rec.autos.simulators

:   I don't understand this attitude at all.  The aero model would be no
: more "fake" than the suspension model.  The rest of the model assumes
: bulk properties - why would it be any different to have an aero model
: assuming bulk properties.  If you are willing to say "suspension element
: X" is a bar with mass m, CofG at x and springyness y.  Why is it more
: of a con to say "drag x, downforce y, at speed z?"  All sim physics is
: "canned" (to use a ludicrous phrase which comes up a lot in this group)
: - people should get over this block and simply ask "is the results of
: what has been canned physically realistic as far as I can tell".  In
: GPL I'd say it's pretty damn good :-)

Exactly.  Most of the top industrial packages for mechanical simulation,
such as ADAMS, could be considered "canned" as well. These babies costs
thousands of dollars, and the results are good enough for their purpose.
It will be along time before we see CFD (computational fluid dynamics) in
real time on workstations, let alone trickling down to a game, and even
then CFD is only ever an approximation of what the aero situation really
is.

Stephen

[snip]

Maybe because "drag x, downforce y, at speed z" is as accurate in
aerodynamics as "max sideforce x, drag y, at speed z" would be
in chassis dyamics?

      _
Mats Lofkvist

: [snip]
:>   I don't understand this attitude at all.  The aero model would be no
:> more "fake" than the suspension model.  The rest of the model assumes
:> bulk properties - why would it be any different to have an aero model
:> assuming bulk properties.  If you are willing to say "suspension element
:> X" is a bar with mass m, CofG at x and springyness y.  Why is it more
:> of a con to say "drag x, downforce y, at speed z?"

: Maybe because "drag x, downforce y, at speed z" is as accurate in
: aerodynamics as "max sideforce x, drag y, at speed z" would be
: in chassis dyamics?

: Mats Lofkvist

  I think I've totally failed to get your point here.  Both the chassis
dynamics and the aerodynamics are both, by necessity inaccurate - you
seem to be suggesting that there is some reason to believe that the
approximations used for chassis dynamics are, for some reason, better.
Have I misunderstood?  We can model neither thoroughly - a model maker
must assume bulk properties.  You seem to be implying that this is
somehow more accurate for chassis dynamics?

--
Richard G. Clegg     Only the mind is waving
Dept. of Mathematics (Network Control group) Uni. of York.

www: http://manor.york.ac.uk/top.html

I think when we talk about "canned" in the sim context it refers to
actions in the sim that are too "modal". For example, the slip angle
exceeds a certain value and the physics engine switches into a different
mode to model sliding friction. This switch into "spin" mode is abrupt
and unrealistic and makes retrieval impossible.

It may be that the GPL physics engine is modal in some respects, but if
it is it hides it superbly well.

- Michael

: : Maybe because "drag x, downforce y, at speed z" is as accurate in
: : aerodynamics as "max sideforce x, drag y, at speed z" would be
: : in chassis dyamics?

: : Mats Lofkvist

:   I think I've totally failed to get your point here.  Both the chassis
: dynamics and the aerodynamics are both, by necessity inaccurate - you
: seem to be suggesting that there is some reason to believe that the
: approximations used for chassis dynamics are, for some reason, better.
: Have I misunderstood?  We can model neither thoroughly - a model maker
: must assume bulk properties.  You seem to be implying that this is
: somehow more accurate for chassis dynamics?

Mats was, I guess, referring to the older "car on a stick" physics model,
where everything amounts to going to a big lookup table to figure out what
to do for a given condition.  But I think he draws an unfair parallel
between that over-simplification of overall car dynamics modelling and the
suitability of a simplified wing model laid on top of the existing GPL
chassis model.  I think that adding wing components to the model, and
coupling their downforce (lift) and drag reactions to the appropriate
chassis pick-up points would be adequate, and indeed sophisticated enough
to match the rest of the GPL model.  Since we're never going to see
real-time CFD in a game, throw in a good table of empirical data for angle
of attack and yaw versus lift and drag, and then just transmit those
forces down through the rigis links to the chassis.  Good luck modelling
the more tricky stuff, though, like car/car interactions on the properties
of the wing.

personally, I'd be quite happy to see such a "crude" wing model grafted
onto the GPL engine.  most of the off-the-shelf engineering simulation
modules are as crude, and get the job done.

Stephen

By the way, the original Flight Unlimited claimed to use a CFD approach to
their aerodynamics model.  I wonder how sophisticated it was?  After all,
it ran on a 486.

Stephen

Of course all the models are by definition simplifications of reality,
it's the proper mix of models that makes a good simulation.

My point was just that I don't think that a _simple_ aerodynamics model
added to GPL is going to simulate cars with wings at the same level as
the current GPL simulates wingless cars, hence I see some merit in the
original posters comment that decent wing modelling would add weight to
an already heavy model.

Still, I'd love to see more aerodynamics added to GPL even without wings.
E.g. ride height dependent drag. And sections like Quiddelbacher H?he
and Pflanzgarten at Nrburgring might get a new dimension if the car
starts to catch air if you are not careful :-)

      _
Mats Lofkvist