Numerical Investigation on Vortex Structure and Aerodynamic Noise Performance of Small Axial Flow Fan ()
1. Introduction
Increasing concern about noise from electrical devices has led to increasing demand for quieter cooling fans. With the increasing of small axial fans released on the market, it is difficult to judge which ones have better acoustical performance. The present study is focused on the mechanism of sound generation due to unsteady behavior of vortex flow of fan internal flow field. Actually fan’s internal flow field has a great influence on aerodynamic noise [1-3]. Powell [4] proposed vortex sound theory reveal the relationship between vortex of the internal flow field and sound, and point out that sound should be produced under the conditions of existence of the vortex when the flow rate is very low. Sound was only generated in the flow field with the region of vortices changed over time. OU-yang Hua et al. [5] studied the flow field of low speed axial fans with different setting angle by PIV and CFD simulation, and analyzed the relationship between inner flow field of fan and noise radiation based on vortex sound theory of low speed homentropic flow, and then predicted aerodynamic noise of fan. It was shown that aerodynamic noise of low speed homentropic flow was mainly induced by stretch and breakdown of vortex in flow. To uniform inflow aerodynamic noise source of low speed axial fan was major caused by blade trailing edge vortex shedding and tip vortex noise, moreover intension of preceding noise source is highly large than the latter.
Nevertheless, few references had elaborated the vortex formation mechanism and flow characteristics of the internal flow field of small axial flow fan whose diameter is less than 200 mm. The main distributions of aerodynamic noise sources and the effect of vortex structure on fan aerodynamic noise were still not very clear, and further study had to be carried out. The commercial FLUENT software is used in this article. The unsteady characteristics of vortex structure in fan internal flow field are analyzed by large eddy simulation. Broadband noise model is introduced to calculate the distributions of broadband noise sources in fan internal flow field, and FH-W acoustic model is introduced to calculate aerodynamic noise affected by the unsteady characteristics of vortex structure. The results will provide a useful reference about further optimizing noise characteristics of the fan.
2. Fan Prototype
A small axial fan RF24S9225H is selected as a study prototype in the paper, as shown in Figure 1. The impeller diameter is 84 mm, the impeller thickness is 18 mm, hub ratio is 0.39, tip clearance is 1.5 mm, number of blades is 7, blade stagger angle is 46.9˚, and the rated rotating speed is 3000 r/m, rated air volume is 0.010 kg/s, rated air pressure is 37.25 Pa.
3. Mesh Generation of Computational Domain and Turbulence Model
3.1. Mesh and Computational Domain
To make the air flow more fully, the arrangement of fan flow field should be reasonable. The center of fan hub is set as the coordinate origin and the length of outlet extension is 400 mm, whose diameter in the computational domain is 200 mm, as shown in Figure 2. Meanwhile, the computational region is divided into 4 parts, nonstructural grid (tetrahedral T-grid) is used in rotating fluid area and surrounding pipeline, while structural grid (hexahedral cooper mesh) is used in the regions of front channel and back channel, as shown in Figure 3. The total grid positions are 2.86 × 106 for fan prototype. The degree of twist of the grid is predominantly between 0.1 - 0.5. The inlet boundary condition is set as a given flow rate, for the outlet the total pressure is taken as a boundary condition. The solid walls such as vane surfaces and hub satisfy the no-slip condition in the computational domain.
3.2. Numerical Calculation
The Mach number of the airflow is under 0.3, hence, the
fluid is regarded as incompressible. Large eddy simulation (LES) can accurately simulate the turbulent field because it can accurately calculate not only the large scale turbulent fluctuation, but also the small scale turbulent fluctuation with the appropriately closed model [6]. In this paper, the steady flow field with standard k-ε turbulence model is the initial solution to the unsteady flow field with the large eddy simulation. The filtered N-S equations with finite volume method are adopted. S-L model with simulating sub-grid scale effects is applied and PISO algorithm to solve the coupling of velocity and pressure is used [7]. Momentum equation is calculated by the second order central difference scheme. The second order implicit scheme is used as time advance, and time step of unsteady-state calculation is set to 2 × 10−5 s.
The noise prediction can be carried out only when the statistically steady-state solution was acquired in large eddy simulation of the unsteady flow field [8]. FW-H equation based on Lighthill acoustic analogy theory is used to simulate sound production and dissemination when the noise is predicted [9]. The equation expression is:
(1)
In the formula: 
is the fluid velocity component along the 
direction, 
is the fluid velocity components perpendicular to the plane f = 0, 
is the surface velocity component along the 
direction, 
is the surface velocity components perpendicular to the plane f = 0, f = 0 is integral surface, f < 0 is sound source zone, f > 0 is far-field zone. 
and 
are the Heaviside function and Diracdelta function respectively:
(2)
(3)
Noise spectrum distribution from the axial fan is acquired after the source data has been obtained in time domain integration and then fast Fourier transform (FFT) processing has been applied. Time step of noise calculation is set to 2 × 10−5 s and cutoff frequency of noise calculation is 20 Hz.
When the monitoring point is set up at 1 m away from the central axis of the impeller along fan outlet, sound pressure level of aerodynamic noise by the numerical calculation is 32.6 dB, compared with 33 dB noise index of fan prototype from fan factory, and the difference is less than 1 dB. It is proved that the numerical calculation method of fan aerodynamic noise is feasible.
4. Results and Discussion
4.1. Distributions of Broadband Noise Sources
In order to clarify the distributions of broadband noise sources in fan’s internal flow field, broadband noise model is introduced after the stability of steady calculation [10]. Sound power level distributions of broadband noise in the surface of fan’s flow field are calculated by broadband noise model, as shown in Figure 4. It can be seen from Figure 4 that the maximum sound power is distributed in the rotating fluid area of fan flow field. Figure 5 shows sound power level of broadband noise of meridian plane (X = 0 mm). It can be seen from Figure 5 that sound power level of broadband noise is mainly concentrated in the rotating fluid area of fan’s internal flow field, where sound power level close to the outlet is the greater than that of other regions. According to the results of Figures 4 and 5, broadband noise sources are
Figure 4. Sound power level distributions of broadband noise on the surface of fan’s internal flow field.
Figure 5. Sound power level of broadband noise of Meridian plane (X = 0 mm).
mainly distributed in the rotating fluid area of fan’s internal flow field.
In order to clarify the distributions of broadband noise sources in the rotating fluid area of fan’s internal flow field, sound power level distributions of broadband noise of different revolution surfaces and meridian plane are analyzed respectively. Figure 6 shows sound power level contour of broadband noise of meridian plane (Y = 0 mm). It can be seen from Figure 6 that the maximum sound power is distributed in the tip clearance region and blade trailing edge region of the rotating fluid area and the maximum sound power at the tip clearance region is greater than that at blade trailing edge region.
Figure 7 shows sound power level contour of broadband noise of revolution surfaces (R = 24 mm and R = 45 mm). It can be seen from Figure 7(a) that the maximum sound power is distributed in blade trailing edge region, while the maximum sound power is concentrated in tip