Building a Vortex Tube

What are you trying to achieve? ...and for what purpose?...We'll need a schematic of your design for analysis....

http://www.google.com/patents/US3173273

Really interesting to see that patent! I was not aware of it when I made and tested several Hilsch vortex tubes the same year (1962) that patent was filed, for my senior college research project. (of course I was not aware of it until this morning, when I saw your post!)

My biggest problem was the high noise level...

The Hilsch vortex tube was discussed many years ago in Scientific American magazine's column "The Amateur Scientist"; two current manufacturers are Exair and Vortec. They are used for cooling electrical panels and personnel suits.

In the early 90s, Collins Radio (now RCI) designed railroad electronics systems. The initial design was going to use these tubes as a cooling method. But that was going to use 80HP~60KW of engine energy, and this idea was thrown out (by the first customer). Then they thought of thermo electric coolers, again once they figured out they were .1 COP they opted for mechanical refer units. Ended up they didn't know how to manage this product line so it was sold to WABCO.

Thanks for answers so far.

Solar Eagle - I don't have a particular purpose other than to do it. My design is basically similar but not as sophisticated as your diagram and my forcing cone in the vortex chamber goes the other way - food for thought - thanks for that - prototype 5 coming up.

Tornado - I know what they are used for and have studied all that I can find on the ones from the manufacturers that you mention. Exair have an agent in Australia, but at almost AUD $400 for the smallest of them, I choose to leave them alone. Thanks anyway.

Ignator - Thanks, I know that there are more energy efficient methods to produce both hot and cold, I just want to see if I can get the results that are claimed.

Nick name - Thanks, I have previously studied that paper until I have become crosseyed, unfortunately it's long on theory but gives no real information on building one.

I think - I did not yet read the paper - that 2 parameters are important:

- input energy = supply pressure and flow ( if over critical nozzle diameter)

- ratio between chamber and tube diameters in order to obtain as high as possible rotation speed.

May be you do not have the right values thus the low separation effect.

Theory coukd help find the right proportions.

Thank you again for your comment.

From memory, that paper gives no nozzle diameters or any other dimensions - so what's the critical diameter?

It does talk about the ratio between chamber and tube diameters, but (again from memory) they were very wide parameters - something like anything from 20:1 to 45:1 as having no effect on performance.

All of my prototypes have fallen inside those margins - the one in the photo is 33:1.

Thanks Lyn,

I have seen all of those sites, that build was my first prototype and I followed the instructions closely, it was pretty much a failure - going from memory as it was about 6 months ago it produced somewhere around a 3°C temperature drop.

None seem to give enough information on the chamber itself, which I suspect is the key.

I suspect the "key" is what makes it so difficult to duplicate and why it isn't widely known.

Maybe this will help:

Ranque-Hilsch Effect Tube ("vortex tube") - Pdbuchan.com

There's a Parts List Page and an Assembly Page.

Thanks again Lyn, that site is pretty much the crux of the problem, all of them claim that they could feel a "slight" change from ambient whereas a Vortex tube that works correctly should give you frost bite at the cold end (-50°C). I seriously doubt that unit could work very well as it has no real vortex chamber. He says "it was uncomfortable to hold at the cold end", but if it were working as should be it would have stuck to his hand. Mine feels uncomfortable to hold and it's only achieving 6°C

Rixter, thanks to you too - that was one of my early concerns as the commercial manufacturers do recommend a high tech filter system (possibly just to sell more product), I borrowed a very good filter/dryer from a spay painter friend, but it made no difference at all. I do get a lot of condensation on the outside of the cold end but as humidity here is around 90% at present, that is to be expected. The cold air coming out of the orifice is relatively dry as is evidenced by blotting paper inserted in the airstream.

Thanks again for suggestions and keep them coming - there might just be a gem among them.

Meanwhile, I'm studying Solar Eagle's offering for clues.

