Hi all,
We are running a model where we take a subset of a large 3D model, rotate it, and restart it. We can do this fairly easily from a time = 0 model with no mesh refinement criterion other than depth dependent setting. We can take subset from paraview, do some postprocessing, re-mesh and interpolate onto a new grid and then make an ascii file that ASPECT can read (following instructions from the manual and cookbook examples).
However, it is more difficult, and a lot messier, when trying to do the same procedure after Time = 5 Myr (for example) rather than Time = 0. The grid is no longer nice and regular as it was at time 0, which makes it more difficult eventually plot onto a new grid.
I thought I would message the forum to see if there is any way to take an irregular grid from paraview and and repackage into the appropriate restart.mesh* files to then restart a modified model.
Basically, the same question asked in another way would be: is it possible to manipulate the restart* files and then repackage and “restart” a new model?
I hope that is clear, thanks for you help.
Cheers,
Phil
However, it is more difficult, and a lot messier, when trying to do the same
procedure after Time = 5 Myr (for example) rather than Time = 0. The grid is
no longer nice and regular as it was at time 0, which makes it more difficult
eventually plot onto a new grid.
Phil, can you say how that happens? Are you deforming the mesh with a free
surface, or why does the mesh change over time?
I thought I would message the forum to see if there is any way to take an
irregular grid from paraview and and repackage into the appropriate
restart.mesh* files to then restart a modified model.
Basically, the same question asked in another way would be: is it possible to
manipulate the restart* files and then repackage and “restart” a new model?
No, that’s not a route worth going. These restart* files are really just
memory dumps of some internal data structures. It’s not possible to manipulate
them in any reasonable way outside the program that understands what these
data structures really are…
Best
W.
Hi Wolfgang,
Sorry for the delay - yes, we are deforming the mesh with a free surface. This makes it difficult to generate the strict formatting of starting a model from an ascii file, yes?
Best,
Phil
Phil,
Yes, then I see. It’s difficult to evaluate unstructured meshes at arbitrary
points. I’m unclear about one thing, though: Didn’t you say that you are doing
the interpolating from the mesh as it is to the ascii input model using
Paraview? Why is it more difficult to do this for one kind of mesh using
Paraview than with another kind of mesh?
Best
W.
Hi Wolfgang,
Actually, we are taking a subset from the model using paraview, then doing postprocessing using matlab (in this case).
We haven’t used paraview to interpolate onto another mesh setup - is that possible? I didn’t think paraview would be able to define a new grid system and interpolate old points onto it.
Cheers,
Phil
I have no idea whether that’s possible. I just read your message incorrectly 
But then my question still stands, if you replace Paraview by Matlab:
“I’m unclear about one thing, though: Didn’t you say that you are doing
the interpolating from the mesh as it is to the ascii input model using
Paraview? Why is it more difficult to do this for one kind of mesh using
Paraview than with another kind of mesh?”
Best
W.
Hi Wolfgang,
By turning off free surface we managed to keep a fixed grid - and have been able to do all our post processing. Thanks for the help!
Phil