Earlier quoted context omitted.
Maybe, that's why they want Greenland so bad. Low temperature, many free spaces, no significant nature and if there is one, not enough citizen who can complain. And if they still need water for cooling, there is probably enough from the melting ice they can use.
All the money they save on water cooling will be spent on massive undersea fiber lines to move the data to and from the island.
45°C cooling design cuts data center water use to near zero
401–410 of 450 posts
Re: 45°C cooling design cuts data center water use to near zero
#402Water use isn't a myth. Our data center is mostly air-cooled, with 100F hot aisles and chilled water going to heat exchangers in every third rack position, although some of our newer GPU racks use chilled water directly. For most of the year that means we don't need chillers - the warm water return is sufficiently hotter than outdoor temp that heat "flows downhill" with a bit of pumping. But we use an evaporative coo…
I disagree if there is an assumption or premise that all datacenters use water for cooling. Datacenters are not built or operated the same, nor have a requirement for water cooling, or evaporative cooling.
Water use is partially a myth for sure, because it's not a certainty or requirement in every datacenter, ever.
If water is used for evaporative cooling, it's used in places that the cooling can't work above a certain temperature. Say, for example the US where it gets over 120 degrees farenheit.
The inability for cooling equipment to perform when it's extremely hot might be supplemented my multiple technologies cooling together, including evaporative cooling. So evaporative cooling might be common in the US or other regions like it.
There are many datacenters that don't need to consume water at all for cooling, except for a water cooler to drink from, and sprinkler systems for fire suppression. They are engineered and cooled just fine.
The demand for datacenters and money in this space will likely solve the water usage, if for no other reason, than to increase flexibility of locations.
If anyone is curious a bit more about how:
Datacenters locations optimize 3 main factors.. Power, Cooling, and Fibre.
We know the climate of location isn't the same everywhere in the world where datacenters are located.
One thing that can get missed among the water hyperventilating crowd is that datacenters are designed for the need of the location's climate, and what it takes to cool. Too many folks spreading the information are usually non-technical, and get their education about datacenters from social media.
There are many places in the world, including in Canada, where cooling is built in 6 months of the year to outdoor air. The temperature isn't high enough in the summer to warrant or require evaporative cooling. They can just vent outside, or capture the heat for other purposes.
In other areas still, proven building approaches such as countersinking the datacenter for some additional geothermal benefit, local power generation (Solar, Natural gas) and a unique path to the same datacenter is possible.
Of course, if foreign companies are wanting to build in a different country, they might by default build how they know for their region and assume it's ok.
There is a massive build out of datacenters occurring in Canada right now, they will be built according to how buildings are built in that climate, not importing cooling technology from regions where it's too hot, or evaporative cooling is the only, or most available option. In temperate climates, considerations have to be made for keeping keep buildings warm (not just cool) year round.
The nice thing is this is not hypothetical.
There is of course, equipment that can have liquid cooling, or we see those fun videos of equipment submerged in liquid operating. These seem to not be the common.
Re: 45°C cooling design cuts data center water use to near zero
#403> In the right geography — somewhere with reliably cool outdoor air Aaahh, there’s the catch. “Save resources on cooling by building your data centre somewhere cold, and pollute the surrounding environment by dumping your waste heat wholesale into that!” Good job Nvidia, I almost thought we had something good there.
Datacenter waste heat even a problem? I only ever heard of nuclear power plants waste heat being a problem when the cooling water is dumped directly into rivers (instead of the ocean).
Re: 45°C cooling design cuts data center water use to near zero
#404> In the right geography — somewhere with reliably cool outdoor air Aaahh, there’s the catch. “Save resources on cooling by building your data centre somewhere cold, and pollute the surrounding environment by dumping your waste heat wholesale into that!” Good job Nvidia, I almost thought we had something good there.
Re: 45°C cooling design cuts data center water use to near zero
#405Points for effort, but this will do literal nothing to appease the opposition, since the "water use" thing is a myth anyway. It's probably good, it sounds like it will be more efficient and efficiency saves resources for other uses, but politically it's completely useless.
Do you have any references for this comment: '"water use" thing is a myth anyway'. I would like to show my kid who keeps complaining to me about AI water usage.
Re: 45°C cooling design cuts data center water use to near zero
#406> In the right geography — somewhere with reliably cool outdoor air Aaahh, there’s the catch. “Save resources on cooling by building your data centre somewhere cold, and pollute the surrounding environment by dumping your waste heat wholesale into that!” Good job Nvidia, I almost thought we had something good there.
