"Consume" water I interpret as requiring a constant stream of water. If little to no water is added, I'd agree it doesn't consume any, even though it uses water.
If you want to get pedantic, even open loop cooling does not consume water. The amount of H20 going in is the same amount that's going out. No water is destroyed.
No, it's per scope. If we're discussing a data center, it's perfectly reasonable to say that the data center is taking water in (that is to say, consuming it) and evaporating water out.
The water isn't being used for its enthalpy of vaporization, so there's a huge difference between that and using water to transport heat in a closed loop. It's pedantic to worry about the amount of water in the closed loop since it is only lost due to shutdown maintenance activities or leaks. Since the goal is for things to work more than 99% of the time this does not constitute a huge consumption. If you are using glycol there is further incentive not to waste glycol additive which is required to be mixed in at a rate proportional to the refill volume.
To be clear, I still consider this laudable and hope to change our house to use a heat exchange eventually, rather than a furnace plus air conditioner pair, per Technology Connections' advisement [1].
"Once we have filled a closed loop with water, we will not need add more very often"
It does use water.
"Consume" water I interpret as requiring a constant stream of water. If little to no water is added, I'd agree it doesn't consume any, even though it uses water.
If you want to get pedantic, even open loop cooling does not consume water. The amount of H20 going in is the same amount that's going out. No water is destroyed.
No, it's per scope. If we're discussing a data center, it's perfectly reasonable to say that the data center is taking water in (that is to say, consuming it) and evaporating water out.
"Consume" doesn't mean destroy either. It's contextual and its pretty clear what it means in this context.
The water isn't being used for its enthalpy of vaporization, so there's a huge difference between that and using water to transport heat in a closed loop. It's pedantic to worry about the amount of water in the closed loop since it is only lost due to shutdown maintenance activities or leaks. Since the goal is for things to work more than 99% of the time this does not constitute a huge consumption. If you are using glycol there is further incentive not to waste glycol additive which is required to be mixed in at a rate proportional to the refill volume.
> zero water > ultra-low Water Usage Effectiveness Those seem like contradictory statements. Ultra low is not zero.
Something with far more impact would be to get AI compute to consume 1% of the energy it does today.
It would have much larger impact than this.
So MICROS~1 has finally discovered the wonders of heat pump systems? Great, what next, photovoltaics?
Thermophotovoltaics (TPV) and Thermoelectrics.
This is also a positive development in HPC sustainability:
"New HPE fanless cooling achieves a 90% reduction in cooling power consumption" (2024) https://news.ycombinator.com/item?id=42106715
Nice! Hopefully Meta and Google are doing something similar or will soon.
[flagged]
Also, in a linked post [0], it looks like this still uses water in a closed loop heat exchange system. It just won't evaporate for cooling.
[0] https://www.microsoft.com/en-us/microsoft-cloud/blog/2024/07...
To be clear, I still consider this laudable and hope to change our house to use a heat exchange eventually, rather than a furnace plus air conditioner pair, per Technology Connections' advisement [1].
[1] https://www.youtube.com/watch?v=7J52mDjZzto 35 minutes
On YouTube, I've seen pools heated by immersion mining rigs and also pools heated by attic heat; but I've never heard of a "zero water datacenter".
Great work.
FWIU Type-III superconductors also have like no loss.
Is there a less corrosive sustainable thermal fluid or a water additive to prevent corrosion?
Is there (solid-state) thermoelectric heat energy reclamation?