Centralized versus distributed systems to reschedule trains in two dispatching areas

Railway dispatchers are in charge of rescheduling trains during operations in order to limit propagation of disturbances occurring in real-time. To help the dispatchers in such task, an advanced decision support system, ROMA (Railway traffic Optimization by Means of Alternative graphs), has been recently implemented to optimize railway traffic within a single dispatching area. This paper presents a novel distributed optimization system to control trains running in a Dutch railway network that is divided into two complex dispatching areas with dense traffic, each one controlled by a single dispatcher with the support of a local ROMA. A coordination level is introduced in order to manage the interaction among the two local ROMAs. An extensive computational assessment of the centralized and distributed systems is performed by using simple and advanced train scheduling algorithms, including dispatching rules adopted during operations. The effectiveness of the distributed system is shown in terms of computation time and delay minimization for practical statistical entrance delay distributions and in presence of an increasing number of blocked platforms in the main station area.