Welcome to Tesla Motors Club
Discuss Tesla's Model S, Model 3, Model X, Model Y, Cybertruck, Roadster and More.
Register

Investor Engineering Discussions

This site may earn commission on affiliate links.
I'd think that as cells grow larger, single ended cooling becomes less usable if you expect high C rates. The temperature gradient becomes a bigger issue the further the cooling point gets from the far side, and single end (i.e. bottom) cooling is going to be incredibly worse than side cooling as the cells get larger, unless they end up wider than they are tall somehow.
Sure, the gradient gets 'worse', but how much worse does it need to get before it becomes 'bad'?
Side cooling requires heat to travel through all the intermediate heat producing layers and the jelly roll can non-interference-fit interface. End cooling gives a direct path for every other layer to a thicker end heat spreader.
Also, there would be the expectation that C rate would decrease as cell size increases (for similar design structural packs with more energy).
 
I'd think that as cells grow larger, single ended cooling becomes less usable if you expect high C rates. The temperature gradient becomes a bigger issue the further the cooling point gets from the far side, and single end (i.e. bottom) cooling is going to be incredibly worse than side cooling as the cells get larger, unless they end up wider than they are tall somehow.

This might have been feasible with a 16850 or 2170 cell that had single tab but I suspect 4680 in high performance applications it won't work.

Probably fine in storage and lower end vehicles however so we might see it in fit example the next gen vehicle, of supercharging speeds aren't a problem for it (i.e. lighter/smaller translating to smaller pack and less needed C rate for x% per y minutes goals - would be still lower kW speed than a 3/Y for example but might be similar "mph" charging rate), so if it's a significant cost savings and the effective charging speed is not greatly impacted, might see use.
With the tabless electrode it doesn't get worse, that is the whole point

It allows you to grow cell diameter "unlimitedly", between quotes because a bigger cells isn't always the cheapest and also manufacturing becomes a pain with a really long jelly roll
 
How thermally conductive is the jelly roll top to bottom vs through it side to side? Is it good enough that the far end in a larger cell like a 4680 still gets better cooling with bottom plate cooling vs side cooling? Obviously these are questions we can't really know the answer to unless Tesla decided to divulge them, but certainly the shorter the can, the easier cooling the top from the bottom will be, whereas side cooling it is instead the diameter that impacts cooling capacity.

So there must surely be a crossover point where bottom plate cooling can't keep up with cooling the far end vs side cooling, if you go too tall. Bottom plate cooling will be favorable more towards fat (large diameter) squat (short height) cells vs side cooling being the opposite (favoring tall skinny cells). Since we haven't seen them use it yet for 4860, which is relatively a fat squat cell, we can only for now assume that the side cooling method is still more advantageous, at least for the necessary cooling performance on current models.

Tradeoffs might lean in the other direction if assumptions are changed (i.e., next gen cheap car might have smaller battery due to lighter vehicle and thus lower C rate, and if so, should have lower heat dissipation needs for a still useful "miles per hour" charging rate, even if the actual kW rate is lower than a 3/Y). Certainly a bottom cooling plate would be obviously cheaper to make than the side cooling extrusion, at least from an outsider's perspective, so the fact they still aren't taking that path implies it is not the ideal solution for their needs, and sufficiently so that the cheaper production cost doesn't outweigh whatever cooling advantages the side cooling method has.
 
  • Like
Reactions: GhostSkater
How thermally conductive is the jelly roll top to bottom vs through it side to side? Is it good enough that the far end in a larger cell like a 4680 still gets better cooling with bottom plate cooling vs side cooling? Obviously these are questions we can't really know the answer to unless Tesla decided to divulge them, but certainly the shorter the can, the easier cooling the top from the bottom will be, whereas side cooling it is instead the diameter that impacts cooling capacity.

So there must surely be a crossover point where bottom plate cooling can't keep up with cooling the far end vs side cooling, if you go too tall. Bottom plate cooling will be favorable more towards fat (large diameter) squat (short height) cells vs side cooling being the opposite (favoring tall skinny cells). Since we haven't seen them use it yet for 4860, which is relatively a fat squat cell, we can only for now assume that the side cooling method is still more advantageous, at least for the necessary cooling performance on current models.

