Update of the python fsi interface and new feature coupling SU2 with Nastran #1124

Nicola-Fonzi · 2020-12-05T18:46:20Z

Proposed changes

Dear all,

I am very new to collaboration on a open source project, but I have been working on an update for the FSI computation in python that I think may be useful also for other researcher. Would you mind giving me a feedback about the modifications?

In few words, I have updated the python interface for fluid structure interaction to the new version of SU2, including the new driver, the new mesh deformation methods and few other modifications in the python scripts. It works with python3.

As far as the C++ code is concerned, I only added the possibility to separate the symmetry boundary conditions for the fluid solver from the ones of the mesh solver. There is an explanation of this modification in the ReadMe file in SU2_PY/FSI_tools, together with an example of application.

The interface is still general and new structural solvers can be coupled by adding it to the fsi_computation.py file. With the present pull request I only include one structural solver, which provides a coupling with the commercial code Nastran.

I added ReadMe files to explain in detail all the modifications and also the keywords to be used in the config files. Finally, there is also a complete test case where the flutter of a flat plate is studied.

I am looking forward for your comments

Nicola

PR checklist

I am submitting my contribution to the develop branch.
My contribution generates no new compiler warnings (try with the '-Wall -Wextra -Wno-unused-parameter -Wno-empty-body' compiler flags, or simply --warnlevel=2 when using meson).
My contribution is commented and consistent with SU2 style.
I have added a test case that demonstrates my contribution, if necessary.
I have updated appropriate documentation (Tutorials, Docs Page, config_template.cpp) , if necessary.

pcarruscag

This looks very interesting especially if it makes it easier to interface other solvers.

One think I must ask is to move all mesh files to the TestCases repository and all the config files to TestCases folder.
You could also try to re-use the mesh files across your different examples, we should try to reduce the size of the repository.

Alternatively you can also create some tutorials (since you already have some readme files), we have a special repository for those, I think a tutorial would give your work more visibility.

I will leave some comments on the code once I go through everything.

SU2_CFD/include/iteration/CIteration.hpp

SU2_CFD/src/solvers/CMeshSolver.cpp

Common/include/CConfig.hpp

Nicola-Fonzi · 2020-12-07T22:06:58Z

Thank you for the feedbacks, I am working on the modifications you suggested. I will perform a couple of tests to be sure I did not break anything.

Hope to commit the new code soon!

…into publish_develop

Nicola-Fonzi · 2020-12-11T10:23:25Z

I was trying to merge the develop to resolve conflicts and I found an error in the compilation. I thought I had broken something so I tried to start fresh new with the commands:
git clone https://github.com/su2code/SU2.git
./meson.py build -Denable-pywrapper=true --prefix=/path/to/my/folder
./ninja -C build install
Even from the master branch of your repo I still find the error attached below. Is it a problem with my compiler? I could always compile with no issues before. The last one I tried from you repo was 7.0.6.

Nicola-Fonzi · 2020-12-11T10:32:25Z

I use openMPI 4.0.4 on a Linux CentOS 7

pcarruscag · 2020-12-11T10:53:01Z

What about compiler version? If you do gcc --version what do you get? (assuming it is gcc)

Nicola-Fonzi · 2020-12-11T10:58:13Z

gcc version 4.8.5

pcarruscag · 2020-12-11T11:08:15Z

I think that was the first gcc with official full c++11 support, and so it might not be full support...
Try pushing the merge anyway so that github can run the tests, I'll see if something can be done about 4.8.5.

Nicola-Fonzi · 2020-12-11T11:27:22Z

Okok, I may also try to update gcc. The only issue is that it is on a server so I should do it locally and it may take a bit more time. I will test the merge and then, with more time, update gcc.

Nicola-Fonzi · 2020-12-11T18:25:20Z

I solved for the compilation issues by upgrading gcc. Thank you for the suggestion!

pcarruscag · 2020-12-11T19:37:31Z

I'll merge this now to try to conciliate it with #902
I will probably ask you to check if I broke anything.

Nicola-Fonzi · 2020-12-11T19:46:18Z

Ok I will now try to run two main test cases I have. But there were no conflicting files after the last commit so I do not think there should be any issue.

Nicola-Fonzi · 2020-12-12T12:17:02Z

Everything seems to be fine. The test cases run smoothly.

