A digital twin of a vacuum filter-bed dryer

Drying active pharmaceutical ingredients in an agitated filter dryer (AFD) is a typical unit operation in drug substance manufacturing. Many simultaneously occurring phenomena, such as the mixing of solids, liquid and vapor transport, heat conduction, evaporation and latent heat, complicate the mathematical description of the drying process. To date, few reliable models capable of accurately predicting the drying performance in an AFD have been developed. In this work, a reduced-order compartment model was developed that relies on (i) experimental data and (ii) Discrete Element Method (DEM) simulation results. The model was fit to experimental results obtained in a lab-scale AFD. The compartment model was used to evaluate various agitation strategies. Simulations at various stirrer speeds showed that low rotation rates significantly reduced the drying time compared to runs without mixing. With increasing speed, the drying time decreased until the bed was well-mixed. After that point, while higher agitation rates no longer affected the drying time, they could lead to higher attrition. Furthermore, it was observed that intermittent mixing could reduce the total mixing time (and potentially attrition) while maintaining high drying rates. It was established that shorter mixing/pause intervals improved the drying performance even if the moisture was not homogeneously distributed after the mixing periods.