Evaluation and application of efficient CFD-based methods for the multi-objective optimization of stirred tanks

Three numerical methods to determine the mixing time in stirred tanks are investigated using Computational Fluid Dynamics (CFD). In a cylindrical tank, the mixing performances of a Rushton turbine (RT) and a pitched blade turbine (PBT) are investigated, using transient methods, i.e., the Unfrozen Flow Field (UFF) and the Frozen Flow Field (FFF) method, and the steady state Mean Age Theory (MAT) approach. The results are compared with experimental data, and the accuracy and numerical effort are judged. The results demonstrate that the MAT method can characterize the complete tracer distribution in the vessel and gives good agreement with experimentally obtained mixing times. Moreover, this method utilizes only a tiny fraction (8%) of the CPU time required by the UFF method. A multi-objective optimization study is performed utilizing the MAT method to find the best trade-off between mixing time and power consumption. The optimum impeller designs are obtained.