Prediction of flowfield and acoustic signature of a coaxial jet using RANS-based methods and large-eddy simulation

The feasibility of using large-eddy simulation (LES) for jet noise predictions has been discussed in several publications. The vast majority of these studies are, however, restricted to fairly simple geometries and moderate Reynolds numbers flows. Recent studies have shown that LES-based methods can be used for realistic flows and complex geometries with promising results. With continuously increased computer capacity and with the possibility to perform computations in parallel on PC-clusters, the possibility of using LES for industrial applications is increasing. However, although, the complexity of configurations for which LES may be used is approaching that of a real jet engine, due to high demands for computational power and long turn-over times LES is still not feasible for industrial use. For low-noise nozzle design, engineering tools with short turn-over time, accurate enough for prediction of noise trends are therefore needed. In the development of such methods, LES can provide data suitable for method validation. In the present work, the high-subsonic flow in an axisymmetric coaxial nozzle/jet configuration is studied. Flowfield predictions obtained using an axisymmetric RANS solver for compressible flows and predicted acoustic signature obtained using a method based on RANS and Lighthill\u27s acoustic analogy are compared with flowfield and radiated sound predictions obtained using LES combined with Kirchhoff surface integration. In the predicted flow statistics, the most noticeable difference between RANS and LES results is that the RANS calculation gives somewhat lower levels of turbulence kinetic energy in the shear layers in the near-nozzle regions and hence the LES jet mixes somewhat faster and thus the potential core region is shorter in this simulation