I would think that condensation, whether in the air stream or on the output pipe would be "stealing some of your cold" as water vapor condensing gives up heat. I don't know how significant this is in this situation.

Another thought: Peltier devices are heat-sinked on the hot side. I was wondering if a heat sink on the outer pipe hot side would reduce the temperature of the cold side.

't know if this is a GA or not, but it is a stimulating good point you raise, if nothing else! Thanks for nothing...I was preparing a budget...

Does this link and website help?

https://www.youtube.com/watch?v=Cu8X_jvpXn8

I forgot to mention that according to the first video, the formula are on his website, which posted early on on the video (keep pencil and paper handy!), that may be exactly what you need to help get yours working as you wish.

I found the video easy to follow, but I only looked at Part 1....

Let us know how you get on,no matter how long it takes....

My work with vortex tubes way back was for respiratory protection with an interest in the 'hot-end' for warming the air, and the search was not for the lowest or highest temperature separation, but more of the amount of air need to drive the tube, and the noise generated.

My simplistic view was that the enormous rotary speed had two prime effects. One was friction against the walls of the tube causing heating and the increase in pressure as well the outer wall due to centrifugal force that also causes a 'vacuum' in the middle where the loss of pressure is accompanied by a drop in temperature. .

The trick was to get the max temperature gradient, with the minimum amount of air going in, subject to enough air coming out to drive the respiratory protective device.

My work was not related to design of the vortex tube as such, but playing with pressures and flows, to drive the particular make of tube under test, in order to predict the working temperatures.

The conclusion i reached is that the supply hoses must be large enough to cope with the max flow, and the supply pressure must be enough to provide this flow when necessary, and in conditions of lesser demand, suitable regulators to adjust the pressure.

I have seen nothing since in terms of theory to prove or disprove the my assumptions on the cause of the temperature difference. I will leave the thermodynamic experts to juggle the formulas.

If my simple theory is correct, the max temp difference could occur at an optimum point. Increasing the speed of the vortex (the centrifugal force) to cause the greatest pressure differential and thus a greater temperature differential might be self defeating because the extra heat generated at the wall by friction at higher surface speed could conduct to the centre therefore contaminate the cold air taken for cooling.

You might find the diameters of the tube (and length), are critical and might need insulating against heat transfer (in or out).

Good luck with your search.

My experiments when I did my research project over 50 years ago lead to two basic conclusions:

1. Injected air molecules have a Gaussian distribution of velocities; the higher velocity (hotter) ones will experience a greater "centrifugal force", so they will migrate to the outside, while the slower (cooler) ones will be pushed to the center.

2. If two air molecules or air streams having exactly the same linear velocity are injected at slightly different distances from center, they will have slightly different angular velocities, with the one closer to the center having a higher angular velocity. Friction (collisions between molecules) will tend to bring the angular velocities closer to each other. Those layers of air near the center will have their velocities reduced (cooled), while those layers of air near the outside will have their velocities increased (heated). A longer, smaller diameter tube should force more energy transfer between layers of air.

Unless I am mistaken, friction with the outside walls will indeed heat the tube, but heating the tube means taking energy from the air, cooling the air. In that case, I suspect it is important that the inside surfaces of the tube be accurately round and highly polished, if hot air output is desired. I did not think of that 50 years ago...

Carrying that last concept one step further, if cold air output is the principal desired outcome, as it usually is, then a rough inner surface should remove more energy from the air to the tube itself. Now we need some insulation to reduce heat transfer from the hot end of the tube to the cold end.

Thanks Horace40, your conclusions pretty much concur with those that I have studied. Small units will operate well on as little as 80psi and 8cfm.

Thanks also to dkwarner, if I didn't have a headache before...I do now.

Prototype number 5.

I shortened the tube, used the same reflector front end but blunted the cone a bit.

New rear end with a distribution chamber sealed by O-rings around the outside of the inner tube and three tangential inputs to the vortex chamber, #4 had only one input.

Results: -2.3°C/+37.6° c

Better than before, but still not good enough.

