Printer Friendly

Analysis of advisory speed setting criteria.

ANALYSIS OF Advisory Speed Setting Criteria


Curve warning signs are often supplemented with an advisory speed plate when the safe curve speed is less than the speed limit. The advisory speed is intended to inform an unfamiliar driver of a possible hazardous situation and recommend a safe speed to drive around the curve. Recent studies and surveys suggest that advisory speeds on curves are generally set too low and are not set consistently from State-to-State or even within a given State. (1,2) As a result the average motorist has little or no respect for the advisory speeds. Drivers using the highway repeatedly become accustomed to the speed that road condition and curvature will allow; consequently, the advisory speed signs do not have much affect on them. However, an unfamiliar motorist who finds it safe to drive 16.1 km/h (10 mi/h) over the advisory speed will be placed in a potentially hazardous situation when encountering the occasional curve posted with a realistic and meaningful advisory speed.

Despite dramatic improvements in tires and vehicle handling characteristics, current criteria for setting advisory speeds have remained essentially unchanged for over 50 years. The study described in this article evaluated the validity of current curve speed criteria for modern vehicles.


Historically, advisory speeds for curves been determined in the field by making several trial runs through the curve at different speeds in a vehicle equipped with a ball-bank indicator. The ball-bank reading is a combined measure of centrifugal force, vehicle roll, and superelevation; as such, it indicates overturning forces on the vehicle. The generally accepted criteria for setting advisory speeds are ball-bank readings of 14 [degrees] for speeds below 32 km/h (20 mi/h), 12 [degrees] for speeds between 32 km/h and 56 km/h (20 and 35 mi/h), and 10 [degrees] for speeds of 56 km/h (35 mi/h) or greater. (3) These criteria are based on tests conducted in the 1930's and are intended to represent the 85th to 90th percentile curve speed. (4)

Another method used to determine the advisory speed is the standard curve formula: [V.sup.2] = 15R (e + f), where V = speed in miles per hour, R = radius of curve in feet, e = rate of superelevation in feet per foot and f = coefficient of side friction. A friction factor of 0.16 is assumed in the nomograph in the Traffic Control Devices Handbook (TCDH). A side friction factor of 0.16 corresponds to the speed at which discomfort begins for an average rider in a 1930 vintage car; this factor may not be valid for modern vehicles. (5) The side friction factors recommended in the design criteria of the American Association of State Highway and Transportation Officials (AASHTO) vary from 0.17 at low speed to 0.10 at the highest speed. (6) However, these values are also based on tests conducted in the 1930's and represent the limit at which a rider will notice a "side pitch" and begin to feel some discomfort. The ball-bank readings of 14 [degrees], 12 [degrees] and 10 [degrees] were found in these earlier studies to correspond to side friction values of 0.21, 0.18, and 0.15, respectively.

The friction factors used in current criteria do not reflect the maximum safe speed but rather an average comfortable speed. Modern cars on dry pavement are capable of generating friction coefficients of 0.65 and higher before skidding. (5) Friction coefficients of 0.40 and higher are typical on wet pavements. Thus, a curve designed for 112.63 km/h (70 mi/h) could be driven well over 160.9 km/h (100 mi/h) before it skidded out. (6)

Data Collection

Curve geometry, spot speeds, and ball-bank readings were measured at 28 curves on two-lane highways (16 curves from Virginia, 7 curves from Maryland and 5 curves from West Virginia). Each site consisted of either a single curve or, in case of a series of curves, the first curve. A few of the sites contained intersecting roads or driveways nearby, but volumes were low and did not have an appreciable impact on the observed speed of traffic. Field studies were conducted in dry weather between 9 a.m. and 4 p.m. on weekdays.

A pair of radar speed readings were taken on 50 free-flowing vehicles at each site. The data collector was positioned to measure the vehicle speeds running at the farthest visible point on the tangent approaching the curve and again at the middle of the curve.

