Cerfacs Enter the world of high performance ...

Abstract:

Predicting landing gear noise is a major concern for an aircraft manufacturer, since it contributes to about 40% of the total aircraft noise during the approach phases. Flight tests and those carried out in anechoic wind tunnels have enabled the understanding of noise generation mechanisms, as well as the design of low noise devices. However, these methods are time consuming and costly to set up. The use of computational fluid dynamics (CFD) is thus emerging as an essential complement to these experimental approaches. The flow around landing gears is complex and highly unsteady, and the noise generated is broadband by nature. Given these characteristics, it is therefore necessary to use unsteady methods with high-fidelity turbulence modeling such as Large Eddy Simulation (LES), to predict these acoustic sources. The lattice Boltzmann method (LBM) is a numerical approach that has recently shown a strong potential for this type of application, thanks to its accuracy, its low restitution time and its ability to handle complex geometries. It is consequently adopted for this thesis. Aeroacoustic simulations require a high level of accuracy since acoustic fluctuations, which are several orders of magnitude smaller than aerodynamic ones, must be properly captured and propagated. Nevertheless, the non-conforming grid interfaces used in LBM have the inconvenience of generating spurious vorticity and acoustics that propagate in the fluid core, which may affect the noise predictions. The PhD objective is to develop new grid coupling models in the “LaBS/ProLB” LBM solver, and to validate them in the context of landing gears aeroacoustics. Two main directions are addressed to overcome these phenomena: 1/ A study of the numerical scheme in the fluid core is performed, highlighting the involvement of non-hydrodynamic modes, specific to the LBM, in the generation of vorticity and of a portion of the spurious acoustics generated at mesh interfaces. After a thorough study of the implication of these modes, an appropriate collision model (H-RR) is chosen to filter them out during a simulation. The stability and accuracy of several LBM schemes including the H-RR one under typical aeroacoustic simulation conditions are also investigated. This study highlights stability issues, as well as questionable precision of many advanced LBM schemes available in the literature. 2/ A direct coupling algorithm between two grids of different resolution is proposed. This algorithm allows to greatly improve the accuracy of the non-conforming grid interfaces, and hence to reduce the spurious acoustic emission produced by the crossing of vortices composing the wakes. Finally, the LAGOON landing gear allows for the validation of these numerical ingredients. An aerodynamic study and then an aeroacoustic one via a coupling with an acoustic propagation code based on the Ffowcs Williams and Hawkings analogy (FW-H) are conducted. The limitations of this analogy in its solid formulation, mostly used to predict landing gear noise, are exposed. Lastly, the effect of extra components of increasing complexity on the noise generated is investigated.

Keywords: Lattice Boltzmann method,Landing gear noise,Grid refinement,Aeroacoustics,CFD,CFD/CAA coupling

Jury:

Advisor	M. Pierre SAGAUT	Aix Marseille Université
Referee	M. Damiano CASALINO	Université de technologie de Delft
Referee	M. Jonas LATT	Université de Genève
Member	M. Alois SENGISSEN	Airbus Operations SAS
Member	M. Éric MANOHA	ONERA
Member	M. Stéphane MOREAU	Université de Sherbrooke
Member	Mme Véronique FORTUNÉ	Université de Poitiers
Invited member	M. Jean-François BOUSSUGE	CERFACS