Wanted: Discussion of Watercooling versus Oil Submersion

The alternative to direct air cooling

Moderators: NeilBlanchard, Ralf Hutter, sthayashi, Lawrence Lee

Post Reply
drownmypc
Posts: 56
Joined: Mon Feb 11, 2008 2:14 pm
Location: chapel hill, nc

Wanted: Discussion of Watercooling versus Oil Submersion

Post by drownmypc » Tue May 06, 2008 6:47 pm

I'm interested in discussing the relative merits of conventional watercooling (water blocks, tubing, pumps) versus oil-submersion (aquarium, mineral oil).

No religion, please! I am not personally advocating either approach (although I have more experience with oil submersion than water cooling).

I would love to hear veterans of each approach weigh in with their thoughts/opinions on these two liquid cooling techniques.

To get things started, here's a laundry list of issues/tradeoffs:

WATERCOOLING (water blocks, tubing, radiators):
  • advantages of water cooling over oil:
    + products exist for water cooling (pumps, tubing, waterblocks)
    + mature method: deep knowledge base for technique, materials, etc.
    + cooling applied only to heat-producing components (GPU, CPU, etc)
  • Pitfalls unique to watercooling (some surprising):
    + galvanic corrosion can harm pump/waterblocks, leading to flow stoppage and burnout. Avoid dissimilar metals.
    + leaky tube connections can drip on components, leading to a catastrophic short. nonconductive fluid helps.
    + algae growth can suddenly stop flow, leading to burnout. Additives, air-tight loops, semiannual cleanings help.
    + kinked tubing can stop flow. quality tubing and tube management helps.
    + a slipped water block can cease cooling, leading to burnout.
  • Requires planning:
    + careful design of cooling loop path.
    + have to choose fluids & additives.
    + system requires maintenance (periodic flushing).
    + water blocks are specialty items (sometimes expensive, can't always be carried over to next system).
    + each water block adds complexity and crowding: cpu, gpu, NB, mobo voltage regulators, DIMMs, gpu RAMs.
    + requires a pump and radiator.
  • Still not absolutely silent:
    + you have replaced mobo and GPU fans with a water pump. But even water pumps make noise (so, get a good one, maybe suspend it or use a pad).
    + PSU has not been silenced, unless you water cool that too (some people do).
    + HDDs have not been silenced with watercooling. (they don't have fans, but they still make some noise).

OIL SUBMERSION (mineral oil in aquarium):
  • minimal risk of catastrophic cooling failure:
    + only way to lose thermal contact is a leaky tank letting out the oil - in which case your biggest worry is NOT a fried computer.
    + everything gets cooling benefit of oil, including cpu, gpu, NB, mobo voltage regulators, DIMMs, gpu RAMs, etc.
    + no need for pump or radiator: the aquarium walls become a large radiative surface. But a rad could still help.
    + submerged parts are absolutely silent - can submerge everything (even PSU) except HDDs .
    + doesn't require maintenance: mineral oil doesn't evaporate, spoil, or support algae growth.
  • pitfalls unique to soil submersion:
    + system harder to move: heavier, cannot be moved quickly, cannot be tipped.
    + spill could be an unholy mess to clean up. Mineral oil doesn't evaporate - so you can't let stuff just "dry out".
    + harder to tinker with system - nasty to attach/detach stuff while submerged; even when removed, components will be forever oily.
    + entire system subjected to hot oil - thus even mobo components that normally do not heat up will feel the 120F op temp.
    + oil can reportedly degrade some plastic components (uncommon).
  • still not absolutely silent (similar to watercooled):
    + Might still have a water pump, if you choose to use a radiator (optional - and reduce noice with same methods as with water cooling)
    + PSU has not been silenced, unless you submerged that too (some people do).
    + HDDs cannot be submerged. Period. Have to use other methods for cooling and silencing.
Opinions? I welcome corrections/additions to these lists.

m^2
Posts: 146
Joined: Mon Jan 29, 2007 2:12 am
Location: Poland
Contact:

Post by m^2 » Wed May 07, 2008 3:52 am

Additionally:
Water:
-Costly
Oil:
-No way to resell oiled parts.
+No dust problem.

