Solar generation programming #154
Description
Description
I really want to switch to RoomMind and leave my custom build nodeRed HVAC controller behind, but it seems to be missing the option to keep solar, wind or other kind off power generation (power overflow) in mind.
My current setup will precondition rooms to a preffered target temperature when there is enough overflow power available (reported by our P1 monitor).
This because power being send back into the grid is wasted power because the return pay is so low it's better to just use it directly for the house (even when no-one is home)..
This automatically keeps the house warm in the winter or cool in the summer minimizing drawing power from the grid relying fully on own generated power.
How I currently programmed it is that each room is given a priority (sequential turn on/off based on said priority) and the hvac max power rating in Watts, the system monitors a power usage/return entity (5m average, 30 datapoints per minute) and when there is an initial of 300W + 15% of the aircon unit max power rating available in the overflow the unit is turned on (based on priority), the system then waits a minute for the power draw to stabilize before turning on the next unit if 15% of the next unit its max rating is available in the overflow.
Another function monitors the overflow and moves the hvacs target temperature closer to the actual room temperature (a correction is allowed every 30s) to reduce the power demand of the currently running units and to maximize their runtime before shutdowns are required because the power generation is no longer sufficient (I target >200W before initiating sequential shutdowns, keeping in mind minimum runtimes and cooldown timers before allowing them to be turned on again).
Moving the target temperature closer to the room almost immidiatly lowers the power draw during temporary cloud coverage for example (5m average also helps here) preventing full shutdowns when the sun is behind a cloud for example minimizing grid draw.
This then goes on fully on its own when enabled per zone and either keeps the room at their current temperature (power generation is to low to increase/lower it but enough to maintain) or brings them at the preffered target and keeps them there all day long (assuming there is enough overflow) preventing the use of extra electricity/gas when the programming kicks in to heat/cool the house before you come home from work because the house fully regulated itself almost for free during the day.
Also when the programming eventually turns off rooms will keep their temperature much longer because not only is the air in the room at temperature but the walls/floors too, preventing quick rise/fall of the temperature when the hvac shuts off.
I tried creating a model that could predict how fast a room would heat up/cool down and incorporate that somehow into the priority/target settings, but unfortunately my skills where not yet high enough to leverage machine learning on history data to get a actual model per room which I then could inference to get solid predictions of the expected delta based on the current weather and what would be needed to heat/cool a room, kinda got it to work but could not trust it's predictions and got to complex for my brain..
So yeah, props to see it incorporated in roomMind!
I hope this makes sense.. ><
Use case
Automated hvac control based on power generation.
Alternatives considered
No response