Analysing Railway Network Capacity Using Macroscopic Fundamental Diagram

Christopher Szymula,Nikola Bešinović,Karl Nachtigall

DOI: https://doi.org/10.2139/ssrn.4103390

2022-01-01

SSRN Electronic Journal

Abstract:The infrastructure occupation of many European railway networks appears to be at its limits, while recent public discussions aim for shifting even more traffic towards the railway system. Thus, the question of the capacity of railway networks is gaining increasing attention. We are assessing the capacity of railway networks by presenting a new petri net (PN) based approach: the railway network occupation model (RNOM). Building upon individual lines, which are then connected to more complex structures, the model uses the networks elementary circuits for occupation analysis. To assess the networks capacity, the model is applied in the enumeration based railway network capacity (RNC) framework. By evaluating the maximum occupation of different scenarios, the networks capacity is identified. The framework provides the networks minimal operable frequency and corresponding train flows as well as the average scheduled waiting times per train. It thereby assesses the capacity and the corresponding network effects, which are also quantified in a network-specific macroscopic fundamental diagram for the railway network (MFD-R). Experiments on different generic networks demonstrated the RNC frameworks applicability and revealed additional insights into the connections between train flow, waiting times and the total number of trains. The results showed that the networks throughput performance is rather sensitive to the networks saturation. The evaluation also showed that the maximum occupation just emerges under certain very specific conditions and might result in unstable operations. Additionally, the results illustrated the dependencies between network capacity and the networks structure. Furthermore, a case study performed at a real life railway network, proves the realism of the results. The scalability to larger problem instances and the relaxation of some model assumptions (i.e. by including aperiodic trains and variable train orders) remain problems to be tackled in the future. However, the presented model can already contribute to more efficient capacity usage and thus support the recent ambitions for eco-friendly transportation.