MULTISCALE FAULT DETECTION AND DIAGNOSIS IN FED-BATCH FERMENTATION

The importance of the FDA PAT guidelines in pharmaceutical process design space can be influenced by the introduction of robust process malfunction and senor fault detection and diagnosis tools. The paper compares a multi-scale multi-block modelling approach with conventional multiway PCA approaches for batch process monitoring. A benchmark penicillin fermentation simulation is used to evaluate the two methodologies. Contributions plots with confidence bounds enhance the fault diagnosis potential of the approaches studied. The methodology is in the process of being evaluated in fermentation and batch cooling crystallisation.