[GeoMechanicsApplication] Extend functionality for user soil model for interface elements by rfaasse · Pull Request #13630 · KratosMultiphysics/Kratos

rfaasse · 2025-07-15T07:14:45Z

📝 Description
This PR extends the existing UMAT laws, using a template parameter and ConstitutiveLawDimension instance, such that they can be used with the correct dimensions needed by the new interfaces.

Note that the existing functionality is kept identical. It might seem counterintuitive, but the existing 2D/interface UMAT laws need to have stress/strain vectors and a D matrix with sizes consistent with 3D models. This is because most UMATs we encounter, assume these sizes (hard-coded), meaning they will write in out-of-bounds memory if the sizes are smaller.

For our newly registered UMAT law, the sizes are as expected.

…mplate parameter

…ument

…e 3d more consistent with 2D)

…impact as possible

…stitutive dimension is merged to master)

…-soil-model-for-interface-element

aronnoordam · 2025-07-15T08:25:51Z

makes sense to me

markelov208

Hi Richard, thank you for the clear PR. I have a non-blocking comment.

applications/GeoMechanicsApplication/custom_constitutive/small_strain_umat_law.cpp

avdg81

Hi Richard,

Thanks a lot for generalizing the primary UMAT constitutive law and for adding a proper UMAT constitutive law for two-dimensional line interfaces. I feel the changes are clear, especially with the comments explaining about the wrong size assumptions in most (if not all) of our UMAT libraries.
I have several suggestions, but none of them is blocking in my opinion.

One more thing: can you please rerun Clang-format before merging this branch into master? Thanks.

avdg81 · 2025-07-15T11:36:24Z

applications/GeoMechanicsApplication/custom_constitutive/small_strain_umat_2D_interface_law.hpp

    /// Pointer definition of SmallStrainUMAT2DInterfaceLaw
    KRATOS_CLASS_POINTER_DEFINITION(SmallStrainUMAT2DInterfaceLaw);

+    SmallStrainUMAT2DInterfaceLaw() = default;


Just out of curiosity: does the default constructor need to be publicly available? I'm aware that the default constructor is used for deserialization, but since the Serializer class is a friend, it will have access to the private members of the befriended class. How do you see that?

Not necessary indeed, moved

avdg81 · 2025-07-15T11:40:43Z

applications/GeoMechanicsApplication/custom_constitutive/small_strain_umat_2D_interface_law.hpp

I've noticed that the save and load member functions supply the incorrect base class: we should use SmallStrainUMATLaw<VOIGT_SIZE_3D> rather than ConstitutiveLaw.

avdg81 · 2025-07-15T11:44:15Z

...ations/GeoMechanicsApplication/custom_constitutive/small_strain_umat_2D_plane_strain_law.hpp

    /// Pointer definition of SmallStrainUMAT2DPlaneStrainLaw
    KRATOS_CLASS_POINTER_DEFINITION(SmallStrainUMAT2DPlaneStrainLaw);

+    SmallStrainUMAT2DPlaneStrainLaw() = default;


I've got the same question here about the access specifier of the default constructor: can we make it private rather than public?

avdg81 · 2025-07-15T11:45:43Z

applications/GeoMechanicsApplication/custom_constitutive/small_strain_umat_2D_interface_law.hpp

+    explicit SmallStrainUMAT2DInterfaceLaw(std::unique_ptr<ConstitutiveLawDimension> pConstitutiveDimension)
+        : SmallStrainUMATLaw<VOIGT_SIZE_3D>(std::move(pConstitutiveDimension))
+    {
+    }


I'd rather move the implementation of this constructor to the .cpp file, and only keep the declaration here.

avdg81 · 2025-07-15T11:46:47Z

...ations/GeoMechanicsApplication/custom_constitutive/small_strain_umat_2D_plane_strain_law.hpp

