PREDICTION OF ACCIDENTS AT UNSIGNALIZED INTERSECTIONS AFTER SAFETY IMPROVEMENT

Abstract

In any roadway system, intersections are the major bottleneck point. In India, in the year 2016, there was a total number of 1,39,389 accidents occurred at intersections and the uncontrolled ones contributed to a major portion of road accidents underscoring the importance of traffic control mechanism at junctions (MORTH 2016).The prime objective of this research is to develop a methodical exploration of the prospects for accepting a more mechanistic approach to unsignalized intersection’s safety based on microscopic simulation of safety performance.For accomplishing the objectives proposed in this thesis, Simulation models of the intersection sites are developed for heterogenous non-lane-based traffic condition in PTV Vissim. A .trj file is then generated by running simulation in Vissim. This .trj file is then used as input to SSAM Software to find the number of conflicts possible in the developed simulation models. A conflict based accident prediction model (APM) is then developed by taking number of accidents as dependent variable and the conflicts generated as explanatory variable using negative binomial numerical method and Bayesian approach. Suitable countermeasures (e.g-speed tables, lane narrowing) for safety assessment was then applied in the developed simulated models in Vissim. Again by using SSAM software new conflicts were obtained and from these new conflicts changed number of accidents was calculated from the already developed conflict based APM to check the effectiveness of applied countermeasures. The results of the study show that speed tables prove to be more effective traffic calming measures than lane narrowing for unsignalized intersection