nuclear costs are inflation

Link post

Why has nuclear power gotten more expensive since 1970? Mostly, because matching inflation requires technological improvements.

Inflation-adjusted retail electricity prices in the US have gone down since 1960. Not only that, but after the increase from initial power line construction finishing, the fraction of electricity costs going to generation has decreased, with more going to transmission. Perhaps that’s because generation has more competition than transmission—California and New York have unusually high electricity prices, and you can thank corruption for some of that.

LCOE (levelized cost of energy) for combined cycle natural gas in the US was estimated to be ~$0.04/​kWh going forwards. In Germany, electric generation in 2021 was ~$0.08/​kWh; half the total cost was taxes/​fees.

Inflation-adjusted cost per ft^2 of building a house in the US has been about flat. Median real wages have also been about flat, with a decrease for people without college degrees and more time spent in school for those with them. And construction costs are the biggest issue overall, so the housing affordability crisis in the US is mostly about increased income inequality.

Inflation rate is a combination of several things combined into 1 number:

  • change in money supply

  • tech progress

  • returns to scale

  • resource depletion

  • capital accumulation/​depreciation

  • institutional improvement/​decline

The average inflation for the whole economy includes a lot of tech progress. Prices being flat when adjusted by (CPI) inflation doesn’t mean “there was no progress”, it means “as much progress as home construction”.

Which has had more tech progress, home construction or nuclear plant construction? I’d argue for the former: electric nailguns and screw guns, and automated lumber production. Plus, US home construction has had less increase in average wages, because you can’t take your truck to a Home Depot parking lot to pick up Mexican guys to make a nuclear plant.

Let’s look at the cost growth of some other things as reference points.

Concrete products increased about 3.9% a year since 1960, a little higher than the CPI inflation rate. Highway construction costs are up >5%/​year over the past 20 years.

If you want to cancel out the technological progress component of the inflation rate, looking at the cost growth of fighter jets seems reasonable, since ~100% of the tech progress on those goes into improved performance instead of reduced cost. And they’ve gotten a lot more expensive: a 1975 F-16 was ~$6.4M and 2021 F-16 was ~$64M, which is ~5.1% cost growth, despite some learning from continued production. Comparing that to the CPI, that would mean ~1.5% average annual cost reduction from topic-specific tech progress, which seems about right to me.

Hinkley Point C is an ongoing nuclear power project in the UK, which has seen massive cost overruns and delays. But if you take prices for electricity generation in 1960, and increase them as much as concrete has, you’re not far off from the current estimated cost per kwh of Hinkley Point C, or some other recent projects considered too expensive.

Yes, at this point some companies have lost their knowledgeable people and forgotten how to build nuclear power plants competently, which means you should applaud the NRC not letting them cut corners, but even the economically disastrous Plant Vogtle reactor project is only 5% inflation since the glory days of nuclear power. Anyway, to whatever extent capabilities have been lost, relative cost increases came first and drove that.

US electricity costs decreased thanks to progress on gas turbines, fracking, solar panels, and wind turbines. Gas turbines got better alloys and shapes. Solar panels got thinner thanks to diamond-studded wire saws. Wind turbines got fiberglass. Nuclear power didn’t have proportionate technological progress.

What about China? They’re building nuclear power.

The PPP/​nominal GDP of China is almost 2x that of the USA. China can construct stuff more cheaply than the US. Also, those plants are heavily subsidized in opaque ways involving government loans, perhaps because the Chinese government has also been rapidly expanding its nuclear weapon arsenal for better deterrence when it goes for Taiwan. If their nuclear power was actually cheaper than coal they wouldn’t be building mostly coal plants, would they?

What about the cheap Kakrapar IPHWR-700 plants?

That project was reported earlier this year as being ~$1.70/​W with current cost overruns, which is weird because that’s half the cost of earlier Indian nuclear plants. You should adjust costs ~3x for PPP, but also, that number is fake; the project isn’t done and costs were shifted to the final reactor with accounting tricks.

What about US Navy nuclear reactors for ships?

The A1Bs used for Ford-class carriers may well be cheaper than the new Vogtle reactor, but the US Navy has never estimated that its nuclear ships were cheaper than using oil, which is more expensive than natural gas from a pipeline. You also can’t use them for commercial power, because they use highly enriched fuel and don’t have a secondary containment structure, plus most places don’t have seawater for cooling.

OK, so for cheap nuclear power we need some sort of better technology, right? Wasn’t that what the billions of dollars the US government spent researching Gen IV reactors for?

No, not really. There’s no reason to expect any of those to be cheaper than current PWR and BWR reactors. The focus of that was making fast reactors more practical. Current nuclear plants are very inefficient with fuel, and if they were used for most human energy needs, current fuel supplies would run out in a couple decades, before natural gas.

Somebody reading this probably wants to say:

but there’s plenty of uranium in seawater!

Maybe you read something about that only being 6x or so as expensive, but whatever number you read is fake. I don’t want to see a comment from you unless you look into how that number was calculated and think real hard about whether it would be the only cost involved. The uranium in a cubic meter of seawater, used in current reactors, produces electricity (~0.1 kWh) worth less than even simple processing of a cubic meter of seawater costs. If seawater processing cost as much as desalination, and fuel was half the electricity cost, that’d be $10/​kWh. Well, at least I guess it’s less dumb than mining He3 from the moon.

I expect solar + wind + 1-day storage from Hydrostor-type systems or metal chelate flow batteries to be cheaper than nuclear power in the US and Europe. That might seem attractive in Europe vs shipping LNG from the US, but seasonal energy storage is a lot more expensive than that; hydrogen can be stored cheaply in underground caverns, but hydrogen from electrolysis is much too expensive for power. I also expect biomass power to be cheaper than nuclear, but Europe and Japan don’t have enough land for that, and land use would be a big issue even in the US.

Current nuclear power is too expensive and doesn’t have enough fuel to be the main source of energy. Breeder reactors would be even more expensive and less safe. And yet...

Let’s suppose war breaks out over Taiwan and LNG carriers to Europe and Japan aren’t able to operate safely due to threats from submarines and long-range missiles, and seasonal energy storage is too expensive, so they have no choice but to use nuclear power. Or, if you prefer, suppose that it’s 50 years from now, AI hasn’t taken over, no unexpected technologies have come up, and the US has run out of natural gas but people don’t want to go back to coal.

Cheaper nuclear plants would then be important, which as I’ve explained above, requires technological improvements to match those that other energy sources got. As for how to do that, as I previously said, I think CO2 cooled heavy water moderated reactors with uranium carbide fuel are reasonable, mostly because you can avoid steam turbines, and TerraPower’s design is terrible and unsafe, which I guess is what happens when investors who don’t understand physics start picking technology. TerraPower’s design actually offends me personally on an aesthetic level; it’s like this when it should be like this. If you must build fast reactors, maybe I’d try to couple a fast neutron area to a thermal reactor area to make control easier by getting some neutron lifetime and some reactivity loss when moderator heats up—but I’m not a nuclear power specialist, I don’t like nuclear reactors, I’m only better than clowns like Kirk Sorensen or the TerraPower people.

I guess you could hope for fusion power too, but it’s not as if that’s a cheaper or less-radioactive way to generate heat than fission. Helion has been in the news lately and hopes for direct conversion from plasma, but they’re going nowhere; here’s a non-technical explanation of why, some semi-technical explanations, and I guess you can message me if those somehow aren’t technical enough for you.

If you’re going to work on some new technology, it’s a lot easier to improve a basic type of thing that’s used now or at least already close to viability. Nuclear power currently isn’t. If Bill Gates or somebody wanted to, for some reason, give me a billion dollars to make a nuclear reactor, I guess I’d do that, but I’d certainly rather work on flow batteries.