Shunting Minimal Rail Car Allocation

Marco E. Lübbecke and Uwe T. Zimmermann

Mixed integer programming; shunting minimization; decomposition; rail transport

We consider the rail car management at industrial in-plant railroads. Demands for loaded or empty cars are characterized by a track, a car type, and the desired quantity. If available, we assign cars from the stock, possibly substituting types, otherwise we rent additional cars. Transportation requests are fulfilled as a short sequence of pieces of work, the so-called blocks. Their design at a minimal total transportation cost is the planning task considered in this paper. It decomposes into the rough distribution of cars among regions, and the NP-hard shunting minimal allocation of cars per region. We present mixed integer programming formulations for the two problem levels. Our computational experience from practical data encourages an installation in practice.

Download as [PDF] [ps.gz]