Parameter identification of thermal model of vanadium redox batteries by metaheuristic algorithms. (20th October 2021)

• Record Type: Journal Article
• Authors: Khaki, Bahman; Das, Pritam
• Is Part Of: Electrochimica acta. Volume 394(2021)
• Publication Date: 2021-10-20
• Subjects: Ant Lion Optimizer -- Battery Parameter Identification -- Battery Thermal Model -- Energy Storage Systems -- Genetic Algorithm -- Particle Swarm Optimizer -- State of Charge -- State of Health -- Vanadium Redox Flow Batteries; Electrochemistry -- Periodicals; Electrochemistry, Industrial -- Periodicals; 541.37
• DOI: 10.1016/j.electacta.2021.139131
• Languages: English
• ISSNs: 0013-4686

Abstract

Abstract: Typically, the battery stack temperature assumes constant in electrochemical or equivalent circuit models of the battery. However, experiments on a Vanadium Redox Flow Battery (VRFB) unit show that the temperature varies significantly during charge and discharge processes and considerably affects the value of the State of Charge (SoC) and other battery internal parameters. Therefore, monitoring and modeling the battery temperature is essential. An electrochemical based thermal model of VRFBs is adapted to the experimental data of a nine-cell VRFB unit in this study to find the optimal values of the coefficients and parameters of the VRFB's electrochemical and thermal models. The electrochemical model of VRFB and its thermal model include many unknown parameters and coefficients that need to be estimated. The objective of the current study is to determine the optimal value of all of these unknown parameters. According to the literature, the empirical methods are not sufficient to accurately determine the model's coefficients and parameters. An Optimization framework is defined to identify these coefficients by minimizing the mean square error between the measured experimental data of VRFB and the electrochemical model-based terminal voltage and electrolyte temperature. Moreover, a VRFB cell's electrochemical and thermal models are highly nonlinear; thus, excellent optimization techniques are needed to find these coefficients and parameters. A few metaheuristic optimizers are tested on the proposed optimization framework, where three of these algorithms have shown consistent and able to converge to reliable solutions. These algorithms are Genetic Algorithm (GA), Particle Swarm Optimizer (PSO), and Ant Lion Optimizer (ALO). The low RMS errors of the estimations by the metaheuristic-based algorithms compared to the measured data, show the accuracy of the proposed parameter identification approaches of VRFB's electrochemical and thermal model.