New Tesla P100D Battery Pack Conceptualized


The P100D Has Arrived, As Well As New Battery Pack Architecture

The P100D Has Arrived, As Well As New Battery Pack Architecture

The latest Tesla splash is their new 100 kwh battery pack unveiled August 23, 2016 in a conference call with journalists.

“The cell is the same – but the module and pack architecture has changed significantly in order to achieve adequate cooling of the cells in a more energy dense pack, and to make sure that we don’t have cell to cell combustion propagation” – Tesla CEO Elon Musk from the conference call on the new 100 kWh options

So Tesla is using the same battery cells as in the P90, but they have somehow figured out how squeeze in more cells and remove heat more efficiently.

How can this feat be accomplished?

Before we conceptualize let’s look at how the current pack is cooled.

The current pack is cooled by a Tesla patented cooling ribbon that snakes thru the cells as shown below.

The current Tesla battery cooling configuration uses a cooling ribbon that snakes thru the cells. Glycol coolant is circulated in the cooling ribbon.

The current Tesla battery cooling configuration uses a cooling ribbon that snakes thru the cells. Glycol coolant is circulated in the cooling ribbon.

The actual configuration is a tad more complicated than shown in figure 1. Teardown photos of the 85 kwh pack have been well documented but the photos are copyrighted. If you are interested in more detail then you can visit this Tesla TMC forum post or this Ricardo Engineering presentation.

The cooling ribbon in Tesla’s 85 kwh pack takes up room. If we wanted to pack the cells tighter, what  could we do?

In our conceptual cooling configuration the coolant is moved from in between the cells to a bottom plate located beneath the cells.

The cooling plate concept is not new. BMW uses it in the i3 and GM uses it in the new Chevy Bolt EV. BMW uses refrigerant directly in the pack cooling plate while GM’s Chevy Bolt EV uses glycol liquid in the cooling plate.

The cooling plate by itself may not provide sufficient heat transfer so, in the concept, thin aluminum fins are thermally connected to the cooling plate to increase heat transfer. These thin aluminum cooling fins transfer heat from the cell to the bottom cooling plate in our concept as shown below:

 Conceptual P100D Battery cooling configuration

Conceptual P100D Battery cooling configuration

Figure 2 shows the basic concept. However there are many permutations one could hypothesize using the bottom cooling plate design. Here’s just a few:

-Use refrigerant in the cooling plate instead of liquid glycol. This would result in a thinner bottom plate than with glycol coolant

– wrap each cell in a thin aluminum shell to increase contact area with the cell

Cells wrapped in thin aluminum cylinder and placed on dimpled bottom plate

Cells wrapped in thin aluminum cylinder and placed on dimpled bottom plate

-eliminate the aluminum fins completely and just use the battery case as the conductor. Once Tesla is making their own cells, we would think that as a definite possibility.

Readers please feel free to comment on the proposed configuration and conceptualize your own version in the comments section.

About the authors: George Bower is a retired mechanical engineer with over 20 years experience in gas turbine power systems.

Co-Author of the piece, Keith Ritter is a mechanical engineer, and licensed professional engineer with over 35 years of experience in heating ventilation and air conditioning systems.

Categories: Battery Tech, Tesla

Tags: , ,

Leave a Reply

60 Comments on "New Tesla P100D Battery Pack Conceptualized"

newest oldest most voted

Nice ideas. Two other options —

1. FULL IMMERSION. Randy Carlson at SeekingAlpha has brought up the possibility of a fully immersed battery pack using an inert cooling liquid like 3M Novec. The advantage would be cell density and cooling that both approach theoretical maximums. The disadvantages would be cost and weight.

2. HORIZONTAL CELL LAYOUT. Cells could be stacked two-high in a horizontal configuration (parallel to the ground) and the resulting pack would still be shorter than in vertical configuration. Also coolant tubes would have more surface area as they could go above and below each layer. Might eventually allow for more cabin headroom due to a shorter battery pack.

Both would be somewhat radical departures, though immersion cooling would certainly be worthy of Elon’s phrasing about “approaching theoretical pack cell density”.


Idea 2 + tubes which allow air flow through the whole pack. Front and back are as open as possible… Only protection against insects… Due to the cylindrical cell design space is lost (air) take that lost space (air) and make it functional.

Furthermore spend less time designing cooling systems and more time developing better cells. What would it cost to reduce inner resistance by 30% ?

It’s nice to see 100kwh but it would also be nice to see a really cheap 60kwh pack without liquid cooling!

Furthermore send me a paycheck. I hate working for free!

