A Distributed Geyser‐Inspired Algorithm for Minimizing Losses in Flywheel Array Energy Storage Systems

• Published in: Concurrency and computation Vol. 37; no. 23-24
• Main Authors: Chen, Jianan, Nusyirwan, Istas Fahrurrazi, Ahmad, Robiah, Razak, Fadhilah Abdul, Jing, Lili
• Format: Journal Article
• Language: English
• Published: Hoboken, USA John Wiley & Sons, Inc 25.10.2025
• Subjects: distributed computing; flywheel array energy storage system; geyser‐inspired algorithm; loss minimization; power allocation
• ISSN: 1532-0626; 1532-0634
• DOI: 10.1002/cpe.70181

ABSTRACT Flywheel array energy storage systems (FAESS), due to their high power density, rapid response time, and long operational lifespans, have come to be recognized as one of the best alternatives for renewable energy storage on a large scale. However, the scarcity of efficient working energy systems results in impeded performance and reliability of the entire system. This paper presents a new distributed architecture of the Geyser‐Inspired Algorithm (GEA), which allows energy loss minimization using a dynamic load assignment among flywheels. This architecture uses dynamic load sharing among flywheels to minimize energy loss. The algorithm works in a distributed way, with each flywheel unit running its own version of the control logic that was inspired by geyser dynamics, enabling real‐time responses to dynamic load changes and system failures. The effectiveness of the proposed GEA is verified through extensive simulations and experimental validation. In simulations, GEA outperforms conventional control strategies such as Proportional Allocation and Round‐Robin Scheduling, showing a reduction in total energy losses by up to 30%, an average State of Charge (SoC) imbalance improvement to 6.2%, and a significantly enhanced real‐time responsiveness with an average response time of about 0.8 s. Moreover, parameter sensitivity analysis demonstrated robust performance across different operational thresholds, with minimal variations in energy loss and response time, confirming the stability and adaptability of the proposed method. Additional validation scenarios, including random load fluctuations and multiple simultaneous flywheel failures, further confirmed the robustness and fault‐tolerance of GEA. Scalability analysis also indicated efficient computational performance, with execution times increasing modestly from 0.85 ms for four flywheels to 4.60 ms for twenty‐four flywheels, underscoring GEA's applicability in larger‐scale energy storage applications. Through the integration of nature‐influenced heuristics and engineering tools in a consolidated manner, our study highlights an avenue through which the design of robust, scalable, and fault‐tolerant control methods in large‐scale electrical energy storage systems is made possible. Given that the point of distribution of the Geyser‐inspired algorithm allows for lesser losses and greater adaptability in the fast‐changing power grid, the distributed Geyser‐inspired algorithm is key in the development of FAESS, a type of battery energy storage system.