Cahn-Hilliard-Navier-Stokes #1013

PierreLaurentinCS · 2024-02-01T23:15:58Z

Description of the problem

The last implementation of the Cahn-Hilliard-Navier-Stokes coupled equations were incorrectly reproducing classical test cases such as the rising bubble or the Rayleigh-Taylor instability
The problem resided in the formulation of the Cahn-Hilliard equation and also major mistakes in the computation of physical properties in the coupling between Navier-Stokes and Cahn-Hilliard
The thought process which eventually led to a change in the initial set of equation led to other additions. Among them : chemical potential smoothing, phase filtering (clipping and tanh) and other minor modifications

Description of the solution

The Cahn-Hilliard equations were fixed and the use of mobility as physical property made it more readable and compact
The coefficients multiplying the energies (bulk and interface) now depend on two parameters : epsilon (proportional to the size of the cell) and lambda which itself depends on the size of the cell and surface tension

How Has This Been Tested?

All the former tests were fixed to account for the new assembly : mms2d and 3d, cahn-hilliard-navier-stokes coupling, time-dependant boundary condition
A new test was added to test the phase filtering
Many canonical tests were performed. This should be illustrated soon with examples.

Documentation

Nothing was changed in the documentation, this will be the work of another PR

Future changes

During the development a gls stabilization was successfully implemented but it was not kept in this version because of the new assembly. This will be added in a future PR.
The fixed angle of contact BC is not functional for extreme angles. New formulations will be tested in future PRs

blaisb

FIrst batch of comments. Could you please add within the example folder the CHN examples you have, even if they are not documented, this way we can start keeping track of them. At least static bubble and rising bubble.

applications_tests/lethe-fluid/mms2d_cahn_hilliard.prm

include/core/interface_property_model.h

include/core/mobility_cahn_hilliard_model.h

source/core/parameters.cc

source/solvers/cahn_hilliard.cc

source/solvers/cahn_hilliard_assemblers.cc

AmishgaAlphonius

Great job @PierreLaurentinCS!!! It is nice to see everything coming together after so many months of hard work!!

I haven't finished my review. I'll look at the assemblers on Monday when I'll have the equations :)

applications_tests/lethe-fluid/cahn_hilliard_navier_stokes_coupling.prm

include/solvers/vof.h

applications_tests/lethe-fluid/cahn_hilliard_navier_stokes_coupling.prm

source/solvers/cahn_hilliard_assemblers.cc

source/solvers/navier_stokes_scratch_data.cc

source/solvers/cahn_hilliard_scratch_data.cc

source/solvers/navier_stokes_base.cc

AmishgaAlphonius · 2024-02-02T21:32:37Z

source/solvers/navier_stokes_base.cc

+              if (this->simulation_parameters.multiphysics.cahn_hilliard)
+                {
+                  this->pcout << "L2 error pressure : "
+                              << std::setprecision(
+                                   simulation_control->get_log_precision())
+                              << error_pressure << std::endl;
+                }


Why is the pressure L2 error norm only displayed for CH?

I thought, since it was not implemented before that there was a specific reason for it, so I left it only for the Cahn-Hilliard simulations

The issue is that for the VOF solver the pressure is sometimes erratic (it will translate by a constant) so generally pressure in VOF is not a very portable variable. I'M not sure if it's a good idea to enable it in a test. We can try and if it is portable we will keep it, if no, we can remove it in the future.

