LAUSR

Abstract Moisture content is a critical quality attribute in drying of pharmaceutical formulations. This paper proposes a hybrid soft sensor for online real-time estimation of the product moisture in batch fluid bed dryers. Major applications include end-point detection, feed-back control, and process optimization resulting from increased process understanding. The proposed soft sensor utilizes commonly available measurements in a hybrid first-principle/empirical mathematical framework with few parameters to calibrate. Each parameter has a physical meaning in the model, enabling quantitative comparison of the drying dynamics of different formulations, products, and equipment. The soft sensor model requires experimental data from few batches for calibration, and historical data from production batches can be used for this purpose when available. Three case studies, two in pilot plant using different formulations and one using historical data from manufacturing batches, are presented in this paper. The results support the proposed soft sensor model as a robust, practical and accurate method for online estimation of moisture in fluid bed dryers.