Principled Uncertainty Quantification for Resilient Infrastructure Management

Simple item page
dc.contributor.advisorDueñas-Osorio, Leonardoen_US
dc.creatorParedes Toro, Rogeren_US
dc.date.accessioned2022-09-23T16:07:02Zen_US
dc.date.available2023-02-01T06:01:16Zen_US
dc.date.created2022-08en_US
dc.date.issued2022-08-15en_US
dc.date.submittedAugust 2022en_US
dc.date.updated2022-09-23T16:07:03Zen_US
dc.description.abstractUncertainty Quantification (UQ) is the prevalent approach to assess urban infrastructure safety and support decision making, as it accounts for randomness and the lack of perfect information about infrastructure components and the environment. This is especially relevant now that urban infrastructure services are becoming more interdependent and automation is becoming integral to their management. Unfortunately, the task of accurate and efficient UQ is believed to be computationally infeasible in general. Understandably, the state-of-the-practice consists of heuristic UQ methods that make computations affordable; however, they give no assurance of correctness or quality of reliability estimates. Thus, the trust in heuristic analyses to reveal and quantify key vulnerabilities in urban infrastructure is limited, especially as the failure modes of infrastructure are poorly understood. For example, system-level collapse can be the result of interdependencies or cascading failures, as opposed to the more traditional view that focused on the integrity of assets and facilities without quantitative system-level effects. This thesis develops and advances UQ methods that enable evidence-based decision making, i.e., for fixed input data and accepted physical models, non-expert users can independently and confidently obtain reliability estimates that are guaranteed to be correct. Thus, this thesis refers to this family of methods as principled UQ methods. While the bulk of the literature devotes attention to the affordability of computations, at the risk of losing correctness, this work focuses on the quality of computations in general applications. We do not abandon the emphasis on speed of computation, as we pay especial attention to the speed of computations in real-world instances. In the process, we stumble upon a surprising finding: principled UQ methods can vastly outperform state-of-the-art heuristic UQ methods in several case studies. We show evidence of this finding by replicating reliability studies on urban infrastructure and fully eliminating the uncertainty brought about by recent heuristic UQ proposals. Moreover, the generality of the techniques we develop and advance allow for resilience quantification; thus, we unify the insights and challenges in both reliability engineering and resilience engineering, especially rare event simulation and estimation, in the context of highly resilient urban infrastructure. All developments in this thesis are exemplified with applications to both synthetic systems and engineered infrastructure systems.en_US
dc.embargo.terms2023-02-01en_US
dc.format.mimetypeapplication/pdfen_US
dc.identifier.citationParedes Toro, Roger. "Principled Uncertainty Quantification for Resilient Infrastructure Management." (2022) Diss., Rice University. <a href="https://hdl.handle.net/1911/113230">https://hdl.handle.net/1911/113230</a>.en_US
dc.identifier.urihttps://hdl.handle.net/1911/113230en_US
dc.language.isoengen_US
dc.rightsCopyright is held by the author, unless otherwise indicated. Permission to reuse, publish, or reproduce the work beyond the bounds of fair use or other exemptions to copyright law must be obtained from the copyright holder.en_US
dc.subjectUncertainty Quantificationen_US
dc.subjectResilienceen_US
dc.subjectReliabilityen_US
dc.subjectStructuralen_US
dc.subjectInfrastructureen_US
dc.subjectFailure Probabilityen_US
dc.titlePrincipled Uncertainty Quantification for Resilient Infrastructure Managementen_US
dc.typeThesisen_US
dc.type.materialTexten_US
thesis.degree.departmentCivil and Environmental Engineeringen_US
thesis.degree.disciplineEngineeringen_US
thesis.degree.grantorRice Universityen_US
thesis.degree.levelDoctoralen_US
thesis.degree.nameDoctor of Philosophyen_US
Files
Original bundle
Now showing 1 - 1 of 1
Thumbnail Image
Name:
PAREDESTORO-DOCUMENT-2022.pdf
Size:
2.81 MB
Format:
Adobe Portable Document Format
Download
License bundle
Now showing 1 - 2 of 2
No Thumbnail Available
Name:
PROQUEST_LICENSE.txt
Size:
5.85 KB
Format:
Plain Text
Description:
Download
No Thumbnail Available
Name:
LICENSE.txt
Size:
2.61 KB
Format:
Plain Text
Description:
Download
Collections
Rice University Theses and Dissertations