Several ball-bank readings were taken at the sites by running tests with a vehicle (Chevrolet Celebrity station wagon, model 1984). The possibility of an inaccurate speedometer during test runs was eliminated by using a moving radar, which displayed the test vehicle speed while negotiating the curve. Before the test run at each site, the ball-bank indicator was mounted on the windshield of the test vehicle and adjusted on a level surface to bring the ball in the indicator to zero (see figure 1). Extra care was taken to drive the car parallel to the center line of the curve so as to avoid any bias in the ball-bank reading from cutting the curve short. The first trial run was made at the posted advisory speed. At least three additional trials were run in 8-km/h (5-mi/h) increments and the corresponding ball-bank readings were recorded. Each curve was also driven three times at a speed judged reasonable to the driver of the test vehicle; and the corresponding ball-bank readings were recorded.

At each site, the degree of curve was determined by measuring the offset in inches from edgeline to the midpoint of a 18.9-m (62-ft) chord. Superelevation was measured along the curve with a carpenter's level. Each of these measurements was taken three times and the average recorded. Table 1 shows the geometric and operating speed of the test sites.

Table : Table 1. - Geometric and operating speed of the test sites
 Site Advisory Degree Super- Operating Speed
 Speed of elevation
 (mi/h) Curve Tangent Curve Speed
 50th 85th 50th 85th
 C1VA 20 21 0.03 43 47 37 41
 C2VA 25 50 0.10 38 42 29 32
 C3VA 15 29 0.10 37 42 32 36
 C4VA 25 25 0.13 45 50 40 43
 C5VA 15 16 0.04 36 41 30 35
 C6VA 25 29 0.03 43 49 32 35
 C7VA 45 9 0.09 58 64 55 58
 C8VA 50 8 0.07 52 56 50 55
 C9VA 50 7 0.07 54 59 51 55
C10VA 40 7 0.04 52 56 51 55
C11VA 40 10 0.06 51 56 46 51
C12VA 35 7 0.09 53 59 50 56
C13VA 40 8 0.10 50 54 48 53
C14VA 35 5 0.09 55 59 55 59
C15VA 25 13 0.03 41 44 37 42
C16VA 20 19 0.10 46 53 36 42
C17MD 40 8 0.05 53 60 49 54
C18MD 25 59 0.10 47 55 26 30
C19MD 25 38 0.07 53 57 32 35
C20MD 25 12 0.04 47 53 42 46
C21MD 25 28 0.09 45 51 32 38
C22MD 25 27 0.09 41 45 37 41
C23MD 30 16 0.03 37 42 33 37
C24WV 35 21 0.10 44 47 38 41
C25WV 35 18 0.13 46 50 41 45
C26WV 45 10 0.08 49 56 44 49
C27WV 35 19 0.10 49 56 41 45
C28WV 50 9 0.05 51 58 50 55

Motorist Compliance

On average, 9 out of 10 motorists exceeded posted advisory speeds. Moreover, zero compliance with posted advisory speeds was observed at almost half of sites. Table 2 shows the percentage of motorist compliance with posted speeds at each site.

Table : Table 2. - Percent compliance with advisory speeds
Advisory Speed, (mi/h) Percent Compliance Range (percent)
 15-20 0
 25-30 8 0-38
 35-40 5 0-32
 45-50 43 0-68

1 mi/h = 1.609 km/h

Motorists generally exceeded the advisory speeds posted below 48 km/h (30 mi/h) by wider margins than they did for advisory speeds posted at 56 km/h (35 mi/h) or higher. The percentage of drivers exceeding the advisory speeds by various ranges are shown in figure 2.

