Tesla/Panasonic Batteries are more energy dense over time. The Evidence.

Tesla/Panasonic Batteries are more energy dense over time. The Evidence.

There have been several tear downs of Tesla battery packs i was just looking at two in particular, both from Electrek, one from Jan 2017 of a 100 KWH pack and the other in Feb 2016 of the 85 KWH pack (both are obviously 18650 cells) so here's what I found.

85 KWH Pack:
80.7 KWH actual
7104 cells
11.4ish watts per cell

100 KWH Pack:
102.4 KWH actual
8256 cells
12.4ish watts per cell

one year apart the cells haven't changed size so the chemistry must have improved by about 9 percent, that's freaking awesome (not to mention all the other improvements). This just makes me even more excited about the possibility of the 2170 cells being even more chemically/energy dense, bigger, and better. what do you guys think? Did I miss something?

andy.connor.e | 15. februar 2017

Where is Electrek getting their data on Teslas batteries?

KP in NPT | 15. februar 2017

@andy I think the data comes from battery tear downs.

andy.connor.e | 15. februar 2017

But you dont know specifically where the data is coming from? Articles can say whatever they want, and ive come across quite a few articles with falsified data or just straight up lies.

andy.connor.e | 15. februar 2017

Some things i noticed.

Nowhere in the article do they say how many battery cells were tested to come up with the average "11.36 Wh". Lithium cells can have massive discrepancies. In a battery pack that has over 7000 cells, you would need to test at least 10% of the cells to get an accurate average. Just for example; if you test 1% of the batteries, is a 1% average enough to implement for the rest of the 99% of the pack? If you want accurate averages, its not.

For some reason they decided to leave out literally every single bit of detail except the number of cells/module, and the number of cells total. Obviously you can calculate the number of modules based on the given numbers, but comparing the 2 articles, they barely gave you any details from the breakdown. This article is exactly what i was concerned about. They just throw numbers at you and did not relay the information like they did in the 85kWh article. There is no Wh/cell estimated or actual, there is no # of modules, and half the article is irrelevant to breaking down the 100kWh battery for actual capacity testing.
"Hughes says that the BMS indicates 98.4 kWh of usable capacity" where did he get that number from? Theres no indication of how he came to that conclusion.

The 85kWh article was very good. Data provided was enough to address the goal. Should be more transparent of how they achieved their numbers such that the tests can be reproduced.
The 100kWh article was not scientific. Not enough data supporting the conclusion. No indication of how they got their numbers. Inaccurate conclusion until transparency of testing is provided.

djharrington | 15. februar 2017

Further to Andy's comments, I don't see any mention of discharge rate used to measure capacity. If a different rate is used, you could easily account for a 10% difference. I'm not saying there hasn't been improvement, but I don't think there's enough material here to prove the case either way.

Frank99 | 15. februar 2017

I'm assuming they're using the capacity as reported by the onboard battery management system:
and dividing it by the number of cells.

Still don't know the discharge rate, variation over production, etc., but it's probably good enough to draw the conclusions that D.Lew was drawing...

andy.connor.e | 15. februar 2017


Not good enough to make any declarative statements. I certainly wont consider this valid information that i would feel comfortable including in an argument or discussion.

Frank99 | 15. februar 2017

Why not? The BMS should have an excellent idea of the cumulative capacity of the 7-8000 cells in the pack, because it's measuring both the energy required to charge the pack as well as the energy drawn from the pack in use. I recognize that battery capacity will be different depending on the loads drawn, but it's hard to argue that there's a better metric for measuring battery capacity in an EV application than, well, measuring battery capacity in an EV application.

Frankly, I can't imagine getting higher quality information on battery performance without signing an NDA.

D.Lew | 15. februar 2017

Thanks everyone for the comments I'll agree its not perfect and maybe a bit simplistic, but I stand by what I wrote and I believe its fairly accurate.

andy.connor.e | 15. februar 2017


Because total energy flow to the battery depends on the charging rate. Theres an article i read where someone got nearly 10-15kWh more in a "100%" charge by charging slow (20 or so amps) at home vs at a supercharger.

If they do not disclose the parameters of how they performed the testing, there can be very significant discrepancies.

Not accurately but just for example, that the 85kWh battery testing was when it was 105 degrees outside, and the 100kWh testing was done when it was 70 degrees. I dont know, but they need to disclose testing parameters and conditions.

Badbot | 17. februar 2017

the way batteries get more dense over time
A. the internet (makes everyone denser )
B. the car crusher squeezes them

cephellow | 17. februar 2017

Some advice for this thread.
Wk057 a.k.a Jason Hughes has done the teardowns and also has root on the car software; he is a very venerated and reliable source of a lot of this information. Look him up, find his site, so that you all may stop having uninformed arguments on this topic. | 19. februar 2017

I agree Wk057 is very knowledgeable. It's less clear if the article writers understood all the issues. One not talked about is how old the packs were when salvaged. It's very likely the 85 was a lot older than the 100 pack, and would explain some of the difference. Older cells will have lower capacity. It's also not clear how the cells were tested (or if they were tested) to come up with the numbers. Using the BMS software numbers may seem valid, but we don't know how Tesla loads the initial numbers for different packs. They may pick values that work for longevity testing that has less to do with actual cell power. This is a super complex area, not easily resolved in the forum. The numbers could be spot on, but I see a lot of reasons why they may be off in either direction. | 19. februar 2017

@Tap: +1

12.4 wh/cell would yield an actual capacity of about 88.5 kWh for my "90" kWh pack then.

9% is what tear downs suggest for the energy density improvement contrary to the 6% that Elon spoke of (90/85 = 1.06). Impressively!

Further improvements in energy density are coming with the 2170 cells built in NV with speculation ranging from <10% to >30% improvement in energy capacity per cc. | 19. februar 2017

Impressive not impressively. I wonder where the iPad autocorrect gets this stuff. Another danger of AI?

djharrington | 20. februar 2017

George, is this the new way we'll talk about our packs: in cc's? Hey, what's your displacement? I like it. It'll confuse the petrolheads. | 20. februar 2017

Well @dj, I'm often confused as well. But, since the dimensions of the cells are expressed in mm, it seems natural to think in terms of watt-hours per cc for energy density.

Frank99 | 20. februar 2017

Hey, we talk about charging in terms of miles per hour, which has certainly got to confuse people, so why not capacity in cc's?