Porous Flow Tutorial Page 04. Adding solid mechanics

In this Page, solid mechanics is added to the thermo-hydro simulation of previous Pages. The equations are discussed in governing equations. Only quasi-static solid mechanics is considered here, without gravity, so the equations read (1) As described previously, is the porepressure, the temperature and is the Biot coefficient. The additional nomenclature used here is

• is the effective stress tensor

• is the total stress tensor

• is the elasticity tensor of the drained porous skeleton

• is the linear thermal expansion coefficient. Note that this is the linear version, in contrast to the volumetric coefficients introduced in Page 1.

Once again, before attempting to write an input file, a rough estimate of the expected nonlinear residuals must be performed, as discussed in convergence criteria. The residual for the Eq. (1) is approximately Corresponding to the choice Pa.m made in Page 02 the choice Pa.m may be made here. This means which is significantly greater than for the fluid equation. Therefore, the displacement variables are scaled by .

Many mechanically-related MOOSE objects (Kernels, BCs, etc) accept the use_displaced_mesh input parameter. For virtually all PorousFlow simulations, it is appropriate to set this to false: use_displaced_mesh = false. This means that the Kernel's residual (or BC's residual, Postprocessor's value, etc) will be evaluated using the undisplaced mesh. This has the great numerical advantage that the solid-mechanics elasticity equations remain linear.

Also, many mechanically-related MOOSE objects require the displacements input parameter. Therefore, it is convenient to put this parameter into the GlobalParams block:

[GlobalParams]
  displacements = 'disp_x disp_y disp_z'
  PorousFlowDictator = dictator
  biot_coefficient = 1.0
[]

To model this thermo-hydro-mechanical system, the PorousFlowBasicTHM action needs to be enhanced to read:

[Variables]
  [porepressure]
  []
  [temperature]
    initial_condition = 293
    scaling = 1E-8
  []
  [disp_x]
    scaling = 1E-10
  []
  [disp_y]
    scaling = 1E-10
  []
  [disp_z]
    scaling = 1E-10
  []
[]

[PorousFlowBasicTHM]
  porepressure = porepressure
  temperature = temperature
  coupling_type = ThermoHydroMechanical
  gravity = '0 0 0'
  fp = the_simple_fluid
  eigenstrain_names = thermal_contribution
  use_displaced_mesh = false
[]

The boundary conditions used here are roller boundary conditions, as well as boundary conditions that model the effect of the fluid porepressure on the injection area:

[BCs]
  [constant_injection_porepressure]
    type = DirichletBC
    variable = porepressure
    value = 1E6
    boundary = injection_area
  []
  [constant_injection_temperature]
    type = DirichletBC
    variable = temperature
    value = 313
    boundary = injection_area
  []

  [roller_tmax]
    type = DirichletBC
    variable = disp_x
    value = 0
    boundary = dmax
  []
  [roller_tmin]
    type = DirichletBC
    variable = disp_y
    value = 0
    boundary = dmin
  []
  [roller_top_bottom]
    type = DirichletBC
    variable = disp_z
    value = 0
    boundary = 'top bottom'
  []
  [cavity_pressure_x]
    type = Pressure
    boundary = injection_area
    variable = disp_x
    component = 0
    factor = 1E6
    use_displaced_mesh = false
  []
  [cavity_pressure_y]
    type = Pressure
    boundary = injection_area
    variable = disp_y
    component = 1
    factor = 1E6
    use_displaced_mesh = false
  []
[]

The SolidMechanics module of MOOSE provides some useful AuxKernels for extracting effective stresses of interest to this problem (the effective radial stress and the effective hoop stress)

[AuxVariables]
  [stress_rr]
    family = MONOMIAL
    order = CONSTANT
  []
  [stress_pp]
    family = MONOMIAL
    order = CONSTANT
  []
[]

[AuxKernels]
  [stress_rr]
    type = RankTwoScalarAux
    rank_two_tensor = stress
    variable = stress_rr
    scalar_type = RadialStress
    point1 = '0 0 0'
    point2 = '0 0 1'
  []
  [stress_pp]
    type = RankTwoScalarAux
    rank_two_tensor = stress
    variable = stress_pp
    scalar_type = HoopStress
    point1 = '0 0 0'
    point2 = '0 0 1'
  []
[]

