fluid shell interaction model. #184

DrChiZhang · 2022-12-15T14:03:02Z

1, add fluid shell interaction dynamics.
2, add new test case of sloshing flow with baffle
3, generalise complex dynamics according with adding third contact body to contact body list.

This reverts commit 65bc702, reversing changes made to 125be31.

merge with master.

This reverts commit 8e36627.

Here, cmake is modified, pay attention when PR.

SPHINXsys/src/shared/particle_dynamics/dissipation_dynamics/particle_dynamics_dissipation.hpp

BenceVirtonomy · 2022-12-15T18:48:54Z

SPHINXsys/src/shared/particle_dynamics/dissipation_dynamics/particle_dynamics_dissipation.hpp

+		Real mass_i = mass_[index_i];
+		VariableType &variable_i = variable_[index_i];
+
+		std::array<Real, MaximumNeighborhoodSize> parameter_b;


These arrays can cause memory errors for multiresolution, where the neighborhood size is larger than the MaximumNeighborhoodSize variable. I think we mentioned this once but I don't remember what our conclusion was with prof. Hu. It's not very efficient either, perhaps @FabienPean-Virtonomy has a better solution?

Yes it happened I had some problems there in the past, it was more related to some system resolution compared to particle spacing. There will be issues for multiresolution definitely.

Two simpler options I see:

The simpler fix would be using a std::vector with reserved memory initial memory (by calculating max neighborhood size)

Using a couple array + vector and adjusting the loops to fill the vector once the array is filled.

Otherwise other solutions could involve a different container outside STL (but require adding dependency), thread local vector (need to be confident about concurrency)

BenceVirtonomy · 2022-12-15T18:51:17Z

SPHINXsys/src/shared/particle_dynamics/dissipation_dynamics/particle_dynamics_dissipation.hpp

+			{
+				size_t index_j = contact_neighborhood.j_[n];
+
+				parameter_b[n] = contact_particles_[k]->DegeneratedSpacing(index_j) * eta_ * contact_neighborhood.dW_ijV_j_[n] * Vol_i * dt / contact_neighborhood.r_ij_[n];


what is "DegeneratedSpacing"? Could we have a more understandable name?

Shell formulation is called Degenerated Approach, therefore, DegeneratedSpacing used here denoting the spacing being degenerated.

What does degenerated mean in this context? Can you explain it please?

What I thought is " shell particle is degenerated (having low standard or behavior) ) from volume particle in the thickness spacing. Then, with shell particle is interacting with volume particle, the degenerated spacing is considered.
If you have other suggestion, please let me know.

Do you mean that you consider the shell particle as a squashed solid particle in the thickness dimension? If yes, I would just call it ThicknessAdjustedSpacing for example, much clearer I think.

Name it "Thinkness" is one of my old choices, while this name is meaningless for volume particles as they do not have thickness. Maybe "ReducedThickness", "ReducedCrossSection", "DegeneratedThickness" or "DegeneratedCrossSection" is more readable.

"ReducedThickness", "ReducedCrossSection" sound good to me

Ok my problem was more with spacing than degenerate, I went to look in the code and it is basically the reduced or degenerated dimension (return 1 for volume, thickness for shells, and area for beams).
I would either go by crossSectionDimension or reducedDimension (or their getXXX variant)

Also why is the naming convention changing ? It was using camelCase before, now part of the new code is in PascalCase. Please change uniformize back to camelCase

It is implemented by virtual function, which I am not so comfortable with, but I don't have a straightforward drop-in replacement to offer at the moment.

SPHINXsys/src/shared/particle_dynamics/fluid_dynamics/fluid_shell_interaction.hpp

SPHINXsys/src/shared/particle_dynamics/solid_dynamics/shell_fluid_interaction.h

BenceVirtonomy · 2022-12-18T10:43:55Z

@ChiZhangatTUM We would like to make a 3d test here too. I was wondering if you already have the elastic beam in a flow case at least in 2d? I mean test_2d_fsi2 case with the shell? I'm asking you to avoid doing something you have done already.
If you have it, I would try to expand it to 3d otherwise build it from scratch.

DrChiZhang · 2022-12-18T11:55:54Z

@ChiZhangatTUM We would like to make a 3d test here too. I was wondering if you already have the elastic beam in a flow case at least in 2d? I mean test_2d_fsi2 case with the shell? I'm asking you to avoid doing something you have done already.
If you have it, I would try to expand it to 3d otherwise build it from scratch.

Sure, I have already done some work.
You can follow tests/user-exampels/test-2d-square-fsi in fluid-shell-interaction branch .

BenceVirtonomy · 2022-12-18T14:18:28Z

@ChiZhangatTUM We would like to make a 3d test here too. I was wondering if you already have the elastic beam in a flow case at least in 2d? I mean test_2d_fsi2 case with the shell? I'm asking you to avoid doing something you have done already.
If you have it, I would try to expand it to 3d otherwise build it from scratch.

