429039 A Hybrid Experimental-Computational Method to Scale-up Powder Dissolution in Viscous Solvent: An Industrial Approach

Thursday, November 12, 2015: 9:30 AM
Ballroom D (Salt Palace Convention Center)
Avik Sarkar1, Rolf Larsen1, Thean Yeoh1, Rick F. Falk2, Pankaj Doshi1 and Mary T. am Ende3, (1)Pfizer Inc., Groton, CT, (2)Bend Research, A Division of Capsugel Dosage Form Solutions, Bend, OR, (3)Pharmaceutical R&D, Pfizer, Groton, CT

Dissolution of solid particulates in a liquid solvent is an important process in several pharmaceutical operations. Our application relates to topical ointment manufacture where a particulate ingredient must be dissolved in a viscous liquid. From a lab-scale process, we seek to scale-up this operation to pilot scale, and finally to commercial scale.

In this work, we developed a hybrid experimental-computational approach to simulate and scale-up the dissolution process. A population balance model (PBM), implemented in-house, served as the framework to track the size and distribution of particles as they evolve with time. Experimental dissolution kinetics data generated from lab-scale reactor vessels (Mettler-Toledo OptiMax-500) was used to calibrate the mass-transfer rate law. Our PBM framework also incorporates the effect of fluid-particle flow on the dissolution rate—the flow behavior for varying agitation conditions at different scales was characterized by performing a series of computational fluid dynamics flow simulations.  The calibrated PBM framework was then used to execute “virtual experiments” in lieu of expensive, actual experiments at the larger scales.

The initial particle size distribution is an inherent source of uncertainty in our process since size distributions of the solute may vary lot-to-lot. The solute powder also is prone to in-situ agglomeration when added to the viscous solvent—the extent of agglomeration is random and unpredictable.  Therefore, we quantitatively captured this size-distribution uncertainty in our modeling framework and developed confidence intervals, instead of point estimates, for the scaled-up dissolution time predictions.

Our work showcases a successful example where confidence-based model predictions have guided process development and reliably addressed scale-up of an industrial pharmaceutical operation.

Extended Abstract: File Not Uploaded