Skip to contentUnited States Department of Transportation - Federal Highway Administration FHWA Home
Research Home
Public Roads
Featuring developments in Federal highway policies, programs, and research and technology.
This magazine is an archived publication and may contain dated technical, contact, and link information.
Federal Highway Administration > Publications > Public Roads > Vol. 72 · No. 4 > The Sound of Safety

Jan/Feb 2009
Vol. 72 · No. 4

Publication Number: FHWA-HRT-09-002

The Sound of Safety

by Kohinoor Kar and Richard S. Weeks

Arizona's findings point to centerline rumble strips for reducing lane departure crashes.

ADOT installed this centerline rumble strip on State Route 64 east of the Grand Canyon as one of the segments evaluated in a study on the effectiveness of this technique in reducing head-on and crossover sideswipe crashes. Photo: Kohinoor Kar, ADOT.
ADOT installed this centerline rumble strip on State Route 64 east of the Grand Canyon as one of the segments evaluated in a study on the effectiveness of this technique in reducing head-on and crossover sideswipe crashes.

Many rural two-lane roadways lack medians, barriers, or other physical measures to separate the travel lanes. As a result, a distinct potential exists for vehicles to cross centerlines, possibly leading to sideswiping or striking oncoming vehicles head-on. According to an Insurance Institute for Highway Safety (IIHS) study, Crash Reduction Following Installation of Centerline Rumble Strips on Rural Two-Lane Roads, these types of centerline crossover crashes account for about 20 percent of fatal crashes on two-lane rural roads.

A promising remedy is emerging, however. A growing number of States are using centerline rumble strips (CLRSs) to warn drivers against crossing over into opposing traffic. For example, the Washington State Department of Transportation (WSDOT) has installed about 1,545 kilometers (960 miles) of CLRSs in various locations. Of the 960 highway miles, WSDOT engineers have conducted preliminary evaluations of 834 kilometers (518 miles) that have been in place 6 months or longer. Preliminary results indicate a 28 percent reduction in all fatal and serious injury collisions, 26 percent reduction in all cross-centerline collisions, and 50 percent reduction in fatal and serious injuries resulting from cross-centerline collisions. According to WSDOT's The Gray Notebook, "It is important to note that these reductions may not be entirely attributable to rumble strips, as other safety improvements may have been implemented under the same contract that installed the rumble strips."

Other IIHS research suggests that CLRSs can reduce the rate of injury collisions on two-lane roads by 15 percent, with similar benefits in terms of fatalities.

CLRSs offer corollary benefits as well. State departments of transportation (DOTs) can install the strips fairly inexpensively at about $2,000 per mile, and they can coordinate the installation with the paving cycle. CLRSs also require minimal maintenance. According to WSDOT, this low cost provides CLRSs with a benefit-cost ratio of about 60:1 based on nationwide average costs.

Today, at least 23 States and 2 Provinces in Canada use CLRSs: Arizona, California, Colorado, Delaware, Florida, Illinois, Indiana, Kansas, Maryland, Massachusetts, Michigan, Minnesota, Missouri, New Hampshire, New York, North Carolina, Oregon, Pennsylvania, South Dakota, Texas, Utah, Washington, and Wyoming; and Alberta and Nova Scotia.

"By installing 5,700 miles [9,173 kilometers] of centerline rumble strips and 1,200 miles [1,931 kilometers] of shoulder rumble strips between 2008 and 2010, the Michigan Department of Transportation [MDOT] is systematically — and cost effectively — upgrading safety on our high-speed, nonfreeway rural road system," says Jill Morena, pavement marking and rumble strip program engineer at MDOT. "This is a systematic treatment for a systemic problem."

Other State and Provincial Efforts Using Centerline Rumble Strips

Rumble strips to prevent lane departure crashes on rural, undivided highways are increasingly popular among States and Canadian Provinces. The following is a sampling of how some States and Provinces are using CLRSs, as found in a literature review performed by ADOT researchers.

