Descriptions of Strategies
The objectives for reducing the frequency and severity of curve-related crashes are to
Exhibit V-1 presents these objectives and their related strategies for improving safety at horizontal curves. Because the AASHTO Strategic Highway Safety Plan is geared toward low-cost, short-term safety improvements, the list of strategies presented in the exhibit is arranged from low-cost, short-term treatments to high-cost, long-term treatments.
Explanation of Strategy Types
The strategies in this guide were identified from a number of sources, including the literature, contact with state and local agencies throughout the United States, and federal programs. Some of the strategies are widely used, while others are used at a state or even a local level. Some have been subjected to well-designed evaluations to prove their effectiveness. However, it was found that many strategies, including some that are widely used, have not been adequately evaluated.
The implication of the widely varying experience with these strategies, as well as the range of knowledge about their effectiveness, is that the reader should be prepared to exercise caution in many cases before adopting a particular strategy for implementation. To help the reader, the strategies in the AASHTO guides have been classified into three types, each identified by a letter:
Related Strategies for Creating a Truly Comprehensive Approach
It is recommended that related strategies be included as candidates in any program planning process to create a truly comprehensive approach to the highway safety problems associated with this emphasis area. There are five types of related strategies:
Objective 15.2 A: Reduce the Likelihood of a Vehicle Leaving its Lane and either Crossing the Roadway Centerline or Leaving the Roadway at a Horizontal Curve
The intent of this strategy is to provide advance warning to a driver that the horizontal alignment of the roadway is about to change and that the driver must alter the path,and possibly the speed, of the vehicle downstream of the warning to negotiate the curve safely. Advance warning of alignment changes should be provided to a driver when changes in alignment are unexpected. This typically occurs in situations where curves are sharper than anticipated or after a long tangent section of roadway.
Advance warning of alignment changes can be conveyed to the driver in numerous ways. The traditional approach is through the use of roadway signing. In the case of a “Curve” sign, the sign not only prepares the driver for a change in alignment, but it also provides information on whether the alignment turns to the left or to the right downstream of the sign. An advisory speed sign can be used to indicate a recommended speed through the curve. Flashing beacons can also be used with the “Curve” and advisory speed signs to draw more attention to these respective signs. Other methods of advance warning that have been used on a more limited basis include warning messages placed on the pavement and rumble strips in advance of the curve. These measures have been used primarily in advance of very sharp curves. In the case of rumble strips, the rumble strips are typically used in conjunction with “Curve” signs and advisory speed signs and are installed to call attention to the advisory speed signs. Note that installation of rumble strips in advance of curves may cause undesirable driving behaviors such as drivers purposely crossing over into the opposing lane to avoid the rumble strips. Also, some motorists (particularly truckers and motorcyclists) do not like the effects (vibration and sound) generated from the rumble strips. Other methods of advance warning involve pavement markings that try to cause a driver to reduce the speed of his/her vehicle through visual deception. These methods can include transverse lines with decreasing spacing or edgelines that give the appearance of a narrowing lane width. Research is underway in National Cooperative Highway Research Program (NCHRP) Project 3-61 to develop a methodology by which horizontal curve information can be conveyed to motorists in a more consistent and reliable fashion.
This strategy focuses on providing drivers with advance warning of the horizontal curve. In some cases it is sufficient just to heighten the awareness of the driver that he/she is approaching a change in alignment. In other situations, advance warning treatments try to influence the speed of the driver on the approach to the curve. Affecting speeds on the tangent sections preceding horizontal curves is particularly important because excessive speed is a significant factor in crashes at horizontal curves. Moreover, research has shown that drivers do not fully adjust their speeds on the approach. The speed at which a vehicle enters a curve relates more to the speed of the vehicle as it approaches the curve (which is based in part on the driver’s response to the preceding alignment) than to the sharpness of the curve (Retting and Farmer, 1998).
This strategy closely relates to Strategy 15.1 A4 in the guide for addressing ROR accidents (Volume 6 of this report), which pertains to enhanced delineation of sharp curves for reducing ROR crashes. Strategy 15.1 A4 focuses on innovative and experimental onpavement markings (nontraditional treatments) that provide advance warning of horizontal curves. Strategy 15.2 A1 pertains to both traditional and nontraditional advance warning treatments at horizontal curves. These treatments do not provide the driver with a view of the curve. On the other hand, Strategy 15.2 A2 focuses on delineation treatments installed along the curve, which provide the driver with a picture of the sharpness of the curve.
Summary of Effectiveness of Nontraditional Treatments at Horizontal Curves
This section provides a brief summary of what is known about the safety effectiveness of nontraditional treatments that provide advance warning to horizontal curves, as presented in the ROR guide. Several variations of nontraditional pavement marking treatments have been experimented with to improve safety at horizontal curves. The two most promising treatments are the pavement arrow (Exhibit V-2) and transverse striping treatments. In general, these nontraditional pavement marking treatments have reduced both speeds and accidents at horizontal curves in experiments conducted by a few agencies. However, it has yet to be determined how effective these nontraditional pavement marking treatments will be when installed on a broader basis and over the long term at a given site.
Effectiveness of Traditional Advance Warning Treatments at Horizontal Curves
The Manual on Uniform Traffic Control Devices (MUTCD) (USDOT, 2003) indicates that horizontal alignment signs (Turn [W1-1], Curve [W1-2], Reverse Turn [W1-3], Reverse Curve [W1-4], or Winding Road [W1-5]) may be used in advance of situations where the horizontal roadway alignment changes. The One-Direction Large Arrow (W1-6) sign may be used on the outside of the turn or curve. If the change in horizontal alignment is 135 degrees or more, the Hairpin Curve (W1-11) sign may be used. If the change in horizontal alignment is approximately 270 degrees, such as on a cloverleaf ramp, the 270-degree Loop (W1-15) sign may be used. Additional warning also may be provided by use of the “Advisory Speed” plaque (W13-1) that is intended to indicate the maximum recommended speed around a curve. The MUTCD states that the Advisory Speed plaque shall be used where an engineering study indicates a need to advise road users of the advisory speed for a condition.
Research suggests that the proliferation of curve warning signs, especially those supplemented with advisory speed plates, may have lessened the average motorist’s respect for the messages that they convey (Lyles, 1980). However, because of tort liability concerns, many highway agencies prefer to use traditional advance warning and curve signs even if research indicates that these signs may be ineffective. The findings from studies that investigated the effectiveness of traditional advance warning signs are summarized in the following paragraphs.
Lyles (1980) examined the effectiveness of five sign treatments for controlling driver speeds in the vicinity of hazardous horizontal curves on rural two-lane highways. Sign treatments ranged from the standard curve warning sign to a regulatory speed zone sign in conjunction with a curve warning sign. The effectiveness of the signs was evaluated based on speeds of motorists as they approached and negotiated the horizontal curves and whether vehicles crossed over center and edgeline markings. Lyles found that no sign, or group of signs, was consistently more effective than another in decreasing the potential hazard at horizontal curves.
Zwahlen (1983) examined the effectiveness of advisory speed plates in causing drivers to reduce their speeds through curves. He concluded that advisory speed signs are not more effective in causing drivers to reduce their speeds through curves than the curve signs alone are, at least not in dry weather, and that further research was needed to determine the effectiveness of advisory speed signs in adverse weather conditions. Zwahlen recommended that advisory speed sign maintenance, especially new installations, be given a low priority.
Ritchie (1972) examined the choice of speed in driving through curves as a function of advisory speed and curve signs. He found that motorists drove faster and produced more lateral acceleration when (a) a curve sign was present, and (b) an advisory speed sign was present, than under the opposite conditions. In addition, motorists exceeded advisory speed signs of 24 to 56 km/h (15 to 35 mph), but motorists did not exceed advisory speed signs of 72 to 80 km/h (45 to 50 mph). Ritchie concluded that advance warning signs serve to reduce uncertainty and allow drivers to proceed with greater confidence.
One of the reasons for the low percentage of compliance with posted advisory speeds on curves may be that the criteria for setting advisory speeds on curves are outdated due to advances in vehicle characteristics. The current criteria for setting advisory speeds on curves have remained essentially unchanged for more than 50 years. Chowdhury et al. (1998) evaluated the validity of current criteria for determining advisory speeds on horizontal curves and concluded that the criteria are not valid for modern vehicles. At most curves, posted advisory speeds were well below the prevailing traffic speed and below the recommended values suggested by the two methods for determining advisory speeds, namely the ball-bank indicator and the Traffic Control Devices Handbook (TCDH) (Institute of Transportation Engineers, 2001).
While the previously mentioned studies suggest that traditional advance warning treatments are not effective in decreasing the potential hazard at horizontal curves, several studies suggest otherwise. Hammer (1968) evaluated the effectiveness of various types of minor improvements in reducing accidents. Two of the minor improvements included in the evaluation were the installation of curve warning signs and advisory speed signs at horizontal curves. Hammer found that curve warning signs reduced accidents by 18 percent at horizontal curves and that installation of both curve warning and advisory speed signs reduced accidents by 22 percent. Leisch (1971) also reported advisory speed signs to be effective in reducing accidents at horizontal curves.
Hanscom (1976) evaluated a slightly different scenario. He evaluated the effects of signing to warn drivers of wet weather skidding hazards at horizontal curves. Three curved highway sections were treated using five experimental sign treatments. The primary measure of effectiveness was mean speed at the critical curve locations. In particular, the target sample was the highest quartile speed group of vehicles arriving in advance of the curve. Significant speed reductions were observed at critical curve locations during conditions of wet pavements when warning signs were supplemented with flashing beacons. Therefore, Hanscom recommended that activated warning signs be used at critical curve locations as a skidding accident countermeasure.
Several other types of traditional advance warning treatments that have not necessarily been evaluated for their safety effectiveness at horizontal curves include oversized warning signs and double-posted signs. The MUTCD (USDOT, 2003) indicates that oversized warning signs may be used where speed, volume, and other factors result in conditions where greater visibility or emphasis would be desired, such as at unexpected or sharp horizontal curves. Agencies have also double-posted warning signs to draw greater attention to warning signs.
