Impact of Manufacturing Variability on Multi–Stage High–Pressure Compressor Performance

The present paper introduces a novel approach for considering manufacturing variability in the numerical simulation of a multi–stage high–pressure compressor (HPC). The manufacturing process is investigated by analyzing three of totally ten rotor rows. Therefore, 150 blades of each of the three rows were 3D scanned to obtain surface meshes of real blades. The deviation of a scanned blade to the design intent is quantified by a vector of 14 geometric parameters. Interpolating the statistical properties of these parameters provides the manufacturing scatter for all ten rotor rows expressed by 140 probability density functions. The probabilistic simulation utilizes the parametric scatter information for generating 200 virtual compressors. The CFD analysis provides the performance of these compressors by calculating speed lines. Post–processing methods are applied to statistically analyze the obtained results. It was found that the global performance parameters show a significantly wider scatter range for higher back pressure levels. The correlation coefficient and the coefficient of importance are utilized to identify the sensitivity of the results to the geometric parameters. It turned out that the sensitivities strongly shift for different operating points. While the leading edge geometry of all rotor rows dominantly influences the overall performance at maximum efficiency, the camber line parameters of the front stages become more important for higher back pressure levels. The analysis of the individual stage performance confirms the determining importance of the front stages — especially for highly throttled operating conditions. This leads to conclusions regarding the robustness of the overall HPC, which is principally determined by the efficiency and pressure rise of the front stages.