Practices for Preventing Roadway Departures (2018)

Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

12 Survey Questionnaire For this synthesis, a questionnaire was sent to DOTs in all 50 states and the District of Colum- bia. The questionnaire, shown in Appendix A in a condensed form, was organized into five parts. Part I was used to obtain information about the person who responded. Parts II through V contained 34 questions, grouped as follows: • Part II related to how state DOTs identify roadway departure crash problem locations and programs for selecting and implementing engineering countermeasures. • Part III had questions relevant to the states’ use (or non-use) of each of the 20 counter- measures listed in Table 4, plus questions relevant to identifying new countermeasures, states’ evaluations of countermeasures, and their need for additional research. • Part IV had two questions to explore how states were addressing vehicle-based technolo- gies, including vehicle-to-vehicle (V2V), vehicle-to-infrastructure (V2I), and autonomous vehicles, as they relate to preventing roadway departure crashes. • Part V had two questions—one to inquire if the state had a particular countermeasure that representatives wanted to feature as an effective countermeasure and another to provide an opportunity to make any further comments related to the application of countermeasures. Forty-one states responded to the survey questionnaire. A complete tabulation of all the responses by state for each question is provided in a series of tables in Appendix B. For many of the questions, there are hyperlinks to files provided by the state. These files have various content, including policies, specifications, drawings, reports, and other relevant technical material, all of which add to the information library for the roadway departure crash issue. They were reviewed, and relevant information was extracted and included in the discussion that follows. The remainder of this chapter will present the primary findings obtained from the state responses to the questions within Parts II through V. Part II. Roadway Departure Problem Identification and Implementation Programs (Questions 1 Through 4) Part II had four questions related to how states identify roadway departure crash problem locations and programs for selecting and implementing engineering countermeasures. These questions asked: 1. If the state had prepared a roadway departure safety implementation plan, 2. If the state had compiled and analyzed roadway departure crash data, 3. If the state had developed any SPFs for roadway departure, and 4. Which implementation strategy (i.e., hot spot, systematic, or systemic) the state followed. C H A P T E R 3 Survey of the State of the Practice

Survey of the State of the Practice 13 The responses to these questions are shown in Tables B1 through B4 in Appendix B and are summarized in the following sections. Roadway Departure Safety Implementation Plans (Question 1) The FHWA has a safety initiative entitled “Focused Approach to Safety” whereby it provides resources to eligible high-priority states to address the nation’s most critical roadway safety challenges, one of these being roadway departure crashes. Since 2009, as part of this initiative, the FHWA, through a contractor and in conjunction with the states, has developed Roadway Departure Safety Implementation Plans (RDSIPs) for 18 participating states. The state-tailored plans include systemic implementation of low-cost engineering treatments aimed at specific crash sub-types. The plans include recommended roadway departure countermeasures, a set of strategies with deployment levels, and an estimate of the funding needed to achieve a substantial and cost-effective annual reduction in roadway departure fatalities (see https://safety.fhwa.dot. gov/fas/ for further explanation and a sample RDSIP). The first question posed to the states was to determine whether they had prepared an RDSIP; they were asked to provide a link to it if one was available. Twenty-one states responded posi- tively, with 11 providing hyperlinks to their RDSIPs or a similar document (see Table B1 in Appendix B). These states were those from the list of 18 participating states noted previously. This synthesis does not summarize and analyze the contents of these RDSIPs. However, they do serve as a resource of engineering countermeasures that those states felt should be imple- mented to address their data-driven crash problems for roadway departure crashes. For exam- ple, Arkansas’s RDSIP identified the following countermeasures to be deployed to address its roadway departure crash problem: • Enhanced signs and markings, including: – Oversized advance curve warning signs mounted on both left and right, – Chevrons, – Advisory speed plates beneath the advance warning signs, – Additional strategies to reduce high-end approach speeds (e.g., speed feedback signs, periph- eral transverse pavement markings), – Raised thermoplastic markings, and – Wider edge lines. • Centerline, edge-line, and shoulder rumble strips. • Alignment delineation. • Wet-weather treatments, including: – High-friction surfaces, and – Pavement grooving. • Guardrail upgrades (40). Each of the states’ RDSIPs were reviewed to see if any state included a countermeasure not already mentioned; none had. Roadway Departure Crash Data (Question 2) The intent of question 2 was to ascertain whether (and if so, how) state DOTs have compiled and analyzed crash data related to roadway departures. Table B2 in Appendix B shows the responses of each state to this question. Thirty-six of the 41 states responding replied “yes” (meaning that they had compiled roadway departure crash data), and of those, 12 provided a link to their crash data and, if available, analysis and report. While the roadway departure crash analyses of several states could be used as examples, the analysis from Massachusetts is provided in Appendix C.

14 Practices for Preventing Roadway Departures An initial task of analyzing roadway departure crash data is to extract the relevant crashes from the statewide total crash database. This requires defining what constitutes a roadway departure crash and the data elements that capture the relevant crash records. Page 8 from the Massachusetts’s analysis (see Appendix C) shows its lane departure definitions and those from the Fatality Analysis Reporting System maintained by NHTSA. Safety Performance Functions for Roadway Departure Crashes (Question 3) SPFs are statistical models used to estimate the average crash frequency for a specific site type (with specified base conditions) based on traffic volume and roadway segment length. SPFs are developed through statistical regression modeling using historical crash data. They could also be developed for a crash type such as roadway departure crashes. Across the country, SPFs have been developed for a variety of analysis purposes. The predicted number of crashes calculated using SPFs is instrumental for a number of activities in the project development process, includ- ing (1) network screening, (2) countermeasure comparison, and (3) project evaluation. More information about SPFs can be found in a number of resources, including the Highway Safety Manual (2) and the website for the CMF Clearinghouse (3). Ten states responded that they had developed SPFs that could be used for the three purposes stated previously relevant to roadway departure crashes; their responses are shown in Table B3 in Appendix B. However, only four states provided hyperlinks to documents that provide that information. And, upon review of those documents, none of the states had developed an SPF specifically for roadway departure crashes. Programmatic Problem Identification and Implementation Strategies (Question 4) At the national level, the Highway Safety Improvement Program (HSIP) has encouraged a traditional approach of improving roadway safety at specific high-crash locations by identifying and analyzing individual crashes at the locations, defining crash patterns, determining appropri- ate countermeasures to reduce future crash potential, and implementing those countermeasures. This approach is frequently referred to as the hot-spot (meaning high-crash-frequency or crash- rate) approach. Two additional approaches were being used to complement the traditional hot-spot approach: • In the systematic approach, the first step is to identify low-cost countermeasures applicable to certain crash types. Then the crash data system is searched to identify highway sections that have targeted crashes at or above a crash threshold that would ensure cost-effective deploy- ment of these countermeasures. • The systemic approach involves widely implemented improvements (i.e., countermeasures) based on high-risk roadway features (e.g., no or narrow shoulders) correlated with specific severe crash types (e.g., roadway departure). It involves identifying the problem, screening and prioritizing candidate locations, selecting countermeasures, and prioritizing projects. It begins by looking at system-wide crash data to analyze and identify systemic safety problems. It then moves to a micro-level risk assessment of locations across the network, which then leads to selecting relevant countermeasures most appropriate for broad implementation. An example of how one state—Arizona—applied a systemic approach on two-lane rural highways with higher potential for run-off-road crashes is presented in Appendix D. This is an example of developing a performance-based practical design for shoulder width and superelevation, two countermeasures discussed in this report.

