2022 Model Y 4680 Structural Pack is "Amazing", Says Munro & Associates

Munro & Associates has just released an analysis of the structural pack for Tesla's 4680 battery cells, having just received a 2022 Model Y SR from Giga Texas.

In their analysis, Cory Steuben (President of Munro & Associates) and Julian Aytes (lead engineer) found that the car's front seats are directly mounted onto the structural pack itself, making the structural pack essentially the vehicle's floor itself.

"It's absolutely mind-blowing to be standing under a vehicle on a hoist and have absolutely nothing for the floor structure. To truly understand how amazing it is to see a vehicle with no floor and the seats mounted to the top of the structure on the pack, you have to go back more than years, but decades."

(Source: Munro & Associates)

The structural pack, including the seats and other components mounted to it, weighs 1,198 pounds, which is "incredible because in a couple of the other EVs we have, the batteries will weigh twice that. Just the batteries. No seat, no carpet, no trim."

According to Elon Musk, the structural pack is "the right overall architecture from a physics standpoint, but still far from optimized," which seems to be a modest take on the speed of Tesla's design improvements.

https://twitter.com/x/status/1544511999392071680

Despite minor manufacturing issues found by the team, the Giga Press casting machines, noted as the world's largest high-pressure die casting machines, have done an amazing job at astronomically decreasing both parts and complexity for the structure of the chassis itself.

"At Munro & Associates, we've seen the development of the automotive industry for the past 30 to 40 years. I've come from a background of benchmarking vehicles where you'd have hundreds of stamped parts where this front giga casting is, and hundreds of parts in the back. The level of refinement and integration is incredible. Tesla is not waiting to integrate the casting for multiple mounting features."

Overall, these improvements have come from decades of constant work and continuous improvement on Tesla's end, and it is clear to see why Tesla is so ahead of many other manufacturers in terms of vehicle and battery structure.

Steuben mentions that the next goal for their analysis is to remove the battery cover and be able to know how Tesla is securing the 4680 cells themselves and take a look inside the battery pack itself.

The full breakdown and analysis is linked below.

 

[sleepydoc]

Ok - I've been trying to get some actual facts on the 4680 batteries. The most solid information I've found is by Jordan Giesige on his Limiting Factor YouTube channel. InsideEVs also has an article that basically summarizes his findings.

Jordan got a 6 month old 4680 cell, had some people at UC San Diego tear it down and analyze it. From that tear down:

• the battery uses a dry battery electrode technology for the anode but not the cathode
• the battery uses a NMC (Nickel-manganese-cobolt) cathode
• the anode is carbon with no silicon
• the cathode is 20-25% thicker than that of the 2170, something he notes is industry-leading
• The case is 4-5x thicker than the case on a 2170 battery (5-600 µm vs 125-250µm) giving the cell a 10-15% weight 'penalty'
• They could not test the actual cell but they constructed some coin cells from the disassembled battery and estimated the energy density to be 272-296 Wh/kg
• Energy density for 2170 batteries is 263-276 Wh/kg

Previous rumors have claimed the 1st gen 4680 cells will be around 98 Wh/cell, 2nd gen 108 Wh and 118 Wh for 3rd gen. Based on a 355g cell weight that corresponds to 276, 304 & 332 Wh/kg energy density. These are only rumors but they do correspond to what Jordan found.

Jordan is also working on getting a battery from the car that Sandy Munro is tearing down to compare and analyze. hopefully he'll be able to actually do some electrical testing on that battery to give actual numbers rather than calculated and extrapolated numbers. (of note, he said in the video that he is paying $800 for the single battery cell!!!)

Jordan noted that if his calculations are correct, Tesla has achieved an equivalent energy density with the 4680 batteries despite using a thicker shell and no silicon electrode. He also noted that the 2170 batteries got a 5-10% boost from using a silicon anode.

As stated above, all of this was from a 6 month old battery. It's currently unknown how much the design changed prior to the production of the current batteries. we also don't know what the roadmap is for design evolution. I suspect they were able to create a battery with equivalent energy density and started production on that with plans to switch to a silicon anode in the future and possibly decrease the wall thickness.

Even though the battery has a thicker wall, making it heavier, if that allows for material savings in the overall battery pack it could yield a net decrease in vehicle weight. So far we don't have any evidence of this.

We also don't know what the peak current and internal resistance of these batteries is. The tabless design was supposed to decrease resistance thereby decreasing power loss, allowing for a larger cell form factor and higher currents. I'm sure this is something Jordan and many others are eager to test once they get an actual cell from Munro. Note that higher peak current doesn't mean more energy, just more power (power is speed of energy transfer.) What's more significant is it would mean lower energy loss for a given power and also lower overhead for cooling.