California analyzed milled CLRSs at no-passing zones. An evaluation of a 37-kilometer (23-mile) segment found a reduction in crashes. During the before period, crashes occurred at a rate of 4.5 per month; during the after period, 1.9 crashes occurred per month. The State uses a pattern of 16.5- centimeter (6.5-inch) by 40.6-centimeter (16-inch) by 1.3-centimeter (0.5-inch) grooves, spaced at 61 centimeters (24 inches).

Colorado uses milled CLRSs at locations with high crash histories in both passing and no-passing zones. Through a research project, the State found a 34 percent reduction in head-on crashes and 36 percent reduction in sideswipe opposite direction crashes. Colorado's experience also has shown that CLRSs have no harmful or damaging effect on pavement life. The State uses a pattern of 13-centimeter (5-inch)-long by 30.5-centimeter (12-inch)-wide by 0.95-centimeter (0.38-inch)-deep grooves, spaced 30.5 centimeters (12 inches) on center, based on another research project.

Delaware evaluated milled CLRS installations using 41-centimeter (16-inch)-wide rumble strips spaced at 30.5 centimeters (12 inches). The State found a 95 percent reduction in head-on crashes based on a before-and-after study, and concluded that effectiveness does not decrease over time. The State did find that noise issues for nearby residents would need to be addressed.

Kansas evaluated the effectiveness of two milled CLRS patterns: one using 16.5-centimeter (6.5-inch) by 30.5-centimeter (12-inch) by 1.3-centimeter (0.5-inch) grooves spaced at 30.5 centimeters (12 inches), and the other using the same groove pattern spaced at 30.5 centimeters (12 inches) and 61 centimeters (24 inches), alternately. Based on the subjective opinions of 247 motorists, the continuous pattern was slightly more effective. Milled centerline rumble strips were found to be a cost-effective measure to reduce head-on crashes.

Massachusetts uses CLRSs at no-passing zones only. The pattern is 16.5-centimeter (6.5-inch) by 45.7-centimeter (18-inch) by 1.3-centimeter (0.5-inch) grooves spaced at 30.5 centimeters (12 inches).

Minnesota evaluated milled CLRSs on 402 kilometers (250 miles) of roadways to ascertain the effects on pavement life, accumulated ice, and motorcycles. The installations were in areas with minimum 80-kilometer (50-mile)-per-hour posted speed limits and head-on crash histories. The State identified noise as an issue based on proximity to residences. Focus groups reported rumble strips were helpful in snow and reduced visibility.

New York plans to place milled CLRSs at two-way no-passing zones (double yellow) only where justified by crash history.

Oregon uses CLRSs at no-passing zones. The State uses a pattern of 17.8-centimeter (7-inch) by 40.6-centimeter (16-inch) by 1.6-centimeter (0.63-inch) grooves spaced at 30.5 centimeters (12 inches). Oregon reports CLRSs as a generally positive experience.

Pennsylvania uses CLRSs at no-passing zones on roads based on road type and average annual daily traffic. The pattern is 17.8-centimeter (7-inch) by 35.6-centimeter (14-inch), 40.6-centimeter (16-inch), or 45.7-centimeter (18-inch) by 1.3-centimeter (0.5-inch) grooves spaced at 61 centimeters (24 inches) and 121.9 centimeters (48 inches). The State has found no adverse effect on motorcycles. Bicycles were not included in this study.

Washington has found no adverse effect on motorcycles. The pattern is 17.8-centimeter (7-inch) by 30.5-centimeter (12-inch) by 1.3-centimeter (0.5-inch) grooves spaced at 30.5 centimeters (12 inches). The reduction in opposite direction crossover crashes was not statistically significant.

