[FreeFEM 3] 3D mesh generation and handling in FreeFEM

[@TheLittlebigman78]

Hello ! Thanks for sharing this. What would be the work around to import a .ply file containing multiple cylinders into FreeFEM ?

[@TuxRiders]

Hi, I'm not sure if it's possible to import it directly into FF. Maybe you need to first convert it to MEDIT (.mesh) or GMSH (.msh) format and then import it. Another workaround would be using the PETSc DM interface. It supports ply files but I'm not sure if this functionality has been implemented in FF PETSc interface yet!

[@akh26.12]

Hello, I am trying to do a structured mesh for a circle, but I am unable to do that using freefem. I did this mesh using matlab, but how can I solve it using freefem? Thank you in advance.

[@TuxRiders]

Hi. You need to save the mesh in a supported exchange format (like .mesh or GMSH format). You may also consider making the mesh in GMSH or SALOME. You should remember that FreeFEM doesn't support any element type other than tetrahedral.

[@tahaaydogan2747]

hi, ı want to find a way to anaylize plate with FEM by using C++. How can ı do it?

[@TuxRiders]

hi. for finite element using C++, you may take a look at MoFEM or deal.ii, great libraries for modeling various physics, including solid mechanics.

[@DanialMolavi]

Thank you for the video.But have you tried importing second-order elements let's say from SALOME? It seems FreeFEM cannot easily read higher order elements (P2) generated in external software.

[@TuxRiders]

I've never tried to import a 2nd order mesh into FreeFEM, but I guess you are right, and FF has problem for importing such mesh files. But, why do you need to do it? does it not suffice to read a normal external mesh (1st order I mean) and define a P2 FE space on it?

[@DanialMolavi]

​@@TuxRiders As far as I understood from FreeFEM documentation, P2 FE space requires 10 computation points on a 3-D tetrahedral element. I think you can only get this number by using 2nd-order element, since 1st-order tetrahedron has only 4 nodes. So I am not sure if it makes sens to assign P2 FE space to an element with only 4 nodes. Do you have any thoughts on this?

[@TuxRiders]

@@DanialMolavi what you said about P2 FE space is totally true, but it doesn't mean you need a 2nd-order mesh to have P2 space. those DOFs you mentioned (like 10 in 3D) are defined mathematically when you create a P2 fespace and do not need physical nodes (more info here: ruclips.net/video/LNeD9vEqN_Q/видео.html&ab_channel=TuxRiders).
the mesh is always 1st-order in all FreeFEM examples, like "mesh Mesh = square(10, 10); fespace SpaceP2(Mesh, P2);", in which a 1st-order square mesh is used to define a P2 space. that's what I meant by importing an external 1st-order mesh and defining a P2 space on it.

[@DanialMolavi]

@@TuxRiders ​ You're right. Normally we do iso-parametric mapping and that is why I was uncertain of this, but as you mentioned, element and FE-space could also be of different orders.
I could resolve my issue. Thank you for your helpful tip :)

[@clementadande1815]

Thank you for this video. I need held to generate a 3D mesh of the segment (0,1)*{0}*{0} with FreeFem++ and substract this mesh to this rectangular parallelepiped (0,1)*(0,1)*(-2,2). Help me, please. Thanks.

[@TuxRiders]

glad to know you found the video useful. I'm afraid I'm not very proficient in doing such binary operations in FreeFEM, so unfortunately I cannot help in this regard.

[@AJ-et3vf]

Great video

[@TuxRiders]

Thanks!

[@mhecht2909]

Very nice video, as always. Nevertheless, still waiting for a Python FEniCS video series. For online applications I need Python for communication with process.

[@TuxRiders]

Thank you. I think I should start recording some videos on FEniCSx after finishing the FreeFEM series.