+    explicit SmallStrainUMAT2DPlaneStrainLaw(std::unique_ptr<ConstitutiveLawDimension> pConstitutiveDimension)
+        : SmallStrainUMATLaw<VOIGT_SIZE_3D>(std::move(pConstitutiveDimension))
+    {
+    }


Also here I'd rather move the constructor's implementation to the .cpp file.

avdg81 · 2025-07-15T12:33:40Z

applications/GeoMechanicsApplication/custom_constitutive/small_strain_umat_law.cpp

-            rValue.resize(mStressVectorFinalized.size());
-
-        noalias(rValue) = mStressVectorFinalized;
+        if (rValue.size() != TVoigtSize) rValue.resize(TVoigtSize);


The size check is also carried out by the resize operation, so we don't need to do it here:

Suggested change

if (rValue.size() != TVoigtSize) rValue.resize(TVoigtSize);

rValue.resize(TVoigtSize);

avdg81 · 2025-07-15T12:34:56Z

applications/GeoMechanicsApplication/custom_constitutive/small_strain_umat_law.cpp

+        for (unsigned int i = 0; i < TVoigtSize; ++i) {
+            rValue[i] = mStressVectorFinalized[i];
+        }


If we can use std::copy_n here, I'd prefer that.

Reverted to just assignment

avdg81 · 2025-07-15T19:35:27Z

applications/GeoMechanicsApplication/custom_elements/line_interface_element.cpp

+std::vector<Matrix> LineInterfaceElement::CalculateConstitutiveMatricesAtIntegrationPoints(
+    const ProcessInfo& rProcessInfo, const std::vector<Vector>& rRelativeDisplacements)


As far as I know, we normally have a reference-to-const ProcessInfo instance as the final function parameter. I would suggest to apply that order here as well.

avdg81 · 2025-07-15T19:37:02Z

applications/GeoMechanicsApplication/custom_elements/line_interface_element.cpp

+std::vector<Matrix> CalculateConstitutiveMatricesAtIntegrationPoints(const std::vector<ConstitutiveLaw::Pointer>& rConstitutiveLaws,
+                                                                     const Properties& rProperties,
+                                                                     const ProcessInfo& rProcessInfo,
+                                                                     const std::vector<Vector>& rRelativeDisplacements)


As far as I know, we normally have a reference-to-const ProcessInfo instance as the final function parameter. I would suggest to apply that order here as well:

Suggested change

std::vector<Matrix> CalculateConstitutiveMatricesAtIntegrationPoints(const std::vector<ConstitutiveLaw::Pointer>& rConstitutiveLaws,

const Properties& rProperties,

const ProcessInfo& rProcessInfo,

const std::vector<Vector>& rRelativeDisplacements)

std::vector<Matrix> CalculateConstitutiveMatricesAtIntegrationPoints(const std::vector<ConstitutiveLaw::Pointer>& rConstitutiveLaws,

const Properties& rProperties,

const std::vector<Vector>& rRelativeDisplacements,

const ProcessInfo& rProcessInfo)

avdg81 · 2025-07-15T20:01:03Z

...sApplication/tests/line_interface_elements/Neumann_single_stage_umat/MaterialParameters.json

I've noticed this is a copy of the MaterialParameters.json file used for the multi-stage analysis. Perhaps we can share this file across both analyses?

I'll wait with this last one until Mohameds integration tests are also in.

rfaasse

Processed review comments

applications/GeoMechanicsApplication/custom_constitutive/small_strain_umat_law.cpp

rfaasse · 2025-07-17T07:44:24Z

applications/GeoMechanicsApplication/custom_constitutive/small_strain_umat_2D_interface_law.hpp

    /// Pointer definition of SmallStrainUMAT2DInterfaceLaw
    KRATOS_CLASS_POINTER_DEFINITION(SmallStrainUMAT2DInterfaceLaw);

+    SmallStrainUMAT2DInterfaceLaw() = default;


