Aerodynamic noise prediction is a major challenge in computational aeroacoustics due to the complexity of phenomena involved such as turbulence and laminar to turbulent transition. Accurate numerical methodologies, capable to provide reliable predictions in a reasonable computational time, are of large interest for the industrial design of Heating, Ventilation and Air-Conditioning (HVAC) systems. The objective of the present research work is to benchmark different CFD/CAA simulation setup (i.e. mesh topologies, boundary conditions) for predicting the broadband noise generated by low speed axial fans to develop guidelines for reliable and computationally affordable simulation.
In previous works the authors investigated the capabilities of the Zonal LES technique coupled with the Ffowcs Williams-Hawkings acoustic analogy for the prediction of the noise generated by an unducted low speed axial fan. The results showed a good agreement with aerodynamic and aeroacoustic experimental data.
Despite the achievements obtained so far, the high physical and numerical complexity of the problem calls for further investigations. The latest developments, presented here, focus on the impact of the mesh topology and the inflow turbulence on the far field noise prediction.
Two computational meshes with different topology are investigated: an unstructured-hybrid mesh, which can be generated with fast and highly automated methods, and a structured-hybrid mesh, which allows better control of the volume mesh around the blade. Both meshes are designed to adequately resolve the boundary layer, providing LES driven values of y+, x+ and z+ on the blade surface for the operating condition considered.
Two different levels of inflow turbulence are studied, one representing an ideal turbulence-free unbounded environment, and one mimicking the experimental measurements environment.
All the aerodynamic and aeroacoustic simulation results presented are benchmarked with experimental data acquired by the authors.