Some new data from research on Tesla model 3 cells

There has recently been released a series of new research reports containing tests on Tesla Model 3 Cells (Panasonic 2170 NCA).
This is the calendar aging test from one of them (25C, 15, 50 and 85% SOC. Checkup once a month):
Using the datapoints from these and putting them in the old charts I ususally post, these match the olds ones quite good. As there is only three points, it do not show the real form of the curve, but all three points match the usual graphs.

For the cyclic tests, they did use rather high currents, not really respresentative to normal EV use. (To the researchers defense, the currents used is sort of the most EV-battery manufacturers current in the specifications but still not close to the regulkar EV usage).
Charged with 0.33C which would match about a 25kW DC charger, or double to four times the usual rate EV owners use mostly. Probably not offsetting the result much, but to be clear this is how it was done.

Discharged with 1C, which would be 78kW, about enough to drive constant at 200kph. This is way above the average power used from a regular EV. Driving at higway speeds at 120kph/80mph or so, we normally use like 1/4 of that power.
The average car often has a average speed longterm of about 50-60kph, meaning we often use 1/8-1/4 of the power in these cyclic tests.
From other tests we can se that lower power reduce the wear, the degradation often reduces to somewhere down to 0.5-0.7C.

In this report the author was a bit surprised over the increased wear at 5-15% SOC and 15-25% SOC. I would say that it it a very high probability of that this is induced by the 1C discharge rate, and that our normal power rates used IRL would make this look different. This is nothing I can promise but from several other research tests we can see that there ususally is a tendency to slightly increase the cyclic degradation at the lowest SOC ranges.

According to this chart, the best cycling range is 55 % down to 35%( see note below about true SOC).

Note: These are “True SOC”. 0% in this chart is where the car already has stopped, and 5% in-chart is about 0% displayed and 55% in-chart is is about 57% displayed.

As I said above, there is a high probability that the low SOC range wear much less with a lower C-rate. Anyway, due to the high impact of calendar aging we most certainly benefit from staying low in SOC.

For the first two years, we would loose about 9-9.5% from calendar aging if staying at high SOC.
During these two years, if we drive 15-20K km annually (10-15Kmiles), and stay in the very low regime cycling (5-25% true SOC, thats 0-20% displayed SOC) we would loose about 1% from ~ 75-100 FCE cycles during these two years/30-40K km.

IRL its not possible to stay that low in SOC without actively stopping the charging, as 50% is the lowest setting (but for reference to low /high SOC).

To reach the same level of cyclic degradation from low SOC cycling according to the chart we would need about 700FCE, or about 280K km, but that is not really possible to do and at the same time stay at 5-25% SOC.

So, a car charged to 80-90%, and used as most EV’s is used, will mostly be above 55% SOC and have a calendar aging close to the 85% graph.
After two years, it will be around 10% degradation if the average cell temp is about 25C.

If the car was charged to 50-55% it would have a calendar aging around 6%, and the cyclic aging would be half the high SOC car, so more or less negligeble.

Link to one report

[Edit]For what its worth, if someone is worried about the low SOC below 20% (I am not, but I’m aware of the classic forum rumors), charging to 50-55% and charging for the daily drives at or above 20% (not talking longer traveling here) all aspect of this report if ticked-in-the-box.

I will not change any of my charging behavior because of this report. There is from time to time small differences in the reports and usually the reason for that can be found by thorougly comparing with other tests. We need much more than one report to state a “fact”.

 

[AAKEE]

It would be a good idea to consider degradation from real usage.

It’s already considered.

-By using real data from for example Teslafi.
-By using real data from teslalogger.

The sum of very many reseach tests /reports tell us how batteries behaves.
The research do not have big differences between individual cells or between different research. It is quite consistent.

I took some time to sum that up in formulas before I got my first Tesla a little more than three year back. I did have to adjust my prediction after a while as my logs showed that the average cell temp was lower than my initial estimate.
Other from that the calculation was spot on.

So about one year ago I started asking people if in forums about their usage (charging schedules, daily usage of SOC, where they live, if they have the car outside or in a heated garage etc.).
So far, I guess > 100 cars has been tested against the formulas. Actually only three is outliers, of which two is living outside in cold climates ( @KenC’s is one of these).
The third one is a model S that most probably had the battery changed or a BMS that are in the blue.
All other has been more or less spot on.

The usual thing is someone stating ”I have only 3% after five years at XXYYY miles”.
After looking into it, the owner has gotten a few things wrong and the real degradation is like the text book example. So far happened ~ 10 times here and likewise on a swedish forum.

One big issue is Tessie.
Firat of all the Tessie makers had a few things wrong. Tesdie ise the energy reported by the car (and SOC numbers) as the base to calculate

For instance, my 2016 P90DL has a usable capacity of 74 kWh. It was 81.8 new. That's roughly 10% degradation over 90k miles and nearly 8 years. I charged to 90% the first year in a hot climate and 80% since then. According to this data, I should have hit 10% degradation after 2 years, which is not the case. For the older NCA based cars, the degradation tends to be 5% in year 1 and 1% each year after. In a recent presentation, Jeff Dahn recommended cycling 75% to 50%. If I remember correctly, that's how he charges his car.