Not necessary indeed, moved

rfaasse · 2025-07-17T07:44:30Z

applications/GeoMechanicsApplication/custom_constitutive/small_strain_umat_2D_interface_law.hpp

+    explicit SmallStrainUMAT2DInterfaceLaw(std::unique_ptr<ConstitutiveLawDimension> pConstitutiveDimension)
+        : SmallStrainUMATLaw<VOIGT_SIZE_3D>(std::move(pConstitutiveDimension))
+    {
+    }


rfaasse · 2025-07-17T08:47:47Z

applications/GeoMechanicsApplication/custom_constitutive/small_strain_umat_law.cpp


        // We compute the stress
-        SmallStrainUMAT3DLaw::CalculateMaterialResponseCauchy(rParameterValues);
+        SmallStrainUMATLaw<TVoigtSize>::CalculateMaterialResponseCauchy(rParameterValues);


rfaasse · 2025-07-17T08:47:55Z

applications/GeoMechanicsApplication/custom_constitutive/small_strain_umat_law.cpp

-            rValue.resize(mStressVectorFinalized.size());
-
-        noalias(rValue) = mStressVectorFinalized;
+        if (rValue.size() != TVoigtSize) rValue.resize(TVoigtSize);


rfaasse · 2025-07-17T09:59:14Z

applications/GeoMechanicsApplication/custom_constitutive/small_strain_umat_law.hpp

     * @brief Dimension of the law:
     */
-    SizeType WorkingSpaceDimension() override { return Dimension; }
+    SizeType WorkingSpaceDimension() override { return mpConstitutiveDimension->GetDimension(); }


rfaasse · 2025-07-17T10:01:52Z

applications/GeoMechanicsApplication/custom_elements/line_interface_element.cpp

+std::vector<Matrix> CalculateConstitutiveMatricesAtIntegrationPoints(const std::vector<ConstitutiveLaw::Pointer>& rConstitutiveLaws,
+                                                                     const Properties& rProperties,
+                                                                     const ProcessInfo& rProcessInfo,
+                                                                     const std::vector<Vector>& rRelativeDisplacements)


rfaasse · 2025-07-17T10:02:54Z

applications/GeoMechanicsApplication/custom_elements/line_interface_element.cpp

+std::vector<Matrix> LineInterfaceElement::CalculateConstitutiveMatricesAtIntegrationPoints(
+    const ProcessInfo& rProcessInfo, const std::vector<Vector>& rRelativeDisplacements)


rfaasse · 2025-07-17T10:05:36Z

...sApplication/tests/line_interface_elements/Neumann_single_stage_umat/MaterialParameters.json

I'll wait with this last one until Mohameds integration tests are also in.

rfaasse · 2025-07-17T10:08:05Z

applications/GeoMechanicsApplication/custom_constitutive/small_strain_umat_2D_interface_law.hpp

…-soil-model-for-interface-element

avdg81

Thank you very much for processing the review suggestions. I have one very minor suggestion, but feel free to not apply it (now). If that's the only thing left, please merge this PR.

avdg81 · 2025-07-17T15:26:32Z

...ations/GeoMechanicsApplication/custom_constitutive/small_strain_umat_2D_plane_strain_law.hpp

 ///@}

-///@} addtogroup block
+///@} addtogroup blocksmal


Nitpicking: this change was probably not intended. No need to revert it when there are no other things to be changed.

