It is well known that unwanted noise from fans can be obtrusive, with the noise being generated from interactions between the fan blades, vortical structures within the flow, and the boundaries and obstacles within the flow, namely flow-induced vibrations, (i.e. aero-acoustics).

ESI has recently undertaken a Research & Development program to develop a technology to drastically reduce the acoustic emissions produced by small to medium sized centrifugal fans by 5 dB to 10 dB in the 1 kHz to the 5 kHz frequency range, with an emphasis on fan wheel diameters between 3” and 20” (75 mm to 500 mm). Preliminary tests showed that the acoustic signature of a flow field can be significantly augmented by Fluid - Fluid Interactions (FFI), and Fluid - Structure Interactions (FSI) which are directly associated with vortical structures within the flow and their interaction with each other and the surrounding boundaries, highlighting the oscillating pressure field and noise they produce. 


Initial testing showed the ability to substantially alter the sound pressure levels (SPL) using surface modifications with SPL changes between 7 dB and 28 dB depending on the selected surface modification. Figure 1 presents the result of 19 different surface arrangements producing a maximum change of approximately 20 dB.

This change in SPL may be better understood by comparing the Power Spectrums that were obtained from miniature omni-directional microphones for just two different geometric modifications, Figure 2, where it may be seen that the surface modifications has resulted in two totally different frequency fields.

Figure 1: Relative sound reduction for different surface modification


Figure 2: Spectrums from surface mounted pressure transducers for two different surfaces

A further, in-depth, program was then initiated and separated the development into two distinct phases, namely; 

Phase 1, where a water tunnel was used to evaluate the performance of modified blades,

Phase 2, where velocity and acoustic measurements were performed using ESI’s Fan Acoustics Test & Evaluation (FATE) Bed (Figure 3).


Figure 3: Diagrammatic of the FATE Bed Facility

Phase 1
An examination of the structure of the blade assembly was first undertaken using a Technological Program developed by ESI and then modified. From this analysis a 2D array of blades were geometrically scaled and fabricated using a 3D printer and evaluated using flow visualization in ESI’s water tunnel.

The resulting images of the flow through the blade assembly were processed and analyzed, and surface modifications that showed the potential for noise reduction were identified and new blade configurations were then prepared for Phase 2.


Figure 4: Water Tunnel Blade Array


Figure 5 : Air Blade Array Connected to FATE Bed

Phase 2

Selected blade configurations base on Phase 1 were rescaled back to the original sizes and a 7 blade array was 3D printed and fabricated for acoustic testing and air flow performance evaluation within FATE Bed (Figure 5).

Miniature hot-wire velocity and turbulence intensity traverses were performed downstream of the blade trailing edge (Figure 6), and the data shows increase in wake turbulence as the blade inlet angle is increased from 60 to 70 degrees.


Figure 6 :Turbulent Intensity Contours Experimentally Acquired Using Miniature Hot-Wire

In addition to the hot wire anemometry measurements, acoustic measurements were also taken at the inlet to the blades and the corresponding results are presented in Figure 7 where it can be deduced that surface modifications can give rise to significant changes in the SPL without altering the frequency spectrum.


Figure 7: Acoustic Measurements Acquired Upstream of the Blade Array Inlet for Several Different Surface Modifications and a Reference Case


Furthermore, it has also been shown that the overall reduction in sound pressure level (SPL) may be significantly altered as depicted in Figure 8 where results are presented for 19 different blade surface modifications, and shows a reduction from 1.2 dB to 6.5 dB is achievable depending on the modification. 

Figure 8: Relative Changes in SPL When Compared to a Reference Case