non repeatable or unrealistic behaviour

[simo-11]

Setup

Beam with

• density 100 kg/m3 to simulate thin walled, issues can be seen also with normal density
• horizontal
• Hz=0, if hz >0, simulation works much better
• rigid supports at (-3,37) and (-1,37)
• mass of 1000000 kg

actions

• Push mass over or relocate support
  • Joint is broken or not broken
• move supports to x=20 and 22
  • Joint is broken or not broken

Add pictures for problematic behaviour

B1 option 2 is quick way to bring this issue up

[simo-11]

Large angular error dominates

[simo-11]

Constraint error looks larger than reported.

[simo-11]

Block bounces from body connected by rigid-plastic joint although restitution is default zero
WeldJoint behaves much better

Bounce was fixed in fix for #6

[simo-11]

Special option set B1 was created to reproduce this.
Model works quite well for cases where mass is dropped away from fixed end due to fixes for #6.
If load is near fixed end velocity difference in y direction gets non-zero although full capacity is not used.

Refactored to use bomb. Can be reproduced with zero speed.

[simo-11]

cdoty gets too high value, should be almost zero

[simo-11]

With fifty velocity iterations solution gives good results.
Try to find way get faster convergence.
Check impulseInitializer.