Tesla Semi Truck Battery Probably Lighter Than You Think – Part 2

Tesla Semi


Tesla Semi

Tesla Semi

IF the Tesla Semi uses standard Model 3 cells, it should be lighter than most might assume.

*This is the second of four articles on Tesla’s semi truck. In this article, we show that the truck battery should be lighter than many think using standard Model 3 cells. Click here to see Part 1.

When estimating battery pack weight, many rely on published numbers for Tesla’s Model S battery weight and kWh’s. Typically, they use Wikipedia’s 1,200 lbs. Model S battery pack weight (not a bad number) and divide by 85 to get a weight per kWh, then just scale that number up to semi truck pack kWh’s.

That gets you a rough estimate, but if you look at the pack in a little more detail, you find that there are some synergies that make the truck pack lighter per kWh than the Model S pack.

It’s accepted that Tesla’s NCA chemistry is somewhere around 240-260 Wh/kg (8.5 lb/kWh). This means 85 kWh of cells weighs 723 lbs. Wikipedia lists the Model S pack as 1,200 lbs. That’s a big difference. Of course, there’s a lot more to the battery pack than just the cells: there’s wires, busbars, cooling tubes, and a big protective case. Each of these subcomponents of the battery pack scale to truck size in a different manner. For example, we can’t just scale up busbar and wire weights because the semi truck is running a much higher voltage. Also, the protective case scales by area and thickness, not kWh. The model S pack is thin and flat with lots of surface area while the truck pack is shaped more like a cube.

Tesla Semi Slide1

Tesla Semi Slide 1

The details of the scaling procedure are in the second part of the article if you care to delve into it. However, to make a long story short, if we scale these subcomponents of the pack up as described we end up with a 21% lower weight pack on a per kWh basis. While the Model S pack weighs in at 13.5lb/kWh, the Tesla semi pack weighs only 10.6lb/kWh. That’s a 21% improvement! Total pack weight estimates at 9,549 lbs. for 900 kWh.

Tesla Semi Slide 2

Tesla Semi Slide 2

Detailed Analysis:

As discussed above, we broke the P100D pack into separate parts and scaled up to truck size.

Tesla Semi Slide 3

Tesla Semi Slide 3

Each part was then scaled to truck size as follows:

Tesla Semi Slide 4

Tesla Semi Slide 4

Our analysis assumes the semi truck pack consists of 48 Model 3 modules. The modules are laid 8 wide and 6 layers deep.

Tesla Semi Slide 5

Tesla Semi Slide 5

Luckily, we have single module weights via Jason Hughes HSR parts catalog and EV West parts catalog. The P100D module is listed at 66 lbs. Subtracting cell weights in the pack from the weight of 16 Modules we derive the weight of the cooling tubes and busbars. That gives us 189 lbs. for the busbars and tubes. Busbars and wiring were arbitrarily set at 50% of 189 lbs., and cooling tubes also arbitrarily set at 50% of 189 lbs.

Tesla EPA documents list the total pack weight of the P100D pack as 1,375 lbs. The difference between the 16 module weights and the pack weight gives us the weight of the case (320 lbs.).

Tesla Semi Slide 6

Tesla Semi Slide 6

Cooling tubes were then basically scaled by kWh. Busbars were scaled by kWh minus 50% because of the higher voltage in the semi truck pack – a big savings. The protective case was scaled by surface area. Note that while the pack is nine times more kWh’s, the case area only increases by a factor of two because of the shape – another big savings.

Tesla Semi Slide 7

Tesla Semi Slide 7

Stay tuned for our subsequent articles in this four-part series.

*Keith Ritter coauthored this article and contributed to the analysis.

Categories: Battery Tech, Tesla

Tags: ,

Leave a Reply

59 Comments on "Tesla Semi Truck Battery Probably Lighter Than You Think – Part 2"

newest oldest most voted

Interesting analysis. Don’t forget once your done with the truck to analyze the new Roadster, because I’m very curious about the math behind fitting the big battery of a S/100 in the much smaller Roadster…twice! Especially after claiming that similar sized Model 3 can’t take more than 75KWh… It’s not just size that seems problematic, averaging pack level energy density of Model S with that calculated in this article for the truck weight would be a whopping ~2400lbs. Seems like a lot of weight for a great handling, under 2 second 0-60 sports car to drag around.

