A Framework for Integrated Decision-making In Railroad Networks

Abstract

. Railroad networks are capital intensive operations that involve interaction between thousands of dynamic entities for e.g. trains, passengers, etc. while constrained by a fixed set of limited resources such as tracks, trains and platforms. Mathematical programming models have been used to allocate and operate resources efficiently and effectively but are applied separately at various levels of the decision making hierarchy namely, operational, tactical and strategic. While the individual considerations of the models at the different levels of decision making may be different (for example, the time horizons and problems addressed), decisions made at various levels affect each other due to common constraints originating from the static nature of the railroad topology, long lead times to add resources, etc. Further, real-time operational disruptions provide feedback that should be incorporated back into the tactical and strategic decision models to improve the model’s performance. One limitation of the current decision models is their stand-alone development with minimal interaction and feedback between decision levels. For e.g., time tables generated using customer demand form constraints for real-time operational planning. Separate treatment of long term strategic decisions and tactical models and modelling them in isolated fashion reduces its practical utility as interactions between these levels are ignored. We show interdependent nature of decision-making at various levels by bringing out the interplay between inter-level model constraints and variables.