Calculate usable battery capacity based on rated miles values

[AWDtsla]

I guess the model (as in math model, not Model S model) they use to predict energy consumption assumes 19in wheels and driving below or at the speed limit. A local P85D owner can beat the rated consumption on his car no problems in here... he tells me the trick is to never exceed ~80kW of power and drive 100kph on the highway. He also has 19in on the car with the low-rolling resistance Michelin tires.

Staying under 80kW implausible. Hills.

100kph would get me at best pulled over for driving suspiciously slow. At worst rear ended. If you do 65mph in a 55 mph zone here, people will pass you on both sides. I have the 19" wheels too.

 

[stopcrazypp]

Also, I could design you several compensation mechanisms right here on the spot. The first one being a lookup table that references current draw with SoC and reports an internal loss number. Simple straightforward implementation.

That would report a "fake capacity" and fail to give you a measure of degradation. A big portion of long term capacity loss is increased internal resistance. If you have a table that back calculates and eliminates that as a factor, then you fail to have a measure that you can meaningfully use to keep track of degradation.

The number itself would also be of limited use for other systems, because the car itself only sees the capacity after such losses. If the current system is already over optimistic about range, this will make it worse.

 

[AWDtsla]

That would report a "fake capacity" and fail to give you a measure of degradation. A big portion of long term capacity loss is increased internal resistance. If you have a table that back calculates and eliminates that as a factor, then you fail to have a measure that you can meaningfully use to keep track of degradation.

The number itself would also be of limited use for other systems, because the car itself only sees the capacity after such losses. If the current system is already over optimistic about range, this will make it worse.

Uhh, no. You use feedback to fill a compensation map applied to the table. It does not drift, it self-tunes.

This is not a problem. This a common design pattern. No rocket surgery, just a little computer science.

 

[ran349]

It would be very easy to get a number for this unreported degradation if Tesla just provided one more value. Along with %SOC and RM, also include kWh. That way, at 100% SOC, it should give you your maximum usable capacity in kWh, and at 0% it should just give 0 kWh, with the bricking buffer not included. That way you could easily compare your reported odometer kWh used vs. the actual battery kWh used. No more mystery about what is not reported. Maybe Tesla doesn't want owners to know this number, but I would like to have it.

 

[jbdvm1988]

Wk057 mentioned that 90 kw batteries had a higher usable capacity in higher vin cars. Does anyone have an estimate of when those cars were produced and what the improved capacity was

 

[Boatguy]

This is an interesting technical discussion, but I'm just trying to figure out how many odometer miles I can drive my S90D. So I really only need to know two things:

a) available power (setting aside the technical arguments about rate of discharge affects on capacity - which I thoroughly understand, but probably aren't an issue in the real world driving down the freeway at a more or less constant 70-80mph)

b) how much power I'll use per mile

For the sale of simplicity, let's assume that I'll drive the remainder of the miles at whatever rate of consumption I've averaged for the last 30mi. So the Energy app will give me the the number to use for b). Or I have an estimate from EVTripPlanner, though it's usually about 10% high.

Now the problem is how much have I got, what is a)? The Energy App asserts that I have 300Wh x RM which is just nonsense. The Energy app's trip display gives me SOC, also not very useful. The OP asserted that I had 285Wh x RM and I've consistently measured that I have 273Wh x RM.

That means if I'm consuming 336Wh/mi, as I did on my recent trip, and say I have 100RM on the dash, then I really can drive another (273/336) * 100 = 81 miles.

For local driving my consumption is more like 305Wh/mi, good for about 90mi.

So my assumption going forward is that on trips, I have about 80% of whatever is displayed, and locally I have about 90%.

Tesla blows a lot of smoke...

