Roadster Owner Based Study of Battery Pack Capacity Over Time

[richkae]

Hey not that I own a Tesla yet, but working on it... I just want to give a shout-out to Richkae (as well I'm sure of the parallel internal efforts of Tesla), for doing this pioneering research. It will aid Tesla, and it will aid us all in future vehicles/battery packs to be totally brainless/transparent charging efforts. The entire community benefits from those who are sharing their logs, and Richkae's ongoing volunteer work.

thx!

Thanks Greg. Bumping this thread to remind people that we always need fresh data!

 

[Kevin Sharpe]

Bumping this thread to remind people that we always need fresh data!

I sent my logs (VIN 272) a couple of months ago but had no response... do you normally acknowledge data submissions?

 

[richkae]

I sent my logs (VIN 272) a couple of months ago but had no response... do you normally acknowledge data submissions?

No, I'm horrible. I just wait for enough new data to roll in an to make it worth making some new charts.
I'll acknowledge them from now on.

 

[Kevin Sharpe]

I'll acknowledge them from now on.

glad it's not going into a black hole :smile: have lots more logs to send you and will encourage the brits to support your study

 

[hcsharp]

...
If you send the raw files, then when improvements are made to the parser, I can re-run the files without asking you to resend the data.
...

I just want to confirm that you would prefer getting the raw files. Correct?

I'm also wondering if you even want my data. More than a year old but only about 12K miles. Range mode charge gets about 246 mi. So if you're looking to study battery degradation, I don't really have any.

I'd like to suggest the possibility that capacity loss may not be related to miles so much as average level of discharge for each charging cycle. It's likely that those who drive a lot will generally be at a lower SOC when they plug in every night. Perhaps you have enough data to compare somebody who, say, drives a lot but charges at work and at home so that they rarely reach a low SOC before charging, to somebody with the same miles but only charges once a day. Maybe you can plot average level of discharge against capacity loss per mile.

 

[richkae]

I just want to confirm that you would prefer getting the raw files. Correct?

I'm also wondering if you even want my data. More than a year old but only about 12K miles. Range mode charge gets about 246 mi. So if you're looking to study battery degradation, I don't really have any.

I'd like to suggest the possibility that capacity loss may not be related to miles so much as average level of discharge for each charging cycle. It's likely that those who drive a lot will generally be at a lower SOC when they plug in every night. Perhaps you have enough data to compare somebody who, say, drives a lot but charges at work and at home so that they rarely reach a low SOC before charging, to somebody with the same miles but only charges once a day. Maybe you can plot average level of discharge against capacity loss per mile.

Raw files are best.
In order to be able to compare things like depth of discharge you need more than just the last raw log file, you need a whole series of log files over the life of the car.
The raw files contain high detail records for a couple of months of data, and low detail records for many years. In order to guarantee that you have high detail for the entire life of the car you have to pull a log file every month or two.
I haven't even done that for my own car ( I've missed a few months ) but I'll try to make a chart.

 

[jory]

Raw files are best.
In order to be able to compare things like depth of discharge you need more than just the last raw log file, you need a whole series of log files over the life of the car.
The raw files contain high detail records for a couple of months of data, and low detail records for many years. In order to guarantee that you have high detail for the entire life of the car you have to pull a log file every month or two.
I haven't even done that for my own car ( I've missed a few months ) but I'll try to make a chart.

what do you mean by "raw files"?

i've been sending files sent through the VMSParser with the -b flag. do you mean that output or something else?

thanks.

 

[strider]

what do you mean by "raw files"?

i've been sending files sent through the VMSParser with the -b flag. do you mean that output or something else?

thanks.

If you're ok w/ sending GPS data on where you've charged, it would be most beneficial to send the "raw" (before it's run through the parser) file. That way if upgrades are made to the parser or other data is desired they can re-run your data automatically.

Personally, I'm only sending in my VMSParser output but I'm keeping all the raw files so if need be I can re-run w/ a new version of the parser or different options. So that's an option as well.

 

[richkae]

If you're ok w/ sending GPS data on where you've charged, it would be most beneficial to send the "raw" (before it's run through the parser) file. That way if upgrades are made to the parser or other data is desired they can re-run your data automatically.

