Optimal allocation of rolling stock in scheduled transportation services

Passenger railway transportation services are operated in cyclic periodicities for urban and long distance traffic. Cabins and locomotives (rakes) are assigned to each service for specified time duration to operationalise timetables. Such resource allocations are generally constrained and incur cost, each in travel time, empty cabin runs and rake compositions. Some constraints are safety related and others are desirable for performance improvement. Resource utilisation is effective, if demands are met; efficient, if utilisation cost is minimum. A multi-commodity network flow model is presented to optimally allocate rolling stock, to fulfil demands at minimum operational cost. However, the model is computationally expensive and a relaxed model using simulation based heuristic is presented. Real data from Indian railways is used for testing. The problem context, model, heuristic approach and solutions are reported. Managerial insights that emerged from this empirical study are interesting; we present them along with limitations and possible extensions.