Railways are confronted with several problems in their daily business. One of these operational problems is delay management. Therein the question of whether a train should wait for a delayed feeder train or depart on time is addressed. Answering this question is not trivial since the determined wait-depart decision may cause serious consequences. While the majority of models in the literature usually take the decision by aiming for minimizing disturbances in the operating procedure, delay management focuses on the impact for passengers. By minimizing passenger delay, delay management differs from the other problems on the operational level and leads to different recommendations for dispatchers.
This thesis puts the scope on railway delay management and its impacts for passengers. It consists of three essays: a literature review on delay management and two models that advance the research in this field. In the literature review, a new classification scheme for operational problems in railways is developed. Literature in delay management and influence from delay management on neighboring areas are discussed. The second essay proposes a stochastic dynamic programming approach taking the dynamic nature of delays and uncertainty into account. Evaluating potential recourse actions derives policies for taking dispatching decisions. The third essay considers the capacity of trains in the decision making process. Rerouting of passengers for broken connections is further assumed and spill effects for passenger streams are measured. A nonlinear model is developed and solved by linearizing it exactly and heuristically.
Both approaches, from the second and third essay, are evaluated in a numerical study on real-world data from the German railway provider Deutsche Bahn. Germany possesses a rather complex and massive railway network that will require further decision support and future research.
Dieser Eintrag ist Teil der Universitätsbibliographie.
Das Dokument wird vom Publikationsserver der Universitätsbibliothek Mannheim bereitgestellt.