Modo
Posts: 486
Joined: Wed Apr 16, 2008 3:32 am
Location: Poland

Post by Modo » Wed May 07, 2008 4:10 am

Oil cooling: Have to wait for bigger SSDs. ;)

Plekto
Posts: 398
Joined: Tue Feb 19, 2008 2:08 pm
Location: Los Angeles

Post by Plekto » Wed May 07, 2008 11:36 am

Note - you could also pretty easily adapt most water cooling setups to run mineral oil.

jaganath
Posts: 5085
Joined: Tue Sep 20, 2005 6:55 am
Location: UK

Post by jaganath » Wed May 07, 2008 1:10 pm

Plekto wrote:Note - you could also pretty easily adapt most water cooling setups to run mineral oil.
you'd probably need a much higher flowrate to make up for poor heat capacity of min. oil (60% less than water).

spookmineer
Patron of SPCR
Posts: 749
Joined: Sat Nov 11, 2006 6:02 pm

Post by spookmineer » Wed May 07, 2008 1:42 pm

And a more powerful pump. Oil is sticky and doesn't transport as well as water. In reality, using oil as a coolant fluid is not an option.

~El~Jefe~
Friend of SPCR
Posts: 2887
Joined: Mon Feb 28, 2005 4:21 pm
Location: New York City zzzz
Contact:

Post by ~El~Jefe~ » Thu May 08, 2008 8:00 am

Yeah, I dont really see this going anywhere except for l33t kiddie sites who have nothing to do but waste money.

it will eventually destroy a few parts of a computer board that could be disolved, gummed up, etc.

you could passively cool a computer better than using any liquid that is passively sitting on components.

shrugs.

what would a pump do? kinda useless, the point of oil was to submerge in a large volume of it so you dont need movement. if you create flow via a pump, 96% steam distilled water and 4% water wetter that I use has massively more heat capacity than any other option and especially oil. It's also cheap and safe.

Modo
Posts: 486
Joined: Wed Apr 16, 2008 3:32 am
Location: Poland

Post by Modo » Thu May 08, 2008 8:53 am

You can get good temperatures with a radiator setup. But this defeats the purpose of true silence, because a pump will still make some noise.

I think the solution is simple: You need an enclosure that has metal walls, preferably a one-piece thing so it would not leak, preferably with ribs on the inside and on the outside for better heat transfer. This would dissipate enough heat to be safe without any outside pumps.

Once SSDs get to bigger capacity, the only things that would really need to stay on the outside are connectors and the optical drive. Voila, there's the truly silent system. Fun to make, for sure. Except that a few companies already build heatsink-enclosures that do the same without an oily mess...

drownmypc
Posts: 56
Joined: Mon Feb 11, 2008 2:14 pm
Location: chapel hill, nc

Post by drownmypc » Thu May 15, 2008 10:28 pm

jaganath wrote:you'd probably need a much higher flowrate to make up for poor heat capacity of min. oil (60% less than water).
I agree. The upside of replacing the water in a conventional pump/tubing/waterblock/radiator setup is that spills won't fry your computer. But the downside is that you lose a lot of cooling capability because oil has poor heat capacity. Not a good trade, in my opinion. If you go to the trouble of having pump/tubing/waterblock/radiator, you might as well use a good cooling fluid (water), and trust that you have no leaks.
jaganath wrote:I think the solution is simple: You need an enclosure that has metal walls, preferably a one-piece thing so it would not leak, preferably with ribs on the inside and on the outside for better heat transfer. This would dissipate enough heat to be safe without any outside pumps.
Bingo. You read my mind. I have a friend who likes to weld. He's going to teach me how (I've always wanted to learn), and the teaching project is going to be an aluminum oil-proof case with fins on inside and out. Technically, I won't be welding -- I'll be "brazing", which is essentially soldering. Aluminum is hard to braze, but we've learned of good aluminum brazing products.

If I could find a good source, I'd do it in copper (twice as conductive as aluminum). Could be pricey though.

proc
Posts: 47
Joined: Thu Sep 07, 2006 4:39 am
Location: Italy

Post by proc » Thu May 15, 2008 11:52 pm

+ HDDs have not been silenced with watercooling. (they don't have fans, but they still make some noise).
Indeed there are some HDDs waterblocks around and you can just slap them to your HDD and entomb it in foam to dump its noise.

