Sign in with
Sign up | Sign in
Your question

Mineral Oil Cooling to Become Mainstream?

Last response: in Overclocking
Share
a b K Overclocking
September 5, 2012 5:25:26 PM

If anybody follows the TH news section, Intel has demonstrated server cooling using submerged mineral oil techniques:
http://www.tomshardware.com/news/Server-Cooling-Hardwar...

I wonder if this will spur on new development in the mainstream/enthusiast markets, and not just the server market. Maybe we'll all one day have aquarium PCs? :D 
a c 328 K Overclocking
September 5, 2012 5:28:04 PM

I just saw this and am bracing myself for some new interest in the subject. The article didn't mention how they used it and most people will automatically assume they can simply fill a bucket with mineral oil and submerge a PC and get the same results. What wasn't discussed was the circulation of the oil and how the oil itself was actually being cooled. All that server heat has to go somewhere and mineral oil isn't very good at dissipating heat once it's absorbed...especially passively into the surrounding air.

I really hate when writers fail to discuss important details such as this, even if they are speculative in nature or to the effect of 'these concepts have to be considered for this to actually work, but these items were not communicated due to secretive process.'

If you listen closely, you can hear the stampede of questions coming...
September 5, 2012 6:15:59 PM

Just throwing a random idea around. It's 2am over here and my brain isn't working well at this time. What about phase change or peltier plates? Or think about it this way, we fill a bar-fridge with mineral oil and build a PC inside. Ignore the part about the compressor frying due to the constant heat inside. Phase-change isn't too popular because of the condensation. But right now condensation won't matter already. Or did I miss something out?
Related resources
a c 235 K Overclocking
September 5, 2012 6:20:45 PM

I don't think I'll be buying any mineral oil any time soon!

It may save running a host of server cooling fans, but Intel doesn't overclock anything as far as I know, so what effects does overclocking bring to the mineral oil table?

Step right up and get your roasted mineral oil!

It's just hoop lah dodo! :lol: 
a b K Overclocking
September 5, 2012 6:21:15 PM

rubix_1011 said:
I just saw this and am bracing myself for some new interest in the subject. The article didn't mention how they used it and most people will automatically assume they can simply fill a bucket with mineral oil and submerge a PC and get the same results. What wasn't discussed was the circulation of the oil and how the oil itself was actually being cooled. All that server heat has to go somewhere and mineral oil isn't very good at dissipating heat once it's absorbed...especially passively into the surrounding air.

I really hate when writers fail to discuss important details such as this, even if they are speculative in nature or to the effect of 'these concepts have to be considered for this to actually work, but these items were not communicated due to secretive process.'

If you listen closely, you can hear the stampede of questions coming...


Yeah, I feel you on that one. I think Intel's going to have clients for this 'service' and I'm betting they're keeping it very quiet due to that. Also, it raised some discussion on new PCBs since traditional PCBs will swell up and disintegrate due to the oil.

I'm wondering what sort of regular consumer products might develop from these sorts of advancements.
a b K Overclocking
September 5, 2012 6:48:04 PM

For all the reasons mentioned above, and then some, mineral oil cooling will never become "mainstream." "Mainstream" requires nothing more than the cheap, stock Intel air cooler. As process nodes continue to shrink, power use (and heat) will only go down. Even if not for ten or twenty years, the day will come when a PC as powerful as what we have today will run on a couple of flashlight batteries and generate no discernible heat.
a c 328 K Overclocking
September 5, 2012 6:52:05 PM

In all honesty, watercooling- even on an enterprise scale is still far more feasible than submerged mineral oil.

I need to find the link to the article where (Google?) was using enterprise-wide water cooling with a nearby reservoir or ocean inlet.
a c 235 K Overclocking
September 5, 2012 10:19:35 PM

I don't see this ever becoming mainstream, it would be interesting to see just how much this entire system has cost Intel in the first place?

After the initial investment it would take quite a while before savings started to show!

Quote:
On the other hand, it can be very difficult or almost impossible to clean and remove mineral oil from the hardware once it has been submerged.


So much for hardware warranty!
a c 150 K Overclocking
September 5, 2012 11:06:14 PM

rubix_1011 said:
In all honesty, watercooling- even on an enterprise scale is still far more feasible than submerged mineral oil.

I need to find the link to the article where (Google?) was using enterprise-wide water cooling with a nearby reservoir or ocean inlet.

