Model S Battery Pack - Cost Per kWh Estimate

[Grendal]

Great job CO. The only question I have is about Elon's comments:

A few months ago Elon mentioned that he thinks battery prices will fall below $200 per KWh in about 3-4 years. Just recently he said that prices will fall below $100 within 10 years. Why would he make such comments if the pricing is already there (almost there for the latter)?

Why would he advertise to the world the price that Tesla is paying? It' makes a lot of sense for him to quote the worldwide average and NOT his prices. The rest of the automotive world is 4 to 5 years behind where Tesla is now. It makes sense that they are also 4 to 5 years behind on pricing too. I have no problem believing that Tesla is paying around $200 per kWh and I hope they are paying less.

Nice analysis, CapitalistOpressor.

Thanks.

 

[sleepyhead]

Does anybody have any experience with what Solar City charges a retail customer for a Tesla sourced grid storage battery pack? What's the retail $/kWh for that? That could give an upper limit to the price range, since there is presumably some margin going to Solar City as well as margin for Tesla on the battery.

Here is an article that I found that says they charge $2,000/KWh, but when you include SGIP and ITC it reduces the effective cost to about $800/KWh:

The Numbers Behind Tesla and SolarCitys Home Energy Storage Play : Greentech Media

In a different article in the comments section, I saw someone write that they installed a PV system through SolarCity and got offered a 10KWh system for an additional $10/month. This sounds like a steal of a deal, so I highly doubt this number.

 

[CapitalistOppressor]

So, I need to read this in much more detail (and the links) but one thing that comes to mind is that Tesla has said they will sell you an option to replace your pack in 8 years for $12k. Granted, they haven't actually sold anyone this option yet; but, if your estimate that today's packs cost $14k is correct, then THAT IS A RIP OFF.

Now, I don't suppose Tesla is actually ripping off its customers. I suspect that there is a reason they have not yet begun selling these replacement pack options. I suspect it has something to do with battery swapping, so you are probably getting a lot more for that $12k than just a pack in 8 years.

Well, it turns out I have a free hour I didn't expect, and access to an actual computer, instead of just my phone, so lets march this discussion forward a bit.

To start with, these costs were announced by Tesla in Q3 of 2012, which is exactly the time that the IEK report, published at the end of Q4 (but referencing events in Q3), discussed an ongoing collapse in battery prices, that was resulting in costs of $120-$200/kWh. So the assumptions at the time that Tesla set their future battery prices would have been based on costs earlier in the year.

Furthermore, until very, very recently, there were negative returns on saved cash as a result of extremely low interest rates. So $12,000 stored by Tesla in 2012 was going to be worth less than $12,000 in the future.

Also, $12,000 isn't such a bad deal for a battery. Even if the cells were $100/kWh, that adds up to $8,900 in my simulated 89kWh pack, with an additional cost to integrate them into the pack of ~$2,000. When you total that together you get ~$11,000 just in manufacturing cost. Paying $12,000 at the consumer level is hardly a rip off.

So setting that concern aside, lets take a quick look at the current 18650 ecosystem to try and better characterize current pricing.

Here is an unambiguous price for a small quantity of 3000mAh cells from a Chinese knockoff brand (Ultrafire) -

Wholesale GLL138 Protected Ultrafire 18650 Rechargeable Lithium lon 3000mAh battery with PCB for LED Camer, Free shipping, $1.2-1.23/Piece | DHgate

The per cell price is unambiguous. $1.20 per cell, with free shipping to the U.S. via Fed Ex if you purchase between 1,000-2,000 cells. The price rises to a whopping per cell price of $1.23 if you order fewer than 1,000 cells. If you purchase in industrial volumes these were much less than $1/cell (I don't have the link on this computer).

Here is a next generation 4200mAh cell from the same company selling for $2/cell -

Tesla Motors Club - Enthusiasts & Owners Forum

Probable Ultrafire knockoff 4000mAh battery selling for $1.08/cell if you purchase 200 batteries. Again, these are far below $1 if you purchase at industrial quantities.

