Turbine Broadband Noise Predictions using Linearised Frequency Domain Navier-Stokes Solvers

Authors

Ricardo Blázquez Navarro - Universidad Politécnica de Madrid
Roque Corral - Universidad Politécnica de Madrid

Abstract

Turbomachinery broadband interaction noise is acquiring outstanding importance on aircraft noise footprint as the intensity of other noise sources diminish. Industrial predictions of fan broadband noise have been traditionally based on the use of the flat plate response to incoming vortical disturbances.While this approach can be valid for plate-like shapes such as a fan, it is unclear how to define an equivalent flat plate for a turbine, and more importantly what the accuracy of such predictions would be.The attractiveness of using flat plate methodologies is mainly their low computational cost for the computation of the broadband noise frequency spectrum. In this work, we propose the use of a computationally efficient linear frequency domain Navier-Stokes solver to retain the influence of the turning, thickness and main geometric parameters on turbine broadband noise. The solver runs on GPUs enabling the computation of thousands or tens of thousands of simulation within an industrial design loop.Firstly, the methodology used to compute broadband noise spectra using a linearised frequency domain Navier-Stokes solver will be presented. The methodology will be used to predict the broadband interaction noise produced by a high-pressure turbine tested within the RECORD European project.It is concluded that the overall agreement with the experiments is good, but an accurate prediction of the turbulence integral length scale is required since it has a major impact on the results. A parametric study is conducted to quantify this effect.Differences of up to 8 dB may arise when comparing detailed simulation results with flat plate results. These are due to multiple effects that are not modelled by semi-analytical methodologies: such as blade thickness and flow acceleration and turning. A thorough discussion of the origin of the discrepancies will be performed.