Adding Drucker-Prager plasticity model #974

joshua-white · 2020-05-28T22:53:45Z

Also a chance to revisit overall design of constitutive/solid

src/coreComponents/constitutive/solid/DruckerPrager.hpp

src/coreComponents/constitutive/unitTests/testDruckerPrager.cpp

src/coreComponents/constitutive/solid/DruckerPrager.hpp

…applied to it instead of a new clean instantiation

joshua-white · 2020-07-13T18:59:11Z

@rrsettgast I am going to get back to this later this week, and try to wrap it up with help from @juliatcamargo and @shabnamjs to add a few extra plasticity models. I'd like to add a material point driver (triaxial loading) for testing and calibration purposes. I recall you had some ideas for how we should structure this?

src/coreComponents/constitutive/solid/DruckerPrager.hpp

sytuannguyen · 2021-02-18T08:52:23Z

src/coreComponents/constitutive/solid/DruckerPragerExtended.hpp

+                         + c5 * deviator[i] * deviator[j];
+    }
+  }
+


The tangent stiffness tensor has only five indenpendent parameters. Does it correspond to a transversely isotropic stiffness tensor? If YES, it will be better to consider the transversly isotropic Ops to compute element stiffness instead of considering the fully anisotropic Ops.

sytuannguyen · 2021-02-18T08:56:33Z

src/coreComponents/constitutive/solid/DruckerPragerExtended.hpp

+  for( localIndex iter=0; iter<20; ++iter )
+  {
+    m_newState[k][q] = m_oldState[k][q] + solution[2];
+


I cannot understand why we could not consider a temporal variable here inside the Newton loop inplace of the m_newState variable. Because after the Newton loop, m_newState will be set back to m_oldState for the next time step so I think it is just something temporal.

sytuannguyen · 2021-02-18T08:59:28Z

src/coreComponents/constitutive/solid/DruckerPrager.hpp

+
+  /// A reference to the ArrayView holding the dilation angle for each element.
+  arrayView1d< real64 const > const m_dilation;
+


m_friction and m_dilation here are not angle

sytuannguyen · 2021-02-18T09:02:00Z

src/coreComponents/constitutive/solid/DruckerPrager.hpp

+
+  /// A reference to the ArrayView holding the new cohesion for each integration point
+  arrayView2d< real64 > const m_newCohesion;
+


Why cohesion is set for each quadrature point when other parameters such as friction, dilation, hardening are for each element? What is the different in their nature?

sytuannguyen · 2021-02-18T09:03:00Z

src/coreComponents/constitutive/solid/DruckerPrager.hpp

+    }
+
+    //printf("%ld %.2e\n",iter,norm);
+


sytuannguyen · 2021-02-18T09:05:53Z

src/coreComponents/constitutive/solid/DruckerPrager.hpp

+  real64 c1 = 2 * m_shearModulus[k] * solution[1] / trialQ;
+  real64 c2 = jacobianInv[0][0] * m_bulkModulus[k] - c1 / 3;
+  real64 c3 = sqrt( 2./3 ) * 3 * m_shearModulus[k] * jacobianInv[0][1];
+  real64 c4 = sqrt( 2./3 ) * m_bulkModulus[k] * jacobianInv[1][0];


We may write the double values in a consistent way. Here we can find 2. 3 1.0 etc.

sytuannguyen · 2021-02-18T09:06:49Z

src/coreComponents/constitutive/solid/DruckerPrager.hpp

+  // begin newton loop
+
+  for( localIndex iter=0; iter<20; ++iter )
+  {


There are many constants in the code that should be moved to the input file

sytuannguyen · 2021-02-18T09:08:43Z

src/coreComponents/constitutive/solid/DruckerPrager.hpp

+
+#ifndef GEOSX_CONSTITUTIVE_SOLID_DRUCKERPRAGER_HPP
+#define GEOSX_CONSTITUTIVE_SOLID_DRUCKERPRAGER_HPP
+


You should add a "_" at the end of this flag to make it consistent with that at the end of this file as well as the flag in other files.

