Just as I was doing my end of the year energy assessment, a new “Electric Vehicle” detection popped up. Somewhere mid-year I deleted my previous Model 3 and Model S detection since they were not terribly accurate, and wrote my own custom EV charge detector, that really only works for my own particular situation. About Oct. a new Model 3 detector from Sense showed up, and has proved to be much more accurate. So when the new Electric Vehicle detection showed up (or was it a revisit of my earlier Electric Vehicle model), I decided I had to do a comparison to see if Sense and EV detections had gotten better.
Here’s the comparison in a bunch of 2 week chunks, backward in time. Red is my Total Usage, blue is the output of my own EV Charging detector (it doesn’t discriminate today between Model 3, Model S, and both, though it could). Green is the Sense Model 3 detector and orange is the Sense “Electric Vehicle” detection which appears to be my Model S. The other interesting thing is that the Tesla Model 3 only has forward-going data, detection that appeared after the detection was added. The Electric Vehicle detection seems to have backfilled for the entire year, leading me to believe it might be a return of the same model.
Looking at the last full two week period in the year, I can see a few things. The Model 3 detection does quite a good job, perhaps even a little more accurate than my detector in two different ways. Mine over estimates the power usage since it just uses the total consumption - not a bad approximation give another usage at 3AM. And it also looks like Sense is better at catching the new Tesla Model 3 pattern that seems to include a “breather” in the middle of charging as shown on the far right. Sense is also doing a good job of detection the Model S charging, though it predicts a far lower power usage than is really happening. Here at the end of the year, Sense is “seeing” about 5kW going to charging when in reality, the number is closer to 20kW.
Back another 2 weeks, it looks like Sense missed the Model S charge on the far left, but caught all the other charging cycles by both the Model 3 and Model S (I’m ignoring the short minimal charge cycles). Notice that Sense cough the back to back charges of each car on the far right.
Here’s an interesting 2 week period. Sense caught most everything except for the Model S charge on the far right, successfully nailed a back to back charge in the middle, but then fabricated a confusing Model S change on the far right.
Two weeks earlier, Sense looks to have done a pretty good job. This is where the Model 3 detection started so we part ways with the green…
I had a few issues with my Sense Monitor in Oct., so I suspect those issues may have contributed to less good detection. I’ll skip commentary on these two graphs.
Now we’re getting to an interesting area. In the next two graphs, the Model S detection seems to actually change. The one for the more recent 2 week period shows about 5kW per detection. 2 weeks earlier, the detection value was about 10kW, which is still about half of what the actual power is. Bottom line is that I think I’m seeing my old 10kW model/detection ending the week of Sept 16, being replaced by a 5kW model (at least today).
And going further back, it looks like the 10kW model runs back to the beginning of 2020.