Alberta uses a pattern of 16.5-centimeter (6.5-inch) by 30.5-centimeter (12-inch) by 1.3-centimeter (0.5-inch) grooves spaced at 30.5 centimeters (12 inches) for its CLRSs. The Province has found no adverse effect on motorcycles.

In Arizona, centerline crossover crashes result in a significant percentage of serious injuries and fatalities on its rural, high-speed, two-way, two-lane roads. Between 2000 and 2005, these crashes comprised 5 percent of all incidents on Arizona's 6,437 kilometers (4,000 miles) of rural highways but accounted for 26 percent of fatal crashes.

To enhance safety, in 2002 the Arizona Department of Transportation (ADOT) began pilot installations of CLRSs on several stretches of rural highways in northern Arizona. ADOT researchers recently evaluated crash data from before and after the installations to determine the effectiveness of CLRSs in mitigating centerline crossover crashes. The results of this ADOT study, "Safety Enhancement Evaluation of Ground-In Centerline Rumble Strips," suggest that CLRSs can be a cost-effective solution to reduce crashes on rural roads. Further, ADOT's experiences, observations, and lessons learned could help other agencies that are considering CLRS installations within their own jurisdictions.

ADOT installed this milled CLRS at the approach to a curve on a rural highway in northern Arizona.
ADOT installed this milled CLRS at the approach to a curve on a rural highway in northern Arizona.

The Arizona Situation

One element of the Arizona Strategic Highway Safety Plan focuses on reducing fatal and serious injury crashes resulting from roadway departures. ADOT already completed programs to install milled shoulder rumble strips on all divided State highways and other qualifying highways to address roadway departure crashes involving fatigued and impaired drivers. According to the Federal Highway Administration, milled rumble strips are popular in many States because they are easy to implement on new or existing asphalt and portland cement concrete pavements and shoulders; they have little or no effect on the integrity of the pavement structure; and they produce greater noise and vibration than rolled or formed rumble strips.

Click here for the table: Centerline Rumble Strip Study Test Sites. The table contains information about the study sites, which were on high-speed, low-volume, rural roads with a high percentage of available passing opportunities.

Like other States, Arizona then began considering applying that success to addressing centerline crossover crashes — both opposite direction sideswipes and head-on collisions. "We were experiencing a high number of head-on collisions between Flagstaff and Page on Route U.S. 89," says ADOT Flagstaff District Engineer John Harper. "We were trying to find ways to reduce these types of crashes."

In 1998, the district installed CLRSs on about 40 kilometers (25 miles) of U.S. 89, a two-lane, two-way, rural highway in northern Arizona. Although the department did not produce a formal report documenting the findings, anecdotal observations were positive and supported the need for a formal study.

A review of research and evaluations by other States formed the basis for selecting a groove pattern that would generate an effective tactile and audible signal warning drivers of imminent crossovers. In addition to studying the safety benefits of CLRSs, ADOT set out to evaluate other issues including maintenance, objectionable sound, two-wheel vehicle control, constructability, driver confusion, and effectiveness in inclement weather.

View Alternative Text
ADOT studied five sites (circled) on high-speed, two-lane, rural highways in northern Arizona.

Study Location

In 2001, ADOT selected 14 highway sections for the study. These segments were typical low-volume, high-speed, two-lane, rural State highways in northern Arizona: State Route (SR) 64, SR 389, and U.S. 160, in addition to U.S. 89. The study "sites were all located in the high desert at elevations of 1,524-1,829 meters (5,000-6,000 feet), where freezing temperatures occur in the winter along with substantial accumulations of snow and ice.

Unlike some studies conducted by other States, ADOT did not base its site selection on crash history. Instead, ADOT's goal was to evaluate the performance of CLRSs in mitigating centerline crossover crashes involving inattentive and impaired drivers on a statewide basis, rather than for limited application at site-specific locations with crash histories. This broad-based approach is consistent with ADOT's aggressive strategy to apply shoulder rumble strips statewide to address roadway departure crashes.

