Argonne Computer Model And The Implications For The 3rd Generation Tesla


Tesla Has Promised A $35,000 (ish), 200 Mile EV In 2017

Tesla Has Promised A $35,000 (ish), 200 Mile EV In 2017

Tesla has promised us an affordable electric car in around 2017 (we will just call it the Model E for now). Being affordable means having access to a battery that is even lower cost than today’s technology will allow. What path do we follow to get to this battery? Panasonic and Tesla have inked a new agreement for many more cells in the future. Tesla also is talking to Samsung about future battery supplies. A “giga” plant may be on order soon.

Panasonic's 18650 Lithium-Ion Cell For Tesla Model S

Panasonic’s 18650 Lithium-Ion Cell For Tesla Model S

Argonne labs has released a computer spread sheet design program that allows designing a battery from the bottom up. The program even designs the cell cooling system and estimates the cost.

The program assumes a battery plant of mammoth scale capable of manufacturing 20,000 to 500,000 battery PACKS per year in the 2020 time frame. What does this computer program tell us about the path that Tesla should take with their new low cost battery?

How large is this battery plant?

"How Big Is This Plant?"

“How Big Is This Plant?” – Figure 1

100,000 Tesla Model Es per year implies a plant capable of 7.4 giga watt hours per year while 500,000 Tesla Model Es implies a plant of 35 giga watt hours per year. That’s a lot! It would take a 1000 Mw nuclear power plant 35 hours to fill up that many batteries.

The computer program assumes a prismatic cell format. However the cost calculations would be relevant for batteries of a cylindrical format as well.

Slide 2

Similar Results Expected – Figure 2

Deal With Samsung SDI Would See Cells Being Delivered At The End Of 2014 As The Model X Is Launched

Deal With Samsung SDI Would See Cells Being Delivered At The End Of 2014 As The Model X Is Launched

When designing the cells it is all about the cathode and anode size, coating thickness and materials. Figure 3 references the cathode chemistry abbreviation (NCA for Tesla S) but also shows the exact chemical makeup and manufacturer. This is a great chart to save for future reference.

Included also on this chart are the material costs for the various chemistries. Note however that cost per pound for the cathode active ingredients doesn’t necessarily translate into the cost for the whole pack. For example, in the case of the NMC cathode, even though the cost per pound is higher than LMO (Volt) the whole pack price for NMC is about the same as for LMO due to NMC’s higher energy density.

"Cathode and Anode Chemistries"

“Cathode and Anode Chemistries” – Figure 3

What drives Battery Costs?

In figure 4 we see that materials and purchased items constitute the majority of the battery pack’s cost. Figure 5 shows that fully half of the material costs are in the cathode, anode and separator.

"Cost Breakdown" - Figure 4

“Cost Breakdown” – Figure 4

"Materials Only Cost Breakdown" - Figure 5

“Materials Only Cost Breakdown” – Figure 5

What do figures 4 and 5 imply? Perhaps an analogy to Henry Ford. In order to lower costs, Ford made his own steel. Iron ore went in one end of the plant and cars came out the other.

Effect of Cathode Size, Thickness and the Number of Cells

As we go to lower power/energy ratio cells (higher AER) cathode coating thickness increases. We can increase capacity (kwh) by increasing cathode active material thickness within a given cell or we can just increase the number of cells as Tesla now does on the Model S. Current day batteries have cathode active material thicknesses of less than 100 microns. If we go to thicker cathode active material thickness instead of increasing the number of cells we can get battery cost down. This is shown in figure 6.

"Effect Of The Number Of Cells"

“Effect Of The Number Of Cells” – Figure 6

We see that cutting the number of parallel cells in half reduces the price of the battery 17%. I believe that Tesla will go to a larger battery than the current 18650 format for Model E.

Effect of Production Rate

The computer program assumes a plant with a production rate of 20,000 to 500,000 packs per year. Once that plant is up and running, what effect does increasing production rate have on final pack cost?

