Isothermal compressible NS assembler #815

AmishgaAlphonius · 2023-07-18T20:07:58Z

Description of the problem

When experimenting with the dam break with an obstacle example, it was noticed that the software performed poorly in mimicking the wave's natural shape after the impact. It was assumed that this is due to the lack of considering the air's natural compressibility.

Description of the solution

In an attempt to solve this issue, an isothermal compressible fluid dynamics Volume of Fluid (VOF) model will be implemented. In a previous PR, an isothermal ideal gas density model was implemented. As a second step, an isothermal compressible formulation of the Navier-Stokes (NS) equations is implemented. Only the continuity equation was adapted to take account of compressibility.

How Has This Been Tested?

Two tests were added:
applications_tests/gls_navier_stokes_2d/mms2d_isothermal_compressible_ns_gls.prm
applications_tests/gls_navier_stokes_3d/mms3d_isothermal_compressible_ns_gls.prm
The restart_01 unit test was updated so that the fluid's density model is set to constant so that it uses the incompressible NS assemblers:
/tests/solvers/restart_01.cc

Documentation

Documentation was updated
doc/source/parameters/cfd/physical_properties.rst

Future changes

Currently, in two phase flow simulations, the isothermal ideal gas density model can only be used in conjunction with the incompressible NS equations. However, it is meant to be used with the isothermal formulation of compressible NS equations to account for weakly compressible flows. In a future update, this change will be implemented.
At the moment, in the momentum equation, the\nabla \cdot (\nabla \mathbf{u})^T term is neglected, however this might be an inappropriate simplification, since in the compressible form, \nabla \mathbf{u} is not necessarily equal to 0. This will be investigated in a future PR and changes will be implemented if it is evaluated necessary.

Comments

Small corrections where made to the fluid dynamics theory:
doc/source/theory/fluid_dynamics/fem_formulation.rst
doc/source/theory/fluid_dynamics/navier-stokes.rst
doc/source/theory/fluid_dynamics/stabilization.rst

hepap

I will finish tomorrow, I still have to check the isothermal_compressible_navier_stokes_assembler.cc

doc/source/parameters/cfd/physical_properties.rst

doc/source/theory/fluid_dynamics/stabilization.rst

hepap · 2023-07-19T18:13:10Z

doc/source/theory/fluid_dynamics/stabilization.rst

  \\
-  & + \sum_{k} \int_{\Omega_k} \left( \partial_t u_k + u_l \partial_l u_k + \partial_k p - \nu \partial_l \partial_l u_k - f_k \right) \cdot \left(\tau_{GLS} \nu \partial_l \partial_l v_k \right) \mathrm{d}\Omega_k  
+  & - \sum_{k} \int_{\Omega_k} \left( \partial_t u_k + u_l \partial_l u_k + \partial_k p - \nu \partial_l \partial_l u_k - f_k \right) \cdot \left(\tau_{GLS} \nu \partial_l \partial_l v_k \right) \mathrm{d}\Omega_k


I am not sure of the sign, but in Laura's proposal, it is +, and in lethe it seems to be -

I based the modification on what was implemented in Lethe; I did ask @blaisb if it was supposed to be a minus, but I'll let him comment on it again here just to make sure that everything is good :)

As far as I know this should be a plus. I guess since it is implemented with a minus we should change the sign in the documentation for now. Let's discuss this afterwards with @blaisb.

include/solvers/isothermal_compressible_navier_stokes_assembler.h

include/solvers/navier_stokes_scratch_data.h

hepap

I checked everything now! I don't have more comment at the moment, but I will redo the weak formualtion (just to be sure I understand well hehe :D)

source/solvers/isothermal_compressible_navier_stokes_assembler.cc

PierreLaurentinCS

Excellent work, I have got nothing to say except for some typos

applications_tests/gls_navier_stokes_2d/mms2d_isothermal_compressible_ns_gls.prm

applications_tests/gls_navier_stokes_3d/mms3d_isothermal_compressible_ns_gls.prm

include/core/density_model.h

include/solvers/isothermal_compressible_navier_stokes_assembler.h

lpsaavedra

Good job! Looking forward to the next PR related to this solver.

