Torbic et al.

DEVELOPMENT OF MORE BICYCLE-FRIENDLY RUMBLE STRIP CONFIGURATIONS

Darren Torbic Research Assistant Pennsylvania Transportation Institute 201 Transportation Research Building, University Park, PA 16802 Telephone: (814) 863-5492 Fax: (814) 865-3039 e-mail: djt106@psu.edu

Lily Elefteriadou, Ph. D. Assistant Professor of Civil Engineering Pennsylvania State University 212 Sackett Building, University Park, PA 16802 Telephone: (814) 863-7923 Fax: (814) 865-3039 e-mail: axe11@psu.edu

Moustafa El-Gindy, Ph.D Senior Research Associate Pennsylvania Transportation Institute 201 Transportation Research Building, University Park, PA 16802 Telephone: (814) 863-7930 Fax: (814) 865-3039 e-mail: mxe15@psu.edu

Paper Number: 01-2358 Presented at the Annual Meeting of the Transportation Research Board January 2001

Torbic et al. ABSTRACT

This research was initiated to develop new rumble strip configurations for the Pennsylvania Department of Transportation (PennDOT) that could alert inattentive/drowsy motorists and be safely and comfortably traversed by bicyclists. Three primary steps were involved in the

development of the new rumble strip configurations. The first step was to evaluate different configurations using simulation for their potential to be bicycle-friendly. The second step was to install several of the configurations that had the greatest potential to be bicycle-friendly and conduct field experiments to further evaluate their effectiveness. The final step was to analyze the field data and rank the configurations that were installed, based on their ability to provide a comfortable and controllable ride for bicyclists and to alert inattentive/drowsy motorists. Based on results of bicycle and motor vehicle tests, two new rumble patterns were recommended for implementation along non-freeway facilities. One configuration was recommended along nonfreeway facilities with higher operating speeds near 88 km/h (55 mph), and a second configuration was recommended along non-freeway facilities with lower operating speeds near 72 km/h (45 mph).

INTRODUCTION To address the problem of run-off-the-road (ROR) crashes, many state agencies are currently using milled shoulder rumble strips (MSRS) to alert motorists who are drifting out of the travel lane. MSRS consist of a continuous pattern ground into the asphalt or concrete shoulder. As motor vehicle tires pass over the MSRS, the motorist receives auditory and tactile warnings of potential roadside hazards. Until recently rumble strips were being installed primarily on freeway facilities where bicycle use is restricted, except in a few states within the United States. However, use of MSRS

Torbic et al.

is being extended to non-freeway facilities on a limited basis. As a result there are legitimate bicyclists concerns that need to be addressed. Bicyclists who have ridden over MSRS complain that the sensation is extremely uncomfortable and that MSRS may cause loss of control of the bicycle. To address these problems, the Pennsylvania Department of Transportation (PennDOT) funded researchers at the Pennsylvania Transportation Institute (PTI) to develop several rumble strip configurations that will both alert inattentive/drowsy drivers and be safely and comfortably traversed by bicyclists (1). Previous studies (2-5) focused on designing rumble strips for cars and, in some instances, trucks and motorcycles. Few studies have considered bicycles in the design of rumble strips. Grder (6) was the first to document tests conducted to verify bicyclists concerns about maneuverability problems associated with rumble strips. Grder, however, did not find evidence to support bicyclists fears that shoulder rumble strips cause them to lose control of their bicycles, but he did report that riding over rumble strips was annoying to bicyclists. Young (7), on the other hand, found it dangerous to ride a bicycle over rumble strips at speeds greater than 8 km/h (5 mph). Ardekani et al. (8) tried to link the feeling and perception of a smooth ride to the width of the smooth surface between grooves and not necessarily to the depth or the width of the depression (within given limits).

RESEARCH METHODOLOGY To develop new rumble strip configurations that both alert inattentive/drowsy motorists and can be safely and comfortably traversed by bicyclists, two different perspectives were considered throughout the research, that of the bicyclist and that of the motorist. From the bicyclists perspective, rumble strips should not shake the bicycle or generate too much vibration as the bicyclist traverses the rumble strips. From the motorists perspective, rumble strips should

Torbic et al.

transmit a sufficient amount of auditory and vibrational stimuli to warn an inattentive/drowsy motorist who drifts from the travel lane. Three primary steps were involved in the development of the new rumble strip configurations. The first step was to develop, evaluate, and rank different configurations for their potential to be bicycle-friendly. A simulation model was developed and validated to evaluate various new designs. The second step was to install several of the configurations that had the greatest potential to be bicycle-friendly and conduct field experiments to further evaluate their effectiveness. The final step was to analyze the data and rank the configurations that were installed, based on their ability to provide a comfortable and controllable ride for bicyclists and to alert inattentive/drowsy motorists.

