Skip to contentUnited States Department of Transportation - Federal Highway Administration FHWA Home
Research Home
Public Roads
Featuring developments in Federal highway policies, programs, and research and technology.
This magazine is an archived publication and may contain dated technical, contact, and link information.
Federal Highway Administration > Publications > Public Roads > Vol. 74 · No. 4 > Twisting Roads Still Spell Trouble

January/February 2011
Vol. 74 · No. 4

Publication Number: FHWA-HRT-11-002

Twisting Roads Still Spell Trouble

by Roya Amjadi and Kimberly Eccles

FHWA and States have joined forces on research to improve safety at curves, where crash fatality rates remain comparatively high.

Fatality rates on curves are increasing, but an ongoing FHWA pooled fund study aims to reveal how rumble stripes, signs, road markings, and other treatments can improve safety on rural roads like this one.
Fatality rates on curves are increasing, but an ongoing FHWA pooled fund study aims to reveal how rumble stripes, signs, road markings, and other treatments can improve safety on rural roads like this one.

According to the National Highway Traffic Safety Administration's Fatality Analysis Reporting System (FARS), an average of 41,515 people lost their lives on U.S. highways each year during the period between 2004 and 2008. A loss of life of this magnitude is clearly a cause for action. Although resources to address these crashes are limited, transportation professionals can effectively tackle the problem by focusing on high-crash locations, systematically determining the major contributing factors using available crash data and roadway inventory files, and identifying potential safety improvements.

However, there are limited data on the effectiveness of some types of improvements, particularly more innovative treatments. These data limitations make it difficult to estimate the reduction in crashes that might result from implementing a given safety improvement. As a result, decisions about which safety improvements to make are challenging, and sometimes less-than-optimal decisions are made.

To improve this decisionmaking, in 2005 the Federal Highway Administration (FHWA) launched the Evaluations of Low Cost Safety Improvements Pooled Fund Study (ELCSI-PFS). The study is one of the largest ongoing FHWA safety-related pooled fund studies, with a total of 28 States contributing. The purpose of the study is to develop reliable estimates of the effectiveness of those safety improvements identified in the National Cooperative Highway Research Program (NCHRP) Report 500: Guidance for Implementation of the AASHTO [American Association of State Highway and Transportation Officials] Strategic Highway Safety Plan.

As part of the study, researchers looked at safety improvements that were already implemented at crash sites and conducted retrospective evaluations to determine their effectiveness. They also are looking at safety improvements that were developed by the participating States, as well as improvements derived from research conducted in the driving simulator at the Turner-Fairbank Highway Research Center (TFHRC) in McLean, VA.

Researchers are evaluating these improvements by Empirical Bayes statistical methodology, using before-and-after crash data for evaluations. FHWA anticipates that more than 30 safety improvements will be evaluated over 10 years of the study.

Motivated by the fact that on average 27 percent (11,071) of total fatalities occur on horizontal curves, the participating State departments of transportation (DOTs) selected horizontal curves as a priority for safety improvements. NCHRP Report 500, Volume 7: A Guide for Reducing Collisions on Horizontal Curves identifies 13 strategies that have been tried to (1) reduce the likelihood of a vehicle leaving its lane and either crossing the centerline or leaving the roadway and (2) minimize the adverse consequences of vehicles leaving the roadway at horizontal curves. For this study, FHWA and the States are evaluating a selection of these safety improvements for horizontal curves, including curve delineation signage, centerline and edgeline pavement markings, centerline and edgeline rumble strips or stripes, friction treatments for pavement surfaces, as well as other new and innovative strategies. The study is considering how these strategies can be applied both individually and in selected combinations to achieve varying degrees of safety effectiveness. Below is a discussion of how some of the participating States are structuring their evaluations of these safety improvements.

