Large Prismatic cells and small cylindrical cells in BEVs

[Olle]

There has been several reports recently of competitors developing BEVs with LG Chem and Samsung batteries with cells that are said to be much bigger than Teslas "giga factory cell". Korean Batteries? Ford To Spend $4.5 Billion In EV - Asia Stocks to Watch - Barrons.com

On Samsung's homepage it said they are prismatic:
Automotive Battery - High Voltage Battery Systems | Samsung SDI

How can Tesla, Samsung and LG Chem come to such vastly different conclusions of what the optimal battery is?

 

[Johan]

To bring this to a more general level the question is really "Given the same underlying facts and data, how can different individuals/organizations reach vastly different conclusions?".

The answer of course is that what may at first seem objective is, in the end, strongly influenced by subjective biases. In this case I'm guessing (like in most other situations in life) the most important biases are:

1) Confirmation bias (a tendency to interpret known and new information in a way that confirms the choices one has already made, or is about to make)

2) Expectation bias (or experimenter's bias) (a tendency to find what you think you're looking for when doing experiments or studying for example the benefits of your own battery design v.s. someone else's)

3) Less/few is better than more bias (people generally, including engineers, will often have a tendency to view a design being "simple" in that there are fewer individual pieces/units or in this case battery cells as inherently better than one with many pieces, in this case cells, regardsless of the fact that in reality it's all one big battery only with a different ratio of surface area, electrolyte, anode and cathode and number of fuse points)

4) Post purchase/post investment bias (a sort of committment and confirmation bias, once the investment decision has been made you will tend to favor a view that confirms that you indeed made the right investment choice)

5) Exclusivity bias (this is not a well described term but something I believe my self to be important). By this I mean that since Tesla and Samsung have already gone down the route of small cylindrical cells then competetiors will inherently bias toward technical solutions that are clearly different, since this gives the apparent effect of having found a different (and better) solution to the problem rather than admitting that your competitor already got it right and now you're just trying to do the same thing (but do it as good or better).

 

[Saghost]

There has been several reports recently of competitors developing BEVs with LG Chem and Samsung batteries with cells that are said to be much bigger than Teslas "giga factory cell". Korean Batteries? Ford To Spend $4.5 Billion In EV - Asia Stocks to Watch - Barrons.com

On Samsung's homepage it said they are prismatic:
Automotive Battery - High Voltage Battery Systems | Samsung SDI

How can Tesla, Samsung and LG Chem come to such vastly different conclusions of what the optimal battery is?

It's RAID all over again. If you've been around computers long enough, you know that the original version of that acronym was "redundant array of inexpensive devices." The idea was, you could take a bunch of cheap, small, semi-reliable drives and combine them to make a large reliable drive for less money than building a dedicated large drive of adequate reliability.

GM, LG, and much of the rest of the industry are on the large drive side. They are addressing the issues of reliability with extraordinary quality control, while building larger cells to drive cost of the cell down.

Tesla took a RAID approach. They have designed a system that can gracefully accept multiple cells failing with minimal adverse effect, and are then using less expensive cells without the extreme measures, and in fact using a riskier chemistry to give better energy density.

Personally, I think the RAID approach is a very clever innovation on Tesla's part, which should offer remarkable reliability as time goes on(assuming something else in the pack doesn't turn out to be relatively unreliable, of course.)
Walter

 

[scaesare]

The other significant factor that Tesla placed much more importance upon was economies of scale.

EV's were largely niche/compliance products for most manufacturers, and the assumption was low-volume/higher cost cells specifically built for automotive use.

Tesla planned on selling a boatload of cars, and thus the economies of scale already achieved by cylindrical cell production for consumer electronics could be even further optimized with the volumes Tesla was anticipating.

Quite frankly, there are a number of trade-offs in using a cylindrical 18650 cell (although a number of unexpected advantages as well), that require additional engineering effort up front to overcome. For traditional automakers it was simply far easier to outsource the battery design/manufacture to an external company, and then install them much like any other outsourced component.