Personally, I'm only sending in my VMSParser output but I'm keeping all the raw files so if need be I can re-run w/ a new version of the parser or different options. So that's an option as well.

The output from the VMSParser provides time your battery has spent at state of charge, charge rate and temperature. It also provides a summary of daily driving distances, the firmware history, and how long each 1000 miles of odometer took. Lastly it provides the battery self reported capacity history.

I was originally hoping to see if any of those factors affected battery capacity, but so far we just don't have enough data. The charge rate and temperature data only comes from the "short term" section of the log - and the log only keeps about 2-3 months of that data - so if you don't keep a complete history of log files ( by gathering and saving one every 2-3 months ) then that data isn't very meaningful.

If we collectively discover some factor that seems important, we can go back and add processing that data to the parser and re-run all the log files. If you stay interested in the results of this thread, then its best if you do that.

The really young cars have not been very interesting so far.
There should be a lot of 3 year old cars out there now - and over the next few months a lot more cars will be providing very interesting data.

 

[Doug_G]

I download my logs periodically, and I believe I have a complete record of my car's activity. If you think that might be helpful I can make the full set available.

 

[tomsax]

If you're ok w/ sending GPS data on where you've charged, it would be most beneficial to send the "raw" (before it's run through the parser) file. That way if upgrades are made to the parser or other data is desired they can re-run your data automatically.

Just to be clear, we believe the log files downloaded from the car contain the GPS coordinates of every location you've ever charged at but it's encrypted.

That data used to be readable, but that portion of the log file has been encrypted since mid 2010 (firmware version 3.5.17 for v1.5 Roadsters, 4.2.34 for v2.x Roadsters). As far as I know that encryption has not been broken, and it would be easy for Tesla to make that encryption unbreakable for all practical purposes. So the odds are we'll never be able to decode that information. Even if it become possible, and there was some interesting reason I can't think of to look at that information, I trust Rich wouldn't disclose it or even decode it without first securing permission for each owner.

So, if you'd like to help us all learn more about the factors that contribute to battery longevity, please collect log files regularly and send either the summary of charge info or your whole set of log files to Rich.

 

[jory]

The output from the VMSParser provides time your battery has spent at state of charge, charge rate and temperature. It also provides a summary of daily driving distances, the firmware history, and how long each 1000 miles of odometer took. Lastly it provides the battery self reported capacity history.

I was originally hoping to see if any of those factors affected battery capacity, but so far we just don't have enough data. The charge rate and temperature data only comes from the "short term" section of the log - and the log only keeps about 2-3 months of that data - so if you don't keep a complete history of log files ( by gathering and saving one every 2-3 months ) then that data isn't very meaningful.

If we collectively discover some factor that seems important, we can go back and add processing that data to the parser and re-run all the log files. If you stay interested in the results of this thread, then its best if you do that.

The really young cars have not been very interesting so far.
There should be a lot of 3 year old cars out there now - and over the next few months a lot more cars will be providing very interesting data.

i keep all the raw log files. please let me (us) know if it becomes desirable to rerun a (new/improved/different) parser on the original files for the battery study.

thanks!

 

[brianman]

I download my logs periodically, and I believe I have a complete record of my car's activity. If you think that might be helpful I can make the full set available.

Do we have an intuition (or confirmation) on whether we'll be able to pull the logs wirelessly at home for Model S? Sounds like it might be worthwhile to pull the logs weekly to my house's file server. If that can be done wirelessly rather than via USB key, it's more likely I'd do it (or rather have the machines do it).

 

[tomsax]

Do we have an intuition (or confirmation) on whether we'll be able to pull the logs wirelessly at home for Model S? Sounds like it might be worthwhile to pull the logs weekly to my house's file server. If that can be done wirelessly rather than via USB key, it's more likely I'd do it (or rather have the machines do it).

In May at the Fashion Island Tesla Store Model S/X event, Ted Merendino told me that owners would be able to get some sort of log info via WiFi at home, but it wasn't clear how detailed it would be. I asked about being able to download log files like we do on the Roadster and the answer to that seemed to be "no".

