Battery Aging Model

edatoakrun said:

Didn't both TickTock's and Azdre/Opossum's LEAFs actually suffer two, rather than 1.6, Phoenix "Summers", the ~3 months with the highest ambient temperatures?

So aren't you actual "tuning" your model to match an incorrect value, a world where Phoenix has two summers in only ~16 months?

Click to expand...

No, because I didn't do any tuning based on those numbers. As I said, the model doesn't account for it, and fractional years will consequently be less accurate than a whole number of years. The model is based exclusively on Nissan's data, with the exception of the solar loading factor which is based on the study of the Prius (probably not a great choice, but I don't have anything else to go on; a scaling factor is included so I can easily increase or decrease the effect of solar loading).

Do you not consider variables in Battery care may be very large factors in capacity loss?

Click to expand...

If you have been following the thread, you will know that this factor is not included in the model yet, and may or may not be depending on whether I have some kind of data or study to reasonably base it on. That data is certainly not available from Nissan, and wasn't mentioned by Nissan as a significant factor in the Casa Grande cars. Armchair quarterbacking not helpful.

I am ready to release version 0.81 of the Battery Aging Model, but I don't have a good place to make it easily available for download from a link in the Wiki. Suggestions?

PS Many thanks to JP White for figuring out how to make a dropdown list including only the cities for which I have an Aging Factor!

Stoaty said:

I am ready to release version 0.81 of the Battery Aging Model, but I don't have a good place to make it easily available for download from a link in the Wiki. Suggestions?

PS Many thanks to JP White for figuring out how to make a dropdown list including only the cities for which I have an Aging Factor!

Click to expand...

How about using dropbox? A 2GB account is free and the public link is only good for a single file shared in the public folder.

Stoaty said:

I am ready to release version 0.81 of the Battery Aging Model, but I don't have a good place to make it easily available for download from a link in the Wiki. Suggestions?

PS Many thanks to JP White for figuring out how to make a dropdown list including only the cities for which I have an Aging Factor!

Click to expand...

I would recommend https://drive.google.com it can import excel, you can share, and a really nice feature is the ability to create forms to collect data such as you requested on the other thread. For example, on my daily log, I created a online form with just the information I need and it automatically adds a row to the spreadsheet with the new information. Any web-attached device can make an entry.

Forgot I already have a dropbox account. Version 0.81 available here now:

https://dl.dropbox.com/u/48149991/Leaf%20Battery%20Degradation%20Model%20Version%20081.ods" onclick="window.open(this.href);return false;

There is brief documentation on the "Documentation" tab of the spreadsheet. It is strongly suggested that you read it before asking questions or making suggestions for improvement. Spreadsheet created in OpenOffice, you need that to open and use it. OpenOffice is available here:

http://www.openoffice.org/download/" onclick="window.open(this.href);return false;

Version 0.82 of Battery Aging Model now available here:

https://dl.dropbox.com/u/48149991/Leaf%20Battery%20Degradation%20Model%20Version%20082.ods" onclick="window.open(this.href);return false;

Changes from version 0.81:

--added Sacramento, CA to list of cities the model can predict for
--additional Tab added "End of Life" for brute strength method of End of Life calculation
--added End of Life calculation (number of years to nearest 0.1 year), with user selectable percentage for end of life (default = 70%) - for Fractional Year Prediction section only
--added Miles at End of Life calculation based on calculated annual mileage and predicted End of Life
--minor updates to Documentation

Version 0.83 of Battery Aging Model now available here:

https://dl.dropbox.com/u/48149991/Leaf%20Battery%20Degradation%20Model%20Version%20083.ods" onclick="window.open(this.href);return false;

Changes from 0.82:

--added El Paso, TX
--added corrected aging factors for Dubai and San Juan Puerto Rico so predictions will now work for those cities
--fixed bug that caused international cities to show bogus solar loading effects (solar loading not implemented for international cities, it now correctly shows no effect)
--percentage remaining capacity that is less than end of life shown in red
--Days Per Week in Sun = 0 now shown with white text on black background to remind you it isn't set
--Documentation updated, including to lower zoom percentage if everything doesn't fit on screen

Version 0.83 of the Battery Aging Model is now available as a Microsoft Excel spreadsheet:

https://dl.dropbox.com/u/48149991/Leaf%20Battery%20Degradation%20Model%20Version%20083.xlsx" onclick="window.open(this.href);return false;

Thanks to JPWhite for saving in Microsoft Office 13 and making a couple of small changes for compatibility.