 

[Olle]

5) Exclusivity bias (this is not a well described term but something I believe my self to be important). By this I mean that since Tesla and Samsung have already gone down the route of small cylindrical cells then competetiors will inherently bias toward technical solutions that are clearly different, since this gives the apparent effect of having found a different (and better) solution to the problem rather than admitting that your competitor already got it right and now you're just trying to do the same thing (but do it as good or better).

That's a big one! And it is one that turned out to be very important for Toyota in putting their huge stake in fuel sells. They are willing to make massive investments into a whole new fuel cell eco system that they probably already know to be inherently worse than a BEV system. Just to be able to show that they took back the initiative.

- - - Updated - - -

It's RAID all over again. If you've been around computers long enough, you know that the original version of that acronym was "redundant array of inexpensive devices." The idea was, you could take a bunch of cheap, small, semi-reliable drives and combine them to make a late reliable drive for less money than building a dedicated large drive of adequate reliability.

GM, LG, and much of the rest of the industry are on the large drive side. They are addressing the issues of reliability with extraordinary quality control, while building later cells to drive cost of the cell down.

Tesla took a RAID approach. They have designed a system that can gracefully accept multiple cells failing with minimal adverse effect, and are then using less expensive cells without the extreme measures, and in fact using a riskier chemistry to give better energy density.

Personally, I think the RAID approach is a very clever innovation on Tesla's part, which should offer remarkable reliability as time goes on(assuming something else in the pack doesn't turn out to be relatively unreliable, of course.)
Walter

Exactly. Another thing was that the available cylindrical cells, 18650, had better energy density than big pouch cells at the time. And easier to cool and heat with the large surface area. Will be interesting to see what energy density these new vehicles with big cells will achieve.

- - - Updated - - -

The other significant factor that Tesla placed much more importance upon was economies of scale.

EV's were largely niche/compliance products for most manufacturers, and the assumption was low-volume/higher cost cells specifically built for automotive use.

Tesla planned on selling a boatload of cars, and thus the economies of scale already achieved by cylindrical cell production for consumer electronics could be even further optimized with the volumes Tesla was anticipating.

Quite frankly, there are a number of trade-offs in using a cylindrical 18650 cell (although a number of unexpected advantages as well), that require additional engineering effort up front to overcome. For traditional automakers it was simply far easier to outsource the battery design/manufacture to an external company, and then install them much like any other outsourced component.

Right. This outsourcing will come back and bite them. They outsourced everything but bodywork and ICE design. I think Daimler and a few others kept transmission in house too. With batteries outsourced and no ICE in the car, what value is left in the companies?

 

[Jeff N]

Automakers have outsourced major aspects of their vehicle manufacturing for a long time. There is a huge 3rd party auto parts industry that supplies brakes, steering, transmissions, radios, navigation, windows, seats, A/C, plastic trim -- just about everything except the body and ICE. Often, these 3rd party parts are co-designed but built more cost effectively by specialty firms that can spread their manufacturing tools and costs across many various automotive customers. It's just basic economics and does not necessarily have to do with commitment to building any particular car model. For example, Toyota started off outsourcing the batteries and transmission manufacturing for their first Prius model although they did the actual non-battery design work. Later on, Toyota invested in the battery maker and brought the transmission making in-house. GM largely did the same with the Volt. They co-designed the motors and probably the inverter but made their own transmission and battery pack while buying cells from LG. The new Volt follows the same strategy but more of the manufacturing has moved to the US (motors, inverter, from Japan, transmission from Mexico, early ICE from GM/Opel in Austria).