So one has to assume Tesla is betting on a significant increase in volumetric/gravimetric energy density to happen in the coming years that would change the math for the trucks as well.

re: Roadster. It’s in on the list. Stay tuned.

I’m very confident Roadster will use a much higher density, much more expensive nextgen battery. Probably solid state.

A lot of SS designs struggle to deliver high power. That’s why they had to use a 200 kWh pack. They probably hope it will be smaller by launch (150 kWh should hit their range target in that size car).

I’m also confident the prototype at the launch event had a different battery, e.g. Toshiba SCiBs.

If true, Solid state could be very well received by the track community, as heating problems have been a function of inertia (both braking regen, and KW out of turns). I guess it depends on how lower maximal power, per KWh, is offset by higher storage?

Another place for hope with 200KWh is the lower C-rate, to push the car around. A fixed KW output (say 300-400HP, or 223-300KW) becomes less taxing on a bigger battery. The comparison to 90KWh of, say, LiPo (cough, Porsche) could be interesting.

On top of George’s excellent contribution, I wonder if the structural nature of Tesla’s car-battery tray would allow for additional weight loss, independent of surface area. Wasn’t it designed to do more than just hold cells?

Tesla’s car packs are structural, but I think they’re mostly just shear panels which are typically quite lightweight. I think the shape issue George raises is a much bigger deal.

“…I wonder if the structural nature of Tesla’s car-battery tray would allow for additional weight loss, independent of surface area. Wasn’t it designed to do more than just hold cells?” Yes. In both the Tesla Roadster and the Model S/X, the battery pack is part of the structure of the car, providing additional stiffness. It certainly seems there is a potential for saving weight on the battery casing for the Tesla Semi Truck, and not merely by improved packaging. I appreciate the estimate here, but I don’t regard that as the last word. I would be surprised if Tesla doesn’t find some methods of shaving weight here and there vs. just stuffing in 7 or 8 or 9 or 10 Model 3 battery packs; some methods that George and Keith did not consider. I guess George and Keith are ignoring the weight of the pumps which circulate the coolant. Perhaps those are external to the pack, but that’s certainly one way in which the Semi Truck’s powertrain is going to weigh less than just sticking in a bunch of parts from the TM3 without modification. Seems almost certain that the Semi Truck will use larger and fewer pumps to circulate… Read more »

“I’m very confident Roadster will use a much higher density, much more expensive nextgen battery. Probably solid state.”

Contrariwise, -I- am quite confident that no vehicle intended to be produced by Tesla in just two years is going to use substantially different batteries than the current 2170 Gigafactory cells. We haven’t yet seen solid state batteries in any consumer product, and that tech will certainly appear in such devices as cell phones and laptops before the prices on those fall enough to put them into PEVs.

Yes, the battery pack(s) in the Roadster Mk II will be quite heavy, just as the pack in the Roadster Mk I was quite heavy. That didn’t stop the Roadster Mk I from having the performance and responsiveness of a sports car, and it won’t stop the Mk II either.

I’m also confident that Tesla has not stacked a double layer of 2170 cells underneath the floor of the Roadster Mk II. There is plenty of space under the hood and behind the rear seat to stuff batteries. The Roadster Mk I had no batteries underneath the floor (they were all behind the seats), and the Mk II may not either.

Tesla Is working with the “New Solid State Lithium Battery” on these New Vehicles which has at Least 3 X the storage Capacity of the Present Li Battery that we are familiar with..Once Tesla is Ready, They will Let it Be Known..”Believe Me Later”

Solid state lithium probably won’t be commercialized until too late for Semi truck or roadster, unless Tesla is ahead of Toyota in that regard, and several industry leaders have indicated Toyota will probably be first to market with an SSB

I think Samsung will be first to market with SSBs and will first appear in their phones. I do not believe Toyota will be the leader on this, that is just Toyota’s excuse for failing to lead now.


