(700c) Impact of Material Attributes on Continuous Mixing Quality: A DEM Study

Digital twins are virtual models designed to accurately reflect the physical objects or processes. In pharmaceutical manufacturing applications, digital twins can aid the design of pharmaceutical production processes, which is especially beneficial when the material availability is low and experimentation at production scale is too costly or even impossible. DEM based digital twins offer detailed insight to the process, but require careful calibration of the material behavior. The digital twin is in principle only valid for the calibrated materials. Thus, a sensitivity study is necessary to quantify how robust or sensitive the digital twin is to changes in the material behavior.

The current work was based on a calibrated and validated DEM model of a cohesive powder in a horizontal continuous mixer. The first part of sensitivity study analyses the influence of the individual DEM contact parameters on the residence time distribution (RTD) of the continuous mixer. The aim of this study is to identify the DEM parameters that have the highest impact on the process outcome and thus require particular care during the calibration procedure.

The second part of the sensitivity study analyses the impact of batch-to-batch variation with a concrete example: Two blends with the same formulation but materials from different batches were experimentally characterized and subsequently used to develop a calibrated DEM model. The technical difference to the first part is that all contact parameters are changed simultaneously to produce suitable DEM contact models for both material batches. The DEM results show that different operating conditions react differently to the batch change. Experimental trials have been performed at nine operating conditions with one material batch, and only one of the two DEM parameter sets match the RTDs of all nine operating conditions at once. These results demonstrate that the presented calibration procedure creates a digital twin that is sufficiently sensitive to detect batch-to-batch variability.

Disclaimer: This abstract reflects the views of the authors and should not be constructed to represent FDA’s views or policies.