Comparison product optimization performance of fed-batch fermentation processes for penicillin g production

Abstract

This paper discusses the development of optimization strategies for a fed-batch penicillin fermentation process. To facilitate the study, a mathematical model of the system was developed. The model was simulated in MATLAB environment and good result was obtained. To provide on-line estimates of the difficult to measure penicillin concentration, Partial Least Squares model was employed. These estimates were then used in the control loop and successful control of the product concentration was established. This provided the opportunity for optimizing the fermentation operation, in particular to further increase the achievable product concentration while satisfying all process constraints. In order to carry out this procedure, two optimization algorithms were selected. First, dynamic optimization using direct shooting method and second is implementation single step ahead Dynamic Matrix Control (DMC). Comparison of these two different approaches was discussed and it was found this single step ahead DMC algorithm shown a best result with an optimization procedure.