lpsaavedra · 2023-07-24T21:00:44Z

doc/source/theory/fluid_dynamics/navier-stokes.rst



-In Lethe, instead of solving for pressure directly, we solve for a reduced pressure :math:`p^*=\frac{p}{\rho}`. This has two advantages, first the scaling of the pressure is more adequate with respect to that of the velocity and, secondly, for single phase simulations, this means that we require a single physical property: the kinematic viscosity. Consequently, unless specified otherwise, you should assume that pressure actually refers to the reduced pressure :math:`p^*`. Although it may seem confusing at first, this is standard in incompressible fluid dynamics.
+In Lethe, instead of solving for pressure directly, we solve for a reduced pressure :math:`p=\frac{p^*}{\rho}`. This has two advantages, first the scaling of the pressure is more adequate with respect to that of the velocity and, secondly, for single phase simulations, this means that we require a single physical property: the kinematic viscosity. Consequently, unless specified otherwise, you should assume that pressure actually refers to the reduced pressure :math:`p`. Although it may seem confusing at first, this is standard in incompressible fluid dynamics.


I am not sure about this change of notation here, since in literature the reduced pressure is always noted as p*.

It is in fact confusing, I'll add a note to specify that this choice was made to lighten the writing in the documentation :)

doc/source/theory/fluid_dynamics/stabilization.rst

lpsaavedra · 2023-07-24T21:11:58Z

doc/source/theory/fluid_dynamics/stabilization.rst

  \\
-  & + \sum_{k} \int_{\Omega_k} \left( \partial_t u_k + u_l \partial_l u_k + \partial_k p - \nu \partial_l \partial_l u_k - f_k \right) \cdot \left(\tau_{GLS} \nu \partial_l \partial_l v_k \right) \mathrm{d}\Omega_k  
+  & - \sum_{k} \int_{\Omega_k} \left( \partial_t u_k + u_l \partial_l u_k + \partial_k p - \nu \partial_l \partial_l u_k - f_k \right) \cdot \left(\tau_{GLS} \nu \partial_l \partial_l v_k \right) \mathrm{d}\Omega_k


As far as I know this should be a plus. I guess since it is implemented with a minus we should change the sign in the documentation for now. Let's discuss this afterwards with @blaisb.

Co-authored-by: hepap <[email protected]>

Co-authored-by: PierreLaurentinCS <[email protected]>

…documentation

This reverts commit 1afb412.

Description of the problem When experimenting with the dam break with an obstacle example, it was noticed that the software performed poorly in mimicking the wave's natural shape after the impact. It was assumed that this is due to the lack of considering the air's natural compressibility. Description of the solution In an attempt to solve this issue, an isothermal compressible fluid dynamics Volume of Fluid (VOF) model will be implemented. In a previous PR, an isothermal ideal gas density model was implemented. As a second step, an isothermal compressible formulation of the Navier-Stokes (NS) equations is implemented. Only the continuity equation was adapted to take account of compressibility. How Has This Been Tested? Two tests were added: applications_tests/gls_navier_stokes_2d/mms2d_isothermal_compressible_ns_gls.prm applications_tests/gls_navier_stokes_3d/mms3d_isothermal_compressible_ns_gls.prm The restart_01 unit test was updated so that the fluid's density model is set to constant so that it uses the incompressible NS assemblers: /tests/solvers/restart_01.cc Documentation Documentation was updated: doc/source/parameters/cfd/physical_properties.rst Future changes Currently, in two phase flow simulations, the isothermal ideal gas density model can only be used in conjunction with the incompressible NS equations. However, it is meant to be used with the isothermal formulation of compressible NS equations to account for weakly compressible flows. In a future update, this change will be implemented. At the moment, in the momentum equation, the\nabla \cdot (\nabla \mathbf{u})^T term is neglected, however this might be an inappropriate simplification, since in the compressible form, \nabla \mathbf{u} is not necessarily equal to 0. This will be investigated in a future PR and changes will be implemented if it is evaluated necessary. Comments Small corrections where made to the fluid dynamics theory: doc/source/theory/fluid_dynamics/fem_formulation.rst doc/source/theory/fluid_dynamics/navier-stokes.rst doc/source/theory/fluid_dynamics/stabilization.rst --------- Co-authored-by: hepap <[email protected]> Co-authored-by: PierreLaurentinCS <[email protected]> Former-commit-id: 73a4237

