initial negative jacobian

[theilenb]

Hi!

I am having trouble running a cylindric model meshed with gmsh. I get some errors stating "Negative jacobian detected at integration point ... of element ...
Did you use the right node numbering?" and then the run aborts with the info febio found some initially inverted elements.

As to now I understood that negative jacobians occur due to too large deformations or ill formed elements. However, when I look at the mentioned elements in gmsh they (or at least most of them) seem to be shaped fairly well, meaning they are not particularly flat and have no intersecting faces.

Exactly which jacobian is detected to be negative here, and how is it computed? I would really like to figure out how to get these elements right, since gmsh is not able to create a hexahedral mesh without these errors.

I attached a feb-file of this problem, although i'm not sure if you are able to look at the individual elements in febio respectively preview/postview. I did that in gmsh.
I should probably point out, too, that i did not use febio's importer but created the .feb file manually from the gmsh-file.

Thanks,
Sebastian

[ateshian]

Hi Sebastian,

I imported your file into PreView and viewed it in wireframe mode (View->Toggle Solid/Wire). I see that inside the cylinder there is a smaller mesh of what seems to be an enclosed structure (a cavity?). Is that intended to be there?

Gerard

[theilenb]

Sorry, I intended to remove the different materials. It represents an enclosure inside of the cylinder. But since its one continuous mesh with different materials (so no contacts) i dont think its the reason for the negative jacobians?

[ateshian]

Hi Sebastian,

I looked at the first element which FEBio says has a negative Jacobian:


**************************** E R R O R ****************************
Negative jacobian detected at integration point 3 of element 1246
Jacobian = -5.26076e-06
Did you use the right node numbering?
************************************************** *****************

If I extract the coordinates of the nodes of that element, create an FEBio file with just that element and import it into PreView, it does look to be inverted (see attached temp.feb file). Perhaps you should double check your mesh generation program?

Best,

Gerard

[theilenb]

You are right, this element looks wrong. Perhaps some elements just were only a bit inverted so to speak, so that it didnt show in preview.

I will try to optimize the mesh-generation, but since i am using an algorithm for hexahedral elements i think i will always have to look out for such mistakes.

Thanks for your help!