PDA

View Full Version : Miscibility of HFC and HC refrigerants



DaBit
29-07-2003, 12:55 PM
I am investigating the feasibility of a using a modified Joule-Thomson refrigeration system for CPU overclocking purposes.

The few articles found about these systems use a mix of hydrocarbons, usually butane, propane, ethane, methane and nitrogen, as the working fluid.

I am investigating whether or not I could exchange some of these components by HFC refrigerants. I am leaning towards a mixture of R134a, R410a (R32/R125), ethane, methane and nitrogen. This to reduce HC content.

One of the things bothering me: will the HFC's and HC's mix in liquid state?

frank
29-07-2003, 08:02 PM
DaBit

Never heard of the Joule-Thomson system but I'm intreaged to know how a non-condensible gas such as nitrogen can be used in a refrigeration system?

Gary
30-07-2003, 12:52 AM
I think DaBit is referring to an autocascade system.

Liquid nitrogen exists, therefore nitrogen is condensible, albeit at a very high pressure and/or very low temperature.

Dan
30-07-2003, 01:29 AM
HFC's and HC's are commonly mixed with no problem. I haven't heard of nitrogen used in an autocascade refrigeration system either.

I have long been fascinated with autocascade. Is that what the Joules-Thompson effect is? I did a search, but couldn't find any references to the Joules-Thompson effect in refrigeration design.

Prof Sporlan
30-07-2003, 02:07 AM
I think DaBit is referring to an autocascade system.
The Prof is acquainted with the Joule-Thompson effect, and wonders if the term 'Joule-Thompson refrigerator' is a bit of a misnomer. 'Autocascade system' is a better term in his humble opinion... :)


One of the things bothering me: will the HFC's and HC's mix in liquid state?
HFCs and HCs are components in many of the refrigerant blends on the market. Mixing them should present no problems when developing the appropriate refrigerant concoction...

DaBit
30-07-2003, 10:02 AM
Originally posted by frank
DaBit

Never heard of the Joule-Thomson system but I'm intreaged to know how a non-condensible gas such as nitrogen can be used in a refrigeration system?

The Joule-Thomson system uses the fact that a high pressure gas stream can be partially liquefied by expanding it. For example, when expanding high pressure CO2, CO2 snow will form.

Both the high pressure gas and low pressure return are passed through a counterflow heat exchanger to bridge the temperature difference.

Basically this is what I decribed here (http://www.phase-change.com/phpBB2/viewtopic.php?t=576). I never heard of these cycles when I wrote that post.
And here (http://www.tu-dresden.de/mw/iem/kkt/mitarbeiter/lib/Alex/CECPortland.pdf) is an article describing the optimization of the gas mixture for a 77K / 20W JT cooler.


Originally posted by Prof Sporlan
The Prof is acquainted with the Joule-Thompson effect, and wonders if the term 'Joule-Thompson refrigerator' is a bit of a misnomer. 'Autocascade system' is a better term in his humble opinion... :)

You might be right Prof, and at least the two look similar.
But the 'usual' autocascade uses phase separators to fractionate (good english?) the refrigerant mixture until only the highest boiling point refrigerant remains, which is then evaporated at the target.

The basic JT-Rankine cycle uses high pressure gas which is partially liquefied during expansion, basically operating in two-phase.

The modified cycle with basically a very large glide refrigerant looks a bit like both. Operation is in two-phase across the entire system, and no phase separators are used, keeping system construction simple. But we are evaporating/condensing different refrigerants along the heat exchanger, which is not all that different from the 'classic autocascade'


HFCs and HCs are components in many of the refrigerant blends on the market. Mixing them should present no problems when developing the appropriate refrigerant concoction...

Any tips & tricks? I am just investigating the possibility.

To get me a guideline for the refrigerant composition, I planned to set up a set of equations which when solved tells me the component fractions needed to get the more volatile component liquid by evaporating the less volatile component. I planned to initially include ltemperature, latent heat of evaporation and specific heat of the liquid and vapour phases only.

Would this lead to a useable 'guideline' mixture which can be optimized by experimentation (I suppose the proposed model is too simple to match reality)?

