Capacity Loss on 2011-2012 LEAFs

[RegGuheert]

Interesting idea, but I don't believe that this is how it works. Phil told us that each Gid corresponded to 80 Wh of energy stored in the battery.

But Phil has also told us that there are innaccuracies in the Hall-Effect devices used to measure the current into and out of the battery. I guess It would be pretty coincidental if the batteries all have *exactly* 24 kWh when new AND the measuerment system in the LEAF was also so accurate that they all read 281 GiDs when new.

I see in the service manual on page EVC-107 there is a LI-ION GRADUAL CAPACITY LOSS DATA CLEAR command to be used when the battery or the controller is replaced with a new one. It would seem that one would need to also tell the controller the measured capacity of the new battery under some conditions, but I don't see a command for that.

Anyway, I could be wrong about all this, but I agree the repeatability of the 281 GID measurement between cars seems too good for it to be absolutely accurate. As such, it must be some representation of relative capacity of the battery versus new.

 

[surfingslovak]

Interesting idea, but I don't believe that this is how it works. Phil told us that each Gid corresponded to 80 Wh of energy stored in the battery.

But Phil has also told us that there are innaccuracies in the Hall-Effect devices used to measure the current into and out of the battery. I guess It would be pretty coincidental if the batteries all have *exactly* 24 kWh when new AND the measuerment system in the LEAF was also so accurate that they all read 281 GiDs when new.

Obviously, all of the sensors and instruments in the car will have inaccuracies. And I agree that such a low manufacturing variance would be quite a feat. Is it possible? Maybe. Is it likely? Probably not. And if you add all the instrument errors and seasonal differences, getting less than 0.3% delta between cars is virtually impossible. To be fair, our sample size of new cars is pretty small as well. Still, this is a good point to raise.

I see in the service manual on page EVC-107 there is a LI-ION GRADUAL CAPACITY LOSS DATA CLEAR command to be used when the battery or the controller is replaced with a new one. It would seem that one would need to also tell the controller the measured capacity of the new battery under some conditions, but I don't see a command for that.

This is just a guess, but I suspect that the consistent Gid readings we see, and the seasonality we observe, are an expression of something the BMS does. We just started recording pack voltages on a 80% and 100% charge. They might give us a clue.

Anyway, I could be wrong about all this, but I agree the repeatability of the 281 GID measurement between cars seems too good for it to be absolutely accurate. As such, it must be some representation of relative capacity of the battery versus new.

To be fair, some forum members reported higher energy drawn from the wall than anticipated, and their range tests indicated more available energy than average. On a related note, it's worth noting that some folks have never reached 281 Gids (e.g. GregH, TickTock), and others have measured more than 281 Gids after a quick charge (jkirkebo). Perhaps it's conceivable that while a Gid is 80 Wh, there is a multiplier that the BMS applies to account for capacity variance. Maybe it's the same coefficient they use to track capacity loss? I wish we knew more. There has been tremendous progress over the last 12 months, and hopefully the trend will continue.

 

[Stanton]

Here are my results after borrowing a SOC/GID meter; I also uploaded to "turbo2ltr" online spreadsheet and will cross-post in the "281" thread. In addition, I did a (timer off) over-night charge to allow for cell balancing with basically no change.

After 80% charge: 217 gids (77.2%)
After 100% charge: 261 gids (92.8%)

~8700 miles after 9 months
only charge to 100% about 5 times/month (rest of time 80% with 6am end-only timer)
never "top-off", never sits at 100% overnight
garaged night/weekends, outside during the day

 

[EVDRIVER]

It would be interesting to map this data against VIN numbers in addition to geography. My VIN is in the very low 400's and I have almost 7k miles with 280. My pack rarely sees temps of 70 and is usually sitting in the mid 60's. I do charge to 80 but to 100 frequently as well. I also hammer the pack very hard and very often but a cool climate in the bay area keeps the temps down.

 

[tbleakne]

Both the 80% and 100% voltages are preset and fixed. The battery holds, what it can hold at those potentials. As it ages, and its storage capacity deteriorates, the maximum Gid count will decline. I believe that we are beginning to see that. The only open question is seasonality. You would think that the battery will hold more energy at 394V when it's warm out, not less. Consequently, the Gid count should be higher in the summer, but that seems to be the opposite of what we are observing.

Your statement matches my understanding. It is true that fairly cold temperatures temporarily reduce the battery's capacity, but temperatures above some optimal point also reduce its capacity. I would like to know more about this mechanism. The maximum power the battery can produce may increase monotonically with temperature.

