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

[STS-134]

Thats not correct.

Lithium batteries is fine down to 0%*.
The 20 and 10% thing is a myth.
A lot of the “saying” on the net about lithium battery care is not correct.

*) 0% is defined as the low voltage limit set for the cell, for Tesla Panasonic cells its 2.50V/cell. The battery shut down before reaching the cell low voltage limit.

Actually, lithium batteries is more than just fine at low SOC.

No problem.

The lower SOC the better from a battery perspective.
Calendar aging is lower the lower the SOC.

Cyclic aging is lower at lower SOC.
Cyclic aging is lower the smaller the cycles is.

This image( taken from research) show calendar aging, low SOC is best so no need to be afraid of 20 or 10%.

View attachment 796478


And for the cyclic part, 0-10% SOC during cycles is the best to keep degradation low.
Below, a picture from another test report.
Lowest degradation will be when cycling at very low SOC.

View attachment 796483

So, no reason to worry at all about low SOC.

Unless you're doing this from 100%, in which case, you have a high upper voltage plus high average depth of discharge, which is really, really bad.

 

[AAKEE]

It is indeed impressive. @KenC, do you use SMT? I was just wondering if you had any data for the NFP and energy buffer of your car?

As on screen range at the battery actually is a battery capacity measurement, we can calculate the Nominal full pack from this value.
And as the buffer is 4.5% of the NFP, we can calculate this as well.

Its a ‘19 LR? 245Wh/mi ?

245 x 310 = about 76kWh?
And the Buffer, 4.5% of this.
(At least, as the range is limited at 310mi)

 

[AlanSubie4Life]

I'm at 93%ile, which means there are 7% better than mine, for the same mileage and model. Yes, same mileage and model. I specifically asked the developer if he was also comparing a 2021 to my 2018, and he said he wasn't.

Funny thing is that was 2000 miles ago, my range is up a bit to 311 estimated, and yet my percentile is down to 93% from 95%. Whatever. Must be limited datapoints or something.

Yes, we discussed this before. You're basically 100th percentile! Pretty sure the developer is just fitting a Gaussian based on his sample data statistics (mean and standard deviation). This creates the silly looking bell curve above. Clearly he is not a statistician - it's not a Gaussian distribution - it has limits, so the results will be incorrect. I'm not a statistician either so I don't know what would be an appropriate distribution.

If Tesla holds about 5% charge in reserve below 0% showing, is it okay to regularly return home from a commute with a 5% charge (for overnight charging)? That’s actually a 10% battery charge remaining, right?

As I recall from @AAKEE's statements (whom I generally trust since he's always got the data), (and now he posted just before I did...got distracted, wrote this a while ago...) there's not a lot of evidence that very low SOC is bad for the battery (we're not talking about bricking it here). There have been threads around here which have hypothesized that there are mechanisms where low SOC can induce stress on the battery (creating a "bathtub" of optimal SOC between 20% and 50% or whatever), but quantifying the effect of low SOC is difficult.

But realistically, unless you have a real stretch of a commute (you did not specify), none of the methods described here of limiting charge to 50% (to reduce capacity loss) make any sense, if it means you're routinely ending at 5%. That's just too stressful - pick a higher SOC to start at, so you end at 20%! 65% (or whatever) to 20% is fine if that's what it has to be. Or you can do 55% to 10% if that's comfortable enough for you.

The main thing is to keep it convenient. All of these contortions may have only a modest impact on your capacity loss, so it's not worth being inconvenienced every day, risking needing a tow if it happens to rain really hard on a given day and kills your efficiency. You lose features below 20% so it's nice to keep it above that.

 

[Steve446]

As on screen range at the battery actually is a battery capacity measurement, we can calculate the Nominal full pack from this value.
And as the buffer is 4.5% of the NFP, we can calculate this as well.

Its a ‘19 LR? 245Wh/mi ?

