Elusive fusion reactors to be commercialised by 2025-2030... Or so they say

[mrk]

I know we've been reading about this for years and years but could the advent of machine learning and AI finally bring us into new energy territory?

https://thenextweb.com/neural/2020/...lve-nuclear-fusion-thanks-to-cutting-edge-ai/

Assuming this is all going as forecast and they do finish a commercial unit by 2025 thanks to AI. The implication of near limitless energy are profound. We won't be seeing cars and transport using it for a long time after because of the need to downsize a fusion reactor, but powering infrastructure would be a huge thing. Considerably cheaper electricity being the by-product for us and manufacturing plants having less power overheads means cheaper products all round as well. There's also a huge benefit in environmental terms since fusion has no radioactive waste and the exact moment the conditions break for fusion to happen, the reaction ends so it's also very safe.

Seems SciFi was wrong, machines/AI won't wipe us out after all.

Although it's worth noting that the CEO of TAE back in 2019 said they expect to start commercialisation by 2023, and now it's 2025.... So maybe this new lease of life thanks to AI requires more funding, hence the media coverage... Who knows

Would be nice though to live the rest of our lives without having to worry about energy costs.

For those still not sure wtf all this is, here's a very recent and easy to understand video covering everything to date.

Looking at the timeline of progress, it does seem like our generation is the one that should finally see the real world use of fusion power and all thanks to AI and quantum computing especially after ITER goes live working full pelt in the 2030s.

Hopefully we are out of lockdown by then

 

[Diddums]

30 years m8. 30 years.

 

[dowie]

We won't be seeing cars and transport using it for a long time after because of the need to downsize a fusion reactor

Eh???

I mean fission reactors have been around for a while now and you don't get people saying we won't be seeing them in cars for a long time... why would anyone expect to see a nuclear reactor in a car in the first place? They're in some navy ships and that's about it...

As for "thanks to AI" - sounds more like that firm got help with inverse problems/optimisation...

What did quantum computing have to do with any of this?

It would be great if there was a commercial fusion reactor created within the next few years - people will get skeptical about this stuff though as claims about fusion being around the corner have been made for a while now.

 

[Zefan]

There's a lot of interesting stuff going on in this space. There are several ways to skin the proverbial cat that is fusion, and - a continuing theme across technology it seems - the private sector has really advanced to a point where they are now leading the charge with a few key technologies that are key to making each implementation viable. I specifically really like Tokamak Energy's concept, which is to create comparatively small reactors which are easier to produce, manage and use in a modular fashion, as opposed to a gigantic hulk of a single unit which has many single points of failure. The success of a small tokamak would mean it's one step closer to being used in transport (obviously beginning with fringe applications first such as military subs, ships, and potentially space).

 

[mrk]

Eh???

I mean fission reactors have been around for a while now and you don't get people saying we won't be seeing them in cars for a long time... why would anyone expect to see a nuclear reactor in a car in the first place? They're in some navy ships and that's about it...

As for "thanks to AI" - sounds more like that firm got help with inverse problems/optimisation...

What did quantum computing have to do with any of this?

It would be great if there was a commercial fusion reactor created within the next few years - people will get skeptical about this stuff though as claims about fusion being around the corner have been made for a while now.

You are confusing fission with fusion in the first bit of that comment I think, nobody in their right mind would pub a fission reactor (even if it were remotely possible) into public transport lol.

There are companies out there working on portable fusion reactors that could be used for transport or deployment in areas without fixed infrastructure.

Quantum computing as per the article and a few others, Google using their computers to analyse fusion iterations and AI being able to spot more efficient ways to analyse in the process and project better ways to do the same tasks. As mentioned in the video linked, AI has helped re-look at older methods with modern technology because even a few years ago today's AI wasn't there to analyse and predict the massive amounts of data these experiments produce in an efficient way.

 

[dowie]

You are confusing fission with fusion in the first bit of that comment I think, nobody in their right mind would pub a fission reactor (even if it were remotely possible) into public transport lol.

There are companies out there working on portable fusion reactors that could be used for transport or deployment in areas without fixed infrastructure.

I'm not confusing them, we don't even have a working fusion reactor yet - I suspect using them for transport is going to be a way off at the moment.

Quantum computing as per the article and a few others, Google using their computers to analyse fusion iterations and AI being able to spot more efficient ways to analyse in the process and project better ways to do the same tasks.

"AI being able to spot more efficient ways to analyse in the process" sorry but that is vague and doesn't sound like what they're reportedly doing at all.

The mention of "AI" seems to set imaginations off - they've had some assistance with optimisation & inverse problems basically... I guess it sounds sexier if a journalist just reports that they're using "AI" and people can then come up with their own narrative re: how... but it isn't necessarily grounded in reality.

As for quantum computers - where do they come into this? Which article covers that aspect?

 

[mrk]

I thought that's what Googles hand was playing in this since a lot of what they've been helping out with in various fields has been through using their quantum computing power to churn through data. Either way, quantum computing in this area isn't brand new and has been talked about for a short while at least since the benefits are said to be compelling.

There's a lot of interesting stuff going on in this space. There are several ways to skin the proverbial cat that is fusion, and - a continuing theme across technology it seems - the private sector has really advanced to a point where they are now leading the charge with a few key technologies that are key to making each implementation viable. I specifically really like Tokamak Energy's concept, which is to create comparatively small reactors which are easier to produce, manage and use in a modular fashion, as opposed to a gigantic hulk of a single unit which has many single points of failure. The success of a small tokamak would mean it's one step closer to being used in transport (obviously beginning with fringe applications first such as military subs, ships, and potentially space).

Yeah there are several reactor designs by private firms that seem to be itching to get to the end first, along with ITER which brings together the collective minds of the world for the joint project. That's still being built in France but with a Qof 10 that's gonna provide a lot of valuable research data to speed things up further for sure.

 

[dowie]

It's more like thousands of GPUs running a bunch of Markov chains in parallel... fortunately google has thousands of GPUs - Still pretty damn cool though.

 

[Angilion]

I'm not sold on the idea that a whole slew of scientists working together with major transnational funding somehow forgot to use modern computing as part of their research. I suppose it could happen if they got serious tunnel vision on the project, but I don't think so it's likely.

A little niggle with that video - fission does occur in nature. Probably extremely rarely, but it has been shown to have occured on Earth. Not recently, thankfully, but a natural nuclear fission reactor was proven to have been running on Earth long ago...a quick look online gets the name - Oklo.

Only a little niggle, though. It's a good overview. It also states that companies have been making claims like the ones referred to in the OP for quite some while now (probably talking it up to procure investment money) and they've done nothing other than cement the image of commercial fusion as something that it always 30 years away.

A niggle with that article - it states that scientists think that a working nuclear fusion reactor might happen by 2025, then it states that most scientists don't think that a commercial nuclear fusion reactor will exist before 2030. The first claim is a bit odd since working nuclear fusion reactors have existed since the 1960s, which is obviously before 2025.

The video is much better than the article.

If there's a demonstration of a working fusion reactor with q>1, that's when to get excited about the possibility of commercial fusion soon afterwards. But only the possibility. Many things work in carefully controlled conditions in a research facility but don't become practical outside them for a decade. Or sometimes never.

While fusion doesn't necessarily generate radioactive waste, in practice it does indirectly generate radioactive waste by gradually making the interior wall radioactive. It's one of the things that researchers have been working on. It's not much radioactive waste, but it's not none.

I'm going to be very optimistic and say that with research building on research we're probably finally at the point at which commercial nuclear fusion might be 20 years away rather than always 30 more years away.

 

[Fusion]

Don't rate the initial article, the author really doesn't sound like he understands nuclear physics at all. I'm with Angilion- the biggest leaps are going to come from the ITER project. I would love to think that smaller private companies are able to offer new insights however- I'm all for anything that would potentially speed up the introduction of this potential game changer.

 

[dowie]

I'm not sold on the idea that a whole slew of scientists working together with major transnational funding somehow forgot to use modern computing as part of their research. I suppose it could happen if they got serious tunnel vision on the project, but I don't think so it's likely.

What's that in reference to? Who forgot to use modern computing?

 

[Angilion]

What's that in reference to? Who forgot to use modern computing?

According to the article, one company has leapt years, even decades, ahead of everyone else because they've used modern computing. I'm not sold on the idea that everyone else forgot to do so, especially not a big team of researchers backed and funded quite well by pretty much every wealthy country in the world. The team currently working on ITER.

 

[dowie]

According to the article, one company has leapt years, even decades, ahead of everyone else because they've used modern computing. I'm not sold on the idea that everyone else forgot to do so, especially not a big team of researchers backed and funded quite well by pretty much every wealthy country in the world. The team currently working on ITER.

Yeah I don't think much of the articles tbh... They've had some assistance from a team at google and access to their advice/resources to essentially do some bayesian inference at scale across multiple GPUs... but I'd assume that other well funded teams have access to plenty of GPUs etc.. too.

Article translates this into a bunch of hype - OMG super duper special AI is being used by one firm...

 

[Nasher]

I met a nuclear physicist last year who had some small involvement in fusion research. He didn't expect it to be ready this century.

 

[Fusion]

It's all very well them talking about AI and data analysis, but someone's still got to physically build a reactor. This is a colossal undertaking (again, see ITER), and something that is highly unlikely anyone would be able to complete by 2025. There will be many aspects of the reactor design and engineering that the ITER team will still be working on/haven't perfected yet. One thing is just the sheer scale of everything. Magnets, transformers. To heat the plasma inside the tokamak, you need a current on the order of around a million Ampere. Not exactly something you can do with parts from Screwfix and B&Q!

 

[Fusion]

I met a nuclear physicist last year who had some small involvement in fusion research. He didn't expect it to be ready this century.

It's my belief that we could absolutely be in full commercial operation this century. We just need to pull our finger out and ensure projects are properly funded. When you think of the potential upside of the technology, it's crazy that we don't spend more on it.

 

[Nasher]

It's my belief that we could absolutely be in full commercial operation this century. We just need to pull our finger out and ensure projects are properly funded. When you think of the potential upside of the technology, it's crazy that we don't spend more on it.

A lot is already being spent on it, but there is only so far it can go until the technology (and experts) is ready to support it.

 

[Fusion]

A lot is already being spent on it, but there is only so far it can go until the technology (and experts) is ready to support it.

Technology gets better by throwing resources at its development. Technology can develop extremely quickly when the impetus is there (look at the Manhattan project in WW2).

 

[mrk]

Look at projects like LIGO, professor Weiss had been thinking about how to test Einstein's gravitational waves since the 1970s even though Einstein himself said we'd never be able to detect them at such a small scale yet the project came to be through funding in technology which lead to his team detecting the waves and earning him the Nobel prize (Noble, if you're Trump ). The detections didn't happen until only a couple of years ago and like Fusion, it had spent decades as a theory until the funding (and so technology) was applied.

Whilst the more mass media articles will always likely be a means to add more investment interest to these projects, the more deeper you read each year on what's been progressing, it seems the closer we are getting and this generation could certainly be the one that sees it start.

Wouldn't it be amazing? We will have lived through the start of several global game changers within our generation.

 

[Fusion]

Look at projects like LIGO, professor Weiss had been thinking about how to test Einstein's gravitational waves since the 1970s even though Einstein himself said we'd never be able to detect them at such a small scale yet the project came to be through funding in technology which lead to his team detecting the waves and earning him the Nobel prize (Noble, if you're Trump ). The detections didn't happen until only a couple of years ago and like Fusion, it had spent decades as a theory until the funding (and so technology) was applied.

Whilst the more mass media articles will always likely be a means to add more investment interest to these projects, the more deeper you read each year on what's been progressing, it seems the closer we are getting and this generation could certainly be the one that sees it start.

Wouldn't it be amazing? We will have lived through the start of several global game changers within our generation.

I see what you're getting at, although Fusion has been proven in practice a long time ago. It was first performed on earth in 1950 as part of Hydrogen/thermonuclear bomb development (it didn't have the best yield- that came in 1952 with the "Ivy Mike" test). Problem is of course, bombs are uncontrolled. Controlled Fusion has been performed, but so far we've not got more energy out than we put in. ITER will hopefully change that.

We've actually had prototype fusion reactors since the 1950s, on British soil (look up "Zeta")!