engineering question: what are the main variables involved in battery degradation?

[ljwobker]

Anyone have a serious, engineering-level writeup of what variables impact battery degradation? I'm looking for something with real numbers and physics, not marketing-esque general statements like "leaving it charged at 100% hurts battery life".

I fully realize that there are a number of variables, but I also fully realize that at some level we're dealing with chemistry and physics, and the degradation of the battery almost can't be a function of more than 3 or 4 dominant variables. Based on the (very limited) good information I've been able to find, the major things appear to be:
1) charge cycles : essentially how many electrons are pushed through the battery chemistry over time
2) temperature : but not at all clear how much of this is operational temps vs. storage temps
3) time spent in some given state of charge : we've been told that "100% for long time is bad"... right... but HOW bad? Does this badness completely go away at 95% charge? 80% charge? What does this curve look like? Is the degradation linear with state of charge? Linear with time spent at that state? Somewhere there has to be a function/curve for this, I'm hoping someone out there REALLY knows this stuff and can provide pointers. ;-)

 

[Jason S]

Vague question will get vague answer.

What is your specific question?

...

What I'm getting at is that you've just ranted about not being able to get an answer to a set of questions like: "The manual says it is bad if I run the car without oil. How bad? For how long? What if there is a little oil? Are there graphs?"

Maybe you should ask something that actually can be answered.

 

[bonnie]

I sent a pm to the OP suggesting exploring some of the threads in this subforum. There are a lot of people who have pulled "engineering-level" information together. We are a technical bunch, after all. I suspect the answers are already here, but the OP had posted originally in the Model S thread (I moved this) and may not have known the wealth of info available here about the batteries.

 

[Jason S]

Empirical evidence based on other cars with similar battery packs is some of the information. Business case evidence given the warranty and price of replacement packs is more information.

But hard data? No where close to available.

I agree with the OP that 3-4 variables should get a good estimate. Those variables all revolve around activities that damage the structure of the battery, but the things that Tesla in particular has done to limit the damage pulls many of the values for those variables out of our grasp and we're back to guessing.

For instance, I've read that Roadster battery pack with ~12K / year charging seems to have only 1-2% capacity loss per year. That gives a benchmark for Model S, but I think the Model S pack is more advanced than the Roadster and therefore would do better. But we won't know for a few years.

 

[TEG]

The main answer is "it depends". The particular cell chemistry and construction plays a part in how they age and wear.

By the way, the "Battery University" web site has a wealth of information...
Charging Batteries at High and Low Temperatures – Battery University

 

[neroden]

The answer is, in the case of chemical batteries, that it's completely and entirely about chemistry. Different chemicals have different behaviors, which is why you want to use different types of batteries for different applications.

For Tesla's battery chemistry, with the temperature controlled and the charge range controlled, number of charge cycles appears to be dominant. There are other chemistries which will just die over the span of years, even if competely unused, but Tesla seems not to have picked one of those.

For future, non-chemical batteries, yet other things will be the key factors in degradation, but even then it's a matter of materials science. Consider capacitors: diffferent sorts of capacitors are optimized for different behavior, and again it's a matter of what materials you choose.

 

[ljwobker]

The question wasn't meant to be vague, it was meant to be very specific. Such as:
"for the battery chemistry used in the model S, what does the degradation curve look like for usage, and what other variables are involved, and what are the relative impacts of those variables?"

I think the problem is that the question might be TOO specific, as I'm basically asking for a function/curve like this:

battery capacity = fn(Av, Bw, Cx, Dy, Ez) where:
A = total charge cycles
B,C,D,E = something else
v,w,x,y,z = constants that describe the relative impacts of A thru E.

It's a very specific question, but I'm 100% aware that the answer is quite complex. What I was really looking for (hence the "engineering level" comment) is the collective knowledge, either via someone who does battery design/engieering for a living, or good data based on group observation, was to try and isolate and/or closely bound what these A-E and v-z numbers actually are. Someone at Tesla has this info, and it may be proprietary and secret (though that seems unlikely)... I think it's just rather that this stuff IS really complex and therefore people don't generally want to explain it. I do very similar work (large scale network systems design) so I'm more than familiar with multi-variable problems and "it depends on a lot of things". But I have a reasonably good idea what those interactions are for what *I* do, so I'm looking for someone who knows roughly the same level for batter chemistry. If there's not "a person" who has that info, maybe we can glean some of it from the data that's available. that's "all" i was asking... ;-)

The battery university link is pretty good and has some of this, so I'll dig around there.

 

[Discoducky]

If you want something to read, check out TM patents. I don't have the link readily available, but someone will.

 

[SCW-Greg]

I think *only* the battery guys at TM fully understand the Model S battery physics as ljwobker is looking for. Because this is a build/configuration/charging system that is completely new, and relatively untested (more than just a few years)... it's all theoretical to a degree. And I would wager that any other battery testing sites/research, assumes dated and small scale individual cell performance, rather than something like the newer (proprietary) Model S pack with more than 7000 cells "balanced" inside.

