Hybrid Computational Aeroacoustics (CAA) methodologies are considered to be an efficient technique for the numerical simulation of aerodynamically generated noise and its far-field radiation. In an hybrid CAA approach, the final acoustic radiation is obtained through a three-step procedure. At first, the aerodynamic noise generating mechanisms are simulated using unsteady CFD techniques to obtain a solution for the aerodynamic fluctuating variables. In a second step, equivalent aeroacoustic sources are defined and the resulting CFD data are mapped from the CFD mesh to the acoustic propagation mesh. The final step consists in the calculation of the acoustic propagation using, in the presence of a non-uniform mean flow, appropiate propagation equations, such as the Linearized Euler Equations (LEE). Flows characterized by a low Mach and a high Reynolds numbers are commonly encountered in engineering applications. In this case, the dominating acoustic wavelengths are much larger than the hydrodynamic length scales and, as a result, the CFD grid has to be finer than the acoustic grid. A common assumption is that the CFD mesh spacing is approximately equal to the Mach number times the acoustic grid spacing. For this reason, an appropriate mapping procedure should be designed such that the CFD results are accurately represented on the acoustic grid. In the absence of a proper mapping routine the acoustic grid needs to be as fine as the CFD grid, which drastically reduces the computational performance. In this paper, a mapping technique using a least squares procedure combined with an anti-aliasing filter is proposed and validated for the two tandem cylinders problem [1]. It is shown that the application of the filter allows the use of coarser grids for the acoustic propagation. This technique results in big reduction of computational cost without reduction in the accuracy of the results.
展开▼
