Limitations of Using Skid Number in Accident Analysis and Pavement Management

Transcription

1 TRANSPORTATION RESEARCH RECORD Limittions of Using Skid Number in Accident Anlysis nd Pvement Mngement }AMES C. WAMBOLD AND BOHDAN T. KULAKOWSKI Pvement frictionl chrcteristics re commonly determined by conducting skid tests in ccordnce with ASTM Stndrd E274. The results of these tests [skid number (SN)] re used in rnking rods to estblish construction priorities nd to estimte tirepvement friction in ccident nlysis. The role of SN in these two pplictions is ddressed. There re serious limittions constrining the use of SN in ccident nlysis s well s in pvement mngement. Use of SN-bsed estimtes of coefficient of friction my led to lrge errors in estimting the brking distnce or initil speed of vehicle involved in n ccident. In pvement mngement, SN is n importnt fctor in evluting trffic sfety; however, other fctors must lso be considered, such s ccident history dt, driving difficulty, rod geometry, nd trffic chrcteristics. A wet pvement index, incorporting ll fctors relevnt for trffic sfety under wet wether conditions, should be developed to provide bsis for pvement mngement decisions relted to highwy sfety. The role of skid resistnce nd its use in wet ccident nlysis is ddressed. Highwy trffic ccidents re mjor concern of ll stte highwy nd trnsporttion deprtments. One of the mny cuses of highwy ccidents is wet slippery pvements. Becuse ll of the forces needed to control vehicle come from the interction between the tires nd pvement, the frictionl demnd, which vries with rod, tire, wether, nd trffic conditions, must be met under ll driving conditions. The bility of tire-pvement interction to provide dequte trction on wet rods is directly relted to the sfe opertion of vehicles. The importnce of wet-pvement trction to sfety ws emphsized by the findings of the Ntionl Trnsporttion Sfety Bord, which reported tht while the pvements re exposed to precipittion less thn 3.5 percent of the time, wet-pvement ccidents ccount for 13.5 percent of ll ftl ccidents (1). Also, the Ntionl Sfety Council's 1988 summry reports more thn 49,000 people killed nd over 1.8 million injured nd totl estimted cost of vehicle ccidents of pproximtely $70 billion (2). This mens tht motor vehicle ccidents ccount for lmost 50 percent of ll ftlities in the United Sttes nd bout 50 percent of the cost, or tht wet-wether ccidents ccount for bout 7 percent of ll ftlities nd their cost. Reserch on pvement skid resistnce is n ctive ongoing effort t both the federl nd stte level. The stndrdiztion of skid resistnce mesurement in the United Sttes nd the routine clibrtion of test equipment hs gretly enhnced the relibility of these mesurements (ASTM E274). Even Pennsylvni Trnsporttion Institute, Pennsylvni Stte University, Reserch Building B, University Prk, P so, there hs only been moderte success in estblishing cler reltionship between skidding ccidents nd pvement condition (3-7). First wht skid resistnce is nd how it is vried re reviewed. Then the drg fctor (DF), quntity used by ccident investigtors, nd how it reltes to ccident reconstruction, re reviewed. Conditions under which skid number (SN) nd DF re nd re not relted re discussed. Finlly, fctors other thn skid resistnce tht ffect wet pvement sfety re reviewed. ESTIMATING FRICTION IN ACCIDENT ANALYSIS One of the most criticl tsks in ccident nlysis is to find the speed of vehicle involved in n ccident t the time or just before the time when the ccident occurred. In order to determine tht speed, n estimte of the coefficient of tirepvement friction is required. In four-wheel brking, the vehicle speed cn be clculted from the following eqution: dv CoA p mv dx +(V+v) 2 +mgf+mg(g+r)=o (1) where m = V = v x = g f = Co= Ar= p G R mss of the vehicle; vehicle velocity; component of wind velocity in direction opposite to vehicle motion; distnce in forwrd direction of vehicle mesured from onset of brking; grvittionl ccelertion; coefficient of tire-pvement friction; coefficient of ir resistnce; vehicle frontl re; density of mbient ir; grde, positive in upwrd direction; nd coefficient of rolling resistnce. Eqution 1 is difficult to solve nlyticlly, especilly when f is considered to be function of speed, which it is under wet conditions. However, the eqution cn be solved numericlly using simple computer progrm. A copy of such progrm cn be obtined from the Pennsylvni Trnsporttion Institute (PTI) nd is referenced in the new FHW A Skid Resistnce Mnul (8).

