Range Loss Over Time, What Can Be Expected, Efficiency, How to Maintain Battery Health

[hanswiththe3]

You have about a 6% degradation so looks normal. Did the Service Center tell you why you got that error?

Hi, this was about a year ago that I got that error and I expect it to be from the excessive supercharging I did. It definitely ruined my battery luckily it was under warranty!

The car would not charge when I got the error so I'm sure it's linked to that.

 

[AlanSubie4Life]

I would take reasonable number for the initial capacity, and then reduce the capacity with the estimated calendar aging according to the NCA graphs.

Yeah, this jumping around is very likely not physical (ruling out effect of temperature). As was pointed out by @scottf200 the software/firmware has also been updated and that can lead to “recoveries” like this. I’d kind of expect that to happen more for the newer 2170L cells, as Tesla accounts for the aging behavior and tweak their firmware to accurately assess energy content of these cells as they age, which may be very slightly different than prior cells. So maybe that explains this. Or maybe they were incorrectly assessing initial capacity of cells with less aging.

Certainly that sort of behavior has been seen on the old cells as well, though I haven’t seen a lot of reports lately.

I wonder if the current value is closer to correct and it was the initial value that was never high enough (which would kind of fit the calendar aging profile). I guess @conv90 will let us know as things progress. Hopefully the new higher levels persist, with standard aging curve from here. Or it could be as you say, some weirdness with cars made in China (no idea where this car was made).

 

[KenC]

I agree. The fluctuations are too extreme to say much about battery degradation with resolution any better than about +- 5%. Btw, I have a very similar capacity chart as yours with a big drop in Winter and recovery in Spring. Not sure what the weather is like where you are but here in NJ USA I am fairly certain the temperature fluctuation has influenced the capacity estimation performed by the BMS. I do not know what the formula is, but I am happy that mine has also appeared to have "recovered" as yours has. It is nice to have this data recorded but I highly doubt it indicates that my battery somehow reversed itself and began becoming "healthier" after a while. It is more likely just a demonstration of how meaningless the data is on any short term scale.

View attachment 802081

I thought TeslaFi, like several 3rd-party apps, used the SOC api that is temperature-affected. The result being that you get estimated range dips in cold weather. I have Stats, and Stats used the SOC api that was temp-affected until Jan of 2021. You can see my Stats estimated range in my avatar. On the left half, my data shows dips in Winter. On the right half, the Stats developer switched SOC apis to the one that isn't temp-affected, and as you can see, no more dips for cold temps.

 

[Steve446]

Percent is unitless though, so it tells you nothing, other than what portion of your maximum energy you have left. No units.

Yes, at least at some other places theres usually a hard battle if the km/miles or percent should be used.
A degraded battery still shows 100%, but it will not even do 100% of the lower-than-EPA practical range it had as new.
The km/mile scale actually show the energy content (In a odd scale, but it does).

So what is the big deal with not showing kWh remaining in the battery? Is this related to Tesla not wanting to declare battery capacity in the car's specifications?

 

[AlanSubie4Life]

thought TeslaFi, like several 3rd-party apps, used the SOC api that is temperature-affected.

Yeah, if that’s true (and I don’t know), it makes TeslaFi useless for this. But I suspect it can’t be true, because I thought @AAKEE ’s TeslaFi plots look pretty steady and he lives in a very cold place. Though maybe he has 5 excess kWh to work with, lol.

So what is the big deal with not showing kWh remaining in the battery? Is this related to Tesla not wanting to declare battery capacity in the car's specifications?

It’s a little weird. I think because kWh are less approachable for most people and also people would become obsessed with it. But people get obsessed anyway, so it’s not clear that not providing the display of this value is helpful.

A more jaded view would be that the fog of uncertainty around rated miles (which there should not be, but people can’t seem to figure out how straightforward they are) makes people less likely to understand when they have reached the warranty threshold. But this happens so infrequently this jaded view doesn’t make much sense.

My theory is that their current method allows them to (mostly) hide vehicle-to-vehicle variation of initial capacity of 1-2% (through the degradation threshold framework), which very much DOES reduce customer complaints. That would be harder to justify with a kWh measurement/display, using kWh that are larger than actual kWh, thus capping at a certain value even when actual kWh capacity is higher. (This is what they do now, but with rated miles.) Can you imagine Tesla owners with 1-2kWh more than other owners, just right out in the open? They would be so annoying!

