TRANSPORTATION RESEARCH RECORD 1311 43

Limitations of Using Skid Number in

Accident Analysis and Pavement
Management
}AMES C. WAMBOLD AND BOHDAN T. KULAKOWSKI

Pavement frictional characteristics are commonly determined by so, there has only been moderate success in establishing a
conducting skid tests in accordance with ASTM Standard E274. clear relationship between skidding accidents and pavement
The results of these tests [skid number (SN)] are used in ranking condition (3-7).
roads to establish construction priorities and to estimate tire-
First what a skid resistance is and how it is varied are
pavement friction in accident analysis. The role of SN in these
two applications is addressed. There are serious limitations con- reviewed. Then the drag factor (DF), a quantity used by
straining the use of SN in accident analysis as well as in pavement accident investigators, and how it relates to accident recon-
management. Use of SN-based estimates of coefficient of friction struction, are reviewed. Conditions under which skid number
may lead to large errors in estimating the braking distance or (SN) and DF are and are not related are discussed. Finally,
initial speed of a vehicle involved in an accident. In pavement factors other than skid resistance that affect wet pavement
management, SN is an important factor in evaluating traffic safety; safety are reviewed.
however, other factors must also be considered, such as accident
history data, driving difficulty, road geometry, and traffic char-
acteristics. A wet pavement index, incorporating all factors rel- ESTIMATING FRICTION IN ACCIDENT
evant for traffic safety under wet weather conditions, should be
ANALYSIS
developed to provide a basis for pavement management decisions
related to highway safety.
One of the most critical tasks in accident analysis is to find
the speed of a vehicle involved in an accident at the time or
The role of skid resistance and its use in wet accident analysis just before the time when the accident occurred. In order to
is addressed. Highway traffic accidents are a major concern determine that speed, an estimate of the coefficient of tire-
of all state highway and transportation departments. One of pavement friction is required. In four-wheel braking, the ve-
the many causes of highway accidents is wet slippery pave- hicle speed can be calculated from the following equation:
ments. Because all of the forces needed to control a vehicle
come from the interaction between the tires and pavement, dV CoA p
the frictional demand, which varies with road, tire, weather, mV dx +~(V+v) 2 +mgf+mg(G+R)=O (1)
and traffic conditions, must be met under all driving condi-
tions. The ability of a tire-pavement interaction to provide where
adequate traction on wet roads is directly related to the safe
operation of vehicles. The importance of wet-pavement trac- m = mass of the vehicle;
tion to safety was emphasized by the findings of the National V = vehicle velocity;
Transportation Safety Board, which reported that while the v component of wind velocity in direction opposite to
pavements are exposed to precipitation less than 3.5 percent vehicle motion;
of the time, wet-pavement accidents account for 13.5 percent x = distance in forward direction of vehicle measured
of all fatal accidents (1). Also, the National Safety Council's from onset of braking;
1988 summary reports more than 49,000 people killed and g gravitational acceleration;
over 1.8 million injured and a total estimated cost of vehicle f = coefficient of tire-pavement friction;
accidents of approximately $70 billion (2). This means that Co= coefficient of air resistance;
motor vehicle accidents account for almost 50 percent of all Ar= vehicle frontal area;
fatalities in the United States and about 50 percent of the p density of ambient air;
cost, or that wet-weather accidents account for about 7 per- G grade, positive in upward direction; and
cent of all fatalities and their cost. R coefficient of rolling resistance.
Research on pavement skid resistance is an active ongoing Equation 1 is difficult to solve analytically, especially when f
effort at both the federal and state level. The standardization is considered to be a function of speed, which it is under wet
of skid resistance measurement in the United States and the conditions. However, the equation can be solved numerically
routine calibration of test equipment has greatly enhanced using a simple computer program . A copy of such a program
the reliability of these measurements (ASTM E274). Even can be obtained from the Pennsylvania Transportation Insti-
Pennsylvania Transportation Institute, Pennsylvania State Univer- tute (PTI) and is referenced in the new FHW A Skid Resis-
sity, Research Building B, University Park, Pa. 16802. tance Manual (8).
44 TRANSPORTATION RESEARCH RECORD 1311

