Page 1 of 1

Comparing meshes : Duplicate vs No-duplicate

Posted: Fri Jun 05, 2015 3:05 am
by amit12kd
Hi,

I am trying to create a couple of meshes using RSurf mesher. In one case, I am using the duplicate nodes at a surface feature of GID and in the other case, I chose not to duplicate a particular surface. In both cases I use the same element size (specifically, I am trying to generate tetrahedral mesh). Would the mesh created by RSurf mesher be same in both these cases (except for the fact the nodes are duplicated in one case and not in the other case)? Furthermore, would the mesh geometry change if I use a linear tetrahedral mesh or a quadratic mesh if I keep size of the elements the same? It doesn't seem to be different but for most cases but it appears to change for some cases. Not sure how the algorithm is adopted by RSurf.

Re: Comparing meshes : Duplicate vs No-duplicate

Posted: Fri Jun 05, 2015 7:31 am
by abelcs
Hello,

the mesh of the surfaces should be the same, independently on if they have the 'duplicate' mesh criteria applied or not. This behaviour is the same for all the surface meshers (Rfast or Rsurf).

It is not the case of linear and quadratic meshes. The meshes should be very similar (even exactly the same in some particular cases), but not identical, although the sizes assigned are the same.

Re: Comparing meshes : Duplicate vs No-duplicate

Posted: Fri Jun 05, 2015 9:47 am
by escolano
If do you want to force that two 'similar surfaces' have exactly the 'same mesh', at least a relationship 1-1 between its nodes and elements,
then you must create a 'separated contact volume' between them.
Another possibility for topological rectangular surfaces is to do a structured mesh fixing the number of divisions on each direction.

Re: Comparing meshes : Duplicate vs No-duplicate

Posted: Fri Jun 05, 2015 4:56 pm
by amit12kd
I am not sure if I understood you correctly. I have the exact same volumes in the two cases I have, shouldn't the the mesher create identical volumetric meshes if element size specified remains the same regardless of the fact that I am duplicating nodes at some surface or creating quadratic or linear elements?

Re: Comparing meshes : Duplicate vs No-duplicate

Posted: Fri Jun 05, 2015 5:19 pm
by escolano
In general, two entities that 'seems equal' (e.g. a copy of one with a translation) are not guaranteed to generate the same mesh (e.g. exactly the same amount of nodes).
There are a lot of things that could affect the mesh generation, e.g. the order of creation of the entities, its orientation, its location, close entities, etc.

Independly if you use other features like duplicate nodes, or if the mesh was set of degree linear or quadratic.

Re: Comparing meshes : Duplicate vs No-duplicate

Posted: Fri Jun 05, 2015 6:40 pm
by amit12kd
Thanks a lot for the clarification. One last followup question regarding the same- so if the mesh could change by just changing the order of the elements, is there an inbuilt feature that I can utilize to have a measure of how different are the meshes that GID creates each time, or is there a systematic behavior to observe mesh creation?