Tessie reports 9.7% degradation. 81.8 new vs 74 now - neither of these is total.

Tessie is wrong.
Tessies initial value is very often wrong, more wrong than right. Before we helped Tessie They/he thought that the value he calculated was net (exluding the buffer).
We know that this is not the case and we know why. Here is the Slightly shortened version:

Teslas system includes the buffer in the range at 100% but hides the buffer progressively until 0% displayed.
And when charging it un-hides the buffer.

When charging from 0% displayed to 100% the energy charged the car will display as charged ( ”+ XY kWh”) is the same as the total battery capacity. This is a ascertained fact that it work this way. It is checked over and over again, so ”we” know this.

Tessie use the + kWh value reported OTA via the API. (Also shown in the tesla app), and calculates the capacity like
[ +kWh / delta SOC = capacity].

This ends up in the calculation showing the total capacity. Earlier Tessie vendors thought they calculated the ”usable capacity” After a discussion and information given in the Tessie vendors thread, the term was changed to ”capacity”.
(Some of this is around here in that thread:
Tessie thread about capacity ).
Before calling my claims ”bullock” you need to either ask @James@Tessie in that thread or use Scan my Tesla and compare like ”we” already have done.

The way Tessie calculates initial capacity causes a lot of confusion. Its much more often right than wrong. In the older cars case Im sure Tessie used a set number for the capacity as Tesdie was not even alive at the time when the cars was new.
And that set number was the net capacity, as Tessie thought that the calculated number was “net capacity” as well. Which it is not.

The numbers Tessie present is very confusing people, and I actually had to ask Tessie in the thread to get a clear picture.
Most people in Sweden in Facebook groups goes around thinking “I want noy have 3% degradation”.
The issue is that the initial value is often too low (way to low) and people thinks “thats the net capacity/without the buffer”

I'm certain 74kWh is the usable capacity.

I know it’s the total capacity.

I only brought this up because of the chart you posted. There is no way to properly test battery degradation in a car over 8-10 years without using the batteries in a car for 8-10 years.

Actually the test data for both calendar aging and cyclic aging can be transfered to real life.

What is typically done in labs is a projection based on some form of coulomb tracking.

The do not “coloumb track” everything. Coloumb counting is one of the methods, mostly when the researchers try to understand why things happen.

The degradation for both cyclic and calendar aging graphs is done by cycling the battery a few times and then charge it full and discharge it in a controlled environment and measuring the energy the battery delivers.
The graphs you see is actual loss of capacity, just like you will see it in the car.

I'm just suggesting that you factor in some real world data instead of relying 100% on this lab study.

You need to understand this ( ):
-It is not “this lab study”. Instead it is litterally hundreds of studies telling the exact (or very close) same thing. Hundreds, literally!
-It is not only the lab research but it is also litterally hundred real life cars checked by calculating the capacity according to the research findings and how the car was used and the result is a very precise hit.

Jeff Dahn is pretty much the GOAT when it comes to battery expertise. Like you, he suggests lower SOCs to limit degradation. For instance, he recommends 30% SOC if you are to leave a EV for months. But I think it's telling that he personally has no qualms about recommending 75%.

If you go back a bit very many persons used Jeff Dahnnas an example of someone stating a different approcah. He had a lot of theoreticall stuff and the net recommendations was not exactly in line with the rest of the world recommendations
I have seen some old things from him and at that point I thought he had some kind of limitations of what he could say due to the work with Tesla.
We can be very certain that if he says 70 or 75% that is not a number taken from the sole purpose of reducing the degradation. Then he have told you what he think is a good compromise between usable range and degradation (which can only be just a compromise that is very generalized). It could also be a limitation from Tesla about what he is allowed to say.

For the 30% its the same. 30% will not give the lowest degradation. We know this from
A Huge numbers of research reports.
This must also be a compromise. Either limitations from Tesla or like making it certain that he will not recommend something that cause the low voltage
battery to run out of juice and get damaged.
30% is not bad. It isnquite good but it is not the number causing the lowest degradation. The lower the SOC down to 0% the lower degradation. But at 0% displayed we are not very long from the HV batt disconnecting to protect itself causing issues on the LV batt so we need a small margin if we leave the car long time.

Tesla has 15-50% for storage of high voltage batteries in stock ( they can be stored for long time, so 15% is to ensure that it stays in safe levels as long as the battery is in storage.

EV Battery Health with Dr Jeff Dahn Dalhousie U - Worth watching the entire presentation if you can.

I suggest to not only listen to a single source, just like not to only read a single research report.
There are more than one person having the knowledge of Jeff Dahn. No shadow over him or his work but we do not know the background to all of hes statements. He most probably has an agreement of what he is allowed to say after working with Tesla.
From what I did see here on TMC the latest information from Jeff Dahn is close
to the data and information I spread here.

 

[AAKEE]

FYI in the EPA test they extracted at least 84kWh from the pack.

2016 90D 384mi at 218.57Wh/mi
Or 405.2mi at 208.02Wh/mi

The data isn’t as nice as newer submissions but that’s the “new” pack capacity.

No idea what buffer percentage is, etc.

