Abstract-We consider the decentralized reactive power control of photovoltaic (PV) inverters spread throughout a radial distribution network. Our objective is to minimize the expected voltage regulation error, while guaranteeing the robust satisfaction of distribution system voltage magnitude and PV inverter capacity constraints. Our approach entails the offline design and the online implementation of the decentralized controller. In the offline control design, we compute the decentralized controller through the solution of a robust convex program. Under the restriction that the decentralized controller have an affine disturbance feedback form, the optimal solution of the decentralized control design problem can be computed via the solution of a finite-dimensional conic program. In the online implementation, we provide a method to implement the decentralized controller at a timescale that is fast enough to counteract the fluctuations in the system disturbance process. The resulting trajectories of PV inverter reactive power injections and nodal voltage magnitudes are guaranteed to be feasible for any realization of the system disturbance under the proposed controller. We demonstrate the ability of the proposed decentralized controller to effectively regulate voltage over a fast timescale with a case study of the IEEE 123-node test feeder.
I. INTROUDCTIONThe installation of rooftop and community solar facilities continues to increase in the United States. In California, for example, approximately 40% of all electricity demand was served by solar energy on the afternoon of to the large inrush current that results during switching operations [4]. Photovoltaic (PV) inverters, on the other hand, do not suffer from these limitations. Their reactive power injections can be actively controlled at a timescale that is fast enough to counteract the fluctuation in demand and PV active power supply. Our objective in this paper is to develop a systematic approach to the design of decentralized controllers for PV inverters, in order to effectively regulate network voltage profile in real-time, while guaranteeing the robust satisfaction of network and individual inverter constraints.Related Work: A large swath of literature treats the reactive power management of PV inverters as a centralized optimal power flow (OPF) problem, which aims to minimize a network-wide objective function (e.g., voltage regulation error) subject to network and resource constraints [5]- [10]. To set its reactive power injection, each PV inverter communicates its local measurements of demand and PV active power supply to a central computer. Using this data, the central computer solves a centralized OPF problem, and subsequently transmits the optimal solution back to each inverter for local implementation. Due to the rapid variation in the active power supply from PV resources, the reactive power injections of PV inverters need to be updated repeatedly over a fast time-scale (e.g., every minute). In the presence of a large number of PV inverters, the imp...