Hi,
I use Sculpt for smoothening the phase boundaries in my microstructure model(image). I got the model working without turning on the periodicity. But, I evetually want the Mesh to be periodic. When I turn on the periodic option, scult fails with ‘ERROR: Unable to resolve some non-manifold conditions in the model’ option. I tried playing aroung with the defeature options and adative meshing. But nothing helped. Any suggestion on how to get this working with periodicity? Thanks!
The Sculpt developer at Sandia National Labs was kind enough to reply to this question. Here is his excerpted response.
The periodic option was implemented for a major tire manufacturer a few years ago to support their diatom-based description of the microstructure. For their use-case, we defined many pseudo-random analytic spheres in a periodic domain. I don’t believe we ever considered the .spn format for periodicity, so not sure if it will work. Note that opposite volume fractions (or material ids) should match exactly across periodic boundaries. You might have them try the periodic_check option.
From the image, it looks like periodicity was not used or was ignored in the input, as it appears that the faces of the RVE are flat. When using periodicity, it should be smoothing across periodic boundaries and won’t project to the RVE face. This results in “ragged”-looking boundaries, but the nodes and faces will match exactly across the period. The smooth=fixed_bbox would be ignored in this case.
As for the non-manifold resolution error, I suspect that is unrelated to periodicity. That can happen when there is some very noisy data in the .spn file and the defeaturing can’t fix it. (I’ve also seen it happen when the spn_xyz_order is not set correctly, and the block regions become distorted). One way to check to see if you have your data the way you expect is to use the “stair=fast” option in the input file. That will just create a stair-step exodus file without smoothing so you can verify the structure of the grains quickly.
Another thought – the non-manifold resolution error may be a result of attempting to resolve periodicity on a non-periodic model. When periodic is on, opposite cells will be copied to build ghost cells for smoothing. If the opposite cells are not periodic, that may also result in the noise I mentioned above.
If you have a customer “.spn” model that they think is periodic, and they would be willing to share, I’d be interested to see if it works.
The image below shows a simple example of the initial use case.
He gives some suggestions of some things you might try. If you can provide us with your model, we can see if there are methods of working with the “.spn” data.
The image in my post was generated without the periodicity option. Only When I turn on the periodicity, I run in to the said manifold problem.
To my knowldge the underlying microstructure is periodic. The periodicity check from sculpt was also successful.
I have attached my spn and input file. Curious of the outcome.
Thanks! Dummy_Micro.zip (42.7 KB)
