Overclock.net › Forums › Cooling › Cooling Experiments › Subambient full submersion phase change cooled pc
New Posts  All Forums:Forum Nav:

Subambient full submersion phase change cooled pc - Page 3

post #21 of 333
Quote:
Originally Posted by technogiant View Post

No problem Terrere.....I've changed my mind and considered so many different methods while thinking this one out I've confused everyone ....including myself....lol

But now I've thought out what I believe to be a fool proof way of recovering the liquid refrigerant I'm confident to go ahead with my favored method which is the full submersion one.

It eliminates many of the problems of the other methods I've considered and the only real drawback was the large volume of liquid refrigerant it would require....which I'm now sure I can safely contain.

Thanks for your input anyway Terrere....please feel free if you think of anything else. smile.gif

Yeah, I'll be tracking the thread and if I have any input that can be of use I will gladly share!
Quote:
Originally Posted by technogiant View Post

Absolutely correct........the additional stuff I'm trying to work out is really just a method of delivering that cooling power as directly, effectively, simply and reliably as I possibly can to my pc components.

thumb.gif I finally realized what you were actually using the AC for and then it all made sense! I think it's a pretty solid design to me, but my experience on cooling systems is more closely linked to industrial solutions. biggrin.gif I definitely will watch to see what comes of this idea!
Dark Space
(18 items)
 
  
Reply
Dark Space
(18 items)
 
  
Reply
post #22 of 333
Cool idea just one problem I can fore see the sealed chamber will have to withstand huge pressures as the liquid will expand to gas.
I cannot tell you exactly how much but I believe it would be 100 times the volume of liquid at a very minimum but could be as high as 45,000, liquid to gas volume.

that's if you had the expansion sack or what ever attached it would have to be the size of a house to equalize the pressure other wise I can see a potential for a huge pressure spike. Before the evaporator started to condense the vapor.
I would put a 1bar safety valve vented to out side your house to be on the safe side. you can always top up the liquid bath later when the system has equalized and running.

I would probably also look at having a pressure chamber for the motherboard chamber. but you also have to remember that under pressure, liquids will change their boiling point to .a higher temp.

I like your idea I would love to run my pc at -25 on load every day too
Edited by feznz - 3/23/13 at 1:50pm
kamikaze
(17 items)
 
  
CPUMotherboardGraphicsGraphics
3770k @ 4.8 asus maximus v extreme Asus Strix 1080 Ti  Asus GTX 770 DCUii 
RAMHard DriveCoolingOS
GSkill 2400Mhz Sandisk ultra pro xspc 240 raystorm Windows 8.1 Pro 
MonitorMonitorMonitorKeyboard
LG UC9938-w AOC 24"  AOC 24" G19 
PowerCaseMouseAudio
enermax maxrevo 1500w nzxt 820 Phantom saitik rat 9 Asus Xonar DX 
  hide details  
Reply
kamikaze
(17 items)
 
  
CPUMotherboardGraphicsGraphics
3770k @ 4.8 asus maximus v extreme Asus Strix 1080 Ti  Asus GTX 770 DCUii 
RAMHard DriveCoolingOS
GSkill 2400Mhz Sandisk ultra pro xspc 240 raystorm Windows 8.1 Pro 
MonitorMonitorMonitorKeyboard
LG UC9938-w AOC 24"  AOC 24" G19 
PowerCaseMouseAudio
enermax maxrevo 1500w nzxt 820 Phantom saitik rat 9 Asus Xonar DX 
  hide details  
Reply
post #23 of 333
Thread Starter 
Cheers Feznz.......You are right about the huge volume and pressures that would involved. The average volume ratio for refrigerants liquids to gas volumes is 1:240....so my 15 liters of liquid refrigerant would become 3600 liters or 3.6 cubic meters. The pressure to contain the gas at ambient temps would be in the vicinity of 100psi.