Could you or anyone else please point out to the specific article of R. Carlson or cite the specific paragraph in which he brought up the possibility of a fully immersed battery pack?
Did he actually called it by the name or is the immersion cooling a logical derivation of a different point he made?

I’d really appreciate a prompt response from you guys.

If the cooling is much improved, that should mean it will perform better for a longer period.
Will be interesting to see if it can do some laps on a circuit without being overheated, finally.

Interesting ideas, but there is no explanation how is dealt with fact that the case of the cell is also a cell negative terminal, so they should be electricaly isolated. Then the thermal conductivity is significantly affected exactly as in standard module, where the ?kapton? material and epoxy coating on the cooling ribbon snake is used to electrical isolation.

My guess is about use of heatpipe thermal management system. The only surface for heat exchange is the bottom of the cell (in fact bottom contact is much more effective for heat transfer than already used side contact cooling). Also heatpipe is much more effective than direct liquid/refrigerant system.

Of course then you need to connect both terminal from the upper of the cell but this is not a big deal.

I hope that we can examine the used design soon 🙂

“…in fact bottom contact is much more effective for heat transfer than already used side contact cooling”

Please explain why.

Seems to me if that system was more effective, then it would already be used, as it would be easier (and less expensive) to set up cooling plates on the bottom of each group of cells than the current system, which uses a flattened tube which has to be curved part way around every individual cell.

Instead of using aluminum cooling finns connected to a cooling base plate, I would consider using aluminum cooling pins of curved triangular shape fitting exactly in the similar shaped places left over when the cells are packed to the maximum. For electric insulation and better fit, a rubber like coating could be applied on those pins prior to cell placement.

Great piece, thanks George S.!

Elon is quoted saying that interpropagation of thermal runaway is also better addressed by the new pack layout.

It will be fascinating to see how the P100D vehicles, using essentially Model 3 Battery packs, will perform under various conditions.

How can you, by any stretch of the imagination, describe these are “essentially Model 3 battery packs” when M3 will have cells made at the gigafactory in a different physical format – and, by then, almost certainly at least a somewhat modified chemistry? I for one expect Tesla to not merely co-locate the production of cells and packs, but also to come up with innovations to lower the cost of cell production.

MAYBE the “pack architecture” and cell layout we see here will be similar in Model 3. But that isn’t nearly enough to say it is essentially the same.

Right? Hel-lo-oh…

Guess you don’t comprehend what you read very well. And, you essentially answered your own question– if that was even the point of your response. *shrugs*

I’m with Terawatt.

Is there a link to the audio or the transciption of the 100kw conference call with Elon and JB?

No unfortunately, it all went down rather quickly (the 100 kWh announcement) and there was a media embargo on it while it was live. I don’t think anyone really was prepared well enough/got a chance.

All of that is hard to believe in this (veritable) day and age.

Why you used bottom cooling plate? Heat is always striving upwards, and chill down. Convection will be better if you install the cooling plate on the top of the battery cell, then you can achieve a much better cooling. Why air-conditioning engineers do not know about this?

Because its only a half truth only related to convection. This is conductive cooling. Rest assured they know what they are doing.

Oh thank goodness, I was worried for a while there.

“…they know what they are doing.”

Perhaps they know what they’re doing in their area of experience, but whether that knowledge and experience is applicable to an EV battery pack is questionable. They apparently don’t know much about the innards of the Tesla battery pack.

Perhaps the batteries are placed upside down, making the bottom cooling plate actually a ceiling…

Also, the lower part of the battery block can be created in the form of labyrinths radiator fins to remove heat from the heated fluid and uses the power of the incoming flow of the wind under the vehicle

Remember that there is a cold season when you don’t want the air to be cooling the pack so cooling fins wouldn’t be desired.

The cold season is the BEST time to air-cool the batteries! Every liquid-cooled system in any car eventually needs air-cooling – otherwise the liquid would keep getting warmer until it reached the boiling point. That’s why ICEVs have huge radiators (and still sometimes boil over).

Air cooled is bad in any temperature. With active cooling, it can be heated to bring the battery up to temperature (ie, while plugged in and charging). Liquid cooling has thermal momentum and some degree of insulation to prevent wild temperature fluctuations. Air cooling is simply inadequate. It might be ok for garden tools, but not for EV.

Yup. The concept of how rapidly waste heat can be transferred out of the pack seems to be lost on many who are posting here. Conduction via flowing fluid in a tube is a far more effective way to transfer heat than either passive radiation or air cooling, which is what far too many comments here are suggesting.

Passive radiation is what the Leaf uses, and I think most of us know that’s simply inadequate.

