Stochastic approach for modeling the effects of intersection geometry on turning vehicle paths

Analytical evaluation techniques for the safety performance of signalized intersections are applicable to limited scenarios and conditions, whereas simulation-based analysis tools are very flexible and promising. This study is part of intensive efforts to develop a microscopic simulation model for the safety assessment of signalized intersections. One important aspect of analyzing driver maneuver is vehicle paths. Broadly varying paths may result in widely distributed potential conflict points with other movements, which may affect the occurrence probability of severe conflicts. Therefore, this study aims to develop a technique to reproduce the variations in the paths of turning vehicles, considering intersection geometry, vehicle type, and speed. Several signalized intersections in Nagoya City, Japan, with various traffic and geometric characteristics were videotaped. The analysis revealed that the paths of right-turning vehicles (left-hand traffic) are more sensitive to the vehicle speed and turning angle whereas those of left-turning vehicles are more sensitive to the intersection corner radius, turning angle, and vehicle speed. For modeling individual vehicle paths, this study applies the Euler-spiral-based approximation methodology where each trajectory is fitted by an entering Euler spiral curve followed by a circular curve and an exit Euler spiral curve. The proposed models are unique since they provide a realistic and rational representation of the variations in turning vehicles’ paths inside intersections.

► Effects of various factors on turning vehicle paths are empirically analyzed.
► Intersection geometry significantly affects the distribution of vehicle paths.
► Vehicle approaching speed significantly affects the distribution of vehicle paths.
► Variations in vehicle paths significantly affect the distribution of conflict points.
► A unique model that can reflect the variations in vehicle paths is proposed.