How to Restart with the Tensor Mechanics Master Action System

[DaijunHU]

Hello!

I am currently working on a two-step solid mechanics simulation, where the stress distribution obtained from the first simulation is mapped to a second model for further computation (both models share identical meshes). I am encountering two issues and would appreciate your assistance:

1. In the first simulation, I used [Modules/TensorMechanics/Master/all] as shown below. However, I am not sure how to carry over the stress components into the second simulation using something similar to initial_from_file_var.
   Is it necessary to record all stress components in AuxVariables and AuxKernels in the first simulation to achieve this restarting?
   Sorry that I couldn’t find clear parameters in the documentation for the Tensor Mechanics Master Action System or Restarting and recovering from previous simulations.
   I wanted to try with initial_from_file_var and initial_from_file_timestep in this block, but the error below actually becomes an obstacle.

2. While running the 2nd simulation using the the following .i file, I encountered an error that left me puzzled:

OnlyMicro80nmRestart.i:34: missing required parameter 'Modules/TensorMechanics/Master/displacements'
Doc String: "The nonlinear displacement variables for the problem."

I actually specified displacements = 'ux uy uz' as shown by the attached part of my .i file, but the error persists. (I should make it clear that I did not set the displacements in my 1st simulation, so when the error occurs I added this line, but still the same error.)

Thank you for your help in advance and look forward to your reply!

Best regards,
Daijun

> [Mesh]
>     file = OnlyMicro80nm_out.e
>     distribution = serial
> []
> 
> [Problem]
>   restart_file_base = OnlyMicro80nm_outcp/LATEST
>   force_restart = false
> []
> 
> [Variables]
>   [./ux]
>     order = FIRST
>     family = LAGRANGE
>     initial_from_file_var = ux
>     initial_from_file_timestep = 635
>   [../]
> 
>   [./uy]
>     order = FIRST
>     family = LAGRANGE
>     initial_from_file_var = uy
>     initial_from_file_timestep = 635
>   [../]
> 
>   [./uz]
>     order = FIRST
>     family = LAGRANGE
>     initial_from_file_var = uz
>     initial_from_file_timestep = 635
>   [../]
> []
> 
> [Modules/TensorMechanics/Master/all]
>   strain = FINITE
>   displacements = 'ux uy uz'
>   add_variables = true
>   generate_output = 'stress_xx stress_yy stress_zz stress_xy stress_xz stress_yz strain_xx strain_yy strain_zz strain_xy strain_xz strain_yz'
> []
> 
> [AuxVariables]
>   [./slip_increment_1]
>    order = FIRST
>    family = MONOMIAL
>     initial_from_file_var = slip_increment_1
>     initial_from_file_timestep = 635
>   [../]
> []

[GiudGiud]

Hello

The stresses are material properties which are not restarted using ICs or using initial_from_file_var parameters.
Your best bet is to use Checkpoint restart
https://mooseframework.inl.gov/application_usage/restart_recover.html

This second error will likely go away if you switched your syntax to the Physics syntax.
If it does not, it's asking for diplacements to be nested under Master/ instead of under Master/all

Guillaume

[DaijunHU]

Hi Guillaume,

Thank you for your reply!
I will try the Physics syntax for the 2nd error.

I am still a bit confused about the Tensor Mechanics Master Action. On the webpage https://mooseframework.inl.gov/syntax/Modules/TensorMechanics/Master/index.html, it is mentioned that it generates aux variables and auxkernels to aid in the output of tensor components. So I think those stress components I added in generate_output = 'stress_xx strain_xx' are not material properties, but a kind of AuxVariables and AuxKernels calculated based on the solid mechanics equations, which should follow the restarting methods of AuxVariables. If I use Checkpoint restart, will all these outputs be automatically inherited into new simulations from the checkpoint file?

If so, is strain_xx inheritable? Because I have chosen in Variables to let the displacement inherit the previous simulation, so the strain should be recalculated based on the displacement field at the beginning of the second simulation?
I'm a bit confused about this, before the Tensor Mechanics Master Action setting came out, I thought my strategy would be to create AuxVaribales and AuxKernels for each stress component and use initial_from_file_var, but not do this for strain_xx since the displacement field is already inherited . Now after using Master/all I am a little confused about this.
I hope you can clear it up for me, thank you very much!

Best regards,
Daijun

[GiudGiud]

these auxiliary variables are only there for visualization purposes. In the equations, the stress/strain material properties are used

[GiudGiud]

so the strain should be recalculated based on the displacement field at the beginning of the second simulation?

it probably should but it's not that simple because there are often history effects (plasticity!) in solid mechanics. So the material properties are not restarted that way. They are restarted properly only using Checkpoint

[DaijunHU]

Thank you for your answer!
If displacements and strain fields are inherited separately, that is, the strain field is inherited as a material property, and the displacement field is inherited manually in Variables and cannot reflect the historical effect, are there any incompatible parts among them?
The displacement field cannot reflect whether there is plasticity in it, but it should still be able to make the material enter the plastic segment according to the material constitutive model we set, becaue the calculation is often driven by the displacements. Is this a kind of duplication or conflict with the inheritance of the strain field?

[GiudGiud]

if you use checkpoint to inherit both, displacements and strain will be consistent.
If you use checkpoint for strain and set the displacements differently, using ICs or another file (with the SolutionUO for example), then they wont be consistent and the results will be affected.