3.0 Battery Longevity

[pharma5]

Hoping these will level off before CAC 160 Ah from the original ESS's when they were new.

I may have missed it in another spot, but does the warranty related to 3.0 upgrade address capacity loss in any way?
(I believe original Roadster, S, and X did not, and Model 3 might have something?).

Thanks.

 

[dpeilow]

Can I please hit informative and dislike at the same time?

Also, when is someone from Tesla going to acknowledge this is an issue?

 

[bolosky]

Also, when is someone from Tesla going to acknowledge this is an issue?

If this is working the way I think it is, then they always knew it was an issue. It really looks to me like the car mostly can't tell what the real battery capacity is unless it's at very low or very high SOC, so it uses an algorithm based on mileage and time to guess at the capacity, and then adjusts it when you range charge or run the battery low.

If this is the case, then Tesla put the parameters into this algorithm, which means they pretty much knew how fast the battery would degrade. In fact, given that range charges seem to move the CAC up a lot, they probably thought it would be worse than it really is.

 

[bolosky]

Just as another illustration, I went to the logs for my car and did a comparison of the rate of degradation for the original and 3.0 batteries. What I did was to plot the CAC minus the starting CAC (156.7 for the original battery, 214.6 for the 3.0) vs. the mileage.

In the case of my car, I got a replacement original battery at just under 30K miles, so there are two lines. The replacement battery was grossly unbalanced when it was installed, so it took quite some time to get back to the new battery CAC, so I just ignored that portion of its lifetime and called 0 for it when it hit 156.7. You'll notice that it kept going up for a few thousand miles before it started to go down for real.

Anyway, the point of this is the dramatic difference in slope. It took about 22K miles for the 3.0 battery to have the same amount of loss that my (second) original battery did in 82K, and there's no apparent end in sight.

I'm starting to wonder whether my original Roadster will last long enough for the new Roadster to make it; I didn't pay the 250K to get a Founders' edition, so I probably need at least four more years. Projecting this slope out four more years gives me a CAC of 98Ah, which not in range mode is more-or-less a first edition Leaf.

 

[ecarfan]

Thanks as always for your extremely informative analysis.

It really looks to me like the car mostly can't tell what the real battery capacity is unless it's at very low or very high SOC, so it uses an algorithm based on mileage and time to guess at the capacity, and then adjusts it when you range charge or run the battery low.

How often do you range charge or run the battery to a low state?

I have not driven my car much in the past few months, so will wait a little while longer before sending you a new log file.

 

[Kerios]

Mine (EU 277), has just hit 201 following little use over the last 6 weeks. That's a 14 point drop roughly in six months, its defintely dropping faster than the original battery and on that basis I presume it has to be something innate to the battery chemistry or charging mechanism - given that 3.0s are showing similar trend from across the world, in a variety of different environments, temperatures, usages and charging regimes. As @dpeilow said, when is Tesla going to acknowledge that there is something up? I'm guessing not at all else they'd probably have put a hold on fitting out new 3.0 packs...

 

[dpeilow]

For me it's not a case of keeping the car going until the New Roadster comes out but about keeping the classic one going as long as reasonably possible. I'm not in a position to be spending $250k on any car but I am/was prepared to spend a reasonable amount of disposable income to maintain a piece of history. I didn't even order the 3.0 battery anyway, it turned up in the SC with my name on it and I decided after a few days that it may have been a case of now or never, so spent 40% of the car's value to get what I hoped would be all over a decade's more use. That is now looking increasingly like a poor, rushed decision.

 

[gregd]

Just as another illustration, I went to the logs for my car and did a comparison of the rate of degradation for the original and 3.0 batteries. What I did was to plot the CAC minus the starting CAC (156.7 for the original battery, 214.6 for the 3.0) vs. the mileage.

In the case of my car, I got a replacement original battery at just under 30K miles, so there are two lines. The replacement battery was grossly unbalanced when it was installed, so it took quite some time to get back to the new battery CAC, so I just ignored that portion of its lifetime and called 0 for it when it hit 156.7. You'll notice that it kept going up for a few thousand miles before it started to go down for real.

Anyway, the point of this is the dramatic difference in slope. It took about 22K miles for the 3.0 battery to have the same amount of loss that my (second) original battery did in 82K, and there's no apparent end in sight.

I'm starting to wonder whether my original Roadster will last long enough for the new Roadster to make it; I didn't pay the 250K to get a Founders' edition, so I probably need at least four more years. Projecting this slope out four more years gives me a CAC of 98Ah, which not in range mode is more-or-less a first edition Leaf.
View attachment 271721

I'm curious why your original battery only lasted 22k miles. It shows a similar slope to the 3.0 battery, if I squint hard enough. What happened to the original?

I have ~46k miles on my original battery, and currently see a CAC of about 139.5. It was about 142.5 when I bought the car 3 years ago, though per the graphs below, perhaps I should peg that a bit higher. Standard charge is down to about 163 miles ideal range now, vs 175. That's about 3 AH of CAC reduction over 18k miles, in 3 years, or a point CAC loss every 2,700 miles since birth. Lifetime, it seems to follow your replacement battery's slope, but more recently the decline seems to have accelerated, more like your original one. That's what has me a bit worried.

Last week, a 95 mile round trip to the airport (with a 1,000 ft elevation gain on the return) left me with about 50 miles ideal left, and that was without the heat or A/C running. Comfortable margin, but I wouldn't want to go a whole lot lower. 25 miles ideal is where my range anxiety starts kicking in. (I also get anxious when my cell phone gets down to about 20% too.) Family is some 150 to 220 miles away, so I was thinking about springing for the new battery later this year, but these graphs keep telling me not to do it. Like @dpeilow, I intend to keep the car "forever", vs springing for the new Roadster. There's something about owning and driving the car that started it all that has me hooked...

Last year, during the annual spring de-leafing ritual, I asked the service folks about the 3.0 battery longevity. They, of course, had heard nothing of the issue. I forwarded them a link to this thread, but never heard anything back. Will follow up again in a few months. Perhaps re-forwarding them the link at the time I make my appointment will at least give them notice that I (we) need an answer. Is the observed decline real, or just an accumulated estimation error that will sort itself out over time? That's not a gamble I want to risk $29k on, given that my original battery isn't in that bad a shape. Yet.

For the curious (myself included), here is my original battery's CAC history. I'm guessing the initial ramp was also partly due to the car not being driven or charged regularly in the several months prior to my purchase at the end of 2014, so things were a probably bit out of balance. But the decline is very noticeable after that. Also note that I retired mid-2016 (40.5 k miles at the time), so the miles driven dropped off significantly after that, from about 250/week to maybe 10/week now, with infrequent distance visits to family. @bolosky, if this data would be of any use, I can send you the spreadsheet.

 

[supersnoop]

I’m under the impression this is all a software issue. Shouldn’t the 3.0 last longer based on the cell chemistry? The CAC calculation just doesn’t seem to be accurate for the 3.0, likely due to the standard charge threshold that doesn’t go high enough for the battery to balance. I’d like to see someone with an OVMS program it to always charge to 85% and see how the CAC responds.

 

[ecarfan]

My guess is that little driving means little variation in SOC means there's little data to estimate CAC, so the system waits and then when it figures something out it has to make a big correction. I assume that the cells themselves are much more smooth, so the reported range is probably bigger than what you actually have.

Interesting hypothesis.

I’m under the impression this is all a software issue. Shouldn’t the 3.0 last longer based on the cell chemistry? The CAC calculation just doesn’t seem to be accurate for the 3.0, likely due to the standard charge threshold that doesn’t go high enough for the battery to balance.

My thoughts as well.

I am beginning to think that the Roadster firmware version that was installed when the 3.0 battery was installed is not accurately determining the CAC. But that is just a guess. The alternative is that the 3.0 battery is degrading faster than I’m sure any of us would prefer.

Before I got my 3.0 battery I was hoping that the revised firmware would allow me to set a specific charge level, or at least let me set it at 10% increments. But that is not what we got.

 

[gregd]

I’m under the impression this is all a software issue. Shouldn’t the 3.0 last longer based on the cell chemistry? The CAC calculation just doesn’t seem to be accurate for the 3.0, likely due to the standard charge threshold that doesn’t go high enough for the battery to balance. I’d like to see someone with an OVMS program it to always charge to 85% and see how the CAC responds.

After reading the various posts about the Model S 90 battery (where Tesla added Silicon to the anode), and the problems with it compared to the original S 85 cells, I've come to the conclusion that predicting long-term battery life is a bit of a crap shoot. We got lucky with the original LCO formulation, aided by Tesla's borderline-paranoid BMS algorithms, and as a result our batteries are generally holding up substantially better than predicted. The S90's, on the other hand, get charge-rate limited after some number of Supercharging sessions, due to concerns over degradation. That procedure is not being applied to the older S-85 cells.

I presume that those in the know about such things have checked their 3.0 battery balance levels, and not found anything amiss. So, if there is an algorithmic issue with the charge measurement, it's likely elsewhere. There have been experiments done with full capacity drain / charge cycles, and I believe the result was an only temporary bump in reported capacity. The degradation curve shifted out in time, not up in capacity. I.e., I don't recall the capacity shifts being such that the battery was effectively put on a different curve. I conclude, therefore, that the reduction is substantially real, and that the only remaining question is where it will level off at.

So the dice have been thrown again with the 3.0 battery, which notably use cells that have not been proven in other automotive roles. 191 was the hope. That hope has been put in jeopardy, if not dashed altogether. Where will it level off, or will it even do so? My checkbook is staying in my pocket until I get an answer, or am forced to do otherwise.

 

[bolosky]

Adding 3 months' worth of data for #33. Thanks, Steve Casner!

This is a car with lots of time but relatively little mileage. It's got about 8 weeks' more time since upgrade than mine, but about 19K fewer miles. For that, I'm 3Ah worse than him. This just reiterates what's been in the data before, which is that calendar life is a bigger factor than mileage. So, you might as well drive your car and enjoy it while it lasts. :-(

 

[bolosky]

I'm curious why your original battery only lasted 22k miles. It shows a similar slope to the 3.0 battery, if I squint hard enough. What happened to the original?

I took the car in for routine service and they said that the battery failed some test, and would I mind if they kept it for a while so they could replace it. The car was under warranty, so I was fine with it.

There's a big drop in the CAC of the first battery at about 18K miles. I think that this was about the time I took it to the track and ran it hard in performance mode. I always wondered if I damaged the battery by running it hard and hot. In any case, after doing that and seeing the effect I decided that I'd never do that again, and that I would only use performance mode for short sprints.

 

[sethr]

Where will it level off, or will it even do so? My checkbook is staying in my pocket until I get an answer, or am forced to do otherwise.

I am now glad that I'm slow to decide about major expenditures - I am now quite sure that I'll be keeping my money as well. We were originally promised 400 miles of range, that seems to have more than evaporated. Disappointing.

 

[ion_1]

Bolosky has 12 roadsters. Although this is good, we need more to have confidence in the trend. So if you have a 3.0, even if it is just installed, please consider sending data. Once we get to 20 or 25 the story will evolve even with such a small sample set.

The worse the data may seem, the more important it is for all (3.0 owners an pending) to see how this story evolves with a larger sample set.

If trend doesn't plateau by 175-180 (almost 20% loss), I will become very seriously concerned.

Many thanks to all those who have contributed..

 

[JRP3]

Although this is good, we need more to have confidence in the trend.

I'd guess 12 is a large enough sample to have confidence in the trend, considering that 100% of reporting cars are showing the same trend.

 

[ion_1]

I'd guess 12 is a large enough sample to have confidence in the trend, considering that 100% of reporting cars are showing the same trend.

Although the general trend is definitely degradation, there is a large variation in driving habits and storage times, so more are needed to have confidence. This is most critical when it comes time to determine when / if a plateau in cac develops.

 

[Young]

Adding 3 months' worth of data for #33. Thanks, Steve Casner!

This is a car with lots of time but relatively little mileage. It's got about 8 weeks' more time since upgrade than mine, but about 19K fewer miles. For that, I'm 3Ah worse than him. This just reiterates what's been in the data before, which is that calendar life is a bigger factor than mileage. So, you might as well drive your car and enjoy it while it lasts. :-(

View attachment 271825 View attachment 271824

Have you tried to fit the curves with exp(-(miles driven)/A) * exp(-(battery age)/B) * (initial capacity) ?

 

[ecarfan]

Although the general trend is definitely degradation, there is a large variation in driving habits and storage times, so more are needed to have confidence. This is most critical when it comes time to determine when / if a plateau in cac develops.

Agreed. I wish our data set was at least twice as large.

 

[Marius]

Just as another illustration, I went to the logs for my car and did a comparison of the rate of degradation for the original and 3.0 batteries. What I did was to plot the CAC minus the starting CAC (156.7 for the original battery, 214.6 for the 3.0) vs. the mileage.

In the case of my car, I got a replacement original battery at just under 30K miles, so there are two lines. The replacement battery was grossly unbalanced when it was installed, so it took quite some time to get back to the new battery CAC, so I just ignored that portion of its lifetime and called 0 for it when it hit 156.7. You'll notice that it kept going up for a few thousand miles before it started to go down for real.

Anyway, the point of this is the dramatic difference in slope. It took about 22K miles for the 3.0 battery to have the same amount of loss that my (second) original battery did in 82K, and there's no apparent end in sight.

I'm starting to wonder whether my original Roadster will last long enough for the new Roadster to make it; I didn't pay the 250K to get a Founders' edition, so I probably need at least four more years. Projecting this slope out four more years gives me a CAC of 98Ah, which not in range mode is more-or-less a first edition Leaf.
View attachment 271721

To make a fair comparison I think you should count with the percentage loss in CAC value rather than the absolute value.
This means a drop of 21,5 in CAC for the new pack is 10% which is the same as a drop of 16 for the older type pack.

But anyway the 3.0 pack degrades way to quick. One should expect newer type cells like Model S chemistry which in NL shows model S cars with 200.000 km with just 7% degradation and 3 years old. This should be the degradation curve for the 3.0.