Operation of shunting yards: train-to-yard assignment problem

Abstract

A multiple shunting yard as a major freight rail yard consists of several sub-yards of limited capacities. Our research question is how to assign the railcars of inbound trains among the sub-yards such that the number of re-assignments is minimized. We denote this problem as train-to-yard assignment problem (TYAP) and prove its strong NP-hardness. Moreover, our computational experiments show that the lower bounds of IBM ILOG CPLEX are poor and not competitive to our lower bounds that can be computed in a fraction of a second. Because of their low computational complexity, our lower bounds can be repeatedly calculated in any enumeration algorithm such as branch and bound or bounded dynamic programming without contributing substantially to the overall runtime.