Personally I think that it's easy to fall into the trap of over-thinking these things and adding too much complexity. I found myself falling into a deep, dark hole this morning looking at Pi servers for Home Assistant before I realised that spending a couple of hundred quid and hours of frustration learning and setting up a system to monitor PV output and automate routines was a fool's errand. The pennies saved will not recompense me for the hours of messing about.I think to stop the battery contributing, I need to set it to charge mode, at whatever %its at, then the Andersen solar mode will take care of charging. I think I'll leave the car set to over night charging, but manually initiate a charge when needed.
I need to run the faffing around a few times manually to work out the algo I want, and will then slowly pick bits to automate as best I can. So many different bits of tech involved I'm worried I'm going to need some kind of windows GUI automation solution with a virtual phone running some of the apps. Only the Tesla really has a proper API access .
As we are in Summer, I let the system charge the 8.2kW battery to 35% on overnight cheap rate power. That then sees the house through the night and the battery ends up at about 10-15% charge in the morning. I then let the PV charge the battery (on a good day its at 100% by 10:30AM). The solar diverter then puts the excess into heating the water cylinder. By then, again on a good day, its probably 1PM. If I am at home I then plug the car in and let the Andersen charger push the excess into the car. Again, on a good day, I can get maybe 10kW in there.
My PV system has generated 40Kw at peak in a day. Normally it's lower than this, but that gives an idea of what I'm playing with (May's total generation was 696kWh, June was 756kWh) . Getting the panels and battery in was the big jump. Everything after that is a game of diminishing returns.