A Comparative Study of MPC-Based Control Configurations of an Industrial Bioreactor to Produce Ethanol

Abstract For a class of continuous stirred tank bioreactors to produce ethanol at large scale, this work evaluates suitable control configurations in a model predictive control framework. The study makes several assumptions, including a non-square plant, a distinction between control and output variables, and the possible sets of available outputs. Since the outputs are obtained in a discrete form, also the effect of the sampling interval was evaluated. Operational-feasible configurations are discussed. Herein, it is highlighted that a conventional configuration of one input-one control variable driven by one output, is the best to maintain all the three states around their nominal values.