Unsteady Simulation of Tonal Noise from Isolated Centrifugal Fan

Noise Prediction by Analytical or Numerical Models

In this study, tonal noise from an isolated centrifugal fan is investigated using unsteady Reynolds-averaged Navier-Stokes (URANS) simulations. Isolated centrifugal fans are often used in ventilation systems and play an important role in producing tonal and broadband noise. The broadband noise can be reduced when the fan efficiency is optimized. However, the tonal noise cannot be effectively reduced. It is therefore of great interest to identify and reduce the tonal noise for this type of fans operating in public environment. The noise is predicted by coupling the URANS and the Ffowcs Williams and Hawkings acoustic analogy. The numerical methodology and mesh generation methods are validated. Numerical simulation is a more convenient way to identify tonal noise than experiments. However, simulation of the noise using advanced computational fluid dynamics (CFD) methods (e.g., large eddy simulation) requires many computational resources. To predict the tonal noise, a potentially convenient method is the URANS. The method can simulate characteristic unsteady structures, which are responsible for the tonal noise generation, with low computational costs. Though it has a drawback to provide the fluctuations that are important for the broadband noise generation.
The aerodynamic properties obtained from RANS and URANS are consistent with the experimental data. The magnitudes of the tonal noise at the blade passing frequencies are well predicted. Moreover, the broadband noise below 350 Hz agrees with the measurement although obvious discrepancies are found at high frequencies, which cannot be resolved by URANS. Recirculating flows, which are responsible for reducing the fan efficiency and increasing the noise generation, are observed between the shroud and the blade trailing edges. It is found that the recirculating flows are associated with the gap that is between the shroud and the inlet duct.