tesr to mesh

[ThomasChauve]

Hello,
Thanks for the tools you provide. I was trying to do transform a tesr to mesh and I got an error.
My tesr is a custom one build for a VTK file. Therefore it can be no written correctly.
Anyway I got this error :

neper -M CI02_539x495_micro.tesr

========================    N   e   p   e   r    =======================
Info   : A software package for polycrystal generation and meshing.
Info   : Version 4.1.2
Info   : Built with: gsl|muparser|opengjk|openmp|nlopt
Info   : Running on 16 threads.
Info   : <https://neper.info>
Info   : Copyright (C) 2003-2020, and GNU GPL'd, by Romain Quey.
Info   : No initialization file found (`/home/chauvet//.neperrc').
Info   : ---------------------------------------------------------------
Info   : MODULE  -M loaded with arguments:
Info   : [ini file] (none)
Info   : [com line] CI02_539x495_micro.tesr
Info   : ---------------------------------------------------------------
Info   : Reading input data...
Info   :   - Reading arguments...
Info   : Loading input data...
Info   :   - Loading raster tessellation...
Info   :     [i] Parsing file `CI02_539x495_micro.tesr'...
Info   :     [i] Parsed file `CI02_539x495_micro.tesr'.
Info   :   - Reconstructing topology...
Info   :     > Reconstructing 1D mesh... 100%      
Info   :     > Reconstructing 0D mesh... 100%      
Error  : You have discovered a bug in Neper!  Please file an issue at
         https://github.com/rquey/neper/issues.  Thank you.

Here is the input file. I just rename the '.tesr' ' to '.txt' as github was not ok to take 'tesr'
CI02_539x495_micro.txt

Therefore I have a couple of question :

1. Is it possible to give a VTK as input in neper ?
2. Is it possible to build a mesh from a 3D raster file ? (from the doc it seems not possible)
3. Overwise do you have over tools to recommand to mesh a 2D/3D microstructure for an experimental raster image ?

Thanks for the great work.
Best
Thomas Chauve

PS: I am using Ubuntu 20.04 and Neper 4.1.2

[rquey]

You just have to filter your tesr before meshing to solve topological issues, using -transform rmsat,grow:

$ neper -T -loadtesr CI02_539x495_micro.txt -transform "rmsat,grow" -o filtered
$ neper -M filtered.tesr -rcl 0.5 -mesh2dalgo fron,dela
$ neper -V filtered.tesr,filtered.msh -datavoxedgerad 0 -datacellcol id -dataelsetcol id -imagesize 600:600 -cameraangle 12 -showelt1d all -dataelt1drad 0.2 -showtesr 1 -showmesh 0 -print filtered-tesr -showtesr 0 -showmesh 1 -print filtered-mesh

(-transform changes only 3 pixels...)

Neper does not take VTK files as input, and it can mesh a 3D tesr by approximating it as a tess (-morpho tesr:file()), see Quey and Renversade, 2018.

[ThomasChauve]

Hello,

Thanks for your quick answer. It works fine.

I just have one more question about the transforming tesr to tess in a 3D case ?
How do you do that ?

I tried from an home made tesr :
CI02_539x495_micro.txt

First to filter it
neper -T -loadtesr CI02_539x495_micro.txt -transform "rmsat,grow" -o filtered_CI02

Then to transformed it to 'tess' using something like :
[neper -T -loadtesr filtered_CI02.tesr -morpho filtered_CI02.tesr -format tess

But it doesn't work. It give me an empty file. I think I am not using correctly the -morpho option.
filtered_CI02.txt

Here is a visualisation of the filtered tesr file :
neper -V filtered_CI02.tesr -datacellcol id -showmesh 0 -print test

Thanks for your help.
Thomas

[janzenchoi]

I'm pretty new to this, so I am not too sure. The volume is meant to be generated by specifying various statistics. The grains are generated first, then the twins, with each having their own set of statistics to specify.

I am more than happy to share the specifics of my process if it interests you.

[mkasemer]

@rquey: yes, it sounds idealized. I think what Dream3D does is essentially what I suggested:

• builds a 1-level tessellation (grain growth, or whatever), giving a set of grains with corresponding crystallographic orientations,
• then for each grain calculates the twin normal and orientation based on the grain's orientation,
• finally introduces a 2-level tessellation into the first with tightly coupled lamella morphology coupled to the 1-level grain state.

I think this could be automated (certainly using external shell/python scripts), but it is system-dependent - i.e., grains in which the twins appear, the twins that appear (normal and xtal ori), etc. will all depend on the crystal symmetry and the material behavior. Here, I don't know what Dream3D is doing: @jazzzmannn, what aspects do you have to feed it for it to generate this microstructure? At minimum, does it know crystal type / material system?

EDIT: I should say that I find annealing twins deceptively complex. There are many twin variations that can form, and which ones that do form are going to be tightly coupled to the crystal symmetry, geometric state of the parent grains, crystallographic texture, material behavior, and the thermomechanical processing route. That is to say, it seems difficult to make a general rule that says "grain X should definitively have twin Y" (and I'm curious how exactly Dream3D makes this call). I would suggest writing a script that does its best to mimic the character of a witnessed microstructure, in terms of the coupling between parents/twins and the statistics concerning which twin variants are more likely (i.e., if there is a bias in terms of a certain variant(s)).

[rquey]

Thanks @mkasemer.

@jazzzmannn - This is not very different from what Neper does, so, if the next step is to approximate the tesr by a Neper tess, I'd simply suggest to work with Neper from the beginning, and directly generate the tess (using external files for what Neper doesn't do natively).

[janzenchoi]

Hello @mkasemer and @rquey,

Sorry for the late reply. The input for generating the microstructure includes various distributions for the grain sizes, grain sphericities, crystallographic orientations, number of twins (per grain), twin thickness, and the twin fraction (i.e., how much of the grain does the twin section).

I think the best route for me now is what you both suggested, in that I should just use Neper for the entire process. Thank you for answering my questions and for the recommendation.

By the way, is Neper able to generate lamellas that don't fully section the grain?

[rquey]

Yes, apply several widths. Example:

neper -T -n 3::from_morpho -morpho "voronoi::lamellar(w=0.5:0.05:1,pos=start)" -o d
neper -V d.tess -datacellcol "scaleid(1)" -datacelltrs 0.5 -print img

It will take some tricks (using external files) to do what you want.