Safety Evaluation of Dilemma Zone Protection System for Rural, High-Speed Signalized Intersections Using Empirical Bayes Method

This study employed the empirical Bayes before-and-after crash analysis for the safety assessment of the continuous wide area (CWA) sensor-based dilemma zone protection (DZP) system deployed at multiple rural, high-speed signalized intersections in Alabama. Six intersections treated with the DZP system were selected as the treatment group, while thirty-three untreated intersections with similar characteristics with regard to traffic, geometry, and speed limit were selected as the reference group. Safety performance functions and crash modification factors (CMFs) were developed using police-reported crash data. Red-light running (RLR) crashes were identified from the crash database by categorizing crashes into two groups (strict definition [SD-RLR] and extended definition [ED-RLR] crashes) for a comprehensive safety analysis. The analysis results showed that SD-RLR and ED-RLR crashes would decrease by 35% (CMF = 0.65, SE = 0.06) and 24% (CMF = 0.76, SE = 0.05), respectively, if a rural high-speed intersection is treated with the CWA sensor-based DZP system.