245 x 310 = about 76kWh?
And the Buffer, 4.5% of this.
(At least, as the range is limited at 310mi)

One of the problems I have observed with values derived from, or by calculation, or associated with my M3's BMS is that, depending on their origin, and the conditions under which they are measured, they can show variance. That can lead to imperfections in the apparent data trend. It might be the same for other Tesla cars and EVs. I sincerely believe that's an issue illustrated by your calculation above ie they are estimations not measurements. I have learnt all that from interpreting data from my previous professional life. Data is not always what it seems. None of this, my remarks or point of view, is aimed at criticizing Tesla, you, or @KenC's data or experience or value. Quite the contrary. I'm just making what I believe to be a valid point

 

[AlanSubie4Life]

Its a ‘19 LR? 245Wh/mi ?

245 x 310 = about 76kWh?

Yes, 245Wh/mi. But actually these vehicles had about 78kWh when new (there are documented readings from SMT for 2019 vehicles), so they simply had a degradation threshold of 76kWh. They didn't show capacity loss until they dropped below 76kWh. (Effectively they started with more like 318 rated miles.)

 

[AAKEE]

I'm at 93%ile, which means there are 7% better than mine, for the same mileage and model. Yes, same mileage and model. I specifically asked the developer if he was also comparing a 2021 to my 2018, and he said he wasn't.

As you can see, I'm at 93%ile, even though I'm at 311 miles estimated for a 2018 LR-AWD.

Interresting to see how you could be “only” at 93’ percentile and the range still is full.
I guess a ‘18 Wont go above this number?

For the other ones, I would say that the very low degradation is not strange.

First thing, keeping low SOC will preserve the battery. We know that. So KenC isn’t just a lucky guy with luck in the battery lottery. He is doing the right thing in terms of preserving the battery.

Also, at least for my car, staying low most of the days during a month will set the BMS a little of to the high side.
I still have full range after 40 K km/ 25K miles and 16 months. Each full charge have shown full range.

As I suspected that the BMS showed the range and NFP a little high( still about 80.5-81kWh where most people started at 80.0-80.5 or lower) I planned to drive the car down to low SOC. 0% wasnt low enogh, so I drove it down to -2%, and this did set the NFP down to 79.4kWh( which is 500km out of 507 new).
But it only took a few days at the normal SOC until the NFP started to increase step by step to 80.5kWh.

Not to take KenC’s good range from him or take the focus from his good battery, but to
emphasize that this behavior is expected from batteries that are kept at low SOC, I post my range/battery degradation from teslafi:
Teslafi only get the whole percentage of the SOC so there is much variation in the range that is not seen in the car, and not reflected in the NFP.
(This is a ‘21 M3P, so 507km EPA-range/“new” range)

 

[AAKEE]

Yes, 245Wh/mi. But actually these vehicles had about 78kWh when new (there is documented readings from SMT for 2019 vehicles), so they simply had a degradation threshold of 76kWh. They didn't show capacity loss until they dropped below 76kWh. (Effectively they started with more like 318 rated miles.)

Thanks!

One of the problems I have observed with values derived from, or by calculation, or associated with my M3's BMS is that, depending on their origin, and the conditions under which they are measured, they can show variance. That can lead to imperfections in the apparent data trend. It might be the same for other Tesla cars and EVs. I sincerely believe that's an issue illustrated by your calculation above ie they are estimations not measurements. I have learnt all that from interpreting data from my previous professional life. Data is not always what it seems. None of this, my remarks or point of view, is aimed at criticizing Tesla, you, or @KenC's data or experience or value. Quite the contrary. I'm just making what I believe to be a valid point

If you look at my teslafi graph you see a lot of variation as teslafi get the current range(as seen on the screen, but with two decimals I think) and the SOC only is sent OTA by the whole percent with no decimals.
Teslafi calculates the 100% SOC range by dividing the current range with the SOC but as it is the whole percentage the resulting number will vary with the rounding.

The NFP for my car is quite steady. I can see the NFP at the same value for a week but the teslafi range jump about 5-6% daily.

