Design of a hermetic centrifugal pump impeller using RSM and evolutionary algorithms with application of SLS technology

Abstract

This study presents the development and validation of a comprehensive numerical optimisation methodology used to improve the energy efficiency of a pump with normal characteristics: volume flow rate, Q nom = 6.3 m3/h, and head, H = 20 mH2O. The methodology was implemented in ANSYS Workbench using ANSYS CFX and optiSLang. The optimisation process is based on data from 853 RANS (SST) calculations on a sample generated by the LHC method, varying the parameters of the blades and flow path. Response surfaces (RSM) were constructed using anisotropic and classical kriging, which were optimised using an Evolutionary Algorithm (EA). The optimised geometry was verified numerically by URANS SST and experimentally. For physical validation, the wheel was manufactured using SLS technology from PA-12 Industrial powder, a strength assessment FSI was performed, and the geometry was checked by 3D scanning. 3D scanning showed a high manufacturing accuracy (deviations of 0.1–0.3 mm). The result is a geometry that increases efficiency while maintaining head, which has been confirmed by experimental validation.