|
OK, if I am reading your data correctly, you have a large variation in readings in the one configuration - like, 12 - 13 per cent variation. That doesn't surprise me from on-road energy consumption testing over a hilly course, but to then calculate averages (and only from a couple of data points) and claim improvements down to decimal places (eg 1.4 per cent) seems a bit optimistic.
I did an analysis of some of your data and I think that, for example, 2 Bar (not 2.5 Bar) tyre pressures gives better results than higher or lower pressures. At minimum you need to use all the data (not just average a couple of the measurements for each configuration).
If I were doing the testing, I think I'd consider variations between your averaged data sets significant only if the difference was ~5 per cent or greater.
Do you have a high resolution, instant power readout, real time? If so, holding one power value and then accurately measuring changes in speed with different tyre pressures may be more accurate. In ICE cars it's certainly much more repeatable than what you have shown here (I get less than 1 per cent variation in consecutive runs).
I've been testing on the road for more than 35 years. If, with the car in an identical configuration, any test technique gives a variation of 12-13 per cent on consecutive runs, then to me it is a highly untrustworthy technique.
For example, my throttle stop testing technique, if done appropriately, gives a maximum variation of 1-2 per cent in consecutive runs - ten times less data spread than you are seeing!
15 years ago when I was testing my Prius and Insight, the on-road fuel consumption data scatter over the course, with the car unchanged, was about 3 per cent.
I think it's great doing the testing, but I think you have to be far more realistic about what you are seeing. For example, I doubt any of your aero mods make any measurable difference on the road - the change in Cd would be far too small to measure in the way you suggest.
|