[GeoMechanicsApplication] Extract list for constitutive matrices and stress vectors #12380

rfaasse · 2024-05-14T12:50:22Z

📝 Description
This PR extracts lists for constitutive matrices and stress vectors. In doing so we add two identical functions to diff order and to the non-diff order element, but we reduce a lot of other duplicated code. If anyone has ideas how to extract something generic here, please let me know. I think it would mean we need to pass on the constitutive laws and maybe make a vector of constitutive parameters and pass that on to make a separate utility function, but that's open for discussion.

…nts inside the function instead of input and use it everywhere the constitutive matrix is being used

…train element and used it in the class (not derived classes yet)

rfaasse · 2024-05-15T14:44:26Z

applications/GeoMechanicsApplication/custom_elements/U_Pw_small_strain_element.cpp

@@ -1943,6 +1887,37 @@ Vector UPwSmallStrainElement<TDim, TNumNodes>::GetPressureSolutionVector()
    return result;
 }

+template <unsigned int TDim, unsigned int TNumNodes>
+void UPwSmallStrainElement<TDim, TNumNodes>::CalculateAnyOfMaterialResponse(


I know that we have two identical functions now for small strain and for diff order, but I still think in total we remove more duplication than we added and the duplication is a lot more localized now.

rfaasse · 2024-05-15T14:47:07Z

applications/GeoMechanicsApplication/custom_elements/U_Pw_small_strain_FIC_element.cpp

        GeoElementUtilities::CalculateNuMatrix<TDim, TNumNodes>(Variables.Nu, Variables.NContainer, GPoint);
        GeoElementUtilities::InterpolateVariableWithComponents<TDim, TNumNodes>(


We checked that these functions don't change anything that the constitutive law uses (i.e. only Nu, while the constitutive law uses Np). Since we changed the order here, I'd like to double check with you: do you think this is correct @WPK4FEM?

When there is a difference between Nu and Np ( e.g. for reason of different interpolation orders of u and Pw ) the constitutive law ( relation between strain and stress ) should be using Nu. For same order interpolation Nu and Np will be equal, still the correct one should be used.

…ist-for-constitutive-matrices-and-stress-vectors

WPK4FEM · 2024-05-16T06:39:18Z

applications/GeoMechanicsApplication/custom_elements/U_Pw_small_strain_FIC_element.cpp

        deformation_gradients, b_matrices, Variables.DisplacementVector, Variables.UseHenckyStrain);
+    std::vector<Matrix> constitutive_matrices;
+    this->CalculateAnyOfMaterialResponse(deformation_gradients, ConstitutiveParameters,


At very first sight this function name is not very descriptive for me. I have to learn what it does reading the function itself.

I agree, I was struggling with the name. Would CalculateAnyOfStressesStrainsOrConstitutiveMatrices be better?

WPK4FEM · 2024-05-16T06:42:38Z

applications/GeoMechanicsApplication/custom_elements/U_Pw_small_strain_FIC_element.cpp

        GeoElementUtilities::CalculateNuMatrix<TDim, TNumNodes>(Variables.Nu, Variables.NContainer, GPoint);
        GeoElementUtilities::InterpolateVariableWithComponents<TDim, TNumNodes>(


When there is a difference between Nu and Np ( e.g. for reason of different interpolation orders of u and Pw ) the constitutive law ( relation between strain and stress ) should be using Nu. For same order interpolation Nu and Np will be equal, still the correct one should be used.

WPK4FEM · 2024-05-16T06:49:18Z

applications/GeoMechanicsApplication/custom_elements/U_Pw_small_strain_element.cpp

+        rConstitutiveParameters.SetStrainVector(rStrainVectors[GPoint]);
+        rConstitutiveParameters.SetShapeFunctionsDerivatives(rDNu_DXContainer[GPoint]);
+        rConstitutiveParameters.SetShapeFunctionsValues(row(rNuContainer, GPoint));
+        rConstitutiveParameters.SetDeterminantF(determinants_of_deformation_gradients[GPoint]);


A more logical order would be SetDeformationGradientF ( as F is the DeformationGradient, this is a repeat ) followed by SetDeterminantF ( which then would have had the expected name SetDeterminantDeformationGradientF ).
When F is given DetF could be computed.

