Is oil cooling realistic?

[aspire]

I know.. aint it great!?!?!? Besides, MilchstrabeStern said he had a laptop that he uses primarily. Experimenting with something new can be fun. So what if it turns out it dont work. 8 or 10 years ago water cooling was on the bleeding edge. Many thought it was a bad idea and many more had EPIC failures.

Sure, maybe it aint the best idea for your primary pc, but if someone wantd to tinker around with building a second pc, I say "come big or go home".

But Oil submersion is nothing new...

It's been around for more than a decade, and since then, there has really been no evolution in it's use. It's just not practical.

8-10 years ago, water cooling wasn't even an industry, the few people who experimented with it used aquarium pumps and car heatercores with homemade water blocks, I don' know about you, but that is most certainly not my definition of "Bleeding Edge". Nowadays there are numerous companies that exclusively make water cooling products and do a **** fine job of it.

Some of the problems with an oil submersion cooled pc include:

Computer weighs too much to move, it coats all your parts in oil, you have to change the oil as it gets dirty, need to be careful not to spill, once the oil becomes hot, it's nearly impossible to cool it down again, and to cool it, your basically running a water cooling loop, so what the point?

And the majority of these problems are impossible to solve to the point of making oil cooling practical, so why bother?

Also I believe the phrase your looking for is "Go big or go home".

 

[MilchstrabeStern]

OK, how did this turn into a 5 page thread? I got the idea on page 1, BAD IDEA, understood.

Leave the drives and other items high and dry.

Thanks for that advice, what's your advice on taking a bath with my toaster?


Anyways. I thought about how to get the aesthetic aspect, without all the hassle. So I thought maybe taking a 10 gallon aquarium, and building a 1" deep jacket around the inside with some glass or plexi-glass. Then I could fill this jacket with water, giving the appearance that the computer is submerged in water.

I then thought about using this water jacket for both convective cooling, and hooking it up to a water cooling system, but I only if I really wanted to overclock. I was also going to use Peltiers on the back side of the aquarium to cool this water (only to room temperature, I do NOT want condensation...).

So how about this idea, eh? Just as long as I am smart about building the jacket, there should be no leaks.

 

[aspire]

I then thought about using this water jacket for both convective cooling, and hooking it up to a water cooling system, but I only if I really wanted to overclock. I was also going to use Peltiers on the back side of the aquarium to cool this water (only to room temperature, I do NOT want condensation...).

So how about this idea, eh? Just as long as I am smart about building the jacket, there should be no leaks.

Don't even bother with Peltiers.

To cool that amount of water were talking Kilowatts of Tec's.

 

[hikaricloud]

But water cooling actually had better cooling results than air cooling...oil cooling has worse...

 

[MilchstrabeStern]

Don't even bother with Peltiers.

To cool that amount of water were talking Kilowatts of Tec's.

The amount of water being cooled is irrelevant. The heat given off by the PC is transferred to the water, which slowly absorbs the energy. Since water is both more conductive and has a much higher specific heat than air, it can hold a lot of energy without a substantial change in water temperature. All the peltier has to do is remove heat at the same rate as it is being produced by the computer. It can be 20 gallons, and the peltier requirement would be the same. Of course this assumes the water is either flowing at a sufficient rate or is well mixed.

In fact, more water is BETTER. One, the water temperature will increase at a much slower rate. Second, more water typically means more opportunity for heat to be transferred elsewhere. Heat being conducted through the glass and convection is slowly transferring this energy to the air. More water = more surface area = more heat transfer.

Now I honestly do not know how much heat is produced by a typical overclocked gaming PC. With the inefficiency of peltier's, you are most likely write in your assessment of the large power requirement.

The point is that it's interesting and somewhat novel, and if I can run it off my power supply as if it were a cooling fan, I'd be happy.

 

[aspire]

The amount of water being cooled is irrelevant. The heat given off by the PC is transferred to the water, which slowly absorbs the energy. Since water is both more conductive and has a much higher specific heat than air, it can hold a lot of energy without a substantial change in water temperature. All the peltier has to do is remove heat at the same rate as it is being produced by the computer. It can be 20 gallons, and the peltier requirement would be the same. Of course this assumes the water is either flowing at a sufficient rate or is well mixed.

