Cylindrical vs Pouch cells

[Fact Checking]

So I got a record number of disagrees regarding my comment about Tesla potentially announcing on the "Battery Investor Day" a decision to not use cylindrical cells and migrate to 'flat' cells such as pouch or prismatic cells.

At the risk of this being OT (although it IMHO matters directly to Tesla investments, cost of goods and cell manufacturing capex and ramp-up speed), here's how I see it:

I don’t see them changing anytime soon. That would require substantially different tolerance for heat in the battery chemistry. Maxwell tech makes things better, but charging at high speeds will still dictate the current form factor for the foreseeable future.

Given higher charging speeds vs. prismatic cells, I think Tesla will choose higher charging speeds until charging is closer to the 5-10 minute range.

Actually, this is a common misconception, cylindrical cells are pretty much the worst-case layout for heat management, because the available heat flux from the innermost 'core' layers dictates the overall cooling requirements - and those layers are a comparatively smaller fraction of the total mass:

To get the heat of those innermost layers out of the cell they external surface of the cell has to be cooled more than an equivalent flat cell would have to be, because the heat as to be disspitated through more layers which are all heat generating as well.

With prismatic, pouch and other types of 'flat' cell designs the mass-to-surface ratio and thus the potential cooling surface is much larger and the size of the innermost limiting layer is similar to the other layers:

I.e. the same chemistry and same mass can be cooled more effectively and more uniformly in flat cells than in cylindrical cells - and the volume can be kept at a more uniform temperature as well. This is a result of the basic geometry of their layout.

Flat cells can also be packed more effectively, as there's no volume loss due to the circle packing loss of cylindrical cells:

Note how wasteful cylindrical cell packing is, compared to flat cells. Plus the cooling channels only touch part of the surface of the cylindrical cells:

Note how in that tear-down image the cooling channels are touching maybe 30% of the circumference of each cell. Cylindrical cells also complicate pack design due to their very granular and not easy to pack nature. I don't think it's an accident that the big Model 3 delay was related to the ... unexpectedly low performance of cylindrical cell based battery pack assembly lines.

Contrast this with the heating and packing efficiency of whole-surface two side contact flat cells.

I believe Tesla's cars have superior charging speeds not because of cylindrical cells, but despite using cylindrical cells.

Whether the better cooling of flat cells, combined with their better volumetric density, matters enough for Tesla to go through the trouble of switching form factors, I have no idea - but I don't think it's nearly as obvious as you appear to make it.

 

[Beltsbear]

So I got a record number of disagrees regarding my comment about Tesla potentially announcing on the "Battery Investor Day" a decision to not use cylindrical cells and migrate to 'flat' cells such as pouch or prismatic cells.

At the risk of this being OT (although it IMHO matters directly to Tesla investments, cost of goods and cell manufacturing capex and ramp-up speed), here's how I see it:



Actually, this is a common misconception, cylindrical cells are pretty much the worst-case layout for heat management, because the available heat flux from the innermost 'core' layers dictates the overall cooling requirements - and those layers are a comparatively smaller fraction of the total mass:

To get the heat of those innermost layers out of the cell they external surface of the cell has to be cooled more than an equivalent flat cell would have to be, because the heat as to be disspitated through more layers which are all heat generating as well.

With prismatic, pouch and other types of 'flat' cell designs the mass-to-surface ratio and thus the potential cooling surface is much larger and the size of the innermost limiting layer is similar to the other layers:

View attachment 427562

I.e. the same chemistry and same mass can be cooled more effectively and more uniformly in flat cells than in cylindrical cells - and the volume can be kept at a more uniform temperature as well. This is a result of the basic geometry of their layout.

Flat cells can also be packed more effectively, as there's no volume loss due to the circle packing loss of cylindrical cells:

View attachment 427563

Note how wasteful cylindrical cell packing is, compared to flat cells. Plus the cooling channels only touch part of the surface of the cylindrical cells:

Note how in that tear-down image the cooling channels are touching maybe 30% of the circumference of each cell. Cylindrical cells also complicate pack design due to their very granular and not easy to pack nature. I don't think it's an accident that the big Model 3 delay was related to the ... unexpectedly low performance of cylindrical cell based battery pack assembly lines.

Contrast this with the heating and packing efficiency of whole-surface two side contact flat cells.

