The Integration of Conflict Probability and Severity for the Safety Assessment of Intersections

The use of traffic crash data-based methodologies for safety evaluation is inadequate due to the shortcomings such as unavailability and low quality of historical crash data. Other than crash data-based analysis, the use of empirical vehicle maneuvers and/or the development of micro-simulation models in conjunction with surrogate safety measures has been shown to potentially complement traditional safety analysis. However, several previous works found that existing surrogate measures for intersection safety assessment, such as post-encroachment time, time to collision and speed, fail to simultaneously represent conflict probability and severity. Thus, this study proposes a measurement of the crash hazard that considers crash occurrence probability as well as expected severity. By utilizing the change in the total kinetic energy before and after the collision, angle of collision and PET, the proposed conflict index is derived and its implications are discussed. Several video-recorded signalized intersections in Nagoya, Japan, were utilized to extract vehicle trajectories, through which conflict characteristics are estimated. The relationship between the estimated distributions of the proposed index and the records of severe crashes at the corresponding sites are compared. The proposed safety measure is successful in similarly ranking different signalized intersection to the severity of crashes that occurred at each site. The author expects the new safety assessment measure can be useful in assisting policy makers in prioritizing different sites for safety improvements by identifying hazardous locations which currently lack accurate and historical crash data.