A comparison of low and high-order methods for the simulation of supersonic jet flows

Abstract

The present work compares results for different numerical methods in search of alternatives to improve the quality of large-eddy simulations for the problem of a supersonic turbulent jet flows. Previous work has analyzed supersonic jet flows using a second-order, finite difference solver based on structured meshes, and the results indicated a shorter potential core of the jet and different levels of velocity fluctuations. In the present work, the results of previous simulations are compared to new results using a high-order, discontinuous Galerkin solver for unstructured meshes. All simulations are performed, keeping the total number of degrees of freedom constant. The results of the current simulations present very similar mean velocity distributions and slightly smaller velocity fluctuations, and they seem to correlate better with the experimental data. The present results indicate that additional studies should focus on the jet inlet boundary conditions in order to improve the physical representation of the early stages of the jet development.