Gaining Traction on Roadway Safety
In a performance study, 17 States collaborated with FHWA to determine how to increase pavement friction through low- or no-cost surface treatments. Check out the results.
Variables such as weather, roadway geometry, visibility, surface conditions (wet or dry), human factors, and other issues complicate attempts to quantify the safety story for any particular roadway. One issue that is fairly well understood is the link between pavement friction and roadway departure crashes. The probability of crashes caused by skidding on wet pavement is reduced when the friction between vehicle tires and pavement is high.
The Federal Highway Administration (FHWA) and National Transportation Safety Board (NTSB) estimate that up to 70 percent of crashes on wet pavement can be prevented or minimized by increasing pavement friction. Although roadway designers cannot control a driver’s response to road conditions, they can reduce the probability of skid-related crashes by adjusting the properties of pavement surfaces.
In 2011, FHWA and 17 volunteer States pioneered a study, Evaluations of Low-Cost Safety Improvements Pooled Fund Study, Phase VI, on using crash data to measure the safety performance of pavement treatments. The researchers developed crash modification factors and benefit-to-cost ratios for nine pavement treatments, such as thin hot-mix asphalt overlays and diamond grinding, to examine their potential in reducing the frequency and severity of lane departure crashes.
In particular, the researchers had initiated the study to shed more light on the effect of pavement surfaces on safety and to determine whether various low- or no-cost surface treatments make a difference in safety performance on wet pavements. The members of the 2011 pooled fund study’s technical advisory committee--27 State safety representatives--voted for pavement safety as a priority concern. To learn more about pavement influence on run-off-road crashes, the members expressed a need for research on lower cost pavement strategies to address crashes due to wet pavement. Also, this study was conducted to build on knowledge gained from older studies, all seeking a better understanding of the effects that various common pavement treatments can have on roadway safety.
“This performance study promises to open a new chapter in pavement design,” says FHWA Associate Administrator Michael Trentacoste, head of the Turner-Fairbank Highway Research Center.
Various participating States--in particular, Pennsylvania, North Carolina, and California--have expressed their satisfaction on providing locations and data for the study and their interest in the results.
“Pennsylvania currently is in the process of implementing the use of the Highway Safety Manual into our design processes, and the development of crash modification factors for these treatments would prove to be very useful when developing the State’s approach to pavement maintenance,” says Christopher B. Speese, manager of safety engineering and risk management, Pennsylvania Department of Transportation (PennDOT). “We hope that the findings from the study show that high friction surface treatment is an effective tool for States to utilize in reducing wet pavement crashes on curves.”
Shawn A. Troy, safety evaluation engineer, North Carolina Department of Transportation (NCDOT), adds: “NCDOT is continuously striving to learn, study, analyze, develop, and implement crash modification factors for safety countermeasures that we have installed across our State. The results of this study will be used as preliminary guidelines for us to begin to understand the potential benefits of these various low-cost pavement countermeasures.”
California also weighs in. “I am very pleased to see that high friction surface treatments clearly provide a substantial safety benefit, as we are aggressively using this treatment strategy up and down the State,” says Robert Peterson, branch chief, Highway Safety and Operational Improvement Program, California Department of Transportation. “At this time, we have over 130 locations where HFST [high friction surface treatment] will or has been placed.”
Surfaces and Safety
Pavement surfaces influence two factors related to roadway safety. First, the frictional properties of pavements affect the resistance of tires to sliding across the surface. Pavement friction helps to keep vehicles on the road when brakes are applied, particularly when the wheels lock up, when a vehicle is navigating curves, or when a driver is steering aggressively. These factors become especially important in wet weather, when a thin film of water on the surface reduces contact between the tire and the pavement.
A second critical factor is the ability of the pavement surface to channel water away from beneath the tire. The texture and porosity of a pavement surface help to provide a path to reduce the potential for hydroplaning. Texture and porosity also affect the splash and spray potential from other vehicles, which can significantly affect visibility in wet weather.
The Study’s Backstory
In 1997, a number of organizations met with experts from driver, vehicle, and highway fields to develop a strategic plan for highway safety. Together with the safety experts, the Standing Committee on Highway Traffic Safety of the American Association of State Highway and Transportation Officials (AASHTO), with assistance from FHWA, the National Highway Traffic Safety Administration (NHTSA), and the Transportation Research Board’s (TRB) Committee on Transportation Safety Management, developed a list of 23 key areas that affect highway safety.