Although compliance with the advisory speeds was poor, drivers did adjust their speeds on curves. Table 3 compares the speed reduction suggested by the posted advisory speed and the actual speed drop. The suggested speed reduction is the difference between the observed speed on the approach and the posted advisory speed. The actual speed drop is the difference between the approach speed and curve speed. Only the average speed reductions observed in West Virginia were near the suggested value. Overall, drivers were advised to slow down an average of 24.1 km/h (15 mi/h); however, the actual speed drop averaged only 9.6 km/h (6 mi/h). The advisory speed at the test sites was thus exceeded on average by 14.5 km/h (9 mi/h).

Table 3. - Average speed reduction on curves

Speed Drop in mi/h

 Suggested Actual
 Virginia 15.8 4.6
 Maryland 18.7 10.4
West Virginia 7.9 4.9
 All Curves 15.1 6.1

1 mi/h = 1.609 km/h

Posted Speeds, Recommended Speeds

and Observed Speeds

To determine whether highway agencies are employing the ball-bank indicator or the TCDH nomograph in setting the safe curve speeds, the actual recommended speed at each site was compared to the speeds derived via these methods.

Table 4 shows the recommended speed of each curve based on the ball-bank indicator criteria and nomograph in the TCDH, which is based on the standard curve speed formula with a friction factor of 0.16. Only West Virginia actually posted advisory speeds consistent with generally recommended criteria. In Maryland, about one-half of the curves were posted according to existing criteria, and most were within 8 km/h (5 mi/h). Less than one-third of the curves in Virginia corresponded to the recommended speed based on the standard ball-bank criteria or side friction value of 0.16; many were set 16 km/h (10 mi/h) or more below the recommended value.

Table 4. - Posted advisory speed and recommended speed
 Site Posted Recommended Speed
 Ball-Bank Nomograph 85th Percentile
 C1VA 20 30 30 41
 C2VA 25 25 25 32
 C3VA 15 25 30 36
 C4VA 25 30 30 43
 C5VA 15 30 35 35
 C6VA 25 30 25 35
 C7VA 45 45 50 58
 C8VA 50 55 50 55
 C9VA 50 50 50 55
 C10VA 40 45 50 55
 C11VA 40 45 45 51
 C12VA 35 55 55 56
 C13VA 40 55 50 53
 C14VA 35 60 60 59
 C15VA 25 30 35 42
 C16VA 20 30 35 42
 C17MD 40 40 45 54
 C18MD 25 20 25 30
 C19MD 25 25 25 35
 C20MD 25 35 35 46
 C21MD 25 30 30 38
 C22MD 25 30 30 41
 C23MD 30 30 35 37
C24WVA 35 35 35 41
C25WVA 35 35 35 45
C26WVA 45 45 45 49
C27WVA 35 30 35 45
C28WVA 50 50 50 55

Table 5 compares differences in the posted and recommended speeds relative to 50th- and 85-percentile curve speed. The recommended speed based on the nomograph better reflects driver behavior than that based on the ball-bank indicator. Even so, the recommended speed would still be below the average speed of traffic. Table 6 shows the sites arranged from the sharpest to the flattest curve, superelevation, friction demanded, and ball-bank readings corresponding to the 85th percentile speed.

Table : Table 5. - Difference in posted advisory speed and recommended speed relative to observed speed
 50th Percentile 85th Percentile Speed
Posted Advisory Speed -8.8 -13.0
Ball-Bank -3.6 -7.8
Nomograph -2.4 -6.6

