Investigation of oxygen transfer process in wave bioreactors using Computational Fluid Dynamics

One of the fundamental criteria for successful cell growth in major biotechnological applications (e.g. vaccine production processes) is the sufficient aeration of the cell culture. Oxygen aeration in typical bioreactors such as stirred tanks is performed mostly through a sparger. However, particular care is required to avoid foaming and high shear forces during sparging that would lead to loss in cell yield. An alternative method is surface aeration during which the oxygen transfer takes place by diffusion through the liquid-air interface. Although this bubble-free aeration seems attractive since it would reduce the cell damage considerably, an adequate liquid-air interface surface must be provided for an efficient cell growth. Wave bioreactors, which were proposed in recent years, overcome this limitation since the rocking motion in the partially filled cellbags generates waves at the liquid surface, greatly enhancing the available surface area for oxygen transfer. For optimal control of cell growth, it is necessary to describe quantitatively the oxygen transfer process in wave bioreactors. Therefore, the unsteady flow has first been computed employing Computational Fluid Dynamics (CFD) in the present work. Corresponding computations for 2 L reactor scale have been performed using the industrial CFD code ANSYS-FLUENT® 6.3. These simulations employ the Volume of Fluid (VOF) method for a correct description of the free liquid surface motion. The initially obtained results demonstrate a very efficient oxygen transfer via diffusion through the predicted liquid-gas interface. This situation is not surprising when the high mass transfer coefficients (kLa) in wave bioreactors are compared with those of classical bioreactors. For instance, kLa is measured as 3.5 hr-1 in a half-filled 2L-cellbag whereas this is around 1.15 hr-1 for the same working volume in a spinner flask (V. Singh, Cytotechnology 30:149-158, 1999). Further simulations and corresponding results will be shown for different reactor scales at the Conference