Welcome to Tesla Motors Club
Discuss Tesla's Model S, Model 3, Model X, Model Y, Cybertruck, Roadster and More.
Register

How I Recovered Half of my Battery's Lost Capacity

This site may earn commission on affiliate links.
I have a 2018 M3 LR AWD that seems to have higher than normal degradation that hasn't flattened out as most seem to. Below is a teslamate chart showing my last 1 year where I've from from 288 miles of range at 34,230 miles down to 266 miles of range at 50,978 miles. The last time I took a screenshot from scan my tesla, it showed my "Full pack when new" was 77.8 kWh and my current "Nominal full pack" was 65.5 kWh. If my math is right, that calculates up to ~16% deg strictly from a kWh perspective. I do not believe I have any cell imbalance as scan my tesla normally reports only 4-6mv of imbalance between the max and min cell voltages. I have been very "nice" to the battery charging less than 10% of my total charge at superchargers and leaving my max charge set at 80% for most of its life. I always have the car plugged in when home, and store it in a garage in Georgia, so nothing to extreme from a temperature perspective.

My car also falls into a manufacturing date range (August 2018) where there are more than 1 TSB that could apply for range affecting battery manufacturing issues. I've asked Tesla service off hand if there is any way to find out if my car is affected by the manufacturing defects since its manufacture date falls squarely within the TSB, but they said no. Not sure what the purpose of the TSB is if there is no way to check for it and no action would be taken anyway unless you hit the normal battery warranty threshold anyway.

If my calculations are correct (admittedly I'm doing straight line math from 0 - 50k miles) at 120k miles my deg would be 103miles/~33% from a mileage perspective or 36% from a kWh perspective. From what I can tell looking at teslaloggers degredation graphs and comparing to what I read online, my deg is a fair amount higher than the fleet average. Does anyone have any suggestions for working with Tesla to investigate further whether my car is affected by one of the TSBs from the August 2018 manufacture date or am I stuck with the continued range loss until I hopefully finally qualify for warranty? I'd really hate to continue to plod along dealing with continued reduction in range waiting for that point and get to 120k miles with only have 29% deg or something silly like that.


View attachment 663413

Seems like your result is probably due mostly to luck (that seems to be the typical reason; there's really not a clear correlation). In any case your result is not extremely bad - it's not the worst that has been reported, especially given your mileage.

It will probably continue to slow down, and I really doubt you'll hit 30% before the mileage limit of the warranty.

My recollection is that Tesla reached out to owners actually affected by the TSB at the time, but I'm not 100% sure about that.

If you actually had blown fusible links in your pack (I think the subject of the TSB as I recall), I'd expect you'd see potentially large balance problems after a discharge (since not all the bricks would have the same number of connected cells, and so would change voltage at different rates, since the current draw from all the series-connected bricks is the same).
 
Last edited:
Can you extract that data from Teslamate to a CSV/Excell/Gsheet/etc?
I've attached a 2018 M3 LR AWD with their/the TeslaFI 'fleet' data so you could draw your own charts for comparison.
Does Teslamate have an option to show their 'fleet' data overlayed on that chart?
Attached is a CSV output going as far back as I've been collecting data in teslamate. They do not have a way of comparing data to the fleet within teslamate. I have compared the data to another self hosted logger called teslalogger. You can view those charts here. I also overlayed the data you sent with my data in excel. The Teslafi data in orange below is obviously much more smooth then your car or my car. Your car is in grey and mine is in blue. Definitely interesting to note how far below the average our two cars are. It seems that at the moment, you are doing a bit better than me, but we are both well below the average.
teslafi_fleet_vs_scottf200_vs_my_car_comparison.JPG
 

Attachments

  • Projected Range - Mileage-data-as-seriestocolumns-2021-05-18 13 09 17.csv.txt
    51.2 KB · Views: 41
  • Informative
Reactions: scottf200
Seems like your result is probably due mostly to luck (that seems to be the typical reason; there's really not a clear correlation). In any case your result is not extremely bad - it's not the worst that has been reported, especially given your mileage.

It will probably continue to slow down, and I really doubt you'll hit 30% before the mileage limit of the warranty.

My recollection is that Tesla reached out to owners actually affected by the TSB at the time, but I'm not 100% sure about that.