2 44 In order to ensure tht the results obtined by solving Eqution 1 re meningful nd ccurte for specific brking conditions, ll prmeters in this eqution must be crefully estimted becuse they re seldom known ccurtely. The most difficult prmeter to estimte is the coefficient of friction between the tires of the vehicle involved in the ccident nd the pvement surfce on which the ccident occurred. The tire-pvement friction is ffected by the type nd condition of the vehicle tires, vehicle suspension, vehicle lod nd lod distribution, wter film thickness, nd pvement surfce chrcteristics, to nme just the most importnt fctors (9,10). In generl, tire-pvement friction hs been estimted either by DF or by SN with no knowledge of how good or bd these estimtes re. Even worse, if DF or SN is not vilble, hndbook vlue of the coefficient of friction is used. Skid Resistnce ASTM Committee El 7 hs the following definition of skid resistnce under considertion: "The retrding force generted by the interction between pvement nd tire under locked, nonrotting wheel conditions." Mesurements of skid resistnce conducted in ccordnce with ASTM Stndrd E274 re reported s SN vlues. SN vlues re mesured t 40 mph nd re equl to the force required to slide the locked test tire divided by the effective wheel lod nd multiplied by 100. If the skid test is performed t nother speed, then the velocity (V) must be reported, i.e., SNv. Pvement SN nd its mny surrogtes rc ccepted worldwide s mesure of pvement frictionl chrcteristics; s SN increses, the pvement frictionl chrcteristics improve. Results of skid resistnce mesurements provide the bsis for pvement mngement sfety decisions. One of the primry objectives of pvement mngement systems is to provide motorists with highwy surfces tht llow them to drive sfely under wet pvement conditions. An importnt prt of this process is the identifiction of sections of rods tht do not provide dequte friction under wet pvement conditions for vehicles performing brking, cornering, lne chnging, nd other mneuvers. Skid resistnce plys the key role in rnking rods ccording to their sfety under wet conditions. A number of studies hve been conducted to determine the correltion between SN nd wetpvement ccident rte (7,11-13). SN represents one of the most importnt fctors for estblishing wet-pvement sfety. However, other fctors must lso be considered when mking pvement mngement decisions: rod geometrics, driving conflicts, visibility, trffic chrcteristics (including percent of trucks), nd verge wet time (verge period of time during yer when the rod is wet). Severl studies hve recently been completed or re currently underwy to devefop n integrted pproch to the problem of evluting the sfety of wet pvements both brod nd in the United Sttes (14-17). Trnsporttion deprtments re ble to provide pvements with good skid resistnce, nd the ASTM locked-wheel skid tester is relible tool to rnk pvements. However, improvement is needed in pplying this rnking, becuse SN only rnks the pvement skid resistnce, not the vehicle demnd for it nor the ctul friction of prticulr vehicle tirepvement interction. Drg Fctor TRANSPORTATION RESEARCH RECORD 1311 In the field of ccident reconstruction, DF is used rther thn coefficient of friction. This fctor is obtined by conducting vehicle brking test with locked wheels. A correct test tkes plce t the ccident site nd involves similr vehicle, similr tires, nd similr wether conditions to those of the ccident. In order to be correct, the test should lso be conducted t the estimted speed of the vehicle or vehicles involved in the ccident. DF then becomes function of mny vribles, including friction, ir resistnce, grde, vehicle