PDA

View Full Version : Refrigeration system proposal; comments please.



DaBit
24-12-2002, 06:48 PM
I am still busy finding the optimal system configuration for a liquid chiller to be used in the PC cooling/overclocking realm. (See also my not yet finished website http://www.icecoldcomputing.com
For the people who have been there before, there is nothing really new.)

The two problems faced are:
- Very low required capacity (50-150W).
- Large difference between lowest load and highest load (~800%).

I have tried both a standard TEV system and a capillary tube for expansion, but both have their problems.

When using a TEV I am facing the problem of a too low load for the TEV to work properly.
With a captube the lowest load causes liquid floodback while the highest load causes excessive superheat.

Hot gas bypass could help, but this reduces the lower temperatue limit. I do nat have product to cool down to -10 °C, for example, but I just want the lowest temperature possible given a certain load.

Now, I had an idea of using the fact that a captube passes less refrigerant when fed with a liquid/vapour mixture, and I posted a topic about that before (http://www.refrigeration-engineer.com/chat/showthread.php?s=&threadid=874), but the answers given there were not satisfactory (or I am a hard student, also possible).

Therefore I have drawn an image this time (excuse for the bad quality; I only had Xfig for drawing).
http://www.arcobel.nl/~dabit/zooi/refsystem.gif

A few moments during operation:

- System startup.
The compressor builds up pressure, the condensed liquid is forming a solid liquid seal at the capillary
tube entrance. Condensing temperature rises, capillary inlet pressure rises, massflow rises.
All refrigerant boils off in the evaporator, chilling the coolant with maximal power. This is
the normal behaviour for a capillary tube system.

- System running with low load applied.
Due to the reduced heat load and low suction pressures, condensing temperature is low, capillary
inlet pressure is low, massflow through the capillary is low. However, *not* low enough to boil off
all the refrigerant in the evaporator. Liquid is trapped into the accumulator, effectively removing it from the
circulating charge. Due to the reduced circulating charge, the capillary tube is fed with
a liquid/vapour mixture, reducing it's capacity even further until the point is reached where
no liquid is returned from the evaporator, and everything is in balance.

- System load increases.
Due to the increased system load, the coolant rises in temperature. The evaporator starves due
to the increased load. Liquid in the accumulator starts boiling off since it is heated with
coolant. This increases the circulating charge in the system, reducing vapour content in the
liquid/vapour mixture fed to the capillary tube, and thus increasing it's capacity. Without
heater coil it would take a long time for the liquid in the accumulator to boil off, giving a
slow reaction to increased load.

- System is powered down.
The small capillary tube from the accumulator to the compressor crankcase slowly returns the oil-rich
refrigerant mixture still left in the accumulator to the crankcase. This should not harm the
compressor, since this is the normal situation for capillary tube systems with the compressor
below the evaporator and a top-to-bottom refrigerant flow, such as household refrigerators.
The massflow through this capillary tube should be very low to reduce its's impact on system
performance.

The Prof taked about this being balancing 4 basketballs on top of each other, but why exactly?

herefishy
27-12-2002, 08:20 PM
I had started a long post, visited DaBit's website, re-edited it several times, then......

I decided I had nothing to say.

Prof Sporlan
28-12-2002, 12:00 AM
Fundamental problem here is you cannot allow refrigerant to trap in an accumulator, or the system low side, or be stored in a receiver (which isn't the problem here) when employing a capillary tube. Systems using a capillary tube require a critical refrigerant charge, which also requires refrigerant not trap or store itself somewhere in the system.

This system, as designed, requires refrigerant in excess of the critical amount. In fact, storing excess refrigerant in the acculumator was done with the Westinghouse Hi-Re-Li system, which used a variation of a thermostatic expansion valve that controlled subcooling from the condenser.

Charging the system so that refrigerant is always trapped in the accumulator, and using a thermostatic expansion valve... the system probably could be made to work ok... if there was a reasonable load to work with... :)

frank
28-12-2002, 03:04 PM
This high load/low load problem would be best served with an electronic expansion valve and an inverter controlled compressor but I expect this is way out of the price range for this diy project :)

Frank

Andy
28-12-2002, 06:35 PM
Hi,:)
If I was designing a system for the above I would fit a D/C compressor such as used in caravans ect, with a simple D/C speed controller (possibly swithching resistor banks) and two expansions valves with MOP function one sized at 75% duty, one at 25% duty with LLSV'S triggered on the glycol return temp on a secondary refrigerant circuit.
Initiation of the LLSV'S could be progressive, along with compressors speed, with basically three steps as follows.
[list=a]
Lowest glycol return temp setting, bring in 25% TEV LLSV and slowest comp speed
Next lowest setting, cut out 25% and bring in 75%, run comp at 50% speed
Next lowest temp setting, increase comp speed to 75%
Highest temp setting, bring in 25% valve with 75% valve and increase comp speed to 100%
[/list=a]

It may be also necessary to decrease the evap pump speed also , but I would try the above first.