If you actually had blown fusible links in your pack (I think the subject of the TSB as I recall), I'd expect you'd see potentially large balance problems after a discharge (since not all the bricks would have the same number of connected cells, and so would change voltage at different rates, since the current draw from all the series-connected bricks is the same).
I don't think mine is slowing down. Looking at the graph I posted of the last year or combing through the data from my car there are large drops and rises, but overall between 24k and 50k miles, its gone from 294 miles down to 267 (27 miles). I would think that would have been past the initial break in period, which coincides with the teslafi fleet data that scottf200 posted. The teslafi fleet data shows going from ~291 at 24k miles down to 276 at 50k miles (15 miles of loss over the same range of mileage). If I straight line plot my data from 24k miles to 50k miles vs the teslafi curve or the teslalogger curve I previously linked to, my battery is on a different trajectory. Since I only have data from 24k miles and not starting at 0 miles, my dataset should actually look better overall since theoretically my battery should be in its more stable phase at this point in its life.

The TBSs I was referring to are linked below. My cars manufacture date falls within the affected date range of both of these TSBs. The first one indicates "upon customer complaint of reduced range". Both seem relatively difficult to diagnose externally. These two also don't seem to indicate a "failure" scenario, but just a reduction of overall range. I would assume that if this was something that could be detected by the telemetry available to scan my tesla or to tesla over the air they would have already addressed these, as you mentioned. I haven't heard of Tesla reaching out to owners for items like this, but as you mentioned...I could be wrong also.

My concern is that depending on how I calculate deg, (mileage or kWh capacity) I usually come up with something between 32-35% at 120k miles. If my car has one of the issues mentioned in the above TSBs, I'd rather not continue to drive a reduced range vehicle hoping that it makes it to the warranty threshold before 120k, or one that will end up at 29% deg at 120k and just miss the cut off. The other side of that coin is if its close to the warranty threshold, I'd rather just stop being "nice" to the battery and maybe that forces another couple % of deg before the warranty expires.
 
my battery is on a different trajectory
I see what you are saying but it is hard to predict what will happen in future.

If you exclude the last drop, which could recover, it does not look so bad.

Hard to know. In any case, 15% capacity loss is not that unusual. Certainly worse than most, but quite commonly reported.
The TBSs I was referring to are linked below

These are both wire bond issues. See above comments. I think you would notice imbalance, since you track that. In order for it to be causing an extra 6-9% capacity loss beyond what you would “expect” you’d have to pop 3-4 wire bonds, probably.

Set up an appointment in the Tesla app if you want to make movement on this. They’ll tell you eventually with a VIN check whether your vehicle is subject to the TSB. As I recall they have a VIN list (not sure though!) so you just need to be persistent and have someone tell you whether or not you are on that list for that TSB.
 
I'm sorry if this has been covered, but: all of this degradation discussion is based on imperfect readings. As the OP suggests, we can help in the readings. Even the teslafi average curve is based on imperfect readings on user's cars. The more people that don't try to help their BMS, the worst the average will look.

When I saw a bigger "supposed degradation" dip on my car in TeslaFi in early 2021, I decided to apply the tricks discussed in the OP. Sure enough, my predicted range adjusted upwards.
Now if you have tried these suggestions, long enough, and you still have degradation then maybe there's something going on... just maybe. I personally do not trust the average any more than I trust the estimate displayed on my car.

cap.png
 
These are both wire bond issues. See above comments. I think you would notice imbalance, since you track that. In order for it to be causing an extra 6-9% capacity loss beyond what you would “expect” you’d have to pop 3-4 wire bonds, probably.

Set up an appointment in the Tesla app if you want to make movement on this. They’ll tell you eventually with a VIN check whether your vehicle is subject to the TSB. As I recall they have a VIN list (not sure though!) so you just need to be persistent and have someone tell you whether or not you are on that list for that TSB.

Yes - what you will want to do is charge to 100% and note the min/avg/max cell voltages and imbalance. Then discharge down pretty low (sub 10% at least, probably closer to 0%) and then also note the min/avg/max cell voltages and imbalance.

Ideally we would be able to read each of the 96 cell groups, but usually the pack is pretty well top-end balanced, so any significant deviations will show up at the bottom end.

What isn't known is just how much deviation at the bottom end is normal or not, and enough to detect how many popped/detached cell fuses you have. With 46 cells per "brick", each cell is about 2.2% of the total capacity. If we can better plot cell voltage to capacity at low SOC (figure out what resting cell voltage = 0-10% in 1% increments, for example), we might be able to better guess if any cells are disconnected.

But - if the rate of capacity loss is continuing, then it's not likely cell fuses unless they are continuing to pop/detach over time.

I'm sorry if this has been covered, but: all of this degradation discussion is based on imperfect readings. As the OP suggests, we can help in the readings. Even the teslafi average curve is based on imperfect readings on user's cars. The more people that don't try to help their BMS, the worst the average will look.

