PDA

View Full Version : capillary tubes and mass flow when fed with liquid/gas



DaBit
10-10-2002, 03:52 PM
Just a little fundamental question about capillary tubes:

Say, we have a capillary tube which is fed with liquid refrigerant only. It causes a mass flow of 10kg of refrigerant when the pressure drop over the tube is 10 bar.

Now, if we use the same capillary tube, and we maintain the pressure difference over the tube, but now it is fed with gaseous refrigerant. How would the mass flow be affected, and by what (approximate) factor?

The reason I ask this, is that if mass flow drops significantly when the captube is fed with a gas/liquid mixture, this fact can be used to build a system with a captube which will work with varying load. All that is needed for this is a well-insulated accumulator after the receiver which can hold the total refrigerant charge present in the system. As soon as load on the evaporator drops, liquid refrigerant accumulates in the accumulator, causing the capillary tube to starve, and thus reducing it's capacity to match the load. When load increases again, evaporator pressure rises, starvation of the captube decreases, and captube capacity increases. At least, that is what I think what happens.

It is just an idea. I have not yet found the ultimate metering solution for a small (25-100W) and highly varying (>400%) load like a microprocessor. Those chips spit out only a few Watts when running idle, but running calculation intensive tasks with highly optimized math cores such as simulation software and games can raise heat production to many times the idle heat production. TEV's have problems with the lowest load situation, captubes have problems with the varying load. So I keep searching for better methods to cope with this situation.

superheat
10-10-2002, 04:23 PM
On a cap tube, gas does not pass easily. The condensing pressure will rise and evap pressure will fall untill the cap tube lets the same amount of ***** pass as the compressor pumps.
An orifice system will let gas pass easier.
I do not know about your application. The important thing to remember is keeping the compressor cool. Your superheat must be in a certian range for the compressor to operate correctly for a reasonable length of time.
I am afraid if you try to modulate the load by undercharging a system, you will ruin the compressor. Modulate your load with fan speed and hot gas bypass. Those are the tried and true methods.

DaBit
10-10-2002, 04:37 PM
Originally posted by superheat
[B]On a cap tube, gas does not pass easily. The condensing pressure will rise and evap pressure will fall untill the cap tube lets the same amount of ***** pass as the compressor pumps.


A falling evap pressure is exactly what I want, since that lowers evaporation temperature.
But is it true that condensation pressure rises? As soon as liquid gets trapped into the accumulator, it is not available on the high side of the system. It also means that evaporator pressure falls, and refrigerant mass pumped by the compressor falls. Rising condensation pressure means condensation, and thus liquid ahead of the captube.




I am afraid if you try to modulate the load by undercharging a system, you will ruin the compressor. Modulate your load with fan speed and hot gas bypass. Those are the tried and true methods.

Hot gas bypass limits low temperature. Fan speed control is indeed able to adapt the system to load a bit, but not that much. On my previous captube system I used it to reduce noise.

superheat
10-10-2002, 06:53 PM
You need liquid ***** to boil inside the evap to absorb heat. You can have a lower evap temp and still get less heat removal.
I would say that you need a large heat sink to absorb the extra capacity of the evap and let the compressor cycle. You want to have a time delay to prevent excess cycling though.

I have to admit, it has been years since I have seen the inside of a computer. Things were simpler back then. You are the engineer on this job. We give you ideas. You figure out how to make them work. I think you need to read a refrigeration book before you get to crazy though.

Gary
11-10-2002, 09:08 AM
A falling evap pressure is exactly what I want, since that lowers evaporation temperature.


Not exactly. What you want is lower chip temperature. This means transferring more heat from the chip to the refrigerant, resulting in lower chip temperature and higher evaporating temperature.

Refrigeration systems involve a series of heat transfer relationships. It is a mistake to focus on single point readings. Low evaporating temperature, by itself, doesn't tell you much. You need to know why it is low.

For example, if the chip temperature is higher than it should be and the evaporating temperature is lower, then there is insufficient heat transfer between them. It is the relationship between the two temperatures which tells you what you need to know.

And to fully evaluate performance, a full set of temperature readings is needed. Various temperatures are then compared to each other in order to analyze heat transfer relationships throughout the system.

Low side:

Evap air/water in temp
Evap air/water out temp
SST or low side pressure
Suction line temp near compressor

High side:

Cond air/water in temp
Cond air/water out temp
SCT or high side pressure
Liquid line temp near condenser

These are the minimum readings needed.

DaBit
11-10-2002, 10:49 AM
Originally posted by Gary
Not exactly. What you want is lower chip temperature. This means transferring more heat from the chip to the refrigerant, resulting in lower chip temperature and higher evaporating temperature.

There is no such thing as 'more heat'. The amount of heat applied to the system is fixed, and thus a more efficient heat transfer should result in lower evaporation temperatures since the TD loss between CPU and refrigerant is lower.


Refrigeration systems involve a series of heat transfer relationships. It is a mistake to focus on single point readings. Low evaporating temperature, by itself, doesn't tell you much. You need to know why it is low.

I agree with you. There are two resons why evaporation temperature would be low in the captube/accumulator system:

- Not enough refrigerant fed to the evaporator. This starves the evaporator, decreases suction pressure, and increases TD between load and refrigerant

- Enough refrigerant is fed to the system to pass liquid to the accumulator. Since this refrigerant is not available to the compressor, the captube will get a mixture with more gas and less liquid. This reduces it's mass flow, and therefore evaporator pressure, up to the point where captube mass flow matches heat input.



And to fully evaluate performance, a full set of temperature readings is needed. Various temperatures are then compared to each other in order to analyze heat transfer relationships throughout the system.


Don't worry, I will bug you soon with all these reading taken from my current TEV system, to see hoe I can optimize it :)
But for this imaginary captube system I do not have readings. I just wondered whether it would work or not :)

BTW: when interested in pictures of the liquid chiller, a few pictures can be found here (forum is in dutch. The first and second page contain pics. I will post pics here as soon as I have loaded the system and have taken temp readings. This makes it a bit more intereseting for you guys):
http://gathering.tweakers.net/forum/list_messages/615287

Gary
11-10-2002, 01:20 PM
There is no such thing as 'more heat'. The amount of heat applied to the system is fixed, and thus a more efficient heat transfer should result in lower evaporation temperatures since the TD loss between CPU and refrigerant is lower.


Actually, the opposite is true.


I agree with you. There are two resons why evaporation temperature would be low in the captube/accumulator system:

- Not enough refrigerant fed to the evaporator. This starves the evaporator, decreases suction pressure, and increases TD between load and refrigerant

- Enough refrigerant is fed to the system to pass liquid to the accumulator. Since this refrigerant is not available to the compressor, the captube will get a mixture with more gas and less liquid. This reduces it's mass flow, and therefore evaporator pressure, up to the point where captube mass flow matches heat input.


These two reasons are in fact the same. Starved evaporator.

I'm not sure where you are going with the cap tube/high side accumulator idea, or why you feel there is a need to match the system to the varying load more than is inherently accomplished.

Is there a problem with the CPU getting too cold under low load?

Gary
11-10-2002, 01:36 PM
Hot gas bypass limits low temperature.

All forms of unloading limits low temperature. That is the purpose of regulating the system to match the load.

Gary
11-10-2002, 02:04 PM
BTW: when interested in pictures of the liquid chiller, a few pictures can be found here (forum is in dutch. The first and second page contain pics. I will post pics here as soon as I have loaded the system and have taken temp readings. This makes it a bit more intereseting for you guys):
http://gathering.tweakers.net/forum...messages/615287


In the last picture, it appears that you have the TEV sensing bulb bundled together with the suction and liquid line. The sensing bulb must sense ONLY the suction line.

DaBit
11-10-2002, 02:30 PM
Originally posted by Gary
Actually, the opposite is true.


Ehm, oops, bad thinking. Of course. I have to stop doing 3 things at the same time.


These two reasons are in fact the same. Starved evaporator.

Why? In the second case the evaporator gets overflooded by the captube, and the liquid coming out of the evaporator gets stored in the accumulator.



I'm not sure where you are going with the cap tube/high side accumulator idea, or why you feel there is a need to match the system to the varying load more than is inherently accomplished.


It was just an idea to use captube starvation to match capacity to load. I want to know whether it works, and when not, why.

The history is this: my first try to chill the CPU used a captube as the metering device. No (low-side) accumulator, no (high-side) receiver.
The problem with this was that I had to adjust captube size and charge to maintain a minimum of 4K superheat close to the compressor inlet when the CPU is running idle and condensing temp is quite high (e.g. because of summer ambient temps). This inherently meant that during a full-load situation superheat rises to about 15-20K. Only a small part of the evaporator is used now, and TD between CPU and refrigerant rises.

This problem is caused by the fairly invariable amount of refrigerant injected by the captube: it's resistance is fixed, so you can only do capacity control by varying condensing temperature and thus the pressure difference over the captube. This method of capacity control is simply unable to cope with a large load variation.

Now I am using a TEV. Way easier to work with (captube sizing is a pain in the ass), and the TEV takes care of load variations. But TEV's have problems with very low loads, so the TEV is useable with my water chiller (I always have energy leaking through the insulation, the pump adds energy to the coolant, and I am cooling more than the CPU only), but not for direct expansion above the CPU.

So far for the standard solutions, except capacity control systems like hot-gas bypass, adjusting pumping capacity of the compressor by varying motor RPM, or just cycling the compressor.
These capacity control systems reduce capacity to protect the compressor, but they won't help to minimize temperature at low load situations.

