Plug In America Tesla Roadster Survey

[shrink]

Hey new owners. If you have not yet filled out the survey for battery life I urge you to do so. Plug In America Tom has done a wonderful job here and one is able to pull the summary data into Excel. It looks like one should have 150 ideal miles in standandard charge after 135,000 from what I can gleam from the data. Of course with more data we will have a far more accurate projection. So please add your data.

I have a question about that regression line. I'm new here, but have experience with the LEAF and in the LEAF world, we were told that battery degradation is non-linear. Degradation would be expected to be fastest initially (I assume in the first year), but then level off in subsequent years, thereby flattening the slope to 80% after 5 years and 70% after 10 years. Now Nissan was way off on that estimate in warm climates, but that's not my point now.

I was wondering if you Roadster owners were also told to expect a nonlinear battery degradation rate? That could potentially make the estimated capacity over time greater, although now I realize the x-axis in that chart is mileage, not time. Hmm...I guess that leaves me back to my one question:

Were you all told that the battery degradation in the Roadster would be non-linear? The collected data points actually look pretty linear by mileage.

 

[djp]

Were you all told that the battery degradation in the Roadster would be non-linear? The collected data points actually look pretty linear by mileage.

The message from Tesla back in 2006 was to expect 30% degradation after 5 years, and I think everyone's thrilled the performance has been a lot better than that. I don't think Tesla has ever made any official statements about linear or non-linear degradation. The general expectation is the battery will slowly drop to 70% then go off a cliff, but we don't have many data points for that yet.

 

[markwj]

I have a question about that regression line. I'm new here, but have experience with the LEAF and in the LEAF world, we were told that battery degradation is non-linear. I was wondering if you Roadster owners were also told to expect a nonlinear battery degradation rate? That could potentially make the estimated capacity over time greater, although now I realize the x-axis in that chart is mileage, not time.

Perhaps people drive the car more when it is new? (Only half joking)

From my understanding, the primary degradation factor is charge-discharge cycles, and that is directly related to mileage. Sure, time has an affect (also directly on charge-discharge cycles via vampire drain), but for roadster at least that seems to be hidden in the overall miles factor. It would be interesting to see the battery health on -very- low mileage roadsters (four years old, but just a few hundred miles on the road) - there are a few of those I've heard about.

 

[djp]

It would be interesting to see the battery health on -very- low mileage roadsters (four years old, but just a few hundred miles on the road) - there are a few of those I've heard about.

I was thinking that the data is a bit biased since it excludes the garage queen Roadsters that have been sitting in storage since 2008. Those owners probably aren't active in the forums and no chance they've installed an OVMS. As a result we're mostly seeing degradation from mileage and we're missing all the data on calendar degradation. Tesla would have the complete set from their nightly downloads and a much better picture of the effects of heat, SOC and time on battery calendar life.

 

[strider]

I was thinking that the data is a bit biased since it excludes the garage queen Roadsters that have been sitting in storage since 2008. Those owners probably aren't active in the forums and no chance they've installed an OVMS. As a result we're mostly seeing degradation from mileage and we're missing all the data on calendar degradation. Tesla would have the complete set from their nightly downloads and a much better picture of the effects of heat, SOC and time on battery calendar life.

From a nerd standpoint it would be interesting but I feel it's safe to assume that calendar degradation is less than cycle degradation. So a properly stored garage queen would see less degradation than us who drive our cars a lot.

 

[wiztecy]

From a nerd standpoint it would be interesting but I feel it's safe to assume that calendar degradation is less than cycle degradation. So a properly stored garage queen would see less degradation than us who drive our cars a lot.

In essences the battery pack replacements are garage queens... they're (they were previously used packs, now called refurbished) unhooked and sitting in a warehouse at a low SOC (I hope 50% and I hope in a refrigerator). Somehow I don't think that's being done.... One thing these packs don't see is the vampire drain from pumps, vms, etc..

 

[shrink]

I feel it's safe to assume that calendar degradation is less than cycle degradation. So a properly stored garage queen would see less degradation than us who drive our cars a lot.

Yes, indeed, that makes a lot of sense. I was actually poking around and found a Plugincars.com article that referenced the Plug-in America Survey and noticed this quote:

It's not all good. German owner Hansjörg von Gemmingen reported in 2012 that he had achieved a record 240,000 kilometers (149,000 miles) on his Roadster. He was a poster boy for effortless long-distance travel in a Tesla. But Die Welt said that, unfortunately, his car "is almost exclusively in the garage" because he now gets only 100 kilometers (62 miles) on a charge.

The above article then links to this one about a recent Li-ion EV study and presentation:

Understanding the life of lithium ion batteries in electric vehicles

I''m sure this group is already well aware, but that article essentially says extended exposure to temperatures over 86F or 30C causes the most degradation, followed by too high or too low SOC, and fast charging. I'm anxiously awaiting my CPO Roadster so I don't know the car that well yet, but I imagine the TMS addresses #1, the default standard charge to 80% and storage mode help with #2, and #3 shouldn't be an issue since the Roadster can't Supercharge.

Am I understanding this correctly?

It's probably too late now, but do you think adding questions to the Plugin American Roadster survey about charging habits (e.g., how many times or how often do you charge in range mode?) be of any use? Regarding temperature, it doesn't seem like climate had an effect in the initial analysis, but there were only 4 cars from Arizona (3 of which had mileage around 40K) and 2 from Texas (about 6000 miles and 14,000 miles on those cars) - so only 6 out of 132. Does the OVMS provide any sort of battery temperature log that might be useful?

 

[markwj]

Regarding temperature, it doesn't seem like climate had an effect in the initial analysis, but there were only 4 cars from Arizona (3 of which had mileage around 40K) and 2 from Texas (about 6000 miles and 14,000 miles on those cars) - so only 6 out of 132. Does the OVMS provide any sort of battery temperature log that might be useful?

A few out of USA also...

OVMS provides detailed temperature logs. Battery, PEM, Motor and Ambiant.

 

[wiztecy]

There's some threads here on Hansjörg von Gemmingen's Roadster and theories / facts. The low 62 miles range now after reaching a milestone of 149,000 miles was mostly attributed to doing Range mode charges after he went down the degradation slope. Meaning that his standard mode charges didn't allow him to do his commute or regular usage of the Roadster which he was originally accustomed to. And its a fact that repeated range mode charges, especially on a pack with high miles, will shorten its life rather quickly.

 

[brianman]

From a nerd standpoint it would be interesting but I feel it's safe to (A) assume that calendar degradation is less than cycle degradation. (C) So a properly stored garage queen would see less degradation than us who drive our cars a lot.

The conclusion (C) doesn't require the assumption (A). The assumption is only necessary if you expand the conclusion to indicate that the garage queen is older than the driven vehicle.

 

[djp]

So a properly stored garage queen would see less degradation than us who drive our cars a lot.

True, but I'm curious how big the difference is between a properly stored Roadster (Storage mode in a cool garage) and an improperly stored Roadster (Standard mode in the heat).
Knowing that can put some bounds on the benefits of babying the battery, like using cool-down mode and limiting to a lower SOC.

We don't have enough data on older stored Roadsters to see that effect in Tom's study. Tesla has nightly logs from all Roadsters so should be able see real world calendar degradation rates.

 

[strider]

That makes sense. I was thinking more from a worst-case degradation standpoint that garage queens in storage mode wouldn't move the needle compared to high-mileage cars.

 

[dhrivnak]

The other issue I see is relatively few garage queens would ever pull the logs and send them in. It seems that it is the drivers who are actively learning the capabilities of the car. As for the slope on the initial graph I posted, yes it is a simple linear regression and it appears to fit. But we will not know for another few years if we continue down a slow gradual slope or we approach a cliff where cells fail in unison after x years and/or y miles. So if you have not added your data please do so so we can answer those questions.

 

[markwj]

We don't have enough data on older stored Roadsters to see that effect in Tom's study. Tesla has nightly logs from all Roadsters so should be able see real world calendar degradation rates.

I've heard that they have data for 2.x roadsters (with gsm enabled), but not 1.5s.

They do get logs for all cars during service visits.

Whichever, they have the best data, and the best knowledge of what that data means, but I haven't seen them release any results.

 

[Doug_G]

Whichever, they have the best data, and the best knowledge of what that data means, but I haven't seen them release any results.

Yes, they do... but it's valuable intellectual property and a really doubt they'll release it.

 

[tomsax]

I'll be speaking at the Plug-In 2013 Conference & Exposition on two panels:

Why Everyone Should Own an Electric Vehicle
An Expert Seminar on Public Education and Outreach
Sept. 30, 2:30-5:30 p.m.

Battery Technology Marches On
Assessing the current and future performance and costs of Lithium Ion and other advances in battery technologies
Oct. 2, 1:30 p.m.-3 p.m.

If you're interested in attending, you can save $100 by registering by Friday, Aug 30th.

 

[tomsax]

My latest article for Charged EVs magazine, about the Roadster battery study, is now available on the web:

Charged EVs | Plug In America research: Tesla Roadster battery longevity exceeds projections

It has an updated chart showing one Roadster with over 100,000 miles on the original battery. The article has more detail than the press release but is a bit less technical than the full study paper, aimed at a more general readership.

 

[hcsharp]

Tom,

Thanks so much for producing this study with PIA. For those of you who feel the same gratitude that I do, don't hesitate to make a generous contribution to Plug In America.

Tom I've noticed that my CAC changes in significant steps, often after I make a long trip (but not always). Last fall it dropped 4 (2.5% of original capacity) after one trip. I can't help but wonder how many data points are just about to correct, or just did, or perhaps are never driven on long trips. For example when I submitted my data to the study, it was right before the above mentioned long trip. I've been too lazy to update it... How do you think this is affecting the data? Some have suggested that it over-corrects and often creeps back up, and it's impossible to account for. Maybe it doesn't matter enough to consider.

 

[djp]

Some have suggested that it over-corrects and often creeps back up, and it's impossible to account for. Maybe it doesn't matter enough to consider.

I've seen the same type of noise in my CAC. After a long trip my Standard range dropped from 304km to 294km and slowly crept back to 301km (with the same changes in the CAC itself). I'd chalk it up to a statistical error term rather than a bias in the results.

 

[tomsax]

The discussion on this thread inspired me to look at how frequency of range mode charging correlates with battery pack capacity.

Looking at the data from the survey (which agrees pretty well with the OVMS data), I have a function that predicts standard mode capacity based on miles driven. For each vehicle in the survey that has at least 20,000 miles on the odometer and has not had a battery pack swap, I calculated how much above or below the expected standard mode capacity they are, then grouped those by how often they report range mode charging.

Points above zero have higher than expected standard mode capacity, points below zero are below average.

This chart seems to be saying that vehicles that are charged in range mode frequently are more likely to report a higher capacity.

There are two caveats to this conclusion: 1) there aren't a lot of data points for frequent range mode charging, and 2) in my experience, the car underreports battery capacity when it's been a while since the last low-to-full range mode charge, so the low values may not reflect actual battery pack capacity.