Bubble in zero gravity conditions

ngj · May 20, 2008, 10:33

Hi Sebastian

I do not know if I should congratulate you then

On the other hand it is always interesting to hear how other programs are performing.
I have been playing around with your setup, and these symmetry conditions seems to cause problems ... maybe that the spurious waves on the interfaces wants to go trough the symmetry plane but is restricted as it is not wave-transmissive?!!? Thus could you try to connect the two symmetry-planes in some way...? I have not tried the wedge type of block myself, but can you connect two faces in a wedge irrespectively of the orientation!?! That might help you to remove the large velocities at the boundary!?!?! As you can see I am only guessing wildly, and then you can do all the work

Have fun, hope you find a solution!

- Niels

sega · May 20, 2008, 12:24

Hey.

I think before I start a wild gambling with the symmetry planes I rather restart the case without the benefits of symmetry.

I will come back with the results ... bet on it!

Greetings.

ngj · May 21, 2008, 04:29

Hi Sebastian

I tried a different kind of setup yesterday, which looks something like this:

So disregarding the small deviation from the correct volume fraction (order of magnitude 0.1% error), I initialized the 'bubble' with a square at rest. After running it, the result came within 2% of the theoretical value, but the circle induced some rather nasty volumes, which gave rise to larger pressures in the 'corners' of the circle. So a better approximation of the circle, as you have already discussed with others, would probably give better results.

On the other hand, this small experiment shows that your problems is highly grid related, as the results are significantly better when a clear representation of the circle is thought into the grid. Some would call it cheating, intelligent, but still cheating

Hope it helps,

Niels

P.S. Cannot give you my configuration as the computer, which I used, is not connected to the internet.

sega · May 21, 2008, 08:19

Your recommended setup is not working by now.
I tried to use the upper right quarter of the computational domain.
But as I have mentioned earlier this symmetry is not working.

The solver is stuck from the first time step with timesteps of e-6 ...

I will try it with the complete computational domain.

Get back to you.

sega · May 21, 2008, 09:33

So, now I tried the static initialized spherical bubble:

Despite the fact that the bubble is moving (any ideas why? - wall effects?) the simulation caluclates a little smaller value of the pressure inside the bubble, as expected.

I think that is due to the "distributed" (not "sharp") interface at the beginning. The liquid phase volume fraction is therefore 0.4 % bigger than expected.
This will lead to a slighlty bigger radius and a theoretical pressure jump of about 11.77 N/m².

So, what magnitude does the pressure have inside the bubble?
How can I tell to which number the "orange" pressure is belonging?

sega · May 22, 2008, 01:45

Ok, I have used the threshold-function to display the interior of the bubble and did a "Reset Range" on the pressure plot.

Looks kind of strange because of the missing parts inside the bubble. Why is that?

Now I guess, that the pressure inside the bubble is about 10.8 N/m². Thats still 8.3 % Error.

sega · May 25, 2008, 03:56

So, now I did a calculation on a mesh which is spherical at the final position of the bubble - like Niels suggested.

But there are still problems.
The pressure is a little bit higher, but not constant inside the bubble.
I think this is due to the mesh inside the bubble-region ...

Have a look at these images:
http://therealsega.th.funpic.de/openfoam/pd.png
http://therealsega.th.funpic.de/openfoam/pd_wire.png

And why is the pressure distorted at the lower left part of the interface?

Greetings

sega · May 25, 2008, 05:52

I tried to give the mesh some arcs in the interiour of the bubble, but the solver is stuck.
Maybe the control volumes are too much contortet.

http://therealsega.th.funpic.de/openfoam/mesharc.png

Niels. Maybe you should look up how you have done your spherical mesh.

ngj · May 26, 2008, 03:31

Hi Sebastian

1. I have had the same problem with moving bubbles ... for some reason the velocity field become asymmetric and then the friction moves the bubble to one of the side.

2. I have formatted my computer and my run went the same way. But I have done the mesh in exactly the way as sketched above in my post from the 21. of May.

3. I have also experienced those weird oscillations on the surface when the resolution becomes finer than some certain threshold.