weight distribution, nd wether. DF is clculted from the following eqution: v2 DF = 30D where V = vehicle velocity before brking, mph; nd D = distnce trveled fter brking, ft. Thus, using Eqution 2 in the form V = (30 x D x DF) 1 ' 2 yields the initil velocity for known stopping distnce, D, nd known DF vlues, or y2 n = - x DF 30 yields the stopping distnce for known initil velocity, V, nd DF vlues. Equtions 2-4 re then used in plce of Eqution 1 nd include ll of the effects of Eqution 1. Skid Number If SN is used, SN/100 becomes n estimte of the coefficient of friction in Eqution 1. SN divided by 100 cn lso be used s n estimte of the reltionship for friction versus speed in the following eqution (18): f(v) = SN(V) = (SN oev) where SN 0 (zero-velocity intercept) nd PNG (percent normlized grdient) re the model prmeters. The vlues of SN 0 nd PNG cn be estimted from locked-wheel SN vlues obtined with ribbed or blnk tires or by conducting severl skid tests t three or more different speeds nd doing log regression to Eqution 3 (19). For mesurements of SN in ccordnce with ASTM E274, stndrd ASTM E501 ribbed tire is used. This stndrd tire does not represent ny specific type of vehicle tire nd in fct is belted tire, wheres presently, most pssenger cr tires re rdil tires. Therefore, SN cn only be considered n estimte of friction between n ctul vehicle tire nd rod surfce, nd by itself neglects ll the other prmeters in Eqution 1. Unfortuntely, circumstnces often do not llow for such tests to be performed, nd then n SN or n SN- (2) (3) (4) (5)

3 Wmbold nd Kulkowski bsed estimte of tire-pvement friction is used. For pssenger crs, n estimte of the men effective coefficient of friction ws proposed (20): = SN 0 (PNG/100) 2 Vf V; 200 f - [ ( )] -] PNG PNG ' 100 ' x 1 - e 100v v where V; is the initil velocity of vehicle. This estimte ws developed for new nd slightly worn pssenger cr tires of similr size nd construction s the ASTM ribbed tire. It tends to overestimte friction for modertely worn tires nd for ll tires when the wter film thickness is more thn 0.05 in. Of even more concern is the cse of using the stndrd SN vlue t speed of 40 mph s the estimte of the coefficient of friction. Here the SN is not known s function of velocity so tht not only is the tire type incorrect, but there is no estimte of the men effective friction. The use of SN, which is mesured in locked-wheel test, hs even more error if the cr is equipped with n ntilock brke system (ABS) nd could result in errors of 100 percent or more on low-friction surfces where brking is ner the pek coefficient rther thn t locked wheel. Estimtion of tire-pvement friction is more complicted for trucks nd buses. Dt on the coefficient of friction for vrious types of truck tires were reported in severl studies (21,22). Truck nd bus tires re designed primrily for highwer resistnce nd usully hve lower friction coefficients thn pssenger cr tires. It is generlly estimted tht the locked-wheel coefficient of friction of truck tires is bout 70 percent of tht for pssenger cr tires. A procedure for predicting the brking distnces of trucks operting on poor, wet rods ws developed by Olson et l. (23). The pek lockedwheel friction (JP) of new truck tire is estimted by fp(v) = l.45fs(v) (7) where fs is sliding coefficient of friction. Truck drivers modulte brkes during brking to void wheel lock nd mintin directionl control. This results in the effective truck tire friction coefficient being higher thn the sliding coefficient of friction, f,, but still lower thn the pek coefficient of friction, fp Moreover, on the bsis of experimentl dt, truck drivers ttined pproximtely 62 percent