Grid sensitivity of flow field and noise of high-Reynolds-number jets computed by large-eddy simulation

Abstract

Three isothermal round jets at a Mach number of 0.9 and a diameter-based Reynolds number of 10 5 are computed by large-eddy simulation using four different meshes in order to investigate the grid sensitivity of the jet flow field and noise. The jets correspond to two initially fully laminar jets and one initially strongly disturbed jet considered in previous numerical studies. At the exit of a pipe nozzle of radius r 0 , they exhibit laminar boundary-layer mean-velocity profiles of thickness 0:2r 0 ; 0:025r 0 and 0:15r 0 , respectively. For the third jet, a peak turbulence intensity close to 9% is also imposed by forcing the boundary layer in the nozzle. The grids contain up to one billion points, and, compared to the grids used in previous simulations, they are finer in the axial direction downstream of the nozzle and in the radial direction on the jet axis and in the outer region of the mixing layers. The main flow field and noise characteristics given by the simulations, including the mixing-layer thickness, the centerline mean velocity, the turbulence intensities on the nozzle lip line and the jet axis, spectra of velocity and far-field pressure obtained from the jet near field by solving the isentropic linearized Euler equations, are presented. With respect to those from previous studies, the results are very similar for the initially laminar jet with thick boundary layers, but they differ significantly for the initially laminar jet with thin boundary layers and for the initially disturbed jet. For the latter two jets, using a finer grid leads to a faster flow development, to higher turbulence intensities in the shear layers and at the end of the potential core, to stronger large-scale structures, and to the generation of more low-frequency noise.

A technique of flux reconstruction at the interfaces of nonconforming grids for aeroacoustic simulations

“…At x =20 r 0 , the spectrum from the nonconforming grid displays slightly higher levels at low Strouhal numbers and collapses more rapidly for St ≥2. The origin of the small differences observed here does not seem to be related to the presence of the nonconforming interface but to a poorer discretization of the jet coherent structures using the nonconforming mesh …”

“…These levels are also lower than those obtained for x >14 r 0 in Figure A. This is most likely due to the mesh resolution that is twice as coarse in the y ‐direction for x >14 r 0 in the nonconforming grid . In the pressure field, acoustic waves propagate from the jet with an angle of about 30° relative to the x ‐axis.…”

“…This is most likely due to the mesh resolution that is twice as coarse in the y-direction for x > 14r 0 in the nonconforming grid. 35 In the pressure field, acoustic waves propagate from the jet with an angle of about 30 • relative to the x-axis. In Figure 22B, no discontinuity of the pressure waves radiated from the jet and no spurious reflection are visible in the vicinity of the nonconforming interface.…”

“…The conforming grid used in the present work is finer than the one used by Bogey and Bailly by factors of 1.3 and 2 in the axial and radial directions, respectively. In addition, since the resolutions of the two grids used in this study differ for x >14 r 0 , small differences between the results from the two simulations are expected, as demonstrated by the grid sensitivity of turbulent jet flows presented by Bogey …”

“…In addition, since the resolutions of the two grids used in this study differ for x > 14r 0 , small differences between the results from the two simulations are expected, as demonstrated by the grid sensitivity of turbulent jet flows presented by Bogey. 35 In each computation, the jet flow is simulated over a time period T = 2 × 10 5 Δt = 2000r 0 ∕c 0 . The flow initialization lasts over 600r 0 ∕c 0 .…”