The evaluation of road-rail crossing safety with limited accident statistics

Safety evaluation is an essential issue in ranking road-rail crossings as candidates for grade-separation. Many small countries like Israel do not possess sufficient data to generate statistical models similar to the US DOT accident prediction formula or others. At the same time, it is desirable to provide estimates stemming from local conditions. For these, available Israeli accident data were artificially enlarged using the unification of accident statistics and information from crossings functioning over a six-year period. A hazard index serves as a basic evaluation tool. The datasets on accidents and crossings are split according to several crossing characteristics (category of warning device, volume of vehicle traffic, volume of train traffic, visibility conditions); the obtained values are combined to supply safety estimates for crossing types defined by these characteristics. The validity of model performance is explored. For Israeli conditions the model provides for a safety evaluation of 168 crossing types. This presents a sufficient base from which to estimate the accident potential of any local crossing when the need for its grade separation is discussed.