Fan muffler/backpressure/turbulence experiment

IsaacKuo · Post by **IsaacKuo** » Thu Jun 01, 2006 7:47 am

I've tried a simple experiment to determine whether a large fan muffler can reduce fan noise, and accidentally found out something interesting.

First, the simple muffler experiment:

I took a piece of carpet and rolled it into a large 2 foot long tube. I tried using an 80mm fan at both 5v and 12v in various locations/orientations. The results were not surprising:

1. Fan in open air was loudest.

2. Fan on one end of the tube was a bit quieter. It made no difference whether the fan was blowing out of or into the tube.

3. Fan within the tube was quietest by a wide margin. The high frequency noise was virtually eliminated. The low frequency noise was greatly reduced.

Now, for the interesting finding--I have long suspected that air bottlenecks inherently generate noise. I confirmed it by using a book to partially block the airflow. Whenever the airflow was restricted, "turbulence noise" would be generated. What's particularly interesting is WHERE this noise came from. The noise seemed to come mostly from the fan--even when the fan was on the opposite side of the tube from the restriction. This was true whether the fan was blowing into or sucking out of the tube.

In the past, I attributed this noise to "backpressure", but I've confirmed that "backpressure" is only half of the story. If the fan is sucking air out of the tube, it's not "backpressure" thats causing the problem. I don't know what the correct term is in that case.

It seems that for a give amount of restriction, the amount of "turbulence noise" generated was similar whether the fan was "blowing" or "sucking". And in both cases, the source of the noise seemed to be the area around the fan regardless of the location of the restriction.

This is important--it means that the "cause" of the turbulence noise may be very far away from where the noise seems to be coming from. A restrictive intake or hard drive cage could be the true cause of noise which is mistakenly attributed to an exhaust fan, or "cavity resonance".

Well, maybe I'm misinterpreting or misremembering the data. I find these results confusing and not intuitive.

At any rate, there's one lesson which is clear to me--avoid airflow restrictions! Airflow bottlenecks can result in extra fan noise where the true causes are not obvious.

nomoon · Post by **nomoon** » Thu Jun 01, 2006 9:09 am

In your experiement, is there any chance that the fan is speeding up when the restriction is introduced? I'm wondering if you are experiencing the "hair dryer effect" where a hair drier actually speeds up and gets louder if you put your hand over the intake or outtake of the hair drier.

Jason

narrasuj · Post by **narrasuj** » Thu Jun 01, 2006 9:23 am

nomoon wrote:In your experiement, is there any chance that the fan is speeding up when the restriction is introduced? I'm wondering if you are experiencing the "hair dryer effect" where a hair drier actually speeds up and gets louder if you put your hand over the intake or outtake of the hair drier.

My brain is being funny and wants me to believe that the increased spinning is due to decreased resistance (due to less airflow). Make any sense or am I just fried from too much sun?

IsaacKuo · Post by **IsaacKuo** » Thu Jun 01, 2006 10:24 am

nomoon wrote:In your experiement, is there any chance that the fan is speeding up when the restriction is introduced?

The fan isn't speeding up much, if at all. The pitch of the fan's "whirring" noise remains mostly the same. The extra "turbulence" noise is of a very different character--much broader.

Try out the difference between a fan in open air versus up against a flat book (to restrict/block airflow). The extra "turbulence" noise is the same as with a tube.

I'm wondering if you are experiencing the "hair dryer effect" where a hair drier actually speeds up and gets louder if you put your hand over the intake or outtake of the hair drier.

No, it's not that. Maybe the "hair dryer effect" only works with centrifugal blowers.

jaganath · Post by **jaganath** » Thu Jun 01, 2006 10:37 am

I have long suspected that air bottlenecks inherently generate noise.

There is no doubt this is the case. As the constriction gets narrower and narrower, the Reynolds number of the fluid will rise until the local flow trips over into turbulence, and it is the turbulence that makes the noise. If the local flow were purely laminar there would be very little noise at all.

