Safety Evaluation of Cable Median Barriers in Combination With Rumble Strips on Divided Roads

CHAPTER 1. INTRODUCTION

This chapter presents background information on the strategy of using cable median barriers and shoulder rumble strips, the goals of the study, and a review of the existing literature on the safety effects of cable median barriers, shoulder rumble strips, and the combination of the two treatments.

Background on Strategy

The United States began using cable median barriers in the 1960s as a treatment for preventing cross-median crashes. Initially, the design of the cable barriers was a low-tension type. Later, it became more common for States to use high-tension cable barriers. Villwock et al. reported the primary advantage of high-tension cable barrier as the ability to withstand multiple hits before requiring repair.⁽¹⁾ Many States have begun to use cable median barriers in place of more rigid (e.g., concrete) barriers. They have cited the following reasons:

• Contain vehicles instead of redirecting them into traffic; up to 95 percent of vehicles are contained.
• Are less expensive to install and maintain except for the danger of snow plow damage.
• Eliminate snow accumulation.
• Are environmentally non-intrusive and aesthetically pleasing and blend in with the terrain.

Although there are significant advantages to cable barriers, States have cited some disadvantages, including the following:

• Multiple studies indicate an increase in crash frequency. One study reported an increase in the injury crash rate, although the period for analysis was limited to 16 months.
• Penetrating the barrier is possible by under-riding or over-riding. The identified causes of under-riding include location of the barrier on a slope and having too large an opening between the cable and the ground caused by the presence of a ditch. One of the causes of over-riding is hitting a post.
• There are conflicting results regarding whether a cable barriers can stop a heavy truck.
• Considerable deflection upon impact prohibits the use of cable barrier in narrow medians. This concern is particularly justified on curves, where the deflection can be larger than on straight segments.⁽¹⁾

The purpose of introducing shoulder rumble strips is to reduce the frequency of run-off-road (ROR) crashes. Although agencies have conducted research into the performance of cable median barriers and shoulder rumble strips (specifically, rumble strips on the outside shoulder) separately, very few studies have looked at the combination of cable median barriers with inside shoulder rumble strips.

Background on Study

In 1997, the American Association of State Highway and Transportation Officials Standing Committee on Highway Traffic Safety—with the assistance of FHWA, the National Highway Traffic Safety Administration, and the Transportation Research Board Committee on Transportation Safety Management—met with safety experts in the field of driver, vehicle, and highway issues from various organizations to develop a strategic plan for highway safety. These participants developed 22 key areas that affected highway safety.

The National Cooperative Highway Research Program (NCHRP) then published a series of guides to advance the implementation of countermeasures targeted to reduce crashes and injuries. Each guide addresses 1 of the 22 emphasis areas and includes an introduction to the problem, a list of objectives for improving safety in that emphasis area, and strategies for each objective. Each strategy is designated as proven, tried, or experimental. Many of the strategies discussed in these guides have not been rigorously evaluated; approximately 80 percent of the strategies are considered tried or experimental.

FHWA organized a pooled fund study of 40 States to evaluate low-cost safety strategies as part of this strategic highway safety effort. The purpose of the FHWA Evaluation of Low-Cost Safety Improvements Pooled Fund Study (ELSCI-PFS) is to evaluate the safety effectiveness of several tried and experimental, low-cost safety strategies through scientifically rigorous crash-based studies. The ELSCI-PFS selected the use of cable median barriers in combination with rumble strips on the inside shoulder on divided roads as a strategy to evaluate as part of this effort.

Literature Review

This review is summarized in three sections: (1) safety effects of cable median barriers, (2) safety effects of shoulder rumble strips, and (3) safety effects of cable median barriers along with shoulder rumble strips.

