Duenas-Osorio, Leonardo2021-08-162021-08-162021-082021-08-13August 202Talebiyan, Hesam. "Interdependent Restoration of Infrastructure Networks with Humans in the Loop: decentralized and strategic decision processes." (2021) Diss., Rice University. <a href="https://hdl.handle.net/1911/111232">https://hdl.handle.net/1911/111232</a>.https://hdl.handle.net/1911/111232EMBARGO NOTE: Submission published under a 6 month embargo; the embargo will last until 2022-02-01This dissertation brings essential elements of real-world decision-making environments into restoration strategies for interdependent networks after contingencies. Existing strategies have been predominantly developed from a centralized perspective. In reality, however, several autonomous decision-makers or agents take actions within and across these networks. Agents usually need to communicate and coordinate to make informed decisions. In practice, however, communications are typically noisy, delayed, or simply nonexistent. Therefore, agents cannot solicit enough information (pertinent to their decisions) from other agents when they need it. In this study, I focus on modeling such a decentralized decision-making environment with imperfect information flow, particularly when facing contingencies from natural hazards---a context in which modeling is known to appeal to community stakeholders. My proposed algorithms presuppose that decision-makers iteratively go through stages seen in the field to devise restoration plans, particularly resource allocation and decision-making stages. As for the first stage, I employ auction theory---given its theoretical and practical appeal---to propose a rigorous decision tool to address the problem of efficient, decentralized resource allocation. As for the second sage, I employ a decentralized approach to solve a family of restoration optimization models subject to budget and operational constraints. The novelty of my approach is the recognition that insufficient information forces agents to use their judgment and domain expertise to speculate other agents' potential decisions. Introducing the concept of Judgment Call (JC), I cover a spectrum of agents' potential mindsets toward judgment with ubiquitous optimism and pessimism at the extremes. Also, I model the agents' interaction during the decision-making stage using Bayesian games. This novel approach models decision-makers' strategic behavior while accounting for their incomplete information and bounded rationality. In addition, it allows me to study the effect of cooperation and competition in the decision processes. The proposed models are demonstrated, first, with ideal yet controllable synthetic networks, and then with the realistic infrastructure of Shelby County, TN, subject to hazard-consistent seismic scenarios. The results show that allocation mechanisms based on combinatorial auctions significantly improve upon equality-driven uniform allocations in terms of resilience. Furthermore, I show that not only are cooperation and optimism the key to gain near-optimal decisions, but also, agents could be offered realistic incentives to act cooperatively.application/pdfengCopyright 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.Infrastructure restorationcommunity resiliencedecentralized decision-makinggame theoryauction theorycivil engineeringmulti-agent systeminterdependent networksThesis2021-08-16