IDENTIFYING HIGHWAY-RAILWAY GRADE CROSSING BLACK SPOTS: PHASE 1

Frank Saccomanno, Liping Fu, C Ren, Lia D’Afonsêca Pedreira de Miranda

2003 · 14 citations · 0 references

Concepts

Abstract

This report presents a risk-based model for identifying highway-railway grade crossing black spots in Canada. This model consists of two prediction components: 1) collision frequency and 2) collision consequence. A graphical approach is adopted to identify crossings with unacceptable risk (high expected frequencies and/or consequences). These crossings are referred to black spots. The model was applied to Canadian inventory (IRIS) and collision occurrence (RODS) data for the period 1993-2001. Poisson and Negative Binomial (NB) frequency prediction expressions were developed for crossings with three types of warning devices (crossings with signs, flashing lights and/or gates). Both Poisson and NB models were found to provide a good fit to the collision frequency data. A weighted consequence score was introduced to represent combined collision severity. The weights used in this combined consequence score were obtained from reported insurance claims for fatalities, personal injuries and property damages. A NB expression was developed for the collision consequence model. The spatial distribution of black spots is discussed with respect to the type of warning device, upgrades in warning device, geographical location, and historical collision occurrence. A list of black spot crossings is provided for the Canadian data based on crossings whose expected number of collisions and/or expected severity score is exceeded at least 0.1% of the time. The regional breakdown of these crossings is discussed with respect to a selected number of attributes and historical collision experience. A GIS platform was developed for the Ontario region and used to illustrate the spatial pattern of expected and historical collision frequency and associated black spots.