Here is a very complete wikibook detailing all the various factors causing noise in axial fans of the PC case type:

Wikibook: Engineering acoustics, noise from cooling fans

Avalanche · Post by **Avalanche** » Thu Jun 01, 2006 10:41 am

Hey Isaac,

Probably what's happening is that when you severely obstruct the airflow, the fan stalls. The fan is happy operating over a limited range of flow rates, but at very low flow rates the flow over the airfoil separates, and the fan looses lift. It makes more noise when it stalls because the airflow is highly turbulent, if you like.

This link has a nice description of axial flow fans, although little explination of stall. That thing he shows in Figure 3 as the "fan curve" actually falls off a lot at low flow rates (far to the left). I can't find a good figure in 30 seconds of Googling, but if you're interested I can look more.[/url]

IsaacKuo · Post by **IsaacKuo** » Thu Jun 01, 2006 10:57 am

jaganath wrote:
I have long suspected that air bottlenecks inherently generate noise.
There is no doubt this is the case. As the constriction gets narrower and narrower, the Reynolds number of the fluid will rise until the local flow trips over into turbulence, and it is the turbulence that makes the noise. If the local flow were purely laminar there would be very little noise at all.

I'm not talking about a severe restriction and the noise seems to come from the fan rather than where the restriction is. This "turbulence noise" occurs even if the restriction is smooth and gradual.

Avalanche wrote:Probably what's happening is that when you severely obstruct the airflow, the fan stalls. The fan is happy operating over a limited range of flow rates, but at very low flow rates the flow over the airfoil separates, and the fan looses lift. It makes more noise when it stalls because the airflow is highly turbulent, if you like.

That sounds like a plausible explanation, although this effect occurs even with relatively modest airflow restriction. It gets worse with how restrictive the bottleneck is, of course.

The flow rate isn't the correct factor, though. It seems to depend upon how much of a restriction there is, not the flow rate. Running at 5v, the flow rate is obviously much lower than running at 12v. Nevertheless, the severety of the "turbulence noise effect" seems to occur at the same levels of restrictiveness. If it were something dictated by the level of airflow, then you'd expect that the restriction would need to be much worse in order to affect the fan at 5v rather than at 12v.

Anyway, this experiment is easy to duplicate. I invite others to try it out, and maybe come up with other things to try that I hadn't even thought of.

Avalanche · Post by **Avalanche** » Thu Jun 01, 2006 11:26 am

The flow rate isn't the correct factor, though.

Sorry, I wasn't very precise. The correcr factor is non-dimensional flow rate, that is, flow speed divided by rotor tip speed. The fan should stall at the same non-dimensional flow rate, regardless of rotation speed.

I think you're point is right, though. Airflow restrictions are bad. The fans do not like to work hard.

Bluefront · Post by **Bluefront** » Thu Jun 01, 2006 2:37 pm

All along I have building cases with the maximum amount of intake area. The problem comes with exactly where to place the intakes. An intake that just allows air to flow directly to an exhaust fan without cooling anything, might keep the fan quieter, but adds little to the cooling effect.

The other thing that has always amused me is the standard ATX form factor,.....you frequently see an exhaust fan, a 120mm PSU fan, and a CPU fan, all sucking air out of one small space. Guaranteed to produce un-necessary fan noise.

cmthomson · Post by **cmthomson** » Fri Jun 02, 2006 6:00 am

You're changing is the static pressure substantially on one side of the fan by adding impedance. This changes the amount of work the fan has to do, which of course changes its efficiency and the amount of turbulance.

In open air, the static pressure difference on the two sides of the fan is nearly zero, the fan is doing a minimal amount of work, and (if the blades are well designed) the air flow paths are nearly laminar.

As the static pressure difference increases, the amount of local stalling around the blade edges increases, which creates turbulance and noise. This wiki page explains stalling of airplane wings, but the concept also applies to fan blades: http://en.wikipedia.org/wiki/Stall_%28flight%29

EDIT: added "difference" in a couple of places.

IsaacKuo · Post by **IsaacKuo** » Fri Jun 02, 2006 6:16 am

That explanation makes perfect sense to me. It's consistent with the way that the noise increase is the same whether the fan is sucking from or blowing into the tube.

So instead of the term "backpressure", I'll use the term "static pressure" to describe this problem. "Backpressure" is inappropriately biased to consider only airflow restrictions on the "exhaust side" of the fan.

