CFD Online Discussion Forums

CFD Online Discussion Forums (
-   OpenFOAM Running, Solving & CFD (
-   -   How to simulate the gravitydriven RayleighTaylor instability (

luckyluke April 18, 2005 09:08

What is the main difficulty in
What is the main difficulty included in this problem? I was told that this problem is hard.
Can I solve it using InterFoam?

thibault_pringuey January 6, 2009 10:53

I have set the gravity-driven
I have set the gravity-driven RT instability using the funkySetFields utility to give the initial sine shape to the interface and running the case with interFoam.

Unfortunately, my solution does not compare satisfactorily with previous work (Lopez et al. 2005 jcp, Popinet: and show additional "ligaments"along the interface.

Modifying the thickness of the interface, the grid refinement and/or the surface tension did not allow to match the references given above.

Has anyone obtained a satisfactory solution on this problem?

egp January 12, 2009 19:37

Thibault, How does this loo

How does this look?

This solution was on a fine grid of 256x1024. In general, I don't like this problem as a validation test case since there is neither an analytical solution or experimental data. In addition, the variation in published results is pretty large.

However, it is good from the point of view that it is a good tutorial on the use of funkySetFields, blockMesh, and interFoam!

egp January 12, 2009 19:43

Hmmm, the image-posting gremli
Hmmm, the image-posting gremlin....

Try these links:

gamma field

pd field

I did do a grid study (256x1024, 128x512, 64x256, 32x128), and the coarser meshes were unable to support thin strands (as would be expected).

Also, surface tension had a pretty big impact on the solution. I used a value close to that for helium & air.

I'll post the cleaned up case later.

egp January 13, 2009 08:16

The case can be downloaded fro
The case can be downloaded from:

fine grid case for the Rayleigh-Taylor problem

Don't forget to run blockMesh. There is a funkySetFieldsDict, but you shouldn't have to run it (0/gamma has the sin-wave I.C.).

Have fun...

thibault_pringuey January 13, 2009 10:28

Thank you very much Eric! Your
Thank you very much Eric! Your solution looks great.

I will study your case files to identify the issue in mine.

BTW, could please let me know how you included links to images in your message?

Many thanks.

egp January 13, 2009 12:51

Thibault, See the "Formatti

See the "Formatting" link on the left hand side of the Message Board browser window? If you follow this link, you can learn how to do all sorts of formatting, like changing the color, <u>underlining text</u>, and making text annoyingly blink!. There are also the commands,


However, the \image command did not work for me the last time I tried. Maybe I was using Safari instead of Firefox.


thibault_pringuey January 16, 2009 06:49

Hello Eric, The differences
Hello Eric,

The differences with my simulation concerned the solvers and schemes employed. Thank you for making me realize this.

I have run the simulation on a 128x512 and a 256x1024 mesh with your settings. I have noticed that there are some "wiggles" along the interface in the region of the neck of the mushroom-shaped gamma distribution. These "wiggles" are amplified on the 128x512 mesh.

Would you know what is causing these "wiggles"? Are they due to spurious currents?

Many thanks.


mike_jaworski January 19, 2009 13:27

Hi Thibault and Eric, I jus
Hi Thibault and Eric,

I just thought I'd drop a note about validation for this instability. There are analytical treatments of the growth rate of the instability when it is still a "small" disturbance and the initial equilibrium can be linearized. The growth rate depends on the wave number of the disturbance (the classic text on this stuff is probably Chandrasekhar's "Hydrodynamic and Hydromagnetic stability"). This could provide validation of the solver for small-scale disturbances before the problem goes non-linear.

For the non-linear growth etc, I'm not sure about the best approach to validating this except by comparison to literature and experiments.

Mike J.

egp January 19, 2009 14:50

@Thibault: I also saw the int
@Thibault: I also saw the interface wiggles in the neck region. It is in the high-velocity region of the flow, and is sensitive to the value of surface tension. Without more study, I don't think I can ascribe it to either numerical instability or something physical such as capillary waves.

@Mike: Good suggestion on testing the linear regime of the instability. However, I'll leave that to someone else! Maybe a young energetic student can chew on this as part of the verification section of their thesis.

sega March 3, 2009 07:49

Hello. Talking about "wiggl

Talking about "wiggles".
Do you mean something like this?

This is a calculation using interDyMFoam.

And what are these strange non-orthogonal cells in the region some distance away from the interface? Or is this due to the visualization within ParaView?

sega March 3, 2009 07:56

Well hope it will work this ti
Well hope it will work this time.
Like this:

sega March 5, 2009 06:34

This behaviour was due to the
This behaviour was due to the usage of CrankNicholson 1 for ddtSchemes.

Using the Euler Scheme solved the problem.
Have a look:

All times are GMT -4. The time now is 01:34.