U.S. Department of Transportation
Federal Highway Administration
1200 New Jersey Avenue, SE
Washington, DC 20590
202-366-4000
1. Report No. FHWA-SA-16-003 |
2. Government Accession No. |
3. Recipient's Catalog No. |
|
---|---|---|---|
4. Title and Subtitle Advancing Innovative Intersection Safety Treatments for Two-Lane Rural Highways: Executive Summary |
5. Report Date December 2015 |
||
6. Performing Organization Code |
|||
7. Author(s) D.J. Torbic, D.J. Cook, J.M. Hutton, K.M. Bauer, and J.M. Sitzmann |
8. Performing Organization Report No. 110818.01.004 |
||
9. Performing Organization Name and Address MRIGlobal |
10. Work Unit No. (TRAIS)
|
||
11. Contract or Grant No. DTFH61-12-C-00023 |
|||
12. Sponsoring Agency Name and Address Federal Highway Administration |
13. Type of Report and Period Covered Final Report |
||
14. Sponsoring Agency Code |
|||
15. Supplementary Notes Project Manager: Jeffrey Shaw |
|||
16. Abstract Intersection safety is a national, state, and local priority. Approximately 26 percent of the fatal crashes that occur in the United States are intersection or intersection-related crashes. The objective of this guide is to advance efforts to improve safety at unsignalized intersections with minor-road stop control along rural two-lane roads, by focusing on strategies that are not yet widespread. The safety effectiveness of three low-cost safety treatments was evaluated to estimate their expected effectiveness in reducing crashes. The low-cost safety treatments included: (1) single luminaire intersection lighting, (2) transverse rumble strips in advance of stop-controlled approaches, and (3) supplementary pavement markings on intersection approaches. The effectiveness of each treatment in reducing crashes was estimated using the Empirical Bayes (EB) observational before-after safety evaluation analysis approach. Analyses were performed to estimate the effectiveness of each treatment in reducing crashes for different severity levels and crash types. An economic analysis of the treatments was also performed. Each of the treatments is effective in reducing crashes of different types and severity levels, and is economically justifiable for most traffic volumes. The information in this report can be combined with information on other strategies to reduce intersection or intersection-related crashes at unsignalized intersections with minor-road stop control along rural two-lane roads. With such information, agencies can make informed decisions about planning and programming safety improvements at intersections under their jurisdiction. |
|||
17. Key Words Highway safety, Countermeasure evaluation, Intersection safety, Crash modification factor, Lighting, Empirical Bayes before-after analysis, Rumble strips, Pavement markings |
18. Distribution Statement No restrictions. This document is available to the public through the National Technical Information Service, Springfield, Virginia 22161 |
||
19. Security Classification (of this report) Unclassified |
20. Security Classification (of this page) Unclassified |
21. No of Pages |
22. Price
|
Form DOT F 1700.7 (8-72)
Reproduction of completed page authorized.
Intersection safety is a national, state, and local priority. Approximately 26 percent of fatal crashes in the United States occur at intersections.(1,2) In the period between 2009 and 2013, the average number of intersection-related fatalities was approximately 7,960 per year .(2)
Crashes in rural areas are often more severe than in urban areas because of higher vehicle speeds, and the outcome of crashes may be more severe, in part, due to longer emergency response times. In rural areas, more fatal and severe-injury crashes occur at stop-controlled intersections than at signalized intersections.(3) At stop-controlled intersections, most crashes are caused by a failure to stop at a stop-controlled approach or the acceptance of an insufficient gap when entering the intersection.(4)
The objective of this study was to increase the deployment of low-cost treatments that reduce motor vehicle fatalities and injuries at intersections with minor-road stop control along rural two-lane roads. Information in this report can be combined with information about other strategies available in references such as the Highway Safety Manual (HSM)(5) and the Federal Highway Administration's (FHWA's) Crash Modification Factors (CMFs) Clearinghouse(6) to reduce intersection or intersection-related crashes at rural unsignalized intersections. With such information, agencies can make informed decisions about planning and programming safety improvements at intersections under their jurisdiction. FHWA sponsored research to evaluate treatments for rural stop-controlled intersections that were in use by highway agencies but for which little quantitative information about their safety effectiveness was available. A wide range of treatments was initially considered for the research. The list was reduced to three treatments based on availability of study sites, crash data, traffic volumes, and other data, as well as on priorities of FHWA. The three treatments for which detailed analyses were conducted included single luminaire lighting, transverse rumble strips, and supplementary pavement markings.
