Discussion Closed This discussion was created more than 6 months ago and has been closed. To start a new discussion with a link back to this one, click here.

Problem: Oil Droplets Seem To Dissolve In Water (Two-Phase Flow, Level Set)

Kristian Tølbøl Sørensen

Please login with a confirmed email address before reporting spam

I have a problem with my "Two-Phase Flow, Level Set" simulation in COMSOL 4.2. I am trying to simulate the formation of oil droplets in water by hydrodynamic flow focusing. The water and oil are defined as two materials of suitable viscosity and density. Walls are wetted with contact angle pi/2, surface tension 0.07. The outlet is "pressure, no viscous stress" at 0 Pa. The mesh is simply physics controlled at the "finer" setting.

The attached .gif shows nice droplet formation generated by model. It looks similar to what is seen experimentally, except the oil droplets seem to dissolve in the water phase! I have not been able to figure out which parameter must be missing to make the two phases immiscible (I actually thought Two-Phase Flow was immiscible per standard).

Can anyone give a quick suggestion as to how I can keep the oil droplets from dissolving in the water phase?

Thanks in advance!

Best regards
Kristian Sørensen
Denmark


17 Replies Last Post 12.06.2017, 07:44 GMT-4
Lechoslaw Krolikowski

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 16.12.2011, 10:07 GMT-5
Hi,

The Two-Phase Flow Level Set solver has two forms: conservative and non-conservative (default). In non-conservative formulation there is a mass loss during simulation caused by numerical errors.
See: CFD Module. User's Guide and Rising Bubble in the Model Library.

BR
Andrzej
Hi, The Two-Phase Flow Level Set solver has two forms: conservative and non-conservative (default). In non-conservative formulation there is a mass loss during simulation caused by numerical errors. See: CFD Module. User's Guide and Rising Bubble in the Model Library. BR Andrzej

Kristian Tølbøl Sørensen

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 21.12.2011, 06:08 GMT-5
I have adjusted my simulation to use the conservative form, this resolves the problem. To anyone else with the same problem: Go to Laminar Two-Phase Flow, expand the Advanced Settings node, choose Conservative form.

Thank you very much Andrzej!

Best regards
Kristian
Denmark
I have adjusted my simulation to use the conservative form, this resolves the problem. To anyone else with the same problem: Go to Laminar Two-Phase Flow, expand the Advanced Settings node, choose Conservative form. Thank you very much Andrzej! Best regards Kristian Denmark

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 17.08.2012, 09:51 GMT-4
I have the same problem but when i switch to conservative form I have convergence problem
I have the same problem but when i switch to conservative form I have convergence problem

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 20.10.2012, 23:32 GMT-4
Just reduce your time step in the time dependent solver and it will work.

However, I ran in conservative form, the droplet still shrinks. I have also played with the scales for dependent variables but not useful.

Did you do anything else besides turn the solver into conservative? please let me know.

Thanks.

Shawn
Just reduce your time step in the time dependent solver and it will work. However, I ran in conservative form, the droplet still shrinks. I have also played with the scales for dependent variables but not useful. Did you do anything else besides turn the solver into conservative? please let me know. Thanks. Shawn

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 23.10.2012, 00:01 GMT-4
I just figure out what exactly to avoid this problem, I am sharing it here for others with the same problem.

Others are correct that the convective terms must be solved in conservative form. However, if the time is large, or if both phases density are not in a close range (e.g. air and water), turning equations in conservative is not enough.Therefore, I defined a pressure point constraint on the droplet. I made these and that worked perfectly.


Other issue is of course the convergence. To fix that one, instead of Level set, simply use Phase-field.

Good luck.

Shawn
I just figure out what exactly to avoid this problem, I am sharing it here for others with the same problem. Others are correct that the convective terms must be solved in conservative form. However, if the time is large, or if both phases density are not in a close range (e.g. air and water), turning equations in conservative is not enough.Therefore, I defined a pressure point constraint on the droplet. I made these and that worked perfectly. Other issue is of course the convergence. To fix that one, instead of Level set, simply use Phase-field. Good luck. Shawn

Ivar KJELBERG COMSOL Multiphysics(r) fan, retired, former "Senior Expert" at CSEM SA (CH)

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 23.10.2012, 01:27 GMT-4
Hi

remains for me is it physical to place a pressure point on the droplet ?
I would rather have it on some "fixed" position, as in presence of gravity the pressure is increasing with the fall towards -y (generally)

One thing is to make a nice simulation with beautiful colours, the other is to model correctly the true physics (or as close as possible.

Remains, that in the phase method, in contrary to the level set, the droplet dissociates, hence we need to add something

--
Good luck
Ivar
Hi remains for me is it physical to place a pressure point on the droplet ? I would rather have it on some "fixed" position, as in presence of gravity the pressure is increasing with the fall towards -y (generally) One thing is to make a nice simulation with beautiful colours, the other is to model correctly the true physics (or as close as possible. Remains, that in the phase method, in contrary to the level set, the droplet dissociates, hence we need to add something -- Good luck Ivar

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 02.11.2012, 17:03 GMT-4
Thanks for the comments Ivar, I agree with you that the model must be physical. However, could tell me how pressure changes over the surface of droplet which is located in external flow? I believe the pressure is remain constant on the surface of the droplet. May be I did not explain my model completely. Anyway, the model still works perfect for 2D cases.

The droplet still shrinks as I reduce its size to less than 100 microns for 2D-axisymmetric models. I think COMSOL needs to think seriously and do something to resolve this issue.