In summary, none of the studies designed to evaluate the effectiveness of traditional advance warning treatments at horizontal curves question the importance of providing a curve warning sign in advance of unexpected or sharp curves, but conflicting results have been obtained on the effectiveness of advisory speed signs. The most recent studies suggest that advisory speed signs do not garner respect from the average motorist. These studies conclude that advisory speed signs do not effectively reduce speeds at horizontal curves.
Before drawing conclusions regarding the effectiveness of advisory speed signs on improving safety at horizontal curves, two issues should be considered. First, of the studies cited above, only Hammer evaluated the effectiveness of advisory speed signs using accident data. The other studies used speed as the measure for evaluating the effectiveness for advisory speed signs. Second, Hanscom is the only reference cited above that recommends targeting the highest quartile speed group of vehicles when evaluating the effectiveness of advance warning treatments based upon speed. He suggests that these vehicles are the vehicles most likely to be involved in accidents at horizontal curves.
The ROR guide presents attributes under Strategy 15.1 A4 that are common to this strategy. The reader is, therefore, directed to that guide for more detailed information related to this strategy. However, three additional points should be considered in addition to what is presented within the ROR crash guide.
First, for higher compliance with posted advisory speed signs, a new set of criteria should be developed for setting advisory speeds. Chowdhury et al. (1998) recommended determining the advisory speed based on a sample of vehicle speeds, but other alternatives should be investigated as well. If a new set of criteria is developed for setting advisory speed, curves currently posted with advisory speed signs will have to be re-evaluated and new advisory speeds will have to be posted, and in some cases the advisory speed signs may be removed completely. A public information effort may be needed to re-educate the driving public until drivers once again respect this type of advisory sign.
Second, an important key to success is identifying sites where treatments of this nature have the potential to improve safety. This strategy targets curves where changes in alignment are unexpected and drivers may need to reduce their speeds to negotiate the curve safely. Third, in several of the studies previously mentioned, speed, or more specifically change in speed, has been used as a surrogate measure for evaluating the effectiveness of a treatment.
Such a relationship should be established by further research. If speed is studied as a surrogate measure, consideration should be given to Hanscom’s recommendation of targeting the highest quartile speed group of vehicles when evaluating the effectiveness of both traditional and nontraditional advance warning treatments.
This strategy focuses on providing the driver with better visual cues to recognize the presence and geometry of the curve. Various methods are available to provide delineation along a curve. Some traditional delineation devices such as chevrons, post-mounted delineators, and delineators placed on guardrail are located outside the roadway shoulder along the curve, while others, such as lane lines or edgelines and raised pavement markers, are placed on the surface of the traveled way. Several nontraditional devices, such as light-emitting diode (LED) in-pavement luminaires and LED barrier-mounted guidance tubes, have also been used for delineation purposes.
Agencies generally implement three levels of delineation based on the context of the location:
This strategy primarily addresses delineation along problem curves, but it may also be applied to most curves. Enhanced delineation of a curve serves two purposes. First, it can provide a better view of the curve on the approach tangent. The degree to which this works well depends in part upon a combination of road factors, including horizontal and vertical alignments, obstructions on the inside of the curve, and the types of delineation devices used. Delineation helps prepare the driver for the approaching change in horizontal alignment. Roadside delineators are particularly effective in providing this advanced view of the curve. In many cases, delineation devices increase the preview sight distance on the approach tangent.
Second, as the driver traverses the curve, the delineation device provides a continuous feature for positive guidance. This helps the driver position his/her vehicle within the proper travel lane while negotiating the curve.
This strategy is related to Strategy 15.1 A6 in the ROR guide, pertaining to better pavement markings at appropriate locations. Strategy 15.1 A6 in the ROR crash guide does not provide detailed information on post-mounted delineators or chevrons. Therefore, a more detailed discussion of these types of roadside delineation is provided below.
Summary of Effectiveness of Better Pavement Markings
This section provides a brief summary of the safety effectiveness of better pavement markings, as presented in the ROR guide. “Better pavement markings” are pavement markings that are more durable, are all-weather, or have a higher retroreflectivity than traditional pavement markings. Raised pavement markings and wider edgelines are two approaches to enhancing delineation at a curve. These treatments are designed to help drivers who might leave the roadway because of inability to see the edge of the pavement along the horizontal curve.
The actual safety benefits of such treatments are difficult to assess. Raised pavement markers provide for increased delineation of the driving path and enhance the ability of the driver to track the roadway, particularly under nighttime, wet-weather, or adverse-weather conditions. Raised pavement markers also can provide tactile and auditory warnings to drivers, similar to rumble strips, when vehicles traverse the markers. When used at isolated curves rather than continuously along the alignment, raised pavement markers may provide greater emphasis to the change in roadway alignment. Several studies have noted significant reductions in accidents because of the installation of raised pavement markers. Despite the noted advantages of raised pavement markers and the positive research, some studies have indicated an increase in nighttime accidents after the installation of raised pavement markers. Therefore, the safety effectiveness of raised pavement markers is questionable. Concerning wider edgelines such as 20-mm (8-in.) edgelines versus 10-mm (4-in.) edgelines, the effectiveness of raised pavement markers in reducing ROR crashes has not been satisfactorily demonstrated in the research literature, although the New York DOT indicates that wider edgelines have the potential to reduce ROR crashes on two-lane roads by 10 to 15 percent.
Post-Mounted Delineators and Chevrons
Post-mounted delineators and chevrons are two types of delineation treatments that are installed outside of the roadway. They are intended to warn drivers of an approaching curve and provide tracking information and guidance to the drivers. While they are intended to act as a warning, it should also be remembered that the posts, placed along the roadside, represent a possible object with which an errant vehicle can crash. Design of posts to minimize damage and injury is an important part of the considerations to be made when selecting these treatments.
In NCHRP Report 440, Fitzpatrick et al. (2000a) report the results of several studies on postmounted delineators. They report that post-mounted delineators reduce the accident rate only on relatively sharp curves during periods of darkness. In addition, highways with postmounted delineators have lower accident rates than highways without post-mounted delineators, and the cost of post-mounted delineators are justified for highways with average daily traffic (ADT) exceeding 1,000 vehicles per day. Fitzpatrick et al. do not quantify the effectiveness of post-mounted delineators in reducing curve-related crashes. Bali et al. (1978) provide similar results.
Krammes and Tyer (1991) evaluated the operational effectiveness of raised pavement markers as an alternative to post-mounted delineators at horizontal curves on two-lane rural highways. They evaluated nighttime speed and lateral placement data from five sites. For both short-term and intermediate-term analyses, vehicle operations with raised pavement markers compared favorably with operations when post-mounted delineators were present. Vehicle operations were not significantly affected on the inside lane of the curve, but significant differences were observed on the outside lane of the curve. Speeds at the midpoint of the curve were consistently 1.6 to 4.8 km/h (1 to 3 mph) higher with the raised pavement markers, and the mean lateral placement of vehicles was consistently 0.3 to 0.6 m (1 to 2 ft) further from the centerline at the midpoint of the curve with the raised pavement markers than with the post-mounted delineators. In addition, the variability in lateral placement of vehicles at the midpoint of the curve was less with raised pavement markers than with post-mounted delineators.
Zador et al. (1987) examined the short- and long-term effects of chevrons, post-mounted delineators, and raised pavement markers on the speed and placement of vehicles traveling on curves on rural two-lane highways. In general, all three delineation treatments affected driver behavior at night. Vehicle paths were shifted away from the centerline on horizontal curves where raised pavement markers and chevrons were installed and toward the centerline on curves where post-mounted delineators were used. Vehicle speed and placement variability were also slightly reduced with the use of chevrons and raised pavement markers. Zador et al. did not conclude that one delineation treatment was superior to the others and indicated that the primary benefit of any of these delineation treatments may simply be that they help drivers better recognize that they are approaching a curve.
Agent and Creasey (1986) investigated the ability of various traffic control devices to delineate horizontal curves so drivers would perceive the curve and slow to an appropriate speed and so drivers would have improved guidance through the curve. The investigation consisted of both laboratory tests and field data collection. The laboratory tests suggested that increasing the height of the post-mounted delineator while maintaining the distance from the post to the pavement edge, and keeping the post spacing constant, made a curve appear sharper than other delineator devices. From speed data, encroachment data, and some accident data, Agent and Creasey found that pavement markings had a greater effect on drivers than post-mounted delineators installed on the roadside did. In addition, chevrons had slightly more influence on speeds and encroachments than other post-mounted delineators did.
Jennings and Demetsky (1985) evaluated the effectiveness of three post-mounted delineator systems in controlling ROR crashes. The post-mounted delineator systems were evaluated based upon changes in speed and lateral placement of vehicles within the travel lane. Jennings and Demetsky found that drivers reacted most favorably to chevron signs on sharp curves greater than or equal to 7 degrees (radius of 250 m [820 ft]) and to standard postmounted delineators on curves less than 7 degrees.
In summary, the safety effectiveness of enhanced delineation at a horizontal curve is difficult to assess because many of the research results are conflicting. Part of the difficulty arises because several of the studies use modifications in speed and lateral placement as surrogate measures to evaluate safety rather than actual crash data. The general conclusions that may be drawn regarding the safety effectiveness of enhanced delineation at horizontal curves are that post-mounted delineators may improve safety at sharp curves and that chevrons are more effective than standard post-mounted delineators are. At this point, no quantitative estimates of the safety effectiveness of enhanced delineation treatments can be made. Zador et al. may have summarized the safety effectiveness of enhanced delineation best by indicating that the primary benefit may simply be that enhanced delineation treatments help drivers better recognize that they are approaching a horizontal curve.
Sight distance is a fundamental element in geometric design. The amount of sight distance provided to the driver is a function of the three-dimensional features of the highway—the cross section (roadside), vertical alignment (grades and vertical curves), and horizontal alignment. At horizontal curves, obstructions that limit the driver’s sight distance come in many shapes and forms. The road surface may be the sight obstruction if the horizontal curve is located on a crest vertical curve. Physical features outside of the traveled way—such as trees or bushes, guardrail or concrete barriers, and the backslope of a cut section—also can limit the driver’s sight distance. As trees and other roadside vegetation mature, the sight distance at a horizontal curve may change. Motor vehicles and other road users can also create temporary sight obstructions. Efforts should be made to ensure that obstructions do not reduce the sight distance at a horizontal curve to less than the minimum stopping sight distance.
