Estimating vehicle accidents on British Columbia roads

Five quantitative relationships that combine some geometric features of two-lane rural highways to accident rates were reported in Special Report 214 of the U.S. Transportation Research Board. This study set out to test the applicability of the accident rate models to the highways of British Columbia. Data limitations reduced the models to those for roadway clearances, bridge widths and horizontal curves. Two segments of road totalling about 300 kilometres and accident data from 1981 to 1985 inclusive were used for the analysis. The analysis found that the SR 214 bridge-accident model was completely at odds with the B.C. data. The horizontal curve model was adequate but due to the arbitrary nature of the constant accident rate per degree of curve, the results must be used carefully. Also, the model worked best for curves of 1 km or more in length. The road clearance-accident model worked well for truly rural two-lane highways, provided the segments of route under study were reasonably long, about 5 km or more. The two-lane highways that have many local passing improvements were found not to behave as expected. The models of SR 214 are a step in the right direction. The application of such models to highway improvement cost analysis still needs considerable research and refinement.