I believe Tesla's cars have superior charging speeds not because of cylindrical cells, but despite using cylindrical cells.

Whether the better cooling of flat cells, combined with their better volumetric density, matters enough for Tesla to go through the trouble of switching form factors, I have no idea - but I don't think it's nearly as obvious as you appear to make it.

This is a very informative post. But a few things...

The cooling on the model 3 design is enough. While flat might offer more cooling, it is not needed. The cylindrical cells waste space but weight is the main issue not size. The reason Tesla tops out in current KWh right now is based on cost and weight. There is no pressing need to reduce size. Finally the cylindrical construction does add to safety with its added void space. So the downsides are small in the real world while the upsides out weight them. That could change but I don’t see it on any model coming out in the next three years.

 

[JRP3]

I believe Tesla's cars have superior charging speeds not because of cylindrical cells, but despite using cylindrical cells.

Cylindrical cells of the same chemistry have higher C rates than pouch and prismatic. Cylindrical cells are inherently resistant to swelling and are structurally sound. Pouch and prismatic will swell and need additional packaging to keep them pressed together tightly.

 

[Fact Checking]

Cylindrical cells of the same chemistry have higher C rates than pouch and prismatic. Cylindrical cells are inherently resistant to swelling and are structurally sound. Pouch and prismatic will swell and need additional packaging to keep them pressed together tightly.

Are there any papers about this by any chance? AFAIK swelling shouldn't really occur if the cells are within specifications, but I could be totally wrong ...

 

[jerry33]

Are there any papers about this by any chance? AFAIK swelling shouldn't really occur if the cells are within specifications, but I could be totally wrong ...

I don't know about papers, but swelling in the prismatic Prius cells has been seen. Admittedly different chemistry, but swelling is swelling.

 

[Beltsbear]

Are there any papers about this by any chance? AFAIK swelling shouldn't really occur if the cells are within specifications, but I could be totally wrong ...

I have seen zero cylindrical cells in laptops swell despite seeing almost 10x as many as pouch cells. I have seen at least six pouch cells swell. I have seen pouches swell in laptops, phones and other devices.

 

[Chocochip]

cylindrical cells are pretty much the worst-case layout for heat management, because the available heat flux from the innermost 'core' layers dictates the overall cooling requirements - and those layers are a comparatively smaller fraction of the total mass

It seems you really want to be right about this, but unfortunately you’re not. I appreciate the effort you put into researching these topics, but just consider the fact that Tesla didn’t make the huge investment into the 2170 cells production lines in the Gigafactory on a whim. In the quote above you point out that the heating generated in the innermost layers of the cell is what determines the cooling requirements, while also correctly pointing out they constitute the smallest fraction of the total mass, in other words the smallest volume fraction that requires cooling.

To get the heat of those innermost layers out of the cell they external surface of the cell has to be cooled more than an equivalent flat cell would have to be, because the heat as to be disspitated through more layers which are all heat generating as well.

Are we sure there are more layers in a cylindrical cell than in a pouch cell?

If yes, then the pouch cell will require more packing material (shell) per volume unit of active material than the cylindrical cell, resulting in a lower overall volume fraction of active material in the battery module.

If no, then the heat needs to be conducted across a similar distance, however with flat layers the core layers have the same volume as the outer layers, meaning there is a significantly higher amount of heat generated in those core layers compared to the cylindrical cell, and that will make cooling more difficult. On top of that, each cell will represent a much larger part of the total battery pack (so, fewer cells in each pack), meaning that each cell that misbehaves will have a greater impact on the pack performance.

With prismatic, pouch and other types of 'flat' cell designs the mass-to-surface ratio and thus the potential cooling surface is much larger and the size of the innermost limiting layer is similar to the other layers:

Again, think in terms of volume. Assuming you aim to keep the median layer — the layer halfway between the shell and the core layer — below 30 deg. C, what volume fraction of the cell will be below that temperature in a cylindrical cell? The answer is 75%. How about in a pouch cell? The answer is 50%. That innermost layer, the one that’s most difficult to cool down, represents a much larger volume fraction of the cell.

 

[skybluecgreen]

So I got a record number of disagrees regarding my comment about Tesla potentially announcing on the "Battery Investor Day" a decision to not use cylindrical cells and migrate to 'flat' cells such as pouch or prismatic cells.

