Skip to main content
eScholarship
Open Access Publications from the University of California

Model-Free Optimal Control of Distributed Energy Resources in Electric Power Distribution Systems

  • Author(s): Arnold, Daniel Brian
  • Advisor(s): Auslander, David M
  • et al.
Abstract

Distributed Energy Resources (DER) have the potential to both improve the manner in which the electric power distribution system is operated and provide services to the bulk electric grid. A key question that has emerged in academic circles is the manner in which DER is governed to provide such utility. While many of the suggested approaches are mathematically elegant and demonstrate positive performance in simulations, they face enormous barriers to implementation due to their reliance on knowledge of network models and burdensome communications requirements. This dissertation focuses on the application of an Extremum Seeking (ES) control approach to manage individual DER, that does not require utility network models and or communication between agents. Results achieved from the application of this technique are shown to be equivalent to those derived from centrally-based optimization strategies that utilize convex relaxation techniques. Initially, I explore the use of Extremum Seeking to manage reactive power resources to perform voltage constrained optimization in balanced radial distribution systems. Next, also for balanced and radial distribution systems, I extend this work to consider the management of four-quadrant-capable DER to enable voltage-constrained distribution substation real power and voltage reference tracking. My approach essentially allows the distribution feeder to act as a controllable (P,V) bus when viewed from the perspective of the Transmission System Operator. Finally, in an attempt to extend my results to 3-phase distribution systems, I derive a linear approximation to unbalanced power flow and explore a new objective of using reactive power resources to regulate and balance system voltages. While the latter-most piece does not specifically consider ES, it lays the necessary foundation for the application of the approach in unbalanced systems.

Main Content
Current View