Tech Improvements Unveiled on Battery Day

[kbM3]

Could you write out the math for me? I am genuinely confused how you arrived at these numbers (meaning I"m trying to figure it out myself and I haven't a clue how to do it).

Replied in another thread to not derail main thread.

Model 3 pack:
Cell count: 4416
Energy content : 80 kWh
Peak charge rate: 250 kW.

At peak charge rate (if it could be maintained) the pack would fill in 80 kWh / 250 kW = 0.32 hr.

New 4680 pack with same number of cells:
Cell count: 4416
Energy Content = 5 X 80 kWh = 400 kWh. ( 5X the energy)
Peak Charge rate = 6 X 250 kW = 1500 kW (6X the power)

At 1.5 MW the 4680 pack would fill in 400 kWh / 1500 kW = 0.27 hrs.

Looking at each cell:
2170 cell energy content = 80 kWh per pack / 4416 cells per pack = 18 Wh per cell.
2170 peak charge = 250 kW per pack / 4416 cells per pack = 56.6 W per cell.

4680 cell energy content = 5 X 18 Wh / cell = 90 Wh Per cell
4680 cell peak charge = 6 X 56.6 W = 340 W per cell.

So the math is the same as the pack calculation.

bottom line English. You can charge a 4680 cell six times as fast, but you’re trying to stuff in five times as much energy, so it just gives you 1.2X leverage.

 

[T3forMe]

I have a couple of questions;
1. since the new 4680 battery takes up 4.8 times the area (21mm dia. vs 46mm dia.), how many of these batteries can fit into say a model 3? Not that this battery will ever be destined for use in the model 3., but you cannot fit 4416, 4680 cells under the floor of this car.
2. Using the model 3 again: Each battery module (aka brick) contains 46, 2170 batteries connected in parallel. Each battery's anode is connected to the positive (+) plain using a fusible link. If one of the 46 cells were to fail by short circuiting, the fusible link opens, isolating that particular cell from the rest of the pack leaving that module is still functional, albeit one cell less. How would the new 4680 batteries be connected together. I have to assume that some batteries will be connected in parallel. If one of these cells shorts, the battery pack design cannot allow for the entire module (brick) to go down. Anyone have any ideas.

 

[SageBrush]

If we assume each horizontal line is $20 / kWh in the below graph

I was staring at that graph too, and I don't think it makes sense if the Y scale is linear. I may track down historical pack prices and curve fit.

 

[SageBrush]

You can charge a 4680 cell six times as fast, but you’re trying to stuff in five times as much energy, so it just gives you 1.2X leverage.

Much along the same lines, I was thinking about the ~ 5x reduction in internal resistance. My first thought was that this enabled crazy fast charging but then I remembered that the parallelism is reduced by ~ the same factor so if I am thinking about this correctly, total pack resistance and therefore charging profile stays about the same.

 

[SageBrush]

My mistake -- deleted

 

[SageBrush]

"This ends up giving you 20% longer range and it is cheaper than what Tesla currently uses. So, far cry from a theoretical 10x energy density improvement."

The 9x theoretical density improvement would occur if 100% of the electrode was silicone. A 0.2x energy density increase works out to ~ increasing the silicone fraction in the cell ~ 2.2%. This also explains why the $/kWh cost is low -- only a small amount of silicone is added.

 

[SageBrush]

I was unsure exactly what that slide said. Does it mean you get a range increase AND a cost decrease together? I guess it does, which is even more mind blowing.

If a current 100 kWh pack is good for 400 miles and costs $100*100, I think they are saying that they have a roadmap for a
150 kWh pack good for 600 miles that costs $150*50

 

[kbM3]

I have a couple of questions;
1. since the new 4680 battery takes up 4.8 times the area (21mm dia. vs 46mm dia.), how many of these batteries can fit into say a model 3? Not that this battery will ever be destined for use in the model 3., but you cannot fit 4416, 4680 cells under the floor of this car.
2. Using the model 3 again: Each battery module (aka brick) contains 46, 2170 batteries connected in parallel. Each battery's anode is connected to the positive (+) plain using a fusible link. If one of the 46 cells were to fail by short circuiting, the fusible link opens, isolating that particular cell from the rest of the pack leaving that module is still functional, albeit one cell less. How would the new 4680 batteries be connected together. I have to assume that some batteries will be connected in parallel. If one of these cells shorts, the battery pack design cannot allow for the entire module (brick) to go down. Anyone have any ideas.

If they redesigned the Model 3/Y (probably not for a while), they would use only about 1/5 of the number of cells. Increasing the cell size greatest benefit is reduced cost and faster production. Per vehicle you only need 1/5 the number of cells.

 

[Cosmacelf]

I have a couple of questions;
1. since the new 4680 battery takes up 4.8 times the area (21mm dia. vs 46mm dia.), how many of these batteries can fit into say a model 3? Not that this battery will ever be destined for use in the model 3., but you cannot fit 4416, 4680 cells under the floor of this car.
2. Using the model 3 again: Each battery module (aka brick) contains 46, 2170 batteries connected in parallel. Each battery's anode is connected to the positive (+) plain using a fusible link. If one of the 46 cells were to fail by short circuiting, the fusible link opens, isolating that particular cell from the rest of the pack leaving that module is still functional, albeit one cell less. How would the new 4680 batteries be connected together. I have to assume that some batteries will be connected in parallel. If one of these cells shorts, the battery pack design cannot allow for the entire module (brick) to go down. Anyone have any ideas.

