PDA

View Full Version : TXV's and LST compressors







DaBit
08-05-2002, 11:07 AM
Flooding the forum again :)

I am busy with a cooling system for my computer (more information can be found here (http://www.refrigeration-engineer.com/chat/showthread.php?s=&threadid=706) and here (http://www.refrigeration-engineer.com/chat/showthread.php?s=&threadid=713)), and I decided to mount a receiver and a kind of expansion valve instead of the captube I am using now.

A simple, small TXV like the Danfoss T2 with the smallest orifice and MOP point seems to be the best option for the only 200W (or less) load, but I foresee a few problems.

First of all I would like to notice that I have never ever seen a TXV in real life, and I have absolutely no experience with them. I know how they work by reading articles on the Internet.

The biggest problem I foresee is that the Danfoss NL11F compressor I use is a Low Starting Torque compressor. This type of compressors needs pressure equalisation between the suction and discharge side to be able to start.

Now, If I got it right, with a TXV mounted the pressure will not be equalised. This would mean that the compressor would not start after the first time.

How do I solve this? I do not need quick equalisation of the pressure. If it takes half an hour it's fine. I have thought about a few solutions:

- Using a TXV with pressure equalisation feature. They seem to exist, but I can't find them.

- Bypass the TXV with an electrical valve which is closed during normal operation, and opens when the compressor is turned off. This is an expensive and probably undoable solution.

- Making a very restrictive bypass between the liquid and evaporator side of the TXV. If it is so restrictive that it takes 30 minutes to equalise pressure, I suppose this bypass won;t affect normal operation. Problem: this bypass must be _VERY_ restrictive, and I have no idea how to solve that.

- Using a constant pressure valve instead of a TXV. Such a valve would maintain a constant evaporator pressure. Is this a feasible solution?

- Mounting a run capacitor on the NL11F. This seems to increase starting torque althought the Danfoss datasheets are vague about this. If it works this is the best solution.

For the range of possible expansion devices I am pretty limited to a part of the Danfoss range and maybe a few Sporlan valves. In The Netherlands you are not allowed to buy those things without a license, so it has to be arranged via-via.

herefishy
08-05-2002, 05:26 PM
It sounds like you are aware of all the available options. I think that I would try the "high torque" configuration on the compressor (with start capacitor). :)

Dan
09-05-2002, 02:28 AM
I am amazed. DaBit, watching you is like watching a refrigeration technician's egg hatch.

I cannot wait until we get into flooded and direct expansion cooling. That's the next consideration, you know.

For a start, what would happen if we just dripped liquid nitrogen on the chip? Maybe it would crack or something. So, okay, it is not just a chip we are cooling. Is it a motherboard we are cooling? All the discussion avoids the actual cooling surface applied to the device or devices you are choosing to cool. I would like to learn more about that end of it.

No matter. I am impressed with your proficiency in the refrigeration realm, regardless of why you seek what you seek.

Prof Sporlan
09-05-2002, 04:45 AM
A simple, small TXV like the Danfoss T2 with the smallest orifice
Keep in mind that Sporlan also makes small capacity TEVs... :) BTW, the abbreviation use for thermostatic expansion valve among the refrigeration cognoscente is "TEV".


Now, If I got it right, with a TXV mounted the pressure will not be equalised.
A TEV will not equalize system pressures unless it has a <b>bleed port</b>.


Using a TXV with pressure equalisation feature. They seem to exist, but I can't find them.
Be careful you do not confuse a TEV with an external equalizer as one that can equalize system pressures. An externally equalized TEV has three connections: inlet, outlet and equalizer. The equalizer connection is used to sense pressure at the outlet of the evaporator via an equalizer line (typically a 1/4 OD copper tube) connecting it to the suction line. An externally equalized TEV is used when there is a measureable pressure drop between the outlet of the TEV and evaporator outlet (a refrigerant distributor being a typical cause)

An internally equalized TEV has only an inlet and outlet connection, and it is designed to sense evaporator pressure via an internal passageway within the valve.

A TEV that can equalize pressures during off-cycle is one that has a bleed port, and that valve can be internally or externally equalized.


Bypass the TXV with an electrical valve which is closed during normal operation
This will work, but is a bit involved. Rube Goldberg would approve of this approach though… :)


Making a very restrictive bypass between the liquid and evaporator side of the TXV
At 200 watts design load, the bypass would have to be extremely restrictive.


