Re: [Jsbsim-devel] Modelling a bouncing ball

[Ron J. <wi...@je...>]

On Friday 08 June 2012 16:31:45 Stephen Dwyer wrote:
> Hello,
>
> In the past, I have had issues with ground interactions in JSBSim, and
> recently have been trying to figure out why.

(...) 

> Is there a way to address this issue? What exactly is going on? Any
> insight would be appreciated.
>
> Thanks,
> -Stephen Dwyer

By default, JSBSim runs at 120Hz. This is pretty low for the way we calculate 
ground interactions. If you set it much higher ( --simulation-rate=10000 ) 
you'll see more accurate results. The down side is it will eat way too much 
CPU time to be practical.

Once the c172x looked like this:
http://www.jentronics.com/fgfs/temp/jsbsim/ground_reactions-stock-120Hz.png
(Graph @120 Hz showing a rebound at ~300 ft/sec)

raising the simulation rate yielded this:
http://www.jentronics.com/fgfs/temp/jsbsim/ground_reactions-stock-10kHz.png
(Graph @10 KHz showing no rebound)

After some tuning I got the 120Hz and 10KHz to match more closely and the 120 
Hz to behave better:

http://www.jentronics.com/fgfs/temp/jsbsim/ground_reactions-mod-10kHz.png
(Graph @120 Hz showing a mild bounce off the ground settling to rest)

http://www.jentronics.com/fgfs/temp/jsbsim/ground_reactions-mod-120Hz.png
(Graph @10 KHz showing a mild rebound settling to rest)

This is the patch that made the change: 
http://www.jentronics.com/fgfs/temp/jsbsim/0001-Enhance-ground-reactions-of-the-c172x.patch

The hints I can give:
- Contact type BOGEY gets out of control if it intersects the ground at a
  steep angles. Use STRUCTURE contacts around the BOGEY to limit the ground
  contact angle under 45 degrees.

- Use as soft a spring coefficient as possible. Harder springs require a
  higher simulation rate to be stable.

http://wiki.flightgear.org/JSBSim_GroundReactions

Good luck,

Ron