of the performnce cpbilities of the rod-tire-vehicle system represented by the pek coefficient of friction (23). The effective truck tire friction cn be estimted by SN, using the following eqution (8,23): f(v) = 0.62 x 1.45 x SN(V) (8) or, equivlently f(v) = SN(V) (9) As in the cse of pssenger cr tires, SN dependence on speed must be known to clculte the coefficient of friction s function of speed, using Eqution 9. Remember tht SN only provides n estimte of the coefficient of friction between n ctul vehicle nd the rod surfce. It is generlly ccepted tht vehicle tire-pvement friction improves s SN increses. Quntittively, however, (6) there my be considerble difference between SN nd the tire-pvement coefficient of friction for two resons. First, s mentioned erlier, the properties of the stndrd ASTM tire used in skid tests re different from the properties of ctul utomobile tires, nd they re bis insted of rdil. With the vriety of cr nd truck tires currently in use, it is difficult to predict how their coefficients of friction will compre with those of the ASTM tire. Previous reserch hs shown tht the ASTM E501 ribbed tire produces friction levels lower thn those produced by comprble, new, nd slightly worn pssenger cr tire of similr size nd construction, s shown in Figure 1 (20). However, when wter film thickness is incresed, the pssenger cr tire produces lower levels of friction thn the ASTM tire, s shown in Figure 2. How the ASTM test tire nd ctul vehicle tires compre depends lso on the type of pvement surfce (24). On polished, wet surfces, the skid resistnce of the ASTM tire ws considerbly lower thn tht of conventionl tires used on motor vehicles. The trend is reversed on higher-friction surfces hving SN greter thn 20. Second, tire operting conditions including lod, infltion pressure, speed, percent slip, wter film thickness, nd vehicle suspension chrcteristics during skid test re different from those of vehicle tire. Current knowledge of the reltionship between the pvement skid resistnce nd vehicle tire-pvement coefficient of friction does not provide method for clculting the error involved in pproximting the tire-pvement coefficient of friction by SN. The following exmples describing ctul cses illustrte the consequences of using n SN-bsed estimte for tire-pvement friction. Exmple 1 The results of brking test performed with Pontic Sunbird on wet section of bituminous pvement were s follows:.70 ng -<>- Poneniier Cr -E!IOI -E524 : V (km/h) FIGURE 1 Coefficient of friction versus speed for the ASTM tires nd the pssenger cr tire when new-site 1, 1980 (1 km = 0.6 mi) (21). 45

4 46 TRANSPORTATION RESEARCH RECORD 1311 Ifil I. -<>- Penger Cr -&-E501 -O-E Cse C: Coefficient of Friction Estimted Using Eqution 6 First, the prmeters SN 0 nd PNG, describing the skid resistnce dependence on speed, must be determined. Becuse ribbed tire SN vlues t two different speeds re known, SN 0 nd PNG cn be clculted directly to be SN 0 = nd PNG = In order to clculte the verge coefficient of friction from Eqution 6, n initil speed of the vehicle must be ssumed. For n ssumed speed of 40 mph, the verge coefficient of friction is then clculted s follows: 101.l x x 40 2 / = r IB V (km/h) FIGURE 2 Coefficient of friction versus speed for the three tires on the brking distnce site, 1981 (1 km = 0.6 mi). x [ 1 - = 0.54 e x 40 ( x 40) rl Hence, the predicted initil speed is V; 5.5(d x!) (34 x 0.54) 112 = 23.6 mph (12) (13) Initil speed, V; = 30 mph; nd Brking distnce, d = 34.0 ft. The results of the skid tests conducted on the sme pvement nd under the sme wether conditions were s follows: Ribbed tire SN t 40 mph, SN 40 = 39.8; nd Ribbed tire