Voronoi tessellation

[sidharthsarmah]

Hi @GiudGiud ,

I want to create a voronoi tessellation of a steel sample and perform diffusion modelling. It has different types of grains and grain boundaries, I dont find lot of text related to VT in MOOSE, can you please suggest.

[GiudGiud]

Hello

There are plenty of options. See the PropertyReadFile to be able to load data using a voronoi distribution. Then once this data is in a FUnction you can use it to assign subdomains, (through FunctionAux + CoupledVarSubdomainModifier), to assign material properties (using the same PropertyReadFile) etc

[sidharthsarmah]

Thanks @GiudGiud

[sidharthsarmah]

Hi @GiudGiud , I am using both triangular and quadratic elements in MOOSE using TRI3 and QUAD4, but does MOOSE not support using both together?

Error: Exodus requires all elements with a given subdomain ID to be the same type.
Can't write both QUAD4 and TRI3 in the same block!

[GiudGiud]

they just need to be in different subdomains. The MeshRepairGenerator will separate them for you

[sidharthsarmah]

hi @GiudGiud , I could not file the source files, please suggest:

[Mesh]
  
  [./mesh]
    type = FileMeshGenerator
    file = PBf.inp
  [../]

  [./combine]
    type = MeshRepairGenerator
    input = 'mesh'
    separate_blocks_by_element_types = true
  [../]
  
[]

*** ERROR ***
A 'MeshRepairGenerator' is not a registered object.

If you are trying to find this object in a dynamically linked library, make sure that
the library can be found either in your "Problem/library_path" parameter or in the
MOOSE_LIBRARY_PATH environment variable.

[GiudGiud]

you are using an old version of moose, it exists:

https://mooseframework.inl.gov/source/meshgenerators/MeshRepairGenerator.html

[sidharthsarmah]

Hi @GiudGiud , should I just update using this?

conda activate moose
conda update --all

cd ~/projects/moose
git fetch origin
git rebase origin/master

[GiudGiud]

what type of variable is cL?
Can you look at the gradient in an auxiliary variable to see if it s not 0

[sidharthsarmah]

its scalar, cL is concentration in mol/m3. just a rectangular block, with left BC=8e-19 mol/m3 and right BC=0, and check the exit flux.

[GiudGiud]

its scalar

which finite element type / family / order

with left BC=8e-19 mol/m3 and right BC=0, and check the exit flux.

what does the solution look like for cL

[sidharthsarmah]

Hi @GiudGiud , I just defined its this way. I used it but its nearing to zero, as 8e-19 is very small, but still got zero, please suggest:

**Element, type=DC2D3
[Variables]
  [./cL]
   family = LAGRANGE
    order = FIRST
    initial_condition = 0.0
  [../]
[]

[AuxVariables]
  [./grad_cL_x]
    family = MONOMIAL
    order = CONSTANT
  [../]
[]
[AuxKernels]
  [./grad_u_x_aux]
    type = VariableGradientComponent
    variable = grad_cL_x
    component = x
    gradient_variable = cL
  [../]
[]

[GiudGiud]

That's why the SideDiffusiveFlux is 0 as well.

you should look into why cL has the spatial profile it has, that determines the gradient

[sidharthsarmah]

Hi @GiudGiud ,
For simple diffusion problem I used this kernels:

[Kernels]
  [./diff]
    type = ADMatDiffusion
    variable = cL
    diffusivity = D
  [../]

  [./time]
    type = ADHeatConductionTimeDerivative
    variable = cL
    specific_heat = 1
    density_name = 1
  [../]
[]

Now, for a coefficient kernels, what should the kernel look like?

  [./gb_Dgb]
    type = ADParsedMaterial
    block = 0
    expression = '(1/(1+cL)'
    args = 'cL'
    property_name = 'Deff'
  [../]

  [./gb_Sgb]
    type = ADParsedMaterial
    block = 0
    expression = '(1/(1+cL)'
    args = 'cL'
    property_name = 'Seff'
  [../]

Please suggest.

[GiudGiud]

should the derivative be with regards to time in that second equation?

I'd create a new kernel rather than create this scaling coefficient that depends on the variable.
I don't think we have the option for a coefficient inside the derivatives in the existing time derivative and diffusion kernels

[sidharthsarmah]

yes, its with time. thanks.