Description
Numerical modeling of a supersonic underexpanded gas jet and subsequent mixture formation is performed in order to study the effects of numerical methodology and mesh resolution for two computational fluid dynamic codes. Accurately modeling supersonic underexpanded gas jets requires advanced computational techniques and highly refined meshes which are both computationally demanding and not suitable for the complex moving geometries encountered in many practical engineering applications. Many commercial codes employ the finite volume method to handle complex moving geometries. However, to date there has been very little published data examining the use of finite volume methods for simulating supersonic jets. The goal of this work is to assess the suitability and minimum resolution requirements for the computation of a supersonic underexpanded gas jet with engineering accuracy using a commercial finite volume method CFD code. Additionally, a 1D code was developed to estimate and predict gas jet nozzle injection and performance parameters. The effects of numerical methodology, modeling, and resolution on the prediction of underexpanded supersonic jet flows are compared using a commercial low-order finite volume solver, Converge™, and a high-order Hybrid Central/WENO-Z based solver developed at SDSU. The impact of grid resolution is clearly evident in the commercial code. Results of a grid convergence study reveal that failing to resolve the grid in the commercial code sufficiently result in drastic differences in the jets behavior, while the Hybrid code displayed only modest differences. Instabilities in the location of the Mach disk were observed in the commercial code but not present in the Hybrid code. The near nozzle shock structure was analyzed and the results of both codes were in good agreement. An analysis of the energy spectrum and Reynolds stresses at a downstream location that is almost fully turbulent displayed expected behavior for both codes. When sufficiently refined, the commercial code does a reasonable job of predicting the underexpanded gas jet within engineering accuracy. However, the commercial code was nearly 17 times more computationally expensive then the Hybrid code, which may limit its application to practical engineering problems.