Yes, I was about to go that way ( EPA test)

The buffer on the bigger packs was a fixed 4.00 kWh on the S/X earlier.
The smaller, from 70-75kWh and down was fixed at 2.4kWh.

 

[James@Tessie]

One big issue is Tessie.
Firat of all the Tessie makers had a few things wrong. Tesdie ise the energy reported by the car (and SOC numbers) as the base to calculate

Tessie is wrong.
Tessies initial value is very often wrong, more wrong than right. Before we helped Tessie They/he thought that the value he calculated was net (exluding the buffer).

The way Tessie calculates initial capacity causes a lot of confusion. Its much more often right than wrong. In the older cars case Im sure Tessie used a set number for the capacity as Tesdie was not even alive at the time when the cars was new.
And that set number was the net capacity, as Tessie thought that the calculated number was “net capacity” as well. Which it is not.

The numbers Tessie present is very confusing people, and I actually had to ask Tessie in the thread to get a clear picture.
Most people in Sweden in Facebook groups goes around thinking “I want noy have 3% degradation”.
The issue is that the initial value is often too low (way to low) and people thinks “thats the net capacity/without the buffer”

There have definitely been confusing numbers and terms used in the past, which is why usable was removed and the capacity was changed to total (thanks for your feedback!) This was quite awhile ago but you still seem to have an issue with the app.

Here's Tessie with my new Model X. The degradation matches the % shown in the Service Mode Battery Health test and the initial capacity matches what is displayed via OBD readers, so I'm not sure what is wrong or confusing here. All data is calculated the same way for all vehicles, so other than the normal BMS fluctuations that each vehicle has, my vehicle is not uniquely correct or special.

I'm very curious how Tessie is a "big issue", "very often wrong, more wrong than right" and "very confusing [to] people". Can you please explain the issue or what is wrong with the data in this screenshot?

 

[dakhlkt22]

It’s already considered.

-By using real data from for example Teslafi.
-By using real data from teslalogger.

The sum of very many reseach tests /reports tell us how batteries behaves.
The research do not have big differences between individual cells or between different research. It is quite consistent.

I took some time to sum that up in formulas before I got my first Tesla a little more than three year back. I did have to adjust my prediction after a while as my logs showed that the average cell temp was lower than my initial estimate.
Other from that the calculation was spot on.

So about one year ago I started asking people if in forums about their usage (charging schedules, daily usage of SOC, where they live, if they have the car outside or in a heated garage etc.).
So far, I guess > 100 cars has been tested against the formulas. Actually only three is outliers, of which two is living outside in cold climates ( @KenC’s is one of these).
The third one is a model S that most probably had the battery changed or a BMS that are in the blue.
All other has been more or less spot on.

The usual thing is someone stating ”I have only 3% after five years at XXYYY miles”.
After looking into it, the owner has gotten a few things wrong and the real degradation is like the text book example. So far happened ~ 10 times here and likewise on a swedish forum.

Have you shared these formulas here? My M3 is only a year old but I'd be curious to see what the formulas predict and to see if I can input data from TeslaFi to match real world data against the predictions.

 

[johnny_cakes]

There have definitely been confusing numbers and terms used in the past, which is why usable was removed and the capacity was changed to total (thanks for your feedback!) This was quite awhile ago but you still seem to have an issue with the app.

Here's Tessie with my new Model X. The degradation matches the % shown in the Service Mode Battery Health test and the initial capacity matches what is displayed via OBD readers, so I'm not sure what is wrong or confusing here. All data is calculated the same way for all vehicles, so other than the normal BMS fluctuations that each vehicle has, my vehicle is not uniquely correct or special.

I'm very curious how Tessie is a "big issue", "very often wrong, more wrong than right" and "very confusing [to] people". Can you please explain the issue or what is wrong with the data in this screenshot?

View attachment 1013322

@James@Tessie Tessie still uses an incorrect original capacity for my 2019 Model 3 SR+. This can be verified by the inconsistent degradation calculation between the kWh estimate and max range estimate. The original range of my vehicle was 386km according to Tesla's website:

The max range estimate according to Tessie is now 336km, so that would imply a 13% degradation. However, using the capacity measurements, Tessie reports a 11.5% degradation if I leave the default 51.6 kWh as the original capacity, effectively under reporting to degradation. Here's an illustration:

Now if I change the original capacity from the default 51.6 kWh to 52.5 kWh as @AlanSubie4Life and @AAKEE suggested in another thread, then everything becomes consistent. See this illustration:

I think this is what @AAKEE is referring to. If the default original capacity is used, then Tessie over estimates the battery health and under reports the degradation.

 

[James@Tessie]

...

Interesting! I wonder if Tesla didn't update the trim and battery config for that car, so it's actually defined as being a different type of car so we're pulling the initial capacity for a different car. What does your vehicle define in the firmware inspector under vehicle_config > trim_badging and efficiency_package?

Out of curiosity, have you run the service mode battery health check? If so, what did it report?

 

[KenBlub]

Interesting! I wonder if Tesla didn't update the trim and battery config for that car, so it's actually defined as being a different type of car so we're pulling the initial capacity for a different car. What does your vehicle define in the firmware inspector under vehicle_config > trim_badging and efficiency_package?