Don't worry my expansion sac isn't that big and I'm not expecting my chamber to go above normal atmospheric pressure...lol

Much of the above discussion has been about how to transfer liquified refrigerant from my computer chamber into a pressure bearing gas bottle. You see while in the chamber the temperature of -25c will maintain the refrigerant as liquid even at normal atmospheric pressure as its boiling point is about -15c.

The liquid will be transferred to the pressure bottle while still cold and in liquid form at atmospheric pressure by the process I've outlined...which I believe is known in industry as "cold line transfer"

Also when the liquid is transferred back from the bottle to the chamber as it will have been in the freezer (temp about -20c)the bottle pressure will be sub atmospheric and will only build slowly as the bottle warms and will then force the liquified refrigerant (not gaseous) slowly into the chamber.

So my chamber will not be at high pressure and the huge gas volume will always be just 15 liters of cold liquid.

The only time I'll be emptying bottle pressure gas into the chamber will be when I initially load it with refrigerant...and yes that will have to be done progressively in pulses that the expansion sac and condensing rate of the chamber can deal with...from a power stand point it will probably take the 3.6kw ac unit about 15 mins to condense out 15 liters of refrigerant......but once that is done it will remain in liquid form apart from the boil off from the pc components during use which I'm sure the 3.6kw ac unit and 250 liter expansion sac will cope with.


Ps...I'm already running my pc in a chamber at -25c 24/7 with air cooled components...so it can be done...this is just going to be a vast improvement over air in the way the cooling is delivered and bring my load temps right down.thumb.gif
Edited by technogiant - 3/23/13 at 3:48pm
post #24 of 333
I had a look at your previous a/c build and basically limited to the air medium to transfer the heat from CPU to evaporator hence the liquid medium.

I think it is a brilliant idea I am keen to see work thumb.gif

I just believe the balloon may prove useless in this case as to condense the gas back to liquid you will have to also get the balloon cold enough to condense the gas to liquid again.

I believe holding the refrigerant under pressure in the cold box may be the solution hence safety valve to replace the balloon. with sufficient expansion volume within the cold box.

I can see this is going to be a tedious filling operation but once up and running I am keen to see the results biggrin.gif

This is pure speculation but I will keep an eye on this thread
kamikaze
(17 items)
 
  
CPUMotherboardGraphicsGraphics
3770k @ 4.8 asus maximus v extreme Asus Strix 1080 Ti  Asus GTX 770 DCUii 
RAMHard DriveCoolingOS
GSkill 2400Mhz Sandisk ultra pro xspc 240 raystorm Windows 8.1 Pro 
MonitorMonitorMonitorKeyboard
LG UC9938-w AOC 24"  AOC 24" G19 
PowerCaseMouseAudio
enermax maxrevo 1500w nzxt 820 Phantom saitik rat 9 Asus Xonar DX 
  hide details  
Reply
kamikaze
(17 items)
 
  
CPUMotherboardGraphicsGraphics
3770k @ 4.8 asus maximus v extreme Asus Strix 1080 Ti  Asus GTX 770 DCUii 
RAMHard DriveCoolingOS
GSkill 2400Mhz Sandisk ultra pro xspc 240 raystorm Windows 8.1 Pro 
MonitorMonitorMonitorKeyboard
LG UC9938-w AOC 24"  AOC 24" G19 
PowerCaseMouseAudio
enermax maxrevo 1500w nzxt 820 Phantom saitik rat 9 Asus Xonar DX 
  hide details  
Reply
post #25 of 333
Thread Starter 
Thanks Feznz, I understand your thought process about containing it in a pressure vessel. But tbh once you go down that route the engineering requirements increase exponentially.

My aim has been to make the system as effective as possible but still within my limited DIY capabilities so I'm trying to design the system to remain at or near atmospheric pressures.

Again I understand your thoughts about about the balloon, but it will not require cooling in itself. Very much as it works now with my current build, while chilling the chamber down the chamber air contracts and causes a pressure drop which then makes the balloon empty into the chamber, the same pressure drop will occur with the new build when refrigerant gas phase changes back to liquid at the evap so the balloon will empty back into the chamber where the contents would be chilled causing more contraction and further discharge from the balloon into the chamber.....it's a continuous process so there is no need to chill the balloon.

I think I will see a larger difference in the volume change in the balloon when comparing the idle to full load conditions with this set up due to the large volume changes that occur with phase change as opposed to just the volume change caused by the difference in chamber temp I see now when going from -30 at idle to -22 at full load.....but I don't see this as something that will be in constant flux rapidly going from balloon full to balloon empty and back again as the chamber itself will have quite a bit of "heat buffering" effect as the components are submerged in a substantial quantity of liquid and for a given heat load it will just reach a steady state volume.
If there were to be large, rapid and continual volume changes that would be a problem as it would effectively transport a lot of heat into the chamber as the balloon isn't thermally insulated, also the balloon would become too cold and affect its flexibility perhaps causing it to crack......I don't see this happening though as I anticipate a steady state volume will be achieved.

This is how I anticipate it will work; as well as the refrigerant liquid/gas (RG) the chamber will also contain a background gas (BG) that will not liquify at the temperatures being used. This is required to maintain the pressure at atmospheric as if it only contained RG then below the boiling point the chamber pressure would be below atmospheric and be crushed.

I'd have sufficient volume of BG so that at the lowest idle chamber temperature the expansion sac would be about empty. So if we consider the volume of the chamber, the vapor space, to be about 100 liters and the largest volume of the expansion sac to be 250 liters then at the coldest idle temps there would be 100 liters of gas the majority of which would be BG and a small proportion being RG say for the sake of argument 99% BG 1% RG.

As heat load is applied obviously the quantity of BG will remain the same but it would be possible for the proportion of RG to increase until the expansion sac was filled.

In those circumstances the new total volume of the system would be 100 + 250 = 350 liters and the fraction/percentage of RG would be 250/350 or 71% RG and 29% BG.

So basically the RG composition could theoretically increase from 1% to 71%.......If you think of this in terms of water vapor and humidity....obviously the higher the humidity the easier and more likely it is that water vapor will condense out.

So I envisage that as the composition of the vapor changes and becomes more saturated with RG the more easily the system will be able to re condense it and a balance point will be reached dependent on the heat load, % saturation and the chambers cooling ability.

But yes I concede that at this point I have no means of telling where the balance point would be but do believe a potential 71% RG vapor content would be sufficient to allow efficient re-condensation. Of course this balance point will also be dependent on the balance of heat load and cooling power.....which again is loaded in favor of success with a 3.6kw ac unit doing the cooling.

But yes granted this part is conjecture/educated guessing at this time.....and time will tell smile.gif

+++++++++++++

Something further I've just considered as relevant to this discussion is the efficiency of the transfer of heat energy to the evap. I think that this process will actually become more efficient as the heat load increases.

If you consider the situation at lower heat load, as the small volume of RG boils off it will hit the much larger body of BG which will be at substantially below the RG's boiling point (the lowest idle chamber temp I've hit is -33c with my current build). So I anticipate it will directly re-condense in the chambers vapor before it hits the evap. This will of course warm the BG that will then be cooled by the evap.....but as you can see at this stage the transfer of heat energy to the evap will be via the BG being cooled as it hits the evap or a gas/evap energy exchange.

At higher heat loads as the proportion of RG saturation increases I envisage that a greater proportion of the RG will be directly re-condensed on the evap surface and so at this point the heat transfer to the evap will be via phase change which will be so much more efficient than a gas/evap energy exchange.

So if the gas/evap interface ever was the limiting factor in terms of chamber temperature in my current build I see this phase change/evap interface being much more efficient and may well lead to lower loaded chamber temperatures which vary with my current build from about -30c idle to -22c at load, I guesstimate the max load atm with my current components to be around 500w - 600w.
Edited by technogiant - 3/24/13 at 6:12am
post #26 of 333
It looks like you have this all worked out pretty good. Hopefully you can find a cheap dual port recovery bottle with the larger 1/2" ftgs. The recovery bottles have a dip tube on the highside so no need to turn bottle upside down.

The heat pipe system that I mentioned before would be separate and completely sealed from the chamber. It's basically a stand alone Gravity Fed Heat Pipe System. See drawing. The gas condenses in the radiator and drains down to the cpu/gpu boiler chamber where it pools and boils off.. the superheated gas rises up back to radiator where it's cooled/condensed.

The piping is large diameter as to not impeded flow in either direction. The cpu/gpu boiler chamber has large volume for capacity. The base would be a boiler enhancement plate for achieving lower core temps. Charge port is soldered to the radiator, all joints can be soldered as pressure is only 80-100psi tops if using r134a.

Anyways.. just throwing some ideas out there.

Any Updates?

GRX
(13 items)
 
  
CPUMotherboardGraphicsRAM
Phenom II X4 965 @ 5.7Ghz Gigabyte 890FXA-UD5 XFX HD 5850 1160/1200 vmods -15c 4GB G-Skill Flares PC3 16000 
Hard DriveOptical DriveOSMonitor
WD Caviar Black 1TB Lite-On DVDR Vista/XP Acer 23.6" 
KeyboardPowerCaseMouse
MS Ultra 750W Benchtop FTW! Logitech 
Mouse Pad
Walmart Special 
  hide details  
Reply
GRX
(13 items)
 
  
CPUMotherboardGraphicsRAM
Phenom II X4 965 @ 5.7Ghz Gigabyte 890FXA-UD5 XFX HD 5850 1160/1200 vmods -15c 4GB G-Skill Flares PC3 16000 
Hard DriveOptical DriveOSMonitor
WD Caviar Black 1TB Lite-On DVDR Vista/XP Acer 23.6" 
KeyboardPowerCaseMouse
MS Ultra 750W Benchtop FTW! Logitech 
Mouse Pad
Walmart Special 
  hide details  
Reply
post #27 of 333
Thread Starter 
Thanks Drew......Yeah I have put a lot of thought into this....probably too much thought and not enough action......but your right....I feel I've thought this through to a point it's feasible to start.

I do like the thermosiphon idea, its nice and "contained". I think the main issue would be getting sufficient liquid flow to return while venting the something like 750 -1000 ml/sec of vapor generated at full load in the opposite direction...as you say you would have to have as wide a bore pipe as possible.

I tried a simple experiment blowing through a pipe connected to the bottom of a water filled container raised above me, even when blowing quite gently not a bit of water back flowed into my mouth...just a simple experiment but shows the potential problems. Even using wide bore pipes these would have to narrow down at the opening to the boiling chamber if I was to use standard water blocks as the boiling chambers....other than that it would involve custom boiling chamber builds.

The other problem that arises is that you would not just have one boiling chamber, but one for each gpu and the cpu. If you were to just use one condenser for all of them then you could potentially have "boil off balance " problems where one chamber would become over filled with liquid and the other under filled.

That could be over come by having one condenser for each boiling chamber...but that increases build complexity/cost

The other alternative would be to have the multiple boiling chamber linked by a bottom tube so the liquid levels between them would equalize.

But this in itself poses problems in that even with the mobo mounted flat the gpu boiler/s would be higher than the cpu boiler, so to have the gpu chamber filled to a sufficient level then the fluid level in the cpu boiler would actually extend into the pipes above and perhaps restrict gas boil off. I'm not sure this would cause a problem as the gas would force through anyway.

The outcome of this is that you would have to have quite a bit of liquid to fill the cpu chamber and pipes above to the level of the gpu, the bottom connecting pipe and the gpu chambers.

