Engineering Acoustics/Noise from cooling fans
| A Wikibookian suggests that this book or chapter be merged with Acoustics/Noise from Cooling Fans.
Please discuss whether or not this merger should happen on the discussion page.
As electric/electronic devices get smaller and functional, the noise of cooling device becomes important. This page will explain the origins of noise generation from small axial cooling fans used in electronic goods like desktop/laptop computers. The source of fan noises includes aerodynamic noise as well as operating sound of the fan itself. This page will be focused on the aerodynamic noise generation mechanisms.
Inside a desktop computer, there may be three (or more) fans. Usually there is a fan on the heat sink of the CPU, in the rear of the power supply unit, on the case ventilation hole, and maybe on the graphics card, plus one on the motherboard chipset if it is a very recent one. The noise from a computer that annoys people is mostly due to cooling fans if the hard drive(s) is fairly quiet. When Intel Pentium processors were first introduced, there was no need to have a fan on the CPU at all, but most modern CPUs cannot function even for several seconds without a cooling fan, and some CPU's (such as Intel's Prescott core) have extreme cooling requirements, which often causes more and more noise. The type of fan used in a desktop computer is almost always an axial fan, while centrifugal fans are commonly used in laptop computers. Several fan types are shown here (pdf format). Different fan types have different characteristics of noise generation and performance. The axial flow fan is the main type considered in this page.
Noise Generation MechanismsEdit
The figure below shows a typical noise spectrum of a 120 mm diameter electronic device cooling fan. One microphone is used at the point 1 m far from the upstream side of the fan. The fan has 7 blades, 4 struts for motor mounting and operates at 13V. Certain amount of load is applied. The blue plot is background noise of anechoic chamber, and the green one is sound loudness spectrum when the fan is running.
(*BPF = Blade Passing Frequency) Each noise elements shown in this figure is caused by one or more of following generation mechanisms.
Blade Thickness Noise - Monopole (But very weak)Edit
Blade thickness noise is generated by volume displacement of fluid. Fan blades has its thickness and volume. As the rotor rotates, the volume of each blade displaces fluid volume, then they consequently fluctuate pressure of near field, and noise is generated. This noise is tonal at the running frequency and generally very weak for cooling fans, because their RPM is relatively low. Therefore, thickness of fan blades hardly affects to electronic cooling fan noise.
(This kind of noise can become severe for high speed turbomachines like helicopter rotor
Tonal Noise by Aerodynamic Forces - DipoleEdit
Uniform Inlet Flow (Negligible)Edit
The sound generation due to uniform and steady aerodynamic force has very similar characteristic as the blade thickness noise. It is very weak for low speed fans, and depends on fan RPM. Since at least of ideal steady blade forces are necessary for a fan to do its duty, even in an ideal condition, this kind of noise is impossible to be avoided. It is known that this noise can be reduced by increasing the number of blades.
Non-uniform Inlet FlowEdit
Non-uniform (still steady) inlet flow causes non-uniform aerodynamic forces on blades as their angular positions change. This generates noise at blade passing frequency and its harmonics. It is one of the major noise sources of electronic cooling fans.
If the fan blades are very close to a structure which is not symmetric, unsteady interaction forces to blades are generated. Then the fan experiences a similar running condition as lying in non-uniform flow field.
Impulsive Noise (Negligible)Edit
This noise is caused by the interaction between a blade and blade-tip-vortex of the preceding blade, and not severe for cooling fans.
Click here to read the definition and an aerodynamic description of stall.
The noise due to stall is a complex phenomenon that occurs at low flow rates. For some reason, if flow is locally disturbed, it can cause stall on one of the blades. As a result, the upstream passage on this blade is partially blocked. Therefore, the mean flow is diverted away from this passage. This causes increasing of the angle of attack on the closest blade at the upstream side of the originally stalled blade, the flow is again stalled there. On the other hand, the other side of the first blade is un-stalled because of reduction of flow angle.
repeatedly, the stall cell turns around the blades at about 30~50% of the running frequency, and the direction is opposite to the blades. This seriese of phenomenon causes unstedy blade forces, and consequently generates noise and vibrations.
