Planning problems in railway traffic where always of great interest to railway companies due to economical reasons. Computer-aided tools are needed for these complex planning tasks in order to use the resources (tracks, carriages, locomotives etc.) efficiently.

Locomotive planning is the assignment of locomotives to a given set of transports. A transport is given by its start and end location and a time window for the departure time. The goal is to use few locomotives but also to avoid passive transports where the locomotives run empty.

In this work, the locomotive planning problem is solved in a case study for the Swedish railway. An existing constraint-based planning system is extended by a constraint model for locomotive planning. This model contains a high-level geometric constraint and a propagation algorithm for this constraint is developed. A heuristic search strategy for good locomotive schedules is presented and tested on examples from the Swedish railway network.

This work was done during a half-year stay at the Swedish Institute of Computer Science (SICS) in Stockholm.