Consumption chart: average vs instant range

[dazzap]

Model Y 2022.4.5.3

Average range shows the predicted range based on last 5, 15 or 30 mile average Wh/mile consumption. So far so good.

The Instant range calculation also gives different predictions for 5, 15 or 30 miles (which are all themselves different to Average). This is bonkers as 5,15 or 30 mile periods has no meaning in relation to an instantaneous calculation. I realise I should have taken pics of each setting, but try it yourself. On Instant range, watch the prediction change depending on what distance you choose.

The manual describes both average and instant exactly how you'd expect

"Instant Range uses only the latest few data points to estimate the projected range, whereas Average Range uses the past 5, 15 or 30 miles (10, 25 or 50 km) of energy consumption to provide a more accurate projected range."

OK, ready for my "Doh!" moment... what have I missed?

 

[pgkevet]

Instant is using the last data points and they have been economic - so exactly as expected..

 

[dazzap]

Instant is using the last data points and they have been economic - so exactly as expected..

I think you've missed the point. Its possible to get THREE different instant predictions depending on 5, 15 or 30 miles. That's not how instant is supposed to work.

The 5/15/30 buttons should be greyed out for instant.

Here's the other two "instant" range predictions! Which one are you supposed to believe???

 

[pgkevet]

What can I say? Doh!

 

[GRiLLA]

I think you've missed the point. Its possible to get THREE different instant predictions depending on 5, 15 or 30 miles. That's not how instant is supposed to work.

The 5/15/30 buttons should be greyed out for instant.

Here's the other two "instant" range predictions! Which one are you supposed to believe???
View attachment 774782

The buttons define how far back the graph goes, not just the projected range, which is why they aren't greyed out. I agree it's a bit odd for instant to change, but if as per the quote it uses the 'latest few data points' I guess the different resolutions change which those are.

 

[MrT3]

Which one are you supposed to believe???

None of them.

Calculating range based on such short term data is never going to be accurate anyway. At a guess I'd assume the instant range figure at 5 miles is taking into account the last 2 miles data (5/2), and 15/5 and 30/10 for the other two. All a bit odd, and pointless.

 

[dazzap]

None of them.

Calculating range based on such short term data is never going to be accurate anyway. At a guess I'd assume the instant range figure at 5 miles is taking into account the last 2 miles data (5/2), and 15/5 and 30/10 for the other two. All a bit odd, and pointless.

But what you're suggesting there is actually just an average over 2, 5, 10, 15 or 30 miles.

There's going to be a simple answer to this.. it's just totally non-obvious but by definition an instant value isn't a really an instant value if it depends on a previous average.

 

[MrT3]

There's going to be a simple answer to this.. it's just totally non-obvious but by definition an instant value isn't a really an instant value if it depends on a previous average.

It will never be instant as the vehicle is reading those parameters at least 3 or 4 times a second so you would never be able to properly read it - so it has to average it slightly. What we don't know is what constitutes the 'latest few data points' but what we can tell from your results is that each data point is different for each of the three time periods. Probably not as large as my exaggerated example but clearly there is a difference.

Did you notice how often the values updated? They may give a clue as to the average duration at each interval.

 

[init6]

OK, ready for my "Doh!" moment... what have I missed?

The search bar

Why is the instant projected range different on the 5/15/30 energy graphs?

Can someone explain why the instant projected range isn't the same across the different scales on the energy graphs? It's pretty clear that the average energy use over the 5, 15, and 30 mile scales would be different, but I'm having a hard time understanding what would make the instant projected...

teslamotorsclub.com

 

[dazzap]

The search bar

Why is the instant projected range different on the 5/15/30 energy graphs?

Can someone explain why the instant projected range isn't the same across the different scales on the energy graphs? It's pretty clear that the average energy use over the 5, 15, and 30 mile scales would be different, but I'm having a hard time understanding what would make the instant projected...

teslamotorsclub.com

Well no actually... I did a couple of searches and didn't have any luck. But, as it happens I don't think anyone properly answered the question and quite a few didnt appear to understand the question!

 

[init6]

The first answer in that thread seems reasonable enough. Why do you disagree?

 

[dazzap]

The first answer in that thread seems reasonable enough. Why do you disagree?

It's nothing more than a guess.. albeit not totally unreasonable. It depends what Tesla mean by a data-point. In the world of telemetry devices record a constant stream of "raw" data at some period, eg 100 milli-sec, 1 meter, whatever. Normally raw sampling is done at a high rate and then averaged values are stored at a lower resolution.

If instant was looking at the most recent set of raw data points the projected range would be the same regardless of what type of average was being displayed. Just like instant mpg the car see's power and speed from which a range could be continuously calculated. Most ICE cars probably dont even try to display a range from the instant mpg because it would be changing so much. So I doubt Tesla is doing that either.

The average range looks like to be calculate from a linear average over 5, 15 or 30 miles. It could be that the instant range is derived from a short duration moving average but over what period and what dataset. Is there a single set of data points spanning a total of 30 miles and at what frequency?

Because Tesla dont document how the instant value is calculated it's use is of really limited value - which is a shame since they presumably feel its of great use to have bothered putting it in the UI. I was hoping someone on the interweb would have inside knowledge or had it explained to them. But this doesnt appear to be the case and in the great scheme of things it's really not important. But I'm a data geek who deals with telemetry on a daily basis so perhaps have a peculiar interest in it.

 

[init6]

Any sub-second sampling instant value is useless. Hence why it's reasonable to expect some sort of averaging.

 

[dazzap]

Any sub-second sampling instant value is useless. Hence why it's reasonable to expect some sort of averaging.

Well of course. However, the instant range value really should be the same regardless of which graph is currently being displayed because if you watch it while driving it updates every few seconds... so what factors would affect the range based on the last few seconds of averaged consumption? Nothing that I can think of... but they are clearly running a calculation that does somehow include the 5/15/30 mile average as well as the actual almost-instantaneous consumption.

 

[init6]

Well of course. However, the instant range value really should be the same regardless of which graph is currently being displayed because if you watch it while driving it updates every few seconds... so what factors would affect the range based on the last few seconds of averaged consumption? Nothing that I can think of... but they are clearly running a calculation that does somehow include the 5/15/30 mile average as well as the actual almost-instantaneous consumption.

The other post posits that the sampling rate is changed based on the 5/15/30 selection so the values would be different.
Ultimately though the instant value is just a guess no matter what aggregation is used. I suspect its just trying to mimic ICE behaviour.