How do heat pumps work if this is true?
Not really. Despite the CPU running super hot, the water in a water cooling loop is a couple degrees C above ambient. Carnot’s law, which provides the theoretical maximum energy extraction not accounting for any real world loss, means you can’t get significant energy from a couple of degrees C temp difference.
notfromhere@lemmy.ml 20 hours ago
Blue_Morpho@lemmy.world 18 hours ago
Heat pumps work by using a high power pump to compress the working fluid from its gas state to liquid. This compression heats the gas to 30 C over ambient. That high temp is then cooled through coils to air temp (30C on a hot day). In the winter it runs in reverse. It takes expands gas which in the case of a heat pump fluid like R-410a boils at -48C, runs that -48C gas through the coils which picks up heat from the outside at -1C (48C hotter than the gas). It then compresses that fluid to 20C which is then released into the house. That’s why heat pumps can freeze up in the winter. The cold cycle is so far below freezing that ice will form on the coils. So heat pumps switch to AC mode every now and then in the winter to warm up their coils to melt the ice off (while turning off the fan in the house). The need for a large temperature difference is why heat pumps don’t work if it gets below -20C.
So a heat pump moves heat but requires a large temp difference which comes from the electric compressor. You also can’t extract significant work from that heat difference once you factor the energy input of the compressor. Otherwise heat pumps would have a device to power part of itself from the heat that it is moving.
notfromhere@lemmy.ml 13 hours ago
So couldn’t we use a heat pump on the few degree above ambient waste heat to then do real work with it?
Blue_Morpho@lemmy.world 1 hour ago
Carnot is 1 - T(coldside)/T(hotside) in Kelvin.
For a heat pump dumping heat a couple C above ambient (typical of heat pumps which is why people coming from furnaces complain that the air coming out isnt hot) that’s n = 1 - 298/300 = 1-0.9933 = 0.67%
That means you’d need a machine that is more than 99.33% efficient to get any work out of that heat difference. For comparison, an engine losses 15-30% of energy just from the friction of polished metal cylinders and cams gliding close to another polished metal wall with a layer of oil in-between. The metal isn’t even touching.
bluGill@fedia.io 23 hours ago
You can get just enough energy for a science fair demonstration. Which scaled up to a data center size is a lot of energy. What the science fair misses is how much energy goes into making the system - You can generate what looks like a lot of power until you realize that the generators and such needed more energy to make than you will get back.