-
Want to remove ads? Register an account and login to see fewer ads, and become a Supporting Member to remove almost all ads.
This is why you can't get 'rated range'
- Thread starter David99
- Start date
darth_vad3r
Well-Known Sith
He said an ex-Tesla engineer told him the 4kWh buffer is misunderstood. I think it still is.
If the BMS reads “4kw” I wouldn’t trust it since an energy buffer should be measured in “kWh”
Here’s the quote from @wk057 ...
I disagree which his TL;DR characterization of what he relayed from the ex-Tesla engineer though.
A 4kWh “window” that is scaled with SoC as the car discharges and is “on average, actually 2kWh” sounds to me like a kludge factor that can be as big as 4kWh to start with and gradually shrinks to zero, thus averaging 2 on the journey from 100% to 0%. This would be the expected case if none of this sliding window was ever used for adjustments. The kludge factor was “to ensure that range calculation would never adjust abruptly” because presumably any readjustment in measurement and estimation would first add or subtract from the kludge buffer and hide the discontinuity from the user.
I do not think this kludge buffer described is anything to do specifically with being able to drive miles below 0 on the display, nor the bottom end anti-brick buffer.
I think of it as a separate invisible battery gauge that can lend (or borrow) power to (or from) the main battery gauge in an attempt to absorb any jitter and keep the main gauge smooth. This is how I interpret the words as relayed from the ex-Tesla engineer.
EDIT: What I wrote above may sound like it starts with 4kWh of capacity (energy), but another interpretation is the opposite, starting at a size of 4, with 0 used, and slowly fill the 4 as necessary.
Or another interpretation is that is a size-4 buffer, that can start at 0, 2, 4 used, but on average as you discharge it’s at 2/4 as it absorbs adjustments. If you think of it as like being able to absorb regen, you don’t want to start at full, otherwise you can’t add.
Last edited:
Another reason why you'll never get rated range: Raven edition
I don't recall my previous car doing this, and I can't check it since I sold it. However I've observed the following:
When charging, the car counts each mile added as 284 Wh, seemingly regardless of SoC.
Measuring charging efficiency by comparing energy added to power at the outlet I'm computing 92-93% efficiency, depending on input power.
When you look at what the BMS is reporting for energy either in nominal or usable kWh, it's significantly less. 94.6% of the value reported in the API. Total charging efficiency becomes ~86.4%
If I divide nominal kWh by rated miles at 100%, I get 284 Wh/mi
If I divide nominal kWh by rated miles at 54%, I get 296 Wh/mi
If I divide usable kWh by rated miles between 54-100%, I get ~292 Wh/mi
So. many. fudge. factors.
I don't recall my previous car doing this, and I can't check it since I sold it. However I've observed the following:
When charging, the car counts each mile added as 284 Wh, seemingly regardless of SoC.
Measuring charging efficiency by comparing energy added to power at the outlet I'm computing 92-93% efficiency, depending on input power.
When you look at what the BMS is reporting for energy either in nominal or usable kWh, it's significantly less. 94.6% of the value reported in the API. Total charging efficiency becomes ~86.4%
If I divide nominal kWh by rated miles at 100%, I get 284 Wh/mi
If I divide nominal kWh by rated miles at 54%, I get 296 Wh/mi
If I divide usable kWh by rated miles between 54-100%, I get ~292 Wh/mi
So. many. fudge. factors.
Do you have the rated miles that go with that? I think it's below the regen total.
darth_vad3r
Well-Known Sith
It could still just be one fudge factor that’s a fudge formula that changes over SoC which makes it hard to analyze using assumptions at different data points
I mentioned this a lot previously. The "EPA" or dash rated range calculation is using a differen WPM depending on what SOC it's currently at. Soooo shady.....
darth_vad3r
Well-Known Sith
So the early charts posted that showed linear drop aren’t accurate, or it changed since then?
You'll have to refresh my memory. Do you mean marketing material? I posted a pic last year of a 100% to 2% trip where I graphed the BMS data using a couple thousand data points. The consumption was just about as linear as I could make it (freeway trip, 254 miles), but the real kicker is the "lying" equation Tesla is dynamically changing as you approach zero is also linear. So that tells us it's not accidental.
Edit: here we go: I think my car is lying to me...
darth_vad3r
Well-Known Sith
That’s the linear black line I was thinking of. So you are sticking with that? It’s linear. That means it’s consistent.
Every X Watts you consume the black line ticks down consistently Y miles. Right?
e.g. if your internal car constant rated consumption is 219 Wh/mi, and the display ticks down at 210 Wh/mi, they save up 9 Wh/mi for the energy buffer to adjust for errors, if any, and to give you miles below zero with that energy left over if not used for errors.
No, the fudge factor is linear. In other words, as it approaches zero they incrementally adjust the wpm calculation the same amount until they hide the 4kw you can't use (but they said you could use at the beginning). That's why it's so insidious, because it's on purpose. If it was "random" we'd chalk it up to a consumption miscalculation or an efficiency loss like you said.
eg. at 100%, they use 290wpm, but at 75% they are using 285wpm, 50% it's 280wpm, and so on and so forth until you get you "zero" and they are using something like 265 or 270, which of course is considerably optimistic and you'd have to use some serious hyper-miling to achieve it. BTW, those wpm numbers aren't exact, I'd have to check the spreadsheet to see what they are exactly.
