Predictability of kLa in stirred tank reactors under multiple operating conditions using an Euler–Lagrange approach

Abstract

In industrial cell culture engineering, the production process consists of a multiscale seed train from lab scale to large scale. The oxygen demand of the cells has to be satisfied in all scales. Computational fluid dynamics (CFD) simulations can provide a tool to predict the mass transfer between the gas phase and the liquid phase. In this work, CFD was applied using an Euler–Lagrange approach for the prediction of the mass transfer coefficient () in stirred tank bioreactors for a wide range of operating conditions. The turbulent dissipation was studied for two different scales that show similar flow behavior. Breakup and coalescence of bubbles was not considered. A standard model was used for the simulation of turbulence and the mass transfer was assumed to be isotropic and turbulence driven. A minimalistic model was found, which was able to successfully predict the mass transfer behavior with high accuracy for the lab‐scale bioreactor (2.3 L) covering a wide range of typical operating conditions. In the given setup, bubbles remained close to the sparger, almost not interfering with the impellers. This supports the assumption of monodisperse bubbles for stirrer speeds between 140 and 260 rpm. Simulation results of an 80 L stirred tank reactor (STR) revealed the need to integrate physical phenomena like breakup and coalescence and a more sophisticated prediction of the initial bubble size distribution. Two‐phase Euler–Lagrange CFD simulations were performed for two differently scaled STRs and the mass transfer coefficient was calculated and compared to experiments in order to evaluate the applied models.