Nonlinear model predictive control of vanadium redox flow battery

Abstract

In this paper, nonlinear model predictive controller (NMPC) is designed for stack voltage control in vanadium redox flow battery (VRFB) and the closed-loop performance is evaluated under different conditions. The controller is designed by using a simplified model of the VRFB system. The flow rate of electrolytes is the manipulated variable and the charging or discharging current is a disturbance. The unmeasured state variables (vanadium ions concentrations) and state of charge of the VRFB are estimated by measuring open-circuit voltage. Numerical simulations show the effectiveness of the NMPC under different operating conditions: when VRFB is balanced and its simplified model is perfectly known, when VRFB is imbalanced and some of its model parameters are not accurately known, and in presence of measurement noise. In all these cases the predictive controller ensures very good tracking performance and rejection of disturbances.