Tradeoffs might lean in the other direction if assumptions are changed (i.e., next gen cheap car might have smaller battery due to lighter vehicle and thus lower C rate, and if so, should have lower heat dissipation needs for a still useful "miles per hour" charging rate, even if the actual kW rate is lower than a 3/Y). Certainly a bottom cooling plate would be obviously cheaper to make than the side cooling extrusion, at least from an outsider's perspective, so the fact they still aren't taking that path implies it is not the ideal solution for their needs, and sufficiently so that the cheaper production cost doesn't outweigh whatever cooling advantages the side cooling method has.

What is important to understand with the 4680 with tabless electrode, is that side cooling is still mostly bottom cooling

The most conductive direction is axially due to the tabless electrode, once the heat reaches the bottom, it travel though the sides of the can to the portion with the cooling ribbon, specially since the 4680 can is significantly thicker than a 18650/2170

Sure, part of the heat will conduct radially, but not the most part due to the thickness of the jelly roll

The limitation is still the interface between the cell and whatever cooling ribbon/plate is being used, which is usually considerably less conductive, if the ratio between the stainless steel to the adhesive is big, it's worth to have the heat travel a longer distance through the stainless steel in exchange for a bigger area for the adhesive
 
I have a salt question for you guys: how are EVs holding up in northern climes where salt is used on highways? In particular I was hoping for some response from Norway since they've made such a big push into EVs.
My understand is the main thing is to lubricate the brake calipers annually in heavy salt areas.
 
I have a salt question for you guys: how are EVs holding up in northern climes where salt is used on highways? In particular I was hoping for some response from Norway since they've made such a big push into EVs.
I’m starting winter number six with my TM3, no corrosion issues noted (except normal brake disk flash).

I did spray the chassis and inside the trunk lid with “Krown” rust spray the summer I got the car in 2018, but there are a plethora of old school TM3s like mine (chrome window trim and chrome door handles) around here with no rust at all, even in the usual tough for rust spots in the area between the rear wheel well and aft edge of the rear pax doors (doglegs).

I do lubricate the brake caliper slider pins once a year (and test the moisture content of the brake fluid every two years via a test strip dipped into the master cylinder reservoir).
 
The calculator states that it's based on averages, however we know that compared to ICE ...

Yes, and that power calculator also states its based on power at the flywheel. Since ICE drivetrains have perhaps ~35% higher drivetrain losses vs Tesla EVs, that decreases wheel horsepower more for ICE. So ICE needs to burn extra gas (and waste more heat) to make the same power at the wheels as a Cybertruck.

Since Sir Isaac tells us F=MA, we can calculate CyberBeasts avg power in the 0-60 mph run:
Then to convert Newtons force into horsepower, we use the formula:
  • Power = Force * Velocity
  • Force 32,833 N * 60 mph or 26.822 m/s = 881 KW
  • that's 1,181 hp for Cyberbeast acceleration, after all losses. :D
So Cyberbeast 0-60 mph tests show 14% better powertrain efficiency vs an ICE vehicle, which would require ~1341 hp at the crank for the same performance.

Note: OP assumed a 'best-case' use of a dual-clutch transmission for ICE power calculation, which is not likely for a 4x4 truck like a Ford Raptor. Instead they use automatic transmissions, which would decrease their driveline efficiency vs a dual-clutch setup.

Whew! Cheers!

Cheers![/S]
 
Last edited:
  • Like
Reactions: jerry33
Yes, and that power calculator also states its based on power at the flywheel. Since ICE drivetrains have perhaps ~35% higher drivetrain losses vs Tesla EVs, that decreases wheel horsepower more for ICE. So ICE needs to burn extra gas (and waste more heat) to make the same power at the wheels as a Cybertruck.

Since Sir Isaac tells us F=MA, we can calculate CyberBeasts avg power in the 0-60 mph run:
Then to convert Newtons force into horsepower, we use the formula:
  • Power = Force * Velocity
  • Force 32,833 N * 60 mph or 26.822 m/s = 881 KW
  • that's 1,181 hp for Cyberbeast acceleration, after all losses. :D
So Cyberbeast 0-60 mph tests show 14% better powertrain efficiency vs an ICE vehicle, which would require ~1341 hp at the crank for the same performance.

Note: OP assumed a 'best-case' use of a dual-clutch transmission for ICE power calculation, which is not likely for a 4x4 truck like a Ford Raptor. Instead they use automatic transmissions, which would decrease their driveline efficiency vs a dual-clutch setup.

