Regenerative braking hysteresis and/or "energy buffer" (miles don't tick up or down)

[Saghost]

Thanks.

If you are curious about your numbers for internal consumption constant and battery gauge rated miles “ticks” you can fairly easily calculate this without driving.

To calculate your increment, go to the energy app, and look at the average consumption in miles and projected range in Wh/mi.

Multiply the two to get a number in Wh that’s an estimate of energy left in your pack. Divide that by the rated miles on your display (ideal on an S would be different, check your display settings). This should give you a number, C, slightly lower than where the “rated” line is in the energy app.

If you repeat this 3 or 6 times (for last 5, 10, 30mi and 10, 25, 50km) you’ll get 6 different numbers that are pretty close to each other (x 1.609344 to get Wh/mi from Wh/km). If you average all 6 numbers I think it should be pretty close to a round number, like for the 3, which is 219, 234, and 245 for SR+, LR RWD, AWD.

I had two historical examples captured in photos, and ran the beginning and end of my commute home today. All came out to 319 or 320 Wh/mile, like this one:

Except the bottom of Mt Washington, when I'd been regening down the mountain for 8 miles:

That shows 328 Wh/mile. Or more likely, that shows that at the same 319/320 the car is otherwise running it should have been showing 172/173 miles instead of 168...

Then to get the ticks that rated miles normally update at ...
If you take that internal constant consumption number, C, and do: 100 Wh / C, you will get a number in miles that your display will tend to always tick down at (e.g. 0.48). Oh, actually you have to cut a percentage off of C ... seems to be aprox 4.5%. So multiply the answer above by 1.047. This might be due to factoring in losses in/out of the battery and/or some other shenanigans going on with the energy buffer.

I don't understand what you're trying to say here - I can do that math easily enough (0.313 or 0.328) but I don't know what this number is supposed to represent or how I'm expected to use it.

 

[GregRF]

LR RWD.

Perhaps its 780 Wh, .1% of a 78 kWh pack (75 useable) giving 230 Wh/mi and 325 mile range at 75 kWh. (Though I never got the bump from 310 to 325.

Sorry, should say 1% not .1%

 

[darth_vad3r]

I had two historical examples captured in photos, and ran the beginning and end of my commute home today. All came out to 319 or 320 Wh/mile, like this one:
[…]
That shows 328 Wh/mile. Or more likely, that shows that at the same 319/320 the car is otherwise running it should have been showing 172/173 miles instead of 168...

Exactly. The regen miles didn’t get added yet so a slightly higher number would give you the same 319 or 320 Wh/mi.

I don't understand what you're trying to say here - I can do that math easily enough (0.313 or 0.328) but I don't know what this number is supposed to represent or how I'm expected to use it.

Ya not much to do with it or use it, but for someone with access to see the API battery range, I would expect to see the miles tick down in ~0.33 mi increments which would represent the underlying pack having been estimated to have used 100 Wh (0.1 kWh ticks).

We’re just trying to understand how it works under the hood, as best we can figure by looking from the outside

 

[Saghost]

Exactly. The regen miles didn’t get added yet so a slightly higher number would give you the same 319 or 320 Wh/mi.

That's after regen had added 22 miles of range continuously over the previous 25 minutes, though. Not sure what the internal logic is, but it definitely dumps it in blocks of 3 or 4 miles, and according to this there was more than the typical block that it "hid" and hadn't shown me.

 

[darth_vad3r]

That's after regen had added 22 miles of range continuously over the previous 25 minutes, though. Not sure what the internal logic is, but it definitely dumps it in blocks of 3 or 4 miles, and according to this there was more than the typical block that it "hid" and hadn't shown me.

Ya 3-4 miles matches my expectation that they add it 1 kWh at a time, or about 3.3 miles for you.

a) If your battery range is at 100.50 to 101.19 (aka “101”) you’d see it jump 3 to “104” (103.80 to 104.49).
b) If your battery range is at 101.20 to 101.49 (aka “101”) you’d see it jump 4 to “105” (104.50 to 104.79).

Randomly, (a) should happen ~70% of the time and (b) ~30% of the time.