The available stopping sight distance on a roadway should be sufficiently long to enable a vehicle traveling at or near the design speed to stop before reaching a stationary object in its path (AASHTO, 2001). Providing at least the minimum stopping sight distance at every point along a roadway is critical for safe operations.
Stopping sight distance is the sum of (1) the distance traversed by the vehicle from the instant the driver sees an object necessitating a stop to the instant the brakes are applied (i.e., the brake reaction distance) and (2) the distance needed to stop the vehicle from the instant the brakes are applied (i.e., the braking distance). Furthermore, drivers have other sight distance needs in addition to stopping for hazards in the paths of their vehicles. A driver needs an adequate view of the roadway alignment and roadway features ahead for safe control and guidance of the vehicle (Gattis and Duncan, 1995). This sight distance to the roadway surface and other appurtenances ahead is referred to as preview sight distance (PVSD). A roadway designed with geometric design features adequate to the design speed would in many cases provide sufficient PVSD; however, a roadway with constrained design features, or a roadway section that does not conform to current geometric design policies, could have inadequate PVSD.
Current design policy does not identify where lack of sight distance may produce a significant safety risk. NCHRP Report 400 (Fambro et al., 1997) suggests that most locations with limited stopping sight distance experience very few accidents. However, limited stopping sight distance may be a greater concern where an intersection or driveway is present along a horizontal curve. Objective 17.1 C in the unsignalized intersection guide provides more detailed information about intersection sight distance.
If the available sight distance is found to be less than the minimum stopping sight distance, the sight obstruction should be removed or the roadway should be realigned to provide at least minimum stopping sight distance. The safety effectiveness of increasing sight distance will be a function of the amount of the sight restriction, the traffic volume exposed to it, and the presence of other conditions contributing to risk. For additional information on stopping sight distance, see NCHRP Report 400 (Fambro et al., 1997).
This strategy focuses on issues related to the safety effectiveness of shoulder rumble strips at horizontal curves (Exhibit V-4). While shoulder rumble strips are designed primarily to reduce ROR crashes, they can also reduce head-on crashes. Head-on crashes may occur when a vehicle leaves the roadway and its driver overcompensates while trying to recover control of the vehicle, sending the vehicle into the opposing traffic lane.
One of the unique issues related to the safety effectiveness of shoulder rumble strips at horizontal curves concerns the departure angle for vehicles that leave the roadway. Studies indicate that an average departure angle for ROR crashes ranges between 3 and 8 degrees (Hall, 1991; O’Hanlon and Kelley, 1974). In these studies, it is not clear whether the analyses included crashes that occurred along tangent sections of highway, crashes that occurred at horizontal curves, or both. Regardless, the research results indicate that ROR crashes typically occur at shallow departure angles. However, the departure angle is a function of the horizontal alignment. If a vehicle drifts to the right along a tangent section of roadway at an angle of 3 degrees, the vehicle follows a certain path onto the roadside. If the same vehicle path occurs along a horizontal curve, the result will be a larger departure angle because of the curvature of the roadway. This has two implications on the effectiveness of shoulder rumble strips at horizontal curves. First, because the vehicle crosses the rumble strips at a greater angle, an inattentive driver has less exposure time to the stimuli (i.e., vibration and noise) generated by the rumble strips. Second, if the shoulder width on the curve is the same as on an adjacent tangent, the driver has less time to regain control of the vehicle before it leaves the shoulder.
This strategy is the same as Strategy 15.1 A1 provided in the ROR guide. The section below provides a summary of the effectiveness of shoulder rumble strips. Following that is a section that presents special issues concerning shoulder rumble strips at horizontal curves.
Summary of Safety Effectiveness of Shoulder Rumble Strips
Shoulder rumble strips have proven to be an effective measure in reducing the number of ROR crashes on freeways. Numerous studies have quantified the reductions in ROR crashes to varying degrees. In general, the studies indicate that ROR crashes were reduced by 20 to 50 percent because of the installation of shoulder rumble strips. The section of this guide on the description of the problem indicates that horizontal curve fatal crashes primarily occur on rural two-lane highways. Because little is known about the effectiveness of shoulder rumble strips on two-lane highways, the ROR guide suggests that one might assume a similar experience to what has been documented on rural freeways—a 20- to 30-percent reduction in single-vehicle ROR crashes. However, the reader should review the considerations that are listed below before making such an estimate.
The ROR guide presents attributes common to this strategy under Strategy 15.1 A1. The reader is directed to the ROR guide for more detailed information. In addition, however, several issues should be pointed out, particularly regarding the installation of shoulder rumble strips at horizontal curves.
Studies concerning the safety effectiveness of shoulder rumble strips have used crash data collected over long segments of highway, meaning that the study segments included both tangents and horizontal curves. No distinction was made in these studies between tangent and horizontal curve sections, and there are no studies that analyze the effectiveness of shoulder rumble strips at horizontal curves only. It might thus be assumed that similar safety benefits apply to the application of shoulder rumble strips along both types of alignments (tangents and curves). However, it should be recognized that the effectiveness of shoulder rumble strips in reducing ROR crashes depends on various elements, including the frequency with which vehicles in the traffic stream run off the road, the vehicle departure angle, the vehicle speed, the shoulder width, and the roadside environment. The vehicle departure angle is of particular interest in this case.
It is, therefore, a complex issue to speculate on the specific effectiveness of shoulder rumble strips specifically on curves. Consider some of the following attributes:
Incompatibility between shoulder rumble strips and bicycle use is a concern in some locales. For example, a moratorium on the installation of ground-in rumble strips where bicycles were allowed was initiated in California until further research on the subject is completed (Bucko and Khorashadi, 2001). The three most comprehensive studies on the effects that rumble strips have on bicyclists were conducted in Pennsylvania, California, and Colorado by Elefteriadou et al. (2000), Bucko and Khorashadi, and Outcalt (2001a), respectively. Each study included bicycle and motor vehicle testing of various rumble strip designs. In general, the rumble strips that provided the greatest amount of stimuli (i.e., noise and vibration) to alert an inattentive or drowsy driver also were the most uncomfortable for the bicyclists to traverse. Likewise, the rumble strips that were the most comfortable for the bicyclists generated the least amount of stimuli in a motor vehicle to alert a drowsy or otherwise inattentive driver. In all three studies, compromises were made when selecting the rumble strip design most compatible for both types of road users.
Shoulder width is a major issue to consider before installing shoulder rumble strips. For further details on designs used by some states, see Appendix 1.
Finally, the impact of rumble strips on pavement performance is an issue that is often overlooked. Because rumble strips reduce the effective structural cross section of the pavement, rumble strips may reduce the overall pavement life or require greater total pavement thicknesses if significant loadings are anticipated to the shoulder or rumble strip area. Elefteriadou et al. (2001) provide some discussion of pavement integrity issues related to rumble strip installation.
For additional information on shoulder rumble strips, Synthesis of Shoulder Rumble Strip Practices and Policies was recently published by SAIC (2001), and FHWA published Technical Advisory for Roadway Shoulder Rumble Strips (FHWA, 2001). Both documents are available from the FHWA rumble strip Web site (http://safety.fhwa.dot.gov/programs/rumble.htm).
It should be noted that the ROR guide addresses the use of mid-lane rumble strips (Strategy 15.1 A3), which serve a similar purpose to shoulder rumble strips except that mid-lane rumble strips are installed in the center of the travel lane instead of on the shoulder.
Centerline rumble strips are installed primarily to reduce head-on and sideswipe crashes along undivided roadways. Their primary function is to alert drowsy or otherwise inattentive drivers that their vehicles are encroaching upon the opposing lane through tactile and auditory stimulation. Centerline rumble strips may also discourage drivers from cutting across the inside of a curve. There is no standard design for centerline rumble strips, but generally the rumble strips are either (1) located along the width of the centerline pavement markings (Exhibit V-5), extending into the travel lane by as much as 0.5 m (1.5 ft) (Exhibit V-6), or (2) placed on either side of the centerline (Exhibit V-7). Some states install rumble strips continuously along the centerline, while other states install centerline rumble using a skip pattern.
Installing centerline rumble strips directly relates to Strategy 18.1 A1 in the guide for addressing head-on collisions, which is centerline rumble strips for two-lane roads. Centerline rumble strips are a relatively new strategy for reducing head-on crashes. Subsequently, little information is available on the safety effectiveness of this type of rumble strip. The section immediately following this one summarizes the effectiveness of centerline rumble strips as presented in the head-on guide, as well as the results of three recent studies on centerline rumble strips. The final section under this strategy discusses strategy attributes that relate to applying centerline rumble strips at horizontal curves.
Summary of Safety Effectiveness of Centerline Rumble Strips
The head-on guide identifies two studies that showed centerline rumble strips to be effective in reducing head-on crashes. Centerline rumble strips were installed on a two-lane, undivided rural highway in Delaware (Perrillo, 1998). During the 36-month before period, there were 6 fatal crashes, 14 injury crashes, and 19 property damage only crashes. During the 24-month after period, there were 0 fatal crashes, 12 injury crashes, and 6 property damage only crashes. It was concluded that the centerline rumble strips reduced the total number of crashes and the severity of the crashes. In California, improvements were made to a 32-km (20-mi) segment of a rural two-lane highway to reduce the number of head-on crashes (Fitzpatrick et al., 2000a). The improvements included replacing the double yellow stripes with centerline rumble strips and raised profile thermoplastic traffic striping. In addition, raised pavement markers were installed between the rumble strips and raised profile thermoplastic. Using 34 months of before data and 25 months of after data, an evaluation showed that the centerline rumble strips and other improvements reduced the crash frequency from an average of 4.5 crashes per month in the before period to 1.9 crashes per month in the after period.