In order to ensure that the results obtained by solving Equa- Drag Factor
tion 1 are meaningful and accurate for specific braking con-
ditions, all parameters in this equation must be carefully es- In the field of accident reconstruction, DF is used rather than
timated because they are seldom known accurately. The most a coefficient of friction. This factor is obtained by conducting
difficult parameter to estimate is the coefficient of friction a vehicle braking test with locked wheels. A correct test takes
between the tires of the vehicle involved in the accident and place at the accident site and involves a similar vehicle, similar
the pavement surface on which the accident occurred. The tires, and similar weather conditions to those of the accident.
tire-pavement friction is affected by the type and condition In order to be correct, the test should also be conducted at
of the vehicle tires, vehicle suspension, vehicle load and load the estimated speed of the vehicle or vehicles involved in the
distribution, water film thickness, and pavement surface char- accident. DF then becomes a function of many variables,
acteristics, to name just the most important factors (9,10) . In including friction, air resistance, grade, vehicle weight distri-
general, tire-pavement friction has been estimated either by bution, and weather. DF is calculated from the following
DF or by SN with no knowledge of how good or bad these equation:
estimates are. Even worse, if DF or SN is not available, a
handbook value of the coefficient of friction is used. v2
DF = 30D (2)

Skid Resistance where

ASTM Committee El 7 has the following definition of skid V = vehicle velocity before braking, mph; and
resistance under consideration: "The retarding force gener- D = distance traveled after braking, ft.
ated by the interaction between a pavement and a tire under Thus, using Equation 2 in the form
locked, nonrotating wheel conditions." Measurements of skid
resistance conducted in accordance with ASTM Standard E274 V = (30 x D x DF) 1' 2 (3)
are reported as SN values. SN values are measured at 40 mph
and are equal to the force required to slide the locked test yields the initial velocity for known stopping distance, D, and
tire divided by the effective wheel load and multiplied by 100. known DF values, or
If the skid test is performed at another speed, then the velocity
(V) must be reported, i.e., SNv. Pavement SN and its many y2
surrogates arc accepted worldwide as a measure of pavement n=- x DF (4)
30
frictional characteristics; as SN increases, the pavement fric-
tional characteristics improve. Results of skid resistance mea- yields the stopping distance for known initial velocity, V, and
surements provide the basis for pavement management safety DF values. Equations 2-4 are then used in place of Equation
decisions. 1 and include all of the effects of Equation 1.
One of the primary objectives of pavement management
systems is to provide motorists with highway surfaces that
allow them to drive safely under wet pavement conditions. Skid Number
An important part of this process is the identification of sec-
tions of roads that do not provide adequate friction under wet If SN is used, SN/100 becomes an estimate of the coefficient
pavement conditions for vehicles performing braking, cor- of friction in Equation 1. SN divided by 100 can also be used
nering, lane changing, and other maneuvers. Skid resistance as an estimate of the relationship for friction versus speed in
plays the key role in ranking roads according to their safety the following equation (18):
under wet conditions. A number of studies have been con-
ducted to determine the correlation between SN and wet-
pavement accident rate (7,11-13) . f(V) = SN(V) = (SNoe~V) (5)
SN represents one of the most important factors for estab- 100 100
lishing wet-pavement safety. However, other factors must also
be considered when making pavement management decisions: where SN0 (zero-velocity intercept) and PNG (percent nor-
road geometrics, driving conflicts, visibility, traffic charac- malized gradient) are the model parameters. The values of
teristics (including percent of trucks), and average wet time SN0 and PNG can be estimated from locked-wheel SN values
(average period of time during a year when the road is wet). obtained with ribbed or blank tires or by conducting several
Several studies have recently been completed or are currently skid tests at three or more different speeds and doing a log
underway to devefop an integrated approach to the problem regression to Equation 3 (19).
of evaluating the safety of wet pavements both abroad and For measurements of SN in accordance with ASTM E274,
in the United States (14-17). a standard ASTM E501 ribbed tire is used. This standard tire
Transportation departments are able to provide pavements does not represent any specific type of vehicle tire and in fact
with good skid resistance, and the ASTM locked-wheel skid is a belted tire, whereas presently, most passenger car tires
tester is a reliable tool to rank pavements. However, im- are radial tires. Therefore, SN can only be considered an
provement is needed in applying this ranking, because SN estimate of friction between an actual vehicle tire and road
only ranks the pavement skid resistance, not the vehicle de- surface, and by itself neglects all the other parameters in
mand for it nor the actual friction of a particular vehicle tire- Equation 1. Unfortunately, circumstances often do not allow
pavement interaction. for such tests to be performed, and then an SN or an SN-
 Wambold and Kulakowski 45