 

[Mrbrock]

Ok - I've been trying to get some actual facts on the 4680 batteries. The most solid information I've found is by Jordan Giesige on his Limiting Factor YouTube channel. InsideEVs also has an article that basically summarizes his findings.

Jordan got a 6 month old 4680 cell, had some people at UC San Diego tear it down and analyze it. From that tear down:

• the battery uses a dry battery electrode technology for the anode but not the cathode
• the battery uses a NMC (Nickel-manganese-cobolt) cathode
• the anode is carbon with no silicon
• the cathode is 20-25% thicker than that of the 2170, something he notes is industry-leading
• The case is 4-5x thicker than the case on a 2170 battery (5-600 µm vs 125-250µm) giving the cell a 10-15% weight 'penalty'
• They could not test the actual cell but they constructed some coin cells from the disassembled battery and estimated the energy density to be 272-296 Wh/kg
• Energy density for 2170 batteries is 263-276 Wh/kg

Previous rumors have claimed the 1st gen 4680 cells will be around 98 Wh/cell, 2nd gen 108 Wh and 118 Wh for 3rd gen. Based on a 355g cell weight that corresponds to 276, 304 & 332 Wh/kg energy density. These are only rumors but they do correspond to what Jordan found.

Jordan is also working on getting a battery from the car that Sandy Munro is tearing down to compare and analyze. hopefully he'll be able to actually do some electrical testing on that battery to give actual numbers rather than calculated and extrapolated numbers. (of note, he said in the video that he is paying $800 for the single battery cell!!!)

Jordan noted that if his calculations are correct, Tesla has achieved an equivalent energy density with the 4680 batteries despite using a thicker shell and no silicon electrode. He also noted that the 2170 batteries got a 5-10% boost from using a silicon anode.

As stated above, all of this was from a 6 month old battery. It's currently unknown how much the design changed prior to the production of the current batteries. we also don't know what the roadmap is for design evolution. I suspect they were able to create a battery with equivalent energy density and started production on that with plans to switch to a silicon anode in the future and possibly decrease the wall thickness.

Even though the battery has a thicker wall, making it heavier, if that allows for material savings in the overall battery pack it could yield a net decrease in vehicle weight. So far we don't have any evidence of this.

We also don't know what the peak current and internal resistance of these batteries is. The tabless design was supposed to decrease resistance thereby decreasing power loss, allowing for a larger cell form factor and higher currents. I'm sure this is something Jordan and many others are eager to test once they get an actual cell from Munro. Note that higher peak current doesn't mean more energy, just more power (power is speed of energy transfer.) What's more significant is it would mean lower energy loss for a given power and also lower overhead for cooling.

I know it’s hard to differentiate the battery material from the casing when discussing energy density but they shouldn’t include the casing in these calculations. That’s like saying an ant weighs 16# if you stuff it inside a bowling ball. If yiu are going to include the casing, then yiu need to include the total volume and weight of the entire battery pack. That means so,etching in the real world. 4680 could be great with advanced chemistry and tables design but it’s worthless in the end from a performance standpoint if they need to beef up the cell walls and overall pack structure for a net loss. Which is what I think we are seeing with 1st Gen packs in the MY AWD. Equal cell energy density which should equate to better pack density since there is less wasted space but the ticker walls, structural foam and potential other structural add ins have negated this gain.

Since part of the benefit of 4680 is cell density within the pack, you have to account for the whole pack. Comparing individual cells from a chemistry standpoint is great but it doesn’t show the bigger picture.

Start with a bare pack minus any removable electronics. Weigh it. Use total pack energy to calculate wh/kg. For example, if a 100kwh pack weighs 500kg, the density is 200wh/kg. Easy peasy lemon squeasy.

 

[AlanSubie4Life]

Jordan is also working on getting a battery from the car that Sandy Munro is tearing down to compare and analyze. hopefully he'll be able to actually do some electrical testing on that battery to give actual numbers rather than calculated and extrapolated numbers. (of note, he said in the video that he is paying $800 for the single battery cell!!!)

We’ll see:

Single "4680" Model Y Cell will come in a small glass jar in epoxy and a custom card that identifies the cell's uniqueness.

 

[Unclepaul]

4680 cells are designed to reduce Tesla's cost to manufacturer battery packs. Far fewer cells to produce, wire, package, install etc. They can produce more battery packs because they require fewer numbers of cell.

The benefits to consumers are yet to be realized.

 

[MP3Mike]

We’ll see:

Single "4680" Model Y Cell will come in a small glass jar in epoxy and a custom card that identifies the cell's uniqueness.

They said if you buying it for analysis that you could get it not potted in expoxy.

 

[AlanSubie4Life]

Previous rumors have claimed the 1st gen 4680 cells will be around 98 Wh/cell,

So is the consensus that it is 672 cells of about 104Wh each, or 768 of 91Wh each?

Or is it 768 of ~100Wh with a bunch of capacity locked out?

Has anyone hooked up SMT to look at voltages or is it not supported on this pack for one reason or another?

Seems like Munro alluded to some other people already having removed the foam, to answer these questions.

 

[sleepydoc]

I know it’s hard to differentiate the battery material from the casing when discussing energy density but they shouldn’t include the casing in these calculations. That’s like saying an ant weighs 16# if you stuff it inside a bowling ball. If yiu are going to include the casing, then yiu need to include the total volume and weight of the entire battery pack. That means so,etching in the real world. 4680 could be great with advanced chemistry and tables design but it’s worthless in the end from a performance standpoint if they need to beef up the cell walls and overall pack structure for a net loss. Which is what I think we are seeing with 1st Gen packs in the MY AWD. Equal cell energy density which should equate to better pack density since there is less wasted space but the ticker walls, structural foam and potential other structural add ins have negated this gain.

Since part of the benefit of 4680 is cell density within the pack, you have to account for the whole pack. Comparing individual cells from a chemistry standpoint is great but it doesn’t show the bigger picture.

Start with a bare pack minus any removable electronics. Weigh it. Use total pack energy to calculate wh/kg. For example, if a 100kwh pack weighs 500kg, the density is 200wh/kg. Easy peasy lemon squeasy.

On one hand, I agree with your comment about the casing. If all we were interested in was a single cell then there would be no question. Ultimately for Tesla what’s relevant is the entire pack and it’s performance in the car. That’s where engineering and design decisions come in to play.

The other thing to note is that if the internal resistance is lower as they claim, that would mean more usable energy per cell because less would be wasted and functionally more range for an equivalent combined total cell volume. We haven’t seen evidence of that yet but we are missing a lot of information, too.

The structural battery pack is a great idea, but it’s also possible that the additional weight required of the battery casings would negate the weight savings in the car structure.

 

[Mrbrock]

Tesla Model Y 4680 Found Just 20 Pounds Lighter Than 2170 Model Y

Tesla claims the combination of 4680 cells and structural pack brings significant weight savings, but that doesn't seem to be the case yet.

insideevs.com

 

[sleepydoc]

So is the consensus that it is 672 cells of about 104Wh each, or 768 of 91Wh each?

Or is it 768 of ~100Wh with a bunch of capacity locked out?

Has anyone hooked up SMT to look at voltages or is it not supported on this pack for one reason or another?

Seems like Munro alluded to some other people already having removed the foam, to answer these questions.

Yes, they mentioned that some other people have removed the foam but I’ve seen no other actual analyses or verifiable reports. At this point I’m waiting for the analyses from Munro and The Limiting Factor. The number of cells should be black and white.

I’ve seen a couple other anecdotal reports of the weight being similar and someone who was able to charge 0-90% in an hour and also noted there was no loss of power when the battery hit zero and no loss of regenerative braking at 100% which would support a larger total capacity with some of it locked out but this is a single anecdote with no other supporting evidence so I’m cautious about over interpreting it.

 

[MP3Mike]

So is the consensus that it is 672 cells of about 104Wh each, or 768 of 91Wh each?

Or is it 768 of ~100Wh with a bunch of capacity locked out?

I thought from what people saw on the tour that the AWD pack had 690 cells. (The LR/fully populated pack would have 828 cells.)

And the rumor is ~98 Wh/cell.

 

[AlanSubie4Life]

I thought from what people saw on the tour that the AWD pack had 690 cells. (The LR/fully populated pack would have 828 cells.)

And the rumor is ~98 Wh/cell.

What’s the series/parallel arrangement for that?

The EPA test docs suggest it isn’t a changed voltage, which implies 96s, but that could be wrong of course (slight change is possible). But they aren’t doing 138s I don’t think. Plaid is 110s.

Using 69 as the multiple seems like a joke?

 

[MP3Mike]

What’s the series/parallel arrangement for that?

The EPA test docs suggest it isn’t a changed voltage, which implies 96s, but that could be wrong of course (slight change is possible). But they aren’t doing 138s I don’t think. Plaid is 110s.

Using 69 as the multiple seems like a joke?

We don't know the series/parallel arrangement yet.

Are you putting it past Tesla/Elon that they would be using 69? Each bandolier shown has 69 cells. There are three bandoliers per module, and 4 modules per fully populated pack. They remove the two outside bandoliers from the AWD pack, so 10 bandoliers of 69 cells each.

I think we need to wait until Munro release more pack tear down details to get more information and/confirm what we think we know. (Or someone needs to get one and setup SMT to get some real data.)

