[SMApp+CLApp] Adding a new triangular thick shell elements + CLs #13921

AlejandroCornejo · 2025-10-27T16:03:45Z

📝 Description
In this PR I am adding:

An improved triangular shell element to the StructuralMechanicsApp. The shell is compatible with linear and finite displacements (corotational). This shell employs the ANDES formulation for membrane and CS-DSG for shear+bending formulation.
A set of shell constitutive laws: plain elasticity and thickness integrated for isotropic plane stress CLs.

The shell has been tested in several examples, improving the behaviour of the current 3 noded thick shell in all cases.

TO-DO's:

add some tests
add printing of variables via CalculateOnIntegrationPoints
add dynamics
add a new shell CL for composite material modelling

This PR needs the Predict issue (#13350) to be merged since I have merged that branch.

FYI @RiccardoRossi @rickyaristio

@loumalouomega

This is the very first step towards attempting to solve #13350 @loumalouomega @AlejandroCornejo @mcgicjn2

applying the ordering change in rve_analysis.py

… fix `IntegrationPointStatisticsTest` with new analysis structure

…alization in `ResidualBasedNewtonRaphsonMPCContactStrategy`

…constitutive-laws

loumalouomega · 2025-10-29T11:39:49Z

@AlejandroCornejo take a look at conflicts, many files not involved look like modified

loumalouomega · 2025-10-29T11:40:30Z

Ah, okay yiou are merging the Predict. This makes difficult the review.

loumalouomega · 2025-10-29T11:41:55Z

...tive/structural_elements_constitutive_laws/thickness_integrated_isotropic_constitutive_law.h

+//     \____/\___/|_| |_|___/\__|_|\__|\__,_|\__|_| \_/ \___\____/\__,_| \_/\_/ |___/\_/ \_/ .__/| .__/
+//                                                                                         |_|   |_|
+//
+//  License:		 BSD License


loumalouomega · 2025-10-29T11:42:31Z

...tive/structural_elements_constitutive_laws/thickness_integrated_isotropic_constitutive_law.h

+
+
+
+


Suggested change

///@}

loumalouomega · 2025-10-29T11:44:59Z

Just one question, do you have a reference or this is your own work?

RiccardoRossi · 2025-11-04T11:18:36Z

@rickyaristio can you take a look at this PR? eventually this shell should replace the existing triangular thick shell..

rickyaristio

I went through most of the CSDSG and ANDES formulation, and everything seems correct. Overall, everything looks good. I only left a few minor comments.

rickyaristio · 2025-11-04T15:13:32Z

...uralMechanicsApplication/custom_elements/shell_elements/cs_dsg3_thick_shell_element_3D3N.cpp

+    TTu(8, 2) = A4;
+    TTu *= (1.0 / A4);
+
+    const double beta0 = 0.5 * (1.0 - 4.0 * std::pow(GetMaterialProperty<double>(POISSON_RATIO, GetProperties()), 2));


I think this is the only difference when comparing the pure membrane case with the other thin shells. Before beta0 is assumed equal to 1. Nice addition!

rickyaristio · 2025-11-04T15:20:52Z

...ve/structural_elements_constitutive_laws/thickness_integrated_isotropic_constitutive_law.cpp

+        const double alpha = 0.1;
+        const double thickness = r_material_properties[THICKNESS];
+        const double t_square = thickness * thickness;
+        const double stenberg_stabilization = (5.0 / 6.0) * t_square / (t_square + alpha * h_max * h_max);


This is the one which comes from the shell (here). With this stabilization, the shells perform better in case of thin problems.

rickyaristio · 2025-11-04T15:39:56Z

...uralMechanicsApplication/custom_elements/shell_elements/cs_dsg3_thick_shell_element_3D3N.cpp

+/***********************************************************************************/
+
+template <bool IS_COROTATIONAL>
+void CSDSG3ThickShellElement3D3N<IS_COROTATIONAL>::CalculateLocalSystem(


Just out of curiosity, why don't you use CalculateAll() and calculation flags, e.g., calculate_stiffness_matrix_flag and calculate_residual_vector_flag for the CalculateLocalSystem(), CalculateLeftHandSide(), and CalculateRightHandSide(), to reduce repetition and keep everything in one place?

rickyaristio · 2025-11-04T16:08:55Z

...ve/structural_elements_constitutive_laws/thickness_integrated_isotropic_constitutive_law.cpp

+        }
+    }
+
+    // Generate locations (from top to bottom: +t/2 -> -t/2)


I may have missed it, but how about SHELL_OFFSET? I think the old Section class defines this one.

