A new approach to oil submersion system.

toarranre

Honorable
Feb 7, 2013
125
0
10,710
Edit: Remember kids, never submerge a hard drive!

I am thinking about building an oil submersed system, mainly for the novelty but also to see if I can make it a bit simpler and more efficient than dumping it in a fish tank with a pump and radiator.
Please note I realize the drawbacks, and I won't be using an expensive system, this project will be about aesthetics and novelty first, practicality and cooling efficiency second.

The main issue for submersed systems is that they remove heat from the components into the oil very efficiently, but then the heat accumulates in the oil unless a radiator system is employed. I read someone comment that for this to work without a radiator and pump your entire tank would need to be a heat sink. So lets submerge a system in a heat sink!

(Numerical references here are very much approximate)
Perspex, while pretty, would be a terrible heat transfer medium, glass fares better at about 5x the thermal conductivity, but steel (200x) or aluminium (800x) would be much better. Copper or silver would, of course, be best but are out of the price range.

I couldn't find any suitable aluminium containers for what I have in mind, but we have tonnes of 18L steel oil drums at work.


So that would do for a prototype. I would plan to cut out a window then use silicon glue to seal a piece of perspex in place, followed by loosely bolting it on for aesthetic and extra sealing. I recon I would also need a frame, slightly thinner than the silicon, between the perspex and container to stop it warping when the bolts are tightened.
As you can see from the less than convincing photoshop job, this is what I would plan for it too look like.



I recon this would do the job for a low power build, but to make sure I need some thoughts on cooling.
I would use fish tank aerators and possibly a small propeller to provide oil circulation but as the oil conducts heat about 6x better than air I am more concerned about the steel to air phase. I want to try and keep everything internal so my plan would be to put ventilation slits in the side of the can above the oil level and on the top then attach a large, slim fan to draw air in from the sides and expel it through the top. Even lower the oil level and solder on a flat radial heat sink, though I can't quite figure out how to do this without actually obstructing airflow considerably.
Thanks for your thoughts on my next crazy project!
 
You could try running some copper heatpipes in the oil up to a heatsink on top of the case, which will dissipate heat passively. If its a low power build, then I imagine that would do for cooling the oil.
But if you want to keep it all internal...
Think the circulation of the oil will be fairly important, and for that I fear you will have to drop your pump requirement. Would plonk a pump in the tank, put some tubing on it with a Y-Fitting to draw oil from various hot spots (CPU heatsink, graphics card if applicable, PSU) and then pump that to the top of the enclosure, where the heat can passively be cooled by the air from the fan.
 

toarranre

Honorable
Feb 7, 2013
125
0
10,710
Thanks for the input, I like the idea of copper pipes out of the oil straight to fan cooling.
I don't mind dropping a pump in, I just don't want to throw a heap of external cooling at it. I don't think just pushing oil to the top would help that much as oil and air are relatively poor heat transfer mediums, though using the same idea but with a semi submerged radial heatsink, (hot oil-aluminum-air) is definitely a winner.



My thoughts are for roughly 6 hours a day it has a huge dissipation surface area, but I would like to push cooling efficiency to the limit for the sake of knowing how far it can go. I will be testing it with heating elements from 100-1000 watts for extended periods of time to see how capable I can make it.
 
Because the case is metal I'd suggest adding heatsinks to the outside of the case . Alumina extrusions bolted through the back of the case . Possibly with some on the inside picking up heat , but they would need a much lower surface area because of the oils density than those on the outside