Table : Table 6. - Friction demanded and ball-bank readings corresponding to 85th percentile speeds
Site Degree of Curve Super- 85th Friction
 elevation Percantile From 85th Ball-Bank
 Speed Percantile Readings from
 Speed 85th
 Percentile Speed
C18MD 59 0.10 30 0.52 26
 C2VA 50 0.10 32 0.5 28
C19MD 38 0.07 35 0.47 21
 C3VA 29 0.10 36 0.34 20
 C6VA 29 0.03 35 0.38 15
C21MD 28 0.09 38 0.38 20
C22MD 27 0.09 41 0.44 23
 C4VA 25 0.13 43 0.41 23
C24WV 21 0.10 41 0.31 15
 C1VA 21 0.03 41 0.38 20
C27WV 19 0.10 45 0.35 18
C16VA 19 0.10 42 0.29 22
C25WV 18 0.13 45 0.29 18
 C5VA 16 0.04 35 0.19 15
C23MD 16 0.03 37 0.22 18
C15VA 13 0.03 42 0.24 20
C20MD 12 0.04 46 0.26 19
C26WV 10 0.08 49 0.20 14
C11VA 10 0.06 51 0.24 12
 C7VA 9 0.09 58 0.26 16
C28WV 9 0.05 55 0.27 14
C13VA 8 0.10 53 0.16 9
 C8VA 8 0.07 55 0.21 9
C17MD 8 0.05 54 0.22 15
C12VA 7 0.09 56 0.17 9
 C9VA 7 0.07 55 0.18 10
C10VA 7 0.04 55 0.21 14
C14VA 5 0.09 59 0.11 6

1 mi/h = 1.609 km/h

Ball-Bank Reading Corresponding to

Observed Speeds

The fact that the recommended speeds based on existing ball-bank criteria are below the prevailing speed of traffic suggests that the current criteria are not valid for modern vehicles. The ball- bank readings corresponding to the 50th- and 85th-percentile curve speeds are plotted in figure 3. The line represents the best fit linear regression. Although higher ball-bank readings were observed at lower speed curves, the ball-bank values do not conform well with existing criteria. The 85th-percentile ball-bank readings are some 5 to 10 [degrees] higher than existing criteria.

On average, the 50th- and 85th-percentile speeds corresponded to ball-bank readings of 14 and 17 [degrees]. However, there is a great deal of scatter in the data; this casts doubt on using the ball-bank indicator to establish safe curve speed even with revised readings.

Side Friction

As mentioned previously, the TDCH nomograph for determining safe curve speed is based on the standard design speed formula using a side friction value of 0.16. Figure 4 shows the friction factor corresponding to the 50th and 85th percentile speeds of each curve computed from the standard curve speed formula in the TCDH and AASHTO design guide. The line in the figure represents best fit linear regression.

The average friction used by 50th-percentile drivers was 0.22; it was 0.29 for 85th-percentile drivers. These values are nearly twice the current value used in establishing advisory curve speeds. Figure 4 shows drivers accept higher side friction on lower speed curves than is currently assumed for road design. However, the friction values used by drivers are at least 50 percent greater than those assumed in road design. For comparison, modern cars on dry pavements can generate side friction values ranging from 0.65 to 0.90 before skidding out.(5) Thus, the friction used on curves is limited more by driver comfort than by the limits of the vehicle-pavement interaction.


The absence of adequate and universally accepted criteria for determining advisory speeds creates the problem of nonuniform and subjective applications; this problem in turn poses a potential safety threat to unfamiliar drivers. The posted advisory speeds have little significance for the motorists. At most curves, posted advisory speeds were well below the prevailing traffic speed.

The study also found noticeable variation in the application of the existing ball-bank criteria from curve to curve and State to State. In most cases, the TCDH ball-bank criteria result in very low and unrealistic advisory speeds. The current criteria of 10, 12 and 14 [degrees] should be revised upwards. Ball-bank readings of 12 [degrees] above 64 km/h (40 mi/h), 16 [degrees] between 48 km/h (30 mi/h) and 64 km/h (40 mi/h), and 20 [degrees] below 48 km/h (30 mi/h) would better reflect observed or average curve speeds.

The nomograph and design speed formula are marginally better than the ball-bank indicator in reflecting driver speed behavior on curves (see table 6), but the current values used for side friction are too conservative. Although not very reliable (as it varied widely for a given speed), friction values of 0.30 at lower speed and 0.20 at higher speed would provide a more realistic determination of safe curve speed.