 

[JonG]

This is an interesting technical discussion, but I'm just trying to figure out how many odometer miles I can drive my S90D. So I really only need to know two things:

a) available power (setting aside the technical arguments about rate of discharge affects on capacity - which I thoroughly understand, but probably aren't an issue in the real world driving down the freeway at a more or less constant 70-80mph)

b) how much power I'll use per mile

For the sale of simplicity, let's assume that I'll drive the remainder of the miles at whatever rate of consumption I've averaged for the last 30mi. So the Energy app will give me the the number to use for b). Or I have an estimate from EVTripPlanner, though it's usually about 10% high.

Now the problem is how much have I got, what is a)? The Energy App asserts that I have 300Wh x RM which is just nonsense. The Energy app's trip display gives me SOC, also not very useful. The OP asserted that I had 285Wh x RM and I've consistently measured that I have 273Wh x RM.

That means if I'm consuming 336Wh/mi, as I did on my recent trip, and say I have 100RM on the dash, then I really can drive another (273/336) * 100 = 81 miles.

For local driving my consumption is more like 305Wh/mi, good for about 90mi.

So my assumption going forward is that on trips, I have about 80% of whatever is displayed, and locally I have about 90%.

Tesla blows a lot of smoke...

Just go to the energy screen and it tells your range based on average consumption over recent miles.

 

[Boatguy]

Just go to the energy screen and it tells your range based on average consumption over recent miles.

With all due respect, if you've been reading this thread you would understand why the "projected range" on that display is worthless.

The SOC in the Trip portion of the Energy app is more accurate, after the first 20mi or so, but try translating that into miles (or km). And if you're not going to a specific destination, it's of no value at all.

 

[Polyport]

So, now I have heard about Remote S, Teslafi, and now TM-Spy. Do they all provide pretty much the same information? Do they all work the same way?

Remote S is an app for your iPhone that could be compared to the official Tesla app in terms of what it provides, with a different UI.
Teslafi is a web service that monitors and collects data from your car
TM-Spy is an iOS and Android app that connects to a CAN bus adapter in your car and gathers raw data the car is providing as you're using it. To use TM-Spy you need to buy a bluetooth CAN bus adapter.

Or in short, to answer your question, no they do not all work the same way or provide the same information. While Teslafi and RemoteS both use the Tesla APIs to retrieve information from your car (albeit for different purposes), TM-Spy is very different, just allowing you to log events that happen as the car is used, directly from the cars data-bus.

 

[JonG]

With all due respect, if you've been reading this thread you would understand why the "projected range" on that display is worthless.

The SOC in the Trip portion of the Energy app is more accurate, after the first 20mi or so, but try translating that into miles (or km). And if you're not going to a specific destination, it's of no value at all.

I must misunderstand your post. You were quoting different levels of consumption and then applying that to the displayed figure in the dash - i.e. If you're using 336wh/m and the dash says 100 you have 81 miles left. My point was in the energy app, on the right hand side against the average consumption figure fir the last 15 miles or whatever, it tells you what it thinks you have left based on that current recent consumption. It does your calculation for you. Of course both methods depend on the car really knowing how many kWh are available which is another point.

 

[Boatguy]

I must misunderstand your post. You were quoting different levels of consumption and then applying that to the displayed figure in the dash - i.e. If you're using 336wh/m and the dash says 100 you have 81 miles left. My point was in the energy app, on the right hand side against the average consumption figure fir the last 15 miles or whatever, it tells you what it thinks you have left based on that current recent consumption. It does your calculation for you. Of course both methods depend on the car really knowing how many kWh are available which is another point.

The problem with the "projected range" that you are referring to is that it assumes that my S90D will consume 300Wh/RM, that is the basis for all the calculations. So it assumes that if you drive one mile and consume 300Wh, you will lose one Rated Mile. So RM will equal odometer miles at 300Wh/mi. But my empirical measurements (posted earlier in this thread) show that if I drive one mile and consume 273Wh, I will lose one Rated Mile. In essence the Energy app is asserting there is more battery capacity than there actually is.

The result is that if it says your Projected Range is 100RM, and you drive so that you are consuming 300Wh/mi, your RM will reach zero after driving only (273/300) x 100 = 91mi. Project Range is only accurate if discounted by about 10%.

Sorry to be using statute miles and not km, it's a legacy of our colonial past.

 

[JonG]

The problem with the "projected range" that you are referring to is that it assumes that my S90D will consume 300Wh/RM, that is the basis for all the calculations. So it assumes that if you drive one mile and consume 300Wh, you will lose one Rated Mile. So RM will equal odometer miles at 300Wh/mi. But my empirical measurements (posted earlier in this thread) show that if I drive one mile and consume 273Wh, I will lose one Rated Mile. In essence the Energy app is asserting there is more battery capacity than there actually is.

The result is that if it says your Projected Range is 100RM, and you drive so that you are consuming 300Wh/mi, your RM will reach zero after driving only (273/300) x 100 = 91mi. Project Range is only accurate if discounted by about 10%.

Sorry to be using statute miles and not km, it's a legacy of our colonial past.

The projected range is based on average consumption not a fixed number so I'm not sure why you keep introducing it. This is not the number on the dash which is based on a fixed consumption rate, im referring to the number on the consumption chart on the right. Mine seems fairly accurate at least. As for miles v km, I'm a miles man being your colonial past

 

[Boatguy]

The projected range is based on average consumption not a fixed number so I'm not sure why you keep introducing it. This is not the number on the dash which is based on a fixed consumption rate, im referring to the number on the consumption chart on the right. Mine seems fairly accurate at least. As for miles v km, I'm a miles man being your colonial past

The problem is with it's assumption about the capacity of the battery as represented by the RM. Look at the first post in this thread. The BMS reports the battery as having a capacity of 85.8kWh and usable capacity of 81.8kWh.

The "rated range" of an S90D is 294. If you look at the projected range graph (sample attached), you'll see a Rated Range line at 300Wh/mi. 294 x 300Wh = 88.2kWh, more than total capacity of the battery!

Now do the projected range calculation yourself. The formula they are using is: Rated Miles / (Avg consumption / 300) = Projected Range. Verify it yourself by doing the math(s).

So if your average consumption was 300Wh, Rated Miles = Projected Range. That is never going to happen because the battery isn't that big.

Now do another calculation. Either Reset your Trip A or do this immediately after charging. Note the RM on the dash. Drive at least 40-50 miles and then immediately record the kWh consumed and the change in RM (i.e., RM consumed). Divide the kWh consumed by the RM consumed. On my S90D, this was 273 for my most recent trip (logs posted earlier), and seems to have drifted down from 275 that I measured six months ago.

Please let me know what your results.

 

[hacer]

You are describing the fraction of usable battery capacity that reaches the drivetrain. Similarly, you could talk about variables that effect motor efficiency. Or air friction from high speeds

So I completely agree that the range will vary with how the car is driven and in what environment -- but it does not change the usable battery capacity.

No he is describing the fraction of the capacity that leaves the battery (not as heat in the battery). This is an important distinction because the consumption display does not measure this lost energy. In fact it can't be measured during driving although it could be modeled. The only accurate way to measure it is with calorimetry on the batteries.
While air drag and other factors affect your range these are measured by the consumption gauge while the internal battery losses are not.

 

[AWDtsla]

No he is describing the fraction of the capacity that leaves the battery (not as heat in the battery). This is an important distinction because the consumption display does not measure this lost energy. In fact it can't be measured during driving although it could be modeled. The only accurate way to measure it is with calorimetry on the batteries.
While air drag and other factors affect your range these are measured by the consumption gauge while TG he internal battery losses are not.

This is not correct. It is measured while driving, and it is reported. How do you think the car displays RM remaining? Otherwise no "fuel guage" at all is possible, and that doesn't exactly make for a practical vehicle.

 

[hacer]

This is not correct. It is measured while driving, and it is reported. How do you think the car displays RM remaining? Otherwise no "fuel guage" at all is possible, and that doesn't exactly make for a practical vehicle.