The safety effectiveness of each treatment was estimated by conducting an Empirical Bayes (EB) before-after crash analysis. The EB before-after crash analyses were conducted in accordance with steps outlined in Appendix 9A of the HSM(5) to estimate the expected reduction in crash frequency due to implementation of the treatment. In all cases, safety performance functions (SPFs) presented in the HSM for stop-controlled intersections on rural two-lane roads were used. The intersection SPFs predict crashes based on the annual average daily traffic (AADT) of the major and minor roads and are adjusted using CMFs for intersection skew angle (CMFSkew), number of major-road left-turn lanes (CMFLTL), number of major-road right-turn lanes (CMFRTL), and the presence of intersection lighting (CMFLighting). (For the safety evaluation of single luminaires, CMFLighting was not included in the predictions.) Target crash type proportions (PR1) and severity proportions (PR2), which were either taken from default proportions presented in the HSM or calculated from crash data obtained for each state, were used to predict the number of total crashes by crash type and crash severity categories. Calibration factors (Cr) were calculated for each state to account for state-to-state differences in crash experience. The general form of the SPFs, including CMFs for intersections on rural two-lane roads, crash type and severity proportions, and calibration for local conditions, is as follows (see Figure 1):
Predicted crashes/yr = {exp[a + b(lnAADTmaj) + c(lnAADTmin)]} × CMFSkew × CMFLTL × CMFRTL × CMFLighting × PR1 × PR2 × Cr
Figure 1. Equation. Model for predicted crashes per year for specific crash types and severity levels, and accounting for local conditions.
where regression coefficients a, b, and c, CMF values, and default values for PR1 and PR2, if used, are provided in Chapter 10 of the HSM.
Using the results from the EB before-after crash analyses, economic analyses were then performed to estimate the benefit-cost ratios of each treatment. This ratio is the annual economic benefit divided by the annualized treatment cost. Benefit-cost ratios greater than one can be used to justify installation of a specific treatment. Crash reduction benefits were calculated using the HSM predictive methodology to estimate the number and severity of crashes that would be expected without the installation of the treatment and then applying the CMFs developed in this research to determine the expected annual change in crash frequency and severity. The economic values associated with crashes at each severity level were those presented in Chapter 7 of the HSM. A range of AADT values for the major and minor roads, representing the range of AADTs observed at the sites included in the analysis, was used to generate a matrix of possible crash reduction scenarios. Crash costs (by severity level) from the HSM were applied to the change in crashes expected due to the treatment to determine the economic benefit of the treatment. The treatment installation, maintenance costs, and service life cycles used in the analyses were gathered from state DOTs. In some cases, upper and lower ranges of treatment costs and/or service life cycles were used to determine the range of benefit-cost ratios that might be expected for a treatment.
Single luminaires are used to reduce nighttime crashes by making drivers aware of the presence of an intersection that may otherwise be difficult to see at night. The luminaire may be pole-mounted near one corner of the intersection or wire-mounted over the intersection (see Figure 2).
For drivers on a stop-controlled approach, increasing the visibility of the intersection may provide additional time to perceive the need to stop. For drivers on the uncontrolled approach, lighting may increase the visibility of vehicles, pedestrians, and bicyclists entering the roadway at that location and thus provide additional time to react appropriately.
Figure 2. Photo. Examples of single luminaire
intersection lighting (Image Credit: Google Earth™
Mapping Service).(7)
The EB before-after crash analysis was based on data from 27 treatment and 61 nontreatment intersections in Minnesota. Data for both 3- and 4-leg intersections were combined in the analysis to maximize the available sample size. Basic intersection descriptives and crash summaries are provided in Table 1.
Intersection Characteristics | 3- and 4-Leg Intersections | ||
---|---|---|---|
Treatment | Nontreatment | ||
Before | After | ||
Number of sites | 27 | 27 | 61 |
Number of years | 5 | 1 to 3 | 8 to 9 |
Number of site-years | 135 | 43 | 509 |
AADTmaj range (veh/day) | 200 to 7,300 | 220 to 5,900 | 145 to 3,900 |
AADTmin range (veh/day) | 30 to 2,000 | 30 to 2,000 | 65 to 2,150 |
Number of total nighttime crashes | 20 | 1 | 27 |
Number of FI nighttime crashes | 7 | 1 | 9 |
Number of PDO nighttime crashes | 13 | 0 | 18 |
The EB before-after crash analysis was performed for nighttime crashes at three severity levels: total, fatal and injury (FI), and property damage only (PDO).