Prototype #6 is on the way.

If the tube can easily be detached from the ends, it would be very interesting to make a series of tubes of different lengths, and graph the resulting temperatures. There should be some optimum length for the tube.

I get the impression that your input air connection is perpendicular (centered). Although it is difficult to thread, if you offset that, it should increase the spin.

I didn't measure input air temp as that was not really a concern to me, and would obviously rise with increased usage.

The input air tube is tangential to the inside of the outer chamber and so is pre-directed in the same direction as the inlets to the inner chamber, and it's not too difficult to thread if done in the mill.

This pic. shows it better.

That looks off-center, but not really tangential, and perhaps not optimal.

It is tangential for some radius... Without knowing the inside dimensions, we can't tell whether its at the optimal position or not, and that is assuming we knew where optimal is... Of course the farther off center, the more difficult it becomes to thread the hole, unless you start with a square block or a casting, or can weld an addition to the cylinder.

The outer edge of the orifice is tangential to the inner curve of the chamber, it is set up, drilled and tapped in a rotary table on a mill, that's as accurate as I can get.

Perhaps this photo with the stub of the tapping drill inserted in the hole shows it more clearly.

Remember that this is the diffusion chamber, not the vortex chamber, the holes of which are also drilled tangential to the inner surfaces.

That looks better. (At least almost.)

Yeah well you have to be careful not to get into the side wall as that will create turbulence where the drill point finishes. Probably better to be a little shy of the wall than too neat.

Yep! stop short with the drill, then go in and finish with a ball end mill. If you can tilt your rotary table, try slanting those tangential injection holes toward the opposite end.

Have you experimented with different sizes of cold exit holes? A smaller hole will obviously reduce the flow, but presumably produce a lower temperature. There has to be a best compromise between flow and temperature. One possibility would be to use a thick O-ring as the orifice, with a threaded ring to compress the O-ring. If you like, I could make a drawing...

Thanks for the input.

By lining up the drill bit accurately, there's no need to do any finishing as it comes through spot on the side of the curve. The holes into the vortex chamber are only 1/16" diameter, so there's little possibility of finishing with a ball end mill.

I have tried different sized holes, the orifice is in a removable sleeve which can be exchanged for ones with differing hole sizes, I get very little difference between 1/8" and 1/4" orifices.

I tried tilting the holes 2° towards the front end on one of my prototypes and it produced less temperature drop than the straight holes, I suspect that it produced less revolutions of the air before it reached the far end.

I could be wrong, but if i understood you correctly, the hole for the jet is drilled with only a 1/16" drill, or did you mean the cold exit holes only?

If yes, that means that the whole unit is tiny.....maybe that is part of the problem? Its simply too small.....

I may have misunderstood you, if so, my apologies....

But size is sometimes important!!

Just a thought!!

Some of the commercial units available are less than 8" long and 3/8" OD tube with up to 8 jets in a 5/8" ID vortex chamber, you wouldn't get many jets around that chamber if they were any bigger than 1/16".

That last prototype with only three 1/16" jets uses in excess of 10cfm when the front nozzle is exhausting most of the air, so the jets are easily large enough to handle the air volume.

Looking at your picture I think that you have the wrong ratio inlet diameter to chamber diameter and section.

Since you have a milling machine you could optimize in order to obtain the highest rotation possible at the tube level.

Your interest is to have the most energy of input flow transformed in a rotating flow with as small as possible losses. Here are two suggestions:

The number of nozzles can be free to choose, you mentioned 8. I think that you can start with 3. In the left design the chamber is big enough to assure an almost allover constant pressure and a small swirl in order to avoid velocity changes at nozzle entry.

In the right design the chamber is reduced progressively as section in order to obtain a constant air velocity at nozzle entry.

In both solutions the pressure drop is concentrated in the nozzle so that the losses are as low as possible.

Thanks for your input nick name.

