Re: [Flightgear-devel] slip angle

[John D. <sf...@av...>]

On 11/22/2007 12:59 PM, Jon S. Berndt wrote:
> .... What happens if
> the combined moment on the vehicle and the translational force causes only
> one gear to go past its ability to hold traction on the ground? 

Well, that's pretty far down on the priority list, but since
you asked: 

That happens sometimes.  It happens more often than I would
like, but it happens.  For example, suppose the student pilot
applies the brakes too much and/or unevenly.  The result is 
that the aircraft turns like crazy.  This is a good recipe 
for a ground-loop.  This is particularly nasty because the
instructor's options are limited.  The instructor cannot
un-apply the brakes.  (This stands in unpleasant contrast
to, say, the student pulling on the yoke;  the instructor
at least has a fighting chance of pushing on the yoke so
as to overpower the student.)

Much higher on the priority list are
 -- parking with the parking brake set and not slipping
  under ordinary non-extreme conditions.
 -- Weathervaning (yaw and roll) during taxi and during
  the initial takeoff run.

> When you
> start programming this to take care of the general case, it just gets out of
> hand.

Maybe it would help to talk to the CDM guys instead of the
FDM guys.
  http://www.google.com/search?q=car-dynamics-model
  http://www.google.com/search?q=car-dynamics-model+nascar

I reckon the car-dynamics guys have a pretty good model of
static friction and quasi-static rolling friction.  After 
all, that is their raison d'être.  Whatever they've got
is surely more than good enough for our purposes.

> it did occur to me that it should be possible to use one set
> of data to do gear modeling, and another set to report values (such as
> sideslip) for instrumentation.

That's a very reasonable compromise for the interim.  The 
airspeed vector (i.e. relative wind vector) exposed via 
the public interface (i.e. the property tree) should be 
the honest-to-goodness airspeed, including nature's wind, 
even while parked.  If the FDM finds it convenient to 
use some notion of "pseudo-wind" as part of a scheme to 
avoid modeling the ground/tire interaction, then that 
should be kept strictly separate ... and maybe kept 
private, internal to the FDM.