Tuning the Battery Aging Model

[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.

Probably not for now. In the future I may add those, but generally want reports on Leafs that are at least a year old, with a current focus on 2011-12 models. These would be in a separate section.

 

[uwskier20]

Manufactured - 03/11
Delivered - 5/18/11
Date of P3227 update - 7/11/13
Location – Snohomish, WA
Miles/kwh – 4.1
Odometer – 42,300
Capacity – 55.96
Date - 9/20/13
Parked in sun - maybe 1 day/week, but garaged in a west-facing garage

 

[ecoobsessive]

Manufactured - 03/2011
Delivered - 5/20/11
P3227 updte - not done
Location - San Antonio, TX
Miles/kwh - 4.7
Odometer - 24,543
Capacity - 49.35
Date - 9/20/13
Parked in sun - 5 days/week (but starting in 2012 I kept car in garage and took the ICE when temps were above 96F)

Health as per LEAF app: 69.27; currently 10 capacity bars - lost bar 11 in June 2013.

 

[Stoaty]

Manufactured - 03/2011
Delivered - 5/20/11
P3227 update - not done

Just remember, those that haven't had the P3227 update done at least 3 weeks prior to the date of the reading won't have their data included.

 

[Stoaty]

Here is version 0.97 of the Battery Aging Model:

https://dl.dropboxusercontent.com/u/48149991/Leaf%20Battery%20Degradation%20Model%20Version%20097.ods" onclick="window.open(this.href);return false;

Changes:

--Changing the Calendar Loss assumption or the Cycling Loss assumption on the Degradation Model tab now correctly changes the Predicted Loss on the Calibration tab. My method is not very elegant, but the brute strength method proved to be the easiest to implement.
--Values temporarily changed on the Model Degradation tab to calibrate the model:
Calendar Loss changed from 6.5 to 6.9
Cycling Loss remains at 2 (which was a change from 1.5 in an earlier release)

With these assumptions and 21 readings currently the model now appears to be fairly well calibrated:

Actual Loss Minus Predicted Loss:

Mean - 0.14%
Standard Deviation - 1.51%

Actual Loss Divided by Predicted Loss:

Mean - 100.08%
Standard Deviation - 10.41%

See attached graph showing some expected scatter, but with Phoenix and Dallas within one standard deviation.

 

[madbrain]

Stoaty,

Here is version 0.97 of the Battery Aging Model:

https://dl.dropboxusercontent.com/u/48149991/Leaf%20Battery%20Degradation%20Model%20Version%20097.ods" onclick="window.open(this.href);return false;

There seems to be a problem in the spreadsheet if the date of reading is less than one year after the date of delivery.
A lot fields go to #NA . I am not sure which one is the first to do so, you know the formula better.

Also, I only have Leafspy lite, which doesn't report Gids. How do I fill out the "actual % Gids" for Full charge and 80% charge ?
Should I use the SOC% after each charge ?

 

[Stoaty]

There seems to be a problem in the spreadsheet if the date of reading is less than one year after the date of delivery.
A lot fields go to #NA . I am not sure which one is the first to do so, you know the formula better.

I think I fixed that problem on the Calibration tab, but forgot to fix it on the Prediction tab. It will be fixed in the next release unless I can't remember how I fixed it.

Also, I only have Leafspy lite, which doesn't report Gids. How do I fill out the "actual % Gids" for Full charge and 80% charge ?
Should I use the SOC% after each charge ?

You can't fill those in if you don't have a Gid meter or equivalent (e.g., Leafspy or Pro). However, it doesn't matter much because only a couple of fields to the right of those will be affected. The rest of the fields will calculate correctly.

PS Don't forget that you have to enter a field twice if you want the Years to EOL and miles at EOF predictions to be correct. I plan to fix that in the future with an approximation which is very close (e.g., gives 5.67 years instead of the correct 5.60 years, a very tiny error), since we would be lucky if the prediction came out within 0.25 years of the actual EOL.

 

[Stoaty]

Here is version 0.97a of the Battery Aging Model:

https://dl.dropboxusercontent.com/u/48149991/Leaf%20Battery%20Degradation%20Model%20Version%20097a.ods" onclick="window.open(this.href);return false;

Changes:

Prediction tab - fixed bug that would leave many fields as N/A if Service Life was less than one year (a one letter change in a formula )

I investigated automating the years to EOL on the calibration tab for the entered readings, but it turns out to be pretty complicated, so I am going to leave it. You can get an accurate value on the Prediction tab, or by entering the row number of the reading on the Calibration tab, it is only the EOL column in the table of readings on the prediction tab that are entered manually (hard coded). They won't change appropriately (and thus the EOL miles will be wrong also) if you change the parameters on the Degradation model tab.