The EB analysis results indicated that installing a single luminaire reduced total nighttime crashes by 71 percent (standard error of 0.29 percent); this safety effect was statistically significant at the 95-percent confidence level. An estimate could not be calculated for FI and PDO nighttime crashes due to the small number of target crashes in the dataset.
Benefit-cost ratios ranged from 0.5 to 35.0, assuming an $8,000 installation cost and $300 annual energy cost over a 20-year life, and using the CMF for total nighttime crashes (0.29) calculated by this research. The benefit-cost ratio exceeded 1.0 for intersections at which the major road carried at least 300 veh/day and the minor road carried at least 150 veh/day, and for all intersections at which the major-road AADT was at least 1,000 veh/day, regardless of the minor-road volume.
Single luminaires should be considered at intersections with a high proportion of crashes occurring during hours of darkness, or simply at intersections with a moderate to high frequency of nighttime crashes. Intersections with patterns of nighttime crashes that suggest drivers are unaware of the presence of the intersection (such as near horizontal or vertical curves or in the case of significant intersection skew) may especially benefit from this treatment.
If there is vegetation near the intersection, foliage should be trimmed and maintained on a regular basis so it does not cause shadows or reduce the visibility generated from the luminaire. Luminaire poles should have a breakaway design and should be located to minimize the risk of being struck by a vehicle.
Annual energy costs can be substantially reduced if solid-state LED luminaires are used, and solar-powered luminaires eliminate the need for a wired power source near the intersection.
Transverse rumble strips are placed in the travel lane perpendicular to the direction of travel. When used on the approach to a stop-controlled intersection, they are often applied in two or more sets of individual closely-spaced strips, and often used to supplement stop-ahead warning signs. They are designed to generate noise and vibration in the vehicle as the vehicle passes over them to alert the driver to the stop condition ahead. Transverse rumble strips may be rolled or grooved into asphalt, formed into fresh concrete, or created as epoxy strips on the surface of the pavement. Figure 3 shows an example of rumble strips milled into asphalt.
Figure 3. Photo. Example of transverse rumble
strips placement (Image Credit: Google Earth™ Mapping
Service).(7)
The EB analysis was based on before and after traffic volumes and crash data from 72 treatment and 126 nontreatment sites in five states-Arkansas, Kansas, Missouri, North Dakota, and Oregon-separately for 3- and 4-leg intersections. Basic intersection descriptives and crash summaries are provided in Table 2.
Three crash types were analyzed: all collision types combined, angle crashes, and rear-end crashes, and each crash type was broken down by severity level. Analysis results are shown in Table 3, for crash types and severity levels with statistically significant reductions in crashes at the 90 percent confidence level or higher.
Assuming a $5,000 installation cost per approach and a 20-year life cycle, benefit-cost ratios ranged from 1.1, for a major-road AADT of 200 veh/day and a minor-road AADT of 90 veh/day, to more than 50 when the major-road AADT exceeded 5,000 veh/day. When a $1,000 installation cost per approach and a 5-yr service life were assumed, benefit-cost ratios over the same range of AADTs ranged from 4.1 to more than 200. These results show transverse rumble strips to be highly cost effective at reducing crashes at rural two-lane intersections, even at very low volumes.