DESIGN OF NEW RUMBLE STRIP CONFIGURATIONS Simulation Model Development The Dynamic Analysis and Design System (DADS) software was used to model the interactive rigid body system of a bicycle, its rider, and the road surface. By introducing spring forces and damping forces between the bicycle tires and the road surface and between the riders hips and the bicycle seat, DADS generated a set of constraint equations to specify the kinematics relationships and a set of differential equations to specify the dynamics of the rider-bicycle-road surface system. To accurately model the interactions between a bicycle, its rider, and the road surface, separate components were developed before the interactions were considered. This model consisted of two components: the bicycle and the rider.

Torbic et al. The bicycle model consisted of three components: bicycle frame front wheel rear wheel

The wheels were connected to the bicycle frame with revolutionary joints, rotating freely with respect to the bicycle frame. The bicycle wheel was modeled as a circle with a linear radius spring distributed around its peripheral. The contact force between the bicycle tire and the rumble strip was proportional to the tire radial deformation and pointed toward the center of the wheel. The rider was modeled as a rigid body system consisting of four parts (9): head and torso upper arms and lower arms lower legs and feet thighs

The torso and thighs and the thighs and lower legs were connected with revolutionary joints. The upper arms and the lower arms were treated as one unit in the simulation. This was a reasonable assumption because often riders keep their arms straight, with relatively no rotational motion between the upper and lower arms. Figure 1 illustrates the bicycle/rider system on a simulated rumble strip configuration.

Simulation Model Validation To validate the bicycle/rider simulation model, field data were compared to simulated data for PennDOTs current rumble strip configuration. The performance measured used in the

comparison was the fast fourier transformation (FFT) of the vibration data. FFT can be used to

Torbic et al.

transfer vibration signals from the time domain into the frequency domain. FFT plots of the vertical acceleration and pitch angular acceleration of the bicycle/rider system were created from the simulation results. Likewise, FFT plots of the vertical acceleration and pitch angular

acceleration were created from the data gathered during field tests. When comparing FFT plots, the important criterion is the frequency of the peak vibration signal, rather than the magnitude of the signal. Comparisons between the FFT plots from the simulation and field tests showed that the peak vibration signals occurred approximately at the same frequencies for both the simulation and field tests. Figure 2 shows the comparison of vertical acceleration measured during the field test and simulation results, for tests conducted at 10 km/h (6 mph). Overall, there was a good comparison between the data from the field tests and the simulation model. Therefore, the bicycle/rider simulation model was used as a tool to evaluate the impact of potential rumble strip configurations on bicycle/rider dynamics.

Testing Potential Configurations with Simulation The major benefit of the bicycle/rider simulation model was the ability to evaluate the dynamics of a bicycle and rider traversing numerous rumble strip configurations. With milled rumble strips, the primary variables that impact the dynamics of bicycles and motor vehicles include: the width of the groove the spacing of the groove the depth of the groove the transverse length of the groove

The simulation focused on varying the width of the groove and the spacing of the grooves. The depth of the groove was not varied. The depth of the rumble strips simulated was held constant at 13 mm (1/2 in.). The transverse length was held constant within the simulation as well.

Torbic et al.

Twenty-six configurations were simulated, varying the width and spacing of the grooves. The groove width was varied from 102 to 203 mm (4 to 8 in.). The spacing between the grooves was varied from 51 to 229 mm (2 to 9 in.). Typically, the spacing of grooves is measured from the center of one groove to the center of the adjacent groove. Rather than referring to the spacing of the grooves as on centers, the spacing of the grooves refers to the length of the flat portion of pavement between grooves. To evaluate the dynamics of the bicycle and rider traversing the various rumble strip configurations, the vertical acceleration and pitch angular acceleration of the bicycle/rider system were measured. Rather than taking an average acceleration or the peak acceleration, the rootmean-square (RMS) of the accelerations was calculated to measure the magnitude of the vibrations. The formula to calculate RMS is as follows:
2 & &i z n

RMS =
Where
& & z

i =1

= acceleration (vertical or pitch angular)

i = ith value acceleration time history

n = number of values Using the simulation model, the RMS vertical acceleration and RMS pitch angular acceleration of the bicycle/rider system were calculated for different speeds for twenty-six rumble strip configurations. Plots of RMS vertical acceleration versus speed were created as well as plots of RMS pitch angular acceleration versus speed. To determine which rumble strip configurations generated the least level of vibration, the areas under the RMS curves were calculated for speeds between 15 and 25 km/h (9 and 15 mph). The areas under the RMS curves for vertical acceleration and pitch angular acceleration were

Torbic et al.

changed into dimensionless percentages with respect to the maximum value of the area. For each configuration, the percentages from the RMS vertical acceleration and pitch angular acceleration were added together and divided by two, providing an average weighted percentage. By sorting the configurations by the average weighted percentage, it was determined which configurations were most bicycle-friendly. The configurations with the lowest average weighted percentages generated the least amount of vibration. Table 1 shows the sorted list of the twentysix rumble strip configurations that were simulated.

