Cooling Theory Flawed?

[chikit]

This may be completely unrelated, but as I was sitting on the jon today doin a number 2 and I was thinking about the theory of cooling, specifically the differences between air and water cooling.

Some mainstream water cooling sets don't outperform the best of air cooling solutions and I was wondering WHY? I mean isnt it that, using the basic laws of physics, water has more particles densely packed together rather than a gas (air)? I was also wondering the this sitting in the pool, how there was a railing leading into the pool and where the metal was out of the water it was hot as hell (sun) and the metal in the water was really cool to touch (obviously). I was thinking since, there is such an obvious difference shouldn't this be the case with water/air cooling computer components?

Then I got to thinking, how can I improve water cooling performance, and how is it that air cooling can do such a great job. Imagine blowing on a heat source, it doesn't cool down that quick. Splash some water on it and its already mild/cool to touch. Because of the fact that water has many more particles per area to transfer the heat energy from the heat source, to the water particles themselves. Heat cannot dissapear, its only transferred.. and is the last form of energy for those of you who studied basic physics before. Water should be the better conductor than air. More so than that, solids should be a better conductor than water. That lead me to thinking of a hypothetical, ideal heat conductor. A never ending, ever lasting long or big, solid. Thats the reason for the growing sizes of heatsinks. Its basically a solid taking away heat energy from the heatsource, only to be trapped in the heatsink, waiting for dissapation from the active air cooling of fans. If that does such a good job already, why isnt water that much better?

Heatsinks are designed to not only be big to retain more heat capacity (in the literal sense) but also with greater surface area for air to pass through. Why can't this work with water? instead of having a water block, why can't we design a heat sink with great surface area, and water pass through to transfer the heat energy away? So I'm thinking an ideal heatsink in my opinion, would be a big heatsink with many fins, in some sort of enclosure so that water can pass through the whole thing. Wouldn't this be better than a waterblock? Its basically having the heatsource transfer the heat in the best way possible - heatsource (solid) to heatsink (solid) and having water actively transport heat away. Solid to solid conductors are the best and fasest arent they?

But then you've got the problem of the heat in the water to transfer to somewhere else, because to cool effectively you need to TRANSFER heat. And the heatsource can never be cooler than the heatsink itself. Would this theoretical 'ideal' cooling solution be better than current water cooling solutions? Basically its has an addition of using a larger surface area heatsink for the water to pass through.

 

[Houndsteeth]

When you are talking about cooling, you need to take the entire system into account. Air cooling is effective because you are able to move more air molecules over the heat source, thus dissipating the heat energy into the air as long as the air is at a lower heat energy state than the heat source. Once equilibrium is attained (where heat source and air molecules are at the same energy level), any energy added into the system is unable to be dissipated, and will build up at the heat source.

In theory, water cooling should be more effective than air cooling because water has a much higher density than air, and is able to dissipate more heat. In reality, some systems are just unable to reach a state of equilibrium that is within a comfortable range for the hardware it is cooling. Whether this is due to the inability to move fluid through the water block at a fast enough rate, or if the ability of the water block to transfer heat to the fluid effectively, or if the radiator is not able to dissipate heat from the system fast enough is very telling of the quality of the hardware you are buying.

Cheap water cooling (~$150 US) is just as good as high end air cooling (~$75 US) in most situations. Mid to high end water cooling (~$250-400 US) is better than any air cooled system, no matter what price, because you are not able to reach that level of performance with air alone.

And yes, the heat source CAN be cooler than the device that dissipates the heat (heat sink). You can use thermoelectric couplers (peltier junctions) in conjunction with a water loop to reach sub-ambient temperatures. Granted, it's not very energy efficient, but for extreme overclocking, it is the only option outside of phase-change cooling, with the possible exception to using evaporative cooling in addition to your water loop.

 

[cb62fcni]

When you are talking about cooling, you need to take the entire system into account. Air cooling is effective because you are able to move more air molecules over the heat source, thus dissipating the heat energy into the air as long as the air is at a lower heat energy state than the heat source. Once equilibrium is attained (where heat source and air molecules are at the same energy level), any energy added into the system is unable to be dissipated, and will build up at the heat source.

In theory, water cooling should be more effective than air cooling because water has a much higher density than air, and is able to dissipate more heat. In reality, some systems are just unable to reach a state of equilibrium that is within a comfortable range for the hardware it is cooling. Whether this is due to the inability to move fluid through the water block at a fast enough rate, or if the ability of the water block to transfer heat to the fluid effectively, or if the radiator is not able to dissipate heat from the system fast enough is very telling of the quality of the hardware you are buying.

