I while back, I observed that a small, light, low-power car would perform
varyingly well against a large, heavy, powerful car, depending on the
layout of the course, and that perhaps some metric needed to be developed
that would allow courses to be indexed.
At about the same time, Jean proposed the 'Evolution Theory of Car
Classing' in which representatives of current cars would be brought to a
test site, Evo school drivers would drive them, and then classing decisions
made based on the results.
The first idea has tremendous merit, but is very difficult to determine
(although Roger had a neat idea about collecting average speed statistics,
which I like because it's so simple - and I'm going to write up something
on this after Nationals)
The second idea is very easy to pull off (the main problem being one of
logistics), but - sorry Jean - is lacking in merit due to the number of
variables involved. (Not to mention perception issues)
But then I started really thinking... let's say we take Jean's idea, and
start eliminating variables. Let's take a selection of cars, and a
selection of drivers, and run them through a series of discrete elements -
0-60, 60-0, 100m skidpad, slaloms of varying lengths and gaps, and so on.
When done, you'd have a sort of data "toolkit" that could be used to string
together an approximation of a given course. For example, the prototypical
ProSolo course: Straight, into slalom, into 180 turnaround, into slalom,
into lane change, into finish. So 0-60 gives you time down the straight,
the appropriate slalom data gives you the slalom entry speed, so working
backwards from the entry point using the 60-0 data gives you braking point
for the slalom. Same deal for the turnaround, etc - skidpad tells max
lateral G, which in turn gives max entrance speed, which can be determined
how is to be reached with 0-60 and 60-0 data. String it all together, and
you get a theoretical "best lap".
Now that "best lap" may or may not actually _be_ the "best lap" possible
("after all" quoth Concord "It's only a model") but it should be consistant
from car to car - the model may break down in places, but it would break
down the same for all cars, and as long as the failings don't favour a
given kind of car, all the relative performances should be the same. Which
might, maybe, make this a useful technique for classing cars. Maybe.
Say, here's another thorny thought. I have 2 cars: they are the same
dimentionally (length, width, height, wheelbase etc) They have
infinite-grip tires. They have the same power/weight ratios, and the weight
distribution is the same. One weighs 1000lbs more than the other. Which car
is faster, or are they tied?
DG
|