Extend porous navier stokes by jgonzalezusua · Pull Request #11236 · KratosMultiphysics/Kratos

jgonzalezusua · 2023-06-07T09:30:44Z

📝 Description
This PR is aimed to include the porous version of Navier-Stokes via VMS for both projections, ASGS and OSS.

Please mark the PR with appropriate tags:

Applications
Feature

🆕 Changelog

Renovated alternative_qs_vms_dem_coupled.cpp, alternative_d_vms_dem_coupled.cpp, qs_vms_dem_coupled.cpp and d_vms_dem_coupled.cpp
Added support for quadratic elements
Included tests for quadratic elements to test second derivatives of the shape functions
Added porosity term in cpp tests of the alternative_qs_vms_dem_coupled.cpp, alternative_d_vms_dem_coupled.cpp, qs_vms_dem_coupled.cpp and d_vms_dem_coupled.cpp elements.

…ic subscales

rubenzorrilla · 2023-06-07T09:44:26Z

applications/FluidDynamicsApplication/custom_elements/fluid_element.h

     * @param rCurrentProcessInfo Current ProcessInfo values (input)
     */
-    void CalculateLocalVelocityContribution(
+    virtual void CalculateLocalVelocityContribution(


Why is this required to be virtual?

This function is derived in the elements I have renovated. I need to initialize the second derivatives of the shape functions.

The point is that by doing so you're introducing the vtable overhead. Considering that you know at compile time which elements are high order as we've a dimension and number of nodes template argument, I think we can achieve this in a more generic and efficient way by using an if constexpr together with an overload of the method that calculates the kinematics. Note that this approach would make possible to keep a unique implementation for all the elements deriving from the FluidElement.

Ok, but we shall consider to implement GetShapeSecondDerivatives in FluidElement and modify 'CalculateLocalVelocityContribution' using an if constexpr.

Do you agree with that, @rubenzorrilla?

@rubenzorrilla, I am still confused. If I remove GetShapeSecondDerivatives, where do I implement the transform of the second shape derivatives?

In a new overload of the CalculateGeometryData.

@rubenzorrilla, this means that we must implement GetShapeSecondDerivatives in the fluid_element and overload the UpdateIntegrationPointData to call the second derivatives from data.

No, you do another version of the CalculateGeometryData (method overload) that does what we're currently doing plus the second derivatives.

It seems that this is not solved yet.

rubenzorrilla · 2023-06-07T09:46:48Z

applications/FluidDynamicsApplication/custom_strategies/schemes/bdf2_turbulent_scheme.h

     * @param b Newton-Raphson right hand side (unused)
     */
-    void Update(
+    virtual void Update(


Why is the virtual required?

Because this function is derived in bdf2_turbulent_schemeDEMCoupled.h, which will be merge to master after this PR.

But this is already virtual in the base class... I mean, it makes no sense to add it also here.

rubenzorrilla · 2023-06-07T09:47:11Z

...ions/FluidDynamicsApplication/custom_strategies/schemes/residualbased_simple_steady_scheme.h

+    void InitializeNonLinIteration(
+        ModelPart &rModelPart,
+        TSystemMatrixType &A,
+        TSystemVectorType &Dx,
+        TSystemVectorType &b) override
+    {
+
+        BaseType::InitializeNonLinIteration(rModelPart, A, Dx, b);
+
+    }
+


This is doing nothing. Please remove it and leave current base class implementation.

Yes, it was for a testing implementation, I forgot to remove it.

rubenzorrilla · 2023-06-07T09:49:43Z

...ynamicsApplication/tests/cpp_tests/test_alternative_qs_vms_dem_coupled_quadratic_element.cpp

@@ -0,0 +1,113 @@
+//    |  /           |


I'd add this test to current test_alternative_qs_vms_dem_coupled_element.cpp. Note that you're testing the very same implementation but with different template instantiation.

rubenzorrilla · 2023-06-07T09:50:28Z

...micsApplication/tests/cpp_tests/test_alternative_qs_vms_dem_coupled_quadratic_element_3D.cpp

@@ -0,0 +1,138 @@
+//    |  /           |


I'd add this test to current test_alternative_qs_vms_dem_coupled_element.cpp under the very same reasoning I posted above.

I thought that it would be better to split these tests in three in order to ease the detection of errors. However, I can add them in only one.

The tests are already split by the macro.

Ok, so you mean to write them in the same .cpp file. I thought you meant to join them in only one test. To me it is ok

Exactly this.

RiccardoRossi · 2023-07-24T14:53:01Z

is this done?

jgonzalezusua · 2023-07-25T08:38:48Z

is this done?

@RiccardoRossi, in theory it is done, but we have decided to move the second derivatives transform to KratosCore. We are waiting for the approval of #11253.

jgonzalezusua · 2023-09-29T10:34:01Z

It is very strange since in my local machine all tests run OK. But in github, the results in fluid cpp tests are different... How is it possible?

rubenzorrilla · 2023-10-02T09:38:56Z

It is very strange since in my local machine all tests run OK. But in github, the results in fluid cpp tests are different... How is it possible?

