Collaborative Research: Power System Flexibility: Metric, Assessment, and Algorithm

This NSF project aims to establish critical metrics and develop a comprehensive evaluation framework to assess the flexibility of power systems. The project will bring transformative change to the power industry by supporting practitioners and researchers to better understand and improve power system flexibility, which is especially crucial due to the volatility and unpredictability of net loads intensified by the increasing penetration of renewable energy.

The proposed work includes i) defining metrics to describe complicated flexibility information, ii) developing a unified mathematical modeling framework with various underlying modeling structures, and iii) designing broadly applicable, efficient, and scalable solution approaches for different metric or application combinations. The intellectual merits of the project include enriching the understanding of system flexibility from different perspectives, enabling the comparison of flexibility across power systems, and providing a safety and resilience enhancement direction for power system planning and operations.

The broader impacts of the project include advancing the theoretical foundations in the interdisciplinary field of optimization and artificial intelligence as well as applying cutting-edge learning techniques into traditional engineering fields.

The study will address challenges in the existing literature by proposing innovative scientific methods in three aspects. (1) The employment of multiple flexibility metrics satisfies the needs of investigating flexibility on individual buses and the whole system, while a single metric is not able to reveal enough information on the high dimensional feasible regions of net loads. (2) The flexibility metric assessment models that involve nonlinearity, discreteness, and nonconvexity, are significantly harder to solve compared to their linear counterparts. To handle the complex cases, the project will use a wide range of modeling techniques, including mixed-integer, two-stage and multistage formulations to reduce the complexity of measuring flexibility. (3) The proposed hybrid mixed-integer programming and deep learning algorithm will significantly improve the efficiency to obtain critical information for flexibility assessment in a time-critical environment and meet the requirement of operations practice.

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.