Development of a resilience model and tool for resilience optimization for interdependent critical infrastructure

Abstract

Critical Infrastructures (CIs) are highly interdependent systems making them vulnerable to cascading failures during disruptions. The resilience assessment of CIs has become a crucial aspect at present yet challenging due to the interdependencies that drive their systematic behaviour, compounded by their dynamic nature. This thesis aims to bridge this gap by developing a comprehensive resilience assessment model for CIs, which incorporates the complexities of interdependencies through the application of Dynamic Bayesian Networks (DBN) and Machine Learning (ML) techniques. The study begins with a systematic review of CI resilience literature, including definitions, interdependencies, climate change impacts, modelling techniques, and resilience frameworks. The study also identifies eight governance measures to enhance the resilience of CIs. Moreover, Climate Change Adaptation (CCA) measures for enhancing resilience of CIs are prioritised based on different stages of infrastructure cycle, using the Analytical Hierarchical Process (AHP), with an expert survey. Then, a systematic literature review and expert survey were conducted to identify and validate the resilience assessment parameters. Then, a novel methodology to quantify the resilience of the CI system, accounting for both the interdependencies of CIs and the dynamic nature of their capacities, is proposed utilising the DBN. A case study from Sri Lanka demonstrated the DBN model’s practical applicability, showing strong agreement with expert assessments. Simulation data generated from the DBN model was used to train and validate six ML algorithms, resulting in a robust resilience prediction tool. SHAP analysis was used to interpret ML outcomes and highlight key drivers of resilience. It further validates the research output identified in the early stages of the present study. The outcomes of this study can be used as a foundation for developing a resilience optimisation tool. It provides actionable insights for policymakers and infrastructure managers, offering a validated, transferable tool to strengthen CI resilience across diverse contexts.