The main objective of this study is the validation of numerical forced response predictions through experimental blade vibration measurements for higher order modes of a blade-integrated disk (blisk). To this end, a linearized and a nonlinear frequency domain CFD methods are used, as well as a tip timing measurement system. The focus is on the blade excitation by downstream vanes, in particular, because this study shows that the correct prediction of acoustic modes is of key importance in this case. The analysis of these modes is presented, both experimentally and numerically, in Part I of this publication. The grid independence study for the aerodynamic work on the blade surface conducted in this part shows a possible prediction uncertainty of more than 100% when a coarse grid is chosen. For the validation of the numerical setup, a study was performed using different turbulence and transition models. The results are compared to the measured performance map, to a 2D field traverse conducted with a pneumatic probe, and to data gained by unsteady pressure sensors mounted in the casing of the compressor. Flow features relevant for the prediction of blade stresses are best represented using the SST turbulence model in combination with the γ–ReΘ transition model. Nonlinear simulations with this setup are able to predict the blade stresses due to downstream excitation with an average difference of 23% compared to tip timing measurements. Single row linearized CFD methods have shown to be incapable of making a correct stress prediction when acoustic modes form a major part of the exciting mechanisms. In summary, this two-part publication proves the importance of acoustic rotor–stator interactions for blade vibrational stresses excited by downstream vanes in a state-of-the-art high-pressure compressor.

This two-part paper investigates the influence of rotor–stator interactions on the blade vibrational stresses of the first rotor, excited by the downstream stator. To this end, aeroacoustic and aeroelastic measurements and numerical setup studies for the solver TRACE are conducted in order to improve the predictive accuracy of blade vibrational stresses. Part I compares tip timing data for resonance crossings of three blisk modes to numerical predictions. Due to the single-row analysis within the linearized version of the flow solver TRACE, unsteady rotor–stator interactions are excluded by default. The findings show that leaving out these interactions in the numerical setup can lead to 97% lower vibrational stress predictions with respect to the absolute value measured. To validate the prediction of rotor–stator interactions by the nonlinear frequency domain method of TRACE, unsteady pressure measurements were conducted at the casing in the inter-row section of the first stage. The results were analyzed using an optimized measuring grid and applying a compressed sensing-based azimuthal mode analysis. Predicted azimuthal mode numbers are in accordance with the experiment, whereas amplitudes deviate from the measurements in part. Part II focuses on the prediction of blade vibrational stresses. To this end, a detailed grid study is performed and comparisons to steady and unsteady measurement data are made. In summary, this two-part paper confirms the importance of rotor–stator interactions for blade vibrational stresses excited by downstream vanes at a state-of-the-art high-pressure compressor.

In this investigation, CFD calculations are conducted to evaluate the differences between five-hole pressure probe-determined flow quantities and the unaffected flow quantities without the probe’s intrusive influence. The blockage effect of the probe is described and evaluated. Furthermore, the influence of this effect is used to estimate the error when using measured stator outflows as forcing functions for the following rotor blades. To compare the flow field, both with and without the probe’s influence, a five-hole pressure probe is traversed numerically at midspan behind each stator row of a 2.5-stage axial compressor. For reproducing the blockage of the probe accurately, the full annulus of the respective stator row has to be modeled. In order to minimize the calculation time, a study to reduce the number of stator passages was successfully performed. To evaluate the flow quantities using the probe, a calibration polynomial is set up numerically. CFD simulations of the probe geometry within a uniform flow field for each pitch and yaw angle, as well as Mach number combination, are performed for this purpose. Moreover, the pressure probe data for the numerical traverses are corrected to account for velocity gradients in the wake region. The comparison of Mach number, with and without the probe’s influence, shows differences both in the width and the depth of the wake. The results of the Fourier-transformed wake profile for both cases are compared and changes in the first harmonic of Mach number of up to −13% identified. Finally, the first harmonic of the flow quantities is used to perform linearized CFD calculations and to evaluate the influence of disturbed forcing functions on the aerodynamic work of the following rotor blade. The average difference in aerodynamic excitation is about −12% with a maximum deviation of more than −30%. The results presented aim to draw attention to intrusive probe influences and their consequences for validating numerical results against experiments. Special attention is given to the discrepancies of forced response calculations with varying gust boundary conditions.