TeslaFi and TripB indicate different consumption

[eschummer]

Not sure if this exact item has come up before. I searched and couldn't find an exact match. My question is this: On a relatively short commute (~88km), the car's trip meter says I consumed 16.0 kWh, but TeslaFi says I consumed 18.55. This happens quite consistently. As a result all other calculated parameters (kWh/km, % efficiency, etc.) are all just a little off. Anybody else observed this? Am I missing something?

Thanks

 

[trayloader]

You might be on to something.
Cosmetics from Teslas´side?

 

[eschummer]

I've been wondering about that. I'm not sure how to get TeslaFi to display consumption over longer periods of time. For example. on the car it shows the lifetime average as 191 Wh/km. I couldn't find anything in TeslaFi to display an equivalent (say the last several months). to compare. I would assume TeslaFi to be correct as it is polling the car's data. On the other hand, I initially had a problem with late polling starts, i.e. the first four or five minutes of a morning commute were missing from TeslaFi. Digging into the actual data showed zero's for that time until data flow started and from then on it seemed to be correct. It had something to do with my sleep settings and the car not waking up to allow polling. Once I cancelled the sleep setting at least the travel times seemed to be correct. But the consumption in kWh? Should that not be hard data? Or does it rely on an estimate based on the percentage battery charge, which in itself is of course only an estimate.

 

[eschummer]

I think I just answered my own question. In settings under "Vehicle" TeslaFi allows you too pick a fudge factor on the consumption to more accurately reflect what you are seeing on your vehicle. 1.0 is the default value, but you can enter anywhere from 0.5 to 1.5 to more accurately reflect what you are seeing. That would lead me to believe that the car's display is actually correct and TeslaFi is an approximation..

I'll try playing with it and will report back to close out the thread...

 

[cvrcv]

You're not missing something, Tesla is. I went on a long rant about this in the Teslafi thread. The car is not correct.

Getting Teslafi and Tesla to match with the fudge factors means nothing.

I should add more details - Teslafi shows measurement by the amount of energy that has gone missing. Tesla reports incomplete consumption.

 

[eschummer]

Hmmm ....I would have thought that the car and TeslaFi are looking at the identical data set. I can see where things like percentage charge and rated miles used, etc, could be off a little because they don't have well defined reference points, but in terms of absolute kWh's used, the drive time, and the km's or miles driven that should be real verifiable numerical values - no?

 

[eschummer]

I've now put in a "fudge factor" of 0.855 to get the car and TeslaFi to agree after 80 or so driven km, but now I'm really puzzled as to what exactly I'm looking at.

Somewhat related (or perhaps not), on a longer trip I used ABRP, and there the indicated (predicted) consumption seemed to correlate quite well with what I was seeing on the car, and I have never experienced anything that gave me pause (other than long lines at certain superchargers..but that's a different story).

 

[cvrcv]

Hmmm ....I would have thought that the car and TeslaFi are looking at the identical data set. I can see where things like percentage charge and rated miles used, etc, could be off a little because they don't have well defined reference points, but in terms of absolute kWh's used, the drive time, and the km's or miles driven that should be real verifiable numerical values - no?

No. The trip meter in the car is just wrong. There's many ways to prove this, which are documented on these forums. For what it's worth, Teslafi isn't calibrated properly for my car by default either, here are directions on how to set your fudge factor properly:

Set fudge factor to 1.0. Look at a long drive for "rated miles used" (measured value), look at "battery usage" for kWh (calculated value). Find the rate mile number for your car (check wk057's threads, among other ways). Now do multiply the miles by this value, and divide by the usage.

i.e. for me: 49.1 rated miles, 15.67 kWh, and P90D is 310Wh/mi 49.1*0.310/15.67=0.971

Now, if you by hand or through Teslafi get the total rated miles used for a trip, you can calculate the kWh used by multiplying by your car's value (i.e. 0.310), as these are the same value in different units. This is almost always guaranteed to be more than what the trip meter reports.

 

[cvrcv]

I've now put in a "fudge factor" of 0.855 to get the car and TeslaFi to agree after 80 or so driven km, but now I'm really puzzled as to what exactly I'm looking at.

Somewhat related (or perhaps not), on a longer trip I used ABRP, and there the indicated (predicted) consumption seemed to correlate quite well with what I was seeing on the car, and I have never experienced anything that gave me pause (other than long lines at certain superchargers..but that's a different story).

What is ABRP?

This should give everyone pause, but it seems most are happy to overlook it. It seems you care about this so it's probably worth understanding.

BTW, you can confirm my calculation above by going in the opposite direction: Do a long charge, carefully noting the total miles added and the total kWh added through the API. Divide, and you get approximately the energy of a rated mile (310Wh/mi for me as above). Each rated mile entering or leaving the car is a unit of energy, there's no way to game that system. It can be miscalibrated, but it's a zero-sum game over many drives in the event there's an adjustment in reported available energy.

 

[eschummer]

Sorry, but my old rusty engineering brain has a hard time accepting "rated miles" as a unit of energy. I can relate to kWh (i can measure that), miles (or km - I can also measure). So my (perhaps erroneous) assumption was that the data set would be based on measurable units and that things like "rated miles", "charge percentage", etc. would be estimates based on some educated assumptions. I don't doubt your formula, the result looks reasonable, but I have a hard time getting my head around the reasoning....