Yes - the only real way to confirm capacity into the pack is to fully discharge and recharge the pack. The procedure would take the highest cell voltage down to a set level (maybe 3.1-3.4V range or so), then recharge to 100% (4.18V-4.20V) and measure the energy used from the wall to charge.

A couple variables would need to be known:
1. Charge current / voltage / power - lower power means less efficient charging.
2. Some way to ensure that HVAC isn't used to heat/cool the pack. Pack temps probably should be within a set range, sentry mode off, cabin overheat off, etc.

With a prop energy meter and knowing those variables, should be able to more accurately determine real range, independent of what the BMS thinks.
 
  • Like
Reactions: AlanSubie4Life
One other datapoint to add to this conversation. I started tracking nominal kWh remaining at around 40k miles. (side note, is Tesla's warranty based off of miles or kWh remaining?) In the below chart I've calculated via straight line math that my battery should have 50.2kWh remaining at 120k miles which is 4KW less than 30% of my pack new capacity.

Since that straight line calculation takes into account the steep initial drop in capacity I also ran a calculation taking into account the amount of pack capacity lost between 40k-50k miles, since that should remove the initial drop. The rate of decline over the last 10k miles is actually higher (0.00028 kwh lost / mile driven) than over the initial 40k miles (0.00023 kwh lost / mile driven). This to me seems to negate the theory that my battery should level off over time or either point to a plateau that occurred sometime after the initial drop but before I started tracking at 40k that skewed the data.

Seems like any way I slice the data, whether via miles or kWh, or either including the initial drop or taking it out, my battery is on a trajectory to have less than 70% retention by 120k miles. I find it difficult to construct a model where it stays above the 70% threshold unless I just assume that all previous and current trends will not continue as they are today.


***added*** if my pack capacity continues to decrease at the same rate it did between 40k-50k miles, it will loose another 19.7kWh by the time it reaches 120k. Given the 65.5 kWh capacity today, if it does lose another 19kWh, it will be at 45kWh and would have long since crossed the warranty threshold. Just another metric trying to reduce any variables like initial drop, or wh/miles conversions.
milesbattery kwhkwh lost / mile drivenwarranty kWh
0​
77.8​
54.46​
40000​
68.6​
0.00023​
46494​
68.4​
0.00020​
50978​
65.5​
0.00024​
estimate 120000​
50.2​
 
Last edited:
Yes - the only real way to confirm capacity into the pack is to fully discharge and recharge the pack.
Yep, and we have done that elsewhere here before for an SR+ and were within 2% of what was expected, accounting for charging inefficiency etc. Fortunately Tesla does post data in the EPA tests on exactly what their charging efficiency is (presumably at 240V/32A) - 88%.

The estimated capacity losses have consistently come very close to real capacity losses in all tests that have been done. Of course, there are occasional special cases where the estimate is off, as outlined at the beginning of this thread. But apparently not the norm - the estimate has to be really really good in general!
 
  • Like
Reactions: GtiMart
whether via miles or kWh, or either including the initial drop or taking it out, my battery is on a trajectory to have less than 70% retention by 120k miles.
Maybe. At the fleet level, capacity loss clearly slows down, but it is difficult to predict what will happen for a particular vehicle. My guess is your capacity loss will slow down - but it is just that, a guess. At some point I would expect the general typical characteristics of the pack to assert themselves.

Estimate errors could play a role too. If you ignore your recent drop from 280 to below 270, your trends would look quite different. It is possible that you will see some recovery there, as you have seen in the past. I assume you’ve tried the methods outlined at the beginning of this thread.
 
I have a 2018 M3 LR AWD that seems to have higher than normal degradation that hasn't flattened out as most seem to. Below is a teslamate chart showing my last 1 year where I've from from 288 miles of range at 34,230 miles down to 266 miles of range at 50,978 miles. The last time I took a screenshot from scan my tesla, it showed my "Full pack when new" was 77.8 kWh and my current "Nominal full pack" was 65.5 kWh. If my math is right, that calculates up to ~16% deg strictly from a kWh perspective. I do not believe I have any cell imbalance as scan my tesla normally reports only 4-6mv of imbalance between the max and min cell voltages. I have been very "nice" to the battery charging less than 10% of my total charge at superchargers and leaving my max charge set at 80% for most of its life. I always have the car plugged in when home, and store it in a garage in Georgia, so nothing to extreme from a temperature perspective.

My car also falls into a manufacturing date range (August 2018) where there are more than 1 TSB that could apply for range affecting battery manufacturing issues. I've asked Tesla service off hand if there is any way to find out if my car is affected by the manufacturing defects since its manufacture date falls squarely within the TSB, but they said no. Not sure what the purpose of the TSB is if there is no way to check for it and no action would be taken anyway unless you hit the normal battery warranty threshold anyway.

