Mortar interface flux match

[srinath-chakravarthy]

A question for all the mortar experts. I am trying to mimic the interfacefluxmatch with mortar. For some reason i am encountering convergence issues with this. I am using ADMortar. Please see below my implementation. Any help will be appreciated.

/*
 * To change this license header, choose License Headers in Project Properties.
 * To change this template file, choose Tools | Templates
 * and open the template in the editor.
 */

/* 
 * File:   EqualNormalFluxConstraint.C
 * Author: srinath
 * 
 * Created on January 28, 2021, 1:21 PM
 */

#include "EqualNormalFluxConstraint.h"

#include "SubProblem.h"
#include "FEProblem.h"

registerMooseObject("electro_chemo_mechApp", EqualNormalFluxConstraint);

InputParameters
EqualNormalFluxConstraint::validParams()
{
  InputParameters params = ADMortarConstraint::validParams();
  params.addClassDescription(
      "EqualNormalFluxConstraint enforces continuity of a gradient component between secondary and "
      "primary sides of a mortar interface using lagrange multipliers");
  params.addParam<MaterialPropertyName>(
      "primary_mat_prop",
      "diffusivity",
      "The material property name providing the quantity to equilibrate on the primary side");
  params.addParam<MaterialPropertyName>(
      "secondary_mat_prop",
      "diffusivity",
      "The material property name providing the quantity to equilibrate on the secondary side");
  params.addParam<bool>("primary_tensor", false, "Is the material property tensor_valued");
  params.addParam<bool>("secondary_tensor", false, "Is the material property tensor_valued");
  
  return params;
}

EqualNormalFluxConstraint::EqualNormalFluxConstraint(const InputParameters & parameters)
  : ADMortarConstraint(parameters),
  _primary_tensor(getParam<bool>("primary_tensor")),
  _secondary_tensor(getParam<bool>("secondary_tensor")),
  _primary_mat_prop_real(!_primary_tensor ? &getNeighborADMaterialProperty<Real>("primary_mat_prop"):nullptr),
  _secondary_mat_prop_real(!_secondary_tensor ? &getADMaterialProperty<Real>("secondary_mat_prop"):nullptr),
  _primary_mat_prop_tensor(_primary_tensor ? &getNeighborADMaterialProperty<RealTensorValue>("primary_mat_prop"):nullptr),
  _secondary_mat_prop_tensor(_secondary_tensor ? &getADMaterialProperty<RealTensorValue>("secondary_mat_prop"):nullptr)

{
}

ADReal
EqualNormalFluxConstraint::computeQpResidual(Moose::MortarType mortar_type)
{
       
  switch (mortar_type)
  {
    case Moose::MortarType::Secondary:
      return _lambda[_qp] * _test_secondary[_i][_qp]; 
    case Moose::MortarType::Primary:
      return _lambda[_qp] * _test_primary[_i][_qp];
    case Moose::MortarType::Lower:
    {
        ADRealVectorValue gradup; 
        ADRealVectorValue gradus;
        auto residual = _lambda[_qp] * _test[_i][_qp];
        if (_has_primary)
        {
            if (!_primary_tensor)
                gradup = (*_primary_mat_prop_real)[_qp] * _grad_u_primary[_qp];
            else
                gradup = (*_primary_mat_prop_tensor)[_qp] * _grad_u_primary[_qp];
            if (!_secondary_tensor)
                gradus = (*_secondary_mat_prop_real)[_qp] * _grad_u_secondary[_qp];
            else
                gradus = (*_secondary_mat_prop_tensor)[_qp] * _grad_u_secondary[_qp];
                
            residual = ((gradup - gradus) * _normals[_qp]  + _lambda[_qp]) * _test[_i][_qp];
        }
                    
        return residual;
    }
    default:
      return 0;
  }
}

[GiudGiud]

@bwspenc who can I tag on mortar problems?
Thanks

[aeslaughter]

@lindsayad

[lindsayad]

@srinath-chakravarthy I believe you need a difference in sign between your Secondary and Primary residual. Right now you are applying the same sign. For something like heat transport for example that would mean that you are applying a heat sink on both sides of the interface (if lambda is positive) or a heat source on both sides of the interface (if lambda is negative).

[lindsayad]

The GapConductanceConstraint class in the heat_conduction module applies continuity of heat flux across a mortar interface. You can see the difference in signs between primary and secondary residuals

[lindsayad]

Well I shouldn't really say "continuity of heat flux". I should say that the heat flux is the same on either side of the interface and we are setting that flux equal to "k * (T_primary - T_secondary) / gap" through the use of the Lagrange Multiplier