Best regards,

Niels

sega · May 26, 2008, 08:36

A guy from the university could decrease the error in his pressure-jump to about 4% ...
He was using Fluent with a TRIANGULAR mesh ...

sega · September 23, 2008, 11:02

Hey!

It's me again. I'm still thinking about this case. As I have recognised there is a difference between the value of volPointInterpolate(pd) and cell(pd).

As the pressure-jump over the interface looks pretty nice (and constant) in volPointInterpolate(pd) it IS NOT when you are looking at the cell(pd)-values itself.

I used some threshold-filters in ParaView to get a feeling about the difference between the values.
It looks like there are some few cells at the inside of the interface who have a pressure like it should be inside the drop. But the number of cells with less than the desired pressure is increasing near the center of the drop.

So, the volPointInterpolate(pd) value suggestes a difference of about 6% to the desired pressure.
While the cell(pd) values range from 2-13% with more than half of the cells in the range of 10-13% error.

So, how trustworthy is the volPointInterpolate(pd)?

maximus · September 23, 2008, 11:08

Seems you missed the pictures.
Curious about what you just said.
Will wait until you load them up :-)...

Grüß
Maximus

maximus · September 23, 2008, 11:09

oops I did it again ....

sorry

sega · September 23, 2008, 11:42

I can see the pictures.
The above one is volPointInterpolate(pd).

??

sega · September 25, 2008, 13:10

Lets get back at this. Obviously the problem is really challenging.

Meanwhile I did a lot of different stuff with the case. Lets have a look at some points I couldn't fix myself.

First of all, I have used some wrong transportProperties all the time. Obviously I have mixed dynamic and kinematic viscosity (in some weird way). Now, the right values are in place.
As a matter of fact these setting should not influence the solution. Wishful thinking!

With kinematic viscosity round about 10^(-6) the simulation times are rather high.
Like Niels I'm now experiencing a movement of the bubble with the long simulation times. Have a look:
http://therealsega.th.funpic.de/openfoam/surfaceTensionOriginalTransportProperti es.mpg
The bubble collides with the wall before it can reach it's final position.

On the way of alleviating the unwanted movement of the bubble I was trying to use a wedge case.
Unfortunately the case reaches a Courant number of 0.5 and the simulation is shutting down.
I have no idea why. The results bevorehand are looking good.
The case with results and logfile are here:
http://therealsega.th.funpic.de/open...onWedge.tar.gz

At the moment I'm simulating in real 3D.
This will take some time.
If anyone has some ideas regarding these problem I would appreciate an answer.
So far, so good.
Have a nice evening. S.

sega · September 26, 2008, 04:38

Hello.

Sorry to bother again. I have tried a different boundary condition for the pressure.
(I'm not talking about the wedge case)

Instead of using pd=0 on all the walls I have used zeroGradient for all walls and pdRefCell=pdRefValue=0.

The bubble is still beginning to move, but the simulation is shutting down with this error-message:

--> FOAM FATAL ERROR : Continuity error cannot be removed by adjusting the outflow.
Please check the velocity boundary conditions and/or run potentialFoam to initialise the outflow.
Total flux : 0.00694063
Specified mass inflow : 3.12571e-133
Specified mass outflow : 7.14486e-18
Adjustable mass outflow : 0

From function adjustPhi(surfaceScalarField& phi, const volVectorField& U,const volScalarField& p
in file cfdTools/general/adjustPhi/adjustPhi.C at line 111.

FOAM exiting

Can this lead anyone to a better understanding whats going wrong?
As all velocity boundaries are set to 0 there is technicaly no out- nor inflow.

sega · November 1, 2008, 10:22

Hey Niels.

Remember the case?
I'm still coming back to this simulation from time to time.

I'm referring to one of your messages above, this one:
In this I got the same problem as you, namely that the kinetic energy is very slow dissipating, the movie is 100s. Note that the smaller sphere is completely steady but the larger once are vibrating.

Now, my interest in this case is are the parasite currents. I have found a setup of transportProperties in which the bubble is not moving away or oscillating until eternity.

To get a feeling about the magnitude of the parasite currents I calculated magU for my case.

So, now I found that magU in my case is sill not constant over time after the liquid has settled down. Here is the timeplot of magU over time:

As I don't have a constant parasite current I'm thinking of using some kind of mean magnitude (which can be found in the image above).

