Convergence problem for porous_flow

[robo-boluo]

Greetings,

I am encountering convergence challenges in a seemingly straightforward porous flow problem. The solver throws either a 'Nonlinear solve did not converge due to DIVERGED_LINE_SEARCH' or a 'Nonlinear solve did not converge due to DIVERGED_MAX_IT iterations 50' error.

Has anyone encountered and successfully resolved this issue before? I would greatly appreciate any insights or solutions you can offer.

Thank you,

Following is my code and inside injection.bh is '1 0 0 0'.

_injection_rate = -0.10 # kg/s/m, negative because injection as a source

[Mesh]
  [gen]
    type = GeneratedMeshGenerator
    dim = 2
    nx = 14 
    ny = 8  
    xmin = -35
    xmax = 35
    ymin = -20
    ymax = 20
  []
  [injection_node]
    input = gen
    type = ExtraNodesetGenerator
    new_boundary = injection_node
    coord = '0 0 0'
  []
[]

[GlobalParams]
  PorousFlowDictator = dictator
  gravity = '0 0 0'
[]

[Variables]
  [f_Al]
    initial_condition = 3.07E-08
  []
  [f_Ca]
    initial_condition = 1.98E-06
  []
  [f_Cl]
    initial_condition = 2.26E-04
  []
  [f_Fe]
    initial_condition = 1.98E-07
  []
  [f_H]
    initial_condition = 2.53E-04
  []
  [f_K]
    initial_condition = 1.02E-05
  []
  [f_Mg]
    initial_condition = 9.87E-08
  []
  [f_Na]
    initial_condition = 2.43E-04
  []
  [f_O2]
    initial_condition = 1.74E-05
  []
  [f_SiO2]
    initial_condition = 1.01E-05
  []
  [f_HCO3]
    initial_condition = 1.56E-02
  []
  [f_SO4]
    initial_condition = 1.8983E-10
  [] 
  [porepressure]
    initial_condition = 1E6
  []
  [temperature]
    initial_condition = 40
    scaling = 1E-6 # fluid enthalpy is roughly 1E6
  []
[]

[BCs]
  [injection_temperature]
    type = MatchedValueBC
    variable = temperature
    v = injection_temperature
    boundary = injection_node
  []
[]

[DiracKernels]
  [inject_H2O]
    type = PorousFlowPolyLineSink
    SumQuantityUO = injected_mass
    fluxes = ${injection_rate}
    p_or_t_vals = 0.0
    line_length = 1.0
    multiplying_var = injection_rate_massfrac_H2O
    point_file = injection.bh
    variable = porepressure
  []
  [inject_H]
    type = PorousFlowPolyLineSink
    SumQuantityUO = injected_mass
    fluxes = ${injection_rate}
    p_or_t_vals = 0.0
    line_length = 1.0
    multiplying_var = injection_rate_massfrac_H
    point_file = injection.bh
    variable = f_H
  []
  [inject_Cl]
    type = PorousFlowPolyLineSink
    SumQuantityUO = injected_mass
    fluxes = ${injection_rate}
    p_or_t_vals = 0.0
    line_length = 1.0
    multiplying_var = injection_rate_massfrac_Cl
    point_file = injection.bh
    variable = f_Cl
  []
  [inject_Na]
    type = PorousFlowPolyLineSink
    SumQuantityUO = injected_mass
    fluxes = ${injection_rate}
    p_or_t_vals = 0.0
    line_length = 1.0
    multiplying_var = injection_rate_massfrac_Na
    point_file = injection.bh
    variable = f_Na
  []
  [inject_SO4]
    type = PorousFlowPolyLineSink
    SumQuantityUO = injected_mass
    fluxes = ${injection_rate}
    p_or_t_vals = 0.0
    multiplying_var = injection_rate_massfrac_SO4
    point_file = injection.bh
    variable = f_SO4
  []
  [inject_Mg]
    type = PorousFlowPolyLineSink
    SumQuantityUO = injected_mass
    fluxes = ${injection_rate}
    p_or_t_vals = 0.0
    line_length = 1.0
    multiplying_var = injection_rate_massfrac_Mg
    point_file = injection.bh
    variable = f_Mg
  []
  [inject_Ca]
    type = PorousFlowPolyLineSink
    SumQuantityUO = injected_mass
    fluxes = ${injection_rate}
    p_or_t_vals = 0.0
    line_length = 1.0
    multiplying_var = injection_rate_massfrac_Ca
    point_file = injection.bh
    variable = f_Ca
  []
  [inject_K]
    type = PorousFlowPolyLineSink
    SumQuantityUO = injected_mass
    fluxes = ${injection_rate}
    p_or_t_vals = 0.0
    line_length = 1.0
    multiplying_var = injection_rate_massfrac_K
    point_file = injection.bh
    variable = f_K
  []
  [inject_HCO3]
    type = PorousFlowPolyLineSink
    SumQuantityUO = injected_mass
    fluxes = ${injection_rate}
    p_or_t_vals = 0.0
    line_length = 1.0
    multiplying_var = injection_rate_massfrac_HCO3
    point_file = injection.bh
    variable = f_HCO3
  []
  [inject_SiO2]
    type = PorousFlowPolyLineSink
    SumQuantityUO = injected_mass
    fluxes = ${injection_rate}
    p_or_t_vals = 0.0
    line_length = 1.0
    multiplying_var = injection_rate_massfrac_SiO2aq
    point_file = injection.bh
    variable = f_SiO2
  []

