Is Tesla making progress in improving battery energy density?

[electracity]

A 30% improvement in energy density over the model 3 solves all the roadster and semi battery issues. The 30% improvement of the batteries in the model 3 over the original model S is somewhat less than expected improvement over 5 years. Jumping another 30% in three years would still leave very reasonable total growth for the decade.

 

[SDRick]

That would freeze sales while people waited for the new tech. Would you buy a new S or X next week if Tesla has talked about significant new battery tech?

Good point!

I hope I am wrong, but I don't see a significant increases in lithium battery density in the next few years. It reminds me somewhat of solar panel efficiency increases. It is one thing to achieve great things in the lab but then another to commercialize it. Even then, historically it's been more incremental than monumental within the same battery technology.

 

[electracity]

If it indeed true that the Semi and the 2020 Roadster have an improved chemistry, why wasn't it obvious in the Model 3?

Perhaps because they needed a cell ready for mass production in 2017? Flip the question. Why would a cell ready for mass production in 2020 only be a 30% improvement over 2012? That would be an improvement rate well below what has been generally expected.

If there is a significantly new cell I would expect to see it in a new pack in the S/X first.

Even then, historically it's been more incremental than monumental within the same battery technology.

True, but plodding along at 6-7% per year gives a doubling in ten years. After twenty years at that pace battery will have half the power density of gasoline. Batteries don't have to be li-on either.

Guessing whether improvements will be density or price/performance is tough.

But I think we are far too early in the era of producing massive amounts of batteries to expect a flattening of the experience curve.

 

[Saghost]

Is there any truth that the roadster test drive demo cars utilized two 100Kw batteries sandwiched on top of each other? I would imagine if there was significant new battery density technology it would have been mentioned.

To use two complete S/X packs isn't possible given the geometry of the car and packs - for one thing, remember the 100 is a 16 module pack - with a double height section up front.

I wouldn't be at all surprised if Tesla used a double height design and existing modules to power the prototype, but they'd need a new pack structure if nothing else.

 

[Rashomon]

There's good bloody reason for our pessimism.

Wikipedia has not changed it's estimate for max Lithium-ion at 265 Wh/kg for years.

Battery University is a similar story.

Elon Musk's claims about improvements were quite close to industry max energy density at the time.

Model 3 cells show a similar energy density to the S cells when taking into account a realistic weight of 65-70 grams given the 2170's volume over the 18650's volume, and the 18650 is 47.5-49.5 grams.

Not to mention the absolute horde of two-bit peddlers claiming "ground-breaking proprietary technology" of some sort or the other and completely failing to prove it when the big players are actually relatively open about their chemistry.

And who can forget all those research articles that mention a genuine breakthrough in one or two or more properties, only to hide well out of sight, a few important and often inherent small prints, usually such as inherent manufacturing cost, lifetime, or efficiency.

Having been following the lithium-ion industry since 1998, I share your pessimism to some degree. However, I recently had lunch with an engineer who used to work for me who's now leading pack design at one of the larger small cell companies -- not one of the big 3. They're going to be at 300wh/kg very soon, and have things in the test pipeline beyond that. I also know the CTO of that company, who was very educational several years ago on the vaporous nature of the industry -- do not take the research lab press releases with any real seriousness. I'm fairly comfortable we'll be seeing 300-350 wh/kg by 2020 -- but perhaps not in the most affordable cells. Tesla seems to be going down the path of the absolute lowest cost cells with adequate performance, which explains why their 21700s are only at about 5.0ah -- disappointing given Musk and Staubel's earlier statements on the Gigafactory cells.

 

[TexasTeslaRacing]

I was referring to the way drag strips measure things. It would be much closer to a pure 0-60 (and quarter mile). But am sure this has been studied and not implemented for historical reasons.

I’m a drag strip regular with a P100D and I wish everyone, including Tesla, would use the actual time it takes to go 0-60, instead of using the 1ft roll out. It’s too subjective. There is up to a three tenths difference between the two. My best 0-60 at the strip was 2.41, however without the 1ft rollout the actual 0-60 was 2.73.

