Mineral Oil Submerged Computer

[PCC- Builder Dan]

If you're considering that the use of air bubbles would make the oil cooler, don't bother.
The oil is taking FAR more heat off the computer components than equivalent volume of air would. Considering that, you'd need to blow so much bubbles that you'd make waves on the oil surface to get anywhere near the temp you're thinking of.

Actually... the bubbler lowered system temps by five degrees C, IIRC. When the oil was at 90 degrees C or so, bubbling 22-ish degree C air through it did cool it off a bit. The oil does soak up all the heat from the components, true, but then where does it go? The plexiglass fishtank is a great insulator. The top is all but sealed off, so the oil can't evaporate and wick away heat that way. All it can do is get hotter, and hotter, and hotter, until enough heat gets through the plexi that the temperature balances out. Pumping very cold bubbles of air through the oil gives the oil another place to store the heat: In the air. Unlike the oil, the now hot bubbles of air can rise to the surface, burst, and work their way out the top of the tank.

In short, bubble cooling the oil does work. A little bit. If we really wanted to keep temps down, we'd have dropped a pump into the tank, and pumped oil through a radiator on the outside, with fans.

[lovespuds1]

The reason that the bubbling of air through the mineral oil is lowering the temperature of the mineral oil is two fold.

1) Heat escapes through the surface of the water because it is the only non insulated surface. The bubbles circulate the oil in the tank so that the oil is transported past the hot components picking up the heat and moving it to the surface where it can escape. (the tank can't be fully sealed or the air would burst the tank if it could not escape, so warm air must be escaping)

2) When transferring heat between interfaces surface area makes a big difference, that’s why heat sinks have all those fins. With no air going in the tank the surface of the oil in the tank stays still. By passing air through the tank the surface of the oil is disturbed and moved this greatly increases its surface area therefore allowing more heat to escape. I know this because the same is true of fish tanks it's a common misconception that bubbling air through a fish tank gives fish oxygen (this actually adds very little oxygen) it's actually the disturbance of the surface that gives the fish more oxygen due to increased surface area at the top of the water.

So why did the graphics card temp rise?. My guess is that the air was actually being sucked into the graphics card cooler. The oil is actually a better conductor of heat than the air is so by sucking air through it you’re actually reducing the oil contact with the graphics cooler and insulting it. This could be tested in two ways move the bubbles away from the graphics card or use an air stone that produces larger size bubbles and move it slightly away from the card. The bigger bubbles will make it go up faster and hopefully keep them away from the graphics card. I know this because of studying physical chemistry (which was painful!) from an effect known as nucleate boiling.

How can you cool the tank more and make this a better long term pc?

Circulating the oil through an external radiator is one way but I doubt it would bring temps down a huge amount as it's to slow. A more efficient way would be to put a large flat shaped radiator behind the motherboard in the tank. Then pass water through the radiator, the water then passes out of the tank to a radiator outside the tank. Why is this more efficient than simply passing the oil through an external radiator?. Simple, water conducts heat much better than oil does and also the water is not as viscous as the oil so will circulate through the radiators much faster than the oil would. This allows it to transfer heat away faster. Inside the tank the air bubbles will circulate the heat and oil passed the radiator. I know this adds more expense.

Also using a more heat conducting top rather than acrylic such as glass would help. Although I know this was not used here because it is difficult to cut to shape. Also make sure cool lighting like leds are used so they dont add heat to the system.

Things to note:

--The air bubbling also increases evapouration of the lower molecular weight factions of the mineral oil.

--Keeping the oil cooler by using the radiator solution will reduce evaporation effects (higher the temperature the higher the evaporation)

The only way to see if the oil corrodes any rubber components is simply to watch this pc for a couple of months regular use.


What do you think?

[FinalStar]

Were either of the computers used as a normal computer before you put them in there? or were they all brand new?

My experience with liquid cooling (even with liquids that say they are none conductive) is if you take a board thats even a little dirty with dust or whatnot, when you start splashing it with stuff, it shorts it out.

Did you have to clean off anything?

[PCC-Jon]

Were either of the computers used as a normal computer before you put them in there? or were they all brand new?

Did you have to clean off anything?

The old hardware was in use for years and was not cleaned (well, we blew the dust off at least!). The new hardware was used for about two weeks, and was also not cleaned.

[washu9]

Just a few tidbits I'd like to add to this discussion.

To: PCC- Builder Dan,
Sorry about the comment on the air bubbles being ineffective. I was thinking of my own area climate. The Environment here can get rather hot during summers (up to 40-43 deg C).

Furthermore, there's usually a buildup of very fine fibers on my desk just in a week time. So in my case, the concern about dirt going into the system through the air pumped through is quite valid.

This is why I think it'd be very ineffective.


To: lovespuds1
Several things I'd like to add to his analysis here.