sytuannguyen · 2021-02-18T09:09:51Z

src/coreComponents/constitutive/solid/DruckerPrager.hpp

+  ElasticIsotropicUpdates::smallStrainUpdate( k, q, strainIncrement, stress, stiffness );
+
+  // decompose into mean (P) and von mises (Q) stress invariants
+


sytuannguyen · 2021-02-18T09:13:38Z

src/coreComponents/constitutive/solid/InvariantDecompositions.hpp

+    volStress += stress[i];
+  }
+  volStress /= 3.;
+


Just one line for this ^^ : volStress = ( stress[0] + stress[1] + stress[2] ) / 3.0

sytuannguyen · 2021-02-18T09:14:44Z

src/coreComponents/constitutive/solid/InvariantDecompositions.hpp

+                          real64 ( & strain )[6] )
+{
+  real64 const tmp = sqrt( 1.5 )*devStrain;
+


2.0 / 3.0 as everywhere

sytuannguyen · 2021-02-18T09:15:43Z

src/coreComponents/constitutive/solid/DruckerPrager.cpp

+{
+
+DruckerPrager::DruckerPrager( std::string const & name, Group * const parent ):
+  ElasticIsotropic( name, parent ),


sytuannguyen · 2021-02-18T09:17:12Z

src/coreComponents/constitutive/solid/DruckerPrager.cpp

+    setApplyDefaultValue( 0.0 )->
+    setInputFlag( InputFlags::OPTIONAL )->
+    setDescription( "Cohesion hardening/softening rate" );
+


We may clarify that m>0 is for hardening, m<0 is for softning

sytuannguyen · 2021-02-18T09:19:04Z

src/coreComponents/constitutive/solid/DruckerPrager.cpp

+    setApplyDefaultValue( -1 )->
+    setPlotLevel( dataRepository::PlotLevel::LEVEL_3 )->
+    setDescription( "New cohesion state" );
+


I tried to find this using VisIt, it's quire hard to find at level3 with so many stuffs

sytuannguyen · 2021-02-18T09:21:20Z

src/coreComponents/constitutive/solid/DruckerPrager.cpp

+}
+
+REGISTER_CATALOG_ENTRY( ConstitutiveBase, DruckerPrager, std::string const &, Group * const )
+}


/* namespace constitutive */

sytuannguyen · 2021-02-18T09:21:41Z

src/coreComponents/constitutive/solid/DruckerPrager.cpp

+  } );
+}
+
+REGISTER_CATALOG_ENTRY( ConstitutiveBase, DruckerPrager, std::string const &, Group * const )


sytuannguyen · 2021-02-18T09:24:57Z

src/coreComponents/constitutive/solid/DruckerPragerExtended.cpp

+
+  registerWrapper( viewKeyStruct::defaultHardeningString, &m_defaultHardening )->
+    setApplyDefaultValue( 1.0e6 )->
+    setInputFlag( InputFlags::OPTIONAL )->


Could you verify this default value of 1e6 hardening parameter for the friction. I tested the model and I found some reasonable values in order of 1e-3 or 1e-8

sytuannguyen · 2021-02-18T09:25:45Z

src/coreComponents/constitutive/solid/DruckerPragerExtended.cpp

+}
+
+REGISTER_CATALOG_ENTRY( ConstitutiveBase, DruckerPragerExtended, std::string const &, Group * const )
+}


/* namespace constitutive */

sytuannguyen · 2021-02-18T09:26:04Z

src/coreComponents/constitutive/solid/DruckerPragerExtended.cpp

+  } );
+}
+
+REGISTER_CATALOG_ENTRY( ConstitutiveBase, DruckerPragerExtended, std::string const &, Group * const )


sytuannguyen · 2021-02-18T09:28:53Z

@joshua-white I added some more detail comments in the DruckerPrager and DruckerPragerExtended classes. Hope it helps.

joshua-white · 2021-02-18T16:55:06Z

@sytuannguyen I'll address your comments in my next push, as I'll have to do at least one or two more to catch up with any merges that go in before this, and for the integratedTests rebaseline.

