CREWS: A Train Scheduling Tool

Abstract

We present a tool, CREWS, for generating applications in personnel scheduling in transportation companies. Systems based on CREWS have been developed for the Portuguese Railways, the Dutch Railways, the Norwegian Railways, and the West Anglia Great Northern Railway (one of the operators of the British Rail). The initial version of CREWS supported long-term scheduling of drivers and guards, but current developments are extending it to support short-term planning as well. CREWS heavily relies on the use of Artificial Intelligence techniques and has been built in the perspective of a white box system, in the sense that the planner can perceive what is going on, can interact with the system by proposing alternatives or querying decisions, and can adapt the behaviour of the system to changing circumstances. Scheduling can be done in automatic, semi-automatic or manual mode. CREWS has mechanisms for dealing with the constant changes that occur in input data, can identify the consequences of the change and guides the planner in accommodating the changes in the already built schedules (re-scheduling). 1. Problem Description Long-term scheduling of crew is typically done several months prior to the execution of the schedule and consists in arranging the tasks that have to be done by personnel into duties (sequences of tasks to be done by one crew member in one day). A set of duties for the crew members of a certain depot with certain qualifications is called a schedule. Crew scheduling is known for its algorithmic complexity. It is usually carried out manually by a small number of planners that acquire most of their knowledge through experience. Besides the human skill of efficiently arranging tasks into duties, crew Transactions on the Built Environment vol 34, © 1998 WIT Press, www.witpress.com, ISSN 1743-3509