I wouldn't be containing this under pressure as you've indicated in the diagram, I'd be allowing it to return to ambient and boil off and be contained in an expansion chamber with the aim of using standard pc rads, water blocks and piping. I think that making pressure bearing rads, custom blocks and pressure piping would be a step too far for my skills. So the problem for me is the amount of liquid needed may be too much to contain in gaseous form. So for me this and the fact that standard water block ports may be too restrictive make it a less desirable option.

But if you have the skills to make a pressure system as per your diagram but just with an interconnecting bottom pipe between the multiple chambers then this would be an effective simple and reliable option.

Tbh the full submersion option has none of these potential problems, it allows for almost nude die cooling with the addition of the highly conductive graphite diamond dust heat spreader I've thought of. The heat spreader can be almost any size I desire and not constrained by the size of a boiler chamber, the build is basically so simple with no concerns about fluid return or venting capacity and just requires circulatory fans to operate it...perhaps two pumps just to cause fluid circulation and enhance the cooling efficiency.

Now I'm satisfied I've thought out how to recover the fluid to a pressure bottle I'm decided on the full submersion method.

That's great info about the dip tube on the recovery bottle though, that would be just perfect for my needs...don't have any links to that sort of thing do you Drew?

As regards updates....still all talk no action at this time....I've decided to do this in an out house/shed rather than in the house...which I have to build first...making preparations to do that currently so hopefully build will be starting in a month or so....the sooner the better for me...getting frustrated at not being able to start straight off.
Edited by technogiant - 4/3/13 at 4:50am
post #28 of 333
This idea makes me very excited. I knew submersion cooling wasn't just scifi/mineral oil. I look forward to seeing how this goes.
    
CPUMotherboardGraphicsRAM
Intel 4770k Asus Z87 Deluxe/Quad EVGA 1080 TI FE G. Skill 
RAMHard DriveHard DriveCooling
G.Skill Western Digital Samsung 840 Pro Water Cooled 
OSMonitorMonitorKeyboard
Windows 10 Pro Samsung 27" LED Acer 22" LCD Saitek Eclipse II 
PowerCaseMouse
Silver Stone 850W Antec Lanboy Air Microsoft Wireless 
  hide details  
Reply
    
CPUMotherboardGraphicsRAM
Intel 4770k Asus Z87 Deluxe/Quad EVGA 1080 TI FE G. Skill 
RAMHard DriveHard DriveCooling
G.Skill Western Digital Samsung 840 Pro Water Cooled 
OSMonitorMonitorKeyboard
Windows 10 Pro Samsung 27" LED Acer 22" LCD Saitek Eclipse II 
PowerCaseMouse
Silver Stone 850W Antec Lanboy Air Microsoft Wireless 
  hide details  
Reply
post #29 of 333
Thread Starter 
Thanks WSJ......I'm just starting to get some things together for this build now.......I've been thinking how to make the containment of the liquid in the chamber while being used as safe as possible. I was thinking of making my own submersion sump out of acrylic sheet so I could get just the size I wanted but found this pressed stainless steel container which is the perfect size.....being seamless one piece construction and made of steel there is no chance of leakage or fracture.

http://www.clickonstore.net/stainless-gastronorm-200mm-deep-23rds-size-p-258.html

The pipe to fill and drain it will no longer enter via a hole at the bottom as I want to maintain the integrity of the bowl....it will enter from above and be placed so the pipe opens at the bottom of the metal container....as it fills by pressure and drains by suction it will still function properly.....so this metal basin will form the submersion sump and be enclosed in a chamber made of acrylic that will contain the gaseous phase.....it's not going to be a pretty build, that's not really my thing, I'm more interested in function, so a cooking pot will do me fine....lol

Going to use a flexible stainless steel pipe to run from the stainless steel basin to the pressure bottle...that's going to be the next purchase..can get these here:-

http://www.southdownsmotorcaravans.co.uk/pdf/sdmc_gaslow_centre_stainless_steel_hoses_brochure.pdf

Being stainless steel it should last about 20 years and not split or crack at low temp...but I'll also be putting a second wider bore pipe over that as a belt and braces approach so if the first pipe were to leak then the liquid would just flow back into the freezer through the second pipe where it would be maintained as a liquid and not explosively boil off, I could then vent it off safely in a controlled manner.

