The computational of quasi-steady and residual turbulence in a stirred tank

When an exothermic chemical reaction is carried out in a stirred tank reactor, hazardous situations may occur when the impeller breaks down, due to less efficient heat removal. One way of achieving reactor safety is to stop the reaction, by means of a killing agent. Since this agent can no longer be spread through the tank by the action of the impeller, residual turbulence has to take care of mixing. Decaying turbulent flow in a stirred tank reactor has been studied computationally. Quasi-steady simulations have been performed first, by using two different methods for modeling the impeller: the snapshot approach and the sliding mesh technique. The quasi-steady flow field served as a starting point for decaying turbulence simulations. The snapshot approach starts from the assumption that, although the flowfield is inherently unsteady, the terms can be neglected. A steady state flow field is therefore solved, using fixed velocity boundaries to model the impeller blades. The sliding mesh technique does solve the complete time dependent Navier-Stokes equations, by defining a rotating grid in the impeller region, and a non moving grid for the baffled region. Theoretical evidence, as well as computed fiow fields show that the snapshot approach gives an incorrect representation of the physics of the fiow in the impeller region, although the overall flow fleld is qualitatively resolved. A modifled snapshot approach, making use of mass sources and sinks as boundary conditions, needs further investigation. The sliding mesh method does give a good representation of the physics. Computational effort is much larger, though. Results show a good agreement between simulations and experiments in the impeller region. The bulk flow field is predicted incorrectly. This is caused by the fact that the fiow field was not fully developed. An estimate for the computational time to obtain a fully developed flowfleld on the computer used was 200 days. A setup for decaying turbulence simulations is made. Starting point was the (not fully developed) sliding mesh flowfield. No quantitatively meaningful results could be obtained.Kramers Laboratorium voor Fysische TechnologieApplied Science