TO-DO! thank you

rickyaristio · 2025-11-04T16:19:06Z

Just one question, do you have a reference or this is your own work?

This shell combines some parts already implemented from other existing shells in Kratos, as far as I know (correct me if I am wrong, @AlejandroCornejo).

membrane+drilling part: https://doi.org/10.1016/S0045-7825(03)00253-6, https://doi.org/10.1002/nme.3109
shear part: https://doi.org/10.1016/S0045-7949(99)00140-6, https://doi.org/10.1002/nme.4506

AlejandroCornejo · 2025-11-14T08:40:09Z

Just one question, do you have a reference or this is your own work?

Hi Vicente, thank you for taking your time to read the PR. I had to merge the predict thing in my branch to properly test if the corotational implementation is OK. Hence we'll wait until the predict is merged to add this PR.

Besides, I have added the 3 main references to implement the 3 noded shell element, regarding the membrane (Felippa), shear+bending (Bletzinger, Rama) and corotational (Haugen and Felippa).

There is nothing "new" regarding the formulation, just a nice combination of different strain-displacement matrices from 2 papers.

loumalouomega · 2025-11-14T08:43:42Z

Just one question, do you have a reference or this is your own work?

Hi Vicente, thank you for taking your time to read the PR. I had to merge the predict thing in my branch to properly test if the corotational implementation is OK. Hence we'll wait until the predict is merged to add this PR.

Besides, I have added the 3 main references to implement the 3 noded shell element, regarding the membrane (Felippa), shear+bending (Bletzinger, Rama) and corotational (Haugen and Felippa).

There is nothing "new" regarding the formulation, just a nice combination of different strain-displacement matrices from 2 papers.

Okay, I think I am not blocking so I will let other to review and if everything OK I will approve (well until predict is not merged this should not be merged)

AlejandroCornejo · 2025-11-14T08:44:02Z

Just one question, do you have a reference or this is your own work?

This shell combines some parts already implemented from other existing shells in Kratos, as far as I know (correct me if I am wrong, @AlejandroCornejo).
* membrane+drilling part:  https://doi.org/10.1016/S0045-7825(03)00253-6, https://doi.org/10.1002/nme.3109

* shear part: https://doi.org/10.1016/S0045-7949(99)00140-6,  https://doi.org/10.1002/nme.4506

That is right, @rickyaristio. Additionally, this new shell (and the subsequent 4 noded one) will be fully compatible with al the constitutive power available in the ConstitutiveLawsApplication via subproperties. This is a "game changer" in many situations (e.g. composite layered with rule of mixtures and not orthotropic elasticity).

AlejandroCornejo · 2025-11-14T08:48:23Z

@rickyaristio I'd like to know your opinion about where to place the k=5/6 reduction factor multiplication. Currently I have implemented this operation in the Constitutive Law. @RiccardoRossi thinks that since this factor is a result of the FE formulation (I agree), this operation must be placed in the element and the CL to be agnostic from this.

However, it seems that if we use a layered composite shell this k should be recomputed somehow (if you have any hint, it would be highly appreciated). But this goes against the idea of "the element should know" because it does depend on the constitutive law to be used!

What do you guys think?

BTW, the Stenberg factor I'll definitely move it to the element since it does not depend on the CL to be used (AFAIK).

AlejandroCornejo · 2025-11-14T11:48:08Z

@rickyaristio I'd like to know your opinion about where to place the k=5/6 reduction factor multiplication. Currently I have implemented this operation in the Constitutive Law. @RiccardoRossi thinks that since this factor is a result of the FE formulation (I agree), this operation must be placed in the element and the CL to be agnostic from this.

However, it seems that if we use a layered composite shell this k should be recomputed somehow (if you have any hint, it would be highly appreciated). But this goes against the idea of "the element should know" because it does depend on the constitutive law to be used!

What do you guys think?

BTW, the Stenberg factor I'll definitely move it to the element since it does not depend on the CL to be used (AFAIK).

For now we have agreed to maintain the addition of k in the constituive law.

