Learning Forum

[marcinstopyra]

Hey everybody!

I try to build a FSI simulation of a disc valve, at the moment I am figuring out the dynamic mesh options on a 2D simplified object and I faced a problem with my mesh deformation.

[Karthik Remella]

Hello,nYou could change to an unstructured 2D tri mesh and you can try remeshing option here. This might help stabilize the simulation.nThank you.nKarthikn

[danbence]

Please don’t take my comments as gospel, but I believe I can shed light on your problem.n nDeforming meshes get really tricky when you have moving walls very close to each other. The source of this difficulty is inflation layers. You want inflation layers to be left untouched by mesh deformations. By their nature, inflation layers are very slender. If they get compressed, you will quickly get negative cell volumes. If they get stretched, you wont get a good capture of the boundary layer (can be seen in your images). The diffusion method of smoothing with a coefficient >0 will try not to move cells near the walls (to preserve inflation)n nYou are trying to stretch the space between two walls in which their inflation layers are touching. In your image you can see fluent pinning nodes near to the wall and then stretching the space between the wall and your beam. This creates skewed cells. Switching on remeshing might help, but it is a still a "big ask" for fluent.n nWhat might work is to split your design into two meshes with a non-conformal interface between them. Put your fixed wall and inflation layers into a small mesh zone. This zone will be static (no deformation). You then interface it to the rest of the mesh. If you switch on remeshing, when the beam is pulled away from the wall, new cells will be generated in the gap you are creating and leave the wall inflation layers untouched.n nAn experienced user may advise against what I am suggesting, however, I may be right n n Dann

[marcinstopyra]

Thanks for your replies, I really appreciate you took time to help menThe thing is that I cannot really switch to tri mesh and remeshing because, my final simulation is going to be 3D object (1/8 of cylinder - base valve of a twin-tube shock absorber), so I need to stay with hex mesh and smoothing. (the tri mesh always resulted in negative cell volumes in my case)nWhen it comes to your idea with interface Dan, I think I will try this approach, maybe thats the solution.nThanks for your help, if anyone have other ideas I would love to hear themn

[Rob]

near enough. In this case the deformation can't remesh as it's hex, and there's a limit to how far that can stretch. If you break up the domain and use non-conformals and tri/tet mesh in certain regions you can get the best of all features. We often tet around the deforming part, and then hex the fixed sections with a non-conformal in there too.