Unfortunately when I had a watercooled pc (with a passively cooled "radiator" and entombed water cooled HD) the noise of the pump was too loud for my liking.

spookmineer
Patron of SPCR
Posts: 749
Joined: Sat Nov 11, 2006 6:02 pm

Post by spookmineer » Fri May 16, 2008 5:12 pm

drownmypc wrote:
jaganath wrote:I think the solution is simple: You need an enclosure that has metal walls, preferably a one-piece thing so it would not leak, preferably with ribs on the inside and on the outside for better heat transfer. This would dissipate enough heat to be safe without any outside pumps.
Bingo. You read my mind. I have a friend who likes to weld. He's going to teach me how (I've always wanted to learn), and the teaching project is going to be an aluminum oil-proof case with fins on inside and out. Technically, I won't be welding -- I'll be "brazing", which is essentially soldering. Aluminum is hard to braze, but we've learned of good aluminum brazing products.

If I could find a good source, I'd do it in copper (twice as conductive as aluminum). Could be pricey though.
Using copper will be a waste of money, aluminium will be just as good.
Eventually (might take a while) the oil will reach a certain temperature when the heat output of the PC equals the heat emitted to ambient (by means of the wall containing the oil).

To keep it a bit simpler than it really is, assume all of the oil has the same temperature, and the wall of whatever material will act as a temperature bridge between the oil temperature and ambient temperature (the temperature inside the wall will drop slightly from the oil temperature where it touches the oil, to ambient temperature where it touches ambient air).
[Note, the temperature of the wall where it touches ambient is not the same temperature as ambient itself - the wall acts as as a radiator at this point, this is where the heat will be extracted from the system]:

Image
In this graph, T1 - Tw1 - Tw2 will be virtually the same, and the big drop happens at T2.

Thermal conductivity of
copper: 401 W/mK
aluminium: 237 W/mK
Means a 4 mm thick copper wall is as thermally conductive as a 2.4 mm thick aluminium wall.
The impact on the oil temperature is negligable, though.

If the PC has a power output of 400 W (which is pretty high end), and the walls are a total of 0.6 m2, a 4 mm copper wall means a delta T of 0.06 °C. Aluminium will mean a delta T of 0.11 °C.
The difference is practically immeasurable.

To put things into perspective, "plastic" (pyrex glass) has a thermal conductivity of 1.0 W/mK. It is a very bad thermal conductor, the result is a delta T of 26.7 °C :shock:
Using metal walls is a good idea but the differences in temperature will be very small when compared to plastic walls.

I even wonder if adding fins to the inside will be a good choice, you will be increasing the area but the thickness of the wall will increase as well, maybe doing more harm than good. In all cases, a wall which means a barrier of less than 1 °C is nothing to be concerned about.
On the outside, where it touches ambient air, you will want to be the area as big as possible, because air is a bad heat conductor.

jonlumb
Friend of SPCR
Posts: 27
Joined: Sat May 10, 2008 12:59 am
Location: Exeter, UK
Contact:

Post by jonlumb » Sat May 17, 2008 1:39 am

Firstly, for those of you in the UK interested in a watercooled setup, the current edition of Custom PC has about 30 pages on watercooling in a mammoth guide.

It covers everything from basics of setup, best practice etc (all the negative points mentioned by the OP are addressed and demonstrated how to prevent). It also reviews a fair few watercooling setups etc.

With regards to Oil, the overriding impression that I get is that unless you are using particularly low power components, it doesn't work that well as a cooler, as the viscosity of the oil inhibits convection.

VanWaGuy
Posts: 299
Joined: Mon Sep 11, 2006 1:01 am
Location: Vancouver Wa USA

Post by VanWaGuy » Sat May 17, 2008 6:41 am

Man, this topic makes me realize that I am getting old.

Really, this topic should have included air cooling too. Why exclude the usually obvious answer two only consider two extreme cases?

When I was younger, I would have loved to experiment with the oil immersion. It would have been fun to play with, but "play" being a key word, come on, having an open aquarium of oil with a Mobo immersed can't really be considered a serious system. I still find the idea interesting, but no way I could live with the inconvenience of that long term. No one even mentioned dropping crap in it, the cat jumping in it, a big tank of oil sitting somewhere in the house just is not practical.

