[GeoMechanicsApplication] Add multi-stage tests for the new line interface elements #12705

avdg81 · 2024-09-30T07:38:13Z

📝 Description

Added two Python tests that cover the behavior of the new line interface element in a two-stage scenario. The first test uses Dirichlet boundary conditions, whereas the second one uses Neumann boundary conditions. Both tests carry out two time steps per stage.

Note that this PR needs to wait for another PR to be merged first, since that PR contains a fix that has been duplicated here. Once it has become available on master, we can remove it from this PR.

So far, the test comprises a single stage with a single time step. It doesn't include any initial relative displacement nor any initial traction.

The prescribed displacements are now interpolated from two tables (one for each direction). The expected values have been updated, too.

So far, this test consists of a single stage with two time steps.

To make it work I had to copy a fix from another PR. Once that one has been merged, we can revert the local copy here.

- Renamed a test case to avoid having non-unique names. - The interface normal stiffness and the interface shear stiffness are now attributes of the test case class.

WPK4FEM

Fast first look.

...anicsApplication/tests/line_interface_elements/Dirichlet_multi_stage/MaterialParameters.json

applications/GeoMechanicsApplication/tests/test_helper.py

applications/GeoMechanicsApplication/tests/test_line_interface_elements.py

applications/GeoMechanicsApplication/python_scripts/geomechanics_solver.py

…stage-tests-for-new-line-interface-elements

…ommon one As a result, the name of one model part had to be changed ("Interface" => "Interfaces").

As a result, the new helper function could be removed (it was copied from another PR).

Also removed an accidentally added blank line.

WPK4FEM

Minor typographical comments left, nothing blocking.

WPK4FEM · 2024-09-30T14:43:39Z

...pplication/tests/line_interface_elements/Dirichlet_multi_stage/ProjectParameters_stage1.json

+            "scaling":         false
+        },
+        "problem_domain_sub_model_part_list": ["Interfaces"],
+        "processes_sub_model_part_list":      ["FixedNodes","AllNodes", "DisplacedNodes"],


Either no space after comma, or always a space after a comma. Preferably not mixed as it is here.

Good catch! Fixed.

WPK4FEM · 2024-09-30T14:44:49Z

...pplication/tests/line_interface_elements/Dirichlet_multi_stage/ProjectParameters_stage2.json

+            "scaling":         false
+        },
+        "problem_domain_sub_model_part_list": ["Interfaces"],
+        "processes_sub_model_part_list":      ["FixedNodes","AllNodes", "DisplacedNodes"],


see comment about comma and space in previous projectparameters file.

Done. Thank you.

WPK4FEM · 2024-09-30T14:45:41Z

...sApplication/tests/line_interface_elements/Neumann_multi_stage/ProjectParameters_stage1.json

+            "scaling":         false
+        },
+        "problem_domain_sub_model_part_list": ["Interfaces"],
+        "processes_sub_model_part_list":      ["FixedNodes","AllNodes", "LoadedSide"],


also here comma and space.

WPK4FEM · 2024-09-30T14:46:20Z

...sApplication/tests/line_interface_elements/Neumann_multi_stage/ProjectParameters_stage2.json

+            "scaling":         false
+        },
+        "problem_domain_sub_model_part_list": ["Interfaces"],
+        "processes_sub_model_part_list":      ["FixedNodes","AllNodes", "LoadedSide"],


no space, space

rfaasse

Thanks for adding these tests (and nice you used the setup method for the test class), I have a few suggestions left for duplication/clean-up, but other than that, this is good to go!

rfaasse · 2024-09-30T15:37:40Z

applications/GeoMechanicsApplication/tests/test_line_interface_elements.py

+        self.normal_stiffness = 3.0e7
+        self.shear_stiffness = 1.5e7


Would it be possible to re-use these in the single-stage, single-step tests?

Yes, of course we can. Thanks, this makes the test code more consistent.

applications/GeoMechanicsApplication/tests/test_line_interface_elements.py

rfaasse · 2024-09-30T15:43:44Z

...pplication/tests/line_interface_elements/Dirichlet_multi_stage/ProjectParameters_stage1.json

+        "linear_solver_settings":             {
+            "solver_type":     "amgcl",
+            "smoother_type":   "ilu0",
+            "krylov_type":     "gmres",
+            "coarsening_type": "aggregation",
+            "max_iteration":   100,
+            "verbosity":       0,
+            "tolerance":       1.0e-6,
+            "scaling":         false
+        },


In the end, I used this solver (a bit less of a hassle to set up and should do the same)

Suggested change

"linear_solver_settings": {

"solver_type": "amgcl",

"smoother_type": "ilu0",

"krylov_type": "gmres",

"coarsening_type": "aggregation",

"max_iteration": 100,

"verbosity": 0,

"tolerance": 1.0e-6,

"scaling": false

},

"linear_solver_settings": {

"solver_type": "LinearSolversApplication.sparse_lu"

}

Thanks for pointing this out. I have simplified the linear solver settings accordingly in the ProjectParameters.json files that have been added in this PR.

rfaasse · 2024-09-30T15:45:17Z

...pplication/tests/line_interface_elements/Dirichlet_multi_stage/ProjectParameters_stage1.json

+        "buffer_size":                        2,
+        "echo_level":                         1,
+        "clear_storage":                      false,
+        "compute_reactions":                  true,


I think we could put this to false. I did some clean-up in the single stage tests (which is of course based of the same example), maybe you could diff and see which settings could be the same? The same holds then of course for the other jsons

I have compared the single stage analysis files with those of the multi-stage analyses. For consistency, I have changed some flags for the multi-stage analyses which had no effect on the obtained results. Thanks for pointing this out.

avdg81 · 2024-10-01T07:41:40Z

Minor typographical comments left, nothing blocking.

