[hsieh]

Hi,

If one generates all-Tet elements first, then, decoposes each Tet into 4 Hex (this should not be difficult to do within OpenFOAM, I assume). What do you think about the quality of the Hex elements? Assuming the initial Tet elements have great quality.

Has anyone done this?

Pei

[mprinkey]

I haven't read the paper but it seems this has been done before:

http://scitation.aip.org/getabs/servlet/GetabsServlet?prog=normal&id=APCPCS00071 2000001002210000001&idtype=cvips&gifs=yes

Here some results from anl:

http://www-unix.mcs.anl.gov/~csverma/tet2hex.html

[hsieh]

Dear Michael:

Thanks for the reply. The website I found mentioned that "most finite element analysts, because of the poor element quality that will in general result, have rejected this solution." I am wondering whethere there is any improvement in this area recently.

Pei

Website: http://www.andrew.cmu.edu/user/sowen...y/hexsurv.html

[mprinkey]

There are always smoothing algorithms that can be applied afterwards. I think that most tet generators have a smoothing pass following the base grid generation. I believe most are based on an elasticity analogy. I know Hrv mentioned doing something similar when doing mesh motion. That would likely improve mesh quality in most cases. The question is whether that would be in anyway superior to just using the polyhedral mesh extracted as the dual of the tet mesh. Unless you were using finite elements, I don't see why tet->hex would be a win over tet->poly.