Total charge kWh being consumed

[ajmorr.is]

I put a charge into the car last night, and thought I’d check how much actually went in. I use Intelligent Octopus and this reported 47.26kWh, the Tesla app and the linked Home Assistant which pulls info from Tesla reported 49kWh (actually 48.82kWh) and my Easee charger reported 50.7kWh.

That’s quite a difference between the charger and IO! So which values do I trust?

 

[Kansalis]

An amount of guess-work and assumption, but this would make sense:

• IO displays what the car reports via the Tesla API as power added to the battery - 47.26
• Home Assistant reports a different stat from the Tesla API, total power used - 48.82
• The charger reports what it actually supplied - 50.7

This seems about right to me, when compared to the stats reported by the various API data points and my charger. Maybe a bit more loss from charger to supplied power than I'd expect.

 

[Kansalis]

Just checked a recent charge of mine. I only charge off peak in Go hours at the moment, so don't have a session with such big numbers, but here's the biggest one from my recent sessions:

• 32.36 kWh added to battery (Tesla API)
• 33.6 kWh used (Tesla API)
• 34.1 kWh supplied by charger (PodPoint app)

 

[speedyranger74]

I have the same issue but different configuration:
Tesla - 14.18 kwh
Rolec charger via ev.energy - 15.01 kwh
Tessie - 14.39 kwh
IO - 11.52 kwh.
My opinion: IO assigned less kwh on smart charging period and the rest will add to the bill but more expensive.
Example for one month (10p for car/off peak and other 41p):
Car - 30 kwh = £3.00
Other - 70 kwh = £28.7
Total - 100 kwh = £31.7

But if the car charging was more...
Car - 70 kwh = £7.00
Other - 30 kwh = £12.3
Total - 100 kwh = £19.3

 

[NewbieT]

Have you totalled your smart meter readings for the same period?

 

[ajmorr.is]

Have you totalled your smart meter readings for the same period?

That's impossible as I would need to know how much every other device in the house uses and subtract from the charging period.

 

[ajmorr.is]

An amount of guess-work and assumption, but this would make sense:

• IO displays what the car reports via the Tesla API as power added to the battery - 47.26
• Home Assistant reports a different stat from the Tesla API, total power used - 48.82
• The charger reports what it actually supplied - 50.7

This seems about right to me, when compared to the stats reported by the various API data points and my charger. Maybe a bit more loss from charger to supplied power than I'd expect.

IO pulls from Tesla, but there's still a difference between the IO figure presented and the one provided by Tesla! Tesla and Home Assistant are both presenting the same figure by the way (although the Tesla app rounds up); on the screenshot I posted of the Tesla app I had the larger of the two blocks selected. The smaller one (before midnight) is 5KWh.

I'd be tempted to believe the charger as the ultimate meter of the power consumed, but I don't know for sure.

 

[kelvin 660]

I've looked at my charging history and this is what I've got...

My charge for the 10th February;
Octopus SMART meter (average for period with no charge) = 1.57 kWh
Octopus SMART meter (average for period with charge) = 27.20 kWh
Less average for no charge period = (27.20 - 1.57) = 25.63 kWh
From TeslaFi = Used: 24.83 kWh Added: 22.91 kWh
Difference between SMART meter and TeslaFi = 3.2%

My charge for the 15th February;
Octopus SMART meter (average for period with no charge) = 1.57 kWh
Octopus SMART meter (average for period with charge) = 27.76 kWh
Less average for no charge period = (27.76 - 1.57) = 26.19 kWh
From TeslaFi = Used: 25.22 kWh Added: 23.71 kWh
Difference between SMART meter and TeslaFi = 3.8%

My charge for the 17th February;
Octopus SMART meter (average for period with no charge) = 1.57 kWh
Octopus SMART meter (average for period with charge) = 22.86 kWh
Less average for no charge period = (22.86 - 1.57) = 21.29 kWh
From TeslaFi = Used: 20.64 kWh Added: 19.52 kWh
Difference between SMART meter and TeslaFi = 3.1%

Anyway, it looks like there is a approximately a 3% difference between what TeslaFi records as used compared to what the SMART meter says has been delivered... I'm not sure how how Tesla app and TeslaFi compare, but I will check next time I charge the car...

 

[NewbieT]

That's impossible as I would need to know how much every other device in the house uses and subtract from the charging period.