 

[tbleakne]

Here are a few data points:
>>>
May 29, 80% charge from 163 Gid to 225 Gid, 387.5 Volts, low temp 57 F

Now I understand. You charged starting at 163 Gid and ended at 225 Gid.
For most of this last winter, I always got 231 Gid (82.2%) for 80% charge.

My latest 80% result:
223 Gid, 388.0V, 63 F. I am down 8 Gid for 80% and 11 Gid (-3.9%) for 100% charging. Not a severe problem now, but worrisome if this decline continues.
I am still trying to get my overnight garage temperature cooler using a fan to rule out temporary temperature loss. I want to do this while the night time minimums are still below 60 F.

I truly feel for the folks in AZ and elsewhere who have apparently lost a whole lot more.
I am also concerned about 91040's results, since his climate and mine should be similar:

Recent gidmeter readings:
7 charges at 100%: 91.1 - 92.5%. Battery: 393 - 394.5V
3 charges at 80%: 76.5 -76.8%. Battery: 386.5 - 388.5V
Mileage: ~23,800
Delivery May 2011

In addition to his much higher mileage, he has had to push his battery to the very low end a number of times.

 

[sparky]

My latest 80% result:
223 Gid, 388.0V, 63 F. I am down 8 Gid for 80% and 11 Gid (-3.9%) for 100% charging. Not a severe problem now, but worrisome if this decline continues.

Tom, my 100% numbers are reduced about the same yours for the last 2 weeks: ~269 'Gids', 393V, averaged over 6 100 % charges.
But, my nine 80 % charges are all 231, 389V±0.5 for the same period. Not sure why it's inconsistent, since I thought the 80% 'Gid' number was just as reliable a predictor for losing a bar; i.e., capacity loss, as the 100%.
Incidentally, my charges have all been at around 60-70 F.

 

[DaveinOlyWA]

one thing i have not tracked regularly is my pack voltage. i have NEVER gotten 281... well at least i am pretty sure of it. i did not track my #'s at first but have hit 280 probably a dozen times. i have had as low as 274 but i am guessing that my not seeing 281 has to do with the timing of the charge complete and when i actually check the car.

also, it could simply be the car itself. i notice that i can actually just sit in the car while its powered on in park and in a few cases, i drop to 279 in a few minutes. so i am guessing the time at 281 would be even shorter.

to reduce confusion, i think we should start another thread that tracks both GID, OAT (outside air temp or whatever you estimate your temp be it outside, garage, etc) Voltage and time between charge completion and when you checked the measurement

 

[vegastar]

I track the gids and voltage before and after the charge, with at leat 30 minutes of rest time where the pack is floating. From my experience there is no further change in the pack voltage after those 30 minutes.

 

[DaveinOlyWA]

I track the gids and voltage before and after the charge, with at leat 30 minutes of rest time where the pack is floating. From my experience there is no further change in the pack voltage after those 30 minutes.

30 minutes might not be enough. some report charging spikes 90 minutes to a few hours later

 

[vegastar]

I usually check the charge stop e-mail to check if the pack is more than 30 minutes floating.

 

[tbleakne]

The NREL report posted by Weatherman in the "Expected . . loss 70%" thread is very interesting reading. I know the specific Li chemistry is important, but they mention Li-MnO4 on their 8th slide. I believe the effects they discuss are the same for us, possibly with different magnitudes. They rate the shelf-life loss mechanism, proportional to t^.5, as well-understood, but the cycle loss mechanisms are less well-understood.

In an attempt to measure the relative contribution of shelf-live loss and cycle usage for us, I setup a very simple model comparison between my own data and that of 91040 (Allen), assuming he, being at a similar altitude in the Los Angeles basin, has a similar micro-climate.
S = shelf-life loss in % per year
C = cycle loss in % per 1K miles
His data is for 13 months, mine for 12 months, close enough to ignore the t^.5 dependence for the shelf life.
91040 total loss:
1.08S + 24C = 7.5%

my total loss
1.0S + 11C = 3.9%

Solving above system of 2 linear equations in 2 unknowns, we obtain:
S = .9 % loss per year
C = .27% loss per 1K miles (1% loss per 3.7K miles)

The system is nearly singular, so these results are somewhat suspect. The shelf-life component is very likely larger. 91040's usage rating probably should be higher than his mileage alone indicates because he frequently takes his battery down to LB and below. The t^.5 dependence means we can expect the shelf-life component to be less in the coming year, but this result does not bode well for the predicted total loss.

