461485 Evaluating Mixing Performance in a Stirred Tank Reactor with Different Impeller Geometries and Reactor Size By CFD

Wednesday, November 16, 2016: 8:30 AM
Union Square 1 & 2 (Hilton San Francisco Union Square)
Rui C. Silva, Filipe A.P. Ataíde and José L.C. Santos, Hovione, Lisbon, Portugal

Stirred tank reactors are ubiquitous in the pharmaceutical industry and their mixing efficiency is paramount on the rate of heat and mass transport processes and on the reaction rate as well. Traditional experimental measurement techniques are restricted to the limited number discrete of points were the measurements are taken. Moreover, measurements are difficult in reactors functioning at extreme conditions involving cryogenic temperatures or high pressures.

The numerical approach circumvents this experimental limitation and allows for the flow pattern to be evaluated to its full extent. Furthermore, CFD minimizes the risk associated with scaling up processes involving complex geometries where classical empirical correlations might not ensure the mixture components are well mixed. CFD also allows to assess the influence of several parameters and operational conditions to define an operational design space, very much aligned with the Quality-by-Design (QbD) approach. This enables an enhancement of process knowledge and to easily troubleshoot potential issues occurring in the production trains.

The objective of this work is to evaluate the mixing performance of a stirred tank reactor by assessing the distribution of a solute starting from a stagnant layer at the bottom of the reactor. Different impeller geometries and reactor scales were tested using Computational Fluid Dynamics (CFD). These simulations were implemented to study a single phase incompressible fluid by means of a RANS-type finite-volume (FV) method. Turbulence was modeled by the two-equation standard high-Re k-epsilon turbulence model, as implemented in OpenFOAM, in tandem with both a Moving Reference Frame (MRF) and Arbitrary Mesh Interface (AMI) approaches to account for the effects due to impeller rotation. New solvers were developed to determine the concentration field of the tracer solute: a steady state solver for MRF and a transient state one for AMI. The passive scalar transport approach was applied to simulate the transport of a scalar quantity in an incompressible fluid flow. The primary assumption of this approach is that the solutes are non-reactive, with no influence on the flow field, acting therefore as numerical tracers in the flow and thus are used to evaluate the mixing efficiency of the impeller.

The steady state and transient flow fields were calculated for given flow conditions using new solvers based on simpleFoam (Semi-Implicit Method for Pressure-Linked Equations) and pimpleDyMFoam (merged PISO-SIMPLE) , respectively. These new solvers include the transport equation for a passive scalar that is solved along with the velocity and pressure fields. The new solvers were used to compare the suitability of either approach (MRF and AMI) in the description of mixing of the passive scalar in typical pharmaceutical reactors. In this study a 95 % mixing uniformity was considered as an indication of a well mixed system.

Extended Abstract: File Not Uploaded