Create the two phase flow process #1530

Yonghui56 · 2016-11-09T13:13:07Z

In this pull request, a volume based two phase flow process is created. Nonwetting phase pressure and capillary pressure are selected as the primary variables. The Liakopoulos experiment is selected as a benchmark for testing.
Note that currently the capillary pressure - saturation, relative permeability --saturation relationships are delivered by curves. Later on these will be updated based on the material properties work @wenqing

endJunction · 2016-11-10T17:57:58Z

Applications/CLI/Tests.cmake

@@ -371,6 +371,16 @@ if(NOT OGS_USE_MPI)
         DIFF_DATA
         h_us_quad_1000.vtu richards_pcs_0_ts_100_t_100.000000.vtu PRESSURE1 pressure
    )
+	 AddTest(


tabs to be replaced with whitespaces. Please change your editor's default choice, since I have found this error already multiple times.

endJunction · 2016-11-10T17:59:41Z

MaterialLib/TwoPhaseModels/CreateTwoPhaseFlowMaterialProperties.h

+namespace TwoPhaseFlowWithPP
+{
+/** Create a porosity model
+ *  @param config  ConfigTree object has a tag of <porosity>


Is this indeed creating a porosity model?

endJunction · 2016-11-10T18:05:38Z

MaterialLib/TwoPhaseModels/TwoPhaseFlowWithPPMaterialProperties.h

+class TwoPhaseFlowWithPPMaterialProperties
+{
+public:
+    typedef MaterialLib::Fluid::FluidProperty::ArrayType ArrayType;


using is preferred over typedef

endJunction · 2016-11-10T18:06:25Z

MaterialLib/TwoPhaseModels/TwoPhaseFlowWithPPMaterialProperties.h

+    void setMaterialID(const ProcessLib::SpatialPosition& pos);
+
+    /**
+     * \brief Compute the coefficient of the mass term by


Is this correct documentation for getPermeability?

endJunction · 2016-11-10T18:08:12Z

MaterialLib/TwoPhaseModels/TwoPhaseFlowWithPPMaterialProperties.h

+    std::unique_ptr<MaterialLib::Fluid::FluidProperty> _gas_density;
+    std::unique_ptr<MaterialLib::Fluid::FluidProperty> _gas_viscosity;
+
+    /** Use porous medium models for different material zones.


The "porous medium models" is from different class. Need to change this to two phase models. (Not only here, but in all similar places.)

endJunction · 2016-11-10T18:09:11Z

ProcessLib/TwoPhaseFlowWithPP/CreateTwoPhaseFlowWithPPProcess.cpp

+    // Process variable.
+    auto process_variables = findProcessVariables(
+        variables, config,
+        {//! \ogs_file_param_special{process__TWOPHASE_FLOW__WITHPP__process_variables__process_variable}


One underscore too much in TWOPHASE_FLOW__WITHPP.
Actually the name is wrong. Please use "TWOPHASE_FLOW_PP" here and below.

endJunction · 2016-11-10T18:10:41Z

ProcessLib/TwoPhaseFlowWithPP/CreateTwoPhaseFlowWithPPProcess.cpp

+    Eigen::VectorXd specific_body_force;
+
+    std::vector<double> const b =
+        //! \ogs_file_param_special{process__TWOPHASE_FLOW_specific_body_force}


needs an underscore before 'specific'

endJunction · 2016-11-10T18:12:05Z

ProcessLib/TwoPhaseFlowWithPP/TwoPhaseFlowWithPPLocalAssembler-impl.h

+#include "MathLib/InterpolationAlgorithms/PiecewiseLinearInterpolation.h"
+#include "NumLib/Function/Interpolation.h"
+#include "TwoPhaseFlowWithPPProcessData.h"
+// using namespace Eigen;


endJunction · 2016-11-10T18:12:16Z

ProcessLib/TwoPhaseFlowWithPP/TwoPhaseFlowWithPPLocalAssembler-impl.h

+#ifndef OGS_TWOPHASEFLOWWITHPPLOCALASSEMBLER_IMPL_H
+#define OGS_TWOPHASEFLOWWITHPPLOCALASSEMBLER_IMPL_H
+
+#include <iostream>


is this needed?