Links to both spreadsheet versions of the Battery Aging Model added to the wiki:

http://www.mynissanleaf.com/wiki/index.php?title=Battery_Capacity_Loss#Factors_Affecting_Battery_Capacity_Loss" onclick="window.open(this.href);return false;

RegGuheert said:

I'm not sure it is worthwhile trying to fit a model to the curves which Nissan gave to TickTock. Frankly, I don't expect to see any LEAFs in Phoenix with 76% capacity left after five years, regardless of miles driven.
>>>>>
As usual, I hope I am wrong about this. OTOH, Nissan's track record for making predicitions in Phoenix is quite poor to date.

Click to expand...

I commend all the detailed work that has been reported on this thread, using solid analysis to apply corrections, but I still agree with this comment posted on pg 1. The curves Nissan gave to TickTock are not data, but rather their model. We should suspect that this model has been massaged to fit with the optimistic projections that Nissan is still standing behind. I am concerned that folks might put more faith in your results than is warranted to the extent that they are based upon their model.

From another thread:

Stoaty said:

surfingslovak said:

It should also not come as a surprise that owners experiencing 15% range loss or more after less than 1 1/2 of ownership feel uneasy about Nissan's projections of 80% original range after five years (76% for Phoenix).

Click to expand...

Especially true because the Battery Aging Model built from Nissan's own data predicts 76% remaining capacity for Phoenix owners driving 12500 miles per year at an average 4 miles/kwh efficiency after only 3.3 years. You can only reach 5 years with 76% remaining capacity if you limit your mileage to 7500 miles per year.

Click to expand...

One clue that Nissan's curves might not be based upon solid science is that they show no seasonal variation, contrary to what we know will happen from the Arrenhius factor. Published government studies clearly show annual variation, with steeper decline during Summer and a leveling off during Winter, such as this graph from the NREL paper which has been displayed before:



For what it is worth, an update of my data is below, based upon 160 days of monitoring the Gid count whenever I charged to 80% with no End Time specified.

Mileage: 11K per year steady.
DOC: on days not shown on the graph,
Upper Limit: usually 40-50% based upon current Gids, which should be 45-55% of current capacity.
Lower Limit: typically 25%, never more than 2% below LBW.
Charges to 100%: 3 during this period
Charges to 85-90%: 2-3 per month.

A lot of people on this forum do not trust the Gids. I am aware of your opinions, and I have found them not perfect by any means, but more than adequately reliable. Assessing the SOC of a Li-ion battery pack is clearly not an easy problem, but Nissan has done a fairly good first attempt. While we don't know the details, it is clear to me that the Gid calculation uses both voltage and coulomb-counting. The voltage varies substantially as a function of power level, so coulomb-counting must be used for minute-to-minute readings while driving. During stops Open Circuit Voltage can be used to re-calibrate the Gids. At the outer-most, slowest level of the algorithm, the coulomb-counting is used to adjust the SOC vs voltage table as the battery capacity declines. The algorithm has problems estimating the energy stored at the very bottom of the battery, especially if one never goes that low. We can agree to disagree on how relevant are errors below VLBW.

For modeling capacity loss the big problem is trying to get a good measure of the temperature history of the battery pack, with only ambient temperature and the temperature bars to guide us. Given that the thermal time-constant of the battery seems to be somewhere in the range of 4 to 8 hours, my experience is that the battery history over a 24-hr period is quite significant. In my inland Southern California climate, the battery swings between 5 and 6 Tbars, so I have attempted to monitor the fraction of time the battery spends at each level with a scale that runs from 0 to 1, with the following interpretation:

Temperature Value Interpretation
0 100% "deep" within 5 Tbars (24-hr average temperature below 65 F)
.2 "barely" 100% within 5 Tbars,
.5 50% at 5 Tbars, 50% at 6.
.8 "barely" 100% within 6 Tbars
1.0 100% "deep" within 6 Tbars (24-hr average temperature above 80 F)



As the last of the warm 6 Tbar weather draws to a close, we can see about 4 Gids of rebound, but I am still looking at 10% capacity loss for the season, which has been confirmed by several charges to 100%. My graph represents half of one "wiggle" on the research graph. Next May I expect to begin another downward decline. Hopefully it will be shallower, especially since I will have my garage A/C available for a full summer season.

tbleakne said:

I commend all the detailed work that has been reported on this thread, using solid analysis to apply corrections, but I still agree with this comment posted on pg 1. The curves Nissan gave to TickTock are not data, but rather their model. We should suspect that this model has been massaged to fit with the optimistic projections that Nissan is still standing behind. I am concerned that folks might put more faith in your results than is warranted to the extent that they are based upon their model.

Click to expand...

Actually, I believe that at least the Phoenix curve is their model, slightly refined from data they got from Carwings. Correct me if I am wrong, but Nissan stated that Phoenix Leafs are on track for 76% capacity remaining after 5 years at 7500 miles per year using the Carwings data they have since the Leaf started selling in Arizona. Of course, this was a real revelation to all of us, because it meant that 12500 miles a year was "high mileage" in Phoenix, and that the predicted EOL (70%) was more like 4.5 years if you drive 12500 miles a year, not 10 years.

One clue that Nissan's curves might not be based upon solid science is that they show no seasonal variation, contrary to what we know will happen from the Arrenhius factor.

Click to expand...

You have to remember that Nissan's curves were to show the overall trend, and are likely a rough summary of their findings. They most likely know about annual variation, but they are looking at the big picture.

A question about your model: the Wiki seems to imply that you are estimating temperature dependence for the calendar-loss component of battery degradation, but assuming no temperature dependence for the cycle component of battery degradation. I have not seen any claims or research that cycle loss is temperature-dependent. However, in the LEAF's case, on the one hand we have Nissan claiming that "excessive" mileage, now defined as over 7.5K mi/year, is a major component of loss in AZ, and on the other hand we have lots of folks in cool climates claiming negligible loss to date, irrespective of their mileage. My impression is that this group includes folks with mileage greater than perhaps 15K miles per year, comparable to folks in AZ with excessive loss.

This suggests that it would be hard to fit a model with no temperature dependence of its cycle component to both sets of data. Do you see this in your results? If not, it would seem the only other likely possibility is that the higher mileage folks in cool climates are not accurately reporting their loss.

tbleakne said:

A question about your model: the Wiki seems to imply that you are estimating temperature dependence for the calendar-loss component of battery degradation, but assuming no temperature dependence for the cycle component of battery degradation. I have not seen any claims or research that cycle loss is temperature-dependent. However, in the LEAF's case, on the one hand we have Nissan claiming that "excessive" mileage, now defined as over 7.5K mi/year, is a major component of loss in AZ, and on the other hand we have lots of folks in cool climates claiming negligible loss to date, irrespective of their mileage. My impression is that this group includes folks with mileage greater than perhaps 15K miles per year, comparable to folks in AZ with excessive loss.

Click to expand...

I really need to flesh out the description of the Battery Aging Model in the Wiki. I am probably going to give it a separate section under Battery Capacity Loss. In my model, cycling loss is also temperature dependent, using the same aging factor as calendar loss. Also, the aging factor that was calculated using the 10 degree C. increase causes doubling of rate of loss was adjusted to fit Nissan's model the graph from TickTock. The adjustment required the high aging factors like Phoenix to be scaled back (roughly 1.8 -> 1.5 for Phoenix on the scale we were using, although the model has the values adjusted to a slightly different baseline of 0.9 for "normal", so the actual value for Phoenix is 1.35 (still 50% greater than basline as the 1.5 factor). The adjustments were all made empirically so that the model reproduced the "known" values from TickTock's graph. All of that definitely needs to be made clear for those who don't want to read through long threads to figure it out.

I put the Battery Aging Model in its own section of the Battery Capacity Loss portion of the Wiki, and rewrote that section to describe the model in more detail:

http://www.mynissanleaf.com/wiki/index.php?title=Battery_Capacity_Loss#Battery_Aging_Model" onclick="window.open(this.href);return false;

As always, comments, criticisms, suggestions, etc. are welcome.

