Estimation of biomass and secondary product in batch fermentation

Abstract

The work described in the paper represents a stage in the development of an industrially realistic, adaptive control strategy for those many batch fermentation processes where biomass and secondary product concentrations are difficult to measure accurately, reliably and rapidly enough to allow good on-line control to be achieved.

The paper describes results achieved in the estimation of biomass and secondary product formation in batch fermentation processes. The first results show that the most serious obstacle to producing a reliable and practicable on-line estimation algorithm is the presence and effect of batch-to-batch variations in the fermentation process. The problem is investigated on data from Sorbose fermentation and it is shown that estimation accuracy is most affected by two particular process parameters, one of which is found to be time-varying. The incorporation of this varying parameter into the on-line estimation scheme compensates well for batch-to-batch variations and good estimation accuracy is obtained even with an incorrect process model. The paper includes a description of the prototype hardware for on-line estimation.