DYNAMIC PCA FOR PHASE IDENTIFICATION OF RIFAMYCIN B FERMENTATION IN MULTI–SUBSTRATE COMPLEX MEDIA

Information regarding when and how a fermentation process changes from one phase to the next is very useful to its modelling and hence control and optimization. In this study, we demonstrated that such information could be obtained by applying DPCA to online measurements of the fermentation process. The process under study is fermentation of Rifamycin B in a multi–substrate complex medium. We compare our observation to the results obtained from the simulation developed for the same system (Bapat et al., in press). The analysis showed that for the first 100 hours or so, the progress of the fermentation experiment in the DPCA score space matched very well to the developed simulation, which had been validated with actual off–line data (Bapat et al., in press). After that (ie. 100 hours onward), there is a significant difference between DPCA analysis result and the simulation result. The reason seemed to be that the simulation did not capture the effects of the secondary metabolism which becomes dominant at later stage of the fermentation.