How to implement non-uniformity simulation of materials?

[752096822]

How to implement non-uniformity simulation of materials?
The parameters such as the elastic modulus of the current model are fixed values. How should I handle the physical parameters changing with the coordinates?

[oldninja]

I have the same question. I thought that one way would be to assign many blocks of materials with appropriate values- Painful but doable.
I could help with the implementation if the development team is interested and can point me in their preferred direction.

[752096822]

For example, how do I implement the following two situations:

1. The material's elastic modulus or other mechanical parameters vary with the coordinate axis (for example, some function gradient materials)
2. How to achieve the overall non-uniformity of the Damage Critical Value (for example, it follows a Weibull distribution at different points)

[JTHesse]

For parameters that are dependent on values like x,y or t we can utilize Meta.parse, as we did in the eval_bc function.

Alternatively, you can also provide the values via the mesh_file for each node. I'm not sure if we should introduce a reference variable:

Mesh Input:

header: x y block_id volume MY_CRIT_VALUE
0.0 0.0 1 1.0E-02 2
0.1 0.0 1 1.0E-02 3
0.1 0.1 1 1.0E-02 4
...

Input deck:

PeriLab:
  Physics:
      Damage Models:
        Damage:
          Critical Value: MY_CRIT_VALUE
          Damage Model: Critical Energy

What is your opinion @CWillberg?

[oldninja]

For example, how do I implement the following two situations:

1. The material's elastic modulus or other mechanical parameters vary with the coordinate axis (for example, some function gradient materials)

As I understand the code right now, you can try to use the approach as shown in this example: examples/RVE/RVE_abaqus.yaml
which requires some pre-processing before setting up the input file on your end. @JTHesse or @CWillberg please correct me if I'm wrong. It is doable, just need to do some preprocessing.

2. How to achieve the overall non-uniformity of the Damage Critical Value (for example, it follows a Weibull distribution at different points)

I think you can follow the same approach as 1.

[oldninja]

For parameters that are dependent on values like x,y or t we can utilize Meta.parse, as we did in the eval_bc function.

That is what I was looking at over the weekend. I think that is a good idea. Would it be possible to do it for the damage models as well?
The alternative approach you suggest would be good too.

[JTHesse]

@oldninja yes you are right, you could use something similar as the RVE example. But this would be, as you already pointed out, very messy.

That is what I was looking at over the weekend. I think that is a good idea. Would it be possible to do it for the damage models as well?

Yes, should be straightforward! We are going to implement this asap.

[CWillberg]

@oldninja write an issue. I will implement it tomorrow in the train.

[oldninja]

Will do. Thank you @CWillberg and @JTHesse !
Now it is done- should be feature request #143