[avar]

In this scenario you could just run the nuclear plant all the time, but just direct its electric production to heating a giant pool of water or whatever.

Then when the grid needs more power you do less of that, and instead send electricity to the grid.

[ajross]

Yes, but the point is that the problem is isomorphic to wind or solar: you get power when you don't want it. So nuclear isn't backstopping any particular need, it has the same drawbacks. The reason gas plants are used as peaker plants is that they can be turned on and off more or less instantly.

[avar]

You're just assigning magical value to "turn it off".

That's only a big deal with fossil fuels due to the fuel cost, the fuel cost of nuclear is marginal.

Yes, I agree that it's stupid to build a (big) nuclear power plant only to use it to boil an Olympic sized swimming pool most of the time.

But that's stupid because we're over-relying on "green" energy that can't provide baseload power.

It's even more stupid with fossil fuels, now you also need an entirely different backup infrastructure, but the fossil one pollutes much more than nuclear.

[hnaccount_rng]

The problem with your logic is, that nuclear is _only_ competitive with a 100% utilization rate/capacity factor. Only with these 90+% capacity factors do you get to competitive rates per kWh produced. Otherwise the initial CapEx is just too large, especially in a non-zero rate environment, as these plants are _already_ scoped/calculated/financed to 40 years. At 90+% capacity factor! That amount of CapEx is basically impossible without either very high earning potential (that’s why TSMC works. Everyone will just pay their price) or government backing (because then you don’t have to care about financing and optionality costs and can just pay for it with taxes). Neither one will be possible, at least in western societies

(And yes this ignores a certain amount of externalities, like the tendency of requiring large scale evacuations in their surroundings every second decade, but we can set that aside, as it’s irrelevant for the economic argument above)

[phil21]

At some point society will realize comparing power pricing to the lowest cost per kwh on a given day is a silly waste of time.

What matters is what is the most realistic mix of power generation and storage for 24x7 reliability. This can of course look very different depending on the situation. Many will argue for distributed storage (e.g. home batteries) - but that just means poor people don't get reliable electric service.

I really don't find that solar is "too cheap to meter" during peak sunlight very interesting. Who cares. What I find interesting is that I can turn a dial on my nuclear power plant to whatever it is I feel like at any time, and have it operating at that capacity within an hour or three.

Since we have such a dial, if you owned both the solar and the nuclear plants you would very likely combine them in a manner that maximizes profits while maintaining continuous service. Short of clouds, regional solar and wind prediction is extremely good to the point that modern nuclear plants may as well be load following. Add in a bit of battery for those minutes (hours max) that surprise you and you're good to go on that front.

You still will need some gas peaker plants for those crazy once-in-a-decade days you don't want to overbuild nuclear capacity for, but you could drastically reduce this infrastructure from what is effectively a 1:1 ratio today.

[avar]

I'm speaking in terms of what we should do. I realize this isn't workable in practice for various reasons.

The main reason is that we price pollution and radiation control into nuclear plants, but fossil fuel plants get a free pass.

[acchow]

Or powering GPU compute centers. When the grid needs more power, these dynamically power down (raise LLM generation prices)

[tuatoru]

Electrolyzing water to make hydrogen for ammonia (fertilizer) and industrial chemical feed stocks, please.