Queueing-based methods for timetable-independent performance analysis of railway nodes

RWTH Publications (RWTH Aachen)

Abstract

Railway transportation is increasingly recognized as a cornerstone for sustainable mobility of passengers and goods. However, existing railway networks are confronted with growing capacity constraints, particularly within node areas such as junctions and stations. In practice, many infrastructure managers rely on timetable-based performance indicators to assess operational quality, which require the assumption of fixed train sequences or timetables in early planning phases. Given that railway infrastructure is designed to last several decades, planning processes must account for uncertainties in future demand and regulatory requirements. Therefore, it is crucial to establish performance metrics that evaluate the infrastructure capacity independently of predetermined timetables. This dissertation introduces three novel queueing-based methods for the timetable-independent performance analysis of railway nodes and stations. The first contribution is a framework that models the capacity allocation process in railway route nodes using Continuous-Time Markov Chains, enabling the application of probabilistic model-checking to derive performance parameters. Furthermore, two extensions are developed: One incorporates phase-type distributions to explicitly model non-exponential arrival and service processes, enhancing modeling accuracy. The second extension presents, for the first time, a queueing-based method to analyze the performance of an entire railway station within a combined multi-channel system. All methods decompose railway nodes by infrastructure routes, facilitating broad applicability without the need for detailed timetable data or extensive expert knowledge. The approaches are validated through case studies, demonstrating their utility for infrastructure managers in assessing capacity constraints and planning optimal layouts for junctions and stations. Given that the presented framework obtains performance parameters on explicitly formulated Continuous-Time Markov Chains, it allows for efficient iterations over input parameters and supports model validation, offering substantial improvements over existing simulation and queueing-based approaches. The results highlight the practicality of the proposed methods in real-world scenarios. By enabling timetable-independent performance evaluation, these methods provide infrastructure managers with powerful tools to support the robust development of railway networks in early planning stages.

Authors 1

1. Tamme Emunds corresponding Aachen

   RWTH Aachen University

   Affiliation as printed

   RWTH Aachen