At the risk of this being OT (although it IMHO matters directly to Tesla investments, cost of goods and cell manufacturing capex and ramp-up speed), here's how I see it:



Actually, this is a common misconception, cylindrical cells are pretty much the worst-case layout for heat management, because the available heat flux from the innermost 'core' layers dictates the overall cooling requirements - and those layers are a comparatively smaller fraction of the total mass:

To get the heat of those innermost layers out of the cell they external surface of the cell has to be cooled more than an equivalent flat cell would have to be, because the heat as to be disspitated through more layers which are all heat generating as well.

With prismatic, pouch and other types of 'flat' cell designs the mass-to-surface ratio and thus the potential cooling surface is much larger and the size of the innermost limiting layer is similar to the other layers:

View attachment 427562

I.e. the same chemistry and same mass can be cooled more effectively and more uniformly in flat cells than in cylindrical cells - and the volume can be kept at a more uniform temperature as well. This is a result of the basic geometry of their layout.

Flat cells can also be packed more effectively, as there's no volume loss due to the circle packing loss of cylindrical cells:

View attachment 427563

Note how wasteful cylindrical cell packing is, compared to flat cells. Plus the cooling channels only touch part of the surface of the cylindrical cells:

Note how in that tear-down image the cooling channels are touching maybe 30% of the circumference of each cell. Cylindrical cells also complicate pack design due to their very granular and not easy to pack nature. I don't think it's an accident that the big Model 3 delay was related to the ... unexpectedly low performance of cylindrical cell based battery pack assembly lines.

Contrast this with the heating and packing efficiency of whole-surface two side contact flat cells.

I believe Tesla's cars have superior charging speeds not because of cylindrical cells, but despite using cylindrical cells.

Whether the better cooling of flat cells, combined with their better volumetric density, matters enough for Tesla to go through the trouble of switching form factors, I have no idea - but I don't think it's nearly as obvious as you appear to make it.

Wouldn't a switch from cylindrical cells to 'flat' mean a radical change in the processes and equipment needed to manufacture them? Wouldn't the investment in Gigafactory 1 become almost worthless and a huge new investment in a new process become necessary?

 

[JRP3]

Are there any papers about this by any chance? AFAIK swelling shouldn't really occur if the cells are within specifications, but I could be totally wrong ...

Prismatic LiFePO4 cells bought in volume from Thundersky and CALB come standard with strapping material even though they are built with hard plastic casing. A Google search of lithium ion pouch or prismatic cell swelling should provide plenty of results.

 

[Fact Checking]

In the quote above you point out that the heating generated in the innermost layers of the cell is what determines the cooling requirements, while also correctly pointing out they constitute the smallest fraction of the total mass, in other words the smallest volume fraction that requires cooling.

This doesn't help your argument though: no part of the cell must overheat beyond the allowed temperature range, and thus the innermost layer will be the "weakest link" that determines how much cooling the cell requires.

The "outer" layers will be over-cooled, just to keep the innermost layers within a safe temperature range.

 

[aubreymcfato]

So I got a record number of disagrees regarding my comment about Tesla potentially announcing on the "Battery Investor Day" a decision to not use cylindrical cells and migrate to 'flat' cells such as pouch or prismatic cells.

At the risk of this being OT (although it IMHO matters directly to Tesla investments, cost of goods and cell manufacturing capex and ramp-up speed), here's how I see it:



Actually, this is a common misconception, cylindrical cells are pretty much the worst-case layout for heat management, because the available heat flux from the innermost 'core' layers dictates the overall cooling requirements - and those layers are a comparatively smaller fraction of the total mass:

To get the heat of those innermost layers out of the cell they external surface of the cell has to be cooled more than an equivalent flat cell would have to be, because the heat as to be disspitated through more layers which are all heat generating as well.

With prismatic, pouch and other types of 'flat' cell designs the mass-to-surface ratio and thus the potential cooling surface is much larger and the size of the innermost limiting layer is similar to the other layers:

View attachment 427562

I.e. the same chemistry and same mass can be cooled more effectively and more uniformly in flat cells than in cylindrical cells - and the volume can be kept at a more uniform temperature as well. This is a result of the basic geometry of their layout.

