Data-Driven Operation and Control of Active Power Distribution Systems with High Penetration of Distributed Energy Resources and Energy Storage

The expansion of large scale temporally changing and spatially separated distributed energy resources (DERs), storage devices and, flexible loads exhibit inherent computational challenges for management of active power distribution systems. Also, time-scale of operation of these devices and configuration requirements such as micro grid formation, involves integrating dynamic system models within the optimization framework. Considering these challenges, this project aims to develop novel algorithms for optimal power flow in power distribution systems, distributed modeling of subnetworks in power distribution system with large number of active devices, and a secondary control framework that can ensure seamless integration with vendor driven controllers. Real-life data and models from the local utility will be utilized to demonstrate feasibility of the methodology. Educational and outreach activities of this project include a) engaging students in problem-based learning from diverse undergraduate and graduate groups including under-represented and minority students, b) designing advanced curriculum on power system control with convex optimization and distributed control approaches, c) providing a platform to motivate and attract students in engineering especially the underrepresented minorities and women, and d) developing a comprehensive dissemination platform through PIs research lab and the center at University of North Carolina at Charlotte.

The project will investigate: a) a novel Receding Horizon Control (RHC) based mixed-integer second order cone programming (MISOCP) model for optimal power flow in power distribution systems that can scale up to integrate thousands of aggregated nodes and provide set points (integer or real) for passive and active devices considering unbalanced distribution system operation, b) a stochastic model predictive consensus framework for distributed modeling of subnetworks in power distribution system with large number of active devices, c) a secondary control framework that provides improved active/reactive power control and can ensure seamless integration with vendor driven controllers in turn enhancing power quality and stability, and d) an implementation platform including communication loops with real-life data and models from the local utility that proves feasibility of the methodology. The proposed optimization framework can provide global solutions for decision control variables and set points, including switches and transformer taps, at all active nodes in power distribution system. Also, the methodology can incorporate stochastic or deterministic changes in the devices such as DERs and energy storage, considering each subnetwork. Moreover, the architecture can be seamlessly integrated with the existing vender driven controllers thus capable of accommodating high in-feed of distributed resources and providing a low-cost solution to exponential increase in decision and grid state variables.

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.