Technical Program

< Go back to the papers' list


Investigation of Different Fan Noise Prediction Methods


Noise Prediction by Analytical or Numerical Models


JUNGER Clemens
Technische Universität Wien

Vienna - Austria
Technische Universität Wien

Vienna - Austria


Fan noise prediction is a necessary stage in all modern fan design processes. For the noise prediction a lot of different methods exist, which can be roughly categorized in three different classes. The first class contains empirical methods, based on geometry and operating conditions. They are very fast and easy to apply. The second class contains semi-empirical methods that are partly based on flow quantities from measurements or simulations. These methods shall represent the actual operating conditions in a better way, but are more time consuming since the flow quantities must be obtained. The third class contains methods that compute the noise radiation directly from fluctuating quantities of the transient flow. These methods give a good insight in the contribution to the overall sound level from different sound sources. Due to the need of transient flow data, this class of methods is highly demanding in computational power.
In our contribution we compare different methods from all three classes and apply them to a low-pressure axial fan. The fan used for the application is a previously published benchmark case. The generic fan is a typical fan to be used in commercial applications. The rotor benchmark fan provides an extensive amount of measurement data including aerodynamic performance (volume flow rate, pressure rise and efficiency), wall pressure fluctuations in the tip gap region, fluid mechanical quantities on the fan suction and pressure side (velocity in three spatial direction and turbulent kinetic energy) and acoustic spectra at different microphone positions. Our numerical approach from class three is based on a forward coupling between a flow simulation with ANSYS Fluent and an aeroacoustic source term and wave propagation computation with multiphysics research software CFS++. The comparison of the different methods and the measurement include averaged sound power level as well as spectral sound power level.