TL; DR - Pulling a snippet of a quote from another thread ... I was wondering if anyone else has experienced this regen hysteresis, and may have gathered data via the API or some 3rd party service like TeslaFi monitoring your drive? or better yet perhaps even the CAN bus?
I recently did a test during a stretch of my commute where I decelerate significantly coming off the highway, and then also have a fairly decent stretch of downhill section.
Observations
I was able to recreate the aforementioned hysteresis to the extent that I did not see the battery range miles (to two decimal places) in the API increase (nor decrease) for a stretch of ~3 miles of driving over ~4 minutes. In fact, it was close enough to exactly 4 minutes that it made me wonder if it might be linked to time and not amount of energy. I had the API being polled every ~10s over the course of 10 minutes to capture this ~4 minute stretch. This stretch was ~1 miles of mainly regen, followed by ~2 miles of driving in the flats ~47mph.
Theory
What I suspect is happening is the infamous "energy buffer" (which is really just a number they track internally, not a separate area of the battery) is incremented when regen is happening, and this only gets transferred from the buffer to the battery gauge displayed range if the buffer gets "too full" for whatever amount that may be (seems an absolute max might be 4 kWh for Model 3 based on people looking at CAN bus, if this is what that is, and then what delay you see would depend on how much was in it to start with, and what their threshold is before 'transferring'. e.g. if they are targetting 2 kWh in the buffer +/- 1 kWh at any time, and you regen 1 kWh, then you'd see that 'transfer' over once the buffer grew from 2 to 3 kWh. 1 kWh would be about 4-5 miles depending on which trim you have).
Likewise, I suspect the opposite happens where, if the energy buffer is "a little bit too full", then when you drive and use power, the power may come from the invisible energy buffer rather than get subtracted from the battery range display.
This seems to be what is happening here with long stretches of regen anyways. While you regen, no range is added so the energy must be being counted somewhere. Then after you stop regen and start using power, no range is being subtracted, so again the energy needs to be being subtracted from somewhere. It seems likely or at least possible anyways, that it is happening for all regen, not just large amounts of regen. This leads me to think that trying to compare (a) changes in displayed battery gauge miles over a trip, with (b) "energy used" (measured by whatever means*) would seem to always have some potential for error of anywhere from 0 to 4 kWh!
RFC/D (request for comments and/or data)
Anyways, I was wondering if someone had some good data on this ... or if I asked the question if someone would go looking through their data for more data points
Maybe from TeslaFi ... or someone snooping the CAN bus?
*Footnote on 'measuring' energy used:
When I say by "whatever means" of trying to 'measure' energy used, this could be, e.g. (a) a recharge event measuring how much AC or DC energy is required to replace the energy used by the trip, or (b) trying to divine the DC energy consumed as reported by the trip meter in distance and consumption (X mi * Y Wh/mi = Z Wh used).
For the latter, (b), I believe that the underlying numbers used to display the trip stats to the user DO get updated during regen unlike the battery range which appears to stall for stretches. I have seen the trip consumption drop significantly after short regen events, especially early in a trip, when these changes would be more noticeable.
I think there is likely an underlying watt-hour-meter used for this along side the odometer to report trip statistics. Or perhaps it is two watt-hour-meters, split into two halves to separately count power delivered from the battery and regen power sent back to the battery.
Tesla Bjørn has a video where someone with a laptop was looking at CAN bus data from his car, and one of the stats I thought I remembered hearing him list off was lifetime regen energy... Turns out I looked up the video ... and it was total AC and DC charging he claimed they could see (for cars after 2016), and if you subtract that from total charge energy which is also available, you are left with the difference being regen. They said ~15% regen is typical for Norway.
Anyways.
Post your anecdotes or your data from the API or CAN bus! or links to such info!
I recently did a test during a stretch of my commute where I decelerate significantly coming off the highway, and then also have a fairly decent stretch of downhill section.
Observations
I was able to recreate the aforementioned hysteresis to the extent that I did not see the battery range miles (to two decimal places) in the API increase (nor decrease) for a stretch of ~3 miles of driving over ~4 minutes. In fact, it was close enough to exactly 4 minutes that it made me wonder if it might be linked to time and not amount of energy. I had the API being polled every ~10s over the course of 10 minutes to capture this ~4 minute stretch. This stretch was ~1 miles of mainly regen, followed by ~2 miles of driving in the flats ~47mph.
Theory
What I suspect is happening is the infamous "energy buffer" (which is really just a number they track internally, not a separate area of the battery) is incremented when regen is happening, and this only gets transferred from the buffer to the battery gauge displayed range if the buffer gets "too full" for whatever amount that may be (seems an absolute max might be 4 kWh for Model 3 based on people looking at CAN bus, if this is what that is, and then what delay you see would depend on how much was in it to start with, and what their threshold is before 'transferring'. e.g. if they are targetting 2 kWh in the buffer +/- 1 kWh at any time, and you regen 1 kWh, then you'd see that 'transfer' over once the buffer grew from 2 to 3 kWh. 1 kWh would be about 4-5 miles depending on which trim you have).
Likewise, I suspect the opposite happens where, if the energy buffer is "a little bit too full", then when you drive and use power, the power may come from the invisible energy buffer rather than get subtracted from the battery range display.
This seems to be what is happening here with long stretches of regen anyways. While you regen, no range is added so the energy must be being counted somewhere. Then after you stop regen and start using power, no range is being subtracted, so again the energy needs to be being subtracted from somewhere. It seems likely or at least possible anyways, that it is happening for all regen, not just large amounts of regen. This leads me to think that trying to compare (a) changes in displayed battery gauge miles over a trip, with (b) "energy used" (measured by whatever means*) would seem to always have some potential for error of anywhere from 0 to 4 kWh!
RFC/D (request for comments and/or data)
Anyways, I was wondering if someone had some good data on this ... or if I asked the question if someone would go looking through their data for more data points
Maybe from TeslaFi ... or someone snooping the CAN bus?*Footnote on 'measuring' energy used:
When I say by "whatever means" of trying to 'measure' energy used, this could be, e.g. (a) a recharge event measuring how much AC or DC energy is required to replace the energy used by the trip, or (b) trying to divine the DC energy consumed as reported by the trip meter in distance and consumption (X mi * Y Wh/mi = Z Wh used).
For the latter, (b), I believe that the underlying numbers used to display the trip stats to the user DO get updated during regen unlike the battery range which appears to stall for stretches. I have seen the trip consumption drop significantly after short regen events, especially early in a trip, when these changes would be more noticeable.
I think there is likely an underlying watt-hour-meter used for this along side the odometer to report trip statistics. Or perhaps it is two watt-hour-meters, split into two halves to separately count power delivered from the battery and regen power sent back to the battery.
Tesla Bjørn has a video where someone with a laptop was looking at CAN bus data from his car, and one of the stats I thought I remembered hearing him list off was lifetime regen energy... Turns out I looked up the video ... and it was total AC and DC charging he claimed they could see (for cars after 2016), and if you subtract that from total charge energy which is also available, you are left with the difference being regen. They said ~15% regen is typical for Norway.
Anyways.
Post your anecdotes or your data from the API or CAN bus! or links to such info!
