Computation of the noise radiated by a hot supersonic jet deflected in a flame trench

Abstract

This study aims to numerically investigate the noise radiated by a reduced-scale rocket engine jet at lift-off conditions including a flame trench. An over-expanded Mach 3.1 hot jet entering a flame duct where it impinges on a deflector before being guided in a horizontal duct, is considered. The computation is performed with a two-way coupled approach on unstructured grids. This methodology relies on a large-eddy simulation of the jet and the acoustic near field, associated with a full Euler simulation of the acoustic far field. The aerodynamic and acoustic results are compared to a previous computation involving the Ffowcs Williams and Hawkings approach and show a better agreement with the measurements conducted at the MARTEL facility. A more careful analysis of the pressure field suggests that the noise is strongly influenced by the flame trench geometry. Nonlinear propagation effects, natively taken into account by the full Euler solver, are finally highlighted and discussed. Based on appropriate metrics, a good agreement with the experiment is obtained.

Numerical Simulation of the Noise Radiated by Free Hot Supersonic Twin Jets

“…The jet configuration for this test case [23] uses twice the same nozzle (exit nozzle diameter Dj) as previous studies [24], [25], integrated in an assembly that has to be adapted from the one designed for single jet. To construct the numerical test case, the single jet nozzle was duplicated, then the two nozzles were inserted in the twin jet assembly (see Figure 1) thus producing a CAD definition permitting the construction of the computational mesh.…”

“…Core regions are filled with tetrahedra, while pyramids allow to link prisms lateral faces to tetrahedra in specific conditions. As in previous studies [24], [25], a geometric step is inserted in each nozzle convergent in order to trigger fluctuating velocity levels at their exhausts. Table 1 summarizes element distribution for the whole 181 642 204 cells grid.…”

“…The CFD code relies on a finite volume method, while the CAA one on a discontinuous Galerkin formulation and solves the full (nonlinear) Euler equations. These two solvers are two-way coupled at their interface ( [24], [25], [32]). The cell localization and the data exchanges are handled by the CWIPI [33] coupling library.…”