  [inject_Al]
    type = PorousFlowPolyLineSink
    SumQuantityUO = injected_mass
    fluxes = ${injection_rate}
    p_or_t_vals = 0.0
    line_length = 1.0
    multiplying_var = injection_rate_massfrac_Al
    point_file = injection.bh
    variable = f_Al
  []
  [inject_Fe]
    type = PorousFlowPolyLineSink
    SumQuantityUO = injected_mass
    fluxes = ${injection_rate}
    p_or_t_vals = 0.0
    line_length = 1.0
    multiplying_var = injection_rate_massfrac_Fe
    point_file = injection.bh
    variable = f_Fe
  []
  [inject_O2]
    type = PorousFlowPolyLineSink
    SumQuantityUO = injected_mass
    fluxes = ${injection_rate}
    p_or_t_vals = 0.0
    line_length = 1.0
    multiplying_var = injection_rate_massfrac_O2aq
    point_file = injection.bh
    variable = f_O2
  [] 

  
[]

[UserObjects]
  [injected_mass]
    type = PorousFlowSumQuantity
  []
  
[]

[Postprocessors]
  [dt]
    type = TimestepSize
    execute_on =  'timestep_begin'
  []
  [tot_kg_injected_this_timestep]
    type = PorousFlowPlotQuantity
    uo = injected_mass
  []
  
[]


[FluidProperties]
  [the_simple_fluid]
    type = SimpleFluidProperties
    thermal_expansion = 0
    bulk_modulus = 2E9
    viscosity = 1E-3
    density0 = 1000
    cv = 4000.0
    cp = 4000.0
  []
[]

[PorousFlowFullySaturated]
  coupling_type = ThermoHydro
  porepressure = porepressure
  temperature = temperature
  mass_fraction_vars =     'f_Al    f_Ca    f_Cl    f_Fe    f_H    f_K    f_Mg    f_Na    f_O2    f_SiO2    f_SO4     f_HCO3'
  save_component_rate_in = 'rate_Al rate_Ca rate_Cl rate_Fe rate_H rate_K rate_Mg rate_Na rate_O2 rate_SiO2 rate_SO4 rate_HCO3 rate_H2O' # change in kg at every node / dt
  fp = the_simple_fluid
  temperature_unit = Celsius
  pressure_unit = Pa 
[]

