Tuning the Battery Aging Model

[Stoaty]

I made an experimental version 0.95f of the Battery Aging Model:

https://dl.dropboxusercontent.com/u/48149991/Leaf%20Battery%20Degradation%20Model%20Version%20095f.ods" onclick="window.open(this.href);return false;

After studying the table on the calibration table and doing some graphing, plus Nissan's focus on how many miles the Phoenix drivers that lost bars had driven, I decided that the assumption of 1.5% capacity loss per 10,000 miles might be too low. In version 0.95f I changed that assumption to 2.0% (on the Degradation Model tab) to see what effect that would have. Unfortunately, I still have to calculate the Corrected Predicted Total Loss for each entry, although it only takes about 10 seconds per entry since it is semi-automated. I haven't figured out a way to automate the correction.

Mean of (Actual Loss - Predicted Loss) = 0.29% (was 2.01% before the parameter was changed)
Standard Deviation of (Actual Loss - Predicted Loss) = 1.53% (was 1.93% before the parameter was changed)
(Maximum of (Actual - Predicted)) - (Minimum of (Actual - Predicted)) = 4.85% (was 6.30%)

Comment: Mean of differences is now close to zero, Standard Deviation is less, range between highest and lowest difference is less. For now I am going to track both models with additional entries (from both Leafs that haven't been reported yet, and repeat reports every 3 months). Another 6-12 months should give enough info to see if the model can accurately predict loss for new entries.

 

[Stoaty]

With the cycling loss set to 2% per 10,000 miles, I re-calculated the End of Life prediction for each of the 17 entries we have so far. Results are not pretty:

3.10
4.40
7.10
8.20
2.70
5.40
8.10
17.80
4.90
7.30
5.00
3.70
4.40
5.60
2.70
4.80
5.20

9 of the 17 are at or below 5 years based on 70%. This suggests that there may be a lot more batteries replaced under warranty than we (or Nissan?) were expecting. Of course, Nissan sets the level at 66.25% (4 bar loss), which drops the predicted at or below 5 years to EOL to only 5 of the 17.

 

[Stoaty]

Based on Nissan's warranty criteria (5 years/60000 miles), only 3 of the 17 entries are on track for warranty replacement (I forgot to consider the mileage limit in the previous post; now added to spreadsheet). The 3 Leafs that should get warranty replacement are in Phoenix, Dallas, and Miami. For the remaining 14 entries that most likely won't qualify:

Average End of Life - 6.5 years
Average Mileage at End of Life - 82,000 miles

Those in Seattle area can expect 110,000 to 120,000 miles until EOL (assuming the model holds, of course).

Certainly a far cry from what Nissan claimed before we purchased. The battery doesn't just have a problem in hot areas, the "average" areas aren't predicted to do all that well either. Stoaty is only slightly above average at predicted EOL of 7.30 years (in spite of babying his battery).

 

[91040]

Should be easy to get a lot of data for this aging model with AltCar (Santa Monica, CA) and PlugIn Days taking place. I plan to print out this form to gather data:

Leaf Battery Aging Model

This information is being used to develop a model to predict battery behavior. All the requested data is required to be of use. Thank you for your help.

Date of Manufacture (month & year) _____/________
Date of Delivery (month, day & year) _____/_____/________
Date of P3227 update _____/_____/________
(If needed. Do not submit data until at least 3 weeks after update.)
Geographic location (City/State/Zip) _______________________________________________
-(If climate significantly different than home:
---Work location (City/State/Zip) ________________________________________
---Days Leaf at work _________
Average miles/kwh for the life of your Leaf (best guesstimate OK) ______miles/kWh
Current odometer (total miles) _______________
Current Ah capacity __________
Date of odometer & Ah reading _____/_____/________
Days/week parked in the sun (fractional days OK) __________
Last 4 numbers of VIN __________ or MNL User Name ____________________

Or post information directly to http://www.mynissanleaf.com/viewtopic.php?f=27&t=14275" onclick="window.open(this.href);return false;

 

[Weatherman]

The 3 Leafs that should get warranty replacement are in Phoenix, Dallas, and Miami.

The only question will be whether it will hit 8 bars before or after I turn it in in a little over 21 months. Ahr readings are still falling rapidly (0.21 in the past 10 days):

Capacity – 55.46
Date - 9/9/13


Capacity - 55.25
Date - 9/19/13

 

[garygTx]

Here is my data:

Manufactured - 5/11
Delivered - 11/12/11
P3227 update - 7/05/13
Location - Garland, TX (Northeast side of Dallas)
Miles/kwh - about 5.4 (as reported on dash - weighted average over life)
Odometer - 20,004 miles
Capacity - 55.22 AHr
Date - 9/19/13
Parked outside at home and work
1 bar lost 8/??/12
Health 61.92%

 

[Stoaty]

Here is my data:
Parked outside at home and work

Are you parked in the sun, or in the shade? Need to know what percent of days you are parked in the sun. Thanks.

