Direct Noise Computation of Low-speed Ring Fans

Direct noise predictions of low speed fans are achieved with both compressible RANS and LBM simulations on an engine cooling axial fan in several configurations with increasing complexity. Both methods yield good tonal noise predictions within the experimental uncertainty at reasonable computational costs on the simplest fan-alone configurations. Only the LBM simulation can efficiently handle more complex geometries such as engine cooling module with possibly upstream rotating obstruction provided a large number of cores are used. Furthermore it can predict the broadband noise up to a cut-off frequency, and propagate acoustics efficiently to the far field. Both predicted noise levels and spectral shapes compare well with measurements on the complete installed fan system. Combining the LBM results with an acoustic analogy shows that the dominant sources are dipolar mainly in the tip region of the blades, with negligible quadrupole sources in the flow. Moreover, the strongest wall-pressure fluctuations for tonal noise sources are found in the tip region of both fan blades and stator vanes. An obstruction placed upstream to control tonal noise is also found to have a negligible self-noise. In the latter case, a complete numerical methodology with a maximum of five LBM simulations has been proposed.