Statistical characteristics mining of measured machining error of multi-stage compressor blisks

Blisks have been widely adopted in various aero-engines due to the advantages such as simple structure and low loss. However, influenced by machining errors, the geometric inconsistency of blisk blades is significant, leading to deviations in the compressor performance from the design and scatter increase. To accurately assess performance uncertainty effects of machining errors using uncertainty quantification methods, ‘statistical characteristics of machining errors’ as uncertainty quantification inputs are particularly critical. This study is the first to highlight measured machining errors’ uncertainty analysis for blisks. Measured machining errors from the front, middle, and rear stages of multi-stage compressor blisks are analyzed regarding their systematic deviations and scatters along the radial direction, and probability distribution characteristics. The results show that due to differences in clamping and fixing methods, the statistical characteristics of machining errors for ‘blisk’ differ from those of ‘single blade’. Additionally, variations in material properties and sizes of blades at different compressor stages lead to differences in the statistical characteristics of machining errors. For different sections, systematic deviations and scatters in machining errors are notably significant near the blade tip, making it challenging to ensure machining consistency. For different stages, machining errors of the rear stage blades are the most scattering. Compared with the design geometry, several phenomena observed in most blades, such as ‘under deflection’, ‘thicker pressure/suction surfaces’ and ‘larger leading-edge radius’, should be improved, owing to their adverse effects in compressors. Furthermore, probability distributions of machining errors exhibit characteristics such as ‘skewness’, ‘bimodality’, and ‘data missing’, indicating that traditional normal distributions are insufficient for accurately characterizing the above distributions. The research results provide a clear demonstration of the machining capabilities of compressor blisks and offer data support for correctly constructing probability models of machining errors, thereby enabling accurate prediction of their performance uncertainty effects.