Out of curiosity, have you run the service mode battery health check? If so, what did it report?

I love my Tessie app and the control it gives me over my 2023 Model 3 LR AWD. BUT…it gave me an initial capacity for my 2023 NCMA of 77.8 vs 78.8 which I believe is the correct number. I think 77.8 might have been the number for the European models. No biggie. I just changed it in the app.

Not a biggie either but I consistently get higher capacity when I multiply avg Wh/mi x miles divided by SOC. Not a big number though. Energy app gives me 78.4 and Tessie 77.5. Probably within margin of error. Still under 2% after almost 7 months.

 

[AAKEE]

There have definitely been confusing numbers and terms used in the past, which is why usable was removed and the capacity was changed to total (thanks for your feedback!) This was quite awhile ago but you still seem to have an issue with the app.

I'm very curious how Tessie is a "big issue", "very often wrong, more wrong than right" and "very confusing [to] people".

To start I’d like to be clear with that I welcome a enlightening discussion and I’m not afraid to do this statements as I (we) always have gotten a good conversation and explanations from you. I’m not here to complain, but I still think I do not need to hold back thoughts.

Well the issue is (from my perspective of course, not to be taken as “true”) that the initial capacity seen in Tessie is considerably lower than the real capacity which we know from the EPA-test, the “Full pack when new”, the nominal full pack and also driving tests online where the delivered capacity is shown.
So one example is model 3 in europe that had four different batteries at the same time. The initial capacity was mostly newer correct and for example the M3P 21 I had, had a 82.1 kWh battery but *any* car using Tessie showed about 77.8 or 78.8, in some cases lower than that.
A common example is we are discussing degradation (and the fact that low SOC is the best, which already is a fact as per the sum of research, to keep it in short terms) The owner using Tessie states I always charge to 90% (or 80%) and my degradation is only 5%. Looking into it, the ”capacity number” Tessie calculates is good and mostly spot on but as the initial number is ~5% off, the real total degradation is about 10%.

There are a lot of these kind of ”issues”.
From a purely battery warranty perspective I do not know from what level Tesla counts the degradation, bit Im sure that I would be dissapointed if the initial capacity that was used in the EPA test was bargained down so I needed to loose 35% before the warranty kicked in.

Here's Tessie with my new Model X. The degradation matches the % shown in the Service Mode Battery Health test and the initial capacity matches what is displayed via OBD readers, so I'm not sure what is wrong or confusing here. All data is calculated the same way for all vehicles, so other than the normal BMS fluctuations that each vehicle has, my vehicle is not uniquely correct. Can you please explain the issue or what is wrong with the data in this screenshot?

View attachment 1013322

The S/X share the same battery, I think.

Well, if we begin with the 2023 EPA application documents for X, the battery delivered 99.9kWh in the the test. (or just 100kWh if we do the rounding correct).

Link to EPA-test ( X, not Plaid
But there is a test protokol for that also)

The 99.976kWh was measured with external high quality equipment (has to, to be valid).
So we know the X can deliver 99.976 kWh when about new.
All
Can be used, the battery actually delivered 99.976 kWh.

The buffer is 4.5% on the new S/X so the net value would be 95.5kWh, if the BMS had 99.976 as the nominal full
Pack. It probably has not, and that value is slightly lower normally.

So the total Battery Capacity is higher than 95.5 kWh as Tessie shows.
The BMS number for a new
Pack is 99.4 kWh for both the model S and S plaid. I would guess the same id valid for the X, but I do not know.

This is my Plaid a few months back:

Full pack when new 99.4kWh.

Nominal full pack 98.4kWh
Nominal remaining 99.0 kWh
These are BMS numbers. I think the terms
are set by scan my tesla, so not Teslas original terms for this, but the terms in confirmed to show what SMT calls them.
(I always got slightly higher
number for nominal remaining than the nominal full pack at full charges, by 0.3-0.7 kWh).

I saw a picture here on TMC from Tessie on a S Plaid and I think Tessie said 95.3 or maybe 95.5kWh initial capacity.

One possible explanation would be that my battery is currently at 103.6% degradation, and the EPA tests for bot S and X and S Plaid was all at about 103 to 104.7% health or -3 to -4.7% degradation.
I know batteries better than that, and the spread in different cells are less and on the pack level, much less.

The model S plaid has a degradation threshold at about 96.2 kWh with the 19” set in the menu. We know this as the range reaches 396 miles (637km) at that point and having a nominal remaining less than 96.2 kWh at 100% will cause the range not to reach the full 396 miles.

I know we discussed this earlier but anyway, can tessie be fooled by the degradation threshold? I think from memory Tessie was able to calculate capacitys above the threshold. ( from my understanding of how the capacity is calculated it should not be limited by that).
It might be a coincidence that your car shows 96.2kWh, but I’m not sure.

As for the Service mode (which I have newer used) I can think of that Tesla uses the degradation threshold as the 0% degradation level in that test.
It would mask the degradation that happens until the range drops, and it would also give Tesla a few percent extra degradation slack from needing to change the battery due to reaching 30%. (Which will not be a very common issue anyway).

So for the real degradation or if we would test the energy or range in a new X or S, driven to the car stops it will be between 98-100 kWh’s worth of range when new and when we only can get 96.2 kWh out of the car we have already lost 2-4% real capacity.

