Experimental Investigation And Numerical Simulation Of Fluid Dynamics In Stirred Tanks

The efficiency enhancement of the impellers is the key to optimum design of stirred reactors. The flow characteristics and mixing performance are mainly determined by the impeller type used in the reactor. The traditional Rushton turbine which is widely used in gas-liquid system has been replaced by some newly developed impellers, such as curved blade disc turbine and wide-blade hydrofoil. However, very few investigations on the mechanism of these new impellers have been carried out. It’s difficult to guide the further improvement of impeller design effectively. For this reason, the flow characteristics and macro mixing characteristics in curved blade disc turbine and multiple impeller stirred tanks by using PIV and CFD approach respectively. The results are of particular relevance to the impeller optimum and it’s applications in multiphase system.The experiments were carried out in a perspex tank of0.48m in diameter stirred by four different disc turbines, including Rushton turbine (RT), concaved blade disc turbine (CD), half elliptical blade disc turbine (HEDT) and parabolic blade disc turbine (PDT). The phase-resolved PIV technique was used to analyze the trailing vortices and elucidate their relationships with turbulence properties under two conditions-same Re and same power input. Results show that a pair of trailing vortices was generated behind each disc turbine but the location, size and strength of vortices are different. The curved blade will cause vortex to stay for a while at the blade tips. Under the same Re, more curved blade results in longer vortex life, smaller and stronger vortices. The TKE and turbulent kinetic energy dissipation near the impeller tip decrease as the blade turns curved. The variation of vortex under the same power input is consistent with that under the same Re. However, as the flat blade turns to concaved blade, the TKE and turbulent kinetic energy dissipation near the impeller tip increase and then decrease as the blade turns more curved.To meet the large-scale reactor development, experiments were carried out in a triple-impeller stirred tank of0.48m diameter. The aspect ratio of the liquid is1.6-1.8. The impeller combination consisted of a HEDT at the bottom and two wide-blade hydrofoils (WH) above HEDT. The effect of impeller spacing and operating style on the flow characteristics and mixing efficiency in the stirred tank has been discussed. Results show that we can get three flow patterns (including five, four and three circulation loops respectively) in the HEDT+2WHU stirred tank by changing the impeller spacing. Reducing the spacing between top two impellers and the submersion depth of the top impeller can help to improve mixing efficiency. Changing the top two impellers to WHD is also conducive to mixing efficiency. The choice of impeller should take other factors such as gas holdup into consideration.The flow in the disc turbine stirred tank and HEDT+2WH stirred tank were investigated by using LES, and were compared with experimental data and the simulation results of standard k-ε model. Results show that LES has more predictive power of the flow field produced by complex impeller and multiple impellers than standard k-ε model, but LES has more requirements for grid style and quality.Last, the mixing characteristics in HEDT+2WH stirred tank were experimentally investigated and numerically simulated by LES. The mixing time in four flow patterns was measured by the chemical decolorization method and laser detecting meter. The tracer concentration distribution in the tank was captured by a high speed camera. The mixing characteristics were also simulated by LES and compared with experimental data and the results of standard k-ε model. Results show that the tracer disperses in circulation loops one by one in the axial direction. Flow pattern has an important effect on the mixing time and tracer concentration distribution. Less circulation loops result in shorter mixing time. The concentration distribution estimated by standard k-ε model is not accurate, while the results of LES are in agreement with experimental data.