Do you think this is applicable?

ngj · November 6, 2008, 13:29

Hi Sebastian

I am the elusive writer on the forum these days, thus sorry for the late response.

Well, I actually prefer the fluctuating curve, as it apparently contain a significant amount of information, which is definitely not present in the average. Then again, it depends on what you are hoping to achieve. Maybe you could simple calculate a number of the statistical moments and see if that gives you anything!?!

Best regards,

Niels

sega · November 6, 2008, 15:00

Hey Niels.

Thanks for your response.
I have discussing this curve in another thread:
http://www.cfd-online.com/cgi-bin/Op...how.cgi?1/9803

Maybe the average is not as bad - if you tread the simulation accordingly.
In this case: Make it continuous instead of interrupted ...

Jimbomet · November 25, 2011, 07:21

Hi Sebastian,
I know this thread is old...but I'm dealing with the same problem of introducing a square or a bubble in a quiescent fluid...(and zero gravity)
I'm not able to find a stable solution for the case of a square (oscillations continue in the shape of the bubble)...

Can I ask you which fluids have you used (in terms of viscosity and surface tension)? And also the order of magnitude of the dimensions of your bubble?
Because, initialising a bubble with funckySetFields I obtain instability in the velocities in all the domain when decreasing the dimension of the bubble to the order of mm...and i thought it could be the presence of high parasite current working with that dimensions but I'm not sure and I don't know how to solve that problem..

Thanks a lot in advance for your help
best regards
Paolo

May 20, 2008, 10:33	Hi Sebastian I do not know	#41
ngj Senior Member Niels Gjoel Jacobsen Join Date: Mar 2009 Location: Copenhagen, Denmark Posts: 1,900 Rep Power: 37	Hi Sebastian I do not know if I should congratulate you then On the other hand it is always interesting to hear how other programs are performing. I have been playing around with your setup, and these symmetry conditions seems to cause problems ... maybe that the spurious waves on the interfaces wants to go trough the symmetry plane but is restricted as it is not wave-transmissive?!!? Thus could you try to connect the two symmetry-planes in some way...? I have not tried the wedge type of block myself, but can you connect two faces in a wedge irrespectively of the orientation!?! That might help you to remove the large velocities at the boundary!?!?! As you can see I am only guessing wildly, and then you can do all the work Have fun, hope you find a solution! - Niels __________________ Please note that I do not use the Friend-feature, so do not be offended, if I do not accept a request.

May 20, 2008, 12:24	Hey. I think before I start	#42
sega Senior Member Sebastian Gatzka Join Date: Mar 2009 Location: Frankfurt, Germany Posts: 729 Rep Power: 20	Hey. I think before I start a wild gambling with the symmetry planes I rather restart the case without the benefits of symmetry. I will come back with the results ... bet on it! Greetings. __________________ Schrödingers wife: "What did you do to the cat? It's half dead!"

May 21, 2008, 04:29	Hi Sebastian I tried a diff	#43
ngj Senior Member Niels Gjoel Jacobsen Join Date: Mar 2009 Location: Copenhagen, Denmark Posts: 1,900 Rep Power: 37	Hi Sebastian I tried a different kind of setup yesterday, which looks something like this: So disregarding the small deviation from the correct volume fraction (order of magnitude 0.1% error), I initialized the 'bubble' with a square at rest. After running it, the result came within 2% of the theoretical value, but the circle induced some rather nasty volumes, which gave rise to larger pressures in the 'corners' of the circle. So a better approximation of the circle, as you have already discussed with others, would probably give better results. On the other hand, this small experiment shows that your problems is highly grid related, as the results are significantly better when a clear representation of the circle is thought into the grid. Some would call it cheating, intelligent, but still cheating Hope it helps, Niels P.S. Cannot give you my configuration as the computer, which I used, is not connected to the internet. __________________ Please note that I do not use the Friend-feature, so do not be offended, if I do not accept a request.