1) In regards to the graphic card, you'll get much bettter temp if you just open the upper lid of the video card heat sink (this is usually possible as most video card maker just use a thin aluminum/tin plate).

All current video card takes air from inside the case, blow it through the fins of the heat sink, through to grille on the slot, and outside the case. Since we got the video card standing vertically on this setup, and the fact that the fan would definitely be unable to move the oil past the oil level; at the very least, you need the oil level to be above the exit grille of the video card (which means putting the DVI connector in oil).

I'd advice removing the cover on the heat sink either way.


2) Since you're considering putting water into the cooling system, I have a more radical idea here.

What if we put both medium in the tank without a radiator (like you said). Oil and water don't mix, and the difference in mass would make water to stay on the bottom of the tank.

You can enhance the heat transfer further by adding fins in the transitional area between oil and water.

Here's the lowdown on my idea. (yes it requires all connector to the motherboard/video card to be submerged along with the system)

___ max oil level


___ motherboard position





___ end motherboard position
___ start fin

oil
___ transitional area ___


water

___ end fin (bottom tank)



Notes:
The width of the fin should cover the whole width of the tank - I'm just thinking a failsafe feature here so that in the event that the motherboard slip down, the fins would prevent it from going to the water side.

Furthermore, the fin should have holes all over it to promote circulation

The height of the water area should at least cover whatever pump you're using.

A pump in the water area will exchange water to an external cooling unit (like a radiator, or a TEC based aquarium cooling unit).

And before anyone ask, no, I don't think you can put any real fish in the water. there'd be barely any oxygen there since its surface is covered in oil. It's one of the reasons offshore oil spills are so deadly to marine animals.

It's a good idea to use a high head pump just in case the water evaporated to the point that oil will go into the pump and on to the radiator. In which case you'd just need to add water to regain normal temp. With this note, I'd also recommend having a tube going on the side of the tank to channel water directly to the bottom part of the tank as I'm quite certain you don't want to get any water going rouge on your precious motherboard/video card.

This is another reason I put the max oil level a length above the motherboard.


CONCERN:
My concern with this setup would be the escaping water vapor.
When the oil heats up the water enough, it will start to evaporate, making tiny bubbles.
These bubbles can be trapped in the system (got stuck on a video card or something) and return to water form after temp goes down enough; and with minor agitation, return to the bottom of the tank.

Now, this is "distilled water" we're talking here, so theoretically it shouldn't be a problem.
What do you think?

[kitizz]

Thanks washu! I'm fairly right with everything except sealing of the CPU, this is relatively new to me, and I've been reading numerous tutorials and I have a rough idea of how it works. But I would feel more confident if I could see a video of this being done. My attempts to find a video of it have currently failed, but while I still look, does anybody else know of any good tutorials or (even better) videos.

Thanks guys, Kitizz

Also, to add onto the water idea; with fans and what-not in the setup, you'd have to be careful that they don't stir up the water into the system, but it's a cool idea

[washu9]

Well that's yet another reason I wanted fins covering the transitional area between the oil and water. It will act as a buffer zone making extra sure that any agitation would not break the surface tension of the water/oil.

In theory, you can shake things up rather harshly and it still would not break surface tension as the force of the waves are buffeted by the fins.

If there's any water going rouge and floating about in the oil, then that means you haven't got enough fins.

Also, if you wanted to be extra sure, just place any fans in the system to face horizontally. I don't believe it to be necessary.

I don't know anyone who's tried this particular oil/water combination yet.
It just came up to me while reading lovespuds1's idea about putting a radiator in.

I believe there's a few video in youtube covering several oil PCs.
Several of them are using mineral oil, while others are using plain cooking oil.
I think there's 1 or 2 video covering some mishap that happened in an oil based setup.

[afie]

Hello people!

I've got a couple of questions to ask (Yes, really!)

First, and mainly, would high-end synthetic motor oil work just as fine? I've got relatively easy (and free) access to low viscosity synthetic motor oil. Would it have the same degrading effect on rubber items? I figure it should be lower, but I'm not an expert in these areas. I know there's no conclusive answer as to what the long term effects of sumbersion causes, but someone might have a clue?

And secondly, what does "Weather Stripping" actually mean? English isn't my first language, so I kind of got stuck on that one ^^

Anyway, it's a great project and I really hope it works out over a longer period of time.
Keep up the good work!

[andics]

Just wondering, would putting in a passive cooler on the video card help? With the bigger surface a passive cooler have, it might have a better transfer of the heat from the vieo card to the oil.

[unholycrap]

I am very curious to see a report on the long term effects (several months) the oil has on a system like this running consistently.

One concern I have is about the lifetime of the submerged fans. Obviously, they are forced to move slower in the oil. Will this cause the motors to struggle and produce heat? Will their lifetime be shortened?