based estimate of tire-pavement friction is used. For passenger there may be a considerable difference between SN and the
cars, an estimate of the mean effective coefficient of friction tire-pavement coefficient of friction for two reasons.
was proposed (20): First, as mentioned earlier, the properties of the standard
- = SN 0 (PNG/100) 2 Vf ASTM tire used in skid tests are different from the properties
f V; 200 of actual automobile tires, and they are bias instead of radial.
-]
With the variety of car and truck tires currently in use, it is
x
[ PNG

(
1 - e100v 1 -
'
PNG
--v
100 ' )] (6)
difficult to predict how their coefficients of friction will com-
pare with those of the ASTM tire. Previous research has
shown that the ASTM E501 ribbed tire produces friction levels
where V; is the initial velocity of a vehicle. This estimate was lower than those produced by a comparable, new, and slightly
developed for new and slightly worn passenger car tires of a worn passenger car tire of similar size and construction, as
similar size and construction as the ASTM ribbed tire. It tends shown in Figure 1 (20). However, when water film thickness
to overestimate friction for moderately worn tires and for all is increased, the passenger car tire produces lower levels of
tires when the water film thickness is more than 0.05 in. friction than the ASTM tire, as shown in Figure 2. How the
Of even more concern is the case of using the standard SN ASTM test tire and actual vehicle tires compare depends also
value at a speed of 40 mph as the estimate of the coefficient on the type of pavement surface (24). On polished, wet sur-
of friction. Here the SN is not known as a function of velocity faces, the skid resistance of the ASTM tire was considerably
so that not only is the tire type incorrect, but there is no lower than that of conventional tires used on motor vehicles.
estimate of the mean effective friction. The use of SN, which The trend is reversed on higher-friction surfaces having SN
is measured in a locked-wheel test, has even more error if greater than 20.
the car is equipped with an antilock brake system (ABS) and Second, tire operating conditions including load, inflation
could result in errors of 100 percent or more on low-friction pressure, speed, percent slip, water film thickness, and vehicle
surfaces where braking is near the peak coefficient rather than suspension characteristics during a skid test are different from
at locked wheel. those of a vehicle tire.
Estimation of tire-pavement friction is more complicated Current knowledge of the relationship between the pave-
for trucks and buses. Data on the coefficient of friction for ment skid resistance and vehicle tire-pavement coefficient of
various types of truck tires were reported in several studies friction does not provide a method for calculating the error
(21,22). Truck and bus tires are designed primarily for high- involved in approximating the tire-pavement coefficient of
wear resistance and usually have lower friction coefficients friction by SN. The following examples describing actual cases
than passenger car tires. It is generally estimated that the illustrate the consequences of using an SN-based estimate for
locked-wheel coefficient of friction of truck tires is about 70 tire-pavement friction.
percent of that for passenger car tires. A procedure for pre-
dicting the braking distances of trucks operating on poor, wet
Example 1
roads was developed by Olson et al. (23). The peak locked-
wheel friction (JP) of a new truck tire is estimated by
The results of a braking test performed with a Pontiac Sunbird
on a wet section of bituminous pavement were as follows:
fp(V) = l.45fs(V) (7)

where fs is a sliding coefficient of friction. Truck drivers mod- . 70 nag

ulate brakes during braking to avoid wheel lock and maintain -<>- Poneniier Car
-E!IOI
directional control. This results in the effective truck tire fric- -E524
tion coefficient being higher than the sliding coefficient of
friction, f,, but still lower than the peak coefficient of friction,
fp· Moreover, on the basis of experimental data, truck drivers :50
attained approximately 62 percent of the performance capa-
bilities of the road-tire-vehicle system represented by the peak
coefficient of friction (23). The effective truck tire friction .40

can be estimated by SN, using the following equation (8,23):