Installation of Selected Rumble Strip Configurations The simulation model served to prioritize potential configurations to be installed at PTIs test track so additional testing could be performed with the selected bicycles and the motor vehicle. A contractor was contacted to install several of the high priority configurations at PTIs test track. However, there were limitations that prohibited the milling machine from cutting the exact dimensions of some of the simulated configurations most likely to decrease the vibration levels experienced by bicyclists. Thus, adjustments were made in the field with respect to the actual dimensions of the rumble strip configurations installed at PTIs test track. Table 2 shows the actual dimensions of the six rumble strip configurations that were installed at PTIs test track for further evaluation. All of the patterns had a transverse width of 406 mm (16 in.). The patterns were installed in approximately 38 m (125 ft) sections.

FIELD TESTING THE NEW RUMBLE STRIP CONFIGURATIONS An experimental plan was developed to gather the appropriate and necessary data during the test track experiments. The test track experiments involved collecting vibration and perception data from volunteer participants bicycling over the six rumble strip configurations and collecting data

Torbic et al.

from a motor vehicle to measure the amount of noise and vibration generated from the six configurations.

Bicycle Testing The following four bicycle types were selected for testing because they represent a large proportion of the bicycles that are in use today:

Standard adult road bicycle: 6 586 x 103 - 827 x 103 Pa (85-120 psi) tire pressure Non-suspended mountain bicycle: 6 276 x 103 - 345 x 103 Pa (40-50 psi) tire pressure Hybrid bicycle: 6 448 x 103 - 552 x 103 Pa (65-80 psi) tire pressure Tandem bicycle: 6 689 x 103 - 758 x 103 Pa (100-110 psi) tire pressure
A non-suspended mountain bicycle theoretically exhibits worse dynamic characteristics as compared to a mountain bicycle with active suspension. A non-suspended mountain bicycle was selected for testing so that the new design considers the worse case scenario.

Bicycle Speed To determine whether speed affects the comfort level of bicyclists and to determine whether speed affects controllability, the bicycle experiments were conducted at three speed levels: Low Speed: 5 to 15 km/h (3 to 9 mph) Intermediate Speed: 16 to 25 km/h (10 to 15 mph) High Speed: greater than or equal to 26 km/h (16 mph)

Torbic et al.

Speed ranges were used instead of specific speeds because it is difficult for bicyclists to ride at a particular constant speed, say 15 km/h (9 mph). It is possible to do so, but it requires the bicyclist to focus heavily on the speedometer.

Bicycle Angle of Approach Out on the roadways it is conceivable that bicyclists traverse rumble strips from all possible angles. Bicyclists may approach rumble strips from straight on, where the rumble strips just begin. Bicyclists may also cross rumble strips at very shallow angles while negotiating from the travel lane to the shoulder or vice versa, or bicyclists may cut across rumble strips at sharp angles to avoid objects or perform turning maneuvers. Three approach angles were selected for testing: 0 degrees (straight on approach) 10 degrees 45 degrees

Type of Bicyclists Participation in the test track experiments was only sought from intermediate and advanced bicyclists. Two issues were considered when selecting the types of bicyclists to be involved in the testing. One issue was safety, and secondly intermediate to advanced riders are most likely to encounter rumble strips out on the roads. The following definitions were used to classify bicyclists as either intermediate or advanced:

Intermediate Bicyclist: Someone who knows how to handle his/her bicycle easily and comfortably. An intermediate bicyclist can ride comfortably and confidently in light or

Torbic et al.

10

moderate traffic environments. An intermediate bicyclist might ride 1,600 to 6,400 km (1,000 to 4,000 mi) per year, usually in good weather.

Advanced Bicyclist: Someone who has many years and thousands of kilometers of bicycling experience, who can do everything the intermediate bicyclist can do, plus ride skillfully, comfortably, and confidently in heavy traffic. An advanced bicyclist typically rides 8,000 km (5,000 mi) per year or more, although the number of kilometers ridden each year alone is not as good an indicator of skill level as is the environments in which the bicyclist rides.

