Skip to main content
Infrastructure Resilience Conference 2018

Full Program »

Measuring the Fragility of Large-Scale Transport Systems

Despite exhaustive risk mitigation measures and reliability improvements, large-scale transport systems remain exposed to unanticipated disruptions. Therefore, countermeasures optimized for particular system states and targeted to the specific disruption scenarios are essential. However, the design of these countermeasures requires awareness of the system state and understanding the expected performance loss.

The fragility of the system operating state can alert about critical working conditions and locate potential weaknesses in the system. To this end, our goal is to develop an indicator of system fragility, and assess the extent to which a disruption affects the system-wide performance.

We develop a probabilistic model of large-scale transport systems by using real-world data of passengers in public transport systems. The model quantifies the fragility of the system and predicts the performance decrease due to a disruption, e.g. the number of passengers directly impacted by a line closure. The model encompasses a statistical model that predicts the passenger demand based on historical data, and a flow model, that determines the movement of passengers and the link demand over time using passenger origin-destination and routing estimates.

The passenger demand at stations and the resulting link flows measure the system loading. Fragility results from the probability of exceeding a certain loading under a specific system condition. The model predicts the probability that a given loading occurs based on its historical frequency. If the loading remains within the desired operating envelope, defined by known station and link capacities, the system is safe. The more the loading sways away from this expected operating regime, the more fragile the system becomes. In addition, we assess the impact of potential threats on the number of passengers not served given the modelled link flows in the undisrupted state.

The system fragility measures and the predictions of the propagation of local disruptions based on the undisrupted state provide the necessary information to develop optimised and targeted countermeasures. The design of these countermeasures will be treated in future work.

Steffen Otto Peter Blume
Future Resilient Systems, Singapore-ETH Centre, ETH Zürich
Singapore

Michel-Alexandre Cardin
Department of Industrial and Systems Engineering, National University of Singapore
Singapore

Giovanni Sansavini
Reliability and Risk Engineering Laboratory, ETH Zürich
Switzerland

 

Powered by OpenConf®
Copyright ©2002-2016 Zakon Group LLC