Simulated Conflict Based Safety Evaluation Models for Hetergenous Traffic in Controlled Intersections

In this paper, an attempt is made to investigate how traffic conflicts identified from microsimulation models can be correlated with explanatory variables which have been traditionally used in accident prediction models. In developing countries with heterogenous traffic streams, availability of accident data is limited especially since accidents are rare events. Such traffic streams normally have some unique attributes like absence of lane discipline, presence of non-motorized vehicles. In urban intersections with such slow-moving traffic streams, conflicts are more useful determinants of intersection safety rather than previous records of accidents since geometry of intersection may be changed from the time to time. Simulated conflict-based safety evaluation models were developed for intersections of Dhaka city. The intersections were modeled in VISSIM after suitable calibration, for 8 hours of peak hour traffic. Surrogate Safety Assessment Model (SSAM) was used to identify the corresponding simulated hourly conflicts from the resulting trajectory files. It was found that hourly simulated conflicts had a significant statistical relationship with observed hourly traffic volume entering the intersection from major and minor roads. Increasing volumes of non-motorized traffic was found to contribute to intersection safety