Large-Eddy Simulation of Supersonic Jet Noise with Discontinuous Galerkin Methods

Abstract

We conduct large-eddy simulations to investigate the role of thermal nonuniformity on the development of instability modes in the shear layer of supersonic $M=1.5$ jets. We develop an arbitrary high-order discontinuous Galerkin scheme, build meshes with over 100 million degrees of freedom, and validate the model against particle image velocimetry (PIV) and microphone data. The analysis focuses on both numerical issues (such as convergence against the polynomial order of the mesh), modeling issues (such as the choice of subgrid model), and underlying physics (such as vortex stretching and noise generation). We consider the use of wall models to capture the viscous sublayer at the nozzle. We conclude that the Vreman subgrid-scale model leads to the most accurate predictions of both the near-field velocity and the far-field noise, while the Smagorinsky model is best in the nozzle and near the lip area. Injection in the shear layer is more effective at reducing sound generation in the near field than injection in the core. The difference in effectiveness is less substantial in the far field. Injection in the shear layer creates an unstable three-dimensional structure that leads to loss of coherency and three-dimensional effects in the core. Three-dimensionality in the core supports vortex stretching leading to fast mixing.