Data-Driven Infrastructure Planning for Uncertain and Changing User Demand

Modern infrastructure systems, including building energy management systems and package delivery networks, must make long-term planning decisions without fully knowing future operating conditions or how users will respond to the services provided. For example, building operators may change energy use depending on weather conditions and installed technologies, while customers may alter delivery pickup choices based on the availability and location of parcel lockers. As energy systems, transportation networks, and logistics services become more interconnected and data-driven, these uncertainties create significant challenges for designing efficient, reliable, and user-centric systems. The research funded by this award develops new mathematical and computational tools to help planners make better infrastructure investment and resource allocation decisions under uncertain conditions and changing user behavior. The research will support the design of responsive infrastructure services with efficient resource allocation and improved resilience, and advance both theoretical and computational frontiers of stochastic optimization. The decision-making framework will be broadly applicable to building energy management, last-mile delivery services, with further extensions to defense, economics, and security systems. The project also will engage students through workshops, project-based learning, and mentorship programs at multiple educational levels to cultivate interest in STEM fields. This project aims to create an integrated hierarchical optimization framework for user-centric infrastructure planning under multi-sourced uncertainty. This includes modeling and investigating interdependence between short-term adaptive user behaviors and long-term infrastructure design under uncertain system conditions. This research aims to develop data-driven approaches to accommodate and manage unknown and multi-sourced uncertainties, including both behavioral and contextual variability, and design efficient computational methods with provable solution quality guarantees. This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.