.3
f(V) = 0.62 x 1.45 x 0.0084SN(V) (8)

or, equivalently .20

f(V) = 0.00755SN(V) (9) .10

As in the case of passenger car tires, SN dependence on

speed must be known to calculate the coefficient of friction 16 32 48 64 80
as a function of speed, using Equation 9. V (km/h)
Remember that SN only provides an estimate of the coef-
FIGURE 1 Coefficient of friction versus
ficient of friction between an actual vehicle and the road speed for the ASTM tires and the passenger
surface. It is generally accepted that vehicle tire-pavement car tire when new-Site 1, 1980 (1 km =
friction improves as SN increases. Quantitatively, however, 0.6 mi) (21).
46 TRANSPORTATION RESEARCH RECORD 1311

Ifil Case C: Coefficient of Friction Estimated Using Equation

I. -<>- Paaenger Car 6 First, the parameters SN0 and PNG, describing the skid
-&-E501
-O-E524 resistance dependence on speed, must be determined. Be-
cause ribbed tire SN values at two different speeds are known,
SN0 and PNG can be calculated directly to be SN0 = 101.1
and PNG = 2.27. In order to calculate the average coefficient
of friction from Equation 6, an initial speed of the vehicle
must be assumed. For an assumed speed of 40 mph, the av-
erage coefficient of friction is then calculated as follows:
.70

101.l x 0.227 2 x 402

/ = -----
.6 200

40
.50 x [ 1 - e0 •0227 x ( 1 - 0.0227 x 40) r l
= 0.54 (12)
.40+---..----..-----r-----.-----.-
0 IB ~ ~ ~ ~
V (km/h)
Hence, the predicted initial speed is
FIGURE 2 Coefficient of friction versus
speed for the three tires on the braking
distance site, 1981 (1 km = 0.6 mi). V; 5.5(d x !) 112
5.5(34 x 0.54) 112 = 23.6 mph (13)
This value is different from the assumed value of 40 mph. In
order to improve accuracy, the process of calculating V; is
•Initial speed, V; = 30 mph; and repeated, until the assumed and predicted values of V; are
•Braking distance, d = 34.0 ft. the same. Repeating the calculation off and V; yields f =
0.67 and V; = 26.3. The prediction error is -3.7 mph or
The results of the skid tests conducted on the same pavement -12.3 percent. The results of predicting initial speed based
and under the same weather conditions were as follows: on the different estimates of the tire-pavement coefficient of
friction obtained in this example are presented in Table 1.
•Ribbed tire SN at 40 mph, SN 40 = 39.8; and This example demonstrates that the use of SN as an estimate
• Ribbed tire SN at 30 mph, SN 30 = 50.3. of tire-pavement friction may lead to large errors in predicting
vehicle speed. The magnitude of the error is likely to be
particularly high when the coefficient of friction is estimated
Case A: Coefficient of Friction Estimated by SN 40 /IOO by SN/100.
This situation is likely to occur in real life. The vehicle speed
is unknown; therefore, a standard ASTM E274 test at 40 mph
is used to estimate tire-pavement friction, f = SN 40 /100 = Example 2
0.398. Because the braking distanced is known, the estimate
of the initial speed is then calculated as follows: The results of several braking tests on both wet and dry sur-
faces of a section of the bituminous skid pad at the PTI test
V; 5.5(d x {) 112 track are presented in Table 2.
The results obtained with an ASTM skid trailer on the same
5.5(34 x 0.398) 112 = 20.2 mph (10) surfaces are presented in Table 3.
On the basis of a regression of the wet E501 (ribbed) tire
The actual initial speed is 30 mph; therefore, the speed pre- tests, the following values were calculated:
diction error is -9.8 mph, or -32.7 percent.
• SN0 = 80.17,
• SN 0 (peak) = 107.7,
Case B: Coefficient of Friction Estimated by SN31,/l00 In
this case, SN measured at the actual vehicle speed is used,
which is not likely to happen in real life. The predicted initial TABLE 1 RESULTS OF PREDICTING INITIAL SPEED
speed, assuming that the braking distance is known, is FOR EXAMPLE 1