There’s a lot Tesla didn’t tell us and series/parallel and fusible links were among that. IIRC we didn’t know how the original Tesla pack was constructed until someone got a salvaged pack and ripped it apart. We might have to wait for the same thing this time. It’ll be considerably harder to do.

 

[davidinmedia]

Great summary, thanks.

What I missed in Teslas presentation (or didn't get it) was actual figures about current price per kWh
and planned capacity for the plaid Model S.

Those graphs without units on one or both axis are more or less sales presentations/castles in the sky.
Not much worth, 50% of what...?

It seems Tesla has some great ideas (and is still far ahead of copmpetition), but the battery day had
a bit more entertainment character than being a "facts day" (->see what we have now!).
And when I hear that FSD is totally rewritten "and will work definitely" I know why I don't buy that feature.

Therefor it was a nice show and Tesla has great plans with it's terrawatt factories.
But all in all it was not overwhelming.

My guess it that Tesla is still a good 20-25% below competition in battery pack cost, due in part at least to quantities purchased. As to the Plaid, I'm guessing that it'll come with a 125 kWh battery using the new tech, with batteries coming from the pilot plant providing enough improvement in capacity to fit the new battery pack in the same space, possibly with some frame lightening. That would get you up to 520 mile range without additional operating efficiencies, which in any case would be hard to come by.

Thinking about it, those pilot-line batteries will get used somewhere, as the line matures, and the specs line up pretty well. Once they get to, say, a 10% run rate, that's still 1GWh/year, or 8000 125-kWh batteries. How many Plaid units will they sell? Maybe a couple of hundred? This would allow them to play with retrofitting the new batteries into the S chassis, as a precursor to the next generation.

 

[SageBrush]

Tesla has reset the bar so far that 10 GWh a year production in 2020/1 is ignored

 

[PSMangelsdorf]

VW is making a real EV from ground up. GM has legacy Frankensteins being sold on discount. Two different things.

In GM's defense, they have a pretty big backlog of ground up new platforms set to release in the next 3 years, and they seem to think the new Ultium batteries are a pretty big step forward. We'll see.

 

[SageBrush]

and they seem to think the new Ultium batteries are a pretty big step forward.

Unlike all the other GM inventions sitting collecting dust that they said were crap when announced

 

[Cosmacelf]

Elon Tweet today:

Q: elon, are you guys making 4680 cells with three different cathodes? or when you talked about a diversified approach were you talking about your external suppliers ?

Elon: Suppliers. We’re only doing high energy nickel ourselves, at least for now. Also, maybe the presentation wasn’t clear that we’ve actually had our cells in packs driving cars for several months. Prototypes are trivial, volume production is hard.

-----------------------------------

I infer from this that they don't have the intellectual property to do the more "normal" cathodes, like LiFePO4, or they are just concentrating their research on ultra high energy cells.

Also of interest that their cells are already being used in beta test cars. But as Elon said, those are just prototype cells - need to figure out how to make those cells cheap enough in high volume.

 

[Olle]

There’s a lot Tesla didn’t tell us and series/parallel and fusible links were among that. IIRC we didn’t know how the original Tesla pack was constructed until someone got a salvaged pack and ripped it apart. We might have to wait for the same thing this time. It’ll be considerably harder to do.

This was definitely the elephant in the room for the physical design part of the presentation. If you want to glue the battery to the top and bottom plates for stiffness and temperature control, how will you then separate them for the series connection to get to 400V?

A few different options:

• Make the whole battery pack 4V instead of 400V
• Divide the structural/thermal/electrical plates into smaller plates for each group and glue with dielectric glue to outer shell
• Make each cell case completely isolated from its electrodes

 

[Cosmacelf]

This was definitely the elephant in the room for the physical design part of the presentation. If you want to glue the battery to the top and bottom plates for stiffness and temperature control, how will you then separate them for the series connection to get to 400V?

A few different options:

• Make the whole battery pack 4V instead of 400V
• Divide the structural/thermal/electrical plates into smaller plates for each group and glue with dielectric glue to outer shell
• Make each cell case completely isolated from its electrodes

Not all cells need to be glued to top and bottom. They could have a honeycomb system where only a fraction of the cells are structural.

 

[Olle]

Not all cells need to be glued to top and bottom. They could have a honeycomb system where only a fraction of the cells are structural.

That's an option too.
If you were to go with the multi plate option (one for each voltage group) you need dielectric coolant (oil?) to not short out the plates or have electrically but not thermally isolating heat exchangers for each plate.
If you already need dielectric coolant, wouldn't it be easier to just flood the cells and not need to worry about thermal plates?

 

[Olle]

There’s a lot Tesla didn’t tell us and series/parallel and fusible links were among that. IIRC we didn’t know how the original Tesla pack was constructed until someone got a salvaged pack and ripped it apart. We might have to wait for the same thing this time. It’ll be considerably harder to do.

Perhaps we can think of the glued top and bottom plates as scaled up PCBs (Giga PCB ). The fuse action could be handled by a thinned out section of the PCB internal conductor per cell. How does that sound?

 

[Cosmacelf]

Perhaps we can think of the glued top and bottom plates as scaled up PCBs. The fuse action could be handled by a thinned out section of the internal conductor by each cell, inside the PCB. How does that sound?

Sounds good in theory. It all comes down to manufacturability and cost. Honestly, we are going to have to wait for tear downs.

 

[kbM3]

The energy content only went up 5X, where volume increased by 5.5X, denoting a volumetric energy density DECREASE for the cell. Does this mean the gains in range were purely gravimetric energy density increases? They do gain in footprint because of extra cell height and no cooling channels, so they can fit more energy in a pack.