Rys et al. (2003) conducted a study to determine the most effective centerline rumble strip pattern for use on Kansas roadways. A survey of the few agencies that currently use centerline rumble strips found that there was no generally accepted pattern as to the types and dimensions of these rumble strips. Using information gathered on some of the patterns currently used, a test section of roadway was prepared by installing 12 different sections of centerline rumble strips of varying dimensions and spacing. Seven test vehicles were chosen to represent an accurate range of roadway traffic, and two measurements were taken in each automobile: interior noise level and steering wheel vibration level. Based on their findings, Rys et al. recommended that further testing be conducted on two of the centerline rumble strip patterns to more clearly determine which is most beneficial: either the continuous pattern that is 305 mm (12 in.) on the center and 305 mm (12 in.) long or the alternating pattern that is 305 mm and 610 mm (12 in. and 24 in.) on the center and 305 mm (12 in.) long. The authors noted that each of these patterns rated high in both noise level and vibration created; both were installed along urban and rural roadways in Kansas for continued testing during the 2003 year.
Mahoney et al. (2003) conducted a before-after study to determine whether centerline rumble strips have an effect on the lateral displacement of vehicles. Data were collected on eight roadway sections (four test and four comparative sections) in rural settings so that an operational analysis, rather than a safety analysis, could be performed on centerline rumble strips. To reduce the effect of outside influences on the lateral placement, tangent segments were chosen with minimal grade, no roadside barriers, and nominal horizontal curvature. The data analysis revealed that the mean lateral placement of vehicles shifted 140 mm (5.5 in.) away from the center of the lane subsequent to centerline rumble strip installation along roadway sections with 3.6-m (12-ft) lanes, and the mean lateral placement of vehicles shifted 76 mm (3 in.) away from the center of the lane subsequent to centerline rumble strip installation along roadway sections with 3.3-m (11-ft) lanes. Introduction of centerline rumble strips also decreased the amount of lateral placement variance that, in previous studies, had been shown to possibly increase traffic safety. The effects of centerline rumble strips on vehicle speed were inconclusive.
In 2001, Colorado DOT completed a before-after evaluation of 27 km (17 mi) of centerline rumble strips installed along a winding, two-lane mountain road (Outcalt, 2001b). The analysis used 44 months of before data and 44 months of after data. The resulting crash data and associated percent changes are shown in Exhibit V-8.
The head-on guide presents attributes common to Strategy 15.2 A5 and Strategy 18.1 A1. The reader is directed to the guide for addressing head-on collisions for more detailed information related to this strategy. In addition, several issues should be highlighted concerning the effectiveness of centerline rumble strips, policy issues, and potential difficulties.
When considering the expected effectiveness of centerline rumble strips in reducing head-on crashes at horizontal curves, similar issues to those discussed in Strategy 15.2 A4 should be considered.
Arizona, California, Colorado, Delaware, Kansas, Maryland, Massachusetts, Minnesota, Oregon, Pennsylvania, Virginia, Washington, and Wyoming are among the states that have installed centerline rumble strips. To learn more about Minnesota DOT’s experiences with centerline rumble strips see http://www.dot.state.mn.us/d3/newsrels/03/10/06_rumble_strips.html. These installations have primarily been on an experimental basis. After agencies have sufficient experience with this new technique, a written policy should result for centerline rumble strips. The policies may include guidelines or recommendations regarding the type of sites at which to install centerline rumble strips, as well as design specifications and pavement thickness requirements.
The possibility of centerline rumble strips adversely affecting motorcyclists and inhibiting passing maneuvers is mentioned in the guide for addressing head-on collisions. However, experiences in Pennsylvania, Washington, and Minnesota suggest that this may be more a perceived problem than an actual problem. In Connecticut, however, centerline rumble strips were installed on a short section (less than 1.6 km [1 mi]) of a state route that carried a high percentage of truck traffic. The centerline rumble strips were removed after approximately 8 months because of complaints about noise.
Other potential disadvantages of centerline rumble strips include decreased visibility of centerline pavement markings, potential drainage problems, and snow removal difficulties. However, experience has not proven these potential disadvantages to be significant or insurmountable. In fact, the opposite may be true in some cases. At least one agency with centerline rumble strip installations has noted that the visibility of centerline pavement markings is not diminished because of centerline rumble strips and that centerline pavement markings are even visible when the rumble strips are filled with water. Likewise, at horizontal curves where greater superelevation can be expected, interruption of drainage flow patterns should be minimal. Concerning snow removal difficulties, no agency has indicated a reduction in pavement life because of centerline rumble strips. But centerline rumble strips have only recently been implemented, so more time is necessary to adequately address this issue. However, at least one agency has received comments that motorists perceived the centerline rumble strips as beneficial during snowy conditions because the motorists were still able to hear and feel the rumble strips in that kind of weather. An added benefit of centerline rumble strips is that they may extend centerline marking life because they decrease the number of vehicles crossing the markings.
Finally, the primary purpose of centerline rumble strips is to reduce head-on and sideswipe crashes. Centerline rumble strips also have the potential to reduce ROR crashes that occur to the left. If vehicles traveling on the inside of a curve cross the centerline, the centerline rumble strips alert the driver as soon as the vehicle encroaches on the centerline. This maximizes the recovery time and distance for vehicles that can run off the road to the left.
Preventing edge dropoffs, also referred to as edgedrops, can reduce both ROR and head-on crashes by enabling a driver to recover an errant vehicle in a more controlled fashion. Edge dropoffs are a significant difference in elevation between the edge of traveled way and shoulder (Exhibit V-9). Edge dropoffs may occur after resurfacing or as the result of weather or vehicle-related settlement and can occur whether the shoulder is paved or not. Edge dropoffs may be more common on curves than on tangents.
Edge dropoffs of more than 10 mm (4 in.) have been shown to contribute to loss of control. Drivers who inadvertently drift onto the shoulder find their right wheel caught against the dropoff. This may induce overcorrecting by the driver, with resultant sudden loss of control or steering into the opposing lane. This behavior may be exacerbated when the driver is tracking a horizontal curve.
The best practice is to always retain the travel lane and shoulder at the same elevation, where they meet. Where this cannot be achieved, such as on roadways with unpaved shoulders, an alternative is to smooth the transition between the traveled way and shoulder surfaces using a wedge of pavement that allows vehicles to safely return to the roadway. For example, during pavement work in the state of Idaho, “moulding shoes” are sometimes equipped on the outside of the pavers to provide safe asphalt slopes. Georgia is also working on a 30-degree asphalt fillet. This strategy is related to strategies under the section in the ROR guide entitled “Apply Shoulder Treatments” (Strategy 15.1 A8).
Reference is usually made to edge dropoffs in the context of the boundary between the traveled-way pavement and the shoulder surface. Edge dropoffs can also occur at the boundary between the shoulder surface and roadside. Efforts should be made to prevent both types of edge dropoffs.
Particular care should be taken to minimize the potential risks of edge dropoffs in work zones. Edge dropoffs can commonly occur in work zones as the result of overlays, pavement replacement, or shoulder construction. The depth of these elevation differentials can vary from approximately 2.54 mm (1 in.), when a flexible overlay is applied, to several meters, when major reconstruction is undertaken. McDonald et al. (2002) reviewed temporary traffic control strategies in numerous states addressing edge dropoff differentials and analyzed crash data and litigation related to edge dropoffs. McDonald et al. also developed recommendations for mitigating edge dropoffs in work zones.
Summary of Effectiveness of Preventing Edge Dropoffs
The ROR guide indicates that little is known about the safety effectiveness of edgedrop treatments because it is difficult to specifically define the percentage of crashes that are caused by edge dropoffs. Regardless of the percentage, it has been proposed that a simple 45-degree-angle asphalt fillet at the lane edge would virtually eliminate this type of crash for shoulder dropoffs (Humphreys and Parham, 1994).
Current design criteria for horizontal curves are formulated to provide comfort to the driver in tracking the curve while keeping vehicles from skidding on wet pavements. The criteria are based upon the standard curve formula that provides that a portion of the lateral acceleration developed by the vehicle will be resisted by superelevation and the remainder by tire-pavement friction. A vehicle will skid during braking and maneuvering when frictional demand exceeds the available friction at the tire-pavement interface.
Much research has been conducted to address curve operations, driver speed, vehicle paths, and safety. Harwood and Mason (1994) evaluated the margin of safety against skidding for a passenger car and truck on a horizontal curve. The margin of safety was defined as the difference between the available tire-pavement friction and the friction demand of the vehicle as it tracks the curve. The authors determined that existing design criteria provide an adequate margin of safety against vehicles skidding off the roadway, assuming vehicles do not exceed the design speed of the roadway and vehicles traverse the curve on a path that follows a constant radius equal to the radius of the curve.
The likelihood of skidding increases when these assumed conditions are violated. Several studies have shown that under real-world conditions both of these assumptions are violated to some degree (Bonneson, 2000; Glennon et al., 1985; Glennon and Weaver, 1972), with the result being that at many curve sites the assumed margin of safety may actually be overestimated. Where this is the case and there is evidence of loss of control because of skidding, several solutions are evident. Solutions may include modifications to the alignment and roadside to control speeds, changing the superelevation along the curve, and/or providing pavement surfaces with better skid resistance. Strategy 15.2 A7, however, focuses upon providing pavement surfaces with better skid resistance.
Summary of Effectiveness of Providing Better Skid-Resistant Pavement Surfaces
This strategy directly relates to Strategy 15.1 A7 in the ROR guide on skid-resistant pavement surfaces. Although further details may be found there, this section provides a brief summary of the safety effectiveness of providing better skid-resistant pavement surfaces, as presented in the ROR guide. New York State has implemented a program that identifies sites statewide that have a low skid resistance and treats them with overlays or microsurfacing as part of the maintenance program. Between 1995 and 1997, 36 sites were treated on Long Island, resulting in a reduction of more than 800 annually recurring wet-road accidents. These results support earlier findings that improving the skid resistance at locations with high wet-road accident frequencies results in reductions of 50 percent for wet-road accidents and 20 percent for total accidents. While these results could be subject to some regression-tothe-mean bias, there is an indication that improving the skid resistance of pavement surfaces reduces wet-road and total accidents. Some states, including California, resurface short roadway segments such as horizontal curves with open-graded asphalt friction courses to improve skid resistance and safety.