In 2003, the National Cooperative Highway Research Program (NCHRP) published the first in the NCHRP Report 500 series of guides to advance implementation of countermeasures, targeting those 23 emphasis areas. Each guide includes an introduction to the problem, a list of objectives for improving safety in that emphasis area, and relevant countermeasures. Each countermeasure is designated as proven, tried, or experimental. Many of the strategies have not been evaluated, and so the majority of them are considered to be in the tried or experimental categories, not proven. Tried countermeasures are those that have been implemented in a number of locations and might even be accepted as standard approaches, but for which valid evaluations have not been found. Experimental countermeasures are those that have been suggested and that at least one agency has considered sufficiently promising to try on a small scale.
With respect to pavement surfaces used as safety countermeasures, NCHRP Report 500 Volume 6: A Guide for Addressing Run-Off-Road Collisions identifies Strategy 15.1 A7 “Provide Skid-Resistant Pavement Surfaces” as a proven countermeasure for reducing run-off-road crashes. Volume 7: A Guide for Reducing Collisions on Horizontal Curves, identifies Strategy 15.2 A7 “Provide Skid-Resistant Pavement Surfaces” as a key strategy for reducing crashes at horizontal curves. The application of skid-resistant pavement surfaces for horizontal curves, however, was not validated as a safety countermeasure.
The NCHRP publication recognizes that, as of 2003, only limited research had been done on site-specific treatments. But both volumes 6 and 7 recognize that the effectiveness of high-friction treatments diminishes over time; therefore a State using this strategy must conduct a dynamic program to target the appropriate sites for new treatments and to maintain the safety benefits from existing treatments.
The Pooled Fund Study Takes Shape
In 2005, FHWA organized a multiphased pooled fund study, Evaluations of Low-Cost Safety Improvements, which at that time involved 24 volunteer States, to evaluate low-cost safety strategies identified by the NCHRP Report 500 guides. The intent of this pooled fund study is to provide reliable quantitative estimates of safety effectiveness for countermeasures in reducing crashes. The pooled fund study is occurring in several phases to be completed in 2017. Crash modification factor and benefit-to-cost economic analyses are being developed for each of the targeted safety strategies identified as priorities by the participating States.
The products will be published in AASHTO’s Highway Safety Manual and FHWA’s Crash Modification Factors Clearinghouse (www.cmfclearinghouse.org). Availability of crash modification factors and benefit-to-cost economic analyses can help State departments of transportation (DOTs) use objective measures to select strategies for improving safety. As of 2014, the Evaluations of Low-Cost Safety Improvements Pooled Fund Study has 38 State members.
CMFs and B/C Factors
A crash modification factor (CMF) quantifies the effectiveness of a safety strategy in reducing or increasing the number of certain types of crashes and their associated severity. A CMF might be a number less than one, equal to one, or greater than one. If it is less than one, it indicates that the strategy is effective in reducing the number of crashes and/or their severity. If the factor is equal to one, there is no improvement in safety, and when the factor is greater than one, the strategy increases the number of crashes and/or their associated severity.
More than one CMF may be developed for any single safety strategy. State DOTs can estimate the reduction in crashes made possible by installing a proven safety strategy at a selected site by multiplying the site’s annual number of crashes by the strategy’s CMF.
A benefit-to-cost (B/C) factor can provide a tool for decisionmakers who are using cost effectiveness when comparing more than one safety strategy, or when the strategy is to be installed on a large scale. The B/C analysis considers the cost of each treatment (typically just the installation cost) and the benefit, quantified in terms of crash reduction and lifespan. The annual benefit (that is, the crash savings) is the product of the total crash reduction per year and the aggregate cost of a crash (all severities combined). The B/C ratio is calculated as the ratio of the annual benefit to the annual cost. For any strategy to be considered beneficial, the B/C ratio must be larger than 1.00.