Motor Vehicle Testing The motor vehicle portion of the test track experiments involved gathering both vibration and noise data while traversing the new rumble strip configurations. Due to the instrumentation needs of gathering the vibration data, only one motor vehicle was used in the testing. The motor vehicle was a 1998 Plymouth Grand Voyager. Vibration and noise data were collected at speeds of 72 and 88 km/h (45 and 55 mph). The minivan was driven at a small departure angle, approximately 3 to 5 degrees. Studies have shown that ROR crashes typically occur at these angles.

ANALYSIS OF THE QUANTITATIVE AND QUALITATIVE MEASURES This section presents results from the analyses to evaluate and assess the relative effects of the six rumble strip patterns installed at PTIs test track on bicycle and motor vehicle dynamics.

Torbic et al. Analysis of Bicycle Measures

11

Twenty-five subjects participated in the bicycle testing at PTIs test track. Thirteen of the participants had intermediate riding experience, while twelve participants had advanced riding skills. A majority of participants riding experience was on-road riding, as compared to off-road riding. Twenty participants were males, and five were females. Very few of the participants indicated any type of recurring pain in particular body parts or any general health related problems, and about two-thirds of the participants had ridden a bicycle over rumble strips prior to participating in the research. Several participants tested multiple bicycle types. In total, twenty-seven bicycle tests were conducted. Eight tests were conducted with the mountain bicycle; nine tests were

conducted with the road bicycle; five tests were conducted with the hybrid, and five tests were conducted with the tandem. Instrumentation requirements prohibited a participant from using his/her own bicycle. Thus, participants rode bicycles provided to them. Ten measures were gathered to assess the effect of the six rumble strip configurations on bicyclists comfort and control of the bicycle: Objective Measure Vertical acceleration Pitch angular acceleration Percentage of time spent off the line Subjective Measure Comfort of wrists, fingers, and elbows Comfort of shoulders and neck Comfort of back Comfort of seat area Comfort of knees, ankles, and feet Overall comfort level Overall control level

Torbic et al.

12

For most of the participants involved in the bicycle testing, data for all measures were gathered. For the other participants, data were collected for all the measures except for the vibration data (vertical and pitch angular acceleration).

Objective Measures of Comfort and Control Vertical Acceleration Analysis of the vertical acceleration data was divided into two parts. One analysis was performed on the data from the mountain, road, and hybrid bicycles, while a separate analysis was performed on the data from the tandem bicycle. Separate analyses were performed because the accelerometer had to be mounted at a different location on the tandem, other than the center of gravity of the bicycle/rider system. Because the reference location of the accelerometer was different for the tandem, the vertical acceleration for the tandem was not directly comparable with the vertical acceleration data from the mountain, road, and hybrid bicycles. Vertical acceleration data were collected for the first five participants on each bicycle type. The RMS vertical acceleration was calculated over all speeds and rumble strip types, but only at a 0 degree traverse angle. There was too much noise in the raw data to accurately calculate the RMS for the 10 degrees and 45 degrees tests. Table 3A presents the rankings of the test configurations based upon the vertical acceleration levels measured on the mountain, road, hybrid, and tandem bicycles across all speed ranges. From the bicyclists perspective, lower vertical accelerations make for a smoother, more comfortable ride and better control of the bicycle. Thus, the rumble strip configurations were ranked according to the levels of vertical acceleration generated. The rumble strip configuration that generated the lowest amount of vertical acceleration received the best ranking, and the

Torbic et al.

13

rumble strip configuration that generated the highest amount of vertical acceleration received the worst ranking. From the bicyclists perspective, test pattern 6 generated the lowest level of vertical acceleration, and pattern 4 generated the greatest amount of vertical acceleration. For the mountain, road, and hybrid bicycles, test pattern 5 generated less vertical acceleration than test pattern 1. However, for the tandem, test pattern 1 generated less vertical acceleration than test pattern 5.

Pitch Angular Acceleration Similar to the analysis for vertical acceleration, the analysis of the pitch angular acceleration data was divided into two parts. Pitch angular acceleration data were collected for the first five participants on each bicycle type. The RMS pitch angular acceleration was calculated over all speeds and rumble strip types at a 0 degree traverse angle. Table 3B presents the rankings of the test configurations based upon the pitch angular acceleration levels measured on the mountain, road, hybrid, and tandem bicycles across all speed ranges. Similar to vertical acceleration, lower pitch angular accelerations make for a smoother, more comfortable ride and better control of the bicycle. Thus, the rumble strip configurations were ranked according to the levels of pitch angular acceleration generated. The rumble strip configuration that generated the lowest amount of pitch angular acceleration received the best ranking, and the rumble strip configuration that generated the highest amount of pitch angular acceleration received the worst ranking. Although the values of the pitch angular accelerations are different for the analysis of the mountain, road, and hybrid bicycles and the analysis of the tandem bicycle, the test patterns received the same rankings. Test pattern 6 generated the least amount of pitch angular

Torbic et al. acceleration across all bicycle types, and test pattern 1 generated the greatest amount of pitch angular acceleration across all bicycle types.

