Shunting operations at flat yards : retrieving freight railcars from storage tracks

In this paper, we study the railcar retrieval problem (RRT) where specified numbers of certain types of railcars have to be withdrawn from the storage tracks of a flat yard. This task arises in the daily operations of workshop yards for railcar maintenance. The objective is to minimize the total cost of shunting via methods such as minimizing the usage of shunting engines. We describe the RRT formally, present a mixed-integer program formulation, and prove the general case to be NP-hard. For some special cases, exact algorithms with polynomial runtimes are proposed. We also analyze several intuitive heuristic solution approaches motivated by observed real-world planning routines. We evaluate their average performances in simulations with different scenarios and provide their worst-case performance guarantee. We show that although the analyzed heuristics result in much better solutions than the naive planning approach, they are still on average 30%-50% from the optimal objective value and may result in up to 14 times higher costs in the worst case. Therefore, we conclude that optimization should be implemented in practice in order to save valuable resources. Furthermore, we analyze the impacts of yard layout and the widespread organizational routine of presorting on the railcar retrieval cost