(09-06-2012, 07:56 AM)NaN Wrote: Do you have access to SWIFT model, data? The only thing I know about SWIFT is, that it is an extended Pacejka to model dynamic effects(rigid ring model?). You will additionally need vertical stiffness, damping?
It is still a semi-empirical model. We can use it to "build" the tire (physically) and add to it Pacejka based slip calculations. What is more, it offers a 6 DOF, just like the new physical model used by isiMotor2.5 (which seems to be slightly more advanced than SWIFT).
We can model stiffness and damping for forces in the major axis: longitudinally, laterally and vertically. As it is basically a ring in contact with the rim, we can ignore flexible nodes of the steel and wedge belts (not optimum, indeed) and still deal with flexible sidewalls.
As for data...I may be able to...er...procure it. We'll see.
The beauty of it is that you can deal with (partly) with physical model of the tire and add MF to calculate slip. I am not too sure though if vDrift will easily support it. I am very much in the beginning of going through portions of the code.
Quote:As to the Pacejka version, 96 maybe? It is using 15 longitudinal, 11 lateral, 18 aligning coefficients.
Ok. Ruud and Eugene have also used 96 at some point - probably still use (though both are using Veble's methods).
Quote:Personally I'd love to experiment with a more simple model if I had some time to spare:
Coulomb and Stribeck with friction transition region(linear?) at origin. And then parametrize the coefficients(maybe tire load as parameter) and try to fit the tire friction curve from Pacejka.
I dunno. How do you deal with the discontinuity when v=0? You'd have to apply a different function (continuous) and decide at what velocity ranges it kicks in.
As I said, I find physical approaches interesting but with a very negative side effect: finding the data (what's worse, the type of data tire companies or racing teams are not so keen on sharing).
So, I'm thinking a semi-empirical might be a good alternative. SWIFT looks good, I don't know however how efficient it will end up being.
Problem is data - if I can procure it, fine, otherwise there'll have to be good old estimations.
Quote:Anyway, I'd suggest you to create a fork on github.
Ok.
Quote:
Btw there has been some talk to add tire slip and load damping/filtering, which could maybe count as a crude dynamic model, but mainly to reduce oscillations.
You mean add to the current model? Has anyone tried it before?
--
And btw, as I am by no means an expert, I will learn a lot from the code you wrote and your threads.
