Regional resilience analysis: Modeling, optimization, and uncertainty quantification

Sharma, Neetesh

Permalink

https://hdl.handle.net/2142/109516 Copy

Description

Title

Regional resilience analysis: Modeling, optimization, and uncertainty quantification

Author(s)

Sharma, Neetesh

Issue Date

2020-12-02

Director of Research (if dissertation) or Advisor (if thesis)

Gardoni, Paolo

Doctoral Committee Chair(s)

Gardoni, Paolo

Committee Member(s)

• El-Rayes, Khaled A
• Sauer, Peter W
• Wang, Pingfeng

Department of Study

Civil & Environmental Eng

Discipline

Civil Engineering

Degree Granting Institution

University of Illinois at Urbana-Champaign

Degree Name

Ph.D.

Degree Level

Dissertation

Keyword(s)

• interdependencies
• risk
• resilience
• stochastic
• optimization
• uncertainty

Abstract

Modern urban society's prosperity depends on the continuous flow of essential resources and services provided by the critical infrastructure. Ensuring the critical infrastructure's reliability and resilience is cardinal to ensure public safety and economic stability. However, past events have highlighted the infrastructure's vulnerability to disruptions caused by natural or anthropogenic hazards. Furthermore, complex interdependencies among infrastructure can cause disruptions to propagate within and across infrastructure, resulting in multi-fold catastrophic consequences on individuals, households, businesses, and communities. The consequences of past disasters have emphasized the need for hazard mitigation and recovery planning for infrastructure. Case studies of post-disaster recovery of different communities worldwide have indicated that successful recovery requires effective governance, intensive planning, community engagement, and intelligent use of resources. However, hazard mitigation and post-disaster recovery of infrastructure represent significant investments. Despite the expected economic advantage of investing in disaster preparedness, communities, businesses, and governments often struggle to budget their limited financial resources toward mitigation and recovery efforts. The uncertainty in predicting the occurrence and impacts of future hazards further increases the complexity of justifying large investments. There is a pressing need for rigorous and accurate models of infrastructure to reduce societal risk and improve regional resilience. This dissertation develops a novel classification of infrastructure interdependencies and a general mathematical formulation for modeling interdependent infrastructure. Specifically, the developed classification partitions the space of infrastructure interdependencies based on their ontological and epistemological dimensions. Under the ontology dimension, infrastructure interdependencies are classified into chronic and episodic. Under the epistemology dimension, infrastructure interdependencies are classified according to their mathematical modeling. The proposed classification better enables us to understand and mathematically model several classes of infrastructure interdependencies. The proposed mathematical formulation models infrastructure as a set of generalized flow networks while using dynamic interfaces to model the interdependencies. Carefully chosen working and benchmark examples illustrate the implementation and the advantages of the proposed formulation in providing accuracy while tackling the computational challenges. The dissertation then develops a rigorous mathematical formulation to model recovery, quantify resilience, and optimize large-scale infrastructure resilience. Specifically, a multi-scale recovery process model is proposed that significantly reduces the computational cost while favoring practical and easily manageable recovery schedules. The proposed resilience metrics then quantify the regional resilience by capturing the recovery process's temporal and spatial variations. A multi-objective optimization problem is then framed to improve regional resilience in terms of the proposed metrics while minimizing the recovery cost. The proposed recovery modeling is also integrated into a stochastic life-cycle formulation to account for the effects of infrastructure deterioration. The proposed approach is illustrated through large-scale examples for the post-disaster recovery modeling of infrastructure. Engineering models for critical infrastructure and measures of the societal impact, if developed in isolation, would not be sufficient to improve community resilience. This dissertation integrates the developed engineering models with existing social science approaches to comprehensively model the impact of hazards on communities and their recovery. Specifically, in combination with a reliability-based capability approach, the developed infrastructure models are used to predict the broad societal impact of hazards in terms of changes in dimensions of individuals' well-being. Some of these concepts are then explained through an example, modeling the dynamics of physical-social systems. Finally, the dissertation also provides an uncertainty propagation formulation for continuous improvement of the developed models and directing further research and data collection efforts. The proposed formulation quantifies the relative importance of engineering and social science models in evaluating the desired community resilience objectives. Specifically, a variable grouping using the interface function values' statistics decouples the regional resilience analysis into the constituent models, reducing the problem dimensions. The computationally intensive models are then identified, and an experimental design is developed for these models to reduce the total computation cost. The uncertainty propagation framework is performed using a global sensitivity analysis based on Sobol's indices.

Graduation Semester

2020-12

Type of Resource

Thesis

Permalink

http://hdl.handle.net/2142/109516

Copyright and License Information

Copyright 2020 Neetesh Sharma. All rights reserved.

Owning Collections