Fans are widely used in electronic devices and are mainly used to dissipate heat. In order to reduce costs, people are constantly reducing fan diameters and increasing speeds, resulting in higher noise levels. In order to choose the right design, precise and economical tools are required. Fans are used under product installation conditions, with the exception of free space or wind tunnel test conditions, so an accurate acoustic field is required**. This article introduces the "Numerical Prediction of Noise Generated from a Box Fan" published at Inter-Noise 2023.
Test method and calculation parameters
*Mention two methods of fan noise sound field: one is to assume that the sound travels in free space, combined with CFD software**; The other is a hybrid approach, where the sound source is extracted from the CFD software and propagated through the acoustic solver. In this test, the unsteady flow field was calculated using the SCFLOW software in the CFD software, and the sound pressure in the free-space sound field was **. We also use ACTRAN software to calculate sound pressure levels. Both software are developed by Hexagon Industrial Software and have good data interactivity.
SCFflow is a comprehensive CFD software based on the finite volume method (FVM) for arbitrary polyhedral meshes. In this study, an incompressible unsteady pressure solver and the WALE model of LES were used, and a fine mesh was employed. The time interval is 360° 4096, i.e. a circle is divided into 4096 steps. The RANS calculation results are used as the initial conditions for LES**. Twenty-five cycles were calculated, with the first 5 laps reaching steady state and the last 20 laps being used for evaluation. The inlet and outlet conditions are total pressure and flow rate, respectively. SCFflow features FW-H sound pressure**, which encrypts the mesh around the side walls of the shroud to capture vortex trajectories.
The commercial software Actran calculates the far-field propagation to the relevant sound source based on the finite element method (FEM). The solver takes into account both the installation effect and the convection effect caused by the average flow. The dipole sound source dominates the small fan case and is generated by the load of the rotating blades, performed with a fixed dipole ring embedded in the acoustic domain**. After completing the force mapping and Fourier transform, the dipole in the frequency domain is obtained. In acoustics**, the setting does not include rotating blades and is replaced with a static dipole. In the far field, the surface of the enclosed domain is non-reflective, and the necessary information is retained. The surface of the strut and casing is completely rigid to the propagation of sound waves. The solver provides acoustic information, such as sound power or acoustic contours.
As with the FW-H**, the dipole source is calculated from the last 20 fan rotations of the CFD**. The sources are arranged in three time intervals, each corresponding to 10 turns (0.).2s）。Through the calculations of the two software, we compared the pressure rise results with the experimental results and visualized the flow field. We also compared the sound pressure levels of the FW-H method and the dipole ring method** with experimental results. The differences in the results of the two methods are discussed, as well as the reasons for the different acoustic properties.
Calculation results
Figure 1 Fan performance characteristics
*The performance of the fan was compared with the experimental results by SCFLOW analysis, as shown in Figure 1, and it can be seen from the comparison that the fan performance** is very accurate.
Baseline fans don't perform well in the low flow rate range, while high-load fans perform well in this range. The high-load fan achieves a higher pressure rise and a wider operating range by shifting the stall point to the lower flow side. The numerical simulation results are in good agreement with the experimental results, indicating that the flow field is well reproduced.
Fig.2 Eddy current visualization using the Q standard.
Vortex structures are mainly found near the wall surface and around the leaf tips and side walls. For two fans with high flow velocities, the tip vortex gradually increases and flows smoothly under the next blade. At medium flow rates, there are differences in flow characteristics between the two fans. The eddy current at the tip of the reference fan interacts with the leading edge of the next blade to increase instability. However, there is no obvious interaction with high-load fans. At low flows, the airflow on both fans is completely stagnant.
Fig.3 Comparison of the sound pressure spectra of the FW-H method and the dipole ring method with the experimental and numerical results at the monitoring points.
For sound pressure, we compared the FW-H method and the dipole ring method, as shown in Figure 3. In the figure above, FW-H is the calculation result of SCFLOW, and Ring Dipoles is the calculation result of the combined dipole ring method of Actran. The dipole ring method is closer to the experimental results, especially in the low frequency part, which may be due to the fact that the FW-H assumes that the sound travels in free space, while the ACDRAN of the dipole ring method is closer to the actual device. Experiments show that when the first blade passes at a frequency (bpf) of 250 Hz, a strong peak occurs, especially a high flow rate =022 o'clock. But neither method is able to ** these peaks.
Fig.4. Comparison of the correlation between radiated sound pressure and flow velocity; Sound pressure levels at monitoring points and radiated power by the dipole ring method** observed experimentally.
We also compared the radiated sound pressure as a function of flow velocity. The experimental results are consistent with the radiated power of the dipole ring method**. For baseline fans, the sound pressure level is the same for medium flow as for low flow; For high-load fans, medium flow is quieter than low flow. The dipole ring method successfully captures this difference. For high-flow fans, the observation point cannot pass the first BPF peak of the value**, but these peaks can be observed by the radiated sound power. This can be related to exterior wall interference or reflections.
It is well known that wall pressure is related to sound production. The sound pressure in FW-H is obtained from the wall pressure, and in the dipole ring method, the sound pressure is calculated from the fluid force on the blade. At low and medium flow rates, the pressure fluctuates greatly, especially near the leading edge of the blade. This is due to the interaction between the blade tip and the vortex. And high and medium flow = 0No strong pressure fluctuations were observed in the fans of 16, and the noise level was reduced.
Conclusion
To sum up, we propose two methods for the noise of box-type fans: one is the FW-H method, which flows and acoustics through CFD software; The other is the dipole ring method, which is a combination of CFD and acoustic software. These methods have been validated on small box fans with two different blade configurations. We also calculated the three flow coefficients of the wind turbine and solved for the unsteady flow field using fine-mesh LES.
The numerical results show that the pressure rise performance is consistent with the experimental results. The two fans have different stall points, and the calculations show this. Comparing the experimental results with the sound pressure levels of the FW-H method and the dipole ring method**, it was found that the results of the dipole ring method were closer to the experimental results, especially in the low frequency range. The dipole ring method can qualitatively capture the relationship between radiated sound pressure and flow velocity in experiments. Finally, we discuss the reasons why different blade structures produce different sound characteristics. At moderate flow rates at baseline fans, the interaction of the blade tips with vortex currents results in strong pressure fluctuations and noise near the leading edge of the wing.