Non-uniform Rotor GeometryEdit
Asymmetry of rotor causes noise at the rotating frequency and its harmonics (not blade passing frequency obviously), even when the inlet flow is uniform and steady.
Unsteady Flow FieldEdit
Unsteady flow causes random forces on the blades. It spreads the discrete spectrum noises and makes them continuous. In case of low-frequency variation, the spread continuous spectral noise is around rotating frequency, and narrowband noise is generated. The stochastic velocity fluctuations of inlet flow generates broadband noise spectrum. The generation of random noise components is covered by the following sections.
Random Noise by Unsteady Aerodynamic ForcesEdit
Turbulent Boundary LayerEdit
Even in the steady and uniform inlet flow, there exist random force fluctuations on the blades. That is from turbulent blade boundary layer. Some noise is generated for this reason, but dominant noise is produced by the boundary layer passing the blade trailing edge. The blade trailing edges scatter the non-propagating near-field pressure into a propagatable sound field.
Velocity fluctuations of the intake flow with a stochastic time history generate random forces on blades, and a broadband spectrum noise.
For some reason, a vortex can separate from a blade. Then the circulating flow around the blade starts to be changed. This causes non-uniform forces on blades, and noises. A classical example for this phenomenon is 'Karman vortex street'. (some images and animations.) Vortex shedding mechanism can occur in a laminar boundary layer of low speed fan and also in a turbulent boundary layer of high frequency fan.
Flow separation causes stall explained above. This phenomenon can cause random noise, which spreads all the discrete spectrum noises, and turns the noise into broadband.
Since cooling fans are ducted axial flow machines, the annular gap between the blade tips and the casing is important parameter for noise generation. While rotating, there is another flow through the annular gap due to pressure difference between upstream and downstream of fan. Because of this flow, tip vortex is generated through the gap, and broadband noise increases as the annular gap gets bigger.
Once a fan is installed, even though the fan is well designed acoustically, unexpected noise problem can come up. It is called as installation effects, and two types are applicable to cooling fans.
Effect of Inlet Flow ConditionsEdit
A structure that affects the inlet flow of a fan causes installation effects. For example Hoppe & Neise  showed that with and without a bellmouth nozzle at the inlet flange of 500mm fan can change the noise power by 50dB (This application is for much larger and noisier fan though).
Acoustic Loading EffectEdit
This effect is shown on duct system applications. Some high performance graphic cards apply duct system for direct exhaustion.
The sound power generated by a fan is not only a function of its impeller speed and operating condition, but also depends on the acoustic impedances of the duct systems connected to its inlet and outlet. Therefore, fan and duct system should be matched not only for aerodynamic noise reasons but also because of acoustic considerations.
Noise reduction of cooling fans has some restrictions:
- Active noise control is not economically effective. 80mm cooling fans are only 5~10 US dollars. It is only applicable for high-end electronic products.
- Restricting certain aerodynamic phenomenon for noise reduction can cause serious performance reduction of the fan. Increasing RPM of the fan is of course much more dominant factor for noise.
Different stories of fan noise are introduced at some of the linked sites below like active RPM control or noise camparison of various bearings used in fans.
Links to Interesting Sites about Fan NoiseEdit
Some practical issue of PC noise are presented at the following sites:
- Cooling Fan Noise Comparison - Sleeve Bearing vs. Ball Bearing (pdf format)
- Brief explanation of fan noise origins and noise reduction suggestions
- Effect of sweep angle comparison
- Comparisons of noise from various 80mm fans
- Noise reduction of a specific desktop case
- Noise reduction of another specific desktop case
- Informal study for noise from CPU cooling fan
- Informal study for noise from PC case fans
- Active fan speed optimizators for minimum noise from desktop computers
- Some general fan noise reduction technics
- Various applications and training in - Brüel & Kjær
 Neise, W., and Michel, U., "Aerodynamic Noise of Turbomachines"
 Anderson, J., "Fundamentals of Aerodynamics", 3rd edition, 2001, McGrawHill
 Hoppe, G., and Neise, W., "Vergleich verschiedener Gerauschmessverfahren fur Ventilatoren. Forschungsbericht FLT 3/1/31/87, Forschungsvereinigung fur Luft- und Trocknungstechnik e. V., Frankfurt/Main, Germany