Save the elastic energy density at each nodal as a auxvariable and then transfer it into subapp.

[Yaode-Yin]

Hello everyone,
I try to implement the phase field fracture model via multiapp method. In the main app, the elastic problem is solved and then I want to save the elastic energy density at each nodal in a aux variable, so that the aux variable is finally transferred into the phase field app.
My question is that how can I save the elastic energy density calculated by ComputeCrackedStress as a aux variable correctly?

[GiudGiud]

Hello

Is it a material property? You can project material properties onto auxiliary variables using auxkernels. This one for "Real" valued material properties
https://mooseframework.inl.gov/source/auxkernels/MaterialRealAux.html
there is a different auxkernel for every type of material property

Auxvariables can then be transferred. Note that you will suffer a projection error most likely between the material property and the auxiliary variable, unless the material property is actually represented exactly in the target finite element basis

Guillaume

[GiudGiud]

for nodal points use a nodal variable, like a first order lagrange.

For Gaussian points, what I mentioned above applies. You project the material property to a variable, then one can evaluate that variable at the Gauss points. They will only match rigorously if the projection step was exact because the material property was actually represented exactly by the finite element family & order you picked for the auxvariable

[Yaode-Yin]

When I try to set elastic_energy as a first order lagrange variable, I receive the following errors.

Nodal AuxKernel 'elastic_energy' attempted to reference material property 'property' Consider using an elemental auxiliary variable for 'elastic_energy'
I don't know why it happens. How can I fix it?

[GiudGiud]

material properties are not defined at nodes.
So a direct projection wont be possible. You can try to make this work
https://mooseframework.inl.gov/source/auxkernels/ProjectedMaterialPropertyNodalPatchRecoveryAux.html

otherwise you can project onto a higher order monomial first, then use the ProjectionAux to project again onto a nodal variable

There's no real reason to use a nodal variable here imo, since material properties are using on quadrature points inside the element not at nodes

[Yaode-Yin]

Yes, you're right! Thanks a lot!
But I still don't know if it is possible to transfer the positive part of the elastic strain energy density at the quadrature points. I think that's what I really need. Can I implement that in the present Transfer system?

[GiudGiud]

Transferring material properties directly, not going through an auxiliary variable, would be a lot of work