I've been putting some more thought to the fluid traps I'll be using to seal the chamber and have been trying to think of a good liquid to use. As the gases I'm using are non polar and hydrophobic in nature then the best would be water...except I don't want it freezing and ice expansion potentially causing damage. Trouble is I can't use standard anti freezes such as glycols or methanol as the gases are soluble to some extent in them which could lead the liquid seals actually becoming a bridge for the gases to diffuse through and escape.

So I've been looking for an ionic solute to use, normal salt, sodium chloride came to mind first...but that is only good down to -23c......there best candidate I've come across so far is calcium chloride, a 30% by mass solution with water has a freezing point of -50c so that's more than enough for my needs and being ionic in nature will deter any gas solubility. And more importantly its readily available.

http://www.reagent.co.uk/calcium-chloride?utm_source=googlebase&utm_medium=free&utm_campaign=googlebase&gclid=CNiL3LfOs7YCFRTMtAodzBsAfQ

It won't pose any corrosion problems as the only materials touching it will be plastics also I'm going to pour a layer of oil over the top of the water in the liquid traps to prevent water evaporating into the chamber and causing condensation problems....may even use an oil that freezes....something like ground nut oil that freezes at +3c so that it will freeze solid and actually bond the lid on when in use.

Other than the calcium chloride solution...which I would hate to drip onto my mobo as that could be an instant kill I've been thinking of using a saturated simple sugar solution like "syrup"....that is polar and so should not allow the gas to dissolve.....it becomes very viscous and eventually forms a "glass" when cold enough that again would seal the lid on solidly.
Edited by technogiant - 4/5/13 at 10:48pm
post #30 of 333
Thread Starter 
I'm getting some stuff together now, I've got the metal drain pan



Uploaded with ImageShack.us

The pic above shows my old x38 socket 775 atx mobo for size comparison.
The basin is described as a 20 liter capacity. Its 15cm deep so should be sufficient to cover the tallest graphics cards.

There is plenty of room on the right for cabling and on the left is room for two D5 pumps.

I may even use some in fill material to occupy some of the unused space so I'm anticipating that with the volume occupied with everything in there the actual fluid volume will be reduced to 15 liters or less.

I've also got some things for a more substantial expansion chamber, the hopper ball I used previously was a little too flimsy for containing flamable gas for my liking, also it was greasy in texture, probably as the material is heavily laden with plasticizers which may have been extracted by the gas and contaminate my components, also it may allow the gases to diffuse out.

So I'm going for a different setup for the expansion chamber with these items.



Uploaded with ImageShack.us

The breather pipe will enter the upturned green water butt through it's base and the black tray will be filled with water. I'm going to cut down the water butt so that it is 15cm taller than the black tray. the water butt will be stuck to the bottom of the tray by its edge with a small gap separating them to allow water to flow in and out of the butt.

In fact the water level will be the same in the tray and the butt. The breather pipe will lead to the computer chamber and so any expansion or contraction of the gas phase will either push water out from or suck water into the upturned butt while not allowing gaseous exchange with the atmosphere.

I'm hoping that the setup will allow for sufficient expansion contraction but its not something I can really gauge until the system is up and running, it should cope with 50-60 liters of volume change.

To prevent water evaporating and entering the pc chamber through the breather I'm going to cover the water surface with a layer of oil...either mineral oil or just vegetable oil, it will float on top and I think stop the water evaporating into the chamber. I can't use all oil as the gases are soluble in it and would diffuse out through it so the water has to be there to maintain the seal.
New Posts  All Forums:Forum Nav:
  Return Home
  Back to Forum: Cooling Experiments
Overclock.net › Forums › Cooling › Cooling Experiments › Subambient full submersion phase change cooled pc