Roadster Owner Based Study of Battery Pack Capacity Over Time

[zack]

What's interesting here is (assuming you did not charge after taking it out of "Storage Mode") that the car, even if it was fully charged at the start of StorageMode, still had 142+ (+30? depending on how you drove) Ideal Miles left after "the last few months" sitting in StorageMode. So ... do you have more info ?

I have no explanation.

[JRP3]

Not sure if you were joking or serious? Aside from the fact that this is hopelessly impractical, there are too many variables to control. Tires, air temp, humidity, battery temp, how it's driven, etc. You could improve the results by repeating the test enough times but for our purposes we want to know within 1 or 2% which would be pretty much impossible.

I was serious, though I realize the difficulty. I would suggest trying to replicate the conditions that Tesla used to get their range numbers. Level ground, moderate temperatures, proper tire pressure, etc. You probably are correct that you could not guarantee results within 1-2%.

[tomsax]

I don't care if the Tesla estimate of the battery capacity is off by a few percent, I just want to know if it's significantly wrong. I log data on every drive and charge segment so I can go back and compare comparable drives over time. I also capture my log files so the detail information from the car's logs adds to my records and my records give context to what's in the car's log.

I generally drive to Portland a few times a year and I have a pretty consistent routine: drive conservatively (60 mph) to Burgerville in Centralia, in case the charger isn't available. The charger has never failed me, so after 30 minutes of charging I have enough juice to drive the speed limit (70 mph) for the second half. I then do the reverse process on the way back. I'll be able to tell if the range is dropping more than what the car's SOC estimate indicates.

BTW, I starting doing my logging by taking notes on my iPhone, but that was a pain because it was difficult to extract data. So, I wrote an app for that. It lets you customize what data you want to enter for charging and drive records and stores everything in a SQLite3 database. You can then export the data either in a tab-delimited file ready to load into a spreadsheet or the raw database file. It's available on the app store, EV Logger.

[richkae]

I believe the log file has data for the volts and amps coming out of the pack every second as you drive.
We should be able to accumulate all the data for the drive and then you would have a bunch of things to put together:
1) battery % charge before the drive ( as calculated by the firmware with some algorithm we dont know )
2) total energy used during the drive
3) battery % charge after the drive ( see 1 )
This should give us another way to calculate the energy capacity of the pack - right?
If you started from 100% and drove to 0% that would be definitive, but short of that if you go for a really really long drive you can minimize the impact of the uncertainty in the numbers from step 1 and 3 in the calculation.

[tomsax]

I believe the battery pack is better modeled as storing amp-hours than kilowatt-hours. When you romp on it, the pack voltage sags, reducing the power you get out of those amps. That's one reason why it's difficult to estimate remaining capacity in energy units. It's even worse trying to estimate miles of range remaining. Not only does how you drive affect how efficiently you use the energy you pull out of the pack, it also affects the amount of energy that can be pulled from the pack.

If people are collecting their log files, we can model the battery pack different ways and see how its behavior changes over time. The log files record info about how you're driving (pedal position, speed) and even some environmental information (temperatures, elevation, etc.). With enough data, we can control many of those variables. So grab your log files every 50 hours of driving and either send them to Rich or keep them to run the battery profiler as we update it to extract more useful information.

[doug]

Yeah, the key is to find the raw value (cell Voltage?) they're converting from to get ideal range. Hopefully that's also logged. Assuming the current conversion formula is accurate, you could correlate it to the raw values and see how the range has changed over time independent of firmware changes.

[JRP3]

I would think they are counting amp hours in and out of the pack to calculate capacity and range.

[vfx]

.... So grab your log files every 50 hours of driving and either send them to Rich or keep them to run the battery profiler as we update it to extract more useful information.

Is this something the Tattler or OVMS can help with?

[doug]

Sorry, I see now that my previous post was a couple hours out of date. Even the mods can't keep up now.

[markwj]

Is this something the Tattler or OVMS can help with?

The charge and drive logs (something like Tom has in his iPhone App) is something we want to do for OVMS. Particularly the charge logs, so we can build up history of charge data and use that to estimate charge duration. The end goal being you being able to say "I leave at 8am - finish by then, please" and OVMS takes care of the scheduling.

But, memory is limited. We can only do this per charge or per drive. Not per second/minute.

Tattler has more memory, but harder to get the data out.