Wholesale 18650 3.7V 4000mAh Rechargeable Battery 18650 battery, Free shipping, $1.08-1.15/Piece | DHgate

Here is another counterfeit Ultrafire 3800mAh selling advertised for between $0-$1/cell -

Tesla Motors Club - Enthusiasts & Owners Forum

Here are actual NCR18650B 3400mAh cells advertised starting at $1/cell (presumably you need to order vast quantities) -

Tesla Motors Club - Enthusiasts & Owners Forum

Here is what appears to be a Panasonic 3100mAh cell that has been wrapped in a new case and is being sold under a house brand -

Tesla Motors Club - Enthusiasts & Owners Forum

Note: This could well be a high performance counterfeit. But the $1-$3 per cell price stands, and the performance specs seem to be on par with the Panasonic cell, and there are numerous references to it being a Panasonic cell.

Here is yet another company advertising the Panasonic 3100mAh cells for $1 -

Tesla Motors Club - Enthusiasts & Owners Forum

And another -

Tesla Motors Club - Enthusiasts & Owners Forum

Again -

Tesla Motors Club - Enthusiasts & Owners Forum

Sorry, I don't feel like fixing the text that TMC inserted for the links. The point of all of this is that paying $2 for a name brand Panasonic NCR18650A 3100mAh cell (which is a less advanced cell than the NCR18650B's) is not something that we should be batting away out of hand. Competing cells are going for much less than $1 at industrial quantities.

Also, the many, many advertisements for $1 Panasonic cells are much easier to find than advertisements for $2 cells. I was attempting to be conservative in the OP, because I could not get anyone to give me a simple quote. But the emphasis should be on "conservative" when discussing $2 cells.

- - - Updated - - -

Just to be clear, Ultrafire seems to be a major generic brand in the 18650 market. I see their cells everywhere on the internets. When thinking about Panasonic/Ultrafire, it might be helpful to think about Intel/AMD.

 

[brianman]

One thing to consider is that at the time when we all thought SuperCharger meant Chargers + Solar Panels, we had a number of: $250'000
Now that we think SuperCharger means Chargers + Solar Panels + 0.5MW Batteries - we have a new number: $300'000.

So could the $50'000 difference be essentially the amount needed to buy 500kWh? (So if these are 85kWh packs, then $8'500 per pack. But they don't need to be packs.)

Elon has said that adding Grid Storage to a SuperCharger location adds about 150K in cost

Your numbers don't match my recollection. I'll try to dig up the quotes.

Quote 1
[I still need to find this one. It might have been at the night-before-SC-announcement interview.
Full Video of Tesla and SpaceX Head Elon Musk at D11 - Liz Gannes - D11 - AllThingsD
I didn't find it here though.]

Quote 2

Cost is approximately what we first estimated, roughly $150K per station in expenditures without solar. And another $150K with solar.

Quote 3

Cost of SC is $150k without solar, maybe $300k with solar on average. Some more, some less.

 

[deonb]

Your numbers don't match my recollection. I'll try to dig up the quotes.

From my quote I assume you mean the original cost vs. new cost? (I agree with the 150k charger + 150k solar, with grid hidden somewhere in there).

The original cost came from the SuperCharger announcement in September of last year.

Here's at least one reference to that:
http://www.businessweek.com/news/20...-000-chargers-for-model-s-drivers-in-highways

Of course it might just be that they're bigger now, and Grid storage has been part of the original $250'000 as well.

 

[techmaven]

The IEK report is based on data that is from Q3 of 2012, so its almost a year old. There might be a few manufacturers that exceed Tesla's volume, but that is not going to substantially affect the price.

But then you appear to go off the rails by projecting that higher per kWh price you derive for the 60kWh pack back up to the 85kWh pack. How does that work, exactly?

Ah, yes, I should have explained that better. There are components that are the same between the 60kWH and the 85kWH pack. Therefore, the $8,000 upgrade price from the 60kWH to the 85kWH pack does not reflect the pricing of those components. Therefore the $320/kWH pricing from the upgrade only captures the increase and therefore does not reflect the true price of the pack as a whole. As a result, I modeled a higher $/kWH for the whole pack.

Furthermore, those changes almost certainly reduce production costs for Panasonic, because a large number of complex components are simply deleted.

The key is obviously just how much does that actually save. Just having to make Tesla special production runs could increase costs. Shipping to Fremont may actually be a significant cost depending on what % had to be air freighted. I therefore took a far more conservative approach to the pricing and I think the truth is between our two numbers.