‘Graphene’ in place of ‘Graphite’ would significantly reduce weight and volume of the electrodes (and therefore increase energy density of the cells) while still using proven Li-ion chemistry. Solid State metallic Lithium is on the way but probably not commercially available for 2019.

Graphene has the same chemistry as graphite; just a more ordered structure. AIUI there was never hope for significantly higher capacity with graphene — only faster charging times. But AFAIK it has been generally acknowledged by know that this didn’t pan out either.

I think you have some good ideas here. Because of the massive scale of the battery pack it probably does make sense to wire it at double the voltage. That also simplified things with charging since the Semi has its own dedicated charge plug that they can probably assume will always support 800+ volts. They have implied that they are using the same motors as the Model 3 and I’m guessing a redesign would’ve necessary to run them at ~800 volts so I guess the inverter steps that down?

An alternative to the inverter stepping down the voltage would be to wire the two motors on each axle in series. Does a semitractor need to be able to use yaw forces to help with steering?

Tesla said each motor/wheel assembly would be individually controlled using “torque vectoring” in lieu of differential gears. This also gives better traction control to help prevent jack-knifing and in snowy conditions.

The prototype had differentials, though. There are some good reasons to use differentials, e.g. the ability to route all torque to one side in low traction conditions.

My understanding was that the semi had 4 motors and two axles, so one motor for each wheel on each side at the rear of the tractor. No motors for the steer wheels. With separate motors and controls on each side of an axle, differentials become unnecessary. The outside wheel in a turn is simply run at a higher speed to account for the longer path the wheel must follow. This also allows anti-jacknife yaw control via the separate motor controls making it a pure software control problem with no torque transfer across a differential to account for. Also remember the different gear ratios for the two pairs of motors, 23:1 for the low speed and 15:1 for the high speed. An interesting control problem and clever optimization to improve operation.

Something similar might be going on with AWD in the S and X models. Anyone know if the front and rear axles have different reduction ratios in those vehicles?

Exactly. See article and Kman video link here to see the all the drive units and driver axles from the reveal. No differential, but 15:1 and 24:1 triple-reduction gear box for each driver wheel motor.


Kman thinks the big unit between the motors is a differential:
The differentials are extremely “Beefy”

I agree. I understand the simplicity of individual wheel motors. But a differential gives you more capability.

Tesla does indeed build cars where the front and rear drive ratios are different.

Elon explicitly stressed that redundancy benefit of using independent motors — so definitely no differentials.

Since peak power output of electric motors is higher than sustained output, temporarily routing more power to the wheel with more traction is perfectly possible *without* a mechanical differential. And the electric drive is not torque-starved to begin with.

I’m pretty sure the large axle assemblies seen on the prototype are *not* differentials. Even ignoring everything said above, the geometry just doesn’t look right IMHO.

We will see how long the packs last with heavy charge and discharge daily for years.

Considering the number of MSs with 200K+ miles, and doing fine, I am going to guess that trucks will do just fine as well.

We will see, Model S do not supercharge every day, they do not recommend that.

Other articles have stated, or at least suggested, that the Tesla Semi Truck will use the type of battery cells which Tesla puts in its PowerPacks. Those are designed to be cycled (if I recall) 80% on a daily basis, far more than Tesla’s car batteries are intended to be cycled.

It’s a different chemistry, I think. Hopefully Gigafactory One will be able to produce 2170 cells with both chemistries, altho there is at least some indication that at present it’s only producing a single chemistry. That Tesla is projecting such a low price for the Semi Truck certainly seems to suggest Tesla believes it can use cells from the Gigafactory for the truck.

Gigafactory started producing NMC 21700 cells for Powerpack 2 / Powerwall 2 even before it started producing NCA cells for Model 3 — so definitely no problem there.