 

[Stoaty]

Here is version 0.97b of the Battery Aging Model:

https://dl.dropboxusercontent.com/u/48149991/Leaf%20Battery%20Degradation%20Model%20Version%20097b.ods" onclick="window.open(this.href);return false;

Changes:

--moved Validation tab to just before documentation, as it is mostly of historical interest at this point
--added to Calibration tab and Prediction tab the following entries:

Qualify for warranty? (based on prediction)
Remaining Service Years (based on prediction)

Notes: End of Life years refers to the number of expected years in service, not the number of years from date of battery manufacture

PS We could use some more entries for the calibration tab that have had the P3227 update at least 3 weeks ago.

 

[Stoaty]

Here is version 0.97c of the Battery Aging Model:

https://dl.dropboxusercontent.com/u/48149991/Leaf%20Battery%20Degradation%20Model%20Version%20097c.ods" onclick="window.open(this.href);return false;

Changes:

--Organizational and cosmetic changes only, no change in functionality
--Moved Calendar Loss and Cycling Loss assumptions from Degradation Model tab to the Calibration tab to make it easier to see the effect of changes in these parameters
--Moved Aging Factor and Solar Loading factors from the Degradation Model tab to Prediction and Calibration tabs (2 different factors for each tab). This makes it easier to see what the factors are, and if desired to change them by entering a number manually (but if you do this, don't save the spreadsheet since manual entry will overwrite the formula).
--Reorganized input (yellow) fields on Calibration and Prediction tabs so they are grouped together more closely and deleted some duplicated fields

 

[vrwl]

--Date of Manufacture 8/11
--Date of Delivery 10/?/11
--Date of P3227 update, if needed 7/2/13
--Geographic location - McKinney, TX 75071
--Average miles/kwh for the life of your Leaf 4.5/kWh
--Current odometer reading 7772
--Current AH capacity reading from LeafDD or Leaf Battery App 56.36
Exact date you took current odometer reading and current AH capacity 9/19/13
Days per week parked in the sun (fractional days OK) 0

One note: My car was only driven 89 miles between October 2011 when it was first placed into service and June 26, 2012 when I purchased it. So I don't know if that will affect your calculations or not.

Based on your need for anniversary readings, I went back to my logs and grabbed this data from 8/17/2013 (as close to the 15th as I could get):

--Date of Manufacture 8/11
--Date of Delivery 10/?/11
--Date of P3227 update, if needed 7/2/13
--Geographic location - McKinney, TX 75071
--Average miles/kwh for the life of your Leaf 4.5/kWh
--Current odometer reading 7505
--Current AH capacity reading from LeafDD or Leaf Battery App 57.57
Exact date you took current odometer reading and current AH capacity 8/17/13
Days per week parked in the sun (fractional days OK) 0

 

[Stoaty]

One note: My car was only driven 89 miles between October 2011 when it was first placed into service and June 26, 2012 when I purchased it. So I don't know if that will affect your calculations or not.

I put the date of service as when you bought the car, since 89 miles in many months is not really "in service". Do you happen to know if the car was sitting on a dealers lot in the sun (or somewhere else in the sun) before you bought it, and if so how many days a week?

Predicted Loss - 14.06%

Actual Loss - 14.93%

 

[vrwl]

One note: My car was only driven 89 miles between October 2011 when it was first placed into service and June 26, 2012 when I purchased it. So I don't know if that will affect your calculations or not.

I put the date of service as when you bought the car, since 89 miles in many months is not really "in service". Do you happen to know if the car was sitting on a dealers lot in the sun (or somewhere else in the sun) before you bought it, and if so how many days a week?

Predicted Loss - 14.06%

Actual Loss - 14.93%

It was originally purchased by Hertz as a rental car, so yes, I would assume it spent the full 8 months out in the sun on a lot somewhere here in the DFW area.

 

[Stoaty]

It was originally purchased by Hertz as a rental car, so yes, I would assume it spent the full 8 months out in the sun on a lot somewhere here in the DFW area.

Assuming the Leaf spent 8 months of its life out of 24 in full sun, that would be an average of about 2 days a week in the sun spread out over its lifetime. That brings your Predicted Loss to 14.76, which is just 0.17% less than Actual Loss.

 

[vrwl]

It was originally purchased by Hertz as a rental car, so yes, I would assume it spent the full 8 months out in the sun on a lot somewhere here in the DFW area.

Assuming the Leaf spent 8 months of its life out of 24 in full sun, that would be an average of about 2 days a week in the sun spread out over its lifetime. That brings your Predicted Loss to 14.76, which is just 0.17% less than Actual Loss.

Ahhh, good accuracy then! Glad I could help. I keep daily driving and charging logs, so I'm happy to provide additional data whenever you need it.

