5 ms·

So, I'm a physicist, and I went to a number of talks from people involved with the JET fusion reactor over the years, though fusion is not my area. And my under

So, I'm a physicist, and I went to a number of talks from people involved with the JET fusion reactor over the years, though fusion is not my area. And my understanding is not that it's difficult to make plasma, or even build a reactor in particular that is the main problem (though instabilities can be problematic), but it's that the internal structure degrades very rapidly and becomes highly radioactive, because you get helium bubbles forming inside the steel which causes fractures and the heavy metals that are often put into steel to increase strength are highly fissionable. So you need to use special types of steel to actually construct the reactor, and these need to have a lifetime that's ~5 years+ and it needs to not have very very radioactive steel at the end of it's usable life. And this is basically an unsolved materials science problem at the moment. So while you might even be able to build a fusion reactor, it's not going to last long enough to make it commercially viable using current technology.

varjag 6y ago
That's a luxury problem at the moment. A fusion reactor that can sustain positive output operation for 5 years is still a dream.
- OJFord 6y ago
  Sort of, but only if you exclude the amortised cost of fabrication from your calculation - it's part of the same goal, unlike say 'they're hard to make, we can't make them fast enough to put a dent in energy production'.
- jnxx 6y ago
  It is a total show-stopper because you don't have the amount of Tritium to power such a reactor.
  - DennisP 6y ago
    We have plenty of lithium. All D-T fusion designs breed tritium from lithium, using the high-energy neutrons from the reaction.
    - galangalalgol 6y ago
      I assume very little lithium is actually needed, will the dramatically increased demand from battery manufacturing affect feasibility?
      - DennisP 6y ago
        Lithium production is a bit strained but lithium resources are fairly high. At Tesla's Battery Day they said just the lithium in Nevada is enough to electrify the entire US fleet. They also plan to do their own recycling, which they expect to work well since they can customize it to their battery. Given the amount of energy that comes from fusion fuel, even extracting lithium from seawater would probably be economical. Beryllium is much less common so that could be an issue. But maybe we can find more beryllium if we look harder; right now we barely use it. If we can't find enough, we could probably design a reactor to use lead instead. Either one acts as a neutron multiplier, and General Fusion's design uses lead instead of beryllium, for the same purpose.
      - pfdietz 6y ago
        DT fusion reactors require 6Li, which is 7.6% of natural lithium. The other 92.4% is 7Li. This "depleted lithium" would work in batteries just fine.
grandinj 6y ago
This++++++. Nuclear reactors are primarily bug generators for chemists and metalurgists. The things that need solving mostly boil down to dealing with the insane chemistry and metallurgical problems that arise when you have a wide variety of incredibly nasty compounds arising that do not normally occur, and consequently require out-of-the-ordinary solutions.
uCantCauseUCant 6y ago
It needs a paradigm shift, away from a fixed structure, towards a structure that repairs part of its "exposed" skin constantly.
pm90 6y ago
Not disagreeing with you, but it might still be ok. I don't believe early fission reactors were cost effective either. If we build a few with heavy subsidies and create an economy around them, we may get more commercial players involved, and a lot more investment dollars chasing the problem.
- justincormack 6y ago
  The early fission reactors were to make bombs not electricity. Large subsidies available there.
  - pm90 6y ago
    I think it was clear from context that I meant fission reactors designed for electricity production. These reactors are significantly different in design from reactors used for producing weapons grade, highly enriched fissile material, and required significant investment in its own right to develop safe, commercial reactors.
Mvandenbergh 6y ago
One of the advantages with the MIT superconducting magnet design is that it allows the vacuum vessel and neutron shielding to be lifted out and replaced wholesale (n.b. this doesn't have to be every 5 years but it does have be done at some point during the reactor lifetime). The ITER design would require essentially disassembling the whole reactor to do that. In both cases, the neutron absorbing first wall will be replaced every 5 years or so which is indeed a substantial challenge requiring robotic operations.
jnxx 6y ago
Applied physicist here. The other thing is: Existing fusion concepts need to use a reaction between Deuterium and Tritium, otherwise it's just infeasible. Deuterium exists in nature and can be produced with some form of isotope separation. Tritium, however, does NOT exist in nature, and can only be conceivably produced in three ways: 1. Inside thermonuclear bombs 2. In heavy-water reactors, e.g. CANDU design, which requires Uranium, so it is not fuel-independent and totally dependent on nuclear fission technology. Getting Tritium that way is also extremely, extremely expensive. 3. possibly, in a breeding blanket of a fusion reactor. This makes fusion technology reliant on breeding reactors which is much more complicated and has many completely unsolved questions in the materials part. Also, existing fission-breeding reactors are also less safe, for example, like the Japanese Monju plant, which was cooled with liquid sodium, a highly reactive metal, which got incensed in a fire. Fusion breeder reactors will probably also require something like sodium cooling because of the high energy densities required. And the existing research projects like ITER have not even started to address these issues - they are purely plasma physics experiments. https://en.wikipedia.org/wiki/Monju_Nuclear_Power_Plant https://en.wikipedia.org/wiki/Monju_Nuclear_Power_Plant
- DennisP 6y ago
  SPARC uses molten FLiBe salt as both coolant and breeding blanket.
- willis936 6y ago
  >And the existing research projects like ITER have not even started to address these issues - they are purely plasma physics experiments. Section 3.2 Test Blanket Module (TBM) Testing Program in ITER https://www.iter.org/doc/www/content/com/Lists/ITER%20Technical%20Reports/Attachments/9/ITER_Research_Plan_within_the_Staged_Approach_levIII_provversion.pdf https://www.iter.org/doc/www/content/com/Lists/ITER%20Techni... Look even closer at the fusion energy research community and you will find a lot of work has already gone into this problem. Assuming physicists sit on their thumbs is a bad bet.
- RobertoG 6y ago
  >> "Tritium, however, does NOT exist in nature" Maybe I don't understand properly, but I'm watching the video linked in a previous comment (1) and around the minute 9:30 professor White says that "what comes into the plant is actually deuterium and lithium". It seems he is saying that you need only an initial quantity of Tritium and then Helium and Lithium is used to created Tritium again. I suppose that it's what you refer as "breeding". He explain it as it's not a big deal. (1) - https://news.ycombinator.com/item?id=24632653 https://news.ycombinator.com/item?id=24632653
pfdietz 6y ago
> the heavy metals that are often put into steel to increase strength are highly fissionable Not fissionable, but many are unacceptable due to formation of long lived activtion products. One additional problem with the alloys used is the degradation of their mechanical properties under radiation exposure. They tend to become brittle. Testing any of these materials will require something close to a working fusion reactor (nothing else duplicates the neutron environment), but that will require the materials. Working around this loop of circular development dependency will be time consuming.