I’ve also seen the claim that Semi will use NMC; which seems very plausible, considering that 1,000,000 miles needs 2000 full cycles — much more than any Model S I’ve heard of yet. Also, the geometry seen in the presentation makes it more likely that they are using vanilla or slightly modified Powerpack/Powerwall modules, rather than anything based on Model 3.

Good work, guys. I’ve always used:

4 kg/kWh for cells
5 kg/kWh for semi pack
6 kg/kWh for S/X/3 packs

These numbers are easy to remember and within a few percent of your detailed analysis.

great minds think alike;)

Do you have thoughts on the Roadster pack?

I added some Roadster2 battery thoughts above. I think it’s a nextgen chemistry. I linked to a Solid Energy spec sheet as one example.

200 kWh really leaped out at me during the unveil. The only reason I can see for such a huge pack is to compensate for low C rate and cycle life. Both are common issues with nextgen solid state chemistries.

Nextgen chemistry is also the only reason I can see for the $250k price. Nothing else in that car is terribly expensive.

Contrariwise, I presume the huge battery pack for the Roadster Mk II is to allow it to produce high levels of power without overheating, thus allowing the car to run at high speeds and with repeated jackrabbit accelerations without having the pack overheat. A very high capacity also will allow the car to be fast-charged quite rapidly, in terms of miles of range added per minute.

Again, I don’t think it’s realistic to expect “next generation” battery tech from Tesla (or Panasonic) as soon as 2020. If it was that close on the horizon, why would Panasonic be investing billions of dollars on current battery tech at Gigafactory One?

Pushmi – “I presume the huge battery pack for the Roadster Mk II is to allow it to produce high levels of power without overheating,” If you use the same chemistry you generate the same amount of waste heat. Spreading it over more cells makes cooling easier (more surface area), but Tesla has easier ways to double cooling effectiveness (e.g. immersion like Kreisel instead of cooling tubes). If Roadster2 used existing cells they’d ship it this year, not 2021. “If (solid state) was that close on the horizon, why would Panasonic be investing billions of dollars on current battery tech at Gigafactory One?” Cost. First solid state cells in phones will be $1000++/kWh next year. That’s $10-20/phone, easy to justify for a premium model. By 2020-21 they’ll be ~$500/kWh. They won’t be $100/kWh until 2025+. Panasonic will have more than recouped their GF investment by then. Fast charging is kind of pointless in a small car with 600+ mile range (probably closer to 800). This isn’t a highway cruiser. The point of the original Roadster was to show the future. It used very expensive cells to make a visceral case that EV drivetrains were superior and would take over the… Read more »
Doggydogworld said: “If you use the same chemistry you generate the same amount of waste heat. Spreading it over more cells makes cooling easier (more surface area), but Tesla has easier ways to double cooling effectiveness (e.g. immersion like Kreisel instead of cooling tubes).” I dunno anything about Kriesel immersion, and anyway I think Tesla would rather stick to what it knows and what has proved to work reliably. But you’re ignoring three factors, including the one (#1) which I think is the most important one: 1. The amount of heat transfer possible in a cooling system. Making the pack bigger doesn’t just spread out the amount of waste heat more, it also gives more room for a greater amount of coolant, and more thermal mass in the coolant, while maintaining the same space between batteries. This allows more heat to be transferred to the radiator per unit of time. 2. Ability to handle more power. With a larger battery pack, more power is available (both for acceleration and for regenerative braking), which means a more powerful inverter and other power electronics. With those being designed and built to handle more power, they will have lower resistance and greater thermal… Read more »

Photos with the doors open, on top of several reports from people who took a test drive, clearly show that the new Roadster has a comparatively high floor level, easily accounting for two layers of cells. It is not nearly as low as the original Roadster.

Limited production numbers are what make this sort of car expensive. Frankly, the announced price sounds like a steal; it would be way more expensive if it used anything significantly deviating from Tesla’s established technology.

Interesting Napkin math guys – but this isn’t lighter than I think.