[AuxVariables]
  [injection_temperature]
    initial_condition = 30
  []
  [injection_rate_massfrac_H2O]
    initial_condition =0.963547651
  []
  [injection_rate_massfrac_H]
    initial_condition =2.29484E-05
  []
  [injection_rate_massfrac_Cl]
    initial_condition = 0.019342814
  []
  [injection_rate_massfrac_Na]
  initial_condition = 0.010748178
  []
  [injection_rate_massfrac_SO4]
  initial_condition = 0.002703787
  []
  [injection_rate_massfrac_Mg]
  initial_condition = 0.001288594
  []
  [injection_rate_massfrac_Ca]
  initial_condition = 0.000409535
  []
  [injection_rate_massfrac_K]
    initial_condition = 0.000399504
  []
  [injection_rate_massfrac_HCO3]
    initial_condition = 0.001531021
  []
  [injection_rate_massfrac_SiO2aq]
    initial_condition = 5.96563E-06
  [] 
  [injection_rate_massfrac_Al]
    initial_condition = 2.6009E-10
  [] 
  [injection_rate_massfrac_Fe]
    initial_condition = 5.38341E-10
  [] 
  [injection_rate_massfrac_O2aq]
    initial_condition = 3.08455E-10
  [] 
  [rate_Al]
  []
  [rate_Ca]
  []
  [rate_Cl]
  []
  [rate_Fe]
  []
  [rate_H]
  []
  [rate_HCO3]
  []
  [rate_K]
  []
  [rate_Mg]
  []
  [rate_Na]
  [] 
  [rate_O2]
  []
  [rate_SiO2]
  []
  [rate_SO4]
  []
  [rate_H2O]
  []
[]

[Materials]
  [porosity]
    type = PorousFlowPorosityConst # this simulation has no porosity changes from dissolution
    porosity = 0.2
  []
  [permeability]
    type = PorousFlowPermeabilityConst
    permeability = '1E-10 0 0   0 1E-10 0   0 0 1E-10'
  []
  [thermal_conductivity]
    type = PorousFlowThermalConductivityIdeal
    dry_thermal_conductivity = '2.3 0 0  0 2.3 0  0 0 2.3'
  []
  [rock_heat]
    type = PorousFlowMatrixInternalEnergy
    density = 2900.0
    specific_heat_capacity = 950.0
  []
[]

[Preconditioning]
  active = typically_efficient
  [typically_efficient]
    type = SMP
    full = true
    petsc_options_iname = '-pc_type -pc_hypre_type'
    petsc_options_value = ' hypre    boomeramg'
  []
  [strong]
    type = SMP
    full = true
    petsc_options = '-ksp_diagonal_scale -ksp_diagonal_scale_fix'
    petsc_options_iname = '-pc_type -sub_pc_type -sub_pc_factor_shift_type -pc_asm_overlap'
    petsc_options_value = ' asm      ilu           NONZERO                   2'
  []
  [probably_too_strong]
    type = SMP
    full = true
    petsc_options_iname = '-pc_type -pc_factor_mat_solver_package'
    petsc_options_value = ' lu       mumps'
  []
[]

[Executioner]
  type = Transient
  solve_type = Newton
  end_time = 7.76E6 # 90 days
  dt = 1E5
[]

[Outputs]
  exodus = true
  csv=true
[]_

[GiudGiud]

Hello

There are troubleshooting instructions for moose in general here
https://mooseframework.inl.gov/moose/application_usage/failed_solves.html
and some instructions to try for porous flow here
https://mooseframework.inl.gov/modules/porous_flow/tutorial_02.html

Guillaume

[1runer]

You don't have gravity and no fluid boundary conditions. So the simulation won't converge to any result.

[1runer]

I'm aware that it should work. But defining a boundary condition makes the convergence much easier.

Does the example file run standalone without MultiApp coupling to the Geochemistry module (as in the original examples)?
Unfortunately, the PorousFlow-Geochemical examples are not straightforward and easy to reproduce. Your non-convergence looks to me like non-physical values in any of the input files.

I guess more experts on the PorousFlow-Geochemistry module can help more than me.

[robo-boluo]

Yes, the example can run standalone by commenting the MultiApp of geochemistry coupling.
I will double check the input file. Many thanks.

[cpgr]

I'll try and take a look today

[cpgr]

Well, the pressure blows up as you are injecting mass but not allowing any to leave with a BC, but I think that the default tolerance is too tight which is the main issue. Try setting something looser like 'nl_rel_tol = 1e-6' or 'nl_abs_tol = 1e-8' etc

[robo-boluo]

This problem has been solved after adding producer and tolerance. Dear Chris Green @cpgr , Appreciate your quick and professional guidance.