Adaptive remeshing techniques

[Jlog50]

Hello all,

I am trying to learn more about remeshing techniques in FEA. I am using ABAQUS, and the programme does not seem to allow adaptive remeshing (i.e. mesh refinement) for 3D models using hex elements.

Is this the same case for all FE software? If so, anyone know why this is? Is it a limitation of the algorithms?

Thanks, I look foward to any comments.

 

[flash3780]

It's much easier to write an algorithm that meshes things with tets. I'm not aware of any software that is able to create an adaptive mesh with hexahedral elements. Since adaptive meshing requires the program to remesh locally on the fly tetrahedral elements tend to be preferred anyway, since they're able to rapidly transition to very small elements.
If you're having trouble with solution times, you might relax your convergence criteria. You can always create a submodel of the area of interest if you don't have the horsepower for an adaptive mesh solution.

 

[Jlog50]

The only reason I preferred the hex elements, is because I am looking at a ball in socket joint which is round surfaces, therefore I felt hex may be a better option than tets?

 

[KevinDeSmet]

By hex you mean cubes, no? I think second order tets will do a pretty good job of capturing round surfaces.

Certified SolidWorks Professional

 

[Jlog50]

Hi all,

Yes I mean cubes. thanks for your advice I will try 2nd order tets and look at the accuracy of results.

Thanks.

 

[flash3780]

First order tets should be avoided for structural analyses (they provide anti-conservative results), but second-order tets aren't terrible elements. As a matter-of-fact, many codes rely solely on higher-order tetrahedral elements (ex. Pro/Mechanica, Cosmos, etc.). The penalty, of course, is that you require more nodes to solve the same problem at a comparable level of accuracy.