High mileage check-in

[AAKEE]

I really feel like this is unlikely to be correct unless you started with an extraordinarily high capacity pack somehow. Even if it were 80kWh it’s only 5% down…

I think I mentioned before that there are some calendar aging tests that show much reduced calendar aging at 10C compared to the data I used for my formulas. We/I also did get data from @KenC about how he ”stores it in the freezer” so to speak.

More or less any car has matched the formulas, and except @KenC’s theres a model 3 in northern Sweden (2019 I think from the memory). The owners statements did not sound right but the energy screen did prove it. After a deeper look, the car was always outside in the freezing cold and clearly was below the already low calendar aging rate I have used.
Using the data with lower calendar aging rate got that car to match the formulas, from a assumed aversge cell temp(no data present so I used the average ambient in that town and added 5C for usage/charging etc as I usually do.

I did put about this ”saving” in calendar aging from lower temperatures:

The reduction in calendar aging is ~ 30% between 25C and 10C.
I did not only use this chart but I used several others, and counted pixels in the chart to get exact values and took tje average from all these that all had about the same reduction.
There was a few research reports that did show higher reduction. Right now I did not find the picture I would like to show, but this is a projected picture of the same data:
10C cut the calendar aging in half in this case.

Also, there is a few reports testing really cold and around -20C you have almost no calendar aging.

I end up with almost knowing for sure how a battery will degrade from 10C to 50C, and with different SOC etc, but the extreme values doesn’t have very many data points (few tests) and slightly difference in the result. Having a few more results pointing into a more precise direction would be good.
Anyway, playing with the thought that we cut it in half from 25C to 10C, and knowing Ken has the car in the freezer so he’s average cell temp probably is much lower than my M3P’s, that almost always was stored in the garage at 10-12C during the winter except when used and at work. I had average cell temp 13.46C or something like that.
Let us play with 5C average cell temp, and that the calendar aging is not higher than 50% of the 25C rate, and even if Ken uses 60%, he will be on the right side of the central graphite peak after a while (calendar aging losses vapacity above the peak).

The 25C rate below the central graphite peak is around ~ 2.75% for one year, so half of this due to below 10C is 1.375%

1.375 x square root (6 years) = 3.36% loss after 6 years. Playing with that the 2018 M3LR delivered 79.2 or so in the EPA test(?) would set us at 76.5 kWh today (calendar aging only.
My formulas that are a bit conservative by purpose on the savings below 10C get it like this for an average of 5C cell temp: (still not very far away).

My takeaway from the research is that there are not any outliers really and no battery lottery in getting a car that degrades less as both the calendar and cyclic aging is predictable.
We just need to understand why we see a specific behavior.

If you do a really big discharge it would interesting to monitor energy expended (with minimal losses in park). Also interested what your park “BMS adjustments” do on that page of the Energy app, does it generally adjust upwards or downwards when you park? How much?

Yes, either a 100-0 drive or a meticulous larger part of the pack drive with initial SOC/energy and as you say, the SOC at park and also the SOC after 30 minutes of no sentry/sleep and also delivered energy from the trip screen. Not too fast, but 55-60 mph or less (variations not an issue but high power will give more losses).

SOC numbers with decimal numbers from SMT.
SOC before, if the car was allowed to sleep a while (30 min or so) + SOC after the drive and after 30 min further + the delivered energy; then we can do an own probably reliable capacity calculation.

I expect it to match OK, and that there is not a big drop in SOC after the drive, but of course I too is curious

 

[AAKEE]

Mine does this both ways lol

Right after I finish charging and the car sits for 30-60 mins I lose 3%

Then when I drive to my destination and it sits for 30 mins I get back 3-4%

Then when I drive home and it sits for 30 mins I lose 3% again

And I’m willing to bet if I left the house and went some where else I would gain 3% again

I never saw 3% loss or win.

When my BMS was fairly off, showing 76 kWh or so capacity (out of 79 kWh real measured capacity), I got back ~ 2% after a drive that took about half the capacity.

If the car has been sleeping before a charge and there is a underestimation from the BMS, the SOC will drop after the charge stops and the OCV is measured.
(If there is an overestimation, the SOC climbs after the charge).

For driving, if the SOC drops a while after the drive, it is a overestimate.
If the SOC regains after the drive, we have a underestimate.

These are all by principle, if we for example do not let the car rest and do several supercharging sessions the SOC can be off for other reasons, so a jump could be by “coincidence” or more correct by other reasons up or down.

But having let the car sleep before the drive, at least a short while will let the BMS have the correct SOC (or very close) before the drive and then for one single drive it by principle will be like that.

Last friday, 100% charge (a couple of very short stops, but 250+ km). The energy calc from delta SOC gave 97.1 kWh which was the same as the BMS was a few days before but the BMS estimate had dropped a bit below 97 kWh at arrival.
See the regain in SOC where the arrow points.
I have dome these checks several times out of nerd interrest and after seeing the consistent behavior I would say that this can be used to bot calculate the real capacity and to indicate a under/overestimation.

