Fusion - breakthrough or another false dawn
Discussion
CraigyMc said:
Mr Whippy said:
"AI" processing will require bucket loads of power...
Where I work there are AI-tasked 6U rack servers that consume 12kW per server. You can't put many of them in a rack because of the power and cooling demands. Power-dense doesn't fully cover it. They have thermals akin to a nuclear reactor.It genuinely won't be long before they start compounding them (using the heat output to generate power that is fed back into the system).
Bigger picture, how much is being put into high temperature semiconductors vs fusion power. The former would revolutionise our system efficiency & open the door to far better use of renewables.
skwdenyer said:
CraigyMc said:
Mr Whippy said:
"AI" processing will require bucket loads of power...
Where I work there are AI-tasked 6U rack servers that consume 12kW per server. You can't put many of them in a rack because of the power and cooling demands. Power-dense doesn't fully cover it. They have thermals akin to a nuclear reactor.It genuinely won't be long before they start compounding them (using the heat output to generate power that is fed back into the system).
Bigger picture, how much is being put into high temperature semiconductors vs fusion power. The former would revolutionise our system efficiency & open the door to far better use of renewables.
At the moment AI is the hot ticket so CIOs and CTOs would get demolished for not investing. How real its value is depends a huge amount on application. There are some straightforward applications where it works well already (eg. design of stressed components, engineering). There are some where it probably will work out better, but isn't there yet for everything (eg. trading) and there are some places where it's currently a total white elephant but nobody will admit that yet.
To answer your question about cooling towers by the way - a lot of old crypto mining outfits that have been sidelined by proof of value rather than proof of work are just reprofiling as AI compute barns. The actual kit in use itself is fairly similar, as are the locations that make sense (cheap, abundant power and cooling = go to Iceland or similar).
I vaguely recall the idea of using waste heat from server farms to heat water was floated somewhere I worked, but the temperature is insufficient to be useful - water from your house boiler comes out at 55 to 60 degrees, you can't get that from a server farm by the time you've run a heat exchanged, pipework etc..
It's the quite common annoying issue that it is low-grade waste heat with insufficient energy to do any useful work.
It's the quite common annoying issue that it is low-grade waste heat with insufficient energy to do any useful work.
Flooble said:
I vaguely recall the idea of using waste heat from server farms to heat water was floated somewhere I worked, but the temperature is insufficient to be useful - water from your house boiler comes out at 55 to 60 degrees, you can't get that from a server farm by the time you've run a heat exchanged, pipework etc..
It's the quite common annoying issue that it is low-grade waste heat with insufficient energy to do any useful work.
But if you pre-heat water with your server farm, you don't need as much energy to raise the temperature up further. It isn't either/or.It's the quite common annoying issue that it is low-grade waste heat with insufficient energy to do any useful work.
skwdenyer said:
Flooble said:
I vaguely recall the idea of using waste heat from server farms to heat water was floated somewhere I worked, but the temperature is insufficient to be useful - water from your house boiler comes out at 55 to 60 degrees, you can't get that from a server farm by the time you've run a heat exchanged, pipework etc..
It's the quite common annoying issue that it is low-grade waste heat with insufficient energy to do any useful work.
But if you pre-heat water with your server farm, you don't need as much energy to raise the temperature up further. It isn't either/or.It's the quite common annoying issue that it is low-grade waste heat with insufficient energy to do any useful work.
An interesting article which claims "It is time to drop the old joke that fusion is 30 years away, and always will be. A poll at the International Atomic Energy Agency’s forum in London found that 65pc of insiders think fusion will generate electricity for the grid at viable cost by 2035, and 90pc by 2040."
https://www.yahoo.com/news/nuclear-fusion-grid-com...
https://www.yahoo.com/news/nuclear-fusion-grid-com...
Caruso said:
An interesting article which claims "It is time to drop the old joke that fusion is 30 years away, and always will be. A poll at the International Atomic Energy Agency’s forum in London found that 65pc of insiders think fusion will generate electricity for the grid at viable cost by 2035, and 90pc by 2040."
https://www.yahoo.com/news/nuclear-fusion-grid-com...
Utter fantasy; generating a net output maybe, no chance of a commercial scale FOAK generating plant. That time scale would be an uphill struggle for a known tech like a PWR fission reactor. Try 2045 and even that would be optimistic.https://www.yahoo.com/news/nuclear-fusion-grid-com...
hidetheelephants said:
Caruso said:
An interesting article which claims "It is time to drop the old joke that fusion is 30 years away, and always will be. A poll at the International Atomic Energy Agency’s forum in London found that 65pc of insiders think fusion will generate electricity for the grid at viable cost by 2035, and 90pc by 2040."
https://www.yahoo.com/news/nuclear-fusion-grid-com...
Utter fantasy; generating a net output maybe, no chance of a commercial scale FOAK generating plant. That time scale would be an uphill struggle for a known tech like a PWR fission reactor. Try 2045 and even that would be optimistic.https://www.yahoo.com/news/nuclear-fusion-grid-com...
The technology is nowhere near the point of even having a trial system that could theoretically be scaled up to do something useful. The usual 30 year timescale remains optimistic having already been demonstrably wrong for several decades. I haven't seen anything that looks like a genuine breakthrough in this field since I was at school and started following this stuff in the 1980s. It remains exceedingly difficult to confine a plasma with magnetic fields, and you can only squish tiddly bits of matter even using super enormous lasers.
ATG said:
The technology is nowhere near the point of even having a trial system that could theoretically be scaled up to do something useful. The usual 30 year timescale remains optimistic having already been demonstrably wrong for several decades. I haven't seen anything that looks like a genuine breakthrough in this field since I was at school and started following this stuff in the 1980s. It remains exceedingly difficult to confine a plasma with magnetic fields, and you can only squish tiddly bits of matter even using super enormous lasers.
On plasma stability they are supposedly using AI to predict the instabilities and compensating. How effective this is in reality remains to be seen.The thing that kills most Magnetic confinement are the "Greenwald density limits". Which forces the reactors to get bigger and bigger to stand a chance of break even. Which makes them more expensive and less efficient.
NIF has achieve a Q-Fusion of 1.54, as steam turbines are 40% efficient typical, and lasers are typically 10-15% efficient, so they'd need a Q-Fusion of around 18 or higher before they would achieve Q-Total of 1.
Then they have to solve the cycling problem as NIF can currently fire about once per day.
ATG said:
The technology is nowhere near the point of even having a trial system that could theoretically be scaled up to do something useful. The usual 30 year timescale remains optimistic having already been demonstrably wrong for several decades. I haven't seen anything that looks like a genuine breakthrough in this field since I was at school and started following this stuff in the 1980s. It remains exceedingly difficult to confine a plasma with magnetic fields, and you can only squish tiddly bits of matter even using super enormous lasers.
Yes but you keep forgetting, AI.It’s not worth £ trillions for nothing all of a sudden
Genuinely though, it’s not an existential issue so it gets chicken feed funding.
If some country genuinely wanted fusion in 25 years they’d likely have it.
But they don’t, there is more money in burning stuff to make steamy water, or making fans to blow in the wind.
I agree it’s depressing being mid 40s yo and the same type of stories cropping up that you were excited about 30 years ago in school.
It’s sad to think what I experienced was the dying light of the technological and positivity boom of the 60s through 90s, and since then we’ve just advanced profit making and optimising what we have.
Ie, Concorde has become 737 Max…
Fusion has become a load of wind turbines and burning gas.
While everyone is terrified of spending actual money on a new idea then all we get is rehashed old stuff.
Caruso said:
An interesting article which claims "It is time to drop the old joke that fusion is 30 years away, and always will be. A poll at the International Atomic Energy Agency’s forum in London found that 65pc of insiders think fusion will generate electricity for the grid at viable cost by 2035, and 90pc by 2040."
https://www.yahoo.com/news/nuclear-fusion-grid-com...
Most people would say that if it is their job. https://www.yahoo.com/news/nuclear-fusion-grid-com...
Sorry to pick your post apart, but you touch on a few of the issues that I was thinking about....
Sadly I am currently of the view that actual power-to-grid is further away now than it was promised getting on for 1/2 a century ago But I would be very happy to be proved wrong!
Mr Whippy said:
Genuinely though, it’s not an existential issue so it gets chicken feed funding.
If some country genuinely wanted fusion in 25 years they’d likely have it.
But they don’t, there is more money in burning stuff to make steamy water, or making fans to blow in the wind.
To be fair to fusion failure to progress: the view for most of my life has been that big tokomaks was the way to go, but they are getting bigger and bigger and still no where near returning power, just consuming more and more. I remember JET being built - actually had a school trip to see the building work.If some country genuinely wanted fusion in 25 years they’d likely have it.
But they don’t, there is more money in burning stuff to make steamy water, or making fans to blow in the wind.
Mr Whippy said:
I agree it’s depressing being mid 40s60 yo and the same type of stories cropping up that you were excited about 3040+ years ago in school.
It’s sad to think what I experienced was the dying light of the technological and positivity boom of the 60s through 90s, and since then we’ve just advanced profit making and optimising what we have.
Ie, Concorde has become 737 Max…
Fusion has become a load of wind turbines and burning gas.
... but other renewables such as wave: which the UK should be rather good at given our coast and tidal range in some estuaries, have been woefully neglected.It’s sad to think what I experienced was the dying light of the technological and positivity boom of the 60s through 90s, and since then we’ve just advanced profit making and optimising what we have.
Ie, Concorde has become 737 Max…
Fusion has become a load of wind turbines and burning gas.
Mr Whippy said:
While everyone is terrified of spending actual money on a new idea then all we get is rehashed old stuff.
there does currently seem to be a lot of actual money being spent on fusion; the smaller and weirder devices seem to be getting backing and maybe making progress. I've seen glimpses of them at times as many are in my part of Oxfordshire and are waving wads of cash at magnet engineers and the like that were otherwise occupied with our particle accelerators!Sadly I am currently of the view that actual power-to-grid is further away now than it was promised getting on for 1/2 a century ago But I would be very happy to be proved wrong!
Caruso said:
hidetheelephants said:
If power independent of weather is needed fission works fine, no need to spend an indeterminate and large sum on a science project.
Even when fission works fine (which is only most of the time) it leaves behind some pretty expensive long term problems. Gassing Station | Science! | Top of Page | What's New | My Stuff