Graduate Thesis Or Dissertation
 

Multi-objective Resilience Optimization of Interdependent Critical Infrastructure Networks

Public Deposited

Downloadable Content

Download PDF
https://ir.library.oregonstate.edu/concern/graduate_thesis_or_dissertations/zs25xh48c

Descriptions

Attribute NameValues
Creator
Abstract
  • Critical Infrastructures (CIs) such as energy, water supply, telecommunications, and transportation are highly vulnerable to cascading failures due to their interdependent operations. A resilient network is crucial to withstand the impact of functionality loss in disruptive events. This research evaluates network expansion as a proactive resilience strategy in which new network components are added to improve redundancy and increase the service level of the network. We present a multi-objective resilience optimization model to evaluate network expansion decisions for interdependent CIs under disruption uncertainty. A network-based graph is developed to model the geographical and functional relationships of two interacting CIs. A resilience score for interdependent CIs is determined from network complexity and unmet demand metrics to represent the topology-based and service-based network performance metrics. The multi-objective approach is used to develop solutions showing the trade-off between increasing the investment and improving resilience. Later, the resilience optimization model is formulated as a two-stage stochastic mixed-integer program with the expected total cost and the expected resilience score as competing objectives. The first-stage decisions involve the optimal selection of candidate nodes and links to add to the network. The second-stage decisions comprise the optimal flow allocation, the unmet demand, and the unused supply of CI commodities post-disruption. The disruption uncertainty is introduced as a set of random parameters corresponding to each disruption scenario. The physical interdependencies between CIs are enforced through the demand constraints. We consider a real-world case study of the interdependent power-water networks in Shelby County, TN under earthquake scenarios with varying degrees of severity. The electrical flow in the power grid is modeled using linear DC power flow equations. The model is solved using the augmented epsilon-constraint method to generate sets of Pareto optimal frontiers. This study confirms that network expansion can improve resilience significantly with low total costs, but the improvement diminishes at higher total costs. Finally, we further evaluate the effect of network topology on resilience through five synthetic interdependent networks having a combination of random and hub-and-spoke topologies with different network sizes and expansion opportunities. We develop a method to generate a manageable number of critical node disruption scenarios. After applying the stochastic resilience optimization model to the synthetic problem instances, the expanded network designs are characterized according to their graph metrics. Random interdependent networks are found to be better connected than the ones with hub-and-spoke structure due to their higher average node degree, higher average clustering coefficient, and lower average shortest path length. Hub-and-spoke structures exhibit scale-free characteristics which make them less tolerant to targeted attacks on the hub nodes. In addition, expansion should be prioritized on the single network with a larger size due to their stronger contribution to the overall resilience. Our study demonstrates the importance of interdependent relationships, network topologies, and disruption types when planning for resilience.
License
Resource Type
Date Issued
Degree Level
Degree Name
Degree Field
Degree Grantor
Commencement Year
Advisor
Committee Member
Academic Affiliation
Rights Statement
Publisher
Peer Reviewed
Language

Relationships

Parents:

This work has no parents.

In Collection:

Items

Thumbnail Title Date Uploaded Visibility Actions
file details TiongAchara2022.pdf 2022-03-14 Public Download