Uncertainty analysis of centrifugal compressor based on adaptive ensemble and nested high-dimensional model representation

To address the high-dimensional and high-accuracy computational demands in aerodynamic uncertainty quantification for centrifugal compressors, this paper proposes an uncertainty quantification method based on the adaptive ensemble and nested high-dimensional model representation (HDMR). The method is applied to the accuracy verification of a test function and to the preliminary aerodynamic uncertainty analysis of High Efficiency Centrifugal Compressor (HECC) taking into account geometric uncertainty factors. The results show that the adaptive ensemble HDMR outperforms individual metamodels by achieving the same accuracy with fewer samples, and among the 20 geometric variables of HECC, the impeller inlet shroud diameter, impeller outlet diameter, and the main blade outlet blade angle at hub side show relatively high sensitivity, with the geometric variations leading to 1%~2% reduction in adiabatic efficiency at the design point. Additionally, uncertainty in geometry tends to have negative effect on aerodynamic performance. The results confirm that the proposed method adaptive ensemble and nested HDMR can be effectively used as a mathematical tool to investigate the flow mechanisms under uncertainty in real centrifugal compressors.