State estimation and model-based control of a pilot anaerobic digestion reactor

Abstract

A state estimator and various model-based control systems have been designed for a real anaerobic digestion (AD) pilot reactor fed with dairy manure. The model used is a modified Hill model which is a relatively simple dynamical AD process model. The state estimator is an Unscented Kalman Filter (UKF) which uses only methane gas flow measurement to update its states. The model and the state estimates are used in different control systems. One of the control systems aims at controlling the methane gas flow to a setpoint. Simulations indicate that the setpoint tracking performance of a predictive control system is considerably better comparing with PI control, while disturbance compensation is not much better. Consequently, assuming the setpoint is constant, the PI controller competes well with the predictive controller. A successful application of predictive control of the real reactor is presented. Also, three different control systems aiming at retaining the reactor at an operating point where the volatile fatty acids (VFA) concentration has a maximum, safe value are designed. A simulation study indicates that the best control solution among the three alternatives is PI control based on feedback from estimated VFA.