 

[James@Tessie]

Well, if we begin with the 2023 EPA application documents for X, the battery delivered 99.9kWh in the the test. (or just 100kWh if we do the rounding correct).

So we know the X can deliver 99.976 kWh when about new.
So the total Battery Capacity is higher than 95.5 kWh as Tessie shows.
The BMS number for a new
Pack is 99.4 kWh for both the model S and S plaid. I would guess the same id valid for the X, but I do not know.

I have this exact Model X and did a full charge_energy_added measurement on day 1 from the factory:

Odometer: 5.4mi
Capacity: 96.2 kWh
Degradation: 0%
Service mode degradation: 0%

We have 13,740 other charge_energy_added readings that confirm this measurement - worst case being 96.06, best case being 96.26.

So it appears it's 13,741 different readings from the car in the real-world that say 96.2 kWh vs. the 1 EPA specification that says 100 kWh. How is it possible that 10,000+ cars are wrong? Or is charge_energy_added wrong? Or is the EPA wrong? The answer has to be here somewhere.

 

[Eric33432]

I have this exact Model X and did a full charge_energy_added measurement on day 1 from the factory:

Odometer: 5.4mi
Capacity: 96.2 kWh
Degradation: 0%
Service mode degradation: 0%

We have 13,740 other charge_energy_added readings that confirm this measurement - worst case being 96.06, best case being 96.26.

So it appears it's 13,741 different readings from the car in the real-world that say 96.2 kWh vs. the 1 EPA specification that says 100 kWh. How is it possible that 10,000+ cars are wrong? Or is charge_energy_added wrong? Or is the EPA wrong? Or something else?

I am curious: Have you determined if the 96.2 kWh battery in MX is the same battery used in the MX Plaid version?

 

[James@Tessie]

I am curious: Have you determined if the 96.2 kWh battery in MX is the same battery used in the MX Plaid version?

11,660 readings for that one being 96.2 as well.

 

[Eric33432]

MX Plaid. Kind of bummed out about this. March 2022 build I think. I only charge to 50% for daily driving, been on 5 road trips using SuC.

 

[AlanSubie4Life]

Or is charge_energy_added wrong?

I suspect this is wrong. How is this calculated/read? I know you have mentioned/described it before. But maybe a refresher…

If it’s just the same value (with extra resolution) presented by the user-facing GUI during a charge event, then it will be incorrect for a brand new vehicle which is above the degradation threshold (that’s my claim anyway).

It’s well documented by comparing the energy screen method with SMT reads that the energy the car shows it has is limited by the degradation threshold.

So I think each mile added will contain more energy than the GUI/API indicates, when the vehicle is over threshold.

The EPA numbers have been pretty well validated by SMT, so I think they are fine.

Could be wrong, but just my guess. I just suspect that looking at charging event stats will be distorted on a vehicle which exceeds the degradation threshold (will look like it has less capacity than it actually does).

 

[DrChaos]

One thing that always makes me look at these studies with some skepticism is their test method. They always charge linear and discharge linear. No EV charges linear and definitely doesn't discharge linear. A German battery expert said that regen braking is one of the most important factors extending the life of EV batteries. Of course this doesn't invalidate the results done using linear discharging, but it reduces the effect on degradation. My point is, that regen braking will reduce the difference in degradation between the cycle tests. If you would factor in the effects of regen braking, the real world difference won't matter much if you charge to 55% and drive to 20% or charge to 80% and drive to 45%.

That's true, but most people's aging without heavy driving is from calendar mechanisms, not cyclic mechanisms, and regen doesn't matter to that. This fact of real driving & regent will make it so that cyclic aging mechanism will be even less important to realistic battery longevity than scientific results (500-1000 full charge cycles) would suggest.

The difference between your two situations will be the overall time of SOC spent below 55% and that will have a significant effect on calendar aging rate, with the benefit being to the first lower SOC strategy.

My EV certainly does charge linearly as at L2 charging up to 55% there is no limitation on the input power from the car's side, so it's 240V x 32A continuously from start to end of charge.

 

[James@Tessie]

I suspect this is wrong. How is this calculated/read? I know you have mentioned/described it before. But maybe a refresher…

If it’s just the same value (with extra resolution) presented by the user-facing GUI during a charge event, then it will be incorrect for a brand new vehicle which is above the degradation threshold (that’s my claim anyway).

It’s well documented by comparing the energy screen method with SMT reads that the energy the car shows it has is limited by the degradation threshold.

So I think each mile added will contain more energy than the GUI/API indicates, when the vehicle is over threshold.

The EPA numbers have been pretty well validated by SMT, so I think they are fine.

Start service mode battery test or drive car to 0% -> charge to full -> read charge_energy_added value in the firmware via the API. That value always matches what's displayed in the car as well. So if that capacity reading is wrong, it's at least consistent in being wrong for everyone in the same way, and has to be used as the basis since that's the only actually measured value we have to go off of (not counting EPA whitepapers since we only care about real-world performance here.)

If the capacity reported by the car is wrong, would be very interesting to know why. Will leave it to the disassemblers and battery chemists smarter than I to figure that out.