The ROR guide presents attributes common to this strategy, under Strategy 15.1 A7. The reader is directed to the ROR guide for more detailed information related to this strategy. The signalized intersection guide also discusses similar treatments under Strategy 17.2 G2.
In conjunction with this strategy, an agency should consider scheduling routine pavement friction tests and creating a pavement friction inventory program. Ideally, this type of program would include the entire roadway network within an agency’s jurisdiction, but at a minimum it should include the highest-volume roadways. Caltrans operates an Office of Pavement Rehabilitation, which includes a program of pavement friction inventory (http://www.dot.ca.gov/hq/esc/Translab/opr.htm). Routine pavement friction tests should be conducted on both tangent and curve sections of a highway. Research conducted by NYDOT in the late 1990s revealed that, under high-volume conditions, significant reductions in friction occurred at curves, compared with tangent sections of the same road segment treated with the same surface treatment.
Finally, drainage is an important issue to consider when implementing this strategy. As the water film thickness on the pavement increases, the likelihood of hydroplaning increases. Therefore, any drainage problems should be corrected in conjunction with this strategy. While checking for and/or correcting any drainage problems, deficiencies in the superelevation and pavement edge profiles should also be checked and improved if found deficient.
Pavement grooving is a technique by which longitudinal or transverse cuts are introduced on a surface to increase skid resistance and to reduce the number of wet-weather crashes. The grooves increase skid resistance by improving the drainage characteristics of the pavement and by providing a rougher pavement surface. Several studies show that grooved pavements reduce wet-weather crashes. However, some potential adverse effects should be considered before this strategy is implemented, including the potential of increased noise pollution, accelerated wearing of pavements, and negative effects on steering.
This strategy is related to Strategy 15.2 A7 in this guide and Strategy 15.1 A7 in the ROR guide. Those strategies focus on improving skid resistance by means of changing the pavement aggregates, placing overlays, or adding texture to the pavement surface. Strategy 15.2 A8 focuses strictly on providing grooved pavement. While pavement grooving is a way to add texture to the pavement surface, its primary objective is to improve the drainage and to mitigate hydroplaning. The grooves decrease the water film thickness on a pavement surface and allow for greater tire-pavement surface interaction during adverse weather conditions. Because pavement grooving is such a unique approach to increasing the skid resistance of a pavement, it is treated separately. The section immediately following this one presents results of studies that evaluated the safety effectiveness of pavement grooving. That is followed by a section that presents attributes unique to pavement grooving that should be considered before this type of treatment is implemented.
Safety Effectiveness of Pavement Grooving
Numerous studies on the safety effectiveness of pavement grooving have been conducted, but none of these studied have controlled for regression to the mean so the results should be considered with caution. Wong (1990) performed a before-after evaluation of the effectiveness of pavement grooving based upon data from one site in California. The site was a two-lane highway with steep vertical grades and sharp horizontal curves. Based upon accident data from a 3-year before period and a 1-year after period, Wong found a 72-percent reduction in wet-pavement accidents, while only finding a reduction of about 7 percent in dry-pavement accidents. Wong concluded that pavement grooving was effective in reducing wet-pavement accidents.
Zipkes (1976) analyzed the frequency of accidents and the percentage of accidents on wet and dry pavement surfaces during a 7-year period to evaluate the effect of pavement grooving. Accident data were obtained for a 44-km (27-mi) section of highway near Geneva, Switzerland. Transverse grooves were cut into the pavement with varying groove distances over a 2-km (1.2-mi) section of highway. Grooving of the polished road surfaces reduced the hazard of accidents when drainage conditions were unfavorable. Zipkes indicated that the advantage of grooving is the reduction of water-film thickness, which leads to better contact between the tire and the road surface for the transmission of forces.
Smith and Elliott (1975) evaluated the safety effectiveness of grooving 518 lane-km (322 lane-mi) of freeways in Los Angeles, while 1,200 lane-km (750 lane-mi) of ungrooved pavement were used as a control. The analysis was conducted using 2 years of before data and 2 years of after data. Only fatal and injury accidents were included in the evaluation. Smith and Elliott found that pavement grooving resulted in a 69-percent reduction of wet-pavement accident rates. Sideswipe and hit object accidents were reduced to the largest extent. Pavement grooving did not change the dry-pavement accident rates.
Mosher (1968) synthesized results from studies conducted by state highway departments on the effects of pavement grooving. Information for the report was obtained from 17 states, including Colorado, Florida, Georgia, Idaho, Illinois, Indiana, Louisiana, Minnesota, Missouri, Nebraska, New York, Ohio, Pennsylvania, Texas, Utah, Wisconsin, and Wyoming. Some sections of highway had longitudinal grooves, while other sections had transverse grooving. Pavement grooving proved very effective, reducing crashes by 30 to 62 percent.
Farnsworth (1968) evaluated the effects of pavement grooving on five sections of California highways. Farnsworth measured the coefficients of friction before grooving and after grooving and found that pavement grooving increased the coefficients of friction, changing the friction values from below critical to above critical. Analysis of accident data revealed a reduction in wet-pavement accidents at each of the sites.
NYDOT evaluated the safety effectiveness of pavement grooving based on the installation of grooves at 41 sites. NYDOT found that wet-road accidents were reduced by 55 percent, and total accidents (dry and wet) were reduced by 23 percent. The results were statistically significant at the 95th percentile. Regression to the mean was not addressed in the analysis.
Pavement grooving involves making several shallow cuts of a uniform depth, width, and shape in the surface of the pavement (Mosher, 1968). Grooves may be cut longitudinally along the pavement (parallel to the direction of travel) or in the transverse direction (perpendicular to the direction of travel). Transverse grooving has been used to a lesser extent than longitudinal grooving, partially because most grooving equipment lends itself more readily to placing grooves parallel to the roadway. Grooves cut in the longitudinal direction have proven most effective in increasing directional control of the vehicle, while transverse grooving is most effective where vehicles make frequent stops, such as intersections, crosswalks, and toll booths. When pavements are grooved, it is important that the pavement contain nonpolishing aggregate.
While studies have indicated that pavement grooving reduces wet-pavement accidents, there have been several concerns associated with pavement grooving (Mosher, 1968). One concern has been the effect that pavement grooving has on the durability of various pavement types. For example, one of the most frequently asked questions by engineers in northern climates is, “What will water freezing in the grooves do to the concrete pavement?” In an examination of grooved pavement in Minnesota after one winter, there appeared to be no deterioration in the grooved pavement because of the freeze-thaw cycles. Concern also has been expressed about grooves in asphalt pavement losing their effectiveness because the material can be flexible enough to “flow” back together, particularly during hot weather. This phenomenon has been observed under certain conditions with a fairly new asphalt pavement or with a pavement with low aggregate content. Concern has also been expressed over the loss of effectiveness because of grooved pavements wearing down under hightraffic conditions.
Complaints also have been received that longitudinal grooves adversely affect the steering of certain automobiles and motorcycles. In general, no severe problems have been encountered. This finding is supported by research conducted by Martinez (1977), who studied the effects of pavement grooving on friction, braking, and vehicle control by computer simulation and full-scale testing. Martinez considered automobiles, motorcycles, and automobile and towed-vehicle combinations in his evaluation.
In Iowa, residents living adjacent to I-380 near Cedar Rapids complained that transverse grooving was the cause of an especially annoying tonal characteristic within the traffic noise (Ridnour and Schaaf, 1987). As a result of the complaints, the surface texture of a section of I-380 was modified. The transverse grooving was replaced with longitudinal grooving. A before-after analysis of the traffic noise levels showed that the surface modification lowered overall traffic noise levels by reducing a high-frequency component of the traffic noise spectrum.
Approximately 51 percent (4,977) of the 9,791 fatal crashes that occurred at horizontal curves in 2002 took place during nighttime hours. To a large extent, these crashes may be attributed to reduced visibility at night.
There is evidence to show that providing fixed-source lighting in urban and suburban areas, where there are concentrations of pedestrians and intersectional interferences, reduces nighttime crashes. The need for lighting on streets and highways in rural areas is much less than on streets and highways on urban areas. The need for lighting on rural highways is seldom justified except in critical areas, such as sharp curves (AASHTO, 2001).
Strategy 15.2 A10: Provide Dynamic Curve Warning System (T)
The purpose of this strategy is to reduce the speed of high-speed vehicles on their approach and as they navigate through a horizontal curve. A typical system combines a radar device with a variable message sign. The system measures the speeds of approaching vehicles and provides messages to drivers who are traveling at excessive speeds to slow down to a recommended, or advisory, speed (Exhibit V-11). Dynamic curve warning systems can also incorporate cameras to provide visual surveillance of curves. These systems can be developed using off-the-shelf technology. The main hypotheses regarding this type of strategy are that a dynamic warning device has a much greater effect on high-speed vehicles than a static curve warning sign and that the dynamic system significantly improves the ability of highspeed vehicles to successfully navigate through the curve.
Several dynamic curve warning systems have also been deployed specifically to reduce the likelihood of a truck rollover crash. In 1998, 207 trucks were involved in fatal rollover accidents on the U.S. highway system (Baker et al., 2001). Truck rollover accidents often occur at exit ramps and at tight curves that require a more reduced speed than the normal travel speed on the freeway. Therefore, many of the dynamic curve warning systems designed to reduce rollover crashes have been deployed at freeway exit ramps. Exhibit V-12 illustrates such a system.
It is common practice to widen the traveled way on horizontal curves to make operating conditions on curves comparable to those on tangents. As noted in the AASHTO (2001) policy, widening the traveled way on horizontal curves is necessary for one of two reasons, either (1) the design vehicle occupies a greater width in negotiating the curve because of offtracking or (2) drivers experience difficulty in steering their vehicles along the center of the lane. Roadway widening, however, can entail more than just widening the travel lanes. It can include widening the shoulders, providing shoulders where none previously existed, providing a buffer zone in the middle of the roadway, or various combinations of the above. By widening the traveled way, drivers have more space within the lane to maneuver their vehicles through the curve, allowing more room for driver error without serious consequences By widening the shoulders or providing a shoulder where one previously did not exist, drivers have more recovery area to regain control of their errant vehicles before encroaching on the roadside.
The section immediately following this provides a brief summary of the safety effectiveness of widening the roadway as presented in the ROR guide.
