Enhancing Anaerobic Digestion Performance with Offset-Free Model Predictive Control

This study presents the development of a nonlinear model predictive controller (NMPC) for an anaerobic digestion process, acting on the dilution rate to track a target methane flow rate. The NMPC framework employs a predictor based on a simplified two-stage anaerobic digestion model, combined with an Extended Kalman Filter (EKF) to estimate both the system states and an additional disturbance state, which primarily accounts for the model prediction bias. The estimation of the disturbance state introduces an integral action into the control strategy, enabling offset-free performance even when the simplified model cannot accurately represent the actual system dynamics. The proposed algorithm was successfully tested in a simulation environment using readily biodegradable soluble wastes, with the Anaerobic Digestion Model No. 1 (ADM1) serving as the plant emulator. Both the integral action and disturbance state estimation proved critical to the controller performance. © 2025 Elsevier Ltd