>>
|
164c83.jpg
Grey Chanting Soda
164c83
Formerly spoilered posts that should have gone in a discussion thread posted here:
(suggestion)
https://en.wikipedia.org/wiki/Geothermal_power
I'm not a specialist, I've just done non-specialist reading on the subject and had some mildly-related development/consulting/security work in the area. The wiki page should explain in sufficient detail for you to know what I'm talking about at least. One key issue that jumps out at me as a semi-layman is that the heat source is a geothermal hotspot that heats water or some other working medium to near boiling or pressurized-steam levels and that this then has to be used or else you potentially run risks of leaks or broader equipment failure issues from heat. Who knows how well the piping, containment vessels and passive radiation features involved are holding up after an apparent attack of some kind, and an unspecified period of management downtime during which there may have been geological instability events that disturbed things. Another issue is that sitting a computing unit near a geothermal plant seems a bit weird because they extract heat from below the ground and use the temperature differential between that and the open air (or perhaps a surface body of water) around to generate movement, and therefore electricity. Because of this a geothermal plant would always be detectably warmer than ambient, whereas most of the time we want computing bits to be in cool, barely-moist-enough-to-avoid-static environments as free from particulates as possible because of how fragile and finicky cheap consumer electronics can be. If the computing is working on future tech optical systems or some other raw-macguffinite technology that is far more tolerant of heat issues than shitty bulk Intel and AMD chips of today that may not be as much of an issue. If you have questions left after reading that or don't feel confident bullshitting some reasonably-plausible tech the tech stuff about it uh... ask specific questions?
Seven01a19!!EyA2IwLwVl
But it all makes sense if the computers are on the cold side of the cooling loop. It's a matter of Carnot efficiency; the closer your energy extraction can get the water to ambient temperature, the more efficient your system is. It's all sources, sinks, and delta between them and absolute zero.
If the cooling water is passed through the core before going down into the crust then the core will be cooled better for being above a geothermal plant. I'm not sure how helpful this would be in a geothermal plant, but in other plant types it would be a form of energy recapture.
(suggestion)
I assumed away thermal energy recapture as sane based on what I know about low-human-interaction server rooms: My assumptions going in were average ambient temperature of 15C, ideal computing core operating temperature of 5C, and therefore it would be a straight up loss having your server room anywhere near the geo plant since by definition you can only asymptotically approach ambient through passive cooling. We don't know what the local fluctuations between night and day temperatures are, what we're using for thermal surface radiative sink (a deep fossil lake like Lake Ontario would be water which is 4C year round), what the local air temperature range is, or even what the computing infrastructure is and its healthy and tolerable ranges of operating conditions for power, temperature, particulates, electrostatic effects and so on.
Seven01a19!!EyA2IwLwVl
With a fluid cooling system 15c/59f is actually within the ideal operating range for a fluid cooled system, while 5c is not.
The key difference is that an air-cooled system convects heat away from the entire unit, while fluid cooled systems conduct AND convect heat away from the hot spots.
With an air cooled system the inside of the unit is going to be warmer than the outside air, meaning that no condensation will form inside. This means that the cooler the air the better (Within reason), as the ambient temperature of the air is the sole determining factor of cooling power.
However, with a fluid cooled system you have a coolant that is likely NOT the same temperature as the ambient air, and is likely to be cooler. The further you get from ambient temperature, the more likely you are to get condensation on the liquid cooling pipes. At 5c you will have serious issues with condensation on the electronics, which is typically fatal. However, at 15c you're working at around room temperature, meaning that condensation won't occur in a structure with ordinary environmental controls.
That's just the effect of the environmental conditions. The real meat and potatoes of the difference between fluid cooling and air cooling is specific heat capacity and specific density. Mass for mass, water takes 4x more heat than air to heat the same amount. However, water is also about a thousand times denser than air, meaning that everything else being equal the cooling of a fan blowing 12cu ft of air (~340L) per minute is equaled by a
pump pushing 0.3L per minute.
Then there's the convection vs conductivity issue. Heat transfers into water a LOT faster than it transmits into air, meaning that an air-cooled system needs to keep pushing warm air off the heat exchange surface to prevent it from forming an insulating layer. As water is an adequate conductor of heat this is essentially a non-issue in water cooling systems, allowing for much lower flow rates.
|