Online Optimization of Power Response and Efficiency of Contra-Rotating Turbines Using Nonlinear Model Predictive Control

This paper presents a nonlinear model predictive control (MPC) architecture for low-head contra-rotating reversible pump-turbines (CR RPTs) in turbine mode. The developed MPC is compared with a baseline feedback controller in a frequency containment reserve (FCR) application. The MPC is designed to optimize the power response and efficiency online, with the former prioritized. A nonuniform time discretization and increasing weighting factors over the prediction horizon are implemented to reduce computational load and emphasize long-term optimality, respectively. Full reserve power step simulations on a 10 MW system demonstrate that the MPC reduces the rise and settling time by an average of 32% and 23%, respectively. The improved performance is achieved with minimal efficiency compromise, with the deficit becoming less than 0.1% within 8.5 s after reaching the setpoint. In FCR scenarios, the root mean square power tracking error is reduced by an average of 27% across various frequency datasets with a negligible reduction in efficiency. The MPC is experimentally validated on a reduced-scale hardware-in-the-loop setup, confirming the improved performance in a physical environment. The results show that the developed MPC enhances the dynamic power response of CR RPTs for grid support applications in low-head pumped hydropower storage systems.