Multi-objective optimization of centrifugal pump impeller based on kriging model and multi-island genetic algorithm
Résumé
Low-specific-speed pumps are extensively used in various fields, but the efficiency is relatively low. An optimization approach is proposed based on numerical simulation, design of experiment, approximation model and multi-island genetic algorithm. The low-specific-speed centrifugal pump IS50-32-160 is determined as the research subject, three parameters containing blade outlet width b₂, blade outlet angle β₂, and blade warp angle φ are selected as the design variables, and then 20 impellers are designed by optimal Latin hypercube sampling method. The commercial software ANSYS CFX 14.5 is applied to conduct the steady numerical simulation to obtain the efficiency and head, which are the optimization objectives. Approximation model is built between the objectives and the design variables based on the Kriging model. Finally, the best combination of impeller parameters is figured out by solving the approximation model with multi-island genetic algorithm. The results of original pump show that the performance curves obtained by experiments and simulations have a good agreement and the head deviation is 3.3% under the design condition. The optimization method improves the hydraulic efficiency by 3.2%. The internal flow in the optimized impeller are steadier and the pressure fluctuation intensity is decreased. This optimization method presented can provide references to the optimization of non-over-load design of low-specific-speed centrifugal pumps.
Origine : Fichiers produits par l'(les) auteur(s)
Loading...