SN t 30 mph, SN 30 = Cse A: Coefficient of Friction Estimted by SN 40 /IOO This sitution is likely to occur in rel life. The vehicle speed is unknown; therefore, stndrd ASTM E274 test t 40 mph is used to estimte tire-pvement friction, f = SN 40 /100 = Becuse the brking distnced is known, the estimte of the initil speed is then clculted s follows: V; 5.5(d x {) (34 x 0.398) 112 = 20.2 mph (10) The ctul initil speed is 30 mph; therefore, the speed prediction error is -9.8 mph, or percent. Cse B: Coefficient of Friction Estimted by SN 31,/l00 In this cse, SN mesured t the ctul vehicle speed is used, which is not likely to hppen in rel life. The predicted initil speed, ssuming tht the brking distnce is known, is V; = 5.5(d x f) 112 = 5.5(34 x 0.503) 112 = 22.7 mph (11) which gives the speed prediction error of mph, or percent. This vlue is different from the ssumed vlue of 40 mph. In order to improve ccurcy, the process of clculting V; is repeted, until the ssumed nd predicted vlues of V; re the sme. Repeting the clcultion off nd V; yields f = 0.67 nd V; = The prediction error is -3.7 mph or percent. The results of predicting initil speed bsed on the different estimtes of the tire-pvement coefficient of friction obtined in this exmple re presented in Tble 1. This exmple demonstrtes tht the use of SN s n estimte of tire-pvement friction my led to lrge errors in predicting vehicle speed. The mgnitude of the error is likely to be prticulrly high when the coefficient of friction is estimted by SN/100. Exmple 2 The results of severl brking tests on both wet nd dry surfces of section of the bituminous skid pd t the PTI test trck re presented in Tble 2. The results obtined with n ASTM skid triler on the sme surfces re presented in Tble 3. On the bsis of regression of the wet E501 (ribbed) tire tests, the following vlues were clculted: SN 0 = 80.17, SN 0 (pek) = 107.7, TABLE 1 RESULTS OF PREDICTING INITIAL SPEED FOR EXAMPLE 1 Sn 40 /100 sn 30 /100 f, eq. 6 Predicted Speed, mi/h Absolute Error, mi/h Reltive Error, %

5 Wmbold nd Kulkowski 47 TABLE 2 RESULTS OF BRAKING TESTS CONSIDERED IN EXAMPLE 2 Vehicle Wet/Dry v,, rni/h d, feet DF Pontic Sunbird Dry B. B6 Pontic Sunbird Wet Trnsit Bus Dry , 53 Trnsit Bus Dry Lincoln Dry BO Cdillc (with ABS) Dry o. Bl Dump Truck Dry TABLE 3 RESULTS OF SKID RESISTANCE TESTS FOR EXAMPLE 2 Tire Condition Speed SN Pek E501 (ribbed) E501 (ribbed) E501 (ribbed) E524 (blnk) E501 (ribbed) Wet Wet Wet Wet Dry 29.B B 90.B B.l PNG = 1.563, nd PNG (pek) = Using these vlues, the clcultions of Cses A, B, nd C in Exmple 1 were repeted to obtin the results presented in Tbles 4-6. OTHER FACTORS AFFECTING WET-PAVEMENT SAFETY The exmples presented in the previous section demonstrted tht using SN or SN-bsed estimtes of tire-pvement friction cn led to very lrge errors in clcultions of speed nd stopping distnce in ccident nlysis. SN lone is not good mesure of wet pvement sfety either. Figure 3 shows considerble sctter of dt in the rtio of wet to dry ccidents versus the SN plot obtined by Hvens in 1979 (24). Tht nothing hs chnged in respect to the reltionship between number of wet ccidents nd SN is demonstrted by Figure 4, obtined from the vlidtion dt set collected for the 1990 study (17). Although the process of identifiction of sections tht do not provide dequte trction under wet-pvement conditions is n importnt prt of pvement mngement, it is but one fctor needed in providing motorists with highwy surfces tht llow them to drive sfely. Mny other fctors ffect sfety under wet-pvement conditions, nd it is only when these conditions demnd prticulr level of trction tht SN then becomes importnt. TABLE 4 RES UL TS OF PREDICTING INITIAL SPEED FOR EXAMPLE 2, CASE A (SN 4 ofl00 IS USED AS f) v, Clculted V 1 Actul Percent Vehicle Condition f mi/h mi/h Error Pontic Sunbird Dry ' Pontic Sunbird Wet. 37B Trnsit Bus Dry Trnsit Bus Dry ' Lincoln Dry Cdillc (ABS) Dry Cdillc (ABS) Dry Dump Truck Wet

