issues in executing code for Dynamic Phase Field Fracture using Explicit Time Integration Scheme.

[zuckarm]

hi all,

i am working on a dynamic fracture problem: a 2d notched plate is subjected to traction on its top and bottom edges. i am using the MultiApps system to separate the solid mechanics problem in the main app, and phase field problem in the sub app. i want to solve the problem using explicit time integration scheme.

i am facing the following errors when i run the code:

Time Step 0, time = 0
Currently Executing
  Finished Performing Initial Setup                                                      [  7.32 s] [  460 MB]

Time Step 0, time = 0

Time Step 1, time = 5e-09, dt = 5e-09
Initial fixed point residual norm: 0.000000e+00

Beginning fixed point iteration 0


Main app solve:

Performing automatic scaling calculation

The 1-th row of A is exactly zero
The 1-th row of A is exactly zero
The 1-th row of A is exactly zero
The 1-th row of A is exactly zero
  Linear solve did not converge due to DIVERGED_PC_FAILED iterations 0
                 PC failed due to FACTOR_NUMERIC_ZEROPIVOT 
 Solve Did NOT Converge!
  Finished Solving                                                                       [  9.26 s] [  813 MB]
Fixed point convergence reason: DIVERGED_NONLINEAR
Aborting as solve did not converge

i have tried a number of edits to the code, but the initial fixed point residual norm is always reported as zero, whether i use an explicit or implicit scheme.
while using an implicit scheme, the nonlinear residuals are zero at all timesteps, due to which the code progresses but it reports values of all variables: displacement, phase field as zero for all timesteps.

can anyone help me in resolving this issue?

problem data:
upload.zip

[GiudGiud]

Hello

is this transfer in the right direction?

	
	[./to_psie_active]
		type			= MultiAppCopyTransfer
		from_multi_app 	= 'fracture'
		variable		= psie_active
		source_variable = psie_active
	[../]

shouldnt it be to_multi_app?

You should make sure both inputs are valid with aritifical data from the other input (setting the transferred auxiliary variables using initial conditions for example)

[zuckarm]

yes, it must be "to_multi_app". i made the changes, but i am still facing same error as before.
i have changed preconditioning too, but i am still getting initial residual as zero.
what could be the reason for initial residual to be zero?

[zuckarm]

i have made a few more changes, and now the code seems to work fine for implicit scheme. however, when i go for explicit scheme, i get the following error:
Initial fixed point residual norm: inf
i have tried reducing the timestep, but to no avail. can anyone please suggest any alternatives?

[GiudGiud]

you have to go kernel by kernel (and BCs too) and check which one is returning the inf number
This interface, which kernels use, will let you save their contribution to the residual in an auxiliary variable
https://mooseframework.inl.gov/source/interfaces/TaggingInterface.html