Click here for the table: Summary of Fatal/Serious Injury Crashes. The table contains information about the crashes at the 14 study sites. The centerline crossover crashes accounted for 36 percent of the total fatal and serious injury crashes occurring during the before period and 22 percent in the after period.

In 2002, ADOT installed CLRSs on 239 kilometers (149 miles) of the roadways identified in 2001. Specifically, the Flagstaff District installed CLRSs on 98 kilometers (61 miles) of highways that already had shoulder rumble strips. The department also installed both centerline and shoulder rumble strips on 142 kilometers (88 miles) of highways that previously had no rumble strips.

The roadways included principal arterials, minor arterials, and minor collectors. All the roadways had two 3.7-meter (12-foot) lanes and either 1.5-meter (5-foot)-wide or 2.4-meter (8-foot)-wide paved shoulders, and were located in flat or gently rolling terrain. All but one segment carry low volumes of traffic. The rumble strip pattern used in the study was based on the standard pattern in use at that time for milled edge-line rumble strips.

Graph. This bar graph depicts the number of fatal and serious injury crashes for each year for 2000-2005 on the road segments that received CLRS installations. On the left-hand side is the term 'Crashes,' with the numbers 0-9 in ascending order. The years are represented by white bars rising from the horizontal plane. The term 'Before,' meaning prior to CLRS installation, appears below the years 2000-2002. The term 'After,' meaning following CLRS installation, appears below the years 2003-2005. For 2000, there were seven crashes. For 2001, there were five crashes. For 2002, there were six crashes. For 2003, there were three crashes. For 2004, there was one crash. For 2005, there were three crashes.
ADOT researchers determined that the number of fatal and serious injury crashes decreased by 61 percent after CLRSs were installed (comparing 3 years before with 3 years after installation).

Crash Analysis

ADOT researchers evaluated data from the roadways for the 3 years before (2000-2002) and 3 years after (2003-2005) installation to determine the effectiveness of CLRSs in reducing centerline crossover crashes.

The study revealed that centerline crossover crashes accounted for 36 percent of the total fatal and serious injury crashes occurring in 2000-2002 and 22 percent in 2003-2005. CLRSs helped cut the number of serious injury and fatal head-on and opposite direction sideswipe crashes from 18 in the "before" period to 7 in the "after" period.

In the before period, 12 of the 18 crashes occurred on tangent segments, 3 occurred on 2-degree curves, 1 on a 4-degree curve, and 2 on 7-degree curves. This distribution suggests that fatigue- and inattention-related lane-departure crashes may be independent of alignment. Four of the 18 crashes were related to fatigue, 6 were related to inattention, and 5 were related to drinking; 3 were reported as unknown.

During the after period, 3 of the 7 serious injury and fatal head-on and opposite direction sideswipe crashes occurred on tangent sections, 1 on a 1-degree curve, 2 on 2-degree curves, and 1 on a 7-degree curve. Four of the 7 crashes were related to drinking; 3 were reported as unknown.

Graph. This bar graph depicts the 'before' and 'after' crash rates in terms of installation of CLRSs. On the left-hand side is the label 'Crashes Per MVMT' (million vehicle miles traveled). In ascending order are the numbers 0.000, 0.005, 0.010, 0.015, 0.020, 0.025, and 0.030. The bar graph consists of two bars. The column on the left, which is labeled 'Before' and covers the years 2000-2002, reaches 0.024. The column on the right, labeled 'After' for the years 2003-2005, reaches 0.012, indicating the CLRSs installed between the periods cut the rate of centerline crossover crashes by 50 percent.
The average rate for fatal and serious injury centerline crossover crashes decreased by 50 percent after installation of CLRSs.

To account for differences in test site lengths and traffic volumes, the ADOT researchers calculated the crash rates using the following equation: Crashes per million vehicle miles traveled (MVMT) = (number of crashes x 1,000,000) / (average annual daily traffic x 365 x segment length).