No. It is still possible to make a fuel gauge, just not a very linear one. The cell voltage and temperature can give you a good idea of the available capacity when it is close to full and when it is close to empty. In between you do the best you can, but you can only model internal losses in the battery, you can't directly measure them. Even in a laboratory you can't make instantaneous measurements of internal loss (with production batteries) because it requires calorimetry which is an (slow) energy measure, not a power measure. The internal loss model is very simple if you always have low currents, but that's hard to do with real-world driving. From my experience, the energy consumption gauge (kWh since last charge) does not seem to include them at all, nor other significant sources of energy loss, like vampire drain.

 

[AWDtsla]

No. It is still possible to make a fuel gauge, just not a very linear one. The cell voltage and temperature can give you a good idea of the available capacity when it is close to full and when it is close to empty. In between you do the best you can, but you can only model internal losses in the battery, you can't directly measure them. Even in a laboratory you can't make instantaneous measurements of internal loss (with production batteries) because it requires calorimetry which is an (slow) energy measure, not a power measure. The internal loss model is very simple if you always have low currents, but that's hard to do with real-world driving. From my experience, the energy consumption gauge (kWh since last charge) does not seem to include them at all, nor other significant sources of energy loss, like vampire drain.

Not thinking hard enough. You don't need to measure the waste heat of each cell, that would be a crazy way to implement a consumer good. (Although right after I wrote this I think I can build this in software just by using the heavy instrumentation Tesla already has on it's coolant loops)

I can use RM to reliably find the chronic underestimation/fudging by the trip meter, drive, after drive, after drive. I'm not using special equipment, just RM as reported by the car. A simple feedback filled compensation table would have righted the reported consumption number, without understanding any of the underlying physics or doing the lab experiment you propose.

There's no excuse for it to be this wrong. It's not impossible. So what does that leave? Incompetence? Or malice?

 

[hacer]

Not thinking hard enough. You don't need to measure the waste heat of each cell, that would be a crazy way to implement a consumer good. (Although right after I wrote this I think I can build this in software just by using the heavy instrumentation Tesla already has on it's coolant loops)

I can use RM to reliably find the chronic underestimation/fudging by the trip meter, drive, after drive, after drive. I'm not using special equipment, just RM as reported by the car. A simple feedback filled compensation table would have righted the reported consumption number, without understanding any of the underlying physics or doing the lab experiment you propose.

There's no excuse for it to be this wrong. It's not impossible. So what does that leave? Incompetence? Or malice?

I've no doubt you can easily make a kalman filter to better estimate the available range. In fact it seems that Tesla has done this with the projected range graph (when you are navigating a trip, where it behaves differently than when you aren't - that may be part of the confusion between those who say its great and those who say it sucks). That doesn't mean that you will correctly be estimating energy consumed (including internal battery losses).

Tesla could do a better job with the energy consumption meter with somewhat complicated models, I don't know why they haven't but my guess would be priorities vs malice or incompetence.

 

[AWDtsla]

that may be part of the confusion between those who say its great and those who say it sucks).

No my guess the difference is between critical thinking and not.

That doesn't mean that you will correctly be estimating energy consumed (including internal battery losses).

The mechanisms in this car are childsplay compared to a modern computer driven fuel injected and turbocharged engine. The amount of physics involved and number of fudge tables (trying to model the physics with limited sensors and limited compute, not to misreport it) is amazing.

Tesla could do a better job with the energy consumption meter with somewhat complicated models, I don't know why they haven't but my guess would be priorities vs malice or incompetence.

This would seem like a #1 priority to me. Does our main product report it's range and consumption correctly??? Remember when Elon promised the end of range anxiety due to nav updates?

 

[ohmman]

There's no excuse for it to be this wrong. It's not impossible. So what does that leave? Incompetence? Or malice?

Malice? That's maybe a little on the strong side.

My guess is that it's just low on the priority list. They view it as good enough, and maybe that's the "coastal California" effect, where they're not dealing with really difficult weather regularly. But I'm doubtful it's the other options. I agree that it can be improved, and hopefully it will be. It won't be perfect, but it can certainly be better.