Liquid cooling

Computer hardware produces a huge amount of heat when operating. Usually this heat is removed by a combination of heatsinks and fans

The grey heatsink conducts the heat away from the processor and the sink’s fins give the heatsink a larger surface area for the air moved by the fan to blow over. Some computers use very large heatsinks in order to do away with the need for a fan entirely, relying only on natural convection currents for cooling.

Some computers do away with fans by pumping water past the heatsink; water is a much better absorber of heat than air* and therefore the system uses less power for cooling.

Green Revolution Cooling have gone one step further – they actually submerge the computing hardware in a special non-conductive liquid. This liquid then circulates, transferring the heat away to an external evaporation tower.

They claim that their cooling system will pay for itself within 1-3 years.

* The specific heat capacity of air is 1.007 joules per gram per kelvin and the specific heat capacity of water is 4.187 J/g/K. This means that water will absorb more than four times the energy of the same amount of air for the same increase in temperature. Green Revolution don’t say what the specific heat capacity of the fluid they use is, but it’s likely to be greater than water’s.

9 thoughts on “Liquid cooling

  1. “Fluid submersion cooling is made possible by GreenDEF™ coolant, a specific but not proprietary formulation of dielectric (nonconductive) white mineral oil that has 1,200x more heat capacity by volume than air.” Does this mean that the oil has a heat capacity of 1,200 J/g/K?

  2. How is specific heat capacity determined? What physical properties dictate how much energy can be absorbed for a given aise in temperature?

  3. It was my understanding that due to all the hydrogen bonding, not much is able to beat water at its specific heat capacity.
    Yep, just checked – not much is able to beat water except ammonia and, weirdly, liquid lithium.

  4. It’s concerning these guys don’t appear to know the science, you say they think their coolant has a specific heat of better than water?

    It looks like they are using a hydrocarbon based oil, there is a video of one of their customers dunking a hair dryer and setting fire to it.

    Oils like PAO have a specific heat (per weight) of 2300 J/kg/K, but are hold around 1200-1300 times more heat per litre volume than air.

    Water holds about 3500 times more heat per litre volume than air.

    I’m not sure about having hydrocarbons in this sort of quantity in a data centre, regardless of how high the flash point is, shorting electrics on 100A breakers can get very very very hot.

    I think 3M Novec based submersion cooling has more promise.

  5. Surely if water has a specific heat capacity ~3500 timesthat of air, as stated by “Pete”, then using incredibly pure water would’ve best? Is this not done because of the difficulty that would be encountered in trying to achieve a suitable purity level? Is there an other reason? How difficult would it in fact be to purify water to the extent that it could be used for cooling? So many questions…

Leave a Reply