The safer assumption is that Tesla is building a completed pack for $200/kWh. I think the evidence is more than strong enough that anyone who wants to claim a per-kWh cost of more than that is going to have a tough time.

Pack integration costs are definitely significant, but at least it is mostly within the control of Tesla themselves. It may also be already highly automated and therefore the incremental cost may be low. Since this is part of the secret sauce, I doubt we're going to get a clearer picture of this until much later. As a sometimes TSLA shareholder, I'm not sure I actually want the company to disclose that much of the secret sauce, so we may have to be content with very rough guessing.

 

[gregincal]

Ah, yes, I should have explained that better. There are components that are the same between the 60kWH and the 85kWH pack. Therefore, the $8,000 upgrade price from the 60kWH to the 85kWH pack does not reflect the pricing of those components. Therefore the $320/kWH pricing from the upgrade only captures the increase and therefore does not reflect the true price of the pack as a whole. As a result, I modeled a higher $/kWH for the whole pack.

However, margins on options for cars are typically much higher than margins on the base car, so this could easily offset the pack integration costs in your model. That might not be the case here, but I wouldn't just assume the margins are the same.

 

[CapitalistOppressor]

Ah, yes, I should have explained that better. There are components that are the same between the 60kWH and the 85kWH pack. Therefore, the $8,000 upgrade price from the 60kWH to the 85kWH pack does not reflect the pricing of those components. Therefore the $320/kWH pricing from the upgrade only captures the increase and therefore does not reflect the true price of the pack as a whole. As a result, I modeled a higher $/kWH for the whole pack.

Ahh, I see what you meant then.

The key is obviously just how much does that actually save. Just having to make Tesla special production runs could increase costs. Shipping to Fremont may actually be a significant cost depending on what % had to be air freighted. I therefore took a far more conservative approach to the pricing and I think the truth is between our two numbers.

There is no need for a separate production run. The only difference is the cap, which is manufactured separately, and then integrated into the cell. The cap of a conventional cell is a highly complex device, with several safety systems to prevent the cell from exploding during use. The cap of Tesla's cell is an aluminum disk that has a gasket ultrasonically welded to it. (Edit: Oops, probably not welded this way, as someone pointed out. Looking at the diagram and photo's, my guess is that it is applied in a press. But thats a guess. However, the cap itself is ultrasonically welded to the tab during cell assembly, which is clearly stated in the patent.)

A Tesla "production run" consists of the worker ultrasonically welding a Tesla cap to the battery at that stage of the production process. This amounts to reaching into a separate bin for the appropriate part.

Here is a video that demonstrates the manufacturing process. I'd suggest watching the whole video to understand how the films are made for the internal guts, but the processes relevant to the cap start around the 3:00 mark. After the necking process, you integrate the cap, but in the video, they are demoing the machines instead of production sequence. You'll see the cap integration maybe a minute later during the spot welder demonstration.

Again, at this point you just reach into a different parts bin. And the Tesla part is vastly cheaper.

Before viewing the video, please look at the relevant patent diagrams. Here is the conventional battery -

Cell cap assembly with recessed terminal and enlarged insulating gasket - diagram, schematic, and image 02

And here is Tesla's battery -

Cell cap assembly with recessed terminal and enlarged insulating gasket - diagram, schematic, and image 01

The cap is a very expensive part of the battery (possibly the single most expensive part), as it takes large numbers of tasks to complete. If you look at the whole video, you will see how the films for the internal guts are just an automated rolling process, and you add chemicals. These cores should be incredibly cheap.

 

[techmaven]

The cap is a very expensive part of the battery (possibly the single most expensive part), as it takes large numbers of tasks to complete. If you look at the whole video, you will see how the films for the internal guts are just an automated rolling process, and you add chemicals. These cores should be incredibly cheap.

Are you sure that the cell cap assembly is all that expensive? I thought that expensive part related to the cap is where there is a requirement for a battery protection circuit module welded on top of the cap and shrink wrapped together - but that is only necessary if one is going to use a battery without a battery controller like in flashlights. Thus a battery vs. a cell. Therefore you have 18650 batteries from manufacturers other than Panasonic that incorporate a Panasonic cell with a welded protection module and they sell that as a complete battery like this:

PANASONIC 18650 PROTECTED NCR18650A 3100mAh FREE Shipping from Florida-USA. 18650 cell-Made in Japan, Protection ICs-Made in Japan

It would be really nice if Tesla was getting a sweetheart deal with the raw cells, especially if they have smart ways to reduce the costs. I do wonder what the hold up is with respect to the NCR18650B.

 

[stopcrazypp]

For people asking why Tesla doesn't just switch to a 3400mAh cell (NCR18650B or similar) in the middle of Model S production (assuming Tesla is not using it in the start of production), switching to a more energy dense cell changes the weight of the pack and thus would require new crash testing and EPA testing. It also changes cell count or the capacity of the modules which would require readjustments in the BMS and also new R&D. This type of R&D is apparently costs enough for Tesla to not develop a 40kWh pack.

You also have to consider that for Tesla to get a good price, they would have to sign a volume purchase contract and they would have to finish that contract first before moving to different cells. Also the 3400mAh cells cost more per kWh so it's more cost effective to use 3100mAh ones at this point.

 

[CapitalistOppressor]

Are you sure that the cell cap assembly is all that expensive? I thought that expensive part related to the cap is where there is a requirement for a battery protection circuit module welded on top of the cap and shrink wrapped together - but that is only necessary if one is going to use a battery without a battery controller like in flashlights. Thus a battery vs. a cell. Therefore you have 18650 batteries from manufacturers other than Panasonic that incorporate a Panasonic cell with a welded protection module and they sell that as a complete battery like this:

PANASONIC 18650 PROTECTED NCR18650A 3100mAh FREE Shipping from Florida-USA. 18650 cell-Made in Japan, Protection ICs-Made in Japan

It would be really nice if Tesla was getting a sweetheart deal with the raw cells, especially if they have smart ways to reduce the costs. I do wonder what the hold up is with respect to the NCR18650B.

I don't know how expensive it is, but it is by far the most mechanically complex part. The rest is mainly just the jelly roll, which is an electrode sheet, anode sheet, and a separator, that is wound up on the winding machine.

According to Tesla's patent, here are the two main safety systems that the cap in the conventional 18650 cell contains -

CID

In a conventional cell, cap assembly 105 is a relatively complex assembly that includes multiple safety mechanisms. In cell 100, tab 109 is connected to assembly 105 via a current interrupt device (CID). The purpose of the CID is to break the electrical connection between the electrode assembly and the positive terminal if the pressure within the cell exceeds a predetermined level. Typically such a state of over pressure is indicative of cell overcharging or of the cell temperature increasing beyond the intended operating range of the cell, for example due to an extremely high external temperature or due to a failure within the battery or charging system.

PTC

PTC 129 is designed such that its resistance becomes very high when the current density exceeds a predetermined level, thereby limiting short circuit current flow.

 

[dmckinstry]

Well, the cable internet provider for our town was out for over a day, and I'm way behind.

Anyway, thanks for the analysis CO.

 

[blakegallagher]

Thanks for your in depth analysis CO .... very helpful ... I get the sense that your numbers truly are conservative and if they are not lower priced they will be by the end of the year. In regards to Elon stating the price in 4 to 5 years that was the "industry price" for sure not his

 

[CapitalistOppressor]

@CO

Great article. Best I've read so far in drilling down on wholesale cost to Tesla to make these batteries.

Couple of points:

1. The 147M batteries should be updated to include updated production slated for >21000. 167M would account for the update.

2. Total batteries used will have to account for the 1/2 mwh available at each Supercharger. I'd love to see a calculation of how many batteries Tesla is going to need just for the initial 100 Supercharger stations. What about when they have 200-300 worldwide in 2014?

3. Under one scenario of a typical swap situation, a new battery pack option might be available at 100kw. Then you might add 20 conventional 100kw packs for each of the 75 service centers for 60kw and 85kw to "swap" for long trips. How many more batteries would that require?

4. Do these changes make any difference in calculations for cost?

Sorry I missed this before. My models for how CES requirements might affect production consistently show that you only need a relatively tiny number of additional batteries relative to the core automotive production.

The reason is that roadtrips which will use CES capacity (either from SuperCharging or SuperSwapping) are rare. Perhaps only 2.5% of vehicle days are spent on these kinds of trips (although they represent a fairly large fraction of total vehicle miles).