14

Objective Measure of Control To obtain an observable objective measure of lateral stability, the bicyclists were asked to ride along a 203 mm (8 in.) wide white line as they traversed the rumble strips. As the participants traversed the rumble strips at a 0 degree angle, a video camera was positioned from a perspective that showed both the bicycle tires and the white line. After data collection the videotapes were played back, and any lateral deviation from the line was noted by recording the amount of time spent off the 203 mm (8 in.) white line. The percent of time spent off the line was calculated by dividing the amount of time deviating from the white line by the traverse time. Lower

percentages of time spent off the line indicate better control while traversing the rumble strip configuration. A baseline for controllability was established by having the participants ride along a 203 mm (8 in.) wide white line that was painted on a smooth portion of the pavement. Videos from 25 of the bicycle tests were used to rank the test configurations based on the objective control measure. Table 4 presents the rankings of the respective test patterns, with the average percent of time off the line combining data across all bicycle types and all speed ranges. Test patterns 3 and 6 caused the least amount of lateral deviations, while test pattern 1 caused the greatest amount of lateral deviation from the lines.

Subjective Measure of Comfort and Control Subjective ratings of comfort and control were assessed following trials where the participants traversed the rumble strips at a 0 degree angle over an extended distance of approximately 13.7 m (45 ft). Participants were asked to rank the level of comfort or discomfort of five key body parts:

Torbic et al. wrists, fingers, and elbows shoulders and neck back seat area knees, ankles, and feet

15

A subjective measure of comfort was obtained for each body section and overall for each rumble strip pattern at each speed level. Subjects were also asked to rate the trial on the relative difficulty or ease in controlling the bicycle when traversing the rumble strip. Based on the subjective results, the ordered list below ranks the body parts most affected while traversing rumble strips: wrists, fingers, and elbows (most uncomfortable) seat area shoulders and neck back knees, ankles, and feet (most comfortable)

Table 5 presents the rankings of the test configurations based on the subjective wrist comfort level, overall comfort level, and overall control level. The average values combine the ratings across all bicycle types and all speeds. From the perceptions of the participants, test patterns 6, 3, and 5 consistently ranked the best from the standpoint of comfort and control. PennDOTs current rumble strip configuration, test pattern 1, was consistently perceived as the worst test pattern from the standpoint of comfort and control.

Torbic et al. Overall Ranking of Test Patterns Based on Bicycle-Related Measures

16

To combine the objective comfort and control measures, the objective control measure, and the subjective comfort and control measures, a normalized value for each of the measures was established. In so doing, each measure listed below received the same weight in calculating the overall ranking of the test configurations based on the bicycle-related measures: Objective Comfort and Control Measures Average RMS vertical acceleration (mountain, road, and hybrid bicycles) Average RMS vertical acceleration (tandem bicycle) Average RMS pitch angular acceleration (mountain, road, and hybrid bicycles) Average RMS pitch angular acceleration (tandem bicycle)

Objective Control Measure Average percentage of time off the line (all bicycles combined)

Subjective Comfort and Control Measures Average comfort level of wrists, fingers, and elbows (all bicycles combined) Average overall comfort level (all bicycles combined) Average overall control level (all bicycles combined)

To obtain an assessment of the relative effects of the different rumble strip configurations based on the subjective ratings of the participants, three measures were utilized: overall comfort level, overall control level, and comfort level of the wrists, fingers, and elbows. The comfort level of the wrists, fingers, and elbows was selected because these are the body parts most affected while traversing rumble strips.