Transverse Rumble Strips (Arkansas, Kansas, Missouri, North Dakota1, and Oregon) | 3-Leg Intersections | 4-Leg Intersections | ||||
---|---|---|---|---|---|---|
Treatment | Non-treatment | Treatment | Non-treatment | |||
Before | After | Before | After | |||
Number of sites | 25 | 25 | 27 | 47 | 47 | 99 |
Number of years | 4 to 5 | 2 to 9 | 9 to 14 | 3 to 5 | 1 to 10 | 8 to 16 |
Number of site-years | 123 | 92 | 313 | 227 | 212 | 1,245 |
AADTmaj range (veh/day) | 245 to 8,900 | 255 to 11,700 | 245 to 8,900 | 165 to 6,700 | 198 to 6,700 | 65 to 15,900 |
AADTmin range (veh/day) | 110 to 6,000 | 120 to 7,000 | 110 to 6,000 | 65 to 4,118 | 85 to 3,547 | 65 to 4,118 |
Number of total crashes | 30 | 31 | 133 | 121 | 227 | 624 |
Number of FI crashes | 12 | 11 | 6 | 68 | 105 | 353 |
Number of PDO crashes | 18 | 20 | 70 | 53 | 122 | 271 |
Number of angle crashes | 3 | 5 | 9 | 73 | 135 | 333 |
Number of rear-end crashes | 4 | 9 | 25 | 16 | 30 | 78 |
1 No control intersections were available in North Dakota; nontreatment sites from Nebraska were used in the analysis. |
Crash Type | Severity Level | Number of Legs | Safety Effectiveness (%) | SE of Treatment Effect (%) | Statistically Significant at: |
---|---|---|---|---|---|
All | FI | 3 | -37 | 20 | 90% CL |
Angle | PDO | -61 | 28 | 95% CL | |
Rear-end | FI | -60 | 29 | 95% CL | |
All | Total | 4 | -13 | 7 | 90% CL |
FI | -29 | 8 | 95% CL | ||
Angle | FI | -25 | 10 | 95% CL | |
Rear-end | Total | -56 | 8 | 95% CL | |
FI | -78 | 8 | 95% CL | ||
PDO | -54 | 10 | 95% CL |
Transverse rumble strips should be considered for stop-controlled approaches to intersections where crash patterns indicate that drivers fail to recognize the stop condition (e.g., angle crashes related to stop sign violations). Rumble strips may be especially effective on the stop-controlled approach to an intersection that is hidden from view due to horizontal or vertical curvature or that follows a long tangent section. The proximity of the intersection to nearby residences or businesses should be considered, as the treatment generates noise that residents may not appreciate. Typically, transverse rumble strips are considered for implementation after less intrusive measures have been tried and failed to improve the crash experience at an intersection.
Transverse rumble strips may have undesirable effects that should be considered prior to implementation including: potential loss-of-control problems for motorcyclists and bicyclists; difficulties associated with snowplow operations; and inappropriate driver responses such as using the opposing travel lanes to drive around the rumble strips .(8)
This study considered two distinct types of supplementary pavement markings: markings on stop-controlled approaches (STOP AHEAD pavement markings) and markings on uncontrolled approaches (supplementary speed limit message [e.g., "SLOW, XX MPH,"] used with intersection ahead symbol [┼, ├, or ┤]), shown in Figure 4. In the first case, the pavement markings are used to alert drivers of the upcoming stop condition, while in the second case, the markings are used to alert drivers of the presence of the intersection and of the potential for vehicles and other road users entering or exiting the roadway at that location.
Figure 4. Photo. Aerial view and street view of
supplementary pavement markings on stop-controlled approach (Image
Credit: Google Earth™ Mapping Service)(7)
and supplementary pavement markings on uncontrolled approach (Image
Credit: Pennsylvania DOT)
The safety effectiveness of supplementary pavement markings was evaluated separately for markings installed on stop-controlled approaches of intersections and those installed on uncontrolled approaches since the two types of installations are very different in their mechanism for potentially reducing crashes.
The EB analysis was based on before and after traffic volumes and crash data from 76 treatment and 140 nontreatment sites in four states-Arkansas, Minnesota, Nebraska, and Vermont -separately for 3- and 4-leg intersections. Descriptive statistics of study sites are shown in Table 4.
Three crash types were analyzed: all collision types combined, angle crashes, and rear-end crashes, and each crash type was broken down by severity level. Data for 3- and 4-leg intersections were analyzed separately. Analysis results are shown in Table 5, for crash types and severity levels with statistically significant reductions in crashes at the 95 percent confidence level or higher.
The EB analysis was based on before and after traffic volumes and crash data from 11 treatment and 28 nontreatment sites in Pennsylvania. Descriptive statistics of study sites are shown in Table 6.
Three crash types were considered in this analysis: all collision types combined, angle crashes, and rear-end crashes. Data for both 3- and 4-leg intersections were combined for analysis. Results are shown in Table 7, for crash types and severity levels with statistically significant reductions in crashes at the 95 percent confidence level or higher.
