China Dives in on the World's First Wind-Powered Undersea Data Center (wired.com)
- Reference: 0179894314
- News link: https://tech.slashdot.org/story/25/10/28/1736211/china-dives-in-on-the-worlds-first-wind-powered-undersea-data-center
- Source link: https://www.wired.com/story/china-dives-in-on-the-worlds-first-wind-powered-undersea-data-center/
Seawater acts as a natural cooling system for the submerged servers. Traditional land-based data centers devote up to 50% of their energy consumption to air conditioning. The underwater design reduces cooling energy demand to less than 10%. The first phase is designed to achieve a power usage effectiveness rating of no more than 1.15. More than 95% of the facility's electricity comes from offshore wind turbines in the East China Sea. The project reduces land usage by more than 90% and eliminates the need for fresh water. The main contractors signed an agreement to launch another offshore wind-powered underwater data center with a capacity of 500 megawatts.
[1] https://www.wired.com/story/china-dives-in-on-the-worlds-first-wind-powered-undersea-data-center/
Project Natick (Score:4, Informative)
Microsoft's Project Natick was an early demonstrator that sea cooling was viable. They did not use wind energy for power, so I guess this can be claimed as a first, but it mostly seems like just a me-too.
Re: (Score:2)
It was also a demonstration project, I think it was mostly treated as backup for one of the land-based data centers, this is supposed to be a production DC.
Cost and Efficiency (Score:5, Informative)
The problems with underwater data centres are: Cost to build a facility at scale that can exist underwater (pressurization, leak detection, etc). Inconvenience of access for operational maintenance and repair. Cost to mitigate against corrosive salt water in your chilled water loops and pumps. Damage to marine ecosystems and the effects of warming localized water. Etc.
We keep trying to do it because the siren song of almost unlimited free cooling is such a draw. A PUE of 1.15 is very impressive, but you can get a PUE of 1.35 in Arizona with only air-cooled chillers (so not consuming any water). Microsoft famously experimented with this from 2018 to 2020 and claimed it as a success due to a low failure rate (they filled the container with nitrogen before submerging), but AFAIK have done nothing since.
Do you need saltwater in your pumps? (Score:2)
> Cost to mitigate against corrosive salt water in your chilled water loops and pumps. Damage to marine ecosystems and the effects of warming localized water. Etc.
Hmm, I don't have details about the design due to paywall, but I don't think running seawater in pumps and loops makes sense. It would make more sense to run treated coolant internally and have it push the heat to a radiator or different cooling system externally. The salt water can corrode the exterior and the radiators, but one would assume they're going to pump as little outside water as possible, from an avoiding clogging with dirt and bacteria perspective. Also, this is the ocean. Unless it's a very
Re: (Score:2)
A lot of data centers today are just sealed cargo containers full of racks of servers. Microsoft first experimented with containerized DC around the turn of the century, they had a DC in Colorado they called 'the trailer park' because it was just a bunch of containers and a double-wide for the techs to hang out in, surrounded by a fence. For cloud operations you can set them up so that something like a fifth or more of all servers could fail and no data would be lost, so the double-wide is no longer neces
Why underwater? (Score:3)
Why under water? You need a fresh water loop to a radiator in the ocean any way and whether that loop is 10 meter or 100 meter doesn't matter much.
Is it because a submerged container doesn't have to worry about weather like a barge does? Why not on land with a sewer pipe to a sufficient deep part of the ocean?
Re: (Score:3)
These don't need piping, just radiators on the outside of the container. Convection will do all the work for you. You don't want piping in the ocean because it rapidly gets clogged with barnacles and other forms of sea life.
Re: (Score:2)
You could use "concentric" pipes in a closed loop (the smaller pipe can just lay on the bottom, no need to be exactly concentric). Can fill the pipes with sea water and then use electrolysis chlorination, nothing will live in there.
Real estate + radiation (Score:2)
> Why under water? You need a fresh water loop to a radiator in the ocean any way and whether that loop is 10 meter or 100 meter doesn't matter much.
> Is it because a submerged container doesn't have to worry about weather like a barge does? Why not on land with a sewer pipe to a sufficient deep part of the ocean?
Just guesses, but I know that when MS did Project Natick, the servers ended up being more reliable. I don't know if that's the nitrogen pumped in or the natural shielding from radiation, but it was surprising.
The other obvious answer is real estate. For many important geographic points, it's cheaper to run electricity and networking to submerge them off the coast than to buy land for the datacenter.
There's also security. It's really hard to to get physical access to a machine that's not meant to
Re: (Score:2)
Why not in the Arctic ? Just use air cooling in the data center, and the Arctic will make cooling the nuke more effective too.
Seems like a double edged sword (Score:1)
While not needing to compress and decompress gasses along with other inefficiencies of a standard data center sounds good. What are the long term issues?
How much energy was required to make, test, transport, assemble, and test was needed? Additionally, a hole in the side of a terrestrial data center can be patched easily. A hole that forms in this means a total loss. What about replacing outdated equipment? Even in nutrient poor water radiators and plumbing will need maintenance due to algae, bacteria
In the 80's was a joke about (Score:2)
...slacker students taking "underwater basket-weaving" to fill credits. Now those skills are in demand!
What warms the ocean more? (Score:2)
What warms the ocean more? Greenhouse gasses from power generation for cooling, or just using the ocean directly as a heat sink.
Re: (Score:2)
> What warms the ocean more? Greenhouse gasses from power generation for cooling, or just using the ocean directly as a heat sink.
While that may matter to you and to me as people that would prefer the biosphere remain inhabitable for the average human, I believe the decision makers have already decided that we're in "keep burning until it's dead" mode, or at least until they can eliminate enough of us non-owner class folks to stop the incessant whining about ruining the environment. Humans don't matter. The biosphere doesn't matter. Profit matters. Continuing in the correct direction to continue accumulating wealth matters. That's the
Re: What warms the ocean more? (Score:2)
On a large scale, the less electricity used to cool would be considered a net global benefit. Locally, probably not.
Re:What warms the ocean more? (Score:5, Informative)
> What warms the ocean more? Greenhouse gasses from power generation for cooling, or just using the ocean directly as a heat sink.
Greenhouse gasses, by orders of magnitude.
The gasses leverage the vast energy available from the sun over a span of decades. The server heat is one and done.
Undoubtedly, the greenhouse gasses released from powering these servers will warm the oceans far more than the servers themselves will.
Re: (Score:2)
They're going to get most of the power from wind, but you're undoubtedly right that the last 5% will create more warming over the next century until it's processed out of the atmosphere than the DC will over the next decade.