guy keren <guy.choo.ke...@gmail.com> writes:
> "running average" is meaningful only if you know the period of time
> it's taking into account ;)
Short answer: no. ;-)
No, I am not daft, just trying to keep you interested. ;-) Read on for a
full explanation.
The averaging period is probably written down somewhere in the manual
but I don't remember it. I made the experiments years ago when I got the
car and learned the computer functionality, purely out of curiosity[1].
You do not need to know the averaging period exactly. What happens is as
follows. You start out in a city and crawl in traffic and stop at
traffic lights and your fuel consumption is lousy. Then you get to a
highway to Jerusalem or Haifa and quickly reach your target cruising
speed and everything is a bliss. As you watch km/l the number keeps
climbing up because for a while the computer still remembers that you
used to burn fuel at traffic lights without moving forward, but with
time it will gradually forget it. At some point it will more or less
stabilize, assuming that you have enough of the highway to drive at a
more or less constant speed for a while. It may fluctuate a bit even on
a perfectly flat road at a perfectly constant speed (basic physics like
wind and road surface affect consumption), but any fluctuations,
including the speed that is not perfectly maintained, will be smoothed
out to a large extent by the averaging (details in note [2] below). If
you do it a couple of times at 80km/h and a couple of times at 100km/h
and notice a difference the result will not be "scientific" but you will
get a basic idea which cruising speed is more economical.
If your road computer gives you an average between resets you can
perform a similar experiment pressing the button twice while driving at
a constant speed. It will measure essentially the same thing. You don't
need to wait until the computer forgets the previous regime, and I don't
need to press buttons. As I said before, there is no reason to expect
similar results for different cars.
What Nadav describes is good for a different question: he assumes,
reasonably enough, that over a long time fuel consumption reflects his
long term average experience, and since he did it or two different cars
he got a comparison of long term mixed regime averages for two
models. This is more relevant to the original question asked in the
thread. We took the discussion on a tangent - apologies to the audience.
[1] I have no objective to minimize my fuel consumption down to the last
gram. If I need to go to place X I will, and if there is traffic I
will be stuck - so what? Normally, I will optimize time rather than
fuel. In many cases both may mean sticking to major roads and
avoiding traffic jams (and traffic lights) as much as possible.
[2] So the averaging period must be longer than the typical fluctuation
time scale but shorter than the time scale over which you maintain a
more or less constant speed - and measure. The exact averaging
period is not terribly important as long as these two conditions are
satisfied. You can validate empirically that this is the case: a)
the computer does "forget" the previous regime; b) the average does
not fluctuate much (you can also watch the momentary values on the
other screen to see how fast and how much they fluctuate).
--
Oleg Goldshmidt | p...@goldshmidt.org
_______________________________________________
Linux-il mailing list
Linux-il@cs.huji.ac.il
http://mailman.cs.huji.ac.il/mailman/listinfo/linux-il