Torbic et al. For each measure, the normalized value was calculated as follows:

17

Normalized Value =

Actual Value Minimum Value Maximum Value Minimum Value

or

(1)

Actual Value Minimum Value Normalized Value = 1 Maximum Value Minimum Value

(2)

For those measures in which a low average value was better from the bicyclists perspective, such as vertical acceleration, pitch angular acceleration, and percent of time spent off the line, equation 1 was used to calculate the normalized value. For those subjective measures for which a higher subjective rating indicated better comfort and controllability, equation 2 was used to calculate the normalized value. In this manner, the test pattern that received the best ranking for the respective measures obtained a normalized value of 0, and the test pattern that received the worst ranking for the respective measures obtained a normalized value of 1. The normalized value for the other test patterns indicated the relative difference between the maximum and minimum values. By summing the normalized values for the 8 bicycle-related measures and dividing by 8, the average normalized value for each test pattern was calculated. A lower average normalized value for the test pattern corresponds to better comfort and control for the bicyclists. Table 6 presents the overall ranking of the test configurations based on the bicyclerelated measures. From the bicyclists perspective, test pattern 6 provides the highest level of comfort and controllability, followed by test pattern 3. Test pattern 2 was next, closely followed by test pattern 5. Test pattern 4 and test pattern 1 received the worst ratings of comfort and controllability from the bicyclists perspective.

Torbic et al. Analysis of Motor Vehicle Measures

18

Three performance measures were collected during the motor vehicle testing at PTIs test track: vertical acceleration of the body frame pitch angular acceleration of the body frame maximum sound level in the passenger compartment

Preliminary analyses showed very little difference in the vibration levels of the body frame generated by the six rumble strip patterns. Therefore, only the maximum sound level in the passenger compartment of the minivan was used to assess the effectiveness of the six rumble strip patterns on alerting inattentive/drowsy drivers. Table 7 presents the rankings of the test configurations based upon the noise level testing for the motor vehicle. The maximum sound level in the passenger compartment was measured at speeds of 72 and 88 km/h (45 and 55 mph). Three runs were made at each speed. In addition to gathering sound levels while traversing the rumble strips, sound levels were measured while riding over the smooth pavement surface. Based on the noise level tests, test pattern 4 generated the greatest amount of noise in the passenger compartment at 72 km/h (45 mph) so it received the best ranking at this speed. PennDOTs current configuration, test pattern 1, received the best ranking at 88 km/h (55 mph). Test pattern 6 generated the lowest amount of noise at both speeds.

CONCLUSIONS From the bicyclists perspective, test pattern 6 was the most bicycle-friendly. However, from a motor vehicle standpoint, test pattern 6 generated the least amount of noise in the passenger compartment in both the high speed and low speed testing. Because test pattern 6 was ranked the

Torbic et al.

19

worst test pattern from the motor vehicle standpoint, test pattern 6 is not recommended for implementation on any type of facility. From the bicyclists perspective, test pattern 3 was ranked the second most bicyclefriendly configuration. Considering test pattern 3 from the motor vehicle standpoint, it

performed well during the high speed testing, ranking second overall with an average maximum sound level of 81.3 dB(A) versus 88.9 dB(A) for the existing configuration. Therefore, test pattern 3 is recommended for implementation along non-freeway facilities with higher operating speeds. During the low speed testing, however, test pattern 3 generated 74.7 dB(A) of noise, less than 7 dB(A) above the ambient noise. As a result, test pattern 3 ranked only fifth overall at this speed. Watts (10) indicated that rumble strips producing 4 dB(A) increases or above can be readily detected by motorists who are awake, but there are no data indicating the sound level difference above the ambient noise necessary to alert a drowsy motorist. 7 dB(A) was not considered adequate enough, so test pattern 3 is not recommended for implementation on nonfreeway facilities with lower operating speeds. The third ranked test pattern from the bicyclists perspective, test pattern 2, was considered for low speed facilities. A close examination of the average normalized values for test pattern 2 and test pattern 5 shows very little difference between the two. Essentially, test patterns 2 and 5 perform equivalently from the bicyclists perspective, but from the motor vehicle standpoint at lower operating speeds, test pattern 5 generates slightly more sound in the passenger compartment as compared to test pattern 2. Thus, test pattern 5 appears to be a better compromise between the bicyclists and the motorists along non-freeway facilities with lower operating speeds than test pattern 2. Therefore, test pattern 5 is recommended for

implementation along non-freeway facilities with lower operating speeds.

Torbic et al.

20

Based on the results of the bicycle testing and the motor vehicle testing, two new bicyclefriendly rumble patterns are recommended for implementation along non-freeway facilities. Test pattern 3 is recommended along non-freeway facilities with higher operating speeds, near 88 km/h (55 mph). Test pattern 5 is recommended along non-freeway facilities with lower

operating speeds, near 72 km/h (45 mph).

FUTURE RESEARCH To solidify and improve the research presented in this paper, additional research is recommended in the following areas. A field evaluation of the newly recommended bicycle-friendly rumble strip configurations is required to ensure that the fully controlled environment (PTIs test track) results are consistent with the field implementation. These rumble strip configurations should be installed in several locations and field evaluations of their effectiveness should be performed. Crash data should be analyzed, and bicyclists should be polled for their evaluation and opinions of these new rumble strip designs. Advancements in milling machine technology are necessary so that contractors are able to install other rumble strip patterns that showed the potential of being bicycle-friendly. Finally, a question that needs to be answered is what level of sound needs to be generated by the rumble strip to alert a drowsy or sleeping motorist.