"Overall, we have had very few complaints, and the centerline rumble strips are well received as a positive safety improvement," says Harper.

Lessons Learned

Crash mitigation. The ADOT study found that serious injury and fatal centerline crossover crashes involving fatigued, inattentive, and impaired drivers may be independent of roadway alignment (tangent versus curve). CLRSs have the potential to reduce these crashes. The crash frequency reduction was comparable to reductions found in studies in other States.

Pavement life. ADOT found no degradation of pavement that could be attributable to the milled CLRSs. Studies in other States support this finding.

Winter maintenance. ADOT found no negative impact of CLRSs on snow and ice removal. Other studies support this finding. Based on observations of CLRSs that have been in place for a number of years, no apparent damage or wear from snowplows is evident. Damage to milled rumble strips would most likely be a function of the plow equipment (design of the plow shoes, cutting edge type) and might be of interest to snowbelt States.

Noise. ADOT's pattern for milled rumble strips generates about 70 decibels of sound at 152 meters (500 feet). This sound level is used to limit rumble strip installations near roadside residences. Normal conversation is 60 to 70 decibels.

Constructability. ADOT found that the CLRS milling process requires dust control, using water to prevent particulates from becoming airborne. Work crews must collect millings for disposal. Crews must apply a fog coat (a light application of an emulsion to seal the exposed aggregate) to the milled area and reapply centerline pavement markings. Also, installation is most efficient when crews coordinate it with other construction and maintenance activities.

Work zone traffic control. ADOT found that during CLRS construction uniformed officers with marked police vehicles command more respect than flaggers. Lane closure by mobile operation traffic control depends on location.

Public reaction. ADOT observed a generally positive public response to CLRSs, except for a few negative comments from bicyclists. The bicyclists contend that motorists may position themselves more to the right in the lane to avoid driving on the CLRSs and thereby crowd bicyclists riding on the roadway. Generally limiting CLRSs to roadways with minimum 1.2-meter (4-foot) paved shoulders will address this concern, because the bicyclists will have a paved shoulder on which to ride. "Over time, we have modified the width and depth of the rumble strip based on motorist feedback," notes Harper.

Graph. This bar graph depicts the crash rate for each year for 2000-2005. On the left-hand side is the phrase 'Crashes Per MVMT' (million vehicle miles traveled). In ascending order are the numbers 0.000, 0.005, 0.010, 0.015, 0.020, 0.025, 0.030, and 0.035. The years are represented by white bars rising from the horizontal plane. The term 'Before,' meaning prior to CLRS installation, appears below the years 2000-2002. The term 'After,' meaning following CLRS installation, appears below the years 2003-2005. For 2000, the crash rate was 0.0328. For 2001, the crash rate was 0.0223. For 2002, the rate was 0.0138. For 2003, the rate was 0.0070. For 2004, the rate was 0.0040. For 2005, the rate was 0.0236.
The crash rates also have decreased (comparing 3 years before with 3 years after CLRS installation). Although the number of crashes in 2005 was half the number of crashes in 2002, the 2002 crashes occurred on higher volume roads—sites 4 and 9. The higher volume roads experienced a greater change in crash rate than the lower volume roads: 88 percent reduction for the higher volume roads compared to 40 percent for the lower volume roads.

Other Studies

ADOT's findings mostly parallel those of other studies. National Cooperative Highway Research Program (NCHRP) Synthesis 339 reported on CLRS practices in the United States and Canada, finding that CLRSs have a greater potential for crash reduction on tangents than on curves due to the departure angle. CLRSs had no adverse effects on pavement life, drainage, or snow and ice removal. Also, the visibility of reapplied pavement markings was not affected.