In the end, I think it is just another layer to create distance from the important metric, and thus reduce customer obsession and questions. Not clear that it accomplishes this objective.

 

[Steve446]

I have not fully undertand:
I was thinking you were charging the car to 55% at night and START to use the car immediately after the end of the charge (or as fast as you can).
Then you say that you had to change habit letting it (instaed) sleep another 1,5 to 2 hours before to use it to go work?

What I wanted to say with my posted graphs (here the update one at today) is that it seems IMPOSSIBLE to determine the ACTUAL kWh capacity even having SMT, because the flutuations are too big.
View attachment 802041

In my example, which is the correct kWh capacity of my march 2021 82.1 kWh "Full Pack When new" Pana Battery? 73,8-73,9 (the february level after 11 months?) OR the 77,7-77,8 kWh (the January level ) Or the 77,8 (the level of NOW)?
OR probably it's something in between? (76,00kW?).
If someone ask to me, which is the level of degradation of your car battery after 14 months? I have to reply : " Im not sure, it's Between -2,6% to -8%"
What I wanted to say is that we are nerds enough to look at the fraction of the Watts, and using millivolt precision to judge things, then we have results where there are 4 kWh difference between readings.

If you really want to, I don't believe that it's so difficult to estimate "degradation" using SMT. If you plot NFP against elapsed time since delivery (I assumed a linear plot) you will get a graph which you can use to estimate pack degradation over the passage of time. I would not add the Full Pack When New to the plot as a starting value since I understand that's hardwired data. Just use the data measured/extracted and, as much of it as possible ie multiple months of data. Of course it will not be perfect but in my opinion this will likely give you a more meaningful value that Full Rated Range Vs Time because with that derived parameter, the "degradation threshold" masks the real curve (refer previous posts from @AlanSubie4Life about this this last point). The NFP plot is the least noisy graph I've seen for my car but, it is what it is and just a guide or estimation - nothing's perfect though

 

[Steve446]

Yeah, if that’s true (and I don’t know), it makes TeslaFi useless for this. But I suspect it can’t be true, because I thought @AAKEE ’s TeslaFi plots look pretty steady and he lives in a very cold place. Though maybe he has 5 excess kWh to work with, lol.



It’s a little weird. I think because kWh are less approachable for most people and also people would become obsessed with it. But people get obsessed anyway, so it’s not clear that not providing the display of this value is helpful.

A more jaded view would be that the fog of uncertainty around rated miles (which there should not be, but people can’t seem to figure out how straightforward they are) makes people less likely to understand when they have reached the warranty threshold. But this happens so infrequently this jaded view doesn’t make much sense.

My theory is that their current method allows them to (mostly) hide vehicle-to-vehicle variation of initial capacity of 1-2% (through the degradation threshold framework), which very much DOES reduce customer complaints. That would be harder to justify with a kWh measurement/display, using kWh that are larger than actual kWh, thus capping at a certain value even when actual kWh capacity is higher. (This is what they do now, but with rated miles.)

In the end, I think it is just another layer to create distance from the important metric, and thus reduce customer obsession and questions. Not clear that it accomplishes this objective.

Yes I think you've made the point quite nicely

 

[Steve446]

View attachment 802183
Ok..I have grappled with this set of graphs everyday for months and each time I see something different....
Obviously it’s the middle graph of NMC batteries that is most worrying (because I think that is the LG75...my battery)
But graph B almost matches graph A...so I was thinking that is the sudden drop off on graph E might just be it’s temporary loss of storage and not long term damage...because every LG battery would fail the warranty period

Good point but bear in mind that this data whilst quite clear in its conclusions is really a guide for establishing a charging and storage strategy and does not necessarily accord with real world conditions. Example - based on the data, I doubt you would store your battery @ 50°C and 70% SoC for 10 months, if you did you would likely be calling in the 70% guarantee after one year... More realistically, this data, as @AAKEE has explained, does suggest that when it's not used, if we keep our car at a lower SoC (20-40%) and charge just before use, we should potentially be able to lower calendar aging. A low storage temperature is just an added bonus.