Cheap water cooling (~$150 US) is just as good as high end air cooling (~$75 US) in most situations. Mid to high end water cooling (~$250-400 US) is better than any air cooled system, no matter what price, because you are not able to reach that level of performance with air alone.

And yes, the heat source CAN be cooler than the device that dissipates the heat (heat sink). You can use thermoelectric couplers (peltier junctions) in conjunction with a water loop to reach sub-ambient temperatures. Granted, it's not very energy efficient, but for extreme overclocking, it is the only option outside of phase-change cooling, with the possible exception to using evaporative cooling in addition to your water loop.

Outstanding reply Houndsteeth!

 

[scifiguy]

Houndsteeth, let me add a little bit to your reply. Water "IS" a more effective heat transfer medium. THe reason that the low end water cooling systems rival the high end air systems is related to size of the water block/radiator and water pumping capabilites. Water will readily absorb heat, but getting it to transfer it back to air requires a careful balance of water flow, air flow and ambient air temperature. Let me give quick example. The thermostat in a car engine is NOT there to just regulate temperature. It's there to also restrict (slow down)the flow of the water in the system. The slower the water moves thru the system, the more heat it can accept and by the same token the more heat it can give back to the air thru the radiator. Up to a point.

Not disputing any of your salient points, just adding a little more.

 

[Groveling_Wyrm]

Scifiguy, I disagree with your analogy of the car radiator and thermostat. Whether the heat is transferred through 1 molecule of water or 1 million molecules of water, the amount of heat transfer coming from the engine will still be the same, due to the thermal conductivity of the engine. In this case the 1 molecule of water would have to absorb the same amount of energy as the 1 million molecules. Which is better?

However, I think that you did not take into account thermal conductivity of the water. The hotter the water gets, the less heat it will absorb, due to increased thermal resistance of the water. If the water absorbs less heat from the engine, then the heat of the engine has no where to go and the engine temperature rises. A computer would also heat up in the same fashion. This is why it is ALWAYS recommended to use more water, rather than less water. There are many different statistical examples of this, such as this EXAMPLE FROM SWIFTECH .

Common sense wise, I don't know of ANYONE in their right mind who would want to heat the water up before evacuating the heat from their computer. That would mean that in the event of a water pump failure, a person would have a LOT less time to power down the system, due to the fact that the system would already be hot.

 

[morerevs]

You can't compare a car's cooling system to a watercooled PC setup. In a computer you'd want the CPU to be as cool as you can get it, but in a car the engine needs to be a certain temp for fuel efficiency and the perfect fitting of a piston to the bore of the cylinder for example. Oil also is made with certain temps in mind aswell as fuel combustion points. The thermostat is there to regulate temps under varying conditions, NOT to improve cooling properties of coolant. If what you're saying was true you'd be paying quite a bit for a new thermostat as it would need complex electronics to balance flow and heat absorbtion. There is a big temperature controlled fan in front of the radiator that takes care of excess heat in case the radiator can't give off sufficient heat when for instance in a traffic jam or on a very hot day and even that is not very sophisticated. Anyways... just sayin'

 

[chikit]

okay back to the topic. would my idea of an imaginary thermalright u-120 extreme in a sealed container with water passing through it work better than a normal copper water block? i'm just thinking, if air cooling works that way, why cant we just use water instead of air. thats if we have sufficient radiators to dissapate the heat.

 

[drcroubie]

okay back to the topic. would my idea of an imaginary thermalright u-120 extreme in a sealed container with water passing through it work better than a normal copper water block? i'm just thinking, if air cooling works that way, why cant we just use water instead of air. thats if we have sufficient radiators to dissapate the heat.

Short answer, yes it would.
of course there would be the obvious technicalities of getting all the water to pass all of the fins evenly (ie not just passing through the centre while the outside water stays still). And then you'd need a radiator at the other end that's just as good (if not better).

i think what everyone's forgetting is that water "cooling" is not actually "cooling" per se. The water is simply used as a medium for transferring the heat from the waterblock to the radiator elsewhere, the point being that the radiator can be bigger than would normally be able to fit directly onto your cpu and thus dissapate more heat. But the inefficiencies come from the water not being able to pick up "all" the heat from the waterblock, and then dissapate "all" the heat to the radiator. Using your theoretical u120 inside a giant waterblock would make for a very-efficient waterblock (despite feasibility issues), and that's why some waterblocks are more expensive than others, as they've had more engineering time spent on making them able to more efficiently transfer heat from the cpu to the water (just as "big air"coolers like u120 and Ninjas are more expensive than "cheap air", because of the more engineering time taken to make them more efficient)