But if I do a full charge and look at the range(as long as the maximum range isnt reached), I get the same number as Scan My Tesla show me, this because each km(or mile) have a fixed energy.
So when actually reading the full charge number we can calculate the NFP into about 0.15kWh precision.

The caviat for the KenC car calculation I did is that the degradation threshold hide energy above that level.

 

[AlanSubie4Life]

I have observed with values derived from, or by calculation, or associated with my M3's BMS is that, depending on their origin, and the conditions under which they are measured, they can show variance.

Once again, lagging @AAKEE here. Haha.

There's two kinds of error, extrapolation error, and estimation error.

The extrapolation error isn't very interesting - just don't pay any attention to extrapolated 100% values until you're above 90% SoC. Problem solved. (You can see exactly what the limits are of the extrapolation error with simple calculations. For example 150 miles (the API actually has access to close to ~0.5mi increments since it can also know the km, so in this case I mean 149.75 to 150.25) at 50% SoC could mean ~150/0.495 = 303 miles or ~150/0.505 = 297 miles. Actual spread is a little wider, after accounting for 149.75-150.25 potential error. The actual value could be anywhere in that range, it's unknown what is "correct". You can get extrapolations that are off by 100 rated miles at low enough SoC (easy to experiment with and people have provided pictures here). Anyway, not very interesting. It's just a limitation of the API.

The estimation error is a different thing and is the BMS estimation error. The error can be as much as a few % of NFP, usually less, but it can vary. But this is present for both SMT and for the rated miles display, and tends to be fairly slowly varying, but can have jumps on occasion (NFP often seems to adjust in "chunks" for unknown reasons - it's not physical).

So the rated miles display displays the same information as SMT; there's not really any additional information provided on your NFP, EXCEPT when your battery is above the degradation threshold (which @KenC 's battery may sometimes be). A little bit better resolution (to within 0.1kWh), but that's it. The estimation error swamps that precision, so it's not really very consequential to have that extra resolution. Obviously SMT provides a great deal of other very useful information, but nothing really additional regarding NFP.

Only some vehicles have significant differences in FPWN and the degradation threshold. (For example 2018, 2019 AWD, and the 2021 AWD (non-P) with "82.1kWh" battery). And once they have capacity loss to below the degradation threshold, SMT and the car are displaying approximately the same information.

that the degradation threshold hide energy above that level.

Yes, and to clarify for folks (it's described elsewhere), this extra energy is still indirectly observable in the car without SMT, but it requires extremely careful correlation of the trip meter to rated miles use, on a long trip. This is because extra energy above that degradation threshold is actually stuffed into all the miles below the threshold (the energy content of each rated mile is expanded when the NFP exceeds the degradation threshold, by the ratio of those two values). So there's not some upper buffer where you can drive for several miles without seeing rated miles count down.

In this specific example, the 245Wh/rmi becomes 78/76 *245Wh/rmi = 251Wh/rmi (this is NOT for displayed rated miles, this is the "charging" rated miles energy content) when the vehicle is new (and the displayed rated miles would be 95.5% of this, 240Wh/rmi, with the trip meter correspondence being ~1% less, about 237Wh/rmi). These values plateau when below the degradation threshold at 245Wh/rmi, 0.955*245Wh/mi = 234Wh/rmi, and ~0.99*234Wh/rmi = ~231-232Wh/rmi. (The 1% factor is "heat loss" uncounted energy and can vary a small amount.) Notably, the car charging constant does not reflect this expansion (the 245Wh/rmi is what you'll see for charging event display, even if the actual effective value is 250Wh/rmi). To be clear: these specific values are for the 2018/2019 AWD Model 3, and do not apply to any other model year or variant, but the basic behavior is the same for other vehicles.

This has all been verified by people on 2021 vehicles and cross checked with SMT, and is documented elsewhere here. (It's long been suspected, but definitive evidence was gathered last year.)

 

[DrChaos]

Thats not correct.

