Does layer in this kind of structure has importance in terms of the results, because I read in somewhere that using just one layer is not enough the capture the bending behavior correctly?
My results are seems okay in terms of stresses and thicknesses but I am not able to validate it.
P.S.:I tried to use surface elements to represent it but due to the complex shape I am not able to manage contact between surfaces, also compound gives me some problem. The Surface Mesh creates Peaky Distortion on the edges. If you have any advice also for that I will be happy.
P.S.: Also from some reading, is this kind of structure has better results when they meshed as surface element?
It’s very thin-walled so it would be more appropriate to model it using surface geometry meshed with shell elements. Instead of compounding, it’s better to merge the surfaces in CAD software then. I can tell you how to do it in FreeCAD, for example.
Shells in CalculiX are internally expanded to solids and have some limitations because of that, but it’s also possible to extrude them using the Thicken Shell Mesh tool if needed (if shells happen to be too problematic in a given case).
To specify the number of solid elements in the thickness direction, you would need proper partitions (i.e. edges in that direction) and you could assign local refinement to them (even with a prescribed number of elements since PrePoMax v2.5.0). However, tetrahedral mesh will still do mostly local refinement so you would rather need a hex mesh which is limited to simple geometries (here, you could try splitting the model into subvolumes with 5/6 faces having 3/4 edges each).
So, all in all, I’d rather prepare a properly connected surface geometry in CAD software and start from that.
I did the surface mesh with compound. I tried to apply the “distributing constraint” with the calculix editor. It is not applying the constraint condition on the surface mesh element. Are there any limitations about it?
I was able to obtain meaningful result in solid elements in terms of “distributing constraints”.
With your comment I tried to split volume, I tried the compound them and they are compounding. Only problem is extrude mesh or sweep mesh is not working with compound mesh. meaning that I am not able to obtain layered mesh in the surface normal direction. If the volume is seperated with 4 edges 5/6 faces it is okay but when it is compounded is impossible to obtain. I don’t know how you are able to connect separate volume mesh together. With a tie contact? If so, it creates higher stress point in the edge points.
Can you explain me how you divide volumes to create layered mesh and how you connect them?
because i divide the solids that are able to be meshed with the sweep or extrude command, but joining them is not working, I know a way that only tie contact, is there another way to do it?
Try a different type of constraint or the Thicken Shell Mesh tool.
Yes, extruded, revolved and sweep meshes don’t work with compounds. Only transfinite meshes. The idea is that many models can be divided into such 5/6 sides subvolumes in CAD software and then compounded in PrePoMax so that you can use this Gmsh algorithm to get a continuous regular hex mesh.
For volume partitioning, you can use FreeCAD, but it’s very tedious manual work there. I can help with some tips/macros if needed. But if you have access to other CAD software, it will probably be much easier there.
Tie constraints are the last resort if you can’t use the aforementioned approach or don’t have time for that. Same as tetrahedral meshes (even worse, especially in such thin-walled cases).
With the same approach with any mesh command operations, after the nodes are in same position, there is a option under the model → node-> merge coincident nodes option to merge the coincide meshes nodes. Then applied the merge part command in FE model fixes the problem of compound mesh problem where we can’t do extrude or swap operation basically.
Basically without any contact operation all model works consistently in my case.
But the nodes must be precisely close to each other so it takes too much time.
Exactly, that’s what limits the usability of this feature. But it’s still helpful, especially for patterned meshes like in the Matej’s video.
It makes all nodes within the specified distance coincident. Midside nodes too for second-order elements. So it’s a full connection, like with compounding leading to continuous meshes.
You can easily test such merging operations and connections in a frequency analysis. Any disconnected regions will visibly separate.
