I previously shared your impression but this report seems to suggest otherwise; see p24 in particular:
> • Battery:PV capacity ratio always at 100% in HI; lower on the mainland (but increasing over time—see bottom right graph)
• Storage duration ranges from 2-8 hours; 50 of the 61 plants have 4-hour duration (other 11 are 5x2 hr, 1x3.7 hr, 4x5 hr, and 1x8 hr)
And there are some detailed case-studies that give other examples too:
The first one (Pine Grove substation) gives a sort of "emergency button" to provide 12 hours of load relief, when they call it in. This seems like it's basically just shaving off extra load that occurs on particularly busy days (~40/year), it's not solving the core demand-curve mismatch. (It looks like the batteries function as essentially very-short-term demand smoothing/arbitrage when they are not being called in).
The second one (Wheatridge) is 4H of 30MW (~10% of the power of the whole facility); I don't have a feel for whether this is closer to the full demand-curve mismatch.
Edit to add: There's actually another case study that might be even better, Slate PV + storage plant in CA, which is ~50% power for 4 hours, which is very substantial.
I think you're right that you can't store much energy overnight -- but that's not really required I believe. My understanding is that if you have a mix of solar and wind in your grid, you tend to be fine overnight since it's always windy somewhere at night. The challenge for solar&wind is supplying the afternoon/early-evening peak.
Previously you'd conceptually have "baseload" which is sized at the daily minimum, and then "peakers" (or these days just rapid-dispatch gas) which are turned on as needed to supply the daily peaks. With solar this isn't an option; the supply curve is fixed. So you either have to massively overbuild to have enough generation to meet the demand peaks, or have some way of shifting the peak solar generation (midday) to the right, to match the peak utilization. So I think your storage ends up needing to look more like "store 25% of the midday generation to be used by 6pm" (numbers made up for the sake of example). Which seems to be the OOM that these projects from the report are achieving.
> • Battery:PV capacity ratio always at 100% in HI; lower on the mainland (but increasing over time—see bottom right graph) • Storage duration ranges from 2-8 hours; 50 of the 61 plants have 4-hour duration (other 11 are 5x2 hr, 1x3.7 hr, 4x5 hr, and 1x8 hr)
And there are some detailed case-studies that give other examples too:
The first one (Pine Grove substation) gives a sort of "emergency button" to provide 12 hours of load relief, when they call it in. This seems like it's basically just shaving off extra load that occurs on particularly busy days (~40/year), it's not solving the core demand-curve mismatch. (It looks like the batteries function as essentially very-short-term demand smoothing/arbitrage when they are not being called in).
The second one (Wheatridge) is 4H of 30MW (~10% of the power of the whole facility); I don't have a feel for whether this is closer to the full demand-curve mismatch.
Edit to add: There's actually another case study that might be even better, Slate PV + storage plant in CA, which is ~50% power for 4 hours, which is very substantial.
I think you're right that you can't store much energy overnight -- but that's not really required I believe. My understanding is that if you have a mix of solar and wind in your grid, you tend to be fine overnight since it's always windy somewhere at night. The challenge for solar&wind is supplying the afternoon/early-evening peak.
Previously you'd conceptually have "baseload" which is sized at the daily minimum, and then "peakers" (or these days just rapid-dispatch gas) which are turned on as needed to supply the daily peaks. With solar this isn't an option; the supply curve is fixed. So you either have to massively overbuild to have enough generation to meet the demand peaks, or have some way of shifting the peak solar generation (midday) to the right, to match the peak utilization. So I think your storage ends up needing to look more like "store 25% of the midday generation to be used by 6pm" (numbers made up for the sake of example). Which seems to be the OOM that these projects from the report are achieving.