Estimating battery capacity based on Supercharging readout

[darth_vad3r]

Has anyone done this / compiled data on this?

TL;DR, using only what the car shows me while supercharging, my SR+ has a capacity* of 52.52688 kWh

Technique:

1. Take a video of your screen while your supercharging session ramps up.
2. When the display ticks up to your rated capacity per hour, read off what the power is at. This should indicate some gauge of what the car itself thinks the capacity is, right?

e.g. In a recent supercharging session I captured the following relevant frames from a video:

52 kW 385 km/hr
52 kW 386 km/hr
53 kW 386 km/hr
53 kW 387 km/hr​

I have the SR+ with rated range of 386km, so from these few datapoints it would seem the car believes that 386km range = 52-53kWh (an hour of theoretical charging at 52-53kW (52kWh) should add 386km of range at 386km/hr).

Using this technique and A LOT more datapoints I managed to figure out that the car's internal number for efficiency appears to be between 136.000 Wh/km and 136.187 Wh/km.

Interestingly, roughly the mid-way point there of 136.08 Wh/km is equal to 219.00000 Wh/mi, so I'm kinda wondering if that's the set constant they're actually using internally on the SR+.

I couldn't narrow the range down any further because there are errors of +/- 0.5 kW and +/- 0.5 km/hr on the screen readouts, and I accounted for min/max on each of these errors to narrow my range. 136.187 Wh/km was the minimum of all of my maximum estimates, and 136.000 Wh/km was the maximum of all my minimum estimates, where a max estimate used the displayed kW + 0.4999 and the km/hr - 0.5, and a min estimate used the displayed kW - 0.5 and the km/hr + 0.4999.

Using 136.08 Wh/km x 386km rated range I get 52.52688 kWh.


*This is not accounting for any supercharging efficiency losses ... which I'm not sure if the readout is accounting for already, or not. Does anyone know? Furthermore, does anyone know what Supercharing losses are typically even if they are already accounted for?

 

[acoste]

I think the number is 225Wh/mile. If I'm right it will show 376km range if you charge it to 100%. New battery is at 54.5kWh.

 

[darth_vad3r]

I think the number is 225Wh/mile. If I'm right it will show 376km range if you charge it to 100%. New battery is at 54.5kWh.

225 you are taking about the “rated” line in the energy screen? That’s not the same as what the car actually thinks you need to estimate a range that is your rated range though, it seems to be lower than that.

 

[AlanSubie4Life]

This is an interesting way to look at things.

- Seems to me it must be post Supercharger losses, otherwise the real Wh/rmi would be too low (your value seems about right).

- I wonder whether it aligns with the trip meter method of calculating this same constant? (Does it align with how many Wh the trip meter says you use in order to use up one rated mile?) There is some evidence the trip meter reads low of course (lots of debate about that), but it still might be set up to be consistent. I would think they might want it to be self-consistent just so people could see that they got out the energy that they put in...

- the way to check the trip meter method was discussed in another thread you started a couple days ago (use trip meter, do a long trip, etc.). Can you compare on your vehicle?

- Coincidentally, if you take the new RWD 325 mile number, and the old 310...

219Wh/rmi * 325/310 =230Wh/rmi (approx)

This 230Wh/rmi number is the value I get on the trip meter which will give me 1 rated mile used per mile driven - for the AWD vehicle type...

So, the 219Wh/mi number would make sense, given what I see for the constant on my P3D (230Wh/rmi using the trip meter method), and the general impression that they may have rescaled the Wh/rmi for the RWD variants, to better align with reality (a lot of people think this is how they “increased” the range of the RWD).

 

[Zoomit]

Has anyone done this / compiled data on this?

Well then, let's dive into the rabbit hole...the current Supercharging conversion parameters, that I had previously calculated, are:

• 3LR AWD/P: 4.08 mi/kWh = 245 Wh/mi = 6.57 km/kWh = 152.2 Wh/km
• 3LR RWD: 4.27 mi/kWh = 234 Wh/mi = 6.87 km/kWh = 145.5 Wh/km
• 3MR: 4.13 mi/kWh = 242 Wh/mi = 6.64 km/kWh = 150.6 Wh/km
• 3SR/+: 4.56 mi/kWh = 219 Wh/mi = 7.33 km/kWh = 136.4 Wh/km

These are based on a limited number of observations so there are definitely rounding errors. Using the @darth_vad3r example, if an SR car is charging at 53kW, you can multiply that by 7.33 km/kWh and should be seeing it report a 388 km/hr charging speed on the charging screen. This matches what he reported above.

