Tuning the Battery Aging Model

[Stoaty]

I don't have your new spreadsheet installed, but the old one (0.83) is dead-on if I use the 9/28/2011 in-service date:
Total Loss = 9.64%
Calendar Loss = 8.29%
Cycling Loss - 1.35%

That gives an error of 1.03%. I agree that since the car had 2000 miles or so on it that is a reasonable in-service date to use.

However, the latest version does a better job:

Predicted Loss - 11.11%
Actual Loss - 10.67%

Actual Loss minus predicted Loss: -0.44%

 

[RegGuheert]

I don't have your new spreadsheet installed, but the old one (0.83) is dead-on if I use the 9/28/2011 in-service date:
Total Loss = 9.64%
Calendar Loss = 8.29%
Cycling Loss - 1.35%

That gives an error of 1.03%. I agree that since the car had 2000 miles or so on it that is a reasonable in-service date to use.

However, the latest version does a better job:

Predicted Loss - 11.11%
Actual Loss - 10.67%

Actual Loss minus predicted Loss: -0.44%

Awesome! Thanks! Good job on the new tuning!

One question: Why do you say actual loss is 10.67% while the app says 9.78%? Do you use a different Ah number for 100% than LEAFSpy?

 

[Stoaty]

Awesome! Thanks! Good job on the new tuning!

One question: Why do you say actual loss is 10.67% while the app says 9.78%? Do you use a different Ah number for 100% than LEAFSpy?

I am not sure what number LeafSpy is using, I have 66.25 AH, which I am pretty sure was the number used by Leaf Battery App before it became LeafSpy.

 

[Stoaty]

Here is version 0.99 of the Battery Aging Model:

https://dl.dropboxusercontent.com/u/48149991/Leaf%20Battery%20Degradation%20Model%20Version%20099.ods" onclick="window.open(this.href);return false;

Changes:

--Found and fixed bug on the calibration tab: EOL graph was showing the EOL figures from the Prediction tab
--On Prediction tab and Calibration tabs, added table for different EOL percentages (which are user settable if desired) and the predicted EOL years, Service Years remaining and miles at EOL
--Added "Qualify for Warranty" for both 65% and 70% capacity remaining EOL calculations in table (since close to 65% is most likely where the 4th bar will be lost)

PS Only a few more changes planned before version 1.0

 

[Stoaty]

Here is version 0.99a of the Battery Aging Model:

https://dl.dropboxusercontent.com/u/48149991/Leaf%20Battery%20Degradation%20Model%20Version%20099a.ods" onclick="window.open(this.href);return false;

Changes:

--No functional changes
--Changed formatting of Annual Table so that impossible values are displayed with white text on black rather than a black box
--Changed Annual Table to Display 5% remaining capacity if less than 5% so that negative values aren't displayed
--Moved EOL table on Prediction tab up to match position on Calibration tab

 

[JeremyW]

So when is the party for the big 1.0?

 

[firetrax]

Hi, great effort on this. Just a couple of quick queries (apologies if they have been answered elsewhere):

- Is there a significant drop in capacity for lower temperatures - the UK Leaftalk forums has reports of people seeing large drops since the summer with a 10C drop in ambient tempatures?

- What was the impact of the software update - did it lead to a drop in reported capacity?

Cheers

 

[Stoaty]

- What was the impact of the software update - did it lead to a drop in reported capacity?

Compared to version 0.83 (the one on the Wiki), the latest version more accurately predicts actual capacity loss. The previous version was too optimistic, suggesting longer times to End of Life than one is likely to see. In the next few weeks I will be editing the Wiki and putting in the more realistic projections. I believe it is unlikely that a Leaf will do significantly better than the current version of the model predicts, but could do worse if calendar aging proves to follow a linear trajectory rather than being proportional to the square root of time (the assumption the model uses).

 

[Stoaty]

So when is the party for the big 1.0?

It will have to wait a couple of weeks until I get back from my trip to Vermont to photograph Fall color.

