In the comments on my submission to the Astral Codex Ten book review contest, several people asked me about what I thought about several high-profile critics of fusion. This is a summary of my responses.
Originally Written: June 2022.
Prerequisites: Book Review of THE FUTURE OF FUSION ENERGY by Jason Parisi and Justin Ball (2019).
Confidence Level: Daniel Jassby is a retired plasma physicist, so this is a debate between people with similar levels of expertise. Sabine Hossenfelder and Elon Musk are both very intelligent people, but not experts in plasma physics.
Daniel Jassby
His Argument: https://inference-review.com/article/the-quest-for-fusion-energy
Requested By: Derek Jones, Charlie P, TheGodfatherBaritone, smopecakes, Matei Ionita, & Seta Sojiro.
My Response:
Jassby’s claims that magnetic confinement fusion (MCF) has stagnated for the last 25 years, while inertial confinement fusion (ICF) has progressed. I do not disagree: “There has been little progress towards a larger triple product since 2000.” I think that this is about to change, while Jassby does not.
There is a simple reason why MCF has not progressed. In 1997, the best MCF experiment in the world was JET. In 2022, the best MCF experiment in the world is JET.[1]JT-60 and several other tokamaks have similar performance.
There’s only so much we can do with JET’s size and magnetic field strength. They could probably get up to $Q = 1.5$ if they really tried, but not the $Q = 5$ that you need for a power plant. Instead, JET has been focusing on other important goals like getting the walls right and increasing how long they maintain the plasma. As soon as we get a better experiment, we will get better results, whether it’s SPARC or ITER.
It is true that computer simulations did not predict JET very well and underestimated the turbulent transport. Computational plasma physics is a lot better than it was in the 1990s, along with everything else involving computers.
Jassby makes a big distinction between beam-thermal versus thermonuclear fusion. I don’t know of anyone else who cares about that distinction. His footnotes for it all point to papers he’s authored by himself.
There isn’t a physical difference between beam ions and plasma ions. An ion typically has to collide a few hundred times before it undergoes a fusion reaction. So by the time fusion occurs, the ions from the beam are thermalized and are indistinguishable from the ions originally in the plasma.
He also seems to use the terms to describe how the plasma is heated. When the plasma is mostly heated by the beam, then Q must be small, and when the plasma is mostly heated by fusion, then Q must be large, … because that’s the definition of Q. The causality is backward. “For a purely beam-thermal system, the maximum theoretical Q is limited to less than 2” should be “When Q is less than 2, the system must be primarily heated by the beam.”
Tritium is very expensive. You don’t want your reactor to have to import it.
Each fusion reaction consumes one tritium and produces one neutron. Each breeding reaction consumes one neutron and produces one tritium. It looks like, in order to sustain this, you’d need perfect efficiency, which is impossible.
Which is why we’re also planning on including a neutron multiplier: beryllium.
$ {}^9_4 Be + {}^1_0 n \rightarrow 2 \ {}^4_2 He + 2 \ {}^1_0 n$
This increases the number of neutrons, which allows the system to sustain itself with less then perfect efficiency.
Jassby does not think that SPARC is a game changer: “improving cost-effectiveness is surely a distraction for MCF research”. I disagree. Cost effectiveness is important on its own. And it means that we can build more reactors faster. It won’t take us 25 years to get the money together to build the next big experiment.
I agree that inertial confinement fusion has made a lot of progress recently: “Progress has been extremely rapid. They crossed $Q=1$ a few months ago.”
The big challenge remaining for ICF is going from 1 shot per day to 1 shot per second. NIF is not trying to do this, because fusion isn’t their main goal. Maybe Marvel or someone else will figure it out. I hope that they do ! But I suspect that this will be harder than getting fusion using a tokamak.
Sabine Hossenfelder
Her Argument: http://backreaction.blogspot.com/2021/10/how-close-is-nuclear-fusion-power.html
Requested By: Derek Jones, Michael, Carl Pham, arbitrario, & smopecakes.
My Response:
Hossenfelder’s background is in high energy physics (i.e. particle physics and cosmology). Her understanding of plasma physics and fusion feels like it is the result of a week or two of research, not like someone who has engaged with the field for years. This is a completely legitimate way of doing popular science and I do it too sometimes.[2]For example, my Brief History of Schooling in America. I do think that you should express more uncertainty about your conclusions.
Hossenfelder does make a few technical mistakes. She describes several ways that a fusion reactor will lose energy, which decreases the efficiency, but doesn’t mention that the tritium breeding provides additional energy, which increases the efficiency. She also seems to act as though experiments and reactors are designed the same way and for the same purposes. ITER should get $Q=10$, but not engineering breakeven. If ITER took off a lot of its diagnostics and didn’t have four different ways of heating the plasma, then it probably could get engineering breakeven. But it wouldn’t be as good of an experiment. We want to get lots of data about what the plasma is doing and we want to try out multiple versions of different subsystems to see which one works best. But the technical mistakes aren’t the most important because she mostly focuses on the messaging.
Most of Hossenfelder’s videos, especially recently, are contrarian. There definitely is a place for contrarians in popular science. Most of her videos are about cosmology or high energy physics. These fields have lots of people who communicate them to the public: Brian Greene and Michio Kaku come to mind. Having one contrarian among many popular scientists is often a good thing. I don’t know if she realizes how few popular plasma physicists there are. Her Youtube channel is much larger than ITER’s (472K vs 63K subscribers), let alone Commonwealth’s (1.7K). For a lot of her audience, this is the most they’ve ever thought about fusion. I don’t like that the contrarian is the only voice a lot of people are hearing.
The core of her argument is about the definition of ‘energy gain’, or $Q$. She claims that plasma physicists are purposefully using misleading terminology to deceive the public.
In the technical literature, this quantity is normally not just called Q but more specifically Q-plasma. This is not the ratio of the entire energy that comes out of the fusion reactor over that which goes into the reactor, which we can call Q-total. If you want to build a power plant, and that’s what we’re after in the end, it’s the Q-total that matters, not the Q-plasma.
This is wrong. In the technical literature, Q is Q. Q-plasma and Q-total are terms that she invented in this video. I don’t know of a single piece of peer reviewed literature that uses her terminology. Her argument seems to be that she doesn’t like how Q is defined and wants plasma physicists to define Q in a different way. Disliking someone’s definitions is not enough to claim that an entire field is misinformation.
The evidence provided is 4 quotes about fusion that are misleading. The first two are by science journalists who have backgrounds in chemistry and astronomy, respectively. One of these starts with a Boris Johnson quote. I agree that science journalism often likes to exaggerate results, but I don’t this reflects badly on the scientists themselves. The third quote is an interview with Norbert Holtkamp, then ITER’s project construction leader. His comments do sound misleading. This quote is from an interview from 2006, which feels kind of cherrypicked. Currently, ITER’s website it not misleading: “ITER will be the first fusion device to produce net energy. [i] Net Energy: When the total power produced during a fusion power pulse surpasses the thermal power injected to heat the plasma.”
The fourth quote is the most potentially concerning. It’s the current Director General of ITER, Bernard Bigot, speaking to the US House of Representatives in 2016. This feels much less cherrypicked than a random interview from 2006. It took some effort, but I found the transcript of the hearing. What we see in Hossenfelder’s video is spliced: the question and the answer are not related. The question is from Grayson’s opening statement. The response is from a discussion between Bigot and Foster about heat flux to the diverter about half an hour later. I’m a lot less sympathetic to Hossenfelder’s argument after discovering this. Accusing someone of misinformation while splicing their video is … not good.
Elon Musk
His Argument: https://youtu.be/CnxzrX9tNoc?t=2849
My Response:
Elon Musk seems to have a cursory understanding of fusion, but hasn’t looked into it in detail. Which is not surprising because we shouldn’t expect him to be an expert about everything. He is both too optimistic and too pessimistic in different ways.
He says it’s 100% technically possible because you just have to increase the scale. That’s not how probabilities work: Would he take a million to one odds on that claim? Getting fusion is also not just an increase in scale. It’s mostly an increase in scale, but whenever you make things bigger, there’s a good chance that something will be different qualitatively too. This is why it’s important to make progressively bigger experiments, instead of jumping from your first small prototype to a full scale reactor. Looking at the history of fusion, we have been surprised by both bad things (e.g. large turbulent transport) and good things (e.g. H mode) when we were just scaling things up. ITER and SPARC are designed for $Q = 10$ instead of $Q = 5$, so even if the experiment performs half as well as we think, it will still be good enough.
The rare fuel claim is interesting. Because it either means that he doesn’t know about the tritium breeding blanket or it means that he thinks that it will never work. Tritium is very rare, but a fusion power plant should make all of its tritium on site out of lithium.
Transforming heat to electricity is done in most power plants today. I don’t think it’s a deal breaker.
Maintenance of the fusion reactor might be a serious problem. We don’t know how much maintenance is needed until we have one running.
Putting these together, Musk thinks that fusion will cost an order of magnitude more than wind and solar. This is not outside the realm of possibility. But it’s much too early to conclude that fusion will not be commercially viable.
References
| ↑1 | JT-60 and several other tokamaks have similar performance. |
|---|---|
| ↑2 | For example, my Brief History of Schooling in America. |