Sn 40 /100 sn30 /100 f, eq. 6

V; = 5.5(d x f) 112
Predicted Speed, mi/h 20.2 22.7 26.3
= 5.5(34 x 0.503) 112 = 22.7 mph (11)
Absolute Error, mi/h -9.8 -7.3 -3.7
which gives the speed prediction error of - 7 .3 mph, or - 24.3 Relative Error, % -32.7 -24.3 -12. 3
percent.
Wambold and Kulakowski 47

TABLE 2 RESULTS OF BRAKING TESTS CONSIDERED IN

EXAMPLE 2

Vehicle Wet/Dry v,, rni/h d, feet DF

Pontiac Sunbird Dry 30 34.B . B6

Pontiac Sunbird Wet 30 44.5 . 67

Transit Bus Dry 26 42. 5 , 53

Transit Bus Dry 42 112. 7 . 52

Lincoln Dry 31 40.0 . BO

Cadillac (with ABS) Dry 46 99 .o . Bl

Dump Truck Dry 30 37.6 . 79

TABLE 3 RESULTS OF SKID RESISTANCE TESTS FOR EXAMPLE 2

Tire Condition Speed SN Peak

E501 (ribbed) Wet 29.B 51. 7 97.1

E501 (ribbed) Wet 40.7 40.1 95.6

E501 (ribbed) Wet 50.1 37. B 90.B

E524 (blank) Wet 39.9 3B.l 92. 7

E501 (ribbed) Dry 39.6 56 .4 107.0

• PNG = 1.563, and measure of wet pavement safety either. Figure 3 shows con-
• PNG (peak) = 0.325 . siderable scatter of data in the ratio of wet to dry accidents
versus the SN plot obtained by Havens in 1979 (24). That
Using these values, the calculations of Cases A, B, and C nothing has changed in respect to the relationship between
in Example 1 were repeated to obtain the results presented number of wet accidents and SN is demonstrated by Figure
in Tables 4-6. 4, obtained from the validation data set collected for the 1990
study (17).
Although the process of identification of sections that do
OTHER FACTORS AFFECTING WET-PAVEMENT not provide adequate traction under wet-pavement conditions
SAFETY is an important part of pavement management, it is but one
factor needed in providing motorists with highway surfaces
The examples presented in the previous section demonstrated that allow them to drive safely. Many other factors affect
that using SN or SN-based estimates of tire-pavement friction safety under wet-pavement conditions, and it is only when
can lead to very large errors in calculations of speed and these conditions demand a particular level of traction that SN
stopping distance in accident analysis. SN alone is not a good then becomes important.

TABLE 4 RES ULTS OF PREDICTING INITIAL SPEED FOR

EXAMPLE 2, CASE A (SN4ofl00 IS USED AS f)

v, Calculated V1 Actual Percent

Vehicle Condition f mi/h mi/h Error

Pontiac Sunbird Dry ' 564 24. 3 30 ·19

Pontiac Sunbird Wet . 37B 22. 6 30 ·25

Transit Bus Dry . 564 26. 9 26 + 3

Transit Bus Dry ' 564 43.8 42 + 4

Lincoln Dry . 564 26 .1 31 ·16

Cadillac (ABS) Dry . 564 41.1 46 ·11

Cadillac (ABS) Dry . 564 24. 5 30 -18

Dump Truck Wet . 378 20.7 30 ·31

48 TRANSPORTATION RESEARCH RECORD 1311

TABLE 5 RESULTS OF PREDICTING INITIAL SPEED FOR

EXAMPLE 2, CASE B [SN(v) NEAREST TEST SPEED IS USED]

V1 Calculated V1 Actual Percent

Vehicle Condition f mi/h mi/h Error

Pontiac Sunbird \let . 517 26.4 30 -12

Transit Bus Dry . 564 43.8 42 + 4

Cadillac (ABS) Dry . 564 41. l 46 -11

Dump Truck \let . 517 24. 3 30 -19

TABLE 6 RESULTS OF PREDICTING INITIAL SPEED FOR

EXAMPLE 2, CASE C [SN(v) IS CALCULATED]

