Conformal vs Non-Conformal Mesh

  • 477 Views
  • Last Post 02 February 2019
joepa_2017 posted this 02 February 2019

I saw a good explanation of the difference between conformal and non-conformal meshes in another post. I can see that a conformal mesh is likely preferred when two surfaces are contacting each other. 

If I have done simulations using non-conformal meshes but have used the mesh convergence tool, would that make the accuracy of the solution approach that of a conformal mesh? Or would that be unfair to say since the mesh convergence tool converges only on the maximum or minimum value? (i.e. perhaps the convergence tool has nothing to do with the nodes where the surfaces meet and form a non-conformal mesh)  Also, if it's relevant, I am using bonded connections where the surfaces contact each other.

Order By: Standard | Newest | Votes
peteroznewman posted this 02 February 2019

I don't make much use of the mesh convergence tool, so I will leave others to comment on its accuracy, but that won't create a conformal mesh where none existed at the beginning.

When I have two bodies of different materials that move together because they are bonded, I first choose shared topology, which creates a conformal mesh, and let the two bodies be connected to the same nodes shared on the common face.  If I want to study the stress in the bonded interface, then I change to bonded contact where I have more options for that study. I say two bodies of different materials because if they are the same material, then I could unite them in CAD.  As you know, sometimes you can get a better quality mesh by splitting a single body into multiple bodies that are easier to mesh and using shared topology to connect them.

Regards, Peter

 

joepa_2017 posted this 02 February 2019

It makes sense that the mesh convergence tool wouldn't create a conformal mesh. I was just wondering if that could compensate for inaccuracies that a non-conformal mesh would introduce.

Also, if a non-conformal mesh is able to generate a solution, is there an obvious reason why a conformal mesh would no longer converge? I guess right now, I'm able to get a solution with a non-conformal mesh, so I wonder if I can keep using a non-conformal mesh; or would it be imperative to implement a conformal mesh?

peteroznewman posted this 02 February 2019

One way to review the results of a non-conformal mesh is to request a Directional Displacement scoped to the two parts. Zoom in on the surface that has the bonded contact. On the legend, grab the line that separates red from the color below and drag it down until red fills one of the parts with red and back off a little so there is a couple of colors leading up to the bonded surface. Now grab the line that separated blue from the color above and drag it up until blue fills one of the parts and back off a little so there is a couple of colors leading up to the bonded surface. You have just created a very sensitive colormap to see displacement discontinuities across the bonded interface.

If you had a conformal mesh, the displacements must be continuous across the interface because there is a shared node. If the bonded contact allows some relative motion, you will see a step in the displacement contour. You can use the colormap to quantify the size of the step.

Two reasons a conformal mesh might not converge when a non-conformal mesh did converge is if the element size was larger on the conformal mesh, or if the element quality was worse on the conformal mesh.

Close