Opened 7 years ago
Closed 6 years ago
#4494 closed defect (fixed)
Incorrect/inconsistent nonlinear solution during simulation
| Reported by: | Owned by: | somebody | |
|---|---|---|---|
| Priority: | critical | Milestone: | Future |
| Component: | Run-time | Version: | v1.12.0 |
| Keywords: | Cc: |
Description
See the attached model in Loop.mo and the attached figure showing the problem.
This model is a four-bar linkage using a RevoluteConstraint joint to break the kinematic loop. At around 3 seconds (in the figure), the position of one of the hinges makes a large jump for no good reason. There is also some chattering between two different solutions.
I suspect there is some issue with the nonlinear solver since it is alternating between the correct and a different (kinematically admissible, but kinetically inconsistent) solution. Looking at LOG_NLS, the values of the iteration variables jump unexpectedly soon after 3 seconds. With LOG_NLS_V, there is a point at which the "System extrapolation" values are dramatically different for no reason, which changes the x0 values and the subsequent solution.
The locations of the jumps change with different solver tolerances. It is easy to see the effect if you simulate with 3D visualization.
Attachments (2)
Change History (3)
by , 7 years ago
comment:1 by , 6 years ago
| Resolution: | → fixed |
|---|---|
| Status: | new → closed |
This seems to be working now with OpenModelica v1.14.0-dev-26497-gc8b2e0e0e6
No idea when it was actually fixed.
Probelmatic model