Author Options:

3D printed vortex tube for spot cooling Answered

These tubes are available as industrial supplies by the dozen but quite costly.
Tried a few designs available on the net but either they are next to impossible to print using FDM or fail to provide any significant temperature difference.
Just to be clear: If amient air temp is 25°C then I would like to get below zero on the cold end.
And well, I would like to be able to salvage a small fridge compressor with considerably less air flow than a shop compressor.

Now to the problem(s) I face:
1: The temp on the cold end is affected by both the length of the tube and by how much air can escape on the hot side.
Makes finding a suitable diameter and length quite tricky if you don't have fancy simulation software at your disposal :(
2: The standard cylindrical design for the "engine" is not really that efficient :(
To make them work a lot of pressure is required, usually they won't do anything below 85PSI and most require about as much air volume as a little air grinder.
3: 3D printed parts with complex insides are a true pain to clean, seal and make sturdy enough to last more than a few minutes under these conditions.

I am currently experimenting with PETG as it is a bit more temp resistant and stronger.
My understanding of these vortex tubes is that the outer vortex, created by the engine, travels up the tube.
At the outlet valve the vortext is split so that the fastest and hottest air escapes while the colder air is directed dow, through the outer vortex and at a significantly lower speed for the rotation.
Scientists can't really agree on the directions and spin of the inner and outer vortex :(
My theory is that the hot air is created by a pressure increase towards the end of the tube.
As this also expands the air, the inner vortex will be choked off so to say.
Only option for it is to go through and by doing so loosing a lot of pressure on the exit - the stream cools down.
The circulation around the vortex spin however means that speeds of well over the speed of sound are reached.
I can only assume that at this speed there will also be some energy exchange happening based on the static electricity created where the two stream "rub" against each other.
So far only models with a finely sanded inner tube actually performed at all.
Will one day have to try using a thick plastic straw as a liner...

1. Do you think a complex, conical nozzle is required?
The commercial ones all use them.
However I tend to think that a slotted approach might be better.
Instead of adjusting the gap of the needle valve two half cylinders with slots can create a finely adjustable outlet.
Way easier to desing if all parts shall be printable....
2. If 1 is not a a real requirement:
How to best design the slots for the correct air flow??
In the direction of the rotation?
Or even against the rotation so the air will be forced out onlyby the pressure but not by direction?
3. Is there any easy workaround for Sketchup that would allow me to create sphere with the required channels along the inner wall to create a fast vortex?
I can do screw, nut and bolts, a sphere is no problem either.
But I can't seem to find a way to comine both techniques so they work on the inside of a sphere :(

Why a sphere you wonder? ;)
I like harmony for these things a an egg or a sphere is way more harmonic than a cylinder.
On top of that it would allow to create a vortex to my specs instead of chance.
Ideally the "threads" would get a smaller pitch with every turn.
This forces the airstream to not olny increase in speed (decreasing diameter) but also to be "compressed" before even entering the tube section.
The inner stream would pass through a tube inside this sphere, after passing through the tiny gap the upgoing airstream created.
My hope is that such a design would allow for the low flow rates of a fridge compressor while being quite small in size as well.

Anyone with good ideas or tips?


The forums are retiring in 2021 and are now closed for new topics and comments.

12 months ago

I made some nice progress in terms of the possible cold end temp.
My last test tube reached minus 3°C with a supply of 22°C.
Only problem still is the massive amount of air required to run the thing :(
My compressor is claimed to be good for 120l per minute - free air delivery however seems to be more in the region of 80 liter per minute max.
You get what you pay for when you decide on the usual hobby level workshop compressor ;)

The relation between length, nozzle and outlet diameters made me change the design.
Instead of the 14mm heavy duty aluminium tube I used so far I will now go for a standard 10mm pipe with 1mm wall thickness.
After all, once finnished and working as I need it to the thing should be made available for everyone...
So far I have only one really working way to decrease the required air amount.
That is by using inlet nozzle sizes that provide a fast air flow while restricting is good enough to keep some pressure.
Problem is that normal FDM printer would need a 0.1mm nozzle to print fine enough for this.
And even then the required cleaning is next to impossible due to the tiny gaps that no longer allow sand paper to pass through.
It took me quite a few design changes until I found the key to the problem.
Go modular ;)
If the swirl / vortex generator has too fine details than simply make it two parts to join later on.
Makes the gaps twice as big and again allows sand paper to be usable.
Oh, how I wish I had a decent resin printer now...

