Use of CompressibleSolid - Stencils refactoring - Transmissibility update #1311

CusiniM · 2021-02-02T08:56:34Z

In this PR we

Introduce the use of the CompressibleSolid in the flow solvers.
Adjust accummulationKernels to use new and old porosity instead of the pvMultiplier.
Remove the old PoreVolumeCompressibleSolid and PoroElastic.
Refactor stencils to allow for the transmissibility to vary throughout the simulation:
- add stencilWrappers that contain only views and can be called in the kernels to update transmissibility
- introduce new stencils to distinguish between embeddedSurfacesToCell connections and faceElementsToCells ones ( Each connection type updates transmissibility in a different way)
Unify SinglePhaseFlowFluxKernels for all stencil types and allow for transmissibility update
Add specific fluxKernels for the poromechanics case in which transmissibility can depend on mechanics unknowns (fractures only for now).
Modify CompositionalFlowSolver and its kernels to work with all subregion and stencil types to have multiphase flow in the fractures.
Make the HydrofractureSolver inherit from the poromechanics one.
Add an integrated test with compositional flow in an embedded fracture.

Rebaseline: https://github.com/GEOSX/integratedTests/pull/145

Tests are passing on Lassen.
Breaks ProppantSolver

…ionalFlowInFractures

klevzoff

Solid work! Haven't gone through the whole thing yet, but here are a few comments

examples/hydraulicFracturing/KGD/KGD_rotated.xml

src/coreComponents/constitutive/solid/CoupledSolid.hpp

klevzoff · 2021-08-18T21:01:44Z

src/coreComponents/constitutive/solid/porosity/PorosityBase.hpp

  struct viewKeyStruct : public ConstitutiveBase::viewKeyStruct
  {
-    static constexpr char const * newPorosityString() { return "newPorosity"; }
+    static constexpr char const * newPorosityString() { return "porosity"; }


I like the change, but can we also rename the function and corresponding member (m_newPorosity)?

I wanted to keep the same naming convention as for the stress. So, I am ok with changing it but I would then change it for both.

I seem to vaguely remember a similar discussion happening around stress, indeed. I'm sure there were good reasons to keep the name newStress. For me personally, the new/old monikers only make sense inside the update function; once the update is done, and for the majority of the code, "new" stress/porosity is just the current value - I would maybe actually prefer curr/prev over new/old (though I've introduced a lost of old variables in the past myself). But consistency across models is also important.

src/coreComponents/fileIO/silo/SiloFile.cpp

src/coreComponents/finiteVolume/CellElementStencilTPFA.hpp

klevzoff · 2021-08-19T06:31:00Z

src/coreComponents/physicsSolvers/fluidFlow/FlowSolverBase.cpp


 std::vector< string > FlowSolverBase::getConstitutiveRelations( string const & regionName ) const
 {
+  // TODO IS THIS EVER USED? WHAT FOR?


Not used currently. I think the idea was to do more intelligent choice of quadrature dimension for each constitutive model based on whether a FEM or a FV solver uses that model. I'm not sure if we're ever gonna go that route though.

src/coreComponents/physicsSolvers/fluidFlow/FlowSolverBase.hpp

src/coreComponents/physicsSolvers/fluidFlow/FluxKernelsHelper.hpp

klevzoff · 2021-08-19T06:47:01Z

src/coreComponents/physicsSolvers/fluidFlow/ProppantTransportKernels.cpp

-                                ElementViewConst< arrayView1d< real64 const > > const & aperture,
-                                CRSMatrixView< real64, globalIndex const > const & localMatrix,
-                                arrayView1d< real64 > const & localRhs )
+  launch< EmbeddedSurfaceToCellStencil >( EmbeddedSurfaceToCellStencil const & GEOSX_UNUSED_PARAM( stencil ),


I don't think all these specializations are necessary, since you've limited the dispatch only to SurfaceElementStencil?

I think I had to add them because the compiler was not happy otherwise but I can try to remove them and check again.

src/coreComponents/physicsSolvers/multiphysics/SinglePhaseReservoir.cpp

klevzoff

This looks great and ready to go IMO (as soon as CI is passed)

src/coreComponents/physicsSolvers/fluidFlow/SinglePhaseFVM.cpp

klevzoff · 2021-08-22T21:08:27Z

src/coreComponents/physicsSolvers/fluidFlow/SinglePhaseBase.hpp

+   * @param jumpDofKey dofKey of the displacement jump
+   */
+  virtual void
+  assemblePoroelasticFluxTerms( real64 const time_n,


These things look a bit hacky, but I don't have a concrete refactoring suggestion at the moment. Unless somebody has one, let's just merge as is and refactor later.

