One More Detection Challenge

[kevin1]

I wanted to highlight one more detection challenge that Sense end-users might encounter - cases where Sense AI or EV detection gets the detection right, but misses on the actual power value. Here are two examples:

1. Variable HVAC - here’s a case from Ed Sellers, where Sense identifies the on/off transitions and flags as detections, but uses constant power (top- from Sense power meter) instead of seeing the variable power (Bottom - accurate measurement). I suspect this is a case where Sense has relaxed HVAC detections to trigger on a broader range of “signatures”, but is stuck seeing each one the same way.

2. The same situation for EV detections. As I mentioned in an earlier posting, Sense EV detections are done differently than Sense AI, but are subject to the same issue. I’m guessing this is more related to variability in charge rate. I was charing our old Model S the other day at 80A, and noticed this strange phenomenon in the Bubble chart. At 80A and 240V, the Model S charges at around 19kW, but the Sense detection is only picking up about 7.4kW.

When I look at the same data in Home Assistant/InfluxDB, where I can compare the Sense detection against the actually reporting of charge rate from the car, I can see that Sense nails the detection, but gets the value wrong. The white line shows the same point in time.

[dannyterhaar]

Since EV charging is such a high amount of power compared to other loads, how useful is Sense in that case ?

I am keeping track of daily miles in a spreadsheet.
Copied from the display in the car, I have used 450kWh from the battery in May.
Let’s add 10% inefficiency of the onboard charger that would mean about 495 kWh fed from L2 to the car.
According to Sense:

37% of my EV charging energy detected by Sense.
And that is the largest electric user I have by far.

I have given up on the idea that this unit is accurate/usable for me