why not oilcooling?
Moderators: NeilBlanchard, Ralf Hutter, sthayashi, Lawrence Lee
-
mattthemuppet
- *Lifetime Patron*
- Posts: 618
- Joined: Mon May 23, 2005 7:05 am
- Location: State College, PA
why not oilcooling?
was chatting with an engineer type friend about building my own passive w/c rad and he asked me why not oilcooling? He said there were several oils that maintained the same viscosity at various temperatures and that oil can carry more heat per unit volume than water. Figured someone on here would have looked into it at some point, though I've no idea how well w/c pumps would work with oil? Just curious
That certainly is a feasible idea...
My house has a solar heating system which uses oil in the solar panels, which runs through a heat exchanger, transferring the energy into water, which is then pumped through the floors. The oil that is used can withstand some rather extreme ranges under pressure...I'm guessing -50F to 250F, and is rather expensive (~$30/gallon). Granted, an oil with such temperature extremes is unnecessary; but I do believe that the only reason why it is used is due to the temperature ranges it can withstand, as it (the oil used in this situation) probably has a much lower specific heat than water.
I also don't believe that in general the problem (as yet) lies in the lack of waters ability to carry the heat (this may change as faster processors are made; I have seen workings of liquid metal cooling systems for computers--the metal had a higher heat conductance than water, but it had a lower heat capacity), it's rather the ability to dissipate the heat elsewhere that's the problem (ie. very large radiator).
I think oil could certainly work, however I don't know exactly what one would really gain by using it.
My house has a solar heating system which uses oil in the solar panels, which runs through a heat exchanger, transferring the energy into water, which is then pumped through the floors. The oil that is used can withstand some rather extreme ranges under pressure...I'm guessing -50F to 250F, and is rather expensive (~$30/gallon). Granted, an oil with such temperature extremes is unnecessary; but I do believe that the only reason why it is used is due to the temperature ranges it can withstand, as it (the oil used in this situation) probably has a much lower specific heat than water.
I also don't believe that in general the problem (as yet) lies in the lack of waters ability to carry the heat (this may change as faster processors are made; I have seen workings of liquid metal cooling systems for computers--the metal had a higher heat conductance than water, but it had a lower heat capacity), it's rather the ability to dissipate the heat elsewhere that's the problem (ie. very large radiator).
I think oil could certainly work, however I don't know exactly what one would really gain by using it.
Compared to water, oil is more expensive, less efficient, and harder to pump. DI water is so cheap its nearly free, even once you count the additives needed to prevent biological growth. "Typical" oils, such as mineral oil, have about half the specific heat of water, making them less efficient conductors of heat energy. There are specialized oils that are more efficient, but the cost is through the roof. And nearly every oil is more viscous than water, which translates to either lower flow rates, or higher pumping energy requirements.
But the nice thing about oil is that youo can do away with the fancy waterblocks and tubing, and just submerge the whole machine in the oil.
But the nice thing about oil is that youo can do away with the fancy waterblocks and tubing, and just submerge the whole machine in the oil.
-
ToasterIQ2000
- Posts: 36
- Joined: Fri Feb 25, 2005 10:36 pm
I've made four attempts to oil cool power supplies.
First I submerged a PSU in a tank with coils of thin walled norprene(tm) tubing wrapping the PSU -- water running through the tubing. For that I used gear oil from the local automotive store. Boy was that one a sloppy first try. The oil softened the hose and would have leaked severly eventually. I found that out because I had to take it apart to replace the soldered in PSU fuse that I blew. Anything submerged in oil is a grand mess to work on. That would be my broadest conclusion here, the general rule, the main lesson I could share with confidance.
My second approach has been to sink the power supply in a copper sided box and use water cooling on the box sides. That is without actively circulating the oil and hoping that passive convection will keep the heat down in the center away from the copper sides. Everything works for a while, but the long term problem has been capacitors swelling and pushing themselves out of the circuit board.
I started working on this after reading several accounts of submerged computers, thinking, 'putting a transformer in transformer oil is nothing new.' But I've always wound up with swollen capacitors pushing themselves out of the solder. I've seen capacitors that have overheated and bloated on the top in aircooled systems. Were I to meet a knowledgable engineer with experience in this I'd ask: is there is some special grade of capacitor spec'd for oil submersion, some layering / winding material that will not soak up oil and expand? ; or is oil not the real problem -- is it really the heat that makes capacitors swell and push out of the board, wheather in air or oil, and caps would not neccessarilly bloat in cooled circulated oil?
My one attempt that is still in use ( one or two days a week when a particular part time emlpoyee is in the office ) features mineral oil, only one full copper side on the PSU tank, a 10+ year old 'no-name' PSU core, and large external radiator for the PSU, CPU, graphics and mainboard. This is a K7 1400 cpu on an asus av7.
Mostly I'm glad that fanless PSU's like the Antec Phantom (3 months or so for me) are now available.
First I submerged a PSU in a tank with coils of thin walled norprene(tm) tubing wrapping the PSU -- water running through the tubing. For that I used gear oil from the local automotive store. Boy was that one a sloppy first try. The oil softened the hose and would have leaked severly eventually. I found that out because I had to take it apart to replace the soldered in PSU fuse that I blew. Anything submerged in oil is a grand mess to work on. That would be my broadest conclusion here, the general rule, the main lesson I could share with confidance.
My second approach has been to sink the power supply in a copper sided box and use water cooling on the box sides. That is without actively circulating the oil and hoping that passive convection will keep the heat down in the center away from the copper sides. Everything works for a while, but the long term problem has been capacitors swelling and pushing themselves out of the circuit board.
I started working on this after reading several accounts of submerged computers, thinking, 'putting a transformer in transformer oil is nothing new.' But I've always wound up with swollen capacitors pushing themselves out of the solder. I've seen capacitors that have overheated and bloated on the top in aircooled systems. Were I to meet a knowledgable engineer with experience in this I'd ask: is there is some special grade of capacitor spec'd for oil submersion, some layering / winding material that will not soak up oil and expand? ; or is oil not the real problem -- is it really the heat that makes capacitors swell and push out of the board, wheather in air or oil, and caps would not neccessarilly bloat in cooled circulated oil?
My one attempt that is still in use ( one or two days a week when a particular part time emlpoyee is in the office ) features mineral oil, only one full copper side on the PSU tank, a 10+ year old 'no-name' PSU core, and large external radiator for the PSU, CPU, graphics and mainboard. This is a K7 1400 cpu on an asus av7.
Mostly I'm glad that fanless PSU's like the Antec Phantom (3 months or so for me) are now available.
-
mattthemuppet
- *Lifetime Patron*
- Posts: 618
- Joined: Mon May 23, 2005 7:05 am
- Location: State College, PA
I'm surprised oil has a lower thermal capacity than water - I would have thought it was higher. Perhaps it's the effect on flow rates? Thinking about motorbikes - in order of performance it goes, air, air+oil then watercooling. I think the oils my friend was talking about were very high performance oils and most likely very expensive. Looks like I'll stick with aircooling for the time being
Oils are good and there are many places to use them but your never going to get one that is significantly going to beat the performance of water. As a coolant water is V good and cheap.
Oils have a lower thermal capacity per unit mass, per unit volume they are alot better than water i would imagine.
Overall most oil systems would be worse because liquid cooling performance (some stuff behaves differently such as the storm block) is proportional to the prandlt number, whos definition escapes me, i think its thermal capacity per unit mass divided by viscosity. Basically how much heat can be moved away from a surface by convection. Water's is much higher than most fluids.
Oils have a lower thermal capacity per unit mass, per unit volume they are alot better than water i would imagine.
Overall most oil systems would be worse because liquid cooling performance (some stuff behaves differently such as the storm block) is proportional to the prandlt number, whos definition escapes me, i think its thermal capacity per unit mass divided by viscosity. Basically how much heat can be moved away from a surface by convection. Water's is much higher than most fluids.