Impact of Input Data Quality on the Uncertainties of a Compressor Blade Parametrization Process

In the wake of the push towards knowledge-based maintenance of jet engines, a process for analysing and evaluating the performance of deteriorated aero engine components was developed. This process automatically generates CFD-simulations from 3D-scans of compressor blades and inputs the calculated performance maps into a gath-path analysis of the jet engine. The uncertainties of such a process are important for the interpretation of its results. Therefore, the uncertainties of a specific part of the process, the geometric analysis, are investigated. Special focus is put on the impact the input data quality has on the uncertainties and how the specific process implementation responds to variations in quality. Test procedures are developed to calculate the uncertainties of individual process steps for specific input data qualities. It is shown, that the noise level of the 3D-scans has a significant impact on the uncertainties. The point-cloud density also has an influence on the uncertainties, which however is small as long as a certain minimum threshold of 189 per profile section is reached. It is also shown that the uncertainties are sufficiently small to capture real geometric variances of deteriorated compressor blades when input data with noise levels of 15 μm and 316 profile points or better is used.