varjag 6y ago

That's a luxury problem at the moment. A fusion reactor that can sustain positive output operation for 5 years is still a dream.

OJFord 6y ago
Sort of, but only if you exclude the amortised cost of fabrication from your calculation - it's part of the same goal, unlike say 'they're hard to make, we can't make them fast enough to put a dent in energy production'.
jnxx 6y ago
It is a total show-stopper because you don't have the amount of Tritium to power such a reactor.
- DennisP 6y ago
  We have plenty of lithium. All D-T fusion designs breed tritium from lithium, using the high-energy neutrons from the reaction.
  - galangalalgol 6y ago
    I assume very little lithium is actually needed, will the dramatically increased demand from battery manufacturing affect feasibility?
    - DennisP 6y ago
      Lithium production is a bit strained but lithium resources are fairly high. At Tesla's Battery Day they said just the lithium in Nevada is enough to electrify the entire US fleet. They also plan to do their own recycling, which they expect to work well since they can customize it to their battery. Given the amount of energy that comes from fusion fuel, even extracting lithium from seawater would probably be economical. Beryllium is much less common so that could be an issue. But maybe we can find more beryllium if we look harder; right now we barely use it. If we can't find enough, we could probably design a reactor to use lead instead. Either one acts as a neutron multiplier, and General Fusion's design uses lead instead of beryllium, for the same purpose.
    - pfdietz 6y ago
      DT fusion reactors require 6Li, which is 7.6% of natural lithium. The other 92.4% is 7Li. This "depleted lithium" would work in batteries just fine.

