Disturbance performance comparison of predictive controllers in SCFs

Abstract

This study performs a comparative analysis between a linear and a non-linear predictive controller (MPC and NMPC, respectively) in controlling the output temperature of a solar thermal collector field (SCF), the main component that dictates the overall efficiency of the solar thermal plant. The controllers are evaluated using a validated non-linear model of a real solar plant facility located in the CIESOL center at the University of Almería, Spain. While the linear MPC is based on a linearized model implemented at a specific working point of the plant, the NMPC uses the full nonlinear dynamic model to predict the outlet temperature under similar conditions. The comparison considers the effect of the computational load of nonlinear controllers due to the nonlinear optimization problem, which can compromise disturbance rejection performance on cloudy days. Therefore, both controllers are explored in an ideal sunny day scenario as a benchmark result, and subsequently, the MPC and NMPC are evaluated in a critically cloudy scenario. The simulation results show that the controllers have similar behavior in both scenarios; however, the NMPC demonstrated to have lower control effort compared to the MPC in cloudy day scenarios. These findings offer insights for future applications of predictive controllers considering effectiveness and performance in disturbance rejection, which can culminate in improving the efficiency of SCFs.