Broadband Noise Modelling and Prediction for Axial Fans

F1 - Noise Prediction and Investigation for Axial Fans

Low noise design of HVAC components receives today increasing attention in the frame of automotive, railway and aerospace industry in order to improve the comfort experienced by passengers. Despite of the great progresses done by CFD the assessment of a reliable procedure for numerical simulation of fan noise is still a challenge and sometimes even the adoption of very CPU-demanding approaches (e.g. LES/DES) is not sufficient to provide good results. This is mainly due to the high complexity of the physical phenomena involved which imposes simulation requirements often resulting in a prohibitive computational cost for 3D complex configurations.
The target of this paper is providing a fast, reliable and computationally inexpensive technique for fan noise prediction based on the integration of aerodynamic models for turbomachinery with the semi-empirical aeroacoustic model for aerofoil geometries of Brooks, Pope and Marcolini (BPM) published by NASA in 1989. According to the BPM model the noise generation mechanisms occurring on an aerofoil surface can be classified in three main categories: Turbulent Boundary Layer - Trailing Edge noise (TBL-TE), Laminar Boundary Layer - Vortex Shedding (LBL-VS) noise and Separation Stall (S-S) noise. The BPM model allows estimating the sound pressure level density spectrum in one-third octave band for a generic aerofoil as function of boundary layer integral quantities via an algebraic process based on the correlation of experimental data.
This approach has been implemented in an in-house tool for axial fan and tested on experimental data from literature. The test-case chosen is a 6 bladed fan with a diameter of 299 mm tested at an operative rotational speed of 3000 rpm providing a nominal volumetric flow rate of 0.59 m3/sec. The aerodynamic simulation of this fan has been performed with two approaches:

• Blade Element Theory (BET) coupled with XFOIL
• CFD steady RANS (ANSYS-Fluent) with Multiple Reference Frame (MRF)

The noise spectra computed with the BPM model show a good agreement between experimental and computational data proving that the BPM model is able to capture sound pressure levels and broadband components over a wide range of frequencies (from 0.1 to 5 kHz), both coupled with BET/XFOIL as well as with CFD. Additionally attention is also focused on directivity effect implementation and interfacing issues of the BPM with the aerodynamic models mentioned above.