From what we know so far, it appears that the Bolt's motor, inverter, and perhaps the enclosing final drive gear box are being made by LG although the motor may have been designed by GM (much like the Volt motors are and were designed by GM but built by Hitachi). The cells and pack are now apparently being made together by LG in South Korea but may eventually be moved to the US. LG is also supplying the radio and entertainment systems instead of some other 3rd party like Bose. None of this necessarily implies anything about GM's commitment to the Bolt or other future BEVs. Nothing prevents the 2nd generation Bolt from moving more of the manufacturing to the US. This deal says more about LG's growth in the automotive parts business than it says about GM or its manufacturing and supplier arrangements.

 

[pbceng]

The traditional OEMs are certainly taking a very different approach to batteries than Tesla. All are outsourcing the cells, some are outsourcing the complete battery. This has the advantage that the tier 1 supplier take the risk, R&D costs etc and the OEM in theory gets a cheaper and higher performance product. The disadvantage is that the OEM looses the knowledge and crucially IMHO some of the control - the tier 1 supplier will try and standardise as much as possible to get the cost and complexity down across multiple customers and of course maximise their return.

Tesla's approach was interesting in that they too a technically less attractive cell design but one that was in very high volume production, modified it to suit their needs and took advantage of the lower cost. In theory pouch or prismatic cells will end up cheaper to produce but that's projected to be a few years off yet and Tesla are still refining the design of their 18650 cells. There are a number of theoretical advantages to the larger pouch of prismatic cells but they can be more difficult to thermally manage and need more protection from surrounding cells in the event of thermal issues. The fact Tesla are retaining the knowledge in-house is potentially a major advantage in a few years time if they survive IMHO.

Should BEVs become more commercially viable than ICE vehicles as some are predicting over the next 15 years then the incumbent OEMs have a huge iceberg approaching but most put short term gain over long term benefit so I don't see that changing!

 

[techmaven]

Tesla is the only major OEM to use NCA chemistry. Everyone else uses something else with lower specific energy, possibly higher tolerances with heat and cold, different lifecycle characteristics, and less volatility. Then there is pack design, with Tesla designing BEV only platforms and most everyone else is converting or sharing a platform, or even their dedicated BEVs will share some design elements with their PHEVs. It is no wonder that there divergence.

 

[jbcarioca]

Automakers have outsourced major aspects of their vehicle manufacturing for a long time. There is a huge 3rd party auto parts industry that supplies brakes, steering, transmissions, radios, navigation, windows, seats, A/C, plastic trim -- just about everything except the body and ICE. Often, these 3rd party parts are co-designed but built more cost effectively by specialty firms that can spread their manufacturing tools and costs across many various automotive customers...

True, but.. even major body parts, sometimes complete bodies, sometimes "body-in-white", often engines are also outsourced or jointly sourced. Factually many vehicles today are more assembly by manufatureis than they are true manufacturing. Several Romance langues call them "montadores" or similar, "mounters", probably a more accurate term than manufacturers. A quick review of Automotive News archives for suppliers and models shows how far this goes. For example check diesel ensines for a paricularly widely-shared technology. BMW has used engines from PSA and Ford.

There is nothing significant in the sourcing for PHEV or BEV. That just continues industry practice since about 1903 or so. Tesla does the same. Nobody can do it all.

 

[jbcarioca]

Without repeating well-documented details, the lessons the NTSB hearings about li-ion in transportation have quite a bit to say about cell format as well as chemistry. If anyone is interested I can upload the entire hearings. During the hearings, which I attended on bahalf of an aviation client, testimony from Tesla, the US Navy (submarine batteries), numerous manufacturers and technical experts was presented. There were a handful of generalizations which were, for me anyway, unforgettable:

1) RAID is a sound principal, but in the battery case, failure containment and individual cell quality control are NOT RAIDable!
Tesla, for example, tests every single cell both in production and prior to installation then maintains continuing records of each cell performance. (This was discussed at some length during the hearings).
2) Other things remaining equal the larger the format the higher the catastrophic failure risk due to the problems maintaining smooth cell chemistry changes, barrier stability and failure protection as cell size increases. Further, the more nearly spherical the cell design the easier smooth cell performance can be.
3) The quality control process in EV and other transportation use needs to exceed anything traditionally known with lead-acid, in-card or other older chemistries.
4) Seemingly riskier chemistries can be deployed safely if steps 1 and 2 are fanatically adhered in all aspects of deployment.

