Behavior-Based Safety Evaluation Model of Vehicles Turning Left at Intersections with a Permitted Left-Turn Phase

Intersections with a permitted left-turn phase have serious safety hazards as left-turning vehicles compete for the right of way with vehicles approaching from the opposite direction. This is a recurring problem in some developing countries. This study develops quantitative methods for assessing the driving safety of left-turning vehicles at intersections, based on data collected at four sites in Xi'an, China. The turning behaviors of left-turning vehicles were classified into three patterns, no-interference, yield, or rush, based on the interactions with the vehicles approaching from the opposite direction. This study established a vehicle-driving safety evaluation model through statistical analysis of kinematic characteristics and obtained the safety level distribution of the three turning behavior patterns. In addition, the intersection was divided into three areas, combined with the driving paths of the left-turning vehicles, and a safety evaluation model based on the entropy weight technique for order preference by similarity to an ideal solution (TOPSIS) method was proposed to calculate the safety level of left-turning vehicles in each area. The results showed that the rush pattern was judged to be the most dangerous turning behavior according to the driving safety level. Further, for vehicles turning left from west to north, the area between the north curb line and the north entry stop line is the most dangerous driving area. This study will help to understand the operational safety characteristics of left-turning vehicles from the spatial perspective and provide ideas for drivers to choose driving strategies and for vehicle manufacturers to design driver assistance systems.