Train Shunting and Service Scheduling: an integrated local search approach

Abstract

Trains have to be maintained and cleaned regularly to ensure high passenger safety and satisfaction. These service tasks must be performed outside the rush hours, when the trains are parked off the main railway network at dedicated service sites. The activities on a service site are currently scheduled by hand; a difficult and time-consuming task that consists of matching incoming and outgoing trains, scheduling the service tasks, assigning trains to parking tracks and routing the trains over the service site. We propose a local search approach for the automated construction of such shunt plans that integrates these four planning aspects. Our heuristic is applied successfully to artificial and real-world planning problems, and outperforms a state-of-the-art mixed integer programming algorithm.