Add DCDD shock capture stabilization term to heat transfer #1141

voferreira · 2024-05-15T20:50:11Z

Description of the problem

Cross-wind flow leads to numerical instability in the heat transfer solver.
Some of this instability was not being sufficiently dampened by the GGLS term we had before, especially at highly predominant convective problems (high Pe).

Description of the solution

Apply the Discontinuity-Capturing Directional Dissipation (DCDD) stabilization term by Tezduyar (2003).
The term identifies crosswind regions and adds to them an artificial diffusivity, dampening the instabilities.

How Has This Been Tested?

Canonical circular advection-diffusion problem in steady state.
Some tests needed to be updated.

Documentation

To be improved

Future changes

Other terms exist that can be added, such as LSIC. For now, I believe this one will suffice.

…dcdd

hepap

Nice addition!

I'm just not sure I understand how the equation 79 of the paper was converted for the temperature. Maybe you can explain it to me?

source/solvers/heat_transfer_assemblers.cc

hepap · 2024-05-16T13:36:50Z

source/solvers/heat_transfer_assemblers.cc

+      const double vdcdd = 0.5 * h * h * u_mag *
+                           previous_temperature_gradient.norm() /
+                           scratch_data.global_T_mag;


Which equation of the paper are you implementing here? Is it 79?

hepap · 2024-05-16T13:38:53Z

source/solvers/heat_transfer_assemblers.cc

+
+      // We remove the diffusion aligned with the velocity
+      // as is done in the original article.  In Tezduyar 2003, this is
+      // denoted s.


Suggested change

// denoted s.

// denoted s in equation (67).

hepap · 2024-05-16T13:47:30Z

source/solvers/heat_transfer_assemblers.cc

+      const Tensor<2, dim> k_corr =
+        Utilities::fixed_power<2>(gradient_unit_vector * velocity_unit_vector) *
+        outer_product(velocity_unit_vector, velocity_unit_vector);


Suggested change

const Tensor<2, dim> k_corr =

Utilities::fixed_power<2>(gradient_unit_vector * velocity_unit_vector) *

outer_product(velocity_unit_vector, velocity_unit_vector);

// Denoted (r.s)^2*ss in equation (70) of Tezduyar 2003.

const Tensor<2, dim> k_corr =

Utilities::fixed_power<2>(gradient_unit_vector * velocity_unit_vector) *

outer_product(velocity_unit_vector, velocity_unit_vector);

hepap · 2024-05-16T13:48:09Z

source/solvers/heat_transfer_assemblers.cc

+      const Tensor<2, dim> k_corr =
+        Utilities::fixed_power<2>(gradient_unit_vector * velocity_unit_vector) *
+        outer_product(velocity_unit_vector, velocity_unit_vector);
+      const Tensor<2, dim> gradient_unit_tensor =


Suggested change

const Tensor<2, dim> gradient_unit_tensor =

// Denoted rr in equation (70) of Tezduyar 2003.

const Tensor<2, dim> gradient_unit_tensor =

hepap · 2024-05-16T13:49:00Z

source/solvers/heat_transfer_assemblers.cc

+        outer_product(velocity_unit_vector, velocity_unit_vector);
+      const Tensor<2, dim> gradient_unit_tensor =
+        outer_product(gradient_unit_vector, gradient_unit_vector);
+      const Tensor<2, dim> dcdd_factor = gradient_unit_tensor - k_corr;


Suggested change

const Tensor<2, dim> dcdd_factor = gradient_unit_tensor - k_corr;

// Denoted rr - (r.s)^2*ss in equation (70) of Tezduyar 2003.

const Tensor<2, dim> dcdd_factor = gradient_unit_tensor - k_corr;

hepap · 2024-05-16T13:50:56Z

source/solvers/heat_transfer_assemblers.cc

+      // solver.
+      const double tolerance = 1e-12;
+
+      // In Tezduyar 2003, this is denoted r


Can you add the same comment as the previous for the equations and terms?