In fact, more water is BETTER. One, the water temperature will increase at a much slower rate. Second, more water typically means more opportunity for heat to be transferred elsewhere. Heat being conducted through the glass and convection is slowly transferring this energy to the air. More water = more surface area = more heat transfer.

Now I honestly do not know how much heat is produced by a typical overclocked gaming PC. With the inefficiency of peltier's, you are most likely write in your assessment of the large power requirement.

The point is that it's interesting and somewhat novel, and if I can run it off my power supply as if it were a cooling fan, I'd be happy.

You would need a separate power supply to run your peltiers. To be able to keep an i7 gpu cool and using a custom machined double sided water block set, (one loop cooling the hot sides of several peltiers, ann another loop being cooled by the cold side), we're talking 3 or 4 450W tec's. Running at around 20-24V. This requires some heavy duty power supplies, first capable of outputting such a high voltage, and second with a high enough wattage rating to power all the peltiers.

It would be cheaper, and a lot less hassle to just use sufficient radiators to maintain a rather small delta between coolant and ambient temps.

Also, how would you be exposing the water to the cold side of the peltier? Just sticking the cold side to the glass of the aquarium is not going to do it.

How are you going to be cooling the hot side of the peltiers? The hot side MUST be cooled with some serious heatsinks or better yet, a water cooling loop of their own.

The most successful use of peltiers in cooling a computer is in direct contact with the chip being cooled, and then water cooling the hot side.

 

[MilchstrabeStern]

Also, how would you be exposing the water to the cold side of the peltier? Just sticking the cold side to the glass of the aquarium is not going to do it.

How are you going to be cooling the hot side of the peltiers? The hot side MUST be cooled with some serious heatsinks or better yet, a water cooling loop of their own.

The most successful use of peltiers in cooling a computer is in direct contact with the chip being cooled, and then water cooling the hot side.

I haven't given it any serious thought. I was just planning on slapping it against the glass and cooling the hot side convectively with heat sinks and fans. The other idea is to just snake some copper pipes between two cold sides of peltiers and pump the water through these pipes. That way it's just a heat exchanger. Having a water cooling loop on the backside of the peltier seems very counter productive. You'd essentially be pumping heat from the water to warm up the water.

I'm not going to do the direct contact approach because I'm a little worried about condensation.

I have old power supplies laying around that can provide 450-500 W specifically for this purpose. Yeah, maybe it's a bit ridiculous, but I also think having multiple video cards, or spending a 1,000 on an overclocked processor is also ridiculous.

Also, it's not like I'd be using the peltiers at full power all the time. Only while I'm gaming, and honestly, how much time for intense gaming am I going to have as a grad student? Not a lot.

Anyways, it's all just an idea. I think it's interesting at least, conversation starter. That way when I invite those HOT cornell grad students over, they'll be sooooooo interested in my computer.

EDIT: I feel like if I start to seriously consider this "case" design, that I should start a new thread soon, because we're no longer talking about oil cooling.

 

[aspire]

I haven't given it any serious thought. I was just planning on slapping it against the glass and cooling the hot side convectively with heat sinks and fans. The other idea is to just snake some copper pipes between two cold sides of peltiers and pump the water through these pipes. That way it's just a heat exchanger. Having a water cooling loop on the backside of the peltier seems very counter productive. You'd essentially be pumping heat from the water to warm up the water.

I'm not going to do the direct contact approach because I'm a little worried about condensation.

I have old power supplies laying around that can provide 450-500 W specifically for this purpose. Yeah, maybe it's a bit ridiculous, but I also think having multiple video cards, or spending a 1,000 on an overclocked processor is also ridiculous.

Also, it's not like I'd be using the peltiers at full power all the time. Only while I'm gaming, and honestly, how much time for intense gaming am I going to have as a grad student? Not a lot.

Anyways, it's all just an idea. I think it's interesting at least, conversation starter. That way when I invite those HOT cornell grad students over, they'll be sooooooo interested in my computer.

EDIT: I feel like if I start to seriously consider this "case" design, that I should start a new thread soon, because we're no longer talking about oil cooling.

The issue is, you need to seriously cool the hot side of a peltier to get any benefit from the cold side. A simple heatsink and fan is not going to do it.

There was a guy on xtremesystems who used two Thermalright Ultra Extreme's with custom mounting plates to cool the hot side of the peltiers while the cold side was against the cpu. It worked alright, but he was pushing nearly a Kilowatt to the Peltiers alone, and cost more than 300 dollars in parts alone.

