Advice Needed; Geothermal Computer Cooling System - Design Phase

[mhumm2]

My goal is a fanless (quiet) cooling system for my computer running at full load (CPU and GPU). It would also stop the dust accumulation and monthly cleanings. I live in Florida, and ws surprised that the cost of a "passive" geothermal system would be less than designing a passive radiator tower. Here's the rough plan so far: The water table where I live, is quite high. I hit ground water at 15 feet! I plan on bailing a well 25' to 30' down so I'm "in water" all year round. Instead of using the ground water as the coolant, I plan to imerse connected aluminum pipes as a heat exchanger then use the ground water to cool it. By building a closed-loop system, I don't have to worry about filtering the ground water (lots of sulfur and particulates) because I can use real coolant. The ground water is also the perfect temp; cold enough to cool the water blocks, but warm enough so condensation won't be a problem. The heat exchanger will probably be 2 or 4 sections of 3 or 4 foot-long 1" aluminum pipe. At some point in the system, I'll have to adapt down to 1/2" tubing to fit standard heat blocks then adapt back up to 1" PVC for cooling efficiency.

Pump Output ---> CPU ---> GPU ---> Heat Exchanger (underground) ---> Pump Input.

Qwestions:

1. Will I need a pressure release mechanism in the system?

2. Should I have a "bubble catcher" to get the air out of the system?

2. Will I need a check-valve or two?

Any other positive comments or suggestions would be appreciated.

[Movieman]

I think your biggest obstacle will be sizing a pump that will lift water 26feet..It can be done but not with any normal pump you see used on regular WC setups

[mhumm2]

Thank you Movieman. Good point. It will take something inbetween my 1.5HP 220vac sprinkler pump and a conventional 12vdc WC pump. I found a couple of commercial pumps in the $100 range that should do the job. The nice part is it's a closed system so I can easily calculate the weight of the water below the pump and subtract the weight of the water above the pump. The net sum would be the "load" on the pump and it would not change since those weights will be constant as the water moves through the system. Would you agree or am I missing something?

[Movieman]

Thank you Movieman. Good point. It will take something inbetween my 1.5HP 220vac sprinkler pump and a conventional 12vdc WC pump. I found a couple of commercial pumps in the $100 range that should do the job. The nice part is it's a closed system so I can easily calculate the weight of the water below the pump and subtract the weight of the water above the pump. The net sum would be the "load" on the pump and it would not change since those weights will be constant as the water moves through the system. Would you agree or am I missing something?

I'm trying to remember the term they use..Lift,rise? for the amount of feet a pump will lift water..That's the key not volume if I remember correctly..They have many types of pumps..
Good luck

[OldChap]

All you need to do is think of the geo loop as a radiator.

Pump head on that loop is determined by the pressure drop in the circuit ....most of the restriction will be as a result of length of pipework.

A sealed system will not need pump head to overcome the height of the installation.



Fitting full flow valves will assist you during maintenance

Yes you will need to bleed the air from the geo loop so a bubble catcher as you call it will be needed in some form or other

I cannot see a need for any pressure release system

You may need to re-think the actual heat exchanger ...alu is not great to mix with typical copper blocks on a system

A copper coil or two might be a better way for the well but the well diameter will perhaps determine this

[madmaxx]

Should be able to find flat copper tubing & a couple Iwakiis. Should do the trick?

Mixing metals = No No and flat or "square" tube is more efficient at transferring heat... IIRC

[mhumm2]

Thanks madmaxx, and OldChap. As far as the pump draw, I agree with you OldChap that since it's closed loop, the water above will offest the weight of the water below. What I don't understand is "pressure drop." i thought the pressure in the system would be virtually static with the pump(s) running. What am I missing?

Bubble catcher; i was thinking of a conventional dual-height disk resevoir placed on the input side (before the CPU). It would be the highest point in the system of coarse and have an air vent on top. I assume the air vents sold for liquid cooling systems are air check valves, it that correct? Is it possible the liquid could escape through the air vent if the level gets too high?

I suppose I could use a coil of copper tubing or some vertical square copper tubes. Hmmm, copper would cost more. More to think about. I understand the dissimilar metals issue, but how can it be an issue if the metals are not in physical contact with one another? I've seen cathodic corrosion in the electronics I work on, but again, that's due to the metals (usually a washer or bolt) coming into contact with the wrong type of metal.

I haven't calculated the volume, yet, but store-bought coolant is going to be expensive. What other options do I have? Can I use the same coolant that vehicles do, like Prestone and such?

[OldChap]

My links will be for UK but so that you get the idea of what I am talking about...

http://www.google.co.uk/search?sourc...=1852&bih=1049

It might be possible to source a bronze circulating pump for this otherwise plastic/ceramic. If you choose to use steel (painted) then you will need to add an anti oxidant. Antifreeze as used in cars could do that mixed to about 10%. It reduces the heat transfer of water a little but protects from rust etc. Using this may also allow the use of alu but you also have to think about the corrosive nature of the groundwater on the outside of your pipes.

