(108d) Modeling the Roll Compaction Process Using Finite Element Analysis

Roll compaction is part of a dry agglomeration process that takes place prior to tabletting. Its end product is agglomerate strips (?ribbons') that are ultimately milled for producing agglomerate granules. Owing to the granules' larger size, this process is more advantageous than direct compaction in many aspects, including reduction in dusting, improvement in flowability, and better die filling, all of which in turn makes capping less likely and production of high quality tablets more likely [1]. Recent experiments, however, have shown that density variations across the ribbon's width are common. Such variations in turn lead to a wide granule size distribution, which ultimately causes the process to lose its benefits.

The ribbon's non-uniform density distribution has been the subject of many previous studies. Those that took an experimental approach include Simon and Guigon [2, 3], who argued that the density distribution in the ribbon was the result of non-uniform material feed. Those that took a computational approach include Johanson [4] and Cunningham [5]. The former study was developed in the 1960s and was derived based on the empirical observation that roll pressure is a logarithmic function of ribbon density. Despite being attractive for its simplicity, this model provides only one-dimensional predictions and does not agree with many experimentally measured observations. The study by Cunningham [5] utilized the Finite Element Method (FEM) for two- and three-dimensional simulations of the roll compaction process. This study showed that FEM is capable of capturing the many characteristics of a roll compaction process. Quantitatively, however, the FEM predictions given in this study do not agree with experimental measurements. This discrepancy is likely because Cunningham did not account for the fact that mechanical properties of pharmaceutical powders are functions of their relative density.

This study revisits the FEM approach. However, unlike previous studies, the material mechanical properties?which are approximated using the Drucker-Prager Cap porous plasticity model?are varied depending on the local relative density using an external FORTRAN subroutine. To better approximate the roll compaction process, an Arbitrary Eulerian-Lagrangian (ALE) meshing algorithm is employed. This meshing strategy allows Eulerian boundaries to be defined at the feed (inlet) and exit (outlet) zones of the computational domain; essentially making the computational mesh act as a deformable control volume that allows material to enter from the inlet and exit from the outlet continuously. Further, an adaptive meshing scheme is also employed to accommodate for significant deformation.

The rollers and feed screw housing are modeled as un-deformable rigid bodies. Coulomb friction is assumed at the powder-roll interface. The effects of screw speed are approximated as a set pressure at the inlet zone. Methods used in approximating the inlet pressure will be discussed. Model outputs are validated experimentally by measuring the ribbon envelop density, as well as the normal and shear stresses acting on the ribbon. Normal and shear stresses predictions are also compared with Johanson's model. Reasons for discrepancies will be discussed.

References:

[1] Kleinebudde, P., 2004, "Roll compaction/dry granulation: pharmaceutical applications," European Journal of Pharmaceutics and Biopharmaceutics, 58, pp. 317-326.

[2] Simon, O., and Guigon, P., 2003, "Correlation between powder-packing properties and roll-press compact heterogeneity," Powder Technology, 130, pp. 257-264.

[3] Simon, O., and Guigon, P., 2003, "Roll press design--influence of force feed systems on compaction," Powder Technology, 130, pp. 41-48.

[4] Johanson, J. R., 1965, "A rolling theory for granular solids," ASME Journal of Applied Mechanics, E32(4), pp. 842-848.

[5] Cunningham, J. C., 2005, "Experimental studies and modeling of the roller compaction of pharmaceutical powders," PhD Thesis, Drexel University.