´óÏ㽶ÊÓƵ

Optional

Abstract: Public sector management shapes daily lives through resource allocation, service delivery, and long-term planning decisions. Operations research offers powerful tools to improve public services, yet gaps remain in identifying problems that matter to stakeholders and could be useful in practice. A priori problem definitions may not fully capture stakeholder priorities or feedback dynamics through which decisions influence and are influenced by constituent responses. Co-design offers a complementary pathway by helping identify relevant questions and key feedback dynamics to incorporate into optimization models. This dissertation explores how co-designed operations research can improve decision-making in two critical public domains: school transportation and coastal adaptation planning.

Excessively long school bus rides can negatively impact student performance and well-being. Collaborative engagement with a rural school district identified reducing travel time as the primary challenge, alongside two interrelated concerns: under-utilization of school buses and localized traffic congestion near schools during drop-off times. This dissertation formulates a school bus routing problem aimed at minimizing student travel time and develops a novel cluster-and-route heuristic to solve this problem. Furthermore, an iterative feedback mechanism models how improved routing strategies influence student ridership decisions, capturing how travel time reductions can collectively address under-utilization and congestion by encouraging mode shift from private vehicles to school buses.

Coastal erosion and flooding pose an existential threat to many small communities that rely on beach nourishment to protect their towns. These communities increasingly face a critical question: if and when to transition from continued nourishment to managed retreat. This dissertation develops a framework to identify optimal nourishment and retreat trigger timings under uncertainty. By explicitly modeling the effects of adaptation actions on observable state variables such as beach width and aggregate property valuation, the framework captures how these changes influence decision-making over the entirety of the planning time horizon.

Both applications demonstrate the potential of operations research to improve upon existing systems. In the school bus routing application, results indicate a 38-39 % improvement in student travel time. In the coastal adaptation application, this dissertation suggests managed retreat two decades earlier than existing benchmarks when accounting for uncertainty and balancing multiple planning objectives.

Thesis Committee: Vikrant Vaze (chair), Klaus Keller (co-chair), Geoffrey Parker, Vivek Srikrishnan (Cornell)