1 Micro-macro approach to robust timetabling Nikola Besinovic a,1 , Roberto Roberti b , Egidio Quaglietta a , Valentina Cacchiani c , Paolo Toth c , Rob M.P. Goverde a a Department of Transport and Planning, Delft University of Technology, b Department of Transport, Technical University of Denmark c DEI, University of Bologna 1 Corresponding author. E-mail: n.besinovic@tudelft.nl Abstract With the increasing demand for railway transportation infrastructure managers need improved automatic timetabling tools that provide feasible timetables with improved performance and short computation times. In this paper we propose a hierarchical framework for timetable design that includes a microscopic and macroscopic model of the network. These two models interact with each other in an iterative manner in order to generate an optimal timetable that is feasible with a high level of precision/detail. An iterative adjustment of train running and minimum headway times is performed in the framework which terminates when a feasible timetable has been generated. The framework always guarantees timetable feasibility also at the microscopic level. Additional timetable performance is realized in terms of stability and robustness. The application to an area of the Dutch railway network shows the ability of the framework to automatically compute a feasible, stable and robust timetable. In this sense practitioners can use this framework both for effective timetabling and post-evaluation of existing timetables. Keywords Timetable, feasibility, robustness, stability, capacity analysis 1 Introduction The recent growth in the demand for railway passengers and freight has raised the necessity for infrastructure managers to improve the efficiency of their networks in terms of increased capacity, higher service availability (e.g. punctuality, regularity) and lower energy consumption. Upgrading the infrastructure (i.e. building new tracks) or the signalling systems (e.g. migrating to European Train Control System ETCS level 2) is a suitable solution to achieve these objectives but very costly to implement. A cost-effective alternative is represented by an effective planning of train services to get timetables that are able to absorb everyday variations in running and dwell times while exploiting network capacity as much as possible. This means allocating as many train paths as possible to the available infrastructure slots while guaranteeing time allowances (i.e. supplements and buffer times) that effectively dampen delay propagation. In this context timetable design must rely on an accurate estimation of realizable train paths and buffer times. Only a robust construction of these time-distance paths allows an effective design of dense timetables that are operationally feasible, i.e., free from conflicts and respecting the constraints imposed by the infrastructure layout and the safety and signalling systems. 041-1 6th International Conference on Railway Operations Modelling and Analysis - RailTokyo2015