Cfd Study of Mixing Process in Rushton Turbine

This paper reports on the CFD study of macro-mixing process of Rushton turbine in stirred tank. The code for mixing calculation is developed in the commercial CFD code CFX4.3. Mixing process simulation on different calculation method, different turbulent model and different tracer adding and detecting position are calculated. The simulation is three-dimensional and the impeller region is explicitly included using a sliding mesh method to account for the relative movement between impeller and baffles. Fluid flow is calculated with a turbulent k-ε and RNG k-ε model using a finite-volume method. The results show that the mixing time highly relies on the flow field, the feeding and detecting position. The improvements of macro-mixing simulation can be obtained by accurate flow prediction and modification to turbulence models applied to stirred tank. NOMENCLATURE C off-bottom clearance, m D impeller diameter, m H liquid height, m p pressure R impeller radius, m Re Reynolds number Sc Schmitt number T tank diameter, m t time, s u axial velocity, m·s v radial velocity, m·s w tangential velocity, m·s ρ density kg·m υ dynamic viscosity m·s θ95 mixing time, s INTRODUCTION Stirred tanks are widely used in the chemical process industry to carry out many different operations. In the design of stirred tanks, detailed information on the flow and mixing phenomena is of great importance. Because of the complexity of fluid mechanics prevailing in stirred vessels, the present design procedures are still closer to an art than science. In order to understand the fluid dynamics and develop rational design procedures, there have been continuous attempts over the past century. Among the attempts, Computational fluid dynamics (CFD) may be the most important method. Macro-mixing denotes the stage of a mixing process which refers to mixture concentration changes down to the scale of physical probes used for measuring local concentrations. It is different from the molecular scale mixing-micro-mixing. Macro-mixing in process equipment is of high importance in the analysis of other processes, such as crystallization or chemical reactions. The intensity of macro-mixing process in stirred tanks has been often characterized by mixing time, i.e. the time necessary to achieve a required degree of homogeneity measured by a mixing index applied. The simulation of macro-mixing process is less popular than fluid field modelling. So there is only few papers reported concentrations obtained from CFD over the whole period of homogenisation process, and the results are often compared with experiments in order to validate the approach. Noorman (1993) compared the experiment and simulation results of a single Rushton turbine homogenisation. The tracer correspondence curve corresponded well with experiment results, but had some difference in detail. Schmalzriedt (1997) compared the results of a single Rushton turbine with literature data, and got the conclusion that the simulation results were highly relied on the turbulence model. Jaworski (2000) reported the simulation results of double Rushton turbine using CFD code FLUENT. The predicted mixing time θ95 was 2~3 times high than the experimental data. The discrepancy came from the under prediction of mass transfer between different circulation loops. Some researchers (Mao, 1997) also reported the process simulation with zonal mixing model. This study is aimed at highlighting the effects of calculation method, turbulence model and different tracer adding and detecting position on the macro-mixing process and mixing time.