Lithium batteries is fine down to 0%*.
The 20 and 10% thing is a myth.
A lot of the “saying” on the net about lithium battery care is not correct.

*) 0% is defined as the low voltage limit set for the cell, for Tesla Panasonic cells its 2.50V/cell. The battery shut down before reaching the cell low voltage limit.

Actually, lithium batteries is more than just fine at low SOC.

No problem.

The lower SOC the better from a battery perspective.
Calendar aging is lower the lower the SOC.

Cyclic aging is lower at lower SOC.
Cyclic aging is lower the smaller the cycles is.

This image( taken from research) show calendar aging, low SOC is best so no need to be afraid of 20 or 10%.

View attachment 796478


And for the cyclic part, 0-10% SOC during cycles is the best to keep degradation low.
Below, a picture from another test report.
Lowest degradation will be when cycling at very low SOC.

View attachment 796483

So, no reason to worry at all about low SOC.

The FCE (Full cycle equivalent?) numbers on the Y-axis are really large. Even 2000 cycles for a non-commercial battery should be a long distance and length of time for ordinary personally owned cars.

 

[Steve446]

Once again, lagging @AAKEE here. Haha.

There's two kinds of error, extrapolation error, and estimation error.

The extrapolation error isn't very interesting - just don't pay any attention to extrapolated 100% values until you're above 90% SoC. Problem solved. (You can see exactly what the limits are of the extrapolation error with simple calculations. For example 150 miles (the API actually has access to close to ~0.5mi increments since it can also know the km, so in this case I mean 149.75 to 150.25) at 50% SoC could mean ~150/0.495 = 303 miles or ~150/0.505 = 297 miles. Actual spread is a little wider, after accounting for 149.75-150.25 potential error. The actual value could be anywhere in that range, it's unknown what is "correct". You can get extrapolations that are off by 100 rated miles at low enough SoC (easy to experiment with and people have provided pictures here). Anyway, not very interesting. It's just a limitation of the API.

The estimation error is a different thing and is the BMS estimation error. The error can be as much as a few % of NFP, usually less, but it can vary. But this is present for both SMT and for the rated miles display, and tends to be fairly slowly varying, but can have jumps on occasion (NFP often seems to adjust in "chunks" for unknown reasons - it's not physical).

So the rated miles display displays the same information as SMT; there's not really any additional information provided on your NFP, EXCEPT when your battery is above the degradation threshold (which @KenC 's battery may sometimes be). A little bit better resolution (to within 0.1kWh), but that's it. The estimation error swamps that precision, so it's not really very consequential to have that extra resolution. Obviously SMT provides a great deal of other very useful information, but nothing really additional regarding NFP.

Only some vehicles have significant differences in FPWN and the degradation threshold. (For example 2018, 2019 AWD, and the 2021 AWD (non-P) with "82.1kWh" battery.). And once they have capacity loss to below the degradation threshold, SMT and the car are displaying approximately the same information.

Thanks for your comments and you have made good points (@AAKEE too). I've clearly opened a Pandora's box. To be unambiguous. The data we see sometimes, is only as good as we think it is. it's often still an estimation or derived parameter and not a measurement. In my previous life, regulatory agencies like FDA and the EMA always had a problem with data measured or estimated in different places in electronic instrumentation but which came from the same original source. It gets noisy. They would only believe that data is the "same" if you could prove it. I'm getting boring, so, my point is that some of the data shown on the forum regards "degradation" (or lack of it), should be interpreted with my above remark in mind. No criticism to Tesla or any of you involved in this discussion. We all have our opinions which makes the discussion objective!

 

[AlanSubie4Life]

it's often still an estimation or derived parameter and not a measurement.

