A computational fluid dynamics study of mass transfer in a large-scale aerated stirred bioreactor

Computational Fluid Dynamics (CFD) is exploited to study mass transfer in a specific stirred aerated bioreactor used in a cell culture process. The focus is on which empirical correlations from the literature can best be used for calculating the volumetric mass transfer coefficient k_La on the basis of the spatially distributed and/or average energy dissipation rate obtained in CFD simulations. This energy dissipation rate plays a key role in many of the empirical correlations which are reviewed in detail. CFD simulations are carried out using the finite volume (FV) ANSYS Fluent software as well as the Lattice Boltzmann (LB)-based code marketed by M−Star. In Fluent, we opted for a two-fluid approach and the realizable k-ε turbulence model, while M−Star models the turbulence by a Large Eddy Simulation and tracks individual bubbles in a Lagrangian way. Gassed power draw, air volume fraction, energy dissipation rate, and (k_La) are calculated in both codes and compared mutually as well as to experimentally measured data and analytical correlations available in the literature. The energy dissipation rate was underpredicted by Fluent, leading to lower breakup rates and an underprediction of k_La. The M−Star simulations also underpredict k_La although predicting much higher levels of energy dissipation. However, using a constant value for k_L and just the volume-averaged a from Fluent or M−Star improved the results significantly, which then are in good agreement with the experimental k_La value.