Using algorithms to provide track managers with consistent and complete early estimates of possible future maintenance interventions: Project STABILITY under the ETH Mobility Initiative

Abstract

In the intervention planning process of Swiss Federal Railways (SBB), it is beneficial to have an early estimate of the required non-routine track interventions, their type, and the expected duration of track possessions. This information is crucial for track managers to effectively plan and allocate resources. Therefore, it would be advantageous for them to have a comprehensive and consistent overview of all the required interventions and related track possessions up to 20 years in advance. Generating this consistent and complete overview requires: - an understanding of the current condition of the components of the track sections and their likely evolution over time, subject to the multiple deterioration processes affecting them, - a general impression of the risk of failure related to the condition of the components, and - an understanding of the possible preventive interventions and which conditions trigger them, as well as the costs and impacts on service of executing them. With a large and increasing amount of information available on these items within the SBB, there is now the opportunity to enhance the SBB intervention planning process to make even more consistent and complete estimates of the future required condition related interventions and associated track possessions. This report presents an algorithm aimed at providing complete and consistent estimates of future track interventions, including the type, time, track possessions, and approximate costs. The algorithm integrates existing data and expert judgment where data is lacking to estimate the future condition of track components and the necessary interventions. Fault trees are used to assess the failure risks of each track section and determine the urgency of the interventions. The integration of the algorithm into SBB's intervention planning process is expected to increase its consistency, completeness, and efficiency. In this document, the algorithm is explained in detail using the Brügg-Zollikofen railway line (25 km) to demonstrate how it would be used to provide an overview of the upcoming interventions. It is shown that the algorithm can generate a complete and consistent overview of the type and time of interventions required in the future, an initial overview of the likely lengths and types of possession time required, and the likely costs of intervention execution. It is also shown that the algorithm can give an indication of the failure risks associated with the track sections and how they might evolve over time. In other words, the algorithm includes all the aspects normally considered by an SBB track manager in the early stages of intervention planning. It is expected that this consistent and complete overview of the upcoming required interventions will enable track managers to better communicate the needs for track possession times early in the train schedule planning process, which in turn will help ensure stable train schedules. Additionally, it generates a solid basis for the initiation of detailed investigations of the track sections by engineering offices, i.e., the investigations that result in the information required to decide as to when an intervention is required, as well as what type of intervention and track possession are required. The requirements in terms of data and computer support to use the algorithm, and the need for the results to be verified by track engineers are discussed, along with how new data is to be integrated into the models used to make the estimates.