Model S Battery Degradation data

[electracity]

Several people have taken the Model S battery packs apart. http://www.teslamotorsclub.com/showthread.php/34934-Pics-Info-Inside-the-battery-pack

Not a whole lot of conclusions in that thread. Tesla could be expanding the voltage envelope as the batteries age. They could be reducing the reserve capacity with age to add to range too. I appreciate that Tesla builds a good battery. It will be interesting what the range plot looks like with more data. Perhaps if a few of the high values were missing from the plot there would be a more believable slope.

 

[glhs272]

Not a whole lot of conclusions in that thread. Tesla could be expanding the voltage envelope as the batteries age. They could be reducing the reserve capacity with age to add to range too. I appreciate that Tesla builds a good battery. It will be interesting what the range plot looks like with more data. Perhaps if a few of the high values were missing from the plot there would be a more believable slope.

Well, the conclusions we can draw from that thread is that there are no "hidden cells" or modules not connected, to be connected at some later time. That would actually be very difficult. They would have to turn on extra cells at every brick (parallel group of cells) in order to have any effect. Or turn on another module entirely, which we can see that is clearly not the case. As far as expanding the voltage envelope, maybe, but I haven't seen it on the upper voltage limit. At the bottom? Well, honestly haven't really checked much on that. But if that were the case we should be able to find out via supercharger(or any L3 charging) data. They clearly have been monkeying with their firmware, so the actual guess-o-meter numbers (rated range) don't really mean much.

What is a more reliable measure of capacity is the 100% depth of charge. If in fact the upper and lower voltage numbers have remained the same, which I think they have, then this should be what we measure by. In my case, my car was one year old at this point. Notice 54kwh extracted from the pack.

Then about 1 year later and 28,188 miles later I tried to get as close to that as possible again. Again I pulled 54kwh from the pack.

Notice show about 3 rated miles less. That's a pretty accurate estimation of range degradation. Perhaps I could have less degradation because I haven't made any effort to balance the pack. But clearly I was able to pull 54.6 kwh out of the pack, with some left to spare.

Now with these pictures I can't prove that the bottom voltage numbers haven't been allowed to move down. But clearly I can still get about the same out of my pack when it was new. I can also say I have never seen the top line voltage allowed to go higher at the supercharger. So if Tesla is opening up the voltage range, they would have to do it at the bottom. Here are a couple of pictures proving that the upper limit of 352 volts hasn't changed. One at about 17,475 miles and another a year later at 45,664.

Note: this is real actual data and not speculation. If you have real actual data that shows otherwise, please show us that.




Edit: actually I meant to show this picture as a total pack capacity (54.9kwh)

Sorry for the big pictures.

 

[FlasherZ]

I just completed a 2,700 mile vacation trip to Florida with the family. It gave me a good chance to exercise the battery and see if there were any "hidden effects" of not taking the battery to lower SOC levels. Many times, I pulled into charging locations with 25-35 miles of range left, then charged all the way back up.

At the conclusion of the trip, I was a few hundred miles over 55,000 - my refurb pack was installed at 19,000ish. 90% charge is still 240 miles, and a fullly balanced trip charge goes to the original range of that pack when installed - 268 miles.

Truthfully, it makes me a bit nervous - I'm not sure if I took the car down to 10 miles if it would do what happened to another driver - immediately drop to zero and say "shutting down"... I just have this concern that I may not have the perfect battery pack and instead, degradation is going to show up elsewhere.

 

[nlc]

The graph is based on displayed range for a 100% charge. But here in France with a lot of tests with friends, we seen that the value displayed can be far of the reality.
The range displayed is based on the estimated capacity of the pack, and in some condition the car cannot precisely knows the real capacity of the pack. For example if you never charge charge to 100% and go close to 0%, the computer cannot knows exactly the energy stored in the pack, and update its internal data.

For example, I have my car since 16 month, 35,000mi , it's a "D" pack, and I am very careful in my use of the car to keep my battery in the better health as possible. On daily use, I never charge more than 70%, and never discharge below 30%. And when I need to do a full charge, I calculate the charge to finish just before I go.
And I found that when I only charge at 70% for a long time, my range decrease gradually. Some month ago a full charge gave me a typical range of 226mi (364Km) !!! But after vacation and some trips with long range with SoC close to 0% and charge to 100%, the computer estimation for a full charge increased to 232mi (374km). It's a 2.7% better range, and was just a computer deviation, not a 2.7% battery degradation.