grandinj 6y ago

This++++++. Nuclear reactors are primarily bug generators for chemists and metalurgists. The things that need solving mostly boil down to dealing with the insane chemistry and metallurgical problems that arise when you have a wide variety of incredibly nasty compounds arising that do not normally occur, and consequently require out-of-the-ordinary solutions.

uCantCauseUCant 6y ago

It needs a paradigm shift, away from a fixed structure, towards a structure that repairs part of its "exposed" skin constantly.

pm90 6y ago

Not disagreeing with you, but it might still be ok. I don't believe early fission reactors were cost effective either. If we build a few with heavy subsidies and create an economy around them, we may get more commercial players involved, and a lot more investment dollars chasing the problem.

justincormack 6y ago
The early fission reactors were to make bombs not electricity. Large subsidies available there.
- pm90 6y ago
  I think it was clear from context that I meant fission reactors designed for electricity production. These reactors are significantly different in design from reactors used for producing weapons grade, highly enriched fissile material, and required significant investment in its own right to develop safe, commercial reactors.

Mvandenbergh 6y ago

One of the advantages with the MIT superconducting magnet design is that it allows the vacuum vessel and neutron shielding to be lifted out and replaced wholesale (n.b. this doesn't have to be every 5 years but it does have be done at some point during the reactor lifetime). The ITER design would require essentially disassembling the whole reactor to do that. In both cases, the neutron absorbing first wall will be replaced every 5 years or so which is indeed a substantial challenge requiring robotic operations.

jnxx 6y ago

Applied physicist here. The other thing is: Existing fusion concepts need to use a reaction between Deuterium and Tritium, otherwise it's just infeasible. Deuterium exists in nature and can be produced with some form of isotope separation. Tritium, however, does NOT exist in nature, and can only be conceivably produced in three ways: 1. Inside thermonuclear bombs 2. In heavy-water reactors, e.g. CANDU design, which requires Uranium, so it is not fuel-independent and totally dependent on nuclear fission technology. Getting Tritium that way is also extremely, extremely expensive. 3. possibly, in a breeding blanket of a fusion reactor. This makes fusion technology reliant on breeding reactors which is much more complicated and has many completely unsolved questions in the materials part. Also, existing fission-breeding reactors are also less safe, for example, like the Japanese Monju plant, which was cooled with liquid sodium, a highly reactive metal, which got incensed in a fire. Fusion breeder reactors will probably also require something like sodium cooling because of the high energy densities required. And the existing research projects like ITER have not even started to address these issues - they are purely plasma physics experiments. https://en.wikipedia.org/wiki/Monju_Nuclear_Power_Plant https://en.wikipedia.org/wiki/Monju_Nuclear_Power_Plant

DennisP 6y ago
SPARC uses molten FLiBe salt as both coolant and breeding blanket.
willis936 6y ago
>And the existing research projects like ITER have not even started to address these issues - they are purely plasma physics experiments. Section 3.2 Test Blanket Module (TBM) Testing Program in ITER https://www.iter.org/doc/www/content/com/Lists/ITER%20Technical%20Reports/Attachments/9/ITER_Research_Plan_within_the_Staged_Approach_levIII_provversion.pdf https://www.iter.org/doc/www/content/com/Lists/ITER%20Techni... Look even closer at the fusion energy research community and you will find a lot of work has already gone into this problem. Assuming physicists sit on their thumbs is a bad bet.
RobertoG 6y ago
>> "Tritium, however, does NOT exist in nature" Maybe I don't understand properly, but I'm watching the video linked in a previous comment (1) and around the minute 9:30 professor White says that "what comes into the plant is actually deuterium and lithium". It seems he is saying that you need only an initial quantity of Tritium and then Helium and Lithium is used to created Tritium again. I suppose that it's what you refer as "breeding". He explain it as it's not a big deal. (1) - https://news.ycombinator.com/item?id=24632653 https://news.ycombinator.com/item?id=24632653

pfdietz 6y ago

> the heavy metals that are often put into steel to increase strength are highly fissionable Not fissionable, but many are unacceptable due to formation of long lived activtion products. One additional problem with the alloys used is the degradation of their mechanical properties under radiation exposure. They tend to become brittle. Testing any of these materials will require something close to a working fusion reactor (nothing else duplicates the neutron environment), but that will require the materials. Working around this loop of circular development dependency will be time consuming.