Predicted Loss: 19.85%

Actual Loss: 16.65%

Comment: You are beating the model by the largest margin of those who reported data - 3.2%. One thing that jumps out at me is that a period of 6 months elapsed between manufacture and delivery. I wonder if the Leaf was sitting in a cooler climate during that time (and at an optimal state of charge) and whether that could account for the difference. Of course, there is going to be a certain amount of variation anyway, just curious.

Edit: Based on the model, you are predicted to qualify for a warranty replacement. However, if you continue to track ahead of the model you probably won't qualify.

 

[SJFlyer]

Here is my data:

Date of Manufacture (month & year) 03/2011
Date of Delivery (month, day & year) 04/30/2011
Date of P3227 update Not applied
(If needed. Do not submit data until at least 3 weeks after update.)
Geographic location (City/State/Zip) Los Gatos, CA 95032
-(If climate significantly different than home:
---Work location (City/State/Zip)
---Days Leaf at work 5
Average miles/kwh for the life of your Leaf (best guesstimate OK) 4.2 miles/kWh
Current odometer (total miles) 36563
Current Ah capacity 54.22
Date of odometer & Ah reading 09/19/2013
Days/week parked in the sun (fractional days OK) 5
Last 4 numbers of VIN 1423

 

[TomT]

It certainly makes a case for "drive the hell out of it and don't worry about it; it is inevitable anyway..."

Stoaty is only slightly above average at predicted EOL of 7.30 years (in spite of babying his battery).

 

[Stoaty]

Date of P3227 update Not applied

While interesting, data is of no use for the analysis unless the P3227 update has been applied. Have to control as many variables as possible, and a software update that improves the accuracy of the capacity gauge would be at the very top of that list. Any reason you haven't gotten the update? Remember that the update also includes compatibility fixes so you won't fry the charger in your Leaf from some EVSE you may encounter during charging at a public charger.

 

[garygTx]

Here is my data:
Parked outside at home and work

Are you parked in the sun, or in the shade? Need to know what percent of days you are parked in the sun. Thanks.

Predicted Loss: 19.85%

Actual Loss: 16.65%

Comment: You are beating the model by the largest margin of those who reported data - 3.2%. One thing that jumps out at me is that a period of 6 months elapsed between manufacture and delivery. I wonder if the Leaf was sitting in a cooler climate during that time (and at an optimal state of charge) and whether that could account for the difference. Of course, there is going to be a certain amount of variation anyway, just curious.

Edit: Based on the model, you are predicted to qualify for a warranty replacement. However, if you continue to track ahead of the model you probably won't qualify.

Most of the time the car is parked in the sun, however I noticed a HUGE typo in my capacity reading. The correct value is 52.22. It was 52.31 yesterday. Low about 80F, high 95F. Charger set on end timer to complete at 4:00 AM.

The car was a dealer demo for the six months before I leased it.

 

[Stoaty]

Most of the time the car is parked in the sun, however I noticed a HUGE typo in my capacity reading. The correct value is 52.22. It was 52.31 yesterday. Low about 80F, high 95F. Charger set on end timer to complete at 4:00 AM.

The car was a dealer demo for the six months before I leased it.

Thanks for the correction. I was wondering how you were doing so well in Dallas. While I was going through the numbers I found a subtle error in calculation that decreased the predicted amount of cycling loss for you by about 1%. Others were affected, but generally by less. Some moved further away from predicted, some moved closer with the correct formula.

Your stats:

Predicted Loss - 18.61%

Actual Loss - 21.18%

Updated overall stats on Actual Loss - Predicted Loss:

Mean - 0.88%
Standard Deviation - 1.48%
Median - 0.71%

There is a slightly larger systemic bias (model still a bit optimistic), but the standard deviation is less after correcting the formulas.

 

[DaveEV]

Most of the time the car is parked in the sun, however I noticed a HUGE typo in my capacity reading. The correct value is 52.22. It was 52.31 yesterday.

That makes sense - I was trying to figure out how you had the same Ah reading as me but your Health reading is about 10% lower...