The problem is the same with CHNS, the pressure is correct up to a constant (but the pressure jump is the Laplace pressure jump). Agreed on the future removal if it's not portable enough, I'll keep it this way (compute pressure error for vof OR chns) for the moment

blaisb

Moar comments

examples/multiphysics/rising-bubble-chns/rising-bubble-chns.prm

source/solvers/cahn_hilliard.cc

blaisb

A little bit moar.

source/core/parameters_multiphysics.cc

blaisb · 2024-02-06T14:56:16Z

source/solvers/navier_stokes_base.cc

+              if (this->simulation_parameters.multiphysics.cahn_hilliard)
+                {
+                  this->pcout << "L2 error pressure : "
+                              << std::setprecision(
+                                   simulation_control->get_log_precision())
+                              << error_pressure << std::endl;
+                }


The issue is that for the VOF solver the pressure is sometimes erratic (it will translate by a constant) so generally pressure in VOF is not a very portable variable. I'M not sure if it's a good idea to enable it in a test. We can try and if it is portable we will keep it, if no, we can remove it in the future.

AmishgaAlphonius

Only some minor comments :)
Great work again!

examples/multiphysics/rising-bubble-chns/rising-bubble-chns.prm

doc/source/parameters/cfd/cahn_hilliard.rst

include/solvers/multiphysics_interface.h

include/solvers/physical_properties_manager.h

…lations

…consistent and consisten SUPG in CH

…ry corrections for epsilon

…sembly

Description of the problem The last implementation of the Cahn-Hilliard-Navier-Stokes coupled equations were incorrectly reproducing classical test cases such as the rising bubble or the Rayleigh-Taylor instability The problem resided in the formulation of the Cahn-Hilliard equation and also major mistakes in the computation of physical properties in the coupling between Navier-Stokes and Cahn-Hilliard The thought process which eventually led to a change in the initial set of equation led to other additions. Among them : chemical potential smoothing, phase filtering (clipping and tanh) and other minor modifications Description of the solution The Cahn-Hilliard equations were fixed and the use of mobility as physical property made it more readable and compact The coefficients multiplying the energies (bulk and interface) now depend on two parameters : epsilon (proportional to the size of the cell) and lambda which itself depends on the size of the cell and surface tension How Has This Been Tested? All the former tests were fixed to account for the new assembly : mms2d and 3d, cahn-hilliard-navier-stokes coupling, time-dependant boundary condition A new test was added to test the phase filtering Many canonical tests were performed. This should be illustrated soon with examples. Documentation Nothing was changed in the documentation, this will be the work of another PR Future changes During the development a gls stabilization was successfully implemented but it was not kept in this version because of the new assembly. This will be added in a future PR. The fixed angle of contact BC is not functional for extreme angles. New formulations will be tested in future PRs Former-commit-id: fbe87ab

PierreLaurentinCS requested review from AmishgaAlphonius and blaisb February 2, 2024 14:48

blaisb requested changes Feb 2, 2024

View reviewed changes

AmishgaAlphonius requested changes Feb 2, 2024

View reviewed changes

AmishgaAlphonius assigned PierreLaurentinCS Feb 2, 2024

blaisb added the Reviewed and waiting for changes label Feb 4, 2024

PierreLaurentinCS added Ready for review and removed Ready for review labels Feb 6, 2024

PierreLaurentinCS force-pushed the cahn_hilliard_navier_stokes branch from 95414e6 to d336c0b Compare February 6, 2024 01:26

blaisb requested changes Feb 6, 2024

View reviewed changes

examples/multiphysics/rising-bubble-chns/rising-bubble-chns.prm Outdated Show resolved Hide resolved

source/solvers/cahn_hilliard.cc Outdated Show resolved Hide resolved

blaisb requested changes Feb 6, 2024

View reviewed changes

AmishgaAlphonius approved these changes Feb 7, 2024

View reviewed changes

PierreLaurentinCS added 18 commits February 8, 2024 11:54

new assembly is settled

0729966

added smoothing term on potential

a550e08

added potential smoothing parameter

cf1022a

indent

b80cda8

well height re-introduced

bcd8028

epsilon set as twice cell_size

bdc2847

return to the old assembly + added pressure computation for chns simu…

096e240

…lations

indent

5844772

correctly added pressur error for chns simulations

01788a6

small modif + indent

01f7f7d

remove relative diffusive flux

516befa

fixed viscosity, phase statistics, mobility parameter, implemented in…

da14417

…consistent and consisten SUPG in CH

relative diffusive flux + mobility

958747e

relative diffusive flux correction

d4bcaae

filtration, W: mobility constant to 1e-7

f370280

temp commit

f9a53b8

mobility works (mat. interac to 0 im prm!!)