Last edited:
darth_vad3r
Well-Known Sith
By "wpm" you mean Wh/mi? If the black line is linear, then it ticks down at a constant Wh/mi, right? That's the slope of the line.
Where in the chart are you seeing different Wh/mi ? Are you still open to sharing the source data? I wouldn't mind analyzing it.
Have you ever driven beyond 0 to see if it's actually at shut-off when it shows 0, or if any/alll of the range that's "lost" is actually still driveable before it shuts off at the anti-brick level?
e.g. like post #187 here:
Calculate usable battery capacity based on rated miles values
There's pictures there showing the energy chart going to -7% and using that "last" ~4 kWh of energy, implying that the anti-bricking buffer of ~4 kWh is completely hidden and below that.
e.g. 100% -> 0% = most of your actual usable capacity, 0% -> -5 to -8% = the remaining portion of your usable capacity that you can use to get total EPA range, and beyond that is the actual anti-bricking portion that's not shown anywhere and internally counted as "true zero" for which you cannot go below.
Last edited:
I can send you the Excel file, no problem. As for the wpm, yes, that is what the graph is showing. Changes to what the BMS is using to get remaining range. They should use the EPA number the whole way down. They don't.
As for brick protection, I don't know any of this -7% nonsense. All I know is that zero should should be what the car tells you at 100% minus what you've driven (at EPA consumption). If Tesla wants to give you an extra 4kw for the idiots out there (at such low SOC, the car is unsafe, not enough reliable power), then that's their business.
Uncle Paul
Well-Known Member
If you were Tesla, what would you do to make the system absolutely perfectly accurate
I believe that the EPA is the one that does the ratings.
I believe that the EPA is the one that does the ratings.
darth_vad3r
Well-Known Sith
True nominal zero is below dashboard zero, just like an ICE car has gas left in the tank below the “E” line. Nominal zero does not include anti-brick buffer, they don’t let you go below nominal zero. They use the EPA number minus a ~5% all the way down, so that there is a ~5% amount available to drive after the dash says zero.
A quick chart to illustrate this. It should look very familiar.
Numbers approximated based on rough memory of people posting, but this is a quick 15-second attempt to chart what’s going on and what people (like @AWDtsla recently) are interpreting as “many fudge factors” that are in fact likely just 3 simple lines.
. Two are parallel with the same slope, the other declines more steeply.y-axis is range (miles) remaining, x-axis is energy (kWh) used, starting at 0 for 100% SoC, ending at 55-ish.
To explain the constants in this example loosely based on Model 3 SR+ (based on recollection),
- 240 represents EPA range displayed at 100% SoC before any pack energy used.
- -(1/.219) represents consumption of 219 Wh/mi, in mi/kWh (slope is y/x aka mi/kWh on this chart)
- -(1/.208) represents a worse* consumption of 208 Wh/mi, in mi/kWh used to tick down the displayed miles more quickly to build up a buffer to use below dashboard zero (this is 219 - 5%).
- (1-0.05) represents setting aside 5% of the range to save for below dashboard zero (208 Wh/mi is the same 95% of 219 Wh/mi, I should have used 219*(1-0.05) for consistency actually instead of 208). The stats apps call this “usable” it looks like, but I believe a more accurate term is “above dashboard zero”. We’ve seen that there is much more usable beyond this “zero”. Nominal is all usable and the EPA test used it all until nominal zero where the hidden anti-brick buffer kicks in.
*Worse here is a “better” or lower consumption #, meaning you’d have to drive at that better-than-rated rate to see the displayed miles above zero tick down at the same rate as real miles driven.
Source of charts: Click here to recreate yourself for your own car in Wolfram Alpha and you can play around with the constants I described above to make your own version.
[EDIT: I updated that link with 219*(1-0.05) used instead of 208 so it’s easier to plunk in your own numbers by changing 240 to your range, and 219 to your car’s internal consumption constant number in Wh/mi, and the end kWh from 53 to 75 or 100 or whatever. Original link for above charts here]
Note that attempting to divide “displayed miles remaining” by underlying pack energy isn’t fruitful here. You need to use “consumed displayed miles” instead.
The formula for kWh (x) from miles (y), based on:
y = 240 - (1/.208)x,
is:x = (240 - y) miles * 208 Wh/mi
So for above numbers, if I see “100 displayed miles” remaining, I take that away from 240, so I’ve used 140 * 208 Wh/mi = 29.1 kWh. The pack should show (nominal 100% capacity in kWh) minus (29.1 kWh used) as the remaining kWh, say 52.5 - 29.1 = 23.4.
If you do it like this, it should be linear, reproducible, and no multiple fudge factors.
If you instead go: 100 * 208 = 20.8 kWh, or 100 * 219 = 21.9 kWh you don’t get the right answer, and might think “this fudge factor is changing the whole way down, I need to use 234 now?!”
Last edited:
Similar threads