The BMS value is most definitely just an estimation based on the vehicle's measurements of energy use. It inherently will always have error, but of course it's very important to Tesla that it be fairly close to "correct," reflecting actual energy content of the battery, otherwise people will run out of energy on the road - the vehicle will say it can make a trip that it cannot. (Can't have situations like with some cell phones where suddenly the SOC drops 50% in a few minutes, with a mild temperature change, due to bad estimation - though modern cell phones are improving in this regard.) All the same, on some vehicles, possibly only transiently, these estimations can have significant error. And software updates can alter the BMS estimations as well. The CAC can also be reset, which is a whole other sort of error, which takes time to settle out.

My impression, based only on my vehicle, is that the BMS estimates are almost always within 2-3% of the actual value (so total variation will be a very tight range of 4-6%), and are often even better than this. (Obviously this is all relative to the trip meter, which is a measurement, which has a tolerance as well, but is presumably what the BMS algorithm relies on for input. You could imagine vehicles where this measurement is "off" being persistently optimistic or pessimistic about actual capacity of the battery, but there'd be no way to really see this, except by driving an identical trip with another identical vehicle (in every way, basically impossible) and noting that your consumption was consistently lower or higher than the "control" vehicle. And you'd consistently do worse (or better) than the Tesla trip planner estimates - but there'd be no way to really see this except with that control vehicle.) That being said, you'd expect a vehicle that had optimistic capacity measurements to show an abnormally high NFP in SMT when new (so that would require SMT or the method above) - but you'd have no way of knowing if that was actually extra energy or just an optimistic estimate.

In @KenC's case, we can't know this since he didn't have SMT, but the delta (the % change) is what would be important. My guess is his vehicle started around 78kWh like most others but is now at ~76kWh. It doesn't really matter whether the actual measurement is correct or not. Just the % change. Is it possible his vehicle was wildly optimistic and started him at 80kWh (meaning he has lost 5% not 2.5%)? Sure. I think unlikely though - but we can't know. Most likely he hasn't had a lot of capacity loss because of how the battery has been stored.

Only Tesla could state and prove that the API (and the car itself) is deriving its rated miles display from the NFP tracked by the BMS (accessible in service mode apparently and also from SMT), but all observations indicate that this is the case. (All we can do is correlate.)

 

[AAKEE]

The FCE (Full cycle equivalent?) numbers on the Y-axis are really large. Even 2000 cycles for a non-commercial battery should be a long distance and length of time for ordinary personally owned cars.

Well, the cycles is only 10% DoD (depth of discharge). Batteries hold up much better with small cycles / DoD. It was not NCA but a miced chemistry of LMO and NMC if my memory didnt fail.
That graph was not ment to guarantee x thousand cycles but to show the principle of lithium battery cycling at different SOC.

There are other research reports but not with such clear point in ome picture.

For NCA(as Tesla/Panasonic) there is a report showing that 600FCE in 10% DoD, thats 6000 10% cycles lost 10% when cycled around 70% SOC (75-65%) but only lost 2% when cycled around 30%(35-25%) SOC.
600FCE should be around 200.000-250.000km or 125-160K mi, for 10% loss.
(This was not panasonics Tesla cells but NCA chemistry if my memory serves).

 

[AAKEE]

there's not a lot of evidence that very low SOC is bad for the battery (we're not talking about bricking it here). There have been threads around here which have hypothesized that there are mechanisms where low SOC can induce stress on the battery (creating a "bathtub" of optimal SOC between 20% and 50% or whatever), but quantifying the effect of low SOC is difficult.



The main thing is to keep it convenient.

Yes, the bathtub picture was made by a TMC memeber,( @Zoomit ) buy guestimating but not really on scientific data. The graph is not completely wrong, it incorporates temperature and high SOC as negative impact on the degradation which is right.
( Battery Degradation Scientifically Explained

Zoomit is clear with that he doesnt really know for sure and that graph was by his intution. Still people attack my posts with that picture(not on this forum ) taking it as the complete truth. I guess the member @EV-Tech Exp isnt following this thread so I wont hit hard, but some of the data he gave in that thread do not have any back up in the scientific research reports, meaning that the tips he gave for adjusting that picture probably is not valid, or at least should be discussed thouroghly before adopting those theories.

I would like to turn the first bolded(by me) statement the other way around:
There is a lot of evidence that very low SOC is not bad at all.

I should start by making this clear: 0% SOC is per definition the minimum voltage of the battery cell specified by the vendor. For Tesla Panasonic NCA this is 2.50V. This is also the most common min voltage for lithium ion batteries that have 4.20V/ cell as maximum voltage = also the definition for 100% SOC.
Going below the minimum voltage and draining the cell until it show 0V is not the same thing as 0% SOC. Doing this can cause damage, but it is random according to the research and some cells dont get damaged at all. Anyway, the BMS protect the battery from this by disconnecting it (with connector relays) before it reaches the minimum value.

The reserachers, are togheter telling us the same thing from hundreds of battery tests: 0% SOC is not unsafe, infact its the best voltage/SOC to preserve LiB’s at during storage.

Its also really safe to say that the lower the SOC the lower a defined cycle wear, like a 10% cycle wear less the lower the SOC is.
10% down to 0% is causing the least degradation and give the longest life. There is a few research reports (or one only?) where the lowest degradation was at 20-10%, followed by 10-0%, but still, the lowest degradation is in the bottom of the SOC.

All degradation examples for wear and life is per Full Cycle Equivalent so the size of a cycle is already taken into account and the comparison is related to how many miles a EV battery can be used.

 

[DrChaos]

Yes, the bathtub picture was made by a TMC memeber,( @Zoomit ) buy guestimating but not really on scientific data. The graph is not completely wrong, it incorporates temperature and high SOC as negative impact on the degradation which is right.
( Battery Degradation Scientifically Explained

Zoomit is clear with that he doesnt really know for sure and that graph was by his intution. Still people attack my posts with that picture(not on this forum ) taking it as the complete truth. I guess the member @EV-Tech Exp isnt following this thread so I wont hit hard, but some of the data he gave in that thread do not have any back up in the scientific research reports, meaning that the tips he gave for adjusting that picture probably is not valid, or at least should be discussed thouroghly before adopting those theories.
View attachment 796571


I would like to turn the first bolded(by me) statement the other way around:
There is a lot of evidence that very low SOC is not bad at all.

I should start by making this clear: 0% SOC is per definition the minimum voltage of the battery cell specified by the vendor. For Tesla Panasonic NCA this is 2.50V. This is also the most common min voltage for lithium ion batteries that have 4.20V/ cell as maximum voltage = also the definition for 100% SOC.
Going below the minimum voltage and draining the cell until it show 0V is not the same thing as 0% SOC. Doing this can cause damage, but it is random according to the research and some cells dont get damaged at all. Anyway, the BMS protect the battery from this by disconnecting it (with connector relays) before it reaches the minimum value.

The reserachers, are togheter telling us the same thing from hundreds of battery tests: 0% SOC is not unsafe, infact its the best voltage/SOC to preserve LiB’s at during storage.

Its also really safe to say that the lower the SOC the lower a defined cycle wear, like a 10% cycle wear less the lower the SOC is.
10% down to 0% is causing the least degradation and give the longest life. There is a few research reports (or one only?) where the lowest degradation was at 20-10%, followed by 10-0%, but still, the lowest degradation is in the bottom of the SOC.

All degradation examples for wear and life is per Full Cycle Equivalent so the size of a cycle is already taken into account and the comparison is related to how many miles a EV battery can be used.

My BMW i3, with Samsung SDI NMC cells, prompts me to plug in and charge if I turn off the drive at a low SOC (<15%?). If I plug it in, it will recharge up to some SOC level, even if I have set charging to happen later only in some low cost interval. The rest of the charging will happen then.

So at least for that battery, the manual and operational software makes it clear that it doses not want to be sitting at a low state of charge for a length of time.

At 50% it would be balanced between lithium ions in the graphite vs in the cathode, and I thought that structural damage occurs on each of those at the extreme of loading with lithium. There are multiple effects, from undesirable chemical reactions happening between electrolyte, additives and the materials over time, or topological structural damage.

