Analysing the resilience of interdependent infrastructure systems with network flow models
File(s)
Author(s)
Goldbeck, Nils
Type
Thesis
Abstract
Critical infrastructure systems form the backbone of modern civilisations and must be resilient against natural and anthropogenic hazards. However, the development of effective resilience strategies is often hindered by the insufficient capability of current analytical methods to capture complex interdependencies between different systems. Such interdependencies give rise to the risk of cascading failures and may delay recovery processes, thus exacerbating the impacts of infrastructure disruption. Advanced modelling and simulation methods are needed to address this problem, especially considering the increasing uncertainty in environmental conditions due to global climate change and additional interdependencies expected as a result of emerging technological trends.
Improving quantitative resilience analysis methods requires the development of models for the stochastic failure and repair of interdependent assets that form a system-of-systems. Moreover, dynamic network effects have to be considered in order to evaluate the impact of component failure on the overall ability of infrastructure systems to meet end-user demand. This thesis presents a novel methodology that combines minimum cost flow assignment models for coupled flows on multiple networks, simulation models for cascading failure and optimisation models for the repair of damaged assets.
The proposed methodology is applied to London's metro and electricity networks to analyse their resilience against flooding. Under certain assumptions regarding hazard exposure and repair resource logistics, the results suggest that doubling the capacity to repair damaged assets could increase resilience by 45~\%. In comparison, protecting all vulnerable assets so that they are 50 % less likely to suffer damage would increase resilience by 64 %.
Another case study applies the methodology to a supply chain that relies on transportation infrastructure. Results indicate the optimal investment into supply chain capacities and recovery capabilities. The model predicts cost savings of 17 % if contingency repair resources can be shared between different parts of the supply chain.
Improving quantitative resilience analysis methods requires the development of models for the stochastic failure and repair of interdependent assets that form a system-of-systems. Moreover, dynamic network effects have to be considered in order to evaluate the impact of component failure on the overall ability of infrastructure systems to meet end-user demand. This thesis presents a novel methodology that combines minimum cost flow assignment models for coupled flows on multiple networks, simulation models for cascading failure and optimisation models for the repair of damaged assets.
The proposed methodology is applied to London's metro and electricity networks to analyse their resilience against flooding. Under certain assumptions regarding hazard exposure and repair resource logistics, the results suggest that doubling the capacity to repair damaged assets could increase resilience by 45~\%. In comparison, protecting all vulnerable assets so that they are 50 % less likely to suffer damage would increase resilience by 64 %.
Another case study applies the methodology to a supply chain that relies on transportation infrastructure. Results indicate the optimal investment into supply chain capacities and recovery capabilities. The model predicts cost savings of 17 % if contingency repair resources can be shared between different parts of the supply chain.
Version
Open Access
Date Issued
2019-03
Date Awarded
2020-05
Copyright Statement
Creative Commons Attribution NonCommercial NoDerivatives Licence
Advisor
Angeloudis, Panagiotis
Ochieng, Washington
Sponsor
Engineering and Physical Sciences Research Council
European Union
Grant Number
EP/L016826/1
Publisher Department
Civil and Environmental Engineering
Publisher Institution
Imperial College London
Qualification Level
Doctoral
Qualification Name
Doctor of Philosophy (PhD)