Bubble rise velocity different in Basilisk and Gerris

[Niklas Hidman]

Dear Basilisk users,

I'm a new Basilisk user and would like to replicate the example case
from Gerris "Spherical bubble rise in a quiescent bath" in Basilisk to
validate that I specify everything correctly in the Basilisk case file.
Even though I think that the setups are approximately the same, the
results differ between Gerris and Basilisk (see attached .png of the
rise velocity). (I have also tried refining the grid further in Basilisk
but the rise velocity is almost the same.)

Perhaps there are some settings that are incorrect or maybe some scaling
of the results that is different? Attached are the two setup-files that
I've used. Can you see where the error might be?

Thank you in advance,

/Niklas

[Stephane Zaleski]

Hi Niklas, 

  Did you find the solution ? 

  Best

Stéphane

[Niklas Hidman]

Hi Stéphane,

The comparison is still not perfect but I found that, in the Gerris case, the harmonic mean of the viscosity was used. So I implemented that in Basilisk as well and the results agreed better (see attached input file and figure) although there is still a difference.

I was a bit surprised to see such a change of results by using a different viscosity interpolation method. Do you think the harmonic mean always gives more accurate results in bubbly flows and in that case why?

Best,

Niklas

[Stephane Zaleski]

Hi

  The harmonic mean is a better approximation in principle, this is explained in 

Tryggvason, G., Scardovelli, R., & Zaleski, S. (2011). Direct numerical simulations of gas–liquid multiphase flows. Cambridge University Press.

  chapter 3, section 4. 

  you may also wan to check whether in the Basilisk code you have a "filtering" activated. It is activated in Gerris by the line:

 VariableFiltered  T1 T 1

  in Basilisk a similar filter is activated if the macro FILTERED is defined. 

  Best

Stéphane

[Niklas Hidman]

Hi again,

Thank you very much for the help and the reference. With FILTERED defined the results agree much better (see attached figure)!

Best regards,

NIklas Hidman

[Vitor Machado]

Hi Niklas, how did you defined "FILTERED" in your simulation, and what you FILETERED? I'm struggling with the same problem regarding terminal velocity of bubble rising. Could you help me?

[Niklas Hidman]

Hi Vitor,

I defined FILTERED before including the two-phase header file as:

#define FILTERED
#include "two-phase.h"

The filtering is breifly discussed in http://basilisk.fr/src/two-phase.h.

If I understand correctly you use a smeared (i.e. filtered) volume fraction field to smoothen the density/viscosity jump at the interface.

Hope this help you.

Best,

Niklas

[Vitor Machado]

Thanks a lot Niklas! It really seem that not only my terminal velocity was corrected, but the rise path is more comparable to experimental results. When the simulation ends, I'll telll you the results!

Thanks again!

Best, 

Vitor