smmaries the change in total VOC when roghness decreases from RCI 8 to 4. This type of data can be sed in road investment decisions to estimate extra ser costs de to road deterioration or determine ser cost savings from road improvements. The two-axle trck experiences the highest, and the bs the lowest, marginal rates of increase with RCI. When a delay of 2 passengers is added, the marginal rate of the bs halves, indicating a high sensitivity to the nmber of passengers. :i AXL.E E.:/.... Cf) f-- lj.j AXLE TRIM AC TRI MAC HDM3 HDM3 HU BAM DR IVER INTEREST DEPRECIATION MAIN TE NANCE TIRE FUEL 8 LUBRICATION FIGURE 1 Comparison of Trimac's VOC data with predictions by ncalibrated HDM-111 and HUB AM. motive data, and nit costs were obtained in srveys and interviews of the trcking indstry, vehicle dealers, and manfactrers. Trimac data (4), fond reliable compared with the srvey data, were sed as the benchmark to calibrate HDM III for smoother road conditions. For rogher roads, data are not readily available and the slope of the voe-roghness relationship was adopted from the Brailian stdy. Sensitivity Analyses The VOC of trcks is abot three times more sensitive to rolling resistance than the voe of cars, regardless of roghness. Cost increments de to deep-textred verss medimtextred asphalt srface wold be similar to increments reslting from an RCI drop from 8 to 6.5. A 2 percent decrease in trck tiliation leads to a comparable effect. A 1 percent increase in payloads from better prodctivity, or from higher trck GVW or volme, can lead to 7 to 9 percent redctions in nit trcking costs. CONCLUSIONS The HDM-III ser cost model is a robst tool for road transportation decisions based on economics, road engineering, and vehicle operation and management principles. Adaptations will be reqired to make the model an accrate simlator of heavy-vehicle operations becase they were not represented in the HDM-III database. The model is relatively easy to adapt, owing to the mechanistic form of most of the relationships, and it will become more sefl once the ongoing research into modeling congested traffic flows is incorporated. Maintenance and depreciation costs need closer examination becase they are interrelated and were developed for a low labor cost environment. If retained, the "adjsted tiliation method" shold be modified to reflect the fact that time savings on a hal do not necessarily translate into additional trips. Relationships for cargo delay and damage de to road conditions need improvement becase these costs are more important in developed contries than developing ones, for which HDM was conceived.

7 7 6 5 RTAC "Pavement \ Management Gide", \ Ottawa, 1977 \ ~_,_ Figre 2.6 for 8km/h --...\, ' \ \ \ ' LEGEND AUTO 6 2 AXLE TRUCK 5 AXLE TRUCK tidm3 > 4 ~ w C> <( 3 :I: ~ 2 HU BAM ---,--+--SASKATCHEWAN ' (1983) ' ', ' , SASKATCH WAN (1984) ', ~ \ ' ' ' ' ' ' ' ' ' ' ' 1 ~r--o...::----"... ~ ,_-+_::~~~..._-----l--~----l ' ' ', ' ' ' RCI FIGURE 2 Percent change in VOe relative to ReI = 8. 6 e 1 w a.. (fl w...j :I: w > AUTO 2 AXLE TRUCK 5 AXLE TRUCK 7 AXLE TRUCK HDM3 HUB AM 4 RC I 6 -' - 'I ' FIGURE 3 Predicted vehicle speeds for free flow conditions. c 8 ~ Cf.) ' I-< ~ RIDING COHFORT INDEX FIGURE 4 Total voe of heavy vehicles. B

8 Bein DEPRECIATIDH & IHTEREST ACKNOWLEDGMENT This research was sponsored by the Roads and Transportation Association of Canada and the Saskatchewan Highways and Transportation while the athor worked for N. D. Lea International Ltd. Valable insights from John B. Cox, Norman D. Lea, and Merv F. Clark are grateflly acknowledged. = HA IHTEHAHCE ~ '\. PARTS CZ)... 