Eng-Tips is the largest engineering community on the Internet

Intelligent Work Forums for Engineering Professionals

  • Congratulations waross on being selected by the Eng-Tips community for having the most helpful posts in the forums last week. Way to Go!

Meshing (Element Type and their Effect) 2

Status
Not open for further replies.

Ramin88

Mechanical
Jul 27, 2012
10
GB
Hi guys
I just had a question regarding the element types and mesh control in abaqus i.e. If i mesh my model with element shape of Hex or if i mesh my model with Tet (Pink color in the viewport)is that gonna effect my result too much...?
How much generally element shape will be effective in the result (Assuming that a model can be meshed with both element shape then which one would be more precise)?

P.s with the Hex shape im not able to mesh the model, the shape will turn into orange color and then it requires to simplify the geometry or partition it.

Regards
Ramin
 
Replies continue below

Recommended for you

Short answer: It depends.
Long answer:

It depends on the problem at hand. If, for example, contact is involved in 3d model and contact pressure is of interest, you'd be better off using linear bricks (c3d8) (not quadratic bricks i.e., c3d20) or modified tetrahedra (c3d10m). Even with modified tetrahedra, you may have to line up the surface with a layer of membrane elements or something. Since the tetrahedra need to be quadratic and another layer may have to be modeled, computational cost increases. If the model is small, any regular machine will solve the model. As soon as more complexities (material/geometric/boundary non-linearities) start piling up along with the number of degrees of freedom (e.g., large number of nodes), you may have to use heavy duty workstations or even servers to solve a model.

Bottom-line: You need to understand the problem. In that, you need to know what your field variable of interest (say, stress) is, where the (stress) concentrations may arise (a sharp corner, for example), where (stress) transitions occur (say, for example, tensile to compressive stresses in a tip-loaded cantilever beam), etc. and if such areas are sufficiently away from the region of interest, then you don't need to bother too heavily on getting the perfect mesh.

I believe hex-mesh offers the highest accuracy to computational expense ratio. Quadratic hex-mesh offers an even higher accuracy but the computational expense is higher too and it can't be used in contact (unless the element is modified).

Simplifying the geometry is an invaluable skill. Partioning is an excellent tool to get the best possible mesh; if possible, use it to get a hex mesh as long as there aren't too many transitions in the region of interest.

And don't forget to do a mesh convergence study (for the field variable at the location of interest).

 
Status
Not open for further replies.

Part and Inventory Search

Sponsor

Back
Top