Nuclear fusion on brink of being realised, say MIT scientists | Hacker News Reader

Nuclear fusion on brink of being realised, say MIT scientists

theguardian.com

44 pointslnguyen78 comments

https://encrypted-tbn2.gstatic.com/images?q=tbn:ANd9GcQV2HAS...

I enjoy the joke about the perpetually shifting fusion acquisition time frame, but it would be fair to add that in terms of actual funding, we have been under the "fusion never" level. I feel like the context for the saying has always been that the proper financing assumption holds.

That said, the material sciences advances required, especially high temperature superconducting magnets, simply were not there in the 70s. So a moonshot project to achieve it may have been a failure at massive costs, discouraging humanity from fusion research for a long time.

First time I heard that fusion was 15 years away was around 35 years ago, in an already old book belonging to my eldest brother. They were talking about ZETA (late 50s).

Wake me when the first power station is being built.

To quote my Physics of Electrical Power Generation lecturer from a few years ago: "Fusion is 15 years away, I think it will always be 15 years away."

It's a good quote, but you should know you're really quoting the zeigeist there, as was your professor.

Article talks about YBCO as if it was new, but it’s almost as old as I am. If it was the only thing holding back fusion, the reactors would’ve had software updates for the Millennium Bug.

Perhaps this time I won’t fall for the Gell-Mann Amnesia effect…

They've found a way to wind it into tapes which make fabricating magnates relatively easy - they have demonstrated world record magnetic capability with these. Fusion is one of the applications but the thing I liked about the research is that there are intermediate technology goals that they are aiming to hit which should fund everything - notably compact medical scanners.

YBCO is not new, but the production of it in the amounts and quality necessary for large superconducting magnets is still difficult. You need to produce long stretches of the material to create large magnets efficiently, and that isn't trivial.

The old superconductors typically used in this kind of magnets have been pushed to the limit already in Nuclear Magnetic Resonance, ~23 Tesla is the strongest existing commercial one and this seems to be a limit for that technology. The first NMR magnets that use YBCO are expected in the next years as far as I remember, I don't think any exist yet.

Now that is interesting. I had thought YBCO was easy to fabricate given I saw some amateur do it on YouTube, but I do accept they had quality control issues — important on large scale production.

The way I’m reading the article they’re talking about a new superconductor where YBCO is one of the materials used. English isn’t my first language though so I could be off base here.

I'm no expert, but 15 years seem like an awfully small timeframe to produce a commercial reactor with a still unproven technology.

The argument is that they can make it small - room sized, with the new magnates. This reduces the every one of the subsidiary technical challenges to doable, basically you don't need big machines to dig big holes, you don't need 30000m^3 of concrete here, there and everywhere, you don't need a supply chain that involves 20000 people in 40 countries... so the co-ordination costs go away as well.

It's pretty like the idea with data analytics or software ; if you can do an iteration / update by yourself in 10 minutes you will be able to solve the problem (for reasonable problems) in a few weeks for sure. If you need to spend a month assembling the data and compiling the code before you can commit a change you are simply not going to ever be able to make progress.

I think the "for reasonable problems" is the question for fusion.

I have no idea of the value of my impression of the work - I saw the MIT PI present on this a year ago and have read up since, but well - that's it. However, I think that they think that the main challenges are solved as per ITER and that they are on a mission to resolve the engineering; and this is the way to do that. They don't think that there is anything left which is a show stopper apart from the need to trouser $50m++ to build the proof of concept.

I figure they could probably build a reactor that produces power - the MIT design looks like it would work if built. Making it profitable without government subsidy is a whole different thing though which I doubt will happen in 15 years.

There are a lot of problems to deal with eg see "Fusion reactors: Not what they’re cracked up to be" https://thebulletin.org/fusion-reactors-not-what-they%E2%80%... There's also a Reddit discussion of that saying the problems aren't quite so bad https://www.reddit.com/r/fusion/comments/67rqqg/fusion_react...

If I were a politician I'd vote for subsidising the thing by say buying the electricity at 30c/kWh to advance human understanding, so I think it could be built. I think MIT estimated $5bn + to build a 200MW generator which is not a huge amount in government terms.

It took sixteen years from our realization that fission based power was possible to the first plant running. Seems possible.

