Charting Battery Capacity/ State of Health Over Multiple Years

[3pete]

I bought my 2014 i3 two years ago and I have been tracking various metrics that seemed relevant to battery capacity/ state of health since then.
Because this seems to be a hot topic on here and because I like charts and graphs, I wanted to share some of what I have found.

First, some disclaimers:
1) This data was manually tracked by me so it may have typos/errors in it but I think I was pretty diligent in entering the data correctly.
2)Because this was manually pulled from the service menu on the car, I only logged data points when I felt like it.  There is a high likelihood that this introduces some skew because it's nowhere close to a random sample. 
3) This is based on ONE vehicle/battery.  It's possible different 2014 i3 batteries behave differently.
4) Some metrics have more data points than others.  For example, I only log trip kWh on trips where I left from 100% charge and remembered to reset the trip meter.
5) Since I bought the car used I only know some of its history.  It seems to have started its life in California and the owner I bought it from had it for about a year in the midwest.

With that out of the way, the big questions are "what is the state of my battery?" and "how fast is it degrading?"  With the i3, there are a few ways to estimate this and none of them are exact. In keeping with the spirit of "all models are wrong, but some are useful", here are the metrics that I have found useful to track in order to answer this question:
1) Batt. Kapa. Max.  This one can be accessed through the service menu and taken from the car as-is.
2) Batt. Ladung kWh/ SOC.  The service menu shows Batt Ladung (charge) in percent and kWh.  By taking kWh divided by % we can get what the car thinks is the 100% kWh level.
3) Trip kWh + Batt Ladung kWh.  The Batt Ladung kWh still comes from the service menu, and to get Trip kWh you reset the trip computer and at the end of your trip take your Trip Miles and divide by Trip Efficiency in mi/kWh.
4) Trip kWh / Remaining SOC.  This is the same trip kWh calculation from #3 divided by the same % from #2.

Charting these over the course of my almost-20k mile ownership of my car yields these charts.  I've added a trend line and a dashed line to represent 13.1kWh and color coded the dots by outdoor temperature because that seems to be relevant:

Some observations I've noticed from this:
1) Batt Kapa Max seems to have the fewest outliers and most consistently represent battery capacity, though it definitely has variation as well.
2) Degradation seems to be non-linear and possibly accelerates in cold weather.  Additionally, these trendlines all suggest a value at 0 miles well above the 18.8 kWh intercept they "should" have.
3) These metrics often disagree with each other on capacity for any given reading but plotted together they seem to agree on a downward trend and a very approximate capacity.
4) My car seems to be very close to the 70% threshold specified in the warranty.  The next few months could be a good indication of whether BMW included any sort of "battery buffer" that is automatically released to keep the capacity at/ above 70%.

Anyway, hopefully this is useful or interesting to some folks.  If you have any questions or comments let me know.  I have some other charts I find somewhat interesting that I'll try to share but if there's any correlation someone has been wondering about, feel free to ask and I can see if I have the data to look at it.

 

[MagicCarpet]

Thanks for posting, that is very good data to consider.

might I ask what pattern of driving is normal and if you use DC 50kW rapid charging or slower methods, whether you do a lot of top ups or run from high SOC to the bottom quarter then recharge?

I have a 2017 and contrary to popular advice usually charge DC 50kW rapid when it gets to around 20% and the charger drops out at 98% so I usually diconnect at around 85% unless doing a long trip next day, being mindful that the 33kWh battery is not stressed to the same extent as rpaid charging the earlier version would be.
I have only 3 snapshots of the service menu parameters - FWIW they were:
Mar.2019 6,000km kappa max 29.2 78% lading 23.3
Aug.2019 kappa max 28.0 93% lading 26.0 temp at charge start 14deg.C end 20degC
Feb.2020 11,000km kappa max 27.6 75% lading 20.5
during that period I was doing 1 charge per week and the REX only ran for mtce cycles. I see no loss of range indication on the instrument diplay when starting out full charge - lately its been showing 200km which is better than the normal 183km but rthis is probably due to longer trips and mild climate.

 

[Steve999]

Excellent work!

I've also been collecting data that I think compliments your information. On my 2015 REX, the plot of Batt Kappa shows a regression line that says initital battery capacity should be around 22KWH (??) with a slope of -0.0002 KWH/mile (btw green data points are the temp. Also, I tend to average 5.2 mi/kwh in eco pro+ used almost all the time). I'm getting a strong correlation with mileage, less so for battery calendar ageing (thanks to covid).

