Quote:
Originally Posted by Vekke
Test method:
Drive car warm (takes 20 km on current weather in my ID3 (warm is looked from idle fuel consumption rolling in neutral at 100km/h speed. Idle consumption is around 2,5 kwh. When that is reached the results has been reliable on the course.
- Set cruise to 103 km/h
- Zero on trip on start
- Fill measurement points on the route from AVG consumption display (highest or lowest values on the route in most cases)
- In the end fill data from end. 5,5km long test.
Make a U turn and repeat to Finish which is same as start to other direction.
Calculate a+b take 4% speedo error into account when comparing to old results.
|
So you drive at 103 km/h for just over 3 minutes, then turn around and drive back the other way at 103 km/h for the same length of time?
And from
just those two runs you expect to get accurate energy consumption data for the car in that configuration?!
To be blunt, that's just ridiculous.
The number of cumulative average readings you take during the 3 minutes is irrelevant - it does nothing to improve accuracy. (I'd assumed
each of those readings was for a full run!)
I am not sure what else to say: nearly everything I can think of (except having the car warmed-up) is as wrong as it could be in this approach:
- Vastly too short a distance over which to measure an average with the accuracy you're after.
- No measuring of baseline scatter (ie variations with the car in one configuration).
- Deliberately, it seems, introducing extra variables (changing load with hills).
- Recording of data that does nothing to improve accuracy (the interim cumulative averages).
I am irresistibly drawn to reflecting on the testing of EM's Jim Michler, who I quote in my big aero book. The testing of different configurations of his truck took place over 4,000km (winter) and 5,600km (summer). Now
that had credibility!