Steady-state simulation using THM

[NorthMagic]

Hello, I encountered a problem at run time with "Vector tag 'TIME' does not exist. This may occur if you have a TimeKernel in your problem but did not specify a transient executioner". When I simulated a single flow channel without junctions, THM using the Steady executioner worked without error. But for a singlel flow channel with junctions, THM using the Steady executioner failed to run.

Is it unsupported to run THM with the Steady executioner for a model including junctions? Do I have to simulate this steady-state problem using the Transient executioner by setting steady_state_dection=true?

[GiudGiud]

Hello

A junction you are using must have a time kernel.

Either:

Find which junction that is and add code to prevent adding the time kernel in a steady solve

Use a transient executioner and let it run to steady state. It's the same as a steady solve in terms of results

[GiudGiud]

I think so

[NorthMagic]

I made a small change to addMooseObjects, just let it omit the addition of ODETimeDerivative. For a single flow channel with junctions by using the Steady executioner, the mentioned problem was solved. But I encountered a new problem. THM got an error with the following:
0 Nonlinear |R| = 2.295593e+02
Linear solve did not converge due to DIVERGED_PC_FAILED iterations 0
PC failed due to FACTOR_NUMERIC_ZEROPIVOT
Nonlinear solve did not converge due to DIVERGED_FNORM_NAN iterations 0
Solve Did NOT Converge!
Aborting as solve did not converge

I don't know what problem it is. Could you please give me some advice?

[GiudGiud]

Can you try adding -pc_factor_shift NONZERO to your petsc options?

There is a troubleshooting page on our website for convergence issues

[joshuahansel]

Also, maybe it could be your initial guess? I know I've tried Steady simulations with THM before and diverged with a bad initial guess, like the velocity being zero.

[NorthMagic]

This unconverged problem was solved by using improved petsc options, which are recommended by the Porous_Flow module. Here is the url: https://mooseframework.inl.gov/modules/porous_flow/solvers.html. Thank you two very much!