(366g) CFD-Based Multi-Objective Optimization of Dual-Impeller Configurations in a Gas-Liquid Stirred Tank

CFD-Based Multi-objective Optimization of Dual-Impeller Configurations in a Gas-Liquid Stirred Tank

Miao-Na Chen, Jia-jun Wang*, Xue-ping Gu, Lian-fang Feng

State Key Laboratory of Chemical Engineering, College of Chemical & Biological Engineering, Zhejiang University, Hangzhou P.R. China 310027

jiajunwang@zju.edu.cn

Abstract Gas−liquid stirred reactors are widely used in the process industries owing to the remarkable advantages, such as strong operation flexibility, excellent mass transfer effect and high mixing efficiency. The configuration optimization of aerated stirred tanks is of great significance since configurations are important factors that affect the flow and mixing in chemical reactors, further influencing the mass transfer and reaction process.

The research on configuration optimization of gas-liquid stirred tanks has developed from early experimental measurement to current computational fluid dynamics (CFD) simulation. Although predictions can be made under any operation conditions with CFD method, the process requires trial and error, leading to heavy workload and time cost. Besides, a large amount of experiments will be involved to take into account multiple design variables, which can generally result in local optimum. How to improve the computing efficiency and obtain the global optimum designs still remain very challenging. However, the fast non-dominated sorting genetic algorithm (NSGA-II) offers an attractive alternative to analyze multi-objective problems in a cost-effective way and find global optimum solutions with parallel-searching and fast convergence.

In our work, a multi-objective optimization methodology based on CFD and NSGA-II approaches was developed for dual-impeller design in a gas-liquid stirred tank. The strategy enabled CFD to make predictions for flow field information. It also utilized NSGA-II to cut down the computational demand and acquire global optimum solutions. A bubble size model was introduced to describe bubble size variation in CFD simulation. Six geometrical parameters of impeller blades were selected as design variables. In view of the mass transfer efficiency and energy saving, the overall interfacial area and power consumption were chosen to evaluate the impeller performance.

Model validation was successfully conducted by comparing the experimental results with the simulated data. A Pareto front was reached after 600 iterations to determine the final dual-impeller designs. According to the optimal designs selected from Pareto front, it was found that the pitched concave blade disk turbine (PCBDT) as the lower impeller and down-pumping pitched blade turbine (PTD) or up-pumping pitched blade turbine (PTU) as the upper impeller achieved the relatively larger gas-liquid contact area with much lower energy consumption.

The influence of impeller combination on hydrodynamic characteristics was also investigated through the comparison of nine unique combinations. The results showed the PTD combinations exhibited the best gas dispersion performance, while the PTU combinations showed the worst gas distribution uniformity. The bubble size distribution of each combination shared the same trend: the bubble size varied along the direction of discharge flow, first smaller bubbles were generated in the impeller region and then the bubbles became larger away from the impeller region. And the bubbles near the free surface and the center of fluid circulation loops were relatively large. Moreover, the optimal PCBDT-PTU (lower impeller-upper impeller) presented the highest peak of local interfacial area while the optimal PCBDT-PTD obtained the most uniform interfacial area distribution. Both optimal designs were energy-efficient in pursuing the best overall mass transfer performance.

Experimental measurements of oxygen transfer coefficient and power consumption by oxygen electrode and torque sensor were utilized to evaluate the optimal design. Compared with the reference design with the largest oxygen transfer coefficient, the optimal PCBDT-PTU has improved the oxygen transfer coefficient by at least 11% with a 25% reduction of power input; while the power consumption of the optimal PCBDT-PTD has been decreased by 27%.

The successful establishment of the multi-objective optimization strategy for the impeller design indicates a new application for CFD-based optimization methodology. The automatic optimization stratery show great potential for many industrial fields involving gas-liquid stirred reactors, like fermentation, PX oxidation, and hydrogenation.