Three-dimensional optimization of non-axisymmetric volute based on a novel global adaptive optimization method

Abstract

This study presents a global three-dimensional multi-objective optimization of a non-axisymmetric exhaust volute in a centrifugal compressor, aiming to enhance efficiency across a wide operating range using a global gas dynamics optimization approach. A three-dimensional parameterization method for the volute is introduced, enabling comprehensive circumferential control and cross-sectional variation during optimization. To reduce computational cost, an adaptive topology high-fidelity surrogate model (ATHFS) is developed and applied, accelerating the optimization process. The constraint nondominated neighbor immune algorithm (C-NNIA) is employed to obtain Pareto-optimal solutions. Optimization results show notable improvements in stage efficiency at Near-stall, Design, and Near-choke conditions, with increases of 0.7 %, 0.92 %, and 1.49 %, respectively, compared to the baseline. The stable operating range is also extended by 3.02 %. Flow field analysis indicates that volute geometry adjustments significantly reduce asymmetry in radial and axial velocity distributions. These results confirm the effectiveness of the global gas dynamics optimization method and the ATHFS model, highlighting their potential for advancing full three-dimensional matching optimization of centrifugal compressor components.