Whew! Cheers!

Cheers![/S]
Um. I looked at this a bit and little bitty alarm bells went off in my head. This is what happens when one who has been through Physics 161, 30-odd years ago, One Remembers Stuff.

First little alarm bell: Yep, F = Force in Newtons = Mass*Acceleration. And the units are kg*m/s^2 -and those are the units for a Newton. So the force is 32,833 Newtons, not "Newton/s".

Next, Power. P(t) = dEk(t)/dt; that is, the power is the time rate of change of the energy. That is a derivative function.

Next, Energy. Kinetic Energy = Ek. Ek(t) = mV(t)^2.

So, we'd like to find the average power.

P(t1) - P(t0) = (E(t1) - E(t0))/(t1-t0) = (m*V(t1)^2 - m(V(t0)^2)/(t1-t0)

Well, V(t0) = 0 (we're stopped), t1 - t0 = 2.6s. m = 3183 kg. But P(t0) isn't zero, it's a derivative function. P(t0-) is zero, since we haven't stomped the pedal yet. Hmm.

So, what I got, then, is 3183kg*(26.8m/s)^2/(2.6s) = 879291W. Let's see, there's 746W/hp, so that would be 1179 HP. Average. A little different number than yours, I guess.

Things that would throw this off: Power on an electric motor is, more or less, torque*(rotational_speed). At t=0, when the motors on the car aren't turning, the power is zero. There's friction and stuff out there, so for a bitty little motor (like the solar powered toy cars my daughter was racing) the power available starts at zero, peaks at some point, then goes back down, smoothly.

What this probably means: Max acceleration at t=0, which gradually reduces as one goes up to the car's maximum velocity.

Fun.
 
  • Helpful
Reactions: Artful Dodger
Um. I looked at this a bit and little bitty alarm bells went off in my head. This is what happens when one who has been through Physics 161, 30-odd years ago, One Remembers Stuff.

First little alarm bell: Yep, F = Force in Newtons = Mass*Acceleration. And the units are kg*m/s^2 -and those are the units for a Newton. So the force is 32,833 Newtons, not "Newton/s".

Next, Power. P(t) = dEk(t)/dt; that is, the power is the time rate of change of the energy. That is a derivative function.

Next, Energy. Kinetic Energy = Ek. Ek(t) = mV(t)^2.

So, we'd like to find the average power.

P(t1) - P(t0) = (E(t1) - E(t0))/(t1-t0) = (m*V(t1)^2 - m(V(t0)^2)/(t1-t0)

Well, V(t0) = 0 (we're stopped), t1 - t0 = 2.6s. m = 3183 kg. But P(t0) isn't zero, it's a derivative function. P(t0-) is zero, since we haven't stomped the pedal yet. Hmm.

So, what I got, then, is 3183kg*(26.8m/s)^2/(2.6s) = 879291W. Let's see, there's 746W/hp, so that would be 1179 HP. Average. A little different number than yours, I guess.

Things that would throw this off: Power on an electric motor is, more or less, torque*(rotational_speed). At t=0, when the motors on the car aren't turning, the power is zero. There's friction and stuff out there, so for a bitty little motor (like the solar powered toy cars my daughter was racing) the power available starts at zero, peaks at some point, then goes back down, smoothly.

What this probably means: Max acceleration at t=0, which gradually reduces as one goes up to the car's maximum velocity.

Fun.
Kinetic is 1/2*m*v²
1/2*3183kg*26.8²=1.143MJ / 2.6 s = 440kW / (746W/hp) = 589 HP average
However, as you point out, HP at 0 rpm is 0.
If torque is constant and bounded by tire traction, then speed increase linearly and peak HP occurs at 60 MPH and is double the average or 1178 hp, which the lack of 1/2 in your ke calc resulted in.
 
Kinetic is 1/2*m*v²
1/2*3183kg*26.8²=1.143MJ / 2.6 s = 440kW / (746W/hp) = 589 HP average
However, as you point out, HP at 0 rpm is 0.
If torque is constant and bounded by tire traction, then speed increase linearly and peak HP occurs at 60 MPH and is double the average or 1178 hp, which the lack of 1/2 in your ke calc resulted in.
Darn it, knew I missed something. And the 538 makes a LOT more sense.
 