Changed the order

WPK4FEM · 2024-05-16T06:51:39Z

applications/GeoMechanicsApplication/custom_elements/U_Pw_small_strain_element.hpp

+    ///
+    /// \brief This function calculates the constitutive matrices, stresses and strains depending on the
+    ///        constitutive parameters. Note that depending on the settings in the rConstitutiveParameters
+    ///        the function could calculate the stress, the constitutive matrix, the strains, or a combination.


Looking at 1896 of the corresponding cpp file, the strain is given. The implementation of CalculateMaterialResponseCauchy may decide to recompute ( for reasons of a different strain measure ) also the constitutive tensor and stress have their own steering through rConsitutiveParameters.GetOptions()

The strain is only given because we always use the line ConstitutiveParameters.Set(ConstitutiveLaw::USE_ELEMENT_PROVIDED_STRAIN); on the ConstitutiveLaw::Parameters we pass to this function. If we don't the strain could actually be calculated by the constitutive law. Therefore, rStrainVectors is a non-const&, because at this point the function is generic and at this point I'd like this to work with any ConstitutiveLaw::Parameters and any constitutive law

WPK4FEM · 2024-05-16T06:55:30Z

applications/GeoMechanicsApplication/custom_elements/small_strain_U_Pw_diff_order_element.cpp

+{
+    std::vector<Matrix> result;
+    const SizeType      VoigtSize =
+        (GetGeometry().WorkingSpaceDimension() == 3 ? VOIGT_SIZE_3D : VOIGT_SIZE_2D_PLANE_STRAIN);


Are the ( ) necessary to encapsulate the ternary operator?

No, removed!

WPK4FEM · 2024-05-16T07:30:56Z

applications/GeoMechanicsApplication/custom_elements/U_Pw_small_strain_element.cpp

@@ -1932,6 +1876,37 @@ Vector UPwSmallStrainElement<TDim, TNumNodes>::GetPressureSolutionVector()
    return result;
 }

+template <unsigned int TDim, unsigned int TNumNodes>
+void UPwSmallStrainElement<TDim, TNumNodes>::CalculateAnyOfMaterialResponse(


For the naming of this function, it computes the indicated things for all i.p.
In other cases we indicated that with a trailing s. That can be done here too: CalculateMaterialResponses

Agreed, either this, or we rename to something like CalculateAnyOfStressesStrainsOrConstitutiveMatrices (see other comment)

…assume any of the stress, strain or matrices vector could be empty or filled and initialized variable_index to fix pipeline build

…ht size

…ist-for-constitutive-matrices-and-stress-vectors

avdg81

Thank you very much for this great effort to compute the constitutive matrices in advance. That was definitely not an easy task, since the code was strongly coupled. As far as I can see you have done an excellent job. I only have a few minor comments and questions. None of them are blocking in my opinion.

applications/GeoMechanicsApplication/custom_elements/U_Pw_small_strain_FIC_element.cpp

applications/GeoMechanicsApplication/custom_elements/U_Pw_small_strain_element.cpp

markelov208

Richard, a very useful step to improve the structure and to make functions more uniform. However, for my understand of the code it will be useful to split CalculateAnyOfMaterialResponse into three functions. What is your opinion? It can be next PR.

rfaasse · 2024-05-16T10:02:14Z

Richard, a very useful step to improve the structure and to make functions more uniform. However, for my understand of the code it will be useful to split CalculateAnyOfMaterialResponse into three functions. What is your opinion? It can be next PR.

@markelov208 I fully agree with this observation and would have preferred 3 functions as well (I would really like to decouple this). However, we are bound by the Constitutive Law interface which does all three at the same time in the CalculateMaterialResponseCauchy via the ConstitutiveLaw::Parameters&, so at this moment, I don't see a way to achieve this without doing the same calculation multiple times.

avdg81

This looks good to go for me.

WPK4FEM

Looks consistent and is a step towards understandable code in the elements

rfaasse added 4 commits May 14, 2024 11:43

Extracted a function to calculate constitutive response

6536da2

Formatting

2cb0d31

Simplified constitutive matrix function

69d643e

Simplified constitutive matrix function more by calculating determina…

687004c

…nts inside the function instead of input and use it everywhere the constitutive matrix is being used

rfaasse linked an issue May 14, 2024 that may be closed by this pull request

[GeoMechanicsApplication] Split the calculation of the inputs for the utility functions for element matrices and the calling of the functions #12319

Closed

markelov208 removed a link to an issue May 14, 2024

[GeoMechanicsApplication] Split the calculation of the inputs for the utility functions for element matrices and the calling of the functions #12319

