Rule-based discrete event simulation for optimising railway hump yard operations

This paper presents a simulation-based optimisation approach for planning railway hump yard operations. A hump yard is used for processing carriages (cars) brought by incoming trains, through a set of classification tracks, into newly formed outbound trains. There are specific constraints on the order in which each operation can be carried out, as well the standing order of cars in each outbound train. The set of decisions to be computed includes (i) the hump (processing) schedule of inbound trains, (ii) the assignment of cars to classification tracks, and (iii) the assembly schedule of outbound trains. The objective is to minimise the average dwell time of cars (the time spent from arrival at receiving tracks to departure). A simple set of rules is used to develop a discrete event simulator. The resulting objective function values vary between 5% and 20% of previously published optimisation formulations, depending on problem constraints. The execution time is between 3 and 5 minutes for a 42-day planning problem.