Abstract
Train accident rates are a critical metric of railroad transportation safety and risk performance. Understanding factors affecting accident rates is also important for evaluating the effectiveness of various accident prevention measures. Accident rates have been the subject of a number of analyses, but these analyses have generally not considered the effect of train length on accident rate. Accident causes can be classified into two groups: those dependent on train length and correlated with the number of cars in the train, and those independent of train length, corresponding to the number of train miles operated. These classifications have implications for the quantitative effect of various changes in railroad operating practices on railroad safety performance. Whether an accident cause is a function of car miles or train miles affects how safety measures that might reduce the likelihood of that cause will affect overall train accident rate. Accident causes have been classified as car mile or train mile related based on expert opinion, but these classifications have not been quantitatively tested. FRA accident data were used to develop a metric to classify objectively accident causes, and 11 causes were reclassified from the previous classification. Based on the results of the study, a sensitivity analysis was conducted to evaluate how changes in train length affect individual trains’ accident likelihood and systemwide accident rate. The concept of causes of car mile versus train mile accidents leads to the premise that, although longer trains are expected to experience more accidents than shorter trains, operation of longer trains results in a lower system-level accident rate.