Experimental validation of a vanadium redox flow battery model via Particle Swarm Optimization

Conference Article


Iberian Symposium on Hydrogen, Fuel Cells and Advanced Batteries (HYCELTEC)





Doc link


Download the digital copy of the doc pdf document


Among all types of energy storage systems (ESS), vanadium redox flow batteries (VRFB) stand out for their high efficiency (70-90 %), long life cycle, safety and the possibility to decouple both energy and power sizes [1]. One of the main challenges in the study and analysis of VRFB consists on the correct caracterization of its performance through mathematical models. The importance of developing mathematical models is reflected in the literature, where numerous studies widely validated and referenced can be found, such as the electrochemical model developed by M. Skyllas-Kazacos [2], who pioneered the use of vanadium in redox flow batteries.

Models can be classified as static or dynamic, depending on whether the effect of time is taken into acount, and as distributed or lumped parameter models, depending on the space dimension [3]. Tools as COMSOL are used to analyze and develop distributed parameter models, to study effects of flow design, components materials or degradation mechanisms [4]. However, for control purposes, the distributed models are rarely used due to its complexity and relevant computational cost, both in resources and time. For that reason, a vast majority of works in the literature concerning VRFB control, use lumped parameters models.

Most of these lumped parameters models use different hypothesis to relax the problem, facilitating its understanding and analysis. This is the case of Skyllas-Kazacos electrochemical models mentioned, where the same flow rate is considered in both semicells,same species concentration inside the cell and tanks, as well as a constant temperature to estimate the open circuit voltage (OCV) by means of the Nernst equation [5]. Therefore, it is necessary to develop an analysis in order to define which variables and parameters will be modelled, and under which assumptions.

Thus, in this work, a complete model that considers the most important effects of a VRFB is presented. It can be divided into four submodels, according to the previous parameters, which are: electrochemical, voltage, thermal and hydraulic ones. This model has been developed in a Matlab-Simulink environment as a tool to analyze the performance of a VRFB in a wide range of possibilities, examining the behaviour of the species and the voltage, temperature and pressure variables, incorporating some improvements with respect to other works. For all submodels, a clear distinction is made between both sides of the system (catholyte and anolyte semicells), not only in terms of dimensions and flow rates, but also on the initial conditions.


nonlinear programming, observability, storage automation.

Author keywords

Vanadium redox flow batteries, Parameter estimation, Particle Swarm Optimization

Scientific reference

A. Clemente, R. Costa, M. Montiel, A. Lozano and F. Barreras. Experimental validation of a vanadium redox flow battery model via Particle Swarm Optimization, 8th Iberian Symposium on Hydrogen, Fuel Cells and Advanced Batteries, 2022, Buenos Aires (Argentina), pp. 115-117.