Establishment of CFD-ANN-NSGA-II model for stirred reactor design

Abstract

Herein, the multiobjective optimization method based on the CFD-ANN-NSGA-Ⅱ model is proposed to efficiently optimize the structural parameters of the stirred reactor, thereby improving its gas–liquid mass transfer efficiency and reducing its energy consumption. Coupling CFD simulations with the ANN-NSGA-Ⅱ model enabled accurate performance predictions. The R² values of the parameters ε, k_La, t, and P/V were 0.940, 0.989, 0.956, and 0.970, respectively, and their mean square error values were 0.0010, 0.0224, 0.0218, and 0.0521, respectively. The optimal reactor structural parameters were H/D = 2.16, d/D = 0.38, w/d = 0.50, and C1/D = 0.30. The ε and k_La values of the structure increased with optimization to 13.5 % and 0.171 s⁻¹, respectively, while the t and P/V values decreased. The study describes an efficient and reliable theoretical method for the multiobjective optimization of chemical equipment, verifying the potential of adopting artificial intelligence in complex fluid system optimization, with important value for engineering applications.