MODELLING OF ACCIDENTS FOR URBAN AREAS

Abstract

Traffic accidents have been considered as a major cause of death in both developed and developing countries. Traffic accidents usually take place as a result of the complex interaction of environment and operational factors in time and space. Number of attempts have been made by various researchers in various parts of the world in developing accident prediction models for the type of accidents considering geometric characteristics, traffic control etc. Stochastic modeling techniques like Poisson and Negative Binomial were found most reliable methods in relating accident and various causative factors as compared to linear regression due to discrete nature of accident. In India the rapid growth in vehicular population and lack of adequate transportation facilities has caused the increase in number of fatalities. Therefore it has become necessary to study relationship between accidents and causative factor so as to take remedial measures. In India models have been developed by considering the vehicle ownership and population as independent variables. But very few attempts have been made to relate accidents with geometric parameter of the road. For this study accident situation in Delhi was studied. An attempt was made to find the relation between accident severity and variables describing the geometry of selected selections. For developing accident prediction models for intersection 81, 3-leg and 72, 4-leg intersections were considered whereas for mid —block section models 45, mid —block sections were considered. The Negative Binomial Regression distribution was found most appropriate for modelling accidents. Models were developed for predicting fatal and injury accident& For intersection, it was found that angle of intersection was most significant variable. It was found that with every one degree increase in angle of intersection increases the likelihood of injury accidents by 4.3% for 3 —leg intersections. Also it was observed that the provision of signal reduces the accidents by 38% for 3 —leg intersections. In case of mid —block section length was most significant. The every lunit increase in section length will increase accident by 32%.