UC Berkeley

Seismic events can severely disrupt regional transportation systems, compromising mobility, economic activity, and emergency response. While the resilience of transportation infrastructure has been widely studied, most existing work focuses either on system robustness or on rapidity of recovery, often using static or small-scale models that limit the spatial and temporal resolution of the analysis. This dissertation addresses these limitations by developing a flexible, regional-scale framework to evaluate the seismic resilience of transportation systems along the dimensions of robustness and rapidity, incorporating both system performance and user-level impacts. The study adopts a mesoscopic, semi-dynamic, agent-based traffic model designed to balance spatial and temporal detail with computational feasibility. The model simulates individual-level vehicle movements across large urban regions using 15-minute time steps and disaggregated origin-destination demand, enabling the capture of dynamic congestion effects and evolving network conditions during and after seismic events. This traffic simulation is embedded within broader resilience assessment frameworks that quantify degradation in system functionality (robustness) and characterize network recovery trajectories over time (rapidity), using performance indicators such as total travel time, vehicle hours traveled (VHT), and recovery curve metrics including area under the curve (AUC) and skewness. Two case studies are used to demonstrate the utility and flexibility of the proposed frameworks. The first assesses the robustness of the Southern California transportation network under a hypothetical magnitude 7.4 earthquake along the Palos Verdes Fault. A hybrid bridge damage modeling approach was used to generate input fragility curves, and economic losses were estimated based on simulated travel delays and trip loss. Results show pronounced spatial heterogeneity in impacts, with disproportionately high delays and losses concentrated near the Ports of Los Angeles and Long Beach. These findings suggest that resilience planning should be guided not only by exposure and structural vulnerability but also by functional importance at the system level and within regional economies. The second case study focuses on rapidity, evaluating the recovery of the Tokyo Metropolitan Area highway system under multiple coordination strategies between two major expressway operators: the Metropolitan Expressway Company (MEX) and East Nippon Expressway Company (NEXCO). A scenario-based coordination framework was introduced to model varying levels of inter-agency coordination, including information sharing, strategy alignment, and pooled resource allocation. Results show that higher levels of coordination significantly improve recovery performance, as measured by both AUC and skewness, and help mitigate spillover congestion onto local roads. These outcomes highlight the importance of governance structures, institutional interoperability, and collaborative response protocols in post-disaster recovery. This dissertation makes three key contributions: (1) it proposes scalable, semi-dynamic traffic modeling frameworks for regional seismic resilience analysis; (2) it operationalizes robustness and rapidity through integrated metrics and simulation outputs; and (3) it provides empirical insights into the spatial distribution of disruption and the value of inter-agency coordination in recovery. Together, these contributions support more targeted investment planning, cross-agency coordination, and data-driven decision-making for hazard mitigation and emergency response.