Meshing in General, Guidelines for good meshing

Altair Forum User

Hy everyone,
I am a newbie and I am trying to get ab better understanding of Meshing in general. On that quest I have been reeding my way throught this Forum but I did only find something very specified requests of how to mesh this speciel form or size or verry general information.  
 

As I understand it, there is not the one concept that works or the one technique that is the best. But every engineer needs guidelines and general idea where a valid solution might to be found.

 

1. What tyypes of elements are there?

 

2.when do I use which elementtype?

As I understood so far, this depends on: the form of my part; the kind of the analysis(stress or displacement); the elementsize i want to use.

 

3. How do I decide the elementsize?

As I understood so far, you have do make a compromise between the numeric mistake, which is getting bigger proportional to the number of elements and therefore antiproportional to the elementsize and the mistake concluding from bigger elements and there rough not fitting shape.

 

I realize that it sounds like I am asking for an all-in-all-solution but i really need more basic information. I would also be very happy if you link me to some helpful explenations.

 

 

Thanks for any help, to be given. That beeing said I admire how productive this forum seems to be.

regards Nikolai

 

Altair Forum User

1. Firstly, it depend to your FEA solver. No rules for all solver.

Size & type of element depend to your computing capacity and your FEA experiences.

It's not so easy to answer your questions. But you can do method 'try-compare-fix-try again'.

FEA is ... an art of numerical computation on which you have to found the best compromise between precision of results and the cost of analysis.

Altair Forum User

Thank you for your promt answer.

I use Radioss and Optistruct as solvers in the Hypermesh enviroment. Do you know rules for those?

Concerning the Method 'try-compere-fix-tryagain':
How do I evaluate the quallity of my mesh and of my results? As I understood, given that element type and applyied loads are chosen corectly, the precission of my result on the one side depends on the nodedensity as an aspect of how precise the geometry is approxmiated and on the other side on how small my elements are which would result in a bigger rounding error in the noumeric soving of the mesh.
Given that when I pick my Elements too big the result is bad for not aproximating percise enough.

But if I pick the elements too small the result is bad for the rounding error has gone up.

 

I've read in another forum that to see of a smaller elementsize is nessary you are to calculate whith smaller elements. If the result stays nearly the same, the earlier elementsize is good.
Do you see my dilemma?

Reguards Nikolai

Altair Forum User

Have a look at this page : http://www.iesweb.com/products/visualanalysis/help/model/plateelements.htm