Abstract:This work focused on exploring a computational fluid dynamics(CFD)method to predict the macromixing characteristics including the mean flow field and impeller capacity for a 45° down-pumping pitched blade turbine(PBT)in stirred tanks. Firstly, the three typical mean flow fields were investigated by virtue of three components of liquid velocity. Then the effects of impeller diameter(D)and off-bottom clearance(C)on both the mean flow field and three global macro-mixing parameters concerning impeller capacity were studied in detail. The changes of flow patterns with increasing C/D were predicted from these effects. The simulation results are consistent with the experimental results in published literature.The pitched blade turbine(PBT)is considered to be a kind of simple and versatile impeller for mechanical agitations in both laboratorial and industrial reactors, since it has a good balance between the pumping and shear capability. PBT(especially down-pumping mode)is widely used in practical applications involving the immiscible liquid mixing, gas dispersion and solid suspension [1][2][3] . Mechanical mixing achieved through both bulk convection and turbulent diffusion is distinguished into macro-mixing(mixing at the bulk scale of the tank)and micro-mixing(mixing at the molecular scale). However, micro-mixing performance is usually dominated by macro-mixing [4] . To improve the mixing quality and optimize the impeller design, the macro-mixing characteristics, such as the liquid velocity distribution and pumping capacity should be well understood.Three global macro-mixing parameters, including power number(N p ), pumping number(N q )and pumping effectiveness(η e ), largely depend on the mean flow fields or flow patterns, which are strongly affected by many impeller design variables [5] . Some qualitative or quantitative results of the mean flow fields produced by PBT impeller and the dependence on some key design variables, especially D and C, were reported in published literature [6][7][8][9] . These studies were usually preformed by the experimental fluid dynamics(EFD)method. Information about the velocity field or turbulent flow is difficult to be obtained by using EFD method, unless by using some complex and expensive instruments such as laser Doppler anemometry(LDA).Tremendous progress has been made in CFD method in past decades, and it has been promoted as a powerful tool to predict fluid flow and model various geometries in stirred tanks. Numerical simulation based on the Reynolds-averaged Navier-Stokes(RANS) equation is the most widespread CFD method to provide quantitative or qualitative information about the velocity field and turbulent field [10, 11] . Effects of many numerical issues on the RANS-based prediction of turbulent flow in the stirred tanks were investigated in published literature [12][13][14] , and these works indicated that RANS simulation along with the impeller rotation model of multiple reference frame(MRF)and the k-ε turbulence model could reasonably simulate both the mean fl...