(778e) Assessing Suspension Homogeneity Using CFD-DEM for Enhanced Content Uniformity of Spray Dried Intermediates

Ensuring a uniform concentration of
API in the feed suspension/slurry of a spray drying process is a crucial step
towards achieving a uniform dosage of the drug product intermediate. To
accomplish this, an adequate feed tank design and process conditions are
required to avoid heterogeneity of the suspension driven by, e.g., density
differences between the solid and fluid phases. However, this heterogeneity can
only be assessed a posteriori and the process optimization can only be
achieved by trial and error. Here we present an alternative to this iterative
and cost-ineffective development strategy by employing computational tools such
as computational fluid dynamics coupled to discrete element methods (CFD-DEM)
that allow offline experiments to be performed for various process conditions.
These tools can also enable optimization of the agitation elements, such as
number of impellers and distance to the bottom.

Here we focus on a spray
drying case study where the feed suspension consists on a solid phase (the
active pharmaceutical ingredient, or API) and the fluid phase is a solvent with
a dissolved binder and other excipients. The process is run in batch mode while
controlling key operation parameters, such as reactor wall temperature and
agitator rotating speed, to obtain a homogeneous distribution of particles in
the volume of the tank. Mixing in stirred tank reactors is enabled by fluid convection
(at a larger scale) and turbulent exchanges (at a smaller scale) [1,2]. Therefore detailed information on the vessel,
baffles and impeller (see Figure 1a), which determined the flow patterns, is
required [2], [3].

A
computational fluid dynamics (CFD-DEM) methodology (see Figure 1b), based on
the OpenFOAM® framework, is used to aid process development and foresee
potential issues related to accumulation zones within the reactor.

Upon choosing the
reactor, its configuration and operating conditions, a steady-state simulation
of the turbulent fluid flow, modeled via the Reynolds averaged Navier-Stokes
(RANS) equations and  shear stress transport (SST) turbulence model is
conducted. The selected transient solver then uses this pre-calculated velocity
field to update particles velocities considering the effect the forces exerted
on them by fluid drag; gravity; and Coriolis acceleration. Furthermore, other
capabilities such as particle-wall and particle-particle collision are
available and may be incorporated into the solution (at the cost of extra
computational effort). This way, the present workflow offers insight on how the
interchangeable parts and operating conditions of the reactor might influence
the steady-state particle concentration distribution.

[1]       A. Ochieng and M. S. Onyango, “CFD
simulation of solids suspension in stirred tanks: Review,” Hem. Ind.,
vol. 64, no. 5, pp. 365–374, 2010.

[2]       G. R.
Kasat, A. R. Khopkar, V. V. Ranade, and A. B. Pandit, “CFD simulation of
liquid-phase mixing in solid-liquid stirred reactor,” Chem. Eng. Sci.,
vol. 63, no. 15, pp. 3877–3885, 2008.

[3]       P. K.
Biswas, S. C. Dev, K. M. Godiwalla, and C. S. Sivaramakrishnan, “Effect of some
design parameters on the suspension characteristics of a mechanically agitated
sand–water slurry system,” Mater. Des., vol. 20, no. 5, pp. 253–265,
1999.