May 21, 2008, 08:19	Your recommended setup is not	#44
sega Senior Member Sebastian Gatzka Join Date: Mar 2009 Location: Frankfurt, Germany Posts: 729 Rep Power: 20	Your recommended setup is not working by now. I tried to use the upper right quarter of the computational domain. But as I have mentioned earlier this symmetry is not working. The solver is stuck from the first time step with timesteps of e-6 ... I will try it with the complete computational domain. Get back to you. __________________ Schrödingers wife: "What did you do to the cat? It's half dead!"

May 21, 2008, 09:33	So, now I tried the static ini	#45
sega Senior Member Sebastian Gatzka Join Date: Mar 2009 Location: Frankfurt, Germany Posts: 729 Rep Power: 20	So, now I tried the static initialized spherical bubble: Despite the fact that the bubble is moving (any ideas why? - wall effects?) the simulation caluclates a little smaller value of the pressure inside the bubble, as expected. I think that is due to the "distributed" (not "sharp") interface at the beginning. The liquid phase volume fraction is therefore 0.4 % bigger than expected. This will lead to a slighlty bigger radius and a theoretical pressure jump of about 11.77 N/m². So, what magnitude does the pressure have inside the bubble? How can I tell to which number the "orange" pressure is belonging? __________________ Schrödingers wife: "What did you do to the cat? It's half dead!"

May 22, 2008, 01:45	Ok, I have used the threshold-	#46
sega Senior Member Sebastian Gatzka Join Date: Mar 2009 Location: Frankfurt, Germany Posts: 729 Rep Power: 20	Ok, I have used the threshold-function to display the interior of the bubble and did a "Reset Range" on the pressure plot. Looks kind of strange because of the missing parts inside the bubble. Why is that? Now I guess, that the pressure inside the bubble is about 10.8 N/m². Thats still 8.3 % Error. __________________ Schrödingers wife: "What did you do to the cat? It's half dead!"

May 25, 2008, 03:56	So, now I did a calculation on	#47
sega Senior Member Sebastian Gatzka Join Date: Mar 2009 Location: Frankfurt, Germany Posts: 729 Rep Power: 20	So, now I did a calculation on a mesh which is spherical at the final position of the bubble - like Niels suggested. But there are still problems. The pressure is a little bit higher, but not constant inside the bubble. I think this is due to the mesh inside the bubble-region ... Have a look at these images: http://therealsega.th.funpic.de/openfoam/pd.png http://therealsega.th.funpic.de/openfoam/pd_wire.png And why is the pressure distorted at the lower left part of the interface? Greetings __________________ Schrödingers wife: "What did you do to the cat? It's half dead!"

May 25, 2008, 05:52	I tried to give the mesh some	#48
sega Senior Member Sebastian Gatzka Join Date: Mar 2009 Location: Frankfurt, Germany Posts: 729 Rep Power: 20	I tried to give the mesh some arcs in the interiour of the bubble, but the solver is stuck. Maybe the control volumes are too much contortet. http://therealsega.th.funpic.de/openfoam/mesharc.png Niels. Maybe you should look up how you have done your spherical mesh. __________________ Schrödingers wife: "What did you do to the cat? It's half dead!"

May 26, 2008, 03:31	Hi Sebastian 1. I have had	#49
ngj Senior Member Niels Gjoel Jacobsen Join Date: Mar 2009 Location: Copenhagen, Denmark Posts: 1,900 Rep Power: 37	Hi Sebastian 1. I have had the same problem with moving bubbles ... for some reason the velocity field become asymmetric and then the friction moves the bubble to one of the side. 2. I have formatted my computer and my run went the same way. But I have done the mesh in exactly the way as sketched above in my post from the 21. of May. 3. I have also experienced those weird oscillations on the surface when the resolution becomes finer than some certain threshold. Best regards, Niels __________________ Please note that I do not use the Friend-feature, so do not be offended, if I do not accept a request.

May 26, 2008, 08:36	A guy from the university coul	#50
sega Senior Member Sebastian Gatzka Join Date: Mar 2009 Location: Frankfurt, Germany Posts: 729 Rep Power: 20	A guy from the university could decrease the error in his pressure-jump to about 4% ... He was using Fluent with a TRIANGULAR mesh ... __________________ Schrödingers wife: "What did you do to the cat? It's half dead!"

