[CIG-SHORT] Fault "id" versus fault "label"

[Brad Aagaard]

Ravi-

You need to set BOTH the fault "label" and "id". The "label" is a string 
and corresponds to the nodeset id in CUBIT or the name of the pset in 
LaGriT. The fault "id" is a tag associated with the cohesive cells. It 
does not need to match the label.

Fault "label"

In order to know which vertices are associated with the fault, we use 
the fault label (which is a string), because some mesh generators use 
strings (LaGriT) to identify groups of nodes while others use integers 
(CUBIT Exodus files).


Fault "id"

For each finite-element cell in the mesh, we set a "material-id" tag 
that corresponds to the material and quadrature scheme for that cell. 
The material-id is set based on the material id for bulk cells and the 
fault id for cohesive cells. When we want to operate on cells associated 
with a material, quadrature scheme, fault, etc., we ask Sieve for a list 
of all cells with material-id equal to some value.

What should happen is that if all materials and faults have unique ids, 
then each bulk or cohesive cell is associated with one and only one 
material or fault. However, if there are nonunique id values, then a 
bulk or cohesive cell ends up being associated with multiple materials 
or faults. In general, this will cause odd, unpredictable behavior.

In version 1.1 we will be adding a check for materials and faults having 
the same id and report this to the user to prevent cells from being 
associated with more than one material or fault.

If you do not set the material id or the fault id, it will use the 
default value. Thus, if you have multiple materials or faults and use 
the default id values, then cells will end up being associated with 
multiple materials and/or faults.

I realize the manual is a little terse on this subject. We will include 
more explanation (perhaps what I have above) when we update it for 
version 1.1.

Brad