REFERENCES 1. Elefteriadou, L., M. El-Gindy, D. Torbic, P. Garvey, A. Homan, Z. Jiang, B. Pecheux, and R. Tallon. Bicycle-Friendly Shoulder Rumble Strips. PTI 2K15. The Pennsylvania State University, The Pennsylvania Transportation Institute, 2000. 2. Franke, K. A. Evaluation of Rumble Strips. Virginia Highway & Transportation

Research Council, 1974.

Torbic et al.

21

3. Tye, E. J. Devices to Prevent Run-Off-Road Accidents. Report No. CA-DOT-TR-12691-76-01. California Department of Transportation, Office of Traffic, 1976. 4. Chen, C. S. A Study of Effectiveness of Various Rumble Strips on Highway Safety. Virginia Department of Transportation, Traffic Engineering Division, 1994. 5. Wood, N. E. Shoulder Rumble Strips: A Method to Alert Drifting Drivers.

Pennsylvania Turnpike Commission, 1994. 6. Grder, P. Rumble Strips or Not Along Wide Shoulders Designated for Bicycle Traffic? In Transportation Research Record 1502, TRB, National Research Council, Washington, D.C., 1995, pp. 1-7. 7. Young, T. Bicycle Use on Highways with Rumble Strips. Test Report U.S. 191/ Hoback Canyon, Teton County, Wyoming, 1997. 8. Ardekani, S. A., S. Govind, S. P. Mattingly, A. Demers, H. S. Mahmassani, and D. Taylor. Detection and Mitigation of Roadway Hazards for Bicyclists. FHWA/TXDOT-96/1394-2F, Texas Department of Transportation, 1996. 9. Wang, E. L. and M. L. Hull. A Model for Determining Rider Induced Energy Losses in Bicycle Suspension Systems. Vehicle Systems Dynamics, Vol. 25, 1996, pp. 223-246. 10. Watts, G. R. The Development of Rumble Areas as a Driver-alerting Device. Report

Supplementary Report 291, Transportation and Road Research Laboratory, 1977.

Torbic et al. LIST OF FIGURES FIGURE 1. Bicycle / rider simulation model. FIGURE 2. FFT of bicycle/rider vertical acceleration at speed of 10 km/h.

22

LIST OF TABLES TABLE 1 Simulated Configurations Sorted by Weighted Average Percentage of RMS TABLE 2 Rumble Strip Configurations Installed at PTIs Test Track TABLE 3A Ranking of Test Configurations Based on Vertical Acceleration TABLE 3B Ranking of Test Configurations Based on Pitch Angular Acceleration TABLE 4 Ranking of Test Configurations Based on Objective Control (All Bicycles) TABLE 5 Ranking of Test Configurations Based on Subjective Ratings (All Bicycles) TABLE 6 Overall Ranking of Test Configurations Based on Bicycle-Related Measures TABLE 7 Ranking of Test Configurations Based on Noise Level Testing

Torbic et al.

23

FIGURE 1. Bicycle / rider simulation model.

Torbic et al.
5000

24

FFT of vertical acceleration (m/s^2/Hz)

4500 4000 3500 3000 2500 2000 1500 1000 500 0 0 10 20 30 40

S im ula tio n Simulation E xp e rim e nt

Field Test

50

Frequency (Hz)

FIGURE 2. FFT of bicycle/rider vertical acceleration at speed of 10 km/h.

Torbic et al. TABLE 1 Simulated Configurations Sorted by Weighted Average Percentage of RMS
Rumble Strip Pattern Width W (mm) 102 102 102 102 102 127 127 127 127 127 152 178 152 152 127 178 152 178 178 178 178 178 203 178 152 Flat L (mm) 127 178 152 203 229 76 127 178 152 229 152 76 229 203 203 51 178 229 127 102 203 178 76 152 127 Area under Acceleration Plots Pitch Angular Vertical Accel. Accel. (m2/s3) (rad*m/s3) 7.27 11.83 16.81 9.68 18.69 28.58 34.58 20.14 28.23 43.42 28.34 63.13 42.04 46.48 40.10 19.94 59.26 31.05 36.05 55.15 57.37 85.27 60.18 79.63 59.10 5.79 4.29 3.52 6.24 5.59 7.38 6.84 13.96 11.83 7.97 21.33 10.68 17.73 17.58 11.62 29.18 16.31 26.22 28.41 22.55 23.26 14.08 24.50 20.78 31.37 24.52
L

