Morphing Air Conditioner into Autocascade System

[mytekcontrols]

Well I hope nobody busts my chops over this, since it isn't actually a PC Cooler. However I thought it would serve as a good example, or should we say slightly different approach to recycling a cheap window air conditioner into a very usable Autocascade Chilling System.

I built the basic refrigeration unit in about 2 days (this is the advantage of trying to use as much of what already exists "as is"). And although this unit was intentionally limited "by design" to only operate down to -30C, it could have just as easily been tweaked for -80C or better.

Using an Autocascade for -30C refrigeration may seem unnecessary, but there are advantages. For instance; ultimate temperature was achieved in only 10 minutes from a warm start condition. Also running pressures were very comfortable at 20 psi suction and 100 psi discharge. You could easily leave your hand on top of the compressor all day long, since it was not hot. Where as utilizing a conventional refrigeration system for -30C, would require operating the compressor suction at a very low pressure or in a vacuum with very little mass flow through the compressor, very hot compressor temperatures, and a very long cool-down period for the evaporator (not to mention hardly any load handling capability).

Now in my application I didn't have an easy way to measure heat load capacity, but judging by the 10 minute cool-down and the large mass being used for the evaporator, I would say it would be good for at least 200 watts or more.

For a complete look at what I am talking about, check out this link: Alcohol_Cold_Trap.pdf

This will give you a complete piping schematic, heat exchanger and cap tube sizing, as well as the refrigerant charge that I used (with exact amounts of each component specified).

For -80C operation, I would recommend increasing the length of the cap tubes by perhaps 3 feet (increases pressure differential and gives us a lower evaporator pressure), eliminate the R-22 and increase the R-123 to 6 ounces instead of the original 3 ounces.

Enjoy

[n00b 0f l337]

Very nice, I've been anxious to view your work. Some questions though, does using a vertical coil not trap liquid? Also, is that 1/4" in 1/2"?

[mytekcontrols]

It's 3/16"od inside of 3/8"od tubing.

n00b asked... does using a vertical coil not trap liquid?

It could if the velocity was very low, or the annular space was large, but in this situation neither is true (it was also much easier to have a vertical orientation due to the size and shape of the space available).

However, you are correct. Ideally it is best to have a horizontal stacking of the heat exchangers, just to insure that there will be no liquid hang-up problems when returning to the compressor.

[n00b 0f l337]

I've been trying to get 3/16", seems like the tighter spacing would be nice. Do you have the lowside in the 3/16? (as in the evap?)
Your gas selection is very nice though, you must have access to many many nice refrigerants.

[n00b 0f l337]

So for -80C, just r123 and r23? You think r23 will condense at all?

[mytekcontrols]

n00b asked... Do you have the lowside in the 3/16? (as in the evap?)

No I used 1/4"od for the evaporator and return line, didn't want too much of a pressure drop since the evap used a fairly good length of it. I suppose if we were talking about making an evap suitable for PC Cooling it could be made out of 3/16"od without a problem.

n00b also asked... Your gas selection is very nice though, you must have access to many many nice refrigerants.

Yeah it's nice that I often times do contract work for a Polycold related service company (M&T Systems). They have R-22, R-123, R-23, R-14, Ethane, and Argon in ready supply.
Lets me be a real mad scientist

[n00b 0f l337]

So how exactly might r123 condense r23...

[mytekcontrols]

So for -80C, just r123 and r23? You think r23 will condense at all?

Most definitely it will (at the final stage), and it will go into solution with the R-123 in the early stage, thereby evaporating at the top (suction side) of the cascade condenser somewhere around -50C. Assuming you have a larger differential pressure due to smaller cap tube flows (7-8 foot x 0.031), the discharge pressure running through the cascade combined with the -50C temperature should have no problem condensing the remaining R-23.

If you keep the suction pressure in the evaporator low (5 psi), you should see an ultimate temperature of -80C. Of course for better load handling, slightly shorter cap tubes (higher flow) with a higher suction pressure, would be preferred (although evaporator temperature will be warmer).

Some of this is off the top of my head, so I might be off a little on the cascade temperature achieved, but I know from actual experience with a very similar sized system that -80C was possible, and reliable.

[mytekcontrols]

Hmmm... I think I should go into a little more theory on why I chose the refrigerants that I did for this project.

First of all the spec I was working towards:

Desired Evaporator temperature: -30C (no colder then -40C... O-Ring leaks)
Maximum Charge Pressure: 150 psi
Fast pull-down and ability to handle large mass.