V1 Calculated V1 Actual Percent

Vehicle Condition f mi/h mi/h Error

Pontiac Sunbird \let • 595 28 . 2 30 - 6

Dump Truck \let . 610 26 .2 30 -13

~ 0 ,7 All drivers in the traffic flow can maintain directional sta-

~ bility of their vehicles if the friction demand exerted by their
8<I 0.6 vehicles does not exceed the available friction . The available
~ friction depends on the pavement skid resistance. Friction
~ 0.5 demand is a function of vehicle speed, road geometry, traffic
> characteristics (including traffic density and percentage of trucks
'if 0.4
in the traffic flow), vehicle characteristics (including vehicle
>-
0::
0 type and its understeer), and driver skills. SN, measured on
~ 0 .3
a given section of road , represents the available friction at 40
~ mph. Whether or not this particular SN is sufficiently high to
31: 0 .2
lL
0
prevent wet accidents cannot be determined unless the de-
~ 0 .1· mand for friction on the section of road under consideration
~ is known. The effects of all important factors on friction de-
0::
0.0+--~-~-~-~-~--~-~-~-~ mand must be considered in pavement management.
I W W ~ M ~ ~ 50 ~ 00
SKID NUMBER

FIGURE 3 Ten-point moving averages of wet-to-dry Vehicle Speed

pavement accidents for 120 test sections in Kentucky
with ADT above 2,700 versus SN. Speed is a critical factor for the balance between friction
demand and supply because it affects both. When speed in-
32 creases, friction demand also increases. For instance, cen-
c
trifugal forces generated during vehicle cornering, which must
2B be counteracted by tire-pavement friction forces to prevent a
!!!z a
llJ
0 24 vehicle from skidding off the road, are proportional to the
u square of vehicle speed. At the same time, the pavement skid
~ a
20 a resistance decreases with increasing speed in an approxi-
l;j a
a
31: ca a mately exponential manner (Equation 5).
16 a
~ a a a
a
0:: 12 a g ca
a ca
a a
a
llJ a
(D
c ace a
:::E a c Road Geometry
:::> 8· ca CCC
a OCIIIDC D °Ee 0 o cn::c 0 a
z a
a
ca =:a:IDDaa ca ca c
4· ca r:D:D ca rn::DC CIJ cc The amount of friction required for safe driving is strongly
c:c a a:u:r:ttD C D ID CDJ CC c a
c 09 D
D
IDt:c " I l l 1111 I i i i ! H·1HR:i a
CDJ affected by road geometry. Friction demand on straight sec-
0
10 20 30 40 50 60 70 tions of road is low if the road is level, if vehicles travel at
SKID NUMBER (SN) constant speed, and if there are no intersections. The demand
FIGURE 4 Number of wet accidents versus SN obtained for friction increases significantly if a grade or a curve must
from the validation data set. be negotiated (12). Page and Butas (25) concluded that pave-
Wambold and Kulakowski 49