What I mean is, if you wanted an accurate reference point a smart meter is manufactured/certified to a tolerance.

I know it won’t give you an answer for that specific charging session retrospectively, but pull all of the breakers, charge your car for 1h and wait 24h for the result.

 

[Kansalis]

IO pulls from Tesla, but there's still a difference between the IO figure presented and the one provided by Tesla! Tesla and Home Assistant are both presenting the same figure by the way (although the Tesla app rounds up); on the screenshot I posted of the Tesla app I had the larger of the two blocks selected. The smaller one (before midnight) is 5KWh.

The Tesla API includes two different data points. One is how much added to the battery & the other is power used. There will always be a difference between them, charging is not a zero loss process. It's quite possible IO is reporting one while the Tesla app and HA are pulling from the other. Without either explicitly stating which they are referring to, it's guesswork to some extent.

 

[ajmorr.is]

What I mean is, if you wanted an accurate reference point a smart meter is manufactured/certified to a tolerance.

I know it won’t give you an answer for that specific charging session retrospectively, but pull all of the breakers, charge your car for 1h and wait 24h for the result.

Do you want to tell my wife that Instagram will be offline for an hour or should I?

 

[ajmorr.is]

The Tesla API includes two different data points. One is how much added to the battery & the other is power used. There will always be a difference between them, charging is not a zero loss process. It's quite possible IO is reporting one while the Tesla app and HA are pulling from the other. Without either explicitly stating which they are referring to, it's guesswork to some extent.

Interesting.

I do understand that there will be losses, but I can't imagine for a 9 hour session at 7.4kW it would be anywhere near the region of 1kW! I'm starting to err more towards the charger being the most accurate in this case.

 

[Kansalis]

Anyway, it looks like there is a approximately a 3% difference between what TeslaFi records as used compared to what the SMART meter says has been delivered... I'm not sure how how Tesla app and TeslaFi compare, but I will check next time I charge the car...

Does your SMART meter report separate usage for your EV charger and other home use, or is that what the "no charge" average is? I'd guess there's some wiggle room in the accuracy of that 1.57 kWh stat & your house base load may be that extra ~3%. There will be some additional loss between what the car thinks is "used" versus the smart meter too, though 3% seems high.

This is my TeslaMate and TeslaFi data from my most recent 10 charges. Interestingly they don't report the same numbers, so one or both must be doing their own calculations somewhere rather than reporting on raw data from the API. TeslaFi also misses a lot of API calls with "API Error - Connection URL Timeout", which probably doesn't help data accuracy. I'd be inclined to think TeslaMate is more accurate.

 

[mspohr]

I put a charge into the car last night, and thought I’d check how much actually went in. I use Intelligent Octopus and this reported 47.26kWh, the Tesla app and the linked Home Assistant which pulls info from Tesla reported 49kWh (actually 48.82kWh) and my Easee charger reported 50.7kWh.

That’s quite a difference between the charger and IO! So which values do I trust?

View attachment 916016View attachment 916017View attachment 916018
View attachment 916019

Did anyone ever tell you that your Avatar is irritating?

 

[kelvin 660]

Tesla's api doesn't report energy used, only energy added to the battery. TeslaFi uses the charge voltage and current draw (once per minute) to calculate energy used, so there will be some errors. I'm not sure about how TeslaMate works out energy used...

This issue needs more investigation so I might have to trip the house breaker and charge the car for a couple of hours to get some reliable data...

 

[Sean.]

Interesting.

I do understand that there will be losses, but I can't imagine for a 9 hour session at 7.4kW it would be anywhere near the region of 1kW! I'm starting to err more towards the charger being the most accurate in this case.

The AC charger in the car is around 94% efficient when running close to its rated draw. So if the gross figure is 9 x 7.4 kW = 66.6 kWh, then you can expect the loss to be around 6% or roughly 4 kWh

 

[Sean.]

Tesla's api doesn't report energy used, only energy added to the battery. TeslaFi uses the charge voltage and current draw (once per minute) to calculate energy used, so there will be some errors. I'm not sure about how TeslaMate works out energy used...

This issue needs more investigation so I might have to trip the house breaker and charge the car for a couple of hours to get some reliable data...

I compared the figures reported by the apps (they will use the same API) and the power as measured by a MID compliant power meter and they were within 1%. The Tesla figure being more pessimistic.

