Flow and noise predictions for a mixer-ejector engine configuration using LES
Paper in proceedings, 2007
Large-eddy simulations of the compressible flow and acoustic field for a mixer-ejector nozzle configuration have been carried out. The configuration consists of two mixing enhancers. Firstly an internal mixer with a lobed trailing edge forcing the core and bypass flows to mix. Secondly, a section allowing ambient air to enter the ejector and to mix with the core/bypass flow. This two-step mixing arrangement creates three-dimensional structures of widely varying length scales. This puts a high demand on mesh resolution and a robust solution technique in order to achieve high quality simulation results. The flow simulation predicts a highly anisotropic turbulence downstream of the mixers, which indicates that turbulence models based on the Boussinesq assumption are unsuitable for this type of flow. The computational domain is discretized using a block-structured boundary-fitted mesh with 219 mesh blocks and approximately 24 × 106 nodes. The choice of a block-structured grid is based on previous experience where this type of mesh has provided good numerical accuracy for both flow and acoustics in combination with the numerical scheme used. However, for a complex geometry such as the mixer-ejector concept, it is a challenge to construct a reasonable block topology for a structured mesh. Nevertheless, this paper shows that this may be done and that a high quality block-structured mesh can be obtained through the use of generalized interfaces with hanging node refinement/coarsening. The Favre-filtered Navier-Stokes equations were solved using a finite volume method with a low-dissipation third-order upwind-biased scheme for the convective fluxes, a secondorder centered difference approach for the viscous fluxes, and a three-stage second-order Runge-Kutta technique in time. A compressible form of Smagorinsky's subgrid-scale model is used for computation of the subgrid-scale stresses.