The Phase VI Focus
The specific goal of the Evaluations of Low-Cost Safety Improvements Pooled Fund Study, Phase VI, was to analyze crash data to determine the effectiveness of various low-cost pavement treatments on improving roadway safety. For the most part, the researchers used the Empirical Bayes before-and-after analysis methodology. The Empirical Bayes method is a set of procedures for developing statistical inferences in which data are used to estimate crash distributions.
The researchers analyzed treatments for both flexible and rigid pavements and developed CMFs for each type of improvement. They developed ratios for those treatments that yielded statistically significant overall crash reductions. This phase of the pooled fund study also sought to identify any potential differences in safety performance for various types of pavement treatments, as this aspect had not been carefully examined under previous research.
In effect, this work was a retrospective study of the effects of pavement treatments on roadway safety. In other words, the researchers did not install pavement safety treatment strategies as part of this phase of the study. Rather, the volunteer States provided the project team with information on the locations of previous pavement treatments, along with crash and roadway data from before and after the treatments that were applied prior to this phase.
Pavement Treatments Evaluated by the Phase VI Study
|Flexible Pavement Treatment Strategies||Concrete Pavement Treatment Strategies|
|Thin Hot-Mix Asphalt (HMA) Overlay||Thin HMA Overlay|
|Open Graded Friction Course (OGFC)||OGFC|
|Ultrathin Bonded Wearing Course (UTBWC)||UTBWC|
|High Friction Surface Treatment (HFST)||HFST|
|Chip Seal (single/double/triple)||Diamond Grinding|
|Slurry Seal||Grooving/Next Generation Concrete Surface|
The project team analyzed the crash data for a minimum of 3 years before the treatment had been applied and a minimum of 3 years following the application. Three years of post-installation crash data, however, were not always available for one of the newer types of treatments, known as high friction surface treatments (HFSTs).
“HFST is not a commonly used surface treatment, but FHWA’s Every Day Counts 2 and HFST teams are responsible for advancing its rapid deployment,” says Joseph Cheung, civil engineer, Office of Safety, FHWA. “The teams provide agencies, contractors, consultants, and industry with the knowledge and skills to apply this recommended countermeasure on horizontal curves, intersection approaches, grades, and other spot locations, as identified, in order to decrease crashes, serious injuries, and fatalities.”
Data collection was the most time-consuming aspect of the study. To help maximize the number of treatment sites available for analysis, the project team started by contacting each volunteer State to determine its most commonly used treatment. The team also asked about the availability of information on that treatment (specifically, site locations, crash data, and roadway information).
The project team developed a “wish list” of data that it hoped to collect for each treatment site. Unfortunately, not all of this information was available for every location. Of note, although the researchers sought data on friction measurement for each treatment site, they found that few States collect that data to the extent necessary for usefulness in this study: namely regular (annual) friction values for each year of the “before” period and each year of the “after” period. The researchers also found that weather data was difficult to include in the analysis, due to a lack of detailed and sufficient data at each treatment site.
The pavement treatments considered in the final analysis included thin hot-mix asphalt (HMA) overlay and the others listed above. These treatments are commonly used by States throughout the country. Some, such as open graded friction courses (OGFC), however, are not commonly used in cold-weather climates. An OGFC treatment is a type of thin HMA overlay, but OGFC uses an open graded or porous asphalt mixture that enables water to drain away quickly from the surface by flowing through the mixture itself. This helps to minimize sheeting or standing water on the surface and the potential for vehicles hydroplaning.
Available Mile-Years Before and After Treatment Strategies in Each State
|Treatment||States||Miles||Mile-Years Before||Mile-Years After|
|Chip Seal||CA, MN, NC, PA||2,557||14,030||9,451|
|Microsurfacing||CA, MN, NC, PA||327||1,789||1,097|
|Slurry Seal||CA, NC||139||710||549|
|Thin HMA Overlay||CA, MN, NC, PA||3,940||20,857||17,952|
|UTBWC||CA, NC, PA||132||821||437|
|Diamond Grinding||CA, MN, NC, PA||150||853||495|
|Thin HMA Overlay||MN, PA||12||71||33|
For the conventional (non-HFST) treatments, the researchers analyzed the data from four States: California, Minnesota, North Carolina, and Pennsylvania. The selection was due to the variety of treatment types used in each State, the number of sites (and mileage) for each treatment, and the availability of crash and roadway data.
