When it snowed yesterday afternoon, it was already clear that the PV modules would be covered in snow today and that virtually no PV power would be generated. However, DESS continues to expect PV production, so that the planning deviates greatly from reality.
It would be ideal to have a way of informing DESS that there will be no PV production the next day, be it due to snow, maintenance work on the PV system or for whatever reason.
And this is the result. If DESS had been informed the evening before that there would be no PV production, it would certainly have recharged from the grid at a favorable time in the early morning.
Good idea to have the possibility to give manual input for the forecasts.
Even for the consumption forecast this would be nice.
Something like “tomorrow in the evening I want to use 4kW additional between 8pm and 10pm”
The problem is the more individual the DESS is designed. The more alternatives and input options there are, the more complicated it becomes. I can completely understand the basic idea.
For this purpose, the Hack of @dognose allows an “Adhoc Charge”.
Currently I’m testing a Node Red flow that will activate an Adhoc Charge between the hours I manually provide.
A better solution would be to override the DESS schedule, but that’s something that will take a lot more time than I have available at the moment.
Nice Idea.
I’ve noted that there is also an option to reset the solar forecast in VRM:
Settings → General → scroll to the bottom.
If you’re feeling lucky, you could try if that would work out. VRM should immediately start to pick up “close to 0” data for any subsequent forecasts and therefore eventually work better on scheduling grid2bat.
but ofc, when the system becomes “snow free”, there will be a certain time of “underestimated solar”, not sure if thats an issue at all, or if another reset would help to remove the 7-days-of-0 from the considered time frame again.
And that’s the reason why I don’t want to reset the forcast completely.
The system should not forget what it has learned so far. But it should be possible to tell the system, that there will be snow within a definable period of time, at least for the next day. Outside that period, the system should continue to use what it has learned so far.