A friend has the inverse case, he is taxi, and always charge its car to 100%, and the car sometime stays very long time at this 100% SoC. It's very bad for the battery, but its car display a typical range of 252mi (406km) !!!
We all had a doubt on the reality of this display, and we was right, 2 time the car completely stopped with 5mi range left on the display !!!

Compared to me, when I arrive to 0mi left, I have 3kW reserve available from the pack before the car completely stops.

Thus to compare the battery degradation in a precise way, it's not possible to use the displayed range. The only precise way is to regularly (for example 1 or 2 time a year, when temperature is close to 20°c) do a full charge, and drive without big acceleration and with a low consumption until the car completely stops. And look how many energy (kWh) you was able to get from the pack (also note the value when you arrive to 0mi, it's interesting to see the reserve energy).
And the best is to do this in a single trip, because if you stop the car the vampire losses will not be taken into account.
It's what we call in France the "Rolland" protocol (it's the name of my friend who exposed the protocol) :biggrin:

I do a complete discharge 3 week ago (not really complete, because I have not driven until the car stops, but when the power limitation was just under 60kW, but I was really close to the end), and I was able to output 69.9kWh from the battery when I reached 0mi left, and 3.1kW more until to be very close to the car stop warning).
Thus 73kW total. I will probably able to output a little bit in better condition, temperature was lot, and I have not driven in an economical way...)
And 69.9kWh before the charge now message should correspond to a full typical range of 235mi, and my car actually displays 230mi, thus 2.1% less than the reality, again it's not battery degradation it's just a computer error.

It can be very interesting to create a topic to follow the evolution of the battery capacity of all user who want to regularly do this measurement

 

[Bangor Bob]

The data fits an entirely plausible degradation curve, similar to what a single cell will show when used carefully. Some initial capacity loss in the first 30-50 cycles, then a very slow taper over thousands of cycles.

If the upper cell voltage limit is controlled (and it is), and high temperatures are avoided (they are), long-durations at high state-of-charge is avoided (by not using Range Charge all the time) and the electrolyte has a good mix of additives to prevent solid-phase electrolyte side-reaction products (unknown), then both calendar life and cycle life can be very long. The real world data seems to be supporting this.

The best way to measure this is very accurate coloumb counting. Coulombic efficiency = Coulombs out/coulombs in. 1.0 is a "perfect" battery. NCA batts can do .9992-ish under ideal circumstances. The .0008 is electrons lost to side-reactions inside the battery, AKA "degradation." At a guess, the current meters inside the car are not accurate enough to definitively determine the CE value for the pack to anything like that level of precision, but if you can measure kWh in, and kWh used over the life of the car, then changes in capacity should show up even despite the cycle-to-cycle variables of charging conversion losses, temperature control, etc.

 

[FlasherZ]

I agree with Bob.

In addition, what the meters inside the car actually measure still isn't known completely. We do know that the meters don't record kWh usage due to electronics draw, A/C & heat, etc. when the car is in park, neutral, or sitting at 0 mph; however, it does record consumption by the heater when the car is moving. It's not instrumented enough to provide reasonable conclusions.

I don't believe it's built into the Wh/mi (and therefore reflected in the rated miles reduction) only because over the lifetime of the car, my driving hasn't changed and my average Wh/mi is being reduced. Even if we could measure that, there are other variables now at play such as the inverter algorithm. Compared to last year, my Wh/mi is considerably down - where I could rarely achieve < 300 in the past on a 30 mile round trip, I can now come home with 250-270 Wh/mi.

Perhaps I'm wrong and they're using improvements in the inverter algorithms to hide battery degradation loss, but we'll never know that unless someone instruments their car with lab-grade gear or unless they perfect and stop improving on the inverter code - then we'll get enough time to see if there's a pattern.

 

[mknox]

I just completed a 2,700 mile vacation trip to Florida with the family. It gave me a good chance to exercise the battery and see if there were any "hidden effects" of not taking the battery to lower SOC levels. Many times, I pulled into charging locations with 25-35 miles of range left, then charged all the way back up.

At the conclusion of the trip, I was a few hundred miles over 55,000 - my refurb pack was installed at 19,000ish. 90% charge is still 240 miles, and a fullly balanced trip charge goes to the original range of that pack when installed - 268 miles.

Truthfully, it makes me a bit nervous - I'm not sure if I took the car down to 10 miles if it would do what happened to another driver - immediately drop to zero and say "shutting down"... I just have this concern that I may not have the perfect battery pack and instead, degradation is going to show up elsewhere.

And I just completed a trip from Toronto with my wife to visit our daughter and son-in-law in northern Illinois. Same deal with the Supercharger stops, but opposite experience with range.

