Hex Meshing of Complicated Part

[MegaStructures]

I am having a lot of trouble hex meshing the below part in Abaqus. This is an issue because this is the first of a handful of research models (industry) I need to make and the remaining models will be complicated as a tube will be added at a random angle.

There does not seem to be a way to partition this model to allow for automatic hex meshing. Is there any manual method that can be used to mesh a model such as this in Abaqus? I'm thinking something similar to meshing each solid separately (all welds and tubes separate) and then manually moving and merging the nodes. Any ideas are welcome.


“The most successful people in life are the ones who ask questions. They’re always learning. They’re always growing. They’re always pushing.” Robert Kiyosaki

 

[FEA way]

It seems like a good idea to mesh these tubes separately and then tie them together or merge their nodes. You can also try using mixed mesh (tie constraint will connect hexahedral and tetrahedral elements) or even pure tetra mesh.

 

[MegaStructures]

Thanks FEA way.. Any recommendations for "matching" meshes on adjacent solids to limit the amount of move in the move merge operation?

“The most successful people in life are the ones who ask questions. They’re always learning. They’re always growing. They’re always pushing.” Robert Kiyosaki

 

[FEA way]

With properly meshed tubes you can try merging nodes automatically using Instance —> Merge/Cut tool. For manual merging it would be easier to use some external mesher/preprocessor.

 

[MegaStructures]

Fair but my entire problem is that I don't know how to "properly" mesh this part

What external pre-processor would you recommend?

“The most successful people in life are the ones who ask questions. They’re always learning. They’re always growing. They’re always pushing.” Robert Kiyosaki

 

[FEA way]

What I meant is that these parts should be meshed in such way that it will be possible to merge their nodes without moving them too much (this could distort elements).

There are two main choices when it comes to preprocessors/meshers - Hypermesh and ANSA. Both should let you complete this task.

 

[Mustaine3]

Just use Assign Mesh Controls to assign tet-meshing to the regions that can't be hex-meshed. The tool will connect the different regions automatically; either with common nodes or with tie constraint.

 

[MegaStructures]

I understand that I could TET mesh the part easily, but the overwhelming majority of the research in my area uses solid HEX meshes. I have not found a single study that recommends TET meshes and would be worried about losing credibility for my research if I resorted to a free TET mesh. There must be some way to HEX mesh this part

“The most successful people in life are the ones who ask questions. They’re always learning. They’re always growing. They’re always pushing.” Robert Kiyosaki

 

[Mustaine3]

The more complex a geometry gets, the less likely it is that it can be meshed with hex-elements. It's actually quite obvious...
Results with parabolic tets are very good. Their disadvantage is the computational effort they need.
In contrast, hex elements can also be used in the wrong way. Most people are not aware of the limitation that linear hex-elements have.

And my tip in the previous post was not to mesh everything with tets, only the cells that cannot be meshed with hex-elements.

 

[MegaStructures]

The areas that can’t be hex meshed are critical areas for measurement, so I need a uniform and high quality mesh there. What are the known limitations of TET meshes in non-linear analysis if any?

“The most successful people in life are the ones who ask questions. They’re always learning. They’re always growing. They’re always pushing.” Robert Kiyosaki

 

[FEA way]

There are several studies that compare these elements in various cases. Not only research papers but also examples in Abaqus documentation. They are mentioned in the chapter "Solid (continuum) elements".