Identification and modelling of noise sources on a realisticnose landing gear using phased array methods applied tocomputational data

Due to the recent development of quieter turbofan engines, airframe noise has started to emerge asthe most important noise source. This is particularly true during the approach/landing phase, whenthe engines are operated at low-thrust levels. In order to meet future noise level regulations, thecharacterization and subsequent reduction of landing gear induced noise is necessary. Wind-tunnelaeroacoustic tests have always been the favoured method for assessing and studying the noise generatedby landing gears, but their prohibitive cost has steered the attention towards numerical methods.Since direct flow noise simulations are still too demanding in computer resources, there is astrong interest in developing coupled CFD-CAA simulations as a tool to model and identify flownoise sources. More recently, they have been coupled with phased array methods in order to conductaeroacoustic studies on scaled-down, or simplified, aircraft components. This project investigates theaerodynamic sound sources on a realistic nose landing gear using numerical phased array methods,based on array data extracted from compressible Detached Eddy Simulations of the flow. Assumingmonopole and dipole modes of propagation, the sound sources are identified in the source regionthrough beamforming approaches: conventional beamforming, dual linear programming (dual-LP)deconvolution, orthogonal beamforming and CLEAN-SC. To assess the accuracy of the employedmethods, beamforming maps from flyover, sideline and forward point of views are obtained andcompared to experimental ones originating from wind-tunnel experiments performed on the samenose landing gear configuration by industrial and academic partners of the ALLEGRA project. Anarray design metric is defined to quantitatively assess the fitness of the employed arrays with respectto the different frequencies and distances separating the beamforming and array planes. A geneticalgorithm based on the Differential Evolution method is used to generate optimized arrays for selectedfrequencies in order to reduce the computational size of the problems solved. The modelledsources are used to generate far-field spectra which are subsequently compared to the ones obtainedwith the FfowcsWilliams and Hawkings acoustic analogy. The results show a good concordance betweenthe numerical phased array beamforming maps and the experimental ones, and a good matchbetween the far-field spectra up to a certain frequency threshold corresponding to the quality of themesh used. The presence of specific noise sources has been validated and their contribution to theoverall generated noise has been quantified. The results obtained demonstrate the potential of numericalphased array methods as a legitimate tool for aeroacoustic simulations in general and as atool to gain insight into the noise generation mechanisms of landing gear components in particular