The Trouble with Fusion (1983) [pdf] | Hacker News Reader

The Trouble with Fusion (1983) [pdf]

orcutt.net

44 pointsakakievich44 comments

I’m still only a few pages in. I don’t have the chance to read it all right now, but I will revisit this tomorrow morning. Here are my early thoughts. Disclaimer: I am an advocate of fusion and am moving into the field (professional electrical engineering support of an academic project).

We need to walk before we can run. D+T fusion is viable. Hull maintenance is a solvable problem. Economic studies come after we can walk. Similarly, actually realizable designs that can reach Lawson performance necessary for p+B11 won’t come before designs that can reach D+T are made.

There are compelling narratives around funding, but ultimately it seems they are simultaneously overly optimistic about our technical advancement and overly jaded about future prospects.

It’s important to note that there is no difference in a reactor design capable of different fusion fuels, other than its Lawson performance. You can fuse D+T just the same as p+B11 if your device is capable of p+B11. There is not magical jump in performance we can make. So why skip over D+T if we need to go through it anyway? If D+T is found to truly not be economically viable (which sounds inherently flawed, since prices go down as technology becomes more advanced) then we could still pursue confinement devices in the hopes of reaching more advanced fuels.

> So why skip over D+T if we need to go through it anyway?

Because much of what you'd end up doing to "do" D+T will be worthless for H+11B. Tokamaks, for example, are useless for H+11B.

Tokamaks may not even be viable for steady state D+T. The bootstrap current effect still needs to be explored further. Electrostatic confinement and inertial confinement are the only known viable options for p+B11. Electrostatic has serious issues related to conduction losses, but perhaps if they were large enough scaling laws may make them viable. It’s not well explored. Inertial confinement is getting a lot of funding and shows serious promise. However, inertial confinement is not a steady state design and its pulsed nature makes it difficult to make a power plant out of. No such engineering cliffs exist for tokamaks and stellarators. The performance just needs to keep marching towards success. Then the discussion of economics can begin in earnest.

Electrostatic confinement is not known to be viable for H-11B. Indeed, Rider showed it probably isn't viable.

For inertial confinement, I understand compression would have to reach densities of as much as 10^6 g/cc for H-11B to be viable.

So I’m a few pages further and I fear I will not be finishing this paper. I am surprised it was published and is still referenced. Where are the citations? Where is the analysis? You can’t just say “fusion reactors need to be ten times bigger than fission reactors”. It’s frankly embarrassing to MIT that they would allow egregious, unfounded claims to be published. Peer review exists to shred such unsupported statements.

This was a semi-popular article. You can find more technical analyses in the literature.

The specific points made in the article have stood the test of time. Lidsky said the power density of a DT fusion reactor would be at least an order of magnitude worse than that of a fission reactor. And if you look at existing reactors and concepts, this is true. Compare to a commercial PWR, in which (thermal power)/(volume of reactor vessel) is about 20 MW/m^3. An order of magnitude worse than that would be 2 MW/m^3. The (thermal power)/(reactor volume) for ITER is about 0.05 MW/m^3. For ARC and Lockheed's concepts, about 0.5 MW/m^3.

Your outrage should not be directed at this article (although such outrage is sadly understandable, if this article is telling you things you don't want to hear.) Instead, it should be directed at the fusion community as a whole, which has downplayed these critiques and glad-handed the issues raised while marching confidently into a dead end.

My outrage is from making scientific conclusions before the science has been done. A hard justification of power density limitations And it’s economic meaning is necessary before you can make a blanket statement like “D+T magnetic confinement will never be economically viable”. The price of electricity is always changing, and public opinion is shifting towards the idea that we have been grossly underpaying for our electricity.

The paper made many good points but I found the power density argument by far the weakest. Yes a fusion reactor will necessarily weight more than a fission reactor of the same output and if we assume a constant price per pound across fission and fusion reactors this means it will be more expensive. But the reactor itself is only a small fraction of the weight of a fission plant. And of course we have every reason to think that the cost per pound of reactors across different reactor types will be entirely different.

I see this argument a lot in discussions around rockets where some people assume that if one rocket weighs less than another per pound delivered to space it must necessarily be cheaper when in fact the effort required to make it lighter almost always means it's more expensive. I think the rise of SpaceX has finally put an end to that argument, at least.

Fission reactor cores and reactor vessels are made of fuel and steel. They are very simple compared to fusion reactors. Fission reactors can therefore be expected to be much cheaper than fusion reactors, even on a per-pound or per-volume basis. They can also be expected to be much more reliable.