For fun, I surveyed all of the used cars I could find online over the past ~6 months that had comfort mode displayed (mostly within 100mi radius of bay area, so temperatures are moderate) and I got a slope of -0.0003 mi loss in range per mile driven ( but R^2 values are poor on regression). This agrees fairly well with your degradation rate of about 0.0002 kwh kappa loss/mi and the loss rates given by the national labs study (AVT-INL) from David Bricknell's book-page 57. Looking at the spread of the values (that looks somewhat like a cone), I think some of it may be related on whether the cars were driven hard/charged hard or driven gently/charged slowly. To represent that, and assuming 3.9 mi/kwh is nominal, I added two lines representing ~2.8kwh/mi usage and ~5.2kwh driver usage rates.

I also collected data on 94aH batteries, but there aren't many high mileage cars and data variation is very high. Even so, the (bad) regression line looks like about 0.0002mi loss per mile driven, or about what eco pro+ driving in my 60ah vehicle is getting (makes sense since battery is 1/3 larger).

Although this data is not taking temperature into account (and is likely skewed downward) or any other driving history for the vehicles since I only got the mileage and estimated ranges from the screens, but it also suggests that people will hit the 30% range loss before 100K miles unless there is a hidden capacity that can be tapped into. At least it shows how ~30 different batteries are performing...

 

[alohart]

I bought my 2014 i3 two years ago and I have been tracking various metrics that seemed relevant to battery capacity/ state of health since then.

Nice job! Thanks for posting this information.

Our 2014 BEV has suffered ~20% actual range loss after only 11k miles, fewer than 5 DC fast charging sessions, only 2 of which were from ~20% to ~80% with the rest being ~10-minute sessions, no extreme battery pack temperatures (Honolulu), the charge level having rarely been allowed to remain at 100% for more than a couple of hours, and having been typically charged to an indicated 90% at a slow 16 A @ 208 V. I almost always drive under the control of ACC in Eco Pro, so accelerations have been gentle. The BMW Connected app indicates that my lifetime driving efficiency is 5.2 mi/kWh.

As far as Batt. Kapa. max is concerned, on 17 November 2015, it was 18.9 kWh with 2.4k miles having been driven. By 19 October 2016, it had dropped to 16.5 kWh after having been driven only 3.8k miles total, so a decrease of 2.4 kWh in 11 months and 1.4k miles. On 4 November 2016, it had increased to 20.9 kWh after a system software update. It has been on a steady decrease since:

14 Jan 2017: 20.0 kWh
16 Oct 2017: 19.6 kWh
17 Dec 2017: 19.5 kWh
12 Feb 2018: 18.7 kWh

Today, Batt. Kapa. max is 15.2 kWh. The decrease from 18.7 kWh to 15.2 kWh has been pretty steady and disappointing considering how few charge-discharge cycles our battery pack has experienced and what an ideal life it has led. For the past 4 years and 7k miles, Batt. Kapa. max has decreased at the rate of 0.0008 kWh/mi, or ~4 times the rate experienced by 3pete and Steve999. However, our i3 has been stored 4 times for periods of 3 to 9 months for a total of 2 years at ~50% charge level with the high-voltage disconnect disconnected. I have read only one obscure suggestion that Li-ion battery cell inactivity (i.e., no charge or discharge) could lead to reduced capacity. If true, maybe that explains our experience. However, despite last storing our i3 more than 2 years ago, the usable capacity has continued to decline during that period.

The sudden increase in Batt. Kapa. max during the 4 November 2016 system software update suggests to me that the battery management system (BMS) had increased the usable capacity by decreasing the unusable buffer size(s). Some versions of BMW's ISTA+ maintenance software support resetting the BMS which usually increases the usable capacity significantly although a slow usable capacity decline seems to occur thereafter. With less than 2 years remaining in our battery pack warranty, I doubt that our usable capacity will decrease enough to make a warranty claim. However, I plan to reset the BMS using ISTA+ after our battery pack warranty expires. I hope that doing so will reveal how much usable capacity loss has been due to battery cell degradation and how much is due to the BMS decreasing the usable capacity.

 

[3pete]

might I ask what pattern of driving is normal and if you use DC 50kW rapid charging or slower methods, whether you do a lot of top ups or run from high SOC to the bottom quarter then recharge?

Since I've had the car, I've followed the ABC (always be charging) mindset. At first that was on 120v so I often left the next day at less than 100% charge. Then I got a Juicebox EVSE and could get to 100% in a few hours no matter what I plugged in at. I have charged at DC fast charge stations probably about 5 (but definitely fewer than 10) times in the past 2 years. All bets are off on what the previous owners of the car did though. It could've had a lot of DC visits in California for all I know!