AmishgaAlphonius requested review from blaisb and hepap July 18, 2023 20:07

AmishgaAlphonius force-pushed the isothermal_compressible_NS_assembler branch from c941844 to 0ec7197 Compare July 18, 2023 20:11

AmishgaAlphonius changed the title ~~Isothermal compressible ns assembler~~ Isothermal compressible NS assembler Jul 18, 2023

AmishgaAlphonius added the Ready for review label Jul 18, 2023

AmishgaAlphonius force-pushed the isothermal_compressible_NS_assembler branch 2 times, most recently from 712e481 to cad9018 Compare July 19, 2023 11:26

AmishgaAlphonius added WIP When a PR is open but not ready for review Ready for review and removed Ready for review WIP When a PR is open but not ready for review labels Jul 19, 2023

hepap reviewed Jul 19, 2023

View reviewed changes

PierreLaurentinCS self-requested a review July 19, 2023 22:34

AmishgaAlphonius force-pushed the isothermal_compressible_NS_assembler branch from 07c55ef to b5f92b6 Compare July 20, 2023 11:11

AmishgaAlphonius changed the title ~~Isothermal compressible NS assembler~~ Isothermal compressible NS assembler [WIP] Jul 20, 2023

AmishgaAlphonius changed the title ~~Isothermal compressible NS assembler [WIP]~~ Isothermal compressible NS assembler Jul 20, 2023

AmishgaAlphonius added Ready for review and removed WIP When a PR is open but not ready for review labels Jul 20, 2023

hepap reviewed Jul 21, 2023

View reviewed changes

source/solvers/isothermal_compressible_navier_stokes_assembler.cc Show resolved Hide resolved

AmishgaAlphonius force-pushed the isothermal_compressible_NS_assembler branch from 7534662 to 9604b31 Compare July 21, 2023 17:09

AmishgaAlphonius requested a review from lpsaavedra July 21, 2023 17:11

AmishgaAlphonius force-pushed the isothermal_compressible_NS_assembler branch from 69c7197 to adfa444 Compare July 24, 2023 12:54

PierreLaurentinCS reviewed Jul 24, 2023

View reviewed changes

AmishgaAlphonius force-pushed the isothermal_compressible_NS_assembler branch from 2a5152c to a8c660f Compare July 24, 2023 15:22

PierreLaurentinCS approved these changes Jul 24, 2023

View reviewed changes

AmishgaAlphonius force-pushed the isothermal_compressible_NS_assembler branch from a8c660f to 209fd95 Compare July 24, 2023 21:07

lpsaavedra approved these changes Jul 25, 2023

View reviewed changes

AmishgaAlphonius force-pushed the isothermal_compressible_NS_assembler branch from 8053009 to 1afb412 Compare July 25, 2023 20:49