Survey of the State of the Practice 15 All 41 states responded that they follow at least two of the approaches, with many following all three; the full results are shown in Table B4 in Appendix B. Six states responded that they also use other approaches; three are noted here: • Sites are identified through network screening and further refined through road safety audits (Alabama). • The United States Road Assessment Program (usRAP) is used as a systemic safety analysis tool to characterize crash risk (Utah—see http://www.usrap.org/ for information about this tool, which is sponsored by the Roadway Safety Foundation). • Safety management plans, which look at segments of roads with hot spots; this approach applies consistent improvements throughout the length of the chosen corridor (Nevada). Part III. Countermeasures Being Used by State Departments of Transportation (Questions 5 Through 24) For questions 5 through 24 (see the survey questionnaire in Appendix A), the states were asked, for each countermeasure shown in Table 4, if they used that countermeasure. If they replied “yes,” they were asked how frequently they used it, with choices of rarely, sometimes, and often. Guidance was not offered as to what level of use (e.g., number of sites or mileage) would apply to the three frequency levels. If they replied “no,” they were asked to provide a reason, with options being: 1. Not aware of the countermeasure, 2. Countermeasure not proven to be effective, 3. Insufficient funding, 4. Negative public feedback, 5. Maintenance concerns, or 6. Any other reason. They were also asked to provide any documents relevant to a policy or guideline, as well as for any comments. The responses to the use of these 20 countermeasures by each of the 41 states are provided in Tables B5 through B24 in Appendix B. In this section, the key findings of the states collectively will be presented for each countermeasure. Each countermeasure will be described, followed by the most recent findings of their safety effectiveness from previous literature, followed by a summary of state practices as identified from the questionnaire responses. Noteworthy comments from individual states regarding the use or non-use of the countermeasures are highlighted as well. Edge-Line Widths Greater Than the Standard 4 In. (Question 5) Description Edge-line pavement marking defines or delineates the edge of the roadway. It provides a visual reference to guide motorists and helps reduce drifting onto the shoulder and roadside area. According to the Manual on Uniform Traffic Control Devices (14), the standard width for a normal line is 4 to 6 in., and a wide line is considered to be at least twice the width of a normal line. Typically, edge lines, especially on non-freeways, are 4 in. wide. Increasing the width provides a better visual perspective and signifies a heightened degree of emphasis. Figures 1 and 2 show the same road section with a 4-in. edge line and an 8-in. edge line, respectively.

16 Practices for Preventing Roadway Departures Literature Findings The most recent and comprehensive research on wider edge lines has given evidence of their safety benefits. In their 2013 report, Carlson et al. (15) cite the following percent reduc- tions in crash types based on data from three states as evidence of the safety benefit of this countermeasure: • Total crashes: 15.0% to 30.1%, • Fatal and injury crashes: 15.4% to 37.7%, • Day crashes: 12.0% to 29.1%, and • Night crashes: –2.4% to 30.7%. State Practices The responses from the 41 states are shown in Table B5 in Appendix B. Thirty-one states stated that they used this countermeasure, and as shown in Table 5, 12 used it often, 10 some- times, and nine rarely, with 10 states indicating that they did not use this countermeasure. (Note: For this table through Table 21, the first column shows the number of states that responded to the question, the last column shows the number of states that indicated they did not use the Figure 1. Roadway with 4-in. edge line (12). Figure 2. Roadway with 8-in. edge line (12). Total States Responding By Frequency of Use Do Not Use Often Sometimes Rarely 41 12 10 9 10 Table 5. States’ responses on use of wider edge lines.

Survey of the State of the Practice 17 particular countermeasure, and the middle three columns show frequency of use. Not all states that responded that they used the countermeasure provided a response as to the frequency of use; hence, for some countermeasures, the sum of the frequency of use and “do not use” will not add to the total number of states responding.) The states that responded that they did not use wider edge-line markings had the following reasons: • Not aware of countermeasure (one state), • Countermeasure not proven to be cost-effective (four states), • Insufficient funding (two states), • Maintenance concern (two states), and • Other (six states). Some of the comments provided for not using this countermeasure were: • “Do not believe wider lines to have the advantage claimed.” • “Wider lines make the shoulder look like a bike lane.” • “Currently prefer to ensure nighttime reflectivity rather than increasing width [of the line].” Advance Curve Warning Pavement Marking (Questions 6, 7, and 8) Description Roadway departure crashes occur most frequently within highway curve sections. One of the low-cost countermeasures being used is to place some type of warning as a pavement marking in advance of the curve. This countermeasure can take different forms. Questions 6, 7, and 8 in the survey (see Appendix A) asked about three types of in-lane pavement markings: 1. Advance curve warning pavement marking, such as illustrated in Figure 3. 2. Speed advisory marking in the lane, such as illustrated in Figure 4, which displays “CURVE-55-MPH.” 3. Speed reduction pavement marking to encourage speed reduction, such as the optical bar illustrated in Figure 5. Figure 3. Example of a curve warning pavement marking (16).

18 Practices for Preventing Roadway Departures Literature Findings The literature review did not uncover any studies that reported on the effectiveness of these measures in terms of changes in crashes, nor are they listed in the CMF Clearinghouse. However, Albin et al. (12) cite the following regarding the effectiveness of speed advisory pavement mark- ings in reducing motorists’ speeds: NCHRP Report 600: Human Factors Guidelines for Road Systems (17) found that the “Curve-55-MPH” text reduced speeds on a rural road by 4 mph citing research by Chrysler and Schrock (18). Another study (19) referenced in NCHRP Report 600 tested the curve arrow with “SLOW” text on a suburban road and found it reduced the percentage of drivers exceeding the speed limit by more than 5 mph during the daytime and late-night timeframes. With regard to the speed reduction marking device (i.e., optical bars), Albin et al. (12) cite two studies (20, 21) that demonstrated the speed reduction benefit, and one study (22) that showed minor changes in speeds that were inconsistent. State Practices Tables B6, B7, and B8 in Appendix B show the state responses for the advance curve warning marking, the speed advisory marking, and the special pavement marking to encourage speed reduction countermeasures, respectively. A summary of the responses from the states for each of these three countermeasures is presented here: • Advance curve warning pavement marking: Only 11 of the 41 states responded that they used this countermeasure, with only two responding with “often,” and the remaining nine responding with “rarely.” The primary reasons for not using this marking were: – Maintenance concerns, specifically the need to re-mark frequently to ensure visibility, and Figure 4. Example of speed advisory marking (12). Figure 5. Example of optical speed bar marking (12).

