(720d) Multivariate Data Analysis of Raw Material Properties from Pharmaceutical Powders for Predicting Compaction Behavior Using Finite Element Method

The development of a robust tableting process in a timely manner is still challenging for pharmaceutical solid dosage manufacturing. This is due to the lack of mechanistic process understanding, fundamental understanding of influence of raw material attributes upon tableting and limited usage of sophisticated process simulation tools. Experimentally determining material attributes that affect tabletability is easier and accurate. However, visualizing the effects of involved parameters such as punch velocity and die wall friction on the powder matrix under compaction and thereafter process optimization is very difficult, expensive and time-consuming. Therefore, numerical simulations of the compaction process based on Finite Element Analysis (FEA) can provide valuable additional information. FEA is used for calculating forces, deformations, stresses and strains throughout a bonded structure, which can later be used to assess possible compaction risks using small amounts of material.

A set of over 50 powders covering both excipients and active pharmaceutical ingredients (API) was extensively characterized using 20 techniques describing material attributes with a potential effect on tablet processability including particle size and shape, density, moisture content, powder flow, compressibility, aeration, surface area and triboelectric charging. Principal Component Analysis (PCA) was then performed to elucidate correlations between the powders and their measured properties. Two principal components (i.e., super properties) could be identified describing 69.0% of the variance within the data set of the characterized powders. PC 1 covered mainly the powder flow related variability, PC 2 explained density related variability and PC 3 was related to particle shape. Furthermore, the (dis)similarity in powder characteristics between all studied powders could be seen and explained. Based on this analysis, blends were composed containing multiple API’s and fillers covering a maximal area in the variability space determined via the PCA. Disintegrant, glidant and lubricant were kept fixed. Blend bulk properties were then characterized with a minimum number of relevant tests, also identified via the PCA of the raw materials.

The structure of these powders changes during compaction from a loose arrangement of particles with various shapes and sizes to a condensed structure that behaves like a continuum. To approximate the in-process mechanical properties for these raw materials and pharmaceutical blends which are of interest towards Direct Compression (DC), a modified Drucker-Prager Cap (DPC) plasticity model was calibrated using a compaction simulator. The mechanical behavior of the materials during compaction from this loose arrangement to the condensed structure was introduced in FEA using the calibrated DPC model. The model parameters were experimentally determined at different local relative density and were varied during simulation using an external USDFLD subroutine. This numerically predicted density distribution is compared with X-Ray Computed Tomography (XRCT) measurements to establish the predictive capability of the model.

Multivariate analysis of large data sets to extract powder characteristics correlated with in-process material behavior and tablet properties contains critical information that allows reducing the number of tests needed for process development in the future and thus less consumption of expensive API during development of tableting processes. This contributes to a better understanding of the impact of powder properties and process settings on the tableting process and final properties of the produced tablets. FEA process simulations provide a detailed and cost-effective means of understanding towards predicting the compaction properties of the formulation material based on the processing parameters. This combination of multivariate data and process simulations can later on be performed at other unit operations such as feeding, granulation and tableting to build end-to-end predictive platforms