I really think you are over-complicating it, especially if it doesn't matter dollar-wise when you save the power. I'd definitely not even mess with it at that granular/complex level. I'm not seeing where moving output from time X to time Y is going to accomplish much.
Maybe you're right and I may over-complicating things.
BUT:
I don't care about the situation dollar-wise, especially since I have a grid base plan where it doesn't matter
when I consume the grid energy. If I need a specific amount of battery charge to make if over the night, I still think it's more wisely to previously let the batteries charge from x% SOC to 100% SOC instead of let them charge from nearly zero SOC to y% SOC (with the same amount of stored energy) before the night starts. This will help the BMS to reset the SOC counter which will usually drift away >10% after 2-3 days not fully charged. Also the batteries are not always discharged to a very low value (which may also have a positive effect).
Yesterday was a bad solar day, about 70KWH down here with the clouds in the valley. GF decided it was a 5 loads of laundry day, and it was brutally hot. I kicked to the grid at 22:00 and back to the inverters at 07:30, first time since February. It's gonna cost me $2-3 to do that effing laundry (she did 5 the day before as well, deep into the evening on battery, I mean REALLY?). Just suck the batteries dry (to some threshold) and flip-over to the grid.
You start with empty batteries, at sun-up you flip the inverters on once the batteries reach a minimum threshold and run all your loads, collecting excess production in your batteries during the day when the sun is shining. When the sun goes down you use that collected energy the rest of the evening possibly into the next morning, possibly until solar production ramps up. If you are short, you may shut down. Your grid exposure is that window however small or large it may be, between shutdown and solar ramp up.
My goal is to fully automate the solar system in a way that absolutely no manual interaction is required. I know that's possible by just using the settings for BackToGrid and BackToBattery as they are designed - but with the mentioned drawbacks.
When your batteries get to 100% you leave power on the table.
That's right, so the goal is to find a grid usage time frame to let the batteries just hit 100% but let the PV production not clipping.
Thus it behooves you use whatever you have stored if needed, to the point of being dry once solar production starts in the morning to maximize storage fill. To my earlier comment, you do not seem to be getting to 100% with similar capacity to me, which I think is disappointing, none-the less I don't see where fiddling with saving battery capacity for use the next day is going to have any significant impact on your electric bill.
As I already mentioned, it's not the goal to reduce the bill with this sort of fine tuning - it doesn't matter for me
when I consume the required grid energy.
I need an average between 105 to 120kWh per 24h with the current temperatures here. But my solar array is only able to produce a max. of 90kWh net usable energy (120kWh with 75% total overall conversion losses) with full sun. So it's clear that I need between 15-25kWh energy from the grid every day (even more on cloudy days) and I have to pay the price for it. I think my daily required energy is higher than yours.
I know, I need to extend my PV size by adding about 10kWp more to my existing 18.4kWp array to be totally independent, even in the hottest summer...
If you were trying to keep a reserve for a power outage/failure, the simplest thing is to raise the "dry" threshold with enough power to handle the number of buffer hours you want. Further you could set your grid-down threshold much lower, say 20% cuts to the grid. If the grid fails, allow the inverters to run it down to 5%, this gives you 15% of your battery in the event the grid fails during non-solar production / low battery hours. Tweaking this window could save you a few pennies. Literally pennies, but for fun maybe.
In the above mentioned two scenarios (let the batteries charge from x% SOC to 100% SOC instead of nearly zero SOC to y% SOC) I have a higher chance to have a reserve in the batteries when I try to keep them between X% and 100% instead of using them on the low end.
At the core I see only two thresholds:
1) Minimum SOC to allow inverter to turn on
2) Minimum SOC at which point the inveters should be off
I'd put at least a 5% gap to keep a cloudy morning from toggling the system off and on excessively.
All that being said . . . I would put an independent sensor/logging device on your mains as a sanity check, if you don't have it already. I rolled my own with timestamps and code, you may want to just put something on the inverter output bus to see if your getting honest numbers.
I measure the whole house energy very accurate since some years and have the data in my smart home system (via Aeotec ZW095 split phase home energy meter, ZWAVE based). So I don't need to rely on the AIO load data which I get in parallel from the SolarAssistant integration and are usually not very accurate.
I'd also spend some effort on your panel output I really think it's low. Do you have shading or anything? It's not yet 12:00 and both banks of batteries are well over 50% from 10 and 8 at 06:30 this morning. Left my EV plugged in, drove the GF's to work, charging will not start until I hit my SOC target based on time of day, may not get there today, might get 10 or so KWH before 17:00. In this case I manage the load, not the power plant.
I have no shading - but the panels are not optimal oriented because of the roof geometry.
The panels have a max. spec power of 18.4kWp (STC) and 13.6kWp (NMOT) but my usual peak is around 14.5kWp (at around 12:30pm) on full sunny days. I think this is not bad for outside (shade) temperatures between 100 and 120 Fahrenheit which we now have here. This is about 78% of the spec max. under STC conditions but 106% of the spec max. under more realistic NMOT conditions!
Usually the PV can add 42% of charge between solar production starts and noon. But the charging rate gets a lot less later in the day because the energy consumption rises a lot. Between 1pm and 6pm I have an average and more or less steady consumption of about 8kW (a central 5 ton heat pump, 2 mini splits (1.5 ton together), a window AC unit (1/2 ton), 1.5kW pool pump, etc.). As a consequence, the SOC it's increasing slower and slower between noon and 3pm and it will drop after 3pm because the batteries are required to be used more an more to cover the load.
Until now, I have not implemented a real smart load management (beside emergency power saving rules like shutting down the pool pump even if it's not running long enough, disconnect the hot water heater, increase the setpoint temperatures of the AC's a bit - these emergency power saving modes only getting active when it's foreseen that the energy will not be enough AND the grid is down).