The trick is not to be too precise about when you start counting from

If we were to work with an assumption that fusion happens soon (<20 years) and power is now cheap, what would we now be able to achieve that was previously prohibitively difficult / expensive?

To name a couple, you enable mass desalination and cracking of water. Solve humanity's access to clean water, enable replacement of fossil fuels for vehicles, solve anthropogenic climate change. Just those two are huge.

Isn't it the promise of renewables already? Prices dropping by double digits every year, exponential growth in every market.

Coupled with emerging energy storage it seems like cheap energy is due to come long before fusion

Renewables are expensive to set up and scale (you're gathering smallish amounts of energy from large areas!), but cheap to pull power out of.

Fusion could potentially be extremely compact, extremely energetic, and not _that_ expensive to set up relative to how much power you get out of it. Renewables could make energy _cheap_, but fusion could make it practically _free_.

> Renewables are expensive to set up and scale

They are cheaper to set up at scale, but also work very well in a micro production / distributed environment (albeit slightly more expensive)

I understand your argument about space, you need a lot of space to harness energy, but we're now starting to see deep sea wind turbines, and Earth happen to have quite a bit of windy areas above deep sea.

Just the North sea itself could cover a massive portion of Europe's energy production (double digit)

Right, I'm absolutely not suggesting renewables aren't a great idea, nor that we shouldn't do them, nor that they aren't vastly superior to fossil fuels along basically every access except space (which, as you say, isn't as much of a problem as it might sound)

Just, fusion could be better _still_, if it gets off the ground. That's not a reason to not focus on what we _can _do now, though.

Fair enough!

Renewable energy doesn't have the same amount of energy density.

Fission already beats renewable by a good margin and Fusion puts some on top of that.

The fuel for fusion is incredibly cheap; hydrogen and helium, the most common elements in the universe. Easily obtained from solar wind if we find the amount on earth is lacking.

Additionally, Fusion does not produce any dangerous waste. Any byproducts that are radioactive decay quickly.

Fusion is also a lot safer than Fission since the reaction cannot exist outside reactor conditions while a uranium rod will happily go prompt critical if it gets the chance.

Fusion is cheap energy.

>Fission already beats renewable by a good margin and Fusion puts some on top of that.

This is not true, I don't think people realise how much the renewable prices are currently dropping: https://www.theguardian.com/environment/nils-pratley-on-fina...

I can see the benefit of fusion though I'm not very convinced by the density argument (we do have space to put solar on our roofs, wind on our shores, batteries in our basements and distribution points, and plenty of spots to pump geothermal)

Uranium produces 80,620,000 MJ/kg [Wikipedia], Deuterium produces about 10% more.

I'm not seeing any Solar Panels or Wind Turbines on the market that can output 80 TJ per Kilogram of Weight over their lifetime.

The article you linked talks about Energy Cost, which is another issue. Though considering Hydrogen and Helium are the most abundant resources in the universe, I'm not holding my breath for any renewable resources consistently beating that.

Cost =/= Density, otherwise we'd be powering electric cars with AAA batteries.

>(we do have space to put solar on our roofs, wind on our shores, batteries in our basements and distribution points, and plenty of spots to pump geothermal)

Yes but the point is that fusion requires a lot less space for a lot more power, doesn't depend on the weather or surrounding geology and is 100% clean.

Almost every application cares more about energy cost than energy density. I think only shipping (ocean, air, and space) care about energy density in any significant way, and the first two can fairly trivially use renewable-electrolysed hydrogen. (Space can also use it, but I wouldn’t say it is trivial because nothing about current space access is trivial).

I'm confident that given time Fusion can outpace renewables in energy cost too.

Uranium is somewhat difficult to obtain; you have to mine it from deep under (using what amounts to slave labor), clean the ore and then carefully and thoroughly enrich it (238 is almost unusuable in reactors, you need 235, which is only present in a small fraction of the ore).

On the other hand, getting hydrogen can be achieved with electrolysis (electrolysis requires less energy than the fusion reaction produces), various other chemical processes or simply solar wind.

Hydrogen is the most common material in the universe, it will and can outpace any other energy source.