Summary of Effectiveness of Roadway Widening
Two strategies in the ROR guide pertain to roadway widening: Strategies 15.1 A5 (improved highway geometry for horizontal curves) and 15.1 A8 (apply shoulder treatments). These may be referenced for further details.
Strategy 15.1 A5, on improving highway geometry for horizontal curves, provides accident reduction factors for widening lanes and/or shoulders on horizontal curves. Widening a lane may reduce accidents by 5 to 21 percent. Widening a paved shoulder may reduce accidents by 4 to 33 percent, and widening unpaved shoulders may reduce accidents by 3 to 29 percent.
Strategy 15.1 A8, on applying shoulder treatments, provides one set of accident modification factors for widening a paved shoulder on a two-lane rural highway and a second set of accident modification factors for various shoulder types and widths. The accident modification factors were developed by a panel of experts charged with developing prediction models on the expected safety performance of rural two-lane highways. The accident modification factors for widening a paved shoulder width vary as a function of shoulder width and average daily traffic. The base case used is a 1.8-m (6-ft) paved shoulder, and the accident modification factors range from 0.87 to 1.50. Regarding the accident modification factors for shoulder type, the base case is a paved shoulder. Depending on the shoulder width, the accident modification factors range from 1.00 to 1.03 for gravel shoulders, from 1.00 to 1.07 for composite shoulders, and from 1.00 to 1.14 for turf shoulders.
Finally, there is concern that widening the roadway may increase operating speeds. Because speed is such a critical factor related to safety at horizontal curves, roadway widening may worsen safety.
Superelevation is one of the key geometric elements of curve design. Designers select a superelevation rate consistent with the design speed, the selected curve radius, and their jurisdiction’s policy for maximum superelevation. Superelevation works with friction between the tires and pavement to counteract the forces on the vehicle associated with cornering.
Many curves may have inadequate superelevation because of vehicles traveling at higher speeds than were originally designed for, because of loss of effective superelevation after resurfacing, or because of changes in design policy after the curve was originally constructed. For whatever reason, curves with inadequate superelevation may pose safety problems, particularly if the actual superelevation is less than the optimal superelevation as recommended by AASHTO policy (AASHTO, 2001).
Accident prediction models indicate that inadequate superelevation increases curve accidents (Zegeer et al., 1992). There is no evidence, however, that safety is adversely affected along a curve where the actual superelevation is greater than that recommended by AASHTO policy. Therefore, research results indicate that safety can be enhanced when the superelevation is improved or restored along curves where the actual superelevation is less than the optimal superelevation. The following section presents the safety effectiveness of improving or restoring superelevation along curves. The discussion is then concluded with a section that presents other issues relevant to this strategy.
Safety Effectiveness of Improving or Restoring Superelevation
Improving the superelevation of a curve can reduce curve accidents where there is a superelevation deficiency (Zegeer et al., 1991). Superelevation deficiency is the numerical difference between the optimal superelevation (as determined from AASHTO policy) and the actual superelevation of a given curve. Based on estimates from Zegeer et al. (1991), an improvement of 0.01 to 0.019 in superelevation (e.g., increasing superelevation from 0.04 to 0.05 to meet AASHTO policy) would be expected to yield an accident reduction of 5 percent. An improvement of 0.02 or greater in superelevation would be expected to yield an accident reduction of 10 percent.
In 2000, an expert panel used the Zegeer work to develop accident modification factors (AMFs) for the superelevation of a horizontal curve on two-lane highways (Harwood et al., 2000). The following relationships were developed based on the expert panel’s judgement:
These relationships indicate that there is no effect on safety until the superelevation deficiency reaches 0.01, which is consistent with the Zegeer work.
The second issue is proper transition from the normal cross slope along the tangent to the fully superelevated cross slope along the curve. For reasons of safety and comfort, the rotation of the pavement should be effected over a length that is sufficient to make such rotation imperceptible to drivers. Normal practice, in the absence of spiral transition curves, is to begin rotating the pavement along the tangent section before the curve and not attain full rotation until into the curve. A portion of the superelevation runoff is typically located on the tangent, in advance of the point of curvature, to minimize peak lateral accelerations and side friction demand. The proportion of the runoff length placed on the tangent varies from 60 to 80 percent, with many agencies using the 2/3–1/3 rule (placing 2/3, or 67 percent, of the runoff length on the tangent). Bonneson (2000) re-evaluated the approach to horizontal curve superelevation/transition design and determined that placing 70 percent of the superelevation runoff on the tangent and 30 percent on the curve was optimal.
Finally, care should be given to provide proper drainage when improving or restoring the superelevation along a curve. The combination of the control line longitudinal profile and superelevation can produce unintended flat spots along the roadway if care is not taken during design of the transition.
This strategy is the longest-term, highest-cost alternative considered for improving the safety of a horizontal curve because it usually involves total reconstruction of the roadway. It may also require acquisition of additional right-of-way and an environmental review.
There are several ways in which the horizontal alignment of a roadway may be modified to improve safety, including
These modification approaches are addressed below.
Safety Effectiveness of Increasing the Radius of a Horizontal Curve
Increasing the radius of a horizontal curve can be very effective in improving the safety performance of the curve. This strategy is also covered under Strategy 15.1 A5 (improved highway geometry for horizontal curves) in the ROR guide. The ROR crash guide presents a table on the percent reduction of total crashes on a two-lane rural highway that would be expected after flattening a curve. The table is based on research conducted by Zegeer et al. (1992) and shows that increasing the radius of curvature can reduce total curve-related crashes by up to 80 percent. An expert panel used the work of Zegeer et al. to develop an accident modification factor for horizontal curvature on rural two-lane highways (see Equation 4). The accident modification factor is a function of the length of the curve, the radius of the curve, and the presence or absence of a spiral.
Safety Effectiveness of Providing Spiral Transition Curves
A spiral transition curve is a horizontal curve with a continuously changing radius that connects a tangent and a circular curve or two circular curves of different radii. A spiral provides a smooth transition between a tangent section and an adjacent circular curve. The smoother transition from tangent to curve and curve to tangent results in the lateral force increasing and decreasing gradually as a vehicle enters and departs the curve. This is intended to minimize encroachment on adjacent traffic lanes and to promote uniformity in speed (AASHTO, 2001).
Research has shown that drivers on unspiraled curves track a path radius substantially sharper than the designed radius. This is primarily because it is not possible for drivers to instantaneously change their path radius from tangent to curve, and, once on the curve, they must “overcorrect” to stay on the roadway. Providing spirals affords the driver the means of tracking a curve that fits the designed alignment.
Council (1998) and Zegeer et al. (1992) reported that spiral transition curves are effective in reducing crashes. The findings are based on studies of spiraled versus unspiraled curves in one state.
An expert panel that developed the accident modification factor for horizontal curvature on rural two-lane highways (Equation 4) judged that there is sufficient evidence to conclude that the presence of spiral transitions on horizontal curves improves safety. The negative sign associated with the spiral variable effectively reduces the value of the accident modification factor, indicating that the presence of a spiral transition at a horizontal curve improves the safety of the curve. An example is presented below that reveals in more detail the effects of curve length, curve radius, and spiral transitions on safety performance.
Safety Effectiveness of Eliminating Compound Curves
Compound circular curves are sometimes advantageous in providing desirable shapes of curves. However, although no quantitative comparisons have been made between the safety at simple circular curves and the safety at compound curves, agencies should be cautious of using compound curves, particularly if the radius of the first curve is significantly greater than the radius of the following curve. An abrupt change in alignment requires considerable steering effort by motorists to travel safely through the successive curves.
If an agency permits the use of compound curves, the designs should meet AASHTO (2001) guidelines. AASHTO recommends that for compound curves on open highways, the ratio of the flatter radius to the sharper radius should not exceed 1.5:1, and on ramps the ratio may be greater, possibly as great at 2:1. Wherever practical, however, smaller differences in radii should be used.
In general, whether the curve is isolated between two long tangents or is located along a stretch of curvilinear roadway, the horizontal alignment should be designed to meet a driver’s expectation. When an alignment fails to meet a driver’s expectation, the alignment should be modified accordingly. Speed profile models may be used to evaluate the conformance of a highway’s geometry with driver expectancy. Fitzpatrick et al. (2000b) have developed a speed profile model that may be used to evaluate the design consistency of two-lane rural roads. This model was developed for use in the Interactive Highway Safety Design Model (IHSDM) (http://www.fhwa.dot.gov/ihsdm/index.htm).
Automated anti-icing systems are a potentially effective tool for keeping roadway surfaces clear of ice and safe for travel in areas of the country with severe winter climates. Anti-icing involves pretreatment of the roadway surface with chemicals before a winter storm arrives, as opposed to deicing, which involves treatment of the roadway surface during or after the storm, when ice may already have formed. The most common automated anti-icing system now in use is one that uses a series of spray-nozzles connected to a chemical storage tank. Using a pump system, a liquid anti-icing agent is distributed along a roadway segment using the nozzles that can be embedded into the pavement or placed along the edges of the road. The system can either be fully automated, relying wholly on sensors in the area to determine the need of an application of anti-icing chemicals, or semi-automated, where the system can be engaged by someone from a remote location in response to a sensor indication. A potential advantage is that automated anti-icing system can be engaged immediately when appropriate conditions are detected. Existing approaches for anti-icing involve a delay until staff can be called out and equipment deployed. Furthermore, truck-based systems treat locations in sequence according to a routing or priority scheme; automated systems can be activated at all appropriate locations simultaneously.
While not yet in widespread use, automated anti-icing systems are beginning to be used in place of exiting anti-icing systems that involved motorized vehicles traveling along roadway lanes while distributing anti-icing agents. Seventeen states have installed anti-icing systems at selected locations. The most prevalent locations for current automated anti-icing systems are on bridges and overpasses where deck surfaces are prone to ice formations sooner than adjoining road sections. One system in California is currently being planned at a location on a horizontal curve. (See Appendix 4 for more detailed information on this system in California.) Horizontal curves are logical locations for anti-icing treatment because tirepavement friction is critical to vehicle control at horizontal curves.