BTW: I haven't found the word 'concoction' in any English-Dutch dictionary....

herefishy
30-07-2003, 02:54 PM
Webster's Compact Dictionary - con-coct' v.t. 1.make of varied indgredients. 2.devise. - - con-coc'tion, n

DaBit
30-07-2003, 03:49 PM
Originally posted by herefishy
Webster's Compact Dictionary - con-coct' v.t. 1.make of varied indgredients. 2.devise. - - con-coc'tion, n

Thanks :)
My English is still far from fluent :)

Prof Sporlan
30-07-2003, 09:41 PM
The Prof learned something today about the Joule-Thompson refrigerator. His engineering texts, BTW, include the 'p' in 'Thompson'. Which is the correct spelling? :confused:

The system employs a gas cooler and desuperheater, like you would have with a trans-critical refrigeration system. And to lower pressure ratio across the compressor, refrigerants are blended together, as one does with an autocascade system.

DaBit
30-07-2003, 11:47 PM
Most of the texts I have seen uses Thomson, without the 'p'.

Anyway, the cycle might be interesting for 'us overclockers', since we are looking for low temperatures at a low load (< 1/16 tons). COP doesn't interest us as much as most of you; size and cost does.

Based on these criteria, autocascade / JT refrigerators might prove an excellent solution.

Prof, could you tell me if the above means of calculating a 'guideline' blend of refrigerants might work? It is fairly easy for me to write some software which performs the task of solving the set of equations.

Also, I begin to feel the lack of a good book on the mathematic side of refrigeration; browsing the Internet only partially answers my questions. And just like a torch or a pipe cutter, math is a powerful tool when used properly.
Which one would you recommend? Ideally it would have to be a complement to Gary's practical books which I intend to buy in the near future.

Prof Sporlan
01-08-2003, 02:50 AM
Prof, could you tell me if the above means of calculating a 'guideline' blend of refrigerants might work?
Without thinking about this subject too hard (the Prof is currently enjoying a fine Paulaner Salvator double bock beer), one would think a traditional autocascade system would be easier to construct and/or more efficient than a J-T refrigeration system for no other reason that autocascade systems seem to be more prevalent. The Prof has seen a number of autocascade systems, but nary a J-T refrigerator!

The Prof recalls an engineer who formerly worked at Revco Scientific who used TK Solver to develop refrigerant blends for his autocascade systems. Nowadays, the NIST RefProp program would be able to simply this effort as it allows one to develop custom refrigerant blends.

DaBit
01-08-2003, 09:39 AM
Originally posted by Prof Sporlan
Without thinking about this subject too hard (the Prof is currently enjoying a fine Paulaner Salvator double bock beer)

I am not comfortable with that specific brand of beer, but I assume that the Prof spend years of research finding suitable beers for himself.

But Prof, one should not enjoy beer behind a computer. The radiation emitted by modern CRT's and TFT's destroy most of the fine flavour. ;)


one would think a traditional autocascade system would be easier to construct and/or more efficient than a J-T refrigeration system for no other reason that autocascade systems seem to be more prevalent.

If you look at the prototype of the JT cooler in the .PDF I posted here, construction of a J-T refrigeration system is easier, and at least it is a lot easier to tune since the only parameters varied are the refrigerant mixture and final expansion orifice. With autocascade systems one must tune the mixture and various expansion devices.

Efficiency of JT refrigerators is not that high, with the prototype mentioned earlier producing 20W of cooling @ 77K with 400W electrical input. I don;t knwo how this stacks up against classic (auto)cascades.

While thinking about it, a combination of both might be optimal. Since an oil separator is needed anyway, it could as well serve as phase separator for the highest boiling point component. Lower boiling point components could follow the JT flow.


The Prof has seen a number of autocascade systems, but nary a J-T refrigerator!

Here in The Netherlands we have a saying ('Wat de boer niet kent vreet hij niet') which would translate to 'a farmer doesn't eat what he doesn't know'.

I have seen this to be one of the reasons why some solutions for engineering problems are popular while other are not.


The Prof recalls an engineer who formerly worked at Revco Scientific who used TK Solver to develop refrigerant blends for his autocascade systems.

TK Solver is very interesting software, but it only works if you know what equations you must solve. I have no idea how to calculate the properties of a blend.


Nowadays, the NIST RefProp program would be able to simply this effort as it allows one to develop custom refrigerant blends.

Seems I need to contact the NIST if they have a relatively cheap version.

Prof Sporlan
01-08-2003, 10:03 PM
I assume that the Prof spend years of research finding suitable beers for himself.

It is an ongoing endeavor to discover the finest beer in the world. :)

DaBit
04-08-2003, 08:56 AM
:)

I will repeat a question: what book covering the mathematics of refrigeration would you recommend?

Ideally it would be a complement to Gary's practical books.

Prof Sporlan
11-08-2003, 02:54 AM
The Prof often refers to ASHRAE's handbooks first when researching a topic on refrigeration. But they aren't necessarily the best one-stop reference for mathematics using in the hvac/r industry.