 

[AlanSubie4Life]

Start service mode battery test or drive car to 0% -> charge to full -> read charge_energy_added value in the firmware via the API. That value always matches what's displayed in the car as well. So if that capacity reading is wrong, it's at least consistent in being wrong for everyone in the same way, and has to be used as the basis since that's the only actually measured value we have to go off of (not counting EPA whitepapers since we only care about real-world performance here.)

If the capacity reported by the car is wrong, would be very interesting to know why. Will leave it to the disassemblers and battery chemists smarter than I to figure that out.

Yeah that value is wrong for a new car which is above the degradation threshold.

It’s ok because the reported value is not what is added to the pack when above the degradation threshold (or at any time due to the 1/0.955 factor). This has been verified by SMT by @AAKEE (I don’t have SMT but he helped to do the observations and calculations and confirm my suspicions) some time back - would have to do some searches to try to find the exact post (I do not have it bookmarked).

At other times it is what is added, inflated by 1/0.955, as you know.

But that’s the reason you see such consistency in new vehicle capacity in your dataset.

What actually happens is the excess energy is shoved into each rated mile, inflating each one by the percentage that the vehicle exceeds the degradation threshold.

Hopefully that makes sense.

The car is not reporting it wrong to itself - if you cross check your results vs. SMT on a new car you’ll see it is close to 99kWh or whatever while you report 96kWh. It’s just the user-facing value which is obfuscated.

If you drive that “96kWh” value vehicle which is actually 99kWh per SMT all the way into the buffer, until it shuts down, you’d see close to 99kWh on the trip meter. Not 96kWh. Or if you drive it to around 0% you might see around 94.5kWh (this does not imply in any way that the 96kWh value Tessie calculates is the usable -it is the full pack within the limits of what is available to the API).

For a vehicle with 95kWh capacity, below the threshold, if you do the same, you’d get 95kWh from Tessie, and on the trip meter: about 95kWh if you drove it to shutdown, and about 91kWh to 0%.

If you were able to gather info from the Trip Meter with the API, and compared to miles used, I think you’d see this behavior (only in vehicles that are brand new AND with NFP above the degradation threshold - not all are as @AAKEE points out). I don’t know whether that data is available in a useful way via the API.

Again; nothing mysterious about it - just the way that Tesla hides some small initial variability in capacity from the new owners and ensures most vehicles display max rated miles when delivered. It’s sort of a top buffer but not really (since (inflated) miles start ticking off right away).

Once below the threshold this fudge factor/capping effect disappears (and you just have the 1/0.955 inflation).

 

[AAKEE]

Here's Tessie with my new Model X. The degradation matches the % shown in the Service Mode Battery Health test

As for the service mode health test…
If we anyway are using the word confusing:
Health test “fake” numbers

That is a model 3 with the 82.1 kWh battery, confirmed via SMT.
The delivered energy varies from 80.7 to 82.07kWh ( Y with that battery also counted).
Most M3Ps showed about 80.5kwh new.
So, we start at - at least 80.5 kWh.

The BMS shows 71.2 kWh nominal full pack and the nominal remaining at 71.5kWh.

71.5 kWh / 0.93 = 76.9kWh as “100%”, if Tesla actually mean health like capacity left/capacity new (thats the word standard for that).

93% would set us at 7% loss, right?
IRL, from the 82.1 new ( remember, 82.07kWh delivered from that battery type
and 82.1kWh full pack when new)

71.5kWh is about 12.9% short of 82.1.
Its about 11% from the degradation threshold, and the energy delivered in the M3P EPA test.

So, the target 76.9kWh for 100% is very low.
I have seen many confusing service mode health tests, and because of this and the other ways of doing it more precise and getting the number self, will be my way.

and the initial capacity matches what is displayed via OBD readers,

The BMS often seems of initially on certain batteries.
With the LG M-50 78.8 kWh the BMS often is way of initially. Its possible that the LG batteries need several cycles to develope the full capacity. Marked 78.8 But often reach 79-79.1 after a couple of months or so according to the BMS. I never dug in to that.
But my MSP had a Capacity at about 98.0kWh at these longer drives when new at the same time as the BMS said 95.7 (it started a progressive
Increase according to the BMS but the capacity was already there.

So the initial BMS value isnt always matching the true capacity.

I will find a few of my older posts later
and send you a few links, but in short the capacity calc I did was like this:

-If the car has been sleeping at least a little
while so the BMS had a decent SoC measure, the true SoC is known to a quite high precision.
-Then we drive and keep track of the delivered energy.
After the drive the car gets to sleep for a while, like 30 minutes or so. This is sufficient for the BMS to get a good measure again. ( checking the SoC with a BMS-tool when parking and after 30 min will show us the probable BMS-off value).
-Well, the Delta of the true SOC before the drive and dito after the drive is the used SOC. For example 50.0%. This needs to be the true SOC, not the displayed SOC.
Used energy / delta SOC = capacity.
So with 50kWh used and 50%’delta, the capacity is 100kWh.

This calc has shown correct when my M3P BMS was way off, and it was
Spot on to the 100-0% drive. The BMS later
Recovered to that exact number!