Re: 45°C cooling design cuts data center water use to near zero
#407> In the right geography — somewhere with reliably cool outdoor air Aaahh, there’s the catch. “Save resources on cooling by building your data centre somewhere cold, and pollute the surrounding environment by dumping your waste heat wholesale into that!” Good job Nvidia, I almost thought we had something good there.
It's not a sleight of hand.
Datacenter locations mainly optimize power, cooling, and fibre.
Cooling is not the same everywhere, even the US has places that are cool enough to not need evaporative cooling water use.
Re: 45°C cooling design cuts data center water use to near zero
#408Earlier quoted context omitted.
You cannot have evaporative cooling without evaporating some water into the atmosphere. A closed circuit cooling tower still has water spraying onto the closed loop process water heat exchanger coil and mixing with atmospheric air to evaporate and cool the process water indirectly instead of evaporating and recirculating the process water that doesn’t evaporate directly like in an open-loop cooling tower. I suppose y…
> I suppose you could condense the evaporated water somehow by using a chilled umbrella or some other ridiculous contraption above the cooling tower, but why would you do that? Specifically to reduce the ongoing demand for water. DCs need to get to net-zero on their energy requirements and their water consumption. They are already losing their political license to operate because they're not. That's independent of th…
You may think these are necessary things to do, but data centers are being built as we speak that are not net zero on energy or water consumption. That’s reality.
It would be ideal if data centers had net zero energy and water consumption, don’t get me wrong. That won’t happen without legislation, and even then legislative efforts will be a patchwork that won’t cover everyone due to some politicians selling out their constituents.
Re: 45°C cooling design cuts data center water use to near zero
#409Earlier quoted context omitted.
> AI and datacenters are here to stay, you can’t fight them, Maybe you can't (or don't want to), but people can absolutely fight data center construction. Your lack of imagination doesn't bind others.
But why? I want cheap tokens. Build a data center on every block (or more realistically in all the empty space we have everywhere).
Re: 45°C cooling design cuts data center water use to near zero
#410Earlier quoted context omitted.
Of old... https://web.archive.org/web/20210115152829/https://www.nrel.... It's got a "heat energy to/from campus" exchange in there. That's a link in March and the air temperature was 31°F. https://web.archive.org/web/20210708150410/https://www.nrel.... is later with air temperature of 68°F.
The diagram you shared has an open loop chiller, are those still needed as a backup?
https://web.archive.org/web/20210708150410/https://www.nrel....
The thermosyphon took it from 90.2°F to 70.8°F ... and then it went to the counterflow cooling towers which took it from 69.8°F to 64.8°F. If that thermosyphon did the majority of the work and there's not too much more in the cooling towers... and that was only rejecting 375kW of heat through evaporation.
Note one of the parameters with it in the dashboard: dry bulb temperature.
https://web.archive.org/web/20251207023030/https://docs.nrel... for more info on the thermosyphon.
I'm disappointed that the dashboard isn't available anymore. It was rather neat to look at.
Some older articles about it:
https://www.nlr.gov/news/detail/program/2023/a-decade-of-gre...
https://www.nlr.gov/computational-science/data-center-coolin... (click through these)
When possible, heat energy from the energy recovery loop is transferred to the building process hot water loop, which provides heat for the office and laboratory spaces within the building. The energy recovery heat exchanger (6) transfers heat from the ERW loop to the process hot water loop.
After re-use potential is exhausted, warm ERW water flows to the fourth floor mechanical room. When temperatures permit, heat is dissipated through a thermosyphon (7), which is an advanced dry cooler that uses refrigerant in a passive cycle to dissipate heat.
Remaining heat is transferred from the ERW loop to a tower water open loop via the cooling tower heat exchanger (8). Cooling towers (9) cool the tower water loop by cascading that water across fill material while drawing ambient air across the fill material. This provides a very energy-efficient way to cool water with sensible (heat dissipated to air without evaporation) and latent (heat dissipated with evaporation) heat transfer. However, this evaporative cooling process requires a continuous source of water.
---https://www.nlr.gov/computational-science/reducing-water-usa...
In August 2016, a prototype thermosyphon cooler—an advanced dry cooler that uses refrigerant in a passive cycle to dissipate heat—was installed. The thermosyphon was placed upstream of the HPC Data Center cooling towers at the ESIF to create a hybrid cooling system. The system coordinates the operation for optimum water and operating cost efficiency—using wet cooling when it's hot and dry cooling when it's not.