离心叶轮非轴对称轮盘面气动敏感性分析

To explore the influence of the non-axisymmetric hub on the efficiency, pressure ratio, mass-flow of choke and flow range of high-pressure-ratio centrifugal impeller, an artificial neural network(ANN)-based full range prediction model of non-axisymmetric hub is adopted to predict the influence of non-axisymmetric hub on aerodynamic performance under 30 shape control parameters. With Sobol sensitivity analysis method, the sensitive parameters that have significant impact on performance are extracted and studied for their influence on the impeller performance. The results show that the control of the efficiency, pressure ratio and mass-flow of choke by hub surfaces mainly depends on the linear superposition of independent effects of each critical parameter. The concave surface of the hub near the impeller inlet is favorable to improve the maximum efficiency and increase the mass-flow of choke, while the convex surface of the hub near the impeller outlet at the channel of pressure side is conducive to the improvement of the maximum efficiency, but it reduces the pressure ratio of the impeller. In contrast, the influence of the hub deformation on the flow range of the impeller is mainly caused by complex high-order interaction of 3 or more parameters. Based on the results of the sensitivity analysis, the values of the control parameters are manually determined, thereby realizing rapid optimization of the non-axisymmetric hub of the centrifugal impeller, increasing the isentropic efficiency by 0.9% and the flow range by 2% with the designed pressure ratio remaining unchanged. This indicates that the sensitivity analysis effectively shortens the optimization period and reduces optimization costs. This research provides directional guidance for the aerodynamic design and optimization of centrifugal impellers in the future.