Accident Analysis at Railroad-Highway Grade Crossings in Urban Areas : Technical Paper

Abstract

The purposes of this research Investigation were to develop mathematical models that measure the relative safety or hazard of urban grade crossings and to establish a priority rating system, based on these models, for determining protection improvements in urban areas. The mathematical techniques of discriminant analysis and regression analysis were utilized to develop discriminant models with linearly assigned probabilities. These models permit potential hazards to be expressed as the probability that a grade crossing is accident prone. A grade crossing where a vehicle-train accident had occurred during a two-year period was considered as a representative member of the population of accident prone crossings. A location which had not experienced a vehicle-train accident for at least five years prior to the date of the field investigation was assumed as a representative member of the population of non-accident prone grade crossings. Data were collected at 295 accident locations and 281 non-accident locations in urban portions of the State of Indiana. The best discriminant model is 74-percent successful in classifying the true group membership of the sample grade crossings. This model expresses potential hazard as a function of protective device, average daily highway traffic, average daily train traffic, degree of effective sight distance, and roadside distractions. A methodology was developed for selecting a minimum level of grade crossing protection and for establishing priorities for the improvement of protection at urban railroad-highway grade crossings.