Flat cells can also be packed more effectively, as there's no volume loss due to the circle packing loss of cylindrical cells:

View attachment 427563

Note how wasteful cylindrical cell packing is, compared to flat cells. Plus the cooling channels only touch part of the surface of the cylindrical cells:

Note how in that tear-down image the cooling channels are touching maybe 30% of the circumference of each cell. Cylindrical cells also complicate pack design due to their very granular and not easy to pack nature. I don't think it's an accident that the big Model 3 delay was related to the ... unexpectedly low performance of cylindrical cell based battery pack assembly lines.

Contrast this with the heating and packing efficiency of whole-surface two side contact flat cells.

I believe Tesla's cars have superior charging speeds not because of cylindrical cells, but despite using cylindrical cells.

Whether the better cooling of flat cells, combined with their better volumetric density, matters enough for Tesla to go through the trouble of switching form factors, I have no idea - but I don't think it's nearly as obvious as you appear to make it.

I don't know anything about cells, but I know that Tesla favors a "one-form-to-them-all" approach. They need a sh!tload of cells for all their products (cars, powerwalls, powerpacks), so probably here the key is finding the sweet spot for all the technical specs you are saying and easiness of manufacturing.

 

[Chocochip]

This doesn't help your argument though: no part of the cell must overheat beyond the allowed temperature range, and thus the innermost layer will be the "weakest link" that determines how much cooling the cell requires.

The "outer" layers will be over-cooled, just to keep the innermost layers within a safe temperature range.

No, you missed the point: a smaller volume is easier to cool down. A small, warm cylinder (the innermost layers) surrounded by a larger, cooler cylindrical sheath (the outer layers and the casing) will cool more effectively than a “sandwich” where the warm, central layers are just as large as the cooler, outer layers.

This, again, assumes we look at the same number of layers. A smaller number of layers in the pouch simply translates to a smaller volume fraction of active material in the module because of the need for more packing material.

 

[Fact Checking]

Again, think in terms of volume. Assuming you aim to keep the median layer — the layer halfway between the shell and the core layer — below 30 deg. C, what volume fraction of the cell will be below that temperature in a cylindrical cell?

No, that's wrong, temperature ranges are safety and longevity related, they must not be exceeded in any of the layers. "Average" temperature is not enough.

So if the max is 30 degrees then the innermost layers must be kept at 30 degrees, and the outer layers at lower temperatures.

but just consider the fact that Tesla didn’t make the huge investment into the 2170 cells production lines in the Gigafactory on a whim.

There's no contradiction: there's various pros and cons of the 21,700 cylindrical cells, which I'm sure Tesla considered originally, and came to the right decision at that time.

Using a standard 21,700 size that many cell suppliers can make in principle, instead of some proprietary pouch format that ties Tesla to a single supplier, must have been a major factor - and indeed they used Samsung cells and were in talks with Chinese 2170 suppliers as well for GF3.

My argument is that if Tesla decides to manufacture cells too, using the Maxwell dry cell process, the above advantage of the industry standard 21,700 cell format of a cylindrical cell go away: they won't be "locked in" a pouch format and won't have unhealthy dependence on the supplier.

This might change the balance of pros and cons in favor of flat cells.

Or not - I don't know: and as I said it in my first comment, I consider the continued use of 2170 cells the more likely decision.

 

[Fact Checking]

No, you missed the point: a smaller volume is easier to cool down. A small, warm cylinder (the innermost layers) surrounded by a larger, cooler cylindrical sheath (the outer layers and the casing) will cool more effectively than a “sandwich” where the warm, central layers are just as large as the cooler, outer layers.

This, again, assumes we look at the same number of layers. A smaller number of layers in the pouch simply translates to a smaller volume fraction of active material in the module because of the need for more packing material.

You also need to factor in the reduction in contact surface: only about 30% of the cylindrical cell's surface is in contact with a cooling channel - the rest is in contact with two other cells in the honeycomb layout, or with an air pocket which is basically a thermal insulation layer.

A rather inefficient layout to get the heat out - but it obviously can be done if cooling capacity is high enough.

Note that I didn't mention thermal properties as a major advantage of flat cells - this was brought up as an argument against my points - which thermal argument is false I believe.

It might be that other advantages of cylindrical cells (such as their mechanical stability against internal swelling) dominate.

 

[Fact Checking]

Prismatic LiFePO4 cells bought in volume from Thundersky and CALB come standard with strapping material even though they are built with hard plastic casing.

Well, there's now numerous other EVs with pouch or prismatic cells and not much of a swelling problem, right?