voferreira · 2024-05-22T00:28:49Z

@blaisb My bad. I will rebase from the next time on.

blaisb · 2024-05-22T12:23:15Z

is this ready to review or? The flags says ready to review but the tests do not pass.

voferreira · 2024-05-22T12:28:33Z

It is ready to be reviewed. There is only one test failing. It is a new test and it is just missing updating. I will update it when I arrive at the office.

blaisb · 2024-05-22T12:37:16Z

@voferreira What about my comment regarding convergence? I did not see any answer for the high Pe case.

blaisb

I'm good with the PR, but there are two points remaining
1- Why was the vof test changed to be without VOF now? If this test is without VOF now, the name of the test should be changed and the reason for this change should be explained in the PR ( I am unsure why such a change would be necessary)
2- What happens to the convergence of the high Pe case?

blaisb · 2024-05-22T12:37:53Z

applications_tests/lethe-fluid/heat_transfer_vof_phase_change_constrain_solid_domain.output

 Pressure drop: 0 Pa
 Total pressure drop: 0 Pa

----------------------


Why is the test case without VOF now? I don't understand, this used to be VOF test case...

It should still be a test for VOF. I guess the results for the same test but without VOF were put here by mistake :) it's almost the same file names! However, I think both should change with this modification.

hepap · 2024-05-22T15:53:25Z

applications_tests/lethe-fluid/heat_transfer_vof_phase_change_constrain_solid_domain.output

 Pressure drop: 0 Pa
 Total pressure drop: 0 Pa

----------------------


It should still be a test for VOF. I guess the results for the same test but without VOF were put here by mistake :) it's almost the same file names! However, I think both should change with this modification.

source/solvers/heat_transfer_assemblers.cc

voferreira · 2024-05-22T17:25:58Z

@blaisb Sorry for the delay about the comment on the high Pe, but I wanted to confirm with Mikael before. The test was not passing when: 1- we were using the previous solution for steady and transient, and 2- when we were testing the term without the velocity component. Now the test is not affected anymore. As for the VOF test, it was changed by mistake and I am fixing it just now.

Co-authored-by: hepap <47506601+hepap@users.noreply.github.com>

blaisb · 2024-05-22T18:14:50Z

Good for me then, i'll wait for the other to approve before merging

PierreLaurentinCS

Very fine addition, good job, just switch the indices in the assembly and we're good

include/solvers/heat_transfer.h

PierreLaurentinCS · 2024-05-22T19:15:55Z

source/solvers/heat_transfer.cc

+  double delta_T_ref =
+    std::max(solution_maximum - solution_minimum, minimum_delta_T_ref);
+
+  // Set minimum value as 1 to prevent overshooting of diffusivity


This comment should be put above the definition of delta_T_ref to explain why we use std::max, not after

source/solvers/heat_transfer_assemblers.cc

PierreLaurentinCS

Everything is fine, good job!

…dcdd

voferreira added 12 commits May 12, 2024 18:32

Add feature

9f3085e

Update test

83d6ff3

Redimensionalize the artificial viscosity

e1a7842

Checkpoint solution

b42ef3c

Checkpoint before rebasing

653f3cd

Merge remote-tracking branch 'origin/master' into heat_shock_capture_…

2c37d7e

…dcdd

Fix extra for indentation

4e1704d

Add printing at bug

766e63c

Bug found, will fix in another PR

97be454

Commit tested and working implementation

541cb82

Make the term explicit in time

3d9e5ca

Update tests

7b29dd8

voferreira requested review from blaisb, hepap and mivaia May 15, 2024 20:50

voferreira added Hackathon Ready for review labels May 15, 2024

voferreira added 3 commits May 15, 2024 16:51

Remove extra file created by mistake

373d922

Fix cavity restart test

f0503c2

Fix test cavity

ca57f4c

voferreira requested a review from PierreLaurentinCS May 15, 2024 21:42

voferreira added 2 commits May 15, 2024 18:06

Fix unnecessary include in the header of heat transfer scratch data

b9b4c6d

Fix test output

6559e0d

hepap requested changes May 16, 2024

View reviewed changes

voferreira assigned mivaia and PierreLaurentinCS May 16, 2024

voferreira added WIP When a PR is open but not ready for review Only documentation missing and removed Ready for review labels May 16, 2024