endJunction · 2016-11-10T18:14:13Z

ProcessLib/TwoPhaseFlowWithPP/TwoPhaseFlowWithPPLocalAssembler-impl.h

+
+    // Note: currently only isothermal case is considered, so the temperature is
+    // assumed to be const
+    // the variation of temperatura will be taken into account in future


formatting. typo in temperature

endJunction · 2016-11-10T18:16:38Z

ProcessLib/TwoPhaseFlowWithPP/TwoPhaseFlowWithPPLocalAssembler-impl.h

+            sm.dNdx.transpose() * permeability * sm.dNdx * integration_factor;
+        Klpc.noalias() +=
+            K_mat_coeff(cap_pressure_coeff_index, cap_pressure_coeff_index) *
+            sm.dNdx.transpose() * permeability * sm.dNdx * integration_factor;


common expressions should be calculated once.

endJunction · 2016-11-16T13:12:45Z

@Yonghui56 You have to answer the comments or change the code. So long I'll set this PR to "needs an update". When ready you can set it to "please review".

ogsbot · 2016-11-16T23:10:56Z

Jenkins: OGS-6/Linux-PRs-dynamic failed: https://svn.ufz.de:8443/job/OGS-6/job/Linux-PRs-dynamic/1721/

ogsbot · 2016-11-16T23:14:41Z

Jenkins: OGS-6/Linux-PRs-dynamic failed: https://svn.ufz.de:8443/job/OGS-6/job/Linux-PRs-dynamic/1722/

endJunction · 2016-11-17T11:26:08Z

ProcessLib/TwoPhaseFlowWithPP/TwoPhaseFlowWithPPLocalAssembler-impl.h

@@ -65,6 +64,9 @@ void TwoPhaseFlowWithPPLocalAssembler<
    auto Mlpc = local_M.template block<cap_pressure_size, cap_pressure_size>(
        cap_pressure_matrix_index, cap_pressure_matrix_index);

+    NodalMatrixType laplace_factor;
+    laplace_factor.setZero(nonwet_pressure_size, cap_pressure_size);


I think the NodalMatrixType is not of size nonwet_pressure_size times cap_pressure_size, isn't it?

@endJunction NodalMatrixType holds the size of [ShapeFunction::NPOINTS] times [ShapeFunction::NPOINTS], while nonwet_pressure_size is equal to ShapeFunction::NPOINTS, so is cap_pressure_size.

wenqing · 2016-11-17T12:49:43Z

ProcessLib/TwoPhaseFlowWithPP/TwoPhaseFlowWithPPLocalAssembler-impl.h

+        // nonwetting
+        double const drhogas_dpg = _process_data._material->getDerivGasDensity(
+            pg_int_pt, _temperature);
+        M_mat_coeff(nonwet_pressure_coeff_index, nonwet_pressure_coeff_index) =


I think matrix type of M_mat_coeff is not needed. You can put the coefficient directly to the matrix calculation like

Mgp.noalias() += (porosity * (1 - Sw) * drhogas_dp ) * + sm.N.transpose() * sm.N * integration_factor;

wenqing · 2016-11-17T12:51:03Z

ProcessLib/TwoPhaseFlowWithPP/TwoPhaseFlowWithPPLocalAssembler-impl.h

+        double const lambda_L = k_rel_L / mu_liquid;
+        K_mat_coeff(cap_pressure_coeff_index, nonwet_pressure_coeff_index) =
+            rho_w * lambda_L;
+        K_mat_coeff(cap_pressure_coeff_index, cap_pressure_coeff_index) =


Similarly, to have a matrix of K_mat_coeff is not necessary.

