On-line reschedule optimization for passenger railways in case of emergencies

A Review of Online Dynamic Models and Algorithms for Railway Traffic Management

“…They are run only once. Most optimization works assume this setup as it translates in solving a problem once and for all [3], [6], [10], [25], [38], [44], [57], [62], [66], [69], [79], and then suggest possible schemes for implementation in reality [16]. Prediction models (i.e., delay propagation approaches) do not control traffic so they somehow fall off those classification [5], [28], [34].…”

“…Some approaches use the (weighted) travel time of passengers [71] or the weighted deviation with regard to the planned schedule. Other interesting objective functions might include (a combination of) passenger travel time [3], [30], deviation from timetable, minimize time to recover operations, and fall back to the original plan [6], [60], maximize punctuality; minimize running cost, or energy spent [1], [2], [22], [39], [51]. Approaches that simulate operations, or use dispatching rules, or simple heuristics have no explicit objective function.…”

“…We consider in this category also all approaches that did not implement any update due to time; in this case, the expected time is fixed once for all, and the setup is corresponding to a complete knowledge of the future, or at least, the optimizer believes he knows completely the future [3], [10], [15], [16], [22], [69].…”

“…While a stream of research considers retiming as a possibility to predict future operations [34], [48], to reduce delays [9], [23] [38], [59]; 2) speed: the updated speed advice are provided to drivers to avoid possible conflicts or save energy [2], [34], [35], [39], [51], [66], [73], [80] due to braking and reacceleration of trains. The two objectives of saving delays and at the same time reducing energy are investigated together by a third stream of research [22], [56], [59], [67], and to a different extent by speed management policies [20]; 3) order: changing the order of trains at shared infrastructure elements is seen as a critical action in the restricted infrastructure capacity of trains [3], [15], [16], [23], [25], [29], [44], [58], [59], [66], [71], [73]; Historically, most research has been dedicated to the meet and pass problem, i.e., finding the place where trains can be held before switches and block sections or single-track areas; a single train at a time can move between two meet locations [10], [51], [60], [66], [69], [74], [76], [78], [81]; 4) local route in a complex interlocking area: changing the path of a train with an adjacent, very similar path, in a local area, such as changing platforms at a station or t...…”

High-speed railway scheduling based on user preferences

“…This approach has been applied successfully in transportation (see Sánchez-Rico et al (2014) and Espinosa-Aranda and García-Ródenas (2012)). Furthermore, it is possible to extend the HSR model by changing the proposed supply model with previous approaches defined in the literature, such as methods based on the alternative graph model (see D'Ariano (2008, Corman et al (2010) and Espinosa-Aranda and García-Ródenas (2013)), or including real-time optimization methods for rescheduling rolling stock (see Almodóvar and García-Ródenas (2013)). …”