Safety Evaluation of Signalized Restricted Crossing U-Turn Intersections

Chapter 7. Conclusions

There are theoretical reasons to believe that signalized RCUT intersections would be safer than similar conventional signalized intersections, and previous research has shown that unsignalized RCUTs are generally safer than conventional unsignalized options. However, there has never been a study of the safety of signalized RCUT intersections. Therefore, the objective of this evaluation was to develop a CMF for signalized RCUT intersections and examine injury crashes, spatial patterns, and other crash variables.

This evaluation collected and analyzed crash data before and after conversion of 11 intersections from conventional to RCUT design. The intersections were in suburban areas along four-lane or six-lane arterials. Available data included more than 2,000 crash reports at the treatment sites over 65 years of intersection operation, including more than 700 injury crashes. Analyses adjusted for changes in traffic volumes. The project team also collected data at 44 potential comparison sites for use in adjusting for simultaneous event and maturation biases.

For most individual sites and groups of sites examined, odds ratio tests showed that there were high-quality comparison sites available, which enhanced the strength of the analyses. Therefore, this evaluation determined the following as the best general estimates of CMFs for conversion of a conventional intersection to an RCUT intersection:

• Overall crashes: CMF = 0.85 (CRF = 15 percent).
• Injury crashes: CMF = 0.78 (CRF = 22 percent).

The SDs of the CMFs were 0.16 and 0.20, respectively. This indicates that the CMF for overall crashes was not significantly different from a neutral value of 1.0 at a 68-percent confidence level but that the CMF for injury crashes was significantly different from a neutral value of 1.0 at a 68-percent confidence level. Regardless, the results support the assumption that a signalized RCUT will generally produce a crash reduction. Also, the fact that RCUTs likely save more injury crashes than overall crashes should not be surprising, since they generally reduce the more severe angle and turning crashes.

The evaluation also produced an estimated B/C ratio for installing an RCUT at the set of test intersections of 3.6 to 1.0 when considering safety and operations or 2.6 to 1.0 considering safety only. When examining the sensitivity of this result to changes in crash costs, the B/C ratio always exceeded 1.0. Installing signalized RCUT intersections at locations similar to those studied should generally lead to positive results in terms of expected crash reductions.

At the individual site level of analysis, 8 of the 11 sites showed decreases in overall and injury crashes after RCUT installation. The three sites with increases (Ohio-Symmes, Texas-Evans, and Texas-Stone Oak) were the only treatment sites with three lanes on both minor street approaches. The only other treatment sites with three lanes on minor street approaches were Texas-Shaenfield, a T-intersection, and the other two Ohio sites with three lanes on one minor street approach and two lanes on the other minor street approach. Therefore, it is likely that signalized RCUTs may be relatively safer when the minor streets are narrower and/or carry lower traffic volumes.

There were clusters of rear-end crashes on the major streets of the RCUTs. An examination of crash types before and after RCUT installation showed that there was generally a conversion from angle crashes to sideswipe crashes.

While this study provided an estimated CMF and B/C ratio for signalized RCUTs, there is still a need for future research to add more knowledge. The project team recommends a followup before–after study with more sites, in more States, over more years, since RCUT installation is accelerating, and such a larger study will be possible in a few years. A larger study may shed more light on the spatial crash patterns and other variables that this study could only touch upon, as well as the circumstances that best favor RCUT installation. A second promising area for future research would be a study with a much larger sample size of sites where a research team could assemble a model of the safety effects of some of the important geometric features of RCUTs. Finally, the project team recommends studies similar to this one for the installation of other types of alternative and conventional intersections. A CMF for signalized RCUTs is only helpful if designers can compare it to CMFs for other intersection forms such as median U-turn, quadrant roadway, and continuous flow intersections, among others.