Decentralised control strategy for effective utilisation of distributed community energy storage in PV-rich LV distribution network

Abstract

Integrating a high-capacity residential-scale photovoltaic (PV) system into a low-voltage (LV) network results in voltage rise. Conversely, voltage drops occur during periods of high demand. Furthermore, PV generation characteristics vary with seasonality and daily weather conditions, complicating voltage rise and drop management. On the other hand, community energy storage (CES) emerges as a solution for future community energy management. This study presents an application of distributed CES to effectively manage voltage rise and drop, addressing a prominent challenge accompanying the widespread adoption of PV. However, upon introducing the CES and its accompanying energy capacity, the control strategy must effectively incorporate the utilisation of the CES’s available capacity year-round. In brief, CES control should adapt to the characteristics of PV by charging the CES during high PV generation. It should also release active power to meet high demand by discharging the CES. This will enable both voltage management and effective CES capacity utilisation in response to PV variations due to seasonality and daily weather conditions. The goal mentioned above can be effectively accomplished by forecasting PV output and using the data to adapt to PV characteristics in a centralised manner. However, due to the challenges associated with centralised control, the present paper proposes a decentralised control algorithm that determines charging and discharging power rates for distributed CES, aligning with the PV and load characteristics in an LV distribution network. The proposed control framework consists of two stages: the first stage regularly updates power references based on clear-sky PV profiles, active power measurements at the LV transformer, and CES active power and energy level. The second stage is activated to mitigate voltage transient caused by fast-moving clouds and small increments in solar irradiance, load between consecutive updates of the first stage. To illustrate and validate the concept, a 13-bus LV distribution network is used, and the proposed control algorithm is implemented on a co-simulation platform that combines MATLAB with a real-time digital simulator.