Clarification regarding extraction of data for every crystallite.

[nitheeshkumar11]

I am working on a simulation involving two phases. I am currently able to obtain the stress and strain outputs for each of the phases. However, I am not able to obtain the data for the entire material or for each crystallite/grain. How can I proceed?

Thanks in advance.

[eisenlohr]

Let's suppose that your two phases are called 'A' and 'B', the material configuration material.yaml, the grid grid.vti, and the resulting file out.hdf5.

I assume you already do something along these lines to calculate the Cauchy stress (from F and P that need to be present in the output file) :

import numpy as np
import damask

r = damask.Result('out.hdf5')
r.add_stress_Cauchy()
r.add_equivalent_Mises('sigma')

The following code condenses the two Cauchy stress arrays of phases A and B (at increment 10):

cauchy = r.view(increments=[10]).get('sigma')      # dictionary of each phase
all = np.concatenate((cauchy['A'],cauchy['B']))
avg = np.average(all,axis=0)

To generate a phase-specific list of each grain (at exemplary increment 10), one would need to place the data onto the underlying grid points:

from collections import defaultdict

materialID = damask.VTK.load('grid.vti').get('material')
cfg = damask.ConfigMaterial.load('material.yaml')

cauchy = r.view(increments=[10]).place('sigma')       # tensor-valued at every grid point

grainCauchy = defaultdict(dict)
for g,m in enumerate(cfg['material']):
    myPhase = m['constituents][0]['phase']
    grainCauchy[myPhase][g] = cauchy[materialID == g]

[nitheeshkumar11]

Thanks a lot.
I was thinking this has something to do with the N_constituents command in the material.yaml file. Can you also please clarify what the command actually does?
What does 'N_constituents 1' mean?

[eisenlohr]

DAMASK, at its core, is a "material point model", meaning that it allows describing the constitutive behavior of the discretization points in the overall boundary/initial value problem. Depending on the scale of that "outer" problem, those points might be representing a bunch of different phases or grains, instead of being a part of a (highly resolved) grain. For the former case, there are different strategies for homogenizing the behavior of individual constituents present within one material point. The variable N_constituents specifies how many separate entities are considered to make up a specific material point. In the case that you are likely dealing with, the spatial resolution is large enough so that every material point corresponds to a volume that only contains a single phase (i.e. it is a subvolume of a particular grain). Hence, N_constituents: 1 and any data split by constituent is addressed as x['constituents'][0] since there is only one entry, located at index 0.

[nitheeshkumar11]

Thank you sir.

[MarDiehl]

@nitheeshkumar11: Did this answer your question? If so, please mark it as solved.