It may not be the right way to go but it is one I would try if it was my problem.
Regards. Andy.

Andy
28-12-2002, 06:41 PM
Hi, Professor:)
you talk about the Westinghouse design that uses a TEV controlling off condenser subcooling. Star uses (or used to use) a design where the subcooling off the LPR controlled the switching of the AKVA (danfoss electronic expansion valve) this was how the refrigerant flow through the system was controlled without floats, much like a HP flaot system. Could something like a mini LPR work in a situation such as this?
Regards. Andy.

Prof Sporlan
28-12-2002, 09:32 PM
Using an LPA and EEV is an interesting approach to control refrigerant flow via subcooling. With DaBit's system, however, we're trying to control a refrigerant flow to satisfy a 100 watt load, which works out to about a 0.1 lb/min flow rate with R-134a. Controlling this type of flow rate would be troublesome for any TEV or EEV.

Andy
28-12-2002, 10:47 PM
Hi, Professor:)
what about varying the system capacity, with a D/C compressor with three steps. 33% 66% and 100% speed. Adding two capillaries one sized at 33%, one at 66%,with solinoide control switching on chiller on temp as my previous post.
I will check up on TEV'S for that duty, but as you say, might be below a TEV capacity.
Regards. Andy.

Andy
28-12-2002, 11:09 PM
Hi,:)
O.K. I've got my head around it now. Load is very small, typically about 25% of the smallest load a TEV is made for. I checked with selection software and found an AKV10-1 would satisfy this capacity at of 0.1kW at 11.9% open, not a very satisfactory match.
Better built a 400 or 500 watt system with a buffer tank for the glycol, run the tank down to design temp then switch off. All components can be selected correctly and you have no problems with liquid flooding back and various other system problems.
OK you you system is not alway running at the ultimate low temp, but by adding a glycol secondary you are not getting the lowest available temps anyway.
It may also be possible to false load the system with another haet sourse such as running water through a heat exchanger.
The other possiblity is a cylinder of CO2 a requlator and a wooden box with some slots cut in in, you could pass the glycol through the box in a coil, that would certainly give low temps at low loads without any of the problems associated with mechanical refrigeration.
Regards. Andy:confused:

DaBit
29-12-2002, 03:30 PM
Originally posted by Prof Sporlan
Fundamental problem here is you cannot allow refrigerant to trap in an accumulator, or the system low side, or be stored in a receiver (which isn't the problem here) when employing a capillary tube. Systems using a capillary tube require a critical refrigerant charge, which also requires refrigerant not trap or store itself somewhere in the system.

That is correct, but what happens when a capillary tube system is undercharged? When undercharging a capillary tube system (or any fixed orifice system), the evaporator pressure drops, and the amount of liquid injected in the evaporator is decreased.

With a normal A/C or fridge this leads to partial use of the evaporator (due to the low amount of refrigerant injected) and high superheat at the evaporator outlet. The system is not performing as it should, and this is, of course, not a desirable situation.

But I am in the situation where the load decreases, making up for the effects of undercharging. And undercharging is what we do when we take refrigerant out of the circulating charge.

I have seen this effect with the capillary tube system described on my web site (Second R134a system (http://www.icecoldcomputing.com/cgi-bin/vdocview.pl?section=phasechange&spath=pc2_1&language=english)). A certain charge X gave me good performance during pulldown, but at an evaporator temperature of ~-8 °C, the superheat at the compressor inlet fell below 3K. Reducing the charge improved this situation, allowing evaporator temperature to drop while maintaining superheat. But this reduced charge was not able to cope well with higher loads.

This effect made me believe the accumulator idea would work.


Charging the system so that refrigerant is always trapped in the accumulator, and using a thermostatic expansion valve... the system probably could be made to work ok... if there was a reasonable load to work with... :)

Yes, but then the 'classic' setup of compressor->condenser->receiver->TEV->evaporator->compressor would work fine either.

The reasonable load is the problem. I agree :)


Originally posted by Andy
[B]Hi,:)
If I was designing a system for the above I would fit a D/C compressor such as used in caravans ect, with a simple D/C speed controller (possibly swithching resistor banks) and two expansions valves with MOP function one sized at 75% duty, one at 25% duty with LLSV'S triggered on the glycol return temp on a secondary refrigerant circuit.