This is cutting edge, and very dynamic in nature, much of which is still somewhat being learned as TM goes forward.

That aside, Richkae had been amassing Roadster logs from willing participants, getting real world life cycles and decay patterns. Though this bears very little in common with the packs the Model S uses.

 

[SteveH]

If battery cycling is important should I not charge my car after driving if it still has enough range for my next planned trip? For example, my standard commute is total 50 miles. I could very comfortably only charge every other day as long as I'm only going to work and back. Would that extend battery life?

 

[neroden]

The question wasn't meant to be vague, it was meant to be very specific. Such as:
"for the battery chemistry used in the model S, what does the degradation curve look like for usage, and what other variables are involved, and what are the relative impacts of those variables?"

I think the problem is that the question might be TOO specific, as I'm basically asking for a function/curve like this:

You're asking for specific details about the behavior of the Panasonic (edit: IIRC) cells which Tesla uses. Which, for the model S, are CUSTOM. Only people inside Panasonic or Tesla would know. However, even they might not know, because you actually verify this stuff by decades of testing, and they haven't *had* decades of testing yet since they're relatively new.

So, yeah, nobody here will be able to answer this question.

 

[Lloyd]

Tesla recommends keeping your car plugged in for maximum battery life. I believe that the more important factor is total energy delivered, rather than the number of short charge cycles.

 

[hcsharp]

If battery cycling is important should I not charge my car after driving if it still has enough range for my next planned trip? For example, my standard commute is total 50 miles. I could very comfortably only charge every other day as long as I'm only going to work and back. Would that extend battery life?

In short, No. This has been discussed at length in other places on the forum, although some of it may be in the Roadster sub-forums. Like Bonnie said, try searching around a little.

- - - Updated - - -

...I think the problem is that the question might be TOO specific, as I'm basically asking for a function/curve like this:

battery capacity = fn(Av, Bw, Cx, Dy, Ez) where:
A = total charge cycles
B,C,D,E = something else
v,w,x,y,z = constants that describe the relative impacts of A thru E.
...

I, too, would love to see testing data for the specific Panasonic chemistry being used in both the MS and Roadster. Thorough data sheets would be quite interesting. Requests for this have been turned down citing confidential proprietary information. We have a lot of good information from sites like battery university and various research linked in these forums, but not always as detailed, accurate, and extensive as you and I would like. Very little that's specific to the newer chemistry that's being implemented.

 

[stopcrazypp]

The question wasn't meant to be vague, it was meant to be very specific. Such as:
"for the battery chemistry used in the model S, what does the degradation curve look like for usage, and what other variables are involved, and what are the relative impacts of those variables?"

The cells being used by the 85kWh Model S is Panasonic's NCR18650A. They are 18650 cells (AKA laptop cells). The chemistry they use is Lithium Nickel Cobalt Aluminum Oxide (LiNiCoAlO2, the marketing term is NNP). Older laptop cells used Lithium Cobalt Oxide (LiCoO2). Details about the battery chemistry is here:
http://powerelectronics.com/mag/601PET06.pdf

The only publicly available information on those cells is the battery datasheet here:
http://industrial.panasonic.com/www-data/pdf2/ACA4000/ACA4000CE254.pdf
Unfortunately, in terms of degradation data (which is what you are interested in), that datasheet only shows 300 cycles of testing.

1) Charge Cycles
There's another Panasonic document here that showed cycle testing to 500 cycles, but the link is now dead. But here's a chart I made from that document:

http://www.teslamotorsclub.com/show...or-Model-S/page2?p=61317&viewfull=1#post61317T

2) Temperature
Here's an excellent paper done on the effects of battery temperature on similar older gen LiCoO2 cells (Figure 1 shows cycle testing vs. temperature):
http://www.che.sc.edu/faculty/popov/drbnp/website/publications_pdfs/web51.pdf
Nasa actually tested the 2900mAh version of the Panasonic cells in the Model S (which uses the 3100mAh version) and has lots of good data here of temperature, discharge speed vs life (although the testing is only to 200 cycles):
https://batteryworkshop.msfc.nasa.g...st_Panasonic_Li-Ion_NCR_18650_Cells_JRead.pdf

3) Time spent at SOC
I haven't been able to find good data on this. The only reference I have seen on this is from battery university (examining only storage at 40% SOC vs 100% SOC at various temperatures):
http://batteryuniversity.com/learn/article/how_to_store_batteries

We do have a thread on Model S battery degradation here (it goes into calendar life mostly, rather than cycle life, so it's a slightly different animal):
http://www.teslamotorsclub.com/show...-does-the-battery-degradation-curve-look-like

 

[hcsharp]

The cells being used by the 85kWh Model S is Panasonic's NCR18650A. They are 18650 cells (AKA laptop cells). The chemistry they use is Lithium Nickel Cobalt Aluminum Oxide (LiNiCoAlO2, the marketing term is NNP). Older laptop cells used Lithium Cobalt Oxide (LiCoO2). Details about the battery chemistry is here:
http://powerelectronics.com/mag/601PET06.pdf

The only publicly available information on those cells is the battery datasheet here:
http://industrial.panasonic.com/www-data/pdf2/ACA4000/ACA4000CE254.pdf
Unfortunately, in terms of degradation data (which is what you are interested in), that datasheet only shows 300 cycles of testing.

1) Charge Cycles
There's another Panasonic document here that showed cycle testing to 500 cycles, but the link is now dead. But here's a chart I made from that document:
<snip>
http://www.teslamotorsclub.com/show...or-Model-S/page2?p=61317&viewfull=1#post61317T

2) Temperature
Here's an excellent paper done on the effects of battery temperature on similar older gen LiCoO2 cells (Figure 1 shows cycle testing vs. temperature):
http://www.che.sc.edu/faculty/popov/drbnp/website/publications_pdfs/web51.pdf
Nasa actually tested the 2900mAh version of the Panasonic cells in the Model S (which uses the 3100mAh version) and has lots of good data here of temperature, discharge speed vs life (although the testing is only to 200 cycles):
https://batteryworkshop.msfc.nasa.g...st_Panasonic_Li-Ion_NCR_18650_Cells_JRead.pdf

3) Time spent at SOC
I haven't been able to find good data on this. The only reference I have seen on this is from battery university (examining only storage at 40% SOC vs 100% SOC at various temperatures):
http://batteryuniversity.com/learn/article/how_to_store_batteries

We do have a thread on Model S battery degradation here (it goes into calendar life mostly, rather than cycle life, so it's a slightly different animal):
http://www.teslamotorsclub.com/show...-does-the-battery-degradation-curve-look-like

While this is helpful data, it's not very applicable to the Model S. These were all full-cycle charges from 4.2v to 2.5v. The MS never charges all the way to 4.2v, not even a range mode charge. And AFAIK it never discharges to 2.5v. Both of these conditions have a dramatic effect on battery life. This chart looks more like what happens in a laptop except the temps are allowed to go higher in the laptop. What we want is the same graph for charging up to about 4.05v (or 4.1v?) which would simulate a std mode charge, with a plot for each of several temperatures, and then another graph with a plot for each of several different levels of discharge (eg. .3Ah, .6Ah ... 2.9Ah) per cycle. Tesla and Panasonic have this data.

 

[dhrivnak]

The standard recommendation is to plug in every day. Many small charge cycles are easier than a few big ones.

 

[stopcrazypp]

While this is helpful data, it's not very applicable to the Model S. These were all full-cycle charges from 4.2v to 2.5v. The MS never charges all the way to 4.2v, not even a range mode charge. And AFAIK it never discharges to 2.5v. Both of these conditions have a dramatic effect on battery life.

I'm well aware of that, but my links are the only publicly available data on the MS cells (and basically all standard cell tests will follow this criteria of "full" cycling). It's a more "conservative" look at battery life, but it'll give a good idea of a ball park figure to expect.
The NASA charts have a higher 2.7V cut off.

This chart looks more like what happens in a laptop except the temps are allowed to go higher in the laptop.

The chart has the batteries at 25 degrees C. I forgot to mention for the temperature effect section, it's important to point out that Tesla's packs are liquid cooled so those effects are largely irrelevant (the Tesla pack won't do worse in hot areas, unlike an air cooled pack like the Leaf).

What we want is the same graph for charging up to about 4.05v (or 4.1v?) which would simulate a std mode charge, with a plot for each of several temperatures, and then another graph with a plot for each of several different levels of discharge (eg. .3Ah, .6Ah ... 2.9Ah) per cycle. Tesla and Panasonic have this data.

Well, it's pointless discussing the data that Tesla/Panasonic might/might not have because they clearly have shown they will not release any of it. Every battery reference I've looked at says using an assumption of scaling by cycles (1 full cycle = 2 half cycles = 4 quarter cycles, etc) is a nice conservative estimate of battery cycling.

If you want data on smaller cycles, there is this test on Sony 18650 cells (1994 and 2002 versions).
The test conditions all have the cells under 4.1V and above 3.0V (same as what Tesla does for the Roadster, see Table 1 for the conditions of all the tests). Test 1 and 2 use ~20% DOD and 32 cycles per day. Test 3 uses 75% DOD and 0.25 cycles per day. Test 4 and 5 don't do any cycling at all.
http://www.dtic.mil/cgi-bin/GetTRDoc?AD=ADA515369

Roadster battery voltages:
http://www.teslamotors.com/blog/bit-about-batteries