ABSTRACT
Regulation of a continuous bioreactor with Saccharomyces cerevisiae is investigated. A number of different sensors are evaluated for this purpose and the process dynamics is investigated around the critical dilution rate. A sensor for reducing gas concentration in exhaust gases is selected for regulating the substrate flow rate. Closed loop identification experiments are carried out to enable identification of the process dynamics near the critical diluton rate. Due to the time-varying nature of this process an adaptive regulator seems to be a promising tool for providing good regulatory and setpoint tracking performance. A simple third order model is used for a model based control design with a Linear Quadratic (LQ)-regulator. The LQ-regulator performs well experimentally, both in an adaptive version where the model parameters are updated on-line, and in a non-adaptive version. During the test the process is exposed to a large disturbance in substrate feed concentration and to a small setpoint disturbance. The proposed regulator is a practical realisation of a productostat where the product in this case is an undesired primary metabolite. Thus, this paper demonstrates a more general principle of utilizing metabolic overflow metabolism for directing fluxes through a desired metabolic pathway. This principle is applicable in the presented form, if a (by-)product can be measured on-line.
