Passenger centric train timetabling problem

• We present a new planning phase for a passenger railway service.
• The new phase integrates the demand (discrete choice theory) and supply (operations research) interaction during the planning.
• We formulate the problem as a Mixed Integer Linear Programming problem.
• One of the objectives (the demand) is incorporated as an epsilon constraint, the result being a Pareto frontier.
• We solve the problem using CPLEX.
• The applicability of the approach is shown on a case study of a Swiss railway company.
• We present quantitative comparison between cyclic and non-cyclic timetable.

The aim of this paper is to analyze and to improve the current planning process of the passenger railway service in light of the recent railway market changes. In order to do so, we introduce the Passenger Centric Train Timetabling Problem. The originality of our approach is that we account for the passenger satisfaction in the design of the timetable. We consider both types of timetable(s): cyclic and non-cyclic. The problem is modeled as a Mixed Integer Linear Programming (MILP) problem with an objective of maximizing the train operating company’s profit while maintaining ε level of passenger satisfaction. The model does not take into account conflicts between trains and does not adjust dwell times at stopping stations among the lines. By solving the model for various values of ε, the approximated Pareto frontier is constructed. The analysis, based on an experiment using realistic data, shows that an improvement of passenger satisfaction while maintaining a low profit loss for the railway company can be achieved. A sensitivity analysis on passenger congestion illustrates a quantitative evidence that the non-cyclic timetables can account better for high density demand in comparison to cyclic timetables.