Many crashes each year result from the vehicle loss of control while traveling on icy roads. Anti-icing systems that have been installed have shown benefit/cost ratios in the range from 1.8 to 3.4 and reductions in the frequency of wintertime accidents from 25 to 100 percent (Friar and Decker, 1999; Barrett and Pigman, 2001; Khattak and Pesti, 2003). However, despite the positive safety evaluations, this strategy is listed as “tried” primarily because this technology has not been in use for a long period of time, nor have the studies specifically addressed installations along horizontal curves.
Strategy 15.2 A15: Prohibit/Restrict Trucks with Very Long Semitrailers on Roads with Horizontal Curves that Cannot Accommodate Truck Offtracking (T)
Longer trucks, particularly single-trailer combination trucks with longer semitrailers, may have difficulty negotiating sharp horizontal curves because they may encroach on an adjacent lane or shoulder because of vehicle offtracking. Vehicle offtracking is the phenomenon in which the rear axles of a truck do not follow the same path as the front axle; at lower speeds, the rear axles of the truck typically track to the inside of the front axle path.
The magnitude of vehicle offtracking is the amount of radial distance displaced from the center of the first axle to the center of the rear axle as a vehicle is making a turn (see Exhibit V-17). Offtracking is a primary determinant of the amount of space a truck or other large vehicle occupies when executing a turning movement; this space, known as the swept path width, is the maximum width of the envelope defined by the front outside corner and the rear inside corner of the truck as it turns (Exhibit V-18). The maximum offtracking and swept path width generally occurs when a vehicle is turning at a low speed. Increasing vehicle speed gradually brings the rear of the truck back toward the path defined by the front axles of the truck. At very high speeds, the rear of the truck can actually offtrack toward the outside of the turn. However, low-speed offtracking is a consideration in the design of any roadway or intersection because traffic or environmental conditions will normally require low-speed travel sometimes on any roadway. For a more complete discussion of vehicle offtracking considerations in roadway design, see NCHRP Report 505 (Harwood et al., 2003).
The greatest concern in vehicle offtracking relevant to horizontal curves is the operation of tractor-semitrailer combination trucks with long semitrailers on roadways with sharp horizontal curvature. The distance from the point of connection of the trailer with the tractor (kingpin) to its rear axle is a critical criterion. A number of states restrict the kingpin-to-rearaxle distance for tractor-semitrailer combinations with semitrailers over 14.6 m (48 ft) in length; on many semitrailers, the rear axles of the trailer can be moved forward to comply with kingpin-to-rear-axle distance limitations. In states where longer kingpin-to-rear-axle distances
for trucks are permitted, it may be desirable to prohibit or restrict trucks with kingpin-to-rearaxle distances that exceed a specified threshold from operating on specific facilities.
California has an active program of identifying roadways with horizontal curves that cannot accommodate trucks with longer kingpin-to-rear-axle distances and establishing appropriate truck advisory restrictions on particular roads (Caltrans, 1989). This program has been in place for nearly 20 years. California establishes these restrictions based on the distance from the kingpin to the center of the rearmost axle of the trailer, known as the kingpin-to-center-of-rearaxle (KCRA) distance. Most other states base their restrictions on the distance from the kingpin to the center of the rear tandem axle set, known as the kingpin-to-rear-tandem (KCRT) distance. The KCRT distance is generally approximately 0.6 m (2 ft) shorter than the KCRA distance.
Along 5,414 km (3,364 mi) or 22 percent of California’s 24,400-km (15,166-mi) state highway system, the California Department of Transportation has established advisory restrictions for trucks with KCRA distances that exceed 12 m (40 ft). This portion of the highway system was selected for restriction based on an analysis of vehicle offtracking on the horizontal curves’ geometrics actually present on those roadways. These restrictions were not based on established accident patterns on the roadways in question, but rather on the potential for collisions when large trucks encroach on adjacent lanes or shoulders.
In California, advisory restrictions on truck use of particular facilities are implemented by signing on the roads in question and maps for truckers published by the California Department of Transportation. An example of a sign used to inform truckers of the restriction is illustrated in Exhibit V-19.
Similar restrictions have been established for motorcoach operators in California, and system maps intended specifically for motorcoach operators are also published.
Combined Discussion of Strategies
Objective 15.2 A focuses on helping drivers stay within the limits of the roadway while negotiating a curve. By contrast, Objective 15.2 B focuses on reducing the severity of curverelated crashes that occur outside of the roadway (i.e., on the roadside).
Despite the countermeasures (i.e., strategies) implemented in pursuit of Objective 15.2 A, some vehicles will still leave the roadway and stray onto the roadside. The strategies for Objective 15.2 B are intended to minimize the consequences to vehicles that travel beyond the shoulder and onto the roadside at a horizontal curve.
Five strategies are designed to reduce the consequences of leaving the roadway:
These strategies are discussed in the ROR crash guide under the section entitled “Combined Strategy: Improving Roadsides,” and the reader is referred to that section for more detailed information on this set of strategies. Although these strategies are fully discussed in the ROR crash guide, it is important to at least identify them here because ROR crashes are so prevalent along horizontal curves. The reader is also referred to the Roadside Design Guide (AASHTO, 2002) for current information and operating practices relating to roadside safety.
Vehicles that encroach upon the roadside while traversing a curve can face conditions different than if they were traveling along a tangent alignment. This issue was addressed under Strategy 15.2 A4 (Install Shoulder Rumble Strips) and Strategy 15.2 A5 (Install Centerline Rumble Strips). As a vehicle leaves the roadway, the departure angle is affected by the degree of road curvature. The concern is for the potentially more extreme condition that can occur when a vehicle leaves the road on the outside of the curve. If a vehicle drifts to the right at a constant angle along a tangent section, the vehicle can have a shallower departure angle than where the vehicle leaves the roadway on the outside of a horizontal curve. The larger departure angle is because of the curvature of the roadway. This is demonstrated, among other things, in Exhibit V-21.
This greater departure angle at horizontal curves has implications on the travel distance and time to roadside objects. Given that an object is located a certain distance from the roadway and that the vehicle follows a certain trajectory, a driver has less distance and time to regain control of his/her vehicle before striking a roadside object located at a horizontal curve than a roadside object located on a tangent.
Another way of looking at it is in terms of the change in the effectiveness of a clear zone (on the outside of the curve) of constant width. Exhibit V-21 shows that not only because of the increased angle of departure, but also because of the curvature of the clear zone adjacent to the curved segment of the road, the distance traveled by an errant vehicle, before reaching the outer edge of the clear zone, will be smaller than if the road were on a tangent alignment. Thus, the probability of an errant vehicle going beyond a constant clear zone is greater along the outside of a curve than along a tangent segment of road. This implies that roadside objects should be further removed from the roadway along the outside of horizontal curves compared with roadside objects located along tangent sections of highway.
There are several assumptions inherent in the analysis, but these are similar to the assumptions currently used in roadside safety analyses. It should also be noted that the opposite conclusion is true for encroachments on the inside of curved roads. However, clear-zone requirements are based on more than just a vehicle that drifts off the road.
The guides for addressing collisions with trees in hazardous locations and utility poles provide supplemental information on reducing the harm done by collisions after vehicles have left the roadway. Strategy 15.2 B2 in this guide directly relates to Strategy 16.1 B1 in the tree guide. Similarly, Strategy 15.2 B3 in this guide and Strategy 16.1 B4 in the tree guide are directly related. The reader is directed to these additional sections for more specific information on the effects of collisions with trees. For Strategy 15.2 B2, if the reader is seeking specific information on utility poles, then the utility pole guide should be consulted.
Agent, K. R., and T. Creasey. “Delineation of Horizontal Curves. Interim Report.” UKTRP-86-4. University of Kentucky. March 1986.
American Association of State Highway and Transportation Officials (AASHTO). A Policy on Geometric Design of Highways and Streets. 2001.
American Association of State Highway and Transportation Officials (AASHTO). Guide for the Development of Bicycle Facilities. 1999.
American Association of State Highway and Transportation Officials (AASHTO). Roadside Design Guide. 2002.
Bali, S., R. J. A. Potts, J. I. Fee, J. Taylor, J. Glennon. Cost-Effectiveness and Safety of Alternative Roadway Delineation Treatments for Rural Two-Lane Highways: Volume I- Executive Summary. FHWA/RD-78/50. FHWA, U.S. Department of Transportation. April 1978.
Baker, D., R. Bushman, and C. Berthelot. “Effectiveness of Truck Rollover Warning Systems.” Transportation Research Record 1779, Transportation Research Board. 2001.
Barrett, M. L., and J. G. Pigman. “Evaluation of Automated Bridge Deck Anti-Icing System,” KTC-01-26/KH36-97-1F, Kentucky Transportation Cabinet, December 2001.
Bonneson, J. A. “Kinematic Approach to Horizontal Curve Transition Design.” Transportation Research Record 1737, Transportation Research Board. 2000.
Bucko, T. R., and A. Khorashadi. Evaluation of Milled-In Rumble Strips, Rolled-In Rumble Strips and Audible Edge Stripe. Office of Transportation Safety and Research, California Department of Transportation. April 2001.
California Department of Transportation (Caltrans). Truck Kingpin-to-Rear-Axle Length State Highway System Evaluation. December 1989.
Chowdhury, M. A., D. L. Warren, H. Bissell, and S. Taori. “Are the Criteria for Setting Advisory Speeds on Curves Still Relevant?” ITE Journal. Vol. 68, No. 2. February 1998.
Council, F. M. “Safety Benefits of Spiral Transitions on Horizontal Curves on Two-Lane Rural Roads.” Transportation Research Record 1635, Transportation Research Board. 1998.
Elefteriadou, L., M. El-Gindy, D. Torbic, P. Garvey, A. Homan, Z. Jiang, B. Pecheux, and R. Tallon. Bicycle-Tolerable Shoulder Rumble Strips. PTI 2K15. Pennsylvania State University. March 2000.
Elefteriadou, L., D. Torbic, M. El-Gindy, S. Stoffels, and M. Adolini. Rumble Strips for Roads with Narrow or Non-Existent Shoulders. PTI 2002-11. Pennsylvania State University. October 2001.
Fambro, D. B., K. Fitzpatrick, and R. Koppa. “Determination of Stopping Sight Distances.” NCHRP Report 400, Transportation Research Board. National Research Council. 1997.
Farnsworth, E. E. “Pavement Grooving on Highways.” Presented at the Pavement Grooving and Traction Studies conference. November 1968.
Fatal Accident Reporting System (FARS). National Highway Traffic Safety Administration, National Center for Statistics and Analysis. http://www-fars.nhtsa.dot.gov/
Federal Highway Administration (FHWA). Technical Advisory for Roadway Shoulder Rumble Strips. 2001. http://safety.fhwa.dot.gov/fourthlevel/pro_res_rumble.library.htm#Papers.
Fitzpatrick, K., K. Balke, D. W. Harwood, and I. B. Anderson. “Accident Mitigation Guide for Congested Rural Two-Lane Highways.” NCHRP Report 440, Transportation Research Board. 2000a.
Fitzpatrick, K., L. Elefteriadou, D. W. Harwood, J. M. Collins, J. McFadden, I. B. Anderson, R. A. Krammes, N. Irizarry, K. D. Parma, K. M. Bauer, and K. Pasetti. Speed Prediction for Two-Lane Rural Highways. FHWA-RD-99-171. Federal Highway Administration, U.S. Department of Transportation. 2000b.
Friar, S., and R. Decker. “Evaluation of a Fixed Anti-Icing Spray System.” Transportation Research Record 1672, Transportation Research Board. 1999. (pp. 34–41).
Gattis, J. L., and J. Duncan. “Geometric Design for Adequate Operational Preview of Road Ahead.” Transportation Research Record 1500, Transportation Research Board. 1995.
General Estimates System (GES). National Highway Traffic Safety Administration, National Center for Statistics and Analysis, 1999. http://www-nrd.nhtsa.dot.gov/departments/nrd-30/ncsa/ges.html
Glennon, J. C., T. R. Neuman, and J. E. Leisch. Safety and Operational Considerations for Design of Rural Highway Curves. FHWA/RD-86/035. Federal Highway Administration, U.S. Department of Transportation. 1985.
Glennon, J. C., and G. D. Weaver. “Highway Curve Design for Safe Vehicle Operations.” Highway Research Record 390, Highway Research Board. 1972.
Hall, J. W. Innovative Treatments for Run-Off-the-Road Accidents. FHWA-NMSHTD-91-02. New Mexico Highway & Transportation Department. 1991.
Hammer, Jr., C. G. Evaluation of Minor Improvements: Part 6, Signs. California Division of Highways, Traffic Department. May 1968.
Hanscom, F. R. “Evaluation of Signing to Warn of Wet Weather Skidding Hazard.” Transportation Research Record 600, Transportation Research Board. 1976.
Harwood, D. W., F. M. Council, E. Hauer, W. E. Hughes, and A. Vogt. Prediction of the Expected Safety Performance of Rural Two-Lane Highways. FHWA/RD-99/207. Federal Highway Administration, U.S. Department of Transportation. 2000. http://www.tfhrc.gov/safety/99207.htm
Harwood, D. W., and J. M. Mason. “Horizontal Curve Design for Passenger Cars and Trucks.” Transportation Research Record 1445, Transportation Research Board. 1994.
Harwood, D. W., D. J. Torbic, Karen R. Richard, William D. Glauz, and L. Elefteriadou, “Review of Truck Characteristics as Factors in Roadway Design.” NCHRP Report 505, Transportation Research Board. 2003.
Humphreys, J. B., and J. A. Parham. The Elimination or Mitigation of Hazards Associated with Pavement Edge Drop-Offs During Roadway Resurfacing. AAA Foundation for Traffic Safety. Washington, D.C. 1994.
Institute of Transportation Engineers. Traffic Control Devices Handbook (TCDH). Washington, DC. 2001.
Jennings, B. E., and M. J. Demetsky. “Evaluation of Curve Delineation Sign.” Transportation Research Record 1010, Transportation Research Board. 1985.
Khattak, A., and G. Pesti. “Bridge Prioritization for Installation of Automatic Anti-Icing Systems in Nebraska,” Mid-Continent Transportation Research Symposium, August 2003.
Krammes, R. A., and K. D. Tyer. “Post-Mounted Delineators and Raised Pavement Markers: Their Effect on Vehicle Operations at Horizontal Curves on Two-Lane Rural Highways.” Transportation Research Record 1324, Transportation Research Board. 1991.
Leisch, J. E. Traffic Control & Roadway Elements-Their Relationship to Highway Safety/Revised. Chapter 12 Alinement. Highway Users Federation for Safety and Mobility. 1971.
Lyles, R. W. An Evaluation of Warning and Regulatory Signs for Curves on Rural Roads. FHWA/RD-80/009. FWHA, U.S. Department of Transportation. March 1980.
Mahoney, K. M., R. J. Porter, E. T. Donnell, D. Lee, and M. T. Pietrucha. Evaluation of Centerline Rumble Strips on Lateral Vehicle Placement and Speed on Two-Lane Highways. The Pennsylvania Transportation Institute, PTI 2003-20, March 2003.
Martinez, J. E. “Effects of Pavement Grooving on Friction, Braking, and Vehicle Control.” Transportation Research Record 633, Transportation Research Board. 1977.
McDonald, T. J., E. Kannel, M. M. O’Brien, and V. Root. Traffic Control Strategies in Work Zones with Edge Drop-Offs. CTRE Project 97-15, Center for Transportation Research and Education, Iowa State University, August 2002. http://www.ctre.iastate.edu/research/detail.cfm?projectID=255
McGee, H., and R. Strickland. “An Automatic Warning System to Prevent Truck Rollover on Curved Ramps.” Public Roads. Vol. 57, No. 4, 1994.
Mosher, L. G. “Results from Studies of Highway Grooving and Texturing by Several State Highway Departments.” Presented at the Pavement Grooving and Traction Studies conference. November 1968.
O’Hanlon, J. F., and G. R. Kelley. A Psychophysiological Evaluation of Devices for Preventing Lane Drift and Run-Off-Road Accidents. Technical Report 1736-F. California Department of Transportation, Division of Highways. September 1974.
Outcalt, W. Bicycle-Friendly Rumble Strips. CDOT-DTD-R-2001-4. Colorado Department of Transportation, Research Branch. 2001a.
Outcalt, W. Centerline Rumble Strips. CDOT-DTD-R-2001-4. Colorado Department of Transportation, Research Branch. 2001b.
Perrillo, K. The Effectiveness and Use of Continuous Shoulder Rumble Strips. FHWA, Albany, New York. August 1998.
Preston, H., and T. Schoenecker. Potential Safety Effects of Dynamic Signing at Rural Horizontal Curves. MN/RC-2000-14. Minnesota Department of Transportation. December 1999. (http://www.lrrb.gen.mn.us/PDF/200014.pdf)
Retting, R. A., and C. M. Farmer. “Use of Pavement Markings to Reduce Excessive Traffic Speeds on Hazardous Curves.” ITE Journal. Vol. 68, No. 9. September 1998.
Ridnour, R., and D. V. Schaaf. Effects of Pavement Surface Texture on Noise and Frictional Characteristics. Iowa Department of Transportation. February 1987.
Ritchie, M. L. “Choice of Speed in Driving Through Curves as a Function of Advisory Speed and Curve Signs.” Human Factors. Vol. 14, No. 6. December 1972.
Rys, M. J., E. R. Russell, and T. S. Brin. “Evaluation of Milled Centerline Rumble Strip Patterns.” Transportation Quarterly. Vol. 57, No. 4, Fall 2003.
Science Applications International Corporation (SAIC). Synthesis of Shoulder Rumble Strip Practices and Policies. 2001. http://safety.fhwa.dot.gov/fourthlevel/pro_res_rumble.library.htm#Papers.
Smith, R. N., and L. E. Elliott. Evaluation of Minor Improvements (Part 8), Grooved Pavement Supplement Report. CADOTTR2152-11-75-01. California Department of Transportation. September 1975.
Stowe, R., “A Benefit/Cost Analysis of Intelligent Transportation System Applications for Winter Maintenance,” Paper No. 01-0158, Transportation Research Board, January 2001.
Strickland, R., and H. McGee. Evaluation of Prototype Automatic Truck Rollover Warning Systems. FHWA-RD-97-124. FHWA, U.S. Department of Transportation. December 1996. (http://ntl.bts.gov/data/124.pdf)
Tribbett, L., P. McGowen, and J. Mounce, An Evaluation of Dynamic Curve Warning Systems in the Sacramento River Canyon, Final Report, Western Transportation Institute, prepared for California Department of Transportation, April 2000.
U.S. Department of Transportation (USDOT). Manual on Uniform Traffic Control Devices (MUTCD). FWHA, U.S. Department of Transportation. 2003. http://mutcd.fhwa.dot.gov/kno-2003.htm
Wong, S. Y. “Effectiveness of Pavement Grooving in Accident Reduction.” ITE Journal. Vol. 60, No. 7. July 1990.
Zador, P., H. S. Stein, P. Wright, and J. Hall. “Effects of Chevrons, Post-Mounted Delineators, and Raised Pavement Markers on Driver Behavior at Roadway Curves.” Transportation Research Record 1114, Transportation Research Board. 1987.
Zegeer, C. V., D. Reinfurt, T. Neuman, R. Stewart, and F. Council. Safety Improvements on Horizontal Curves for Two-Lane Rural Roads—Informational Guide. FHWA/RD-90/074. Federal Highway Administration, U.S. Department of Transportation. October 1991.
Zegeer, C. V., J. R. Stewart, F. M. Council, D. W. Reinfurt, and E. Hamilton. “Safety Effects of Geometric Improvements on Horizontal Curves.” Transportation Research Record 1356, Transportation Research Board. 1992.
Zipkes, E. “The Influence of Grooving of Road Pavements on Accident Frequency.” Transportation Research Record 623, Transportation Research Board. 1976.
Zwahlen, H. T. Warning Signs and Advisory Speed Signs-Reevaluation of Practice. Final Technical Report. FHWA/OH-84/003. FHWA, U.S. Department of Transportation. June 1983.