Master thesis and internship[BR]- Master's thesis : Application and calibration of a viscous throughflow model to axial flow compressor blades subject to geometric variability[BR]- Integration Internship

Abstract

This study presents the assessment of a viscous throughflow model applied to high subsonic flow for two modern compressors designed by Safran Aero Boosters. Blades forces terms resulting from the averaging of the Navier-Stokes equations are computed using correlations from the literature. In particular, the optimal incidence angle, the profile loss and the deviation angle are estimated with empirical models. However, these correlations are not adapted for modern blade geometries and high Mach number flows. Therefore, the first aim of this work has been to assess the validity of the correlation predictions using a blade-to-blade approach. The three correlations have been investigated separately. The comparisons between the correlation predictions and the results of the cascade simulations used as reference revealed the necessity to calibrate the correlations. As a result of this observation, the optimal incidence angle is first corrected by adding a correction factor. The predictions for the profile losses are accurate for the two compressors once the angle of incidence is corrected. This study mainly focuses on the calibration of the deviation angle. Three strategies are investigated to calibrate the deviation angle correlation. The first calibration tunes the empirical coefficients of the correlation via an algorithm. The second calibration revises the terms included in the correlation and adds a contribution. These two strategies prove successful. The third method, based on a genetic algorithm, attempts to generate a polynomial function depending on the geometric parameters to characterise the deflection angle. This third method still needs to be tuned for a higher number of cases. The first two calibrated correlations are implemented in the throughflow model, and the resulting performance maps are analysed by comparison with higher-fidelity results. The modified correlations yield significant improvements in performance predictions. A sensibility analysis is finally carried out based on the stagger angle variability. The comparison with higher-fidelity results demonstrates the validity and reliability of the throughflow model with the two enhanced correlations.