Hi,
On Montag, 26. April 2004 00:13, Arnt Karlsen wrote:
> ..ok, time for another of my stupid questions: Mathias, does your
> new model code include tire creep? And tire sidewall flex?
No.
At the moment the gear model I use is mostly unchanged.
One problem here is the stiffness of the tire model. This is not a special
characteristic of our tire model, I would expect that *every* tire model
which models what it is called like, is kind of stiff. Stiff in the sense of
differential equations which is what we definitly have here. The effect of a
stiff ODE (=ordinary differential equation) is that you see some overshooting
in the state values. This is what you see when you arrest the brakes and
watch the aircraft jiggling. This effect of numerical timestepping couples
then back to the tire physics. The short time loss of enough weight on the
wheel while the aircraft is jiggling, reduces the friction of the wheel for a
short time. When outer forces are applied (wind, thrust) for example your
c172 begins to twist.
What we attack at the moment is this part of the problem.
And this part is sufficient to eliminate the jitter in the gear model and to
prevent the aircraft from twisting with brakes appiled.
The other part is the tire model itself. What I have in mind here is this
Pacejka model. This is a simple parameter fitting generated formula which is,
according to the references I found, often used in professional (car)
simulations.
This model is also something velocity dependent and, like Andy already told,
it is harder to produce zero velocity with such a model. Even what I
described above slightly moves because of this characteristic, but the
movement is extremly small (about 1e-5ft/s).
I expect that this tire model contains some kind of sidewall flex. *BUT* I
think this is only included in the sense that such effects might show up
somewhere and the parameters fitted contain them in some way.
Anyway I expect that this model is much closer to a real tire than what we
have now.
An extension to that model to make it also position dependent is something
which makes the problem only stiffer and we are back at the beginning...
What we do at the moment is the first part of the problem. This works as
expected on my development branch. But it is a long way until this ends in an
enduser distribution ...
One thing at one time but time will come ...
> ..a wee virtual demo on tire side wall flex: Take a parked, say Cessna
> 172, out at the leading edge of the very wing tip, and push it straight
> aft, then release, and repeat.
>
> ..once you get your repeat pushes close to the system resonance
> frequency, you will find wee pushes generates quite a yaw oscillation,
> and that it will swing around a point somewhere near the nose wheel.
> System here, is, tire side wall flexibility (some people prefer doing
> calculus on its inverse, "tire sidewall stiffness"), against _parts_ of
> the wings + tail + nose etc masses.
What you describe here will most likely not happen with someting only velocity
dependent.
Keep it in your head and try to move your aircraft around its nose with that
method when we have a better tire model. It you don't get it turned, we can
look if this is doable :-)
> ..a wee virtual demo on tire creep: park an automobile sideways on a
> slope, so it leans with one side down, buth with the front (and rear)
> horizontal. Mark the position of at least the up side tires. Lock the
> steering wheel, say by turning off the ignition and taking the key
> out of the lock. Now get your fat ass outta the car and push it. ;-)
:)
> ..15 feet forward, then back to those marks, will do fine.
> Note how far down the tires crept. Tire creep. ;-)
This is what I expect to show up with the Pacejka model. Let's see when this
is done ...
Greetings
Mathias
--
Mathias Fröhlich, email: [EMAIL PROTECTED]
_______________________________________________
Flightgear-devel mailing list
[EMAIL PROTECTED]
http://mail.flightgear.org/mailman/listinfo/flightgear-devel