Furthermore, the battery pack certainly does need heating in very cold weather. The best way to do that is, just as with cooling, to circulate a fluid (liquid or gas) thru the battery pack. The Leaf does have a battery heater, but obviously it can’t work as well as the heating systems in battery packs which circulate a fluid.

The battery pack photo illustrating this article on the home page is not a Tesla automotive pack. It’s the battery tray used in a Powerwall or Powerpack stationary storage product.

I suspect the photo was taken at the recent Gigafactory grand opening event during the tour of the stationary storage product assembly line.

/true, those are 21-70s out of a PowerPack…but we have to put something on the homepage, (=


By the way, I’m pretty sure those are 18650s. They aren’t making cells yet at the GF so those are still imported. The math also works out if you count them up and assume that they are lower energy-density 18650 NMC cells (which the stationary products use). If they were 21700s, the energy density would be implausibly low since that tray goes into a 6.4 kWh Powerwall.

Actually, the they are 21-70s (not built, but assembled at the GF)- confirmed from the tour. What you are seeing is a module tray (16 in a PowerPack). Here is a better look at it:

They weren’t yet making any cells at the Gigafactory at the time of the tour and I’ve seen nothing to indicate that Tesla was importing 21-70 cells from Panasonic (which has not previously made that size) or Samsung for Tesla Energy products.

If those are truly 21-70 size cells my guess is they would be dummy cells, maybe injection molded from plastic and spray painted silver. They must have had hundreds or thousands of dummy cells made, surely, to figure out such configrations.

Jay, I believe you were misled. Powerpacks and Powerwalls all still use 18650s, not the new 2070 size. You’ll notice the configuration in the pic is exactly the same as 85kwh cell modules (though not the same chemistry. Engineers on the tour told me that they intend to stick with 18650s for Stationary Storage while only the model 3 gets the new cells, at least at first…

Then again, what do I know? I missed the change to 21-70! 🙂

I have a hard time understanding why cells are cylindrical instead of rectangular like the flat camera batteries. That little space between round cells is approximately 1/5 of the volume taken within a battery pack!

I believe (someone please correct me if I am wrong) that Tesla used the 18650 cylindrical battery because they were a very common and cost effective solution. They are most commonly used in laptop computers (well, I guess Tesla is using more of them now…)

Yes. And the cylindrical form factor of the cells lets Tesla put “goop” in between the cells for better passive cooling; the round shape naturally creates spaces to vent the heat. Flat cells stacked together (prismatic or pouch cells) create much more of a problem with venting heat. That’s why, for example, the Dreamliner airliner’s li-ion battery packs kept catching fire; they were too tightly packed and had no cooling system.

The Volt battery pack from GM, and perhaps packs from other EV makers, do use flat-sided cells, but GM uses a refrigerant-based cooling system, which can transfer heat faster, but uses much more energy driving a compressor than Tesla’s cooling system, which needs only water pumps.

GM uses coolant. Specifically, 2014 SparkEV used bottom plates with coolant, which is very similar to Bolt.

BMW uses refrigerant. Since refrigerant would (could) move more heat, I doubt there’s large energy penalty compared to coolant based system.


My understanding, based on a lot of reading on the subject, is that the Volt 1.0 uses only a refrigerant based system for battery pack cooling — which does qualify as “coolant” — and the Volt 2.0 uses both refrigerant and glycol/water (antifreeze) cooling loops, presumably switching between them as needed.

If this is incorrect, please cite your source(s).

Mea culpa… I was wrong here, and you are right, Sparky.

On the positive side… I learned something today! 🙂

The original reasons Tesla went with 18650 cells has little to do with any technical merits over other formats. They were simply the best available from both a cost and energy density point of view. At a time where making batteries was not a part of Teslas plans I don’t think they really made a big effort to try and figure out whether other formats could theoretically have advantages. That said, it’s not like pack density is irrelevant in other applications. Laptop users want as much battery life as possible too. Cylindrical cells had come to dominate because of faster and thus cheaper manufacture. This doesn’t necessarily mean they will remain so. It isn’t for me to say that pouch cells are impossible to make as cost-effectively – and I don’t think it’s for anyone to say when we consider that entirely new chemistries may appear some day. However, it seems to me that the cylindrical format probably isn’t wasting space since the cooling requirements even the cylindrical cells cannot in practice be packed as densily as their physical shape allows. I think pack density will continue to increase mainly due to increases in cell density. If internal resistance (and… Read more »
EV makers other than Tesla chose to use larger format flat cells (pouch or prismatic) because it was cheaper per kWh to make fewer, larger cells. However, in practice this did not work so well for them, as the larger cells have lower volumetric energy density, requiring more space inside the car, which meant either a larger heavier car or else a smaller (and thus lower capacity) battery pack. Furthermore, when you have a single cell which holds perhaps 1/2 kWh, then losing a single cell is a significant loss of capacity. Tesla’s packs are designed to be able to have a few cells go bad, but cut those cells out of the circuit, so the pack experiences very little loss of functionality or capacity. As far as the form factor, cylindrical vs flat: As you yourself have noted, Tesla went with the 18650 form factor primarily because it was the cheapest way to get a lot of kWh of batteries. Perhaps Tesla made a virtue of a necessity by using the space between cylindrical cells for venting heat. However, I seriously question your claim that Tesla didn’t test other types of batteries before settling on the 18650 form factor.… Read more »

