Post of the Week 2021-05-13
My thoughts today led me to reading about nuclear wastes. It turns out that what we call “nuclear waste” is a pretty vague term.
A light-water reactor generates different kind of “waste” from a molten salt reactor. A traditional light-water reactor doesn’t consume fuel efficiently. It leaves behing partially consumed fuel that is not economically attractive to refine for further use. These partially spent rods continue to generate heat and are sometimes stored in cooling pools. Wastes from light-water reactors contain plutonium, which can be used to create atomic weapons.
A thorium-based molten salt reactor produces waste in a fluid form. This waste has low heat content which makes it easier to store. The only weapons-capable content of this waste is uranium-233 (though some argue it’s hard to weaponize). Unlike light-water reactors, the waste from molten salt reactors does not contain plutonium. According to this paper, there are still not enough experiments to fully understand how to process and handle waste from molten salt reactors.
It is frustratingly hard to find information on the merits of molten salt reactors. The Union of Concerned Scientists, which seems perfectly legitimate, has a lengthy PDF (released this year) that points out shortcomings of molten salt reactors, such as them being producing radioactive gases, requiring on-site chemical processing, and more risk of having weapons grade nuclear material stolen (U-233?). On the other hand, the document from UCS does not include “nuclear battery” designs, such as the Toshiba S4, which is positioned as both efficient and has low nuclear proliferation risks.
I can’t help but wonder if people’s reluctance to accept nuclear power is affected by the availability bias. In theory, a comparison between nuclear power and fossil fules should go like something this:
A = ((Benefit-of-Nuclear/year) - (Cost-Per-Nuclear-Problem * Chance-Of-Nuclear-Problem/year)) * years
B = ((Benefit-of-Coal/year) - (Cost-Per-Coal-Problem * Chance-Of-Coal-Problem/year)) * years
Where the decision is to prefer nuclear power if A > B, or coal otherwise.
Coal’s Chance-of-Problem is 100%. It is constantly producing harmful pollutants into the air.
Nuclear’s harm is mostly contingent on high profile disasters, such as a meltdown or a potential for weaponization. That is, the harm of nuclear power is harder to assess since it is contingent on probabilities, and we don’t know what those probabilities are. Since we don’t know what those probabilities are, people are likely to use the availability heuristic to assess those risks. Unfortunately, this heuristic leads to lousy predictions and overestimates the costs of nuclear power. In other words, when people try to choose between nuclear and coal power, they see the image of Chernobyl vs a (relatively harmless) smoke stack. They do not adjust those two mental images via probabilities. That’s just not how the human brain (System 1) works.
It seems frustratingly impossible to move forward on nuclear power, because there isn’t an outright perfect design. Each design makes different trade-offs between complexity, waste creation, and proliferation concerns. This makes it impossible for engineers to agree on a single design as the best one. But surely they are all better alternatives than burning fossil fuels.
This also points out a silly problem in the green energy movement. Variants of green energy are not the enemy – fossil fuel is. I guess the problem is that the challengers are always competing over scraps from the fossil energy industry. With so little left, different green energy solutions end up competing with each other. Whatever criticism they have for each other inevitably become supporting evidence for fossil fuels. I suspect the solution to this competition problem is to apply some sort of multi-armed bandit algorithm over different green energy alternatives so that we might optimally allocate resources to exploration.