Aeroacoustic Simulations of an Axial Fan with Modelled Turbulent Inflow Conditions
E2 Noise Prediction by Analytical / Numerical Methods
Evaluating and reducing the emitted acoustics is an important part in the development of new generations of fans. Especially for fans, which operate in applications close to humans (e.g. air conditioning, kitchen hoods, etc.), low noise generation is requested. In such applications the acoustics may differ significantly compared to measurements or simulations of the standalone fan due to the turbulent character of the inflow conditions. For axial fans these turbulent inflow conditions are often caused by upstream heat exchangers. Due to the complex geometry and dimensions of typical heat exchangers, it is economically not possible to properly resolve the turbulence generating geometry in scale resolving aeroacoustic simulations (LES).
The FRPM (Fast Random Particle Mesh) tool, developed by the DLR Braunschweig is able to reconstruct turbulent fluctuations based on statistical turbulence properties (e.g. turbulence intensity and length scale) derived from RANS simulations. Using this method it is possible to synthesize the time dependent turbulent fluctuations caused by complex geometries, such as heat exchangers, using precursor RANS simulations. The reconstructed turbulence can then be used to model the inflow conditions due to the presence of a heat exchanger in scale resolving aeroacoustic simulation of fans. This approach allows the evaluation of the effect of turbulence generating geometric features on the acoustics of the fan with feasible computational effort.
The developed approach is tested by simulating a ducted axial fan, which was experimentally investigated at the University of Siegen. In the experiment the turbulent inflow conditions are generated by a turbulence grid. Besides the extensive experimental investigations, the setup of this experiment allows the simulation of the fan and the turbulence generating grid with reasonable computational effort. The simulation results with the modelled inflow turbulence show good agreement with the measurements as well as a simulation of the entire setup. However, due to the missing of the turbulence generating geometry as well as additional upstream geometries in the simulation with reconstructed turbulence, the effect of these features on the acoustic propagation cannot be represented.