I have been following the conversation with Rocco, I will now contact him to understand how to avoid duplication and how to best integrate our works

TobiKattmann · 2020-12-16T13:25:20Z

gcc version 4.8.5

yep I had to move to gcc 5.3.0 recently from 4.8.5 actually quite recently... which would be in line with 7.0.6 working for you. I don't recall exactly what feature that was 🤔 Beginning of October with change to 7.0.7 and the SIMD stuff

pcarruscag · 2020-12-16T13:44:57Z

Guilty as charged.

TobiKattmann

I know I am a bit late to the party. Thanks for the contribution and adding the tutorial as well 💐
It would be nice to have a testcase in tutorials.py to safeguard it against future changes. Maybe it is wise to wait for after #902 if that changes sth (but I have no clue tbh).
Usually not everything needs its own testcase but as this adds a tutorial, so I vote here in favor of testcase.

(btw I stumbled upon this today and thought I already heard that name of the first author somewhere already :D . Big fan of Mr Brunton 👍 )

TobiKattmann · 2020-12-16T13:40:50Z

SU2_CFD/src/python_wrapper_structure.cpp

+      for (unsigned long iPoint = 0; iPoint < geometry_container[ZONE_0][INST_0][iMesh]->GetnPoint(); iPoint++) {
+
+        /*--- Overwrite fictitious velocities ---*/
+        su2double Grid_Vel[3] = {0.0, 0.0, 0.0};
+
+        /*--- Set the grid velocity for this coarse node. ---*/
+        geometry_container[ZONE_0][INST_0][iMesh]->nodes->SetGridVel(iPoint, Grid_Vel);


I guess because it is automatic memory allocation with fixed values the compiler optimizes this su2double Grid_Vel[3] = {0.0, 0.0, 0.0}; away. Otherwise I would argue that Grid_vel could be created outside of at least the iPoint loop

You are right!

TobiKattmann · 2020-12-16T13:49:31Z

SU2_CFD/src/solvers/CMeshSolver.cpp

  /*--- Symmetry plane is clamped, for now. ---*/
  for (iMarker = 0; iMarker < config->GetnMarker_All(); iMarker++) {
    if ((config->GetMarker_All_Deform_Mesh(iMarker) == NO) &&
+        (config->GetMarker_All_Deform_Mesh_Sym_Plane(iMarker) == NO) &&
        (config->GetMarker_All_Moving(iMarker) == NO) &&
        (config->GetMarker_All_KindBC(iMarker) == SYMMETRY_PLANE)) {

      BC_Clamped(geometry, numerics, config, iMarker);
    }


Me: Is the symmetry boundary clamped?
Su2:

... but no

(not your issue but i found it funny that above explicitly is stated Exceptions: symmetry plane, the receive boundaries and periodic boundaries should get a different treatment)

ahaha yes I also found it quite funny when I first saw it. But it is fine now to clamp it... As the sym plane for fluid and mesh are now separated, one will explicity use the deform mesh sym plane marker if she/he wants to obtain a symmetric deformation. Otherwise, the code will clamp deformation.

That was because the FEA solver did not use to have a symmetry BC.
And then when I added it, it was better to clamp the symmetry plane for my testcases :)
But now that Nicola decoupled the physical symmetry from the deformation symmetry all is well xD

Nicola-Fonzi · 2020-12-16T14:41:29Z

The most critical part is the unsteady FSI, but that actually takes its time... You think it would be fine to compare the results after just few iterations for a testcase? We could use one of the simulations in the tutorial.

Thanks for the interest in the youtube video!

pcarruscag · 2020-12-16T14:44:13Z

All testcases run for only a few steps, not entirely to convergence, the purpose is really just to detect change.

Nicola-Fonzi · 2020-12-16T14:47:22Z

Ok, I will prepare a test case. I will take the Mach=0.1 case from the tutorial and change it slightly. Now the coupling between fluid and structure occurs after 100 steps, I will just couple them immediately.
The only issue with #902 I see at the moment is with the file fsi_computation.py. The others are in separate folders so they should not create conflicts...

