Development and application of a probabilistic risk assessment model for evaluating advanced train control technologies

Abstract

Quantitative research on risk and train control has traditionally relied exclusively on historical accident data. In this thesis, a model is developed that enables a probabilistic rather than statistical assessment of safety improvements achieved with advanced train control technologies. This probabilistic assessment is performed through fault-tree analysis of four accident categories: head-on and rear-end collisions, track-fault-related derailments, and collisions with maintenance-of- way equipment. Risk is defined as the aggregate consequences (human casualties) over accident categories. A mathematical model is built, which predicts the frequency and consequences of each accident category as functions of operational (e.g., speed, traffic mix and volume), and physical parameters (e.g., terrain, curvature) as well as train control capabilities. Accident frequencies are attached to metrics of exposure such as meets and passes for head-on collisions. Accident consequences are related to speeds, terrain, train type and occupancy. Train control can alter the speed of the train(s), thereby reducing both the frequency and severity of accidents. The model is first applied on single lines for sensitivity analysis to key parameters such as speed, traffic volume, and block length. Results indicate that risk grows more than linearly with these factors. The risk reduction achieved with advanced technologies ranges from about 30% to 90% according to the capabilities of the train control system. The model is then used on a simple example corridor representative of high-density railroad lines in the United States. The aggregate risk reduction for this corridor varies from 47% for a "bare bones" system to 55% for a more advanced system. However, risk reduction of individual accident types on single lines of the corridor can exceed 90%. The model is finally applied to a "Composite Corridor" used in a recent study of collision safety. Results corroborate those of the study, further demonstrating that train control technology is an important determinant of railroad safety whose influence varies according to operational and physical parameters. Potential applications include evaluating actual corridors in order to assess the safety benefits of implementing advanced train control systems. In addition, this model provides a framework for including additional safety-related factors, and other parameters related to further capabilities of train control technologies as necessary.