Short abstract The research aims to contribute to advance multimodal transport resilience assessment in the city of Lisbon, focusing on the early detection and measurement of vulnerabilities in the public transport network. Abstract A resilient transport system is an essential requirement for cities and transport operators to deliver safe and secure mobility services in presence of disruptive events. Emerging risks and threats such as climate change along with the current pandemic represent additional challenges for advancing urban transport resilience assessment for sustainability.The measurement of resilience in engineered systems was previously defined as “the ability to anticipate, prepare for, recover, learn, and improve from external disturbances.” The research reported in this paper aims to contribute for the early detection and measurement of vulnerabilities in the public transport network in the city of Lisbon using available big data. More specifically, it aims to enhance existing resilience assessment methods to answer the following questions:• How different types of failures and attack strategies will impact on the transport network’s connectedness?• What are the main topological vulnerabilities in the public transport network? And how can these be prevented?To tackle the above research questions, the city of Lisbon public transport (bus, underground, railway, inland waterways, etc.) was modelled and examined as a multiplex network which enables the analysis of intralayer, interlayer and the entire system of modal connections. This helped to better understand the integration of different transport modes as well as its characteristics and resilience performance. Several issues were examined and evaluated such as: how different stations and stops may have similar properties such as bridging communities with higher centrality; how the network may decrease its performance; simulation of performing robustness tests for each case. To assess the impact of each strategy, the evolution of different metrics for each simulated removal was graphically represented. To understand how connectedness is affected, several indicators were used such as the average path length (also known as the average geodesic distance), average degree, number of isolated components, and the giant strongly connected component size. To understand the average path length change, the shortest paths remaining in the simulated network were averaged out. Considering the methodology used, the research is novel in what concerns the assessment of node and edge targeted percolation impact in multimodal networks for transport resilience assessment.To compare different strategies, we used a normalized version of the discrete Area Under Curve (AUC) metric. We were able to compare side by side the robustness of each modality layer, regardless of their size, as well as the whole transport network.Considering the simulation of different targeting strategies, we also observed that resilience tests needed to remove about half the network nodes, to leave all the remaining nodes completely disconnected; node removals are more effective than removing edges in this topology. Results also indicate that networks of this type are much more susceptible to node attacks than edge attacks, and different strategies are clearly more efficient in one over the other. The attacks that involve recalculating metrics are usually the most dangerous. However, we identified a specific context where, counter-intuitively, that is not the case. Lastly, we simulate cascading events that showed that neighbour nodes failure is more dangerous than line failure. Comparing different attack strategies, we concluded that the metrics’ evolution throughout the test is similar according to strategy even in different topologies, with some justifiable exceptions. This can constitute a valid contribution in terms of the resilience assessment of transport networks on other urban contexts, even in multimodal scenarios.In the research reported, we propose a method for modelling and assessing multimodal transport networks that allow a more comprehensive analysis of transport resilience focused on prior knowledge. The approach allows a better understanding of how particular features of the network topology play a role in their resilience, as well enables to identify the potential nodes/locations where the transport network would get more stress. One important output is an automatic report that shows how specific failures in the transport network would affect the mobility of its users.This research reported in this paper is anchored in the pioneer research and innovation project - Integrative Learning from Urban Data and Situational Context for City Mobility Optimization(ILU), in the field of artificial intelligence applied to urban mobility that joins the Lisbon city Council and two national research institutes. Research results are important for theory and policy purposes.
Aparício et al. (Tue,) studied this question.