For HFST, the team analyzed the data from eight States that employ this treatment: Colorado, Kansas, Kentucky, Michigan, Montana, South Carolina, Tennessee, and Wisconsin. The selection of States was due primarily to the number of treatment locations and availability of data for those sites.
Next, the researchers summarized the mile-years before and mile-years after treatment available for the various conventional treatments in the relevant States. The numbers indicate the extent of the various treatments that were analyzed, giving the project team confidence in the adequacy of the sample sizes for the analysis. Mile-years are the number of miles where the strategy was implemented, multiplied by the number of years of data before or after implementation. For example, if a strategy was implemented at a 9-mile (14.5-kilometer) segment and data are available so far for 3 years since implementation, then there is a total of 27 mile-years of after-period data available for the study.
Summary Statistics for HFST Site Data
|Site Type||Sites by State||Sites by Road Classification|
|Ramps||KS – 2
KY – 2
MI – 6
MT – 1
SC – 6
WI – 1
|Urban – 17
Rural – 1
|Curves||CO – 2
KS – 2
KY – 28
MI – 1
MT – 1
SC – 1
TN – 4
|Urban – 4
Rural – 35
Note that many more miles of treatments were available for asphalt pavements than for concrete. The majority of the treatments considered are more commonly applied to asphalt pavements. Of the concrete pavement treatments, diamond grinding is the most common in the four States in the non-HFST list, with a limited assortment of slurry seal, microsurfacing, and bituminous overlay (that is, thin HMA, open graded friction course, and ultrathin bonded wearing course) treatments. This distribution is significant, however, since diamond grinding tends to preserve the concrete pavement surface as concrete, while bituminous treatments effectively cover up the concrete surface permanently.
As mentioned earlier, the analysis methodology employed for this study for the conventional treatments was Empirical Bayes. This methodology is very robust for crash data analysis as it accounts for the so-called “regression to the mean” through the inclusion of reference sites in the analysis. In the regression-to-the-mean phenomenon, variables do not tend toward their extreme values, but rather toward the average value.
Reference sites are those with similar characteristics (such as pavement type, traffic, roadway classification, number of lanes, shoulder type, and median type) to the treatment sites, but they have not received the treatment being studied. By analyzing the crash data from reference sites in conjunction with data from actual treatment sites, the researchers could account for any changes in crash rates due to weather conditions or other changes not related to the application of the treatment.
Effect of Pavement Treatments on Wet and Dry Road Crashes
|Treatment||Wet Road Crashes||Dry Road Crashes|
|Thin HMA Overlay||Two-lane roads
CMF = 1.256
CMF = 1.181
|Multilane and Freeways
CMF = 0.865 (multilane)
CMF = 0.797 (freeway run-off-road)
|All other roads combined|
CMF = 0.685 (freeway)
|Two-lane and multilane roads
CMF = 1.108 (multilane)
CMF = 1.181 (two-lane)
|All other roads combined|
|Chip Seal||All roads
CMF = 0.373 (run-off-road)
CMF = 0.950 (two-lane)
CMF = 1.206 (multilane)
CMF = 0.937 (two-lane)
CMF = 0.862 (two-lane)
CMF = 1.142 (two-lane)
|All other roads combined|
|Slurry Seal||All roads
CMF = 0.802 (two-lane)
CMF = 0.947 (freeway)
CMF = 0.694 (two-lane)
CMF = 0.905
|All other roads combined|
|Diamond Grinding||All roads
CMF = 0.869 (freeway)
CMF = 0.959
|All other roads combined|
|Note: Color-coding indicates whether the treatment resulted in a benefit (orange), disbenefit (yellow), or neither (white).|
For the HFST sites, identifying suitable reference sites for the various treatment locations was difficult, preventing the use of the Empirical Bayes methodology. HFST is typically applied specifically as a safety treatment to problematic locations, sometimes due to an unusual geometric feature of the roadway at that location, making it difficult to find reference sites with similar characteristics.
Therefore, for the HFST sites, the researchers employed two analysis methodologies: the naïve before-after method and the comparison-group method. The naïve methodology simply looks at the number of crashes at the treatment site before and after application of the treatment. Treatment sites are those that received the safety improvements. The other methodology uses a comparison group of untreated sites (in addition to the treated sites) to compensate for external causal factors that could affect the change in the number of crashes. Comparison groups are sites that did not receive safety improvements, but they do have road features and traffic characteristics similar to the treatment sites.
Results of the Analysis
Next, the researchers determined the effects of all of the treatments on wet and dry road crashes by roadway type. They also determined which CMFs are statistically significant according to the Empirical Bayes before-and-after analysis for various treatments. They plotted the results in tables and color-coded the cells to indicate whether the treatment resulted in a benefit (orange), disbenefit (yellow), or neither (white).
Some of the key results for each treatment type include the following:
- Thin HMA Overlay: Benefit reducing wet road crashes on multilane roads and freeways, but a disbenefit for both wet and dry road crashes on two-lane roads.
- OGFC: Benefit reducing wet road crashes on freeways, but a disbenefit for dry road crashes on both two-lane and multilane roads.
- Chip Seal: Significant benefit reducing wet road crashes on multilane roads, and small benefit for wet and dry road crashes on two-lane roads, but a disbenefit for dry road crashes on multilane roads.
- Microsurfacing: Benefit reducing wet road crashes on two-lane and multilane roads, but a disbenefit for dry road crashes on two-lane roads.
- Slurry Seal: Benefit reducing wet road crashes on two-lane roads.
- Ultrathin Bonded Wearing Course: Benefit reducing wet road crashes on freeways and wet and dry road crashes on two-lane roads.
- Diamond Grinding: Benefit reducing wet and dry road crashes on freeways.
CMFs Generated by Naïve and Comparison Group for HFST*
|Methodology||Total Crashes||Wet Road Crashes|
|*The CMFs here were multiplied by a factor of 1.25 to account for a regression-to-the-mean bias.|
The scope of Phase VI did not allow for the study of why some of the above pavement treatments were not effective in reducing crashes on two-lane rural roads. One assumption might be that two-lane rural roads are not maintained as regularly as other roadways that are on a higher priority list for DOTs and other transportation agencies. Less maintenance may impact safety negatively and may have influenced the study results to some extent.
For HFST, the researchers determined the CMFs generated by the naïve and comparison-group analysis sites. These results show a substantial benefit for crash reduction for all types of HFST treatments. These two methods are likely biased toward underestimating the CMFs for lower values, and thereby exaggerating crash reductions, because regression to the mean is likely at play and is not accounted for in these methods. Nevertheless, the potential for crash reduction is significant and has been adjusted by a factor of 1.25 to account for a regression-to-the-mean bias.
The Highway Safety Manual suggested an approximate method for resolving this problem. That report states that “for a large [regression-to-the-mean] bias, where only a few sites with the highest crash frequency were treated out of the total population and few years of before crash data were included in the evaluation study,” the biased CMF should be corrected by multiplying it by a factor of 1.25. This recommendation seemed appropriate for this evaluation, so the researchers applied this correction of 1.25 to the biased CMFs.
Results of Disaggregate Analysis
The properties of pavement treatments change over time. Traffic and weather play a significant role in wearing down the surfaces, generally leading to a reduction in pavement texture and friction. The cause can be a complex interaction of factors, but intuitively it seems that aggregates abrade, polish, and ravel or are broken off the surface; ruts can form, causing formation of a water puddle; asphalt binders can bleed to the surface; and porous surfaces can become clogged. Although there have been few studies to confirm this link between age and safety, the project team sought to evaluate whether any correlation could be found in the data analyzed. Therefore, for some of the pavement treatments, the researchers investigated the possible changes in safety benefits as treatments age.
Statistically Significant CMFs for Wet Road Crashes on
|Estimated CMF (Standard Error) by Period After Treatment|
|All Years||Year 1||Year 2||Year 3||Year 4|
(all two-lane roads)
(all two-lane roads)
The researchers then determined the statistically significant CMFs generated from selected analysis of treatment effects over time. Single-layer chip seals show a noticeable reduction in benefit (increase in CMF) over the first 3 years after installation. This result is not entirely surprising as chip loss (raveling) and bleeding are the two common failure mechanisms that result in reduced surface texture and friction, particularly within the wheel paths. The team did not observe the same trend for double- or triple-layer chip seals, possibly due to the fact that the failure mechanisms may take longer to manifest themselves with these more robust treatments.