@rrsettgast, @corbett5, and @TotoGaz, I was finally able to pass the cuda-debug build. Travis times out whenever nothing is output to the log for 10 min. I tried using travis_wait, but that does not work well with docker run. The other suggestion I found online was to add the following to the .travis.yml script:

- while sleep 5m; do echo "... still building ..."; done &

which just prints a brief message every five minutes. In this case it builds with no problems. The cuda-debug build takes 9 min longer than develop (52 min vs 43 min) but all other builds are basically identical. For example, cuda-release is 1h23m on the branch vs 1h25m on develop. I've also compared on Lassen and the cuda debug build times are identical, so this is something specific to Travis.

Here are the results for this PR:

Here are the results for the latest develop build:

corbett5 · 2021-02-18T17:57:10Z

@joshua-white nice, seeing as that's still not the longest build that's a fine change by me.

TotoGaz · 2021-02-18T18:33:36Z

.travis.yml

@@ -39,6 +39,7 @@ geosx_linux_build: &geosx_linux_build
    # but that would not have solved the problem for the TPLs (we would require extra action to copy them to the mount point).
  - CONTAINER_NAME=geosx_build
    # Now we can build GEOSX.
+  - while sleep 5m; do echo "... still building ..."; done & 


This means that any job really stuck would last to the build limit (3 hours if I'm correct).
I'd be in favour to making the timeout longer if we can.

@TotoGaz Good point. I'd propose we merge this branch as a working patch, and then come back and revisit other options in a separate PR. Maybe we can get the travis_wait option to work with more effort. @corbett5 Also suggested maybe migrating from make to ninja for the travis build script, and we could update things like this at the same time.

joshua-white added 5 commits May 25, 2020 17:10

Setting up Drucker-Prager files

1083cda

Merge branch 'develop' into white/feature/druckerPrager

b003096

Laying our parameters and arrays

f012092

Setting up proposed interface

bac62c6

adding SaveConvergedState()

3e9773f

joshua-white requested a review from rrsettgast May 28, 2020 22:53

joshua-white marked this pull request as draft May 29, 2020 00:19

Replacing base pure virtuals with error-throwing implementations

2223657

rrsettgast reviewed May 29, 2020

View reviewed changes

src/coreComponents/constitutive/solid/DruckerPrager.hpp Outdated Show resolved Hide resolved

src/coreComponents/constitutive/solid/DruckerPrager.hpp Outdated Show resolved Hide resolved

src/coreComponents/constitutive/solid/DruckerPrager.hpp Outdated Show resolved Hide resolved

Finished basic DP algorithm

ebdb6d6

joshua-white commented May 30, 2020

View reviewed changes

src/coreComponents/constitutive/solid/DruckerPrager.hpp Outdated Show resolved Hide resolved

joshua-white commented May 30, 2020

View reviewed changes

src/coreComponents/constitutive/solid/DruckerPrager.hpp Show resolved Hide resolved

joshua-white commented May 30, 2020

View reviewed changes

src/coreComponents/constitutive/solid/DruckerPrager.hpp Outdated Show resolved Hide resolved

joshua-white commented May 30, 2020

View reviewed changes

src/coreComponents/constitutive/unitTests/testDruckerPrager.cpp Outdated Show resolved Hide resolved

joshua-white commented May 30, 2020

View reviewed changes

src/coreComponents/constitutive/unitTests/testDruckerPrager.cpp Outdated Show resolved Hide resolved

rrsettgast reviewed May 31, 2020

View reviewed changes

src/coreComponents/constitutive/solid/DruckerPrager.hpp Outdated Show resolved Hide resolved

rrsettgast and others added 4 commits June 8, 2020 15:46

fix testDruckerPrager.cpp to use the constitutive model that had XML …

49e9011

…applied to it instead of a new clean instantiation

Merge develop

9ff9c82

Catching up with develop changes

3d84316

Merge branch 'develop' into white/feature/druckerPrager

fa1d341

joshua-white linked an issue Jul 9, 2020 that may be closed by this pull request

Reimplement PoreVolumeCompressibleSolid to fit in solid model hierarchy #816

Closed

joshua-white mentioned this pull request Jul 9, 2020

Reimplement PoreVolumeCompressibleSolid to fit in solid model hierarchy #816

Closed

Merge branch 'develop' into white/feature/druckerPrager

64c1712

joshua-white added 4 commits August 3, 2020 19:17

Merge develop and resolve conflicts

98a7c8c

Updating array setValues calls

5dbbe2f

Updating interface to use c-array style

06ee95f

Adding two-invariant utilities

954290e

corbett5 reviewed Aug 11, 2020

View reviewed changes

src/coreComponents/constitutive/solid/DruckerPrager.hpp Outdated Show resolved Hide resolved

Partial progress restructuring solid class hierarchy

a89e771

sytuannguyen reviewed Feb 18, 2021

View reviewed changes

src/coreComponents/constitutive/solid/DruckerPrager.hpp Outdated

}

//printf("%ld %.2e\n",iter,norm);

Copy link

Contributor

sytuannguyen Feb 18, 2021

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

rm

sytuannguyen reviewed Feb 18, 2021

View reviewed changes

joshua-white added the ci: run CUDA builds Allows to triggers (costly) CUDA jobs label Feb 18, 2021

TotoGaz reviewed Feb 18, 2021

View reviewed changes

joshua-white added 4 commits February 18, 2021 14:28

Merge develop (poroelastic fractures commit)

975922d

Final review edits

af1c707

Fix typo in Sneddon tutorial

9bd0db9

Updating integratedTests hash

5a350ca

joshua-white merged commit ec566dc into develop Feb 19, 2021

joshua-white deleted the white/feature/druckerPrager branch February 19, 2021 02:54

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Adding Drucker-Prager plasticity model #974

Adding Drucker-Prager plasticity model #974

joshua-white commented May 28, 2020

joshua-white commented Jul 13, 2020

sytuannguyen Feb 18, 2021 •

edited

Loading

sytuannguyen Feb 18, 2021 •

edited

Loading

sytuannguyen Feb 18, 2021

sytuannguyen Feb 18, 2021

sytuannguyen Feb 18, 2021

sytuannguyen Feb 18, 2021

sytuannguyen Feb 18, 2021

sytuannguyen Feb 18, 2021

sytuannguyen Feb 18, 2021

sytuannguyen Feb 18, 2021

sytuannguyen Feb 18, 2021

sytuannguyen Feb 18, 2021

sytuannguyen Feb 18, 2021

sytuannguyen Feb 18, 2021 •

edited

Loading

sytuannguyen Feb 18, 2021

sytuannguyen Feb 18, 2021

sytuannguyen Feb 18, 2021 •

edited

Loading

sytuannguyen Feb 18, 2021

sytuannguyen Feb 18, 2021

sytuannguyen commented Feb 18, 2021

joshua-white commented Feb 18, 2021

corbett5 commented Feb 18, 2021

TotoGaz Feb 18, 2021

joshua-white Feb 18, 2021


		/// A reference to the ArrayView holding the dilation angle for each element.
		arrayView1d< real64 const > const m_dilation;


		/// A reference to the ArrayView holding the new cohesion for each integration point
		arrayView2d< real64 > const m_newCohesion;


		#ifndef GEOSX_CONSTITUTIVE_SOLID_DRUCKERPRAGER_HPP
		#define GEOSX_CONSTITUTIVE_SOLID_DRUCKERPRAGER_HPP

		ElasticIsotropicUpdates::smallStrainUpdate( k, q, strainIncrement, stress, stiffness );

		// decompose into mean (P) and von mises (Q) stress invariants

Adding Drucker-Prager plasticity model #974

Adding Drucker-Prager plasticity model #974

Conversation

joshua-white commented May 28, 2020

joshua-white commented Jul 13, 2020

sytuannguyen Feb 18, 2021 • edited Loading

Choose a reason for hiding this comment

sytuannguyen Feb 18, 2021 • edited Loading

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

sytuannguyen Feb 18, 2021 • edited Loading

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

sytuannguyen Feb 18, 2021 • edited Loading

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

sytuannguyen commented Feb 18, 2021

joshua-white commented Feb 18, 2021

corbett5 commented Feb 18, 2021

Choose a reason for hiding this comment

Choose a reason for hiding this comment

sytuannguyen Feb 18, 2021 •

edited

Loading

sytuannguyen Feb 18, 2021 •

edited

Loading

sytuannguyen Feb 18, 2021 •

edited

Loading

sytuannguyen Feb 18, 2021 •

edited

Loading