Also, demand and generation are positively correlated as people use less power on cloudy days and less AC in the winter. You can look at how much oversupply and storage people need to live off grid and realize that’s a worst case situation isn’t 7x oversupply or 30 days of storage it’s roughly 3x oversupply of solar and 3 days of storage. At grid scale that’s roughly 6c/kWh for solar plus batteries which last longer the less their used though that’s also a high when you add other energy sources and grid transmission.
The good news is markets are great at minimizing cost functions. In the near term lots of natural gas makes up the difference, but the cost breakdown is such that offsetting 1kWh of natural gas works out even if 50% of solar generation is wasted. Aka 2x oversupply of solar shows up from market forces it’s that cheap. Bring on storage and things start to look very solar heavy even with zero subsidy or price reductions.
People living off grid still continue to consume the products of grid-enabled industries, from manufacturing to farming to logistics to infrastructure. Heating/cooling and lighting a cabin is much more trivial than paving a road coming to that cabin, feeding its inhabitants, building solar panels for them, or ensuring dense enough population to make research and manufacturing viable. In fact, we don't need to extrapolate minor part of personal consumption to the whole society, we have aggregate energy consumption numbers.
"Market forces" can't generate electricity on their own. Instead of taking the bottom-up view, fraught with wishful thinking and unstated assumptions, why not take the top down, working from the potential generation capacity under ideal assumptions? That's exactly what the paper I linked did, and the results are… not great for 100% renewable
6% is already a large fraction of grid demand and that’s an average. It’s critical for finding anything that’s even vaguely accurate.
“Why no take the top down” because we want to minimize costs. It doesn’t matter how much over production happens vs how much energy storage happens, what matters is how much each costs.
If you really want to do a high level analysis you need to cover a wide range of major inputs. Hydro flexibility and the costs to increase that flexibility, as in how much can it ramp up and how much can it average over a year. Next is the actual cost of Wind in each location for various turbines and the time of day for production this varies quite a bit and becomes a massive optimization problem. Next is solar production at each location as well as how panels are aimed in various wealthier conditions, again a major optimization problem. Next actual energy output vs demand, Aka actually modeling what happens to both for various weather events. Next is cost of grid storage capacity, as in how much capital needs to sit around. Next is how much to utilize grid storage as in a charge vs discharge cycle, the first vs second are different for each technology and a mix of several is likely. Next grid interconnect costs as in how much does flexibility cost both in dollars and resistive losses. Next optimized Solar and wind capacity vs grid demand as in at each set of seasonal and weather conditions to minimize costs. Finally changes to costs over time or based on deployment.
It’s possible to model all of that, but in turns out that’s what markets do. The bottom up approach isn’t wishful thinking it’s what actually happens as people build A due to A being profitable up to point X when building B is more profitable etc etc.
I wonder how much dispatchable generation has to be kept around to only have two 12 hour blackouts per year due to lack of generation.
Also, "good grid" means "perfect national transmission", which is a wildly optimistic assumption. The US is currently split into multiple grids, with ties being able to handle less than 1% of their generation capacity.
Moreover, that's just the US. Here is what the link has to say about countries less blessed with landmass and latitude:
> Indeed, in smaller countries, substantial gaps (>30% of demand for >20 h per year; pale orange curves in Fig. 4) remain in systems even with 12 h of energy storage and annual generation that is 3x annual demand.
Or this, directly referring to extensive overbuilding required to meet demand:
> For instance, historical solar and wind resources data in Germany reveal that there were nearly 2 weeks in which dispatchable generation had to cover practically all of the demand because of a period with very low solar and wind power availability (called “dark doldrums”)27. Although with vast enough wind and solar capacity it might still be possible to meet demand in all hours, the required capacity increases exponentially after a point that depends on the renewable resources of that country, and it is this geophysically-dependent point that will largely determine the cost-effectiveness of highly-reliable, renewables-based electricity systems.
[0]: https://pubs.rsc.org/en/content/articlelanding/2018/ee/c7ee0...