As noted earlier, a September 2003 study by IIHS found CLRSs reduced all injury crashes by 15 percent and opposite direction injury crashes by 25 percent, based on installations at 98 sites on 338 kilometers (210 miles) in California, Colorado, Delaware, Maryland, Minnesota, Oregon, and Washington. Most sites in the study were limited to no-passing zones. There was no significant difference in crash occurrence between day and night.

A Georgia Institute of Technology study in a driving simulator found that drivers react faster to shoulder rumble strips than to CLRSs, and 27 percent of them made initial leftward (toward the opposite lane) corrections upon contacting CLRSs. This might have been due to simulator conditions, however.

ADOT was concerned that CLRS constructability could be affected by retroreflective pavement markers but found that it could install the milled strips and recessed markers without conflict.
ADOT was concerned that CLRS constructability could be affected by retroreflective pavement markers but found that it could install the milled strips and recessed markers without conflict.

Pennsylvania State University field research found that CLRSs had a statistically significant effect on lateral vehicle placement away from the centerline, and decreased the variance of lateral placement. The researchers found no effect on speed.

Recommendations

Many States have installed and evaluated CLRSs, and most evaluations have found statistically significant positive benefits. ADOT's Flagstaff District has documented no adverse effects related to installation of CLRSs.

Based on their own findings and those of other States and studies, the ADOT researchers believe that milled CLRSs have the potential to prevent and reduce opposite direction crashes on undivided highways. The researchers recommend prioritizing installations on high-volume, high-speed rural roads, and coordinating placement with scheduled highway projects.

The researchers recommend a standard CLRS width of 30.5-centimeters (12 inches). Further, a distinct spacing pattern should be used for centerlines to help drivers distinguish them from shoulder rumble strips. The researchers further recommend using a pattern similar to that used by Michigan and Missouri, which consists of a pair of grooves at 30.5-centimeter (12-inch) spacing separated by a 61-centimeter (24-inch) gap. This pattern has a frequency and amplitude that generates effective tactile and audible stimuli, according to research by NCHRP, the University of Massachusetts, the Kansas DOT, and the Transportation Research Board.

Because CLRSs cause high-decibel sound, DOTs should avoid using them in residential areas. The researchers recommend that DOTs install CLRSs on continuous lengths of highway, avoiding placement in intersections or adjacent to left-turn lanes to eliminate noise produced by the turning traffic.

Finally, guidelines for appropriate use of CLRSs should be based on a logical, written policy to ensure effective and consistent use.

Shown here are milled CLRSs installed in a two-way passing zone on an Arizona roadway.
Shown here are milled CLRSs installed in a two-way passing zone on an Arizona roadway.

Future Installations

This study found that CLRSs effectively reduce crossover crashes involving fatigued and inattentive drivers. Based on the results of this study, ADOT currently is finalizing a CLRS policy, developing installation guidelines, creating CLRS groove pattern standard details, and defining the elements of an implementation program.

ADOT installed milled CLRSs at this left-turn pocket.
ADOT installed milled CLRSs at this left-turn pocket.

Kohinoor Kar, Ph.D., P.E., PTOE, is a transportation safety engineer at ADOT. He has a doctorate in civil engineering, specializing in transportation engineering, from Michigan's Wayne State University. He is experienced as a practitioner, researcher, instructor, and consultant. He serves on several national and State safety research advisory committees.

Richard S. Weeks, P.E., PTOE,is a senior project manager at the Phoenix office of AECOM. He has more than 30 years of planning, design, and operations experience in traffic safety, with emphasis on human factors.

The authors gratefully acknowledge the cooperation of ADOT's Flagstaff District in supporting installation of CLRSs for this evaluation.

For more information, contact Kohinoor Kar at 602-712-6857 or kkar@azdot.gov, or Richard S. Weeks at 602-337-2606 or richard.weeks@aecom.com.

Disclaimer: Any statements expressed in this article are those of the individual authors and do not necessarily represent the views of the agencies they are affiliated with.

ResearchFHWA
FHWA
United States Department of Transportation - Federal Highway Administration