And water cooling. I was the same about it. Years ago, I learned water and electronics do not mix. Period. You can try to be cute, and cheat fate, but eventually, it is going to be a problem. Yeah, it too has high geek appeal, but why bother? I have toyed with ideas for really simple home made water blocks (because sometimes at night, some geeks can't turn off the stupid ideas) but still, why bother?

Look at that list of disadvantages. You can avoid them all by just using air.

Air is not cool, not much geek appeal, but it meets the requirements, is simple, and just works.

Do a bit of research on this site (or many others) and you learn right away that if you get a case with decent air flow, put in decent parts, and you can get a great performing, cool running system with very little maintenance required, very low catastrophic failure chances, and no Oily cat! (A lot of store-bought systems have crappy cases, poor air flow, and still manage to be kept cool. You really don't need exotic technology to do way better than those systems!)

On water or oil, why bother? Maybe you could push your system clock rate a little higher. Who cares? I used to Overclock, but any more, technology advances so fast, if you want a faster system, just wait 6 months.

I have 7 computers in the house. If I do not want to be constantly tinkering with them, I need something simple that works and scales well. I do not have time to clean Algae from 7 systems or make sure that the cats are staying out of 7 oily aquariums.

Plekto
Posts: 398
Joined: Tue Feb 19, 2008 2:08 pm
Location: Los Angeles

Post by Plekto » Mon May 19, 2008 3:47 pm

A couple of thoughts on oil in a water system:

It'll still cool a bit more effectively than air. But it's dead silent as well, plus has no real problems with leaking or problems with corrosion or whatnot(eventually the fluid will become thermally conductive).

The idea here would be to have a setup with no fans and no worries about what happens if it leaks(other than the mess, of course). Plus, it would work well hooked up to a radiator for spot heating ;)

cercle
Posts: 13
Joined: Fri Jun 27, 2008 8:34 am
Location: Paris,France

OIL&WATER coolings!

Post by cercle » Sat Jul 19, 2008 10:52 am

I don't like water in my computer ,oilcooling is not efficient. But :!: :oil and water don't mix together !So diping a water cooled computer into oil would separate/protect it from water, and moreover it would slow down sound wave of the watercoolingpump?silencing+shortcircuitproof.The rad plus it's (silent) fan are standing out of oil ,so you have the best of the two worlds:watercooling+under oil silent pump+oil shortcircuit safety :roll:

Gator
Posts: 16
Joined: Thu Jul 26, 2007 2:16 pm
Location: Gainesville, FL - USA

Re: Wanted: Discussion of Watercooling versus Oil Submersion

Post by Gator » Sat Jul 19, 2008 5:45 pm

drownmypc wrote: ...
WATERCOOLING (water blocks, tubing, radiators):
  • Still not absolutely silent:
    + you have replaced mobo and GPU fans with a water pump. But even water pumps make noise (so, get a good one, maybe suspend it or use a pad).
    + PSU has not been silenced, unless you water cool that too (some people do).
    + HDDs have not been silenced with watercooling. (they don't have fans, but they still make some noise).
...
Opinions? I welcome corrections/additions to these lists.
However, unlike an HSF, the watercooling pump, radiator, and any radiator cooling fans do not necessarily need to be in the same location as the rest of the computer. :D It is possible to run cooling lines and place the pump, radiator and radiator cooling fans at some distance, possibly resulting in true silence at the computer location.

drownmypc
Posts: 56
Joined: Mon Feb 11, 2008 2:14 pm
Location: chapel hill, nc

Post by drownmypc » Tue Jul 22, 2008 3:13 pm

spookmineer wrote:If the PC has a power output of 400 W (which is pretty high end), and the walls are a total of 0.6 m2, a 4 mm copper wall means a delta T of 0.06 °C. Aluminium will mean a delta T of 0.11 °C.
The difference is practically immeasurable.
WOW. Nice work, spookmineer. I was aware of the thermal conductivities of copper vs aluminum, but I had no idea that they would perform nearly identically in this scenario. I guess it is because the thermal resistance of copper and aluminum are both quite low compared to the thermal resistance of air which is the final destination of the heat energy.
I guess the analogy is that of choosing between two resistors, a 1ohm and a 2ohm, to hook up in series with a 2000 ohm resistor. It doesn't matter whether you choose the 1 ohm or 2 ohm, the 2000 ohm is going to dominate.
I'd be interested in knowing the formulas you used in these calculations. Any pointers to websites ?
VanWaGuy wrote:come on, having an open aquarium of oil with a Mobo immersed can't really be considered a serious system.
VanWaGuy, I totally agree. I think well-engineered air cooling is the least expensive, least hazardous, and most reliable solution. Oil cooling is just for fooling around. I enjoy thinking about the science and engineering aspects of it, and wondering how far you can take it. The gee-whiz factor is a big part of it. And that's why I didn't offer up air cooling for discussion -- I just don't find it that interesting to discuss.
Gator wrote:the watercooling pump, radiator, and any radiator cooling fans do not necessarily need to be in the same location as the rest of the computer.
Excellent point!

