Tesla Battery Investor Day

[JRP3]

Not the large format cells I've seen. They do have a BMS but that can't stop a shorted cell.

 

[JRP3]

Here is a pic of the Bolt battery cell connections, these are pouch type cells with no internal safety mechanisms. Maybe they are counting on the tab welds acting like a fuse under excessive current, though it looks as if even if the welds burned through there would still be contact.

 

[mongo]

Other OEM's using larger cells don't bother with small bond wires as a safety feature. I have some BYD LFP packs using I think 180 Ah cells in series for 24V and some Toshiba Scib LTO packs using 20 Ah cells, (2P for 40Ah), 12S for 24V. Both high C rate chemistries using large bus bars for connections directly to the cells. I don't think they'd work well as fuses. I'm not sure I've seen anyone else using small wires the way Tesla does.

Don't other cells have internal fuses? Or other short circuit protection?

So here's the thing. Those packs are all 1P or 2P with cells the same Ah as the pack (or half). So the max fault energy is the same (or half) as a single cell fault. Tesla combines multiple cells in parallel, so the group fault current is multiples higher than a single cell fault. 100kWh Model S/X pack is 86p96s, one cell shorting itself (10.8Wh) versus one cell shorting 86 (928.8 Wh).

End result is Tesla's design provides lower fault current and adding fuses inline to the other designs (that have parallel cells) would not accomplish enough to bother.

With the 4680 at 5x the energy, there would still be 17 cells or so in parallel for a 100kW pack, so individual fusing drops fault energy/power/current by a factor of 10 or so (fuse tolerance).

 

[Cosmacelf]

So here's the thing. Those packs are all 1P or 2P with cells the same Ah as the pack (or half). So the max fault energy is the same (or half) as a single cell fault. Tesla combines multiple cells in parallel, so the group fault current is multiples higher than a single cell fault. 100kWh Model S/X pack is 86p96s, one cell shorting itself (10.8Wh) versus one cell shorting 86 (928.8 Wh).

End result is Tesla's design provides lower fault current and adding fuses inline to the other designs (that have parallel cells) would not accomplish enough to bother.

With the 4680 at 5x the energy, there would still be 17 cells or so in parallel for a 100kW pack, so individual fusing drops fault energy/power/current by a factor of 10 or so (fuse tolerance).

Huh. Yet more reasons why the other OEMs have gone with large cells. Small cells have their own challenges. I wonder if Tesla were started today whether they would still have chosen cylindrical. In other words, have large format cells caught up to cylindrical in terms of price/performance? Ignore Tesla's latest innovations, of course (you'd have to presume Tesla would have applied similar innovations to large format had they been working on those).

 

[mongo]

Huh. Yet more reasons why the other OEMs have gone with large cells. Small cells have their own challenges. I wonder if Tesla were started today whether they would still have chosen cylindrical. In other words, have large format cells caught up to cylindrical in terms of price/performance? Ignore Tesla's latest innovations, of course (you'd have to presume Tesla would have applied similar innovations to large format had they been working on those).

If anything, I though it showed a benefit to small cells. Limitation of fault energy is a big plus. Even if your cell never fails, in a crash, it could get shorted.

 

[Cosmacelf]

If anything, I though it showed a benefit to small cells. Limitation of fault energy is a big plus. Even if your cell never fails, in a crash, it could get shorted.

Now that you mention it, I recall that's one reason the other OEM engineers did not want to use Tesla's much more energy dense NMC cells. They were too volatile and dangerous. Tesla saw that problem and engineered around it. And you're right, engineering a safety fix for a much larger NMC cell wouldn't be easy.

 

[mongo]

Agreed that a common cooling plate couldn't service all the cells, however one could be attached to each parallel group.

But as I said in my original post, and your comment underscores, some overcurrent protection mechanism would have to be implemented...