It could be a memory error. Are you running the tests in FullDebug?

jgonzalezusua · 2023-10-02T09:43:30Z

It is very strange since in my local machine all tests run OK. But in github, the results in fluid cpp tests are different... How is it possible?

It could be a memory error. Are you running the tests in FullDebug?

Yes, I am running in FullDebug. That's why is so strange.

jgonzalezusua · 2023-10-13T17:02:05Z

Now, tests should work. I have checked the tests with qs_vms.cpp and they give similar results. The difference stems from the calculation of element size. In our case, we are using AverageElementSize instead of MinimumElementSize.

jgonzalezusua · 2023-10-14T10:44:07Z

@rubenzorrilla tests are runnning. We are ready to merge.

rubenzorrilla

The issue with the virtual seems to be still pending. Aside of this everything looks good to me.

jgonzalezusua · 2023-10-16T12:34:05Z

@rubenzorrilla, I have removed the virtual from fluid_element.h

rubenzorrilla · 2023-10-19T05:44:49Z

applications/FluidDynamicsApplication/custom_elements/alternative_d_vms_dem_coupled.cpp

+    if(Dim == 2){
+        if (NumNodes == 9 || NumNodes == 6 || NumNodes == 4)
+            mInterpolationOrder = 2;
+    }
+    else if(Dim == 3){
+        if (NumNodes == 10 || NumNodes == 27)
+            mInterpolationOrder = 2;
+    }


As these are template parameters, in this case you can use if constexpr so you evaluate the if at compile time.

(I'd do this in a future PR considering that this is already taking a while...)

Yeah, you are right

rubenzorrilla

No more changes in base elements. Thanks.

jgonzalezusua added 7 commits June 6, 2023 17:12

Merge branch 'master' into extend_porous_Navier_Stokes

e2afe0c

IRegistering quadratic elements

70e9a1a

Fixing terms and including transform of second shape derivatives

7ea6cb2

InitializeNonLinearIteration to update the subscale velocity in dynam…

81db336

…ic subscales

including virtual to derive the bdf2 Update function

2573e3e

Including quadratic element tests

e8183f7

Including porosity in tests

15dc987

jgonzalezusua requested a review from a team as a code owner June 7, 2023 09:30

jgonzalezusua added Applications Feature labels Jun 7, 2023

rubenzorrilla reviewed Jun 7, 2023

View reviewed changes

jgonzalezusua added 3 commits June 7, 2023 12:07

removing InitializeNonLinIteration

165a63d

remove useless virtual

e087a21

Joining tests

4775934

jgonzalezusua added 2 commits September 27, 2023 12:55

Merge branch 'master' into extend_porous_Navier_Stokes

1c629c6

including second derivatives implementation from KratosCore

5170481

jgonzalezusua added 3 commits October 13, 2023 11:50

including all terms in bilinear elements

8ba76b4

Merge branch 'master' into extend_porous_Navier_Stokes

cd762a7

Fixing fluid cpp tests

2399f0a

jgonzalezusua requested a review from a team as a code owner October 13, 2023 16:29

jgonzalezusua requested a review from rubenzorrilla October 14, 2023 12:24

rubenzorrilla reviewed Oct 15, 2023

View reviewed changes

removing virtual

72b3815

jgonzalezusua requested a review from rubenzorrilla October 16, 2023 15:15

jgonzalezusua added 4 commits October 17, 2023 10:15

include the correct virtual and removing KRATOS_WATCH

298b44a

removing last virtual of this PR and removing minor comment

67a11d7

changing fluid_fraction_gradient definition

c3e6c90

shorting tests

9de59e2

rubenzorrilla reviewed Oct 19, 2023

View reviewed changes

rubenzorrilla approved these changes Oct 19, 2023

View reviewed changes

jgonzalezusua merged commit ea7f22a into master Oct 19, 2023

jgonzalezusua deleted the extend_porous_Navier_Stokes branch October 19, 2023 08:00

Conversation

jgonzalezusua commented Jun 7, 2023 • edited Loading Uh oh! There was an error while loading. Please reload this page.

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!

jgonzalezusua Jun 8, 2023 • 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!

jgonzalezusua Jun 9, 2023 • 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!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

jgonzalezusua Jun 7, 2023 • 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!

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!

jgonzalezusua Jun 7, 2023 • 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!

RiccardoRossi commented Jul 24, 2023

Uh oh!

jgonzalezusua commented Jul 25, 2023

Uh oh!

jgonzalezusua commented Sep 29, 2023

Uh oh!

rubenzorrilla commented Oct 2, 2023

Uh oh!

jgonzalezusua commented Oct 2, 2023

Uh oh!

jgonzalezusua commented Oct 13, 2023

Uh oh!

jgonzalezusua commented Oct 14, 2023

Uh oh!

rubenzorrilla left a comment

Choose a reason for hiding this comment

Uh oh!

jgonzalezusua commented Oct 16, 2023

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

rubenzorrilla left a comment

jgonzalezusua commented Jun 7, 2023 •

edited

Loading

jgonzalezusua Jun 8, 2023 •

edited

Loading

jgonzalezusua Jun 9, 2023 •

edited

Loading

jgonzalezusua Jun 7, 2023 •

edited

Loading

jgonzalezusua Jun 7, 2023 •

edited

Loading