The glass is going to severely insulate the liquid from the tec and doesn't act as a good heat transfer medium. A flat surface does not offer much in the way of surface area for the cold to transfer to the hotter liquid.

"You'd essentially be pumping heat from the water to warm up the water. "

This quote makes no sense whatsoever. You obviously don't seem to be following me.

You either run a single loop with the water block and cpu sandwiching the tec. The cold side is in contact with the cpu and the waterblock is in contact with the hot side. The colder you keep the hot side, the cold the cold side becomes.

Or, you run two separate loops. You then use two water blocks one that mounts on each side of the tec. One loop just removes heat from the tec, while the other is cooled by the tecs and actually cools your components.

The problem with your "old power supplies" is that they don't offer more than 12V outputs and thus your only going to be able to run any tecs you buy at half their power ratings. Also your power supply will probably lack the necessary power to run more than a single high power tec. When all is said and done, your going to be talking about a good 2kw of electricity to run everything.

Just go for a straight water cooling loop, save yourself a ton of trouble, save some money, and have a lot less headaches in the future.

 

[MilchstrabeStern]

"You'd essentially be pumping heat from the water to warm up the water. "

This quote makes no sense whatsoever. You obviously don't seem to be following me.

You either run a single loop with the water block and cpu sandwiching the tec. The cold side is in contact with the cpu and the waterblock is in contact with the hot side. The colder you keep the hot side, the cold the cold side becomes.

Or, you run two separate loops. You then use two water blocks one that mounts on each side of the tec. One loop just removes heat from the tec, while the other is cooled by the tecs and actually cools your components.

I misunderstood your second point here. I somehow thought you were using the same cooling water, the water you are trying to cool to begin with, to cool the hot side of the tec. I was really hoping this wasn't the case, lol.

I understand the idea of having direct contact between the tec and the water block and have seen that done before. Perhaps this is the easiest, most affordable implementation. I've read somewhere that this works better than water cooling alone, except you have a major problem if, for some reason, the tec quits on you. Then you're stuck with an insulator between your CPU and water block. I would need a way to keep water vapor away from the pelt though if I pursue this option.

Yes, the flat pelts are very inconvenient for transferring heat from cylindrical objects, like pipes. An easy way to get around this problem is just to use a CPU or GPU water block on the pelt.

For removing heat from the hot side of the pelt I know the solution! I'll just slap like 3 more pelts on that baby in parallel, stick a ginormous heat sink on it, then stick it out my window when it's 10 degrees out. Yeah...

Whatever, it's all about the journey anyways, right? Chances are I will just stick to regular air cooling. I'm just trying to keep things interesting here. I think having that water jacket will look cool enough by itself that I won't end up using it for cooling at all. Water cooling alone is generally overkill from what I've read, I'm only interested in doing it because it's different, not because it's better.

EDIT:

A note about the old power supply: It has a 12V output @ 12 amps, so that's 144 W right there, most small tecs I've seen have been right around 144 W. Yes, the bigger ones are around 400+, I know.

There's also 5V @ 25 amps, which is another 125 W. So yeah, I'd be rather limited, but I can probably do 1-2. I honestly am probably just going to buy a few on ebay and just play with them. I've always wanted to anyways. I'll set up a calorimeter and see how much energy I can move with different configurations (assuming I ever have time).

 

[aspire]

A note about the old power supply: It has a 12V output @ 12 amps, so that's 144 W right there, most small tecs I've seen have been right around 144 W. Yes, the bigger ones are around 400+, I know.

There's also 5V @ 25 amps, which is another 125 W. So yeah, I'd be rather limited, but I can probably do 1-2. I honestly am probably just going to buy a few on ebay and just play with them. I've always wanted to anyways. I'll set up a calorimeter and see how much energy I can move with different configurations (assuming I ever have time).

The problem with your logic on powering tec's is that most take 20V inputs. That being the case, if you feed it 12V it's only going to be running at 60% capacity so you can basically cut it's wattage rating in half. Then factor in efficiency ratings say 80% to be generous, and your at 48% of what it's claimed efficiency is.

You can't run a higher voltage tec with a lower voltage and expect it to compensate by drawing more amps. Electronics just don't work that way..