I had always thought that a "rated mile" was simply based on an EPA cycle guesstimate of the total number of miles available on a full charge, and you can then see how efficiently you are driving by comparing your "real" miles to your "rated miles" consumption. It's been somewhat artificial in my view - it's comparable to the EPA "rated" fuel consumption in MPG for an ICE car. Take a full tank of gas and and multiply by the MPG and see how many "rated" miles the car can go.

Thanks for the explanation though, I will try what you suggest and see where it comes out.

 

[cvrcv]

Sorry, but my old rusty engineering brain has a hard time accepting "rated miles" as a unit of energy. I can relate to kWh (i can measure that), miles (or km - I can also measure). So my (perhaps erroneous) assumption was that the data set would be based on measurable units and that things like "rated miles", "charge percentage", etc. would be estimates based on some educated assumptions. I don't doubt your formula, the result looks reasonable, but I have a hard time getting my head around the reasoning....

That's just how it's calculated internally. Actually Wh, %, rated miles, ideal miles can all be directly translated to each other via multiplication. I'm leaving out any funny business like having the snowflake icon and reporting reduced range, or things that might happen as the battery degrades. So you see people switching from rated miles display to %, they change nothing except their perception.

I had always thought that a "rated mile" was simply based on an EPA cycle guesstimate of the total number of miles available on a full charge, and you can then see how efficiently you are driving by comparing your "real" miles to your "rated miles" consumption. It's been somewhat artificial in my view - it's comparable to the EPA "rated" fuel consumption in MPG for an ICE car. Take a full tank of gas and and multiply by the MPG and see how many "rated" miles the car can go.

It is based on EPA rating, you can work backwards from the EPA rating to the value used internally if you compensate for charging efficiency and vampire drain (as observed when tested), using a standard amount of miles driven per year. i.e. EPA rating for my car is 360Wh/mi adding both additional factors. But the number used in the computer is 310.

 

[mknox]

Basically, Tesla only records the energy coming OUT of the pack when the car is ON. It does not take into account energy used while the car is off or charging... things like charger losses, vampire drain and energy used by the HVAC when turned on remotely are not accounted for. Further, if you park and charge with the HVAC on, Tesla will only show you what went into the pack and not the additional energy used for heating/cooling.

 

[cvrcv]

Basically, Tesla only records the energy coming OUT of the pack when the car is ON.

Sorry, it even falls short of this. The location of the shunt resistor does not account for this either, unless someone wants to claim a 90% efficiency coming out of of the pack at a small fraction of C is normal. If this were true the cells would basically be eating themselves. Anything that isn't a I^2R heat loss is energy put into cell chemistry degradation.

 

[mknox]

Sorry, it even falls short of this. The location of the shunt resistor does not account for this either, unless someone wants to claim a 90% efficiency coming out of of the pack at a small fraction of C is normal. If this were true the cells would basically be eating themselves. Anything that isn't a I^2R heat loss is energy put into cell chemistry degradation.

Fair enough. I was generalizing to the larger issues that the layperson might notice.

 

[Nifty-Stuff]

Wow, this was a great read, thanks guys!

I came in here looking for info because over the past 600 miles I have the following:

• In car shows 251 Wh/mi
• Teslafi shows 159 Wh/Mile

?!?!

After reading through this thread, I went into my TeslaFi settings to look at the "kWh & Wh/m(km) Factor" setting and noticed my car was set as a Standard Range Dual Motor and not my actual "Long Range Dual Motor".

... would this account for such crazy discrepancy in the reported Wh/mi?!?

I was hoping that changing the setting would retroactively change all the data, but I didn't see any impact. I guess I'll need to see what TeslaFi.com shows once I do a few more drives.

If it's still totally off, I guess I'll need to consider seriously tweaking the "factor" setting.

I went on a long rant about this in the Teslafi thread

@cvrcv , you seem to know your stuff! I couldn't find your "rant" using search, so if you wouldn't mind pointing me to it (or any other related posts / threads) that would be epic!!! Thanks!

 

[emir-t]

@Nifty-Stuff I think your Teslafi could be Wh/km. because 251/159=1,578 which is very close to mile to km ratio.

@mknox Exactly. Add to that Tesla only accounts for consumption when the car is moving. So even being stuck in traffic or waiting in traffic lights while A/C running is consumption not accounted for. So I think Teslafi correction factor is good and should be used. Yet it makes Teslafi, which only works over API, less accurate than the Tesla on board computer. I'll compare Teslafi vs. onboard consumption next time I make a drive with lots of stops and waiting.

 

[cvrcv]

Wow, this was a great read, thanks guys!

I came in here looking for info because over the past 600 miles I have the following:

• In car shows 251 Wh/mi
• Teslafi shows 159 Wh/Mile

?!?!

After reading through this thread, I went into my TeslaFi settings to look at the "kWh & Wh/m(km) Factor" setting and noticed my car was set as a Standard Range Dual Motor and not my actual "Long Range Dual Motor".

... would this account for such crazy discrepancy in the reported Wh/mi?!?

I was hoping that changing the setting would retroactively change all the data, but I didn't see any impact. I guess I'll need to see what TeslaFi.com shows once I do a few more drives.

If it's still totally off, I guess I'll need to consider seriously tweaking the "factor" setting.


@cvrcv , you seem to know your stuff! I couldn't find your "rant" using search, so if you wouldn't mind pointing me to it (or any other related posts / threads) that would be epic!!! Thanks!

Setting the wrong car will definitely get you a wrong answer. However for more up to date info this thread is probably better: This is why you can't get 'rated range'