CAREER: Overcoming Nonlinearities, Uncertainties, and Discreteness to Mitigate the Impacts of Extreme Events on Electric Power Systems

Georgia Tech Research Corporation, Atlanta GA

Investigators

Abstract

This NSF CAREER project aims to develop algorithms for optimizing the planning and operation of electric power systems in the context of extreme events such as wildfires, hurricanes, evacuations, etc. The project focuses on three key computational challenges: nonlinearities associated with the physical models of electric grids, uncertainties from wind and solar generators and failures of system components, and discrete choices such as where to upgrade infrastructure. The project will bring transformative change by providing operators with the computational tools needed to accurately model heavily stressed power grids. This will be achieved by combining new machine learning techniques with advanced nonlinear optimization algorithms and novel power system modeling methods. The project’s intellectual merits include the development of new solution algorithms for the optimization problems encountered in power systems during extreme events. The broader impacts of the project include mitigating the impacts of climate change as well as educational efforts to develop video game style simulations focused on power system resiliency. In the spirit of citizen science, the players' solutions to these simulations will form a crowdsourced dataset that will be used to train the machine learning models in the project's research efforts, closing the loop between research and education. The goal of this project is to develop the fundamental theory and algorithms for addressing the heavily stressed conditions inherent to power systems during extreme events. Accurately modeling these heavily stressed conditions yields stochastic mixed-integer nonlinear optimization problems that are intractable with existing theory and algorithms. Existing approaches address these challenges using assumptions that are inapplicable for the atypical conditions inherent to extreme events, resulting in large errors and resiliency plans that fail to adequately reduce the impacts of extreme events. This project will develop new algorithms that can accurately model power flow nonlinearities, uncertainties, and discrete decisions without sacrificing computational speed and reliability. To accomplish this, the project will improve and combine alternative power flow models, mixed-integer programming solvers, machine learning techniques, and nonlinear optimization to create tailored theory and algorithms for resiliency applications. 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.