rec.autos.simulators

Sidb19.dof is a 5kb file... but yet if I tyr and open it in -  say
notepad...  it will make my system slooooooooooooowwwwwwwwwwwww  down...

It even take ages to virus scan some if these files....

Me thinks something is wrong...

ed_

Something if clearly wrong here... it took me 10mins to install version
0391! (i.e the damn computer used 8+ minutes to install a single 5kb
file!! -  and 2 mins on the rest of them..

Just loading racer takes ages too...  - 5-10 mins to be exact.

What gives?

ed_

OH! Just incase... this on a Win98SE, Celly 700, 128mb Pc-100 Ram, V4500,
DX7.0a system...

...

I'd say a lot more even, since with the relatively small time steps,
the chances that your triangle will hit are huge. I'd say the impact
of 4-600 triangle would be more in the hundreds. :)
Now physics will be even faster.

I'll have a look. First I'll have to view the Australian GP. Just got
back from work, and gosh, I need some time off, to explore some new
ideas for the suspension which is bugging me.

Sounds good. I hope though somehow getting Pacejka data will be more
easy in the future, since seemingly more people are using it. WSC is,
for example. It's just the numbers to plug in are hard to find (which
you know ofcourse, since you're trying to get at Pacejka numbers in
another way).
Sounds like a nice model, although very intensive. Good luck with
anyway!

Ruud van Gaal, GPL Rank +53.25
Pencil art    : http://www.marketgraph.nl/gallery/
Car simulation: http://www.marketgraph.nl/gallery/racer/

Must be the loading of the same file. For me, on PII400, 256Mb
TNT2U32Mb, loading Racer takes about 10 seconds or so. Installing it
takes even less.

Have you tried to install it on another partition or harddisk? (or
even OS?)

Ruud van Gaal, GPL Rank +53.25
Pencil art    : http://www.marketgraph.nl/gallery/
Car simulation: http://www.marketgraph.nl/gallery/racer/

  Yes, you're probably right.  Maybe real-time brush tire modelling IS possible
:0)!!  I'll have to play with that one.

  Sounds good.  Are you going to try the A-arm kinematics?  Ask Chris about it,
he'll probably help.

  Thanks :0)
Todd Wasson
---
Performance Simulations
Drag Racing and Top Speed Prediction
Software
http://PerformanceSimulations.Com

...

Perhaps if you can reduce the patch subdivision in real time, based on
how much things vary across the tire, you can have a good improvement.
Other than that, I'm going the Pacejka route.

He's given me a little hint about suspensions. About energy
conservation. I'm going to try that real soon. I'd say a spring with
wheel movement should contain it's energy level. Although I should not
take into that the movement of the wheel in world space; just relative
to the suspension. The funny thing though is where to leave the energy
burned as heat in the damper. That would seem to make the energy level
go down.
Which also means that when hitting a bump, the energy level in the
spring/wheel combination would go up, or not hmm. I'll go investigate
this with some new set of view elements which can visually display
some parameters 'live'. Will be cool.

Ruud van Gaal, GPL Rank +53.25
Pencil art    : http://www.marketgraph.nl/gallery/
Car simulation: http://www.marketgraph.nl/gallery/racer/

  Interesting.  I've got no idea where or why energy should be calculated.

Todd Wasson
---
Performance Simulations
Drag Racing and Top Speed Prediction
Software
http://PerformanceSimulations.Com

It seems at least one of the reasons why WSC can run so low as 100Hz.
Normally, you'll have trouble with the high frequencies that tire
rates (and F1 spring rates as well) generate.
In other words, your Euler integration steps create errors that are
too big. Somehow, by calculating the energy in the system, you can
restrict the steps/errors so the energy involved in the wheel moving
(relative to the suspension, I guess) and the stored energy in the
spring itself, doesn't grow so that energy seems to be put in out of
nothing.
This limiting of energy growth seems a nice thing to look into. If it
can prevent 'over-bouncing' the wheels, it would make things just a
bit smoother (and at a smaller integration step). I believe this will
be necessary for it to get as low as 100Hz.
But the thing I don't get is this; dampers reduce the energy in the
wheel/spring combination, because it disappates (sp?) energy into heat
(into its oil). So energy *can* disappear. Now when you hit a bump and
things start moving, energy *must* be put in somewhere.
Not quite figured out how this should work yet.

Ruud van Gaal, GPL Rank +53.25
Pencil art    : http://www.marketgraph.nl/gallery/
Car simulation: http://www.marketgraph.nl/gallery/racer/

Take the dampers away, and ideally the whole car-wheels-springs system
should conserve energy when not in contact with the ground. The damper
are indeed a source of disippation, and the energy should decrease,
that's what they're there for :).

I don't think conservation of energy per se can be easily included into
numerical calculations; however, checking for unexpected too large
growths in energy might help with some instabilities that occur in
numerical integration, as this is usually a clear intication that the
timestep is too large.

If you are really striving to keep the timestep over or at about 1/100
s, I think it's hard if you want to treat wheels as an inertial object
with regard to suspension travel. A timestep of 1/100s allows the
frequencies of the order of 100Hz to be modelled at best regardless of
the numerical method used. Based on an estimate from the Millikens'
book, this coincides about the characteristic time that is present for
the wheels of a Champ Car taking into account the tire spring rate.
While this indicates that the calculations may work while the car is
firmly on the ground, any jumps and subsequent contacts will most likely
create instabilities.

-Gregor

  My biggest problem was in the rotational stability of the tires, so I run
that part on its on high frequency loop inside the main loop.  I suppose the
suspension and all the other parts could be done the same way, with their own
frequencies.  Still, I'm sure Chris' methods are quite a bit more advanced and
efficient than mine.  I'm not much of a mathematician :0)  

  Interesting.  I wouldn't have thought of that in a million years.  Let me
know how it goes, I have no idea how this would work.

  So you'd basically do some sort of energy check and limit wheel movement
based on this?  I'm completely lost!  lol

Todd Wasson
---
Performance Simulations
Drag Racing and Top Speed Prediction
Software
http://PerformanceSimulations.Com

  With a suspension frequency of 100Hz, I'd say he'd need a lot finer
resolution than 1/100s time step.  I tried the Champ Car data from Millikens'
book in my model.  With 300Hz sampling rate, it worked fine as long as the
wheels stayed on the ground and didn't change surface slopes too abruptly.
Dropping the car from a meter or so caused some pretty serious bouncing, as you
predicted.  It took fairly heavy damping to control.

Todd Wasson
---
Performance Simulations
Drag Racing and Top Speed Prediction
Software
http://PerformanceSimulations.Com

...

Ah, interesting. Ofcourse, that can be done, not even too difficult.

Yep, well, in the end a lot of it comes from just growing the physics
model, I've noticed. You can add more and more stuff, and get a lot of
functionality in the beginning, but new parts will enhance the
feeling. It's great.
Just did Pacejka modeling yesterday night btw. Phew, it worked after
just 3 hours! :)) It's just strange to use it as a braking curve as
well. That doesn't seem right.

...

It would involve detecting strange changes in energy. Note that a lot
of apps do this 'type' of programming. Excel for example was/is
debugged using multiple different algorithms to do the same thing. 1
is the smart way, the other (non-release) is the slow but robust way.
If the 2 algo's disagree, you can print a message.
The rigid body classes from Chaney also has an Energy() function. It's
meant as a test; the total energy of the car while flipping and flying
in the air should remain constant. If not, you know there's some funny
business. It's hooking in another approach which should match.

...

Yep, limit velocity rather than movement. So 1/2mv^2 + potential
energy (mg) (relative to the suspension) remains equal.

Ah well, it's running ok now at 250Hz, but there's a big problem
ahead. Now that I have Pacejka running I have aligning moment and can
have some fun trying force feedback, the biggest problem is the
geometry of the track. I have a small (Carrera) track which I want to
release, but the polygon subdivision is too small, so you get real
bumps as opposed to smoothly varying gradients. The same is true for
the Oval, although it is a bit better there.
So I'll have to do splines to define the road surface, and create a
patch surface from the road. Thinking about Hermite splines to start
with, but will have to dive into patch/ray intersections soon. WSC
does multiple spline types, ofcourse. ;-)
Wonder how it does bumps, like the famous bump in Monaco. That's
currently the biggest thing I see in finally getting acceptable
laptimes on the tracks. But it's a big challenge, all those splines. A
lot of work for the track editor.

Ruud van Gaal, GPL Rank +53.25
Pencil art    : http://www.marketgraph.nl/gallery/
Car simulation: http://www.marketgraph.nl/gallery/racer/

Thanks!

-----= Posted via Newsfeeds.Com, Uncensored Usenet News =-----
http://www.newsfeeds.com - The #1 Newsgroup Service in the World!
-----==  Over 80,000 Newsgroups - 16 Different Servers! =-----