This paper examines the capability of both the Proper Orthogonal Decomposition (POD) and the Singular Value Decomposition (SVD) to identify the zones on the surface blades of a centrifugal fan that contribute the most to the sound power radiated by moving blades. The Computational Fluid Dynamics (CFD) OpenFOAM~? source code is used as a first step to evaluate the pressure field at the surface of the blade moving in a subsonic regime. The fluctuating component of this pressure field makes it possible to directly estimate both the loading noise and the sound power that is radiated by the blade based on an acoustic analogy of Ffowcs Williams and Hawkings (FW&H). In the second step, the estimated loading noise is then employed to evaluate the radiated sound power using the POD and SVD approaches. It may be noted that the sound power reconstructed by the two latter approaches, when relying solely on the most important acoustic modes, is similar to the one predicted by the FW&H analogy. It is also noted that the contribution of the modes in the radiated sound power does not necessarily appear in ascending order in the decomposition (i.e., in descending order of energy). Moreover, the highest radiating SVD modes are mapped onto the blade surface so as to highlight the zones that contribute the most to the noise. It is then expected that this identification will be used as a guide in the design of the blade surface to reduce the radiated noise.
展开▼
