Quote:
Originally Posted by Vekke
If you are not able to test something accurately does not mean someone else cannot do better. O fcourse even these results there is variation. Last run 0,1kWh change came from temperature. The other variation came from start which reason is unknown to me at this point.
https://youtu.be/JhcDhwnwU0I
That 2nd run to east was off from start. Almost came to same result still.
5 times 16,4 and one time 16,3=98,3/6=16,38
4 times 14,1 and one time14,2=14,12
--> Average consumption in my test route 15,25kWh
What is the accuracy if you don`t take the temp changes into account?
I say its under 0,1 kWh accurate based on this set.
 |
1. If these are all records of your running average, taken one after the other on the one run, then only the figures at the end matter - otherwise, as had already been said, you're double counting. (That what you seem to have done in your calculation, but just confirming that for others who will see all the figures and think they're all for different runs.)
2. You cannot go from three significant figures (your readings) to four significant figures (your average). If you don't want to take my word for it, do some research - eg
https://courses.lumenlearning.com/in...icant-figures/ ("The final answer in a multiplication or division problem should contain the same number of significant figures as the original number with the fewest significant figures.")
3. Therefore, your average is 16.4 one way, and 14.1 the other way. Therefore, on resolution alone, the smallest variation you can measure is 0.6 per cent one way and 0.7 per cent the other way. (If I were doing the testing, I'd in fact say that I couldn't expect to see any changes of less than 1 per cent, but I am very conservative.)
4. I think you'd need to do what I always do - do the runs standard (to set the baseline) and then make the aero change and then immediately redo the runs in modified form. Then, for best accuracy, remove the mods and redo the runs, ensuring the measured results match the first series of runs in that session. ie A / B / A.
6. And maybe stop listing the running averages that add nothing to the accuracy of anything?