Ryan,
Given that this is an old thread, let me add some general suggestions that Sense can use to create some machine learning guidance on EVs.
The chargers in all modern (post 2006 cars/trucks) are all power factor corrected, meaning there is a perfectly resistive looking load on the line. It has not inductive or capacitive component to it.
All chargers slowly ramp up to full current, whatever that will be, over the period of 5-30 seconds. No clunk, ringing waveforms, overshoot with high current transients. I know your 1 Megasample per second A/D will catch those kinds of things with motors, like older refrigerators etc.
The machine learning aspect should realize that after several HOURS of steady consumption, this is not and oven, probably not a hot tub (since the connection is made typically in the evenings after 9-10 PM, when electric power is cheap and plentiful.
When the duration parameter is logically AND’ed with a high current level (>25 amps)
machine learning can conclude it’s an EV. No other load charges for hours without cycling like a heater once reacing setpoint. A hot tub will blast away until a reasonable temperature is reached. Most hot tub are heated with gas not electricity, but there are always outlying exceptions for “solar powered mansions”.
Many EVs taper their current draw as the battery fills up. The new Chevy Bolt EV starts that taper at the 50% full point, and slows down the fill rate (reducing the current gradually until the thing is done. Many owners stop their at 80% or some “nearly full” condition. Perhaps Sense employees can connect with Bolt EV owners to characterize things, especially since an EV is the largest single load in the household in most cases.
Training Sense to use these will go a long way to satisfy your growing clientele which are adopting EVs by the thousands. There are happily over 1.2 million plugin cars on US roads today. Not all are plugged in daily, some not at all (like municipally owned Volts, where there is no accountability to do so. There have been Volts for sale that have hardly used the on-board charging equipment at all during their first ownership!
Hope this helps. We, as a community, are looking forward to getting our energy consumption logged and correlated correctly!
Ron
@freund.ron,
As an owner of 3 EVs, I appreciate your analysis and applaud your inputs. I would be surprised if Sense hasn’t already been running a duration ANDed with high balanced 240A current level to tag those usage regions for the EV data science guys so they have concentrated, tagged data to train on. I think that team is still searching for the elusive early (first minute or two) on-signature, and really can’t use the duration / current criteria as an input feature since is is so time-delayed from the needed detection. And the folks at Tesla don’t make that super easy for Sense since they do tweak EV charging profiles over time with OTA updates.
Good news: We’re rolling out revised Chevy detection models which should help to resolve issues with detected charge ending early, duplicate EV devices, and provide more accurate tracking. Models will be slowly rolling out over the coming days, so you will likely not see immediate improvements. As always, your mileage may vary depending on other factors in your home.
New models don’t roll out to everyone immediately, so it’s possible they haven’t rolled out to your home. Even after they roll out, Sense will still need to see more charging cycles since it has been refined to look for different things. Keep checking and let me know if it gets detected within the next week or so.
Hi Ryan - so Electric Vehicle showed up the day after you said to look for it - However it doesn’t seem to be very accurate. When compared to what my Charge Point app shows as its usage, it doesn’t seem to match up with the app very often. Do you know if you plan on integrating the api’s from the actual charger for more accurate readings?
Ill have to keep an eye on it as it goes Hopefully over time Sense will add more integrations with ChargePoint, Juicebox and other smart devices such as Casetta light switches and such so the app becomes even more useful!
Now i just need to figure out what 2 “pumps” are and where they are lol they are energy hogs
Hi @brucewhitneyvt - if you are consistent in charging frequency and time that you charge your EV, this should help Sense build the device model for your EV. If this issue persists, please reach out to us at support@sense.com so we can take a deeper look.
hi Justin -
During the winter months I plug in nightly. The car is programmed to charge during non-peak hours (I have TOU rate) 9 PM to 1PM following day and charge just prior to departure time ( 8 am weekdays) so frequency and time is generally consistent. However, the hot water heater is often coming on in the same time period because its on Aquanta and has TOU times and knows my shower timing is generally 7:45 :).
I also have cold climate heat pump which I would love to get detected and that has not happened yet.
We have an ERV running 60% of the time - would that being making detection more difficult?
It’s possible that the water heater and ERV are interfering, but difficult to say without actually seeing your data. I’d suggest reaching out to the Support team so they can take a closer look.
The EVSE needs a ground but not a neutral when operating with 240V. There are several safety checks an EVSE makes to ensure safety. Here is a document that describes the checks made by and EVSE.
My Bolt EV is detected fine, but then disappears as the charging rate begins tapering down at a out 75% charge. Any suggestions to help Sense understand?
Thanks,
Joe
What year do you have? I think starting with the 2019 it does a general taper and 2018 and below used a staircase dropdown. But… I think that was changed with the recent recall that applied a software update to try and prevent the fire issues and all the Bolts with the update now perform a taper. I’d imagine that the taper looks slightly different though because the battery capacity is higher in the later model Bolts.
I’ve only had my Bolt since June and don’t have an L2 charger so this is all just conjecture. I do plan on getting an L2 charger at some point, but for now L1 charging and using an HS110 is adequate.
Any Gen2 Volt owners getting varying levels of detection based on a manually started charge versus the rate-based automatic started charge?
If I plug in manually and start a charge immediately (it’s either an off-rate period, or I override regardless) I seem to be getting solid detections. However, I’ve noticed more often than not if the car is plugged in earlier in the evening but then automatically starts charging at 7PM (When our electricity rates go to off-peak) I’m not getting a detection - this means that the entire ~15kwh charge ends up in “Other”.
Hopefully over time Sense will add more integrations with ChargePoint, Juicebox and other smart devices such as Casetta light switches and such so the app becomes even more useful!
.