We are probably less than 15 years away from seeing a fusion power station begin to contribute to the grid.
Also, Tyler Cowen points to an MIT news article.
Developing the new magnet is seen as the greatest technological hurdle to making that happen; its successful operation now opens the door to demonstrating fusion in a lab on Earth, which has been pursued for decades with limited progress. With the magnet technology now successfully demonstrated, the MIT-CFS collaboration is on track to build the world’s first fusion device that can create and confine a plasma that produces more energy than it consumes. That demonstration device, called SPARC, is targeted for completion in 2025.
In the informal Zoom meetup I had with readers about ten days ago, I asked what they were optimistic about. One of them mentioned that machine learning was helping to solve problems in engineering that were nearly impossible to solve without it. One example he gave was fusion power. Fusion power is not practical now, because the amount of energy needed to generate fusion power currently exceeds the power that can be usefully extracted from it. But scientists and engineers are gradually improving the ratio of energy output to energy input.
That would be great, but people have been saying that for at least 40 years. I’ll believe it when I see it.
+1
“We are probably less than 15 years away from seeing [insert some rosy tech prediction here].”
I’ve been promised that a lot of really cool stuff was right around the corner for at least the last 20 years. Basically none of it has come to fruition.
This time is different?
Piling on: “Economical fusion power is twenty years away … and always will be.” I think I first heard that more than 20 years ago.
The basic ideas of magnetic or inertial (laser) confinement have been around for over half a century, and the huge technical problems with both approaches have been known for just as long.
If we could make the magnets or lasers both strong and efficient enough, then, theoretically (with some caveats), one can say with reasonable confidence that practical fusion reactors are doable. But if we can’t, then they probably aren’t.
So, a lot of the “always 20 years in the future” predictions are wild guesses about what we might be able to do with magnets and lasers. People knew we were *way* off from what was needed, but they didn’t know if it was even possible to get to where we needed to be, or how long that would take if it was possible.
Well, it’s been slow going, slower than people expected for a long time. The latest results are indeed really impressive, but at the most optimistic, they are perhaps still just on the very edge of working at all at the absolute most lenient threshold for ‘working’.
My impression is that most likely we’ll need at least another doubling in strength and efficiency to make it, and, assuming we can even maintain current rates of progress, that is still really far away. “15 years until contributing to the grid” is just nuts, and I’ll put lots of my money where my mouth is: Ridley can come and have a go if he thinks he’s hard enough.
At this point, I no longer believe the predictions enough to care.
If you’ve got 40+ years of life left, I’d say the odds are over 50% some functioning fusion reactor.
No positive contribution to any grid in the next 20 years (90% est.).
But a much better magnet is a nice breakthru.
We’ve had fusion reactors on a small scale since Philo Farnsworth invented one in the 1960s. Their main use has been neutron generation; even small amounts of fusion produce huge amounts of neutrons. That’s the operating principle of the neutron bomb, a weapon designed to kill the population of a city while leaving the buildings intact.
Neutrons have two really annoying properties. First, they are absorbed by most atomic nuclei, which often produces radioactive isotopes. This means that fusion isn’t cleaner than fission power in practice. Second, they are neutral so they cannot be contained by magnets.
For these reasons, I’m highly doubtful that we’ll see fusion power in the relevant timeframe (roughly, “in time to save us”).
Oliphant was generating neutrons, tritium, and alpha particles in Cavendish’s lab via accelerated deuterium ion fusion way before that – *ninety* years ago. And if you don’t have a fission reactor handy, accelerators are still mostly how it’s done for practical and research applications. There is no difference in principle between what he did and the postage-stamp-sized ‘neutristor’.
Fusors never really amounted to anything. No basis for a reactor, for sure. But the fact you can build one cheap on your tabletop as a hobbyist (with some paperwork) is undeniably cool.
Even for neutron generation, the best ever fusor (about 100x more than normal) got up to about 10 million usable flux (per cm2 going through a sub-steradian aperture), but reactors make 100 million times more, and the best accelerators 10 billion times more.
It’s basically just lucky for us that Mother Nature is the best non-proliferation enforcer, and these fluxes and efficiency of capture are low enough that it would take 50 years to use accelerators to turn cheap depleted Uranium into enough Plutonium 239 to make a bomb, and that you still need to enrich Uranium (hard and expensive) and burn it in actual fission reactors to do it. If things were a few orders of magnitude different, then our lives, if we still had them, would be a lot different.
If only Mother Nature was as good at counterproliferation where bioweapons are concerned.
Will fusion ever be cheaper than solar and batteries? Will it be cheaper than fission power plants?
Very precisely controlled exceedingly strong superconducting magnets don’t sound like the kind of thing that can be massed manufactured very easily.
Calling that thing from MIT a news article is a misnomer. It’s a press release for the MIT/CFS team. It’s just how the game is played: you have to put out these puff pieces to keep the funding pipeline going both on the academic side of things (it would be interesting to know the nature of the relationship between MIT and CFS) and from investors. Nothing it says is completely untrue, but it’s not news. A reporter would hardly fail to mention, for instance, that achieving break-even is not such a novelty. The JET tokamak (Culham, UK) came close to break-even in its last deuterium-tritium (D-T) campaign, and the JT-60 tokamak (Japan) would have exceeded break-even if it had been licensed to work with tritium, both at the end of the 1990s. A reporter would also have to mention that CFS grows (probably thanks to Bayh-Doyle Act) straight out of MIT’s research tokamak Alcator C-Mod, which was upgraded many times before being finally discontinued in 2016. It has always emphasized large magnetic fields and, despite its small size, holds the current record for plasma pressure in a magnetic confinement fusion device.
Ridley’s article, though, is starry-eyed bullshit. He obviously does not know what he’s talking about. There are fundamental physics constraints on fusion that prevent miniaturization of viable reactors to “shipping container dimensions”, at least with any reasonably foreseeable magnet technology, which determines achievable plasma pressure (instabilities appear in magnetic confinement systems when plasma pressure becomes large enough in relation to magnetic field pressure). These constraints are the cross-sections of the nuclear reactions involved.
For energy production, the only fusion reaction that’s going to be practicable for a long time is D-T. Its cross-section exceeds D-D and D-³He by many orders of magnitude until ion temperature reaches 300-500keV, 50-100 times larger than today’s records. Other fusion reactions touted by some of the many recent fusion energy startups the MIT press release mentions, such as p-¹¹B, are even worse in this respect. (A real reporter would try to dig up who are the investors in CFR, which at least makes sense scientifically, and in the many other fusion startups that are, to put it bluntly, fraudulent. This is especially relevant given that the Theranos trial is starting.) Tritium does not occur naturally in meaningful quantities and has to be bred on site by irradiating lithium with neutrons from the fusion reaction. The energy spectrum of fusion neutrons from D-T reaction and the cross-sections of these breeding reactions dictate that the breeding blanket, which needs to wrap around the plasma chamber in order to capture the neutrons, cannot be made thinner than 3 feet and retain good efficiency. This puts limits on the magnetic and vacuum systems.
Neutron fluxes themselves present an important and as yet unsolved materials problem. Even in DEMO, a conventional tokamak design, they are many times higher than in fission reactors, and fluxes increase quadratically as one tries to make a fusion reactor with a given power smaller. High-temperature superconductors are extremely unlikely to tolerate such fluxes, so the magnets must be put outside the breeding blanket to shield them properly. It is also very likely that reactor components subject to these fluxes will have to be replaced regularly over the reactor’s operating period.
All that said, there are no fundamental physics constraints to building a viable fusion reactor and a power station based on it. At this point the semi-empirical scaling laws are sufficiently well-understood that engineers know what size reactor will work. Since the discovery of high-confinement modes in mid-80s it has been known that building a practical tokamak reactor is possible, and research has concentrated on ironing out the myriad details. The problem with fusion reactors is their extremely complicated design and correspondingly high capital costs, much higher than for fission reactors. I am not familiar with any recent estimates of capital costs, but you can form a rough idea if you consider a fusion power plant as a fission power plant that must (a) put the whole active zone into a vacuum vessel of about the size of an RBMK reactor, (b) have a superconducting magnet of around 10,000 tons in the immediate vicinity of the active zone, with associated cryogenic plumbing, heat shielding etc., that must be available for maintenance and modular replacement together with the vacuum vessel due to neutron damage, (c) reprocess tritium from the lithium blanket in-place and (d) manage to put the steam-generating plumbing somewhere so that balance-of-plant (turbines, condensers etc.) have steam to work with. All this makes for a lot of additional complications compared to a fission reactor, so I don’t see how a fusion reactor is going to be cheaper or more economical.
Candide III sounds like he knows what he’s on about. I don’t, except an anecdote, which Matt Ridley also seems well aware of.
At Cambridge Uni, 1986, lectures in electrical engineering, or power production, or similar.
Lecturer:
“You’ve probably heard that nuclear fusion is 30 years away. I was a student in the 1950s, and I was told ‘nuclear fusion is 30 years away’. I can confidently tell you it is way more than 30 years from now.”
His explanation was, if you say “if you say it’s 50 years away, no one will fund it, if you say it’s 10-20 years away, people will ask for some kind of evidence, like proof of concepts. 30 years is the magic number..”
Well, his name has passed into obscurity but 2016 came and went without any nuclear fusion so I like his explanation of how far off nuclear fusion is.
And so, nuclear fission seems like a good idea. We made it work, actually very easy to make it work, then we made it super expensive because lots of people were frightened of it.
So now it’s too expensive in the West which is a miracle.