People with far more knowledge about this: when a data center “uses” water, what happens to it? Does the act of cooling servers with water “use up” the water or can it be cycled back into the water system? And if it is theoretically possible for it to be cycled back into the system as opposed to being dumped like sewer water, why isn’t it?
Because that costs money. It’s really that simple.
In many states with functioning infrastructure, a property like this that’s using that much water, would be required to build and maintain their own treatment plant to cover it.
But we’ve got regulatory capture, so… Good luck with that one.
They use evaporative cooling systems. The water absorbs the heat, turns into water vapor, and is vented into the air. In theory I assume it could be collected, but that would require a lot more post processing to cool it back down or compress it to get back to liquid for.
Most high rises have cooling systems. They run ambient temp water to all of the floors, and tenant hvac units cool their server rooms by pumping heat into that water.
It then goes through the loop and gets cooled back down to ambient with large radiator systems outside.
It’s closed loop.
There’s no reason this can’t be applied to datacenters.
I’d imagine DCs are doing it the cheapest way they can. But I think comparing the high rises to a DC doesn’t really work just due to the scale difference and the amount of heat needing removed. I’m sure there’s a way that it could be made something of a closed loop for DCs, but I’m guessing it would be a bit different if a process compared to high rises. I wouldn’t be surprised if a DC removed as much heat as a year of every high rise in NYC in a day or less…
Agreed, but i would think it’s mostly an upfront cost. Datacenters have massive footprints, the roofs could be covered with heat exchangers, also covered with solar panels. It just takes some regulation to encourage it.
If you want to get really into it, you could figure out a way to reclaim all that heat energy.
If like to see a true engineer mock or what it would entail. I think it would require a dedicated facility to pull it off given just how much heat needs removed and the volume of water used. The problem is there is an insane amount of heat energy from these DCs. It has to go somewhere, and water is a very good medium for that. That heat still has to go somewhere, so removing it from the water is much harder. If you had a huge setup of heat exchangers, they could probably do it with enough time and space, but time isn’t an option, because they run continuously. It would probably end up being just a huge holding tank that lets the hot water cool over time.
I don’t think anything like solar is going to help. It would be good environmentally to offset some of their power consumption, but I think it would be negligible in their overall power draw and wouldn’t have any effect on their cooling. It’s not like they are cooling airspace, which they also are, but it’s the components themselves.
The problem isn’t about the cooling systems “using” the water in the sense that the water “dissapears” from the water cycle per-se. The problem is that when a Data Center is built, the water that the population of that zone used, is mostly diverted from them to the data center, where it is evaporated and thus the population have access to less and less water.
Evaporated is ‘used’. It’s arguably more ‘used’ than it is when it goes down the sewer to local wastewater treatment, as that water is often put back into groundwater through infiltration fields. Depleting aquifers is the main concern of data center water usage.
I have a water-cooled PC and I only add water to it about once a year when the reservoir is around 2/3 full. My loop is water-tight, not air-tight, so water is slowly lost to evaporation.
I would think enterprise loops would be air-tight but perhaps the cost of implementing that is less than paying for water to be added
People with far more knowledge about this: when a data center “uses” water, what happens to it? Does the act of cooling servers with water “use up” the water or can it be cycled back into the water system? And if it is theoretically possible for it to be cycled back into the system as opposed to being dumped like sewer water, why isn’t it?
Because that costs money. It’s really that simple.
In many states with functioning infrastructure, a property like this that’s using that much water, would be required to build and maintain their own treatment plant to cover it.
But we’ve got regulatory capture, so… Good luck with that one.
They use evaporative cooling systems. The water absorbs the heat, turns into water vapor, and is vented into the air. In theory I assume it could be collected, but that would require a lot more post processing to cool it back down or compress it to get back to liquid for.
Most high rises have cooling systems. They run ambient temp water to all of the floors, and tenant hvac units cool their server rooms by pumping heat into that water.
It then goes through the loop and gets cooled back down to ambient with large radiator systems outside.
It’s closed loop.
There’s no reason this can’t be applied to datacenters.
I’d imagine DCs are doing it the cheapest way they can. But I think comparing the high rises to a DC doesn’t really work just due to the scale difference and the amount of heat needing removed. I’m sure there’s a way that it could be made something of a closed loop for DCs, but I’m guessing it would be a bit different if a process compared to high rises. I wouldn’t be surprised if a DC removed as much heat as a year of every high rise in NYC in a day or less…
Agreed, but i would think it’s mostly an upfront cost. Datacenters have massive footprints, the roofs could be covered with heat exchangers, also covered with solar panels. It just takes some regulation to encourage it.
If you want to get really into it, you could figure out a way to reclaim all that heat energy.
If like to see a true engineer mock or what it would entail. I think it would require a dedicated facility to pull it off given just how much heat needs removed and the volume of water used. The problem is there is an insane amount of heat energy from these DCs. It has to go somewhere, and water is a very good medium for that. That heat still has to go somewhere, so removing it from the water is much harder. If you had a huge setup of heat exchangers, they could probably do it with enough time and space, but time isn’t an option, because they run continuously. It would probably end up being just a huge holding tank that lets the hot water cool over time.
I don’t think anything like solar is going to help. It would be good environmentally to offset some of their power consumption, but I think it would be negligible in their overall power draw and wouldn’t have any effect on their cooling. It’s not like they are cooling airspace, which they also are, but it’s the components themselves.
NREL ESIF Data Center — Golden, Colorado https://maps.google.com/?q=NREL+ESIF+Golden+CO
Sandia National Laboratories HPC — Albuquerque, New Mexico https://maps.google.com/?q=Sandia+National+Laboratories+Albuquerque+NM
Munters SyCool Deployment (Representative Location) — Phoenix, Arizona https://maps.google.com/?q=Phoenix+AZ
Switch SUPERNAP Campus — Las Vegas, Nevada https://maps.google.com/?q=Switch+SUPERNAP+Las+Vegas
Equinix SV10 IBX — San Jose, California https://maps.google.com/?q=Equinix+SV10+San+Jose+CA
The reason is cost
The problem isn’t about the cooling systems “using” the water in the sense that the water “dissapears” from the water cycle per-se. The problem is that when a Data Center is built, the water that the population of that zone used, is mostly diverted from them to the data center, where it is evaporated and thus the population have access to less and less water.
Evaporated is ‘used’. It’s arguably more ‘used’ than it is when it goes down the sewer to local wastewater treatment, as that water is often put back into groundwater through infiltration fields. Depleting aquifers is the main concern of data center water usage.
evaporation is also part of the water cycle
I would also like to know the answer to this.
I have a water-cooled PC and I only add water to it about once a year when the reservoir is around 2/3 full. My loop is water-tight, not air-tight, so water is slowly lost to evaporation.
I would think enterprise loops would be air-tight but perhaps the cost of implementing that is less than paying for water to be added
I believe it evaporates