Robust Nonlinear Model Predictive Control for Two-Stage Anaerobic Digesters

Abstract

Two-stage anaerobic digestion (AD) processes have been proposed to improve the operation of conventional single-stage AD systems. In this study, both standard and robust nonlinear model predictive controllers (NMPCs) were applied to a two-stage AD process treating tequila vinasses. NMPC provides a systematic methodology to handle nonlinearities and constraints on the manipulated and controlled variables, which are commonly found in the two-stage AD processes. In contrast, the robust NMPC framework aims to determine the control actions that minimize the offsets in the controlled variables while taking into account uncertainties that are likely to occur in the process, leading to the poor performance of the standard NMPC controller. Results showed an acceptable and superior closed-loop performance of the proposed robust NMPC controller with respect to the standard NMPC in the presence of disturbances in the inlet streams and set-point changes, including uncertainties.