Comparison of RANS-based jet noise models and assessment of a ray tracing method

Abstract

Four models to predict jet mixing noise available in the literature are reviewed and compared. Two models are derived from the Lighthill Acoustic Analogy (LAA) and the other two use the Linearized Euler Equations (LEE) with added source terms. All models use input from a computational solution of the Reynolds-Averaged Navier–Stokes (RANS) equations and empirical functions to model turbulent correlations. The models are concerned with the sources of sound related to small turbulent structures. Results for an observer at 90 $$^{\circ }$$ to the jet axis are presented, where effects of sound-flow interaction are negligible and the small structures are considered to act as the dominant source. A range of subsonic jets issued from different nozzle geometries with different exit velocities and temperatures is considered. In addition, a ray tracing method to compute the effects of sound refraction is reviewed. The comparison of the numerical predictions with experimental data shows that using the LAA with an improved description of turbulence statistics results in a model as accurate as the ones based on the LEE that relies on less empirical coefficients and has a simpler mathematical derivation. The results corroborate the use of RANS along with LAA to model jet mixing noise.