5 ::: (.Ll HAIHTENAHCE LABOR 3.CRE~ 1 TIRE FUEL & OIL B RID ING COMFORT INDEX FIGURE 5 Seven-axle trck voe components. "' II '-' 25 o::; E- 2- A><!.E TRUCK.,_, 2 7-AX!.E TRUCK ~ 3-AX!.E TRUCK E-,.,_, o::; <I 5-AXLE TRUCK 15 '-' ~ 1 ::.,_, c.!l :: 5 <I :i::: '-' <l<1 3-AXLE BUS NO DELAY 3-AXLE BUS DELAY RIDING COHFORT INDEX FIGURE 6 Relative change in VOe of heavy vehicles. Trcking cost data from rogh rotes are rgently neeaed for planning. This data wold enable verification of the slope of the cost-roghness fnctions critical in planning and programming road maintenance and rehabilitation. The HDM- 111 assmptions on road srface textre shold also be checked in fll-scale experiments sing representative trcks and tires, so that the rolling resistance base of speed, fel, tire, and time-related trcking costs in HDM-111 can be regarded with confidence. A similar conclsion has been reached by an independent, in-depth research (12). A concerted effort wold be desirable among the Canadian road infrastrctre administrators, the trcking indstry, and the research commnity to carry ot the adaptations and calibrations. REFERENCES 1. Statistics Canada. Trcking in Canada. Cataloge Annal, Ottawa, Ontario, 1984, Traffic World. Spplement, Dec. 5, P. Bein, J. B. Cox, and N. D. Lea. Adapting the Brail/UNDP Cost Research to Canadian Conditions. Research Series, Roads and Transportation Association of Canada, Ottawa, Ontario, Trimac Conslting Services Ltd. Operating Costs of Trcks in Canada, Transport Canada, Policy and Coordination Branch, TP921E, Ottawa, Ontario, N. A. Irwin and R. A. Barton. Economics of Trck Sies and Weights in Canada. Final Report. Roads and Transportation Association of Canada, Ottawa, Ontario, Jly T. Watanatada, C. G. Harral, W. D.. Paterson, A. M. Dhareshwar, A. Bhandari, and K. Tsnokawa. The Highway Design and Maintenance Standards Model HDM-1/1. Volme 1: Model Description, Volme 2: User's Manal. Highway Design and Maintenance Standards Series, Johns Hopkins University Press, Baltimore, Md. and London, England, P. Bein. Adapting HDM-3 User-Cost Model to Saskatchewan Pavement Management Information System. In Transportation Research Record 1229, TRB, National Research Concil, Washington, D.C., A. Chesher and R. Harrison. Vehicle Operating Costs: Evidence From Developing Contries. Highway Design and Maintenance Standards Series, Johns Hopkins University Press, Baltimore, Md. and London, England, R. Bond, A. R. Williams, and G. Lees. An Approach Towards the Understanding and Design of the Pavement's Textral Characteristics Reqired for Optimm Performance of the Tyre. In Proceedings, Symposim on Physics of Tyre Traction. General Motors Research Laboratories, Detroit, Mich., P. C. Crtayne, A. T. Visser, H. W. d Plessis, andr. Harrison. Calibrating the Relationship Between Operating Costs of Bses and Road Roghness on Low-Volme Roads. In Transportation Research Record 116, TRB, National Research Concil, Washington, D.C., G. G. Hayes and D. L. Ivey. Textre, Skid Resistance and the Stability of Atomobiles in Limit Manevers. Special Technical Pblication 583. American Society for Testing and Materials, Philadelphia, Pa., D. C. Biggs. ARFCOM-Models for Estimating Light to Heavy Vehicle Fel Consmption. Research Report ARR152. Astralian Road Research Board, Vermont, Victoria, Sept U. Hammarstriim and B. Karlsson. VETO-A Compter Program for Calclation of Transport Costs as a Fnction of Road Standards (in Swedish). VTI Meddelande 51. Swedish Road and Traffic Research Institte (VTI), Linkiiping, J. P. Zaniewski. Fel Consmption Related to Road Characteristics. In Transportation Research Record 91, TRB, National Research Concil, Washington, D.C., 1983 pp P. Claffey. Fel Consmption Related to Road Characteristics: Discssion. In Transportation Research Record 91, TRB, National Research Concil, Was!1ington, D.C., C. R. Bennett. The New Zealand Vehicle Operating Costs Model. In Transportation Research Record 116, TRB, National Research Concil, Washingtor:, D.C., Pblication of this paper sponsored by Committee on Application of Economic Analysis to Transportation Problems.