An alternative approach to determining safe curve speed would be to sample vehicular speeds. A sample of 10 vehicles could be used to estimate the average curve speed to within [plus or minus] 5 km/m (3 mi/h). This approach is currently being investigated as are several other alternatives for recommending safe speeds on curves. These alternatives include prediction models of curve speed based on degree and length of curve and use of the G-analyst, an accelerometer that provides a direct measure of lateral acceleration.


(1) Motorists Compliance with Traffic Control Devices. Publication No. FHWA-RD-89-103, U.S. Department of Transportation, Federal Highway Administration, Washington, DC, 1989.

(2) Institute of Transportation Engineers Technical Committee 41-M. "Review and Effectiveness of Advisory Speeds," ITE Journal, September 1978.

(3) Traffic Control Devices Handbook, U.S. Department of Transportation, Federal Highway Administration, Washington, DC, 1983.

(4) R.A. Moyer and D.S. Berry. "Marking Highway Curves with Safe Speed Indications," Proceedings of Highway Research Board, Vol. 20, 1940.

(5) David R. Merrit. "Safe Speeds on Curves: A Historical Perspective of the Ball-Bank Indicator," ITE Journal, September 1988.

(6) A Policy on Geometric Design of Highways and Streets, American State Highway and Transportation Officials, Washington, DC, 1990.

(7) "Road Test Digest." Car and Driver, Vol. 36, May 1991.

(8) D.W. Harwood, J.M. Mason, W.D. Glauz, B.T. Kulakowski, and K. Fitzpatrick. Truck Characteristics for Use in Highway Design and Operation, U.S. Department of Transportation, Washington, DC, 1990.

Mashrur A. Chowdhury is a doctoral student in civil engineering at the University of Virginia. Mr. Chowdhury conducted this study as a graduate research fellow at the Turner-Fairbank Highway Research Center of the Federal Highway Administration (FHWA).

Davey L. Warren is a traffic research engineer in the FHWA's Intelligent Vehicle-Highway Research Division. Before October 1990, he was project manager of highway safety research regarding speed and speed limits. Mr. Warren is a 1975 graduate of the FHWA Highway Engineer Training Program; since then, he has been involved in research on traffic signal operations and safety, work zone traffic control, passing lanes, and variable message signs.

Howard Bissell, P.E., is a traffic research engineer with the FHWA's Office of Safety and Traffic Operations Research and Development. He manages the office's Traffic Control for Safety research program which supports the Manual on Uniform Traffic Control Devices. He has worked in traffic engineering positions for State, city, and county governments; the Highway Research Board; and in private consulting before joining the FHWA. Speed control is part of his research program area. Figure 1. - Ball-bank indicator mounted on the windshield. Figure 2. - Percent exceeding posted advisory speeds by various amounts. Figure 3. - Ball-bank readings corresponding to 50th- and 85th- percentile speeds. Figure 4. - Side friction corresponding to 50th- and 85th- percentile speeds.
COPYRIGHT 1991 Superintendent of Documents
No portion of this article can be reproduced without the express written permission from the copyright holder.
Copyright 1991 Gale, Cengage Learning. All rights reserved.

Article Details
Printer friendly Cite/link Email Feedback
Author:Chowdhury, Mashrur A.; Warren, Davey L.; Bissell, Howard
Publication:Public Roads
Date:Dec 1, 1991
Previous Article:Portland cement concrete pavement: rehabilitation challenges and recommendations.
Next Article:Fatigue testing for thermex reinforcing bars.

Related Articles
Relationship between fish consumption and serum chlordane levels.
Drivers face speed action.
Microbiological Risk Assessment in Food Processing. (Bookshelf).
AICPA, CICA offer tools to aid in corporate security and privacy.
Data is first step in strategy plans: understanding the forces driving the life/health market can make all the difference in sales.

Terms of use | Copyright © 2017 Farlex, Inc. | Feedback | For webmasters