Exploring the relationship of conflict characteristics and consequent crash severity

This paper outlines the development of a modelling approach to measure the occupant injury severity of two vehicle crashes based on the characteristics of the conflict. A combination of statistical analysis and numerical analysis based on Newtonian Mechanics is utilised to develop this modelling framework. The Australian Crash In-Depth Study (ANCIS) database is used to estimate and validate the model. Three sequential steps were undertaken to develop the model. In the first step, the driver reaction before a crash was determined according to the type of conflict, weather condition, driver's gender and speed limit of the vehicles. A Binary Probit Model was utilised to measure the driver reaction before the crash. In the second step, the speed change of the subject vehicle (ΔV_s) was estimated using conflict characteristics and driver reaction in conflict. A Log-Gamma regression model was used to estimate ΔV_s. Newtonian mechanics was used to estimate the kinetic energy applied to the subject vehicle according to the mass and estimated ΔV_s. In the third step, a Log-Gamma regression model was used to estimate the Injury Severity Score (ISS) of the conflict based on estimated kinetic energy of the subject vehicle, the impact type of the crash, presence of airbag, presence of seat belt and age of the occupant. The modelling approach can be incorporated into a micro simulation model to provide more accurate safety assessment of road locations based on conflict analysis.