Another problem I encountered was pressure...
8 bar does not sound like much but if your printer does not print properly it is not nice to have a part failing at tese pressures.
To safe a lot of time I reverted back to using UV curing glue to get the parts together.
Sadly I am running out of UV transparent PLA LOL
Anyway, the original desing got quite a bit bigger with the insert option.
Makes it a bit harder to create a pressure safe vortex chamber.
If in doubt make it thicker and use more perimeters LOL

I am getting confident that I can reach my goal of 30° temp difference from the supply temp.
If I can solve the problem of printing a vertical thread properly without the need for support I might even get a bit more out if.
Right now the difference on how plastic and metal expand with temp limit the prototyping a bit.
But once I have the basics solved it would be no problem to use silicone and some set screws in the final design.


1 year ago

Volume vs pressure...
I now learned many ways to desing vortex tubes that simply won't work in my desired parameters.
Most however produce either some nice cold air output or some exteme heat on the hot side.
I call it a prove of concept only for the time being.

However, in my experiments with different motor shapes and slot arrangements I noticed something really odd.
A few of my tubes actually suck in air badly from the cold end!
And I don't mean a bit, I mean more like te whole thing is some mental venturi nozzle.
Funny thing is that those few also use way less air from the compressor - despite having simial hole and channel sizes as the other models.
There seems to be a destinct relation between the tube diameter and gap between engine outlets and the tube for the inner vortex.
Especially if the design provides for really fast spinning motion.

Theory so far:
If the outer vortex has enough free space in the lower section the tube then it creates a low pressure zone in this area.
In commercial designs for the engine insert it seem the tube for the inner vortex always has the same outer dimensions and lenght - at least judging by the images I could find online.
Only things that differ depending on cooling capacity / temp difference are the diameter on the hole for the inner vortex and the size for the slots in the engine.
Key seems to be to have the input opening for the inner vortex right at the height where there is still a positive pressure from the upgoing vortex.
However, in my theory I think this approach is counterproductive as it won't really address the problem of too much air being used.
If the upgoing vortex could be created with the intention to have a strong low pressure zone in the lower end it could be benefitial.
Instead of wasting the hot air from the upgoing stream it could be diverted back into the low pressure zone.
Would mean we could control the air volume and amount of hot air simply by changing the pressure and air volume from our compressor.
Problem is that we need a double walled tube for this.
Bigger problem is that the ring gap on the outside can't cause a pressure difference.
What goes in hot air must equal what the low pressure zone is able to take in - without the need to suck it in...
Why do I think this should work much better?
1. : We really only want the hot air not any of the colder air in the vortex.
Means whatever is left will be used to produce only cold air for the cold end output.
2. : The hot air entering the low pressure zone will be subject to getting much colder.
3. : We can deal with the produced heat by means of heat sinks/cooling fins or if in doubt with some heat pipe design and a fan.
Means what goes in only comes out as cold air.
Also means way less air volume is required, making the system much more efficient.

Why do I think it will be impossible?
I only know that Solidworks should be able to simulate such complex air flow things - but I don't have it and neither the money nor the required time.
Trial and error only get so far until you run out of filament...