 

[Stoaty]

Stoaty needs help with an annoying problem in the Battery Aging Model spreadsheet from someone who knows OpenOffice (or other spreadsheets in general). I have not been able to get around the problem that End of Life calculations (EOL years, EOL miles and remaining service life) do not update properly after making a change in one of the parameters. Requires a workaround to either press Control-Shift-F9 (recalculate all formulas in spreadsheet) or enter the changed parameter a second time (with exact same value). Since Autocalculate is on, formulas should be recalculated correctly without this workaround. Any experts out there with ideas about what might be causing this problem or how to work around it without the above kludges?

 

[Stoaty]

I have removed the Documentation tab from the Battery Aging Model in the next release and have started on documentation which is a bit more detailed. Initial draft here (not completed):

https://dl.dropboxusercontent.com/u/48149991/Battery%20Aging%20Model%20Documentation.pdf" onclick="window.open(this.href);return false;

I still have to do the Calibration tab documentation.

Since I have been working with the spreadsheet so long it is fairly evident to me what does what, but thought it might not be so for others. Please post any comments, requests for clarification etc. to this thread.

 

[thimel]

Stoaty needs help with an annoying problem in the Battery Aging Model spreadsheet from someone who knows OpenOffice (or other spreadsheets in general). I have not been able to get around the problem that End of Life calculations (EOL years, EOL miles and remaining service life) do not update properly after making a change in one of the parameters. Requires a workaround to either press Control-Shift-F9 (recalculate all formulas in spreadsheet) or enter the changed parameter a second time (with exact same value). Since Autocalculate is on, formulas should be recalculated correctly without this workaround. Any experts out there with ideas about what might be causing this problem or how to work around it without the above kludges?

I found a rather kludgey solution. On the prediction sheet, take the formula in cell Q7 and copy the text of the formula by highlighting it in the formula bar. Hit escape. Then select cell Q3, click in the formula bar and paste the formula. This should end up with the exact same formula in Q3 that was in Q7. (it should still reference cell F13 for example.) Now the new formula will not update properly but the one in Q7 will work. If you can't get this to work, let me know and I will make the modification and email you the spreadsheet.

That was NOT the first thing I tried. It looks like a bug in OpenOffice to me, but I could not find any reference to such a bug via Google. Below is more detail on what I did for anyone still interested.

I opened Stoaty's spreadsheet in OpenOffice and confirmed the problem occurs as he explained. I then had OpenOffice save it in Excel format and opened it with Excel. It worked properly there. So, one solution was to use Excel.

I went back to OpenOffice and studied the problem. The root cause is that cell Q7 (End of Life Years) on the prediction tab does not update. It uses VLOOKUP with a reference to a table on the end of life tab. I checked that that table updates like it should. It does. It acts like the VLOOKUP in cell Q7 does not know it changed.

I then tried a number of fixes. Short story: none of them worked until I got to the one above.
Long story: I changed the table by hand. Cell Q7 updated properly. I modified cell Q7, adding references to user input cells or to cells in the end of life table. I then checked the response to changes in the input cell. Q7 would update properly with regard to the extra arithmetic I added, but still did not properly pick up the new value for VLOOKUP.
I then tried just copying the formula as described in my first paragraph giving the fix. I saved it to the right of the table on the end of life tab (updating the reference to cell F13 to come from the prediction tab). It worked properly! I then tried it in cell Q10 on the prediction sheet. It still updated properly! I tried deleting the cells Q7-Q9 so my new formula would be in the right place. Broken again! I finally realized that it must be the first one that doesn't get updated and put the formula in cell Q3.

 

[Stoaty]

I found a rather kludgey solution. On the prediction sheet, take the formula in cell Q7 and copy the text of the formula by highlighting it in the formula bar. Hit escape. Then select cell Q3, click in the formula bar and paste the formula. This should end up with the exact same formula in Q3 that was in Q7. (it should still reference cell F13 for example.) Now the new formula will not update properly but the one in Q7 will work.

Amazing! Got it. Fix works fine here, I added an extra touch: formatting Q3 using white text on a white background (so you can't see this cell with an out of place, wrong result). I will include this in the next release. Thanks!!!

PS I tried various things similar to this, but couldn't get it to work. Glad you found a solution.

 

[Stoaty]

Here is version 0.98 of the Battery Aging Model:

https://dl.dropboxusercontent.com/u/48149991/Leaf%20Battery%20Degradation%20Model%20Version%20098.ods" onclick="window.open(this.href);return false;

Changes:

--A bit more rearrangement of the entry fields on the Prediction and Calibration tabs to make it easier to follow
--Removed Documentation tab; there is now a separate PDF file (still being worked on) with documentation
--Worked around OpenOffice bug that caused a failure of Years to End of Life to update properly (thanks, thimel). You no longer need to press Control-Shift-F9 to make it show the correct value.

Newly identified bug:

If you make a change to one of the parameters on the Prediction tab and then press Control-Z to compare the before and after numbers, the EOL years do not update properly. Solutions?