Range shorter than EPA (if you want to protect battery)

[omgwtfbyobbq]

There's something to larger depth of discharge really hurting capacity. Using ~50% of capacity only reduces capacity by ~10+% after 3000 cycles, but using 100% reduces capacity by ~15% after only 300 cycles.

http://www.teslamotorsclub.com/show...atteries/page6?p=340581&viewfull=1#post340581

 

[stopcrazypp]

There's something to larger depth of discharge really hurting capacity. Using ~50% of capacity only reduces capacity by ~10+% after 3000 cycles, but using 100% reduces capacity by ~15% after only 300 cycles.

http://www.teslamotorsclub.com/show...atteries/page6?p=340581&viewfull=1#post340581

First of all, the two cells were different cells (the graph on the left are prototype cells of smaller capacity, the one on the right is the actual high capacity 3100mAh production version). A true test would require the same cells (preferably from the same batch) and doing multiple batches of tests.

Second of all, the usage you get out of the shallow cycle is less (voltage range is 0.45V for the left, 1.7V for the right) and that must be accounted for. Looking at the battery graph in the next post, at 1lt, 3.6V represents a 900mAh discharge out of about 3000mAh which normalizes to about 30% capacity. That means you have to adjust the 3000 cycles to be 1000 cycle equivalent.

Third is that even under the baseline 100%DOD (which doesn't happened anyways even at "100%" charge), your pack will still last well past 100k miles.
http://www.teslamotorsclub.com/show...ook-like/page3?p=204756&viewfull=1#post204756

 

[omgwtfbyobbq]

I think those are all valid points. With that said, I think calendar life loses are higher than some other batteries, especially since the BMS is a little more lenient with thermal management.

http://www.teslamotorsclub.com/showthread.php/22172-Model-S-Overall-Thermal-Management

Real world, depending on location, I imagine that someone will see a ~3+% decline per year in capacity due to aging, and another 3+% decline in capacity per year due if they used a larger percentage of the battery, versus a ~1+% decline with shallower cycling. I think it's prudent for owners to minimize the energy used per cycle (within reason) if they would like to maximize battery life. Similarly, high discharge (100%), but low c-rate cycling also has much less impact on capacity.

http://ma.ecsdl.org/content/MA2011-02/17/1282.full.pdf

Ultimately it's a matter of preference. I'd rather have the battery in my MS last as long as the battery in my Prius before requiring replacement (200k+ miles) rather than have it replaced at 100k+ miles.

 

[stopcrazypp]

Real world, depending on location, I imagine that someone will see a ~3+% decline per year in capacity due to aging, and another 3+% decline in capacity per year due if they used a larger percentage of the battery, versus a ~1+% decline with shallower cycling.

You have to be careful on that. Calendar and cycling losses are not additive. The NREL battery degradation model (which has been matched to actual cell behavior of the same NCA chemistry) says to take which one is worse, but not to add the losses together. The reason for this is because cycling suppresses electrolyte film growth, which is the main cause of degradation from storage (see page 11 here):
http://www.nrel.gov/vehiclesandfuels/energystorage/pdfs/45048.pdf

Ultimately it's a matter of preference. I'd rather have the battery in my MS last as long as the battery in my Prius before requiring replacement (200k+ miles) rather than have it replaced at 100k+ miles.

I agree with this. If you want your battery to well past 200k+ miles, it makes sense to care about these things. But if you have the typical 100k+ miles perspective, then stressing over this is pointless.

 

[brianman]

I agree with this. If you want your battery to well past 200k+ miles, it makes sense to care about these things. But if you have the typical 100k+ miles perspective, then stressing over this is pointless.

Can you specify a timeframe here? Thanks.

 

[stopcrazypp]

Can you specify a timeframe here? Thanks.

Time frame is a bit different as that depends on calendar life, which is unlikely to be worse than cycle life unless you rarely use the car (or you store it at high SOCs and/or high temperatures).

And in that regard, here's a long term test on 18650 battery cells of the chemistry similar to what the Roadster uses:
http://www.dtic.mil/cgi-bin/GetTRDoc?AD=ADA515369

Basically there was ~7% loss in the first year and ~1% loss per year thereafter.