The researchers observed no clear trend (benefit or disbenefit) for diamond grinding during the first 4 years after treatment. For OGFC, they observed an increasing benefit for freeways in the first 4 years, while they observed a decrease in benefit for two-lane roads. Insights from this study suggest an apparent relationship between CMFs and annual average daily traffic, and sometimes precipitation, urban versus rural settings, and crash frequencies.
Results From Benefit-Cost Analysis
By considering the cost of each treatment (typically just the installation cost) and the benefit (quantified in terms of crash reduction and lifespan), agencies will be equipped to better justify selection of one treatment over another. The study team based treatment unit costs on ranges published in a recent Strategic Highway Research Program 2 research report on pavement preservation treatments, Preservation Approaches for High-Traffic-Volume Roadways (S2-R26-RR-1).
Summary Results of the B/C Analysis
|Treatment (State)||Road Type||B/C Ratio|
|Chip Seal (All)||Two-Lane||0.69|
|Chip Seal (CA only)||Two-Lane||2.06|
|Diamond Grinding (All)||Freeway||5.95|
|Thin HMA (NC only)||Multilane||3.01|
|OGFC (NC only)||Freeway||9.15|
|Slurry Seal (All)||Two-Lane||2.25|
The results show high B/C ratios for some of the treatments, such as OGFC on freeways, diamond grinding for all types on freeways, ultrathin bonded wearing course for all kinds on two-lane roads, and thin HMA on multilane roads: all but some chip seals have a B/C ratio greater than one.
What Does It All Mean?
This study represents a first-of-its-kind evaluation of the effects of various low- and no-cost pavement treatments on roadway safety. Although the results did not reveal that one particular treatment is substantially better than another, they did show some potential differences in benefits among the treatments, particularly when cost is considered through the benefit-cost analysis. In the end, virtually all of the treatments provided a benefit for wet road crashes on all road types, with the exception of thin HMA overlays on two-lane roads.
Keep in mind that these treatments are not necessarily interchangeable from a pavement standpoint. Different treatments may be better suited for different climates, traffic levels, and underlying pavement types. The intent of this study was not to judge which treatment is best for any particular circumstance. Nor was it in the scope of this study to consider other major contributing factors for crashes, such as light conditions, road alignments, or speed.
This study does provide CMF and B/C factors for agencies to consider in optimizing pavement management and pavement preservation programs for safety considerations.
HFST clearly provides a substantial safety benefit, even when weighed against its higher cost. However, this study does not purport to endorse such a treatment for all circumstances. Use should be focused on applications where the benefits are most clear, on ramps and curves, until subsequent studies can be conducted to evaluate the benefit for other applications.
FHWA’s Offices of Safety Research and Development and Infrastructure Research and Development would like to encourage State DOTs and other transportation agencies to improve the collection and storage of critical crash and surface property data, so that in the future it will be available to improve and compare CMFs under different circumstances and in different climate and aggregate-type regions.
Roya Amjadi is a civil engineer at FHWA’s Turner-Fairbank Highway Research Center (TFHRC) in McLean, VA. She manages the Development of Crash Modification Factors Program and the Evaluations of Low Cost Safety Improvements Pooled Fund Study. She has a bachelor’s degree in mechanical engineering and a master’s degree in civil engineering from Cleveland State University.
David Merritt is a project manager with The Transtec Group of Austin, TX, where one of his specialties is pavement surface characteristics. He received his bachelor’s degree from Northern Arizona University and master’s degree from The University of Texas at Austin, and is a registered professional engineer in Texas.
James Sherwood is a civil (highway research) engineer at TFHRC. He currently monitors projects on pavement friction management and flooded pavements, and performs staff research in the data analysis program for safety performance of pavements. He received his bachelor’s degree from the Massachusetts Institute of Technology.
For more information, see the technical report on this study and a TechBrief (both to be posted August 2014) at www.fhwa.dot.gov/research/tfhrc/projects/safety/comprehensive/elcsi/index.cfm or contact Roya Amjadi at 202–493–3383 or email@example.com, David Merritt at 512–451–6233 or firstname.lastname@example.org, or James Sherwood at 202–493–3150 or email@example.com.