Enable adaptive time stepping for contact and periodic rve boundary condtions - #2105
Enable adaptive time stepping for contact and periodic rve boundary condtions #2105m-frey wants to merge 3 commits into
Conversation
There was a problem hiding this comment.
This test should somehow assert that the timestep is actually reliably and predictably adapted in this simulation. Maybe by writing the reaction forces of a DBC and comparing that to a reference with a non-constant timestep?
There was a problem hiding this comment.
Good idea, I modified the testcase.
There was a problem hiding this comment.
@m-frey Why not check the time step size directly with a result test? That tests the quantity of interest directly (instead of relying on a secondary quantity such as reaction forces).
And that's how we do it for adaptive time stepping based on error estimation, as far as I know.
| } | ||
|
|
||
| /*----------------------------------------------------------------------* | ||
| | Reset contact after failed trial time step | |
There was a problem hiding this comment.
This documentation belongs into the header file to be processed by doxygen.
| SPECIAL: true | ||
| QUANTITY: "num_iter_step_15" | ||
| VALUE: 3 | ||
| TOLERANCE: 1e-20 |
There was a problem hiding this comment.
The following result description would check for the time step size directly:
| TOLERANCE: 1e-20 | |
| TOLERANCE: 1e-20 | |
| - STRUCTURE: | |
| SPECIAL: true | |
| QUANTITY: "time_step_size" | |
| VALUE: 0.123456789 | |
| TOLERANCE: 1e-14 |
Note: This is not actual 4C-ready code. It requires to expose the time step size of the final time step within the result test, which has not been done for the solid time integration, yet.
There was a problem hiding this comment.
It would be great if you could add a test description.
| SOLVER 9: | ||
| SOLVER: "UMFPACK" |
There was a problem hiding this comment.
If I am not mistaken, this is unused. I would recommend removing all unnecessary inputs. Same e.g. for the IO/MONITOR STRUCTURE DBC condition, you don't need this feature to test the time adaptivity, right?
There was a problem hiding this comment.
@lauraengelhardt I assume this is my fault: #2105 (comment)
Description and Context
Description
This PR enables adaptive time stepping for contact and the constraint framework by implementing the missing
reset_step_state()functionality.Changes
reset_step_state()for the contact model evaluator.reset_step_state()for the constraint framework.Related Issues and Pull Requests