September 23, 2008, 11:02	Hey! It's me again. I'm sti	#51
sega Senior Member Sebastian Gatzka Join Date: Mar 2009 Location: Frankfurt, Germany Posts: 729 Rep Power: 20	Hey! It's me again. I'm still thinking about this case. As I have recognised there is a difference between the value of volPointInterpolate(pd) and cell(pd). As the pressure-jump over the interface looks pretty nice (and constant) in volPointInterpolate(pd) it IS NOT when you are looking at the cell(pd)-values itself. I used some threshold-filters in ParaView to get a feeling about the difference between the values. It looks like there are some few cells at the inside of the interface who have a pressure like it should be inside the drop. But the number of cells with less than the desired pressure is increasing near the center of the drop. So, the volPointInterpolate(pd) value suggestes a difference of about 6% to the desired pressure. While the cell(pd) values range from 2-13% with more than half of the cells in the range of 10-13% error. So, how trustworthy is the volPointInterpolate(pd)? __________________ Schrödingers wife: "What did you do to the cat? It's half dead!"

September 23, 2008, 11:08	Seems you missed the pictures.	#52
maximus New Member Maximus Join Date: Mar 2009 Location: Germany Posts: 3 Rep Power: 17	Seems you missed the pictures. Curious about what you just said. Will wait until you load them up :-)... Grüß Maximus

September 23, 2008, 11:09	oops I did it again .... so	#53
maximus New Member Maximus Join Date: Mar 2009 Location: Germany Posts: 3 Rep Power: 17	oops I did it again .... sorry

September 23, 2008, 11:42	I can see the pictures. The a	#54
sega Senior Member Sebastian Gatzka Join Date: Mar 2009 Location: Frankfurt, Germany Posts: 729 Rep Power: 20	I can see the pictures. The above one is volPointInterpolate(pd). ?? __________________ Schrödingers wife: "What did you do to the cat? It's half dead!"

September 25, 2008, 13:10	Lets get back at this. Obvious	#55
sega Senior Member Sebastian Gatzka Join Date: Mar 2009 Location: Frankfurt, Germany Posts: 729 Rep Power: 20	Lets get back at this. Obviously the problem is really challenging. Meanwhile I did a lot of different stuff with the case. Lets have a look at some points I couldn't fix myself. First of all, I have used some wrong transportProperties all the time. Obviously I have mixed dynamic and kinematic viscosity (in some weird way). Now, the right values are in place. As a matter of fact these setting should not influence the solution. Wishful thinking! With kinematic viscosity round about 10^(-6) the simulation times are rather high. Like Niels I'm now experiencing a movement of the bubble with the long simulation times. Have a look: http://therealsega.th.funpic.de/openfoam/surfaceTensionOriginalTransportProperti es.mpg The bubble collides with the wall before it can reach it's final position. On the way of alleviating the unwanted movement of the bubble I was trying to use a wedge case. Unfortunately the case reaches a Courant number of 0.5 and the simulation is shutting down. I have no idea why. The results bevorehand are looking good. The case with results and logfile are here: http://therealsega.th.funpic.de/open...onWedge.tar.gz At the moment I'm simulating in real 3D. This will take some time. If anyone has some ideas regarding these problem I would appreciate an answer. So far, so good. Have a nice evening. S. __________________ Schrödingers wife: "What did you do to the cat? It's half dead!"

Thread Tools	Search this Thread
Show Printable Version Email this Page	Search this Thread: Advanced Search
Display Modes
Linear Mode Switch to Hybrid Mode Switch to Threaded Mode

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
GRAVITY	Lorenzo	FLUENT	0	September 19, 2007 10:10
gravity-table.scm scheme file for trans gravity	Shankar	FLUENT	0	May 31, 2006 04:30
Turbulent boundary conditions for bubble column	anjai	FLUENT	12	October 17, 2005 06:34
gravity driven flow- boundary conditions?	Shankar	FLUENT	4	November 26, 2003 15:45
Boundary Conditions for Bubble Column	cfd-novice	FLUENT	0	April 14, 2003 01:50

