Numerical study of flows related to aerated stirred tanks

Abstract: The overall purpose with this work is to investigate the bubbly flow in a aerated stirred tank using numerical simulations. Aerated stirred tanks are commonly used in chemical processes for producing for example insulin and antibiotics. The main requirement of these tanks, is to provide an optimal environment for the microorganisms found inside, with homogeneous mixing of air. The flow in a aerated stirred tank is complex and turbulent with stagnation points, swirling motion and recirculation zones. Introducing bubbles in this environment creates a wide range of bubble sizes. For the aerated stirred tank, Large Eddy Simulation (LES) is used for the continuous phase and the two-way coupled Eulerian-Lagrangian model for the dispersed phase. A break-up and coalescence model has been incorporated. Since the bubbles inside a bioreactors are many and with a wide range of sizes as well as largely varying inter-droplet distance, the underlying assumptions of the model can have significant error. However, the approach offers computational efficiency and allows one to include bubble breakup and coalescence models. The moving blades, representing the rotating impeller, is modeled by using the Volume of Solid (VOS) model. The averaged radial and tangential liquid velocities decreased with increasing gas volume fraction. Additionally, for the axial velocities the gas redirected the radial jet upwards and the symmetry of the ring vortices vanished. Although, low gas flow rate, the periodicity from the impeller is decreased and interfere with the creation of the trailing vortex pair behind the impeller. Including bubble break-up and coalescence model in the aerated stirred tank, induces small spherical bubbles. For large bubbles, shapes become important and can be modeled using the Volume of Fluid (VOF) model. Numerical simulations has been performed for deformable air-bubbles in a straight channel. Bubble features such as aspect ratio, equivalent diameter, velocity and path are compared against experimental data obtained by using shadow-graph technique. The VOF model is capable to predict the different bubble features and shows a promising future for studying the detailed interaction between the different phases inside a bioreactor. The limitations of the Eulerian-Lagrangian model arises when the inter-particle distance is small. From the simulations of Lattice Boltzmann Method (LBM), both drag- and lift-coefficients were obtained for cases with strong particle-particle (so called four-way) interaction. A novel approach of handling large spherical bubbles combined with the Eulerian-Lagrangian model in the Large Eddy Simulation (LES) framework, has been developed and utilized.