A Constrained Extended Kalman Filter for State-of-Charge Estimation of a Vanadium Redox Flow Battery With Crossover Effects

One of the main issues with vanadium redox flow batteries (VRFBs) is that vanadium ions travel across the membrane during operation which leads to a concentration imbalance and capacity loss after long-term cycling. Precise state-of-charge (SOC) monitoring allows the operator to effectively schedule electrolyte rebalancing and devise a control strategy to keep the battery running under optimal conditions. However, current SOC monitoring methods are too expensive and impractical to implement on commercial VRFB systems. Furthermore, physical models alone are neither reliable nor accurate enough to predict long-term capacity loss due to crossover. In this paper, we present an application of using an extended Kalman filter (EKF) to estimate the total vanadium concentration in each half-cell by combining three voltage measurements and a state prediction model without crossover effects. Simulation results show that the EKF can accurately predict capacity loss for different crossover patterns over a few hundred cycles.