Moisture soft sensor for agitated pan dryers using a hybrid modeling approach

A hybrid soft sensor model is developed to provide real-time estimations of the moisture content in the Active Pharmaceutical Ingredient (API) wet cake for an agitated pan drying process. The soft sensor is calibrated using data from 5 batches. The estimation accuracy of the soft sensor is demonstrated with additional 21 commercial scale batches. Dynamic global sensitivity analysis has been performed to investigate the significance of input process parameters on the variability of soft sensor estimations over the whole drying process. The results indicate that the soft sensor acts as a powerful tool for real-time estimation of the moisture content, which can consequently be used to monitor and control the API drying process and enable Quality by Design (QbD) approach for this important pharmaceutical process.