As he did, you can multiply that by the EPA range (aka rated range) and come up with an estimate of the battery capacity needed to drive that distance:

• 3LR AWD/P: 245 Wh/mi * 310 mi = 75.9 kWh = 152.2 Wh/km * 499 km
• 3LR RWD: 234 Wh/mi * 325 mi = 76.1 kWh = 145.5 Wh/km * 523 km
• 3MR: 242 Wh/mi * 264 mi = 64.0 kWh = 150.6 Wh/km * 425 km
• 3SR/+: 219 Wh/mi * 240 mi = 52.7 kWh = 136.4 Wh/km * 386 km

Next, and this is critical, let's replace the published EPA range in the previous calculations with the strict (55% city + 45% hwy * 0.7) results from the EPA tests:

• 3LR AWD/P: 245 Wh/mi * 308.4 mi = 75.5 kWh
• 3LR RWD: 234 Wh/mi * 333.8 mi = 78.1 kWh
• 3MR: 242 Wh/mi * 263.8 mi = 63.9 kWh
• 3SR/+: 219 Wh/mi * 247.2 mi = 54.2 kWh

Now it's helpful to review the usable capacities as reported by Tesla in their EPA Applications (ref, ref):

• 3LR AWD/P: 79.2 kWh
• 3LR RWD: 78.3 kWh
• 3MR: 63.8 kWh
• 3SR/+: 54.5 kWh

What trends do you see comparing the previous two lists? The LR RWD, MR and SR/+ estimates for usable battery capacity match the EPA Applications, within rounding tolerances. The 3 LR AWD/P estimate does not. To be consistent, if they have the same usable capacity as the LR RWD cars, the AWD/P cars would need to show 254-257 Wh/mi [78.3/308.4 or 79.2/308.4] or 3.7-4.9% worse energy consumption on the charging screen. That would translate to a ~4% lower charging speed on that screen to match the other cars.

But, stay with me now, the EPA ranges for the LR RWD and SR+ cars were voluntarily reduced by Tesla, and the original MR range was increased from 260 to 264. Let's look at those:

• 3LR RWD: 333.8/325 = 2.7%
• 3MR: 263.8 / 260 = 1.5%
• 3SR+: 247.2/240 = 3.0%

In summary, it's inconsistent and messy:

• 3LR AWD/P:
  • The charging screen speed is 4% higher than a speed derived from the EPA consumption tests (Again, this assumes they have the same usable capacity as LR RWD cars)
  • The published EPA Combined range is 0.5% above the strict test results [310/308.4]
• 3LR RWD:
  • The charging screen speed matches the EPA consumptions tests
  • The published EPA Combined range is 3% lower than the strict test results [325/333.8]
• 3MR:
  • The charging screen speed matches the EPA consumptions tests
  • The published EPA Combined range was 1.5% lower but now matches the strict test results
• 3SR+:
  • The charging screen speed matches the EPA consumptions tests
  • The published EPA Combined range is 3% lower than the strict test results [240/247.2]

To address the question about Supercharging losses, it's critical to know 1) the battery usable capacity and 2) how much of that usable capacity is outside of the 0-100% indicated SoC. Tesla holds margin below 0 mi range to ensure the car doesn't quit with more than 0 mi indicated range left. This is something like 2%. More importantly we would need to resolve the discrepancies between what's been measured by owners from the screen, what's been measured independently on a test bench, what Tesla states internally and what is documented in the EPA applications. I'll save that for another day. Hint: The results don't match and, like above, the trends are inconsistent between the different battery sizes.

 

[acoste]

225 you are taking about the “rated” line in the energy screen? That’s not the same as what the car actually thinks you need to estimate a range that is your rated range though, it seems to be lower than that.

Once I calculated this and the energy screen used the same number. But it may have got changed.

- Coincidentally, if you take the new RWD 325 mile number, and the old 310...

219Wh/rmi * 325/310 =230Wh/rmi (approx)

These are arbitrarily picked numbers.

EPA measures the range on a dyno with all accessories off. Then they multiply the number by 0.7 and that's going to be the rated miles. For the long range RWD this number is 334 miles. Tesla picked 310 miles to match this with AWD numbers. Then this year they changed it to 325.

Interestingly the EPA multiplier for the SR+ is not 0.7 any more but 0.75.