My car has a hair over 51,000 miles and is on the original A-pack battery. Prior to the trip, I was getting 218 to 220 miles @90% and around 242 at 100%. After the trip, I now get 216 miles @90% and 240 miles @100%. I didn't gain anything and, in fact, slid backwards a bit. When the car was new, I got 245 miles on the "Daily" setting and exactly 265 miles on the "Trip" setting (or whatever those two settings were called).

My experiences compared with most others is making me kinda sad. Today, I flipped my settings over to "Energy" from "Miles" so that my dash just shows the % SOC. The lowering numbers I used to see there were depressing me.

- - - Updated - - -

Compared to last year, my Wh/mi is considerably down - where I could rarely achieve < 300 in the past on a 30 mile round trip, I can now come home with 250-270 Wh/mi.

Not here. My Wh/mi numbers have remained very consistent over the past 2+ years. I easily see around 260 after my daily 90 mile commute, and have since the car was new. EDIT: In summer. My winter numbers are obviously higher, but consistent year over year.

 

[Bangor Bob]

mknox, the battery guessometer factors-in recent driving history? And the majority of your trip to Illinois and back was likely on 400-series and interstate highways at speed and high Wh/mi consumption? There are good odds you'll get those two miles back as you resume your "normal" drive cycle for a few weeks. Your average of 260Wh/mi indicates a lower-speed commute. That said, your car does seem to be off the curve at 90% of new miles at 80,000km vs. 95% on the trend-line, or 93-ish% actual measurements at that mileage.

Certainly worth keeping an eye on, but you're a long way from qualifying for a warranty replacement. Not sure if that's reassuring or not...

(FWIW, the guess-o-meter on my Volt varies from 32 to 45 miles per charge based on recent driving, but a full charge always yields a "consumed energy" of 9.7kWh from the battery. If that number starts changing then I know I'm seeing degradation as GM tells us the upper and lower SOC charge/discharge limits don't change with time.)

 

[nlc]

After the trip, I now get 216 miles @90% and 240 miles @100%. I didn't gain anything and, in fact, slid backwards a bit. When the car was new, I got 245 miles on the "Daily" setting and exactly 265 miles on the "Trip" setting (or whatever those two settings were called).

IMO it's because you rarely go to very low SoC and the computer is not able to follow correctly the energy loss with degradation. When you do a deep cycle, the computer can have new data and adjust internal values, and the 100% typical range decrease. That's normal...
But I am impressed with the display you have, with less mileage my 100% typical range displayed is 230mi...

 

[mknox]

mknox, the battery guessometer factors-in recent driving history? And the majority of your trip to Illinois and back was likely on 400-series and interstate highways at speed and high Wh/mi consumption? There are good odds you'll get those two miles back as you resume your "normal" drive cycle for a few weeks. Your average of 260Wh/mi indicates a lower-speed commute. That said, your car does seem to be off the curve at 90% of new miles at 80,000km vs. 95% on the trend-line, or 93-ish% actual measurements at that mileage.

What's interesting here is that my driving history and the type of driving on this trip are very similar. I have a 90 mile round-trip commute that is almost all freeway. I go against the flow of traffic which means I am driving between 65 and 75 MPH, and usually with my Cruise Control on. That's pretty much how my trip to Illinois went, and I was getting almost exactly the same Wh/mi readings on my trip as I do in my daily driving. The only significant difference on my trip is that I was doing wider charge/discharge cycles.

- - - Updated - - -

IMO it's because you rarely go to very low SoC and the computer is not able to follow correctly the energy loss with degradation. When you do a deep cycle, the computer can have new data and adjust internal values, and the 100% typical range decrease. That's normal...

That is true. In my day-to-day, I will consume about 35% of my battery in the "nice" weather, and 50+ % in the winter. On my trip, I was typically going from 90 to 20% SOC. Others, though, have reported that a few low to hi SOC sessions will actually increase range numbers, not decrease them as I saw.

 

[Maarten ST]

new data on the blog post Tesla Model S battery degradation data | Steinbuch

 

[ZsoZso]

My car is about 14 months old, I had the first long-range trip about a year ago and at that time max charge has shown 420km rated range. We are going on a ski trip today, so I charged to 100% again (usually only charge to 80% daily), and I was shocked to see the rated range is only 394km now. I thought that was a very large drop (bit over 6%) for a year, but since I have driven ~37,000km I see that it matches somewhere around the average from the posted graph.

I hope it does not keep falling 6% every year...