Supplementary Pavement Markings Installed on
Stop-Controlled Approaches (Arkansas, Minnesota, Nebraska, and Vermont) |
3-Leg Intersections | 4-Leg Intersections | ||||
---|---|---|---|---|---|---|
Treatment | Non-treatment | Treatment | Non-treatment | |||
Before | After | Before | After | |||
Number of sites | 35 | 35 | 54 | 41 | 41 | 86 |
Number of years | 3 to 5 | 1 to 16 | 8 to 20 | 3 to 5 | 2 to 16 | 13 to 20 |
Number of site-years | 169 | 321 | 812 | 192 | 398 | 1,393 |
AADTmaj range (veh/day) | 90 to 4,700 | 102 to 5,700 | 90 to 8,900 | 105 to 3,900 | 134 to 4,800 | 105 to 6,600 |
AADTmin range (veh/day) | 40 to 2,395 | 40 to 3,000 | 40 to 5,800 | 25 to 1,766 | 25 to 1,500 | 25 to 2,510 |
Number of total crashes | 25 | 32 | 293 | 49 | 100 | 497 |
Number of FI crashes | 11 | 15 | 128 | 28 | 57 | 23 |
Number of PDO crashes | 13 | 17 | 162 | 18 | 43 | 227 |
Number of angle crashes | 3 | 3 | 47 | 23 | 46 | 257 |
Number of rear-end crashes | 1 | 2 | 50 | 7 | 12 | 63 |
Crash Type | Crash Severity | Number of Legs | Safety Effectiveness (%) |
SE of
Treatment Effect (%) |
Statistically Significant at: |
---|---|---|---|---|---|
All | Total | 3 | -67 | 7 | 95% CL |
FI | -76 | 7 | 95% CL | ||
PDO | -72 | 7 | 95% CL | ||
Angle | Total | 3 | -92 | 5 | 95% CL |
FI | -88 | 7 | 95% CL | ||
Rear-end | Total | 3 | -95 | 4 | 95% CL |
FI | -96 | 5 | 95% CL | ||
PDO | -97 | 3 | 95% CL | ||
All | Total | 4 | -66 | 4 | 95% CL |
FI | -69 | 5 | 95% CL | ||
PDO | -77 | 4 | 95% CL | ||
Angle | Total | 4 | -74 | 4 | 95% CL |
FI | -71 | 5 | 95% CL | ||
PDO | -88 | 3 | 95% CL | ||
Rear-end | Total | 4 | -89 | 3 | 95% CL |
FI | -86 | 5 | 95% CL | ||
PDO | -95 | 2 | 95% CL |
Supplementary Pavement Markings Installed on Uncontrolled Approaches (Pennsylvania) | 3- and 4-Leg Intersections | ||
---|---|---|---|
Treatment | Non-treatment | ||
Before | After | ||
Number of sites | 11 | 11 | 28 |
Number of years | 4 to 5 | 2 to 12 | 8 to 17 |
Number of site-years | 52 | 58 | 404 |
AADT maj range (veh/day) | 1,889 to 14,188 | 1,917 to 16,267 | 629 to 15,258 |
AADT min range (veh/day) | 184 to 5,377 | 183 to 4,242 | 173 to 5,377 |
Number of total crashes | 108 | 162 | 460 |
Number of FI crashes | 61 | 79 | 269 |
Number of PDO crashes | 47 | 83 | 191 |
Number of angle crashes | 69 | 125 | 275 |
Number of rear-end crashes | 14 | 20 | 58 |
Crash Type | Crash Severity | Safety Effectiveness (%) | SE of Treatment Effect (%) | Statistically Significant at: |
---|---|---|---|---|
All | Total | -46 | 5 | 95% CL |
FI | -49 | 7 | 95% CL | |
PDO | -50 | 6 | 95% CL | |
Angle | Total | -38 | 7 | 95% CL |
FI | -42 | 8 | 95% CL | |
PDO | -35 | 10 | 95% CL | |
Rear-end | Total | -69 | 7 | 95% CL |
FI | -76 | 9 | 95% CL | |
PDO | -75 | 8 | 95% CL |
The objective of this study was to increase the deployment of low-cost treatments that reduce motor vehicle fatalities and injuries at intersections with minor-road stop control along rural two-lane roads. The safety effectiveness of three low-cost safety treatments was evaluated using the EB observational before-after safety evaluation analysis approach to estimate their expected effectiveness in reducing crashes:
In addition, economic analyses were performed to estimate the benefit-cost ratio of each treatment, incorporating safety effectiveness results from the EB analyses. The safety and economic analyses show that the treatments are effective in reducing crashes of different types and severity levels and are economically justifiable for installation at intersections with patterns of crashes that suggest drivers are unaware of the presence of the intersection. The results of this research can be combined with information on other strategies to reduce crashes at intersections with minor-road stop control along rural two-lane roads. With this information, agencies can make informed decisions to plan and program safety improvements at intersections under their jurisdiction.