Optimizing of working conditions of vanadium redox flow battery based on artificial neural network and genetic algorithms

Abstract

The integration of electrode compression in a vanadium redox flow battery (VRFB) with optimized operating conditions is essential for achieving the maximum net discharge power. This study presents the development of a 3D steady-state model for VRFB aimed at maximizing net discharge power (Pnet) by optimizing key working conditions, including electrode compression ratio (CR), applied current density (Iapp), and electrolyte inflow rate (Qin). The comprehensive impacts of CR, Iapp, and Qin on VRFB charging and discharging curves, energy efficiency (EE), and system efficiency (SE), and Pnet were investigated. Subsequently, the optimization of CR and Qin under a fixed applied current density range of 20–400 mA/cm2 was carried out employing a hybrid methodology integrating artificial neural network (ANN) and genetic algorithms (GA). Results reveals that increasing CR does not linearly improve system efficiency and net discharge power, with an optimal CR identified to maximize net discharge power. The optimal CR varies with operating conditions and is influenced by factors such as Qin and Iapp. Notably, both CR and Qin negatively impact net discharge power due to increased pump losses, while Iapp and state of charge (SOC) positively affect net discharge power through enhanced electron transfer and higher active substance concentration, respectively. The optimal CR is identified around 65 % for Iapp exceeding 60 mA/cm2. Furthermore, the optimal Qin correlates directly with increasing Iapp, as higher Iapp rates necessitate greater Qin to sustain active species supply. This study offers an optimization approach for addressing coupled operating conditions and provides insights for achieving high net discharge power in VRFB operation.