Distributing airflow evenly

  • B
  • Thread starter Grey Area
  • Start date
  • #1
15
1

Main Question or Discussion Point

HI all. A bit out of my depth here...degree is in chemistry, but that's not helping me with the real world problem I'm trying to solve.

I have a single air inlet that's being used as a cooling stream. It is channelled by a duct (or ducts, there are a few available designs) onto a piece of work - but mostly from a single direction. There is a school of thought that distributing it evenly from more directions (for example eight, or at least four cardinal compass points) would provide better results

There are some simple "distribution" devices consisting of a ring with holes on the internal face, but since the air in these designs mostly enters radially (I think that's the right word...from rim to centre along the radius) it tends to exit preferentially from the holes nearest the inlet (at least, according to my "wet finger" test it seems to - the stream from holes on the opposite side seems "weaker"...if science would indicate otherwise, please advise.

There are some restrictions when trying to solve this problem...it's impossible for the air to enter axially, but possible to have it enter radially or tangentially
It's on a very small scale...the ducting system has to operate in a cylindrical volume of about 25x80x80 millimetres - and there are other things like the actual work area that impinge on that volume.
The cooling needs to cool the workpiece, but not the hot tool doing it!
(It's for 3D printing)

Questions I cannot answer...
Would having the air enter tangentially have any effect? Some people I've discussed it with keep telling me air doesn't have a "memory"...but if I turn on a blower, the air goes in front of it...not behind and I know it's a mistake to think of the duct as "empty" before the blower starts...the air is already in there, so it gets pushed out almost immediately (save a bit of compression perhaps) by more air coming in...

Does it depend if the sum of the cross sectional area of the vents is equal, lesser or greater than the cx area of the input? I'm assuming that under constant air flow some equilibrium is reached - is it possible that the ring would achieve a constant pressure which would result in all the vents achieving the same flow?

I've been toying with all sorts of designs with varying sizes of holes, even with a moving carriage in side that spins (though I don't know if it actually wouild) to block some holes and thus give others a fair "chance"...but I realised I don't know enough about what's going on to know how (or even if it's necessary to) achieve an even distribution...so thought I'd find someone to ask.

Any ideas? I HAVE Googled this, but the language used in anything that seems relevant seems to start at a point somewhere higher than I can currently grasp...

Thanks in advance!
 

Answers and Replies

  • #2
jrmichler
Science Advisor
1,026
968
A diagram or two would be very helpful here. A general rule is that the amount / rate of cooling is proportional to the velocity of the cooling air at the surface of the part and also to the temperature difference between the part surface and the cooling air.

If you are blowing air on one side of a hot part, you are cooling that side of the part, and depending on conduction through the part to cool the other side. The cooling will be slower than blowing air on both sides of the part.
 
  • #3
Tom.G
Science Advisor
3,063
1,813
Some people I've discussed it with keep telling me air doesn't have a "memory"...
But like anything physical it does have inertia. Quoting Merriam-Webster dictionary: a property of matter by which it remains at rest or in uniform motion in the same straight line unless acted upon by some external force.

But that doesn't address your problem of unequal flow from the holes. To a first approximation, the total area of the holes will have to be less than the cross sectional area of any part of the ducting. (although you can make up for that to some extent with a higher source air pressure)

One way to think of it is if you have a window screen laying down with a tilt from one end to the other. If you pour a small amount of water on the screen, it will flow right thru where it is poured. If you pour a large volume of water on it, the water can't get thru the nearby holes before more water arrives; it spreads out to other holes to get thru. The same thing happens in your cooling ring. You must have enough pressure to force the needed amount of air thru the holes, and enough fan capacity to maintain the needed volume at the needed pressure.

Hope that helps clarify things for you a bit. There are others here (@Chestermiller comes to mind) that can come up with the math necessary to supply you with actual numbers, providing you know how much thermal input you have and the relevant temperatures.

Cheers,
Tom
 
  • #4
15
1
Thank you for the advice.

As you've describe it is kind of as I imagined it...the "air memory" thing I thought was particularly in evidence in things like tornados and the vortex cannons that area favourite of science shows in America; an equilibrium state that persists...with of course other forces acting on it to maintain and eventually degrade it.

Glad that I was right about the importance of cross sectional area and pressure too...once the "distribution device" is fully pressurized, if all the holes are the same size, they'll all experience the same flow..so in point of fact if the "(sum of outlets cx area) < inlet cx area" is met, the current "ring" devices will deliver the air equally.