 

[Dave EV]

Telsa needs to reclaim this battery from you and study it (can't imagine you'd let them pry it away though!). I'm still curious how many batteries there are out there like this. Obviously we usually see the worst case reports; few people take the time to report their perfect batteries - they probably don't even appreciate what they have.

Don't forget that KenC has another significant benefit - he lives in Maine, so his battery temperatures will be significantly cooler than most other batteries in the USA. You can see from AAKEE's post that the rate of capacity loss roughly follows Arrhenious' equation as expected (for every 10C increase in temperature, the rate of chemical reactions will double).

The challenge is designing a cost effective battery that can withstand annual temperatures in Phoenix, Miami or Palm Springs while sitting at 90% SOC and still last 15 years and 225k miles before getting to 70% capacity remaining...

 

[AlanSubie4Life]

Don't forget that KenC has another significant benefit - he lives in Maine, so his battery temperatures will be significantly cooler than most other batteries in the USA. You can see from AAKEE's post that the rate of capacity loss roughly follows Arrhenious' equation as expected (for every 10C increase in temperature, the rate of chemical reactions will double).

The challenge is designing a cost effective battery that can withstand annual temperatures in Phoenix, Miami or Palm Springs while sitting at 90% SOC and still last 15 years and 225k miles before getting to 70% capacity remaining...

Yes, it makes a big difference. Need to air condition my garage!

 

[Dave EV]

Yes, it makes a big difference. Need to air condition my garage!

Just get a heat pump water heater and stick it in your garage if it's not there already. Free air conditioning!

 

[Bouba]

The challenge is designing a cost effective battery that can withstand annual temperatures in Phoenix, Miami or Palm Springs while sitting at 90% SOC and still last 15 years and 225k miles before getting to 70% capacity remaining...

There is now a lot of data showing that early adopters have (or will) achieved those figures…and I expect that many of them would be in hot places

 

[My Corona]

There's no problem using all of the battery's capacity, but doing it habitually? Not so good. You usually want to avoid < 20% and > 80% and especially < 10% and > 90%.

Thats not correct.

Lithium batteries is fine down to 0%*.
The 20 and 10% thing is a myth.
A lot of the “saying” on the net about lithium battery care is not correct.

*) 0% is defined as the low voltage limit set for the cell, for Tesla Panasonic cells its 2.50V/cell. The battery shut down before reaching the cell low voltage limit.

Actually, lithium batteries is more than just fine at low SOC.

No problem.

The lower SOC the better from a battery perspective.
Calendar aging is lower the lower the SOC.

Cyclic aging is lower at lower SOC.
Cyclic aging is lower the smaller the cycles is.

This image( taken from research) show calendar aging, low SOC is best so no need to be afraid of 20 or 10%.

View attachment 796478


And for the cyclic part, 0-10% SOC during cycles is the best to keep degradation low.
Below, a picture from another test report.
Lowest degradation will be when cycling at very low SOC.

View attachment 796483

So, no reason to worry at all about low SOC.

Very grateful. I appreciate the explanation (and time!).

 

[AAKEE]

There is now a lot of data showing that early adopters have (or will) achieved those figures…and I expect that many of them would be in hot places

Teslalogger.de/degradation (seems down right now) imply that the degradation found in research also is valid for early model S.

We also know that battery manufacturers reduce the Cobalt as cobalt helps stabilizing the battery chemistry but do not help with increasing the capacity. I think the thrend is to reduce cobalt and increase the nickel part to increase the capacity.
Research probably finds ways to keep the stability/resistance to degradation during the reduction of the cobalt share. Still it is possible that some batteries wont hold up as good as the early model S 18650.

Summarized;
-We probably can not be sure that the low degradation seen in pictures and the Tesla battery survey is really valid.
-The new batteries we get today and got in early Model 3 do not have the same chemistry as the first Model S. Because of that hey will not behave exactly the same.