If my calculations are correct (admittedly I'm doing straight line math from 0 - 50k miles) at 120k miles my deg would be 103miles/~33% from a mileage perspective or 36% from a kWh perspective. From what I can tell looking at teslaloggers degredation graphs and comparing to what I read online, my deg is a fair amount higher than the fleet average. Does anyone have any suggestions for working with Tesla to investigate further whether my car is affected by one of the TSBs from the August 2018 manufacture date or am I stuck with the continued range loss until I hopefully finally qualify for warranty? I'd really hate to continue to plod along dealing with continued reduction in range waiting for that point and get to 120k miles with only have 29% deg or something silly like that.


View attachment 663413
Was that the TSB about missing connections between cells? If SMT says you had 77.8kWh, wouldn't that mean, that your cells were all connected originally? Or are there other TSBs about batteries around that time?
 
If SMT says you had 77.8kWh, wouldn't that mean, that your cells were all connected originally
FPWN in SMT is a hard-coded value and doesn’t have any bearing on the pack’s actual capacity when new. Think of it as the design capacity denominated in BMS (not EPA) kWh (these kWh are apparently not equivalent, much to my chagrin - 78kWh BMS seems to be equivalent to about 79-79.5kWh EPA).
 
  • Informative
Reactions: scottf200 and KenC
I'm sorry if this has been covered, but: all of this degradation discussion is based on imperfect readings. As the OP suggests, we can help in the readings. Even the teslafi average curve is based on imperfect readings on user's cars. The more people that don't try to help their BMS, the worst the average will look.

When I saw a bigger "supposed degradation" dip on my car in TeslaFi in early 2021, I decided to apply the tricks discussed in the OP. Sure enough, my predicted range adjusted upwards.
Now if you have tried these suggestions, long enough, and you still have degradation then maybe there's something going on... just maybe. I personally do not trust the average any more than I trust the estimate displayed on my car.

View attachment 663552

imperfect readings dont matter. this is a) averaged fleet data so overreading and underreading gets averaged and b) inthe grandscheme of things noone here cares if they have 3% or 4% degradation. we care that we have 10% degradation though.
 
  • Like
Reactions: AlanSubie4Life
You may be right. I'm basically saying that we have no idea if the BMS discrepancy stops at 3-4% or if it goes to 10% or more. I'm saying that using the average and comparing, or using your own value, is not a good way to measure degradation. The only way is to do the proper test, emptying and filling the battery, as I believe AlanSubie4Life said.
 
You may be right. I'm basically saying that we have no idea if the BMS discrepancy stops at 3-4% or if it goes to 10% or more. I'm saying that using the average and comparing, or using your own value, is not a good way to measure degradation. The only way is to do the proper test, emptying and filling the battery, as I believe AlanSubie4Life said.

it doesnt matter if the disreptency is 20%. if you cannot access those 20% then you may as well have 20% degradation.

Emptying and filling the battery is a myth. Anyone who does it (enough tesla warriors on youtube) has NEVER shown that the BMS was off in any way. I am very welcome to see a video. dont get me wrong, there are a few where people show that they have "3% less degradation doing a discharge test" - but those videos have such comically poor maths if you calculate it yourself you can see that the BMS is spot on.
 
I want to believe the BMS is precise, like everyone does. However, I readjusted my BMS by around 4% without trying too hard. That leads me to believe the BMS is not that precise. It might be the best in the industry but it's still a guess. My understanding, although I have not verified it scientifically, is that I never lost that 4% in actual capacity. I think assuming capacity loss based on the BMS is not the right way to go, that is all. People get overly stressed about it on thousands of posts here, and other forums, for possibly no good reason.
 
  • Like
Reactions: Dave EV
Emptying and filling the battery is a myth. Anyone who does it (enough tesla warriors on youtube) has NEVER shown that the BMS was off in any way.
He is referring specifically to measuring energy put into the pack, I think. Not just cycling the pack to see if the BMS “corrects.” (Generally “correcting” the BMS seems to be a myth, though there are exceptions.)

That measurement of added energy has been done, and it has been shown that the BMS is very accurate. If there are good YouTube examples of this with all the calculations, that would be great to see linked (I haven’t looked).

As you say, no one cares much about a couple %, 3% or even 4%. But 10-15% starts to matter!

And that capacity loss has been measured by measuring AC energy added to the pack in controlled conditions (no Sentry, AC, etc.), accounting for charging losses (which are well known, consistent, and published by Tesla), and it matched the BMS estimate very well (within a couple %).

