1384523

11234

Supervisory control of fed-batch fermentation processes is a difficult problem because the process is inherently non-linear, operated at unsteady state, and often poorly modeled. This problem is exacerbated in laboratory and pilot-plant applications due to lack of historical data from similarly operated batches. In this paper, we propose an approach for the supervisory control of fedbatch fermentation during process development. Transitions in the process are explicitly modeled and characterized using simple multivariate rules. Online data is then used to identify the occurrence of transitions and thus track the process across different phases. Supervisory control actions such as sequence control, phase-specific regulatory controllers and alarm settings are then subsequently executed based on this knowledge of the current phase. One key advantage of this approach is that it does not require detailed process knowledge or extensive process data and is thus well suited for application in process development. This approach has been implemented as an online expert system, called iProphet, and successfully tested on two laboratory-scale process development case studies—a bacterial (Escherichia coli) fermentation and laboratory-scale yeast (Pichia pastoris) fermentation. The approach and results from the two case studies are presented.

367

Fermentation , Expert system , Process monitoring , pharmaceuticals

Studia Iranica

382

382