Diagrams...never my forte...but here are two of the items that are actually used...the straight one;

https://makerware.thingiverse.com/thing:2086191

and the "ring" one

https://www.thingiverse.com/thing:1823404

I've been having fun with some of the ducts, smoke matches and my 5mW laser behind a Fresnel lense...shows up the flow quite nicely...but the videos I took are a bit rubbish...I'll post back here if I get anything better...

oh and thanks jrmichler for the info on cooling efficiency...particular about the velocity of the cooling air...I assume there's some optimum point where both volume of air and velocity of air contribute to the greatest cooling? GO either side and either the low velocity overcomes the effect of the higher volume or the low volume overcomes the effect of higher velocity?

One thing a lot of us in the 3D printing community have also seen is the effect of colour; white and black are completely different animals...black needs the fan full on too cool sufficiently, on white you can turn it down (NEED to turn it down) to less than half the speed.

Thanks again!
 
  • #5
15
1
One more question, sorry..

I know it's commonplace to "rifle" taps with a spinning pattern to "focus" the stream of water...is there any similar principle that could be applied to cooling air? I have read about turbulent flow in heating systems, but that seems to be specific to material that remains within a pipe rather than being expelled out of one. Thanks again.
 
  • #6
jrmichler
Science Advisor
1,026
968
.I assume there's some optimum point where both volume of air and velocity of air contribute to the greatest cooling? GO either side and either the low velocity overcomes the effect of the higher volume or the low volume overcomes the effect of higher velocity?
This is correct. A small volume of air at high velocity will be a needle jet that only cools a tiny area. A large volume of air at low velocity will mostly flow around the part. Somewhere in between will be the best cooling. Try to hit as much of the part as possible with air moving as fast as possible without blasting the part into the next room.

A general rule of thumb for designing the air distribution nozzle is make the cross sectional area of the inlet as large as practical, then shrink the distribution holes until they flow more or less equally. There is a broad optimum, so the thumb test is plenty good enough. Stick your thumb in there, and if feels like all holes are flowing more or less the same, you are good to go.
 
  • #7
15
1
Thank you jrmichler. In my designs I was thinking of using broad "slots" to maximise velocity but at least cover a wide surface area. The inlet of the fan is 175 square millimetres, so if I split it into four vents each 40 millimetres (so 160 total to build up some pressure) they could be something like 2mm x 20mm...,the low profile would help in the confined volume too.

Though I was also thinking of one or maybe two larger vents to compare efficiency, which is where my idea of some sort of "rifling" of any vent came in...but I don't know if it would achieve anything.

Mixing tactics a bit, but does air experience diffraction as it goes through a grating? That could be another way to "spread" the flow, thought the individual slots might have to be narrower than a 3D printer could print (all these bits are made for the printers, by the printers...it's very incestuous!)

Thanks once more. If I manage to produce something that people regard as successful, I'll post details back here.
 
  • #8
19,920
4,095
HI all. A bit out of my depth here...degree is in chemistry, but that's not helping me with the real world problem I'm trying to solve.

I have a single air inlet that's being used as a cooling stream. It is channelled by a duct (or ducts, there are a few available designs) onto a piece of work - but mostly from a single direction. There is a school of thought that distributing it evenly from more directions (for example eight, or at least four cardinal compass points) would provide better results

There are some simple "distribution" devices consisting of a ring with holes on the internal face, but since the air in these designs mostly enters radially (I think that's the right word...from rim to centre along the radius) it tends to exit preferentially from the holes nearest the inlet (at least, according to my "wet finger" test it seems to - the stream from holes on the opposite side seems "weaker"...if science would indicate otherwise, please advise.

There are some restrictions when trying to solve this problem...it's impossible for the air to enter axially, but possible to have it enter radially or tangentially
It's on a very small scale...the ducting system has to operate in a cylindrical volume of about 25x80x80 millimetres - and there are other things like the actual work area that impinge on that volume.
The cooling needs to cool the workpiece, but not the hot tool doing it!
(It's for 3D printing)

Questions I cannot answer...
Would having the air enter tangentially have any effect? Some people I've discussed it with keep telling me air doesn't have a "memory"...but if I turn on a blower, the air goes in front of it...not behind and I know it's a mistake to think of the duct as "empty" before the blower starts...the air is already in there, so it gets pushed out almost immediately (save a bit of compression perhaps) by more air coming in...

Does it depend if the sum of the cross sectional area of the vents is equal, lesser or greater than the cx area of the input? I'm assuming that under constant air flow some equilibrium is reached - is it possible that the ring would achieve a constant pressure which would result in all the vents achieving the same flow?

I've been toying with all sorts of designs with varying sizes of holes, even with a moving carriage in side that spins (though I don't know if it actually wouild) to block some holes and thus give others a fair "chance"...but I realised I don't know enough about what's going on to know how (or even if it's necessary to) achieve an even distribution...so thought I'd find someone to ask.

Any ideas? I HAVE Googled this, but the language used in anything that seems relevant seems to start at a point somewhere higher than I can currently grasp...

Thanks in advance!
I'm having trouble visualizing this. Any chance of providing a schematic diagram?
 
  • #9
russ_watters
Mentor
19,425
5,563
I'm having trouble visualizing this too -- and unless I missed it, I didn't see anything about how the air gets out.

I'm an HVAC engineer, by the way...
 
  • #10
15
1
In the few minutes while I create it, prepare yourselves for the worst "technical" diagram you have ever seen...back soon...

Actually I might be able to spare you that fate...it's still my file but this is a (reasonably) accurate mockup of the working area of our 3D printers...in this specific case, the ANet A8.

Pointy bit with the red wires going into it is the heated block - gets up to around 200 degrees centigrade, melts the plastic which is fed in through a tiny hole in the top (0.4mm diamter)

The black box is the air inlet, fed by a (I think) 24v blower fan - takes in air axially and expels it downwards where it enters the duct, which is the "design target" of this post. The duct is designed to cool the workpiece (plastic) more rapidly, thus setting it before it flows out losing detail. Prints that are cooled well are sharply defined, prints that are not tend to be a bit "wobbly" looking.

The duct pictured is a circular one...it was my opinion that it would put more air out through the holes immediately in front of the inlet, and that less would come out "round the back"...and yes, ideally the nozzle should be in the centre...people tend to modify their printers so not all fan designs "fit" perfectly without a bit of tweaking...

The original duct as supplied with the printer is a simple ducks bill type affair...though there have been several variations since...I may add a few to the 3d workspace if I can find time...(edit:: there are now a couple on the left)...

The whole assembly pictured raises up and down a layer at a time. Depending on the complexity of what is being printed, the fan could either by effectively sitting on a "plane" of plastic, or in mid air with only a very small surface in front of it under the nozzle.

Because of this I'm starting to think there is no "perfect" fan duct solution...that would require selecting a different duct depending on the ratio of "plastic to air" under the nozzle at any given time, which is impractical. But is there a design that can be the best "average" performer, regardless of whether it is sitting on mostly plastic or mostly air?

Remember the "trick" is to cool the work, not the nozzle...so I feel improved directional control may be the way to go, but I'm not skilled in this, I just have a tendency to look at things and think..."what if"?...

Such as...

What if the holes in the ring design were replace with a single circular vent, angled down at "x" degrees?
What is the performance difference between a 2mmx18mm vent and a 6x6mm vent (if any)?
Would "rifling" the vent narrow the focus of the air - and is that beneficial?
There are designs where the air is split between two "letterbox" vents at 180 degrees to each other (though facing slight downwards). What effect is the mixing of the two streams having on cooling? DOes it slow the air down and thus reduce efficiency?Would it be better if they were angle to "just miss" each other?
And many others...I'll stop now though or I start to sound like a 5 year old constantly asking "Why?"...and people lose patience :)

Here's the link to the 3D representation of the workplace and duct...it's a copy and I believe you can edit it from this link, so feel free to have a bit of fun if you're so inclined...

https://www.tinkercad.com/things/f8wulZEZHYU-copy-of-anet-a8-nozzle-and-fan/editv2?sharecode=HIDb51VJnTISx_9QCADXcBZ7lRRi2skFNAFHX3C1M6I=
 
Last edited:
  • #11
15
1
The good thing about HAIVNG a 3D printer is you could provide some science-backed suggestions for improvement, I can (probably) turn it into a working prototype within a couple of days (I'm not the fastest, but I'll not burden you with why at this stage) so I could include your suggestions in with my own testing (which is more based on "whacky idea, looks cool, but probably doesn't work"...
 
  • #12
15
1
All gone very quiet...can anyone access the link? I know some people have problems. I can produce a (poor quality) diagram instead if it helps...
 
  • #13
russ_watters
Mentor
19,425
5,563
All gone very quiet...can anyone access the link?.
No, you have to sign up to view it. Can you post an image or pdf?
 
  • #14
15
1
No, you have to sign up to view it. Can you post an image or pdf?
Sorry Russ, didn't realise...I'll get on with an image straight away...
 
  • #15
15
1
OK took some screenshots of the tinkercad page...hope they help..

The setup;
4hjpnqC.jpg


The other design options;
JUL9rGp.jpg
 

Attachments

Related Threads on Distributing airflow evenly

Replies
3
Views
858
  • Last Post
Replies
2
Views
3K
  • Last Post
Replies
3
Views
3K
  • Last Post
Replies
2
Views
2K
  • Last Post
Replies
4
Views
12K
Replies
0
Views
7K
  • Last Post
Replies
14
Views
5K
  • Last Post
Replies
7
Views
13K
Replies
8
Views
11K
  • Last Post
Replies
6
Views
1K
Top