(A fission reactor vessel will also last the lifetime of the power plant. A DT fusion reactor, not so much.)

(The average density of ITER is about the same as the average density of an unfueled PWR reactor vessel, btw.)

The analogy with rockets would be for those advocating air-breathing launchers. Make the launcher much more complex, just so one can save on LOX. Make the reactor much more complex, just so one can not pay for uranium. But LOX is cheap, and fuel is not a major cost driver for fission power.

Yes, the arguments about the relative complexity of fusion reactor were well taken and they present serious challenges, unlike the power density issue.

But at the same time fission reactors are very expensive. The reactors themselves cost something on the order of $10 billion and the cost of the fuel and steel that go into the reactor is a very small fraction of that. And as far as I can tell the reason for that price is that the building is built to very exacting standards. It has to be built to those because even a reactor that's been shut down is putting out roughly 10% of the thermal power it was generating when it was active due to secondary decay, which fades away over time. But that means that the cooling system cannot ever fail, which means things get very expensive. And that's the same reason why instead of using the $.10 off the shelf screws you get off the shelf NASA uses special $100 aerospace grade screws.

The paper talks about safety as if it can be separated from price but in a world where we care about making unsafe things safe the two are inextricably linked. If a cooling failure results in a meltdown that will make the systems very expensive. If a fusion containment failure just stops output and causes excess wear on the inner lining, which has be replaced periodically anyways, then that's a different issue.

Which factor, over-engineering for safety or design complexity, will dominate, I don't know.

> unlike the power density issue.

I reject that assertion. The argument you tried to give against that power density argument (using magnet advances) just showed you didn't understand the point Lidsky was making.

> And as far as I can tell the reason for that price is that the building is built to very exacting standards.

Fusion reactors will also have to be built to very exacting standards -- not because of safety concerns, but because any malfunction in the radioactive, hands-off part of the plant will be economically disastrous. Lidsky goes into this point as well.

> The (thermal power)/(reactor volume) for ITER is about 0.05 MW/m^3.

https://www.f4e.europa.eu/understandingfusion/iterdevice.asp...

500/800 = 0.625 (plasma volume, not reactor volume)

In an interview with the Omega Tau podcast a while back, it was mentioned by ITER personnel that the DEMO plant that would be the successor to ITER was probably going to be designed as a 2000 MW reactor. Since the plasma is in shape of a torus, that volume increases pretty quickly with the size of the confinement vessel.

Also, per unit mass of fuel, magnetic confinement fusion is still posited to be thousands of times more efficient than fission power.

Using plasma volume is clearly not appropriate, since the point of the comparison is to compare cost of the equipment, by comparing the volume and size of the equipment between the two approaches (and ignoring that fusion equipment is much more complex and sophisticated than that in a fission reactor.)

The fuel comparison is silly, since fuel is not a major cost driver for either approach. The misconception that fuel costs matter is one of the errors that leads people to think fusion is desirable.

Not to mention D+T has 2500 times higher power density than p+B11.

This old critique has aged fairly well. It was too optimistic about advanced fission, and the suggestion to move to advanced fuels for fusion was mostly shot down by Lidsky's student, Todd Rider.

https://pdfs.semanticscholar.org/fce7/a35629488ec030d983025a...

https://dspace.mit.edu/handle/1721.1/11412

A similar critique was being made around the same time by Pfirsch and Schmitter in Europe.

https://pure.mpg.de/rest/items/item_2131865/component/file_2...

There are a few parts that have been overtaken by advances. Improvements in superconductors have driven the minimum size for a fusion reactor far down, for instance.

Lidsky's argument is entirely unaffected by that advance. The power of his argument is that one can just ignore the plasma physics.

Plasma physics tells us the relationship between the minimum workable size and the containment field strength. Change the field strength and the size changes.

Yes, and Lidsky's argument works even if one assumes arbitrarily good plasma performance. He was not basing his argument on magnetic field strength or beta limits, or even any particular device geometry. Assume 100T magnetic fields and beta=1; his argument still applies.

What may be confusing you is that very low power density designs, like ITER, may be even worse than Lidsky's bound. So they could be improved somewhat by better magnets or plasma physics tricks. But once Lidsky's bound is reached, further improvements of that sort are no help.