Renewables are simply different methods for capturing energy from the fusion happening in the sun with some efficiency.

which renewables are truly clean? Wind turbines require so much material and energy to manufacture that they will compensate this only over their whole lifetime. No clue about solar panels.

China is building tons of new coal plants to cover its current needs. It doesn't seem that renewables are that great, when not subsidized (which is the only true measure of effectivity)

I think your numbers may be off for wind turbines. Random googling gives

>for an offshore wind turbine 0.57 years of expected average energy production are necessary to recover all the energy consumed for manufacturing, operation, transport, dismantling and disposal https://www.wind-energy-the-facts.org/energy-balance-analysi...

>China is building tons of new coal plants to cover its current needs

Not so sure about this one: https://www.theguardian.com/environment/2017/mar/22/coal-pow...

Sure they are still building but at a rate that has been drastically shaken up by both the price of renewables and their own environmental policies.

Now regarding the environmental impact of producing solar panels and wind turbines, they clearly are carbon negatives

http://www.drawdown.org/solutions/electricity-generation/roo...

http://www.drawdown.org/solutions/electricity-generation/win...

Though I'm sure they can't have no-impact, you will get some other form of pollution, but at the moment what matters most if reducing our greenhouse effect. It's cheap, it's already there so there's no reason to delay it I think.

https://www.theguardian.com/environment/2012/feb/29/turbines...

If energy is basically free, how will it impact crypto prices?

I believe the proof-of-work calculations automatically become more difficult as more computing power is dedicated to solving them. The net result is that the amount of energy consumed will probably increase inversely proportionally to any reduction in power prices.

Will the concept of money still mater as much if at all if all energy is free?

Let me rephrase - “will anything be scarce if electrity is free?”

The answer to that is obvious: yes. So why wouldn’t we need money?

Agreed, hence "as much". My thoughts were that the main cost of 'stuff' is the energy used in mining or preparing it. If you have free (cheaper) electricity, you don't need oil, metal production has it's biggest cost removed, etc.

Variable costs are eliminated so the amortized fixed cost of hardware is the only ingredient for price.

Basically the price (demand) for hardware goes up and should equilibriate when it reaches the previous cost of energy.

What will be realized first: affordable graphene production, nuclear fusion, or general artificial intelligence?

We can make graphene very cheaply and easily now, the issue is making large continuous high quality sheets of it. Same for carbon nanotubes.

GAI and fusion are different issues in that we have clear elementary understand of fusion (and a large chunk of the technology worked out). We have little to no understanding of cognition and much of the technology that we use for AI now may (or may not) prove to be irrelevant to GAI.

I'll go one step further - humans might sort their species out and stop with all the pointless tribalism and unite for the common good.

"stop with all the pointless tribalism "

I'll challenge that, because I think tribalism (living in small groups) is the way more akin to our genes. And much from the rising mental problems of humans today comes from living on your own in a big anonymous pool of strangers.

But those small groups could then indeed work (more) together for the common good.

If we are looking at what's in are basic genes as a marker on what we should be doing, I don't think we should be messing about with fusion. Or fission come to think of it. I agree loneliness and mental health are big problems today however don't think the answer is having an adversary to motivate yourself.

"If we are looking at what's in are basic genes as a marker on what we should be doing, I don't think we should be messing about with fusion. "

Why?

I didn't say we should live stone age style. And using technology is very akin to our genes, afaik. So I don't see any problems with that. What matters more than the tools are the social interactions I believe. And there we could do better.

I'm not sure if the Earth is large enough to distribute 8 billion humans into sufficiently separated small groups, and such a scheme isn't even on the political horizon.

I didn't say there need to be a big space separation, nor clearly divided groups, just that the general people you interact with are in general familiar and not randomly strangers who change all the time. So a more tribal way of living in the cities is possible, I believe.

But .. actually there is also enough space on earth left ... poplated areas are very dense compared to big open lands.

The funny thing is that you're announcing you're part of the tribe that doesn't like tribalism.

I'd say it was more of an opinion rather than a tribal belief or following. I guess it depends of the definition of tribalism.

In short, I'd like humans to just get on with each other and treat the world we share with other species with greater respect. If that puts me in a tribe, then so be it!

