Dynamic Aggregation of Energy Storage Systems Into Virtual Power Plants Using Distributed Real-Time Clustering Algorithm

• Published in: IEEE transactions on industrial electronics (1982) Vol. 68; no. 11; pp. 11002 - 11013
• Main Authors: Zhang, Runfan, Hredzak, Branislav, Fletcher, John
• Format: Journal Article
• Language: English
• Published: New York IEEE 01.11.2021; The Institute of Electrical and Electronics Engineers, Inc. (IEEE)
• Subjects: Algorithms; Balancing; Batteries; battery; Clustering; Clustering algorithms; distributed clusters; Distributed generation; Energy storage; Estimates; Estimation; Heuristic algorithms; microgrid; Microgrids; multiagent; Power demand; Power plants; Real time; Real-time systems; reduction of power losses; Renewable energy sources; state of charge (SoC); Storage batteries; Storage systems; Virtual power plants; virtual power plants (VPPs); voltage regulation
• ISSN: 0278-0046; 1557-9948
• DOI: 10.1109/TIE.2020.3038074

Energy storage systems are widely used for compensation of intermittent renewable energy sources and restoration of system frequency and voltage. In a conventional operation, all distributed energy storage systems are clustered into one fixed virtual power plant and their state of charges are maintained at a common value. In this article, it is proposed to dynamically cluster the energy storage systems into several virtual power plants based on the energy storage systems' power demands and capacities. This results in reduced network power losses. The proposed dynamic clustering algorithm enables to cluster agents (energy storage systems) based on their preselected feature states (local power demands and energy storage capacities). To determine the clusters, the distance of the agents' current feature states from the average estimates of the states is determined in all clusters. The algorithm also provides average estimates of designated auxiliary states that can be used for control purposes. Presented RTDS-based real-time implementation results verify that clustering energy storage systems (batteries) into dynamic virtual power plants can reduce the network power losses.