…ionalFlowInFractures

francoishamon

Great work! Everything looks very good to me. I just made small comments to help with the cleanup before merging.

src/coreComponents/unitTests/fluidFlowTests/testSinglePhaseFVMKernels.cpp

src/coreComponents/physicsSolvers/multiphysics/HydrofractureSolver.cpp

...hysicsSolvers/fluidFlow/integratedTests/compositionalMultiphaseFlow/deadoil_3ph_corey_2d.xml

src/coreComponents/physicsSolvers/fluidFlow/SinglePhaseFVMKernels.cpp

src/coreComponents/finiteVolume/EmbeddedSurfaceToCellStencil.hpp

src/coreComponents/constitutive/solid/porosity/PressurePorosity.hpp

src/coreComponents/constitutive/permeability/ParallelPlatesPermeability.hpp

rrsettgast · 2021-08-27T03:40:49Z

src/coreComponents/constitutive/permeability/ParallelPlatesPermeability.hpp

+    real64 const perm = 0.25 * ( oldHydraulicAperture*oldHydraulicAperture*oldHydraulicAperture +
+                                 oldHydraulicAperture*oldHydraulicAperture*newHydraulicAperture +
+                                 oldHydraulicAperture*newHydraulicAperture*newHydraulicAperture +
+                                 newHydraulicAperture*newHydraulicAperture*newHydraulicAperture ) / 12;


We should note that this is essentially a choice on integration schemes.

yeah, I think we may want an option here somehow. I have kept this one coz it was used in the previous version of the code and some integratedTests do not converge with the fully nonlinear model. They may just require a smaller timestep and we could just revisit this.

src/coreComponents/constitutive/permeability/PermeabilityBase.hpp

rrsettgast · 2021-08-27T03:43:11Z

src/coreComponents/constitutive/solid/CompressibleSolid.hpp

  }

+  GEOSX_HOST_DEVICE
+  void updateFromPressureAndAperture( localIndex const k,


update what from pressure and aperture?

oh, I thought the constitutive model in general and so all models it "contains".

State update?

rrsettgast · 2021-08-27T03:43:58Z

src/coreComponents/constitutive/solid/CompressibleSolid.hpp

   * @brief Catalog name
   * @return Static catalog string
   */
-  static string catalogName() { return string( "CompressibleSolid" ) + PORO_TYPE::catalogName() + PERM_TYPE::catalogName(); }


no variation in poro type?

Not for now since we only have 2 porosity models and one is only for poromechanics which uses PorousSolid. It may change though if, for example, we introduce geochemistry.

rrsettgast · 2021-08-27T03:47:44Z

src/coreComponents/constitutive/solid/CoupledSolid.hpp

+   * @return number of gauss points per element
+   */
+  GEOSX_HOST_DEVICE
+  localIndex numGauss() const { return m_porosityUpdate.numGauss(); }


Not really happy with explicit calling this "Gauss". It is really number of material points per element, and the fact that they are used in some sort of integration isn't really in scope from the constitutive model point of view....unless I am not thinking about this correctly?

src/coreComponents/constitutive/solid/CoupledSolid.hpp

rrsettgast · 2021-08-27T03:50:25Z

src/coreComponents/constitutive/solid/PorousSolid.hpp

+   * @param stiffness the stiffness array
+   */
+  GEOSX_HOST_DEVICE
+  void getElasticStiffness( localIndex const k, real64 ( & stiffness )[6][6] ) const


Is this called?

yeah, the EFEM kernel needs it. It's not great coz it means it will only work with Elasticity but that's the way it is for now.

rrsettgast · 2021-08-27T05:26:14Z

src/coreComponents/constitutive/solid/porosity/PorosityBase.cpp

  registerWrapper( viewKeyStruct::newPorosityString(), &m_newPorosity ).
    setPlotLevel( PlotLevel::LEVEL_0 ).
-    setApplyDefaultValue( 0.2 ); // will be overwritten
+    setApplyDefaultValue( 1.0 ); // will be overwritten but it's important for newly created faceElements.


Yeah, when creating a fracture element it is convenient if it's porosity is automatically set to 1.0. Otherwise one would have to set that value. For the cell elements is not a problem coz porosity is properly set usually either through a field specification or by defining the referencePorosity.

src/coreComponents/constitutive/solid/porosity/PorosityBase.hpp

Initial commit. Adapting kernels to include fractures.

6d1705c

CusiniM mentioned this pull request Feb 2, 2021

Permeability and stencils refactoring idea. #1312

Closed

CusiniM added 5 commits February 2, 2021 15:17

Kernels done. Working on stencil.

27472ff

Introduced updates and passThrough for permeability.

2a06d1b

PermeabilityBase and FracturePermeabilityBase + a model for each.

b70bca8

builds with all new perm models.