 

[Zoomit]

Interestingly the EPA multiplier for the SR+ is not 0.7 any more but 0.75.

The 0.7 multiplier is only used to calculate 247.2 mi from the actual 353.1 mi test result. This itself is a 45%/55% combination of city and highway test results [367.7 * 0.55 + 335.3 * 0.45]. Tesla can lower it however much they want below that.

If you want to back out a multiplier from the 240 mi range they chose, it would be 0.68 [240/353.1].

 

[AlanSubie4Life]

These are arbitrarily picked numbers.

Yeah, I know. It is arbitrary, just as the Wh/rmi chosen by Tesla is largely arbitrary. Obviously the Wh/rmi are based roughly on EPA miles and the capacity of the battery, but is actually at Tesla’s discretion, as they can 1) implement a reserve energy below zero miles, 2) make the Wh/mi indicator in the car read low or high as they see fit, etc. The Wh/mi indicated could be off by 10x and it would make zero difference - it is just a number. It could say you use 1000MWh/mi, for all the difference it makes.

I just thought it was likely that the scalar Wh/rmi on the RWD vs the AWD, if they are different, is likely to be in the ratio 310/325 because those are the two numbers Tesla has chosen. But it is not required to be that relationship.

We’ll see. Hopefully someone can say what it is on the SR and the LR RWD. I know it is somewhere between 230 and 235Wh/rmi for the AWD (P3D).

Even though the value is arbitrary, knowing the actual scalar used IS useful, because once you know it, it tells you what you need to get for efficiency to achieve the rated miles in a single drive (to the extent the state of charge indicator is linear, of course). I have to do better than ~230Wh/mi to get 310 miles out of a single charge on my car.

 

[ewoodrick]

The absolutely best way to get the numbers is just take some trips to the car.

If you take some trips, then you may find that it really isn't that big of a deal. If you drive 75 or more, you aren't going to make your numbers. If you slow down, you'll get your range. It's a lot easier than everyone wants to make it.

As soon as you stop worrying about how big the battery exactly is and how much degradation it doesn't have, then you can start enjoying the car.
Driving from Cape Canaveral to above Atlanta, two stops. Because of time, we couldn't eat at the location that we wanted to and had to wait a little while charging at the second stop. 10 hour drive, about 525 miles. In a LR RWD Model 3.
I pretty much drove it as an ICE.
And actually drove down the day before. It was a quick trip to see a SpaceX launch, planned at the last minute and at least the third trip the car had made down there. Oh and last weekend, made basically the same length trip to Dayton/Xenia Ohio. 2,000 miles in 10 days, at least 1900 on FSD.

The car works, you just need to put the Excel sheets down and drive it!!!

 

[AlanSubie4Life]

The absolutely best way to get the numbers is just take some trips to the car.

If you take some trips, then you may find that it really isn't that big of a deal. If you drive 75 or more, you aren't going to make your numbers. If you slow down, you'll get your range. It's a lot easier than everyone wants to make it.

As soon as you stop worrying about how big the battery exactly is and how much degradation it doesn't have, then you can start enjoying the car.
Driving from Cape Canaveral to above Atlanta, two stops. Because of time, we couldn't eat at the location that we wanted to and had to wait a little while charging at the second stop. 10 hour drive, about 525 miles. In a LR RWD Model 3.
I pretty much drove it as an ICE.
And actually drove down the day before. It was a quick trip to see a SpaceX launch, planned at the last minute and at least the third trip the car had made down there. Oh and last weekend, made basically the same length trip to Dayton/Xenia Ohio. 2,000 miles in 10 days, at least 1900 on FSD.

The car works, you just need to put the Excel sheets down and drive it!!!

Everyone is different. For me, and for some others, I want to know how many Wh/rmi there are. It helps me easily understand how many miles I am likely to get. I don’t have to think about battery capacity at all that way (in fact, doing so would lead to error)! It just turns it into solving for the unknown in the ratio.

 

[darth_vad3r]

The absolutely best way to get the numbers is just take some trips to the car.

If you take some trips, then you may find that it really isn't that big of a deal. If you drive 75 or more, you aren't going to make your numbers. If you slow down, you'll get your range. It's a lot easier than everyone wants to make it.