Or, human might sorth their species out and away from the list of Earth species, effectively rendering question of nuclear fusion, graphene production, AI, and social constructs irrelevant.

According to the 2014 arxiv.org paper on the ARC reactor

https://arxiv.org/pdf/1409.3540.pdf

the concept (which would produce 190MW(e)) uses 90 tonnes of beryllium.

According to the USGS, the current total world annual production of Be is 220 tonnes, and total global resource is estimated at 100,000 tonnes.

The world uses ~20 TW of primary energy, so roughly 100,000 ARC reactors would be needed to supply the world, using about 100x as much Be as the estimated amount available for mining.

Unless Be supplies can be drastically increased, this concept is at best a niche player.

For comparison, Olkiluoto 3 fission plant has been under construction for 13 years already. It's currently estimated to enter energy production a bit over one year from now.

The delays there have been the usual big project problems - big builds, corporate drama, quality control. none of those apply to the MIT plan (they did apply to ITER, which was one of the arguments against it, but at the time it was the best option because the small strong magnets didn't exist).

Fusion always 15 years from being realized. Always will be.

Sadly, that joke is at least 30 years old too.

There are lots of companies in the game now.

Tri Alpha Energy, General Fusion, Lockheed Martin and there are quite a few more.

There is the Wendelstein 7X stellarator and ITER.

The increase in fusion magnetic energy has actually been faster than Moore's law.

See this video at 18:30 https://www.youtube.com/watch?v=KkpqA8yG9T4&t=3990s

The massive improvements in superconductive magnets are really making a huge impact. Check out ARC and SPARC from MIT.

Humanity is going to get usable energy generated from fusion.

Sadly, that joke is at least 30 years old too.

When I was in high school, 45 years ago, I visited prospective colleges. Some grad students were discussing the fusion work they were doing, in a big building, with lots of shiny hardware. They probably told me that practical fusion was 15 years away.

Humanity is going to get usable energy generated from fusion.

Yeah I heard that in high school. I'm now retired and am still hearing it. Still, stranger things have happened. The Chicago Cubs won the World Series a few years ago!

Past performance is not indicative of future results.

Sure it is. It is the best indicator we use.

Certificates, diploma, driver license, etc. are all based on that.

Scientific research is inherently a field where progress could be modeled as punctuated equilibria. Just because there are long periods where little progress is made on a hard problem does not mean that it will never be solved, or that the solution is perpetually "15 years away". But it does mean that attempting to linearly extrapolate progress over a period of time is likely to be incorrect.

"where little progress is made on a hard problem does not mean that it will never be solved"

Never said that. But if fusion scientists say since 50+ years, that they are allmost there, then it is a pretty good indicator, that there probably also won't be a working fusion plant in the next 15 year's.

That would be true if we had no other data to go on, but we do: the fusion triple product, which has to exceed a threshold for net power. The JT-60 tokamak in Japan reached a triple product in 1999, using D-D fuel, that would have barely exceeded breakeven if they'd been using D-T. Fifty years ago we were nowhere near that close.

Ha.

Wonder when the first company that was targeting fusion was started.

It seems there weren't any until the 2000s sometime. Now there are lots. People with the money to sink 10s of millions into the idea think that it's close.

Danny Hillis will bet you that someone will do this by 2020:

http://longbets.org/605/

They have lots of skin in the game.

Hi. Fusion company founder from 1995 here (Fiat Lux Research, funded by DFJ). We were probably the first in the internet age, but didn't have a website (stealth mode engaged!). We definitely weren't the first. Companies which were earlier than us that I am aware of include Bogdan Maglich's Migma company (1974), Paul Koloc's Prometheus 2 (1976) and Bussard's EMC2 (1985). You could also include companies with divisions targeting fusion like General Atomics (1974) and ITT (1964).

Wow.

Thanks heaps for that. I had no idea.

What do you think of the new wave of fusion companies? Do you think they will get somewhere?

There is a subreddit for fusion. reddit/r/fusion.

Would you be interested in answering questions, doing a sort of AMA there?

I think this says more about the credulity of VCs and their general failure at anything approaching heavy metal technologies than anything else. Transatomic, a startup which is vastly less ambitious than any fusion startup (molten salt reactors are old tech) is a great example. Big oops on this one.

