Let's say I get a 20kW (DC) solar setup up and running. Let's also say that I'm able to directly charge the 85kWh battery with this DC power with some decent amount of efficiency. We'll say 85% charging efficiency for now, which I think is very low and safe guestimate.
20 * 0.85 = 17kW usable DC power. Where I plan to setup this system, calculations come out to an average of 5.3 hours per day of peak output. Bit higher in the summer, bit lower in the winter.
So, 17*5.3=90.1 kWh. 90.1kWh. Let's jump down to the winter number, say, 3.5 hours/day (about 1/3rd loss). Still 59.5 kWh/day. That's almost 1800 kWh generated per month in the winter and around 3000 kWh in the summer (when, coincidentally, it will be needed more). In the winter I don't think my personal use will spike much over 2000kWh at the most, so, this is fine. In the summer a good chunk of daytime output will likely go to climate management. If it looks like power would ever be wasted (full pack) I'd probably have the system setup to dump power into the A/C and bring the home temp down a notch or two to make night time usage of the battery a little less.
59.5 kWh/day comes out to an average constant load of just under 2.5kW. Even using only 80% of the SoC range on an 85kWh pack, it would take about 27 hours to run out of power with zero power input, starting with a full charge. I doubt that the array will ever generate zero power in a 27 hour period. Even 20% of expected in the winter is 11.9 kWh which would be almost an additional 5 hours of run time for 32 hours total. At that rate let's say we're full near sundown on day 1. By sundown on day 2 we've consumed 48.1 kWh net, or 70% of the available power. That leaves another 8-ish hours before the pack would be below the cut off point. Say, 6PM to 2AM remaining.
At that point, when the battery hit the cut off point, my idea would be to have a ~20kW AC-DC grid charger kick in to bring the pack back up a bit. Preferably this setup would be smart enough to estimate the amount of time left without sun input and other factors to make sure to not miss out on any solar generated power from the pack being full already. Basically, I'd keep a grid interconnect for just this purpose.
In the summer this whole potential problem is moot since there would normally be enough power generated during the day to more than fill the entire pack by sun down, even accounting for some poor weather days. Winter will be the biggest hurdle, and I plan on actually doing some more detailed analysis of the energy needed during that timeframe and the array size needed to maintain it.
Honestly, I may utilize two 85kWh packs in parallel (or similarly suitable configuration) if needed. I doubt my power needs are quite that high though. Likely less expensive to just make the home more efficient overall.
I'd probably keep a grid connection also with a transfer switch setup to move my HPWC from off-grid to on-grid when absolutely needed, which probably won't be often.
Overall I plan on building up a reasonably complex setup. It will surely take some time, some trial and error, but, I'll get there.
One plan I've been throwing around is to actually start the build out of the system, building everything *except* the solar array, and using grid-tied charging to essentially simulate solar power. That way I can run the system for a bit under different test conditions and determine the exact needs for the array. If it turns out to be too ridiculous (I don't think it will be) then I at least have a pretty awesome backup setup or something that could be used for ToU arbitrage or something else useful later if desired.
I'm in the process of obtaining an 85kWh pack from a salvage now, also.