wenqing · 2016-11-17T13:04:45Z

ProcessLib/TwoPhaseFlowWithPP/TwoPhaseFlowWithPPProcess.cpp

+              std::move(secondary_variables), std::move(named_function_caller)),
+      _process_data(std::move(process_data))
+{
+    DBUG("Create Two-Phase flow process.");


Create two-thase flow process with primary variables of capillary pressure and gas pressure, or simplyCreate TwoPhaseFlowWithPPProcess``

wenqing · 2016-11-17T13:10:45Z

MaterialLib/TwoPhaseModels/CreateTwoPhaseFlowMaterialProperties.cpp

@@ -0,0 +1,99 @@
+/**


Since files in TwoPhaseModels are for a set of process related parameters, I would like to move them to ProcessLib/TwoPhaseFlowWithPP.

In MaterialLib, I think, it should only has physical chemical models or parameters of fluids, solids and porous medium.

wenqing

Looks good

endJunction · 2016-11-17T15:47:52Z

Applications/CLI/Tests.cmake

+         EXECUTABLE ogs
+         EXECUTABLE_ARGS Twophase_Lia_quad2.prj
+         TESTER vtkdiff
+         ABSTOL 1e-1 RELTOL 1e-1


Why is the tolerance so loose? On my computer I can set 1e-9 and 1e-12 for the absolute and relative tolerances.

You have also the gas_pressure with 1e-8 and 1e-13, and saturation with 1e-14 and 1e-14
absolute and relative tolerances (on my computer -- might be slightly looser on other machines), which you can add to the ctest; Just another lines with

Lia_20.vtu twophaseflow_pcs_0_ts_200_t_20.000000.vtu gas_pressure gas_pressure

taking 1e-8 and 1e-12 as the tolerances.

endJunction · 2016-11-17T15:57:35Z

Squash after adding additional ctest vtkdiff lines; I'll merge then.

wenqing · 2016-11-17T16:43:07Z

Calculation of velocity and saturation as secondary variables can be added later on.

endJunction · 2016-11-18T14:46:52Z

Good! When the tests are finished I'll merge it.

ogsbot · 2020-06-19T12:37:43Z

OpenGeoSys development has been moved to GitLab.

See this pull request on GitLab.

endJunction reviewed Nov 10, 2016

View reviewed changes

endJunction force-pushed the master branch from 6be30e3 to 95d89c8 Compare November 12, 2016 16:25

endJunction added the needs an update label Nov 16, 2016

Yonghui56 force-pushed the pr_twophasewithpp_curve_test branch from a9a855c to 06763fd Compare November 16, 2016 23:07

Yonghui56 force-pushed the pr_twophasewithpp_curve_test branch from 478f72e to 3c1a687 Compare November 17, 2016 10:10

endJunction reviewed Nov 17, 2016

View reviewed changes

wenqing reviewed Nov 17, 2016

View reviewed changes

wenqing approved these changes Nov 17, 2016

View reviewed changes

endJunction removed the needs an update label Nov 17, 2016

endJunction reviewed Nov 17, 2016

View reviewed changes

Yonghui56 force-pushed the pr_twophasewithpp_curve_test branch 5 times, most recently from bdf8afe to 538306d Compare November 18, 2016 13:25

Yonghui56 force-pushed the pr_twophasewithpp_curve_test branch 2 times, most recently from 61cd686 to 657a074 Compare November 18, 2016 14:43

Create twophase flow process with p-pc as primary variable

e9dfc49

Update the Tests/Data

990d02a

Yonghui56 force-pushed the pr_twophasewithpp_curve_test branch from 39821e0 to 990d02a Compare November 18, 2016 14:47

endJunction merged commit d15070c into ufz:master Nov 20, 2016

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Create the two phase flow process #1530

Create the two phase flow process #1530

Create the two phase flow process #1530

Create the two phase flow process #1530

Conversation

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment