Modeling Swirling Jet Flows Using a Hybrid RANS/LES Methodology

This study evaluates whether unsteady simulations of swirling jets that resolve a large range of scales can better predict the enhanced mixing in swirling flows particularly since two-equation RANS models have been shown to be very poor at accurately predicting both the axial and swirl velocity mixing rates in swirling flows. Unsteady, three-dimensional, hybrid RANS/LES results for a swirling jet are presented. The configuration chosen has a moderate swirl number of 0.23 for which an experimental database with detailed mean and turbulence measurements exists (Gilchrist and Naughton 1 ). The simulation results show very good comparison with the experiment and demonstrate that an unsteady framework employing hybrid RANS-LES methodology is able to resolve the disparate length scales of mixing seen in the axial and swirl directions. Furthermore, the hybrid RANS/LES methodology reduces computational cost, and makes the methodology practical from an "engineering" perspective on present day computational resources. The jet flow field shows that the swirl decays very rapidly by x/D~5, and further downstream the jet behaves similar to a non-swirling one with a linear growth rate and exhibiting self-similarity in the axial velocity profiles. Results from a second-order and a higher-order (4 th order) spatial scheme do not show much difference in prediction of the mean flow, but the turbulent fluctuations, particularly the axial component, is better predicted by the higher-order scheme.