  • Like
Reactions: mongo
Being at big pedantic, but power at 0 rpm isn't zero, useful power yes, but battery power is equal I²R losses, which are quite high at full torque

But I think Tesla quotes mechanical motor power, since Plaid does well above 1020 hp from the pack, if they don't, we need to consider the losses
 
  • Informative
Reactions: EVCollies
Being at big pedantic, but power at 0 rpm isn't zero, useful power yes, but battery power is equal I²R losses, which are quite high at full torque

But I think Tesla quotes mechanical motor power, since Plaid does well above 1020 hp from the pack, if they don't, we need to consider the losses
And I guess I'll play pedantic right back at you.

Power is Not Energy. The simplistic view is that P = dE/dt, where E (as correctly pointed out, above) is E = (1/2)*mass_of_car*d(v(t)/dt)), and d(v(t))/dt = a(t). After doing all the substituting, we got P(t) = 1/2*mass_of_car*a(t)

But, as you correctly point out, there are losses that are vaguely related to i(t), and the power losses in the car are something like i(t)^2/R, where R is the resistance in the battery, wires in and outside of the motor, and in transistors everywhere 🥴.

So, we finally get P(t) = 1/2*mass_of_car*a(t) + i(t)^2/R. Whee. And we don't know i(t), we don't know R, we don't know the power at t = 0 (but it's not zero, since a(0) isn't zero), and so on.

Likewise, for an electric motor, the equations of state for a rotating object are torque = I*d^2(w(t))/dt^2. (There're no Greek letters around here.. but that's angular acceleration in rads, that stuff there. But the energy in a stopped shaft might be zero; but the Power present that accelerates that shaft is not zero.

So, when I wandered in here, and @Artful Dodger was making a valiant attempt at the math: We don't have a clue about the instantaneous acceleration, energy, and power on a Tesla under acceleration; nor do we know the dynamic (rolling) resistance and all that jazz. All we can make an attempt at is the average power, which, after a couple of false steps, appears to be around 590 HP.

Now, the specs on the Beast Mode, from Tesla, says that the motor is capable of 845 HP. Wonderful. But I'll bet a quarter.. no, make it $50, that power level is only achieved with the car on a dynamo at a particular, fixed speed. Nothing wrong with that: ICE guys rate their motors the same way.

So, it might be that max power is achieved at 30 mph; or 90 mph; or some other $RANDOM number. All we're saying, here, is that if we had 590 HP, constant, starting at t=0 to 2.6 seconds (ha!) we'd get the same time in a quarter mile.

Back when I used to inhabit a Prius forum, far away, there was a poster who actually owned a recording accelerometer. He used to amuse people by measuring the acceleration in a straight line and going around curves. And there was a point to his madness: What with the always-on hybrid in that car with the dual motor-generators and a planetary gear set, the rpm of the gasoline engine could be set arbitrarily with respect to the wheel rpm; that is, there was a infinitely-variable transmission built into the car. Given that was the case, when one floored the gas pedal, the ICE didn't go to max revs - it went to the revs where it could deliver maximum power. At the same time, in the same architecture, the maximum power available from the battery was also applied to the motor generators, giving this relatively lethargic car more acceleration than one would think.

Now, there's no "transmission", per se, in a Tesla. And I'm sure the miracle workers at Tesla in the motive power division have a riot making more power out of an electric motor than people might think would be possible. But, no question: The power out of that motor is not going to be a constant.

Might be fun if somebody with a recording accelerometer took some Tesla on a 0-to-60 "drag" race to see what a(t) actually is. From that, and the mass of the car, we could probably get a closer approximation to P(t).

Fun.
 
Might be fun if somebody with a recording accelerometer took some Tesla on a 0-to-60 "drag" race to see what a(t) actually is. From that, and the mass of the car, we could probably get a closer approximation to P(t).
You mean sort of the like the power graph Tesla shared at the Plaid event?

OIP.jpeg


The Cybertruck won't be as flat as the Plaid, but probably better than the 2019 Performance.
 
You mean sort of the like the power graph Tesla shared at the Plaid event?

View attachment 998024

The Cybertruck won't be as flat as the Plaid, but probably better than the 2019 Performance.
Interesting. That's P(v); but not "v(t)". I'm going to have to think about it a bit to work out P(t) which, natch, will be different.

The interesting bit is how all the curves flat-top, then trend down. That matches with what I'd expect with an electric motor, but also with the maximum power output of the battery pack (as compared to the maximum the motor can sustain).