As for the battery cells I linked in the previous page (which may not model the Tesla accurately, but it's the closest we have), the cycle life is as follows:
0-300 cycles: lose ~7.8% per 100 cycles
300-500 cycles: lose ~1.35% per 100 cycles

That means during the early life of the battery, 1% loss from cycling will take only 13 full cycles, which is about 3500 EPA miles, which most people will easily pass in a year. Of course in the real car you will get more miles than that (because of the limiting in SOC and also lower average charge/discharge rates), but that's the ball park.

 

[brianman]

@stopcrazypp - Let me clarify. You were giving rough guidance on what kinda of vehicle usage should warrant "extra attention" to charging habits. That guidance had raw mileage numbers but no timeline. That's not as useful as I'd like. I was hoping you could clarify that a bit more. 200k miles in 80 years is very different from 200k miles in 4 years, for example.

 

[stopcrazypp]

@stopcrazypp - Let me clarify. You were giving rough guidance on what kinda of vehicle usage should warrant "extra attention" to charging habits. That guidance had raw mileage numbers but no timeline. That's not as useful as I'd like. I was hoping you could clarify that a bit more. 200k miles in 80 years is very different from 200k miles in 4 years, for example.

Again in most common usage (above 5k miles per year driven), the miles tends to matter more and that will vary based on how common you drive.

Again, calendar loss is ~7% loss in the first year and ~1% loss per year, and taking 70% "end of life" means about 24 years of life (adjust as appropriate for your "acceptable" end-of-life number). The conditions were 75% DOD cycling for 25% of the time and each cell kept at 4V (~90%SOC) for 75% of the time. The cells were temperature controlled to ~23°C (ambient temperature of the lab) using an aluminum heat sink.

This number is changed if you keep the pack at high SOCs and high temperatures (Model S is aided by active cooling though). The wikipedia article on li-ion batteries gives a general idea in terms of loss per year (although this data is likely first year losses which tend to be higher):
(Temperatures) 0°C 25°C 40°C
100 % SOC : 06% 25% 35%
40-60%SOC: 02% 04% 15%
http://en.wikipedia.org/wiki/Lithium-ion_battery#Battery_life

The basic idea is to take the worst number (like most warranties, take whichever comes first: miles or years).

 

[brianman]

Again in most common usage (above 5k miles per year driven), the miles tends to matter more and that will vary based on how common you drive.

This helps ballpark the timeline. Thanks.

 

[omgwtfbyobbq]

You have to be careful on that. Calendar and cycling losses are not additive. The NREL battery degradation model (which has been matched to actual cell behavior of the same NCA chemistry) says to take which one is worse, but not to add the losses together. The reason for this is because cycling suppresses electrolyte film growth, which is the main cause of degradation from storage (see page 11 here):
http://www.nrel.gov/vehiclesandfuels/energystorage/pdfs/45048.pdf

That may not be strictly accurate. Based on slide 14, it looks like low %dod cycling does inhibit the high t^.5 resistance growth due to calendar life aging, but higher %dod cycling does not inhibit that nearly as much.

I think that the Qsites equation covers capacity loss at both high and low %dod cycling for this chemistry, where low %dod suppress resistance growth, while high %dod don't do such a good job of that, and modelling no cycling is fairly straight forward and represents an upper bound to capacity loss based on temperature.

With all of that said, I think this does lend itself to what the OP is talking about. Specifically, it looks like the battery could last for a really long time, perhaps a 300k-400k, or longer, if the owner lives in a cool climate (or can mod the MS cooling system to keep the batteries cooler at the expense of range), and keeps the average %dod in the 35%-50% range.

On the other hand, like you mentioned, it may only last 100k miles in warmer climates, and in that context higher %dod cycling may not matter as much because calendar life capacity loss is so high anyway.

What's really crazy is that capacity loss due to cycling appears to be minimally affected by the number of cycles per day based on the page 18 graphs. It's mostly dependent on average temperature and average %dod (and I think charge/discharge c-rate). I'm really curious to see how these cars would hold up when used as cabs in cooler climates assuming the owners kept the %dod around 35%. It might be possible to get 500k+ out of the pack.