Numerous other Geo builds use plastic water pipe underground. Not sure how this would pan out for a single well type though.

EDIT: Take a look at something like an EK multi res reservoir, there are just pipe connections and blanks on it, no valves. Placing a res at the top of a system just makes life easier but is not strictly necessary.

Pressure drop: resistance of pipe and any fittings to flow will cause this.

Even if you use distilled water it does not take long before it has taken on contaminants, thereby making it conductive and thus connecting the dissimilar metals

[AMDforME]

Even though I fail to see the point in this system... I'll offer a couple comments.

1. A sprinkler system typically uses a high pressure pump so you will need appropriate pressure reduction via duct area restrictions or some other device

2. A pump still needs to be have the proper lift capacity (maximum head) for 25+ feet or whatever height there is from the well to the pump or it will cavitate and stop pumping at some point. The water volume in the system has nothing to do with pump lift capability nor cavitation. Yes a closed system most definitely will cavitate if the pump doesn't have sufficient lift capability.

http://www.mopsdirect.us/skus/wp/Danner_Flow_Chart.asp

3. You could have the pump below ground for minimal lift but you'd need proper electrical and pump protection

4. The power you are going to consume to run the pump(s) in this geo system will be far more than any PC cooling system many times over, should that matter to you

5. Metal piping has the best heat transfer but it corrodes over time even with coolant thus you may want a micro filter to remove most of the crap

6. This system is a lot of time, money and effort when a proper air-cooled, phase change or refrigerant system could also be quiet, reliable and almost maintenance free - just saying...

[OldChap]

???

?? 2. What does the height of the well to the pump have to do with it?? A closed circuit will not be working to lift the column of water.... just circulate it

4. I ran with a 40W pump no issues. http://www.grundfos.com/products/fin...ct/alpha2.html

6. Agreed, You could likely achieve your goal with a Mo-Ra or Supernova ..... but where is the fun in that??

[lowfat]

This is a sealed loop, you definitely wouldn't need a pump capable of 25 ft of head. Any single Iwaki would have no problem. A pair of DDCs would even none sufficient.

A geothermal loop will also get sub ambient temps, something that is impossible for normal loops.

Plus it could be perfectly silent, not just quiet. Another thing that really isn't possible for normal loops. If you think phase can be quiet you are nuts, even a 'quiet' one is very loud.

If he uses copper piping for thr loop he should be fine, shouldn't ever have to worry about corrosion.

[CrazyNutz]

Run about 30 feet of copper (do not use aluminum) tubing down then back up. Put a fill port at the top (perhaps one on the inlet, and outlet). You don't need a monster pump either as lowfat said, It's closed loop so you have an equilibrium.

[mhumm2]

I can't believe this is finally coming together. First of all, to AMDforME -- Thank you for the input and to address your not understanding why question, the answer is (in order of priority):

1. To run my system silently
2. To mitigate dust cause by moving air through the case
3. Bragging rights. Please don't dismiss this one. How many people do you think have a geothermally-cooled computer system? I don't know either, but NOT MANY!!! I also want to see if I can make this work. And from the inputs I'm getting on this forum, I'm inspired. Thank you all.

As to the power considerations, the only electrical load will be 1 or maybe 2 pumps vs. two-120mm, two-80mm, and one-60mm fan. And that's not to mention the two fans in my PSU which will also benefit from cooler case air temps. I really don't expect a big increase in load capacity; at least not "many times over" I think.

Thank you for explaining the pressure drops caused by fittings, and transitions of all kinds. That makes total sense.

OldChap, "Where is the fun in that?" Exactly sir! For I know this man's heart and it is young!!!

CrazyNuts, Wow, 30 feet of copper tubing?!? I understand that with copper I don't have to worry about dissimilar metals in the system and I don't have to worry about corrosion. The copper will corrode (oxidize) by turning black then stop. What about diameter of the tubing. I know that the more surface area the more cooling so would you recommend 1/2" copper tubing ?

I think the well pipe will be a 4" schedule 40 pvc pipe. I expect to have about 4 or 5 feet of water at the bottom of the well pipe. So a 5-foot cylinder is the space limitation for the geo-radiator. What do you think about spiraling the tubbing to fit in the pipe? What's the maximum radius for 1/2" copper tubing? I also think the input to the geo-rad should be at the bottom of the spiral and the output at the top. Does everyone agree? This way natural convection is aiding the effort and not fighting it.

More to come...

[shazza]

Good Luck with your project! I remember someone doing something similar - can even remember pictures of a backhoe doing some trenching. I can't recall if he finished the project (but I think he did). I just looked through the work logs, but can't find the project - maybe someone else with a better memory can recall it (it would be a project from more than 3 years ago ...)