As soon as you stop worrying about how big the battery exactly is and how much degradation it doesn't have, then you can start enjoying the car.
Driving from Cape Canaveral to above Atlanta, two stops. Because of time, we couldn't eat at the location that we wanted to and had to wait a little while charging at the second stop. 10 hour drive, about 525 miles. In a LR RWD Model 3.
I pretty much drove it as an ICE.
And actually drove down the day before. It was a quick trip to see a SpaceX launch, planned at the last minute and at least the third trip the car had made down there. Oh and last weekend, made basically the same length trip to Dayton/Xenia Ohio. 2,000 miles in 10 days, at least 1900 on FSD.

The car works, you just need to put the Excel sheets down and drive it!!!

Read my sig. We are enjoying the cars, and enjoying the calculating. If you don’t enjoy it, skip this thread.

Everyone else, thanks for the posts. Need to catch up, glad this got some rabbit holes going

 

[acoste]

I did my calculations again based on Tesla Bjorn's X P100D charging video and the charging power on the car's display is the power that is delivered by the charger, but it's not equal to the net charging speed. The display showed 110kW while the real charging speed was 100kW. At lower rates however, like around 50kW the difference was minimal (low currents, less heat loss).

 

[darth_vad3r]

I did my calculations again based on Tesla Bjorn's X P100D charging video and the charging power on the car's display is the power that is delivered by the charger, but it's not equal to the net charging speed. The display showed 110kW while the real charging speed was 100kW. At lower rates however, like around 50kW the difference was minimal (low currents, less heat loss).

I think it would be more accurate to say the car’s display is what the outboard charger is delivering to the BMS, but not the total power delivered from the outboard charger.

I believe Bjorn has one video at a hypercharger where he commented on the fact the accessory draw (heat/AC etc) was showing on the charger output but subtracted out from the charging display on screen. End of his Model 3 range drive video maybe?

Is the video you are referencing with accessories on or off?

What method do you use to calculate “net charging speed”? Is 110kW in your example on the car’s screen, and 100kW is calculated (by what method?)

Thanks

 

[acoste]

I think it would be more accurate to say the car’s display is what the outboard charger is delivering to the BMS, but not the total power delivered from the outboard charger.

I believe Bjorn has one video at a hypercharger where he commented on the fact the accessory draw (heat/AC etc) was showing on the charger output but subtracted out from the charging display on screen. End of his Model 3 range drive video maybe?

Is the video you are referencing with accessories on or off?

What method do you use to calculate “net charging speed”? Is 110kW in your example on the car’s screen, and 100kW is calculated (by what method?)

Thanks

In this video below watch the Tesla's power. 10kW can't be the accessories.
At 16% 7:02s the charging speed shows 110kW and stays there until 46% 23:35s
// timer is not the video timer but the charger timer shown in the video

Bjorn says at 1:20 the usable battery capacity is 92kWh.

92kWh*(46%-16%) = 27.6kWh
23:35 - 7:02 = 993 sec = 16:55 = 0.2758 hours

charge speed = 27.6kWh / 0.2758 hours = 100kWh

 

[acoste]

Bjorn says this car has 92kWh available capacity.

But Jason here says it is 98.4kWh
Pics and Info: Inside the Tesla 100kWh Battery Pack | wk057's SkieNET

I'm not sure if they are talking about the same model or who is right. If it is 98.4kWh, then my calculation changes to 98.4*0.3/0.2758 = 107kW.

Maybe 98.4kWh is the stored energy but only 92kWh was available due to heat losses when Bjorn tested it? I don't know.

 

[AlanSubie4Life]

Bjorn says this car has 92kWh available capacity.

But Jason here says it is 98.4kWh
Pics and Info: Inside the Tesla 100kWh Battery Pack | wk057's SkieNET

I'm not sure if they are talking about the same model or who is right. If it is 98.4kWh, then my calculation changes to 98.4*0.3/0.2758 = 107kW.

Maybe 98.4kWh is the stored energy but only 92kWh was available due to heat losses when Bjorn tested it? I don't know.

Impossible to do with the X in the video unless you have an identical car....but....

I really think measuring the Wh/rmi with a long trip will help determine whether or not this Supercharger screen “matches” the trip meter, etc.

I understand the original purpose was to “estimate battery capacity”, but in the end all we can determine is the number of “kWh” which represents a full charge. This may or may not align with the actual full battery capacity for aforementioned reasons. I put kWh in quotes because the readout on the trip meter is somewhat arbitrary and might not represent true kWh.

I do think that since Tesla is charging by the kWh, there is some reason to think that the kWh they charge you for there represent true kWh.

@Zoomit answered this before (can’t remember the answer), but there are a couple relevant questions:

1) Which kWh does Tesla charge for? Only the ones delivered to the pack in the form of rated miles, or the ones from the DC side of the Supercharger (by necessity due to I^2*R and other battery charging losses these will be different by a bit)?

2) As asked above, if you calculate kWh delivered by Supercharger and compare to rated miles added, does that Wh/rmi constant align with that calculated from observing the trip meter?

3) If these Wh/rmi values are consistent, what do they imply about the capacity of the battery pack? If they are not consistent, does it simply mean Tesla is charging for kWh from the DC charger, or does it mean the trip meter doesn’t capture everything, or what? To some extent it is hard to answer these questions without knowing how the various Wh/rmi constants differ, and how the implications of the calculated constants on the total battery capacity compare to the battery capacity declared in the EPA tests (again the EPA number is for a full discharge and it is likely that a full rated miles discharge will not discharge the entire EPA quantity due to the below 0 buffer, and perhaps for other reasons).

The only piece of data I have is that in a P3D the trip meter Wh/rmi appears to be 230-235Wh/rmi...have not tried any measurements at a Supercharger.

 

[Zoomit]

Impossible to do with the X in the video unless you have an identical car....but....

I really think measuring the Wh/rmi with a long trip will help determine whether or not this Supercharger screen “matches” the trip meter, etc.

I understand the original purpose was to “estimate battery capacity”, but in the end all we can determine is the number of “kWh” which represents a full charge. This may or may not align with the actual full battery capacity for aforementioned reasons. I put kWh in quotes because the readout on the trip meter is somewhat arbitrary and might not represent true kWh.

I do think that since Tesla is charging by the kWh, there is some reason to think that the kWh they charge you for there represent true kWh.

@Zoomit answered this before (can’t remember the answer), but there are a couple relevant questions:

1) Which kWh does Tesla charge for? Only the ones delivered to the pack in the form of rated miles, or the ones from the DC side of the Supercharger (by necessity due to I^2*R and other battery charging losses these will be different by a bit)?

2) As asked above, if you calculate kWh delivered by Supercharger and compare to rated miles added, does that Wh/rmi constant align with that calculated from observing the trip meter?

3) If these Wh/rmi values are consistent, what do they imply about the capacity of the battery pack? If they are not consistent, does it simply mean Tesla is charging for kWh from the DC charger, or does it mean the trip meter doesn’t capture everything, or what? To some extent it is hard to answer these questions without knowing how the various Wh/rmi constants differ, and how the implications of the calculated constants on the total battery capacity compare to the battery capacity declared in the EPA tests (again the EPA number is for a full discharge and it is likely that a full rated miles discharge will not discharge the entire EPA quantity due to the below 0 buffer, and perhaps for other reasons).

The only piece of data I have is that in a P3D the trip meter Wh/rmi appears to be 230-235Wh/rmi...have not tried any measurements at a Supercharger.

Those, right there, are a hard set of questions to answer! I feel we're chipping away at them but definitely requires more data and some critical thinking.

Here's my post that I think you referred to. Exactly how many miles driven per hour of charge?

It basically says that the charging screen power indication is power to the battery, not the whole car. I would expect that Tesla charges based on the power output of the Supercharger, not just the power going into the battery. That's just my speculation that they'd want to bill for the all the power provided, including charging losses and accessory usage.

 

[AlanSubie4Life]

Thanks.
I never supercharge (haven’t been on a long enough trip yet), so I am lacking info here. I guess for someone who has paid attention (or maybe it can be gleaned from videos?), it would be good to know the following (along with the vehicle type):

1) Billing info for a particular charging session - how many kWh billed, rate, etc. I guess they just tell you the dollar cost right on the screen...along with $/kWh?

2) For the same session, I think the car tells how many miles were added on the screen, report that. In the older video above it looks like actually it reports kWh. Whatever it says, anyway, report that.

3) Report the actual number of rated miles added (so note them at beginning and end of session).

4) Report accessory use if any during the session.

5) If you know for sure how this works (is accessory use charged, etc.), just report that!

If we also know the trip-meter-based Wh/rmi constant for the same vehicle, I think we should be able to draw some conclusions?

Not sure if there is anything else we would need to know to draw some conclusions...probably!

