Mesh Plate with Hole

Coreform Cubit Meshing
#1

Hello. I am following along this Abaqus tutorial, and I was curious if I could recreate the mesh they made with Cubit. I tried webcutting and setting the same size that they specified. I am not sure if there’s a particular scheme to get what they produced. Attached are images.

# {width = 100}
# {height = 50}
# {radius = 10}
create surface rectangle width {width} height {height} zplane 
move surface 1 x {width/2} y {height/2}
create surface circle radius {radius} zplane 
move surface 2 x {width/2} y {height/2}
subtract surface 2 from surface 1
block 1 surface 3
block 1 name "plate-with-hole"
webcut body all with plane xplane offset {width/2}
webcut body all with plane yplane offset {height/2}
merge vol all
surface all size 2
mesh surface all

Abaqus Produced

image
image936×478 139 KB

Coreform Cubit Produced
image
image936×478 8.6 KB

#2

I think I could be having a senior moment here, but what is the issue with your mesh? Using the Mk 1 Human Eyeball I’m not seeing anything ‘wrong’ with your mesh?

#3

Mainly to note is its lack of symmetry compared to the meshing performed in Abaqus. Plus farther away from the hole, some of the elements are still pretty distorted.

#4

The workhorse quad-meshing algorithm in Coreform Cubit is “paving” – in fact, it’s the seminal algorithm that led to development of Cubit in the first place.

To recover a mesh similar to Abaqus, you can apply the polyhedron meshing scheme to the surfaces:

surface all scheme polyhedron
surface all size 2
mesh surface all

image

or, if you want to guarantee symmetry - mesh one of the surfaces using polyhedron and then copy mesh onto the others:

surface 6 scheme polyhedron
surface 6 size 2
mesh surface 6

copy mesh surface 6 onto surface 7 source curve 10 source vertex 4 target curve 11 target vertex 1
copy mesh surface 6  onto  surface 8  source curve 10  source vertex 4  target curve 14  target vertex 3  
copy mesh surface 6  onto  surface 9  source curve 10  source vertex 4  target curve 13  target vertex 2

image

And, of course, you could add a smoother (e.g., centroid area scheme) as well:

surface 6 scheme polyhedron
surface 6 size 2
mesh surface 6

set smooth tolerance 1e-6
surface all smooth scheme centroid area pull free 
smooth surface all

copy mesh surface 6 onto surface 7 source curve 10 source vertex 4 target curve 11 target vertex 1
copy mesh surface 6  onto  surface 8  source curve 10  source vertex 4  target curve 14  target vertex 3  
copy mesh surface 6  onto  surface 9  source curve 10  source vertex 4  target curve 13  target vertex 2

image

#5

Thank you for taking the time to do that. That helps a ton. In general, I have found circular geometries benefit from the polyhedron approach. Do you agree that the scheme should be changed for those topologies appearing in a geometry?

#6

when i take a look at the mesh quality and think about solid mechanics, i would say it’s not worth the effort/time to change the scheme manually. The quality is nearly the same. And if you need smaller element size to be more accurate you can always refine or mesh with a smaller size.

pave scheme, scaled jacobian

grafik
grafik1048×496 33 KB

last example from greg, scaled jacobian

grafik
grafik1124×477 39.4 KB