Safety Effects of Cable Median Barriers

Villwock, Blond, and Tarko studied the effects of cable barriers on rural interstate highways.⁽¹⁾ They used data from eight States that had installed both low- and high-tension cable median barriers. Their data included 70 mi of treated roadway and 360 mi of control roadway (i.e., sections that did not receive cable barriers). They used a combination of before–after EB analysis with negative binomial and logistic regression to determine the effect on three types of crashes—single-vehicle, multiple-vehicle same-direction, and multiple-vehicle opposite-direction (cross-median or crossover crashes). Their results showed that the cable barriers affected the crash types differently. The overall crash modification factor (CMF) for single-vehicle crashes was 1.83, suggesting an increase in this crash type. The CMF for multiple-vehicle opposite-direction crashes was 0.06, suggesting an effective reduction of this crash type. They did not see a significant effect on multiple-vehicle same-direction crashes.⁽¹⁾

Cooner et al. evaluated installations of cable median barriers in Texas.⁽²⁾ Cross-median crashes resulted in a substantial percentage of interstate highway fatalities, which led to widespread installation of cable barriers. This research team had difficulty conducting a rigorous safety evaluation of the cable barrier; owing to a changeover in the management of crash data, the data were unavailable for the years preceding and following the installation. They relied on a simple before–after examination by the Texas Department of Transportation Traffic Operations Division of 407 mi that received cable barriers. This examination, which did not account for the potential effects due to regression-to-the-mean, examined at 1 year before and 1 year after installation at the treated sites only. They observed that head-on fatal crashes decreased from 14 to 1 during the study period.⁽²⁾

Chandler reported on the benefits of cable median barriers in Missouri.⁽³⁾ In 2002, the Missouri Department of Transportation (MoDOT) began to install cable barriers on freeways with median widths of less than 60 ft, focusing on two main interstate highways. From 2002 to 2006, a total of 179 mi of cable barriers were installed, with most of the installations occurring in 2005. From an examination of fatalities on one interstate highway, Chandler concluded that the installation of 179 mi of cable median on freeways nearly eliminated cross-median fatalities. Figure 1 and Figure 2 show a comparison between miles of cable barriers installed and cross-median fatalities. The cross-median fatalities decreased from a high of 24 per year to 2 per year by 2006, a decrease of 92 percent. Chandler concluded that cable median barriers were an effective safety tool in Missouri.⁽³⁾

Hunter et al. evaluated the safety of three-strand cable median barriers using data from a section of I-40 outside Raleigh, NC.⁽⁴⁾ The research team used data from 1990 to 1997 to conduct the analysis. The treatment population was the 8.5-mi section of Interstate 40, and the reference population was identified as the entire North Carolina Interstate System (except for those sections treated with cable median barriers). The research team also identified a secondary reference population to account for discrepancies in traffic volumes between the entire reference population and the treatment segment. They identified this subpopulation as a subset of sites where the annual average daily traffic (AADT) exceeded 50,000 vehicles per day. They developed several regression models, assuming a negative binomial error structure for many of the models, to estimate the effects of the installation of the cable median barriers. The models accounted for year-to-year variations (e.g., AADT and weather conditions).

Results showed that the installation of the three-strand cable median barriers was followed by a reduction in fatal and serious injury crashes. ROR left, hit-fixed-object, and rear-end crashes increased with the installation of the cable median barriers. The models predicted that ROR head-on crashes (i.e., the primary target for the treatment) would decrease with installation of the cable median barriers. However, the research team was not able to confirm this hypothesis because this particular crash type was very rare.⁽⁴⁾

Milton and Albin reported on the experience with using cable median barriers in Washington State.⁽⁵⁾ The authors calculated the total societal benefit of cable barriers as $420,000/mi annually. Their examination of the Washington State Department of Transportation bid history indicated a cost of cable median barriers at $44,000/mi, with an annual maintenance cost of $2,750/mi.⁽⁵⁾

Safety Effects of Shoulder Rumble Strips

Griffith studied the safety effects of rumble strips on freeway shoulders in California and Illinois.⁽⁶⁾ These States installed shoulder rumble strips on the inside and outside shoulders of the highway in both directions. Griffith examined crash data for 280 mi of freeway in Illinois and 120 mi of freeway in California with rumble strip installations. The author conducted a before–after analysis to determine the effect on crashes. Because the author selected the installation sites based on their listing in the resurfacing schedule rather than based on high crash frequency, there was no need to account for regression-to-the-mean bias. Thus, Griffith used a before–after design with a comparison group to account for trends and found that shoulder rumble strips reduced single-vehicle run-off-road (SVROR) crashes by 18.3 percent on all freeways and 21.1 percent on rural freeways. The author did not conduct a separate analysis to determine the effect of installing rumble strips on only the inside shoulder.⁽⁶⁾

Torbic at al. conducted one of the most recent studies on rumble strips under NCHRP Project 17‑32.⁽⁷⁾ The authors of the report included a review of prior research on the safety impacts of shoulder rumble strips. Table 1 shows an adaptation of table 4 of their report, which summarizes 20 studies that calculated the effect of shoulder rumble strips on crashes (including the study by Griffith previously discussed).^(6,7) Although there was no specific breakout of inside versus outside shoulder rumble strip effects, most studies in the list used roadway facility types that were divided (i.e., interstate highways, freeways and other divided highways). Most studies used SVROR crashes as the target crash type. The most common study design was before–after, including naive before–after and before–after with a comparison group.