Sure, I have already done some work. You can follow tests/user-exampels/test-2d-square-fsi in fluid-shell-interaction branch .

Ahh, I see. Could you make a draft PR from that branch? That way we can all see the changes there too.

Xiangyu-Hu · 2022-12-19T08:39:48Z

SPHINXsys/src/shared/particle_dynamics/dissipation_dynamics/particle_dynamics_dissipation.h

+	};
+
+	/**
+	 * @class DampingPairwiseFromShell


Which test cases uses this class?

Xiangyu-Hu · 2022-12-19T08:41:27Z

SPHINXsys/src/shared/particle_dynamics/dissipation_dynamics/particle_dynamics_dissipation.hpp

+				variable_[index_i] += parameter_b[n] * (variable_i - variable_k[index_j]) / (mass_i - 2.0 * parameter_b[n]);
+			}
+			// backward sweep
+			for (size_t n = contact_neighborhood.current_size_; n != 0; --n)


I do feel that we need not repeat so much codes here.

Xiangyu-Hu · 2022-12-19T08:43:10Z

SPHINXsys/src/shared/particle_dynamics/fluid_dynamics/fluid_dynamics_complex.cpp

@@ -36,7 +36,7 @@ namespace SPH
 					Vecd nablaW_ijV_j = contact_neighborhood.dW_ijV_j_[n] * contact_neighborhood.e_ij_[n];

 					// acceleration for transport velocity
-					acceleration_trans -= 2.0 * nablaW_ijV_j;
+					acceleration_trans -= 2.0 * nablaW_ijV_j * this->contact_particles_[k]->DegeneratedSpacing(index_j);


this will decrease efficiency of the original code a lot, as it needs another value from particle j.

Xiangyu-Hu · 2022-12-19T08:47:15Z

SPHINXsys/src/shared/particle_dynamics/fluid_dynamics/fluid_dynamics_complex.hpp

@@ -88,7 +88,8 @@ namespace SPH
 				Neighborhood &contact_neighborhood = (*this->contact_configuration_[k])[index_i];
 				for (size_t n = 0; n != contact_neighborhood.current_size_; ++n)
 				{
-					sigma += contact_neighborhood.W_ij_[n] * contact_inv_rho0_k * contact_mass_k[contact_neighborhood.j_[n]];
+					Real mass = this->contact_particles_[k]->ParticleMass(contact_neighborhood.j_[n]);


Same issue on decreasing efficiency. We should have a discussion on this.

Xiangyu-Hu · 2022-12-19T08:48:08Z

SPHINXsys/src/shared/particle_dynamics/fluid_dynamics/fluid_shell_interaction.h

@@ -0,0 +1,136 @@
+/* -------------------------------------------------------------------------*


This should not be long to fluid dynamics.

Xiangyu-Hu · 2022-12-19T08:51:58Z

SPHINXsys/src/shared/particle_dynamics/fluid_dynamics/fluid_shell_interaction.h

+		* @brief Abstract base class for general fluid-shell interaction model. 
+		* @note  Here, 
+		*/
+		template<class BaseIntegrationType>


I am thinkin the current formulation is too complicated. We need a discuss on this.

Xiangyu-Hu · 2022-12-19T09:16:23Z

1, add fluid shell interaction dynamics. 2, add new test case of sloshing flow with baffle 3, generalise complex dynamics according with adding third contact body to contact body list.

Dear Chi, I think that we need a discussion on the pull request to clarify several issues.

DrChiZhang · 2022-12-19T09:51:04Z

1, add fluid shell interaction dynamics. 2, add new test case of sloshing flow with baffle 3, generalise complex dynamics according with adding third contact body to contact body list.

Dear Chi, I think that we need a discussion on the pull request to clarify several issues.

Sure, I will be office from Tuesday to Thursday.

DrChiZhang · 2022-12-19T09:53:54Z

Hi all, As we have lots of issues on this PR. I will close it without merging in the following days.

BenceVirtonomy · 2022-12-19T10:34:06Z

Hi all, As we have lots of issues on this PR. I will close it without merging in the following days.

You don't need to close it Chi, unless you want to start from scratch. You can set it back to draft so that we can still see the changes.

DrChiZhang · 2022-12-20T10:02:34Z

I will add it to the user_examples of this current branch in the next commit, and you can modify and add it to test with proper validation.

Poiseuille flow with shell failing

DrChiZhang added 27 commits November 25, 2022 11:47

update for checking out.

f8fade7

Merge branch 'merge_eigen_with_master' into fluid-shell-interaction

5950b9b

add shell fluid interaction.

ec0e3a9

Merge branch 'merge_eigen_with_master' into fluid-shell-interaction

80a1659

Sloshing test run, will check the accuracy.