Survey of the State of the Practice 19 – Concern about vehicles and motorcycles sliding on the markings, particularly if thermo- plastic is used. Three states stated that they felt signs could be more effective, and one state commented that it preferred to use other countermeasures such as doubling-up signing, fluorescent sheet- ing, and sign-mounted flashers. • Speed advisory marking in lane: Only nine of the 41 states stated that they used this measure, and all of these states indicated that it was rarely used. The reasons raised for not using this measure included: – Maintenance requirements, as noted previously, – Concerns about vehicles such as motorcycles sliding on markings, particularly if they are long-life markings such as thermoplastic, and – Preference for using warning and regulatory signing. • Special pavement marking to encourage speed reduction: Twenty-two of the 41 states responded that they were using or had used this type of marking, but all stated that they did so rarely. Comments from four states concerning their experiences appear to confirm the literature findings: – “Pilots [were tried] on a two-lane secondary (shoulder) and four-lane undivided highway. Probably will be replaced with next overlay due to limited success on speed reductions.” – “This is a spot treatment that has been used rarely. I believe the [FHWA] ELCSI Pooled Fund Study determined these to have limited safety benefit, so we do not push for their installation.” – “We utilize the speed reduction markings shown in the MUTCD [Manual on Uniform Traffic Control Devices]. We have installed these at three locations throughout the state with mixed results.” – “We’ve experimented with optical speed bars with hit-and-miss success.” As with many pavement markings subject to heavy traffic, maintaining this device at a high level is a maintenance issue. Some comments shed doubt on the cost-effectiveness of this device [such as “Showed an initial change in driver behavior (speed reduction) but not long term”], and two states responded that they tried it but that it did not seem to be effective. One respondent said the state preferred to use alternate countermeasures such as speed feedback signs. Dynamic Curve Warning System (Question 9) Description Another countermeasure for preventing roadway departure crashes on highway curves is identified generically as a dynamic curve warning system, which can have different forms using supplemental beacons or messages that activate when a motorist approaches the curve at a high speed. Examples of two of these are shown in Figures 6 and 7. Another system that has gained more use recently is the sequential dynamic curve warning system (SDCWS), which is a series of horizontal curve chevron signs with solar-powered flashing lights embed- ded in the signs, as shown in Figure 8. It was this system that was suggested by question 8 in the survey. Literature Findings The purpose of these systems is to reduce vehicle speeds in horizontal curves, and Albin et al. (12) cite several studies that have shown that they do (23, 24, 25). With respect to the SDCWS, Smadi et al. (26) also showed a speed reduction and a simple before–after crash reduction of 7% to 91% at seven locations.

Figure 6. Example of a speed warning actuated sign (12). Figure 7. Example of a dynamic warning system (12). Figure 8. Example of sequential dynamic curve warning system (12).

Survey of the State of the Practice 21 State Practices Table B9 in Appendix B displays the responses from the 41 states that responded, and Table 6 summarizes these. Twenty-five states indicated that they used this countermeasure, but over half of them stated that it was rare that they did so. The reasons the other states gave for not using this countermeasure were: • Maintenance concerns (seven states). • Insufficient funding (three states). • Countermeasure not proven to be effective (two states). Comments from two states were: • “Power source and upkeep of the devices are typically a concern with these treatments, as most districts will shy away from the use of solar devices, as they tend to get stolen.” • “Would want to identify criteria for consideration of this countermeasure; otherwise, they would be requested at any location where there was a serious crash, regardless of the cause.” Flashing Beacons on Warning Sign (Question 10) Description As specified in the MUTCD (14), flashing beacons can be placed over warning signs to increase their conspicuity and heighten the degree of warning to the motorist. These can be as simple as adding a continuous flashing beacon to an advance curve warning sign or using one that is activated by a motorist traveling at higher than the designated speed, which is usually that shown on the speed advisory plaque; an example of the latter is shown in Figure 9. Total States Responding By Frequency of Use Do Not Use Often Sometimes Rarely 41 0 7 18 16 Table 6. States’ responses on use of dynamic curve warning systems. Figure 9. Example of flashing beacon on warning sign, Augusta, ME (12).

22 Practices for Preventing Roadway Departures Literature Findings The literature search did not reveal any studies of the safety effectiveness of this countermeasure, but as cited by Albin et al. (12), when flashing beacons are installed with curve warning signs and chevrons, the CMF Clearinghouse lists a 37% to 76% reduction in various crashes based on a 2009 study from Italy (27). State Practices The responses of all states are shown in Table B10 in Appendix B. As noted in Table 7, all but one state responded that they used this countermeasure, with 12 responding with “rarely,” and the remaining 28 responding with “sometimes” or “often.” The two comments shown here would seem to reflect the states’ use of this countermeasure: • “These are installed on a case-by-case basis after a thorough engineering study.” • “This is a spot treatment that is used when standard traffic control devices have proven to be ineffective in solving the problem.” The one state that responded that it did not use this device indicated that maintenance con- cerns were the reason. Rumble Strips (Questions 11, 12, and 13) Description There are three types of rumble strips used to alert drivers that they: 1. Are about to encroach into opposite-direction traffic (the centerline rumble strip), 2. Have reached the right side of their travel lane (the edge-line rumble strip), and 3. Have encroached onto the shoulder (the shoulder rumble strip). When the centerline or edge-line pavement marking is placed over the strip, it is referred to as a rumble stripe. Examples of these are shown in Figures 10, 11, and 12. In the questionnaire, the states’ uses of these three applications were treated separately, but they are combined for this summary. This countermeasure has become commonplace over the last two decades. It originated with shoulder rumble strips used for Interstates, and in the last 10 to 15 years has expanded to other road types, with the use of centerline rumble strips for two-lane roads and shoulder and edge-line rumble strips for non-freeway roads. Literature Findings The literature on all aspects of rumble strips is extensive. The FHWA has a website devoted to rumble strips (9). Part of the website discusses the safety effectiveness of these three types of rumble strips. It states the following: Eleven states and one national study have analyzed the effectiveness of centerline rumbles in reducing crashes. These studies conclude that crossover crashes were reduced 18% to 64%, with most studies showing 40% to 60%. Total States Responding By Frequency of Use Do Not UseOften Sometimes Rarely 41 9 19 12 1 Table 7. States’ responses on use of flashing beacons on warning signs.

Survey of the State of the Practice 23 (Courtesy of Tracy Lovell, Kentucky Transportation Cabinet). Figure 10. Example of a milled centerline rumble stripe. (Courtesy of Tracy Lovell, Kentucky Transportation Cabinet). Figure 11. Example of a milled edge-line rumble stripe. (Courtesy of Tracy Lovell, Kentucky Transportation Cabinet). Figure 12. Example of a milled shoulder rumble strip.

24 Practices for Preventing Roadway Departures On rural freeways, edge-line rumble strip studies show that single-vehicle run-off-road fatal and injury crashes can be reduced by nearly 29%. For shoulder rumbles, 14 state and two multistate studies report reductions in single-vehicle run-off-road freeway crashes of 14% to 80%, with most reporting reductions in the 30% to 40% range. The three states that restricted their crash analysis to crashes caused by distracted or drowsy driving (the true target crashes for rumble strips) report 40% to 80% reduction in those crashes. (9) State Practices Tables B11, B12, and B13 in Appendix B show the state responses for shoulder rumble strips, edge-line rumble strips, and centerline rumble strips, respectively. Table 8 combines the responses from the 41 states that responded regarding their use. As seen by the data, nearly all states used centerline rumble strips, a large majority were using shoulder and edge-line rumble strips, and all three types were used often by a majority of states. Some of the comments made by the states for each of the types are shown here: • Shoulder: – “Often used on divided highways; seldom used on conventional roads due to noise complaints.” – “Policy is to install on rural limited-access and have on some urban limited-access road- ways. Having difficulty making more systemic or systematic due to noise concerns.” – “Typically, any major roadway, new or reconstruction, would include shoulder rumbles in rural locations.” • Edge line: – “We really like this one because of the added benefit of improved visibility of striping.” [Note that this would apply to the rumble stripe design.] – “Used when shoulder width insufficient for milled rumble strip.” – “Some concerns over conflicts/interactions with bicyclists.” – “Standard practice for all rural paving projects.” • Centerline: – “Most districts have concerns about the reduced life span of the centerline pavement joint with CLRS [centerline rumble strips]. Districts typically prefer when this treatment is used that HSIP funds cover an overlay, as well, which is cost prohibitive.” Raised (Profiled) Pavement Marking (Question 14) Description A raised (also known as profiled) thermoplastic pavement marking is a less costly but less effec- tive treatment to produce the same result from that of a centerline or edge-line rumble strip. As illustrated by Figure 13, it is a pavement marking line composed of a thicker-than-usual thermo- plastic with ridges added at a prescribed spacing. This design and material increase its visibility and produce a vibration, albeit less than that of a rumble strip, to alert motorists. It is not used in states where there is snow because it is easily damaged by snowplowing operations. Rumble Strip type Yes, Use Frequency of Use Rarely Sometimes Often Shoulder 41 3 3 35 Edge line* 30 7 5 17 Centerline 40 5 13 22 *One state did not indicate its frequency of use. Table 8. States’ use of shoulder, edge-line, and centerline rumble strips.

Survey of the State of the Practice 25 Literature Findings While this device has been used by some states for many years, there had not been any published research evaluating its effect on crashes until an FHWA-sponsored study was completed. The study used data from two-lane and multilane roads in two states—Florida and South Carolina— to examine the effects for specific crash types, including total, fatal plus injury, run-off-road, head-on, sideswipe-opposite-direction, sideswipe-same-direction, wet-road, nighttime, and nighttime wet-road crashes. Only nighttime wet-road crashes, the principal target crash type, experienced a material change in yielding a CMF of 0.908, which was not unexpected since this was the primary target crash type. Although the estimated CMF was based on a small sample of crashes and was not statistically significant at the 95% confidence level, it was consistent between the two states, which suggests the use of this device may be justifiable (28). State Practices The responses for all 40 states that responded are shown in Table B14 in Appendix B, and Table 9 shows the tally for frequency of use. Sixteen of the states reported that they used this device, and nearly 44% (seven states) of those said they did so often. While various reasons were given for not using this device, the most prominent response was maintenance concerns—specifically that they are scraped off by snow- plows. However, in states where this is not an issue, the thicker thermoplastic material provides a longer service life than that of standard pavement markings. SafetyEdge (Question 15) Description Pavement edge drop-off on highways has been linked to many serious crashes, including fatal collisions. To mitigate vertical drop-offs, the FHWA advocates installing the SafetyEdge paving technique during paving or resurfacing projects. This countermeasure allows drivers who drift off highways to return to the pavement safely (29). Figure 13. Example of profiled pavement marking (12). Total States Responding By Frequency of Use Do Not UseOften Sometimes Rarely 40 7 5 4 24 Table 9. States’ responses on use of raised profile markings.

26 Practices for Preventing Roadway Departures The SafetyEdge is constructed with a paver attachment that enables the pavement edge to be paved and compacted to a finished 30-degree angle to promote a safe return to the travel lane after a roadway departure. After paving, the SafetyEdge is backfilled and graded flush with the paved surface. Where the compacted material covering the pavement SafetyEdge settles or erodes, the angled edge is easily traversable by vehicles attempting to re-enter the roadway when compared to a vertical or near-vertical pavement edge drop-off (29). Figure 14 shows a typical pavement cross-section with the SafetyEdge for a pavement resur- facing project. Literature Findings The most recent and comprehensive evaluation of the safety benefits of SafetyEdge was con- ducted by Donnell et al. (30). Their study estimated CMFs for this paving technique on two- lane rural roads. Using an empirical Bayes observational before–after evaluation, they found that this countermeasure was associated with statistically significant reductions in fatal and injury (FI), run-off-road (ROR), opposite-direction, and drop-off-related crashes. Further- more, their economic evaluation found that the SafetyEdge paving technique is cost-effective, with the benefit–cost ratios ranging from 590:1 to 1180:1 for ROR crashes and from 730:1 to 1460:1 for FI crashes. State Practices Table B15 in Appendix B displays the responses from all states. The use of SafetyEdge is becoming more prevalent in the United States, as evidenced by the responses summarized in Table 10, which shows that 85% of the 41 states responding were using this treatment, with a majority stating it as being used often. Several states have adopted it as part of their design stan- dards and guidance, especially for resurfacing projects. Table B15 provides hyperlinks to several state policies, standards, and specifications Six states responded that they had not yet adopted this paving technique because of insuf- ficient funding, maintenance concerns, or negative public feedback. One state responded that it was working to incorporate this procedure into its standard practice but was receiving some resistance from contractors based on liability. Figure 14. Cross-section showing SafetyEdge (29). Total States Responding By Frequency of Use Do Not UseOften Sometimes Rarely 41 22 9 4 6 Table 10. States’ responses on use of SafetyEdge.

Survey of the State of the Practice 27 Shoulder Widening on Curved Section (Question 16) Description Shoulders, which can be either paved or non-paved but are usually stabilized, provide an opportunity for an encroaching vehicle to return to the travel lane. They are an integral part of the cross-section design, with recommended widths that vary depending on several factors, but nota- bly the roadway type (see https://safety.fhwa.dot.gov/geometric/pubs/mitigationstrategies/ chapter3/3_shoulderwidth.cfm). Shoulder widening (depicted in Figure 15) was identified as a countermeasure for roadway departure crashes by Albin et al., who noted that it “. . . is particularly important in horizontal curves where vehicles typically use more of the travel lane than in straight sections” (12). Hence, providing a shoulder where one did not exist or, more commonly, widening an existing narrow shoulder, will provide a recovery area, allowing the driver to regain control in the event of a roadway departure. Literature Findings While the safety effects of varying shoulder widths have been examined for decades, the lit- erature search did not identify any studies specific to widening shoulders on curves. The most recent relevant safety information is reported in the Highway Safety Manual (2) and presented with modification in Albin et al. (12) and Table 11. Figure 15. Shoulder widening on the inside and outside of the curve (12). Shoulder Width Average Annual Daily Traffic (AADT) (vehicles/day) <400 400–2,000 >2,000 0 ft +10% Between +10% and +50%, depending on AADT +50% 2 ft +7% Between +7% and +30%, depending on AADT +30% 4 ft +2% Between +2% and +15%, depending on AADT +15% 6 ft 0% 0% 0% 8 ft or more –2% Between –2% and –13%, depending on AADT –13% Crash types: Single-vehicle run-off-road, multiple-vehicle head-on, opposite-direction sideswipe, and same-direction sideswipe. Table 11. Percent change in crashes relative to providing a 6-ft shoulder on rural, two-lane roadway segments (modified from HSM, Table 13-7).

28 Practices for Preventing Roadway Departures Earlier in this chapter, under the section Programmatic Problem Identification and Imple- mentation Strategies, an example was presented of how Arizona applied performance-based practical design as a decision-making approach and evaluation for a shoulder width (and super- elevation) improvement project as a countermeasure to a high occurrence of roadway departure crashes. In that case study, provided in Appendix D, Arizona analyzed the effect on safety for two widening alternatives (1 ft to 5 ft and 1 ft to 8 ft) on a 24-mile corridor of a two-lane rural road. It was found that while both alternatives would result in a significant reduction in total crashes, the first alternative was selected for full application because of its higher benefit–cost ratio. State Practices The responses from the states are shown in Table B16 in Appendix B. As shown in Table 12, 30 of the 41 states responded that they widened shoulders on curves, with 10 responding that it was rarely done, 17 that it was sometimes done, and three that it was done often. Some of the comments from those states that responded positively were: • “Typically, it has been done more in the past for erosion control issues rather than safety.” • “Plan on doing more in the future through systemic process.” • “We have done shoulder widening around curves to mitigate truck off-tracking issues.” • “Would be done as part of a paving project.” The reasons states gave for not using this countermeasure were varied: • Not aware of countermeasure (two states), • Countermeasure not proven to be effective (two states), • Insufficient funding (four states), • Maintenance concerns (one state), and • Insufficient right-of-way, especially on more rural roads (one state). High-Friction Surface Treatment (Question 17) Description A roadway must have an appropriate level of pavement friction to ensure that drivers are able to keep their vehicles safely in the lane. Poor pavement conditions, especially wet pavement, have been identified as one of the major contributing factors in roadway departure crashes. When a pavement surface is wet, the level of pavement friction is reduced, and this may lead to skidding or hydroplaning. To address this problem, state agencies are using several pavement friction enhancement treatments, one of which is labeled as high-friction surface treatment (HFST). HFST products consist of a thin layer of binder—usually urethane, silicon, or epoxy— topped with specially engineered, durable, high-friction aggregates. The aggregate systems have a long-lasting skid resistance, and also make the pavement overlay much more resistant to wear and polishing. An example of an application, with a close-up of the surfaces, is shown in Figure 16. Total States Responding By Frequency of Use Do Not UseOften Sometimes Rarely 41 3 17 10 11 Table 12. States’ responses on use of shoulder widening on curve sections.

Survey of the State of the Practice 29 Literature Findings The safety benefit of this countermeasure has become well established, as reported by Albin et al. (12) and the FHWA website for this countermeasure (see https://safety.fhwa.dot.gov/ roadway_dept/pavement_friction/high_friction/). Kentucky used this countermeasure on 30 curves in 2009 and observed a crash reduction of 70% to 75% at these curves. A case study report from Kentucky for this countermeasure is provided in Appendix E. State Practices Table B17 in Appendix B provides the responses of all 41 states. The use of this pavement treatment as a countermeasure has become widespread among the states, as shown by the sum- mary of states’ use or non-use in Table 13. Ninety percent of the 41 states responding indicated that they were using this countermeasure. Several states responded that they were starting to experiment with this countermeasure. For example, Indiana DOT was executing its first HFST projects in fiscal year 2017, investing roughly $1 million, almost exclusively for horizontal curves. It planned to continue the program in future years. There is still some hesitancy by a few states to use this countermeasure, with one state citing insufficient funding and another questioning its durability. Pavement Grooving (Question 18) Description Grooving is a pavement treatment in which narrow grooves are saw cut into the pavement surface, typically in the direction of traffic, and typically 0.75 in. apart (see Figure 17). The Figure 16. Example of HFST (12). Total States Responding By Frequency of Use Do Not UseOften Sometimes Rarely 41 3 17 17 4 Table 13. States’ responses on use of HFST.

30 Practices for Preventing Roadway Departures grooves increase pavement macrotexture, thereby increasing or restoring pavement friction. Grooving is typically used on concrete pavements, but it can be used on asphalt pavements (31). It is especially effective in reducing wet-weather crashes by improving the drainage characteris- tics, thereby serving as a roadway departure countermeasure. Literature Findings Albin et al. cited a New York State DOT evaluation of pavement grooving that found that wet-pavement–related crashes were reduced by 55%, and the total for both wet and dry pavement crashes was reduced by 23% (12); however, a reference was not provided. Under the FHWA’s Evaluation of Low-Cost Safety Improvements Pooled Fund Study, Merritt et al., employing an empirical Bayes before–after methodology, evaluated the effects on various crash types on several types of pavement friction improvements, including diamond grooving. For diamond grooving, there was an overall benefit (significant at the 5% level) for both wet- and dry-road crashes, which resulted in a significant overall benefit for total crashes (31). State Practices The state responses for this countermeasure are shown in Table B18 in Appendix B. As shown in Table 14, 39 states responded, with a nearly equal number responding with yes and no. Of those who used this countermeasure, the majority responded with “rarely.” Two noteworthy comments from those states that used this countermeasure were: • “Both HFST and pavement grooving have been used in areas where a significant proportion of crashes are wet-pavement–related.” • “Not part of a safety program but part of pavement group repair strategies.” Figure 17. Example of pavement grooving (12). Total States Responding By Frequency of Use Do Not UseOften Sometimes Rarely 39 2 7 10 20 Table 14. States’ responses on use of pavement grooving.

Survey of the State of the Practice 31 The reasons cited by states that did not use this countermeasure included: • Not aware of the countermeasure (10 states), • Maintenance concerns (five states), and • Not proven to be cost-effective (two states). Cable Median Barrier (Question 19) Description As stated in a 2016 NCHRP synthesis report, generic low-tension cable barriers had been used in several states for many years, and it was not until the year 2000 that a high-tension cable bar- rier was first installed in the United States (32). The questionnaire in this synthesis did not label this countermeasure as “high-tension,” but based on the responses, the respondents presumably assumed that it was. The most common applications today are in medians of divided roadways. Cable barriers function by capturing or redirecting impacting vehicles to prevent these vehicles from intruding into the opposing traffic lanes. As such, it is a countermeasure with the objective of preventing head-on crashes with opposing traffic. There are five propriety high-tension cable barrier sys- tems deemed eligible for federal funding by the FHWA (32); Figure 18 illustrates one of these. Literature Findings The overall safety benefit from the use of cable median barrier systems has been well docu- mented by several studies. A summary of these studies was documented by the Louisiana Depart- ment of Transportation in its Cable Median Barrier Systemic Review (33). The conclusion drawn from the many before-and-after studies of changes in crashes and severities is that these systems are highly effective in reducing fatal and severe injury crashes, but result in an increase in total, property damage only, and minor injury collisions. State Practices Responses from all states are shown in Table B19 in Appendix B. As shown in Table 15, all but four states responded that they used cable median barriers, with 32 responding with either “sometimes” (11 states) or “often” (21 states). Figure 18. Example of a high-tension cable median barrier system (32).