Increasing the nominal voltage of the cell will reduce the amount of heat produced internally.

Tesla is already working on that idea.

So reducing the amount of heat created internally in the cell and better cooling will lead to a denser more efficient pack.

Well, the *nominal* voltage matters less than the *actual* voltage..! 😀

Cell voltage is not easy to change though – it follows directly from the chemistry of a cell. That’s why you can easily design for a particular cell capacity (by changing the cell size), whereas the voltage is simply given by the chemistry.

Alaa said:

“Increasing the nominal voltage of the cell will reduce the amount of heat produced internally.

“Tesla is already working on that idea.”

Nothing in the article you linked to supports your assertion that higher voltage cells emit less waste heat. That may be true, but the article doesn’t support your claim.

I’m surprised to see two experienced engineers suggest that better cooling can be had by putting a cooling plate against just the bottom of the cells. That’s going to be worse cooling, not better; that is, the heat wouldn’t be transferred as fast with that configuration, because (a) it would put the heat transfer farther away from the center of mass of the cell… where the most heat will build up, and (b) it would limit the surface area where the heat transfer was taking place to only the cross-section of the cell, rather than the larger surface available at the sides of the cell. If you look at teardown of the original Tesla Model S85, you can see there are gaps between some of the rows of cells (photo linked below). Tighter packing can be achieved by simply moving the rows closer together. And since flattened tube (shaped more like a ribbon) that carries the coolant past each cell is only in contact with about 25% of the height of the cell, better cooling could be achieved by running two or even three tubes past each cell, instead of just one. Or, perhaps they could simply use a wider… Read more »
I think that they were suggesting a bottom cooling in addition to the cooling from the side contact aluminum fin between the cells or the full individual aluminum cylinders around each cell, not bottom surface cooling on its own. On my own I proposed to ditch the cylinders for curved triangular aluminum pins instead matching the ones that naturally occur between cells when packing them to the max. That would make use of a normally lost opportunity to carry away heat and not give extra spacing between the cells. It would also be quiet good because the pins would be massive meaning the plane cross section which is a measure of heat carrying capacity would be rather high. The pins would be in direct contact with the bottom plate where coolant flows to further carry away the heat outside the pack. That curved triangular pins pattern could be made by taking a 70 mm thick Aluminum plate and drilling 18 mm diameter 65 mm deep holes in it at every supposed cell location. That would be for a test item and later that same pattern could be produced industrially by using a mold or some other technique. Or perhaps the… Read more »

How about a top and bottom plate with hollow pins? The coolant could then flow through the pins and the plates.

Yes also possible but what is saved in complexity by not having the tubes inside the pins will come at the price of having a more difficult fluid leak prevention. With the cylinders the flow is only between the 3 bottom plates, the cells are completely free of Fluid above and can even be inserted long after the 3 bottom plates have been soldered together for complete leak safety or glued together for still excellent safety. Only one fluid entrance and exit remain.

2014 SparkEV also used cooling plates at bottom. For 2015/2016, they changed to channels between cells like the Volt. Based on some forum comments on DCFC, 2014 seem to DCFC (ie, cooling) just as well as 2015.

The Spark EV is not a high-performance car, and cannot be charged at Tesla Supercharger speeds. So the battery pack cannot be charged or discharged nearly as rapidly (in kWh per minute) as the Tesla Model S.

Therefore, it doesn’t need nearly as robust a cooling system.

I used a similar concept for cooling the cells in the battery pack on my Institute’s Formula Student car. Coolant was passed through the bottom plate of the pack, while the heat from the cells was conducted to the bottom plate through the partitions between each row of cells. Worked like a charm, plus we were able to have an incredible packing efficiency because of that. The partitions were a part of the battery housing itself, so need to wrap any aluminium fins/ribbons around the cells. Saves the complexity in assembly, but difficult to manufacture