Curve Delineation

One low-cost strategy to enhance safety is improved curve delineation. This study aims to increase drivers' awareness as they approach or navigate through curves by providing more conspicuous signing and more effective lane markings. Improving delineation can be particularly helpful for motorists at night and/or in adverse weather conditions. One option for improving delineation is the use of pavement markings with greater durability, all-weather functionality, and higher retroreflectivity. Other options include post-mounted delineators, chevrons, raised pavement markers, and wider edgelines.

By day, this curve might appear easy enough to navigate, but darkness makes executing the turn much more dangerous. FHWA is helping States find the best options for improving night visibility and safety in general.
By day, this curve might appear easy enough to navigate, but darkness makes executing the turn much more dangerous. FHWA is helping States find the best options for improving night visibility and safety in general.

To determine the safety effectiveness of these changes in curve delineation, TFHRC obtained geometric, traffic, and crash data for 89 curves in Connecticut and 139 curves in Washington State that had these treatments. All sites were on two-lane rural roads, but specific treatments varied by site, including new chevrons, horizontal arrows, advance warning signs, improvement of existing signs, and installation of new signs using fluorescent yellow sheeting. Daily traffic volumes at the sites in Connecticut ranged from about 1,000 to 20,000 vehicles. In Washington, the traffic volumes ranged from about 250 to 15,000 vehicles per day.

The Connecticut Department of Transportation (ConnDOT) used fluorescent yellow sheeting to improve signing at horizontal curves. This included installing new signs or replacing existing ones. Included were warning signs (for example, curve ahead or suggested speed limit) and curve delineation signs (chevrons or horizontal arrows).

ConnDOT identified candidate locations for treatment through a regular safety program called the Suggested List of Surveillance Study Sites. The list uses crash data, traffic volumes, and roadway characteristics to identify intersections and road segments with higher than expected crash rates. District engineers reviewed the locations to identify those on horizontal curves. The engineers then treated the curves with improved signing. ConnDOT received positive feedback from enforcement officials who asked to have similar improvements at other curves.

Graph. This graph depicts the growth in the number of fatal horizontal curve crashes as a share of total U.S. highway fatalities for the years 2004–2008. The years are plotted on the x-axis, while the percentages are plotted on y-axis. Yearly data points are shown as dots, with a green line connecting them. For 2004, the data point is marked 25.4 percent. The line rises to the right to a dot marked 26.1 percent for 2005; rises to the right to a dot marked 26.9 percent for 2006; rises to the right to a dot marked 27.9 percent for 2007; and curves down to the dot marked 27.1 percent for 2008. There is a yellow trendline across with the positive slop of .52 starting at 25.12 percent at 2004.
Despite an overall downward trend in highway fatalities across the United States, the trend of fatalities that occur at horizontal curves is increasing (as shown here), indicating that curves remain treacherous territory

Unlike Connecticut, the treatments in Washington involved the State DOT installing only chevrons (W1-8 signs) on horizontal curves, without any other delineation or signing. This treatment included installing chevrons for the first time at some locations and adding additional chevrons to sites that already had a few.

Using the empirical Bayes method, TFHRC investigated the effects of the changes in delineation through aggregate and disaggregate crash analyses (that is, comparing crashes at each site and at groups of sites before and after the strategies were installed). The aggregate analysis revealed an overall 18 percent reduction in injury and fatal crashes, along with a 27.5 percent reduction in total crashes and a 25 percent drop in lane-departure crashes in dark conditions. All of the results were statistically significant at the 95 percent confidence level. Total crashes on curves and lane-departure crashes on curves (that is, crashes involving a vehicle leaving its lane) decreased by about 9 percent, but this change was not statistically significant.

A more detailed review revealed that crash reductions were greater at locations with higher traffic volumes, sharper curves (radii less than 492 feet, or 150 meters), and with more hazardous roadsides (roadside hazard rating of 5 or higher, on a scale of 1 to 7, with 7 being the most hazardous). The analysis also indicated greater crash reductions at locations with higher levels of improvement; that is, on curves where the DOTs either added more signs or substituted signs with more retroreflective material.

