Tesla’s new Solar Roof costs less than a new roof plus solar panels

Workers, fuel, repairs, decomissioning etc. are very cheap, much cheaper than equivalent coal fired plant, for example. Levelised cost of electricity, which includes all of these plus construction costs, is still the lowest for nuclear energy, much lower than wind and solar even if you don't build any batteries at all.

> Further Nuclear only has a ~90% capacity factor they can’t operate 24/7 365 for 50 years without turning off.

??? How is this relevant for the discussion at all? If you have 2000 plants, then with 90% capacity factor, you'll have 10% of plants down at any given time, thus you simply need to have 10% more plants than what you need. The crucial thing is that with nuclear, down time can be scheduled, while with wind and solar, overcast sky and windless days are completely beyond your control.

> Nuclear also has the opposite problem, it needs to scale up production in the day to cover peak demand, and even more to reach peak seasonal demand.

You simply need to build enough for peak seasonal demand. The nuclear plants can scale up and down as needed, albeit slower than gas plants, so you just scale up to cover peaks, and have aluminium plants enjoy cheap electricity when you overshoot.

> Run the numbers assuming exactly the right amount of power based on an annual average and your shifting energy across months.

My 2000 plants for $200B/year was already 12% above average use, but sure, I could double my numbers to 4000 plants and still be below 10% of US federal government spending. Imagine how big of a boon would such plentiful and cheap off-peak energy be for US heavy industry or data centers.

> Also, pretending we suddenly need to replace all sources of energy from PV power that’s then stored in batteries is crackpot territory. Part of those 101 quads are fuel used to generate electricity at sub 40% thermal efficiency. We don’t need generate electricity to replace the energy from fuel used to generate electricity.

Sure, but note that the nuclear numbers also scale just as well, so even if the increased efficiency might make the whole batteries business a bit closer to the realm of feasible, it will make the nuclear approach even cheaper too. The inefficiencies you mention will probably amount to something like 30% of current energy use, so they don't change the calculations substantially: utility scale batteries still infeasible.

> PS: The article makes it’s own estimate at 2.25 trillion for 900GWh of batteries being sufficient for 100% rentable electricity generation using some other set of assumptions.

Yes, their estimates are 10 times higher than mine. That should make you make think really hard about feasibility of large scale batteries. For 2.25 trillion, South Koreans can build 900 GW worth of nuclear generating capacity -- which one you think is more sensible choice, one hour of 900 GW from batteries, or continuous 900 GW from nuclear plants, if your goal is "decarbonizing the grid"?

Yes, this exactly means that operating costs are very cheap: if 70% of the cost per kWh is amortization of construction costs, and only 30% is operating costs, and the kWh from nuclear plants are still the cheapest, or exactly means that the operating costs of running a nuclear power plant are very, very low, much lower than coal plants.

> Looking at pure construction costs in a steady state environment without loans it’s a significantly smaller fraction which continues to increase with age. This is why several viable nuclear power plants have been decommissioned early, their operating costs are to expensive even after construction has been paid off.

Which ones? What were their operating costs?

> So if you want 2.2 TW 24/7 you need to build ~1,112 of them not 1,000 of them.

Sure, but as I noted, we can build twice the amount we need rather easily, while building enough of utility scale batteries to ride out the variations in solar/wind production is just infeasible. Of course nuclear isn’t perfect, but I thought your alternative is solar/wind, which has all the same problem, but worse - instead of predictable and controllable 90%, you get 40-50% of installed capacity on average, and it can be all down at times beyond your control.

> Up to a point, your talking days of downtime not hours so you still need peaking power. The cheapest approach is for a steady state of workers smoothly moving from one project to the next. That’s hard to pull off, and you end up needing even more generation to cover scheduling issues and unexpected problems.

That’s okay, since we can build twice our peaks and still be firmly in the realm of feasible.

I note, however, that you gave up on the idea of utility scale batteries being feasible approach to work around the inherent unpredictability of solar and wind. Good.

1) solar is distributed. In the past couple years, California has had massive fires caused by transmission lines. Right now, millions of people are without power because the utility can't safely deliver electricity. Distributed solar obviates the need for huge transmission lines, and is more durable in a disaster.

2) utility scale solar is cheaper than nuclear here.

3) There are large earthquakes here that make it less obvious that we should run 70% of our economy on nuclear.