The problem of mesh refinement

[PEI0214]

Check these boxes if you have followed the posting rules.

• Q&A General is the most appropriate section for my question
• I have consulted the posting Guidelines on the Discussions front page
• I have searched the Discussions forum and my question has not been asked before
• I have searched the MOOSE website and the documentation does not answer my question
• I have formatted my post following the posting guidelines (screenshots as a last resort, triple back quotes around pasted text)

Question

Hello,
When I use mesh refinement and MeshModifiers together, the following happens. This doesn't seem normal. When the grid is 8-node cells, this phenomenon does not occur.

Grid and result

The 6-node grid and results are as follows:

The 8-node grid and results are as follows:

The setting of boundary conditions and mesh refinement are the same. It's all full restraint on the bottom and pressure on the top. Why does this happen?

[GiudGiud]

Hello

This looks like the solve is failing and the final value for mesh displacements in unphysical.
Can you run these other cases:

• no mesh modifier, no mesh refinement
• mesh modifier, no mesh refinement
• no mesh modifier, mesh refinement
  To see if the interaction of the two is the problem.

Also can you please paste your whole input

[PEI0214]

Is there more information to learn about it? Can you run the input file above? It has problems in the calculation of refined mesh.

[PEI0214]

Use MeshModifiers for six-node units after refinement. The constraint on the hanging node seems to have failed.

[GiudGiud]

?
how so?

[PEI0214]

This will get the wrong result.

[GiudGiud]

Can you try setting the boundary conditions to use preset= false

And is the nodeset properly covering the newly appeared nodes from refinement ? What does the output show

[PEI0214]

Can you try setting the boundary conditions to use preset= false

I tried setting preset = false, but it didn't affect the result.

And is the nodeset properly covering the newly appeared nodes from refinement ? What does the output show

How to check the nodeset? It should be correct.

[GiudGiud]

So first

    enable_neighbor_refinement = false

is a bad idea. As mentioned in the documentation, neighbor refinement is there to avoid problems.
Let's keep it to the default

Looks like the block restriction of the displacements is the problem. Using this makes it work for me

[Variables]
  [disp_x]
    # block = '${new_domain_blocks}'
  []
  [disp_y]
    # block = '${new_domain_blocks}'
  []
  [disp_z]
    # block = '${new_domain_blocks}'
  []
[]

# Present empty equations before subdomains are changed
[Kernels]
  [n1]
    type = NullKernel
    block = '1 2'
    variable = disp_x
  []
  [n2]
    type = NullKernel
    block = '1 2'
    variable = disp_y
  []
  [n3]
    type = NullKernel
    block = '1 2'
    variable = disp_z
  []

[PEI0214]

In my simulation, the mesh needs to be gradually activated. If I don't set block restriction of the displacements, the inactive mesh will also have displacements, which leads to poor results. Is there a way to avoid this? While using displacement block restriction, the mesh should still be refined properly.

[GiudGiud]

the inactive mesh will also have displacements, which leads to poor results.

It should not, not with just the Nullkernels as the equation for displacements

[PEI0214]

Does this require creating a kernel for zero displacement?

[GiudGiud]

I dont think so. See the input snippet I posted.
But you can try that if you think that will help. Having kernels that push the displacements to 0 in a region where they should be 0 will not hurt