Direct Noise Computation with a Lattice-Boltzmann Method and Application to Industrial Test Cases

The Direct Noise Computation (DNC) approach is challenging in Computational Aero-Acoustics (CAA) since the computation of the pressure fluctuations requires the resolution of unsteady and compressible flows. Indeed, most numerical methods such as the finite volume and element methods face limitations to efficiently resolve the unsteady compressible Navier-Stokes equations for turbulent flows and the computational costs is extremely high. Instead, steady-state analysis with Reynolds-Averaged Navier-Stokes (RANS) turbulence modeling is generally preferred, however this prevents from a direct computation of the noise as only the mean pressure field is obtained. The use of artificial viscosity is also required to achieve numerical stability, which dampens the high frequency physics. The lattice-Boltzmann method (LBM) is, on another hand, extremely efficient to deal with unsteady, low-speed compressible, and highly turbulent flows. The subsonic aeroacoustics is, thus, one of the major advantages of the method, which is steadily becoming a popular alternative to the traditional Navier-Stokes solvers. This Lagrangian method works at mesoscopic scale based on probability distribution functions to resolve the Boltzmann transport equation which, by means of the Chapman-Enskog expansion, reproduces the hydrodynamic limit and hence the compressible Navier-Stokes equations. This work aims at validating the LBM-based CFD solver XFlow on three categories of noise sources: constrictions, subsonic jets, and bluff bodies. First, the numerical methodology of LBM applied to aeroacoustics will be described. Three aeroacoustics test cases used in the industry to validate CAA methods will then be studied: a duct flow past a thick orifice plate, a subsonic jet through straight pipe, and the flow past a 2-wheels landing gear. Finally, a conclusion on the application of LBM to aeroacoustics analysis and the results found will be drawn.