Although R-22 seems like it has the right boiling point (-40C at 1 atmosphere) to do the trick, it would require running the compressor suction close to 0 psi (considering pressure drop, and heat transfer losses). With the small compressor I was proposing to use, this would have required a fairly low mass flow through the system in order to maintain such a low suction pressure. Low mass flow equates to more time required to transfer heat of the evaporator, through the compressor, and out through the air cooled condenser.

I really wanted this thing to be up an fully operational within a few minutes (down to alcohol trapping temperature).

So using another refrigerant of a lower boiling point in an autocascade, would allow me to have a higher mass flow (faster heat transfer). I had R-23 available (-84C at 1 atmosphere). By balancing the proportion of R-23 and R-22, and adjusting the cap tube flows, I could achieve -30C with a very high mass flow.

The R-123 (+28C at 1 atmosphere) that was also used improves the heat transfer in the warmest stages, including the energy put into the compression process, heat due to electrical resistance in the motor windings, and rejection of heat at the air-cooled condenser. This should greatly prolong the life of the compressor.

On the other hand if we wanted to run the system closer to R-23 boiling point temperatures (-84C), then the R-22 becomes a detriment, acting like a contaminate in the R-23 condensate leaving the final stage. Basically the more R-22 that is present (in solution) with the R-23, then the warmer will be the boiling point of the condensate. Essentially you are creating a pseudo refrigerant, something that boils off in the range between R-22 and R-23.

So to achieve a more pure form of R-23 condensate, we would either need additional or better phase separation (e.g; 2nd phase separator and cascade), or to keep it simple, a substitute for the R-22 that has a much warmer boiling point like R-123. Of course if this were a conventional cascade, R-123 wouldn't do us much good since it does have such a high boiling point. But with autocascades, we now have the advantage that all the refrigerants are working together in a mixture, and as such presents a whole different picture.

In autocascades as we move upstream in the serially connected heat exchangers, we will see an ever decreasing temperature. We will also see some refrigerants being condensed, perhaps in a sub-cooled state before being separated out, and then go on to be evaporated at the next stage up. The more sub-cooled a refrigerant is, the higher will be the amount of other refrigerant gases that will dissolve into it. When such a sub-cooled condensate is later evaporated, it will do so at a much lower temperature then its own boiling point. So if the conditions are right, R-23 gases will dissolve into sub-cooled R-123 condensate. In fact beyond a doubt, the R-123 circulating in an autocascade will be extremely sub-cooled by the time it is extracted at the first phase separator. This is good for 2 reasons. First of all, being in a such an extremely sub-cooled state, precludes much of it getting past the first phase separator (less chance of contaminating the R-23 gases that are passing through). Secondly, a good amount of R-23 will have dissolved into the sub-cooled R-123, and when evaporated, yield a temperature somewhere between -40 to -50C which should be sufficient to condense the R-23 on its way to the evaporator (assuming we have a high enough discharge pressure).

So hopefully that helps clarify my choice of refrigerants for the cold trap chiller configuration, and also for a means of achieving a -80C system that might be suitable for PC Cooling.

[esdee]

mytekcontrols please DO make more threads like this
they are extremely helpful for us!

thank you!

[Jack]

Very nice! And a very complete story, and understandable

You mention a Suction Accumulator in your pdf file, but I don´t see it anywhere in your pictures?

What kind of phase seperator are you using? It looks like an ordinary accumulator?

An unit like this would make an awesome first stage for a classical cascade, lets say tuned to -60 with ethylene in the second stage

[n00b 0f l337]

The rotary compressor has an accumulator on it. How large of a load would you call the unit is taking in trapping alchohol.
As for the gas selection, thank you very much for the explanation. Especially with 3/16" with 3/8" working as a good HX, i'm even more tempted to pursue r23 (would cost me $800) vs ethylene (would cost me $300 or so).

I've personally converted A/C units like that into chillers though, and have the condenser and setup for the rotarys is quite nice. Not to mention economical. Dropping into a new nicer case, and possibly a larger unit (9k btu, around 3/4hp rotary) might make some very nice autocascades.

Ultimately though, would r22 help in the unit at all? Or just in the long run result in warmer evaporator temperatures.

[[XC] gomeler]

Thank you for the most interesting thread, was a pleasure to read

[n00b 0f l337]

Is r11 a possible substitute for r123? Aside from environmental concerns.

[mytekcontrols]

Thanks for the kudos

Now let me see if I can answer all the questions:

You mention a Suction Accumulator in your pdf file, but I don´t see it anywhere in your pictures?

N00b answered this one correctly. Yes it is an integral part of the LG rotary compressor, and is not something I needed to add.

What kind of phase seperator are you using? It looks like an ordinary accumulator?

