Disruption management on high frequency lines : measuring the effectiveness of recovery strategies

Abstract

Incidents in urban rail systems are common. They vary in cause and severity, and can lead to minor or major disruptions on the network. Disruptions affect both the customers and the operating staff. To control and lessen the impact of an incident on the rail network, controllers implement corrective actions. Controllers rely mainly on experience, personal judgment and intuition to decide what recovery strategy to deploy for a given disruption. The recovery strategy deployed has a strong impact on both the service quality and the time to recovery. However, there is no systematic feedback loop to evaluate specific choices made in the control room. The scarcity of numerical data directly retracing operational actions makes ex poste aggregate analysis very difficult. The objective of this research is to increase our knowledge about the impact of recovery strategies deployed on high frequency lines. A crucial step is to build a new dataset that accurately retraces controller's actions. The process is based on a comparison between observed train movements and scheduled train movements. Actual train movements can be obtained through various vehicle location databases. This research develops an efficient merger of several vehicle location databases to create a reliable and complete vehicle location dataset. Building upon the reconstructed recovery strategy database, the research describes a framework to evaluate the effectiveness of recovery strategies. The framework includes a comparison between measures of recovery strategy characteristics and measures of recovery effectiveness. In particular, this methodology includes the definition of recovery effectiveness indices (REI) that take into account the impact of the disruption both for passengers and for the crew. For passengers, the research defines an integrated index based on a calculation of excess waiting time. For crew, the study focuses on lateness evaluated at crew relief points. The framework is applied to a case study based on the Piccadilly Line, a high-frequency line of the London Underground. In the context of the case study, the comparison of recovery strategies and effectiveness metrics suggests that an incremental implementation of cancelations compared to an aggressive cancelation strategy can have a positive overall impact on passengers. Even though most of the research is applied to the Piccadilly Line, both the proposed framework and the conclusions of this thesis are transferable to other metro lines and systems.