The overall goal of this research was to quantify the safety performance of alternative traffic control strategies to mitigate right-turning vehicle-bicycle crashes at signalized intersections in Oregon. The ultimate aim was to provide useful design guidance to potentially mitigate these collision types at the critical intersection configurations. This report includes a comprehensive review of more than 150 scientific and technical articles that relate to bicycle-motor vehicle crashes. A total of 504 right-hook crashes were identified from vehicle path information in the Oregon crash data from 2007-2011, mapped and reviewed in detail to identify the frequency and severity of crashes by intersection lane configuration and traffic control. Based on these efforts, a two stage experiment was developed in the OSU high-fidelity driving simulator to investigate the causal factors of right-hook crashes at signalized intersections with a striped bike lane and no right-turn lane, and to then identify and evaluate alternative design treatments that could mitigate the occurrence of right-hook crashes. Experiment 1 investigated motorist and environmental related causal factors of right-hook crashes, using three different motorist performance measures: 1) visual attention, 2) situational awareness (SA) and 3) crash avoidance behavior. Data was collected from 51 participants (30 male and 21 female) turning right 820 times in 21 different experimental scenarios. It was determined that the worst case right-hook scenario occurred when a bicycle was approaching the intersection at a higher speed (16 mph) and positioned in the blind zone of the motorist. In crash and near crash situations (measured by time-to-collision) the most common cause was a failure of the driver to actively search for the adjacent bicyclist (situational awareness level 1), although failures were also determined to occur due to failures of projection (i.e. incorrectly assuming that the bicycle would yield or that there was enough time to turn in front of the bicycle). Elements of driver performance and gap acceptance collected in the first stage simulator experiment were field validated to provide additional confidence in the findings. The research reviewed 144 hours of video and identified 43 conflicts where the time-to-collision (TTC) measured less than 5 seconds. When field observations of scenarios most similar to those in the simulator were isolated, the analysis indicated that the distribution of the TTCs values observed in the simulator were consistent with those observed in the field. Experiment 2 evaluated several possible design treatments, (specifically: signage, pavement markings, curb radii, and protected intersection designs), based on the visual attention of motorist, their crash avoidance behavior, and the severity of the observed crashes. Data was collected from 28 participants (18 male and 10 female) turning right 596 times in 22 scenarios that were used. The resulting analysis of the driver performance indicators suggest that while we can measure the various driver performance metrics robustly, and all of the treatments had some positive effect on measured driver performance, it is not yet clear how to map the magnitudes of the differences to expected crash outcomes. Additional work is recommended to address the limitations of this study and to further consider the potential effects of the right-hook crash mitigation strategies from this research. (Author/publisher)
Abstract