In short:
- A captube system cannot cope with the load variations
- A TEV system cannot cope with the low load.

My idea was that by adding a accumulator on the low side, it would be possible to cope with both low load and load variations by using the fact that massflow through the captube would become smaller when it is fed with a gas/liquid mixture. When load is low, the unevaporated liquid refrigerant (caused by the low load situation) gets trapped in the accumulator, and would cause a shift to more gas/less liquid at the captube entrance, and thus a lower refrigerant feed to the evaporator. The system should stabilize to the point where no liquid gets trapped into the accumulator. At this point the mass flow through the captube matches the low load. At least, that is what I hope ;)


Is there a problem with the CPU getting too cold under low load?

No. The colder the better. At least until we get near the -60 °C (-76 °F) mark; below this temperature the CPU fails. But -60 °C is not a realistic temperature for DIY-ers like me.

DaBit
11-10-2002, 02:40 PM
Originally posted by Gary


In the last picture, it appears that you have the TEV sensing bulb bundled together with the suction and liquid line. The sensing bulb must sense ONLY the suction line.

It does only sense the suction line. At the place where the TEV bulb is mounted, the liquid line is insulated so no thermal contact between those two exist.

It is mounted after the suction gas<-> liquid line heat exchanger, though. This way I should be able to protect the compressor from flooding while maintaining 3-4K temp difference between coolant and evaporating refrigerant.

Prof Sporlan
12-10-2002, 02:57 AM
My idea was that by adding a accumulator on the low side, it would be possible to cope with both low load and load variations by using the fact that massflow through the captube would become smaller when it is fed with a gas/liquid mixture. When load is low, the unevaporated liquid refrigerant (caused by the low load situation) gets trapped in the accumulator, and would cause a shift to more gas/less liquid at the captube entrance, and thus a lower refrigerant feed to the evaporator.
This concept is valid, though not likely practical to implement. Getting a the correct two-phase flow mixture ahead of the capillary tube for low load operation by simply allowing refrigerant to trap in the low side in the correct amount..... Mmmmmm.... the Prof would liken that to balancing four basketballs on top of one another.... :)

Far simpler to use a TEV and hot gas bypass valve for capacity control. The hot gas bypass valve will limit how low you will be able to run evaporator temperature, but it will allow you to control flow at low load conditions.

Gary
12-10-2002, 11:03 AM
It is mounted after the suction gas<-> liquid line heat exchanger, though. This way I should be able to protect the compressor from flooding while maintaining 3-4K temp difference between coolant and evaporating refrigerant.


Mounting the sensing bulb downstream from the suction/liquid heat exchanger ensures that the evaporator is fully flooded, but makes it very easy to flood the compressor under low load conditions. It may be advantageous to install a second suction/liquid heat exchanger downstream from the sensing bulb in order to protect the compressor.

DaBit
13-10-2002, 12:35 PM
Originally posted by Prof Sporlan
This concept is valid, though not likely practical to implement. Getting a the correct two-phase flow mixture ahead of the capillary tube for low load operation by simply allowing refrigerant to trap in the low side in the correct amount..... Mmmmmm.... the Prof would liken that to balancing four basketballs on top of one another.... :)

And we all know how hard that is :)

But where exactly does this concept fail? The way I explained it, it balances itself. Why does it sound good in theory, but not in practice?


Mounting the sensing bulb downstream from the suction/liquid heat exchanger ensures that the evaporator is fully flooded, but makes it very easy to flood the compressor under low load conditions. It may be advantageous to install a second suction/liquid heat exchanger downstream from the sensing bulb in order to protect the compressor.

I will run the liquid line along the suction line after the bulb, just as you told me a couple of months ago. I first tried it this way to see how hot the discharge line would become. Since this seems to be reasonable, I can improve SG<->LL heat exchange.

DaBit
24-10-2002, 04:27 PM
Does anyone have a guess how the mass flow through a capillary tube is altered when it's feeding is changed from 100% liquid to, say, 50%/50% (by volume) liquid/gas, given the same pressure difference over the tube?

Thus, a capillary tube passes 10kg/h at a given pressure difference X when fed with 100% liquid. How much would it pass when it is fed with 50/50 (or another ratio if it fits better) liquid/gas, given the same pressure difference X?

superheat
24-10-2002, 06:52 PM
I would quess that it passes nearly the same volume, but the density changes so the mass flow rate changes.

In reality that will probably not happen, because the condensing pressure would biuld until mostly liquid passes.
If you design a system to pass gas, you are wasting compressor work. Use a smaller compressor that passes the right amount of liquid. The gas is dead weight. It does nothing but take up space. The higher velocities might even wear out parts faster. Gas and liquid togather would act like a sand blaster.