AmishgaAlphonius and others added 18 commits July 26, 2023 12:10

Add isothermal incompressible NS assembler

a0f558e

Optimize ICNS assembler and add comments

7a681d6

Update doc, add test and final touches

d82f5d6

Small corrections to doc

3e927b4

Revert to assembler with density and dynamic viscosity

b59e238

Apply suggestions from code review

0f31e53

Co-authored-by: hepap <[email protected]>

Update mms and other corrections

e6e0c9e

Update restart_01 unit test (precise constant density model)

f71ce64

Update include/solvers/navier_stokes_scratch_data.h

7639dd5

Co-authored-by: hepap <[email protected]>

Indent

2d6f699

Correct gather velocity history

314d507

Apply suggestions from code review

1a55175

Co-authored-by: PierreLaurentinCS <[email protected]>

Remove unnecessary lines

78f7182

Small correction to doc

53f643d

Adjust notation in stabilization doc

07f1179

Add note on the notation of the pressure and reduced pressure in the …

9acdd1c

…documentation

Edit assembler

716ef8d

Revert "Edit assembler"

a79f502

This reverts commit 1afb412.

AmishgaAlphonius force-pushed the isothermal_compressible_NS_assembler branch from 4dcd573 to 1aaeec5 Compare July 26, 2023 16:11

Remove unnecessary lines

db2cd26

AmishgaAlphonius force-pushed the isothermal_compressible_NS_assembler branch from 1aaeec5 to db2cd26 Compare July 26, 2023 16:16

hepap approved these changes Jul 26, 2023

View reviewed changes

AmishgaAlphonius added Reviewed and ready to merge Rebased and removed Ready for review labels Jul 26, 2023

lpsaavedra merged commit 73a4237 into master Jul 27, 2023

AmishgaAlphonius deleted the isothermal_compressible_NS_assembler branch July 27, 2023 15:30

AmishgaAlphonius self-assigned this Aug 1, 2023

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Isothermal compressible NS assembler #815

Isothermal compressible NS assembler #815

AmishgaAlphonius commented Jul 18, 2023 •

edited

Loading

hepap left a comment

hepap Jul 19, 2023

AmishgaAlphonius Jul 20, 2023

lpsaavedra Jul 24, 2023

hepap left a comment

PierreLaurentinCS left a comment

lpsaavedra left a comment

lpsaavedra Jul 24, 2023

AmishgaAlphonius Jul 25, 2023

lpsaavedra Jul 24, 2023



		In Lethe, instead of solving for pressure directly, we solve for a reduced pressure :math:`p^=\frac{p}{\rho}`. This has two advantages, first the scaling of the pressure is more adequate with respect to that of the velocity and, secondly, for single phase simulations, this means that we require a single physical property: the kinematic viscosity. Consequently, unless specified otherwise, you should assume that pressure actually refers to the reduced pressure :math:`p^`. Although it may seem confusing at first, this is standard in incompressible fluid dynamics.
		In Lethe, instead of solving for pressure directly, we solve for a reduced pressure :math:`p=\frac{p^*}{\rho}`. This has two advantages, first the scaling of the pressure is more adequate with respect to that of the velocity and, secondly, for single phase simulations, this means that we require a single physical property: the kinematic viscosity. Consequently, unless specified otherwise, you should assume that pressure actually refers to the reduced pressure :math:`p`. Although it may seem confusing at first, this is standard in incompressible fluid dynamics.

Isothermal compressible NS assembler #815

Isothermal compressible NS assembler #815

Conversation

AmishgaAlphonius commented Jul 18, 2023 • edited Loading

Description of the problem

Description of the solution

How Has This Been Tested?

Documentation

Future changes

Comments

hepap left a comment

Choose a reason for hiding this comment

hepap Jul 19, 2023

Choose a reason for hiding this comment

AmishgaAlphonius Jul 20, 2023

Choose a reason for hiding this comment

lpsaavedra Jul 24, 2023

Choose a reason for hiding this comment

hepap left a comment

Choose a reason for hiding this comment

PierreLaurentinCS left a comment

Choose a reason for hiding this comment

lpsaavedra left a comment

Choose a reason for hiding this comment

lpsaavedra Jul 24, 2023

Choose a reason for hiding this comment

AmishgaAlphonius Jul 25, 2023

Choose a reason for hiding this comment

lpsaavedra Jul 24, 2023

Choose a reason for hiding this comment

AmishgaAlphonius commented Jul 18, 2023 •

edited

Loading