Closed

rfaasse linked an issue May 14, 2024 that may be closed by this pull request

[GeoMechanicsApplication] Split the calculation of the inputs for the utility functions for element matrices and the calling of the functions #12319

Closed

rfaasse added 3 commits May 14, 2024 16:13

Changed signature such that it's clear what is changed by the function

7fa8f04

Created new function for list of constitutive matrices in upw small s…

eb20b9f

…train element and used it in the class (not derived classes yet)

Used new function in derived classes as well

2295b5d

rfaasse changed the title ~~Geo/12319 extract list for constitutive matrices and stress vectors~~ [GeoMechanicsApplication] Extract list for constitutive matrices and stress vectors May 15, 2024

rfaasse added 2 commits May 15, 2024 14:25

Removed redundant determinant calculations

76acc15

Change signature of function to make it more symmetric

f7315bd

rfaasse commented May 15, 2024

View reviewed changes

Merge remote-tracking branch 'origin/master' into geo/12319-extract-l…

1ffdc87

…ist-for-constitutive-matrices-and-stress-vectors

rfaasse marked this pull request as ready for review May 15, 2024 15:07

rfaasse requested review from WPK4FEM, avdg81 and markelov208 May 15, 2024 15:07

Remove unused variable and some redundant comments

f7bf495

WPK4FEM reviewed May 16, 2024

View reviewed changes

rfaasse added 3 commits May 16, 2024 10:30

Changed implementation of CalculateAnyOfMaterialResponse function to …

0c3ca03

…assume any of the stress, strain or matrices vector could be empty or filled and initialized variable_index to fix pipeline build

Make sure the inputs are filled with zero vectors/matrices of the rig…

4a68a07

…ht size

Merge remote-tracking branch 'origin/master' into geo/12319-extract-l…

155f48b

…ist-for-constitutive-matrices-and-stress-vectors

avdg81 reviewed May 16, 2024

View reviewed changes

markelov208 reviewed May 16, 2024

View reviewed changes

Processing review comments

8007aaa

rfaasse requested a review from avdg81 May 16, 2024 12:02

rfaasse requested review from WPK4FEM and markelov208 May 16, 2024 12:02

avdg81 approved these changes May 16, 2024

View reviewed changes

WPK4FEM approved these changes May 16, 2024

View reviewed changes

markelov208 approved these changes May 16, 2024

View reviewed changes

rfaasse merged commit d759e7f into master May 16, 2024
11 checks passed

rfaasse deleted the geo/12319-extract-list-for-constitutive-matrices-and-stress-vectors branch May 16, 2024 13:29

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

[GeoMechanicsApplication] Extract list for constitutive matrices and stress vectors #12380

[GeoMechanicsApplication] Extract list for constitutive matrices and stress vectors #12380

rfaasse commented May 14, 2024 •

edited

Loading

rfaasse May 15, 2024

rfaasse May 15, 2024

WPK4FEM May 16, 2024

WPK4FEM May 16, 2024

rfaasse May 16, 2024

WPK4FEM May 16, 2024

WPK4FEM May 16, 2024

rfaasse May 16, 2024

WPK4FEM May 16, 2024

rfaasse May 16, 2024

WPK4FEM May 16, 2024

rfaasse May 16, 2024

WPK4FEM May 16, 2024

rfaasse May 16, 2024

avdg81 left a comment

markelov208 left a comment

rfaasse commented May 16, 2024 •

edited

Loading

avdg81 left a comment

WPK4FEM left a comment

		GeoElementUtilities::CalculateNuMatrix<TDim, TNumNodes>(Variables.Nu, Variables.NContainer, GPoint);
		GeoElementUtilities::InterpolateVariableWithComponents<TDim, TNumNodes>(

[GeoMechanicsApplication] Extract list for constitutive matrices and stress vectors #12380

[GeoMechanicsApplication] Extract list for constitutive matrices and stress vectors #12380

Conversation

rfaasse commented May 14, 2024 • 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

Choose a reason for hiding this comment

Choose a reason for hiding this comment

avdg81 left a comment

Choose a reason for hiding this comment

markelov208 left a comment

Choose a reason for hiding this comment

rfaasse commented May 16, 2024 • edited Loading

avdg81 left a comment

Choose a reason for hiding this comment

WPK4FEM left a comment

Choose a reason for hiding this comment

rfaasse commented May 14, 2024 •

edited

Loading

rfaasse commented May 16, 2024 •

edited

Loading