[jnagy]

Comments and questions are welcomed on the article Coe et al., IKA-FLOW : A Flexible Body Overset Mesh Implementation for Fish Swimming, OpenFOAM Journal, https://doi.org/10.51560/ofj.v3.89.

[mco]

Please let me know if you have any questions or comments, or need any clarification about the code.

[rkandukuri]

in figures.31-34 , does the background mesh only part refer to a body fitted mesh or the overset assembly with the component meshes blanked out ? If it's the latter, whats the rationale behind including that ? I'm sorry If I missed that bit in the article ?

[mco]

Hi rkandukuri,

The background mesh only refers to the background mesh with the overset component mesh not rendered (using a threshold). This is to show the hole and the corresponding flow field for just the background mesh.

The rationale is that we found it interesting comparing the flow fields of the two and thought a reader might also be interested.

A comparison with a body fitted mesh would be good as well, but it was quite difficult to get the diffusion functions to work with the inflation layer requirements we had (35 layers). Without going into too much detail, we were looking at how fish propulsion scales from Reynolds 1e3 to 1e7, which is why the high number of inflation layers was needed. This is what prompted the development of the overset meshing route.

Please let me know if you have any other questions.

Michael

[Tianyz]

Originally Posted by mco

Please let me know if you have any questions or comments, or need any clarification about the code.

Hi Michael,

Your work is amazing, I'm using overset mesh based on OpenFOAM. But it's only for a rigid body, your contribution fills this empty. I still reading your code and want to talk with you further. Since your solver is focusing on prescribed motion (one way) currently, I'm wondering if you plan to extend it to two-way coupling in the future?

Best,
Tian

[mco]

Hi Tianyz,

Thank you for the comment. I did have a plan to implement the forces class inside the solver to get the average forces and then change the amplitude variable to balance the thrust and drag. This is outlined in the great work done by Audrey Maertens in Maertens 2017.

Currently, I'm working on another project developing heat transfer solvers. If you would like to implement something as described above, send me a private message.

Hope that helps,

Michael

[sunweiyi2020]

Hi jnagy,

About the overset mesh, I have one question. Can it be used to deal with the deformation of elastic structure connected with the bottom boundary? I try to use overset mesh on this situation. But the simulation diverged quickly.

[mco]

Hi sunweiyi2020,


With an overset mesh, you need to make sure there are enough cells between the hole, overset mesh, and the boundary of your background mesh. I believe a minimum of 3 is the required. So you would want at least 3 overset mesh cells between your hole and any boundary in you background mesh.


I'm not sure what your case is exactly, but it sounds like the hole mesh would be right at the boundary. This is not going to really work. If you can give a screenshot of your mesh, it would be easier to give a more exact answer.

[sunweiyi2020]

Hi mco,

This is my mesh. Please check it. The arc is the structure, a thin membrane. It is connected with bottom boundary. I try it with overset mesh but the result is so bad. [IMG]file:///C:/Users/Coastal%20Eng%20Lab/Desktop/mesh.png[/IMG]
If this link is not available, please get the mesh figure in this link https://github.com/NUGUY2022/mesh.git.

[mco]

In this case, it might be easier to just use a regular morphing mesh. If you look under the src/fvMotionSolver/pointPatchFields/derived folder in OpenFOAM, you will see motion solvers for patches. You can convert one of these to suite your needs.


Another option would be to code it all in the dynamicMeshDict using the new coding feature. An example of this is in the /tutorials/mesh/moveDynamicMesh/twistingColumn folder.


Are you calculating forces on the patch and moving it as an FSI simulation or are you prescribing the motion?

[sunweiyi2020]

Sorry for late, I just check the forum.
Yes, you are right. I am calculating forces on the patch and moving it as an FSI simulation. But large mesh deformation occurred on my case, resulting in divergency. I am not familiar with these motion solvers in src/fvMotionSolver/pointPatchFields/derived. Could you give me any advice for selecting the most suitable solvers? Regarding others' suggestion, they recommend RBFMeshMotionSolver to me. But it seems that wave2foam doesn't support it. Do you have experience to integrate it into interFoam or wave2foam?

[mco]

Sorry, I'm not familiar with those two solvers. No reason not to try them out. You can play around with the diffusion equations in the dynamic mesh dict and lowering your time step.



What is happening with the mesh when it diverges? Is it a mesh problem or a numerics problem? If the forces are diverging, where on the patch is this happening?


Michael

[sunweiyi2020]

I have uploaded the crashed mesh on the link https://github.com/NUGUY2022/mesh.git.
It is obvious that the mesh diverges rather than a numerical problem, because the case with small deformation can run without any problems. Now, I use the displaceLaplacian solver for mesh motion. I will try different mesh motion solvers for this case. In addition, do you think it can be solved by optimizing mesh?

[Tianyz]

Originally Posted by mco

Please let me know if you have any questions or comments, or need any clarification about the code.

Hi, Coe

Sorry to reply to you late. I was tightening some other things. And I got one question for you. Form your paper, you said that "Besides fish motion, this library can be used to deform an overset mesh to accommodate any motion desired", what if I want to simulate a FSI case with a deformable structure with other motionsolver, say displacementLaplacian or 6DOF Motion Solver? should I recompile it? Thank you for any hints.

Best,
Tian

[Forrest Abbott]

Another approach is to use mesh adaptation techniques, such as mesh refinement or coarsening, to dynamically adjust the resolution of the mesh near the deforming boundaries.