Charging efficiency vs Vampire Loss

[insaneoctane]

I've been gathering data to monitor my charging efficiency. I chose to roll it myself vs just use TeslaFi because I wanted to include my EVSE Current Transformer taps which includes the actual power used, not just what my Tesla is telling me. So I'm pulling my charging data via the Tesla API daily. Anyway, I'm noticing the weekends (when I don't drive my car as much) are showing far more "loss" between what Tesla claims was used vs my measurements. This is showing in my charging efficiency calculation, which I don't think that's accurate. I suspect it's more due to energy that the Tesla is pulling while being left plugged in to run pumps, feed vampire losses, etc, that don't show in Tesla's reporting. So I'm wondering if anyone has any ideas on how to isolate these vampire losses in order to more accurately measure true charging efficiency?

Date    Day    Odometer    kWh, indicated    kWh actual    State @ measure    Mileage    Wh/mi    Charging Efficiency
5/21/2018    Mon    1029.6    2.9    3.4    Complete    3.4    987.4    85.85%
5/22/2018    Tue    1151.4    26.4    29.7    Complete    121.9    243.3    89.04%
5/23/2018    Wed    1262.8    25.2    27.9    Complete    111.4    250.4    90.37%
5/24/2018    Thu    1396.6    32.8    37.2    Complete    133.8    277.7    88.26%
5/25/2018    Fri    1516.7    26.8    29.9    Complete    120.1    249.3    89.51%
5/26/2018    Sat    1556.5    13.3    15.0    Complete    39.7    376.3    88.93%
5/27/2018    Sun    1599.4    5.7    6.7    Disconnected    42.9    156.3    85.00%
5/28/2018    Mon    1602.7    8.9    10.2    Complete    3.3    3039.6    87.41%
5/29/2018    Tue    1628.7    8.4    9.6    Complete    25.9    370.8    87.35%
5/30/2018    Wed    1759.0    29.4    32.8    Complete    130.4    251.4    89.70%
5/31/2018    Thu    1879.3    27.6    31.2    Complete    120.3    259.5    88.44%
6/1/2018    Fri    1993.6    24.7    27.7    Complete    114.3    242.2    89.24%
6/2/2018    Sat    1994.3    0.5    0.7    Complete    0.7    1110.1    67.93%
6/3/2018    Sun    2002.6    4    4.6    Complete    8.3    557.0    86.04%
6/4/2018    Mon    2016.8    4.2    4.8    Complete    14.1    336.4    88.25%

 

[AWDtsla]

Use "energy added" from the API to check efficiency, you should get a very consistent number unlike the above. You'll hit ~92%, I'd bet.

 

[insaneoctane]

Use "energy added" from the API to check efficiency, you should get a very consistent number unlike the above. You'll hit ~92%, I'd bet.

I am using energy added for the "kWh indicated".

Here's (hopefully) better formatted text.....I missed the 10 minute window to fix original message...

Date          Day    Odo    kWh,indic actual   State@meas  Mileage  Wh/mi    Charging Efficiency
05/21/2018    Mon    1029.6    02.9    03.4    Complete    003.4    0987.4    85.85%
05/22/2018    Tue    1151.4    26.4    29.7    Complete    121.9    0243.3    89.04%
05/23/2018    Wed    1262.8    25.2    27.9    Complete    111.4    0250.4    90.37%
05/24/2018    Thu    1396.6    32.8    37.2    Complete    133.8    0277.7    88.26%
05/25/2018    Fri    1516.7    26.8    29.9    Complete    120.1    0249.3    89.51%
05/26/2018    Sat    1556.5    13.3    15.0    Complete    039.7    0376.3    88.93%
05/27/2018    Sun    1599.4    05.7    06.7    Disconne    042.9    0156.3    85.00%
05/28/2018    Mon    1602.7    08.9    10.2    Complete    003.3    3039.6    87.41%
05/29/2018    Tue    1628.7    08.4    09.6    Complete    025.9    0370.8    87.35%
05/30/2018    Wed    1759.0    29.4    32.8    Complete    130.4    0251.4    89.70%
05/31/2018    Thu    1879.3    27.6    31.2    Complete    120.3    0259.5    88.44%
06/01/2018    Fri    1993.6    24.7    27.7    Complete    114.3    0242.2    89.24%
06/02/2018    Sat    1994.3    00.5    00.7    Complete    000.7    1110.1    67.93%
06/03/2018    Sun    2002.6    04.0    04.6    Complete    008.3    0557.0    86.04%
06/04/2018    Mon    2016.8    04.2    04.8    Complete    014.1    0336.4    88.25%

 

[AWDtsla]

I am using energy added for the "kWh indicated".

Hmm, well you should discard small charges as they will be noisy, but there's still quite a bit of variance left. The other thing is that you need to measure only the charging session, and do so immediately after the charge. Energy Added can go backwards after it's done charging. Power consumed can also go up for any loads like preconditioning, basically anything above 240V@1A

I used to get efficiency per charge down to about +/- 0.1%, at least on the firmware from 2 years ago, when not getting to current limited charging i.e. too cold or SoC >90%

 

[insaneoctane]

The other thing is that you need to measure only the charging session, and do so immediately after the charge. Energy Added can go backwards after it's done

My data pull happens right after the charging status indicates "complete"

 

[mongo]

I assume so, but does your data logger monitor the line voltage along with the current?

 

[insaneoctane]

I assume so, but does your data logger monitor the line voltage along with the current?

Yes

 

[mongo]

Yes

Near the EVSE? Not that I'd expect a lot of line loss...

Edit: What is the load of the EVSE itself?

 

[AWDtsla]

Looking at it again, if you only look at >24kWh you are within +/- 1%. Which is not really all that bad. I don't know how accurate you want to try and get it. Try only looking at charges >40kWh.

 

[insaneoctane]

Looking at it again, if you only look at >24kWh you are within +/- 1%. Which is not really all that bad. I don't know how accurate you want to try and get it. Try only looking at charges >40kWh.

What numbers are you using to calculate +/-1% ?

 

[AWDtsla]

What numbers are you using to calculate +/-1% ?

Efficiency. Just drop all rows under 24kWh first. It's actually points not %, but close enough.

 

[insaneoctane]

Those efficiency numbers are just calculated by dividing kWh indict / kWh actual ...

 

[davedavedave]

I used teslafi for 2 days and saw a lot of vampire drain - it doesn’t let the car sleep. Heisenberg Principle at work!

 

[SageBrush]

I plan to compare my vampire drains with smartphone on and car used as recommended vs smartphone off and car off for most days of the week in terms of rated miles lost while car is parked.

We'll see.
It might turn out that I don't value the immediate step into car and go feature from home enough to tolerate the vampire drain, particularly if the start-up sequence is under 30 seconds.