6 48 TRANSPORTATION RESEARCH RECORD 1311 TABLE 5 RESULTS OF PREDICTING INITIAL SPEED FOR EXAMPLE 2, CASE B [SN(v) NEAREST TEST SPEED IS USED] V 1 Clculted V 1 Actul Percent Vehicle Condition f mi/h mi/h Error Pontic Sunbird \let Trnsit Bus Dry Cdillc (ABS) Dry l Dump Truck \let TABLE 6 RESULTS OF PREDICTING INITIAL SPEED FOR EXAMPLE 2, CASE C [SN(v) IS CALCULATED] V 1 Clculted V1 Actul Percent Vehicle Condition f mi/h mi/h Error Pontic Sunbird \let Dump Truck \let , <I 0.5 > 'if 0.4 >- 0:: : 0.2 ll :: I W W M SKID NUMBER FIGURE 3 Ten-point moving verges of wet-to-dry pvement ccidents for 120 test sections in Kentucky with ADT bove 2,700 versus SN. 32 c 2B!!! z llj 0 24 u 20 l;j 31: c 16 0:: 12 g c c llj (D c ce :::E c :::> 8 c CCC z OCIIIDC D Ee 0 o cn::c 0 c =::IDD c c c 4 c r:d:d c rn::dc CIJ cc c:c :u:r:ttd C D ID CDJ CC c c 09 D IDt:c "Ill 1111 I iii! H 1HR:i D CDJ SKID NUMBER (SN) FIGURE 4 Number of wet ccidents versus SN obtined from the vlidtion dt set. 70 All drivers in the trffic flow cn mintin directionl stbility of their vehicles if the friction demnd exerted by their vehicles does not exceed the vilble friction. The vilble friction depends on the pvement skid resistnce. Friction demnd is function of vehicle speed, rod geometry, trffic chrcteristics (including trffic density nd percentge of trucks in the trffic flow), vehicle chrcteristics (including vehicle type nd its understeer), nd driver skills. SN, mesured on given section of rod, represents the vilble friction t 40 mph. Whether or not this prticulr SN is sufficiently high to prevent wet ccidents cnnot be determined unless the demnd for friction on the section of rod under considertion is known. The effects of ll importnt fctors on friction demnd must be considered in pvement mngement. Vehicle Speed Speed is criticl fctor for the blnce between friction demnd nd supply becuse it ffects both. When speed increses, friction demnd lso increses. For instnce, centrifugl forces generted during vehicle cornering, which must be countercted by tire-pvement friction forces to prevent vehicle from skidding off the rod, re proportionl to the squre of vehicle speed. At the sme time, the pvement skid resistnce decreses with incresing speed in n pproximtely exponentil mnner (Eqution 5). Rod Geometry The mount of friction required for sfe driving is strongly ffected by rod geometry. Friction demnd on stright sections of rod is low if the rod is level, if vehicles trvel t constnt speed, nd if there re no intersections. The demnd for friction increses significntly if grde or curve must be negotited (12). Pge nd Buts (25) concluded tht pve-

7 Wmbold nd Kulkowski ment ccident rtes re significntly higher on curves thn on ny other type of geometric lignment. The effect of curvture on wet-ccident rtes ws found to be prticulrly strong on pvements hving SN vlues less thn 25. When SN is less thn 25, wet-pvement ccident rtes re significntly greter for both uphill nd downhill slopes steeper thn 3 percent thn for fltter terrin. Trffic Flow Trffic volume, in generl, does not hve significnt influence on wet-ccident rtes (26). However, under specil circumstnces, nmely, on undivided highwys hving SN vlues less thn 25, wet-ccident rtes were found to increse significntly when the verge dily trffic ws greter thn 15,000 vehicles per dy (25). Trffic composition, in prticulr, the percentge of trucks