Tri-Alpha was, as far as I can tell, the most legitimate of the fusion companies. My pal Dave just retired from it; he worked on their data efforts. FWIIW my original comment was quoting him.

If we invested in space technology, the same money we invest into fusion, we'd be only 15 years from going to the moon too.

However there was the space race and lots of funds were allocated to going to the Moon.

But will someone investing in fusion make someone else money, esp. since the electricity is free and some of the resources aren't scarce?

We don't do a lot of things, at least in the modern US, because we have tie economic incentive to everything vs doing for society, even at a 'loss'.

It's the golden timeframe for world-shattering technology.

At least money is being raised and people are trying, it's easy to be a cynic about something that's been promised as around the corner for the last 50 years.

Fusion with net positive output would change everything for the world. Really hope they get over the line.

But there's been progress. The sun's been doing fusion for a while, we made fusion bombs in the 50s, the JET torus near Oxford produced fusion output at 16 MW from an input of 24 MW of heating in 1997, and one of the MIT's proposed designs is projected to produce 200MW net power output if built. (https://www.youtube.com/watch?v=KkpqA8yG9T4&feature=youtu.be...)

So it's coming along even if commercial viability may still be a while off.

I first heard about fusion maybe 30 years ago. At the time I was told that the cliche of fusion energy is that it's always just around the corner.

And this article seems to say, well it's.......

This prediction will always be wrong... until it won't.

By the way, "on the brink of" in the title is pretty stupid.

Ever wonder why that is?

I'd suggest the optimistic estimates are both for scientists and investors. Optimism is used as a fuel to keep scientists going when the future is uncertain, and any breakthrough, however small, is a chance to recharge that fuel. On the other side, (in our society) scientific research requires funding. It's easier to acquire funding if you suggest the research may take 20 years rather than suggesting it'll take 100 years. This also encourages scientists to be optimistic in their predictions, as if they aren't optimistic they risk losing their funding, which greatly reduces the chance of rapid progress.

For those reasons, I'd suggest these long estimates (as a definition of 'long', let's say in this case anything over 5 years) are just tools that help to get the work done, rather than being something with a high probability of being true.

Well, we're mostly software developers - I'm sure we all understand being an hour away from completing a project for the past several days of development.

Scale that up, and...

Sure, but there's a difference between setting short term targets and long term targets.

Setting short term targets is only realistic when you're doing something that either has been done before or where it's a variation on something that has been done before. Perhaps you underestimate the time it takes to do something by being unaware of the underlying complexity of a task (or by telling people what they want to hear, if they can't handle pessimistic estimates), but they still have a reasonable chance of being completed on time.

Long term targets work differently. If you want the time taken to be as short as possible, and you still give an estimate within a window of time which you have little to no hope to predict, you're basically saying "I don't know".

To put it another way, I could have a reasonable guess at what I'd be doing one week from now. On the other hand, I don't really know what I'll be doing 20 years from now. Any estimates I put within a range that is outside my ability to predict would basically boil down to me implying I haven't got a clue. With that said, I'd still put a timescale on an baseless estimate if it served a purpose. It's the purpose behind pulling the number out of thin air which is the point I was trying to highlight.

I hope we crack the concept of fusion soon, it would solve a lot of our problems (atleast most of the energy related ones, ie safety, pollution, availability of fuel)

isn't it meaned to always be 50 years away?

The 10s runtime is surprisingly close to a continuum runtime btw.

I'm sure it would be useful to have that technology around and I'm quite excited by the prospect of it, especially as the volume seems to be drastically reduced compared to ITER.

But I'm wondering how relevant fusion is to reverse climate change. It seems like renewables and storage are already fitting the bill, at a much lower cost.

Edit: happy to hear any arguments

Renewables are cheap when backed up by natural gas plants, but wouldn't be especially cheap if you built enough storage to run civilization on them.

I certainly think we should build renewables as fast as we can, because they're available now and a long way from the market share that requires a lot of expensive storage. But down the road, it's entirely possible that advanced nuclear power sources, whether fusion or advanced fission like MSRs, will be more economical than building enormous amounts of battery storage. (We do have cheap hydro storage but that's geographically limited.)

Reply on news.ycombinator.com