Using a constant pressure valve instead of a TXV.
Like the TEV, a contant pressure expansion valve will also hold a pressure differential unless it has a bleed port.


For the range of possible expansion devices I am pretty limited to a part of the Danfoss range and maybe a few Sporlan valves.
The Prof might argue Sporlan offers more varieties of small capacity TEVs than the other manufacturers, despite that fact no one can really claim to have a nominal 200 watt version… :)

DaBit
09-05-2002, 01:06 PM
It sounds like you are aware of all the available options. I think that I would try the "high torque" configuration on the compressor (with start capacitor).

If it is sufficient, this is definitely the way to go. However, I am not sure *if* it is sufficient. I tried to calll Danfoss, and they just told that the cannot help me since I am not an official licensed refrigeration engineer. And since I lack the elemntary refrigeration equipment, I have to be pretty sure before I make a design change since if it does not work, I'll have to wait several weeks.


I am amazed. DaBit, watching you is like watching a refrigeration technician's egg hatch.

My English is not good enough to fully understand this, but I will take it as a compliment :)


I cannot wait until we get into flooded and direct expansion cooling. That's the next consideration, you know.

I know, but hoses are easier to handle than copper pipes, and with indirect cooling I can almost effortless add devices to the chain, which I cannot with direct expansion on the CPU die.



For a start, what would happen if we just dripped liquid nitrogen on the chip? Maybe it would crack or something. So, okay, it is not just a chip we are cooling. Is it a motherboard we are cooling? All the discussion avoids the actual cooling surface applied to the device or devices you are choosing to cool. I would like to learn more about that end of it.

Okay, sit back and relax. What I am cooling is currently only the CPU. Physics state that the speed of a transistor implemented on a CMOS process is almost lineairly with absolute temperature. Thus, at 150K it would run twice as fast than at 300K. I use this to increase speed, mainly because it's fun to do it. A faster chip is just like a faster car; once you are used to it, the fun is over. But designing and building the cooling devices is just fun. Pick a goal, and go for it.

But, the CPU is not the only part on the mainboard that influences speed. You also have the chips driving the CPU (called the Northbridge), the RAM and the video adapter. And to compare with cars again: it does not make sense to put very large tyres under your car if you don't upgrade the engine too. So, I have to cool multiple chips to keep everything balanced.

And that is where the coolant loop comes in the picture. Running liquid refrigerant to all chips and letting it evaporate over there is quite a hassle. It can be done with multiple TXV's or liquid distributor, but it is not exactly easy. Especially not when you consider that the heat generated by some chips is only a single Watt. How would you throttle refrigerant? Thus, to solve this, I decided to use a secondary coolant loop. The coolant gets chilled by a water chiller, and the cold coolant is pumped to the diverse chips on the mainboard. Adding an extra device is means cutting a hose in two pieces, adding some copper, and put the thing back in service again. The disadvantage is that I loose a couple of degrees.

People often tell me to just buy a faster processor, but the fact is that it often takes the processor manufacturers about 9-12 months to release a processor which is faster than my overclocked one. Mu current processor (AMD XP 1800+, which runs at 1533MHz) is 6 months old, and it runs at somewhat above 2GHz. AMD still has to release a proc which runs at 2GHz+ rate.

For the primary cooling system I chose R134a. My initial goal is -20 &deg;C (-4F) on the coolant, at load. In this area, R134a is suitable. And, more important, Europe uses it in refrigerators, so repair companies can fill my system. With R22 of R404a or something, this is not the case. Also, second hand R134a compressors can be found in multiple places.

For the secondary circuit, I chose a 30/70 methanol/water mixture. I chose this mixture because it has a large heat capacity, decent heat transfer capability, low viscosity at low temperatures, and it is easy to obtain methanol since it is used in model engines. The freezing point of the current mixture is -30 &deg;C. Water/glycol does not perform nearly as well (tried that) as water/methanol. It becomes thicker at lower temps which impacts flow, and has a lower heat transfer capability. Twice as bad, thus.

Now, back to the nitrogen: this is a funny experiment, but it is very impractical to keep such a system running as a normal cooling installation. Adding a nitrogen compressor could work, but all this becomes too dangerous for me. Too low temps, too high pressures. Ethane evaporating at -80 &deg;C might be an option someday, but I think it will stay a dream. I am not (completely) crazy, and I want to stay alive.


No matter. I am impressed with your proficiency in the refrigeration realm, regardless of why you seek what you seek.

Thanks :)
I am still learning, and you guys are of great help. I only wished I had found this forum a few months before. That would have saved me a lot of trouble. But I learn from my mistakes, dump the debris, and continue.



Keep in mind that Sporlan also makes small capacity TEVs... BTW, the abbreviation use for thermostatic expansion valve among the refrigeration cognoscente is "TEV".


I know, but where can I obtain one in The Netherlands? I suppose you are not willing to ship me one, are you? I do not have a favour for Danfoss, I just happen to have a catalogue of them here.

Oh, and thanks for help on the technical terms. I just use what I pick up somewhere, which is not always correct.


A TEV that can equalize pressures during off-cycle is one that has a bleed port, and that valve can be internally or externally equalized.

I meant an internally equalised valve with bleed port :)
I don't think an external equalised valve would work much better.

Question: does it hurt to use an externally equalised valve, even when it is not strictly nescessary? I could do this to avoid problems, and running an extra 1/4" line is not that much work either.


The Prof might argue Sporlan offers more varieties of small capacity TEVs than the other manufacturers, despite that fact no one can really claim to have a nominal 200 watt version…

What type of valve would you recommend me, and where could I obtain one in The Netherlands? I am willing to use Sporlan, but obtaining them is a problem.

Gary
09-05-2002, 01:29 PM
Keep in mind that Sporlan also makes small capacity TEVs... BTW, the abbreviation use for thermostatic expansion valve among the refrigeration cognoscente is "TEV".

Actually, both TXV and TEV are commonly used throughout the industry. In an effort to standardize terminology, it seems the powers that be have settled upon TEV as the officially accepted abbreviation. Still, you are as likely to see one as the other.

Abe
11-05-2002, 12:00 AM
Da Bit

If you "fail" to secure a valve, give me a bell, I can obtain one for you in the UK and ship it over to you, ( providing Im not breaking any laws) I doubt very much I would be.
I need exact type of valve you require. My local wholesaler stocks Danfoss, but I can obtain Sporlan as well.

Regards
Abe

DaBit
11-05-2002, 10:27 AM
Aiyub:

I have not yet been able to obtain one. Yesterday I was kicked out of a HVAC scrapyard when I told the guy that the TEV he found was not suitable for me since it was R404a and a too large orifice. The pig told me that only a valve and a few meters of 3/8" pipe was not worth enough to justify searching for them, and then he kicked me out of the door, literally. But then, I never heard a positive story about that guy from other people, so it was expected.

I have two adresses I have to visit before I consider getting a valve in Holland as 'impossible'.

There are three valves which are possibly useable:

- Sporlan FJ-1/8-VC-ODF (But I will let the Professor comment on this valve :) )
- Honeywell TMVBL-00206 + VD 0.3
- Danfoss T2 valve with orifice 00
- Danfoss TUA valve with orifice 00

But if everything fails, I will be glad to accept your offer :)

Prof Sporlan
11-05-2002, 05:38 PM
Question: does it hurt to use an externally equalised valve, even when it is not strictly nescessary?
An externally equalized TEV can be used for any application.... its only drawback is you have to run an equalizer line.

An internally equalized TEV should only be used on small capacity single circuit evaporator coils having minimal pressure drop.


- Sporlan FJ-1/8-VC-ODF (But I will let the Professor comment on this valve :))
There is a problem with this model number. The 'FJ-1/8' indicates a 1/8 ton R-134a TEV, which is fine for your application. The 'VC' designation, however, indicates an R-22 thermostatic charge. The 'VC' is probably a mistake. The actual thermostatic charge will be marked with a black epoxy on the power head, or metal stamped if it is an old valve. Make sure there is an 'FC' and/or 'JC' marked on the power head. If you find a 'VC', however, someone has swapped the head so that the valve could be used for R-22 service, and you'll need to find a R-134a power head for the valve.

BTW, the Danfoss T2 or TUA with the 00 cartridge should work also, provided these valves have the appropriate R-134a thermostatic charge.

DaBit
11-05-2002, 08:23 PM
I was actually looking at new valves, since obtaining one second-hand is even harder. Those valves are of too less value to buy second hand for most professionals. For the rare refrigeration DIY guy like me it is another story of course.

I think I have read the table wrong. Thanks for correcting me.

RogGoetsch
13-05-2002, 02:01 AM
I sometimes design very small refrigeration systems and completed a 100 watt, R-12/saline chiller (tube-in-tube HX) for a bio-medical mfg co. in '94. I redid that design for R134a a few years later.