in the trffic flow, hs significnt effect on friction demnd. This is becuse the stopping distnces of trucks re 1.3 to 2.8 times longer thn the stopping distnces of pssenger crs (26). Driving difficulty is nother importnt chrcteristic of trffic flow. The criteri determining driving difficulty include number of ccess points per segment of rod, presence of turn lnes, type of surrounding lnd use, trffic signliztion, nd the rodwy cross section. Vehicle Type Figure 5 shows stopping distnces of buses nd vrious configurtions of trucks versus the stopping distnce for typicl pssenger cr from 60 mph (96 km/hr) on dry rod (27). The friction demnd for buses nd ll types of trucks is higher thn for pssenger crs if equl stopping distnce is required for ll vehicles. The friction demnd is lso higher for vehicles with lower degrees of understeer (20). Driver Skills Few drivers cn operte their vehicles t 100 percent efficiency, i.e., using 100 percent of the vilble friction. Olson Buses L..1 0 Crs Stopping Distnce, ft FIGURE 5 Stble stopping distnce from 60 mph (96 km/hr) on dry rod. RnQe for 11 Typicl 11 Vehicles et l. (23) found tht truck driver efficiencies rnge from 62 to 100 percent nd tht most drivers hve little or no prctice in emergency brking situtions. The concern over driver brking skills will be llevited when ntilock brke systems become commonly used. In order to fully evlute the risk of wet-pvement ccident, ll fctors influencing the demnd for friction nd vilble friction must be considered. Severl reserch studies re currently under wy in which mthemticl models of pvement sfety index re being developed. These sttisticl models estimte the probbility of wet pvement ccidents s function of the vribles ffecting the blnce between friction demnd nd vilble friction. CONCLUSION In generl, the use of SN 40 s coefficient of tire-pvement friction would be expected to underestimte the speed or overestimte the stopping distnce. In the cse of truck or bus, this ws not true; however, if the 75.5 percent rule is used for truck tires, then the bus velocity would hve lso been underestimted. Using SN(V) t the speed of the ccident is n improvement, but it still underestimtes the speed. Similrly, clculting men friction for SN(V) over the speed of interest yields n even better estimte, but still underestimtes the speed. This is to be expected becuse n ASTM tire ws found to exhibit lower SN thn regulr tire. In the cse of pvement mngement, rnking rods on the bsis of SN only is not wise. Certfoly long, stright rod with n SN vlue of 30 nd ADT of 1 is of no concern when there re so mny rods with high ADT vlues, curves, intersections, etc., t the sme or even higher SN vlues, which re thus more likely to hve wet-wether ccidents. Rods with good SN vlues re not problems with regrd to skidding, but rods with low SN vlues (e.g., below 20) need to be ddressed. For cses in between, the mjor problem re, it is recommended tht n index tht gives the probbility of wet-wether ccident should be used in rnking which rod should be repired first on sfety bsis. REFERENCES 1. Sfety Effectiveness Evlution: Selected Stte Highwy Skid Resistnce Progrms. Report NTSB-SEE Ntionl Trnsporttion Sfety Bord, Wshington, D.C., Sept Accident Fcts, Ntionl Sfety Council, Wshington, D.C., R. R. Hegmon. Wet Wether Accidents nd Pvement Skid Resistnce. Public Rods, Sept. 1981, p J. C. Wmbold, J. J. Henry, nd R. R. Hegmon. Skid Resistnce of Wet Wether Accident Sites. The Tire Pvement Interfce, STP 929. ASTM, Phildelphi, P., June 1985, pp J. R. Hosking. Reltionship between Skidding Resistnce nd Accident Frequency: Estimtes Bsed on Sesonl Vrition. Reserch Report 76. Trnsport nd Rod Reserch Lbortory, Crowthorne, Berkshire, U.K., J. M. Zuiebck et l. Frictionl Requirements Necessry to Reduce Skidding Accident Frequencies. Report SAI LA. FHWA, U.S. Deprtment of Trnsporttion, R.R. Blckbum, D. W. Hrwood, A. D. St. John, nd M. C. Shrp. Evlution of Accident Rte-Skid Number Reltionships, Effectiveness of Alterntive Skid Reduction Mesures, Vol. 1. Report FHWA-RD FHWA, U.S. Deprtment of Trnsporttion, Nov B. T. Kulkowski et l. Skid Resistnce Mnul. Pennsylvni Trnsporttion Institute, University Prk, P., July

