Frequency Regulation with Heterogeneous Energy Resources: A Realization using Distributed Control

TL;DR

This paper demonstrates a real-world distributed control system coordinating heterogeneous energy resources for frequency regulation, achieving accurate tracking, economic benefits, and market participation potential.

Contribution

It introduces a practical implementation of distributed control for frequency regulation using diverse devices and real-time optimization algorithms in a power distribution grid.

Findings

Successful coordination of 69 active devices and 107 passive devices.

Accurate real-time regulation signal tracking over 40-minute periods.

Potential for up to $53K annual revenue from ancillary services market participation.

Abstract

This paper presents one of the first real-life demonstrations of coordinated and distributed resource control for secondary frequency response in a power distribution grid. We conduct a series of tests with up to 69 heterogeneous active devices consisting of air handling units, unidirectional and bidirectional electric vehicle charging stations, a battery energy storage system, and 107 passive devices consisting of building loads and photovoltaic generators. Actuation commands for the test devices are obtained by solving an economic dispatch problem at every regulation instant using distributed ratio-consensus, primal-dual, and Newton-like algorithms. The distributed control setup consists of a set of Raspberry Pi end-points exchanging messages via an ethernet switch. The problem formulation minimizes the sum of device costs while tracking the setpoints provided by the system operator. We demonstrate accurate and fast real-time distributed computation of the optimization solution and effective tracking of the regulation signal by measuring physical device outputs over 40-minute time horizons. We also perform an economic benefit analysis which confirms eligibility to participate in an ancillary services market and demonstrates up to $53K of potential annual revenue for the selected population of devices.