Commonwealth Fusion Systems was formed out of the work presented here: https://www.youtube.com/watch?v=KkpqA8yG9T4

A lot has changed since 1983. The superconducting materials that CFS depends on weren't even discovered until the late 80s, and it took a very long time for that to become practical. Fast-forward to today and you can buy this stuff on Alibaba!

Lidsky's arguments against fusion apply to the ARC design. Its overall volumetric power density is 0.5 MW/m^3, 40 times less than a PWR reactor vessel.

What defines the volume to get to that number? From what I can tell from the ARC specs the plasma volume is 141 m3, and the expected power output is 200-300 MWe, which makes it 1.5-2 MWe/m3.

Do you have some numbers on a PWR reactor vessel? I was trying to look up some details on the APR-1400, but could not find any.

The volume is the volume of the reactor, including blanket, magnets, and the structure needed to support the JxB forces on the magnets, not the volume of the plasma. See table 11, page 30, in the ARC paper:

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

For PWRs:

https://ocw.mit.edu/courses/nuclear-engineering/22-06-engine...

(take the dimensions given for the primary reactor vessel, compute the volume as a cylinder with spherical end caps, and divide that into 3400 MW(th). The result is slightly below 20MW/m^3. Note also the power density of the core itself is given as greater than 100 MW/m^3.)

We know what the issue is--fusion never got a useful amount of funding.

Now, it is entirely possible that even given an enormous amount of funding--fusion might still not work. However, engineers with lots of money are remarkably clever and effective beasts (see: radio, semiconductors, plastics, steel).

Fusion has been funded at "Fusion Never" levels for almost 40 years while we subsidize every other significant energy source to the tune of billions or trillions of dollars every single year. (For example, how much money got poured into the technology that became "fracking"?)

No, that's not what the issue is. That's an effect, not a cause. The cause is that fusion turned out to be less promising than had been thought, and that led to budgets being tight. Lidsky's devastating critique was part of that (tokamaks not being as good as early hopes implied was another.)

> The cause is that fusion turned out to be less promising than had been thought

So did a whole lot of chemical rocketry. Instead of whining about it--we spent a lot of money on engineering, we made the Saturn V, and we went to the moon anyway.

The issue is that fusion only has one end point--providing energy. Researching chemical rocketry made better weapons--so we funded the snot out of it. The DOE spent billions on the Unconventional Gas Research Programs and lined a lot of pockets.

Funding isn't guaranteed to make progress, but lack of funding practically guarantees lack of progress.

Not a good analogy. Chemical rockets are the only real way to get to space. So if you make them better, even incrementally, you have a win. It also helped that launchers were very far away from fundamental economic limits on their performance. Expendable launchers, unlike power plants, are expended.

But fusion is competing against a plethora of other approaches to production of energy that actually work, and are being used, and are arguably superior.

Fusion also uses components that are mature due to their use in these other approaches. DT fusion, which Lidsky is addressing, will produce its energy as heat. This heat has to be turned into power using turbines and generators, a mature technology. And it's a mature technology that's a major part of the cost of coal and nuclear power plants, and is a big reason why those power plants are no longer competitive.

(This echoes an argument from the mid 20th century, when it was pointed out that fission power would, at best, be only slightly less expensive than power from coal, due to all the common elements the power plants shared. And it turned out nuclear fission was more expensive than that, as one could not do the nuclear parts too cheaply. DT fusion reactors promise to be much larger and more complex than fission reactors, for the fundamental reasons Lidsky and others gave, so one can reasonably expect their economics to fail even more.)

(This is also why the focus on thorium and SMRs to try to keep fission alive is probably hopeless.)

("Those who cannot remember the past are condemned to repeat it")

The complaint about funding is a red herring. It has the presumption that if funding had been available, fusion had a real chance of succeeding. But as Lidsky points out, it didn't have a real chance of succeeding. Even if the program had produced a reactor, no one would have wanted it.

> This heat has to be turned into power using turbines and generators, a mature technology. And it's a mature technology that's a major part of the cost of coal and nuclear power plants, and is a big reason why those power plants are no longer competitive.

Erm, natural gas uses turbines and generators and nobody seems to be whining about that. It isn't the turbine driving the cost in coal (nuclear is a different story).

First, we still can't engineer a superconductor. We have barely doubled magnetic field strength since 1970ish. And the big advance of superconductivity in graphite with slight offsets demonstrates just how little we know. Superconductors would have had a massive improvement in basic science with funding (this was one of the huge losses in not funding the Superconducting Supercollider).