Thank you for your comments. I have processed them.

- Reuse attributes `normal_stiffness` and `shear_stiffness` in the single-stage tests. - Extracted a method that asserts the results of a multi-stage test. - Simplified the linear solver settings. - Made the analysis flags more consistent with those of the single stage tests.

avdg81 · 2024-10-01T08:57:02Z

Thanks for adding these tests (and nice you used the setup method for the test class), I have a few suggestions left for duplication/clean-up, but other than that, this is good to go!

Thank you very much for your thorough review. I have processed your suggestions to the best of my abilities. Could you please have another look, just to double-check that I didn't miss anything or misinterpreted a suggestion?

rfaasse

Looks good to go to me!

avdg81 added 11 commits September 27, 2024 16:20

Added a rudimentary integration test

061a4b9

So far, the test comprises a single stage with a single time step. It doesn't include any initial relative displacement nor any initial traction.

Improved indentation of json file

91ec03e

Extended the integration test to carry out two time steps

0614097

The prescribed displacements are now interpolated from two tables (one for each direction). The expected values have been updated, too.

Some preparations before adding a second stage

5818099

Added a second stage

4ac57d4

Added a new test using a Neumann boundary condition

23e5862

So far, this test consists of a single stage with two time steps.

Added the second stage

dba5d54

Both tests now use the "is restarted" flag

d8ec7a5

To make it work I had to copy a fix from another PR. Once that one has been merged, we can revert the local copy here.

Minor refactoring work

40573a3

- Renamed a test case to avoid having non-unique names. - The interface normal stiffness and the interface shear stiffness are now attributes of the test case class.

Compacted the test code by extracting a helper method

030a5ea

Compacted the code for the test with Neumann BCs

1113c89

avdg81 added the GeoMechanics Issues related to the GeoMechanicsApplication label Sep 30, 2024

avdg81 requested review from rfaasse and WPK4FEM September 30, 2024 07:38

avdg81 self-assigned this Sep 30, 2024

avdg81 linked an issue Sep 30, 2024 that may be closed by this pull request

[GeoMechanicsApplication] Add integration tests for the new line interface elements #12693

Closed

WPK4FEM reviewed Sep 30, 2024

View reviewed changes

avdg81 added 6 commits September 30, 2024 13:10

Merge remote-tracking branch 'origin/master' into geo/12693-add-multi…

61b6428

…stage-tests-for-new-line-interface-elements

Replaced two copies of MaterialParameters.json by referencing the c…

947041e

…ommon one As a result, the name of one model part had to be changed ("Interface" => "Interfaces").

Retrieving the traction vectors has been unified

dc8c2af

As a result, the new helper function could be removed (it was copied from another PR).

Retrieving the relative displacement vectors has been unified

351b919

As a result, the new helper function could be removed (it was copied from another PR).

Reuse the method that checks two vectors for equality

f6b84fe

Also removed an accidentally added blank line.

Updated the test documentation

bcc8372

avdg81 marked this pull request as ready for review September 30, 2024 12:35

avdg81 requested a review from WPK4FEM September 30, 2024 14:08

WPK4FEM previously approved these changes Sep 30, 2024

View reviewed changes

rfaasse reviewed Sep 30, 2024

View reviewed changes

Removed a few redundant spaces

ed54c04

avdg81 dismissed WPK4FEM’s stale review via a9a2944 October 1, 2024 08:53

avdg81 requested a review from rfaasse October 1, 2024 08:57

rfaasse approved these changes Oct 1, 2024

View reviewed changes

avdg81 enabled auto-merge (squash) October 1, 2024 09:35

avdg81 merged commit ac2b1a3 into master Oct 1, 2024
9 of 11 checks passed

avdg81 deleted the geo/12693-add-multistage-tests-for-new-line-interface-elements branch October 1, 2024 10:34

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

[GeoMechanicsApplication] Add multi-stage tests for the new line interface elements #12705

[GeoMechanicsApplication] Add multi-stage tests for the new line interface elements #12705

avdg81 commented Sep 30, 2024

WPK4FEM left a comment

WPK4FEM left a comment

WPK4FEM Sep 30, 2024

avdg81 Oct 1, 2024

WPK4FEM Sep 30, 2024

avdg81 Oct 1, 2024

WPK4FEM Sep 30, 2024

avdg81 Oct 1, 2024

WPK4FEM Sep 30, 2024

avdg81 Oct 1, 2024

rfaasse left a comment

rfaasse Sep 30, 2024

avdg81 Oct 1, 2024

rfaasse Sep 30, 2024

avdg81 Oct 1, 2024

rfaasse Sep 30, 2024

avdg81 Oct 1, 2024

avdg81 commented Oct 1, 2024

avdg81 commented Oct 1, 2024

rfaasse left a comment

[GeoMechanicsApplication] Add multi-stage tests for the new line interface elements #12705

[GeoMechanicsApplication] Add multi-stage tests for the new line interface elements #12705

Conversation

avdg81 commented Sep 30, 2024

WPK4FEM left a comment

Choose a reason for hiding this comment

WPK4FEM left a comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

rfaasse left a comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

avdg81 commented Oct 1, 2024

avdg81 commented Oct 1, 2024

rfaasse left a comment

Choose a reason for hiding this comment