Remarkably, the MILP models and solution algorithms developed for the train dispatching problem can be readily exported to other relevant contexts, such as aircraft ground management at airports, see for example, .
The dispatching problem on multitrack territories: Heuristic approaches based on mixed integer linear programming
Article first published online: 1 OCT 2013
Copyright © 2013 Wiley Periodicals, Inc.
Volume 62, Issue 4, pages 315–326, December 2013
How to Cite
Boccia, M., Mannino, C. and Vasilyev, I. (2013), The dispatching problem on multitrack territories: Heuristic approaches based on mixed integer linear programming. Networks, 62: 315–326. doi: 10.1002/net.21528
Note that, while G is undirected, routes have an orientation. So, formally, routes are directed paths of a new graph obtained from G by orienting its edges.
With some abuse of notation, we write to denote that the directed edge ( u, v) belongs to r and also to denote that node w belongs to r.
- Issue published online: 28 OCT 2013
- Article first published online: 1 OCT 2013
- Manuscript Accepted: JUL 2013
- Manuscript Received: JAN 2013
- Russian Foundation for Basic Research. Grant Number: 12-07-33045 mol_a_ved. and 12-07-13116 ofi_m_RZD
- Dispatching problem;
- Mixed Integer Linear programming
Trains running through railway lines often accumulate some delay. When this happens, rescheduling and rerouting decisions must be quickly taken in real time. Despite the fact that even a single wrong decision may deteriorate the performance of the whole railway network, this complex optimization task is still basically performed by human operators. In very recent years, the interest of train operators to implement automated decision systems has grown. Not incidentally, the railway application section (RAS) of INFORMS has issued a challenge devoted to this problem concomitantly with the INFORMS Annual Meeting 2012. In this article, we describe two heuristic approaches to solve the RAS problem based on a mixed integer linear programming formulation, and we report computational results on the three RAS instances and on an additional set of instances defined on a more congested network. Computational results on the challenge test bed show that our algorithms positively compare with other approaches to the RAS problem. © 2013 Wiley Periodicals, Inc. NETWORKS, Vol. 62(4), 315–326 2013