LAUSR

Abstract A soft sensor for on-line risk-based monitoring was applied for a 700-L Streptococcus thermophilus cultivation using a biochemical model that was coupled with a compartment model, the latter to account for mixing effects. The process risk, defined as the likelihood of not achieving the target biomass production per batch, was calculated continuously during the cultivation process. A Monte Carlo simulation accounted thereby for uncertainties in the model parameters. In the present cultivation, the estimated process risk was to lose ca. 3.5% of the total production capacity. The compartment model allowed the prediction of the spatial distribution of the pH in the bioreactor. The compartment model was based on a computational fluid dynamics (CFD) simulation and its computational speed (