 

[Bouba]

Good point but bear in mind that this data whilst quite clear in its conclusions is really a guide for establishing a charging and storage strategy and does not necessarily accord with real world conditions. Example - based on the data, I doubt you would store your battery @ 50°C and 70% SoC for 10 months, if you did you would likely be calling in the 70% guarantee after one year... More realistically, this data, as @AAKEE has explained, does suggest that when it's not used, if we keep our car at a lower SoC (20-40%) and charge just before use, we should potentially be able to lower calendar aging. A low storage temperature is just an added bonus.

Agreed..but storing a battery @ 70% and 50 degrees for 10 months spread over an eight year period is a different matter

 

[Steve446]

Agreed..but storing a battery @ 70% and 50 degrees for 10 months spread over an eight year period is a different matter

I'm glad I don't live somewhere "hot"! Nice and temperate Western France is fine for me

 

[conv90]

Yeah, this jumping around is very likely not physical (ruling out effect of temperature). As was pointed out by @scottf200 the software/firmware has also been updated and that can lead to “recoveries” like this. I’d kind of expect that to happen more for the newer 2170L cells, as Tesla accounts for the aging behavior and tweak their firmware to accurately assess energy content of these cells as they age, which may be very slightly different than prior cells. So maybe that explains this. Or maybe they were incorrectly assessing initial capacity of cells with less aging.

Certainly that sort of behavior has been seen on the old cells as well, though I haven’t seen a lot of reports lately.

I wonder if the current value is closer to correct and it was the initial value that was never high enough (which would kind of fit the calendar aging profile). I guess @conv90 will let us know as things progress. Hopefully the new higher levels persist, with standard aging curve from here. Or it could be as you say, some weirdness with cars made in China (no idea where this car was made).

My car is a Freemont made in USA with the 82.1 E3LD batt (1154423 -00 T). It's a 1st Quarter (delievered in Italy in March 2021). I really don't know if MIC or Made in USA.
I have connected SMT 140 km after the delivery and the capacity never passed 79,6 - 79,7 kWh. only one time passed 80,1 kWh but only when charged to 100% and it returned to 79,6 after 5 km of drive after charge.
The only doubt I have is that SMT showed at the beginning that over 50-53 kWh where already used (and a full cycle charge was done). So probably the car was charged to 100% before to go on Boat and then left discarged to 35% (during the trip to Europe) when I've picked up at the delivery.
So after 2 or 3 days SMT was showing : 2 discharge cycles and 3 charge cycles.
My battery stayed 79,6 kWh for about 4 or 5 months with same around 500-504 km range and it has NOT a thresold effect, because my battery always stayed at 79,6 well under the threshold.

 

[Apprunner]

Hi, this was about a year ago that I got that error and I expect it to be from the excessive supercharging I did. It definitely ruined my battery luckily it was under warranty!

The car would not charge when I got the error so I'm sure it's linked to that.

Did you get a new battery? Also, Supercharging wouldn't cause that problem. Its actually generally very safe for the battery with the end result being slightly higher degradation vs. at-home level 2 charging.

 

[AAKEE]

So probably the car was charged to 100% before to go on Boat and then left discarged to 35% (during the trip to Europe) when I've picked up at the delivery.
So after 2 or 3 days SMT was showing : 2 discharge cycles and 3 charge cycles.
My battery stayed 79,6 kWh for about 4 or 5 months with same around 500-504 km range and it has NOT a thresold effect, because my battery always stayed at 79,6 well under the threshold.

I think there is rules for overseas shipping that set the max SOC for the shipping, in general max 50% SOC.
walleniuswilhelmsen use 50% as max.

 

[AAKEE]

View attachment 802183

But graph B almost matches graph A...so I was thinking that is the sudden drop off on graph E might just be it’s temporary loss of storage and not long term damage...because every LG battery would fail the warranty period

The losses shown is (forever) lost capacity. No recoverable degradation.


The LG cells Tesla use may have a newer chemistry, perhaps changing the calendar aging compared to this test.

Otherwise, with normal temperatures we will probably not see a noticable difference. In very hot climates, its possible that the LG batteries degrade more than Panasonic do.

As most batteries is not held at 50C most of the time, they will not follow the 50C degradation curve. I do not think there is a overhanging danger that the battery do not not hold up the eight years.

 

[AlanSubie4Life]

around 500-504 km range and it has NOT a threshold effect, because my battery always stayed at 79,6 well under the threshold.