 

[darth_vad3r]

I'm not sure* [I am now, see edit] if the display on the "Supercharging" screen may change based on your display settings similar to the way the normal L1/L2 charging status screen shows either (a) km or (b) % and associated (a) km/hr or (b) kW, along with the always-displayed amps/volts: x/y A and z V.

With my display set to "km", the Supercharging screen shows me my range at the top middle (same number as that by the battery icon).
Below the car/battery figure it shows: x kW, y km/hr, +z km

[*EDIT: Oh look at me being thorough ... one of my videos during supercharging I switch display mode to % ... in that mode above the car is X %, below the car is x kW, y km/hr, +z kWh ... now if I had stayed in that mode until charge completion I could have mapped that +kWh to the billled kWh
EDIT2: nevermind on that last wish, I forgot that my test site here is in Canada billed per minute not per kWh anyways ]

Note that +z km seems to be just a mapping of kWh added converted to a (rated?) km amount, and it does not always equal "current km range displayed" minus "starting km range displayed". This is probably due to rounding is my guess, but who knows, it could be also due to some non-linear scaling of the displayed km range. I think it may be a bit of both?

e.g. if I started with "56 km" range displayed, internally that could be 55.5km to 56.499km. The Supercharging display screen starts at +0km.

During charging I might see 77km range, +21km added ... this matches initial conditions of 56km / +0km.
Then I can see it tick up to +22km but it stays at 77km, fine ... that could be due to rounding. It's gotten to 21.5 or more, but the range is at 77.49 or less. If I started at 55.5km / +0km, then now I could be at 77.0km / +21.5km (aka "22"). OK, but ... it ticks up again to 78km/+22km ... still fine I guess, could be 77.5km / +22km, and then ... "79km / +22km" ... huh? What's happened here?

How can +22km show 77, 78, and 79km range?

IMO, the trip meter and the charge display screen can be "trusted" more reliably than the "range" indicator as far as trying to extrapolate figures like capacity (whether usable or 0-100% or whatever, not actual physical capacity which nobody but Tesla can know since that includes some brick protection in the bottom 2-3% 5% and some top protection as well most likely since that's what they were doing initially in the Roadster I believe as well, never allowing charge over 95% or 97% or some number).

As far as I'm concerned we're always talking about the number available to us as drivers from "100%" displayed and charge stops, down to either "0%" or "car shuts off below 0%".

... but measuring differences in the displayed range I think throws a monkey wrench into the equation if it adapts non-linearly based on unknown paramters

 

[darth_vad3r]

IMO, the trip meter and the charge display screen can be "trusted" more reliably than the "range" indicator as far as trying to extrapolate figures like capacity (whether usable or 0-100% or whatever, not actual physical capacity which nobody but Tesla can know since that includes some brick protection in the bottom 2-3% 5% and some top protection as well most likely since that's what they were doing initially in the Roadster I believe as well, never allowing charge over 95% or 97% or some number).

As far as I'm concerned we're always talking about the number available to us as drivers from "100%" displayed and charge stops, down to either "0%" or "car shuts off below 0%".

This (potentially totally outdated because based on original Roadster 18650 cells and chemistry) info is what I was partly remembering and referring to, which said max actual charge allowed was 95% real SOC, and max discharge before shutoff was 2% SOC: A Bit About Batteries

"The other factors affecting cycle life are tied to how the cell is used. In particular:

1. Avoiding very high and very low states of charge. Voltages over 4.15V/cell (about 95 percent state of charge [SOC]) and voltages below 3.00V/cell (about 2 percent SOC) cause more stress on the insides of the cell (both physical and electrical).

[...]

There is a huge difference in cycle life between a 4.2V/cell charge (defined by the manufacturers as “fully charged”) and a 4.15V/cell charge. 4.15 volts represents a charge of about 95 percent. For this reduction of initial capacity (5 percent), the batteries last a whole lot longer. Unfortunately, further reduction of charge has a much smaller benefit on cycle life. Understanding this tradeoff, Tesla Motors has decided to limit the maximum charge of its cells to 4.15 volts, taking an initial 5 percent range hit to maximize lifetime of the pack. We also limit discharge of our battery pack to 3.0V/cell and will shut down the car when the batteries reach this level."
​

So, again with my above caveats, I'd say a decent guess for current chemistry in the 2170 is something similar-ish to their self-admitted strategy in the Roadster in 2006, and that Tesla's usable capacity from "displayed 100%" to "car shuts off below 0%" is probably 95% -> 2% actual theoretical SOC).