Whoops, seems like the start of a search command that ended up here. I'll fix it when we add the 3D umat interface version 👍

rfaasse added 15 commits July 9, 2025 13:34

Added (unused) template parameter for the SmallStrainUMAT3DLaw

3b9961b

Removed static constexpr VoigtSize and replaced its usage with the te…

f22c6dc

…mplate parameter

Corrected the voigt size in the template instantiations

87dfb95

Formatting

75a16b4

Replaced all usages of VOIGT_SIZE_3D in the law with the template arg…

d25f981

…ument

Added the constitutive dimension to the umat 3d law (not using it yet)

2ea848a

Actually use the constitutive dimension in the appropriate places

a1aa4fa

Make some of the copy operations dependent on the voigtsize (i.e. mak…

0897c8f

…e 3d more consistent with 2D)

Revert some of the changes to make the generalization as small of an …

ecb60fe

…impact as possible

Renamed 3d UMAT law to be more generic

e92ce50

Added integration test (which will start working when the correct con…

13b5dcc

…stitutive dimension is merged to master)

Merge remote-tracking branch 'origin/master' into geo/13568-link-user…

6b8d122

…-soil-model-for-interface-element

Used the newly created constitutive interface dimension

f382f41

Formatting

f74c66c

Added some comments to clarify the UMAT predicaments

6612e03

rfaasse linked an issue Jul 15, 2025 that may be closed by this pull request

[GeoMechanicsApplication] Link the user soil model for interface element #13568

Closed

rfaasse added 2 commits July 15, 2025 09:18

Added some more comments to clarify the UMAT predicaments

8cbf8b7

Cleaned up integration test files

62d2ff7

rfaasse self-assigned this Jul 15, 2025

rfaasse changed the title ~~[GeoMechanicsApplication] Link user soil model for interface element~~ [GeoMechanicsApplication] Implement functionality for user soil model for interface elements Jul 15, 2025

rfaasse added the GeoMechanics Issues related to the GeoMechanicsApplication label Jul 15, 2025

rfaasse changed the title ~~[GeoMechanicsApplication] Implement functionality for user soil model for interface elements~~ [GeoMechanicsApplication] Extend functionality for user soil model for interface elements Jul 15, 2025

Added clarification for voigt size

f2730f3

rfaasse marked this pull request as ready for review July 15, 2025 07:42

rfaasse requested a review from a team as a code owner July 15, 2025 07:42

rfaasse requested review from aronnoordam, avdg81 and markelov208 July 15, 2025 07:42

markelov208 previously approved these changes Jul 15, 2025

View reviewed changes

applications/GeoMechanicsApplication/custom_constitutive/small_strain_umat_law.cpp Outdated Show resolved Hide resolved

avdg81 reviewed Jul 16, 2025

View reviewed changes

Processing review comments

a4555fd

rfaasse dismissed markelov208’s stale review via a4555fd July 17, 2025 10:07

Removed redundant template argument

a76c91b

rfaasse commented Jul 17, 2025

View reviewed changes

Merge remote-tracking branch 'origin/master' into geo/13568-link-user…

3cbde48

…-soil-model-for-interface-element

rfaasse requested review from avdg81 and markelov208 July 17, 2025 13:51

Formatting

ff7a22a

rfaasse enabled auto-merge (squash) July 17, 2025 13:57

Updated linear elastic interface UMAT

001a5a6

avdg81 approved these changes Jul 17, 2025

View reviewed changes

rfaasse merged commit c8d5975 into master Jul 17, 2025
10 checks passed

rfaasse deleted the geo/13568-link-user-soil-model-for-interface-element branch July 17, 2025 16:58

	if (rValue.size() != TVoigtSize) rValue.resize(TVoigtSize);
	rValue.resize(TVoigtSize);

		std::vector<Matrix> LineInterfaceElement::CalculateConstitutiveMatricesAtIntegrationPoints(
		const ProcessInfo& rProcessInfo, const std::vector<Vector>& rRelativeDisplacements)

Conversation

rfaasse commented Jul 15, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

aronnoordam commented Jul 15, 2025

Uh oh!

markelov208 left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

avdg81 left a comment

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!

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!

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!

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!

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!

Choose a reason for hiding this comment

Uh oh!

rfaasse left a comment

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!

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!

Choose a reason for hiding this comment

Uh oh!

avdg81 left a comment

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

rfaasse commented Jul 15, 2025 •

edited

Loading