IBM has been watercooling their servers for years!
September 5, 2012 11:53:56 PM

rubix_1011 said:
In all honesty, watercooling- even on an enterprise scale is still far more feasible than submerged mineral oil.

I need to find the link to the article where (Google?) was using enterprise-wide water cooling with a nearby reservoir or ocean inlet.


heres the google article
http://www.tomshardware.com/news/google-seawater-coolin...
here's the ibm article:
http://www-03.ibm.com/press/us/en/pressrelease/38065.ws...


the google has a different use of the term watercooling, as it is cooling the air of the servers/serverroom, not directly cooling the server components.

From thermodynamics perspective, the same amount of heat is generated by the servers- the change is what is the efficient way to get rid of that heat.

The leap forward with watercooling changes the energy storage medium from air to water, which is more efficient to perform a transfer of energy back out.
Then, the additional leap forward here with the mineral oil is complete submersion so you don't need to worry about tubing and individual pumps and leaks, as well as probably even better heat transfer due to direct complete contact

In the google case, they are using seawater to do a water-based heat exchange of air, but if they do the same heat exchange but on water or oil as the direct energy medium, it could be even more efficient.
a c 235 K Overclocking
September 6, 2012 10:30:11 AM

raytseng said:
Then, the additional leap forward here with the mineral oil is complete submersion so you don't need to worry about tubing and individual pumps and leaks, as well as probably even better heat transfer due to direct complete contact.


Well actually you do have those concerns, mineral oil has to flow to move the heat away from the component producing the heat, if the mineral oil stays static it holds the heat in the area.

So there is pump and tubing involved in mineral oil cooling, click on boiler1990s link, look at the bottom picture to the left you see a mineral oil stream, look lower left and you'll see tubing, that's an active flowing system.

There has to be some kind of flow to circulate the mineral oil, and if you have a mineral oil leak from your containment vessel, pump, or tubing, you have a major clean up problem on your hands.
a c 328 K Overclocking
September 6, 2012 1:46:35 PM

^Correct; just because the server/PC is submerged doesn't mean that mineral oil can passively disperse the heat it's absorbing. You still need flow and you need a heat exchanger. Mineral takes longer to absorb heat and also takes longer to dissipate it- otherwise known as thermal conductivity of a liquid.
a c 190 K Overclocking
September 6, 2012 4:50:59 PM

**Then, the additional leap forward here with the mineral oil**
Forward, I don't think this word means what you think it means...
from water to oil is backwards, a regression in efficiency, a McFail without fries
in laymans terms 'A stupid move'
I commented on .co.uk sites article and I see that Rubix, Ryan and Boiler all share my view, we are in for a lot of 'imma Oilkewl ma Riggz yo' threads,
We have a specific section of the sticky to put those threads in as we are bored of repeatedly explaining how bad oil is and why,
Here is a link for that
http://www.tomshardware.co.uk/forum/276288-11-mineral-l...

(I think this thread should be added btw Rubix)
One last time, and I shall never comment on another oil involving thread,

Oil is bad, do it for a prank on old systems for proof of concept if you want,
but don't try it with a new top of the game build and then cry when it all goes wrong,
We won't care by then :) 
Moto
a c 235 K Overclocking
September 9, 2012 12:30:19 PM

^ Quote from Jar Jar Binks, from Star Wars: Episode I - The Phantom Menace (1999)

Quote:
"Hmmm... yousa point is well seen. "
a b K Overclocking
September 9, 2012 5:02:08 PM

amuffin said:
IBM has been watercooling their servers for years!


Not just years, try decades. They used distilled water and they ran pretty good even with minor leaks.
a c 150 K Overclocking
September 9, 2012 6:27:45 PM

nforce4max said:
Not just years, try decades. They used distilled water and they ran pretty good even with minor leaks.

Exactly! I put years because I couldn't remember how many decades....should have just put decades LOL
a b K Overclocking
September 10, 2012 9:22:48 PM

amuffin said:
Exactly! I put years because I couldn't remember how many decades....should have just put decades LOL


I remember one of the old stories that one of my teachers had told the class at the college I used to attend had an IBM mainframe back when they were still top of the line. The system had an issue with coolant and would often run low every two weeks. IBM often had to send a maintenance tec to work on the system, they never did find the leak and he often used is tie to check how much was in the tank. They kept some of the drives after all these years, try 14inch platters! :sol:  They even had a Univac system decades ago only to have dumped it in a field somewhere out of town. :pfff: 
!