(680d) Material Property Database and Its Applications in Predictive Batch and Continuous Unit Operation Modeling

This work includes GSK’s efforts in collaboration with Rutgers to build a database with relevant material properties. An important objective of a database is not just to compile the material data but link it to different applications. Towards that end, a bottom up approach was identified, where the unit operations of interest were listed. The required predictive capabilities and the desired output for each of these unit operations were identified. The material attributes that impacted the performance and the output of these unit operations were identified and were decided to be included in the database. Materials were characterized for their flow (characterized using two different shear cells), density, permeability, basic flow energy, wall friction and particle size. A Brookfield Powder flow tester and an FT4 rheometer were used to characterize the flow of powders. The database enabled comparing the output from two different shear cells as well as two different attachments (sample holders) for the same shear cell. The database was linked to a variety of applications; predicting feeder variability, demonstrating mixing rules, surrogate identification and identifying redundant material properties.

The comparison of the Brookfield powder flow data obtained from the shear cell test using two different sample holders demonstrated that the data were not interchangeable. The discrepancy in the data between the two cells was more pronounced for good flowing materials. A characteristic line was shown to exist between the measurements from the two cells, allowing the data to be used interchangeably. A comparison between the repeatability of the measurements from FT4 and Brookfield was also conducted and the reproducibility of each of the measurements was determined. An important goal for such a database is to identify surrogates for the desired unit operation as well understand the redundancy in several material properties. The ability to merge different databases was also investigated. Merging different databases, compiled within the same or different organizations, can help expand the given database rapidly and cost effectively, saving resources and the need to repeat testing for some common excipients. It can also help in standardizing the methodology for data collection. A similar approach will be demonstrated where two different databases containing similar material properties were analyzed and approaches will be demonstrated to combine them, if feasible.