distilled water + oil submerged, would this work?

Several problems:

1) Mineral Oil is less dense than water, thus it would float on the water, not the other way around as you describe it.
2) The turbulence of the fluids inside the tank, which would be necessary to adequately cool the components, would cause the fluids to mix. Pure distilled water may not conduct electricity, but water is easily contaminated and will conduct electricity quite easily. This is bad for any component that comes in contact with water while charged.
3) I'm not even sure what your trying to accomplish by doing this. Water is so much better at removing heat without heating up itself that it just makes more sense to use regular water cooling than some weird oil/water hybrid system that doesn't even obey physics. If you really wanted to do water cooled oil cooling, you should put the computer in a tank with oil only and have a pump (or two or three) pumping the oil to a liquid-liquid heat exchanger outside of the tank and use water (or ice water) as your second fluid. This is overly complex and inefficient, though, and you'd see better performance for less money with straight up water cooling.

And to answer your question as to why water is so prone to contamination is because of its molecular structure. It dissolves a lot of substances quite readily. They don't call water the "universal solvent" for nothing. The reason completely sealing it into a chassis doesn't work is because the water will actually dissolve (very) small amounts of either the chassis or the computer itself, thus making it electrically conductive. It's not much, but enough to ruin a rig right quick. Ever drink water from a soda can? It tastes slightly metallic. That's because small amounts of the can are now in the water, and that's what you're tasting. The reason you don't get that taste from those metal water bottles is because the inside of those are lined with an insoluble plastic which is commonly used in bottling.

wouldnt be worth it, if u did get it to work...

just better todo the normal oil submersion thing.
or watercooling

when u pump water back in, it could push the oil out of the way, and get some water on ur comp gear?
nice idea not sure if will be worth it though, not sure if itll even cool the oil down much

Not sure if what you want to do will work, but I know that Puget Systems sells do-it-yourself Mineral Oil Aquarium kits that might save you alot of time.

Heres the link, http://www.pugetsystems.com/submerged.php
They also explain alot about mineral oil builds.
Edited by WizrdSleevz - 7/21/12 at 11:37pm

This is all true... except I feel I should inform the poster of this particular post of something:

Submerged cooling is not the same as a liquid cooling loop. With submerged cooling, the components are sitting in a bath of the liquid, with direct contact, rather than the liquid being tubed and flowing through metal heat exchangers (waterblocks).

There are a few reasons to do such a thing, which cannot be done with traditional liquid cooling (or in some cases, can not be practical with traditional liquid cooling). The most common substance to use with submerged cooling is mineral oil.

One of the downsides is that many people believe the high viscosity of mineral oil will cause pumps used to circulate the liquid will prematurely burn out. Another issue is that it is very difficult to clean out a radiator that has been filled with oil.




Now, to address the issue more directly:

No, in addition to the reasons posted above, even if you were to separate the oil and water in such a manner that they do not mix at all AND only the oil touches the electronics, it still wouldn't work. The reasons being:
1) the oil portion would not circulate, thus creating hot spots and removing much of the benefit of submerged cooling.
2) the non circulating oil will not effectively move heat to the surface that is in contact with the flowing water.
3) tiny air bubbles will get trapped between the two layers of liquid, causing poor transfer of heat between the two.
4) the surface area between the oil and the water will be minimal, causing poor heat transfer.

IF you are thinking about this in order to save a pump, don't worry! A high quality pump can handle mineral oil without issue. There are many pumps designed to move things besides just water.

If you are worried about oil getting stuck in your radiator... Well, tough luck there. Best thing I can tell you is that submerged cooling has it's costs. One of them is sacrificing a radiator to never be used with non-oil liquids.

Yea, I had a submerged build for about a year. The pumps will be fine submerged. I had 2 DDC 3.25s in mine, running in parallel, pumping the oil out through a 2x2x120mm rad. This thing is a monster and it's currently serving on the rebuild which is just traditional water cooling. (check sig for build log)

The oil circulation in the tank is easily accomplished with some 60mm fans. Smaller diameter fans seem to work better than the larger diameter fans, according to Puget Systems (I used one of their DIY kits, it was a freaking life saver). The oil does a damn good job of muffling the noise of the fans, so you can get the highest rpm fans you can find. A small bubble bar adds a nice aquarium feel to it, while adding the benefit of being able to see how the oil is moving in the tank. Also, in a submerged build, you'll be using typical air cooling heatsinks so you'll have your cpu and video card heatsinks circulating the oil as well. The fans will be fine, as long as they're powerful enough to get started in the oil. I had the fan on my PSU spinning slow enough to see the fan blades. It was probably spinning at around 30-40rpm. Didn't do much, but certainly looked cool. That PSU and fan are still going strong in the current build.

You need to have the fans to circulate the oil, because the natural convection created by the hotter oil wanting to rise in the cooler oil is simply not enough to get the heat out.

I'm curious as to how air bubbles would get trapped between the oil and water. Air will always float to the surface of the water. Surface tension is not enough to keep any air bubbles that might miraculously show up at the fluid interface from rising through to the top.

Larger fans typically are lower static pressure and lower rpm fans. They rely on their surface area to move large ammounts of air quietly. If they had the same speed and pressure, they would be significantly louder than their smaller counterparts.

I believe this is the reason smaller fans usually do better in the vicious oil. Panaflos, San Ace, or Delta server fans should do even better, albeit at a potentially higher cost and noise level, and certainly aren't as flashy as LED fans.


