Optimizing Train Shunting Operations at Kijfhoek

Abstract

With an ever growing economy, the freight train shunting yards have become more and more congested. It is crucial that research is done on solving the industrially sized Freight Shunting Problem (FSP). In this thesis, we consider the FSP at Kijfhoek. Kijfhoek, being the only shinting yard with a hump in the Netherlands, is of growing importance. It serves the largest port in Europe, the port of Rotterdam. This makes this shunting yard not only vital for the economy of the Netherlands, but also for the economy of Europe. The Kijfhoek shunting yard, owned by ProRail, is sublet to multiple railway operators, including Deutsche Bahn, currently allowed to use 37 of 43 classification tracks. As the need for shunting increases, ProRail is investigating to facilitate other railway operators, reducing track availability for Deutsche Bahn. Therefore, a more efficient way of shunting is needed. We will research Deutsche Bahn’s shunting problem at Kijfhoek to optimize the process and investigate the effect of reduced track availability. In this thesis, we devise a mathematical model of the Kijfhoek-specific FSP and use historical data from Deutsche Bahn, producing larger problem instances than researched before. We solve the resulting FSP by using Simulated Annealing (SA) to produce optimized shunting plans. We conclude that our SA algorithm can find good, feasible shunting schedules in a reasonable time. We are capable of creating solutions that have more than 95% of the cars depart with the first possible train, minimizing delays. We show that, when classification track availability drops below 29, shunting performance decreases significantly.