 

[Stoaty]

Solving above system of 2 linear equations in 2 unknowns, we obtain:
S = .9 % loss per year
C = .27% loss per 1K miles (1% loss per 3.7K miles)

Interesting, but over 6,500 miles and 8 months I lost only 3 Gids (for 80% charge, rarely charge to 100%). That is 3/224 = 1.3%. Using your formula, I would expect a 2.75% loss. I lost 3 Gids over 5 months for 100% charge (haven't charged to 100% for 3 months so can't give an exact amount of loss for 8 months). That is 3/272 = 1.1%. If linear and extrapolated to 8 months that would be 1.8%. Either way, I think the loss per 1,000 miles is too high. My guess is that the calendar life losses (as a function of temperature) are significantly more than the cycling losses, unless you live in a very cool climate.

 

[abasile]

Solving above system of 2 linear equations in 2 unknowns, we obtain:
S = .9 % loss per year
C = .27% loss per 1K miles (1% loss per 3.7K miles)

Interesting, but over 6,500 miles and 8 months I lost only 3 Gids (for 80% charge, rarely charge to 100%). That is 3/224 = 1.3%. Using your formula, I would expect a 2.75% loss. I lost 3 Gids over 5 months for 100% charge (haven't charged to 100% for 3 months so can't give an exact amount of loss for 8 months). That is 3/272 = 1.1%. If linear and extrapolated to 8 months that would be 1.8%. Either way, I think the loss per 1,000 miles is too high. My guess is that the calendar life losses (as a function of temperature) are significantly more than the cycling losses, unless you live in a very cool climate.

You do cycle your pack much more gently, avoiding high/low SOC extremes, than 91040.

Also, at 15K miles, our LEAF reported 280 gids (as I've posted elsewhere). At this point, at least with a pack that isn't cycled too deeply, I think temperature effects really tend to dominate. While this is bad news in hot climates, it could be very, very good news for many LEAF drivers in relatively cool areas like mine; their packs might last much longer than initially expected.

 

[Stoaty]

You do cycle your pack much more gently, avoiding high/low SOC extremes, than 91040.

Do you cycle your pack gently? I understand you still have outstanding number of Gids... of course, that cool mountain living probably isn't hurting your battery capacity any.

 

[abasile]

You do cycle your pack much more gently, avoiding high/low SOC extremes, than 91040.

Do you cycle your pack gently? I understand you still have outstanding number of Gids... of course, that cool mountain living probably isn't hurting your battery capacity any.

During Year 1, we were relatively gentle in terms of cycling. However, these days we're often cycling between 80% and LBW, as the car is going down the hill (where it's hotter) much more frequently. Year 2 will not be as gentle. We'll see what effect that has.

 

[surfingslovak]

The NREL report posted by Weatherman in the "Expected . . loss 70%" thread is very interesting reading. I know the specific Li chemistry is important, but they mention Li-MnO4 on their 8th slide. I believe the effects they discuss are the same for us, possibly with different magnitudes. They rate the shelf-life loss mechanism, proportional to t^.5, as well-understood, but the cycle loss mechanisms are less well-understood.

Tom, interesting model, very neat. That said, and I'm sure you are aware of that, it's difficult to make any projections based on two vehicles. I won't have much time this week to revisit the discussion, but I collected all the data that has been posted recently, and tried to correlate Gids to various parameters. I could be wrong, but I saw a small correlation between Gids and car mileage (correlation factor of 2.5%). Surprisingly, seems to be a much weaker relationship to vehicle age (correlation factor of -0.5%). This is likely because most of the cars were about one year old, and none of them have seen less than six months of use. Although I counted more than 10 times as many cars, it was still a very small sample.

 

[Herm]

However, these days we're often cycling between 80% and LBW, as the car is going down the hill (where it's hotter) much more frequently.

That is still a fairly gentle treatment of the battery. LBW is at the 8.5% SOC point?

 

[surfingslovak]

That is still a fairly gentle treatment of the battery. LBW is at the 8.5% SOC point?

Herm, it's between 17 and 18%, and well ahead of the knee on the discharge curve.

 

[Cheezmo]

I just measured my GIDs for the first time last night after a 100% (timer override, starting near 80%) charge.

Delivery late March 2012, Plano, TX. ~2500 miles. Mostly charge to 80%. Have QC'd 4 times in one day, 2 times on a couple of days.

265

Will take some 80% and other measurements over the next few days.