 

[AlanSubie4Life]

Are you putting it past Tesla/Elon that they would be using 69

Not exactly, but I couldn’t make the voltage work. The voltage just doesn’t work, at least for any way I can think of. Is there something I am missing? I could see 69 being split into 3 groups of 23 series, which would sort of work. But that would lead to a low voltage pack (4x23 =92, rather than 96 (or 110 for Plaid)). So that would be a 92s9p (828). But that seems a bit of a high count given what we know.

I guess I haven’t seen the pictures of bandoliers removed so not sure how that would work when going to a smaller pack. This is accomplished with fewer cells in parallel of course. But from the description I don’t see how that could work. You have to drop in multiples of 92 and that won’t work with 69 unless you drop 3 sets of 92 (4 sets of 69). So maybe they remove a bandolier from each module (not just the outside), that would make sense, and fill in the empty space with structural cans?

I am probably just thinking about that wrong.

I think we need to wait until Munro release more pack tear down details to get more information and/confirm what we think we know. (Or someone needs to get one and setup SMT to get some real data.)

Yes, certainly the answer will be known soon. I was just curious if we could guess it. We do know there are four modules now (not five like Plaid) which helps narrow the solution space slightly.

 

[MP3Mike]

Not exactly, but I couldn’t make the voltage work. The voltage just doesn’t work, at least for any way I can think of. Is there something I am missing? I could see 69 being split into 3 groups of 23 series, which would sort of work. But that would lead to a low voltage pack (4x23 =92, rather than 96 (or 110 for Plaid)). So that would be a 92s9p (828). But that seems a bit of a high count given what we know.

Do we know the upper voltage for the AWD pack? (I haven't seen that.)

Tesla went with lower voltage packs for the Model 3 SR+ packs, and even on the smaller Model S packs, so I don't see why they wouldn't do that with the AWD pack as well. (The 4680 LR pack may be ~450 volts, or maybe even higher?, like the current Model S pack, so the lower voltage 4680 AWD pack may be a little under 400 volts?)

 

[AlanSubie4Life]

Tesla went with lower voltage packs for the Model 3 SR+ packs,

Did they? Thought it was 96s31p.

so I don't see why they wouldn't do that with the AWD pack as well.

Yes definitely possible. I just thought we had not seen a different voltage yet for 3/Y. Excluding LFP.

Anyway 92s is conceivable. But if composed of fully populated bandoliers of 69 not sure how we end up with a pack with at least 70kWh ( but not 828 cells which seems like too much given what Tesla has said). Can do 92x8 but not divisible by 69. Could dummy up a third of a bandolier per module and that would work. ((2X3 + 1x2) X 23 per module)

 

[ZenRockGarden]

We know that the MYAWD is a step slower than the MYLR, so a slightly lower total voltage would be possible on it, with the long term plan to add in the missing cells to get to full pack size and voltage....

 

[MP3Mike]

The next 4680 pack tear down video from Munro is available:

It shows the gap on the sides, but it appears that they are saying there are no missing bandoliers. So it is likely 816 or 828 cells.

 

[sleepydoc]

We know that the MYAWD is a step slower than the MYLR, so a slightly lower total voltage would be possible on it, with the long term plan to add in the missing cells to get to full pack size and voltage....

Do you mean the MY AWD LR is a step slower than the MY P? The long range model is AWD. Currently you can pay an extra $2k to get a performance boost on it and decrease the 0-60 time from 4.8 to 4.3 seconds vs 3.5 for the performance model.

 

[uscbucsfan]

Do you mean the MY AWD LR is a step slower than the MY P? The long range model is AWD. Currently you can pay an extra $2k to get a performance boost on it and decrease the 0-60 time from 4.8 to 4.3 seconds vs 3.5 for the performance model.

No, it's slower than the MYLR. It has a 5 second 0-60 and no option for Acceleration Boost at the moment.

 

[AlanSubie4Life]

So it is likely 816 or 828 cells.

I don’t see how 816 is possible if all the modules are the same.

That would be 204 per module, which is fine in theory.

But 816 has to be 102s8p (unless they went to a very high voltage pack, 136s6p, which seems extremely unlikely).

So if all the modules are the same that 102s is not divisible by 4. And obviously each module is not 102s2p; the BMBs are relatively low voltage.

828 would be a lower voltage pack (92s9p) (92/4 = 23 per module) which is arguably unfortunate. Thicker wires. And theoretically slower charging depending on charger current limits. Small differences though.

I am not sure we can tell from the number of cells how many real cells there are though. Obviously there are only so many possibilities and looking at the connecting structure (which may be possible from the video but no time) would eliminate some.

It’s probably possible to reason through the depop options and figure out which ones are possible, but have not bothered. It seems like depop would be easiest by just removing a parallel string and leaving number of series the same. (So 23 at a time per module, for example, could take the pack to 736, 92s8p, 4x 23s8p)

We’ll know soon enough! I’m sure someone has already figured it out without looking.