Modeling time table based tram traffic TN-1 Modeling time table based tram traffic Daniel Lückerath, University of Cologne, Cologne University of Applied Science, Germany Oliver Ullrich, Ewald Speckenmeyer, University of Cologne {lueckerath|ullrich|esp}@informatik.uni-koeln.de In mid-sized cities, tram networks are major components of public service infrastructure. In those networks with their typically dense schedules, multiple lines share tracks and stations, resulting in a dynamic system behavior and mounting delays following even small disruptions. Robustness is an important factor to keep delays from spreading through the network and to minimize average delays. This paper describes part of a project on simulation and optimization of tram schedules, namely the devel- opment and application of a simulation model representing a tram network and its assigned time table. We begin by describing the components of a tram network, which consist of physical and logical entities. These concepts are then integrated into a model of time table based tram traffic. We apply the resulting simulation software to our hometown Cologne's tram network and present some experimental results. 1 Introduction Tram networks are important parts of public transport infrastructure, which is exemplified by the 745,000 passengers that are transported in Cologne's tram network every day as described in [5]. Especially mid-sized cities often have mixed tram networks, i.e. networks where trams travel on street level (thus being subject to individual traffic and corresponding traffic regulation strategies) and on underground tracks. In such dense networks robustness is an im- portant factor to minimize average delays. Robust- ness measures the degree on which inevitable small disturbances, e.g. obstructed tracks due to parked cars, have impact on the whole network. With robust time tables delays are kept at a local level, whereas with non-robust time tables they spread through the network and might subsequently cause delays of other vehicles as described in [2, 3]. In this paper we present the simulation module (first described in [4]) which is part of a larger project to generate and evaluate robust time tables in order to minimize the impact of small delays. We develop a model and implement an application to simulate time tables of mixed tram networks in order to evaluate given time tables before their implementation in the field and to compare time tables generated by optimi- zation methods (as in [7, 8]) with respect to their applicability. In addition we want to show that the adopted simulation engine can be applied to real world problems. A more detailed description of our project and in particular our optimization approach is presented in the accompanying paper “Simulation and optimiza- tion of Cologne's tram schedule” (see [7]). We begin the remainder of this paper by describing the basics of time table based tram traffic (section 2), followed by a short discussion of our model repre- senting the physical and logical entities of the tram network (section 3). The resulting software is then applied to Cologne's tram network (section 4). We close with a short summary of the lessons learned and give some remarks on further research (section 5). S03W S03E S04E S05W S05E S25W S25E © Utopian Transport Authority Gare du Nord Piccadilly Gamla Stan Wall St Line 2 Line 1 120s 120s 90s 90s 30s 30s 90s 90s Tram 0212, line 2 (pass. exchange) Tram 0102, line 1 (moving) Utopia City Link • Commute and Conquer Figure 1. Part of a tram network