7a67f49

small modifs

5db2f8c

PierreLaurentinCS added 23 commits February 8, 2024 11:54

implementation of tanh filter (non functionnal)

85ef6b7

modified stf term in chns assembly

1280959

rising bubble MVP

9752743

cleanup starts from there

92784d0

cleaned tanh filter

9d0f76b

cleaned gls stabilization in ch assembler

13fa932

cleaned spring constant, well height, and epsilon and applied necessa…

e1c0ea2

…ry corrections for epsilon

implementation of mobility as physical property 1

9d60b04

fixed automatic epsilon scaling and added do-nothing filter

a01651c

removed uncessary mobility parameters and unsued variables in chns as…

2d9711f

…sembly

added clip filter and added the correct epsilon term in the assemblers

3dc1f01

added tests for different chns filters

98e0ba0

fixed angle of contact BC

74c2f47

indent

d9e7040

fixed epsilon, fixed 2d,3dmms and coupling

f30228f

removed comments, fixed time-dependent ch BC

f954c80

indent

fb5aa4c

fixed remaining warnings and unused features

69f69bf

removed unnecessary cahn_hilliard example folder

9a83d17

apply part 1 of first batch of comments

07c6485

applied part 2 of first batch of review comments

b2b19ad

applied second batch of comments

94785c1

applied second batch of comments

7d1680f

PierreLaurentinCS force-pushed the cahn_hilliard_navier_stokes branch from d336c0b to 7d1680f Compare February 8, 2024 16:58

PierreLaurentinCS added the Ready for review label Feb 8, 2024

blaisb approved these changes Feb 9, 2024

View reviewed changes

blaisb merged commit fbe87ab into master Feb 9, 2024
8 checks passed

blaisb deleted the cahn_hilliard_navier_stokes branch February 9, 2024 03:28

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Cahn-Hilliard-Navier-Stokes #1013

Cahn-Hilliard-Navier-Stokes #1013

PierreLaurentinCS commented Feb 1, 2024 •

edited

Loading

blaisb left a comment

AmishgaAlphonius left a comment

AmishgaAlphonius Feb 2, 2024 •

edited

Loading

PierreLaurentinCS Feb 6, 2024

blaisb Feb 6, 2024

PierreLaurentinCS Feb 6, 2024

blaisb left a comment

blaisb left a comment

blaisb Feb 6, 2024

AmishgaAlphonius left a comment

Cahn-Hilliard-Navier-Stokes #1013

Cahn-Hilliard-Navier-Stokes #1013

Conversation

PierreLaurentinCS commented Feb 1, 2024 • edited Loading

Description of the problem

Description of the solution

How Has This Been Tested?

Documentation

Future changes

blaisb left a comment

Choose a reason for hiding this comment

AmishgaAlphonius left a comment

Choose a reason for hiding this comment

AmishgaAlphonius Feb 2, 2024 • edited Loading

Choose a reason for hiding this comment

PierreLaurentinCS Feb 6, 2024

Choose a reason for hiding this comment

blaisb Feb 6, 2024

Choose a reason for hiding this comment

PierreLaurentinCS Feb 6, 2024

Choose a reason for hiding this comment

blaisb left a comment

Choose a reason for hiding this comment

blaisb left a comment

Choose a reason for hiding this comment

blaisb Feb 6, 2024

Choose a reason for hiding this comment

AmishgaAlphonius left a comment

Choose a reason for hiding this comment

PierreLaurentinCS commented Feb 1, 2024 •

edited

Loading

AmishgaAlphonius Feb 2, 2024 •

edited

Loading