As far as I am concerned, the main advantages of a TXV (okay: TEV, darn those cognoscente, anyway!) in a small system are where load fluctuations are anticipated or to take advantage of the higher regions of the compressor curve during initial pull-down.

A capillary tube system is optimized for a certain set of conditions and would therefore seem fine for your system if properly sized. Therefore I think you are probably barking up the wrong tree if you are trying to solve a performance problem by the switch to a TEV (and thereby creating another problem with the compressor starting torque issue.)

It would be interesting to put the thing on a P-H (pressure-enthalpy) diagram. I started to read through all the previous posts related to this and gave up before I could get clear on the operating conditions of whatever is the current version.

In any chiller design, fluid characteristics are prime. A static fluid boundary layer will kill you. Even laminar flow is terrible compared to turbulent flow, hence the suggestions in other posts to get that coolant moving around and through the tubing bundle.

Good Luck!

Rog

zolar1
13-05-2002, 08:33 AM
A tornado effect could be beneficial with respect to the alcohol/water misture inside the 'chiller'

DaBit
13-05-2002, 10:15 AM
As far as I am concerned, the main advantages of a TXV (okay: TEV, darn those cognoscente, anyway!) in a small system are where load fluctuations are anticipated or to take advantage of the higher regions of the compressor curve during initial pull-down.

If possible I want to take advantage of both. The shorter the inital pulldown takes, the better it is for me. I am also facing load variations. Currently I have adjusted the refrigerant charge in such a way that I have a 6K superheat under minimum load conditions (when the CPU is idling). This increases to about 20K when load increases. It will become even worse when I add the graphics adapter to the coolant chain. Then, I have a 80W difference between, for example, running Word (system idles 95% of the time) and playing a game (CPU load 100%, gfx card load 100%).



A capillary tube system is optimized for a certain set of conditions and would therefore seem fine for your system if properly sized.

I agree that my capillary tube is probably not sized well. But, as I said before, I do not have the possibility to play with it since I lack basic refrigeration engineering equipment like a vacuum pump and oxy-acetylene torch. I have to do that work somewhere else.



Therefore I think you are probably barking up the wrong tree if you are trying to solve a performance problem by the switch to a TEV (and thereby creating another problem with the compressor starting torque issue.)

The starting torque thing can be solved by switching the start relay and adding a start capacitor. Not really a big deal.



It would be interesting to put the thing on a P-H (pressure-enthalpy) diagram. I started to read through all the previous posts related to this and gave up before I could get clear on the operating conditions of whatever is the current version.

What information are you missing?


In any chiller design, fluid characteristics are prime. A static fluid boundary layer will kill you. Even laminar flow is terrible compared to turbulent flow, hence the suggestions in other posts to get that coolant moving around and through the tubing bundle.

The chance that you get a laminar flow over the pipes in any liquid chiller are almost zero. But you are right about the coolant moving too slowly and forming a static fluid boundary around the evaporator pipes.

I will switch the evaporator for a coaxial design using a 22mm (19mm I.D) outer pipe and a 9mm (7mm I.D.) inner pipe (this is almost 3/8", which would be 9.5mm O.D). I will do this anyway since the PVC causes me way too much trouble. It seems to be leak-free for the moment, but I wonder how long it will stay leak-free.
The inner pipe will be fitted with copper fins which will serve as both spacers (so the inner pipe stays centered within the outer pipe) and contact area enlargers. They also provide a more turbulent coolant flow which is nescessary since coolant speed through the evaporator is still not as large as I would like it to be. This is due to the quite large diameter of the outer pipe and the rather low coolant speed (500-800L/hour flow. Calculations show this will equal to about 60cm/second through the pipe).
Here in The Netherlands I can buy 15mm O.D. pipe and 22mm O.D. pipe. There is no size between those, and 15mm is too thin.

Total length of the (folded) coaxial construction will be somewhat more than 3 meters. This length is not based on any knowledge, but it is more a convenient length for me.

On the suction line from the coaxial evaporator I will mount two suction gas<->liquid line heat exchangers with the TEV bulb mounted on the suction line between the two heat exchangers, just as I was being told. The heat exchangers will consist of a 1/4" liquid line parallel to the 3/8" suction line, brazed together with silver and with copper wire wound and brazed around the two tubes to further increase heat exchange. This is a simple construction which should suffice.

I will try to use an externally equalised TEV to avoid problems with pressure drop over the evaporator. The TEV should be able to cope with the load variations, and be able to maintain better operating conditions than the captube I am using now.


Good Luck!

Thanks!

RogGoetsch
13-05-2002, 05:21 PM
For basic P-H diagram need press & temp, inlet & outlet of comp & metering device.

If I were building this, I would use the 9mm in the 15mm, forget the fins, put liquid through inner & refrigerant through annular space. This would allow me to measure temps along the outer tube beneath the insulation & monitor performance. I would be able to see how far along the HX I had saturated refrigerant vs superheat. I would skip the liquid-to-suction HX's since you will have beaucoup superheat off the evaporator alone, (presuming a 200 watt load @ 5degreesC average TD) and there is almost nil thermodynamic advantage. The efficiency improvement referred to is in evaporator surface utilization and net refrigerating effect per lb of refrigerant circulated (since cooler liquid refrigerant has less flash gas.)

An educated guess is that 1 meter of the above will give you about 100 watts at 5degreesC TD, hence my unconcern for the extra HX's. Oh, you are probably trying for much less than 5C TD, given your "as cold as possible" criterion. Okay, go for it! Turbocharge that puppy!

Rog

DaBit
13-05-2002, 05:57 PM
For basic P-H diagram need press & temp, inlet & outlet of comp & metering device.

I cannot give you pressures since I have no pressure meters installed. I will measure temperatures for you tonight.


If I were building this, I would use the 9mm in the 15mm, forget the fins, put liquid through inner & refrigerant through annular space. This would allow me to measure temps along the outer tube beneath the insulation & monitor performance.

the 15mm piping has an I.D. of about 13mm. This leaves only 2.75mm between the 9mm evaporator pipe and the outer pipe. This restricts coolant flow too much (the pump driving the coolant flow is a centrifugal one. These pumps produce less flow when the pressure drop increases). Monitoring performance is also possible with 22mm piping.


I would skip the liquid-to-suction HX's since you will have beaucoup superheat off the evaporator alone, (presuming a 200 watt load @ 5degreesC average TD) and there is almost nil thermodynamic advantage. The efficiency improvement referred to is in evaporator surface utilization and net refrigerating effect per lb of refrigerant circulated (since cooler liquid refrigerant has less flash gas.)

And it allows me to have a smaller superheat setting on the evap alone without the risk of damaging the compressor. Also, cooler refrigerant reduces the risk of flashing before the TEV, and it improves efficiency since the liquid refrigerant itself need not to be cooled to evaporating temperature (well, at least less). A second advantage is that the suction gas<->liquid line heat exchangers increase superheat, so the total superheat in the evap is lower, which is desirable, especially in this situation. The second HX raises superheat even further, preventing liquid returning to the compressor.

Anyway, it does not hurt, wheter it improves performance or not. I doubt performance would lower from performing SG<->LL heat exchange. But then again, I am a refrigeration newbie without practical experience, so I could be wrong.


An educated guess is that 1 meter of the above will give you about 100 watts at 5degreesC TD, hence my unconcern for the extra HX's. Oh, you are probably trying for much less than 5C TD, given your "as cold as possible" criterion. Okay, go for it! Turbocharge that puppy!

The extra HX-es are not hard to build, and if they give me any gain, fine. If they don't, too bad. But 5 degrees TD is quite a lot.

Where does your educated guess come from? I would like to know more about it.

Gary
14-05-2002, 01:59 AM
Dabit, 5C TD is an accepted design standard. Rog's educated guess is probably correct. Your evaporator will be more than enough, and the HX's are also not needed. But we are indeed turbocharging this puppy and going for 'as cold as possible'. Don't forget to get a TXV with MOP, or we could overload the compressor during pulldown. :)

Now that I think about it, the compressor is quite a bit oversized, so there shouldn't be any danger of overloading it.

DaBit
14-05-2002, 10:00 AM
Dabit, 5C TD is an accepted design standard. Rog's educated guess is probably correct. Your evaporator will be more than enough, and the HX's are also not needed. But we are indeed turbocharging this puppy and going for 'as cold as possible'.

When I look at the extra amount of energy you have to put into a cooling system to obtain those extra 5 &deg;C, I think that maintaining a 5 &deg;C TD is a pure waste. With evaporators drawing energy from air I can understand it, but liquid is so much more efficient in transporting heat. And indeed, I want to go as cold as possible.


Now that I think about it, the compressor is quite a bit oversized, so there shouldn't be any danger of overloading it.

Ehm, can you explain how you came to this conclusion? Just curious.

Question for all of you: I can possibly obtain expansion valves of the brand Egerhof or something. It would be the type TER-L1, R134a thermostatic charge, no MOP. I got this information through the phone from a guy talking local language (and not my local language), so I might have understood the name of the manufacturer wrong. It seems to be a German manufacturer. Does anyone of you know more about these TEV's?

Prof Sporlan
15-05-2002, 02:28 AM
Question for all of you: I can possibly obtain expansion valves of the brand Egerhof or something.

Egelhof is a German valve manufacturer. They primarily manufacture TEVs for automotive applications, but they also manufacture solenoid valves, and they have an electric expansion valve, though the Prof hasn't had a chance to look at it. Its division in France was recently acquired by Honeywell, which the Prof understands also manufactures TEVs. Honeywell has also developed some business relationship with Egelhof Germany. Honeywell also recently acquired Flica, another German valve manufacturer involved with TEVs, solenoid valves, and filter-driers.

Egelhof and Flica have little presence in the US market, and Danfoss, of course, tends to dominate the European market with respect to refrigeration valves.

The Prof hasn't committed Egelhof TEV model numbers to memory (as he has with other TEV manufacturers... :D), but he'll look up the TER model when he gets a chance...

Gary
15-05-2002, 04:04 AM
Ehm, can you explain how you came to this conclusion? Just curious.


Judging from the previous data, the compressor and condenser were very lightly loaded. On the other hand, with an oversized evaporator the startup load will be much higher. I would get the TXV with MOP just in case. Once the temperature drops down, the condensing unit should handle the load easily.

zolar1
15-05-2002, 06:30 AM
Why couldn't he just buy a $30 heat exchanger for his project? They are rated for about a 2 to 10 ton A/C unit.

RogGoetsch
15-05-2002, 07:57 AM
Actually, my educated guess is probably wrong. The only question is how far wrong.

Of course engineers always calculate everything to twelve decimal places and then double it for a safety factor!

The interesting thing about small chiller design is that there is almost no data to work from. At least in the ASHRAE Handbook, the US AC&R bible, there was no data on heat transfer across tubing smaller than 5/8"OD when I needed it. You have to make some assumptions, build, measure and adjust.

My "educated guess" is an extrapolation from data in a 12-year-old computer file of a length of tube-in-tube heat exchanger with 28% less surface area per foot of length than you are proposing. But the more I think about it, the less I can remember how far apart my data points were. I suspect I may have erred by a factor of two. If I can unearth the original drawing in the near future, I will let you know. But you're going to build it anyway, and there are many variables that will differ between the two systems, so you may get twice the performance. (How's that for hedging!)

As far as the 5°TD, in our business, if you are serious about efficiency you use a direct expansion system. By the time you are considering a chiller, there are criteria other than efficiency driving the design. So how about we fix you up with some nice, little, chip-sized, copper pancakes with a cap tube feeding each one......!

Rog

DaBit
15-05-2002, 10:27 AM
Egelhof and Flica have little presence in the US market, and Danfoss, of course, tends to dominate the European market with respect to refrigeration valves.

You can say that! It's Danfoss everywhere. But then, try to obtain one as a DIY-er.... I really hate those laws. What harm can I possibly cause by buying a TEV?


The Prof hasn't committed Egelhof TEV model numbers to memory (as he has with other TEV manufacturers... ), but he'll look up the TER model when he gets a chance...

I tried checking out their website, without much luck. I will do some research myself, but any information you can give me is welcome. The fact that it is egeLhof instead of egeRhof, searching works better.


Judging from the previous data, the compressor and condenser were very lightly loaded. On the other hand, with an oversized evaporator the startup load will be much higher. I would get the TXV with MOP just in case. Once the temperature drops down, the condensing unit should handle the load easily.

I doubt that the compressor I am using right now will be able to overload the condenser. I have never seen the fan controller raising fan speed because the default speed (which is quite slow, just enough te ensure the fans keep on running) was insufficient (I am using 2 Papst 24V 120mm fans on the condenser, controlled by a simple DIY proportional controller). Even with too much refrigerant in the system and during pulldown the condenser was able to handle the load without the slightest problem.

If the compressor is lightly loaded, I would like to use that situation to drop the evaporator temperature.


Why couldn't he just buy a $30 heat exchanger for his project? They are rated for about a 2 to 10 ton A/C unit.

Well, I tried to obtain a plate heat exchanger, but they are way oversized for the application. According to a SWEP (a Dutch plate HE mfgr) engineer, I would not have a lot of success with them. If you are interested, I will translate the mail to English and post it here. But even if they were suitable, obtaining one is the harder problem.


Of course engineers always calculate everything to twelve decimal places and then double it for a safety factor!

Yup, until the marketing guy drops by and says it is going to be too expensive ;)


At least in the ASHRAE Handbook, the US AC&R bible, there was no data on heat transfer across tubing smaller than 5/8"OD when I needed it.

I am interested in the data on heat transfer across 5/8" tubing. At least it gives me an idea of the quantities of the diverse parameters.


But you're going to build it anyway, and there are many variables that will differ between the two systems, so you may get twice the performance. (How's that for hedging!)

Yes, I am going to build it anyway. But I can still vary the length of the evaporator and some more parameters.

I am getting pretty close to the point of writing a simulation program for these problems. As an electronics engineer (digital IC design) I simulate everything until I am 99.8% sure that it will work. My first steps into the realm of refrigeration are more like stepping into the complete dark, unable to do forecasts, simulations or even an educated guess.


As far as the 5°TD, in our business, if you are serious about efficiency you use a direct expansion system. By the time you are considering a chiller, there are criteria other than efficiency driving the design.

I do have other criteria like ease of use and easy of experimentation. It is easier to direct a liquid flow to some parts which need chilling than it is to direct refrigerant to those places. And I would like to have more fun of it than only the current generation of processors etc. Maybe I step over to dual processor next year, and then it is easier to direct a stream of liquid to the new part than it is to redo the captube sizing.


So how about we fix you up with some nice, little, chip-sized, copper pancakes with a cap tube feeding each one......!

I am not sure it would give me lower temperatures. The fact is that the energy density on the current processors (well, at least my overclocked one) is somewhat above 1 MW/m<sup>2</sup>. Water is very well able to get rid of that amount of heat, water/methanol preserves most of the good heat transfer properties of water, but I have seen everyone trying to cool a processor with direct expansion fail. The surface area of the evaporator is usually way too small to be efficient. The heat generating surface is about 1 cm<sup>2</sup>, and it produces about 100W in my setup.



I have made a 'sneak preview' <a href="http://www.arcobel.nl/~dabit/zooi/heatex3_preview.jpg" target="_top">picture</a> of my new coaxial heat exchanger. It uses 9mm tubing for the evap, 22mm tubing for the shell. Copper fins are attached to the evap pipe to center the evap pipe into the shell and increase contact area.

mr-cool
20-05-2002, 08:26 PM
Hi DaBit

The solution to your problem using a LST compressor and a TEV!!!

Fit a low pressure switch on the low side of your system and set it so that the evaporator is clear of ice (this setting will depend of your choice of refrigerant) before the compressor turns on again, you can play around with the pressure settings untill you are getting sufficiant cooling and a long enough off time for the system to equalise

NOTE....I have NOT read all the threads so if someone has already state this .........sorry

Kind Regards

Stephen

Keep It COOL!!!!!!!!!!!!!

PS Anyone know how to change the the Apprentice engineer bit that was about **** years ago but I must admit I still learn something new every day

DaBit
20-05-2002, 10:14 PM
I thank you for your answer, but reading the rest of the replies might help in understanding what I want to accomplish.

Gary
21-05-2002, 12:17 AM
I doubt that the compressor I am using right now will be able to overload the condenser.

The question is not whether the compressor will overload the condenser, but whether the new evaporator will overload the compressor at the start of pulldown. The MOP TXV is designed to prevent this.


If the compressor is lightly loaded, I would like to use that situation to drop the evaporator temperature.


I suspect the new evaporator will do this.


I have made a 'sneak preview' picture of my new coaxial heat exchanger.

Where is it?

Dan
21-05-2002, 01:33 AM
Gary, the word "picture" was a link.:)

DaBit
21-05-2002, 09:03 AM
The question is not whether the compressor will overload the condenser, but whether the new evaporator will overload the compressor at the start of pulldown. The MOP TXV is designed to prevent this.

I am still trying to obtain a TEV, preferrably with MOP. The MOP will be set too high for the compressor (which highest allowable evaporation temperature is -10 &deg;C), but R134a valves with MOP point below 0 &deg;C are rare.

I doubt this will cause much trouble, since during startup with capillary tubing the comressor will be overloaded also.


Where is it?

Ah, well, I'll post another one. This picture is of bad quality since my digicam cannot cope too well with low light environments combined with a small diaphragm:

http://www.arcobel.nl/~dabit/zooi/heatex3_preview2.jpg

The previous pic can be found here: http://www.arcobel.nl/~dabit/zooi/heatex3_preview.jpg, in case you missed it. It shows a detail of the inside of the tube.

Gary
22-05-2002, 01:46 AM
I am still trying to obtain a TEV, preferrably with MOP. The MOP will be set too high for the compressor (which highest allowable evaporation temperature is -10 °C), but R134a valves with MOP point below 0 °C are rare.


The MOP should be above 0 °C. About 10 °C would do nicely. But as you said, it probably isn't necessary.

The new evap looks good so far. :)

DaBit
22-05-2002, 09:05 AM
The MOP should be above 0 °C. About 10 °C would do nicely. But as you said, it probably isn't necessary.

Question: if the compressor is designed to run with evaporating temperature between -10 &deg;C and -30&deg;C, why is it that the MOP should be above 10 &deg;C? It would seem logical to me that the MOP should be -10 &deg;C or lower to prevent compressor overload in any circumstances.

Not that it is a problem, but I am just curious...


The new evap looks good so far.

Thanks :)
Computational Fluid Dynamics simulations conducted at the University of Delft showed that the heat exchange rate between copper and water with a speed of 1m/s (I am running at 0.6m/s) is about 4000-5000W/m<sup>2</sup>/K. So, the approximately 0.1 m<sup>2</sup> area would guarantee a low TD between copper and liquid.

When I set the heat transfer coefficient for evaporating R134a to 2000W/m<sup>2</sup>/K (just an educated guess), and when I incorporate some losses over the copper tubing, this evaporator will give me 2-3K temp difference between liquid and evaporating refrigerant. I hope these quick calculations are correct :)

Gary
22-05-2002, 01:48 PM
Question: if the compressor is designed to run with evaporating temperature between -10 °C and -30°C, why is it that the MOP should be above 10 °C? It would seem logical to me that the MOP should be -10 °C or lower to prevent compressor overload in any circumstances.


The compressor is designed to pull a freezer down (to it's normal operating range) from room temperature. But that's with a cap tube, which is self limiting. The TXV is capable of opening up and delivering a much heavier load.

On the other hand, it is doubtful that the small volume of water mixture contains enough BTU's to cause an overload. The MOP is simply a precaution, and the setting should be somewhere between room temp and operating temp.

If MOP is not available, we can use other strategies (if needed) to limit load. For example, we could delay the start of the water pump.

The idea is to deliver as heavy a load as possible for quick pulldown, but without overload. Something just short of a hernia. :)

DaBit
22-05-2002, 03:34 PM
On the other hand, it is doubtful that the small volume of water mixture contains enough BTU's to cause an overload. The MOP is simply a precaution, and the setting should be somewhere between room temp and operating temp.

How well are compressors able to handle a temporary overload? After all, to change the temperature of the coolant from initial condition (say: 25 &deg;C) to operating conditions (say: -20 &deg;C) the compressor needs to pull 342kJ (ca. 325 BTU's) of energy out of the coolant (2kg * 3800J/kg/K for water/methanol * 45 degrees). Not a terrible amount of heat, but it is still a small compressor.

By the way: with the new evap this number decreases since less coolant is contained in the system.


For the curious: a guy I know is conducting computational fluid dynamics simulations of the waterjackets we use in cooling our processors, using a software package named ProCAST.

One of the victims chosen for simulation is this waterjacket:
http://www.hyperreal.org/~geert/test/koel009.gif

This one of the resulting pictures (red = 41 &deg;C, purple = 22 &deg;C):
http://www.hyperreal.org/~geert/test/koelblok_simulatie013.gif

The vectors display heat flux. The simulation is still running, and a stable situation has not yet been reached. So, don't look at the temperatures too much, they are not right yet. Heat flux from the 'processor' into the waterjacket is 1MW/m<sup>2</sup>, which is only a tiny bit more than what an moderately overclocked real-world processor would produce. Heavy overclocked processors would produce even more heat.

On this image, you can see the boundary between copper and water which limits heat flux:
http://www.hyperreal.org/~geert/test/watertemp_koel009_03.gif

Too bad he doesn't have the parameters for evaporating R134a :(

Gary
18-06-2002, 02:23 PM
How is your project coming along, DaBit?

DaBit
18-06-2002, 02:27 PM
I was just preparing another post. Hold on...