So you probably see it go +3 about twice as many times as +4.

The “more than typical hid” can also be explained by all the rounding errors in the inputs to the calculation. 356 and 155 (55.2 kWh) could be 355.5 and 154.5 (54.9 kWh), or 356.49 and 155.49 (55.4 kWh)

168 mi on the display could be (167.5 to 168.49) + (0 to 3.3 mi of regen).

So extremes are:
Max 55.4 kWh / 167.5 mi = 330.7 Wh/m
Min 54.9 kWh / 171.79 mi = 319.6 Wh/mi

You really need a lot of data samples to narrow the range

 

[darth_vad3r]

LR RWD.

Perhaps its 780 Wh, .1% of a 78 kWh pack (75 useable) giving 230 Wh/mi and 325 mile range at 75 kWh. (Though I never got the bump from 310 to 325.

See attached for the table.

Thanks for the data.

Hmm, interesting. I don't know how they've accounted for the range increase under the hood, just twiddlng the constants I presume.

BUT ... your numbers do imply you have range over 310.
The min, max and average range I get from your data is:
308.3, 316.8, 312.9 miles.

If I had to guess what 100% would be, looking at the numbers, where they've ticked down and said the same percent for several different battery ranges ... I'd guess 100% would give you 314, and 282, 251 for 90 and 80%. Any of these look familar?

Also the last mile of your drive toggled between 70 and 69% a lot (5 changes), while the battery range stayed flat at 218.14 miles that whole time. This implies you were hovering right around 69.5% SOC.
218.14 / 0.695 = 313.87, matching my guesstimate using other methods.

To compare:
100% 325 would be 293 at 90%, and 260 at 80% <- a car with 325 range
100% 314 would be 282 at 80%, and 251 at 80% <- my guess for your car
100% 310 would be 279 at 80%, and 248 at 80% <- a car with 310 range

Going to crunch the other numbers on the 'constant' later...

 

[darth_vad3r]

Yes, that number that you never look at, haha... I agree, that IF that value is going to update down at all it would update every 0.1 mile. I thought about this as I was driving in to work this morning but I don't really have a good downhill that isn't out of the way to actually experiment with.

I’ll keep it on “since last charge” next time I drive home from work and see if I can notice!

So I saw the trip meter kWh number tick down today. I was keeping a close eye on it, and it had ticked up way before my downhill stretch so I didn’t get it to tick down then (I estimate only 1-200 Wh of regen there), but I managed to notice a tick up shortly before an intersection. I really let the regen kick in hoping I’d see it go back down. It did.

So the trip meters seem to immediately account for regen energy without a hysteresis buffer to go through, unlike the battery gauge distance display.

 

[GregRF]

Thanks for the data.

Hmm, interesting. I don't know how they've accounted for the range increase under the hood, just twiddlng the constants I presume.

BUT ... your numbers do imply you have range over 310.
The min, max and average range I get from your data is:
308.3, 316.8, 312.9 miles.

If I had to guess what 100% would be, looking at the numbers, where they've ticked down and said the same percent for several different battery ranges ... I'd guess 100% would give you 314, and 282, 251 for 90 and 80%. Any of these look familar?

Also the last mile of your drive toggled between 70 and 69% a lot (5 changes), while the battery range stayed flat at 218.14 miles that whole time. This implies you were hovering right around 69.5% SOC.
218.14 / 0.695 = 313.87, matching my guesstimate using other methods.

To compare:
100% 325 would be 293 at 90%, and 260 at 80% <- a car with 325 range
100% 314 would be 282 at 80%, and 251 at 80% <- my guess for your car
100% 310 would be 279 at 80%, and 248 at 80% <- a car with 310 range

Going to crunch the other numbers on the 'constant' later...

TeslaFi shows that on that date (over a year ago) 80% charge was 250.63, projected 100% at 313.29.

From the start of my drive this morning:

Hit a perfect net 0. the kWh meter never moved from 0.

 

[darth_vad3r]

TeslaFi shows that on that date (over a year ago) 80% charge was 250.63, projected 100% at 313.29.