Sure, but then you need as many cooling plate sections you have battery groups, and the cooling sections need to be separate from the pack. If the chemistry does not change greatly, that is 96 sections each with an inlet and outlet to connect.
Contrast that with a single cooling section and electrical separator. Or even make the bottom of the case is the cooling plate (not great when you want to heat the cells, but then so is having the drive unit as the heater)

 

[Doggydogworld]

With the 4680 at 5x the energy, there would still be 17 cells or so in parallel for a 100kW pack,...

More like 12 cells per group, right? 4680 should have 85-90 Wh each, so 1100-1200 cells in a 100 kWh pack

 

[ammulder]

When they talked about "X% more range" in the presentation... did they ever state the context for that? Was it more range for the same weight of pack? Or what?

Let's say all the improvements get made. I'm assuming in some cases they'd sell smaller packs with equivalent range rather than add more than some baseline range to their low-end vehicles. Maybe in other cases they'd fix the pack size or weight to minimize design changes. And in other cases maybe they'd hold the cost fixed and offer substantially the same vehicle with much greater range.

I just walked away confused by what "X% more range" means in practice.

 

[mongo]

More like 12 cells per group, right? 4680 should have 85-90 Wh each, so 1100-1200 cells in a 100 kWh pack

Oh! good catch, I was talking about the S/X with 86P to start, but using the 5X gain relative to the 2170. The 18650 to 2170 is an extra 50% gain or 2/3 the cells for 12 per group.

The 75kWh LR 3/Y would drop from 46P to 9P.

 

[JRP3]

Tesla's much more energy dense NMC cells.

NCA.

 

[MC3OZ]

https://twitter.com/YingShirleyMen1/status/1285699586485624833

I posted this link in the main thread but it is more relevant here...

My assumption is the Manganese-Nickel option referenced on Battery Day may be LMNO.

As Shirley says I recommend reading the paper...LMNO is new as such there scope for improvement.

These quotes from the article I found particularly interesting:-

The high-voltage spinel LNMO (LiNi0.5Mn1.5O4) has attracted wide attention since it can deliver a high mass-specific energy density and a high operating voltage (4.7 V) [13,14]. More importantly, it does not contain expensive cobalt or an excessive amount of lithium, which makes LNMO cost-effective and suitable for applications in the field of power batteries and large-scale energy storage
...
The high operating voltage may not be harmful to the LNMO cathode itself but causes severe decomposition of carbonate-based electrolytes and other cell components.

....
These post-mortem analyses indicate that the degradation mechanism of LNMO-Graphite full cell is not solely caused by the failure of the LNMO cathode; the graphite anode degrades even more after ultra-long cycling.

it is just a hunch but perhaps LMNO performs well in a 4680 cell with a silicon anode.
We are not sure the Manganese-Nickel 4680 option is LMNO, but it sure seems like an interesting area of further study...

There may be an electrolyte/anode combination that works well with LMNO, it ticks a lot of the other boxes. including price.

 

[JRP3]

It seems they still need to solve the high voltage electrolyte problem to work at full potential.

 

[micheljull]

Yeah, it doesn't tell that.
From the rest of the presentation, there is a 6x power increase for a 5x energy increase, which would seem to imply a potential 20% charge rate increase on a C rate basis.

Yes exactly, or a 1-5/6=17% charge time decrease

 

[micheljull]

The Y axis graph is labled "Supercharge Time Increase"
Both lines start at 0. That does not mean the cells charge instantaneously, nor at the same rate, but only that 21mm is the reference point for each type. From there, the tabless has mininal change in charge time as diameter increases, the tabbed has significant.

The tabbed has a much better electrical and thermal path, so how could it charge at the same rate as tabless at 21mm?

Right again. If the Y axis was labeled “Supercharge Time” the tabless curve would be shifted up and the tabbed line would be shifted higher up. It seems this perfectly explicit graph has left most people clueless !

 

[Christine69420]

I found this infographic about existing and planned battery cell production in Europe. Somewhere in the twitterverse - if anyone knows more about the source please post.

