100kWh pack - geometry, pack density and ... immersion cooling?

[Alketi]

I saw this mentioned elsewhere and it has some interesting implications.

Let's start with what we know:

- Tesla uses "coolant" running through a ribbon between the individual cells, as described in these patent drawings

As can be inferred from the drawings this limits the surface area that's in contact with the temperature controlling substance.

We also know that extended periods of high current draw cause thermal limit protections to kick in, i.e. the battery cells are not cooled fast enough to dissipate the heat they generate.

News from yesterday

- The 100 kWh battery pack uses the same cells, but a new module and pack architecture

- It's using a new cooling system and electronics, which JB Straubel described the upgrade as a “significant change”.

- Elon said the pack architecture is now "quite close to theoretical limit"

- Elon Musk described the new P100D pack as “complex” and limited to a quantity of 200 per week.

What this may imply

A significant change to the cooling system, plus a cell arrangement that's close to the theoretical limit.

To get a cell arrangement close to the theoretical limit, the space between the cells needs to be shrunk to a minimum. It logically follows that the tubing between the cells would be the limiting factor in reaching that theoretical packing limit. Thus, the tubing would need to be removed.

This leads to the possibility is that for the new 100kWh pack Tesla has switched over to immersion cooling with an electrically inert substance.

In home brew PC builds this is done, for example, by submersing all the electronics in a fish tank filled with mineral oil. At an industrial scale its accomplished with better chemicals, such as the 3M Novec product.

Immersion cooling/heating should bring two advantages -- increased pack density and increased cooling capacity. The second point might alleviate one of the long-standing criticisms from the performance crowd about "track capability", aka running at high speeds for extended periods of time. It would also bode well for the Model 3.

Thoughts?

 

[Fiver]

I wonder if this pack (the 100) is a test of the cooling system we'll see in the model 3 (with it's larger cells).

/edit Not test really, as clearly they believe it's ready for prime time in a production vehicle. I guess I mean I wonder if this IS the cooling system we will see in the 3.

 

[andrewket]

Thoughts?

A well thought out theory. I'd give it good odds that the cooling system has been changed, and slightly worse odds (but still good) that it's immersion based.

 

[Alketi]

I wonder if this pack (the 100) is a test of the cooling system we'll see in the model 3 (with it's larger cells).

/edit Not test really, as clearly they believe it's ready for prime time in a production vehicle. I guess I mean I wonder if this IS the cooling system we will see in the 3.

I read too many articles yesterday to have the exact quote in front of me, but Musk said something to the effect of what they learn building/deploying this will apply to the Model 3. [That's a paraphrase]

 

[RubberToe]

Since you have to cool the batteries, it makes sense to try to pack them in the smallest possible footprint while still cooling them. Agree that this seems likely for the Model 3. I wonder if the individual cells, instead of being up and down, might be lined up front to back. Cooling tubes running in the space between 4 proximate cells. Better cooling and closer spaced cells, no immersion required.

Side to side would work too, and keep the 16 modules in place.

RT

 

[JeffK]

I read too many articles yesterday to have the exact quote in front of me, but Musk said something to the effect of what they learn building/deploying this will apply to the Model 3. [That's a paraphrase]

There have been some other recent advancements in battery tech so maybe this is to see how well they handle real world charge and discharge rates and longevity.

 

[LargeHamCollider]

Immersion cooling would add unnecessary mass, also you'd still need the fluid to be moving. Here's a copy and paste from the ST investors thread that I think will clear some things up.

Actual tear downs and tests have shown the actual capacity of the 85kWh pack to be 81kWh under normal conditions, the 90kWh pack is almost certainly near 86kWh actual capacity. Electrek reports that the cells themselves have not changed. What HAS changed is how tightly the cells are packed together. The existing Tesla modules hold 444 cells and the cell containing area measures about 28cmx68.5cm or 1918cm^2. Circle packing theory tells us that the most dense arrangement of circles (and therefore cylinders) covers 90.69% or pi/(2*SQRT(3)) of an area. Maximum diameter of an 18650 cell is 1.86cm (No! It's not 1.80cm!) making cross-sectional area 2.72cm^2 which means that the theoretically maximal number of cells that can fit in a module is (1918cm^2*.9069)/2.72cm^2 ...which is 640 cells. Tesla only fits 444 cells in each module so in theory they could increase the pack capacity by (640/444-1)*100% or 44%... That's a lot of room for improvement.
Tesla did not update the range number of the P90D like they did for the 90D after the facelift so I think using the Model X' increase in range will give us the best estimate of the capacity increase. 289/250 = a 15.6% increase in capacity, this should correspond very closesly to the increased cell count with it being likely that the extra mass causes a small hit to efficiency. Tesla's current packs are 96s/74p and I think it's likely that the new packs will want to keep the voltage the same so they will stick with 96s. That means that we're looking for a multiple of 96 that is approximately 16% greater than seventy four. 74*1.16 is 85.84. This points to the new pack being 96s86p, which (I just checked) is exactly what Yggdrasill suggested above. I did not rely on Yggdrasill's number at all in my estimate because his module dimensions are wrong, that said we came to the same conclusion... I believe it is the correct conclusion with >95% confidence.

We know the actual capacity of the 90 packs is about 86kWh, I spent a long time trying to verify this after they were launched, multiply 86 by 86/74 and you get (drum roll) 99.95kWh! So Tesla is no longer rounding stated pack capacity wayyyyy up. This is good. This means that the 100 pack will produce 16.2% more power AND it will charge 16.2% faster at all SoCs. I think this is a pretty big deal...