So in a CES scenario, you only maybe need 3-6% additional production above what is already in the cars. This is one reason why it is affordable for Tesla. Even if a finished pack is costing $17,000, building enough packs to cover your 6% requirement means a per car cost of ~$1,020 to provide the storage capacity to support roadtrips. Tesla is charging $2,000 for access to the SuperCharger system (which also needs to pay for chargers, solar, etc).

These numbers shift around quite a bit depending on how you design the CES system, but you never get big increases in battery production as a result.

- - - Updated - - -

Thanks for your in depth analysis CO .... very helpful ... I get the sense that your numbers truly are conservative and if they are not lower priced they will be by the end of the year. In regards to Elon stating the price in 4 to 5 years that was the "industry price" for sure not his

The IEK report described an ongoing collapse in prices starting in Q3 of 2012, with a quoted range of $120-$200/kWh at the time (these low prices were already enough to describe it as a collapse).

Right now you can easily find competing 18650 batteries that are purchasable in small quantities (not industrial scales) for ~$70kWh. In my opinion, this is very similar to what we saw with the solar price collapse which put Solyndra out of business.

I very much believe that Tesla is already on the verge of delivering a pack for under $100/kWh. If the multiple advertisements for $1 NCR18650A 3100mAh Panasonic cells are correct, then Tesla is probably able to build the Model S pack right now for less than ~$110/kWh. So if not now, then by 2014.

 

[Citizen-T]

Well, it turns out I have a free hour I didn't expect, and access to an actual computer, instead of just my phone, so lets march this discussion forward a bit.

To start with, these costs were announced by Tesla in Q3 of 2012, which is exactly the time that the IEK report, published at the end of Q4 (but referencing events in Q3), discussed an ongoing collapse in battery prices, that was resulting in costs of $120-$200/kWh. So the assumptions at the time that Tesla set their future battery prices would have been based on costs earlier in the year.

Furthermore, until very, very recently, there were negative returns on saved cash as a result of extremely low interest rates. So $12,000 stored by Tesla in 2012 was going to be worth less than $12,000 in the future.

Also, $12,000 isn't such a bad deal for a battery. Even if the cells were $100/kWh, that adds up to $8,900 in my simulated 89kWh pack, with an additional cost to integrate them into the pack of ~$2,000. When you total that together you get ~$11,000 just in manufacturing cost. Paying $12,000 at the consumer level is hardly a rip off.

I'm not so sure. If the battery packs are $14k today, and we take the low end of Elon's 8-10% yearly price/performance improvement projection, then we arrive at a cost of something like $7k for a pack in 8 years.

It is true that there was a negative return on stored cash until recently, but I'm sure in coming up with this price they didn't assume that would continue indefinitely or that it would offset battery cost improvements.

Furthermore, I seem to remember that Tesla gets to keep your old battery when they eventually do the upgrade which would still retain some value (quite a but actually). Let's assume it isn't much, how does $2k sound? If we subtract the residual value of the old battery from that $7k then we are looking at a battery that costs Tesla net something like $5k. Again you paid $12k and tied your cash up for 8 years.

And that was using conservative numbers. This makes me think that there is more to this story.

Sent from my Droid RAZR using Tapatalk 2.

 

[CapitalistOppressor]

I'm not so sure. If the battery packs are $14k today, and we take the low end of Elon's 8-10% yearly price/performance improvement projection, then we arrive at a cost of something like $7k for a pack in 8 years.

It is true that there was a negative return on stored cash until recently, but I'm sure in coming up with this price they didn't assume that would continue indefinitely or that it would offset battery cost improvements.

Furthermore, I seem to remember that Tesla gets to keep your old battery when they eventually do the upgrade which would still retain some value (quite a but actually). Let's assume it isn't much, how does $2k sound? If we subtract the residual value of the old battery from that $7k then we are looking at a battery that costs Tesla net something like $5k. Again you paid $12k and tied your cash up for 8 years.

And that was using conservative numbers. This makes me think that there is more to this story.

Sent from my Droid RAZR using Tapatalk 2.

Yes. The key thing to keep in mind is that the price for the upgrade was set before or during the price collapse, and I'm still not sure how certain we can be of the price that Tesla is paying right now. Also, I don't think Tesla has actually implemented the program they announced. They might have been surprised by events and are re-evaluating the policy. The fall in the battery prices might also be a reason to finally decide to finally implement the swapping system that they've been thinking about doing.

Anyways, I've been making the case that prices might be very low right now, but I have not been able to find an authoritative source that will confirm the cost of the stock versions of the cells that Tesla is using. There are a ton of ads on Alibaba that make me think they are $1-$2/cell, and the prices for competing cells list clear prices showing higher power densities (3800mAh-4200mAh) at prices that are less than $1 per cell.

The IEK report was released in December, and referenced prices as low as $120/kWh in Q3 of 2012. The 3800mAh Ultrafire cells are running for half of that price right now, and Chinese companies making counterfeit Ultrafires are even lower. That is just a massive fall in prices over the course of less than a year. It's just nuts.

I assume that explains why higher quality, lower density 3100mAh cells from Panasonic might be going for $1, but the implications are just staggering. High quality cells selling for ~$90/kWh would make ICE uncompetitive. It's such a big story that I find it really hard to believe.

Edit: Also to respond more directly to the concern over replacement packs.

Recall also that at the time Tesla announced that policy they were on the edge of death, and implementing all kinds of policies to try and gin up revenue, including making service contracts mandatory.

Getting an extra 12k per customer might have seemed like a good idea. It also generated a lot of positive media reports, because whether it was going to be profitable for Tesla or not, it provided a lot of reassurance that you would not be stuck shelling out $30k for a new battery. So a lot of moving parts went into that announcement.

 

[Bgarret]

Funny that the Barron's article references that the current battery costs were unlikely to drop from the current $400/kWh to $200/kWr by the time of Gen III. There was a quote by a GM battery guy hypothesizing 20-40% drops from current prices (though I'm not sure if the Barron's folks had the wisdom to ask him what were the current costs). If CO is correct, and batteries are about to be $90-100/kWr, then Tesla and EV's are about 5-10 years ahead of what is being passed off as conventional wisdom. Can't wait for that Genie to come out of the bottle.

 

[CapitalistOppressor]

Funny that the Barron's article references that the current battery costs were unlikely to drop from the current $400/kWh to $200/kWr by the time of Gen III. There was a quote by a GM battery guy hypothesizing 20-40% drops from current prices (though I'm not sure if the Barron's folks had the wisdom to ask him what were the current costs). If CO is correct, and batteries are about to be $90-100/kWr, then Tesla and EV's are about 5-10 years ahead of what is being passed off as conventional wisdom. Can't wait for that Genie to come out of the bottle.

I just read the article. Screw Barrons. I'm gonna do an article with the data I have. Elon was right to hang up on them.

- - - Updated - - -

Important quotes from Elon -

"I have no interest in an article that debates what we consider to be an obvious point -- which is that there is a dramatic reduction in battery costs," Musk said, after a few questions. "You clearly do not understand the business. My apologies. I am terminating the interview."

Before ending our interview, Musk said that improvements will cut the cost of the Model S's battery to $10,000-$12,000.

- - - Updated - - -

Note the use of "there IS a dramatic reduction in battery costs".

 

[marvinat0rz]

@CO, thanks for some really great and insightful analysis.

This forum has a lot of incredibly good digging. But there is one question I keep asking myself - does it really help Tesla's case that we're all digging and unearthing Tesla's competitive advantages and pointing out where the existing auto industry is wrong? Part of the success of Tesla hinges on the existing auto industry having their heads up their asses and not catching on that battery technology is, in fact, right on the border of making cheap electric cars viable. I would hate to see Tesla get more competition right away because this forum has helped educate them. Executives and board members can remain ignorant for a long time when there is disruptive change going on, but *someone* inside the established carmakers has got to see the writing on the wall right now.

This is not meant as criticism of your analysis, it's just that this worries me a bit - I have felt hesitant posting my thoughts occasionally because of this.

 

[lolachampcar]

marvinat0rz,
I've chewed on this point a lot and also tried to make sense of Elon's enthusiasm for other manufacturers getting deeper into the game. I think it all comes down to acceptance. If the majors march forward with this as fast as they possibly can, they will accelerate population wide acceptance of EVs at just about the time Tesla needs broad based support for GenIII. That line of thinking leads me to see majors going all in as a good thing.

As for the forums producing good research, you can bet the majors know how to pay consultants for just this kind of research. What CO is doing is educating those beyond the majors which might just have the net result of producing more converts.