 

[firetrax]

Thanks Stoaty. Sorry for not being clear but I was actually asking about the impact of the P3227 Leaf software upgrade on reported capacity numbers.

Could that be the reason some people have seen a large drop in reported capacity since the summer, or is it likely to be the cooler autumn temperatures in the UK?

- What was the impact of the software update - did it lead to a drop in reported capacity?

Compared to version 0.83 (the one on the Wiki), the latest version more accurately predicts actual capacity loss. The previous version was too optimistic, suggesting longer times to End of Life than one is likely to see. In the next few weeks I will be editing the Wiki and putting in the more realistic projections. I believe it is unlikely that a Leaf will do significantly better than the current version of the model predicts, but could do worse if calendar aging proves to follow a linear trajectory rather than being proportional to the square root of time (the assumption the model uses).

 

[Stoaty]

Thanks Stoaty. Sorry for not being clear but I was actually asking about the impact of the P3227 Leaf software upgrade on reported capacity numbers.

Could that be the reason some people have seen a large drop in reported capacity since the summer, or is it likely to be the cooler autumn temperatures in the UK?

We don't have enough info to say whether the P3227 update might be causing this. How cool are "cooler" autumn temperatures?

 

[firetrax]

They're about 14C now Vs mid/high 25-30C in the summer

Thanks Stoaty. Sorry for not being clear but I was actually asking about the impact of the P3227 Leaf software upgrade on reported capacity numbers.

Could that be the reason some people have seen a large drop in reported capacity since the summer, or is it likely to be the cooler autumn temperatures in the UK?

We don't have enough info to say whether the P3227 update might be causing this. How cool are "cooler" autumn temperatures?

 

[Stoaty]

They're about 14C now Vs mid/high 25-30C in the summer

I don't think that would be low enough (57 degrees F.) to cause much of a drop. Tony Williams is the expert here, somewhere on MNL he has estimated the effects of temperature.

 

[DaveinOlyWA]

I am seeing the best temps for recovery being in the upper 60s. Got a few readings in the mid 240s. But lately with batt temps in low 60s back to mid to upper 230s

 

[klapauzius]

I have seen my capacity going up lately...just slightly, but the precipitous drop from 65 ah (June - pre update) to 60 ah now has come to a stop it seems.
Has anyone corroborated the cars AH readings with some real world measurements?

 

[RegGuheert]

I have seen my capacity going up lately...just slightly, but the precipitous drop from 65 ah (June - pre update) to 60 ah now has come to a stop it seems.
Has anyone corroborated the cars AH readings with some real world measurements?

Was it hot in Seattle this summer? (It hasn't been hot here, but our car is now a bit below 60 Ah.)

 

[surfingslovak]

Has anyone corroborated the cars AH readings with some real world measurements?

The Ah values seem to align well with GIDs on a full charge, but I don't think that anyone has tried to correlate Ah to usable range, if that's what you were asking.

 

[Stoaty]

Has anyone corroborated the cars AH readings with some real world measurements?

The model also shows predicted Gids at 80% and 100% charge (based on the percent loss of capacity in AH). I can tell you that for my Leaf, the predicted Gids at those charge levels are very close to actual (generally within 1-2 Gids). Don't know about others.

 

[klapauzius]

Yes, it was hot. I think we had a couple days with temps >80 F.
Garage temperatures for the summer see below.

 

[klapauzius]

Has anyone corroborated the cars AH readings with some real world measurements?

The model also shows predicted Gids at 80% and 100% charge (based on the percent loss of capacity in AH). I can tell you that for my Leaf, the predicted Gids at those charge levels are very close to actual (generally within 1-2 Gids). Don't know about others.

Yes, but the GIDs live within the cars "brain". Likewise the predictions of your model are based on the internal states of the car. It is reassuring that your model represents a "real" world model, so in a way it can serve as a sanity check on the Leafs computations.

But since the Leaf-AHs are temperature dependent in exactly the opposite way a real world battery would behave, I am curious how real "AH" compare to "LEAF-AH" for a given temperature.