Elon Musk confirmed yesterday that Tesla will buy batteries from 3rd party producers - all they can get. So several of these would be potential Tesla suppliers.

Tesla in Grüneheide we know about.

Panasonic is already a supplier to Tesla. The Norwegian factory is in it's early planning stages. Counties are running all over eachother with site offerings. The 2000 jobs they will be offering are very popular.

CATL is also a Tesla supplier. Will their Erfurt plant sell to Tesla?

LG Chem is another Tesla supplier. Do their existing factory in Wroclaw make suitable batteries for Tesla?

Freyr in Mo i Rana plan to start building their pilot production line later in 2021. AFAIK they will sell ther product on the open market.
Source: FREYR Battery Norway | Decarbonizing transportation and energy…

Morrow in Arendal is in it's early stages yet. Planned completion is 2024. They will be making lithum-sulphur batteries.
Source: Morrow Batteries to build 32 GWh battery factory in Norway - electrive.com

Northvolt in Skellefteå/Salzgitter will start production this year. They will be supplying BMW.
https://northvolt.com/production/

 

[MC3OZ]

I found this infographic about existing and planned battery cell production in Europe. Somewhere in the twitterverse - if anyone knows more about the source please post.


View attachment 631545

Elon Musk confirmed yesterday that Tesla will buy batteries from 3rd party producers - all they can get. So several of these would be potential Tesla suppliers.

Tesla in Grüneheide we know about.

Panasonic is already a supplier to Tesla. The Norwegian factory is in it's early planning stages. Counties are running all over eachother with site offerings. The 2000 jobs they will be offering are very popular.

CATL is also a Tesla supplier. Will their Erfurt plant sell to Tesla?

LG Chem is another Tesla supplier. Do their existing factory in Wroclaw make suitable batteries for Tesla?

Freyr in Mo i Rana plan to start building their pilot production line later in 2021. AFAIK they will sell ther product on the open market.
Source: FREYR Battery Norway | Decarbonizing transportation and energy…

Morrow in Arendal is in it's early stages yet. Planned completion is 2024. They will be making lithum-sulphur batteries.
Source: Morrow Batteries to build 32 GWh battery factory in Norway - electrive.com

Northvolt in Skellefteå/Salzgitter will start production this year. They will be supplying BMW.
https://northvolt.com/production/

Tesla plans to have a run rate of 200 GWh of internal 4680 cell production by the end of 2022.

For example, this might be split as follows:- Fremont 10 GWh, Berlin 60 GWh, Austin 130 GWh.

So it is possible Tesla might have all the cells they need for Berlin and Austin vehicle production on site.

Eventually bumping Fremont up to 50 GWh would be enough to cover al Fremont production.

In China they might use LG and CATL as 4680 cell producers because sufficient cells are available at the right price.
IMO plans for China are the hardest to guess.

In Europe and the US they still can (and IMO will) use external cell providers for energy storage products for many years.
Occasional delays or shortages of energy storage cells are not as critical.

 

[Mo City]

Tesla plans to have a run rate of 200 GWh of internal 4680 cell production by the end of 2022.

For example, this might be split as follows:- Fremont 10 GWh, Berlin 60 GWh, Austin 130 GWh.

So it is possible Tesla might have all the cells they need for Berlin and Austin vehicle production on site.

Eventually bumping Fremont up to 50 GWh would be enough to cover al Fremont production.

In China they might use LG and CATL as 4680 cell producers because sufficient cells are available at the right price.
IMO plans for China are the hardest to guess.

In Europe and the US they still can (and IMO will) use external cell providers for energy storage products for many years.
Occasional delays or shortages of energy storage cells are not as critical.

Remember the big deal it was when Panasonic said Giga Nevada would ramp to 54 GWh?

Talk about getting left in the dust.

We can only guess how close Austin will be to a terafactory in 2025, or so.