[FluidProperties]
  [the_simple_fluid]
    type = SimpleFluidProperties
    bulk_modulus = 2E9
    viscosity = 1.0E-3
    density0 = 1000.0
    thermal_expansion = 0.0002
    cp = 4194
    cv = 4186
    porepressure_coefficient = 0
  []
[]

[Materials]
  [porosity]
    type = PorousFlowPorosity
    porosity_zero = 0.1
  []
  [biot_modulus]
    type = PorousFlowConstantBiotModulus
    solid_bulk_compliance = 2E-7
    fluid_bulk_modulus = 1E7
  []
  [permeability_aquifer]
    type = PorousFlowPermeabilityConst
    block = aquifer
    permeability = '1E-14 0 0   0 1E-14 0   0 0 1E-14'
  []
  [permeability_caps]
    type = PorousFlowPermeabilityConst
    block = caps
    permeability = '1E-15 0 0   0 1E-15 0   0 0 1E-16'
  []

  [thermal_expansion]
    type = PorousFlowConstantThermalExpansionCoefficient
    drained_coefficient = 0.003
    fluid_coefficient = 0.0002
  []
  [rock_internal_energy]
    type = PorousFlowMatrixInternalEnergy
    density = 2500.0
    specific_heat_capacity = 1200.0
  []
  [thermal_conductivity]
    type = PorousFlowThermalConductivityIdeal
    dry_thermal_conductivity = '10 0 0  0 10 0  0 0 10'
    block = 'caps aquifer'
  []

  [elasticity_tensor]
    type = ComputeIsotropicElasticityTensor
    youngs_modulus = 5E9
    poissons_ratio = 0.0
  []
  [strain]
    type = ComputeSmallStrain
    eigenstrain_names = thermal_contribution
  []
  [thermal_contribution]
    type = ComputeThermalExpansionEigenstrain
    temperature = temperature
    thermal_expansion_coeff = 0.001 # this is the linear thermal expansion coefficient
    eigenstrain_name = thermal_contribution
    stress_free_temperature = 293
  []
  [stress]
    type = ComputeLinearElasticStress
  []
[]

[Preconditioning]
  active = basic
  [basic]
    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      lu           NONZERO                   2'
  []
  [preferred_but_might_not_be_installed]
    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 = 1E6
  dt = 1E5
  nl_abs_tol = 1E-15
  nl_rel_tol = 1E-14
[]

[Outputs]
  exodus = true
[]

Finally, some mechanics-related Materials need to be defined

  [elasticity_tensor]
    type = ComputeIsotropicElasticityTensor
    youngs_modulus = 5E9
    poissons_ratio = 0.0
  []
  [strain]
    type = ComputeSmallStrain
    eigenstrain_names = thermal_contribution
  []
  [thermal_contribution]
    type = ComputeThermalExpansionEigenstrain
    temperature = temperature
    thermal_expansion_coeff = 0.001 # this is the linear thermal expansion coefficient
    eigenstrain_name = thermal_contribution
    stress_free_temperature = 293
  []
  [stress]
    type = ComputeLinearElasticStress
  []
[]

An animation of the results is shown in Figure 1.

Figure 1: Displacement (magnified by 100 times) and effective hoop-stress evolution in the borehole-aquifer-caprock system.

The dynamics of this model are fascinating, and readers are encouraged to pause and play with parameters to explore how they effect the final result. In fact, this model is very similar to the "THM Rehbinder" test in PorousFlow's test suite. Rehbinder (Rehbinder, 1995) derived analytical solutions for a similar THM problem, and MOOSE replicates his result exactly:

Figure 2: Comparison between MOOSE and Rehbinder's analytical solution.

Figure 3: Comparison between MOOSE and Rehbinder's analytical solution.

Figure 4: Comparison between MOOSE and Rehbinder's analytical solution.