摘要：AbstractTrain scheduling is commonly the third step in the classical hierarchical approach to railway services planning and perhaps is the phase more related to user's perception about quality of service due to the passengers’ direct perception of timetables. Unfortunately, disruptions appear frequently due to an increase of the demand or as a consequence of fleet size reductions. Both circumstances give rise to un-supplied demand at certain stations, which generate passenger overloads in the available vehicles. Design strategies that guarantee reasonable users’ waiting time and maintain admissible operation costs are then an important topic in this field. Traditionally, two main off-line strategies to deal with such kind of situations have been used: Short-turning and deadheading. The first one is usually used when only a few of high demand stations should be attended. With this strategy, some vehicles perform short cycles in order to increase the frequency in certain stations of the lines. The second one consists of skipping stops (deadheading) at those stations with less demand, diminishing travel times and allowing for a fast service in conflictive stations. In both cases, different approaches to this problem have been developed considering frequencies (periodic timetables) and supposing uniform demand behaviour. This paper proposes a tactical model to determine optimal policies of short-turning and non-stopping at certain stations, considering different objectives like minimizing the passenger overload and preserving certain level of quality of service. The model enables to obtain both periodic and non-periodic timetables and, in contrast with previous works, it is able to use a dynamic behavior of the demand along the complete planning horizon (usually one day). Computational results for a real case study and comparisons with previous approaches are provided.