Torbic et al. also conducted a survey on the rumble strip installation practices of transportation agencies. The results indicated that 93 percent of agencies installed shoulder rumble strips on both the inside and outside shoulder when installing on a median-divided roadway. Thirty-five percent of agencies indicated that they used different policies for inside shoulders versus outside shoulders. These policies included smaller offsets on the inside (median) shoulder and continuous rumble strips on the inside (median) shoulder compared with intermittent gaps on the outside shoulder.⁽⁷⁾

Torbic et al. conducted an evaluation of the safety effect of shoulder rumble strips according to roadway type, crash severity, placement of the rumble strip (i.e., on edge line versus not on edge line), and offset distance. The analysis did not attempt to separate the effects of inside shoulder versus outside shoulder rumble strips. When developing the dataset, the authors assumed the rumble strips on the inside shoulder were installed at the same time as rumble strips on the outside shoulder. Although SVROR crashes were a target crash type, the analysis of SVROR crashes did not distinguish between those to the left and right of the road.⁽⁷⁾

Torbic et al. conducted the safety evaluation using both the EB before–after study method and a generalized linear modeling cross-sectional approach. Ultimately, they used the EB results because EB was the preferred analysis method in resources such as the Highway Safety Manual.⁽⁸⁾ They combined the results of their safety evaluation with two other reliable studies to create CMFs for shoulder rumble strips. For rural freeways, the authors estimated that shoulder rumble strips led to an 11-percent reduction in SVROR crashes (standard error (SE) 6) and a 16‑percent reduction in SVROR fatal and injury crashes (SE 8). Their analysis did not indicate whether the sites that were treated with shoulder rumble strips had cable median barriers.⁽⁷⁾

Safety Effects of Cable Median Barriers with Shoulder Rumble Strips

Monsere et al. investigated the effects of more than 21.9 mi of continuous three-strand cable median barriers installed on Interstate 5 in Oregon in December 1996 (approximately 7 mi) and April 1998 (approximately 18 mi).⁽⁹⁾ In addition, the Oregon Department of Transportation installed milled shoulder rumble strips in fall 1998 in the same area. The crash analysis used two primary datasets: (1) reported crashes (recorded in the State’s computerized crash record system) and (2) maintenance logs that documented cable barrier strikes.

For the analysis using State-reported crash data, the authors could identify no suitable reference group, so a simple before–after study was conducted. They used 3 years of before and after data (with the exception of the transition period identified as January 1997 through April 1998). Results showed a decrease in fatal and serious injury cross-median crashes between the before and after time periods—three crashes in the before period and none in the after period and an increase in reported crashes striking a barrier. Although the analysis could not identify left and right side separately, there were 7 crashes in the before period and 60 in the after period. For the analysis using maintenance logs, the authors estimated the effectiveness of the barriers by examining the type of damage the cable median barriers sustained during the crash. Results showed that 105 potential crossovers (of 231 barrier impacts) were contained by the barriers between December 1996 and April 2002.⁽⁹⁾

The authors acknowledged that the State transportation department installed milled shoulder rumble strips around the same time as the second cable median barrier installation, which could have contributed to some of the results from the analysis.⁽⁹⁾

Summary of Literature Review

Overall, the studies seemed to indicate a reduction in cross-median crashes and an increase in crashes involving the striking of a barrier following the installation of cable median barriers. Regarding the effect of shoulder rumble strips, it was evident that they were effective in reducing SVROR crashes. The study by Monsere et al. was the only one that looked at the combination of cable barriers and shoulder rumble strips.⁽⁹⁾ However, that study was a simple before–after evaluation and therefore did not account for possible bias due to regression-to-the-mean and the effect of changes in traffic volume. In addition, the sample of crashes was much too small to obtain robust results.⁽⁹⁾