Jumping up and down not consistent, I haven’t seen but it must depend on something else. Maybe the voltage vs state of charge curve has changed, without the BMS catched that change.

 

[gottagofast]

I never saw 3% loss or win.

When my BMS was fairly off, showing 76 kWh or so capacity (out of 79 kWh real measured capacity), I got back ~ 2% after a drive that took about half the capacity.

If the car has been sleeping before a charge and there is a underestimation from the BMS, the SOC will drop after the charge stops and the OCV is measured.
(If there is an overestimation, the SOC climbs after the charge).

For driving, if the SOC drops a while after the drive, it is a overestimate.
If the SOC regains after the drive, we have a underestimate.

These are all by principle, if we for example do not let the car rest and do several supercharging sessions the SOC can be off for other reasons, so a jump could be by “coincidence” or more correct by other reasons up or down.

But having let the car sleep before the drive, at least a short while will let the BMS have the correct SOC (or very close) before the drive and then for one single drive it by principle will be like that.

Last friday, 100% charge (a couple of very short stops, but 250+ km). The energy calc from delta SOC gave 97.1 kWh which was the same as the BMS was a few days before but the BMS estimate had dropped a bit below 97 kWh at arrival.
See the regain in SOC where the arrow points.
I have dome these checks several times out of nerd interrest and after seeing the consistent behavior I would say that this can be used to bot calculate the real capacity and to indicate a under/overestimation.
View attachment 1056658


Jumping up and down not consistent, I haven’t seen but it must depend on something else. Maybe the voltage vs state of charge curve has changed, without the BMS catched that change.

You have never seen that before? Think something is wrong? It does it every day

 

[AlanSubie4Life]

You have never seen that before? Think something is wrong? It does it every day

Everything is fine. It is possible you looked too fast.

 

[gottagofast]

Everything is fine. It is possible you looked too fast.

Worrying that the person that does battery research all the time has never seen that before

 

[AlanSubie4Life]

Worrying that the person that does battery research all the time has never seen that before

Definitely post the pictures. Worthwhile. Can do the trip meter monitoring mentioned above concurrently. Starting miles, start percentage, energy used, end miles, end percentage etc. Every piece of data painstakingly and slowly gathered.

 

[kpanda17]

And mines at 8% in 24 months, it’s been over year since 100% charge though so that might change things

For 2170 NMC batteries, mine, I charge to 100% once every month. This is a rebalancing of every cell in the pack, basically discovering every cell to understand its charging status and capacity. No need for LFP as they are always recharged to 100%. Rebalancing will reclaim and report on the true capacity of your battery back by the BMS.

 

[AAKEE]

Worrying that the person that does battery research all the time has never seen that before

I have very good data that no of my cars doesnt do like that.
Because of the logging with teslafi and teslalogger I could have seen it.

I wouldn’t worry though.

What car did you have, age, miles and how did you charge it mostly?

 

[gottagofast]

I have very good data that no of my cars doesnt do like that.
Because of the logging with teslafi and teslalogger I could have seen it.

I wouldn’t worry though.

What car did you have, age, miles and how did you charge it mostly?

22 model 3 performance, I had been charging to 80% from Tesla recommendation until I started reading here about 50% and less SoC so ive been charging to 50-70% for the last 5 months

Miles right now is 34k

 

[Bouba]

I think I mentioned before that there are some calendar aging tests that show much reduced calendar aging at 10C compared to the data I used for my formulas. We/I also did get data from @KenC about how he ”stores it in the freezer” so to speak.

More or less any car has matched the formulas, and except @KenC’s theres a model 3 in northern Sweden (2019 I think from the memory). The owners statements did not sound right but the energy screen did prove it. After a deeper look, the car was always outside in the freezing cold and clearly was below the already low calendar aging rate I have used.
Using the data with lower calendar aging rate got that car to match the formulas, from a assumed aversge cell temp(no data present so I used the average ambient in that town and added 5C for usage/charging etc as I usually do.

I did put about this ”saving” in calendar aging from lower temperatures:
View attachment 1056627
The reduction in calendar aging is ~ 30% between 25C and 10C.
I did not only use this chart but I used several others, and counted pixels in the chart to get exact values and took tje average from all these that all had about the same reduction.
There was a few research reports that did show higher reduction. Right now I did not find the picture I would like to show, but this is a projected picture of the same data:
10C cut the calendar aging in half in this case.
View attachment 1056631

Also, there is a few reports testing really cold and around -20C you have almost no calendar aging.
View attachment 1056632

I end up with almost knowing for sure how a battery will degrade from 10C to 50C, and with different SOC etc, but the extreme values doesn’t have very many data points (few tests) and slightly difference in the result. Having a few more results pointing into a more precise direction would be good.
Anyway, playing with the thought that we cut it in half from 25C to 10C, and knowing Ken has the car in the freezer so he’s average cell temp probably is much lower than my M3P’s, that almost always was stored in the garage at 10-12C during the winter except when used and at work. I had average cell temp 13.46C or something like that.
Let us play with 5C average cell temp, and that the calendar aging is not higher than 50% of the 25C rate, and even if Ken uses 60%, he will be on the right side of the central graphite peak after a while (calendar aging losses vapacity above the peak).