RiccardoRossi and others added 30 commits May 15, 2025 12:14

Attempting to solve #13350

9cec182

This is the very first step towards attempting to solve #13350 @loumalouomega @AlejandroCornejo @mcgicjn2

Merge branch 'master' into FixUpdateOrderInAnalysisStage

cb9175b

Update rve_analysis.py

ba48231

applying the ordering change in rve_analysis.py

Update structural_mechanics_prebuckling_analysis.py

3297295

Merge branch 'master' into FixUpdateOrderInAnalysisStage

0a347c3

adding shell elastic CL

5f9bb0c

Adding references

a900c8d

clean CL

31eb5ec

shell CL elastic ok

eab177b

ok

fc6bcea

final version of elastic CL

2da87f0

Add time_scheme setting to solver settings in test_serializer.json

d417f3a

adding some more error msgs

783b4d5

adding base 3N shell element

4bf58bc

Merge branch 'master' into FixUpdateOrderInAnalysisStage

ca17632

f

32c68f3

building B

d7c5e5f

B ok

9a8c633

B averaged

a033c27

adding clone

4be1bda

add init material

b99b55f

minor

1fdd0d4

add dofs

4088abd

minor

c1ad6e7

[FluidDynamicsApplication] Refactor FluidAnalysisWithoutSolution to…

afbfbdb

… fix `IntegrationPointStatisticsTest` with new analysis structure

on it

f8cacf6

compiles

38dbfff

adding to Init

5ac29f8

[ContactStructuralMechanicsApplication] Improve handling of DOF initi…

1eada2a

…alization in `ResidualBasedNewtonRaphsonMPCContactStrategy`

Remove unused include directive from contact strategy headers

d1f5183

AlejandroCornejo added 3 commits October 28, 2025 18:03

using the node 1 ref coordinates as local system origin

7092835

using temp matrix in LHS

2e8682a

Merge branch 'FixUpdateOrderInAnalysisStage' into CLApp/adding-shell-…

0e5d225

…constitutive-laws

loumalouomega reviewed Oct 29, 2025

View reviewed changes

RiccardoRossi requested a review from rickyaristio November 4, 2025 11:18

rickyaristio reviewed Nov 4, 2025

View reviewed changes

AlejandroCornejo added 6 commits November 7, 2025 09:03

Merge branch 'master' into CLApp/adding-shell-constitutive-laws

f5adf59

Merge branch 'master' into CLApp/adding-shell-constitutive-laws

1c6481e

complete comments and references

216b555

minor

11dcba9

using -=

4363bb2

some comments

19b4a7b

AlejandroCornejo added 5 commits November 14, 2025 11:29

optimizing the GetDofsList

8fc3c5e

reverting change in old shell

a9143be

adding one test for the linear triangle with elastic CL

54f3848

adding corotational test

69b5f10

nice optimization

64c4e1c

AlejandroCornejo added 2 commits November 19, 2025 14:07

solving a bug when plasticity

850fffc

remove unused vars

dab9667

[SMApp+CLApp] Adding a new triangular thick shell elements + CLs #13921

Are you sure you want to change the base?

[SMApp+CLApp] Adding a new triangular thick shell elements + CLs #13921

Conversation

AlejandroCornejo commented Oct 27, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

loumalouomega commented Oct 29, 2025

Uh oh!

loumalouomega commented Oct 29, 2025

Uh oh!

loumalouomega Oct 29, 2025

Choose a reason for hiding this comment

Uh oh!

loumalouomega Oct 29, 2025

Choose a reason for hiding this comment

Uh oh!

loumalouomega commented Oct 29, 2025

Uh oh!

RiccardoRossi commented Nov 4, 2025

Uh oh!

rickyaristio left a comment

Choose a reason for hiding this comment

Uh oh!

rickyaristio Nov 4, 2025

Choose a reason for hiding this comment

Uh oh!

rickyaristio Nov 4, 2025

Choose a reason for hiding this comment

Uh oh!

rickyaristio Nov 4, 2025

Choose a reason for hiding this comment

Uh oh!

rickyaristio Nov 4, 2025

Choose a reason for hiding this comment

Uh oh!

AlejandroCornejo Nov 14, 2025

Choose a reason for hiding this comment

Uh oh!

rickyaristio commented Nov 4, 2025

Uh oh!

AlejandroCornejo commented Nov 14, 2025

Uh oh!

loumalouomega commented Nov 14, 2025

Uh oh!

AlejandroCornejo commented Nov 14, 2025

Uh oh!

AlejandroCornejo commented Nov 14, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

AlejandroCornejo commented Nov 14, 2025

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

7 participants

AlejandroCornejo commented Oct 27, 2025 •

edited

Loading

AlejandroCornejo commented Nov 14, 2025 •

edited

Loading