TFHRC conducted an economic analysis to determine the relative cost effectiveness of curve delineation improvements. Excluding labor, the unit cost of the fluorescent yellow signs used in Connecticut ranged from $30 to $160 depending on the size of the signs. Smaller signs such as chevrons and advisory speed signs were cheaper than larger signs such as curve warning signs. The unit cost of the chevrons in Washington was about $100.

Figure. This figure shows eight warning signs from the MUTCD. The signs are arranged into two rows. In the top row, the sign at left is labeled W1-1 and is a yellow diamond-shaped sign featuring a black arrow indicating a sharp 90-degree curve to the left. The next sign, labeled W1-2, is a yellow diamond-shaped sign with a black arrow bending in a slight angle to the left. The next sign, W1-3, is a yellow diamond-shaped sign featuring a black arrow that starts out pointing up, then makes a sharp 90-degree curve to the left, and then points up again. Next is W1-4, a yellow diamond-shaped sign featuring a black arrow that starts out pointing up, then makes a slight curve diagonally to the left, and then continues up again. The first sign in the second row, W1-5, is a yellow diamond-shaped sign featuring a black arrow that starts out pointing up, then curves to the right and then to the left and back to the right, and then points up again, indicating a series of curves ahead. Next is W1-10, a yellow diamond-shaped sign featuring a black arrow that starts out point up and gently curves to the right; a small black line intersects the arrow head near the bend, indicating a side road intersecting the primary road. The next sign to the right, W1-6, is a yellow horizontal rectangle-shaped sign featuring a large black arrow simply pointing left. The final sign in the figure, W1-8, is a yellow vertical rectangle-shaped sign featuring a black chevron pointing left.
Connecticut installed these types of signs, as shown here from FHWA's Manual on Uniform Traffic Control Devices, to improve safety on horizontal curves.

The researchers then computed the average annualized costs of the signs assuming 5-year service lives, and compared them to the estimated annual benefits. They determined that improving curve delineation with signing upgrades is very cost effective, with a benefit-cost ratio exceeding 8:1.

Curve Multistrategy Safety Improvements

Although the pooled fund study is providing some quantitative measures of the safety benefits of enhanced delineation on curves, many other treatments also are promising. The NCHRP Report 500 series lists treatments and strategies that were experimental and/or had limited use. One critical research need is to evaluate these new options as well as other treatments that emerge. Each strategy needs to be evaluated alone and in combination with others to determine its effectiveness in various situations.

Some previous evaluations during this pooled fund study were retrospective, in that they looked at strategies that had been in place for years. In 2008, however, the study took a different approach to overcome various sample size constraints. Between 2008 and 2010, six States completed installation of safety strategies at various horizontal curve sites for build-to-evaluate or prospective evaluations. The States -- Florida, Iowa, Kansas, Kentucky, Missouri, and Virginia -- and TFHRC researchers selected the sites and the most appropriate (in terms of constructability within cost constraints) safety strategies by analyzing statewide crash data. The strategies included surface friction treatments on curves (two- and four-lane roads), surface friction treatments on ramps, in-lane pavement markings for curve warnings, larger and brighter chevrons, and edgeline rumble stripes on curves.

Improving Curve Delineation Through Signing

Crash Type Reduction in Crashes (Percentage Estimate) Standard Error
Injury and Fatal Curve Crashes 18.0% 8.6
Nighttime Curve Crashes 27.5% 7.3
Nighttime Lane Departure Crashes on Curves 25.4% 7.8

Source: Safety Evaluation of Improved Curve Delineation (FHWA-HRT-09-045).

The States are installing single treatments at some sites and combinations of treatments at others. In-lane pavement markings and surface friction treatments are two notable experimental strategies: only a handful of agencies nationwide have used these treatments for this type of application. FHWA facilitates communication among the participating States so they can share information on their progress and learn from each other's efforts. In 2013, after 3 years of data collection, TFHRC and the States will evaluate the data and identify any effects of the treatments, comparing data from 3 years before and 3 years after completion of the safety improvements. A summary of the efforts by four of those States follows.


