Model Aided Multiple Correlation Analysis between Preculture and Main Fed Batch Culture

Abstract

For an industrial antibiotic process the course of physiological state of both, preculture and main fed batch culture, is characterized by the kinetics of pH and acid fed F_A to control pH, respectively. These kinetics could be described not by manageable causal dynamic models but by nonlinear functions of time t. A fast and global convergent algorithm was developed to identify the parameters of these nonlinear functions for a great number of fermentation runs (>30). The model parameters for the kinetics of pH(t) of the preculture as well as the carbon feed F_C (t) and acid feed F_A(t) of the main culture are correlated by multiple regression analysis. The result gives a data-based link between preculture and main culture. It provides a quite general approach for preculture dependent control of main fed batch culture.