Proposal: Vasline cooling

[madd0ct0r]

ok.

I find myself drawn, again and again, to immersion cooling solutions.

The basic idea covered in a thread further down the page.

Basically, take 1 PC, string out any optical disk drives and HDDs and submerge the rest in mineral oil.

The idea being that the oil completely surrounds the heat sinks; is more thermally conductive then air (so heat transfer happens faster); and has a higher specific heat capacity then air (ie, it takes more thermal energy to warm the same volume)
Mineral oil is transparent, doesn't go rancid and shouldn't dissolve your hardware. It also seems like a simple thing to setup (compared to water cooling). Noise reduction is always a pleasant bonus.

The best quality discussion I've seen is here. Their first build is now well into it's second year with no failure. Even the fans are still turning!


Problems:

Wicking - Due to the capillary effect, oil get pulled up the gap between a cable's insulation, and it's core. Or the gaps between the copper fibres.
This means, unless precautions are taken, your mouse, keyboard and power cord will eventually start leaking oil.


Cooling the oil - most oil rigs seem to be in clear (glass or plastic) boxes to show it off. The oil-box-air interface isn't exactly efficient, and the stable running temp can be scarily high. The Puget system ran (stably) at 88degress without additional cooling.

The normal solution is just to pump the oil through a radiator (normally with low speed fans on it). The cooled oil is then returned to the tank.


Weight. - an oil filled PC is heavy. Ever picked up a fishtank?




I propose a metal case, with internal and external fins to allow heat to be dumped as fast as possible.

New Idea (I think) is to replace the oil with Vaseline (Petroleum Jelly)
This would have the advantage of solidifying when the PC was cool, so no risk of leaks, and no risk of wicking.
(And instant shock absorber if you have to move your pc)

Vaseline melts at 35-60 deg C
This would make it a phase change material, as well as having a high temperature capacity when it's liquid.
I'm just worried about potential heat gradients in the initial phase.
Anyone got any good ideas for testing it?

[d_stilgar]

I would say that you build two identical super cheap computers and run one with mineral oil and one with Vaseline. Or start off doing tests with mineral oil and later replace it with Vaseline and do tests again.

[Luke122]

I would say "ewww".

Immersion cooling, while very.. umm.. "cool", is (at least in my mind) completely impractical, messy, and cannot be easily undone. Once something has been submerged in oil, the re-useability of that item is greatly diminished.

Still, I whole-heartedly support you in your quest, and look forward to seeing this project!

[x88x]

Hmm, an interesting proposal. I've never heard of wicking being a problem in a total submersion system, but I guess I've never actually seen a writeup from anyone who has run one for a long period of time before.

I do have a couple concerns with regards to using vaseline though. The main one is that is has such a high melting point (~75C, according to wikipedia). Even if we assume you have a variant that has a melting point of 35C, it would be of concern. The thing to keep in mind is that the entire volume of vaseline is not going to raise in temperature at the same time. I cannot find a reference to its conductivity rating, but even if it were very high, it would still take a long time for the entire tank to raise above the melting point. The vaseline would melt, starting at the points of heat generation (CPU, GPU, etc), and gradually spreading. I could definitely be wrong, but I do not think that it would melt fast enough to properly dissipate the heat from these elements. Additionally, any fans/pumps/etc that you put in to move the fluid around (to disperse the heat) would have to be started after all of the fluid melts. One way that you could get around this is by incorporating a heating system to warm the vaseline to its melting point before starting the computer.

Another issue I see arising is actually one of the benefits that you mentioned. If a small crack were to form while the system was turned off, yes, it would be great that it would not leak all over the place, but if you did not discover the crack before turning it on, it would be discovered as the jelly around it melted...most likely once the system was fully functioning, and possibly causing damage to the components (depending on how well they fare out of the fluid). It occurs to me, though, that you mentioned that you wanted to make a metal case, so this might not be much of an issue.

All that said, I am by no means an expert on the subject, and I'll be very interested to see the results of your testing if you go ahead with it.

[billygoat333]

you would have to keep the case thin, or else the like x88x said, the outside of the vaseline would stay rock solid and you wouldnt have very good thermal transfer at all, basically causing the cpu to be insulated for a while, causing high initial temps. I would think that it would cause the cpu to heat up really fast, with no cooling til its at its breaking point, which even if the vaseline melted at that point, you are still stressing the cpu. (if that makes sense. basically it seems like it would be too much of an insulator before it started actually conducting the heat away from the hotspots.)


Maybe find an old junker pIII system or something for free and experiment!

[madd0ct0r]

Well, my current computer (and the one before that) was free, but I can't afford to blow up another one for a while.

following wikipedia's references:
http://www.inchem.org/documents/icsc/icsc/eics1440.htm
gives a melting point of 35-60 deg C, which makes sense for a blend of hydrocarbons.

http://www.absoluteastronomy.com/top...l_conductivity
gives a thermal conductivity table. Vaseline isn't specified, but LPG and Mineral oil is, clearly showing that thermal conductivity go's DOWN as the hydrocaron chains get longer.

http://wapedia.mobi/en/Petroleum
Section 3 gives us all the equations I need to model it.


@Billygoat333 - You're right. It's the initial start up phase I'm worried about, not the steady state equilibrium with nice convection currents set up.

@x88x - And just like that, a benefit becomes an issue! ;-) hope some of the stuff at the start of the post allows you to speculate a little more.

[mDust]

This is a very interesting idea. I honestly don't see it working out, but if you do have some hardware you don't mind putting at risk, then test it. Make sure to be running some software that tracks cpu, gpu, and system temps in real time. If the cpu hits 80C = FAIL. Same with 100C for the gpu.
I think the lack of fluidity will cause a hot bubble to form before there is enough convection taking place to cool the components. It really depends on the melting temp of the vaseline...the lower the better.
The oil cools well because as soon as it absorbs heat from the parts, it rises out of the way and cool oil replaces it. I'm not entirely sure vaseline is capable of that unless it's preheated...but then that defeats the purpose of cooling your system. Experiment though...maybe you'll surprise me.

And as for getting mineral oil off of computer parts: soapy distilled water works very well. It's quite easy. I've washed the dust and crud off of motherboards, graphics cards, and other pci devices with soapy distilled water, a distilled rinse, and a 99% isopropyl bath as the final rinse. The isopropyl is optional but it evaporates much faster than water and I didn't have several days to wait. Just make sure it's completely dried out from under the PCB components before reinstalling it. Everything that worked before I washed it still works today.
I don't know how to get vaseline off...maybe just isopropyl and a brush? Good luck with that.

[x88x]

I don't know how to get vaseline off...maybe just isopropyl and a brush?

Maybe with a heatgun first to melt the vaseline?

[Spawn-Inc]

Maybe with a heatgun first to melt the vaseline?

i was thinking a blow torch but i think that would be safer