Selecting Element Types in LS-Dyna

Suresh J

Hi,

 

I'm doing a study on which element types would opt for static analysis. Could anyone please help me in selecting the element type (Shell/Hex/Tet) and element formulation in ls-dyna. Also, I'm getting different results when I make the mesh more finer. Should there be any mesh size variations to be maintained between Hex and Tet elements or same element size can be used for both element types with different elforms? Please let me know if anyone can help me on this.  

 

Thanks,

Suresh

tinh

Hi,

Are you trying Implicit Lsdyna code?

if you want to study elem types, you should compare results with other solver (example: nastran or optistruct)

ofcourse results will be different when changing mesh size. I cannot say should or should not

you are studying about these and you should make all cases then compare results to choose good formulation /emoticons/default_wink.png' srcset='/emoticons/wink@2x.png 2x' title=';)' width='20'>

Suresh J

Hi Tinh,

Thanks for your reply.

yes, I'm using LS-Dyna implicit code.

I heard that the tet element size should be half of the hex element size (if hex element size is 2mm then tet should be 1mm). Is it true? 

Altair Forum User

Altair Forum User said:

yes, I'm using LS-Dyna implicit code.

I heard that the tet element size should be half of the hex element size (if hex element size is 2mm then tet should be 1mm). Is it true? 

Where do you found this 'rule'? By respecting this 'rule' what will you got? Same max stress? same stiffness? same displacement? or what?

tinh

Hi , I think you can make a simple model to compare hex-2mm and tet-1mm, make two cantilever beam (rectangular section) and mesh them with hex and tet

apply same BCs and run analysis, compare the results

normally tetra4 (or tria3) is stiffer than hex8 (quad4) due to constant-strain so someone wants to mesh tetra model smaller, but not always ratio 1/2 i think

when using hex8 or quad4 elforms you should also care about locking (full IP), locking also make model stiffer, so elem size should be smaller than 1-IP form

 

Suresh J

Hi Tinh,

I have made a simple model and tested as you said but the results are still different (deflection & stress variations are more).

I didn't get your 2nd point 'when using hex8 or quad4 elforms you should also care about locking (full IP), locking also make model stiffer, so elem size should be smaller than 1-IP form'. Could you please explain it.