Combination of Tetrahedron and Hexahedron-Elements ???

[JulesVerne77]

hi,
I have to model a structure with a complex geometry.. for first estimations, I used linear brick elements.. now, in the more detailed model, I doubt that these element type will give me good results due to high bending moments at local regions.. Alternatively I can mesh all with tet10 elements, but this will increase the calculation time enormously.. and I assume that I will get some convergence problems (I perform a nonlinear static analysis with contact definitions (in I-DEAS))..

So now my question, is it possible to use tet10 elements in the critical areas (regions with high bending moments) and brick elements (hex8) for the rest.. so I hope I will get sufficient results and a better convergence..

I don`t know, if the combination of tet and hex elements will produce other problems? Does anyone have some experiences in this problem? I would be glad, if someone could help me.. thanks..

 

[corus]

At the boundary between the brick and tet elements you will have to tie the nodal freedoms together. This will give you bad results at that face. This may be ok though if it's far enough away from the region you are interested in.
My experience with tet elements in contact problems is - don't even bother. As you say, there are convergence problems with quadrilateral elements due to problems associated with the mid-side node forces. It would be far more efficient in both solution time and your time to spend more time on meshing the complex area with linear brick elements than just simply meshing with tet elements and hoping it will work. Using a fine mesh with linear brick elements should give you better results than a coarser quadrilateral brick element mesh when looking at the contours.

 

[JulesVerne77]

thanks chorus,
I will try it..