25

Percentage of Vert. Accel. (actual/max) 0.08 0.12 0.18 0.10 0.20 0.30 0.36 0.21 0.30 0.46 0.30 0.67 0.44 0.49 0.46 0.21 0.62 0.33 0.38 0.58 0.60 0.90 0.63 0.84 0.62 1.00

Percentage of Pitch Angular Accel. (actual/max) 0.18 0.14 0.11 0.20 0.18 0.24 0.22 0.44 0.38 0.25 0.68 0.34 0.57 0.56 0.68 0.93 0.52 0.84 0.91 0.72 0.74 0.45 0.78 0.66 1.00 0.78

Wtd Avg Percentage 0.13 0.13 0.14 0.15 0.19 0.27 0.29 0.33 0.34 0.36 0.49 0.50 0.50 0.53 0.57 0.57 0.57 0.58 0.64 0.65 0.67 0.67 0.71 0.75 0.81 0.89

203 127 94.83 1 in. = 25.4 mm; 1 ft = 0.305 m

Torbic et al. TABLE 2 Rumble Strip Configurations Installed at PTIs Test Track Test Groove Width Flat Portion between Cuts Pattern (mm) (mm) 1* 2 3 4 5 6
1 in. = 25.4 mm

26

Depth (mm) 13 13 10 13 10 6.3

178 127 127 127 127 127

127 178 178 152 152 178

* PennDOTs current standard

Torbic et al. TABLE 3A Ranking of Test Configurations Based on Vertical Acceleration Average RMS - Vertical Acceleration Test Pattern (m/s2) Mountain, Road, and Hybrid Bicycles Best 6 11.315 3 13.176 2 16.951 5 19.323 1 21.909 Worst 4 21.942 Tandem Bicycle Best 6 8.939 3 10.364 2 12.451 1 16.804 5 17.965 Worst 4 19.082 1 ft = 0.305 m

27

TABLE 3B Ranking of Test Configurations Based on Pitch Angular Acceleration Average RMS Pitch Angular Acceleration Test Pattern (rad/s2) Mountain, Road, and Hybrid Bicycles Best 6 13.288 3 15.809 5 18.994 2 21.230 4 21.711 Worst 1 30.593 Tandem Bicycle Best 6 10.912 3 13.171 5 15.407 2 16.404 4 17.829 Worst 1 22.006

Torbic et al. TABLE 4 Ranking of Test Configurations Based on Objective Control (All Bicycles) Percentage of Time Off the Line Test Pattern Difference Average (Pattern Base) Smooth 8.14 % Best 3 12.28 % 4.14 % 6 12.6 % 4.46 % 4 16.44 % 8.30 % 5 19.22 % 11.08 % 2 19.56 % 11.42 % Worst 1 25.35 % 17.21 %

28

Torbic et al. TABLE 5 Ranking of Test Configurations Based on Subjective Ratings (All Bicycles) Test Pattern Average Rating* Wrist, Fingers, and Elbows Best 6 14.3 3 13.7 5 11.9 4 10.5 2 10.1 Worst 1 7.1 Overall Comfort Level Best 6 14.8 3 14.5 5 12.1 2 11.0 4 10.0 Worst 1 7.3 Overall Control Level Best 6 14.3 3 13.4 5 11.5 2 10.8 4 9.5 Worst 1 7.4
* Comfort scale: very uncomfortable (0) and very comfortable (25) * Control scale: uncontrollable (0) and no effect on handling (25)

29

Torbic et al.

30

TABLE 6 Overall Ranking of Test Configurations Based on Bicycle-Related Measures Test Pattern Sum of Normalized Values Average Normalized Value Best 6 3 2 5 4 1 0.025 0.919 3.972 4.015 5.303 7.772 0.003 0.115 0.497 0.502 0.663 0.972

Worst

Torbic et al. TABLE 7 Ranking of Test Configurations Based on Noise Level Testing Test Speed Avg. Max. Sound Level Difference Pattern (km/h) dB(A) (Pattern Smooth) Best 4 72 83.6 15.2 1 72 80.0 11.6 5 72 79.3 10.9 2 72 78.4 10.0 3 72 75.2 6.8 Worst 6 72 74.7 6.3 Smooth 72 68.4 1 2 3 4 5 Worst 6 Smooth 1 mph = 1.61 km/h Best 88 88 88 88 88 88 88 88.9 83.7 81.3 81.2 79.1 78.2 65.2 23.7 18.5 16.1 16.0 13.9 13.0

31