(551b) Advances in Discrete Element Modeling of High-Shear Wet Granulation Process Using Rocky-DEM

High-shear
wet granulation process is widely used in the pharmaceutical industry to
improve the bulk density and flow characteristics of a formulation. There are different particle-level phenomenon and mechanisms at
play during a granulation process, such as nucleation, agglomeration and
breakage. The challenge is to ensure that these mechanisms occur similarly at
different scales of the granulator to keep consistency in the quality of the
granules. The ability to accurately predict the stress and collisional behavior
in a granulator, would not only help our fundamental understanding of the
process but also help alleviate experimental costs during scale-up studies.

Discrete
element method (DEM) is a valuable tool to model such particle-level
interactions, but in the past, has suffered from practically restrictive run
times. In order to circumvent this, assumptions on particle size and shape
(modeled as spheres or glued-spheres) were routinely made. However, more
recently, advances in DEM technology, such as that provided by software Rocky-DEM,
have addressed some of these limitations. Rocky-DEM provides a step-change in
the ability to perform large scale DEM simulations significantly faster by
utilizing the large number of cores available in NVIDIA® graphics card (GPU
processing). The simulation run times using the GPU were compared with
traditional and more expensive CPU solvers.  In addition, several capabilities unique
to Rocky-DEM in comparison with other commercially available DEM codes were utilized
in this study; a few of them include the representation of Avicel using non-spherical
particle shapes (polyhedral shape representation) and the ability to include cohesive
forces and breakage without loss of mass or volume.

Rocky-DEM was
used to predict the movement of Avicel particles in a Diosna® granulator at 50%
volume fill (Figure 1).  The simulations were validated by comparing results to
experimentally-obtained surface velocity profile measurements using
video-imaging.   The validated DEM model was then used to establish a
relationship between various parameters on simulation measurements of bulk
properties (such as average velocity and stress distributions in the powder). 
Some of the variables included the particle stiffness (low and high), particle
shape (spherical vs Avicel-shaped), particle size (0.8, 1.8 mm), impeller speed
(Froude number vs tip speed), granulation type (wet vs dry particles), and
granulator scale (1L, 10L and 150L). Thus, by numerically predicting granular
pressures, velocities, and stresses in different regions of the granulator,
this study provides guidelines on achieving reproducible particle environments
with changes in scale and impeller speeds, and establishes the effect of
particle shape and size. The Rocky DEM platform enabled the utilization of
actual particle shapes and sizes in these studies and therefore are a closer
reflection of a real-case scenario.

Figure 1. A
cross-section of the granulator colored by the particle initial fill position
to show mixing profiles