Recompute grain statistics after "quick surface mesh" and "laplacian smoothing"

[stopk...@gmail.com]

Hi Mike,

I have a DREAM.3D microstructure that I convert to a triangle geometry, then smooth using "Laplacian Smoothing", then export .stl files for each grain to mesh using gmsh and simulate using ABAQUS using tetrahedral elements. I would like to compute the "updated" grain statistics, essentially a fresh version of a features.csv file that considers the smoothened grain boundaries. I realize that properties will either not change at all (e.g., grain orientations), change slightly (volume, shared area between grains with a boundary normal to one of the three global directions, since voxelated), or change significantly (shared area between grains with a smooth vs. voxelated boundary at 45 degrees to one of the three global directions).

I use a Python script to extract a .vtk file of the triangle geometry from the ABAQUS output file and I can likely further modify into a format that DREAM.3D can read. My question is whether DREAM.3D is capable of computing updated statistics on triangular geometry data. If not, I think I can manage this in some Python scripts...

Best regards,

Kris

[Michael Jackson]

Hello Kris,

    I don’t think dream3d is entirely capable of this. You might be able to read the triangle mesh back into DREAM.3D and then use the “Sample Triangle Geometry to Regular Grid” filter to get the data back onto an Image Geometry. As long as you can bring the “Face Labels” back in order to convert those to FeatureIds then I think you can do the round trip.

 

--

Michael Jackson | Owner, President

      BlueQuartz Software

[e] mike.j...@bluequartz.net

[w] www.bluequartz.net

--
You received this message because you are subscribed to the Google Groups "dream3d-users" group.
To unsubscribe from this group and stop receiving emails from it, send an email to dream3d-user...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/dream3d-users/1d526b35-51bd-48e0-b50e-7dbd93babee6n%40googlegroups.com.

[stopk...@gmail.com]

Hey Mike,

I would prefer not to sample the tetrahedral data onto an image grid since I already have the microstructure model in a voxelated format. I will try to see if I can read in the data into DREAM.3D and then export the "Face Labels".

Best regards,

Kris

[Sean Donegan]

Hi Kris,

Some of the statistics can be computed directly from the triangle mesh in DREAM.3D. In particular, the filter Find Feature Volumes from Triangle Geometry will compute feature volumes directly from the surface mesh representation, while the filter Find Feature Shapes from Triangle Geometry will compute the shape statistics (omega3s, axis lengths, axis Euler angles, and aspect ratios). I wrote these filters a long time ago, and while I recall them working let me know if they give you strange results. 

If you want to compute the areas of all grain boundaries, you can first uniquely identify each boundary by running Generate Triangle Face Ids (which will generate a feature Id for the grain boundaries) and then running the normal Find Feature Sizes filter (which should now work on any geometry, including triangles). 

There's a few other tricks you can play to compute statistics directly on the surface mesh, so let me know if you have any other ideas. Note that the above are for working on the surface mesh, so you might want to compute them before volume meshing. 

hope that helps,

-Sean

To view this discussion on the web visit https://groups.google.com/d/msgid/dream3d-users/faa4ed54-1789-4641-8238-d9bcc012d864n%40googlegroups.com.