The 25C rate below the central graphite peak is around ~ 2.75% for one year, so half of this due to below 10C is 1.375%

1.375 x square root (6 years) = 3.36% loss after 6 years. Playing with that the 2018 M3LR delivered 79.2 or so in the EPA test(?) would set us at 76.5 kWh today (calendar aging only.
My formulas that are a bit conservative by purpose on the savings below 10C get it like this for an average of 5C cell temp: (still not very far away).
View attachment 1056639

My takeaway from the research is that there are not any outliers really and no battery lottery in getting a car that degrades less as both the calendar and cyclic aging is predictable.
We just need to understand why we see a specific behavior.

Yes, either a 100-0 drive or a meticulous larger part of the pack drive with initial SOC/energy and as you say, the SOC at park and also the SOC after 30 minutes of no sentry/sleep and also delivered energy from the trip screen. Not too fast, but 55-60 mph or less (variations not an issue but high power will give more losses).

SOC numbers with decimal numbers from SMT.
SOC before, if the car was allowed to sleep a while (30 min or so) + SOC after the drive and after 30 min further + the delivered energy; then we can do an own probably reliable capacity calculation.

I expect it to match OK, and that there is not a big drop in SOC after the drive, but of course I too is curious

Hi AAKEE, some of the graphs on higher temperatures where done by mathematical extrapolation and not experimentation. This could be open to errors. For example, extrapolation would let you believe that 100% charge is worse than 80% SOC....but experimentation shows that 100% is better

 

[AAKEE]

Hi AAKEE, some of the graphs on higher temperatures where done by mathematical extrapolation and not experimentation. This could be open to errors. For example, extrapolation would let you believe that 100% charge is worse than 80% SOC....but experimentation shows that 100% is better

I did not use any extrapolated data, only real tests and for my formulas i used only Panasonic NCA tests.
I also did own tests with Panasonic 2170, the early model 3 cell.

Not all tests shows that 80% is worse than 100, there are several examples with about the same rate between 65-70% to 100%.

Most tests only use a few test points and they often connect the points in the chart with lines that are ”extrapolated” because of this. In this cases only the test points are used.

 

[M3BlueGeorgia]

Idk I think I need to do a 100% charge something seems out of whack to me

You can successfully extrapolate total EPA range from a routine 90% charge.

 

[AlanSubie4Life]

My takeaway from the research is that there are not any outliers really and no battery lottery in getting a car that degrades less as both the calendar and cyclic aging is predictable.
We just need to understand why we see a specific behavior.

Yeah. Seems possible I suppose. I'd be surprised at an average cell temp of 5C. Average temp in Portland, Maine is 8C, but that could be a bit colder inland presumably.

Of course, it's not just the average cell temp - if it's at 15C 50% of the year that will add up even if there's zero loss the rest of the year. I guess I am not sure if you account for that in your model (it does seem like it would make a bigger difference in the model results when you get to very cold temps, though it would matter at any temperature).

To me it just still seems like a bit of an outlier. Certainly a huge component of it is the cold temperature, and @KenC keeping it at low SOC. There are always outliers, as can be seen in the research data (for one reason or another). Maybe not a huge outlier, but still seems like it is an outlier beyond the expect low calendar aging. I'm not aware of any other Model 3 reports from 2019 which do not show loss of range (you mentioned the one other). And there are plenty of cars in Canada, Norway, Sweden, etc. Of course we only get a very small sample of them.

How much would cyclic aging be for ~67k miles, of course with generally shallow cycles?

 

[Bouba]

Summer in the South of France you see a lot of Scandinavian Teslas....so while they may live in their battery preserving deep freeze...they do occasionally come south of the Arctic Circle

 

[AAKEE]

In my world there are no outliars

How much would cyclic aging be for ~67k miles, of course with generally shallow cycles?

This is rather new cells cycled (actual model 3 2018.)

Time which causes calendar aging will most probably protect from cyclic aging rate, like we see the reduction in rate below.

Using 10% daily would set us at about 8% after 1000FCE, thats around 250K mi.

So worst case around 2% from 55-45% (in-car-displsyed, 57-47%). But calendar aging would protect and as it seems, for low SOC cycles we more or less can disregard the cycles. Probably lessens calendar aging in the same rate it causes degradation. Remember SEI that is the nice kind of loss, also protects from further like losses.

1year from delivery, 1 year three months from build date, and 21K km, the NFP follows the calendar aging graph (average temp + average SOC adjusted) like a charm.