One 8,760 line Excel spreadsheet later, I have your answer: On 51 days in calendar 2021, my solar production was less than the sum of my partial and peak consumption. So I would not have had enough stored to discharge throughout peak and partial peak periods. We are in sunny (but sometimes coastal overcast) Oakland Calif, so 13% shortfall days is to be expected, 51 out of 356. Over that year, our solar production was 97% of our consumption kWh.Just so I understand your situation; how may days last year did you experience where your solar was unable to charge your batteries to carry you through the peak time(s)? Will you really see much benefit grid-charging your Powerwalls?
The cost savings is probably less than $100 per year. But since my True-Up is in the $100 vicinity (by design), small improvements result in large percentage changes in the final bill.
To find that 51 day answer, I downloaded the SAM-8760 file, (365 X 24 lines of hourly consumption data) from my Enphase micro inverter system, as well as daily solar production for the year. An afternoon on Micro$oft's support site and a few inglorious kuldges got me a "pivot table" showing the count of 51 days to answer your question. A sunny day with low consumption might leave a high PW state of charge, so the exact kWh saved would need an even more complicated calculation. Hence the SWAG <$100 above.
Folks with less solar and/or higher consumptions (think heat pump winter heating) would see much more savings, of course.
SW