Finite Strain and in echelon dynamics test

[amassaf]

Hello,

We have trouble working with dynamics in finite strains when using initial conditions on displacement. In fact we can see the correct displacement at the first increment while the strain and stress computations are wrong. This in echelon test works fine in small strain even when coupled with phase field fracture. I wonder if it's because of the strain initialization at the ComputeFiniteStrain ? Thanks for any help.

[selarem]

Any help on this issue ?

[amassaf]

When using an initial condition like :

[Variables]
  [./disp_x]
  [./InitialCondition]
      type = FunctionIC
      function = initial_cond_func_x
    [../]
  [../]
  [./disp_y]
  [./InitialCondition]
      type = FunctionIC
      function = initial_cond_func_y
    [../]
  [../]
[]

[Functions]
[./initial_cond_func_x]
    type = SolutionFunction
    solution = s0
    from_variable = 'disp_x'
[../]
 [./initial_cond_func_y]
    type = SolutionFunction
    solution = s0
    from_variable = 'disp_y'
[../]
[]

[UserObjects]
  [./s0]
    type = SolutionUserObject
    mesh = Finitehalf_elastic_960_out.e #half_elastic_960_out.e
    timestep = LATEST
    execute_on = initial
  [../]
[]

And with the Material block as:

[Materials]
 [./elasticity_tensor]
    type = ComputeIsotropicElasticityTensor
    youngs_modulus = 3.09e9
    poissons_ratio = 0.35
  [../]
[./elasticStrain]
    type = ComputeFiniteStrainElasticStress
  [../]
[]

We got correct displacements but stress and strain are null which is strange.

[amassaf]

could this be due to the initialization in the 'ComputeFiniteStrain.C' or 'ComputeIncrementalStrainBase.C'

[GiudGiud]

Hello

You need to use a proper restart using Checkpoints to restart stateful material properties.

You can't use solution user objects to restart just some variables. Unless you put in more objects from the ProjectedMaterialPropertyStorage (or some name like this) action.

[amassaf]

Ok thanks! I did this using Checkpoint and restart_file_base but I got this error:

*** ERROR ***
The stateful material property 'elastic_strain' was declared in ComputeFiniteStrainElasticStress "elasticStrain" but was not declared in that object on checkpoint.

This is not currently supported due to ambiguity.

Please contact the development team on GitHub if you desire this capability.

I don't even understand what this capability is. Both the original file and the new restart file have a ComputeFiniteStrainElasticStress object.

[GiudGiud]

Does it have exactly the same name? And same suffix/prefix (if those are used) for material properties?

[selarem]

Now the error is :

*** ERROR ***
The stateful material properties in ComputeLinearElasticPFFractureStress "elastic" that are being restarted do not match the stored properties in the same material object from the checkpoint.

Checkpointed stateful properties:
 - hist

Current stateful properties:
 - hist
 - stress

We use the same material in both computations.

[GiudGiud]

Some option must be different.
Can you paste the Materials and GlobalParams blocks in both simulations?

[selarem]

Everything is the same only the Modules:
The first (to import) is quasistatic:

[Physics/SolidMechanics/QuasiStatic]
  [all]
    strain = small 
    add_variables = true
    extra_vector_tags = 'kernel_residual'
    additional_generate_output = 'stress_xx stress_xy stress_yy'
  []
[]

and the main that imports :

[Modules]
  [./TensorMechanics/DynamicMaster]
      [all]
        add_variables = true
        hht_alpha =-0.333
        newmark_beta = 0.44 #0.3025
        newmark_gamma =0.83
        strain = small 
        density = 1180 
        additional_generate_output = 'stress_yy stress_xy stress_xx'
       extra_vector_tags = 'kernel_residual'
      []
  [../]
[]

Some option must be different. Can you paste the Materials and GlobalParams blocks in both simulations?

[GiudGiud]

the executioner is a Transient for both cases too?