That first diagram is pretty much how my last prototype is constructed. It has 3 x 1/16" jets and I have tapered the input ends of them (with a dremel) similarly to that picture.

The second one would be considerably more complex to machine, and I doubt that the last jet would see as much air flow as the first one, so I haven't approached it from that angle. The available air supply is ample to provide more air than can escape through the jets, so pressure and flow through all 3 should be reasonably close

As for appropriate ratios, I'm sure that is the crux of it, but no-one has yet provided that precise information and it does not appear to be freely available. I have tried a number of differing ratios with varying results, none of which yield the low temperatures that are attributed to these devices.

Thanks again to you horace40.

Commercial devices work from as low as 80 psi upwards, so pressure, while important does not appear to be critical provided it is not below required level. I can easily maintain 100 psi while running the device.

Inlet air temperature will have some effect, but again, commercial devices work off normal compressors like mine. One manufacturer states that the output air temperature will vary proportionately to the input air temperature, so a few degrees should not make that much difference for my purposes.

Not sure about your theory that the average in and out air temperatures are the same as the coldest temperature out is not when the hottest is also being exhausted, and this would logically be the case if that theory were correct. Commercial devices can deliver hot air at around 120°C, but at that setting there is almost no cold air being exhausted.

Hello Spades #36

I'm not sure about my theory either. I had to pick something that made sense to me based on what I knew at the time. There does not appear to any definitive formula as yet.

Please bear in mind that I was interested in the cooling/heating effect in respect of the application where the untreated coming through the hoses to the suit was more or less at ambient temperature and this air was being breathed. If it was too cold or too hot something had to be done to produce air at a comfortable temperature. Hence (one solution) was the inclusion of a vortex tube in the airline.

Information about how much the inlet air was cooled or heated was exactly what we wanted to know. My work was connected with 'efficiency' and a desire to obtain air at the correct temperature and flow (to be breathed) using the minimum total flow possible.

Regarding the average temperature I am quite happy to learn about anomalies in my theory. If there is little or no cold air coming out when it running 'hot' then perhaps the temperature gradient is not 'straight' as I assumed, but maybe 'squared' (or even 'cubed') at the hot edge. This is where thermodynamic maths experts could help.

Our particular product was very sensitive to air flow and made it uneconomical to use when probably 80% of the pure was used to drive the vortex and was dumped to atmosphere as exhaust waste.

I wish you well with your experiments.

To Spades #23

The inlet temperature and pressure is important and so is the flow. My post #14 explains my background and where I come from on this.

It could help if you said what you want most from the vortex tube. Is it cold air, or hot air, efficiency, or economy, or weight, or cost, or whatever ??

Then it might help to do some thermodynamic energy balance calculations to point you in the right direction. These are a bit beyond me but the CR4 expert, if they read this, might be able to help.

Assuming you have a working tube, my simple view is that the inlet air splits into hot and cold over a range of temperatures and assuming there is a temperature gradient - from dead centre to the outer wall -(it might be parabolic but without gravity say a straight line for easy arithmetic) - the outlet temperature (hot or cold) will depend on the split between hot and cold on the temperature gradient. ie, the cold end average (you have measured) will be a mixture of very cold and some warm air, whereas the hot end average (you have measured) will be some very hot air mixed with some warm air.

My logic is the average temperature of all the exit air cannot be higher than the average of all the inlet air going in, and assuming adiabatic expansion (at the best) - from which the pressure drop enables the lowest average temperature to be calculated - from which the proportions of hot and cold might enable a temperature range to be calculated.

Exactly how these proportions are achieved is related to the dimensions of the tube, and inlet pressure and temperature. Hence the need to focus on priorities.

I can picture this happening in my head but my thermodynamic maths relating to expansion, momentum, velocity, friction and heat, etc, are not up to the job.

We abandoned the work ourselves because results led us to believe it could only be done by modification to the cooling suits not made by us. That would have breached the regulations and violated any warranties.

What was the temperature of the inlet air?