PDA

View Full Version : R507 in an R134a compressor: system design



DaBit
22-04-2003, 04:48 PM
Hello everyone. This thread is more or less a followup on this (http://www.refrigeration-engineer.com/chat/showthread.php?s=&threadid=1055) thread.

Since my last post in that thread some things have happened. I had a long chat with an airconditioning installation/repairman who was just curious about my adventures. After a few hours of chatting about the difficulties of refrigerating a PC, he gave me a lot of interesting stuff, including a refrigeration manifold with 4 valves, a few solenoids, sight glasses, schraeder valves, expansion valve for R404a/R507, high side pressostate, pipe and 1kg of R507.

I did not have to pay a single dime for it. The only thing he wants in return is that I keep him updated.

Of course, I am very thankful for his help, especially the refrigeration manifold, since now I will finally be able to provide Gary with the required measurements.

Since my current system has it's problems, I will start all over. It's not that much work to redo the piping etc. since I am getting pretty quick in it. The target is: -35 °C / -31F secondary coolant temperature @ 150W load.

I definitely want your comments on the design, and possible improvements.

Here is the first shot at the new chiller design with R507:
http://www.icecoldcomputing.com/misc/R507_syst.gif

Since I will use R507 in a compressor designed for R134a, I have to adjust the operating conditions to fall within the compressor's limits. This means:
- Suction pressure 0.8 .. 2 bar absolute. (R507: -51 °C .. -31 °C / -60F .. -24F)
- Condensing pressure < 16 bar absolute. (R507: 33&deg;C / 91F)

When I keep pressures within these limits, the amount of work to be done for the compressor should be approximately the same as the amount of work to be done for R134a with -10 &deg;C .. -30 &deg;C evaporating and 55 &deg;C .. 60&deg;C condensing.

Since I am shooting for -35 &deg;C / -31F secondary coolant temp, I have to evaporate at approximately -40 &deg;C / -40F. This is within the suction pressure limit.

To keep the condensing pressure within the given limit, I will use a liquid cooled condenser with 4-8L (1-2 gallons) of water and an air-cooled radiator (a heater core from a European car called Opel Ascona). The large amount of water functions as a buffer, and it is sufficient to keep condensing temperature low during startup, when extra heat is dumped into the condenser. When the system is at operation conditions, the amount of generated heat is low enough for the air cooled radiator to get rid of with very little rise in temperature.

I cannot find a minimal suction gas temperature in the Danfoss datasheet, but I can increase suction gas temperature by performing suction gas<->liquid line heat exchange. This seems to be very good for the COP in R507 systems anyway; even more than R134a benefits from it. Can anyone confirm or deny this? I know this 'medal' has two sides, and we discussed the benefits of a SG<->LL HX earlier.

To keep everything safe, I will mount a pressostate (is that the correct word), which will cut out the compressor when condensing pressure reaches 18 bars.

To use a low amount of refrigerant, I will use a receiver just large enough to hold the entire amount of refrigerant in case a pipe becomes clogged. Or course the receiver should not run empty when suction pressure is high.

The expansion valve will be a Danfoss TUAE stainless steel R404a/R507 valve, designed for -60 .. -20 &deg;C (-76F .. -4F) operation, -20 &deg;C MOP, external pressure equalisation. This will protect the compressor a bit during startup.

The refrigerant distributor. I plan to use a few pieces of 0.8mm / .031" I.D. captube to distribute refrigerant over multiple 1/4" evaporator coils. I still need to find out how long I should make those tubes, but a rough guess would be 30cm / 1ft. I just need a bit of pressure drop to obtain a decent distribution of refrigerant, but how much pressure drop I need: I don't know.

In the evaporator, I will try to cram as much 1/4" pipe into the (copper) shell as possible, divided into 3-4 circuits. 12 meters / 36ft must be possible, maybe even more. My current evaporator uses 3 meters of 3/8" pipe, and Gary told me to use at least 4 times that amount to minimize the TD between refrigerant and secondary coolant. Translated to 1/4" pipe this is even a longer piece of pipe to obtain the same amount of surface area.

Then, there is the secondary coolant circuit. This circuit uses a water/methanol mixture and a small pump to minimize the pump motor heat dumped into the secondary coolant.

Please note that theoretically the circulation direction of the secondary coolant is wrong (not counterflow), but in practice I need a bit of head above the pump's inlet to keep it running smoothly at such low temps. And I need the refrigerant to flow from top to bottom.

Comments, anyone? I would really appreciate tips on the evaporator and refrigerant distributor, and any general tips.

Prof Sporlan
22-04-2003, 07:10 PM
The refrigerant distributor. I plan to use a few pieces of 0.8mm / .031" I.D. captube to distribute refrigerant over multiple 1/4" evaporator coils. I still need to find out how long I should make those tubes, but a rough guess would be 30cm / 1ft. I just need a bit of pressure drop to obtain a decent distribution of refrigerant, but how much pressure drop I need: I don't know.
Since the Prof has done a fair amount of work modeling distributor tube pressure drop, he should be able to help in this area... :)

When sizing a conventional refrigerant distributor, it is a good idea to size the tubes such that a pressure drop of about 10 psi is created which helps distribute the flow. Both 3/16" OD and 1/4" OD tubing are commonly used for distributor tubes, though with small capacity refrigeration units, one normally does not get much pressure drop with this size tubing.

In DaBit's situation, if we assume a -40°F evaporator, 100°F liquid entering the TEV, an R-507 refrigerant flow rate equivalent to 150 watts, and let's assume we have 3 circuits each being fed by 12" of 0.031" ID cap tubing. The Prof's model would predict a 70 psi pressure drop across this tube. This is much higher than desired. One must keep in mind that flow leaving the TEV will be two-phase, and that is why cap tubing isn't used as distributor tubes unless it is also acting as the expansion device. Using cap tubing with a 0.064" ID, however, should work ok in the above scenario, though one would want to verify pressure drop under the high load condition.

Andy
22-04-2003, 09:43 PM
Hi Dabit,
Few things I have seen that would concern me.

1/ 16 barg head pressure, not sure that is a good idea, what about a cascade system to limit you head pressure, or at least chilled water fed to the condenser section.

2/ Evaporator design, why not just put a trap in the suction line at the lowest point and connect on the TEV sensing bulb on the horizontal above the IMS fluid level, that is how most coils Iv'e seen have been circuited.

3/ you can buy small distributor heads for say 3/16 distributor pipes danfoss make these, this would aid distribution a bit, for a minimal cost.

All the best on your new build, keep us posted.
Regards. Andy:)

DaBit
23-04-2003, 10:13 AM
Originally posted by Prof Sporlan
In DaBit's situation, if we assume a -40°F evaporator, 100°F liquid entering the TEV, an R-507 refrigerant flow rate equivalent to 150 watts, and let's assume we have 3 circuits each being fed by 12" of 0.031" ID cap tubing. The Prof's model would predict a 70 psi pressure drop across this tube.

Prof, can you enlighten me in the model used or give me a link to a descriptive .PDF? I am very interested in a model which predicts mass flow through a tube, especially in the two-phase scenario.


This is much higher than desired. One must keep in mind that flow leaving the TEV will be two-phase, and that is why cap tubing isn't used as distributor tubes unless it is also acting as the expansion device.

OK, got that. But as always I am curious:

You once told me that the capacity of the TEV is, amongst other variables, dependant on the square root of the pressure drop over the TEV.

Let's assume a 15-bar pressure drop, and do some math:
- Nominal TEV capacity, orifice #0 @ 15 bar / 217 psi pressure drop = ~450W / 0.12 tons.

Now, pressure drop over the TEV with the captube distributor would become 15 - 4.8 bar = 10.2 bar (~150psi). This would reduce TEV capacity to 450W * sqrt(10.2 / 15) = 371W / 0.10 tons.

In my case, I can allow this pressure drop since I am asking for 150W. I can even envision that the TEV can control the mass flow better since a larger pin stroke per Watt of refrigeration is obtained this way.

Wouldn't the 70psi pressure drop give a better distribution of the refrigerant mass flow? It would limit pulldown speed, but it would also limit heat dumped into the condenser during pulldown.

In my profession, electronics, when I am designing a power amplifier and I need multiple bipolar transistors to distribute the load, I also need distribution resistors. And the larger I make them, the better the load gets balanced, though there is a point of diminishing return.

Or, does a TEV need a certain pressure drop to function correctly? Based on looking at the design of a simple TEV like I am using, the TEV is a proportional controller without integrating of differentiating terms in it, and I cannot find a reason why a lower pressure drop over the TEV would impact anything but full-open capacity.


Using cap tubing with a 0.064" ID, however, should work ok in the above scenario, though one would want to verify pressure drop under the high load condition. [/B]

I think I have some lying around. In case I do not, can I also use shorter pieces, eventually multiple in parallel? The shortest I can use is approximately 12cm / 5"



Originally posted by Andy
Hi Dabit,
Few things I have seen that would concern me.

1/ 16 barg head pressure, not sure that is a good idea, what about a cascade system to limit you head pressure, or at least chilled water fed to the condenser section.

Head pressure would be 16 bara / 15 barg max. The NL11F is allowed to condense R134a at 60 &deg;C / 140F for a short amount of time, 55 &deg;C / 131F for a longer amount of time. This equals to condensing R507 at 32 &deg;C. Using these pressures assume that 1.5 m/s air flow is directed over the compressor and that the compressor compartment is not warmer that 38 &deg;C / 100F.

In a normal static cooled refrigerator condensing temp also rises to 50-60 &deg;C, and those last for years.

Thus, I am staying (more or less) within the compressor's limits. Then, why would the 16 bara pressure hurt me?

About the cascade: I will eventually build one, but then I need more knowledge and experience to successfully complete that project. At this moment, I simply don't know enough of refrigeration. I already had an offer for a free 1.5HP R404a compressor to use in the high stage. However: low stage refrigerant (R170 seems the best for our purposes) and low stage compressor are stilll a problem.


2/ Evaporator design, why not just put a trap in the suction line at the lowest point and connect on the TEV sensing bulb on the horizontal above the IMS fluid level, that is how most coils Iv'e seen have been circuited.

Can you explain this a bit more? Do you mean routing the suction line upwards, going horizontal above the water/methanol fluid level, and putting the TEV bulb there?


3/ you can buy small distributor heads for say 3/16 distributor pipes danfoss make these, this would aid distribution a bit, for a minimal cost.

I think 3/16" pipes form a too little restriction to ensure proper refrigerant distribution. The circulation rate of the refrigerant will be 3.5-4kg/hr, which is not much.


All the best on your new build, keep us posted.
Regards. Andy:)

I will, thanks :)

Prof Sporlan
23-04-2003, 02:51 PM
Prof, can you enlighten me in the model used
The Prof has used both the Lockhart-Matinelli and Dukler correlations for two-phase flow, but he favors the Dukler correlation. Both correlations are rather involved, though the ASHRAE Fundamentals Handbook does a nice job presenting both in its Two-Phase Flow chapter.


Wouldn't the 70psi pressure drop give a better distribution of the refrigerant mass flow?
Marginally, perhaps, over that of a 10 psi pressure drop


It would limit pulldown speed, but it would also limit heat dumped into the condenser during pulldown.
This is really the problem you have with very high distributor pressure drop... you limit pulldown capacity, and effectively limit the range the TEV can operate. But with an oversized TEV, this will be less of a problem.

Andy
23-04-2003, 07:12 PM
Hi Dabit,

Head pressure would be 16 bara / 15 barg max. The NL11F is allowed to condense R134a at 60 °C / 140F for a short amount of time, 55 °C / 131F for a longer amount of time.
R134a @ 55 deg C is a condensing pressure of 14.9 bara, on R507 you would need to be below 31 deg C not to exceed this pressure. You could do this by running mains water to waste through a water cooled condenser, this would ensure a cold suply of say 12 deg C water. That would be the way I would be thinking of going.


Can you explain this a bit more? Do you mean routing the suction line upwards, going horizontal above the water/methanol fluid level, and putting the TEV bulb there?

Exactly, but I would fit a little p shaped trap t the lowest point on the suction to aid oil return.


I think 3/16" pipes form a too little restriction to ensure proper refrigerant distribution. The circulation rate of the refrigerant will be 3.5-4kg/hr, which is not much.

Personally I would fit the 3/16 distributor, then reduce to the capillary size you require, you may need a larger capillary that first envisaged anyway.

Kind Regards. Andy:)

Gary
23-04-2003, 07:40 PM
To my mind, the way to develop a system is to evolve it, perfecting one variable at a time, as opposed to jumping from one radical system design to the next radical system design. I would use the new evaporator design on your current system, but I would make it counterflow, even if this meant feeding the refrigerant in from the bottom. The advantages of top feed are minimal compared to the disadvantages of parallel flow. Laying it on it's side would give you the best of both.

If you are intent on using R507, I would strongly recommend that you go to the R404A compressor you mentioned and stay with the air cooled condenser. The TEV will dictate the capacity, and it's MOP charge will prevent any overload of the condenser. In a future cascade system, this system would be your low stage, and the R134A compressor would be used for your high stage.

The LP access shraeder should be in the suction line near the compressor.

DaBit
24-04-2003, 04:30 PM
Originally posted by Prof Sporlan
The Prof has used both the Lockhart-Matinelli and Dukler correlations for two-phase flow.

Unfortunately Google does not show many useful links, not even when I use Martinelli instead of Matinelli. It seems I definitely need to get a copy of the ASHRAE Fundamentals handbook. If you happen to know a link to a good description of the correlations, I'm in for it.


Marginally, perhaps, over that of a 10 psi pressure drop.

OK, I will use the larger sized tubes.


Originally posted by Andy
R134a @ 55 deg C is a condensing pressure of 14.9 bara, on R507 you would need to be below 31 deg C not to exceed this pressure. You could do this by running mains water to waste through a water cooled condenser, this would ensure a cold suply of say 12 deg C water. That would be the way I would be thinking of going.

Staying below 31 &deg;C should be possible during normal operation. Staying below 31 &deg;C during pulldown is a bit harder, but not impossible.

Running with tap water is fine for testing, but when finished, I would like my PC to consist of one enclosure containing all the necessary components. This should not be a problem. The older copper car heater cores like the ones found in a Suzuki Swift are cheap (at the junkyard), fairly small (size in the order of 10"x6"), and they can get rid of a tremendous amount of heat. I used to run a 172W TEC, which generates ~300W of heat, combined with such a radiator (albeit from another car) and a 200gph/800L/hr pump, and my water temperature was 27 &deg;C / 80F with an ambient temp of 23 &deg;C / 73F. When necessary, I have the space and possibility to fit 2 or 4 of them. And 4 slow running fans make less noise than a single fan running at high speed.

And just in case condensing pressure rises too much, the compressor is turned off since the high-side pressostate (correct English word?) breaks the mains to the compressor, cycling it on and off in such a way that the condensing pressure stays below the set maximum.


Exactly, but I would fit a little p shaped trap t the lowest point on the suction to aid oil return.

Can you make a rough drawing and post it here? (or mail to dabit@trybit.com so I can post it for you, if you do not have server space somewhere). In this case a picture says more than 1000 words.


Personally I would fit the 3/16 distributor, then reduce to the capillary size you require, you may need a larger capillary that first envisaged anyway.

I will see if I can get any.


Originally posted by Gary
To my mind, the way to develop a system is to evolve it, perfecting one variable at a time, as opposed to jumping from one radical system design to the next radical system design. I would use the new evaporator design on your current system

I was already planning to do some testing with R134a first, since my supply of R507 is very limited. Every single gram of refrigerant is administered in The Netherlands, so obtaining some is pretty hard.

But in fact, I am not changing much. Besides adding Schraeder valves for pressure measurements, I change the evaporator and condenser, and I switch back from captube to R134a TEV. This looks like a large change, but the only not directly necessary change is the switch to a watercooled condenser. However, I do not see how this could ever be a disadvantage. A liquid-cooled condenser is very useful to control condensing pressure when troubleshooting, since controlling water temperature is easy.

Then, I can get TD data for both the condenser and evaporator, and when measured TD's (now including SST and SCT based on pressure) are fine, switching to R507 is not more than replacing the TEV and recharging.

If R134a tests show a too large TD over both the evap or the condenser, I know I have to improve before using R507.


but I would make it counterflow, even if this meant feeding the refrigerant in from the bottom. The advantages of top feed are minimal compared to the disadvantages of parallel flow. Laying it on it's side would give you the best of both.

What about refrigerant and oil trapping in the evaporator when feeding the refrigerant from bottom to top (or from side to side)? Wouldn't that fool the TEV? In case it doesn't, bottom-to-top feed

The disadvantage of the parallel flow is probably overestimated, given the small TD (< 0.5K) between sec. coolant in / sec.coolant out, which is the result of the relatively high secondary coolant flow rate.


If you are intent on using R507, I would strongly recommend that you go to the R404A compressor you mentioned and stay with the air cooled condenser.

Two problems:
- I do not thrust my current condenser with the higher condensing pressures since it is a steel tubes/steel fins R12 model, not even allowed to operate with R22.
- The compressor mentioned is a 1.5HP model. Quite a bit too powerful, large and probably noisy for this application.


The TEV will dictate the capacity, and it's MOP charge will prevent any overload of the condenser. In a future cascade system, this system would be your low stage, and the R134A compressor would be used for your high stage.

I am not too sure that R507 would be a good low-stage refrigerant. It's saturated pressure at < -70 &deg;C evaporation is still very low. And < -70 &deg;C evaporation will be my cascade target.


The LP access shraeder should be in the suction line near the compressor.

There is already one at the compressor service port (which is directly connected to the hermetic shell). I planned an extra Schraeder after the TEV to have an indication of pressure drop over the evap + distributor + suction line. Then, I have three pressure measurement taps: one HP, one after the TEV, and one at compressor suction pressure. Then, when things fail, I can see if the pressure drops are as expected.

Gary
24-04-2003, 09:30 PM
A liquid-cooled condenser is very useful to control condensing pressure when troubleshooting, since controlling water temperature is easy.

Therein lies the flaw in your reasoning. The air is cooling the water, therefore the water must necessarily be warmer than the air, in which case you are better off cooling directly with the air. Unless you are using evaporative cooling for the water, or domestic cold supply water which is cooled by the Earth, you are better off without it.


Then, I can get TD data for both the condenser and evaporator, and when measured TD's (now including SST and SCT based on pressure) are fine, switching to R507 is not more than replacing the TEV and recharging.

There is a lower limit to the pressure your compressor is physically capable of pumping. Roughly speaking, I would say this is about 15 inches of vacuum. At this point, the temperature can go no lower and you would want to consider going to a higher pressure refrigerant. Some are fairly cheap and all have disadvantages. When you see the price tag on refrigerants used in commercial cascade systems you are going to be shocked.


The disadvantage of the parallel flow is probably overestimated, given the small TD (< 0.5K) between sec. coolant in / sec.coolant out, which is the result of the relatively high secondary coolant flow rate.

That is not the TD. That is the dT. It is small because it isn't transferring much heat, due to a very large TD (difference between the refrigerant temperature and the secondary coolant temperature). Direction of flow is very important for good heat transfer. The disadvantage of parallel flow is very substantial. Until you have designed and built an efficient evaporator, maximizing heat transfer, all other factors are essentially meaningless.

DaBit
25-04-2003, 11:25 AM
Therein lies the flaw in your reasoning. The air is cooling the water, therefore the water must necessarily be warmer than the air, in which case you are better off cooling directly with the air.

True. But given the 3-4 times higher heat load on the condenser at startup, I have only a few choices to keep condensing pressure during startup low:
1) Use an excessively large condenser, dimensioned for startup load.
2) Accept the compressor on-off cycling generated by the high-side pressostate.
3) Use a buffer which forms a temporary storage for the generated heat.

Since my current condenser is unable to handle startup load with the low condensing pressure required, option 1) would involve scavenging the junkyards for a larger one. And they get very large pretty soon (airconditioning in the 3-ton range), so finding a good one is not that simple.

Option 2) sounds bad. I doubt I can live with that, although that is more a psychological consideration than a technical one. Besides that it would lenghten pulldown time.

Option 3) is just what I intended with the liquid cooled condenser. The 1-2 gallons of water would provide enough heat storage capacity to handle the startup load.

What about the following idea: I precede the conventional air-cooled condenser with a large coil submerged in (moving) water, like this:

compressor discharge -> water-submerged coil -> condenser-> receiver

This gives me the required heat storage capacity to handle the incidental startup load (which only occurs after hours of standstill). After startup, the refrigerant condenses at the coldest point, which will be the air-cooled condenser, since that one cools the fastest when load decreases.

Does this sound plausible?


There is a lower limit to the pressure your compressor is physically capable of pumping. Roughly speaking, I would say this is about 15 inches of vacuum. At this point, the temperature can go no lower and you would want to consider going to a higher pressure refrigerant.

And also the suction gas cooling of the hermetic motor is compromised.


Some are fairly cheap and all have disadvantages. When you see the price tag on refrigerants used in commercial cascade systems you are going to be shocked.

Hmm. I haven't seen prices yet, and I haven't investigated the properties of low-stage refrigerants yet. But when thinking cascade for my purposes, two refrigerants come to mind:

- Ethane (R170) or an ethane/propane mix. Flammability is the disadvantage, but even when a leak occurs it is hard (read: almost impossible) to get concentrations between the lower and higher explosion limit since only a few grams of the refrigerant is used, the compartment containing the cooling has a large volume, and a steady air flow is directed through it. And even when an explosion occurs, the released energy is at most high enough to wreck the enclosure, without releasing high velocity particles. I did some calculations and simulations based on 40 grams (< 0.1 lbs) of propane in the system, which is a gross overcharge.

- CO2, but there are problems with the stuff going solid.



That is not the TD. That is the dT.

When do you say TD (temp. difference), and when do you say dT (delta-T)?


It is small because it isn't transferring much heat, due to a very large TD (difference between the refrigerant temperature and the secondary coolant temperature).

And the relatively high flow rate. Some math:

- Assumed effective flow rate: 400L/hr - 100gph (1200L/hr - 300gph pump restricted by flow resistances)

- Heat capacity of 50/50 water/methanol @ -30 &deg;C / -22F : 3376 J.(kg<sup>-1</sup>. K<sup>-1</sup> (number taken from Coolpack)

Per second, 400/3600 = 0.11L passes the evap coil. At a Qe of 150W, this means that the water/methanol cools down 150 / (4128 * (400/3600)) = 0.40K

Thus, T<sub>water, out</sub> = T<sub>water, in</sub> - 0.40K

Now, what temp difference between refrigerant and sec. coolant would be realistic? When the TEV is set at 4K superheat, it will not get better than 4K difference between sec. coolant and boiling refrigerant since the gas at the evap exit cannot get warmer than the secondary coolant temp. 6K is more likely, assuming a 2K temp drop over the refrigerant->copper interface + copper thermal resistance + copper->sec. coolant interface. And even 2K is probably not easy.

Assuming the 6K temp difference, in the counterflow situation the TD between sec. coolant and boiling refrigerant would be 5.6K, in the parallel flow situation itr would be 6K.

With large chillers where water enters the evaporator at 12 &deg;C and exits at 7 &deg;C, the difference between counterflow and parallel flow is dramatic.

But still, with counterflow I save 0.4K
The question: does this 0.4K advantage weight up against the possible oil and refrigerant trapping in the evaporator due to bottom-top or side-side refrigerant feed?

Doesn't the bottom-top feed screw up TEV operation when refrigerant gets trapped?


Until you have designed and built an efficient evaporator, maximizing heat transfer, all other factors are essentially meaningless.

OK, some more tips? Until now I have tried:

- 1.5m (4.5ft) of 12mm (1/2") pipe in a 80mm(3.15") shell. It was leaky, and didn't work too well anyway.

- 8m (24ft) of 1/4" pipe in a 125mm (5") PVC shell. This evap had problems with a too low sec.coolant speed over the evaporator coil, a high pressure drop over the coil, and problems with keeping the methanol inside.

- 3m(9ft) of finned 3/8" pipe in a coaxial construction. A too low evaporation surface area and too much heat ingress (even with 11mm Armaflex) due to the high surface area of the outer shell are the problems.

A plate heat exchanger is still the best evaporator possible for this purpose, but they are so damn expensive.

Thus, I am back at the coil-into-shell idea, but this time with multiple 1/4" coils, guides to force water over the coils, and a full copper construction.

Any tips are very welcome.

Prof Sporlan
25-04-2003, 02:44 PM
Unfortunately Google does not show many useful links, not even when I use Martinelli instead of Matinelli.

DaBit, you are, of course, correct with your spelling of Martinelli, and the Prof didn't find much useful in his google search either. Perhaps he might just have to set up a web page on how to model two-phase refrigerant flow one of these days... :)

DaBit
25-04-2003, 03:33 PM
Originally posted by Prof Sporlan Perhaps he might just have to set up a web page on how to model two-phase refrigerant flow one of these days... :)

This web page would score at least one visitor :)

Gary
25-04-2003, 07:58 PM
When do you say TD (temp. difference), and when do you say dT (delta-T)?

When the temperature of something changes, it is a delta-T. When the temperature of one thing is compared to the temperature of another thing, it is a TD.


Since my current condenser is unable to handle startup load with the low condensing pressure required, option 1) would involve scavenging the junkyards for a larger one. And they get very large pretty soon (airconditioning in the 3-ton range), so finding a good one is not that simple.

I haven't seen a complete set of numbers for your current system, so I have no way of judging the relative sizes. Given that the current condenser is matched to the current compressor, it should be able to handle startup pressures, especially since the evaporator isn't picking up a full load. Without those numbers, I can't tell what the problem is from here.


And also the suction gas cooling of the hermetic motor is compromised.

Not necessarily so.


Now, what temp difference between refrigerant and sec. coolant would be realistic? When the TEV is set at 4K superheat, it will not get better than 4K difference between sec. coolant and boiling refrigerant since the gas at the evap exit cannot get warmer than the secondary coolant temp. 6K is more likely, assuming a 2K temp drop over the refrigerant->copper interface + copper thermal resistance + copper->sec. coolant interface. And even 2K is probably not easy.

Your current evaporator is nowhere close to the TEV superheat limitation. When you reach that point, it will be cause for celebration. And there are ways to overcome that limitation, also. Let's cross that bridge when we come to it.


But still, with counterflow I save 0.4K

You are assuming a Qe of 150W.


Doesn't the bottom-top feed screw up TEV operation when refrigerant gets trapped?

The potential problem would be oil trapping, and the velocity in this situation should be more than sufficient. Prof Sporlan would be better able to enlighten us on this. :)

In any case, if you lay it on its side, you should be able to get both top feed and counterflow, just as you are doing with your current evaporator.

Also, you may want to add a storage tank to the secondary coolant loop similar to what you show in your condenser water loop. This would give you a place to add coolant and remove air, as well as provide some flywheel effect.


Any tips are very welcome.

Never make solder connections inside the evaporator.

Counterflow is better than crossflow, which is better than parallel flow.

Control and monitor the variables. Change one variable at a time, evaluate it, perfect it, then move on to the next variable.

Prof Sporlan
26-04-2003, 01:21 AM
What about refrigerant and oil trapping in the evaporator when feeding the refrigerant from bottom to top (or from side to side)? Wouldn't that fool the TEV?
Refrigerant trapping can occur in bottom fed evaporator coils, or for that matter, any evaporator circuited in such a manner that requires refrigerant to flow uphill, if refrigerant velocities are not sufficiently high. When this happens, refrigerant traps in an area of the coil which eventually gets pushed along in the form of a slug. This can cause the TEV to hunt in long cycles if the slug is able to influence coil outlet temperature.

The Prof would not be too concerned about refrigerant trapping in this case. The 1/4" OD tubing should provide sufficient velocity at a 150 watt load, even if 3 or 4 circuits are employed.

DaBit
26-04-2003, 10:00 AM
Originally posted by Gary
I haven't seen a complete set of numbers for your current system

The numbers shown in the previous thread are the best I can do at the moment. When I add the schraeder valves, I will be able to provide you with more data.


, so I have no way of judging the relative sizes. Given that the current condenser is matched to the current compressor

The current condenser runs up to ~40 &deg;C condensing with R134a and the Danfoss TEN2 valve, orifice #0X. During normal operation, only the few top rows of the condenser are warm, the rest stays cool.

The condenser itself is scavenged from an old R12 refrigerator.



Not necessarily so.


Why not? If there aren't much suction gas molecules around the motor, how would the windings be cooled?


Your current evaporator is nowhere close to the TEV superheat limitation

I know, and that's why I am doing the evaporator-homework again.


When you reach that point, it will be cause for celebration. And there are ways to overcome that limitation, also. Let's cross that bridge when we come to it.

OK.


You are assuming a Qe of 150W.

Yes. It will be in the 150-200W range, based on heat output of the cooled components, and heat ingress through the insulation. I did some simulations with CoolPack, and they match measured values within 20% or so. Close enough to be useful.

During pulldown, the load will be much higher, but pulldown is just that, pulldown. A situation which happens only once a day, at most.


In any case, if you lay it on its side, you should be able to get both top feed and counterflow, just as you are doing with your current evaporator.

But not gravity-supported flowback of oil to the compressor, since the axis of the coil is horizontal, when the evap is laid on it's side.


Also, you may want to add a storage tank to the secondary coolant loop similar to what you show in your condenser water loop. This would give you a place to add coolant and remove air, as well as provide some flywheel effect.

Total sec.coolant loop volume will be 1-2 liters (0.25 - 0.5 gallons). This is dictated by the volume of the piping, hoses and evaporator. Based on experience I can say that I do not need more thermal storage.

Removing secondary coolant is done with a plug connected to the lowest point of the system. Adding coolant is done through a plug connected to the highest point of the system. The piece of pipe connecting to the fill-plug is normally not filled with liquid, and servers as an expansion chamber for the secondary coolant to cope with the volume of secondary coolant increasing and decreasing due to temperature changes.


Never make solder connections inside the evaporator.

I guess brazing (those SilPhos rods) the connections inside the evaporator, and soldering (using SnAg) the end caps of the outer shell is fine. The 500 &deg;C diference in melting temp should keep me out of trouble.


Originally posted by Prof Sporlan
The Prof would not be too concerned about refrigerant trapping in this case. The 1/4" OD tubing should provide sufficient velocity at a 150 watt load, even if 3 or 4 circuits are employed.


Thanks Prof, that's what I wanted to know.
Provided that I have enough 1/4" pipe here, I will construct the evaporator today.

I will construct the evaporator coils with a horizontal axis (coils like this: --^^^^--), and mount them into the 80mm copper shell. I will show you some pictures, on which I would like your comments.

Another question: Currently I have an (U-shaped, with oil draining tube) accumulator mounted in the suction line to prevent liquid slugging of the compressor directly after startup. Should I maintain that accumulator?

DaBit
29-04-2003, 04:11 PM
OK, I am done constructing the evaporator. I crammed 12 meters (36ft) of 1/4" tubing into a 80mm / 3.15" copper shell, divided over 3 coils of ~4 meters each.

The refrigerant distributor is made of 3 pieces of 1.2mm/.047" capillary tube, 24cm/9.4" long.

I do not have an exactly nice counterflow since two coils are mounted top-to-bottom. Well, I didn't have much choice if I also wanted to keep circuit lengths approximately equal and do it with the tools I posess.

As promised some pictures (links; I do not want to bother the accidental visitor with the pictures). Sizes are 20-35kbyte/picture:

<a href="http://www.arcobel.nl/~dabit/zooi/R507evap_coilsjustbent.jpg" target="_blank">The just-bent evaporator coils</a>

<a href="http://www.arcobel.nl/~dabit/zooi/R507evap_2coils.jpg" target="_blank">2 out of 3 evaporator coils</a>

<a href="http://www.arcobel.nl/~dabit/zooi/R507evap_3coils_suction.jpg" target="_blank">3 coils.</a> Here you can also see that two coils are not mounted in a stricktly counterflow arrangement.

<a href="http://www.arcobel.nl/~dabit/zooi/R507evap_coilsinshell.jpg" target="_blank">coils into shell</a>

<a href="http://www.arcobel.nl/~dabit/zooi/R507evap_refinlet_side.jpg" target="_blank">Distributor tubes</a>

<a href="http://www.arcobel.nl/~dabit/zooi/R507evap_refinlet_soldered.jpg" target="_blank">Distributor tubes, soldered into the 3/8" line from TEV</a>

Pipes are all a bit long, but it is easier to shorten them than to lengthen them.

I pressure-tested the evaporator with 21 bars of pressure. Yesterday I put 5 bars of pressure in it and closed the pipes. Tonight I will see whether the evaporator is leaky or not.

Gary
29-04-2003, 11:07 PM
What stops the water from going straight down the center, rather than around and between the coils?

Why not make all three coils coaxial, with the second coil wrapped around the first and the third wrapped around the second?

Why not drill three 1/4 inch holes in the end caps and bring the tubes out through those holes, so that all brazing can be outside the evaporator, similar to what you did with the cap tubes?

DaBit
01-05-2003, 09:25 AM
Originally posted by Gary
What stops the water from going straight down the center, rather than around and between the coils?

Sorry, this is a detail I forgot to mention, and of which I did not take a photograph. I blocked most of the inside of the inner coil with a capped piece of 22mm (~7/8") pipe since I noticed this flaw myself. This prevents flowing of the liquid directly through the center to the other side. There is approximately 5mm (0.2") space between the 22mm pipe walls and the inner coil.


Why not make all three coils coaxial, with the second coil wrapped around the first and the third wrapped around the second?

This is a loss of water/pipe surface contact area since the coils fit too well inside each other.

The second reason why I didn't do that was the length of the coils. To keep refrigerant distribution and heat load per coil the same, I thought it might be a good idea to keep the lengths of the coils approximately equal.


Why not drill three 1/4 inch holes in the end caps and bring the tubes out through those holes, so that all brazing can be outside the evaporator, similar to what you did with the cap tubes?

Bending the 1/4" pipe in small radiuses is still a problem for me. I do not have the tools, and using a cylindrical thing to bend around only works when you have enough space to work.

Besides, the end caps are soldered with normal (plumbing) solder. The melting point of the brazing material used and that solder differs ~500 &deg;C / ~1000F, which is enough to leave the brazing intact while flowing the SnAg solder.

I did the captubes externally so I can easily alter the direction of the evaporator. Initially I will place it horizontally, but if required I want to be able to mount it vertically. And I want the 3/8" -> 3x captube distributor to be vertical to ensure an as good distribution as possible.

I pressure-tested the finished evaporator at 21 bars (~300psi), and I left a test pressure of 5 bars (72 psi) in it for 36 hours. Test pressure came out unaltered, so I can reasonably assume that the evaporator does not leak.