 

[Stoaty]

Here is version 0.96 of the Battery Aging Model:

https://dl.dropboxusercontent.com/u/48149991/Leaf%20Battery%20Degradation%20Model%20Version%20096.ods" onclick="window.open(this.href);return false;

Changes:

--added several more entries to calibration tab (thanks for submitting!)
--I decided that the 2% degradation rate per 10,000 miles will be the new standard, since it fits the data significantly better than the old assumption
--Found and corrected a subtle bug in the calculation of solar loading on the Calibration Tab (it was using the Solar Loading Factor from the prediction tab, which meant that everyone on the Calibration tab got the same solar loading factor). However, with the current data the change in the Predicted Loss was very small.
--Found and corrected a subtle bug in the calculation of Corrected Cycling Loss. The formula was using battery age rather than years in service, so the change was fairly small for most (about 0.5%), but for garygTx it made a difference of 1% since his Leaf mostly sat around for 6 months before he took delivery. This correction actually decreased the standard deviation of Actual minus Predicted Loss. The correction was made on both the Prediction and Calibration tabs.
--Work in Progress on the Calibration Tab to fully automate the calculation of Corrected Cycling Loss, so that it will be updated if the Parameters on the Degradation Model tab are changed. Right now I have to manually enter a number and get the calculation, which is manually entered in the Corrected Predicted Total Loss column. Since the number is then hard coded, it won't change appropriately if other parameters are changed. Scroll down to see the "Weatherman" table at the bottom of the Calibration Tab. This portion is not working yet.

 

[Devin]

I somehow seem to be the luckiest (?) person on here in terms of beating the estimate. Nice to know that under current patterns I wouldn't get a new pack. I'm planning to transfer to CSUN next year which has just installed two DCQC stations so I may see my car's results swing the other direction and become eligible for warranty coverage.

 

[madbrain]

Manufactured - 6/2012
Delivered - 10/20/2012
P3227 update - 7/3/2013
Location - San Jose, CA; Leaf spends 4 days a week in Santa Clara, CA
Miles/kwh - 3.5 miles/kWh lifetime (from Carwings)
Odometer - 9304 miles
Capacity - 61.49 Ahr
Health : 86.14%
Date - 9/19/13
Parked in sun - 4 days/week . In west-facing garage 3 days/week

 

[KJD]

Manufactured - 11-2011
Delivered - 12-08-2011
P3227 updte - 08-29-2013
Location - Salt Lake City, UT 84105
Miles/kwh - 4.8
Odometer - 16092
Capacity - 59.37 ah (89.61%)
Date of reading - 09-20-2013
Parked in sun - 0 days per week

Car is a 2012 SL model.
GID count of 210 - 80% chg today is the same as 2 days before the P3227 update.
OAT was 52f this morning.
Car is parked outside at night June, July and Aug. At this point not sure if that makes any difference.
Charging is 80% end only timer set for 7AM

 

[palmermd]

Here is an example with my data:

Manufactured - 03/11
Delivered - 6/17/11
P3227 updte - 7/15/13
Location - West Los Angeles, CA; Leaf spends 4 days a week in Canoga Park, CA
Miles/kwh - 5.6
Odometer - 20720
Capacity - 57.07
Date - 9/10/13
Parked in sun - 4 days/week

I realize you prefer data after the update, but here is a point before. I'll send it in again later for a second point.

Manufactured - 02/11
Delivered - 3/30/11
P3227 updte - not yet done. setting appointment today, and will resubmit report a month or so after completion.
Location - Stockton, CA 3/11 to 7/12 then Sacramento, CA 7/12-present (both climates are identical so should not matter)
Miles/kwh - 4.2
Odometer - 18,770
Capacity - 57.45 (lost 1 point since June 18: was 58.43)
Date - 9/20/13
Parked in sun - 1 day per week average

 

[Stoaty]

KJD:

Predicted Loss - 10.02%
Actual Loss - 10.38%

palmermd (pre-P3227):

Predicted Loss - 13.11%
Actual Loss - 13.28%

Comment: I think the Battery Aging Model is pretty well dialed in, with the possible exception of hotter climates and/or larger losses. However, plotting the Actual Loss/Predicted Loss vs. Predicted Loss didn't show any clear trend (see attached, only includes post-P3227 Leafs). We need more readings from hot climates.

PS I may not respond to every single reading submitted, but keep those cards and letters coming! I am adding them to the spreadsheet, and you can look at the results when I release an updated version.

 

[stjohnh]

Stoaty,

Do you want information on the MY2013s made in May or later? They typically have capacities in the 60-61 range when new, and over the next several months increase in capacity. AFAIK, it is not known why their capacities are reported low, though testing by several of us show that they have normal range, require 21KwHr to recharge, and except for the low reported capacities, seem entirely normal. (Boomer has suggested the name "warm" batteries for these).