What options are left to explore?
Efficiency of course.
It is clear that there is a minimum requirement for pressure and volume for these commercial tubes.
And there seems to be no problem in the industry is these system require this.
Without a nice SLA printer I am limited to what a FDM machine can produce in terms of details and structures that would otherwise require support.
Trying to produce consistent channels or outlets below the 1mm range is painfully prone to errors.
The slightest change in the filament, a bit of drooping or spiderweb thread from the nozzle all cause troubles when it comes to printing finer than the machine is rated for.
Try to print a single perimeter cylinder with a 0.4mm nozzle that only has a thickness of 0.2mm and you know how hard things can be...
And of course with FMD you won't get the smooth finnish required and can't just use sandpaper in a twisted channel of just 2mm in diameter.
As said, I learned a lot of ways that just won't work with filament.
But if we check simple things like a 3D printed whistle is is clear that there are clever design options out there to produce a powerful air stream with little air volume and pressure that goes in.
Will have to warm up the remaining brain cells one day to figure out how to combine a whistle with a vortex tube without the noise.

Pressure increase by clever design?
If you limit for example a tube diameter then a compressor won't get enough air through this tube to cause the air tank to run empty.
With that also comes the fact that the pressure stays high and that whatever comes outof the tube does so with quite some force.
Going back to my Repulsine project I realised that it should be possible to create a inlet channel system that creates multiple, rotating streams from a single input.
Imagine a flat tri-filar pancake coil...
Now stretch it a little bit and it almost looks like a snail housing with three spins instead of just one.
The required pressure comes from the gap size between outlets and tube for the inner vortex.
No complicated engine desings anymore, just tubes with different wall thickness - or inserts to go inside the engine.
If you check a section cut of such a design it would show that the outlets are actually rather huge in terms of their area.
Instead of about 1mm² they would be almost 4mm².
Ever seen how easy you can lift heavy things with a littler blower and a plate creating an air cushion? ;)
Only problem is that our air cushion is forced to go around a tube and up through it.
With the small gap it would do so quite quickly... ;)
My struggle with that at this point is to make it printable so it can last long enough before blowing apart or starting to leak through cracking layers.
And I might have to use something better than printed plastic for the actual tube.
Once it works good it just gets too hot for normal filaments, even PETG :(

Jack A Lopez
Jack A Lopez

1 year ago

Do you remember hurricane balls?

It was this toy made from two ball bearings joined together, and put in motion by a stream of moving air.

I recall wondering if that toy could be made into a practical turboexpander,


a machine that intentionally removes energy from moving gas, by way of the gas doing work on the turbine.

Also the spiral-shaped countercurrent, tube inside a tube, heat exchanger seen in a typical Hampson-Linde cycle refrigerator,


has some superficial similarity to the inner and outer gas streams in a Ranque-Hilsch. Basically both are a streams of gas, moving in opposite directions, with one giving heat to the other.

I guess what I am saying, is that maybe the Ranque-Hilsch tube is too simple, and maybe it could be improved by adding some moving parts (making it more like a turboexpander) or by adding a wall to separate the spiral layers of hot gas and cold gas (making it more like a countercurrent heat exchanger).

Actually I am not sure about that last suggestion of adding a wall, to make the gas streams stay in their own lanes. In fact I am kind of confused about why the cold-air stream is giving up heat to the hot-air stream. That by itself seems like voodoo.


Reply 1 year ago

I've read about these expanders before but don't think I can utilise their features with just compressed air.
And the H-L cycle is actually the grandfather of your normal refrigeration compressor systems.
Both work really well with things like Propane as you only need moderate pressures to change it from a gas into the liquid stage.
Air is a mix of gases that require more pressure than any of my compressors can handle to liquify.

But I see once more that we often think in similar directions ;)
I experimented a bit with a spining "vortex tube" inside my tube.
Basically just a simple and stretched spiral running on cone shaped "bearings".
Idea was to use the upgoing vortex to make it spin so the inner and downgoing vortex would get some extra compression.
Sadly it did not work out as planned as my compressor won't deliver enough airflow at the required pressure to do proper tests.
And of course I highly doubt it work without some serious finetuning LOL
After all, the shape must fit the vortex stream so the thing only sping with the friction of the air moving through it but not by the airflow like a turbine or fan would.

My main goal still is to develop a working cooling tube that does not require these insane air volumes...