One-line monitoring, state and parameter estimation, adaptive/computer control and dynamic optimization of a continuous bioreactor
In this research program, various important aspects for computer-based adaptive control and optimization strategy for a continuous bioreactor, have been investigated. The study was carried out in the following four phases: (a) development of a new method for on-line monitoring of biomass concentration; (b) measurement of kinetics of growth of S.cerevisiae; (c) dynamic bioreactor simulation studies; and (d) experimental control of a continuous bioreactor. An interesting observation was made in absorption of light by yeast cells at "high" concentrations leading to a new equation: log (T/T0) = K log (C/C0), while developing a suitable method for on-line monitoring of yeast cell concentrations. A consistent theoretical explanation was developed starting from the fundamental assumptions of Beer-Lambert's law. This equation was shown to be valid for several optically sensitive solutions [with negative deviations] and thus has the potential of becoming a law. The underlying reasoning behind this phenomenon may improve our present day understanding about absorption of electromagnetic radiation by various substances. Based on this new concept, a novel spectrophotometric technique has been developed and successfully implemented for on-line monitoring of a wide range of yeast cell concentrations in a continuous bioreactor [which has been considered a difficult task in the literature due to lack of reliable instrumentation]. To the author's knowledge, this is the first successful method for on-line monitoring of "high" biomass concentrations which could be implemented for process control applications. This approach may lead to a new generation of instruments in spectroscopy. Extensive batch and continuous experiments were carried out on a well-defined medium using S. cerevisiae at different initial glucose concentrations. The biomass yield was found to be a function of the inhibitory environment of the bioreactor. Four new correlations have been proposed to explain the inhibitory kinetics of ethanol fermentation. These experimental results are expected to have a significant influence in formulating the fermenter design variables and control strategy for optimizing the productivity of ethanol fermentation process. Based on extensive simulation studies, an algorithm [called the SE algorithm] was successfully formulated using state equations: (a) for on-line estimation of important unmeasurable states and critical time-varying parameters; and (b) for adaptive control and dynamic optimization of a bioprocess. Based on simulation studies, a numerical technique was also developed to improve the convergence of the extended Kalman filter algorithm. The SE algorithm was implemented for on-line state estimation and dynamic optimization of a lab-scale [450 mL working volume] continuous ethanol fermenter. An IBM PC along with an OPTO board were used for on-line data acquisition and for execution of the optimization algorithm. A number of experiments were carried out to verify the performance and true adaptive nature of the algorithm. The experimental results clearly illustrate the successful development and implementation of computer-based adaptive control and dynamic optimization strategies to a continuous bioprocess.