On the occasion of the public defense by Joris Wagenaar, a seminar is organised on Railway Optimization.
Speakers are Prof. Ralf Borndörfer (ZIB Berlin) & Prof. Richard Lusby (DTU Copenhagen).
Richard Lusby: Case studies in Integrating Planning Problems for DSB S-train
Providing a rail service typically requires solving several interdependent planning problems. Due to the size and complexity of each of these, they are usually solved sequentially, based on the planning horizon required. Without considering the future consequences of a given planning problem, an undesirable consequence of this sequential approach is that sub-optimal, or even infeasible, solutions are encountered at the later stages of planning. In this talk we explore the possibility of integrating planning problems to mitigate this undesirable behavior and ensure that planning problems at later stages are not unnecessarily restricted by the solutions to earlier planning problems. We present two case studies from the suburban passenger train operator in Copenhagen, DSB S-train. The first considers integrating, to a certain extent, the line planning and timetabling problem. The proposed methodology builds from scratch a line plan, minimizing passenger travel time and operator cost, for which a robust timetable exists. Computational results show that the Mixed Integer Programming based approach is able to significantly improve the robustness of the timetable over a standard sequential approach. The second case study focuses on integrating rolling stock scheduling with train unit shunting. If not solved simultaneously, rolling stock schedules can be infeasible from a shunting perspective. Computational results confirm the necessity of the decomposition based integrated approach for short-term planning; high quality solutions to the integrated planning problem are obtained on instances where a conventional sequential approach ends in infeasibility.
Location: EUR Woudestein campus, Mandeville building
Registration: send an e-mail Cheryl Blok (email@example.com)