Modify shock absorption term

de61e99

blaisb added Reviewed and waiting for changes and removed Ready for review labels May 22, 2024

blaisb reviewed May 22, 2024

View reviewed changes

hepap reviewed May 22, 2024

View reviewed changes

voferreira and others added 2 commits May 22, 2024 13:29

Apply suggestions by Hélène

15113f6

Co-authored-by: hepap <47506601+hepap@users.noreply.github.com>

Fix mistake on VOF test

f53b129

blaisb self-requested a review May 22, 2024 18:14

blaisb approved these changes May 22, 2024

View reviewed changes

voferreira added Ready for review and removed Reviewed and waiting for changes labels May 22, 2024

PierreLaurentinCS requested changes May 22, 2024

View reviewed changes

PierreLaurentinCS added Reviewed and waiting for changes and removed Ready for review labels May 22, 2024

voferreira added 2 commits May 22, 2024 18:04

Fix coherence between reference, documentation and implementation

39a821e

Indent

33cc47a

voferreira requested a review from PierreLaurentinCS May 22, 2024 22:10

hepap approved these changes May 23, 2024

View reviewed changes

PierreLaurentinCS approved these changes May 23, 2024

View reviewed changes

Merge remote-tracking branch 'origin/master' into heat_shock_capture_…

bff2742

…dcdd

voferreira added Rebased Reviewed and ready to merge and removed Reviewed and waiting for changes labels May 23, 2024

blaisb merged commit 372edfb into master May 23, 2024
8 checks passed

blaisb deleted the heat_shock_capture_dcdd branch May 23, 2024 22:47

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Add DCDD shock capture stabilization term to heat transfer #1141

Add DCDD shock capture stabilization term to heat transfer #1141

voferreira commented May 15, 2024 •

edited

hepap left a comment

hepap May 16, 2024

hepap May 16, 2024

hepap May 16, 2024

hepap May 16, 2024

hepap May 16, 2024

hepap May 16, 2024

voferreira commented May 22, 2024

blaisb commented May 22, 2024

voferreira commented May 22, 2024

blaisb commented May 22, 2024

blaisb left a comment

blaisb May 22, 2024

hepap May 22, 2024

hepap May 22, 2024

voferreira commented May 22, 2024

blaisb commented May 22, 2024

PierreLaurentinCS left a comment

PierreLaurentinCS May 22, 2024

PierreLaurentinCS left a comment

	const Tensor<2, dim> gradient_unit_tensor =
	// Denoted rr in equation (70) of Tezduyar 2003.
	const Tensor<2, dim> gradient_unit_tensor =

	const Tensor<2, dim> dcdd_factor = gradient_unit_tensor - k_corr;
	// Denoted rr - (r.s)^2*ss in equation (70) of Tezduyar 2003.
	const Tensor<2, dim> dcdd_factor = gradient_unit_tensor - k_corr;

Add DCDD shock capture stabilization term to heat transfer #1141

Add DCDD shock capture stabilization term to heat transfer #1141

Conversation

voferreira commented May 15, 2024 • edited

Description of the problem

Description of the solution

How Has This Been Tested?

Documentation

Future changes

hepap left a 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

voferreira commented May 22, 2024

blaisb commented May 22, 2024

voferreira commented May 22, 2024

blaisb commented May 22, 2024

blaisb left a 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

voferreira commented May 22, 2024

blaisb commented May 22, 2024

PierreLaurentinCS left a comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

PierreLaurentinCS left a comment

Choose a reason for hiding this comment

voferreira commented May 15, 2024 •

edited