That is to say - if the BMS says you are missing 10% of your energy, you indeed are missing 8-12% of your energy, in most circumstances. And you can measure this and confirm it by charging your car from near empty, measuring extremely carefully, accounting for all the charging losses, to determine how much energy was added to the pack. It will match the BMS estimate.

If you’re really worried about charging loss inaccuracy, you can do a side-by-side test using the same charging equipment of an older Model 3 with capacity loss, and compare it to a Model 3 without capacity loss, with the same full pack when new value. Starting from the same SOC %, in carefully controlled conditions (no accessories!), the one without capacity loss will take considerably longer to charge! ~10% (11%) longer, in the case of 10% capacity loss on the older vehicle. You don’t have to rely on the BMS at all!

Or you can just trust the BMS (which is generally pretty accurate, but occasionally has small (less than 4% usually) deviations, for the reasons mentioned in this thread)!

Summary: Range loss means real capacity loss, with a similar % for each. But you don’t have to trust some crazy dude on the Internet - you can directly measure it yourself, if you wish. All the information needed is available.
 
Last edited:
  • Like
Reactions: DaddysCar
Update: I went back to my standard procedure of shallow(ish) cycling with an anchor around 50% (optimizing to maximize most sitting time spent closest to 50%) and I'm gaining RM and taking names again. At one point I dropped to low 290s, now I'm just shy of 300 (299.75). It really seems like winter threw a monkey wrench, together with the changes in firmware in re: to the temperature dependency of the readings. I also think that once I started to SuperCharge liberally I probably took a real hit of 5-8 RM but it's since stabilized -- I was at 309+ RM before SCing. I anticipate the range will continue to improve with higher summer temps, if anyone's interested I can update when I observe anything meaningful.
 
seems like winter threw a monkey wrench,

Yes, winter has a real impact on available energy. Hence, the BMS tells you have less energy - because you do! Of course it's an estimate, but it does the best it can (given the battery temperature can change a lot from drive to drive, and even changes a lot WHILE you drive, it is constantly having to correct for that and re-estimate). Cold batteries simply cannot provide the same amount of energy, even if they are 100% healthy, and not showing the "blue snowflake." It's physics; Gibbs free energy and all that jazz is a function of temperature, I think is the reason...

I'm gaining RM and taking names again
😂
 
Last edited:
Well that was quick. I put in a case with service indicating that my car was loosing range more quickly than others with a similar age/mileage and the desire to validate if the two TSBs for range loss for M3s around my manufacture date applied to my car. I got an nearly immediate response:

"Your vehicle doesn't qualify. The 8 year battery warranty only applies if you are loosing more than 30%. Your battery does have an 8 year warranty. At this time, there are no issues with your battery. We have closed your ticket out with us."

In other news, my car has dropped another 2 miles of displayed range down to 264 at 51,347 now (I know..I know..it may go back up). I have a couple of long trips coming up where I may try charging the battery all the way to 100% (which I have avoided ever doing up to this point) just to see if that makes any difference. To the original point of this article, I feel like I already get OCV readings at different battery charge levels because I have scheduled charging setup to start at 2AM. My car regularly sits at various battery levels (depending on what level I arrive at home with) until charging starts at 2AM already (frequently more than 6 hours of idle time). The car should have plenty of time to take readings at various levels of charge just based on my normal charging habits. The one thing I have never tried is to fully charge it to 100%. Hopefully I'll give that a shot soon before a long trip and be able to see if that has any impact. Worst case it degrades my batter a little more and gets me closer to 30% anyway.

That raises another question. If someone truly believes they are headed for 30% degradation before the warranty expires, why wouldn't someone (super)charge to 100% all the time. I've never seen anything in the warranty that voids the warranty based on charging habbits. Just curious if anyone knows how Tesla handles something like that if someone "abuses" their battery leading to premature degredation? I suppose Tesla is banking on even those habits not causing serious degradation.
 
why wouldn't someone (super)charge to 100% all the time
I’ve heard that early owners who supercharged a lot eventually had their rates throttled a bit, so that is one reason. The other reason is that it’s expensive if you don’t have free supercharging. It’s also really inconvenient to Supercharge if you’re not road-tripping.
The car should have plenty of time to take readings at various levels of charge just based on my normal charging habits.
Just make sure it is reliably sleeping, with the contactors open. I doubt it will matter, but I guess cars that have real problems sleeping can have estimates that become fairly far off (I hear - no experience with having estimates that are wrong). Unusual and probably not the issue in your case.

At least it sounds like they confirmed that your car is not covered by the TSB (I assume that is what they meant by “your car does not qualify”).
 
Last edited: