A Decision Support System for real-time track assignment at railway yards

In this thesis, we study the Real-Time Track Assignment Problem (RT-TAP), a real-time assignment problem that arises from the high percentage of stochastic arrivals of freight trains and the large quantity of last-minute parking requests at railway yards. We show that the RT-TAP is NP-Hard and provide a mixed integer program for solving the RT-TAP by minimizing the total weighted delay of trains. Because of its computational complexity, we develop a problem specific Genetic Algorithm (GA) and compare it with a First Scheduled First Served (FSFS) heuristic. Smaller instances show that there is no optimality gap between the Brute Force (BF) approach and the GA. The heuristic approaches are tested on two Real-Time simulations where we consider 74 inbound trains and 9 tracks. In order to define the effect of the input on the two models, we excluded in one of the simulations the track length i.e. all trains can be assigned to all tracks. Although we saw that the output of the GA remained the same, the FSFS heuristic was not able to show results as good as the GA. Therefore, we conclude that we developed a Decision Support System (DSS) that cannot only stabilize, but also improve the current decision-making process with regards to real-time track assignment at railway yards.