Yep. Definitely you were below the threshold at all times. Some possibilities:

1) Current estimate is wrong.
2) Initial estimate was too low, as well as the estimate when you reached your minimum. Resolved by software.
3) Some sort of environmental factor changing which altered capacity (seems unlikely).

its possible that the LG batteries degrade more than Panasonic do.

Every LG product I have owned has degraded. First was the notorious initial retina MacBook display years ago. Second was my OLED TV screen, which while initially beautiful, now suffers from catastrophic burn-in. Glad I don’t own an LG battery.

 

[Bouba]

The losses shown is (forever) lost capacity. No recoverable degradation.


The LG cells Tesla use may have a newer chemistry, perhaps changing the calendar aging compared to this test.

Otherwise, with normal temperatures we will probably not see a noticable difference. In very hot climates, its possible that the LG batteries degrade more than Panasonic do.

As most batteries is not held at 50C most of the time, they will not follow the 50C degradation curve. I do not think there is a overhanging danger that the battery do not not hold up the eight years.

Hi AAKEE, the graphs A and B show damage to the batteries but they are almost identical...yet the NMC battery in graph E falls off a cliff...almost reaching the Tesla replacement guarantee in only 10 months!
In summer here a car can easily get to fifty degrees (if you didn’t put cabin overheat protection on)...of course only for a few hours a day and only at the height of summer...but it would add up over the eight year warranty period.
Just out of interest, I built a car porch to protect my Tesla from the sun

 

[AAKEE]

Hi AAKEE, the graphs A and B show damage to the batteries but they are almost identical...

No.

Figure a, b and c shows the anode potential. This is only a part of the battery but they did find a clear connection between low anode potential and high calendar aging.
You need to read this if you start to dig into deeper mechanisms :
Calendar aging

a,b and c do not represent any damage, only the potential which could be/probably is tested on a brand new cell. So these pictures do not show any damage information at all.

The loss found in the calendar aging tests is to the absolute most part not reversible. The SEI (Solid Electrolyte Interphase) is causing the dominant part of calendar aging(if my memory doesn’t fail on me). The growing thickness of SEI is permanent.
There might be research(I’ve seen some) about reverting SEI with special methods, but this will not happen in hour Teslas today so we should think of SEI as peemanent damage.

 

[conv90]

Yep. Definitely you were below the threshold at all times. Some possibilities:

1) Current estimate is wrong.
2) Initial estimate was too low, as well as the estimate when you reached your minimum. Resolved by software.
3) Some sort of environmental factor changing which altered capacity (seems unlikely).



Every LG product I have owned has degraded. First was the notorious initial retina MacBook display years ago. Second was my OLED TV screen, which while initially beautiful, now suffers from catastrophic burn-in. Glad I don’t own an LG battery.

I'm for the 2 scenario (or a mix of the whole 3 scenarios).
It's because I'm receiving feedbacks from italian friends with my same battery and same age and all are experiencing raise in the NFP (or calculated range at 100%).
A freind of mine (with a 2021 3rd Quart Model 3 Perf , just sent me a screen with a NFP of 79,5 kWh (it was near 77 some weeks a go).
What the 2) scenario is strange is: How tesla can change the software to show more miles/kWh if the constant (around 159 Wh/km) is the same??.

 

[Steve446]

No.

Figure a, b and c shows the anode potential. This is only a part of the battery but they did find a clear connection between low anode potential and high calendar aging.
You need to read this if you start to dig into deeper mechanisms :
Calendar aging

a,b and c do not represent any damage, only the potential which could be/probably is tested on a brand new cell. So these pictures do not show any damage information at all.

The loss found in the calendar aging tests is to the absolute most part not reversible. The SEI (Solid Electrolyte Interphase) is causing the dominant part of calendar aging(if my memory doesn’t fail on me). The growing thickness of SEI is permanent.
There might be research(I’ve seen some) about reverting SEI with special methods, but this will not happen in hour Teslas today so we should think of SEI as peemanent damage.

Damage information? With respect AAKEE, surely the data does show (irreversible?) degradation. Let's use the publication's Figure 2 with NMC stored @ 50°C. When 70-80% SoC is used, at the end of the author's 9-10 months storage, the relative capacity appears to be around 83% of the original starting value.

 

[AAKEE]

What the 2) scenario is strange is: How tesla can change the software to show more miles/kWh if the constant (around 159 Wh/km) is the same??.

The constant is the same, it is the estimated capacity that changes.