The phase separator is simplicity in itself. Due to the small volume of the system, I was able to use the 1st bullet strainer as a phase separator (and it still serves the original purpose of straining any particulate matter before it feeds the cap tube). Take a look closely in both the piping diagram and the first photo in the PDF file and you'll see how it's setup.

How large of a load would you call the unit is taking in trapping alchohol

I have roughly calculated it as 50 watts based mostly on insulation losses. However I think it would be quite capable of handling 200 watts at pretty close to the same temperature (-30C). If optimized for -80C, I would think that 75-100 watts would be about it. Of course getting a little bit bigger air conditioning unit would improve this (I'm presently using the smallest LG unit for my application).

Ultimately though, would r22 help in the unit at all? Or just in the long run result in warmer evaporator temperatures

I have noticed R22 does help with autocascades having 2 or more cascade heat exchangers (my example only has one), and especially when trying to achieve -100 or better temperatures. A typical blend for a -100C 1 HP 3 cascade system would consist of: 10 oz R123, 6 oz R22, 60 psi R23, 40 psi R14 (gases are added to pressure of previous refrigerant). A system like this would require additional expansion volume because the compressor and stack volume alone would not be enough to maintain a reasonable balance pressure (total system volume should be at least 1/2 a cubic foot, with a charged pressure of 125 psi).

Is r11 a possible substitute for r123? Aside from environmental concerns

Yes R11 works just as good, and perhaps better for low temperature operation, since it is an excellent solvent for oil (insures that the oil is carried back to the compressor). I used R11 on an earlier system I built several years ago that was of similar size, and it worked great! Although I was also using R13 instead of R23, but these do have very similar characteristics.

[mytekcontrols]

Jack stated... An unit like this would make an awesome first stage for a classical cascade, lets say tuned to -60 with ethylene in the second stage

Yes! This would certainly do the trick, and it would be a far better situation for the first stage compressor (cool running).

[gosmeyer]

Thanks mytekcontrols,
Very informative and well done.

[n00b 0f l337]

*Note to self, buy a charging scale*
Thanks again mytek, your information is incredibly helpful. Too bad its hard to find small quantities of r23 or the like, I can only find 20lbers. You seem to have so much experience that you can properly estimate (are you even estimating?) the charge and temps. But lets stay with what you have now, applying that to a 200 watt load, with the 1/2hp rotary, what sorta temps when aiming for as low as possible? Would r22 be helpful at that point? Or still r123 (or r11) then r23.

[Polizei]

Excellent thread... I love seeing people who do stuff like this for a living helping us hobbyists.

What was the point of this project, if I may ask? Condensing alcohol? For what? Moonshine?

[Jack]

The phase separator is simplicity in itself. Due to the small volume of the system, I was able to use the 1st bullet strainer as a phase separator (and it still serves the original purpose of straining any particulate matter before it feeds the cap tube). Take a look closely in both the piping diagram and the first photo in the PDF file and you'll see how it's setup.

From the photo I gather that it's just a filter/drier, where the liquid will flow down to the captube and the gas will exit via the pipe at the top?

*Note to self, buy a charging scale*
Thanks again mytek, your information is incredibly helpful. Too bad its hard to find small quantities of r23 or the like, I can only find 20lbers. You seem to have so much experience that you can properly estimate (are you even estimating?) the charge and temps. But lets stay with what you have now, applying that to a 200 watt load, with the 1/2hp rotary, what sorta temps when aiming for as low as possible? Would r22 be helpful at that point? Or still r123 (or r11) then r23.

Mytek, perhaps you could design a nice unit for us... with a 1/2hp rotary and coldest temp, because I wouldn't know what lenghts of captube to use for instance.
Thanks

[n00b 0f l337]

Provided your insulation is perfect, and no extra heat is added, using the exact same size cap tubes for both "stages" is correct. But since its not perfect, you can see in cascades we use a shorter cap tube on first stage. Here its not as much needed as theirs less heat to deal with from compressor and such. I think 9ft would be about right though. I've found (w/ r22, r290, r507a, r402a, isceon 89, and a few others) that in a single stage 9' 6" is good for 185 watts, 9' is good for 230, and 8' is good to 300++. Based on that I think we could run 9' nicely of .031.
The cheap phase sep probably works very well too, since if you have more liquid then gas, itll fill the bottom and act a little like a receiver I'd imagine. (bit theoretical in my head of course). I'm considering going and picking up a window ac now, the compressors not the best part, but the huge oversized condenser with fan can be both quiet, and offer incredible performance.
How the phase sep will perform compared to something more commerical, or even a temprite 304a (a non-floated oil separator thats pretty small), or a small peice of pipe, who knows.

