Publications Repository - Helmholtz-Zentrum Dresden-Rossendorf

A non-baffled laboratory-scale stirred tank reactor, mechanically agitated by a gas-inducing turbine, was experimentally and numerically studied. The system under investigation comprises air as gas phase and isopropanol as liquid phase at room temperature. The initially stationary isopropanol and air were brought into motion by the rotating impeller. The X-Ray cone beam tomography measurements were taken at five different stirrer speeds with thresholds of 50 rpm starting from 1000 rpm at which the gas inducement occurs for the given operating conditions. The final stirrer speed was 1200 rpm at which the central vortex virtually reaches the impeller. Additionally, video observations, performed with a digital camcorder, were taken to study the unsteady behaviour of the central vortex.
CFX 10.0 numerical software was used to carry out the computational fluid dynamics analyses of the stirred tank reactor. A full three dimensional approach was adopted in order to capture the unsteady behaviour of the central vortex. Transient and steady state numerical calculations were performed. Four steady state simulations, at stirrer speed from 200 to 800 rpm, were conducted to obtain an initial guess of the flow field and the phase distribution for the steady state and the transient simulations at 1000 rpm. The numerical predictions above 1000 rpm used the previous simulation results as an initial guess. Starting from 1000rpm, five steady state simulations were performed at stirrer speed thresholds of 50 rpm to be compared with the experimental observations. The transient numerical predictions were compared with visual observations, since the X-Ray cone beam tomography provides an average phase distributions more suitable for comparison with the steady state predictions. The tetrahedral mesh with above 1500000 elements was globally refined since a detailed view in the whole geometry is required. The inhomogeneous two-phase flow model with the particle transport model was applied to the system with momentum transfer described by the drag force and turbulence transfer modelled by Sato enhanced eddy viscosity model. The gas phase was modelled as dispersed fluid with a mean diameter of 1 mm and the liquid phase as continuous fluid. Different turbulence models were considered for the liquid phase but the k-ε turbulence model was found to be the most computationally stable and was used in the simulations. The gas phase turbulence was modelled using the dispersed phase zero equation. The flow was regarded as buoyant and implemented using the density difference model.
The CFD predictions closely mimic the experimental observations for the central vortex depth as well as for its spread out. The results demonstrate the X-Ray cone beam tomography and the CFD capabilities to capture the two-phase flow in detail, which can provide valuable information for the industry. In particular the special gas phase distribution, which can also have time-dependent behaviour, can have a crucial impact on the reactor performance. This can be in detail predicted by the computational fluid dynamic software, which can prove to be an essential tool for the reactor optimisation and scale-up.