 

[NewbieT]

Columns in white Tesla API. Columns is Yellow calculated. My car wasn't warming the battery during this charging.

Issues with the P=IV calc method
1. Starting and stopping charging: My charge stopped part way through a measurement interval. So did it charge for 20 seconds or 40 seconds in that window. Say an App polls and takes readings at 07:55:00 and 08:00:00 (I wouldn't because Tesla's servers get very busy 'on the hour' and timeout). Car stopped charging 07:56:00. Does the app assume the car was or wasn't charging in that 5 minute? At 7kW that 4 minutes creates a variance of +/- 0.46 kWh. Similar issue when the car starts to charge and is ramping up.
2. Variance. The calculation assumes a constant voltage and current in between the measurement intervals (which it won't be). Example 07:13 where Tesla and calc vary by c20%. There's no way around this other than poll more frequently (and there's no point - see conclusion).
3. Measurement intervals/frequency/resolution. Mine were set at 1 min. I'd be surprised if these systems/Apps are polling at that rate for fear of Tesla firewalling them on rate limits (x lots of cars). Lower resolution = lower accuracy. I got a few duplicate records (sorry deleted in pic) which suggests Tesla's internal car/polling/logging/data transfer intervals aren't that frequent.

Over an entire charging session with 1 minute resolution, the P=IV method seems to be pretty close to Tesla's reported 'charge energy added', so I would use this instead of wasting time calculating it.

This doesn't answer how accurate the car's measurements are, or address charging overheads (one for scanmytesla?), it just says to me that the charge_energy_added figure is as good as you will get from the car (without attaching measuring devices to the car).

I suspect TeslaFi's 'Added' is calculated from the increase in battery range and converting that back to kWh based on a calculated battery size. Example: my range increaed by 4.233% in this example. (If!) the battery is 75kWH that's 3.18kWh added (my battery got warmer). This method assumes (calculates) a battery size and has no idea as to the balance of the battery / battery temp / state of the BMS.

 

[kelvin 660]

Tesla's api doesn't report energy used, only energy added to the battery. TeslaFi uses the charge voltage and current draw (once per minute) to calculate energy used, so there will be some errors. I'm not sure about how TeslaMate works out energy used...

This issue needs more investigation so I might have to trip the house breaker and charge the car for a couple of hours to get some reliable data...

So I decided not to put the home in darkness and use the granny charger with a Tapo P111 energy monitor to monitor the actual energy supplied to the charger. The results are;

Charged for 3hrs 58 mins (3.966hrs) and Tapo recorded 7.075kWh
TeslaFi recorder Used: 6.83 kWh Added: 6.16 kWh
Tesla recorded 6kWh (added)

Tesla is only showing charge added to the battery (rounded down) to 6 kWh. Not very accurate!
And TeslaFi is 7.07 / 6.83 ~ is 3.5% reading low (as per my other data)
And Tapo recorded what was actually used (7.075 kWh)

So if you are like me, you need to be adding an additional 3.5% when calculating actual energy supplied to the car when using data from TeslaFi. So for example, since I have owned the car TeslaFi says I spent £ 307.71 charging, but in reality I've actually spent £ 318.48 for 15,700 miles. Not such a big difference...

Sanity check;

Energy supplied = Energy added to the battery + Energy used to operate the computer + other losses. The computer takes ~ 233 watts when running.
Therefore 7.075 kWh supplied = 6.16 kWh to the battery + (3.996 hrs x 0.233 kW computer) = 0.931 kwh + other losses

Therefore other losses = 7.075 - 6.16 - 0.931 = 0.016 kwh. So it seems that other losses (ac to dc conversion) is very low!

 

[MrBadger]

How is kWh used and added calculated? / Knowledge base / TeslaFi

KWh added is obtained directly from the TeslaFi API. It is calculated by the difference between the starting energy_added field and the ending energy_added field of the charging session.kWh used by the charger is calculated by TeslaFi and not directly obtainable by the API. Each data point is...

support.teslafi.com

Also BMS energy guesses can be very variable in accuracy. Not unheard of seeing a -ve efficiency (now capped at 0%) or >100% (iirc now capped at 100%) efficiency on some TeslaFi reported charges. Normally the very low or very high efficiency charges indicate a BMS recalculation during that period so start or end battery energy estimates are not with the BMS in the same state.