determining the battery SOH trends

[bumcurriculum]

I am wondering if there is a common best practice to determine the SOH trend of my car. I think I understand the basic math (as described at Tesla battery degradation and health), also I have Teslafi which provides the famous Battery Report. However in my understanding the result of the battery report is impacted by "temporary" factors like BMS Calibration and cell imbalance, not just the actual battery degradation.
What might be the best practice to minimize the impact of the temporary factors on the SOH calculation? I don't necessarily want to see exact results, just to see a long term trend of real battery degradation.

(in my current example for my 2018 MS100 Teslafi showed that battery range always moving around 470km, then 1-2 weeks ago it dropped to 465km, and now it consistently moves around 465km. I am wondering if this is real degradation, or cell imbalance/BMS calibration issue. Reading through quite a lot of discussion on TMC in related topics, my understanding is that there is no universally acknowledged step-by-step method to rebalance/recalibrate)

E.g. is it possible to check cell imbalance without SMT? Maybe in the service menu?

 

[AAKEE]

I am wondering if there is a common best practice to determine the SOH trend of my car. I think I understand the basic math (as described at Tesla battery degradation and health), also I have Teslafi which provides the famous Battery Report. However in my understanding the result of the battery report is impacted by "temporary" factors like BMS Calibration and cell imbalance, not just the actual battery degradation.

The easy way is to use the energy graph calculation (probably in your link).

Energy app calc, Teslafi, Tessie and looking straight in to the BMS data with SMT all suffers from dependence on the BMS estimate.
But the BMS is most often not far away from the truth.

Battery degradation is predictable from the environment and average SOC (and slightky from the miles driven).
This means that if your displayed range, normally slowly decreasing suddenly looses 30km it is not a real sudden degradation. Either the BMS was off before and corrected, or just got of by 30km.

If you know the normal range and only see a slow reduction the BMS probably is correct. So in that case, the reading from SMT, Tessie, Teslafi etc is not far away.

If you suspevt the BMS being off, only reliable test is a 100-0% drive like this:
I had my BMS go rogue on my M3P once.
Showed 75.7-76 kWh for a while, coming from 79-79.4 kWh.
So I made a 100-0% drive, and the delivered energy showed that the real capacity was around 79kWh or maybe slightly above.

There also is a possibility to drive less than 100-0%, but you need to drive a substantial part of the battery. From that, the capacity can be calculated.
I have described this in a few posts. I’ll try to find these and get you the links.
(My new MSP showed only 95.7 kWh new, and I was suspicious that the BMS was off. Doing the test like above I could calculate that the capacity was 98-98.3 kWh. One month later the BMS reported above 98 kWh, and the top number was 98.4kWh)

What might be the best practice to minimize the impact of the temporary factors on the SOH calculation? I don't necessarily want to see exact results, just to see a long term trend of real battery degradation.

(To reduce errors from BMS, do like the test above.)
To see trends, I would use the BMS value, abd accept a few smaller errors.

I could calculate and show you how your battery should behave from new and several years ahead.

I could probably calculate the capacity and degradation for your car as it is today. I probably will not end up far away feom a energy graph calculation.

What car do you have? Model, year /age?
To what SOC do you charge daily?
What is the normal SOC in the end of the day?
Do ypu have the car outside? Where do you live (area for checking the climate)

(in my current example for my 2018 MS100 Teslafi showed that battery range always moving around 470km, then 1-2 weeks ago it dropped to 465km, and now it consistently moves around 465km. I am wondering if this is real degradation, or cell imbalance/BMS calibration issue.

That sort or clears some of the questions above.

You should ecpect the car to slowly loose range, the older the slower. But people often state that loss comes in batches.

Reading through quite a lot of discussion on TMC in related topics, my understanding is that there is no universally acknowledged step-by-step method to rebalance/recalibrate)

Slight Imbalance probably do not affect displayed range.

The basics around balancing the cells is well known and the BMS need to see the battery from several SOC levels is also well known - no problem with knowing how to perform these.

 

[bumcurriculum]

The basics around balancing the cells is well known and the BMS need to see the battery from several SOC levels is also well known - no problem with knowing how to perform these.

Thank you very much for the detailed response!

Regarding the above, I have seen multiple, contradicting descriptions, so I am not sure which procedures are currently considered as best practice.
-balancing: above 80, 90, 93 or only at 100%? During sleep, idle or charging?
-for the BMS to collect information at different charge levels, should the car idle or sleep for a couple of hours?

 

[AAKEE]

Here is one way to calculate the real capacity, without the need to rely in the BMS estimate:

Capacity calc

 

[AAKEE]

Thank you very much for the detailed response!

Regarding the above, I have seen multiple, contradicting descriptions, so I am not sure which procedures are currently considered as best practice.
-balancing: above 80, 90, 93 or only at 100%?

Make sure to have at least 90%.
Sleep is needed, and the rate for balancing is not fast as the resistor does not allow for high currents. Several hours needed to fix just a little imbalance.

-for the BMS to collect information at different charge levels, should the car idle or sleep for a couple of hours?

When driving, the battery voltage droops from the load and when disconnecting the battery it takes some time for the battery voltage to recover.

When charging the battery will show a higher voltage just after the charge, and it will take a while for the battery voltage to droop and show the real ocv voltage.

A couple of hours or three should be good enough.

There are examples of the time to reach about the OCV resting voltage in other research but I did just grab the forst pic I found:

Actually, the harder you drive the more time is needed for the voltage to recover.
Tge faater you charge the more time is needed to get the correct reading.

Wait longer means more exakt voltage but I would say that 2-3 hours is good enough, and yes: Sleep.

 

[bumcurriculum]

Can you maybe suggest the best thread which gives the most complete view about battery "science" in terms of Tesla BMS, interpreting SMT values, etc,?

 

[bumcurriculum]

If you suspevt the BMS being off, only reliable test is a 100-0% drive like this:
I had my BMS go rogue on my M3P once.
Showed 75.7-76 kWh for a while, coming from 79-79.4 kWh.
So I made a 100-0% drive, and the delivered energy showed that the real capacity was around 79kWh or maybe slightly above.

How can you perform a 100-0% drive in practice? Just drive until the car stops and then call a trailer?

 

[AAKEE]

Can you maybe suggest the best thread which gives the most complete view about battery "science" in terms of Tesla BMS, interpreting SMT values, etc,?

Just search for something like calendar aging or cyclic aging or degradation etc. on my nickname. (Probably doesn't sound very humble, but I think its one of the best way to...)

How can you perform a 100-0% drive in practice? Just drive until the car stops and then call a trailer?

When the car shows 0% you still have the buffer to drive on. For my M3P (and all newer Tesla LR/P) it is 4.5% of the total capacity so around -4.7% if the car would have displayed - numbers. (SMT shows true SOC and SOC in negative numbers, togheter with the cell voltage so I have been down to -1.8/-2% without any issues.

But read the link about calculating the capacity from a partial drive, its a less long drive and no worry for 0%. (but it should at least be around 50% of the SOC.

 

[AAKEE]

How can you perform a 100-0% drive in practice? Just drive until the car stops and then call a trailer?

Today I charged full and drove 254km
Started with 100% and arrived with 54.4% which after ~ 45 minutes after the drive was 54.5% (values with SMT to get 1/10%).
The used energy in percent was 45.5%.

Did not have time to let it sleep longer but the most adjustment happens within 30 minutes.

The used energy was 42418 Wh (”trips”, km driven since last charge x average consumption.

42.418 kWh / 0.455 = 97.6 kWh.
In this case, as the car was allowed to sleep after the drive the SOC is measured quite correct, so we can be sure that the used energy SOC wise is close to 45.5%.

I did not take any picture of the finished charge, but the nominal remaining was 97.6 kWh when the charging just finished.
Nomimal full pack is 97.1 kWh.

(My BMS is rather okey calibrated as the SOC doesn’t really change when parking after long drives, and also matches my capacity calc.)

You can use this calc as well.
I did provide an optional way to use the displayed range instead of SMT SOC readings in a thread here on TMC. Just needs to be found

 

[bumcurriculum]

Here is one way to calculate the real capacity, without the need to rely in the BMS estimate:

Capacity calc

just read that, and understood the main point, its quite a brilliant method to calculate the real capacity even without SMT.

Is there maybe some similarly clever solution to estimate whether cell balance is off (without SMT) and its time to perform the balancing voodoo?

 

[AAKEE]

just read that, and understood the main point, its quite a brilliant method to calculate the real capacity even without SMT.

Is there maybe some similarly clever solution to estimate whether cell balance is off (without SMT) and its time to perform the balancing voodoo?

No, I guess you need SMT (or Equivalent) to se cell voltages.
In general, you can rely on the car/BMS to take care of this by itself.
I have never looked into service mode, if there are any data.

 

[bumcurriculum]

In general, you can rely on the car/BMS to take care of this by itself.

I charge only to 55% on a daily basis, and although there are multiple different rumors about when a MS FL starts balance, with this charging regime it almost surely won't. I am happy to do the balancing when it is not balanced.
Probably I will need to get SMT sooner or later anyway

 

[AAKEE]

I charge only to 55% on a daily basis, and although there are multiple different rumors about when a MS FL starts balance, with this charging regime it almost surely won't. I am happy to do the balancing when it is not balanced.
Probably I will need to get SMT sooner or later anyway

I started 55% 3.5 year back. Of course, I selldom see more than one month with 55% only but during all times, I never seen the imbalance increase.

My M3P always had 4mV at 55%.
My MSP always is at ~10 mV at 55%

Charged full yesterday, went 250km away. 12mV at 100%
Charged from 55 to 75% with the UMC, 2.3kW slo’charge.

I saw 4mV this morning, so the pack did most probably balance itself during the charge, or during the four hours from reaching 75% until we drove.

 

[bumcurriculum]

For now I decided to not focus so much on determining the real SOH better than the BMS does, but to help the BMS calibrate itself, and leave it to determine the SOH (and maybe help it to balance). I started from 90% SoC and used the car through multiple days, leaving the car to "rest" for hours roughly after every 10% discharged, until it went down to 7% SoC. Then I set the charge limit to 100%, and connected to the home charger.

It charged until 94%, where it stopped, this is the charging power diagram in the last hour as measured:

During the following 2 hours the car was idling(did not fell asleep), not taking any power from the charger, but the reported SOC going up gradually from 94 to 99%, and also the "teslafi battery range" has increased from 537.42km to 574.62km.

Now Teslafi Battery Report shows a further 6km decrease in the "Estimated Battery Range at 100%" compared to the last charge 6 days ago. However that report shows that it was a charge up to 94%, so I am not sure if it was considering the magical range increase during the two hours which happened after the finish of the charge.

Tomorrow I will drive approx. 300km, after that I plan to charge to 100% again and see what happens...

 

[bumcurriculum]

Since my last post a couple of things happened, some of them are a bit concerning.
The next day after a 309km drive I charged to 100%, then the teslafi "Estimated Battery Range at 100%" increased a bit to 464km. (earlier it was fluctuatin aroung 470, in the last couple of week it is more around 460)

Then during the weekend there was bit longer road trip with multipe DC charging. The car arrived at home Sunday evening with 8% SoC on the dashboard.


I let the battery to stabilize, and in about 1 hour the SoC dropped to 0%. I was not monitoring the car, so I didn't realized this, only after 2 hours later, when it still was at 0%. (during this 3 hours the car didn't go to sleep, although sentry was not enabled)
Then I charged it until morning with 5kW. Charging was stopped at 46%, and the car was immediately driven 32km, during which the SoC went down to 40%.
In the first 30 minutes of parking the SoC increased from 40% to 49%, then after further 30 mins it went even to 50%. After that it was slowly discharging, as sentry was accidentaly left on, so the car didn't sleep during this whole 9 hours parking period, which ended at 47% SoC.

These SoC fluctuations are not optimal I would say

• might this mean that somehow the battery is not healthy, and starting to break?
• this inaccuracy of the dashboard SoC potentially causes planning issues. I have already recognized, that at low SoC (below approx. 30%-20%) the built in navigation is too optimistic, e.g. if I plan to arrive with 10%, then the forcasted arrival SoC decreases gradually during the drive and I need to significantly decrease my speed to reach the destination. Might this be caused by the same BMS issue which is causing these fluctuations during Idling/Sleep? Is there anything I can do to help the BMS to learn the characteristic of the battery better?

I have the car for half a year approx., and drove 20 000km. Daily charging scheme is charging 55%, then discharging to 35%. (for road trips I charge to 90%. Only one time did I charge to 100% before last week. 25 charges above 80% altogether. Maybe not enough for the BMS to do its magic?)

I use off-peak charging, and usually during the night the car charges for an hour then rest for another hour, then charge and so on. The car is doing this itself, without external intervention, when off-peak charging is enabled. With such pattern I would think that the BMS can take the voltage samples at least from the 35-55% range at every %5 approx.

As I mentioned, to help the BMS, last week I let the battery discharge from 90% to 7%, with roughly 10% drives, letting the car rest for hours between the drives. I also performed two 100% charges(the first ended at 94% as described above, car didn't charge further, then after some rest it jumped to 100% without taking any power from the charger) I was hoping that this will help the BMS to be more accurate, but apparently it started to make things much worse somehow.

 

[AAKEE]

Since my last post a couple of things happened, some of them are a bit concerning.
The next day after a 309km drive I charged to 100%, then the teslafi "Estimated Battery Range at 100%" increased a bit to 464km. (earlier it was fluctuatin aroung 470, in the last couple of week it is more around 460)

Then during the weekend there was bit longer road trip with multipe DC charging. The car arrived at home Sunday evening with 8% SoC on the dashboard.


I let the battery to stabilize, and in about 1 hour the SoC dropped to 0%. I was not monitoring the car, so I didn't realized this, only after 2 hours later, when it still was at 0%.

If the SOC drops after a longer drive like that it is probably because of a overestimation of the capacity. Several cycles without sleep might increase the faulty calculation from the BMS about the SOC.

(during this 3 hours the car didn't go to sleep, although sentry was not enabled)
Then I charged it until morning with 5kW. Charging was stopped at 46%, and the car was immediately driven 32km, during which the SoC went down to 40%.
In the first 30 minutes of parking the SoC increased from 40% to 49%, then after further 30 mins it went even to 50%.

This also implies that the BMS overestimates the capacity.

After that it was slowly discharging, as sentry was accidentaly left on, so the car didn't sleep during this whole 9 hours parking period, which ended at 47% SoC.

These SoC fluctuations are not optimal I would say

• might this mean that somehow the battery is not healthy, and starting to break?

I think it is time to perform a BMS Calibration. In general I do not think the common usage to ”up” the displayed range is wort it. It only (might) increase the displayed range but not the real drives le range, so its more of blowing up the ego than helping

But in your case, it seems like the BMS is fairly off.
Search for BMS calibration here on TMC or maybe ”How I got back 50% of my range”

What is done is showing the real capacity to the BMS, helping it adjust the estimated capacity to better match the real world.

You could look at the SOC/displayed rabge after a little longer drive.
If the car was let to sleep before the drive the initial SOC should be about correct and when arriving, take notes of both the SOC and the displayed range. Then let the car sleep (no sentry etc) and then check the SOC and range about one hour after the car was parked.

If the SOC /range dropped there is an overestimate. Range is used as it has a higher resolution than SOC.

You can use Teslafi to see this. The variations in SOC is graphically shown in the upper part of Teslafi.

 

[bumcurriculum]

Search for BMS calibration here on TMC or maybe ”How I got back 50% of my range”

Thank you, I appreciate your support!
Are you referring to this method:

Tesla battery management system (BMS) calibration

Had this site bookmarked on my computer for at least 1 1/2 years from memory: https://tesla-info.com/guide/tesla-bms-calibration.php This thread may make it easier for others to search for information provided in that guide and it seems to be updated regularly as seen here: Tesla battery...

teslamotorsclub.com

I had the understanding that when I did the 90%->7% discharge during the course of 5 days, letting it sleep for hours between the 7%-10% discharges, I was pretty much performing this procedure. But apparently things got worse. (not necessarily from these steps, but happened so.)
Or should I maybe repeat this multiple times?

 

[AAKEE]

Thank you, I appreciate your support!
Are you referring to this method:

Tesla battery management system (BMS) calibration

Had this site bookmarked on my computer for at least 1 1/2 years from memory: https://tesla-info.com/guide/tesla-bms-calibration.php This thread may make it easier for others to search for information provided in that guide and it seems to be updated regularly as seen here: Tesla battery...

teslamotorsclub.com

I had the understanding that when I did the 90%->7% discharge during the course of 5 days, letting it sleep for hours between the 7%-10% discharges, I was pretty much performing this procedure. But apparently things got worse. (not necessarily from these steps, but happened so.)
Or should I maybe repeat this multiple times?

Yes, I saw your earlier steps. I think the BMS can be a little grumpy and not adjust that quick.
(My Plaid arrived with a BMS thinking 95.7kWh. I made a test drive and calculated the capacity to 98.0-98.2kWh. One month later the BMS showed 98.3kWh. It canntake time. I did not specially do any calib but had long drives anyway so the car still got to see abd sleep with 100% and below 10%.
One charge to high and one sleep is not enough.)

Use Teslafi to show if the SOC changes after drives. (taking notes is better but Teslafi also does it). When the SOC doesnt change after a long drive, the BMS is on track.

Bms calib



See the 250km drive below: charge, sleep on hour and then drive. When arriving, before the new sleep the SOC seems flat, right? Betwen the second ”D” stopped to open a electric gate) until the ”S” (sleep).
In this case the BMS updated the SOC up 1% during the long sleep. It can not be seen in this picture but it did. So a slight underestimate was present.

I do not have any overestimates, but this is
A slightly higher underestimate. See the climb just before the sleep. If the BMS estimate is off enough the BMS will notice this quite quick after parking. So a initial adjust will be done.

An overestimate will be about the same, but downwards. Parking with fpr example 10%, but seeing 11% shortly after and 12% after a few hours.
Or if a smaller one, only see the adjust after some hours of sleep.

 

[bumcurriculum]

I kept following the "BMS calibration" regime for the last couple of weeks now, and the car "lost" 4km again from the "Estimated Battery Range at 100%" reported after last nights 100% charge. In summary, it went down from 470km to 457km in just one month.

I am getting a bit concerned now, though the car still has battery warranty until 2026.

 

[AAKEE]

I kept following the "BMS calibration" regime for the last couple of weeks now, and the car "lost" 4km again from the "Estimated Battery Range at 100%" reported after last nights 100% charge. In summary, it went down from 470km to 457km in just one month.

View attachment 1058385

I am getting a bit concerned now, though the car still has battery warranty until 2026.

A BMS calibration, if performed correct helps the BMS to better calculate/estimate the capacity. Which means that it can go both up, and down depending on if it was an over- or under-estimate before. (Or stay put if the estimate was about right.)

If you provide info about the car, model, year, mileage and how it has been charged since new, I can give you an estimate on what to expect in form of degradation - capacity- range today.