Aren't heat pipes the same principle?
Except heat pipes connect to the hot parts directly and use no fan.
I know I'm being naive - I just don't know how.
Hitachi is mounting refrigerators on the rear doors of racks to suck out heat at a much lower cost. The Hitachi Data Centre Solutions Group idea, announced at yesterday's Datacenter Dynamics conference in London, is to treat a rack as a quasi-refrigerated cabinet and have its heat turn refrigerator liquid into a gas. Air is …
Heat Pipes do indeed work that way, and are much more efficient. However, you would need servers especially built for the purpose, which is pretty well what you won't get in a co-lo environment. What you need is a solution that will cool general purpose servers.
I don't think this can be the full cooling story. It will have a narrow working band in that it can only cool servers where the exit air temperature is higher than the outside air temperature, and the air temperature coming out the back panels cannot be lower than the coolants boiling point (which has to be high enough to condense at outside air temperatures).
So I see this something that reduces the air-conn power bill and will need to be supplemented with normal air-conn equipment.
Did they hire the old 1960's Control Data Corp cooling engineers? Does this have desuperheating and a hot gas bypass for cool load leveling to prevent condensation?
In answer to AB, 'heat pipes' have to get rid of their heat somewhere, usually ambiently. This provides a water heat exchange mechanism to remove the heat at a physically removed location..
@StephenD " What does this mean? I'm pretty sure they haven't produced a power source with greater than 100% efficiency"
In refrigeration the COP is the amount of cooling you get (in kW) divided by the amount of energy required to run the refrigeration system (also in kW), which is usually the compressor and fans. COP is not an efficiency it is just a ratio of paid energy input to energy transferred.
In refrigeration you generally design the system so that you transfer a whole lot more heat from inside the refrigerated environment (to outside) than you put electrical energy in, thus the COP is typically bigger than 1.
In this case, because there is no compressor, they are claiming that the only paid energy input is the fan, thus because the fan energy is relatively small the COP is apparently large.
However, this means that the cabinet can't be refrigerated below the temperature of where they are condensing the refrigerant gas. Which is why in the example, they are sending the vapour off to a chilled water heat exchanger. The real cost is the cost of creating that chilled water. That is the COP which is relevant here; which is likely going to be on the order of 2-4.
So it is rubbish.
what it says - that only 1 kJ is required to remove 43.5 kJ of heat from the system, per second. It's not producing energy, it's simply doing 1kJ of work per second to decrease overall system entropy - it likely isn't the whole story as at the other end, the water is evaporating due to heat, so 3rd law of Thermodynamics isn't violated ;).
Also, for many years you could have had a home heating with apparent >100% efficiency - installing a heatpump to remove energy from the surroundings and heat up your home during winter is much cheaper than using the same energy for heating directly. And, given the current craze about CO2, is also more 'enviromentally friendly'.
They're using 1 kW of power to move 43.5 kW worth of heat from one place to another, not creating it.
The same is true of a garden variety A/C unit; a typical portable A/C might be able to remove 14,000 BTU/hr, which is the equivalent of about 4 kW, while only consuming 1.5 kW from the wall outlet.
With the current cost some companies have for electricity (ask Google), it is becoming a better option to develop servers that never heat up. Big companies like Google will eventually have to build their own nuclear power plants to supply power for their datacenters (Wind Farm is already in place right?). The day someone creates a processor that never increases in temperature, even at full capacity, IT costs will go down considerably.
And of course, the person would win a Noble Prize, easily.
Wonder if they could sell it at a premium to the Docklands technorati? It might be a bit cloudy and furry, but could be marketed as ionised water or something. There's always a market for shite.
Talking of shite - Ofcom missed a trick in not selling their brand of bottled water. I can see the ad line now
"Ofcom are proud to introduce Eau de Ofcom....The water crafted to embody our corporate ineptitude - the only bottled water with more gas than content, and for the discerning Eurocrat, the new limited edition Eau de DG Enterprise with real Heinz (Zourek) best clagnuts and dangleberries for that real distinctive fresh EU bullshit flavour.