For the new MSP, with the BMS still very
Low, the calc was spot on with the value the BMS adjusted to in about 1 month or so.

 

[AAKEE]

Start service mode battery test or drive car to 0% -> charge to full -> read charge_energy_added value in the firmware via the API.

I guess you saw my post ( or will) about the health test - just as in many other posts here using that feature ; we do not
Know what Tesla mean the health number. Its not consistent, and it mostly shows lower degradation than we can se by other means.

To the 96.2-thing.

The added energy comes from the fixed energy per mile added.

It is probably not a coincidence that we see 96.1-96.2 as the degradation threshold.
( Nominal full pack 94.6, 627 km out of 637km full range.)

If the BMS has 96 kWh estimate initially (not connected to the degradation treshold to be clear)
a charge from 0-100% will
Show 96 kWh added, regardless of the real energy in the battery.
The +kWh is coupled to the range displayed.

So the BMS kind of set the limit for calculations built on that number I guess.

 

[AAKEE]

For instance, my 2016 P90DL has a usable capacity of 74 kWh.

Did you see the answer from James@tessie?
(Answer)

He confirms your 74kWh is total capacity.
And that Tessie usually find that battery to deliver 84.1kWh - close to the EPA test measure of 84.2 if I do rbembrr it correct.

From the full new value (not present in the BMS for older MS, but well known), your
Car has lost 13.8%.
From the EPA test value, and Tessie normal finding, 12%.

So the square root of (eight years) times ~5% seen in the graphs that points to 14% is not far of from what we find here.
The discussion is often about where to count zero from, but the initial EPA test
Value and the nominal full pack is often agreeing so its not that unfare to use 85.8kWh.

I know its not fun to learn this thinking “10%” and learning that the real value is higher, but I still thinknit is better to know how it actually is.

There is a common issue in threads, trying to kill the common battery myths, by informing about the findings from the research community ( its a lot and it is agreeing and it is valid). Every now and then the statements I or someone else does, that is built on science is questioned by someone that truly believe that he/her has very low degradation despite having used high SOC for long time.

If the questioning is not challanged, other people around may get stuck in the world of battery myths, so to stop that from happening this usually ends in up a slightly broken heart between the owner “with low degradation”’and he’s or her’s battery.

It is newer my meaning to step on anyones toes, but it would be nice to kill the myths or at least the extent of the myths.
But the myths are hard established and the believe in these are so strong. Also, of course -anyone with a firm believe for having a low or acceptable degradation will be disappointed when learning it is not that low (the numbers doesnt matter… its the downward change that do).

 

[randall_s]

It’s already considered.

-By using real data from for example Teslafi.
-By using real data from teslalogger.

The sum of very many reseach tests /reports tell us how batteries behaves.
The research do not have big differences between individual cells or between different research. It is quite consistent.

I took some time to sum that up in formulas before I got my first Tesla a little more than three year back. I did have to adjust my prediction after a while as my logs showed that the average cell temp was lower than my initial estimate.
Other from that the calculation was spot on.

So about one year ago I started asking people if in forums about their usage (charging schedules, daily usage of SOC, where they live, if they have the car outside or in a heated garage etc.).
So far, I guess > 100 cars has been tested against the formulas. Actually only three is outliers, of which two is living outside in cold climates ( @KenC’s is one of these).
The third one is a model S that most probably had the battery changed or a BMS that are in the blue.
All other has been more or less spot on.

The usual thing is someone stating ”I have only 3% after five years at XXYYY miles”.
After looking into it, the owner has gotten a few things wrong and the real degradation is like the text book example. So far happened ~ 10 times here and likewise on a swedish forum.

One big issue is Tessie.
Firat of all the Tessie makers had a few things wrong. Tesdie ise the energy reported by the car (and SOC numbers) as the base to calculate




Tessie is wrong.
Tessies initial value is very often wrong, more wrong than right. Before we helped Tessie They/he thought that the value he calculated was net (exluding the buffer).
We know that this is not the case and we know why. Here is the Slightly shortened version:

Teslas system includes the buffer in the range at 100% but hides the buffer progressively until 0% displayed.
And when charging it un-hides the buffer.

When charging from 0% displayed to 100% the energy charged the car will display as charged ( ”+ XY kWh”) is the same as the total battery capacity. This is a ascertained fact that it work this way. It is checked over and over again, so ”we” know this.

Tessie use the + kWh value reported OTA via the API. (Also shown in the tesla app), and calculates the capacity like
[ +kWh / delta SOC = capacity].

This ends up in the calculation showing the total capacity. Earlier Tessie vendors thought they calculated the ”usable capacity” After a discussion and information given in the Tessie vendors thread, the term was changed to ”capacity”.
(Some of this is around here in that thread:
Tessie thread about capacity ).
Before calling my claims ”bullock” you need to either ask @James@Tessie in that thread or use Scan my Tesla and compare like ”we” already have done.

The way Tessie calculates initial capacity causes a lot of confusion. Its much more often right than wrong. In the older cars case Im sure Tessie used a set number for the capacity as Tesdie was not even alive at the time when the cars was new.
And that set number was the net capacity, as Tessie thought that the calculated number was “net capacity” as well. Which it is not.

