An algorithm to improve the efficiency and effectiveness of determining multi-component intervention programmes for railway networks that result in the lowest disruptions to service: A case study of Zollikofen-Brügg railway network: Project STABILITY under the ETH Mobility Initiative

Abstract

To ensure the stability of the train schedules, it is crucial to determine the optimal component-level intervention programs, their associated required possession windows, and their execution costs. This is a key step in the later phases of the intervention planning process. Network, production, and capacity managers at SBB would, therefore, benefit from an efficient and automated method for determining optimal intervention programs for up to 15 years ahead of the execution. Determining optimal component-level interventions in each planning period requires knowing: • what interventions are potentially required to be executed on asset components, how much their execution disturb the service, and how much do they cost to be executed, • where is the location of each component in the railway network, • how the execution of each component-level intervention is dependent on the execution of an intervention on another one(s), • what are the allowable possession windows that interventions can be executed while disturbing the service, • how long the execution of interventions can be delayed without an excessive increase in the failure risks, • what is the available budget for executing interventions. The availability of this information within the SBB, has now provided the opportunity to synthesize them from different databases to improve the later stages of the intervention planning process by improving the ability of the network managers to have a consistent and complete information regarding the optimal component-level intervention programs, and their required possession windows. This report proposes an automated methodology to determine optimal component level intervention programs taking into account the possession windows required for executing interventions, the impacts of execution on service, intervention costs, the interdependencies between the component-level interventions, the failure risks associated with the serviceability of assets, and the availability of resources, e.g., available possession windows. The structure of the optimization model used in the methodology is based on the network flow theory and evaluates different candidate intervention programs with respect to their net benefits. The model determines the optimal intervention programs that yield the maximum net-benefit. This methodology is used to demonstrate how it can be used to determine optimal component-level intervention programs a 25km railway network located in the canton of Bern between Brügg and Zollikofen. To ensure compatibility, the candidate interventions and the initial estimation of the costs, failure risks, and service impacts are assumed to be the ones described in the WP5 of the STABILITY project. The topology of the physical railway network is defined using available online maps. The interdependencies between component-level interventions are defined based on the location of the components. It is assumed that there are 273 days available per year in each planning period for execution of interventions to account for the harsh weather conditions. To evaluate the potential of the proposed methodology a comparison between two cases are made: one that considers interdependencies between the interventions to determine the optimal component-level intervention programs and another that does not consider these interdependencies. The results indicated that considering interdependencies can reduce the severity of service disruptions needed to execute interventions.