Boeing ignored those first two steps in initial B787 li-ion deployment, as did the US Navy in submarines. The Navy learned. Boeing refused to deal with the format issue, so developed a massively separate failure containment process for the B787 main batteries, soon following the hearings.

Regardless of what new technology might emerge we can rely on Tesla to be the industry safety issue. Candidly, I ordered my own Model S soon after returning from these hearings. Tesla probably will continue to be the industry leader in new battery technology, but it certainly will be the leader in battery safety.

 

[dow]

FYI, recent coverage on flat/prismatic cell usage and its supposed superior characteristics:

Tesla: A Gigafactory Challenge - Tesla Motors (NASDAQ:TSLA) | Seeking Alpha

This is credited for being part of how LG and GM beat Tesla to market with a $38k ($30k after rebates), 200 mile EV.

 

[techmaven]

FYI, recent coverage on flat/prismatic cell usage and its supposed superior characteristics:

Tesla: A Gigafactory Challenge - Tesla Motors (NASDAQ:TSLA) | Seeking Alpha

This is credited for being part of how LG and GM beat Tesla to market with a $38k ($30k after rebates), 200 mile EV.

Linking to Seeking Liars is bad enough, but linking to Frank? That's like pointing to Weekly World News. You deserve a demerit and to sit in the corner and contemplate the damage you have perpetuated upon humanity.

 

[dow]

Linking to Seeking Liars is bad enough, but linking to Frank? That's like pointing to Weekly World News. You deserve a demerit and to sit in the corner and contemplate the damage you have perpetuated upon humanity.

Hah! Sorry, didn't realize there was history there.

Was genuinely curious how we saw the LG/GM battery strategy argument being made there. So, that's one vote for "it's wrong" - though, I am curious, is there a short answer for why? Is it that the cylindrical/18650 approach remains superior, period (and we expect the model 3, and maybe future models will thus continue with it)? Or is it more expensive and heavier, but safer, in people's opinion? Or...?

 

[jbcarioca]

Hah! Sorry, didn't realize there was history there.

Was genuinely curious how we saw the LG/GM battery strategy argument being made there. So, that's one vote for "it's wrong" - though, I am curious, is there a short answer for why? Is it that the cylindrical/18650 approach remains superior, period (and we expect the model 3, and maybe future models will thus continue with it)? Or is it more expensive and heavier, but safer, in people's opinion? Or...?

Prismatic cells offer flexility in oackaging but are more expensive, tend to swell in use, have poorer thermal management, and tend to have shorter life. However, technological advances are reducing everything negative except for costs.
Types of Battery Cells; Cylindrical Cell, Button Cell, Pouch Cell
has a bit of perspective. BTW, the larger the format the greater the thermal management problem.
Tesla chose 18650 format originally, according to Tesla engineer testimony before the NTSB, because of easy cell isolation in the event of catastrophic failure, low cost, easier thermal management and easier manufacturing (it has been by far the most common Format).
Although lots of people favor large format prismatic cells because of easy design integration and simple connections the negatives still make it unclear whether or not Tesla will adopt them fir the Model 3. The combination of swelling propensity in prismatic cells with swelling propensity with silicon anodes makes this a difficult technical decision, especially when reducing costs is even higher priority than increased energy density, according to JB.

still, technical advances are coming quickly...

 

[Jeff N]

Just read the link. He's got a lot of dated or misleading information there.

