I just started reading this, and I have a few years of metalworking experience. Mostly with steel but some aluminum, and I've made some watertight boxes, so I thought I'd throw my 2c in.

Making an accordion type of fin/vane thing will be very difficult to seal, even if it's steel. All gaps in the base, where the fins meet the bottom will leak like a sieve. If your fins are 1"deep, and 1" wide you'll be able to get a weld in there, but if you go below 3/8" or even 1/2" wide and keep a 1" depth it'll be functionally impossible to weld.

Copper would be a better solution if you can bend it right, as you can simply bake and solder the whole thing together at once.

If I was doing this I would buy 2 - 6 large (whatever the side dimensions are by 2", with a 1/4" base) aluminum or copper heatsinks, and glue them together with thermal epoxy. Have one side as the wet side, the other as the dry side.

I was considering making a new thread but guess i can just as well reawaken this one.

I'm considering putting a pc in a old steel amunition box submerged in oil as a useable experiment. I still have to check the actual size of the case but think it is close to 30x20x10 cm.

I have a matx motherboard with llano 100w tdp apu (i'm unable to lower the voltage with my motherboard), so im not sure if i would use that apu with a cheap mitx motherboard or find something else to submerge.

For the oil im thinking about the usual mineral oil or maybe the less considered coconut oil. coconut oil melts between 24 and 26 degrees celcius, i remember finding someone who sucecfully used it for submerging a psu for at least a year but can't find it again.

The advantage of being solid at room temperature would be that its easy to transport without having to worry about spilling the oils, and furthermore if its transported against gravaty along cablesdue to capalery effects it would solidify somewhere up the cable in stead of making a mess beside the pc.

I'm wondering however if with a cpu the first second or whatever that the oil is still solid would cause overheating isues or would it maybe actualy cool more because of extra energy absorbed in the melting process. And if i submerge fans along with the motherboard would the initialy stalled fans cause any electrical problems.

as for the heat loss with the metal box, i saw some calculations before in the thread but didnt really understand them, i came at a heat loss of 160W not considering radiative heat loss.

A =0,16m2 (4 sides of the box, top might be closed of with air in between and it stands on its bottem)
U = 25 w/m2k (steel)
T= 40 (60 degrees celcius for the oils and 20 degrees ambient)
H= 160 W

So basicaly im wondering if my idea is realistic, would a 100w tdp apu, motherboard and psu be cooled enough in a steel case as i described or should i go for a more energy friendly pc, for what kind of oil you think i should go.

The whole thing is just an idea so don't expect me to try some of the stuff within the month:P

You should disregard the top of the case for dissipation purposes unless you have totally filled it to the brim without air bubbles. Assuming that the remaining surfaces can still dissipate enough heat you should have no trouble. I would recommend getting a wattmeter, you'll likely find your system uses far less power than you think on average. Oil can also absorb longer, higher watt spikes than air without significant rise in temperature. (total heat being the same)

As far as coconut oil, I would be concerned that solid oil would not transfer heat quickly enough to melt most of the oil. Think of a cookie pan of bacon grease with heat on one corner. You'd only have localized melting, with possible overheating.

Your plan is epically bad for the following reasons:

1) The oil must be liquid at room temp. Convection and not conduction is the primary means of heat transfer, and the oil won't melt until long after you've deep-fried your components.
2) Coconut oil will go bad very quickly and need constant replacement.
3) Even if you could get the coconut oil to work, the liquid oil would still wick up your cables and leak everywhere.

I did a lot of investigation into submerged computing, and at this point, it just sounds like a bad idea. Only very low-power systems can operate without pumps and/or radiators keeping things cool, and at that point, you might as well just go with watercooling. Unless you're looking for novelty, I'd say forget about it.

No need to invent the wheel twice:

http://www.pugetsystems.com/submerged.php

Ok thanks for replies so far, guess i won't be using coconut oil but something always liquid like mineral oil tho i would have to close the box of for health reasons as there is the posibility that the vapours might be damaging for your health.

As for the wicking along cables, it's probably posible to prevent it by having no cables from the oil to the outside.
The hdmi connectors etc. of the motherboard would not have to be submerged and the sata cables would only have to lead to the top/lit of the box so wicked oil should fall back in the box.

And i'm not planning to reinvent the wheel, just planning to make a smaller wheel of a diferent material based on the same idea. I'm not planning to show it of in a transparant aquarium (only as a box without fans) so i have a bit more posibilities, but most examples i found actualy used a transparent case

Wicking shouldn't be a problem if you have a few solid cables. Tight groups like the ATX connector or braided cables would wick, but a SATA or HDMI cable wouldn't.

you could probably make a drip loop in any cables suffering from wicking, if it becomes a problem.

I did some measurements, the box is 9x28x16 cm so it won't even fit a mitx board. Even if i don't submerge the entire motherboard i still would not be able to close the lid of the box. So guess i will have to find some other metal container.

I also measured how much energy my current llano pc uses, with normal use 33-45W with short peaks up to 70W, with max loads i use it for (archicad along with a couple other programs) its 36-93. with startup(including ssd and 2 extra hdd, after startup i unmounted the hdd's) it had a short peak of 125 W.

so at max load (for safety i asume an average of 90W) at a minimal sized case 17cmx17cmx9cm considering only direct heatloss trough 4 sides and not radiation the formula for oil temp would be:

H=90w
A=0,0884m^2
U=25

90/0,0884/25=40,72

T=40,72 which would mean an oil temp of about 61 degrees celcius.

But i would guess 61 degrees for the oil (even if only in the most extreme situation) would not leave enough temperature diference between apu and oil to release enough heat to the oil.

So if i want to try this i guess i would need a bigger then minimum sized case, lower power pc for this or get a mitx board that could undervolt the apu

Or does the heat radiation of a case acount for enough heatloss that it probably would actualy be posible with current energy consumption?

Radiation does contribute significantly, particularly if you're using a material with a high emissivity, but 60C is actually pretty low for aquarium PC's under constant load. I think your figure is overly optimistic and that you'll need a much bigger, much heavier case to do the job. Also, you are completely SOL if you miscalculate and end up with insufficient cooling, or if you ever want to upgrade to more power-hungry components.

If I were you, I'd use something like a watercooling loop, with the radiator exhausting into the bottom of the case and the APU waterblock (or is that "oilblock"?) intaking from the top. The pump will be submerged, so you won't hear a thing if it's properly decoupled, and active circulation is an absolute necessity in submerged PC's, anyway (oil is far more viscous than air and therefore more prone to boundary layer formation). The end result will also be lighter, more compact, more reliable, and far more flexible.