[XorNot]

But it's the half gigawatt of battery which is the problem - it's nowhere near enough. Roughly batteries scale 1:4 power-to-energy. So 0.5 GW is only about 2 GWh of actual storage. But you need that for almost 18 hours a day since solar tends to have about 6 productive hours a day. So your batteries are covering you for maybe 0.1 GW of constant draw - presuming nothing goes wrong (i.e. a week of regional cloud cover).

On top of that the solar plant capacity factor is somewhere between 10% - 30% in most locales, so the sticker plate capacity of 5 GW is going to be under 2.5 GW at best (and that would be a 50% capacity factor).

I've never been able to find a way to square an actual "no fossil fuels grid" with the supposed cheapness of solar or wind - it always feels like people are quoting selectively useful $/GW values and then not giving a full accounting of the assumptions behind them - i.e. GW type quotes originate with thermal powerplants which have capacity factors which are essentially "whatever you want if you pay us".

[ben_w]

I'm less optimistic about the current state of affairs than the person you're conversing with, but I think more than you.

The LCOE values I've seen place batteries+PV at ~ nuclear… but nuclear is more expensive than almost anything else.

I anticipate further reductions in the price of batteries from the learning curve and demand driven by electric cars where they're already cheap enough to replace ICEs, such that the cost of batteries for electricity time-shifting will be OK fairly soon (as in: 5-10 years)but that's a forecast and not a guarantee.

There's also the possibility of a global power grid — the maths works out just fine, few hundred billion USD and a year or two of global aluminium production, we have to spend more than that on upgrading the last (hundred) miles even if we never build the global interconnects — but basically only China has both the interest and the capabilities to attempt something like that as part of a future belt-and-road initiative, everyone else will definitely not get past the "talking about it" stage.

[XorNot]

I mean that to me seems like the obvious answer if we're doing solar: don't really do storage, just wrap the generating capacity around the Earth and have it follow the sun.

Nation-states would likely still retain strategic reserves of thermal powerplants, but they wouldn't be run, and the budgeting for them would be under national defense and interpreted through that lens (i.e. you can buy it down with strategic alliances and diplomacy).

[svnt]

Why wouldn’t the right answer just be don’t grid-tie and switch back to DC for everything. Almost everything gets simpler and cheaper.

[ben_w]

Switching everyone to DC is expensive; for one, we'd have to replace all the AC transformers on the existing grids.

[jaggederest]

You don't need constant draw though. My numbers are based on actual grid usage of a power plant, not the idea that all powerplants have 100% coverage 24/7. It's designed to replace the existing usage of a half gigawatt peaking plant. Zero of them run at midnight, most only run for shoulder hours in Cali/Texas. They're the majority of the new generation being planned right now, so if the drop-in costs are less, they'll be replaced with solar. As the costs further go down, more and more plants and base generation will be replaced with solar. I expect solar to get cheap enough that e.g. thermal sand storage starts to be viable, which will use essentially the same infrastructure as a nuclear plant but without the NRC approval, radiation safety controls, and backup infrastructure.

> I've never been able to find a way to square an actual "no fossil fuels grid" with the supposed cheapness of solar or wind - it always feels like people are quoting selectively useful $/GW values and then not giving a full accounting of the assumptions behind them - i.e. GW type quotes originate with thermal powerplants which have capacity factors which are essentially "whatever you want if you pay us".

I'm telling you right now that LCoE for replacing natural gas is here. Coal has been dead for a while, new hydroelectric plants have massive siting concerns, wind is already too expensive compared to solar + battery, oil has been dead this entire century for electricity. What else is left? Only marginal things like geothermal which are entirely location-based.

https://www.eia.gov/todayinenergy/detail.php?id=61424#

The grid will be solar. Very soon, in fact. Within 20 years, which is lightning fast in grid terms.

[XorNot]

The comparison you're drawing though is still hyper-optimized to a specific use-case: a marginal peaking plant i.e. an infrequently used one with a very low required capacity factor (and also presumably operating right at the peak of wholesale pricing on the daily cycle).

But that's a utility being injected into a grid which already has widespread stored-fuel powerplants. I'm not contesting batteries work under some circumstances, I'm contesting whether they actually work when they are doing more then displacing load-handling at the edge. The grid runs 24/7: there's a massive difference between running batteries for 2 hours, and then recharging because you can buy power any time of the day you want, versus their being near zero dispatchable generation on the grid.

Because a gas generator is more then happy to sell you power and run a little longer to do so at any time of day. If that gas generator doesn't exist though, then once your battery is empty it's empty until the renewables pick back up. And that's the answer I'm still not seeing - the question isn't "can you optimize the grid" the question is "can you eliminate stored-fuel power plants entirely". It's fairly obvious that batteries can help in some circumstances given that gas plants have start up times in the tens of minutes, and power prices going negative is bad for them.

EDIT: Basically, are we actually displacing any fossil fuels off the grid, or just optimizing it's expansion - given that an infrequently used peaker plant can become a frequently used peaker plant quite easily, but a solar farm can't do the same.