Predictive Scheduling of a Penicillin Bioprocess Plant

In this paper a predictive strategy for the reactive scheduling of a multi-stage bioprocessing plant is outlined. In the procedure, the various batch stages of the bioprocess are dynamically re-allocated to the appropriate processing units in response to the biological variability inherent in each stage. Forecasts of the process productivity and consequent completion times for the tertiary stages of industrial penicillin fermentations are used in conjunction with a genetic algorithm to solve the scheduling problem. Initial results using data from a commercial penicillin plant demonstrate that the predictive scheduling framework could deliver increased production and, consequently, major financial benefits. Copyright © 2002 IFAC