ment accident rates are significantly higher on curves than on et al. (23) found that truck driver efficiencies range from 62
any other type of geometric alignment. The effect of curvature to 100 percent and that most drivers have little or no practice
on wet-accident rates was found to be particularly strong on in emergency braking situations. The concern over driver
pavements having SN values less than 25. When SN is less braking skills will be alleviated when antilock brake systems
than 25, wet-pavement accident rates are significantly greater become commonly used.
for both uphill and downhill slopes steeper than 3 percent In order to fully evaluate the risk of a wet-pavement ac-
than for flatter terrain. cident, all factors influencing the demand for friction and
available friction must be considered. Several research studies
are currently under way in which mathematical models of
Traffic Flow
pavement safety index are being developed. These statistical
models estimate the probability of wet pavement accidents as
Traffic volume, in general, does not have a significant influ-
a function of the variables affecting the balance between fric-
ence on wet-accident rates (26). However, under special cir-
tion demand and available friction.
cumstances, namely, on undivided highways having SN values
less than 25, wet-accident rates were found to increase sig-
nificantly when the average daily traffic was greater than 15,000 CONCLUSION
vehicles per day (25). Traffic composition, in particular, the
percentage of trucks in the traffic flow, has a significant effect In general, the use of SN 40 as coefficient of tire-pavement
on friction demand. This is because the stopping distances of friction would be expected to underestimate the speed or
trucks are 1.3 to 2.8 times longer than the stopping distances overestimate the stopping distance. In the case of a truck or
of passenger cars (26). bus, this was not true; however, if the 75.5 percent rule is
Driving difficulty is another important characteristic of traffic used for truck tires, then the bus velocity would have also
flow. The criteria determining driving difficulty include num- been underestimated. Using SN(V) at the speed of the ac-
ber of access points per segment of road, presence of turn cident is an improvement, but it still underestimates the speed.
lanes, type of surrounding land use, traffic signalization, and Similarly, calculating a mean friction for SN(V) over the speed
the roadway cross section. of interest yields an even better estimate, but still underes-
timates the speed. This is to be expected because an ASTM
tire was found to exhibit a lower SN than a regular tire.
Vehicle Type In the case of pavement management, ranking roads on the
basis of SN only is not wise. Certafoly a long, straight road
Figure 5 shows stopping distances of buses and various con- with an SN value of 30 and ADT of 1 is of no concern when
figurations of trucks versus the stopping distance for a typical there are so many roads with high ADT values, curves, in-
passenger car from 60 mph (96 km/hr) on a dry road (27). tersections, etc., at the same or even higher SN values, which
The friction demand for buses and all types of trucks is higher are thus more likely to have wet-weather accidents. Roads
than for passenger cars if equal stopping distance is required with good SN values are not problems with regard to skidding,
for all vehicles. The friction demand is also higher for vehicles but roads with low SN values (e.g., below 20) need to be
with lower degrees of understeer (20). addressed. For cases in between, the major problem area, it
is recommended that an index that gives the probability of a
Driver Skills wet-weather accident should be used in ranking which road
should be repaired first on a safety basis.
Few drivers can operate their vehicles at 100 percent effi-
ciency, i.e., using 100 percent of the available friction. Olson REFERENCES
1. Safety Effectiveness Evaluation: Selected State Highway Skid Re-
~Cars sistance Programs. Report NTSB-SEE-80-6. National Transpor-

-----.L..1
tation Safety Board, Washington, D.C., Sept. 1980.
2. Accident Facts, National Safety Council, Washington, D.C., 1979.
~ Buses
RanQe for
3. R. R. Hegmon. Wet Weather Accidents and Pavement Skid Re-
11 11 sistance. Public Roads, Sept. 1981, p. 68.
Typical
Vehicles
4. J. C. Wambold, J. J. Henry, and R. R. Hegmon. Skid Resistance
of Wet Weather Accident Sites. The Tire Pavement Interface,
STP 929. ASTM, Philadelphia, Pa., June 1985, pp. 47-60.
5. J. R. Hosking. Relationship between Skidding Resistance and Ac-
cident Frequency: Estimates Based on Seasonal Variation. Re-
search Report 76. Transport and Road Research Laboratory,
Crowthorne, Berkshire, U.K., 1986.
6. J. M. Zuieback et al. Frictional Requirements Necessary to Re-
duce Skidding Accident Frequencies. Report SAI-260-77-547-LA.
FHWA, U.S. Department of Transportation, 1977.
7. R.R. Blackbum, D. W. Harwood, A. D. St. John, and M. C.
Sharp. Evaluation of Accident Rate-Skid Number Relationships,
0 400 500 Effectiveness of Alternative Skid Reduction Measures, Vol. 1.
Stopping Distance, ft Report FHWA-RD-79-22. FHWA, U.S. Department of Trans-
portation, Nov. 1978.
FIGURE 5 Stable stopping distance from 60 mph 8. B. T. Kulakowski et al. Skid Resistance Manual. Pennsylvania
(96 km/hr) on dry road. Transportation Institute, University Park, Pa., July 1990.
50 TRANSPORTATION RESEARCH RECORD 1311