The electronics are not a problem since I am well capable of designing a speed controller. The problem is the DC compressor. They are so weak that I am better off using a normal 1/4HP 220VAC compressor and calculate the capillary and charge for the lowest possible load condition.
Besides being weak, this is also a costly solution. A powerful DC source is needed, a speed controller, and a DC compressor.

The second problem is the TEV's. Lowest capacity is still too high.


Originally posted by Andy
[B]Hi,:)
O.K. I've got my head around it now. Load is very small, typically about 25% of the smallest load a TEV is made for. I checked with selection software and found an AKV10-1 would satisfy this capacity at of 0.1kW at 11.9% open, not a very satisfactory match.


Better built a 400 or 500 watt system with a buffer tank for the glycol, run the tank down to design temp then switch off. All components can be selected correctly and you have no problems with liquid flooding back and various other system problems.
OK you you system is not alway running at the ultimate low temp, but by adding a glycol secondary you are not getting the lowest available temps anyway.

No, but designing a system to feed multiple evaporators of which some are only loaded with 4-5 Watts is a real pain in the ass. Using a secondary water/methanol or glycol loop to distribute the cold is way easier.

I agree that building a 400-500W system which uses compressor cycling is probably the easiest solution, but this is not comfortable to sit next to, it adds a lot of weight, and it is not really the solution I am looking for. It is more an alternative when everything else fails.


It may also be possible to false load the system with another haet sourse such as running water through a heat exchanger.

Or a resistive heater which is controlled by water/methanol temperature. A waste of energy, but this is not my main concern.


The other possiblity is a cylinder of CO2 a requlator and a wooden box with some slots cut in in, you could pass the glycol through the box in a coil, that would certainly give low temps at low loads without any of the problems associated with mechanical refrigeration.

This would cost a LOT of CO2, and is not really a solution.

Andy
29-12-2002, 05:13 PM
Hi, DaBit:)
what about using two capillaries to feed the evaporator, one of these capillaries could be controlled by a liquid line solinoide valve. This solinoide could be triggered to operate on high secondary refrigerant temp, cutting out when the secondary drops to a given point where the superheat at the compressor is as low as you want to go (or simply by switching on suction temp by means of a simple domestic fridge stat).
You could develop a system with a suction accumulator and boil off coil (this is what a Low Pressure Receiver is) but for a small system such as this it would be time consumming.

Control of a LPR would be quite simple.

Two capillaries, one operational all the time the other by a solinoide valve controlled on the liquid off the LPR heat exchanger temperature (ideally this would be say a 20K diff, but it could be a trigger temperature of say +5 deg C) the diff is a measurement accross the subcooler, how many degrees of subcooling.

I have often used this for Large Low Pressure Receiver systems, but something as small as this is completely different.

Good Luck, Regards. Andy:)

DaBit
30-12-2002, 12:08 PM
Originally posted by Andy
[B]Hi, DaBit:)
what about using two capillaries to feed the evaporator, one of these capillaries could be controlled by a liquid line solinoide valve. This solinoide could be triggered to operate on high secondary refrigerant temp, cutting out when the secondary drops to a given point where the superheat at the compressor is as low as you want to go (or simply by switching on suction temp by means of a simple domestic fridge stat).

I have been thinking about that solution too. But my idea was to make it a binary system using capillaries with half the restriction each time.

Thus, say we make a 3-bit system. We use three capillary tubes where the smallest one passes a certain mass X at given conditions. The second tube passes mass 2X, the third passes mass 4X. This would allow me to dial in a massflow between 0X and 7X. The 0X setting would be useful to prevent refrigerant condensation in the evaporator during off cycles.

Switching the correct solenoids could be done using a simple microcontroller (I love the Atmel AVR series 8-bit micros) based on compressor suction line temperature near the compressor. If the temperature near the compressor suction connector becomes too low (sign of liquid returning), the mass flow is reduced. An even better method would be to sense the pressure and temperature at the evaporator outlet, and base the massflow on the superheat calculated from these two variables. However, this drives up cost quite a bit due to the required pressure transducer and signal conditioning electronics.

The main problem with this solution is the distribution of refrigerant over the solenoids/capillary tubes. Using normal piping would give me an inbalanced refrigerant distribution. Maybe this could be solved by mounting the solenoids at the other end of the captubes, the low pressure side.


You could develop a system with a suction accumulator and boil off coil (this is what a Low Pressure Receiver is) but for a small system such as this it would be time consumming.

I should get myself some good books about this stuff instead of re-inventing the wheel each time ;)

But why would using an LPR be time-consuming? I have build the LPR already since I do not like to accidentally feed the compressor with liquid, which is something that even happens with the current TEV system during low load conditions. I want to install the LPR anyway to prevent damage.

