Implementation and Development of a Stochastic Turbulence Model Based on RANS Formulation for the Prediction of Aeroacoustic Noise

In the computation of aeroacoustic noise, both the Lighthill analogy and the linearized Euler approaches require the definition of source terms involving instantaneous flow fluctuations, which are generally obtained from either Direct Numerical Simulation (DNS) or Large Eddy Simulation (LES). However, these approaches are not economically viable in terms of computational resources, as they require very fine grids to deliver accurate results. Therefore, the Stochastic Velocity Field generation model (SVFG) has has been applied in this paper. The SVFG model is based on the concept of the Stochastic Noise Generation and Radiation (SNGR) without sound propagation and linearized equations. The SVFG model uses time-averaged quantities from the Reynolds Averaged Navier-Stokes equations (RANS) to generate a synthetic time dependent turbulent flow field. The turbulent fluctuations are modeled using a stochastic description of the three-dimensional turbulent motion with a discrete set of Fourier modes. This synthetic turbulent field represents many of the characteristics of real turbulence. Nevertheless, it still has some imperfections; although it exhibits the expected correlation length and the required ratio of length scales, it does not predict the convective properties of shear flow turbulence, as the approach generates homogenous and isotropic turbulence. These properties are shown in this paper with the test case of an axial-symmetrical subsonic jet. The SVFG model is used to generate the turbulent flow field, which then is used to compare with actual experiment measurement and other prediction methods. The results of the comparison show strengths and weaknesses of the model. Since the SVFG approach is relatively low cost when compared to both LES and DNS, it offers an attractive alternative to derive the turbulent flow field.