CFD Online Discussion Forums

CFD Online Discussion Forums (http://www.cfd-online.com/Forums/)
-   OpenFOAM Running, Solving & CFD (http://www.cfd-online.com/Forums/openfoam-solving/)
-   -   InterFoam mesh dependency (http://www.cfd-online.com/Forums/openfoam-solving/107619-interfoam-mesh-dependency.html)

nimasam October 2, 2012 15:01

InterFoam mesh dependency
 
Hello Friends
i made a simulation by InterFoam for bubble rising in stagnant liquid,
it seems solvers result is mesh dependent and make mesh refined does not make it free from mesh?
has any body similar experience with interFoam?

ata October 4, 2012 02:45

Hi
Do you have contact angle?

Bernhard October 4, 2012 03:45

This is probably due to spurious currents. It is known that they get worse on mesh refinement.

nimasam October 4, 2012 07:16

5 Attachment(s)
Quote:

Originally Posted by ata (Post 384819)
Hi
Do you have contact angle?

no i dont have

however to make it much more clear i put here the result of an spherical bubble rising in stagnant liquid for different mesh,shape is the same but final position in the same time is deferent!

Also, to calculate the terminal velocity of bubble, i calculate the mass gravity of bubble center, then i calculate the slope of mass gravity position in different time,

terminal velocity for different mesh is:
case: terminal velocity (m/s)
40: 0.036
80: 0.030
160: 0.024
320: 0.016

nlinder October 4, 2012 08:39

Hi Nima,

I have several (!) cases where I experience the same problems. Droplets, bubbles with or without contact to the wall, hex tet or poly mesh, you name it.

I did not yet find out where it exactly comes from, but I hope to figure it out some time :)

Just want to let you know, that you are not the only person having these problems! I'll let you know if I have anything new.

Edit: Check also the pressure-field in some cases, they differed extremely from each other due to mesh refinement!

regards
Nicklas

ata October 6, 2012 06:31

Hi Nima
Is this a symmetry BC? If yes this is numerically similar to normal contact angle.Would you please examine with the complete 2-D domain and let me know the results.

nimasam October 6, 2012 08:08

nope it is axisymetric modeling, so frontAndBack are wedge and axis is empty!
you can find the case in above attachment, but i will try it with 2D simulation

alberto October 7, 2012 03:32

As suggested by Ata, could you try your case considering a planar 2D simulation with the whole bubble, and see if the problem persists?

nimasam October 8, 2012 18:45

3 Attachment(s)
Hello
i made a simulation of 2D bubble rising, it seems problem is there:
case: terminal velocity {m/s}
40: 0.0486
80: 0.0411
160: 0.0377

ata October 9, 2012 06:57

Hi
Would you please attach picture of bubbles on the mesh simultaneously.

nimasam October 9, 2012 08:38

whats your mean ata? do you want surface plot with edge to see mesh?
the mesh is uniform hexahedral

ata October 9, 2012 08:44

Hi
Yes. I want to see your mesh resolution.
An other issue. How much is your density and viscosity ratio? How much is the capillary number?

nimasam October 9, 2012 10:33

2 Attachment(s)
Hi
it has been solved with finer mesh for 2D case, this time i examined (320x1600) and terminal velocity was 0.0362 {m/s}, so based on previous result in above post, the geometry with 160x800 is suitable for simulation, so the cell size is about 2.8e-5 {m} but question is remained,
1-why axisymetric modeling shows dependent result?
2-is there any way to solve this problem for axisymetric domain?
3-is there any method which can avoid us from this very fine mesh and give us reasonable result in coarser mesh?

ata October 10, 2012 05:18

Hi
1-Because in the axisymetric case you really simulate a bubble near the wall with 90 degree contact angle.
2-AFAIK nope.
3-Use a more precise scheme.


All times are GMT -4. The time now is 03:06.