A Google search of lithium ion pouch or prismatic cell swelling should provide plenty of results.

Most of them suggest that "cell swelling" is a condition that occurs on overcharging, or with poor chemistries, which Tesla's chemistry and a good battery management system should avoid?

 

[Chocochip]

You also need to factor in the reduction in contact surface: only about 30% of the cylindrical cell's surface is in contact with a cooling channel - the rest is in contact with two other cells in the honeycomb layout, or with an air pocket which is basically a thermal insulation layer.

The casing is aluminium, which has a very good thermal conductivity (about one order of magnitude higher than steel). This means the casing will be able to very efficiently absorb heat from the cell and transfer it to the cooling ribbon, even with a limited contact area. I also believe (although I’m not sure) that the space between the cells is not empty, but rather filled with some sort of thermoconductive paste or filling.

On the point of the thermal properties of a cylindrical battery cell module not being the main topic: as I already mentioned, when Tesla started building the GF and purchasing production equipment, they were launching a completely new product, the Model 3. Considering the scale of production they were preparing for, they had the opportunity to figure out the optimal battery pack design and choose the ideal battery cell geometry for it, as the scale of the project would have justified any supplemental costs required to create a proprietary design. They went with a somewhat obscure format (at that time), the 2170 cylindrical cell. It wasn’t a legacy format for them. Why not continue with the 18650s, which were being produced at a much larger scale and therefore more convenient to get production equipment for? Or any of the pouch or prismatic cell formats already in production at both Panasonic and the Korean and Chinese manufacturers?

If indeed the 2170 cells were easier or cheaper to get or to implement in modules, wouldn’t all the legacy OEMs have gone for those as well? Those guys were already trying to simply build EVs on ICE platform in order to put the least amount of effort into it, so logically they would have gone for the most “convenient” tech. They went for pouch cells. Tesla was trying to get the best technology, and they went with 2170 cells. They probably had a very good reason.

Also, as you made several comments on my previous explanation: please read carefully what I actually wrote and what assumptions I made. I was trying to give theoretical examples to make it simpler to “visualise” the heat flow in the cells. The whole argument was based on geometry and physics principles, which are universally valid, and not on a detailed knowledge of battery cell manufacturing or temperature tolerances, which I simply don’t possess.

Finally, I regret having been sucked into this discussion on battery cells, which is clearly off-topic for this thread. I will stop now.

 

[JRP3]

Well, there's now numerous other EVs with pouch or prismatic cells and not much of a swelling problem, right?

Less volatile chemistry, and the pouches are tightly clamped. Volt pack, not labeled but just above the cooling tubes are threaded rods and nuts clamping it all together:

Bolt pack is similar, starting at the 24 minute mark:

Most of them suggest that "cell swelling" is a condition that occurs on overcharging, or with poor chemistries, which Tesla's chemistry and a good battery management system should avoid?

Tesla has a more volatile cell chemistry. I've never seen NCA in pouch or prismatic form. LiFePO4 is a stable, high C rate chemistry, and under normal use they can swell if not clamped and end up being damaged.

 

[JRP3]

Finally, I regret having been sucked into this discussion on battery cells, which is clearly off-topic for this thread.

Not in this thread

 

[Sudre]

The inner of core does not have to be the tap for the positive and the outer does not have to be the tap for the negative. The current collectors lay opposite each other the same as the prismatic cell and the taps can be anywhere as designed along the sandwich. There are many youtube videos on how cells can be made.

You can get the idea of which one is faster to create and why Tesla (IMO) is using cylindrical.

As others have said. Heating is not an issue for any style if the right amount of cooling is supplied to any cell. Obvious to me because the Model 3 charges faster than anything on the market.

Still there may be an argument for a hybrid type cylindrical cell that has heat sink(s) inserted in the roll so the finished product looks more "eye" shaped or flattened roll. This would give the cell more surface area contact with the cooling plates and suck some heat out of the center through the heat sink(s). (A bit hard to describe the creation method)
There is also no reason a cylindrical cell can't have a heat sink in the middle.
also.... what if there was a way to make the separator a type of heat sink?

I don't see prismatic cells ever compressing tight enough or being created fast enough. Second video. Sure you can speed up that second video's machine but you can also speed up the cylindrical one just the same.

 

[22522]

It is bad when faults propagate. A cylindrical cell limits fault propagation better than a pouch.