September 26, 2008, 04:38	Hello. Sorry to bother agai	#56
sega Senior Member Sebastian Gatzka Join Date: Mar 2009 Location: Frankfurt, Germany Posts: 729 Rep Power: 20	Hello. Sorry to bother again. I have tried a different boundary condition for the pressure. (I'm not talking about the wedge case) Instead of using pd=0 on all the walls I have used zeroGradient for all walls and pdRefCell=pdRefValue=0. The bubble is still beginning to move, but the simulation is shutting down with this error-message: --> FOAM FATAL ERROR : Continuity error cannot be removed by adjusting the outflow. Please check the velocity boundary conditions and/or run potentialFoam to initialise the outflow. Total flux : 0.00694063 Specified mass inflow : 3.12571e-133 Specified mass outflow : 7.14486e-18 Adjustable mass outflow : 0 From function adjustPhi(surfaceScalarField& phi, const volVectorField& U,const volScalarField& p in file cfdTools/general/adjustPhi/adjustPhi.C at line 111. FOAM exiting Can this lead anyone to a better understanding whats going wrong? As all velocity boundaries are set to 0 there is technicaly no out- nor inflow. __________________ Schrödingers wife: "What did you do to the cat? It's half dead!"

November 1, 2008, 10:22	Hey Niels. Remember the cas	#57
sega Senior Member Sebastian Gatzka Join Date: Mar 2009 Location: Frankfurt, Germany Posts: 729 Rep Power: 20	Hey Niels. Remember the case? I'm still coming back to this simulation from time to time. I'm referring to one of your messages above, this one: In this I got the same problem as you, namely that the kinetic energy is very slow dissipating, the movie is 100s. Note that the smaller sphere is completely steady but the larger once are vibrating. Now, my interest in this case is are the parasite currents. I have found a setup of transportProperties in which the bubble is not moving away or oscillating until eternity. To get a feeling about the magnitude of the parasite currents I calculated magU for my case. So, now I found that magU in my case is sill not constant over time after the liquid has settled down. Here is the timeplot of magU over time: As I don't have a constant parasite current I'm thinking of using some kind of mean magnitude (which can be found in the image above). Do you think this is applicable? __________________ Schrödingers wife: "What did you do to the cat? It's half dead!"

November 6, 2008, 13:29	Hi Sebastian I am the elusi	#58
ngj Senior Member Niels Gjoel Jacobsen Join Date: Mar 2009 Location: Copenhagen, Denmark Posts: 1,900 Rep Power: 37	Hi Sebastian I am the elusive writer on the forum these days, thus sorry for the late response. Well, I actually prefer the fluctuating curve, as it apparently contain a significant amount of information, which is definitely not present in the average. Then again, it depends on what you are hoping to achieve. Maybe you could simple calculate a number of the statistical moments and see if that gives you anything!?! Best regards, Niels __________________ Please note that I do not use the Friend-feature, so do not be offended, if I do not accept a request.

November 6, 2008, 15:00	Hey Niels. Thanks for your	#59
sega Senior Member Sebastian Gatzka Join Date: Mar 2009 Location: Frankfurt, Germany Posts: 729 Rep Power: 20	Hey Niels. Thanks for your response. I have discussing this curve in another thread: http://www.cfd-online.com/cgi-bin/Op...how.cgi?1/9803 Maybe the average is not as bad - if you tread the simulation accordingly. In this case: Make it continuous instead of interrupted ... __________________ Schrödingers wife: "What did you do to the cat? It's half dead!"

November 25, 2011, 07:21	problem with stable bubble	#60
Jimbomet New Member Paolo Join Date: Nov 2011 Posts: 7 Rep Power: 14	Hi Sebastian, I know this thread is old...but I'm dealing with the same problem of introducing a square or a bubble in a quiescent fluid...(and zero gravity) I'm not able to find a stable solution for the case of a square (oscillations continue in the shape of the bubble)... Can I ask you which fluids have you used (in terms of viscosity and surface tension)? And also the order of magnitude of the dimensions of your bubble? Because, initialising a bubble with funckySetFields I obtain instability in the velocities in all the domain when decreasing the dimension of the bubble to the order of mm...and i thought it could be the presence of high parasite current working with that dimensions but I'm not sure and I don't know how to solve that problem.. Thanks a lot in advance for your help best regards Paolo