What I wonder is if they are using any more conservative design principles compared to the Model 3. Trucks have to endure being pushed to their capacity limits much more than just a small car.

re: napkin. This is the 21st century: It was an Excel spread-napkin;). That HAS to make it more accurate. As Jay would say…\science:-) That’s a good question: on the mechanical side – if fatigue from long-hours at design-HP have been factored in. I did notice from the Kman videos that Tesla had some very large coolant hoses going to each motor/inverter drive unit – a lot bigger than they have inside the S or X. Cooler equal happier electronics/bearings. Kman noted standard semi-truck suspension items. The axles structurally appeared similar to standard semi-driver axles. I suspect they have a few experienced heavy-equipment driveline engineers on-staff and/or are working with Dana to sort out those details. Same for the power electronics. Regarding pack life – I’m sure every EV and battery manufacturer, including Tesla, GM, LG, Panasonic, etc. has massive battery testing labs and complete test data on the cells and packs going through all kinds of operating cycles, including daily deep discharge/recharge cycles. So they probably have that figured out. And why it is unlikely they would go with an exotic new cell chemistry without a lot of long-term life testing. But the ultimate proof-of-concept is to get a few… Read more »

I hear the words, but if that is so why does every Tesla model that arrives has problems with the doors?

Other companies seem to get them to work ok, as did one German who for 50 Euros will give you a stainless steel rack for your Model S, whereas Tesla just keeps putting the same old crap back in for $800 Per door each time.

As far as reliability goes – never a Tesla strongpoint in the first place, I’d think ultimately sales of these trucks is going to be an uphill battle.

That foray into ‘Flamethrowers’ ultimately will be seen as the DUMBEST thing Musk has ever done.

Besides sounding like he was smoking too much pot before releasing the truck “it will blow your skull out of the universe” – that kind of thing – anyone considering this truck will be thinking:

Oh, yeah, Tesla – thats the company that resells those ‘flamethrowers’ for 9 times the price – we wonder what next junk toys they’re going to come out with next?

The not-a-flamethrower is not a Tesla product.

Nissan did ask the testing, how did that work out for them?

“…the Tesla semi pack weighs only 10.6lb/kWh.”

Thank you George Bower!

This is the sort of refinement to the estimates which I hoped to see, when I posted what Domenick Yoney called my “Napkin Math 1.0” estimate of a generic (not specifically Tesla) BEV Semi Truck feasibility study.

My ballpark estimate for battery weight was 11.5 lbs/kWh, but I expected the actual weight to be lower using 2170 battery cells from Tesla’ Gigafactory One.

George, thank you very much for taking the time to “drill down” on a more accurate estimate! (And I don’t care if some think the phrase “accurate estimate” is an oxymoron. I think it’s entirely appropriate in certain cases… such as this one.)

It isn’t likely they’re using higher voltages as they’re using same motors.

More likely 4 parallel independent systems each running a motor with enough interconnection to balance charge / discharge imbalances.

This is supported by pictures of the charging port that shows 8 interconnects, likely in pairs.


i have a question we know they are using the 21700 Cells for the model 3, we assume they are using the same 21700 Cells for the Tesla Semi but we base the math on the model S 100D pack witch uses the 18650 cells what am i missing?

These calculations are based on the assumption that Semi uses Model 3 NCA cells, which is quite possibly wrong… I don’t remember where the claim that it uses NMC cells instead came from, and thus how official it is — but it certainly sounds plausible, considering the promised 2000 full cycles, with regular Megacharging. This is reinforced by the geometry of the pack as seen in the animation shown during the unveil (and on Twitter): it looks very much like it simply uses two stacks of Powerpack 2 / Powerwall 2 module pairs standing on edge… Assuming that the Powerpack 2 / Powerwall 2 actually uses pairs of modules in each pod. (Is there any information available on that? I haven’t seen anything thus far…) Judging by photos of the Powerpack 2, my estimation is that each pod is less than 85 mm high, with 40 mm gaps between them. Assuming the modules in the Semi are packed tightly instead — like in the original Roadster or Smart ED — and the width is around 2.2 m (as estimated by someone on the TMC forum), that would fit around 26 module pairs in each stack; or maybe a few more.… Read more »