So TeslaFi basically agreed with me ... did you always have more than 310 miles projected 100% rated range, or was this after the advertised update that told people it was 325?

Hit a perfect net 0. the kWh meter never moved from 0.

Now I’m wondering what the “SLO” trip meter is for?

Since Last O...rangutan?

 

[GregRF]

So TeslaFi basically agreed with me ... did you always have more than 310 miles projected 100% rated range, or was this after the advertised update that told people it was 325?

Sadly no, been a bit all over the place. Probably has to do with normally only charging to 80% and my commute only uses about 11% of the battery. Different firmware and the occassional long trip make up most of the bumps.

Now I’m wondering what the “SLO” trip meter is for?

Since Last O...rangutan?

My old college town of San Luis Obispo

 

[AlanSubie4Life]

Sadly no, been a bit all over the place. Probably has to do with normally only charging to 80% and my commute only uses about 11% of the battery. Different firmware and the occassional long trip make up most of the bumps.
View attachment 447971



My old college town of San Luis Obispo

Haven't been following along. Does anyone know if the TeslaFi "Estimated 100% Range" historical plot always agrees with what hand calculations from the %/rated miles extrapolation to full range would be (probably you'd want to read the rated miles & % from the API for accuracy of the extrapolation)? Or is TeslaFi adding any secret sauce to it beyond that?

 

[GregRF]

Haven't been following along. Does anyone know if the TeslaFi "Estimated 100% Range" historical plot always agrees with what hand calculations from the %/rated miles extrapolation to full range would be (probably you'd want to read the rated miles & % from the API for accuracy of the extrapolation)? Or is TeslaFi adding any secret sauce to it beyond that?

No secret sauce I'm aware of, but it only adds to this battery report the extrapolated range from the mileage and battery percentage at the end of a charging session.

 

[darth_vad3r]

No secret sauce I'm aware of, but it only adds to this battery report the extrapolated range from the mileage and battery percentage at the end of a charging session.

Whereas I was looking at driving data and picking off a transition point to try and get better precision on the percent divisor. E.g. 69.5 and 70.499% yield highly variable results which will account for some portion of the variation on their charts. It is +/- 4 mi on the LR for 70, 80, 90% charge targets accounting for that +/- 0.5 variance alone, let alone any rounding on the miles (which should be more accurate with TeslaFi using API numbers with 2 decimal places). You’d hope the end of charge SoC would be more accurate though ... but I think we’ve all likely seen the battery gauge change an hour after a charge ends, so who knows.

 

[darth_vad3r]

Sadly no, been a bit all over the place. Probably has to do with normally only charging to 80% and my commute only uses about 11% of the battery. Different firmware and the occassional long trip make up most of the bumps.

Yep. 80-69-80-69... good for the battery, ‘bad’ for “range display anxiety” (or even just range display accuracy, without any anxiety at all )

My old college town of San Luis Obispo

Heh, I was WAY off . I just thought “since last” because I glanced and it looked like SLC, and when I take notes I use S/LC for Since Last Charge and S/x:yy for the time

 

[GregRF]

Whereas I was looking at driving data and picking off a transition point to try and get better precision on the percent divisor. E.g. 69.5 and 70.499% yield highly variable results which will account for some portion of the variation on their charts. It is +/- 4 mi on the LR for 70, 80, 90% charge targets accounting for that +/- 0.5 variance alone, let alone any rounding on the miles (which should be more accurate with TeslaFi using API numbers with 2 decimal places). You’d hope the end of charge SoC would be more accurate though ... but I think we’ve all likely seen the battery gauge change an hour after a charge ends, so who knows.

It seems like a decent enough method. I only having it polling roughly once per minute. On the charge session today it took 5-7 minutes of charging to add 1% to the battery. It completed the charge on its 5th minute at 80%. So I think the car is charging up to right under the switch to 81%

Yep. 80-69-80-69... good for the battery, ‘bad’ for “range display anxiety” (or even just range display accuracy, without any anxiety at all

Heh, I was WAY off . I just thought “since last” because I glanced and it looked like SLC, and when I take notes I use S/LC for Since Last Charge and S/x:yy for the time

Not Salt Lake City?