I *think* the steep decline at the beginning of my batt kapa max chart from around 40k to 43k might have something to do with switching from L1 to L2 charging. Not that I think L2 hurt the battery capacity, but more so that I'm skeptical of the numbers charged off of L1 because the battery wasn't always charged to 100. It seems like the recent times I've charged to less than 100% before leaving are also the times I've had a ridiculous outlier (you can see a couple after 54k and 56k in that chart). If anyone has any other thoughts on that steeper 40-43k slope I'd be interested to hear them.

@Steve -- I'm also getting Y axis points roughly around 20kWh and somewhere around .0002 loss per mile on a linear fit the data I have (but again, I think it's clear that a linear fit might be a good approximation but is NOT exactly correct for this). It's interesting that Art has actually seen the car display a Kapa Max value above 18.8kWh.

If I see anything else interesting throughout the next few months, I'll try to update.

 

[Star63]

I've been wondering how important it really is not to charge the battery full in order to minimize degradation.
According to another thread, the max. voltage of the 60 Ah battery is 392 V while the 120 Ah battery has 402 V.
This means that the cell voltages are about 4.08 V vs. 4.19 V. The "fully" charged 60 Ah battery has almost 10% capacity reserve to help the battery to last longer. The 120 Ah battery has very little such reserve.

- Is the newer chemistry so much better that the battery can be almost fully charged (without speeding up the degradation rate)?
- A part (about 8% ?) of the capacity difference between 60 Ah and 120 Ah batteries seems to come from the different voltages allowed.

Have I misunderstood something here?

 

[agzand]

A 120 ah battery will go through 50% lower cycles during the life of the car (as well as warranty period), so it can be pushed further at 100% charge. A 60ah battery regularly goes through a deep daily charge/recharge cycle, but a 120ah battery is unlikely unless it is a super commuter with >100 miles commute a day.

In a PHEV (like my Honda Clarity) only 2/3 of the battery capacity is usable (12 kWh out of 17), because the daily charge/recharge cycles are very deep.

 

[bwilson4web]

Thank you!

Bob Wilson

 

[3pete]

It's been 5 months since my original post so I figured it'd be a good time to post an update to this data. Plus, the weather has turned warmer for the season so the winter decline in Batt Kapa Max seems to be over for this year.

Same disclaimers from original post apply (not a random sample of data, one car only etc.).

And to highlight the new data since last post, here is the past 5 months only. The trends are definitely flatter and even upward in graph #2 and #4 which is very likely due to the warmer weather being on the right side.

At this point I am seriously considering taking the car in to ask BMW about the capacity because recent months of Batt. Kapa Max readings have mostly been below 13.1 (which I calculate to be the 70% warranty threshold--anyone please correct me if they know/ suspect otherwise).

Additionally, the "Electrified" app shows my battery State of Health at 68% which both lines up with what the Kapa Max reading shows and is also below 70% last time I checked.

Anyone have any thoughts/ recommendations on whether I should do that or not? I realize I'd have to pay for the test if it passes, but I also feel pretty strongly that the capacity is below the 70% warranty threshold (even if just by a few percent)

 

[3pete]

As a quick update here, I DID take my car in to get checked and my car magically gained more capacity from the trip (full saga here: https://www.mybmwi3.com/forum/viewtopic.php?f=10&p=69332#p69332) .

That was about a month ago, so I have a little bit of data here to update the same charts I've been showing. I took out temperature to replace with before/after check to focus on that. Orange points are the ones after the update.

All metrics seem to show a noticeable jump upwards but #4 seems to be the least significant. Interestingly this metric is the closest to 73% which is supposedly what the test returned as my car's actual battery capacity.

I'll keep updating as time goes on or if anything interesting happens.

 

[3pete]

An update from a few months later:

My Batt. Kapa Max values seem to be holding around 15 or so and the other metrics I track seem to agree that there was a bump in battery capacity from the software upgrade. This makes me think that BMW indeed programmed some "conservative" parameters into the initial BMS which have since been loosened up in an update.

That's completely understandable since they were dealing with an unproven technology but they should offer a software upgrade to anyone with an older i3 if that is indeed the case.

I'm open to other theories if anyone has any thoughts?

 

[alohart]

What is the system software version (a.k.a., integration level) of your i3 after the update? If the BMS has been updated in recent software versions, I might be willing to pay our BMW dealer to update the system software on our 2014 BEV whose Batt. Kapa. max value has dropped to ~15 kWh after having been driven only 12k miles.

 

[3pete]

What is the system software version (a.k.a., integration level) of your i3 after the update?

During the capacity check I went from: I001-17-11-543 to: I001-21-03-530.

If the BMS has been updated in recent software versions, I might be willing to pay our BMW dealer to update the system software on our 2014 BEV whose Batt. Kapa. max value has dropped to ~15 kWh after having been driven only 12k miles.

I can't say definitively that the software upgrade changed the BMS logic, but it seems to make sense to me. Art, yours would be a very interesting case with such low mileage and such an ideal climate for an EV. I wonder if you could convince your dealer to upgrade/ check "for science" to report back to the BMW mothership. With the chip shortage and so few new cars to sell, I'm sure they're bored and looking for something to do :lol:

 

[alohart]

During the capacity check I went from: I001-17-11-543 to: I001-21-03-530.
Thanks! It would really be nice to know what changes in each software update. The fact that BMW is apparently still releasing an i3 software update every 4 months makes me wonder what is changing. I'm guessing that the changes affect mostly recent i3's and not old ones like mine.

I can't say definitively that the software upgrade changed the BMS logic, but it seems to make sense to me. Art, yours would be a very interesting case with such low mileage and such an ideal climate for an EV. I wonder if you could convince your dealer to upgrade/ check "for science" to report back to the BMW mothership. With the chip shortage and so few new cars to sell, I'm sure they're bored and looking for something to do :lol:

… or looking to separate me from my money

When I return in mid-November, the annual state safety inspection will be due. I might as well let our BMW dealer perform this inspection because it's a fixed cost no matter where it's done. When I talk to the service advisor, I could plead my case and learn the cost for a software update if they won't do it at no cost. Another local i3 owner did get a capacity check including a software update at no cost, so there's hope. I don't really need more range, but it would be nice to have some additional range on occasion.

 

[Arm]

I bought my 2018 i3S with 43500 miles on it on 9/4/2021. At the time, I checked the Kappa and it read 29.2 kWh. It was bang in the middle of summer and it was a hot day. I don't recall what the SOC was at the time or if the car was charging. But I do know the temperature was well in the 90's.

I checked my vehicle today (11/13) and I got the following:

Kappa = 28.8
OA Temp = 69.5F (in a garage)
Coolant Temp = 118 C (not plugged in...can't make sense of this reading)
Batt Temp = 71.6 F
SOC = 93%
Current Miles = 44500

I take the approach of almost always having it plugged in. I feel it's better for the battery to be conditioned the way it was intended by design. That said, I do not know how the car was treated prior to my purchase. I do know the overall condition of the car was good and I am assuming the car did a lot of freeway miles given the fact that it averaged 15k miles a year. So I would assume there were quite a few fast charging sessions involved given it's a BEV. The original purchase location of the car was Palm Springs 3 years ago which is roughly 120 miles away from where I am.

 

[3pete]

Another update since it's been about six months since the last one. I added a vertical line to indicate where the battery test & software update happened (even though it's pretty obvious in most charts).

Everything is trending back downwards, but that also coincides with colder temperatures, which does seem to have some impact on these metrics.

Other than that, I find it interesting that #4 is the metric that is least impacted by the test/update and even though it has the largest +/- from datapoint to datapoint, it is the flattest trend line AND seems to agree with the value BMW told me their test returned on battery SOH.

FWIW, Electrified and Mi3 report SOH as 74%, down from 80% at the time of test (vertical line).

 

[gt1]

Thanks for sharing.
Did you notice any practical change in the range after the update?
How do you collect the data? With so many points it has to be automated. I graphed the data through BMW Connected and smart home software for a while, but BMW stopped reporting Batt. Kappa Max through the API.

 

[3pete]

Did you notice any practical change in the range after the update?

Range seemed to increase with the update. But it's worth noting that I've never run the car to 0 and I base all range assumptions from the GOM, which is obviously calculated off Batt Kapa Max.

The other thing worth noting is that even though the change looks drastic, the battery is so small in the 60ah i3 that 2kWh more capacity/approx 8 miles of range is a very noticeable difference even though it doesn't really impact what you can use the car for that much.

How do you collect the data? With so many points it has to be automated.

I wish it was automated but I collect the data manually. I started because I wanted a reference point when I got the car then thought it could be interesting to track periodically in an online spreadsheet and realized it's actually not too time consuming to do. I only log data "when I feel like it" which might skew the data compared to an automated approach and obviously I'd prefer the reliability of data from an automated solution, but this is the best I could come up with given the technology constraints BMW gave us.