For some time, the Iowa Department of Transportation's (Iowa DOT) highway safety improvement program has focused on low-cost but effective safety improvements, including strategies to reduce fatal crashes on curves. According to Tom Welch, a retired State transportation safety engineer for Iowa DOT, a 2004 crash analysis found that 11 percent of all primary highway fatal curve crashes occurred on just 30 curves in Iowa. On those curves, Iowa DOT installed low-cost improvements that included paved shoulders with rumble stripes and larger, brighter chevrons. The agency also removed fixed objects outside the curves. State and county engineers are making low-cost systematic improvements at other curves and accelerating improvements at high-crash curves.

For TFHRC's six-State pooled fund study, Welch says he expects there will be a sufficient sample of each of five strategies, and combinations of them, to conduct a scientific evaluation of their safety effects by 2013. "It is the hope that the results of these evaluations, if positive, will encourage other agencies to implement these strategies," he says.


The Kansas Department of Transportation (KDOT) is in the process of selecting curve sites for its part in the pooled fund study. The agency's analysis of rural, two-lane, undivided, nonintersection, nonanimal crashes revealed that more than 20 percent occurred at horizontal curves. KDOT mapped any curves with more than three of this crash type in a 5-year period. Knowing the number of crashes, annual average daily traffic, radius, and length of each curve, KDOT estimated safety performance functions (SPF) and ranked the top 140 curves. "So now we have a strong list of candidate locations," says KDOT Safety Engineer Steven Buckley. About 30 of the identified curves appear in Kansas's "5 percent report" to FHWA, which describes the State's roadway sites with the most pressing safety concerns.

Summary of Economic Analysis Results

State Sites Years Total Crash Reduction (per year) Total Crash Reduction (per site-year) Crash Savings (per site-year) Range of Benefit-Cost Ratios
All 228 4.58 11.373 0.050 $4,347 12.7 to 67.9
Connecticut 89 3.05 6.686 0.075 $6,546 19.1 to 102.3
Washington 139 6.76 4.687 0.034 $2,938 8.6 to 45.9

Source: Safety Evaluation of Improved Curve Delineation (FHWA-HRT-09-045).

"Many strategies can't or shouldn't be applied systematically -- some are cost prohibitive while others are rendered inef-fective if overused," Buckley says. "For these reasons, the question becomes which type of strategy to apply where. This question is best answered with real-world research -- thus our participation in the build-to-evaluate phase of this pooled fund study."


The Kentucky Transportation Cabinet's (KYTC) contribution to the pooled fund study will include friction treatments, in-lane pavement markings, improved curve warning signs, and edgeline rumble stripes. Many of the candidate locations will be selected from the State's Roadway Departure Safety Implementation Plan (RD Plan), which was developed in cooperation with FHWA and identifies sites with roadway departure crash histories that could be addressed by proven low-cost countermeasures.

KYTC developed a list of the 10 curved ramps that had the highest number of single-vehicle crashes over a recent 3-year period. According to Scott Pedigo, a KYTC safety engineer, 258 of the 286 crashes (about 90 percent) occurred on wet pavement. For the pooled fund study, the agency plans to install a high-friction treatment at several of these ramps. The RD Plan also lists more than 200 other locations on State rural roads where surface friction treatments should be considered.

A yellow left curve and chevron signs in New Britain, CT, mark the way for drivers, which can be especially important in low-light or wet weather conditions, such as shown here.
A yellow left curve and chevron signs in New Britain, CT, mark the way for drivers, which can be especially important in low-light or wet weather conditions, such as shown here.

Further, Kentucky plans to deploy enhanced signs and markings at approximately 23 curves selected because of their crash histories. Enhanced signing likely will include larger signs fabricated with diamond-grade, fluorescent yellow sheeting. The State expects implementation costs to be no higher than $5,000 per curve.

In addition, Kentucky is in the second year of a pilot effort to install edgeline rumble stripes on two-lane roadways as they are being resurfaced. In the first year, the department installed approximately 47 miles (76 kilometers) of edgeline rumble stripes and planned to install more than 150 miles (240 kilometers) of edgeline rumble stripes in 2010. These sections will be included in the pooled fund study's analysis of edgeline rumble stripes on curves. If these pilot studies are successful, edgeline rumble stripes could be installed on nearly 1,000 miles (1,600 kilometers) of Kentucky's rural roadways in the future, at hazardous locations identified in the RD Plan.

Photo. Shown here is a rural, two-lane road with a nonpaved shoulder (as indicated by a tape measure placed between the edgeline stripe and the pavement edge) and no signage indicating the horizontal curve ahead.
Photo. Shown here is a rural, two-lane road with safety enhancements, including the addition of a paved shoulder with an edgeline rumble stripe and several chervron signs placed along the road’s curve.
On this two-lane rural road in Stoddard County (Top), MoDOT added a paved shoulder, installed edgeline rumble stripes, and placed chevron signs along this curve (Bottom), improving safety for drivers.


The Missouri Department of Transportation (MoDOT) also is paying particular attention to lane-departure crashes. "This crash type is one that MoDOT believes can be reduced in a variety of ways, but especially through engineering improvements,"says MoDOT Traffic Safety Engineer John P. Miller. So far, the primary engineering safety improvement has been use of rumble stripes on improved shoulders and chevrons in curves. Generally MoDOT has installed rumble stripes on improved shoulders on Missouri's most traveled roadways, but it also uses this approach, along with chevrons, on high-risk rural roads as a curve improvement.

"In Missouri, a large percentage of run-off-road crashes occurs in curves," says Miller. "Using strategies like rumble stripes on improved shoulders for a system of routes will help reduce severe crashes on these roads overall."

Miller notes that a recently completed project on a high-risk rural road involved both chevron installation and rumble stripes on improved shoulders. The project was fairly inexpensive, about $55,000, and involved a series of curve improvements on the route.

Missouri is just now beginning to see the benefit of its overall lane-departure strategy, as fatalities from this kind of crash on the most traveled roadways went down 41 percent between 2005 and 2009. Because this strategy is successful, Miller says, MoDOT can continue to implement it on more roadways and especially on curves with chevron installations. And, as with the other States, these findings will support the TFHRC pooled fund study.

Researchers at TFHRC use this Highway Driving Simulator to evaluate potential low-cost safety improvements that could reduce crashes at horizontal curves on rural roads.
Researchers at TFHRC use this Highway Driving Simulator to evaluate potential low-cost safety improvements that could reduce crashes at horizontal curves on rural roads.

Improving Nighttime Visibility of Curves

Another aspect of the pooled fund study is to evaluate innovative treatments to improve the nighttime visibility of curves. One component of the study involved experiments using TFHRC's driving simulator. Simulation is beneficial because it can provide objective measures of safety effectiveness without having to physically build the safety improvements at many locations.

Researchers used the simulator to evaluate two sets of low-cost safety improvements for rural areas. They programmed in the software a set of improvements to enhance the visibility of curves on rural, two-lane, undivided roads at night. The simulated improvements were (1) edgelines and (2) standard reflectorized as well as "streaming" post-mounted delineators. The latter consist of light emitting diodes (LEDs) mounted on posts and programmed with time delays such that the individual LEDs simulate a stream of light traveling around a curve at night, highlighting the curve for motorists.