74fee31

Merge remote-tracking branch 'upstream/develop' into feature/composit…

585f497

…ionalFlowInFractures

CusiniM changed the title ~~Compositonal multiphase flow in fractured media.~~ Permeability constitutive model - Stencil weights update - compositional flow in fractures. Feb 23, 2021

CusiniM changed the title ~~Permeability constitutive model - Stencil weights update - compositional flow in fractures.~~ Permeability constitutive model - Stencil weights update - Compositional flow in fractures. Feb 23, 2021

CusiniM linked an issue Feb 23, 2021 that may be closed by this pull request

Permeability and stencils refactoring idea. #1312

Closed

CusiniM added 6 commits February 24, 2021 10:26

wip stencilWrapper

d3cb19f

Merge remote-tracking branch 'upstream/develop' into feature/composit…

3efb05d

…ionalFlowInFractures

porosity model + permeability fixes.

1a07359

Almost compiling but there is still an error in SinglePhaserFVM.cpp

b4b119f

compiling!

9de1161

adding proper tensorial calculation.

a3f3763

CusiniM requested review from francoishamon and klevzoff March 10, 2021 23:09

CusiniM added 2 commits March 10, 2021 15:17

added ocnstant permeability

57e9bba

vectorial porosity ref.

66439c9

rrsettgast self-requested a review March 11, 2021 18:28

castelletto1 mentioned this pull request Mar 15, 2021

Kernels for poroelastic solvers #1353

Merged

CusiniM added 4 commits March 22, 2021 11:58

Merge remote-tracking branch 'upstream/develop' into feature/composit…

05ad484

…ionalFlowInFractures

Merge remote-tracking branch 'upstream/develop' into feature/composit…

d5f5212

…ionalFlowInFractures

fixed compilation error.

fa70551

compressible_1d is working.

0c50336

CusiniM changed the title ~~Permeability constitutive model - Stencil weights update - Compositional flow in fractures.~~ Permeability/Porosity constitutive models - Stencil weights update - Compositional flow in fractures. Apr 2, 2021

CusiniM requested a review from castelletto1 April 2, 2021 20:47

CusiniM added 2 commits April 2, 2021 14:51

Fixed restart issue. Added possibility to have a fracture porosity.

aa1da2a

Fixed some more xmls.

1fd5398

CusiniM and others added 5 commits August 16, 2021 14:32

fixed HydrofractureSolver.

9602657

fixed xmls.

a6260ca

fixed all xmls.

272c2a3

fixed compilation errors + formatted xmls.

5a28b05

adding back refPoro scaling of comp residual + fixes to xmls.

9e41948

klevzoff reviewed Aug 19, 2021

View reviewed changes

Addressed Sergey's comments.

6a7565d

klevzoff approved these changes Aug 22, 2021

View reviewed changes

francoishamon mentioned this pull request Aug 24, 2021

Fix VTK crash with empty subregions #1543

Merged

Matteo Cusini and others added 8 commits August 24, 2021 12:34

explicit capture + removed old porosity, pvMult wrappers.

118ada0

fixing some doxygen errors.

3b37092

Merge remote-tracking branch 'upstream/develop' into feature/composit…

839fbeb

…ionalFlowInFractures

fixed xml files

6436096

changed function name.

af4681e

fixed junction flow.

064ddf9

fixed all doxygen errors.

be2bb52

fixed poroelastic compositional integratedTest.

a9cd757

francoishamon approved these changes Aug 26, 2021

View reviewed changes

CusiniM and others added 3 commits August 26, 2021 12:16

Francois's comments.

4fff753

integrated test with comp flow in frac.

02b02a3

added integrated tests. Updated integratedTests tag.

96d0193

CusiniM added ci: run CUDA builds Allows to triggers (costly) CUDA jobs and removed flag: ready for review labels Aug 26, 2021

udpated integratedTests tag.

fbb83dd

rrsettgast approved these changes Aug 27, 2021

View reviewed changes

CusiniM added 2 commits August 26, 2021 23:29

small fixed based on Randy's review.

37b934f

change function name.

60bd63d

CusiniM merged commit a23824a into GEOS-DEV:develop Aug 27, 2021

klevzoff mentioned this pull request Aug 30, 2021

[Bug] Broken import of referencePorosity from external meshes #1553

Closed

CusiniM deleted the feature/compositionalFlowInFractures branch October 14, 2021 18:52

Use of CompressibleSolid - Stencils refactoring - Transmissibility update #1311

Use of CompressibleSolid - Stencils refactoring - Transmissibility update #1311

Uh oh!

Conversation

CusiniM commented Feb 2, 2021 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

klevzoff left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Uh oh!

klevzoff left a comment • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

Choose a reason for hiding this comment

Uh oh!

francoishamon left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

CusiniM Aug 27, 2021 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Reviewers

CusiniM commented Feb 2, 2021 •

edited

Loading

klevzoff left a comment •

edited

Loading

CusiniM Aug 27, 2021 •

edited

Loading