The FFT (fast fourier transform) was effectively useless in the 1970s and 1980s--until Moore's Law made it not so useless. Similarly, computational dynamics made huge advances since the 1980s--to the point where non-simple toroids are now the standard.

Knowledge advances in a "front". If you throw money at a point (especially a fundamental one), it drags related knowledge forward as well. Look at steel, for example. Steel has been considered "mature and well-understood" (hah!) practically since 1910--but there was so much money being thrown at it that it continuously advanced for almost a century. Once steel moved forward, architecture and construction moved forward. Then we got new applications like cars. Then we got new tooling like heavy presses. Then we could use more exotic materials like titanium. I can go on and on.

Natural gas uses combustion turbines, not steam turbines (except as a bottoming cycle in combined cycle plants, but that produces only 1/3 of the output of the plant).

What combustion turbines allow you to do is avoid heat exchangers, and also operate at a temperature much higher than a steam turbine because no solid material needs to be at the temperature of the working fluid. A simple cycle gas turbine (without regeneration) has no heat exchangers. Heat exchangers are expensive; transfer of heat across a solid/fluid interface is not as fast as we'd like it to be.

This kind of neglects the fact that "rocket science" was just "harder than thought" for american scientists. Stuff started to work when they let German scientists build it.

Which is no surprise. Germany already had dozens amateur (space) rocket clubs way back in 1930, long before people in other countries were even mostly aware of the possibilty of sending rockets way up there.

The US had some catching up to do in the science department, which they accomplished in the end, though rockets loosely based on those initial designs are still in use today.

This is different to today's fusion situation, where we don't know whether there will be any better design at all - because nobody did built something that could be adapted to fit our purpose before us.

Didn't the feasibility of fusion increase with advances in magnets. I thought the much stronger magnets being available today allowed for much smaller (volume per watt generated) systems, smaller even than the Iter design.

I'm a layperson like everyone else here, so I'm just going to rely on the fact that fusion research is ongoing, well-supported by government and private sector investment, and is an active area of research for hundreds or thousands of intelligent and sane institutionally supported academics, as reasonable and practical justifications for my belief that fusion is not quite as problematic as laid out in Lidsky's dated and superseded work.

However, for a sketch of how to address the talking points of the more strident objectors one encounters in the wild, one can perhaps turn to https://fire.pppl.gov/fusion_critic_response_stacey.pdf for some ideas. Cheers!

In fact, section III of that document specifically rebuts The Trouble with Fusion, based on the state of knowledge 16 years later.

This is all you really need to read from Stacey:

"Based on our present understanding, D-T tokamak fusion reactors project a cost-of-electricity that is about 50% larger than the projected cost-of-electricity from advanced light-water reactors in the middle of the next century."

We all know what happened to the projected cost of fission reactors -- the projections turned out to be hopelessly optimistic, because of complexity and loss of experience. Fusion would face these problems in even worse form (indeed, ITER's cost ballooned 4x or more past the initial projections.)

The experience with fission has enabled us to calibrate the optimism bias in these projections, with damning results.

Simply being competitive with fission is no longer good enough for fusion to succeed. It has to be significantly better than fission.

There has never been any serious expectation of getting usable power from fusion. All the reactor designs worked on in mainstream research would destroy themselves in a short time by high-energy neutron flux.

Fusion research is, instead, a jobs program for high-neutron flux physicists, to provide a pool to draw on for weapons work.

There are interesting commercial projects for designs that do not suffer from high neutron flux, such as those pursuing pB reactions. I read of another where neutrons are emitted in a place some distance from where the expensive machine parts are.

You can tell if a fusion process is serious by whether they have an answer to the neutron problem. Tokamak doesn't.

High neutron flux is useful not only for weapons but for civilian purposes too, for example for "burning" nuclear waste, or for fusion-fission hybrid reactors. I wonder why the Department of Energy does not invest more in this area.

Fusion-fission hybrids combine the worst features of both. There is no user "pull" for the concept. If you want power with fission, just build a fission reactor; that's going to be simpler, cheaper, and altogether more sensible. If you want to dispose of waste, just seal it in dry casks and wait a century or three before deciding what to do with it. That will also be much simpler and (due to nonzero interest rates) cheaper.

> Fusion-fission hybrids combine the worst features of both.

Maybe, but maybe not.

Worst features for fusion reactors: 1. they don't exit now and they won't exist for the next 50 years; 2. they produce lots of neutrons, which make the surroundings radioactive

Worst features for fission reactors: 3. they can go Chernobyl, 4. they produce long-living radioactive waste, 5. they are horribly expensive 6. proliferation concerns

How do these things look for a fusion-fission hybrid:

1. fusion reactors don't exist. Well, they do exist but they are well below the breakeven point. For a hybrid, the fusion part has (a very) negative energy balance, but it's more than made up for by the fission part, so being above breakeven is not a concern. The technology to manufacture the fusion part of a hybrid exists today (and has existed for decades)

2. fusion reactors produce lots of neutrons. For a hybrid, this is actually the point of the fusion half

3. fission reactors can go Chernobyl. This is so because the current fission reactors are powered by a chain reaction. This chain reaction threads the very fine line between subcritical and supercritical, in other words a classical fission reactor sits in a very narrow region between a bomb and a fizzle. The fission reactor in a hybrid gets its neutrons from its fusion partner, not via a chain reaction. The beauty of not having a chain reaction is that you can't have a supercritical chain reaction, or a Chernobyl event

4. fission reactors produce long-lived nuclear waste. I agree with you that this is not the big deal that's made up to be by environmental groups, but the fact that you can burn it via a fusion-fission hybrid is a nice bonus point

5. fission reactors are expensive. this is fundamentally a consequence of 3, that they present the danger of going boom. And as long as the fission reactors get their energy from a chain reaction, this danger exists. If you have a design that cannot go supercritical because it does not rely on a chain reaction, this is going to be inherently passively safe.

6. proliferation concerns. Here I simply have no idea how fussion-fission hybrids compare with classical fission reactors. That's why I mentioned the Department of Energy. If they develop and run these new reactors, then proliferation concerns become moot.

Besides all these points, the fusion-fission hybrids have another advantage: they can burn U-238 [1], which makes up 99% of the uranium on Earth. This means not only you have more fuel available, but you don't have to go through the stupendously expensive process of enrichment. Or it can burn Thorium-232, which is 3 times more abundant than uranium. In other words, not only the construction costs would be much lower, but the operation costs too.

Oh, and here's another advantage. Because classical fission reactors are based on a chain reaction that has to be very narrowly confined between supercritical and subcritical, at any given point only a very tiny fraction of the fuel is burning. Nuclear advocates don't like to dwell on that, but they like to point to the flip side of this coin, that the fuel lasts for a very long time (years). However, if you could burn the fuel faster, you can get the same power from a smaller reactor. We could be talking a factor of 100. Since construction costs don't scale linearly with size, a reactor that's 100 times smaller could easily be 1000 or 10000 times cheaper. And we could end up being able to send gigawatt-size reactors to Mars, rather than the kilowatt-size currently envisioned by NASA [2]

[1] https://en.wikipedia.org/wiki/Nuclear_fusion%E2%80%93fission...

[2] https://en.wikipedia.org/wiki/Kilopower

You are referring to TAE's design. The people behind that were told 20+ years ago the concept would not work.

https://www.researchgate.net/profile/Wallace_Manheimer/publi...

http://w3fusion.ph.utexas.edu/ifs/ifsreports/919_wong.pdf

Some tokamaks do. MIT's ARC, for example, uses jointed superconducting tapes that let the reactor be opened up to replace the inner wall. That's done annually, and the wall is 3D-printed. Surrounding the inner wall is molten FLiBe salt, which acts as coolant, neutron absorber, and breeding blanket. Commonwealth Fusion is attempting to commercialize the design.

The concerns and criticism make too much sense; as much as I want to believe a comic/movie density power solution (like arc reactors) can work...

I think the criticisms would have gone over better if something more than the implicit: 'look for better ideas' proposal had been included as a path.

It's been over 30 years; is there anything else as an idea in the field of power generation? (Preferably something we could also use in space)

"Fusion in a magnetically-shielded-grid inertial electrostatic confinement device"

> Theory for a gridded inertial electrostatic confinement (IEC) fusion system is presented that shows a net energy gain is possible if the grid is magnetically shielded from ion impact. A simplified grid geometry is studied, consisting of two negatively-biased coaxial current-carrying rings, oriented such that their opposing magnetic fields produce a spindle cusp. Our analysis indicates that better than break-even performance is possible even in a deuterium-deuterium system at bench-top scales. The proposed device has the unusual property that it can avoid both the cusp losses of traditional magnetic fusion systems and the grid losses of traditional IEC configurations.

https://arxiv.org/abs/1510.01788

Reply on news.ycombinator.com