32 Practices for Preventing Roadway Departures Some of the comments provided by the states are summarized here: • The results of a study of high-tension cable barriers, completed by Wayne State University (34), show that cable median barriers have been highly effective at reducing crossover crashes in Michigan. After the barriers were installed, crossover crash rates on those freeway segments fell by 87%, and the barriers successfully contained 97% of the vehicles that hit them. Cable barriers have improved overall safety at the locations where they have been installed. The most serious crashes—fatal and severe injury crashes—decreased by 33% after cable median barriers were installed, according to statistical analysis. Since their installation, cable median barriers are estimated to have saved 20 lives and prevented over 100 serious injuries in Michigan. • About half of one state’s 1,200 centerline miles of the Interstate system have cable median barriers. • They are implemented on high-volume and narrow-median stretches. Minnesota currently has about 500 miles implemented. The reasons given by the four responding states for not using this barrier type were maintenance concerns (three states) and that the crash frequency was deemed to not be a significant issue. Increase Clear Zone (Question 20) Description AASHTO’s Roadside Design Guide defines a “clear zone” as the total roadside border area, starting at the edge of the traveled way, available for safe use by errant vehicles (35). This area may consist of a shoulder, a recoverable slope, a non-recoverable slope, and a clear run-out area. The desired minimum width is dependent on traffic volumes, speeds, and roadside geometry. By creating a clear zone, roadway agencies can increase the likelihood that a roadway departure will result in a safe recovery rather than a crash and can mitigate the severity of any crash that might occur. The AASHTO guide provides suggested values for the design clear zone, which can be from at least 7 ft to as much as 30 ft. The question posed in the survey was about increasing the clear zone beyond the minimum required. State Practices The responses of 39 states can be seen in Table B20 in Appendix B. As shown in Table 16, 22 states responded that they do increase clear zones beyond what would be required, with 13 replying with “sometimes,” seven with “rarely,” and only two with “often.” Some comments from states that responded that they used this countermeasure were: • “Designers are encouraged to remove large hazards beyond the clear zone. This can often be difficult and hard to get approval [for].” Total States Responding By Frequency of Use Do Not UseOften Sometimes Rarely 41 21 11 5 4 Table 15. States’ responses on use of cable median barriers. Total States Responding By Frequency of Use Do Not UseOften Sometimes Rarely 39 2 13 7 17 Table 16. States’ responses on increasing clear zones.

Survey of the State of the Practice 33 • “The designer may choose to increase the clear-zone width on the outside of horizontal curves where accident histories indicate a need or where specific site investigation shows a defini- tive accident potential. This may be cost-effective where increased banking or other accident countermeasures are not feasible.” • “This is not used that often but is used where there is sufficient ROW [right-of-way] and crash data to support [it].” The reasons given by states for not using this countermeasure are shown in Table 17; the primary reason was insufficient funding. Increasing the clear zone usually requires obtaining more ROW, which can increase the project cost substantially. One state’s comment was “many times the project type, repaving, for example, is not scoped for acquiring the necessary ROW to increase the clear zone.” From the reasons cited and from the comments received, the cost of acquiring additional ROW limits the use of this countermeasure. Flatten Side Slope (Question 21) Description Steeper and cut-type slopes are more hazardous and have been shown to significantly affect the severity of run-off-road crashes. The AASHTO Roadside Design Guide (35) considers side slopes steeper than 1V:3H as critical slopes, meaning that a vehicle could become unstable to the point that the risk of it overturning is increased. Roadside slope improvement, or flattening, has as its objective the provision of a forgiving environment for an errant vehicle. Flatter slopes lower the likelihood of vehicles overturning. Literature Findings As reported by Albin et al., the CMF Clearinghouse contains several CMFs for slope flatten- ing (12). Flattening side slopes from the critical slope of 1V:3H to just 1V:4H can realize a 42% reduction in injury crashes, while a side slope improvement of 1V:4H to 1V:6H can realize a 22% reduction in injury crashes. State Practices The responses from the 39 states that answered this question can be seen in Table B21 in Appendix B. As shown in Table 18, 30 states responded that they flattened slopes, but only three indicated that this was done often. Reason No. of States Citing Reason Not aware of countermeasure 2 Countermeasure not proven to be cost- effective 2 Insufficient funding for countermeasure 9 Negative public feedback 5 Maintenance concerns 3 Other 5 Table 17. States’ reasons for not increasing clear zones as a countermeasure. Total States Responding By Frequency of Use Do Not UseOften Sometimes Rarely 39 3 18 9 9 Table 18. States’ responses on use of side slope flattening.

34 Practices for Preventing Roadway Departures Some of the comments made were: • “Mostly done during reconstruction. Not an active program to complete this.” • “We hope to use this more in the future to help move the cable barrier from the shoulder to the center of the ditch, to help eliminate nuisance hits.” • “Typically 3R [resurfacing, restoration, rehabilitation] or widening project driven.” As for the nine states stating that they did not use the countermeasure, the reasons given were: • Not aware of countermeasure, • Countermeasure not proven to be cost-effective, and • Insufficient funding. Tree Removal (Question 22) Description One of the most common causes of fatal and severe injury crashes, especially on rural roads, is vehicles leaving the road and striking a fixed object. In fatal crashes involving a fixed object, trees are the objects most often struck. About 8% of fatal crashes involve crashes into trees (5). Elimination of trees with greater than a 4-in. diameter within what should be the clear zone for a given roadway (as illustrated in Figure 19) would be the obvious strategy or countermeasure. However, trees contribute to roadway aesthetics, and their removal evokes deep sentimental and environmental concerns among agencies and stakeholders (5). Literature Findings The literature contains few studies on the safety benefits of a tree removal program. In 2009, Clemson University researchers reported on their evaluation of roadside collision data and clear zone requirements in which trees and other fixed objects were considered (37). In South Caro- lina, trees account for 25% of all fatal crashes, compared to 8% nationally, and 72% of the tree- related crashes occurred on curve sections. Their analysis of sites with and without adequate clear zones led them to conclude that South Carolina could realize a notable decrease in fatal and injury crashes if the recommended clear zones were provided. State Practices The responses from 40 states that replied to this question can be seen in Table B22 in Appen- dix B. As shown in Table 19, all but one state responded that they used this countermeasure, with 18 doing so sometimes, 16 rarely, and four often. Figure 19. Example of tree within clear zone (36).

Survey of the State of the Practice 35 From the comments provided by the respondents, it appears that it is difficult to get approval to remove trees and that it is done on a case-by-case basis and not through any general policy or guideline. One respondent commented that they needed better guidance on how this issue relates to speed—more specifically, at what speed levels does tree removal become more critical. The representative from the state that does not use this countermeasure commented that this is not a problem because they have a predominately desert climate. The literature search on this issue uncovered an FHWA document entitled Noteworthy Prac- tices: Roadside Tree and Utility Pole Management that provides examples of successful tree (and pole) management practices from several states (38). Increase Sight Distance on Curves (Question 23) Description A Guide for Reducing Collisions on Horizontal Curves lists “increasing the radius of a horizontal curve” as one strategy for reducing the likelihood of a vehicle leaving its lane and either crossing the roadway centerline or leaving the roadway at a horizontal curve (41). Increasing the sight distance on a curved section of the road can be achieved by increasing its radius through realignment, as depicted in Figure 20. The line superimposed on the aerial image illustrates the old alignment. Literature Findings In a recent study of curve realignments that increase sight distance (39), the researchers devel- oped CMFs for the following three crash types: • Total crashes: 0.315, • Injury and fatal: 0.259, and • Run-off-road plus fixed object: 0.216. Total States Responding By Frequency of Use Do Not Use Often Sometimes Rarely 40* 4 18 16 1 *One state responded “yes” but did not indicate frequency of use. Table 19. States’ responses on use of tree removal. Figure 20. Example of horizontal curve realignment (39).

36 Practices for Preventing Roadway Departures They also developed CMFunctions that showed that safety benefits may be greater for curves with a larger central angle and where the difference in radius between the before and after period conditions is larger. Their economic analysis revealed a benefit–cost ratio range of 1.75:1 to 4.38:1 (39). State Practices Table B23 in Appendix B shows the responses from the states, and Table 20 shows a summary of their use or non-use. Somewhat surprisingly given the cost involved, which could include ROW acquisition, 37 of the 40 states replied that they made this improvement. Only one state indicated that it was done often, while 21 states said that it was done sometimes and 14 rarely. The most frequent comment was that this type of improvement was done on a case-by-case basis and as part of a reconstruction. The primary reason for those states not using this countermea- sure was insufficient funding. Superelevation Improvement (Question 24) Description Superelevation is the rotation of the pavement on the approach to and through a horizontal curve. It is intended to assist the driver by counteracting the lateral acceleration produced by tracking the curve. Selection of a maximum superelevation rate is based on variables such as cli- mate, terrain, highway location, and frequency of slow-moving vehicles. Inadequate supereleva- tion can cause vehicles to skid as they travel through a curve, potentially resulting in a roadway departure crash. Trucks and other large vehicles with high centers of mass are more likely to roll over at curves with inadequate superelevation. Superelevation is occasionally inadequately designed or is lost over time due to settling or overlays. Correcting an inadequate superelevation is considered a countermeasure for roadway departure crashes. Literature Findings The literature review did not uncover any recent studies that examined the safety effects of improving superelevation; however, the Highway Safety Manual provides a function for calcu- lating CMFs for horizontal curves for two-lane rural roads based on superelevation variance (2). State Practices Table B24 in Appendix B shows the responses from all of the states, and Table 21 summarizes the responses for frequency of use. All but three of the 40 states made superelevation improvements Total States Responding By Frequency of Use Do Not UseOften Sometimes Rarely 40* 1 21 14 3 *One state responded “yes” but did not indicate the frequency of use. Table 20. States’ responses on use of increasing sight distance on curves. Total States Responding By Frequency of Use* Do Not UseOften Sometimes Rarely 40 4 15 16 3 *Two states that responded “yes” did not provide response for frequency of use. Table 21. States’ responses on use of superelevation improvement.

Survey of the State of the Practice 37 as a roadway departure countermeasure. Based on the comments provided, this improvement appears to be done on a case-by-case basis and frequently is part of a 3R-type project. The only reason offered for not using this countermeasure was that was not considered cost-effective. Other Countermeasures Being Used by States (Question 25) The intent of this question was to identify countermeasures being used by the states that were not one of the 20 mentioned in questions 5 through 24. The responses from each state are shown in Table B25 in Appendix B. Other countermeasures obtained from the responses are: • Passing lanes: Alaska noted that passing lanes on rural two-lane roads help with aggressive driving, which can lead to roadway departure crashes. • Fluorescent yellow sheeting: Noted by Alabama, this device was mentioned earlier in this report, but was not included as a countermeasure because it is considered a standard traffic control device. • Culvert extensions: Noted by Hawaii, this is a roadside improvement that might prevent roll- overs or vehicles hitting a fixed object. This improvement to the roadside was also mentioned in follow-up discussions with Alabama as one of many countermeasures it was using as part of its roadway departure safety improvement program. • LED in-pavement lighting: Noted by Colorado and by Ohio as being used on interchange ramps and rural curves. • Motorcycle barrier attenuator: Noted by Utah, this attaches to a standard roadside guardrail to protect errant motorcyclists from impacting the guardrail post during crashes. Research to Evaluate Safety Effectiveness of Countermeasures (Question 26) Table B26 in Appendix B shows the responses provided by 18 states. A summary of these responses by countermeasure is as follows: • Add 2-ft shoulder on two-lane roads: – Alabama – a 25% to 35% reduction. • Rumble strips: – Michigan: a before-and-after crash study performed for the 2008–2010 installation locations found significant reductions across all crash severities and lane departure crash types, including: � 50% reduction in head-on crashes, � 46% reduction in run-off-road crashes, � 51% reduction in fatal crashes, and � 41% reduction in incapacitating injury crashes. – Mississippi: statistically significant difference in the number of roadway departures between the period before construction and the period after construction. – Minnesota: found no evidence to implicate centerline rumble strips as a hazard to two- or three-wheel cycles. – Vermont: crash analysis from the two sites evaluated demonstrated a reduction in the total number of crashes and the proportion associated with centerline crossover events. • Cable median barrier: – Louisiana: cable median barriers reduced fatal and serious injury crashes by almost 30% and 20%, respectively. – Michigan: fatal and serious injury crashes decreased by 33%.

38 Practices for Preventing Roadway Departures • Wider (6-in.) edge line: – Minnesota: 6-in. edge lines were an effective countermeasure for overall crash reduction and run-off-road right crash reduction. • Slope improvement: – Nevada: for a single 30-mile project, simple before–after analysis in the 3-year period showed a 55% reduction in total crashes and a 75% reduction in injury crashes. • High-friction surface treatment: – South Dakota: a total crash reduction of 78% over 4 years for application at four horizontal curves. Hyperlinks to the research reports can be found in Table B26. Countermeasures Needing More Evaluation (Question 27) Table B27 in Appendix B displays the comments made by the 23 states that felt more research was needed for one or more of the countermeasures. The responses are summarized as follows: • Seven states believed that more research is needed for the use of rumble strips, with concerns for: – Pavement maintenance for centerline rumble strips, – Pavement marking over the strip, – How or whether to use in combination with raised pavement markers, – Sinusoidal rumble strip effectiveness (from two states), and – Use with narrow shoulders. • Three states mentioned the need for developing CMFs for tree and other fixed-object removal. • Several states mentioned that further research was needed for delineation treatments, including for: – Wider edge lines, – Flexible tube delineators, – Larger (6 × 8 in.) delineators for low-volume roads, and – Delineation of hazards that cannot be shielded or moved. • A few states commented on the need for further research on signing for: – Horizontal alignment, – Dynamic curve signs, – Higher intensity sheeting, and – Fluorescent sheeting. • A few states commented on the need for further research related to shoulders, specifically: – Shoulder widening, and – Narrow-shoulder sign treatments. • Other research needs mentioned by at least one state included: – Slope flattening, – Clear zone widening, – Embedded LEDs, – SafetyEdge, and – High-friction surface treatments. Information provided in this synthesis may address some of these concerns and obviate the need for further research. Evaluations of the Non-Safety Impacts of Countermeasures (Question 28) The intent of this question was to determine whether there were any issues with regard to durability, service life, maintenance, or another non-safety effect for any of the countermeasures that might influence whether or how a countermeasure would be used. The comments from the

Survey of the State of the Practice 39 11 states that responded that they have used or intend to use the countermeasures are shown in Table B28 in Appendix B. Of note are the following: • Georgia conducted research to determine how well pavement joints located at the center of the roadways had been holding up to centerline rumble strips. Based on that research, it was pursuing approval of a detail calling for two offset rumble strips to be placed on each side of the roadway centerline to avoid deterioration of the pavement along the joint. • Michigan had a comprehensive analysis of its cable median barrier program performed. The sponsored research showed significant positive safety benefits from use of this countermea- sure, but given the cost of installation plus considerable maintenance costs, it conducted a life-cycle cost evaluation. This evaluation consisted of a time-of-return (TOR) analysis, which is defined as the amount of time that must pass after implementation, typically gauged in years, for the expected benefits of the initiative to equal the costs of the initiative. Engineering, construction, and maintenance costs were considered as part of the TOR analysis, as well as the benefits realized by reductions in severe crashes. The evaluation revealed that the TOR for cable median barrier installation in Michigan was 13.38 years (34). • Kentucky performed durability analysis of pavement markings, raised reflective pavement markers, and rumble strips as alternatives for providing wet-nighttime delineation. Some of the findings were: – Snow-plowable markers provided the most effective wet-nighttime delineation. Durability issues associated with the steel-casting marker made the recessed marker preferable over the life of the pavement. – Wet-reflective tape placed in a groove provided both dry and wet-nighttime delineation and remained durable. – The performance of thermoplastic material installed on the pavement surface supported its future use, but it will not provide wet-nighttime delineation. Inconsistent performance of inlaid tape argues against its expanded use. Poor performance showed that future use of wet-reflective tape should not be considered. – A cost analysis, considering durability of the materials, showed that the cost of using more durable materials over the life of the pavement was not dramatically more than the cost of traffic paint. – Edge-line rumble stripes and centerline rumble strips enhanced wet-nighttime delineation and should be incorporated into resurfacing projects on two-lane roads where pavement width permits. • Both Virginia and Washington were evaluating the life cycle of their high-friction service treatment sites. Part IV. Questions Related to Vehicle-Based Technologies (Questions 30 Through 31) While the scope of this synthesis was to focus on engineering countermeasures, states were asked to comment on how they were addressing two vehicle-based technologies—(1) V2V and V2I technologies, and (2) autonomous vehicles. While these are not engineering measures, it was suggested to explore how these vehicle technologies might affect states’ programs for addressing the roadway departure crash problem. These were meant to be exploratory questions and were not intended to be part of an in-depth inquiry. Actions States Pursuing Related to V2V and V2I Technologies (Question 30) Thirty-five states answered this question, with 15 responding that they took no special action, six simply providing a contact person for follow-up, and 15 responding with a variety of

40 Practices for Preventing Roadway Departures comments, which are shown in Table B30 in Appendix B. The responses from those states that provided more substantive information are highlighted here: • “We have created a Transportation, Systems, Management and Operations Division to incorporate new technology and develop programs” (Connecticut). • “Multidisciplinary team meeting to discuss and determine actions needed for future implementation” (Kentucky). • “We have formed an Autonomous Vehicle Technology Team, which is investigating issues that affect this technology” (Louisiana). • “We are watching this evolution closely and watching for those infrastructure improvements that may be required, such as wider edge lines or Lidar sensing units along the roadway” (Nevada). The responses indicate that states were mostly monitoring these technologies and assessing what actions would be necessary with regard to roadway infrastructure. Actions States Are Pursuing Related to Autonomous Vehicles (Question 31) This question asked what special actions were being taken by states to ensure that lane bound- aries [defined by lane lines (for multilane facilities), edge lines, or centerlines] were well defined. For this question, 16 states provided a response; these are shown in Table B31 in Appendix B. In general, based on these 16 responses and the fact that the remaining 24 states did not provide a response, it appears that states were not taking any special actions at the time. The responses from these two states would seem to apply to most of the states: • “Future of automated and autonomous vehicles has not to this point changed the manner in which we establish and maintain those traffic control devices.” • “Simply applying our pavement marking policy and doing our best to maintain the longitu- dinal markings. But we were doing this before autonomous vehicles.” Part V. Case Examples and Final State Comments (Question 32 and Question 34) Examples of How Countermeasures Were Effective in Reducing Roadway Departure Crashes (Question 32) States were asked to provide examples of any countermeasures they applied that were espe- cially effective in reducing roadway departure crashes and that they would like to see featured in the synthesis. Sixteen states responded to this question; their statements are shown in Table B32 in Appendix B. Countermeasures that some states felt had been especially effec- tive were: • High-tension cable median barriers, • Shoulder widening with rumble strips, • Rumble strips of all types, and • High-friction surface treatments. Comments from two state DOT respondents in support of these countermeasures are shown here: • “Internally, the agency [Indiana] recently completed two simple before–after in-service per- formance studies of our now 10-year-old Interstate cable median barrier program and 4-year-old rumble stripE program. Both concluded that the two treatments were highly

Survey of the State of the Practice 41 effective at reducing risk of severe crashes, particularly fatal events. We expected high effec- tiveness (CRF [crash reduction factor]) with cable barrier relative to reducing risk of fatal crashes on freeways, but were stunned by apparent positive effectiveness of rumble stripE projects. Overall, Indiana DOT has installed some 1,300 ‘run-miles’ of edge-line and centerline rumble stripE since 2011. In terms of highly relevant crashes to that counter- measure, of the 54 projects with sufficient ‘after installation’ time to make a reasonable before–after assessment, only one site experienced higher frequency of fatal crashes after vs. before. One outstanding example of sheer before–after effect was on a 17-mile stretch of high-volume, high-speed two-lane highway—one of the early projects, with combination of centerline and edge-line rumble stripe—where 11 fatal crashes took place in the 4 and a half years just prior to installation but none in the 3 years since.” • “Mississippi had a number of very serious and/or fatal cross-median crashes in 2007 on a stretch of high-speed urban Interstate in the Jackson area. One example is this story: http:// www.wdam.com/story/7130388/arkansas-man-faces-murder-charge-in-fatal-i-220-wreck. This necessitated our first installation of high-tension cable barrier along this Interstate high- way. Even before the system was fully constructed in 2008, the system was saving lives. From the initial success stories, MDOT [Mississippi DOT] has been able to systemically install cable barriers on all high-speed, controlled access, divided highways where the site conditions allowed the installation of this treatment. The initiative to install cable barrier through Mis- sissippi won MDOT an award in 2012. While no countermeasure is completely crash proof, it is the belief of MDOT that this treatment has significantly reduced crashes on high-speed, controlled access, divided highways across the state.” Also, Kentucky provided a link to a project case study for its evaluation of a high-friction sur- face treatment; a one-page summary is provided as Appendix E. The state treated a horizontal curve that experienced 53 wet-weather crashes over a 3-year period with HFST. After the treat- ment, there were only five wet-weather crashes over a period of 3.18 years. Responses to this question from three states—Alabama, Georgia, and North Carolina—were of particular interest not because of examples of effective countermeasures, but of examples of approaches they are using to further the implementation and evaluation of roadway depar- ture countermeasures. Follow-up interviews were conducted to acquire additional information beyond what was provided in their responses shown in Table B32. Summaries of their programs are provided in Appendix F. Other Issues Related to the Application of Countermeasures (Question 34) For the last question, states were asked to raise any other issues related to the topic that had not been covered by responses to the other questions. Table B33 in Appendix B provides the full responses from the states that choose to respond. What is interesting in these responses is that they are all different, with no repetition among the states. Some of the states’ comments are shown here: • “An overall push to implement as many of the systemic measures as possible as part of main- tenance or new construction is likely the most effective long-term solution to bringing down the total number and severity of crashes.” • “Driver inattention, fatigue, DUI, etc. are major contributing factors.” • “We need to address the noise issue that influences the use of standard rumble strips.” • “Driveways and embankments are a huge issue—yet very costly to fix on a systematic basis.” • “Due to our small number of crashes on local roads and low volumes, it is difficult some- times to convince towns that chevrons and curve signs should be installed. We are working

42 Practices for Preventing Roadway Departures on a new systemic approach that we hope will help us identify high-risk areas and provide justifications.” • “Many facets of continuing superelevated shoulders adjacent to the roadway or having a sloped rollover, namely: – Sensitivity of various slope rollovers (2%, 4%, 6%, 8%), – Effects of slope rollover location with respect to total shoulder width and paved shoulder width, – Effects of slope rollover with respect to interaction with roadside crash barrier, and – Effects of slope rollover with respect to roadway slopes outside of shoulder.”