So, I know we can’t easily train the sense. Is there anything I can do to account for known always on items?
first example: I have a hardwired radon ventilation system that runs 24/7. No plugs. It is a pretty fixed draw, sure it’s a motor and does vary a bit but it is quite reliable. Can i create a fake device in sense and assign it what I know to be that motor’s energy consumption? I know sense won’t monitor it, but at least I can label a portion of my unknown always ons. I can toggle the breaker it’s (by itself) hardwired into and determine energy change to measure it.
Second example: we have outdoor lights on a timer. They’re actually zwave switches on a specific schedule and using led bulbs (typical lights outside a home). I know exactly the energy usage from 6p-6a that they represent. I’d love to quantify that. Not sure if you’d pull them out of always on to their own bubble, or divide always on up into pie slices and shade the ones we claim to know than sense can’t ID yet?
We don’t have any functionality like that in the app at this time (outside of using smart plugs, at least, but as you note those aren’t applicable to all loads). However, we’re doing a lot of thinking currently about Always On and we’re looking into possibilities along the lines of what you’re suggesting.
Always On is a magic function that Sense continues to fine tune. It represents the near low point of usage in your house over a long period of time (now 48 hours) minus any contributions to the power usage total at that point in time from smartplugs (that power is associated with the idle of the device on the smart plug). I say near low point, since occasionally Sense might drop some data and it will look like your house is using zero energy for a few seconds, so it uses a number that is statistically far enough away from the lowest to ensure it is a real minimum usage value. Always On changes steadily based on a rolling calculation.