[lowfat]

Good Luck with your project! I remember someone doing something similar - can even remember pictures of a backhoe doing some trenching. I can't recall if he finished the project (but I think he did). I just looked through the work logs, but can't find the project - maybe someone else with a better memory can recall it (it would be a project from more than 3 years ago ...)

http://www.xtremesystems.org/forums/...mal-PC-Cooling

[OldChap]

...or a bit smaller....

http://www.xtremesystems.org/forums/...ght=geothermal

[shazza]

Thanks guys - that's the one I remembered, lowfat.

[CrazyNutz]

CrazyNuts, Wow, 30 feet of copper tubing?!? I understand that with copper I don't have to worry about dissimilar metals in the system and I don't have to worry about corrosion. The copper will corrode (oxidize) by turning black then stop. What about diameter of the tubing. I know that the more surface area the more cooling so would you recommend 1/2" copper tubing ?

You reach the water level at 15feet so 30 feet down and back up will give you 30 feet under water. If you think that's overkill, well you are on xtremesystems right? However you would still be doing just fine with 20 feet.

Also the copper tubing will last a long time. Probably far longer than you care to run a water cooling loop through.

And I would use 1/2 inch tubing.

[mhumm2]

Shazza and OldChap... yep I read both of those from beginning to end. CrazyNutz... when you're right, you're right. 15 * 2 = 30. Of course the water level at the new well site has to cooperate. If I don't hit water, i have to go deeper. When I baled the well for my sprinkler system, I hit water at about 18 feet, but kept going. I hit hard pan at 30 feet and broke through adding 10 more feet before setting the well foot. The water I pump through my system doesn't stain my driveway or house (no iron deposits) and it doesn't smell bad either from sulfur like shallow well can. For this project, the less depth I have to bale, the better.

I went to Lowe's today and made some measurements (and did some price checks ). If I use 3/4" copper pipe and put 6 pipes together (each pipe is two-feet long), that would be 12 feet of geo-rad and I'd only have to bale the well deep enough to get about 3 feet of standing water. Six, 3/4" copper pipes fit quite nicely inside a 3" PVC pipe. I think the next step is to figure out how big the georad needs to be in order to dissipate the heat I want removed from my system.

I would like to remove 60 Celsius degrees (which is different than the measurement of 60C); 30 from my CPU and 30 from my GPU. The break down doesn't really matter since the CPU and GPU water blocks will be in series. 60 Celsius degrees = 108 Btus. According to Wiki, the temperature conductivity of copper is 231. So I have some number to crunch.

[bds71]

question: for a geothermal loop, would the ambient temperature underground be enough to overcome the amount of heat being introduced? ie: how much heat would you need to introduce to an underground loop/well to raise the temperature of xx gallons of water xx degrees? if you ran a passive loop to a well, how long would it take to heat up 20 gallons (for example) of well water? would it stabalize at some point?

[OldChap]

Generally at a given depth underground (depending on where you live) the temperature remains the same summer and winter. with this in mind all you need is enough area of pipe in contact with the ground/water to dissipate the heat generated. There will be heat losses in the pipe going to the well if ambient is below the coolant temperature so what remains is a certain number of watts to dissipate through say a coil in a well. There will be a heating effect on the well water by the coil and there will be a cooling effect on the well water as heat is dissipated into the surrounding ground. The greater the difference between the temperatures of coil and well water or between well water and surrounding earth then the greater the ability to dissipate heat.

Conversely, if there is very little difference between the temperatures then the ability to dump heat is reduced.

As the earth/well water is not moving this is a passive cooling system and as such yes it will reach equilibrium provided the initial design is sufficient for the needed head dump. By this I mean that what might be achieved by a 10ft coil immersed in a stream could take 100 ft of coil just buried.

The clever bit when building such a setup is to figure out the heat load capable of being dissipated by a slug of water of x diameter into ground of y composition and whether the ground too can dissipate the heat.

I think it is easier to figure when using long pipes under a large area of ground but maybe that is just me.

How long to heat well water? The theory goes 1w = 1cc water by 1 deg C in 1 second but in reality it takes longer again because of the temperature differential between the two liquids and the losses to the surroundings

[bds71]

...provided the initial design is sufficient for the needed head dump.

i did a (brief) search on google to see if i could find anything, and i couldn't find anything simplified enough or specific enough to what i'm looking for. how would i calculate how much (underground) copper i would need for a given wattage of heat? (for example: 700W for OC 3930+GTX690) assume buried copper rather than well/stream.

I plan to imerse connected aluminum pipes as a heat exchanger then use the ground water to cool it.

OP: how much pipe (Al or Cu) are you planning on using, and what is your heat load? how did you come by these numbers?