 

[ChadS]

Here is a chart of battery capacity vs miles driven:
View attachment 6797
You'll notice one data point that stands out as having a low capacity after only 11000 miles. That car sat for a very long time unused, and the data was submitted after it changed owners after that long period of disuse.
I believe that the battery needs to be cycled a few times to discover its true capacity, as the computer is being conservative.

I believe that this is evidence that there is not a lot of capacity loss due to age, it is primarily a function of miles driven. The data point that stands out in this chart is actually a car that has driven 51k+ miles in 1.5 years.

Thanks Rich; this is a great chart. My data is not in it because of an oddity in my log file that the parser doesn't like.

I'd be the lowest spot on the chart, though--by a wide margin after you toss out the anomaly that Rich mentions. I am at 85% capacity (166 miles in standard mode; haven't done range mode in a while, but it's barely over 200 miles) after less than 3 years and 35k miles. Tesla has done a recent bleed test and says everything seems fine...it is just a lot more degradation that I was hoping for. I can probably live with it if the degradation slows down (which all the charts say should happen)...but so far, it seems fairly linear. Too bad I have not kept good records.

I live in Seattle, and it's garaged, so it doesn't get too hot or cold. I've never raced it. I plug it in whenever it is at home. It has never sat for a long time at a low charge (or at a high charge--when I charge in range mode, I time it so I leave when it gets full). I have only gotten it in to the bottom 10% maybe 4 times (and just barely so, with one exception), and every time plugged in very shortly thereafter.

So I don't think it's anything I'm doing...I think it's just part of the variation of packs. But if anybody has any ideas for things to change, I'm all ears.

 

[bolosky]

It has never sat for a long time at a low charge (or at a high charge--when I charge in range mode, I time it so I leave when it gets full). I have only gotten it in to the bottom 10% maybe 4 times (and just barely so, with one exception), and every time plugged in very shortly thereafter.

It's possible that your battery has gotten badly out of balance. While it tries to balance at a full standard mode charge, I suspect that it doesn't work all that well. You should try range mode charging it and let it sit plugged in at full range mode overnight. (If you know how to get to the diagnostic screens, there's one under ESS that shows when it's balancing; leave it plugged in until it says it's not.)

I had my battery replaced with one that had been unused for a long time, and it was WAY out of balance. It took several days of sitting at full range mode charge to balance it, but it added quite a bit of mileage when it did (something like 15 miles or so in standard mode).

This might not be your problem, but if it is then it's easy to fix.

And you're not going to damage your battery by doing this once (or even a couple of times a year). You just don't want to range mode charge every day or anything.

 

[wiztecy]

>You should try range mode charging it and let it sit plugged in at full range mode overnight.
Talking to Tesla, unless your battery pack is really bad and out of balance (like ones sitting in storage) you get the same result letting your car sit in Std. Mode say for the weekend. Anytime the SOC is over 80% the battery will initiate its equalization process.

 

[ChadS]

Thanks for the idea bolosky; I tried it last night. But as wiztecy assumed, it didn't make a difference (at least, not more than a couple of miles). I got 210 miles after letting it sit in Range mode all night, just about 85%.

 

[hcsharp]

Chad I'm curious if your capacity loss has been linear over time. Did you lose capacity faster in the first year than in the most recent year?

edit: I re-read your post #135 and see that it has been mostly linear. Where did you see "all the charts" that suggest the degradation will slow down? I think that's true with laptops and other equipment with poor battery management, but I've never seen anything that suggests that will be the case with a Tesla pack.

 

[ChadS]

The one chart I remember for sure was a Nissan chart about expected degradation of the Leaf battery--this was a while back, before the Leaf was released. I remember thinking at the time that the curve looked exactly like a chart I'd previously seen of Tesla's battery, although I'm not positive where I saw the Tesla chart. It probably was on these forums somewhere, about 2 years ago. Not sure if it was official Tesla, or somebody's projection.

I've seen a third similar chart too--I am sure it was lithium, but not sure of more than that. Could have been non-auto.