My current LPR consists of a U-shape made out of 7/8" pipe, with the inlet and outlet connecting to the different legs of the U. For pressure equalisation between the legs, the legs of the U are connected at the top with a piece of 3/8" tubing. A piece of 1/4" tubing fed through the U is my boil-off coil. And a piece of capillary tube mounted at the bottom of the U slowly drains off liquid and returns it to the compressor's oil sump/carter (which word is correct? English is still not my native language).

Not exactly a difficult thing to build, and the U-shape definitely prevents liquid coming from the evaporator spraying into the line going to the compressor suction connector.



Control of a LPR would be quite simple.

Two capillaries, one operational all the time the other by a solinoide valve controlled on the liquid off the LPR heat exchanger temperature (ideally this would be say a 20K diff, but it could be a trigger temperature of say +5 deg C) the diff is a measurement accross the subcooler, how many degrees of subcooling.

Pretty good idea, but this scheme boils off refrigerant to perform subcooling on the liquid, while I'd rather use the cold suction gas to perform subcooling since that improves performance. Using a part of the secondary coolant flow to heat the boil-off coil is a better idea since that makes the boiling off of the refrigerant in the LPR dependant on the actual system load. Whether to switch on the extra capillary tube(s) could be measured elsewhere.


I have often used this for Large Low Pressure Receiver systems, but something as small as this is completely different.

I know, and it is quite hard to get right too. The low load/huge load variation/lowest obtainable secondary coolant temperature goals are hard to meet.

Thanks a lot for the input.

Andy
30-12-2002, 12:42 PM
Hi, DaBit:)
your binary system sounds good, but it does go to lenghts that even valve manufacturers don't normally go to. Think of an electronic expansion valve, which will give very close control of superheat even on NH3 which is notoriously hard to control on DX. An electronic expansion valve is just a direct action (except for the big ones) solinoide with an oriface seat. We used to use them on LPR's now we use a direct action solinoide with an oriface plate after it. If electronics is your thing I would build a EEV system, all you need is the controller, which you could build yourself (or a small PLC with an analog input) a pressure transducer and a temperature sensor. The valve could be just a solinoide with a capillary after it.
[QUOTE]Pretty good idea, but this scheme boils off refrigerant to perform subcooling on the liquid, while I'd rather use the cold suction gas to perform subcooling since that improves performance.

The older LPR's used exactly that with the boil of coil attached to the suction to the compressor from the LPR. The reason this was discontinued was to reduce the discharge temperatures on the compressors and to reduce the system refrigerant charge.
Remember by fully flooding the eveporator you are increasing the evaporator capacity by 20 to 30% also you are reducing the pressure drop in the suction line to the LPR (called wet return), the vapours entering the compressor are again much denser if the boil off coil is in the bottom of the LPR, allowing the compressor to displace more refrigerant, basically increasing the compressor capacity.
I personally would use a small LPR with a boil off coil, with one solinoide switched on subcooling, with another say 25% capillary feeding all the time to induce flow, this is the simplest and cheapest way I could think of performing the necessary function, but again it's you baby and it looks like you have done a lot of the ground work on your idea already.

Best a luck, Regards. Andy:) [QUOTE]

DaBit
01-01-2003, 04:42 PM
Originally posted by Andy
[B]Hi, DaBit:)
your binary system sounds good, but it does go to lenghts that even valve manufacturers don't normally go to.

I know, but that system can easily be done in my garage, while building or modifying a TEV/EEV is not possible with the tools I posess.


If electronics is your thing I would build a EEV system, all you need is the controller, which you could build yourself (or a small PLC with an analog input) a pressure transducer and a temperature sensor. The valve could be just a solinoide with a capillary after it.

And then doing PWM on the solenoid to modulate flow?



The older LPR's used exactly that with the boil of coil attached to the suction to the compressor from the LPR. The reason this was discontinued was to reduce the discharge temperatures on the compressors and to reduce the system refrigerant charge.

I am not having any discharge heat problems. Discharge line stays < 60 &deg;C, so this is not a problem for me. With NH3 of R22 you run into discharge temp problems much earlier than with R134a or eventually R290.


I personally would use a small LPR with a boil off coil, with one solinoide switched on subcooling, with another say 25% capillary feeding all the time to induce flow, this is the simplest and cheapest way I could think of performing the necessary function, but again it's you baby and it looks like you have done a lot of the ground work on your idea already.

I will investigate this further. I have done the ground work, but I do not have a degree in refrigeration technology, but in electronics. This means that I just bother you people to check my ideas. I have had a tremendous benefit from this forum and the people populating it.