Jørgen Thorlund Haahr

A Matheuristic Approach to Integrate Humping and Pullout Sequencing Operations at Railroad Hump Yards

This article presents a novel matheuristic for solving the hump yard block-to-track assignment problem. This is an important problem arising in the railway freight industry and involves scheduling the transitions of a set of rail cars from a set of inbound trains to a set of outbound trains over a certain planning horizon. It was also the topic of the 2014 challenge organized by the Railway Applications Section of the Institute for Operations Research and the Management Sciences for which the proposed matheuristic was awarded first prize. Our approach decomposes the problem into three highly dependent subproblems. Optimization-based strategies are adopted for two of these, while the third is solved using a greedy heuristic. We demonstrate the efficiency of the complete framework on the official datasets, where solutions within 4–14% of a known lower bound (to a relaxed problem) are found. We further show that improvements of around 8% can be achieved if outbound trains are allowed to be delayed by up to 2 h in the hope of ensuring an earlier connection for some of the rail cars.

Integrating rolling stock scheduling with train unit shunting

In this paper, we consider integrating two important railway optimization problems, in particular the Rolling Stock Scheduling Problem and the Train Unit Shunting Problem. We present two similar branch-and-cut based approaches to solve this integrated problem and, in addition, provide a comparison of different approaches to solve the so-called Track Assignment Problem, a subcomponent of the Train Unit Shunting problem. In this analysis we demonstrate, by way of a counter example, the heuristic nature of a previously argued optimal approach. For the integrated problem we analyze the performance of the proposed approaches on several real-life case studies provided by DSB S-tog, a suburban train operator in the greater Copenhagen area. Computational results confirm the necessity of the integrated approach; high quality solutions to the integrated problem are obtained on instances where a conventional, sequential approach ends in infeasibility. Furthermore, for the considered instances, solutions are typically found within a few minutes, indicating the applicability of the methodology to short-term planning.

Optimization methods for the Train Unit Shunting Problem

We consider the Train Unit Shunting Problem, an important planning problem for passenger railway operators. This problem entails assigning train units from shunting yards to scheduled train services in such a way that the resulting operations are without conflicts. The problem arises at every shunting yard in the railway network and involves matching train units to arriving and departing train services as well as assigning the selected matchings to appropriate shunting yard tracks. We present an extensive comparison benchmark of multiple solution approaches for this problem, some of which are novel. In particular, we develop a constraint programming formulation, a column generation approach, and a randomized greedy heuristic. We compare and benchmark these approaches with two existing methods, a mixed integer linear program and a two-stage heuristic. The benchmark contains multiple real-life instances provided by the Danish State Railways (DSB) and Netherlands Railways (NS). The results highlight the strengths and weaknesses of the considered approaches.