3443ced

add dam break impacting thin structure.

d5e664c

Merge branch 'master' into fluid-shell-interaction

681eb49

Fluid-sehll with Riemann solver.

125be31

Merge branch 'master' into fluid-shell-interaction.

65bc702

Revert "Merge branch 'master' into fluid-shell-interaction."

8e36627

This reverts commit 65bc702, reversing changes made to 125be31.

fluid-shell model with transport and surface indicatior.

f2a5922

bug fix for fluid force on shell.

b6ceedd

hydrostatic FSI with shell model, add damping from shell class.

471c930

flow induced vibration of filament attached to square cylinder.

b707560

not work tests.

39d41a4

small modification.

89a9bc6

merge master.

fb290cb

merge master.

c6d04f8

Merge branch 'master' into fluid-shell-interaction

624fbd1

merge with master.

Revert "Revert "Merge branch 'master' into fluid-shell-interaction.""

9a08163

This reverts commit 8e36627.

modification after merge with master.

ff9f30f

Here, cmake is modified, pay attention when PR.

small modification.

6c2d127

remove examples before merging wiht master.

2011946

Merge branch 'master' into fsi-merge-with-master

235b67f

small changes.

d793dff

all tests pass with new test of sloshing flow with baffle.

5c7cfed

Small change before PR.

c678a18

DrChiZhang requested review from BenceVirtonomy, Xiangyu-Hu and FabienPean-Virtonomy December 15, 2022 14:03

DrChiZhang requested a review from DongWuTUM December 15, 2022 18:11

BenceVirtonomy reviewed Dec 15, 2022

View reviewed changes

add veriable size reverse.

69d626e

Xiangyu-Hu requested changes Dec 19, 2022

View reviewed changes

This comment was marked as outdated.

Sign in to view

add new test.

8799ebb

BenceVirtonomy requested a review from YuVirtonomy March 12, 2023 20:38

YuVirtonomy and others added 7 commits March 31, 2023 11:00

solid wall with poiseuille

f385aa1

shelll with poiseuille

311d7e6

was using wrong radius_mid_surface

ec0b095

format

30e2d15

U_f should be 0.5* Re / diameter * mu_f

0532bbd

inflow condition at the end of Acoustic time step

6544d38

Merge pull request #228 from Xiangyu-Hu/feature/poiseuille

9fefef0

Poiseuille flow with shell failing

DrChiZhang mentioned this pull request Oct 11, 2023

Draft PR: one-sided fluid-shell interaction #453

Closed

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

fluid shell interaction model. #184

fluid shell interaction model. #184

DrChiZhang commented Dec 15, 2022

BenceVirtonomy Dec 15, 2022

FabienPean-Virtonomy Dec 19, 2022

BenceVirtonomy Dec 15, 2022

DrChiZhang Dec 16, 2022

BenceVirtonomy Dec 16, 2022

DrChiZhang Dec 16, 2022

BenceVirtonomy Dec 17, 2022

DrChiZhang Dec 18, 2022

BenceVirtonomy Dec 18, 2022 •

edited

Loading

FabienPean-Virtonomy Dec 19, 2022

BenceVirtonomy commented Dec 18, 2022

DrChiZhang commented Dec 18, 2022

BenceVirtonomy commented Dec 18, 2022

Xiangyu-Hu Dec 19, 2022

Xiangyu-Hu Dec 19, 2022

Xiangyu-Hu Dec 19, 2022

Xiangyu-Hu Dec 19, 2022

Xiangyu-Hu Dec 19, 2022

Xiangyu-Hu Dec 19, 2022

Xiangyu-Hu commented Dec 19, 2022

DrChiZhang commented Dec 19, 2022

DrChiZhang commented Dec 19, 2022

BenceVirtonomy commented Dec 19, 2022

DrChiZhang commented Dec 20, 2022

This comment was marked as outdated.

		@@ -0,0 +1,136 @@
		/* -------------------------------------------------------------------------*

fluid shell interaction model. #184

Are you sure you want to change the base?

fluid shell interaction model. #184

Conversation

DrChiZhang commented Dec 15, 2022

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

BenceVirtonomy Dec 18, 2022 • edited Loading

Choose a reason for hiding this comment

Choose a reason for hiding this comment

BenceVirtonomy commented Dec 18, 2022

DrChiZhang commented Dec 18, 2022

BenceVirtonomy commented Dec 18, 2022

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

Xiangyu-Hu commented Dec 19, 2022

DrChiZhang commented Dec 19, 2022

DrChiZhang commented Dec 19, 2022

BenceVirtonomy commented Dec 19, 2022

DrChiZhang commented Dec 20, 2022

This comment was marked as outdated.

BenceVirtonomy Dec 18, 2022 •

edited

Loading