(221d) Enhancing Powder Blending Processes through Calibrated Digital Twins and Machine Learning

This presentation highlights advancements in powder blending optimization using calibrated Digital Twin technologies. A rigorous calibration procedure for Discrete Element Method (DEM) simulations was conducted and validated on a 100-liter bin blender, ensuring simulation accuracy aligns closely with experimental results. Leveraging this calibrated dataset, a robust machine learning model was trained, effectively predicting blending performance across varying operating conditions. Further, to democratize access and usability of these advanced simulations, a customized web application was developed, integrating predictive modeling capabilities into a user-friendly platform. This approach enables rapid, accurate decision-making and significantly streamlines process development, showcasing the transformative potential of combining digital twins, machine learning, and accessible digital tools in pharmaceutical manufacturing.