Its possible the actual 0-60 of the base new roadster is 2.2-2.3 seconds. Maybe the special upgrade package makes it an actual 0-60 of 2.0sec flat. That would be even more impressive.

 

[JKolodziejski]

Having been following the lithium-ion industry since 1998, I share your pessimism to some degree. However, I recently had lunch with an engineer who used to work for me that's now leading pack design at one of the larger smaller cell companies -- not one of the big 3. They're going to be at 300wh/kg very soon, and have things in the test pipeline beyond that. I also know the CTO of that company, who was very educational several years ago on the vaporous nature of the industry -- do not take the research lab press releases with any real seriousness. I'm fairly comfortable we'll be seeing 300-350 wh/kg by 2020 -- but perhaps not in the most affordable cells. Tesla seems to be going down the path of the absolute lowest cost cells with adequate performance, which explains why their 21700s are only at about 5.0ah -- disappointing given Musk and Staubel's earlier statements on the Gigafactory cells.

SK Innovation's NCM 811, by any chance?

I am guessing the volumetric density is going to be 800 Wh/L at 300 Wh/kg, due to a consistent Li-ion density of approximately 2 2/3 kg/L, and judging from the difference between the older types and upper limits of NMC 811.

(that is, 250Wh/kg & 670Wh/L and the theoretical maximum manufacturable NCM 811's with 350+ Wh/kg & 950+ Wh/L, from another source I can't find again...)

http://pushevs.com/2017/09/02/sk-innovation-start-producing-ncm-811-battery-cells-soon/

Also see the links at the bottom of that page.


I've also found some interesting consistencies:

- From wk057's post, an 18650 from the 100kWh pack has 12.4 Wh of usable energy.

- 12.4 Wh results from a 3.6 to 3.65 V nominal at 3400 to 3450 mAh recorded for Panasonic's top 18650 batteries.

- From a Panasonic datasheet, an 18650 is said to weigh 49 grams , give or take half a gram. Dividing 12.4 Wh by this number gives 253 Wh/kg.

- Increasing 12.4 Wh by the 2170's volume difference of +46% and accounting for Straubel's "15% more efficient" statement, that is, 12.4 x 1.46 x 1.15, gives a result of 20.8 Wh.

- A figure that got bandied around sometime ago was the "5750 mAh" from a presentation somewhere.

- As it turns out, dividing 21 Wh / 5.75 Ah gives... 3.65V.

- Given that a 2170 cell is highly likely to have a mass of 65-70 grams by noting the difference between the masses of the 18650s and big old 26650s, circa 20 to 21 Wh at 70 to 65 grams means a specific energy range of 20/0.070 to 21/0.065 or 285 to 320 Wh/kg & therefore 760 to 850 Wh/L.

So there is my new assumption... 300 Wh/kg and probably 800 Wh/L it is.

...And I'll just leave this here as well. Would love to know if anyone could find out the source of that image halfway down the page...

Because it explicitly states that the 2170 cell by Tesla has 300 Wh/kg and costs $170 per kWh.

https://www.dnkpower.com/teslas-mass-production-21700-battery/

 

[JKolodziejski]

@Rashomon

I really wish posts could be edited... the numbers are correct but the sentences are misleading.

1) "12.4 Wh of usable energy" should be "12.4 Wh of absolute energy".

2) "Dividing 12.4 Wh by this number gives 253 Wh/kg." refers to 12.4/0.049, not 12.4/49.

 

[Rashomon]

SK Innovation's NCM 811, by any chance?

I am guessing the volumetric density is going to be 800 Wh/L at 300 Wh/kg, due to a consistent Li-ion density of approximately 2 2/3 kg/L, and judging from the difference between the older types and upper limits of NMC 811.

(that is, 250Wh/kg & 670Wh/L and the theoretical maximum manufacturable NCM 811's with 350+ Wh/kg & 950+ Wh/L, from another source I can't find again...)

http://pushevs.com/2017/09/02/sk-innovation-start-producing-ncm-811-battery-cells-soon/

Also see the links at the bottom of that page.


