Fan Idea - captured, swept *back* blades

[NeilBlanchard]

Hello:

As somebody pointed out earlier, since the AeroCool's blades are basically straight, it is going to "flare" the air out more, since the air "slips off" the ends of the blades. Most fans have blades that are swept forward and so the air "slips" towards the center hub, and so they "hold" the air in a straighter column as it blows away from the trailing edges. The air coming away from the AeroCool is going to be more conical; getting wider as it blows away from the trailing edges.

This brings up a point that I was contemplating a while back: since the tips of the blades are moving faster, they would be able to move more air (or possible the same amount of air, more quietly) if they were swept back -- and there was a cylindrical rim at the blade tips that prevented the air from "slipping off" the ends of the blades?

Like this:



The red cylinder captures the ends of the blades -- it is attatched to the blades, and spins with the fan, like an outer rim. The cyan blades are swept back -- and they should be curved; I am showing them as straight. (Due to my limited skills with the modeling software, which is DataCAD. There is a very powerful macro, called Sweep, that would probably do this, but I don't know how to use it...)

The front and back face plates are connected by the eight rectangular struts, and these are not touching the outside of the red cylinder.

[Tzupy]

Interesting idea, but I guess the turbulence resulting from the air colliding with the 'shroud' could impare the airflow, and make more noise.

[NeilBlanchard]

Hello:

You may have point -- but on the other hand if things are smoothly filleted and if the blades are curved, it might not be any worse than having the tips of the blades (on a regualr, typical fan) cutting through the air...and no air "escapes" around the end of the blade (aka turbulence) -- this may be a quieter and more efficient design?

The other point I should make, is that the fan blades would not have to have the gap at the tip (as all typical fans do) and therefore the blades at slightly longer! And since the tips are moving the fastest, and can have the greatest surface area -- the fan could possibly push more air at lower RPM's.

Please notice that the motor support struts "cross" the trailing edges of the fan blades at nearly 45 degrees, so the "shadow" effect is minimized, and this would also make things quieter. Some (or many?) existing fans also do this (the Yate Loon 120mm do!), but some have them more or less parallel -- which is not very ideal. BTW, one of the struts would have to be a bit wider to accomodate the wires...the one that is in the "nine o'clock" position in the upper image, would be the best place to locate these?

[Edit] I'd be really happy if some R&D department wants to use my idea -- please consider it an "open source" idea; which means that any refinements or tweaks should be shared with all of us!

[Tibors]

Actually that shroud should reduce noise. See this .pdf about Y.S.Tech TMD fans. (I once had a more technical explanation from their site, but they redesigned their website.) This not strange either. Air "escaping" at the tip of a fanblade is very turbulent. With the rightly engineered shape of the connection between the blade and the ring it shouldn't be hard to reduce that turbulence.

[Tzupy]

Well, IMO the only way to find out how well this idea works is to build several protoypes (slightly different) and test them. But if a R&D department of a fan manufacturer is going to make this idea work, they won't treat it as 'open-source', they'll patent it...

[BrianE]

This is an interesting idea, but I'm wondering if maybe it's not as great in practice as it is in theory. The only reasons I say this is because you don't see a lot of fan blade applications like this in the real world, and another thing I started thinking about was naval screw propeller applications. Some of them are shrouded, but I don't think any of them have this type of design... if anyone would have thought of this as a way to reduce noise it would have been the military submarine makers probably decades ago. Maybe the problem is that it creates less noise to have the water slip off the tips of the blades rather than having it collide with the outer hub? I don't think it's an unreasonable stretch to compare the two because they both work using fluid dynamics.

Then again I'm hardly an expert on this - feel free to criticize.

[nici]

Anyone ever thought of a fan without a center-hub, with the blades all the way to the middle, and a frame like that around the blades that spins with them, and a small motor on the side spinning the whole thing.. Think belt-drive in mini-scale. You could have a small hub in the middle with some quality bearings, or some kind of huge bearing around the spinning thingie and no need for a hub..
Just an idea, im thinking this is too complicated to do with no real benefits though

[Splinter]

Nici - Yes, there are fans like what you describe, called 'TMD' or 'Tip Magnetic Drive' fans.

First link I could find[/url]

[StarfishChris]

Click the link in Tibors's post I don't think there would be much difference when driven with a belt instead of the magnets, though.

[nici]

Damn.. i missed that link Anyway, YS-Tech seems to on to the second generation on these already with only one coil. They dont have any size information available though. No information on availability either, i would certainly de interested in trying these... That design is smarter than a small motor rotating the thing with a belt, but basically that is exactly the kind of fan i thought about.

Meanwhile, its patent impeller with diversion ring of T.M.D. FAN can divert the flow field to a specific direction and efficiently decrease the possibility of turbulence. It presents a 5% decrease in acoustic noise at a same flow rate comparing to traditional fan.

T.M.D. FAN reduces 75% of traditional motor hub area. This revolutionary breakthrough not only increases 30% of airflow and static pressure, but also brings the flow field directly to the centralized heat source. As a result, the thermal resistance (θ) of a Cooler with T.M.D. FAN is 8% less than with traditional axial fan

[StarfishChris]

YSTech TMD Fan
70 x 25
Rim-drive fan almost eliminates center hub and corresponding hole in airflow typical of axial fans, but is available only in high speed (5800 and 4500 RPM) models that whine or buzz. They also vibrate a lot. Even at 5V, they are not really quiet. But their special characteristics and nominal cost are useful in some applications. Dec 24/02

Unfortunately it seems the second generation is still only available as 70mm and high(er) speed, though whether it's any good at 5V is unknown. My guess is some serious decoupling will be required

[pony-tail]

If you were to use a belt drive the motor would be spinning at a very high rpm probably 10 times the fan rpm and you may have high frequency noise (like a WD raptor bearing noise)

[RonG]

This article discusses air conditioner fans, but has some ideas that might be applied to smaller ones. Specifically, they talk about using foam to close the gap between the shroud and the blades of the fan, and letting the fan blades wear out a groove in the foam if there is some overlap. They also write about extending the shroud ( approximately one diameter ) and putting a cone over the center hub.

http://www.fsec.ucf.edu/bldg/baihp/pubs/develop/

[PS]

Neil-

One thing I'm still having trouble figuring out about your idea is the "swept back" thing. You see, the normal "swept forward" design makes sense to me. But if you imagine a swept back design, that would bring the angle of tilt of the tips of the fans to be nearly parallel to their direction of motion, which would have them slicing through the air moving no air at all, right?

So I can understand doing away with the "swept forward" design, because, by the same token, it brings the angle of tilt of the tips to be nearly orthogonal to the direction of motion, which creates the undesirable turbulence.

So why not just leave the blades straight, and utilize the moving conical section in your design? You should see if you can get some samples made.

Also, it's too bad I don't have any program designed for studying fluid flow -- we could have gotten some computers analyzing these models. Nevermind I wouldn't even know how to use such a program...

-PS

[NeilBlanchard]

Hello:

Interesting article Ron. From that article:

# Fan rotation speed is a major factor in sound propagation
dB 1 = dB 2 + 50log 10 (rpm 1/rpm 2)
Thus, a fan moving at a 20% slower speed should exhibit a 5 dB drop in sound level
# Other factors:
- Vortex shedding: turbulent eddies in the wake of the fan blade tips
- Turbulence due to the obstructions in the intake or exhaust wake
- Fan motor vibration
- Harmonic resonance associated with the number of blades
- Fan interaction with compressor noise

The first point is of particular interest, of course -- and it points in the direction of getting more air flow from lower RPM's.

And the point that you mentioned, as well as the "vortex shedding" both seem to indicate that the captured blade ends is a good thing.

As for why any sweep is needed -- from what I can glean from quiet fans fans that I've seen, is that the leading edge needs to have a gentle angle of attack (to minimize turbulence above the top surface of the blade), and then the surface needs to curve to accelerate the air. The angle of the trailing edge is what is going to determine the velocity of the air flow.

What the sweep does is it allows this acceleration to be less abrupt -- since the length of the curve (of the section, or airfoil) of the blade is longer that is possible with a straight blade (with no sweep), the acceleration can either take longer and/or get to a greater angle at the trailing edge. (I'll try and draw this in DataCAD later on, after I boot to Windows; I'm running Linspire right now.)

The number of blades is important -- I think there is a big benefit to having an odd number of blades. This seems to be borne out by the fact that almost all fans have 3, 5, 7, or 9 blades. The only exception I can thing of is the AeroCool with 16 blades.

Why does this matter? My best guess is that an even number would cause the harmonics of the sound produced by each individual blade to sum up and increase the sound pressure level -- because the even harmonics are going to reinforce each other. I.E. the peaks of the second harmonic will line up with of every other peak of the fourth harmonic. The odd harmonics are going to be out of phase with each other: the peaks of the first harmonic lines up with the trough of the third, and therefore it subtracts from the sound, rather than reinforcing it.

[Edit] I was thinking more on this: it may have more to do with the RPM and the harmonics produced by each blade. If there are an even number of blades, and the fan turns at 1000RPM, then the pitch of the fan (1kHz) will have strong pitches at the even multiples: 2kHz, 4kHz, 8kHz, etc., to be emphasized, and these are the same frequencies as the harmonics. If there are an odd number of blades, then the frequencies are distributed over a greater number of nodes. [/Edit]

Now I could be all wet on this! But I think, at least that there is some good reason to have an odd number of blades. It also seems that the larger the fan, the more blades it can have and still be practical, and so it can provide a larger surface area actually doing the work of pushing air.

Lower RPM fans can have more blades and more surface area. High RPM fans need to have fewer blades in order to avoid "cavitation" when air cannot move fast enough to move into the low pressure areas after each blade passes by...

[PS]

I also read that article... it was indeed interesting, but I only understood some fraction of it. But yeah it would be interesting to see if I could make some of the adjustments they talked about in the article on just any ol' fan in my computer. The adjustments that I think any of us could make would be the conical hub thing and the foam around the inside edge of a fan. In fact -- after finals, maybe I'll try to do some sort of adjustments with some louder fans and see what happens.

But also that thing about asymmetrically spaced fan blades was also interesting, they were talking about having asymmetric blades so that harmonics wouldn't even be created, right?.

As for numbers of blades vs. surface area vs. rpms, I don't think that it's really quite that simple. There could be fewer blades, but if each blade is significantly wide, you will still have a good amount of surface area. I've read that higher numbers of blades on a fan make a fan create higher pressure and perform better in restrictive environments. I'd imagine that this is because you have air accelerating off of the fan blades, and when you have more fan blades, you would have more of these flows of accelerated air, and so it's more likely that one of these flows will find its way through the restrictions.

Neil, I was having trouble understanding what you meant by "forward-swept". I finally realized what you meant when I was reading the most recent SPCR review of the Antec 500W Phantom, and I was looking at pictures of the fan inside.

Before I was talking about the curve of the fan blade in a different dimension (like if you were looking at a blade's smallest edge; that is, if you were looking down the lenght of the blade, you know how it's concave?). So I thought you were suggesting that they make that curve convex instead, which wouldn't make sense to me, as I pointed out earlier. In fact, I would either still argue that this should be straight, or that this specific curve (don't know what it's "technically" called) should not only remain concave, but should also follow the path of a cycloid (it may actually already be standard to be cycloidal).

But as far as the sweep goes... I'm still having trouble imagining why it would make a difference. As far as quiet fans go, did you see the post awhile back about quiet AC Fans? Did you see the shape of the blades of that Sunon? I personally haven't seen any DC fans like it. Nevermind that I haven't seen many fans to begin with. So maybe a symmetric sweep would be worth thinking about as well?

[Edited for spelling]

[NeilBlanchard]

Hello:

As for numbers of blades vs. surface area vs. rpms, I don't think that it's really quite that simple. There could be fewer blades, but if each blade is significantly wide, you will still have a good amount of surface area. I've read that higher numbers of blades on a fan make a fan create higher pressure and perform better in restrictive environments. I'd imagine that this is because you have air accelerating off of the fan blades, and when you have more fan blades, you would have more of these flows of accelerated air, and so it's more likely that one of these flows will find its way through the restrictions.

[snip]

But as far as the sweep goes... I'm still having trouble imagining why it would make a difference.

Well, if you have just three, broad blades, then the angle of each is shallow, and you won't get as much velocity for a given RPM. If you have more blades (5, 7, 9) then the area can be kept the same, and with each increase, the angle of each blade is increased, as well, so the velocity can be increased for a given RPM. Up to a point, of course. Too many blades doesn't have enough space for the air to pass through well...

As for the sweep affecting all this: as the fan rotates, the length of the section on each blade is longer for a swept blade than it is for a straight blade. And yes, a general concavity is a good thing; though the curve that works better may be closer to an ellipse or a parabola, than an arc.

I still need to try and draw some things to show what I am referring to...

[bobo5195]

I’m all a bit new to this really and I still haven’t got most of this thread but from my knowledge of fluid dynamics.

Small and big fans have a number of differences fluid dynamically. Turbulence which is going to be something very important in this analysis is quantified by the Reynolds number. Re = rho * v * L / mu. rho being velocity and mu being viscosity of the fluid. Now it is quite obvious that velocity and length change with fan size. So a smaller fan will probably spin faster for its size (V is not linearly proportional with L) than a large air compressor fan. So analysis of large fans might not be so useful in this case as there are important differences.
[for those who don’t know a high Reynolds number means turbulent flow. Re of the order of 10 to the power 4 or 5 implies turbulent flow. Its not a hard and fast rule but its reasonably accurate. The Reynolds number measures fluid inertia; speed compared to the frictional effects of viscosity; mu. At high speeds the air can become turbulent and wash all over the place because it has enough inertia to overcome viscous effects. Also V and L for this kind of case are arbitrary length and velocity scales. They are whatever you feel like calling them.]

The second thing is getting a computer on this seems simple but it is not. Turbulent flow modeling is hard. Quantum physics level hard. Simply put you cannot just model turbulent flow in an FEAA program because the scales are too small. There is not enough computing power. Approximations have to be used and these require a little bit of knowledge about how things work to get right.

The programs to do this kind of things are expensive. I have access to them and might have a look but it is not as simple as getting a fan and pressing analysis. All of them are going to take a day plus to learn for someone who knows what they are doing but they can most definitely do this kind of thing. Working out the noise (wind noise not mechanical) from it is a bit more complicated but should be do able.

[NeilBlanchard]

Hello:

You are way out in front of me! It sounds like it is probably easier to do so R&D prototypes and just test them -- 3D prototyping is not a far-fetched thing, these days. Those 3D "printers" that can produce full-size plastic objects of amazing detail:

http://www.zcorp.com/products/printers.asp


Model it, make an STL file, "print" it, and test it! BTW, these guys will produce 3D items for you -- you don't have to buy a "printer"...

[StarfishChris]

Be wary - they may sell those printers cheap, but they'll really hit you hard with the in-er... plastic granules?

Just be careful if you print one of those fans out, though. If you do it as a single object, it won't spin that well...
(You could even stick one of your slot car motors in it for testing though it'll be more Delta than Nexus if they're anything like the ones we use)

[bobo5195]

I was having a look at getting some rapid prototyping done today actually. Those cheapy 3d inkjet style printers arent so good for what you are after. They produce very weak material especially if your after fans that wont smash if you put your finger in them. the finish is not that good either. Seems a little expensive and it might not produce a big enough object.

I was quoted $250 for a 120mm fan prototype im guessing that this could be taken down to sub $200 for a 80mm fan made using FDM printing (ABS plastic inkjet) i think $300 - 400 for a SLS - M E T A L (yes you can rapid prototype metal) is a good bet for a fan. You can buy a SLS machine to do it but it will cost about $ 0.5 million but rapid prototyped cases :):):).

CFD software can also model accoustics in fan flow. As it happens this can be done quite easily. College even has as far as i can guess a "supercomputer" to play on for this kinda stuff (unix remote login for one of the programs, im guessing an ibm power box of some sort). This thing has been done on ansys which i have acess to. I think you could pick up something to model what you are after for about $1000 for the program.