Modeling reliability of train arrival times

Abstract

This paper deals with the scheduling of trains so as to minimize train trip times while maximizing reliability of arrival times. The amount of risk of delay (RD) associated with a schedule is used as the reliability component of a constrained schedule optimization model. The paper outlines the model developed to quantify the risk of delays to individual trains, as well as to specific track segments and to the schedule as a whole. The risk model, which deals with single track operations, can be used to estimate the likely impact on reliability of arrival times, changes in train frequencies, and operating practices; track and station infrastructure investment strategies; and train technology upgrading. An application of the model to the optimization of schedules on a track corridor is described. The results obtained using the model are compared with the schedules used by train operations planning staff, in terms of overall delay and timetable reliability. The results highlight the significance of including a measure of timetable reliability, such as risk of delays, in the objective function for scheduling optimization.