rec.autos.simulators

Pretty much what I was going to say.  AFAIK there's little if any
difference, on a smooth track, between a roll couple distribution
due to a certain spring combination, and the same RCD arrived at
by a combination of springs and ARBs.

What ARBs allow you to do is vary the RCD independently of the
springs, in situations where there are other factors (such as
keeping the front up during braking, or increasing compliance
to cope with bumps) which affect the preferred spring rates.

I believe that, as a generalisation, rally cars run soft springs
and heavy ARBs, using the ARBs to limit roll; whilst touring cars
run stiffer springs and fairly light ARBs, using the ARBs as fine-
-tuning devices.

<goes away to re-read GPLF...>

Hmm, can't immediately find that bit - but it doesn't make sense
to me:

If adding a front 'bar increases the front load transfer, then
the rear weight transfer must be reduced because the total
load transfer is only a function of CG height, track width and
lateral acceleration.  Even if it makes the car roll less,
that should actually *reduce* the total load transfer...

Jonny

Hi

This all sounds entirely logical (and very handy from an implementation
point of view :) )
but it does seem to differ from what Doug Milliken says so I'm not quite
sure what to think.
I've also read the bit in RCVD about this and again it seems imcompatible
with what you
have said. Am I somehow not understanding what Doug says and you do agree?

An angle, you mean it rolls off in a straight line? or in a circle?
I had kind of assumed from what I had read in RCVD that it would
be a circle but angle implies straight line?

I do like the sound of what Haqsau says here, sounds very easy to implement!

    Paul

Disagree is probably a bit too strong a word.  Doug is working with
traditional ideas derived from traditional tire testing, and within that
context he is perfectly correct.  What I am trying to say is that those
traditional models are inadequate and have led to false assumptions and
poorly implemented tire models in sims.  Unfortunately I don't have the
necessary data to propose an adequate alternative.  I'm really trying to
encourage programmers and any lurking engineers alike to think outside
the box and come up with something new.  A wheel is basically a rim with
a couple of highly damped springs under it connected to a contact patch
whose size varies with pressure.  Anything that strays far from that
concept is probably going to give you the wrong answer at some point.
What I have seen of Pacejka (sp?) seems to ignore this and instead just
try to model existing tire data.  That seems fine on the surface, except
that as I have tried to point out existing data is not formulated in a
very practical way.  Camber thrust, such as it is, could easily be
explained by the wheel, spring and contact patch model, and looking at
it that way it should also be clear what the limitations of camber
thrust are.  Sorry, I guess I am not really giving you a very clear
answer, I'm just trying to stir the pot a little, get people to throw
away the old models, and see what new things they can come up with.
Again, I would love to see somebody take on a real time FEA tire model.
I know non-real-time FEA models are being done in the industry and at
universities, and I personally believe that we now have enough computing
power to do a simplified (i.e. small number of elements) FEA model in
real time in a computer game.  Codemasters is already doing real time
FEA damage simulation in Pro Race Driver, for example.  Not being very
helpful, am I?  ;o)

Regards,
Hal

FEA stands for Finite Element Analysis, which you have undoubtedly heard
of.  I guess FEM (Finite Element Method) is the more common
abbreviation.  Anyway I was not referring to any specific model,
although I have seen papers on some models somewhere (SAE?) but I was
referring in general to the idea that the tire model ought to be a
problem that could be solved through application of FEA/FEM.  I don't
even think the whole tire has to be modeled, just a sector from slightly
in front of the contact patch to slightly behind, and then put
appropriate boundary conditions on that sector based on more traditional
models.  Plus with racing slicks you wouldn't have to model the tread,
so you could conceivably get away with fewer elements.  Just thinking
out loud though, I have never actually tried to apply FEA to the tire
model myself.