The numbers Tessie present is very confusing people, and I actually had to ask Tessie in the thread to get a clear picture.
Most people in Sweden in Facebook groups goes around thinking “I want noy have 3% degradation”.
The issue is that the initial value is often too low (way to low) and people thinks “thats the net capacity/without the buffer”

I know it’s the total capacity.

Actually the test data for both calendar aging and cyclic aging can be transfered to real life.

The do not “coloumb track” everything. Coloumb counting is one of the methods, mostly when the researchers try to understand why things happen.

The degradation for both cyclic and calendar aging graphs is done by cycling the battery a few times and then charge it full and discharge it in a controlled environment and measuring the energy the battery delivers.
The graphs you see is actual loss of capacity, just like you will see it in the car.

You need to understand this ( ):
-It is not “this lab study”. Instead it is litterally hundreds of studies telling the exact (or very close) same thing. Hundreds, literally!
-It is not only the lab research but it is also litterally hundred real life cars checked by calculating the capacity according to the research findings and how the car was used and the result is a very precise hit.

If you go back a bit very many persons used Jeff Dahnnas an example of someone stating a different approcah. He had a lot of theoreticall stuff and the net recommendations was not exactly in line with the rest of the world recommendations
I have seen some old things from him and at that point I thought he had some kind of limitations of what he could say due to the work with Tesla.
We can be very certain that if he says 70 or 75% that is not a number taken from the sole purpose of reducing the degradation. Then he have told you what he think is a good compromise between usable range and degradation (which can only be just a compromise that is very generalized). It could also be a limitation from Tesla about what he is allowed to say.

For the 30% its the same. 30% will not give the lowest degradation. We know this from
A Huge numbers of research reports.
This must also be a compromise. Either limitations from Tesla or like making it certain that he will not recommend something that cause the low voltage
battery to run out of juice and get damaged.
30% is not bad. It isnquite good but it is not the number causing the lowest degradation. The lower the SOC down to 0% the lower degradation. But at 0% displayed we are not very long from the HV batt disconnecting to protect itself causing issues on the LV batt so we need a small margin if we leave the car long time.

Tesla has 15-50% for storage of high voltage batteries in stock ( they can be stored for long time, so 15% is to ensure that it stays in safe levels as long as the battery is in storage.
View attachment 1013303


I suggest to not only listen to a single source, just like not to only read a single research report.
There are more than one person having the knowledge of Jeff Dahn. No shadow over him or his work but we do not know the background to all of hes statements. He most probably has an agreement of what he is allowed to say after working with Tesla.
From what I did see here on TMC the latest information from Jeff Dahn is close
to the data and information I spread here.

So I've posted this twice, but I'll post it again. "74 kWh * 75% charge / 310 wh/mi = 179 miles, which is what my BMS reports as my range at 75% charge". The BMS thinks I have 74 kWH based on the rated range. That's not a total. It's the usable capacity based on rated range. This is as straight forward as it gets.

Did you see the answer from James@tessie?
(Answer)

He confirms your 74kWh is total capacity.
And that Tessie usually find that battery to deliver 84.1kWh - close to the EPA test measure of 84.2 if I do rbembrr it correct.

From the full new value (not present in the BMS for older MS, but well known), your
Car has lost 13.8%.
From the EPA test value, and Tessie normal finding, 12%.

So the square root of (eight years) times ~5% seen in the graphs that points to 14% is not far of from what we find here.
The discussion is often about where to count zero from, but the initial EPA test
Value and the nominal full pack is often agreeing so its not that unfare to use 85.8kWh.

I know its not fun to learn this thinking “10%” and learning that the real value is higher, but I still thinknit is better to know how it actually is.

There is a common issue in threads, trying to kill the common battery myths, by informing about the findings from the research community ( its a lot and it is agreeing and it is valid). Every now and then the statements I or someone else does, that is built on science is questioned by someone that truly believe that he/her has very low degradation despite having used high SOC for long time.

If the questioning is not challanged, other people around may get stuck in the world of battery myths, so to stop that from happening this usually ends in up a slightly broken heart between the owner “with low degradation”’and he’s or her’s battery.

It is newer my meaning to step on anyones toes, but it would be nice to kill the myths or at least the extent of the myths.
But the myths are hard established and the believe in these are so strong. Also, of course -anyone with a firm believe for having a low or acceptable degradation will be disappointed when learning it is not that low (the numbers doesnt matter… its the downward change that do).

That's believable. I initially had higher rated miles than advertised (I think 282@100% compared to 269), but chalked it up to BMS calibration or changes in rated miles calculations over time and did not use OBD2 for some time. 282 miles * .310 kWh/mi = 87.42 kWh. 269 miles * .310 kWh = 83.39 kWh. Could I have actually had 87 kWh at some point? Seem high. 83 kWh seems reasonable.

Maybe I'm confused over terminology, but if total = usable + buffer, the math doesn't work. Assuming rated miles = usable energy / rated efficiency. 74 kWh / .310 kWh/mi = 239 miles, which is my 100% rated range. I would hope the 239 miles my S shows at 100% doesn't include miles below 0%.

Also, the 81.8 kWh usable figure came from Tesla’s hacked Battery Management System exposes the real usable capacity of its battery packs