Nicola Fonzi added 4 commits December 5, 2020 15:14

Version to publish

5b5cfa4

Removed unnecessary files

e3e50fd

Included ReadMe files and test cases

72ae6f7

Removed old headers in cfg files

8e0683f

pr-triage bot added the PR: unreviewed label Dec 5, 2020

Nicola-Fonzi added 2 commits December 5, 2020 20:05

Removed old lines meson

32a9e7d

Modifications as suggested by code factor

660d597

pcarruscag reviewed Dec 6, 2020

View reviewed changes

SU2_CFD/include/iteration/CIteration.hpp Outdated Show resolved Hide resolved

pcarruscag reviewed Dec 7, 2020

View reviewed changes

SU2_CFD/src/solvers/CMeshSolver.cpp Show resolved Hide resolved

pcarruscag reviewed Dec 7, 2020

View reviewed changes

Common/include/CConfig.hpp Show resolved Hide resolved

Nicola-Fonzi added 5 commits December 7, 2020 15:45

Test case moved to tutorials

426b66d

Removed duplicate code

052e306

Fix to time iter

43273d2

Fixing access to protected variables

cfae101

Fixing access to nodes

ffe424c

Nicola-Fonzi added 13 commits December 8, 2020 10:28

Small fix

68cec1e

Use of sym plane definition already in CFEMSolver

7c3253d

removed unused line

fc9f12a

introduced debug

69877b9

Try with new debug

27d5d41

Try with new debug

f151228

Fixing problem with react

49d9651

Merge branch 'publish_develop' of https://github.com/Nicola-Fonzi/SU2 …

6de6cef

…into publish_develop

Test to understand LinSysReact

16103a0

another test to understand

97efbf9

small fix

003c753

another one

83fd036

removed one

6862a8c

Removed unnecessary output and condensed velocity initialisation

f8da484

Nicola-Fonzi and others added 2 commits December 11, 2020 12:41

Merge remote-tracking branch 'upstream/develop' into publish

95f0a99

Merge branch 'develop' into publish

554e3d7

pcarruscag added the changelog:feature label Dec 11, 2020

pcarruscag approved these changes Dec 11, 2020

View reviewed changes

pr-triage bot added PR: reviewed-approved and removed PR: unreviewed labels Dec 11, 2020

pcarruscag merged commit 6691b70 into su2code:develop Dec 11, 2020

pr-triage bot added PR: merged and removed PR: reviewed-approved labels Dec 11, 2020

RoccoBombardieri mentioned this pull request Dec 11, 2020

[WIP] pyBeam integration into SU2 as a subproject #902

Closed

5 tasks

TobiKattmann reviewed Dec 16, 2020

View reviewed changes

pr-triage bot added PR: reviewed-approved and removed PR: merged labels Dec 16, 2020

Nicola-Fonzi mentioned this pull request Dec 29, 2020

Small update of the python fsi interface #1148

Merged

5 tasks

Update of the python fsi interface and new feature coupling SU2 with Nastran #1124

Update of the python fsi interface and new feature coupling SU2 with Nastran #1124

Uh oh!

Conversation

Nicola-Fonzi commented Dec 5, 2020 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Proposed changes

PR checklist

Uh oh!

pcarruscag left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Nicola-Fonzi commented Dec 7, 2020 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

Nicola-Fonzi commented Dec 11, 2020

Uh oh!

Nicola-Fonzi commented Dec 11, 2020

Uh oh!

pcarruscag commented Dec 11, 2020

Uh oh!

Nicola-Fonzi commented Dec 11, 2020

Uh oh!

pcarruscag commented Dec 11, 2020

Uh oh!

Nicola-Fonzi commented Dec 11, 2020

Uh oh!

Nicola-Fonzi commented Dec 11, 2020

Uh oh!

pcarruscag commented Dec 11, 2020

Uh oh!

Nicola-Fonzi commented Dec 11, 2020

Uh oh!

Nicola-Fonzi commented Dec 12, 2020

Uh oh!

TobiKattmann commented Dec 16, 2020

Uh oh!

pcarruscag commented Dec 16, 2020

Uh oh!

TobiKattmann left a comment

Choose a reason for hiding this comment

Uh oh!

TobiKattmann Dec 16, 2020

Choose a reason for hiding this comment

Uh oh!

Nicola-Fonzi Dec 16, 2020

Choose a reason for hiding this comment

Uh oh!

TobiKattmann Dec 16, 2020

Choose a reason for hiding this comment

Uh oh!

Nicola-Fonzi Dec 16, 2020

Choose a reason for hiding this comment

Uh oh!

pcarruscag Dec 16, 2020

Choose a reason for hiding this comment

Uh oh!

Nicola-Fonzi commented Dec 16, 2020

Uh oh!

pcarruscag commented Dec 16, 2020

Uh oh!

Nicola-Fonzi commented Dec 16, 2020

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants

Nicola-Fonzi commented Dec 5, 2020 •

edited

Loading

Nicola-Fonzi commented Dec 7, 2020 •

edited

Loading