Element formulation...
 
Notifications
Clear all

Element formulations for 3D elements

4 Posts
2 Users
0 Reactions
8,324 Views
Posts: 19
Topic starter
(@erick)
Student
Joined: 6 years ago

There was a post earlier regarding shell element formulations. I was wondering if you can provide insight as to how element formulations for 3D elements work. What formulation would you use for something like a table, a car door, human tissue, etc.?

3 Replies




Negative Volume
Posts: 668
Admin
(@negativevolume)
CEO
Joined: 6 years ago

First off, different types of elements require different element formulations (ELFORM): hexahedral elements (ELFORM=-2 to 3), tetrahedral elements (ELFORM=4, 10, 13, 16, 17), and pentahedral elements (ELFORM=15, 115). Ideally, you'd like all of your 3D elements to be hexahedral as the are the most stable and accurate, but a lot of times you can have really complex geometries where tetrahedrals are required. Example: meshing a table top will be relatively easy to do with hexahedral elements, but meshing an organic shape like a human heart will be much more difficult to do with hexas so you will need to use tetrahedral elements.

The main drawback to using tetrahedral elements is volumetric locking which yields a much stiffer response. You can read more into this here:

https://www.dynasupport.com/tutorial/element-locking/a-pathological-case-of-volume-locking-in-triangular-elements

With regards to element formulations, it's generally best practice to use ELFORM=1 for hexahedral elements, ELFORM=13 for tetrahedral elements, and ELFORM=15 for pentahedral elements (although these are mainly used for transition elements between hexas and tetras). A *Hourglass card (HG) will be necessary for an ELFORM=1 hexahedral element which can drastically change the response characteristics of that material. This can be frustrating if you spend time validating a material model, only to have the response completely dependent on hourglass formulations. Generally, use HG=1,2,3 or HG=6 for soft materials under high rates, HG=4,5 for more structural components or if you are having stability issues with HG=1,2,3. The qm value for the hourglass will matter as well and usually this is between 0.1 and 0.5. Further details on what hourglasses to use can be found below:

https://www.dynasupport.com/howtos/element/hourglass

ELFORM 2, -2, and -1 are all selectively reduced fully integrated hexahedral elements and are much more computationally expensive (and stiff most of the time) than ELFORM 1. -2 and -1 were recent improvements to 2 and have shown some promise. If you are having issues with finding a suitable combination of ELFORM 1 and Hourglass, then you can try any of these and see if it responds any better. 

For tetrahedral elements, ELFORM 13 alleviates some of the volumetric locking concerns and has faster runtime than ELFORM 10. 

A really helpful look into all of this can be found below. Let me know if you have any specific questions regarding any of this!

https://www.dynamore.de/de/download/papers/forum11/entwicklerforum-2011/erhart.pdf

Reply




Posts: 19
Topic starter
(@erick)
Student
Joined: 6 years ago

What about when you have more than one type of 3D element in the same part?

Reply




Negative Volume
Posts: 668
Admin
(@negativevolume)
CEO
Joined: 6 years ago

Great question! Generally, it's best to avoid this whenever possible but sometimes there is no other option. *Control_Solid has an option called ESORT which automatically sorts tetrahedral and pentahedral elements to a default formulation. ESORT=1 - 4 will do this. Unfortunately, you only have the option to sort tetrahedrals to ELFORM=10 and not 13. 

So if you have a 3d part with hexahedrals, tetrahedrals, and pentahedrals and you assign this part ELFORM=1 and have ESORT=1 under *Control_solid, then the hexas will be ELFORM=1, the tetras will be ELFORM=10, and the pentas will be ELFORM=15. 

Does this make sense?

Reply







Share: