Date of Award

Spring 2026

Abstract

Urban infrastructure systems face unprecedented compound threats from aging physical stock, climate-driven hazard escalation, and large-scale systemic shocks, yet the analytical tools available to decision-makers remain inadequate in both empirical scope and equity orientation. This dissertation advances urban resilience science along two complementary empirical frontiers.The first investigation develops a deep reinforcement learning framework for socially equitable, flood-aware bridge portfolio maintenance optimization. Using Proximal Policy Optimization, a portfolio of 19 flood-vulnerable bridges in Suffolk County, Massachusetts is optimized over a multi-year planning horizon integrating HAZUS-based flood hazard simulation across four return-period scenarios (10, 50, 100, and 500 years), HEC-18 hydraulic scour depth modeling, and a novel Social Impact Index derived from the CDC Social Vulnerability Index. Results demonstrate a 52% reduction in total portfolio cost relative to status quo practice and a 93.3% improvement in equity-weighted failure probability relative to the do-nothing baseline. A Pareto frontier quantifies the equity–efficiency trade-off across a continuously adjustable vulnerability influence parameter, enabling transparent and auditable decision-making. The second investigation estimates the causal influence of urban morphological structure on multi-dimensional pandemic resilience across 52 major US cities. Four synthetic morphology metrics—crowdedness, Accessibility, the Non-Residential Centricity Index, and the Development Uniformity Index—are derived from a 12-million-building database and applied within a Frisch-Waugh-Lovell residualization framework to isolate morphological effects from structural confounders across 12 resilience outcomes spanning economic, social, and health dimensions over the full COVID-19 disruption period. Key findings identify the Development Uniformity Index—the spatial autocorrelation of building age at the city scale—as the strongest predictor of economic resilience loss across both the robustness and recovery phases of the pandemic shock, a result without precedent in the empirical urban resilience literature. A consistent robustness-recovery asymmetry is documented across crowdedness and Accessibility, and the analysis demonstrates that urban pandemic resilience is genuinely multi-dimensional, with cross-domain trade-offs that single-index composite approaches cannot capture. Together, these investigations establish that the physical configuration of the built environment is an active structural determinant of resilience outcomes across multiple scales, and that formal integration of social equity into both infrastructure optimization and resilience measurement is essential for equitable and effective urban governance.

Document Type

Dissertation

First Advisor

Fei Han

Second Advisor

Erin Bell

Third Advisor

Weiwei Mo

Department or Program

Civil and Environmental Engineering

Degree Name

Doctor of Philosophy

Download

Share

COinS