[wdrzal]

Lets me be a real mad scientist

Thanks for sharing

Lets just keep the MAD out of your statement and treat this as the highly technical science that it is. Always keeping safety #1.

I realize that was just a off the wall post,But don't won't our members to think anything else other than this is science combined with mechanical skills.

I've been working for years to promote safety and quality craftsmanship ,thus creating professional builds that are safe.

I only singled that statement out because of the interest in your thread and work. Please keep in mind we have people of all stages of their learning curve.

Thanks again
Walt

[mytekcontrols]

From the photo I gather that it's just a filter/drier, where the liquid will flow down to the captube and the gas will exit via the pipe at the top?

Actually it's just a "strainer" with nothing more then a screen inside. And yes the basic idea is that the larger internal volume of the strainer as compared to the 3/16"od tubing feeding into it, will allow the velocity to decrease, thereby letting the heavier liquid drop away from the gas. Same principle of any phase separator, this one just happens to be rather small (and cheap).

Of course an even better phase separator has the refrigerant entering from the side, taking advantage of centrifugal force (not just gravity) to separate the liquid from the gas.

The cheap phase sep probably works very well too, since if you have more liquid then gas, itll fill the bottom and act a little like a receiver I'd imagine. (bit theoretical in my head of course).

Yes I would definitely have to agree with you, but then any phase separator should, and does do this (unless the cap tube flow rate exactly matches the amount of liquid generated).

Mytek, perhaps you could design a nice unit for us... with a 1/2hp rotary and coldest temp, because I wouldn't know what lenghts of captube to use for instance.

I would love to tell you that I design systems by carefully calculating the various parameters (CFM, flow, heat transfer, ect.), but I don't. In fact it's been so long ago that I even did it that way that I have completely forgotten how. Don't laugh too hard, or judge me an idiot, but I just do this stuff by intuition. However I have found that in autocascade systems the length (flow rate) of the cap tubes is not as critical as some would have you believe.Which is good, since it would be very difficult to accurately calculate the proper size for cap tubes in the final stages anyway. Reason being; you won't have a constant liquid seal on the entrance of the cap tube especially when under load (standard practice and calculations in sizing cap tubes usually assumes you do have a constant liquid seal).

I'm not saying that it makes no difference at all, but it certainly isn't anywhere near as critical as the type, quantity, and number of refrigerants to be charged into the system in order to meet a given spec.

But just so you are at least in the ball park (and also to prevent wasted refrigerant from having to make multiple trials), I do believe I saw somewhere on this forum a calculation that will get you close (sorry I don't remember where).

Another method that works quite well when developing a brand new design, is to run several very small cap tubes in parallel. These can then be pinched off one at a time while testing under load to determine the best size for your new system. It is also a tremendous help, especially when dealing with a system using several cascades, to thermocouple the In's and Out's of all the heat exchangers and plot them in real-time. This is how you can develop a "feel" for what happens when changing flow, and/or adding refrigerants. If you choose to only monitor the evaporator, you will end up with a very incomplete picture.

What was the point of this project, if I may ask? Condensing alcohol? For what? Moonshine?

No it wasn't a fancy alcohol still

The purpose was to protect a vacuum pump when pumping down a refrigeration system that had previously been flushed of contaminates by using alcohol as a solvent.

At the service facility I contract to, we sometimes have to restore a large autocascade that had previously suffered a compressor burn-out, and gone acidity. Alcohol is cheap, has a zero ODP, and is a fairly good solvent. The problem is that it can really foul up the vacuum pump oil when we later evacuate the system for charging (hard to drain it out of all the nooks and crannies).

So as it turns out, -9C will condense it out of the gas stream even when pumped down to 10mmHg. -30C is even better. So the project is a "Cold Trap" for alcohol.

[n00b 0f l337]

Again, amazing advice. You are an incredible help to this community mytek. I'm going to pickup one of these ac's as the condenser setup is exactly what size I've been wanting, and the rotarys already placed perfectly. I do not have r23 just yet, but I'm also going to get a quote as I think its nearly the perfect autoc gas from most reading. I'll probably turn this ac into a chiller with a 16plate hx, but in the future I'm going to try what you have built, with our style evaporator and flexible line, and also use my lever arm pipe bender to make a tighter almost plateHX sized heat exchanger. I have one last question, do you use a pour-then-expand insulation foam, or a spray foam?

[mytekcontrols]

I have one last question, do you use a pour-then-expand insulation foam, or a spray foam?

Once again I am lucky to work for a service and repair facility, since they have a very nice spray foam machine that I can use.