4 ms·
you misunderstand how the "+4" storage design works for solar. a 1MW solar plant nameplate capacity will only produce 1M for an hour or two a day, the hours be
by cagenut 4y ago
you misunderstand how the "+4" storage design works for solar. a 1MW solar plant nameplate capacity will only produce 1M for an hour or two a day, the hours before and after that peak will be some fraction, the sum total off the area under the curve works out to between 3 and 5 hours depending on latitude and weather[1].
so when you design a utility solar system with "+4" storage, what you're really doing is creating a "one days worth of full production" buffer. that can be used to run the output at a fixed rate while the buffer builds and empties every day (its never that simple but thats the basic principle). for example a 100MW farm with 400MWH of storage can in the simplest sense produce 16MW constantly (all through the day and night). in practice there's plenty of other stuff on most grids so they don't do a full battery cycle every night, but rather use the buffer to be able to meet day ahead and dispatchability contracts for a very cloudy day or a lightly cloudy week.
now don't get me wrong, obviously that still doesn't put it in the same reliability category as nuclear, but it closes like 80% of the gap in practice. its not better, but it is clearly on track to be 'good enough'.
[1] https://www.nrel.gov/gis/assets/images/solar-annual-ghi-2018-usa-scale-01.jpg https://www.nrel.gov/gis/assets/images/solar-annual-ghi-2018...