8 50 9. R. R. Hegmon. Tire-Pvement Interction. Technicl Pper SAE, Wrrendle, P., 1987, pp A. G. Veith. Tire-Rods-Rinfll-Vehicles: The Trction Connection. STP 793. ASTM, Phildelphi, P., 1983, pp H. W. Kummer nd W. E. Meyer. NCHRP Report 37: Tenttive Skid Resistnce Requirements for Min Highwys. HRB, Ntionl Reserch Council, Wshington, D.C., E. Frber, M. S. Jnoff, S. Cristinzio, J. G. Blubugh, W. Reisener, nd W. Dunning. NCHRP Report 154: Determining Pvement Skid Resistnce Requirements t Intersections nd Brking Sites. TRB, Ntionl Reserch Council, Wshington, D.C., A. B. Moore nd J. B. Humphreys. High Speed Skid Resistnce nd the Effects of Surfce Texture on the Accident Rte. In Skid Resistnce of Highwy Pvements, STP 530. ASTM, Phildelphi, P., 1973, pp OECD Rod Sfety Reserch: A Synthesis. Orgniztion for Economic Coopertion nd Development Scientific Experts Group, Pris, Frnce, Anti-Skid Tests in U.K. Interntionl Rod Federtion, Pris, Frnce, April D. L. Ivey nd L. I. Griffin III. Development of Wet Wether Sfety Index. Report FHWATX F. Texs Stte Deprtment of Highwys nd Public Trnsporttion, Austin, Nov B. T. Kulkowski et l. Development of Methodology to Identify nd Correct Slippery Pvements. Reserch Project Pennsylvni Deprtment of Trnsporttion, University Prk, P., Aug J. J. Henry, K. Sito, nd R. Blckburn. Predictor Model for Sesonl Vritions in Skid Resistnce: Comprehensive Finl Report, Report FHWA-RD-83/005. FHWA, U.S. Deprtment of Trnsporttion, J. C. Wmbold. Obtining Skid Number t Any Speed from Test t Single Speed. In Trnsporttion Reserch Record 1196, TRB, Ntionl Reserch Council, Wshington, lj.c., TRANSPORTATION RESEARCH RECORD J. J. Henry. Comprison of the Friction Performnce of Pssenger Cr Tire nd the ASTM Stndrd Test Tires. Specil Technicl Publiction 793, ASTM, Phildelphi, P., 1983, pp P. S. Fncher, R. D. Ervin, C. B. Winkler, nd T. D. Gillespie. A Fctbook of the Mechnicl Properties of the Components of Single-Unit nd Articulted Hevy Vehicles. Report DOT HS NHTSA, U.S. Deprtment of Trnsporttion, I. Guskov, R. Rice, S. Pugliese, nd R. Glgnski. An Evlution of Methods to Investigte Truck Tire Wet Trction. Report DOT HS NHTSA, U.S. Deprtment of Trnsporttion, P. L. Olson, D. E. Clevelnd, P. S. Fncher, L. P. Kostynick, nd L. W. Schneiter. NCHRP Report 270: Prmeters Affecting Stopping Sight Distnce. TRB, Ntionl Reserch Council, Wshington, D.C., June W. Bergmn. Skid Resistnce Properties of Tires nd Their Influence on Vehicle Control. In Trnsporttion Reserch Record 621, TRB, Ntionl Reserch Council, Wshington, D.C., 1977, pp B. G. Pge nd L. F. Buts. Evlution of Friction Requirements for Cliforni Stte Highwys in Terms of Highwy Geometrics. Report FHWA/CA/TL-86/01. FHWA, U.S. Deprtment of Trnsporttion, Jn D. C. Mhone nd S. N. Runkle. Pvement Friction Needs. In Highwy Reserch Record 396, HRB, Ntionl Reserch Council, Wshington, D.C., 1972, pp R. W. Row!inski nd S. F. Willims. NHTSA Hevy Duty Vehicle Brke Reserch Progrm-Report No. 1: Stopping Cpbility of Air Brked Vehicles. DOT HS nd 739. NHTSA, U.S. Deprtment of Trnsporttion, April Publiction of this pper sponsored by Committee on Surfce Properties-Vehicle Interction.