The results of the experiment indicated that edgelines offered a small potential safety benefit. However, combining standard reflectorized post-mounted delineators with edgelines offered a somewhat greater benefit. This result does not imply that edgelines are not needed; edgelines provide continuous delineation of travel lanes, especially at close range. Of all the treatments explored, the streaming post-mounted delineators with edgelines offered the most promising potential safety benefit. However, no applications exist for streaming post-mounted delineator on two-lane rural roads, nor are the cost implications of developing such a system known. Therefore, the TFHRC researchers recommend further study including economic analysis and field validation of this potential safety improvement.

Looking to the Future

TFHRC researchers also are looking at the potential safety improvements associated with combining center--line and edgeline rumble stripes on curves with no shoulders or narrow ones. The research is focused on two-lane rural roads in Missouri and Pennsylvania with shoulders less than 4 feet (1.2 meters) wide. All installations are to be completed in 2012.

Also, in 2011, TFHRC will begin an evaluation of the role of pavement performance on curves nationwide. The goal is to identify the extent and severity of a pavement's contribution to curve crashes. The researchers will look at pavement type, age, skid number, and condition. These findings will lead to the identification of applicable low-cost safety countermeasures where pavements are identified as contributors to crashes at curves. This research will be based on the findings of an in-house study involving the FHWA Office of Safety Research & Development and the Highway Safety Information System (HSIS.)

Overcoming Data Challenges

A major barrier to improving safety on horizontal curves is the lack of inventory information about exactly where curves are located and their characteristics. Some agencies have partially overcome this challenge by (1) using crash data to identify crash clusters or (2) using sign inventories to identify locations that are signed with chevrons. Other approaches include (3) visually identifying curves by using geographic information system data or other maps; (4) working with district, county, or other local agency staff to locate curves from memory and local knowledge; and (5) working with enforcement staff to identify hazardous curves based on field observations. Although each of these methods has known weaknesses, in the absence of a comprehensive curve inventory, they do provide some ability to make targeted improvements.

Compiling a horizontal curve inventory would not only make targeted improvements possible in response to crash problems, but it also would make systematic improvements feasible. For example, a curve inventory that contains the location and some characteristics (such as degree of curvature, length of curve, and length of spiral) could help DOTs identify and program systematic improvements at horizontal curves. The agencies could apply strategies used only at some curves to those with similar characteristics. In addition to chevrons and advance curve signs, a DOT could apply lane curve warnings at sharper curves. DOTs also could use inventory and crash data analysis to systematically upgrade all curves in a given jurisdiction.

"If we are going to successfully reduce fatalities by half within 20 years, we will need to learn to do more with less," says KDOT's Buckley. "We're engineers. We're engineered to build things. But why rebuild one curve and address maybe 10 crashes when we can re-sign, mark, or surface 10 curves for the same price and address maybe 100 crashes? That, at its core, is the value of low-cost safety improvements and related research."

By day, this curve might appear easy enough to navigate, but darkness makes executing the turn much more dangerous. FHWA is helping States find the best options for improving night visibility and safety in general.
Here, the TFHRC simulator depicts a two-lane rural road at night with curve delineation and post-mounted delineators on both sides of the road.

Roya Amjadi is a research civil engineer (highway) at TFHRC and has more than 20 years' experience in transportation engineering. She manages the Evaluation of Low Cost Safety Improvements Pooled Fund Study and Rural Safety Innovation Program. She has a B.S. in mechanical engineering from the University of Iowa and an M.S. in civil engineering transportation safety from Cleveland State University. Currently, she is working toward a master's in statistics at George Mason University.

Kimberly Eccles, P.E., is a senior engineer with Vanasse Hangen Brustlin, Inc. (VHB). She has more than 12 years of experience in transportation safety, specializing in safety evaluations, and is the principal investigator for the pooled fund study. She manages the VHB satellite office in Raleigh, NC. Eccles has a B.S. in civil engineering from Michigan State University and an M.S. in civil engineering from North Carolina State University.

For more information, contact Roya Amjadi at 202-493-3383 or, or Kimberly Eccles at 919-834-3972 or, or visit

United States Department of Transportation - Federal Highway Administration