I've also found some interesting consistencies:

- From wk057's post, an 18650 from the 100kWh pack has 12.4 Wh of usable energy.

- 12.4 Wh results from a 3.6 to 3.65 V nominal at 3400 to 3450 mAh recorded for Panasonic's top 18650 batteries.

- From a Panasonic datasheet, an 18650 is said to weigh 49 grams , give or take half a gram. Dividing 12.4 Wh by this number gives 253 Wh/kg.

- Increasing 12.4 Wh by the 2170's volume difference of +46% and accounting for Straubel's "15% more efficient" statement, that is, 12.4 x 1.46 x 1.15, gives a result of 20.8 Wh.

- A figure that got bandied around sometime ago was the "5750 mAh" from a presentation somewhere.

- As it turns out, dividing 21 Wh / 5.75 Ah gives... 3.65V.

- Given that a 2170 cell is highly likely to have a mass of 65-70 grams by noting the difference between the masses of the 18650s and big old 26650s, circa 20 to 21 Wh at 70 to 65 grams means a specific energy range of 20/0.070 to 21/0.065 or 285 to 320 Wh/kg & therefore 760 to 850 Wh/L.

So there is my new assumption... 300 Wh/kg and probably 800 Wh/L it is.

...And I'll just leave this here as well. Would love to know if anyone could find out the source of that image halfway down the page...

Because it explicitly states that the 2170 cell by Tesla has 300 Wh/kg and costs $170 per kWh.

https://www.dnkpower.com/teslas-mass-production-21700-battery/

Well, we now know the Tesla/Panasonic cells, assuming they are nominally 3.6 volt, are rated at 5.0Ah, or have 18Wh each, not 20wh. This is only a small improvement on the current 18650s when you take the volume increase of the new cells into account. How do we know this? We have the cell count and capacity of M3 packs. The pack for the long-range M3 has 4,416 of them, and is nominally 79-80 kWh, with something like 75kWh usable. The standard pack is about 53.5kWh, with 2,976 cells, and a usable capacity (probably) of around 50 kWh. Efficiency on the M3 is very good to get the 220 mile EPA range. This strongly suggests that these cells in a MS/MX will not give a substantial increase from the 100kWh currently available from 18650s.

 

[electracity]

This strongly suggests that these cells in a MS/MX will not give a substantial increase from the 100kWh currently available from 18650s.

Which perhaps suggest Tesla is sandbagging the M3 until the S/X packs can be updated.

 

[jkn]

Well, we now know the Tesla/Panasonic cells, assuming they are nominally 3.6 volt, are rated at 5.0Ah, or have 18Wh each, not 20wh. This is only a small improvement on the current 18650s when you take the volume increase of the new cells into account. How do we know this? We have the cell count and capacity of M3 packs. The pack for the long-range M3 has 4,416 of them, and is nominally 79-80 kWh, with something like 75kWh usable. The standard pack is about 53.5kWh, with 2,976 cells, and a usable capacity (probably) of around 50 kWh. Efficiency on the M3 is very good to get the 220 mile EPA range. This strongly suggests that these cells in a MS/MX will not give a substantial increase from the 100kWh currently available from 18650s.

If they can fit slightly longer M3 cells into MS/MX back, total cell volume and capacity will increase. This may be impossible.

Sooner or later they get new chemistry with higher capacity. It will be used first on low volume product. M3 will get it last.

 

[JKolodziejski]

Well, we now know the Tesla/Panasonic cells, assuming they are nominally 3.6 volt, are rated at 5.0Ah, or have 18Wh each, not 20wh. This is only a small improvement on the current 18650s when you take the volume increase of the new cells into account. How do we know this? We have the cell count and capacity of M3 packs. The pack for the long-range M3 has 4,416 of them, and is nominally 79-80 kWh, with something like 75kWh usable. The standard pack is about 53.5kWh, with 2,976 cells, and a usable capacity (probably) of around 50 kWh. Efficiency on the M3 is very good to get the 220 mile EPA range. This strongly suggests that these cells in a MS/MX will not give a substantial increase from the 100kWh currently available from 18650s.

That was my assumption on my geometrical analysis of the Semi in that thread - roughly 18.25 Wh. It's also very much 46% larger than the result for the the 18650, 12.4 Wh. I would assume 3.65V average, it seems to give more accurate figs.

However a lot of people are assuming that the 2170 actually has less mass than would be suggested by it's volume - roughly 65 grams as opposed to 70. That still means 18.2/0.065 = 280 Wh/kg.

Using a mere 4% rule of thumb from early 2017 onwards, that means 8% by beginning of 2019, so, 280 x 1.08 = 302 - 303 Wh/kg. so 300 Wh/kg is a pretty good assumption for then, but probably not now.

So 750 kWh usable now, and 800 kWh usable later with a little extra weight and a little higher efficiency?

 

[HillCountryFun]

2. I don't think they've simply stacked two Model 3 packs. I think they've stacked Model 3 battery modules in some configuration in the car.

I agree with the statement above. Here are my thoughts:

• Model S/X < 100KWh - Original 18650 cells and original Pack design
• Model S/X 100 KWh - Original 18650 cells and revised Pack design
• Model 3 - 2170 cells and second-generation pack design
• 2020 Roadster and Tesla Semi - 2170 cells and third generation pack design
• I expect new Model S's and X's to use 2170 cells and third generation pack design and a 2022 Release Date (2012 + 10)

We're all wondering how the 2020 Roadster can contain 200KWh and everyone was surprised at the range of the Tesla Semi. Both of these questions are resolved when you include the 2170 cells AND the next generation of pack design.

My guesses on Third Generation Pack Design Improvements

• Lighter weight per KWh
• Better cooling
• Higher Super/Mega Charging rates

I'm sure Tesla will incorporate cell advancements which occur over the next couple of years but the 2020 Roadster and Semi packs are doable today without any future advances.

 

[JKolodziejski]

I agree with the statement above. Here are my thoughts:

• Model S/X < 100KWh - Original 18650 cells and original Pack design
• Model S/X 100 KWh - Original 18650 cells and revised Pack design
• Model 3 - 2170 cells and second-generation pack design
• 2020 Roadster and Tesla Semi - 2170 cells and third generation pack design
• I expect new Model S's and X's to use 2170 cells and third generation pack design and a 2022 Release Date (2012 + 10)

We're all wondering how the 2020 Roadster can contain 200KWh and everyone was surprised at the range of the Tesla Semi. Both of these questions are resolved when you include the 2170 cells AND the next generation of pack design.

My guesses on Third Generation Pack Design Improvements

• Lighter weight per KWh
• Better cooling
• Higher Super/Mega Charging rates

I'm sure Tesla will incorporate cell advancements which occur over the next couple of years but the 2020 Roadster and Semi packs are doable today without any future advances.

First of all, it's 18650.

Secondly, the Semi pack is less controversial than the 2020 Roadster's, especially in volume.

• 800kWh at ~80% cell mass proportion, 280Wh/kg x 0.8 = ~ 225 Wh/kg = 3,550 kg (7,800 lb).
  
• Tractor mass at 9100 kg (20,020 lb).
• Pack mass proportion of = 3550/9100 = 39% of total. Original Roadster was 38%.
• 1.6 miles per kWh at max load. = 500 miles.
  
• 3/5 of consumption at empty load. = 800 mile.

Volume of 800 kWh at 2.65 times water density = 225 x 2.65 = ~ 600 Wh/L.

800,000/600 1+1/3 m^3, or 1300-1400 litres, approx. only 350-370 US gallons?.

 

[HillCountryFun]

800kWh at ~80% cell mass proportion

I think your pack size is a little small. To achieve their range at their stated efficiency (500 @ 2KWh/mi) and ensure the pack never bricks I suspect the total pack size is more like 1.05 to 1.1 MWh. Also, it's not clear where the pack is installed but assuming it is low on the Semi, it will require quite a bit of volume. Time will tell -

 

[mongo]

I think your pack size is a little small. To achieve their range at their stated efficiency (500 @ 2KWh/mi) and ensure the pack never bricks I suspect the total pack size is more like 1.05 to 1.1 MWh. Also, it's not clear where the pack is installed but assuming it is low on the Semi, it will require quite a bit of volume. Time will tell -

Hi,
The presentation said < 2kWh/mile not =2kWh. From the physics calculations, 1.6 kWh/ mile is potentially achievable.
Pack is directly below the cab floor, Tesla's Twitter feed has a video of component locations.

 

[SO16]

Get an EV for everyday use and rent a Prius for that one trip a year. Your life will be a whole lot better.

Ditto!!!!!

People who deal with a less quality ride for that once in a year “what if” scenario, seems silly. Rent for that one time a year. It’s like people who complain about long fill ups for Tesla for that once in a year road trip but then are fine with being inconvenienced 51 other weeks a year with not being able to fill up at home.

 

[HillCountryFun]

Hi,
The presentation said < 2kWh/mile not =2kWh. From the physics calculations, 1.6 kWh/ mile is potentially achievable.
Pack is directly below the cab floor, Tesla's Twitter feed has a video of component locations.

Thanks for the note on the video, that's about the size I was expecting. I'm not sure about your physics calcs so I still think the Semi pack size is closer to 1 MWh.

I think we're going to see a new pack Architecture with the Semi and the new Roadster.

 

[JKolodziejski]

I think your pack size is a little small. To achieve their range at their stated efficiency (500 @ 2KWh/mi) and ensure the pack never bricks I suspect the total pack size is more like 1.05 to 1.1 MWh. Also, it's not clear where the pack is installed but assuming it is low on the Semi, it will require quite a bit of volume. Time will tell -

Daimler E-FUSO Vision One (slightly streamlined, flat face, 11 tonne payload) => 300 kWh / 220 miles = 1.4 kWh/mile.

Daimler unveils heavy-duty all-electric truck concept with ‘up to 220 miles’ range


Cummins AEOS (34 tonnes overall, 75k lbs, Class 7 (drayage?) ) => 140 kWh / 100 miles = 1.4 kWh/ mile.

Besting Tesla's Reveal By Just Days, Cummins Unveils AEOS Electric Semi


BYD Class 8 (Cheap, Chinese-made conventional flat-face truck) => 188 kWh / 94 miles => 2.0 kWh / mile exact.

http://www.byd.com/usa/wp-content/uploads/2016/08/T9-final.pdf


Freightliner Supertruck (65k lbs, most similar use case and possibly similar all-length drag efficiency to the Tesla Semi.)

https://www.trucks.com/2015/06/25/freightliner-supertruck-is-super-efficient/

=> 48-52% thermal efficiency, 12.2 US-mpg, => 3.22 miles / L or per 10.8 kWh(th) @ 0.48 to 0.52 = > ~ 3.22 miles / 5.2-5.6 kWh mech.
=> (5.2 to 5.6) kWh / 3.22 miles = 1.61 to 1.73 kWh / mile after engine.

... and a 92% efficient battery power equivalent to "48-52% efficient 12.2 US-mpg Freightliner Supertruck" = 1.61 to 1.73 / 0.92 = 1.75 to 1.88 kWh / mile.


And this.

Tesla Semi is reasonable, part 1 | Selenian Boondocks

"<2 kWh" was stated for a reason.

1.5 to 1.7 kWh per mile at 56-65 mph is realistic.

 

[Brando]

I think what you are saying is that the maximum gross vehicle weight is 80,000 pounds, and that includes the tractor and its engine + fuel or motor + battery pack. Put swap out the diesel stuff for motor and batteries, and the difference in weight has to come out of the cargo.

Perhaps so, but is it really that significant a loss in net cargo, compared to the net improvement in operating efficiency?

I suspect many cargo trips are not at maximum 80,000 lbs. weight but are limited by volume.