Manufactoring Rate For "Giga" Plant

Manufacturing Rate For “Giga” Plant – Figure 7

Future Tesla "Giga" Lithium-Ion Battery Facility Could Be Located On Land Recently Bought Beside Production Line

Future Tesla “Giga” Lithium-Ion Battery Facility Could Be Located On Land Recently Bought Beside Production Line

The effect can be seen in figure 7. Increasing production rate from 20,000 to 100,000 packs per year would lower pack price by 12%. This may seem like a small amount but we must remember that this is for changing the rate ONCE THE PLANT IS BUILT. The biggest step change in cost comes with building the “giga” plant in the first place. Once the plant is built, production rate has a more modest effect on price.

Cathode Chemistry and Safety Issues

The combustion rates for Tesla Roadster, Model S, Volt, A123 and NMC are shown in figure 8.

"ARC Safety Testing of Cathodes" - Figure 8

“ARC Safety Testing of Cathodes” – Figure 8

The combustion rates are given in degrees C/ minute. This would be how fast the fire gets hot right after ignition. Tesla Roadster has the highest combustion rate at 350 degrees C/minute followed by the Model S chemistry at 275 degrees C/min. While Model S is an improvement over the Roadster it is still orders of magnitude higher than the Volts LMO chemistry at 2.5 degrees C/min. NMC chemistry is quite a bit better than the Tesla S with a combustion rate of 50 degrees C/min.

Tesla Roadster And Model S Noted As Gen 1 And Gen 2

Tesla Roadster And Model S Noted As Gen 1 And Gen 2

It is interesting to note on this Sandia chart that the Tesla Roadster is referred to as Gen 1, Tesla S is referred to as Gen 2 and NMC is referred to as Gen 3. Is this just a coincidence or could Tesla be considering NMC for their Model E? It is probably a long shot but it is a possibility.

The biggest draw backs is that NMC would probably not offer a much higher energy density than the current Model S chemistry (NCA). However NMC would offer an improved energy density over LMO (Volt) cathode chemistry (approximately 25% smaller pack for Volt with NMC instead of LMO).


How much does a battery made in this “giga” plant battery cost?

"What Cost Is Argonne Predicting?" - Figure 9

“What Cost Is Argonne Predicting?” – Figure 9

As shown in figure 9, Argonne predicts a battery pack price of 135$/kwh with an uncertainty of +26% and -18%.


Big Wrap-Up - "Path To Tesla Model E Battery"

Big Wrap-Up – “Path To Tesla Model E Battery”

The path to Model E battery is shown in figure 10. The single most influential factor is construction of a “giga” plant capable of producing 100,000-500,000 packs per year. Secondly Tesla must have access to best pricing on the cathode active materials.

Can Tesla Really Get Its Third Gen To Market On Time And As Promised?

Can Tesla Really Get Its Third Gen To Market On Time And As Promised?

In case of NMC chemistry this means a good price on Ni and Co since the Cathode, anode and separator represent half of the battery material cost and materials cost are the biggest driver for a plant of this scale. Second order effects can also bring down battery costs. Raising production rate could bring costs down 10-20%. Changing to a larger format cell can also be expected in the Tesla Model E battery design. Changing to a larger format would decrease cost by 17%.

Does all this sound doable?

Can Elon Musk pull this off?

Even more importantly: can this new battery be done with battery chemistry that exists today??

….or must we wait for a breakthrough in battery chemistry still to come.

Reference: Argonne

Categories: Tesla


Leave a Reply

15 Comments on "Argonne Computer Model And The Implications For The 3rd Generation Tesla"

newest oldest most voted

Nice breakdown of the 140 page Argonne National Lab report George on the existing battery chemistry. Hard to challenge their data without a lot of work but at least it puts some factors into perspective. Whether the gen III gets a new chemistry is unknown but it surely seems that the planning had to be done with existing chemistries. Keep them coming George.

$135 per kWh probably does not include capital costs of gigafactory. This would be interesting to know how much exactly the planned production capacity will cost?

Elon Musk said in conference call that Tesla is not planning larger cell sizes, because 18650 is close to optimal size.

I think this may be too pesimistic, the assumption is 74kwhr pack and the Model S is available with a 60kwhr battery pack now, range 208 miles. I think 50kwhr is more like it, the “E” will be smaller and less powerful, this makes the pack ~$7,000! The big question remaining is how much do they let you use? If it’s 90% the pack costs $7500, if it’s 60% it’s $10,000.
If you go up from the Leaf (closer to the right size) the numbers are about the same, Nissan let’s you use 22kwhr and range at 100% charge is ~85 miles, 50kwhr useable would put you right at 200.
I predict the “E” will have a 50kwhr pack that costs less than $10K.

I was thinking the same thing. The 74kWh number seems to be completely off base. I’m also thinking that it will have a 50kWh pack.

However, the rest of the report is focused on price per kWh. To get a 50kWh pack to <$10,000, you only have to get below $200/kWh. The analysis is pretty thorough, and this was a great recap.

Great Catch Dave. I was going to say the exact same thing.

I love to see the government’s theoretical attempts in predicting the future of EV batteries. But, I trust the predictions of someone like Elon Musk much more, since he’s already been doing it for two EV generations. I would prefer Argonne Labs use their knowledge and computing power to identify a new battery chemistry that could double the current energy density. That would be a real contribution to the future.

Fantastic article George.

If there is one thing I might add, is that in the last quarterly Tesla conference call, Musk stated that there is one battery step-change before the cheaper car gets released. I’m guessing that step- change is likely to be to Li-S batteries in mid 2016. I believe that the lower cost of materials for Li-S batteries, as well as their performance (350Wh/kg, 800Wh/l), will win out over other highly evolved Li-Ion chemistry (Envia, et. al.) on a cost basis, even as it provides 30% less weight and 25% less volume.

Good job George! I’m glad your head didn’t explode, assimilating all that data – mine would have 🙂

Fantastic article. Thank you George for summarizing the report.

One thing that I have been wondering about is how Elon can claim the E will cost so little. Incremental downsizing from the Model S and stripping out some of the tech doesn’t get us to that goal (though it helps). Until now, it had to be an article of faith that Tesla would figure it out. The report shows that there is a path that could lead us there.

I am also somewhat surprised about the fire characteristics of the various chemistries. I hope Tesla is paying a lot more attention to that for the next generation and maybe even for the next big rev of the Model S.

Looking at a typical market applications chart:

This from 2010. The EV applications as a percentage of the total lithium battery market was less than %1. Thus, the cost advantage was with consumer lithium batteries, and hence the advantage to Tesla for using packs made from those batteries.

Will that change completely in 3 years?

On the giga factory price side, I think the past could hold an answer. When the first commercial nuclear reactor started to produce its electricity it got its uranium 235 from an enrichment facility paid for by the military. Likewise it would be possible to make humvees powered with batteries builded in a giga factory paid for by the military. From then on the giga factory could also start supplying the car manufacturers.
On the 200 miles range side, I really doubt that this is sufficient to replace a standard car. Take a Prius for instance it has 400 miles of autonomy, not 200 miles. If the car is to be a pure EV, 200 is not enough, it needs to be 400 or it must be equipped with a range extender of some type. An Aluminum air single use cell, a liquid fuel based combustion unit or something else, but it needs something.

@Priusmaniac: I disagree. 200 miles is plenty of range for day to day driving. For the 5% long distance use case, by 2017 Tesla will have blanketed North America, Europe and parts of Asia with SuperChargers. That will take you between cities if you want to do it, and given Tesla has already increased SCs from 90 kW to 120 kW, with discussions to increase the speed to 135 kW and beyond, it doesn’t appear to make sense to carry the extra batteries every day when you just need the distance periodically. Most people will trade a free fillup for waiting around 15 to 20 minutes, especially when you have to stop to eat anyway.

200 is more than enough for EVs to hit “mainstream” (aka 100k+ sales for Tesla’s Model E). 400 miles is for people who go on road trips frequently and they don’t make up much of the population (they can chose a hybrid).