9. R. R. Hegmon. Tire-Pavement Interaction. Technical Paper 87024. 20. J. J. Henry. Comparison of the Friction Performance of a Pas-
SAE, Warrendale, Pa., 1987, pp. 197-204. senger Car Tire and the ASTM Standard Test Tires . Special Tech-
10. A. G. Veith. Tire-Roads-Rainfall-Vehicles: The Traction Con- nical Publication 793, ASTM, Philadelphia, Pa . , 1983, pp. 219-
nection . STP 793 . ASTM, Philadelphia, Pa., 1983, pp. 3-40. 231.
11. H . W. Kummer and W. E. Meyer. NCHRP Report 37: Tentative 21. P. S. Fancher, R . D. Ervin , C. B. Winkler, and T . D . Gillespie .
Skid Resistance Requirements for Main Highways. HRB, Na- A Factbook of the Mechanical Properties of the Components of
tional Research Council, Washington, D.C., 1967. Single-Unit and Articulated Heavy Vehicles. Report DOT HS 807-
12. E. Farber, M. S. Janoff, S. Cristinzio, J. G . Blubaugh, 125. NHTSA, U.S. Department of Transportation, 1986.
W. Reisener, and W. Dunning. NCHRP Report 154: Determining 22. I. Gusakov, R. Rice, S. Pugliese, and R. Galganski. An Eval-
Pavement Skid Resistance Requirements at Intersections and Brak- uation of Methods to Investigate Truck Tire Wet Traction. Report
ing Sites. TRB , National Research Council , Washington, D .C. , DOT HS 806-577. NHTSA , U.S. Department of Transportation,
1974. 1984.
13 . A. B. Moore and J. B. Humphreys. High Speed Skid Resistance 23 . P. L. Olson, D . E . Cleveland, P. S. Fancher, L. P. Kostynick ,
and the Effects of Surface Texture on the Accident Rate. In Skid and L. W. Schneiter. NCHRP Report 270: Parameters Affecting
Resistance of Highway Pavements, STP 530. ASTM, Philadel- Stopping Sight Distance. TRB, National Research Council ,
phia, Pa., 1973, pp. 91-100. Washington , D .C., June 1984.
14. OECD Road Safety Research : A Synthesis. Organization for Eco- 24 . W. Bergman . Skid Resistance Properties of Tires and Their In-
nomic Cooperation and Development Scientific Experts Group, fluence on Vehicle Control. In Transportation Research Record
Paris, France , 1986. 621 , TRB, National Research Council, Washington, D.C., 1977,
15 . Anti-Skid Tests in U.K. International Road Federation, Paris, pp. 8-18 .
France, April 1988. 25. B. G. Page and L. F. Butas. Evaluation of Friction Requirements
16. D . L. Ivey and L. I. Griffin III. Development of a Wet Weather for California State Highways in Terms of Highway Geometrics.
Safety Index . Report FHWATX77-2211F. Texas State Depart- Report FHWA/CA/TL-86/01. FHWA, U .S. Department of
ment of Highways and Public Transportation, Austin, Nov. 1977. Transportation , Jan. 1986.
17. B. T . Kulakowski et al. Development of a Methodology to Identify 26. D. C. Mahone and S. N. Runkle . Pavement Friction Needs. In
and Correct Slippery Pavements. Research Project 88-06. Penn- Highway Research Record 396 , HRB, National Research Coun-
sylvania Department of Transportation, University Park, Pa., cil, Washington, D.C., 1972, pp . 1-11.
Aug. 1990. 27. R. W. Row!inski and S. F. Williams. NHTSA Heavy Duty Ve-
18. J. J. Henry, K. Saito, and R. Blackburn. Predictor Model for hicle Brake Research Program-Report No. 1: Stopping Capa-
Seasonal Variations in Skid Resistance: Comprehensive Final Re- bility of Air Braked Vehicles. DOT HS806 738 and 739. NHTSA,
port, Report FHWA-RD-83/005 . FHWA , U .S. Department of U .S. Department of Transportation, April 1985.
Transportation , 1983.
19. J . C. Wambold. Obtaining Skid Number at Any Speed from Test
at Single Speed . In Transportation Research Record 1196, TRB, Publication of this paper sponsored by Committee on Surface Prop-
National Research Council, Washington, lJ.C., 1988. erties- Vehicle Interaction.