Execute MeshModifers on Mesh Adaptivity

idaholab / moose

Multiphysics Object Oriented Simulation Environment

https://www.mooseframework.org

GNU Lesser General Public License v2.1

1.72k stars 1.04k forks source link

Execute MeshModifers on Mesh Adaptivity #5894

Open friedmud opened 8 years ago

friedmud commented 8 years ago

It could useful to be able to run MeshModifiers after the mesh has been adapted. Think about adding a nodeset for instance... the nodeset won't be projected properly after mesh adaptivity so it could be useful to re-run the MeshModifer at that point to re-add the nodeset to any new nodes.

I suggest we add an execute_on = 'adaptvity' option for MeshModifiers....

YaqiWang commented 8 years ago

how about meshChanged method?

friedmud commented 8 years ago

meshChanged() is a callback for an object to reset its own internal data. It doesn't "re execute" an object...

Plus you would want this ability to be toggleable per MeshModifier for ALL MeshModifiers... so we need to actually do this in FEProblem

permcody commented 8 years ago

This isn't a bad idea however if you think the main use-case is adding new nodes to an existing nodeset we should probably just add that capability specifically to the framework. I don't think this option makes sense for very many of our modifiers. On Sat, Oct 31, 2015 at 7:50 AM Derek Gaston notifications@github.com wrote:

meshChanged() is a callback for an object to reset its own internal data. It doesn't "re execute" an object...

Plus you would want this ability to be toggleable per MeshModifier for ALL MeshModifiers... so we need to actually do this in FEProblem

— Reply to this email directly or view it on GitHub https://github.com/idaholab/moose/issues/5894#issuecomment-152738212.

friedmud commented 8 years ago

Current MeshModifiers it makes sense for:

AddExtraNodeset
AssignElementSubdomainID
BoundingBoxNodeSet
OrientedSubdomainBoundingBox
SubdomainBoundingBox

mlesueur commented 8 years ago

Add this one to the list Derek: ImageSubdomain I am following this thread because I really want to use adaptivity with the mesmodifier above. But since it isn't implemented, I am currently working with my own solution.

friedmud commented 8 years ago

@mlesueur good one!

mlesueur commented 8 years ago

Any update regarding this topic ? @friedmud

friedmud commented 8 years ago

Nope - no update. Still a good idea though ;-)

mlesueur commented 8 years ago

I am using the ImageSubdomain meshmodifier to create meshes from CT-scan.

Right now, if I want to fully capture the image, I have to go to a initial mesh refinement equal to the image resolution and then use adaptivity to coarsen my mesh in order to simplify the simulation.
If this function is implemented, I would not be constrained to use a uniform refinement at the beginning (which is very expensive) but I could just start with an adaptive refinement before using the meshmodifier to separate the blocks.

But you were the one pushing for this update, I suppose you don't disagree with its usefulness ^^

permcody commented 8 years ago

We can handle your request without fully implementing this feature.

mlesueur commented 8 years ago

Wow that would be great Cody, do you need me to help you in any way to implement this ?

GiudGiud commented 1 month ago

This should be done with MeshModifier (coming soon) user objects now. We'll nest them under a dedicated MeshModifier syntax.

It's not clear how to do this with MeshGenerators for now.