Nearly 65% of road accidents in India occur on two-lane undivided rural highways. These highways carry mixed traffic, with fast-moving and slow-moving vehicles sharing the carriageway. Further, the geometric and road environment characteristics make these highways differ widely from the rural highways in developed countries. The present study is an attempt to developaccident prediction models (APM) for these rural highways, taking into account the effect of traffic composition. Data of 2815 accidents occurred on nearly 200 kilometres of two-lane undivided rural highway segments, overa period of three years, is used to develop the models. It is found that,in addition to traffic exposure, proportions of buses, trucks, motorized two-wheelers and non-motorized vehicles in the traffic significantly influence accident occurrence. Based on the models, measures like segregation of motorized two-wheelers and non-motorized traffic, controlling driveway density, provision of wider shoulders and straightening of curves are suggested for reduction of accidents on these highways.
Abstract