Estimating the robustness of long-haul train plans

In Australia, and elsewhere, the movement of trains on long-haul rail networks is usually planned in advance. Typically, a train plan is developed to confirm that the required train movements and track maintenance activities can occur. The plan specifies when track segments will be occupied by particular trains and maintenance activities. On the day of operation, a train controller monitors and controls the movement of trains and maintenance crews, and updates the train plan in response to unplanned disruptions. It can be difficult to predict how good a plan will be in practice. The main performance indicator for a train service should be reliability – the proportion of trains running the service that complete at or before the scheduled time. We define the robustness of a planned train service to be the expected reliability. The robustness of individual train services and for a train plan as a whole can be estimated by simulating the train plan many times with random, but realistic, perturbations to train departure times and segment durations, and then analysing the distributions of arrival times. This process can also be used to set arrival times that will achieve a desired level of robustness for each train service.