spookmineer
Patron of SPCR
Posts: 749
Joined: Sat Nov 11, 2006 6:02 pm

Post by spookmineer » Tue Jul 22, 2008 4:27 pm

drownmypc wrote:I'd be interested in knowing the formulas you used in these calculations. Any pointers to websites ?
I found out there is a mistake in the calculation :/


For a refresher I just googled thermal conductivity and the wiki has most answers.

H = dQ / dt = k * A * dT / x

H = amount of heat flow (in example, 400 W)

= dQ/dt = amount of heat flow: 400 joules / 1 second

k = thermal conductivity (depending on material, aluminium = 237 W/mK)
A = surface area (0.6 m2)
dT = temperature difference
x = thickness of the wall

For copper:
H = 400 (W) = 401 (W/mK) * 0.6 (m2) * dT / 0.004 (m)
dT then equals 0.006 °C (not 0.06 °C as I wrote above).
For aluminium then 0.06 *237 / 401 = 0.011 °C.

So... for perspex the temperature difference is not 26.7 °C but only 2.67 °C. I rushed it before and entered 0.04 for wall thickness...

Here an easy calculator to see some results.

Rusty075
SPCR Reviewer
Posts: 4000
Joined: Sun Aug 11, 2002 3:26 pm
Location: Phoenix, AZ
Contact:

Post by Rusty075 » Fri Jul 25, 2008 8:00 pm

As someone who actually has built and operated a submerged oil PC, my semi-random thoughts:

-You'll need to include "oil vapor" as one of the downsides, if you're thinking about an un-sealed system. It smells terrible, and potentially has health impacts.
-The viscosity of even the thinnest mineral oils will destroy a conventional aquarium-type water pump fairly quickly. You'll want to invest in one that is specifically designed for oils if you're thinking of swapping water for oil in a convention circuit system.
-But, you can just use conventional case fans to circulate the oil within the enclosure of a submerged system. I had good success with plain old 80mm fans. They'll spin slowly, but they'll spin.
-After nearly a year in the oil none of the parts I submerged were any worse for the experience. In fact the NIC card used in that machine is still in use today.
-I was a bit worried about the high voltage coils in the PSU, but it didn't seem to mind being submerged at all.
-Another con you don't think of until it's in use: Capillary action will draw the oil up any wire or cable that protrudes from the tank. It took me months to figure out how I kept getting a little drop of oil on my mousepad.

cercle
Posts: 13
Joined: Fri Jun 27, 2008 8:34 am
Location: Paris,France

Water C. in Oil

Post by cercle » Sun Jul 27, 2008 6:14 am

To Rusty075 .HI ,Cercle here,I don't like water in a pc,but I need to cool silently a hot/powerful graphic card.That's why I think about dipping a watercooled pc into oil.OIL and WATER don't mix so shortcircuit could be avoided :a strong water jet on pc could go through oil and destroy it by shortcircuit.That should be exceptional,most of leaks on watercooling could be repaired.My thanks for returning your experience,I am looking for a test of this combination,maybe a website ,"spcr"?,could use some pcs'part of its stock in order to test this.I should,I guess ,ask Mr Chin to test this? :shock: :roll: :wink:

drownmypc
Posts: 56
Joined: Mon Feb 11, 2008 2:14 pm
Location: chapel hill, nc

Post by drownmypc » Mon Jul 28, 2008 6:55 am

Rusty075 wrote:You'll need to include "oil vapor" as one of the downsides, if you're thinking about an un-sealed system. It smells terrible, and potentially has health impacts.
I've had an unsealed system running for almost 6 months. No smell, no perceptible drop in oil level, air quality (my basement) seems fine. Plus, I found spec sheets on mineral oil that claim no appreciable evaporation rate (at room temp - but our oil is hotter - so may not apply), and no health risk other than aspiration (taking of liquid form into lungs).
Rusty075 wrote:But, you can just use conventional case fans to circulate the oil within the enclosure of a submerged system. I had good success with plain old 80mm fans. They'll spin slowly, but they'll spin.
I concur. My system simply use the same four air fans (volt reg on mobo, GPU, CPU, northbridge) to circulate the oil. They spin slowly - but they spin.
Rusty075 wrote:After nearly a year in the oil none of the parts I submerged were any worse for the experience. In fact the NIC card used in that machine is still in use today.
I concur. My system is doing fine at 6 months. [Edit: still fine at 9 months]
Rusty075 wrote:Another con you don't think of until it's in use: Capillary action will draw the oil up any wire or cable that protrudes from the tank. It took me months to figure out how I kept getting a little drop of oil on my mousepad.
Yup. My SPU and HDD are outside the oil. There is an extremely fine film of oil on their cables. All the other cables (sounds, keyboard, mouse, video, USB) all seem clean, which is curious.
Last edited by drownmypc on Fri Dec 19, 2008 1:45 pm, edited 1 time in total.

theycallmebruce
Posts: 292
Joined: Sat Jul 14, 2007 10:11 am
Location: Perth, Western Australia

Post by theycallmebruce » Wed Nov 19, 2008 9:21 pm

Just checking back in on this thread. Top marks for doing this "for real", drownmypc, and for the excellent detailed analysis of the results.

drownmypc
Posts: 56
Joined: Mon Feb 11, 2008 2:14 pm
Location: chapel hill, nc

Post by drownmypc » Fri Dec 19, 2008 1:47 pm

theycallmebruce wrote:Just checking back in on this thread. Top marks for doing this "for real", drownmypc, and for the excellent detailed analysis of the results.
Thanks, Bruce. It has been a lot of fun. I added an update to my other thread about problems (algae, mineral scale, evaporation) I had in my water tank. (I have a double hull arrangement, oil in the inner tank, water in the outer tank - the water part is giving me trouble). Link is here: viewtopic.php?t=45613&start=80 (PC in aquarium full of mineral oil).

rill2
Posts: 10
Joined: Thu Dec 25, 2008 12:16 pm
Location: Netherlands

Lava-lamp PC?

Post by rill2 » Sat Dec 27, 2008 11:51 am

I would definitely go for submerged if I could make my aquarium-pc into a lava-lamp pc (have blobs moving around the aquarium). Then the sheer awesomeness would blow any and all disadvantages out of the mineral oil.

drownmypc
Posts: 56
Joined: Mon Feb 11, 2008 2:14 pm
Location: chapel hill, nc

Re: Lava-lamp PC?

Post by drownmypc » Sun Dec 28, 2008 4:50 pm

rill2 wrote:I would definitely go for submerged if I could make my aquarium-pc into a lava-lamp pc (have blobs moving around the aquarium).
That's a pretty cool idea! The colored blobs in lava lamps are some kind of waxy/oily substance that won't mix with the water it floats in. The wax and the water parts are formulated to have very similar densities. There's a procedure described for it here: http://www.oozinggoo.com/ll-form5.html.

If you tried this, I think you would face two challenges:

(1) You will have to find some way to prevent the various chemicals from evaporating. If the water evaporates, it's density will increase, throwing off the balance. If the tetrachloroethylene in the wax evaporates, the density of the wax will drop, throwing off the balance. Real lava lamps are hermetically sealed (well, MINE is, with an ordinary bottlecap, in fact), probably to prevent evaporation.

(2) In addition to precisely balancing the densities, you may have to precisely balance the temperatures as well. If the whole tank stays too cool, the wax stays at the bottom, and if the whole tank stays too hot, the wax stays stuck at the top. Real lava lamps are tuned to heat at the bottom and cool at the top and generally stay within a proper range. Might be hard to get your oil PC to heat the tank to just the right temp and stay there.

All my comments are just speculation. I was a physics major in college, so I'm just making educated guesses here. I haven't actually spoken to a lava lamp expert.

Post Reply