As for the air bubbles, I do't mean to imply that they will get trapped for a particularly long period of time, but even briefly is long enough to momentarily deaden heat transfer in a given spot. The turbulence in the water would pretty much ensure that tiny microbubbles would find their way to the boundary of the two liquids constantly, assuming the whole thing isn't 100% filled and then vacuum sealed or something.

At any rate that's all irrelevant as it wouldn't work for other reasons as we both stated

Proper submersion can have significantly better results than water cooling. While the water itself has better thermal properties than oil, you can't bathe the whole rig in it.

With water, you need to rely on small metal heat exchangers (waterblocks). There can only be attached to so many surfaces from which to remove heat. The surface area in contact with the heat producing components is actually pretty small it it relies on the heat getting pulled off by the water quickly enough that it doesn't get to build up in the areas that aren't exposed to the waterblock. Furthermore, you have 2 layers of between the head producing component and the heat-transfer material (water) in the form of metal (waterblock) and TIM.

With oil, none of that is true. You go straight from component to transfer materiel (oil), you get significantly larger surface area (If you toss a heatsink on the cpu, gpu, etc.) and everything is being cooled at the same time, eliminating any and all hot spots.


You also get the benefit of the very large volume of oil acting as a buffer to resist changes in temp. It's like freezing a lake vs freezing a cup of water. They'll both eventually freeze, but the lake will take a LOT longer. If that lake is also getting heated (or in this case the oil getting cooled by a rad), rock-steady temps can be achieved everywhere other than the point of heating and cooling.










If you really want to "improve" this idea, there are a couple ways. None of them involve mixing fluids.

The first way is to use a fluid created by 3M specifically for this task (but for supercomputers). It flows like water, has even better thermal properties, no tension, and is totally non conductive, non capacitive, non-corroding, and won't break down over time. It just costs about a thousand bucks to fill a small tank.


The other method is to combine tubed water cooling with oil submersion. Directly pipe cold water to the most important parts as you normally would with liquid cooling, but then put the whole thing in oil. Warning - the oil will eat normal liquid cooling tubes.

Yes and no. Mineral oil is far less effective at removing heat than water is. It's specific heat capacity (the property that really matters) is like 1/5 that of water. Without breaking out my heat transfer textbook, you'll need significantly more surface area (it's not a 1:1 trade-off between specific heat and surface area), so you would absolutely need to run an air cooling heatsink in your submersion build. You may be able to get results comparable to water cooling, but you won't be able to surpass it. If you're not running a heatsink, you're not going to be able to get enough heat out of the components fast enough and things are going to break. I had my cpu heatsink come off the cpu because a clip had broken. My temperature immediately shot up to 60°C (and climbed up to 62°C by the time the shutdown sequence started) and the cpu underclocked and saved itself from total destruction. Heatsinks are not optional in oil submerged builds.

I'm currently running about one liter of water in my rig. When I was running a submersion build, I was running 4 gallons of oil. I simply couldn't turn the oil in the tank over fast enough. I'm getting about 1.6gpm flow rate, which is slightly higher than what I was getting with the oil rig. I'm moving 100% of the fluid through the radiator every 10 seconds. It took about 4 minutes to move 100% of the oil through the radiator. This means that my water is almost always around the same temperature; conversely, the oil heated up noticeably and my temperatures were about on par with air cooling after about 20 minutes of use. Gaming and other heavy load situations compounded the problem. Because the oil would heat up, my component temperatures would also rise, and my gaming sessions were limited to about an hour and I had to let the computer idle for at least an hour between games or else everything would get too hot and I would run the risk of the tank cracking due to the heat it was never designed for (happened twice, which is why I'm not running oil cooled anymore). Now, perhaps there were things that could have mitigated this, like better heatsinks and better fan radiators, and maybe I'll revisit that experiment in the future with those improvements.

Now let's talk about system maintenance and upgrades. Warranties do not exist in the realm of oil submerged cooling. If you get a new part, you better make sure it works before dunking that thing in oil. Companies are going to laugh at you if you ask for an RMA on a part you dunked in oil. Maintenance is a complicated issue. There's no dust to worry about clogging up your heatsinks, so that's nice. But if you want to add, subtract, or change anything, it's going to be an all-day affair. Oil will be all over everything. You're going to want to let it drip back into the tank before you do any actual work, which is going to take a while. So there's an hour or so gone before you've even done anything. And if your PSU isn't mounted to the motherboard tray somehow, you've got a crap-ton of wiring mess to deal with, too, and you're going to get covered in oil up to your elbows. You don't want the oil to get on anything that isn't your computer, so you need to wash your hands and arms constantly. So after you've washed your hands for the 10th time in 5 minutes, and waited an hour for the excess oil to drip back into the tank, you're ready to do whatever it is you wanted to do to begin with. So do that and then give the computer an air-cooled test run, because you don't want to have to go through this process all over again if something doesn't work out as expected. If all is well, put the computer back in the oil. If the PSU isn't mounted to the mobo tray somehow, you're going to need to get your arms covered in oil again to arrange the wires and stuff. Also, be careful because acrylic edges can be very sharp and cut you quite deep. Blood is not something you want inside the tank.

"Better thermal properties than water" is rather dubious and misleading. Its specific heat is not as good as water's, which is what makes water so good as a coolant. The fluid you're talking about (Novatec, I believe) works so great because its boiling point is slightly above room temperature. This means that the fluid is always at a known temperature (37°C I believe) and the heat can move through the IHS of the cpu fast enough to make the use of an air cooling heatsink unnecessary. You also need to hermetically seal the tank or else you're going to be breathing that stuff, and it's very bad for you if inhaled.