(233g) Towards a General Model for Twin-Screw Wet Granulation: Validation of a PBM Model through Linking Calibrated Model Parameters to Process Conditions

Verstraeten, M., Ghent University
Lee, K., Pfizer Inc.
am Ende, M. T., Worldwide Research and Development, Pfizer Inc.
Doshi, P., Worldwide Research and Development, Pfizer Inc.
De Beer, T., Ghent University
Nopens, I., Ghent University
In a newly developed three compartmental PBM model with a novel kernel for aggregation, prediction of both unimodal and bimodal behaviour for different L/S ratios with one single aggregation kernel structure is achieved. The use of one single kernel to achieve this is a major step forward as it points out that similar mechanisms are occurring but at a different rate for different L/S ratios. The model structure and calibration is presented in another abstract.

In the work presented here, the link between operational conditions (screw speed, mass flow rate, and L/S ratio) and the calibrated model parameters of the novel PBM is investigated. Since the experimental data is gathered using a three level full-factorial design of experiments (DOE), the effect of the experimental conditions screw speed, mass flow rate, and L/S ratio on the calibrated model parameters can be investigated. Separate relationships were retrieved and built into the PBM. The latter can now predict granule size distributions taking the experimental conditions as input.

For the validation, two new independent experiments were performed that were not used during model calibration. The L/S ratio of the centre point was perturbed in both directions leading to different levels of bimodality.

The predictions of the validation experiments were found in good agreement with the experimental data. This is a proof that the model is applicable in a whole DOE space, and that the PBM for TSWG of this hydrophobic compound can be considered as validated. Further, this means that the addition of mechanistic knowledge into the aggregation kernel is a vital part of modelling the correct behaviour inside the granulator. It significantly increased the predictive power of the model.