The table of battery aging factors in the Wiki has now been expanded to show predicted remaining battery capacity at 1 year, 2 years, 3 years, 5 years, 10 years and the number of years to end of life (70%). Table was created assuming 12500 miles per year and 4 miles per kwh efficiency, but does not include possible losses from solar loading (if you park in the sun).

http://www.mynissanleaf.com/wiki/index.php?title=Battery_Capacity_Loss#Battery_Aging_Model" onclick="window.open(this.href);return false;

Of course, if you want an individualized prediction for your annual mileage, driving efficiency and amount of time parked in the sun you can download the Battery Aging Model spreadsheet and enter your own numbers.

Stoaty said:

http://www.mynissanleaf.com/wiki/index.php?title=Battery_Capacity_Loss#Battery_Aging_Model" onclick="window.open(this.href);return false;

Click to expand...

This following quote from the wiki is missing an important piece of the puzzle; the heater or air conditioner use will lower the "useable battery capacity" available for range autonomy. 4 miles/kWh multiplied by 21 = 84 miles, however with the climate control on, the same 4 miles/kWh could equal 76 miles of range autonomy (76 / 4 = 19kWh), or any range between 76 and 84 miles, as suggested in the quoted Nissan bulletin.

"One deduction from this table is that Nissan expects a range of usable battery capacity of 19-21 kwh when the car is new. It would be surprising if manufacturing tolerances are that large, so it may be due to variations in the time between manufacture and when the buyer takes delivery, or more likely to give some leeway for some dealers who store unsold Leafs at 100% SOC in the hot sun. Another possibility is that up to 1 kwh may be due to pack imbalance. A fourth possible explanation for the range in the chart is variability in the economy meter due to instrumentation accuracies (i.e. Gids). "

TonyWilliams said:

Stoaty said:

http://www.mynissanleaf.com/wiki/index.php?title=Battery_Capacity_Loss#Battery_Aging_Model" onclick="window.open(this.href);return false;

Click to expand...

This following quote from the wiki is missing an important piece of the puzzle; the heater or air conditioner use will lower the "useable battery capacity" available for range autonomy. 4 miles/kWh multiplied by 21 = 84 miles, however with the climate control on, the same 4 miles/kWh could equal 76 miles of range autonomy (76 / 4 = 19kWh), or any range between 76 and 84 miles, as suggested in the quoted Nissan bulletin.

"One deduction from this table is that Nissan expects a range of usable battery capacity of 19-21 kwh when the car is new. It would be surprising if manufacturing tolerances are that large, so it may be due to variations in the time between manufacture and when the buyer takes delivery, or more likely to give some leeway for some dealers who store unsold Leafs at 100% SOC in the hot sun. Another possibility is that up to 1 kwh may be due to pack imbalance. A fourth possible explanation for the range in the chart is variability in the economy meter due to instrumentation accuracies (i.e. Gids). "

Click to expand...

Thanks Tony, I will add that to the Wiki later this weekend.

TaylorSF used my SOC meter for a bit to record his commute. i put it on a google doc

https://docs.google.com/spreadsheet/ccc?key=0ArmkVCszTek9dC16a1Q4MjYxNHdJOWJva2s3NGViSnc" onclick="window.open(this.href);return false;

this should give us a bit of help in long term range degradation. also HAVE to mention; he is still on his OEM tires still has plenty of tread left! :shock:

Interesting model... I really hope it is accurate on the low temperature environments. Living in Tromsø/Northern-Norway, with probably a similar temperature profile as Juneau in Alaska, according to the model I should see the Leaf battery aging more slowly than I have ever dreamed of. Good for Tromsø Leaf owners. For the time being there are at least 6 Leafs up here in Tromsø (probably more, since I only can tell them apart by the colors and they are all present plus at least one with the same color as mine ;-) ). Unfortunately I don't know anybody with a Gid-meter, so I really don't know in what kind of state the battery was last March when I bought the car new or in what kind of state it is now. Anyhow I haven't seen any range reduction yet, but I would hardly be able to identify reduction below 10 % anyway. There are a lot of Northern lights here in the winter... I hope they don't affect the battery chemistry negatively...

DaveinOlyWA said:

this should give us a bit of help in long term range degradation. also HAVE to mention; he is still on his OEM tires still has plenty of tread left! :shock:

Click to expand...

That's an interesting data point! Can you please tell us what tire pressure he uses? TIA!