This also means that the 100D should get 340 miles range... which is more than a Hellcat, it should beat a Hellcat to 60 too unless the Hellcat has a pro driver or racing slicks...​

 

[jdevo2004]

I suspect the cooling solution looks a lot like a pocket air mattress. Where the pockets are filled with batteries and the air mattress is filled with coolant. Of course the holes are much closer together for the battery pack version. I also suspect that that both the positive and negative terminals are routed to the top of the battery to allow the bottom of the battery to be cooled as well.

 

[emir-t]

Immersion cooling would add unnecessary mass, also you'd still need the fluid to be moving. Here's a copy and paste from the ST investors thread that I think will clear some things up.

Actual tear downs and tests have shown the actual capacity of the 85kWh pack to be 81kWh under normal conditions, the 90kWh pack is almost certainly near 86kWh actual capacity. Electrek reports that the cells themselves have not changed. What HAS changed is how tightly the cells are packed together. The existing Tesla modules hold 444 cells and the cell containing area measures about 28cmx68.5cm or 1918cm^2. Circle packing theory tells us that the most dense arrangement of circles (and therefore cylinders) covers 90.69% or pi/(2*SQRT(3)) of an area. Maximum diameter of an 18650 cell is 1.86cm (No! It's not 1.80cm!) making cross-sectional area 2.72cm^2 which means that the theoretically maximal number of cells that can fit in a module is (1918cm^2*.9069)/2.72cm^2 ...which is 640 cells. Tesla only fits 444 cells in each module so in theory they could increase the pack capacity by (640/444-1)*100% or 44%... That's a lot of room for improvement.
Tesla did not update the range number of the P90D like they did for the 90D after the facelift so I think using the Model X' increase in range will give us the best estimate of the capacity increase. 289/250 = a 15.6% increase in capacity, this should correspond very closesly to the increased cell count with it being likely that the extra mass causes a small hit to efficiency. Tesla's current packs are 96s/74p and I think it's likely that the new packs will want to keep the voltage the same so they will stick with 96s. That means that we're looking for a multiple of 96 that is approximately 16% greater than seventy four. 74*1.16 is 85.84. This points to the new pack being 96s86p, which (I just checked) is exactly what Yggdrasill suggested above. I did not rely on Yggdrasill's number at all in my estimate because his module dimensions are wrong, that said we came to the same conclusion... I believe it is the correct conclusion with >95% confidence.

We know the actual capacity of the 90 packs is about 86kWh, I spent a long time trying to verify this after they were launched, multiply 86 by 86/74 and you get (drum roll) 99.95kWh! So Tesla is no longer rounding stated pack capacity wayyyyy up. This is good. This means that the 100 pack will produce 16.2% more power AND it will charge 16.2% faster at all SoCs. I think this is a pretty big deal...

This also means that the 100D should get 340 miles range... which is more than a Hellcat, it should beat a Hellcat to 60 too unless the Hellcat has a pro driver or racing slicks...​

This is pretty much parallel with assumptions and guesses I made on the new 100kWh pack here;

My take on the new 100kWh pack

However I think they still are rounding it up like they do with 85 an 90 packs in the same rate too as I explained in the thread I started. This site is pretty good to fiddle with circle packing;

Circles within Rectangles

So if space in between cells were 0 and they were touching each other in a hexagonal pattern we could fit 629 of them in a module. That means Tesla was only using 70% of the space. (We have to assume hexagonal packing for our guesses even though the actual pack is triangular since the site only does rectangular or hexagonal)

However we don't want cells touching each other and if I recall correctly the distance between them were 2.2mm. Using that we get 90% utilization. That must be the sweet spot Tesla chose to dissipate adequate heat and pack enough cells with their current cooling system. Rest is just coolant tubes, loops, turning points etc.

I don't know what this new cooling system is but if it can halve cells' distance from each other(which would be drastic, meaning it can get rid of equal or better heat as before in this tighter packing) Tesla could fit, with the same 90% util. rate for other stuff 498 cells, instead of 444.

If the new layout is indeed 86 cells per group as you suggested that means 516 cells per module. Meaning they had to have cells almost touching each other (highly unlikely) or they can dramatic improvements in the area 'other stuff' covers as well. However if it was packing 82 I suggested, That means new pack has 492 cells per module now.

If it was 82p 96s as my guess; 3300mAh * 82 * 3,7V * 96 = 96,1kWh. Allowing for a 97% DoD for protection that is 93kWh of usable energy. Enough on Tesla's game to call it 100D.

Your guess is also not impossible and would solve the rounding up problem, it would however mean an incredible cooling solution.

Once people start taking deliveries we could measure the peak output of amps and divide it by the old C rate we know, which is 6.14C (1500Amps for a 244Ah pack = 6,14C) 82*3,3*6,14 = 1661Amps, my guess, 86*3,3*6,14 = ~1750Amps your guess. This is also assuming Tesla uses the same C rate hence the same voltage drop. They could opt for less C rate but also less voltage drop.

Another easy way would also be measuring the usable energy difference. If 10%, my guess, 15%, your guess. Official EPA rates would be a hint at this too.

p.s. This datasheet of Panasonic 18650 cells suggest the width to be 18.15mm, not 18.6mm. I calculated accordingly.

 

[Evbwcaer]

The most space efficient way would be to fit all the batteries as tightly as you could, and then run coolant through the left over space. The remaining space is what you would get if you packed four circles into a square.

I think this is the same as the air mattress example.