With the actual chemistries being used, the Tesla/Panasonic cells and packs have been and probably will still be cheaper than the pouch cells that other car companies are using. There are lots of engineering and business trade offs between the different approaches. Tesla's approach has been very good for them so far and they have all the information, are smart, and have apparently decided to continue their existing approach but with the usual tweaks that come from experience with the last generation.

I say this as someone who closely follows and appreciates the engineering approaches being followed by both Tesla and GM with their plugin cars.

 

[McRat]

I tore down a 2012 Volt battery which is water-cooled 3P x 96S configuration. They use very thin (0.004") aluminum plates with water racetracks in them to cool the pouches. They use the same system in the new 2P x 96S packs, and dropped weight, increased capacity, and reduced pack volume.

Going without my notes, it seems they are 370gm x .15 liter? and 55wH on the older 3P pouches? I'm not sure people realize these aren't primitive tech even by today's standards. As these pouches are now passing 100,000 miles of EV use with no significant degradation, it makes you wonder what the new packs are capable of with even larger kWh sizing. A Volt with 100k pure EV miles is akin to a Model S 85kWh reaching 500,000 miles?

The Bolt will be using the tech from the Gen 2 Volt pouches apparently. So 400,000 miles might be doable. We won't know for at least 10 years though.

 

[David99]

JB once said he was surprised how much people discuss the cell format when really what's important is what is inside the cell.

 

[McRat]

JB once said he was surprised how much people discussion the cell format when really what's important is what is inside the cell.

Apparently the LG chemistry doesn't suck. The untouched 2012 pouches are reading 3.85vdc right now. Never been trickle charged. And the fact they aren't degrading significantly at 1700 full charge cycles (keep in mind regen probably increases that wildly) and 4 years is also good.

They run at 7C discharge through their SOC window. The charge at 3.7C through their charge window.

Perhaps the Tesla batteries are better, I'm not sure. The Tesla packs however aren't wildly better, and nobody will know the charge cycles they will endure without significant losses for many years to come.

 

[Jeff N]

I tore down a 2012 Volt battery which is water-cooled 3P x 96S configuration. They use very thin (0.004") aluminum plates with water racetracks in them to cool the pouches. They use the same system in the new 2P x 96S packs, and dropped weight, increased capacity, and reduced pack volume.

Going without my notes, it seems they are 370gm x .15 liter? and 55wH on the older 3P pouches? I'm not sure people realize these aren't primitive tech even by today's standards. As these pouches are now passing 100,000 miles of EV use with no significant degradation, it makes you wonder what the new packs are capable of with even larger kWh sizing. A Volt with 100k pure EV miles is akin to a Model S 85kWh reaching 500,000 miles?

The Bolt will be using the tech from the Gen 2 Volt pouches apparently. So 400,000 miles might be doable. We won't know for at least 10 years though.

After 100k EV miles (300k miles overall) plus almost 4 years of use, its implausible that there has been "no significant degradation". I say that as someone with a 5.25 year old Volt with 78,400 EV miles and 125,000 miles overall. It has held up remarkably well but is probably showing 5-10% or so range loss. It's hard to know for sure since I have non-oem tires on it and I haven't carefully measured the total charge consumption recently. We haven't yet heard recently from Erick about any range loss on his car with 100k and 300k overall. However, the Volt uses cells that are constructed and have chemistry tuned more towards power-density whereas Tesla cells are tuned for energy density. Power dense cells are known to have general longer cycle lifetimes.

We don't know the cell chemistry of the Bolt cells although they have been described as "nickel rich" and will be constructed for energy density. They are also cooled differently. Instead of active cooling fins with coolant flowing inside, the Bolt EV pack has a coolant-filled bottom plate with passive aluminum fins extending up from it that interleaved with the new thicker cells. This likely allows denser cell packing and a cheaper cooling system that is somewhat less effective but good enough since the new cells are said to be tolerant of warmer temperatures.

 

[McRat]

Erik posted in that Facebook link above (comments) that he hasn't seen degradation. Whether it's true, who knows?