Thanks!

jaganath · Post by **jaganath** » Fri Jun 02, 2006 9:13 am

In open air, the static pressure is nearly zero

I don't think that can be true.

http://en.wikipedia.org/wiki/Static_pressure

The air pressure inside a latex balloon is a static pressure and so is the atmospheric pressure.

The static pressure in that case would be equal to the atmospheric pressure.

Isaac, instead of re-inventing the wheel, you might want to purchase a college-level textbook on aerodynamics and fluid mechanics.

IsaacKuo · Post by **IsaacKuo** » Fri Jun 02, 2006 9:30 am

What wheel am I reinventing? I just want to know the appropriate term to use instead of "backpressure".

In aeronautical terms, this effect seems similar to a "compressor stall"--refering to the intake fan of a turbine/jet engine.

cmthomson · Post by **cmthomson** » Fri Jun 02, 2006 10:48 am

jaganath wrote:
In open air, the static pressure is nearly zero
I don't think that can be true.

http://en.wikipedia.org/wiki/Static_pressure

Well, you're right, I was being sloppy. It is the static pressure difference on the two sides of the fan is nearly zero (ie, until the fan is ducted and has to overcome impedance it does very little work).

I've edited my earlier post.

datapappan · Post by **datapappan** » Mon Jun 05, 2006 7:16 am

Why is there more noise from the fan when the air path is obstructed?

First, refer to Comairotron site, to understand the concept of operating point.

When you obstruct the path, system impedance increases, and given the same rpm, the fan will give less air flow. This in turn, drives the fan along an increasing noise curve, as can be seen on page 4 of this document.

As for the fan reving up, I have no clue, please fill in.

/ datapappan

IsaacKuo · Post by **IsaacKuo** » Mon Jun 05, 2006 7:46 am

Thanks datapappan! Those links are incredibly informative, and exactly what I needed to know. I guess this quote says it succinctly:

"Noise figure is higher in the lower airflow area where a fan is
operating with higher system impedance because
eddy airflow is significantly higher."

There's the answer for me--the noise is "eddy airflow" or "turbulence" noise; the cause is excessive "impedance".

Thanks again!

pcy · Post by **pcy** » Fri Jun 09, 2006 5:02 am

Hi,

I have recently discovered exactly the same thing; and I also concluded that the principal effect was that the constiction was stalling the blades of the fan.

My experiment was part of the development of a quiet case. It's close to production and I'm hoping Mike C. will accpet one for review once I have units to spare.

I was designing the CPU cooler duct - it's set up with the CPU fan on the exhaust side of the heatsink, with a duct so that the hot air goes straight out of the case without re-circulating. This also eliminates the case exhaust fan...

The CPU cooler duct acts as a muffler, of course. It's made of dense acoustic foam.

Overall, I found that the fans were very sensitive to restrictions in the airflow. Trying to make the mufler more effective turned out to be counte-productive in many cases, because the increase in airflow impedance forced the fan to work harder (for constant cooling) and the overal noise level increased.

When I put a restriction over the fan (120mm running at around 1000 rpm) initially it reduced noise, but if I overdid it the noise suddenly increased significantly. The restriction in question was a flat plate made of a sandwidch of different acoutic materials, with a 3" hole in it. The noise increased when the plate came closer than 15mm to the fan.

It seems that the total cross sectional area of the exit duct is surprisingly critical. My results suggest you need around 5000 mm*2 (~8 sq ins), for a straight duct to carry 12 CFM without significant airflow impedance, but if you double the section you can safely put right angle bends in the duct which then acts as a low frequency muffler quite well.

I also found that the heatsink itself has huge impedance. Though the fans (I tried several) had nominal airflows of 30 - 50 CFM, the actual flow was 15 CFM or less. I measured the airflow by measuring the temperature rise of the air flowing through the CPU cooler and using the known power output of the CPU and the thermal capicity of air.

The formula I endid up with was:

1 Watt = 1.8 CFMC (CFM x temp rise in C)

Can anybody confirm this is (aproximately) correct.

In my case the noise was all relatively low frequency - 100 to 500Hz. I can find materials to block it, but can anybody recommend something to absorbe it? Almost all the noise is coming out of the exhaust airpath - I can't block that, obviously.

Peter