--
Shawn
Thanks for the comments Ivar, I agree with you that the model must be physical. However, could tell me how pressure changes over the surface of droplet which is located in external flow? I believe the pressure is remain constant on the surface of the droplet. May be I did not explain my model completely. Anyway, the model still works perfect for 2D cases. The droplet still shrinks as I reduce its size to less than 100 microns for 2D-axisymmetric models. I think COMSOL needs to think seriously and do something to resolve this issue. -- Shawn

Ivar KJELBERG COMSOL Multiphysics(r) fan, retired, former "Senior Expert" at CSEM SA (CH)

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 03.11.2012, 07:38 GMT-4
Hi

I do believe COMSOL's developpers have though seriously about it ;)
but it's rather our use of level set and phase set that is not correct: my first test the other day also showed that level set fallong droplet worked out nicely, while my first attempt with the phase set had the droplet to dissolve before it had gone far ;)
So I need to go back to the theory and library examples to take a new "learning session", I'm not often in these physics, as I'm mixing many physics, I tend to forget the specialities, of one and the other, if I do ont use them over a month or two

Now I do not agree that the pressure is constant all around, in all generality, as the fluid velocity changes around a falling droplet, hence the local forces too and the normal force = pressure wil then change all around the droplet circumference

--
Good luck
Ivar
Hi I do believe COMSOL's developpers have though seriously about it ;) but it's rather our use of level set and phase set that is not correct: my first test the other day also showed that level set fallong droplet worked out nicely, while my first attempt with the phase set had the droplet to dissolve before it had gone far ;) So I need to go back to the theory and library examples to take a new "learning session", I'm not often in these physics, as I'm mixing many physics, I tend to forget the specialities, of one and the other, if I do ont use them over a month or two Now I do not agree that the pressure is constant all around, in all generality, as the fluid velocity changes around a falling droplet, hence the local forces too and the normal force = pressure wil then change all around the droplet circumference -- Good luck Ivar

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 20.11.2012, 18:19 GMT-5

Ivar,

I agree with your statement regarding the pressure-change around moving objects in "general".

But as I mentioned, I believe that it is reasonable to consider it constant around micron size droplet when it has practically no speed. Again, this is physics dependent, and with my case it is totally justified!

Good luck

Shawn
Ivar, I agree with your statement regarding the pressure-change around moving objects in "general". But as I mentioned, I believe that it is reasonable to consider it constant around micron size droplet when it has practically no speed. Again, this is physics dependent, and with my case it is totally justified! Good luck Shawn

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 22.01.2013, 02:52 GMT-5
Hi,
I am using the conservative form, however the time required to solve the simulation increased very much. is there any way to decrease the solution time?
Thanks
Hi, I am using the conservative form, however the time required to solve the simulation increased very much. is there any way to decrease the solution time? Thanks

Please login with a confirmed email address before reporting spam

Posted: 9 years ago 15.09.2015, 03:42 GMT-4
I didn't find the advanced node to choose Conservative form. Can anybody help me.
Regards,
A K Jana
I didn't find the advanced node to choose Conservative form. Can anybody help me. Regards, A K Jana

Please login with a confirmed email address before reporting spam

Posted: 9 years ago 15.09.2015, 13:07 GMT-4
click "laminar Two-phase flow, level set " at settings,you will find "advanced settings", then you will find it
click "laminar Two-phase flow, level set " at settings,you will find "advanced settings", then you will find it

Please login with a confirmed email address before reporting spam

Posted: 9 years ago 15.09.2015, 13:14 GMT-4
Hi, Ivar:
My model is similar to what we are discussing here, but I have problem that when I use phase field method, it seems the drop(it is an organic) dissolve in the water. I cannot figure out what's wrong. Actually, I also meet another problem. In my model, I use three fluids: air, water and organic drop. The organic drop first stay in the air, and then it fall downward in to the water. I tried both level set and phase field. It seems there have some problem with the initial interface. I am not sure if comsol can solve this kind of simulation. Do you have any ideas? Or others who know this, can you help me? Thank you so much.

Best,
Esther
Hi, Ivar: My model is similar to what we are discussing here, but I have problem that when I use phase field method, it seems the drop(it is an organic) dissolve in the water. I cannot figure out what's wrong. Actually, I also meet another problem. In my model, I use three fluids: air, water and organic drop. The organic drop first stay in the air, and then it fall downward in to the water. I tried both level set and phase field. It seems there have some problem with the initial interface. I am not sure if comsol can solve this kind of simulation. Do you have any ideas? Or others who know this, can you help me? Thank you so much. Best, Esther

Please login with a confirmed email address before reporting spam

Posted: 9 years ago 03.12.2015, 13:09 GMT-5
Dear Shawn.F

Many thanks for your clarifying. Indeed, I am trying to simulate a droplet on a flat surface and I followed all instructions bout still I faced a problem with the convergence and lot of errors. can you please guide me how to solve my problem.

Thanks.
Dear Shawn.F Many thanks for your clarifying. Indeed, I am trying to simulate a droplet on a flat surface and I followed all instructions bout still I faced a problem with the convergence and lot of errors. can you please guide me how to solve my problem. Thanks.

Please login with a confirmed email address before reporting spam

Posted: 9 years ago 05.01.2016, 08:52 GMT-5
Hi Esther:

Did you find out what was wrong in your model? I am currently stuck with the exact same problem, I am investigating droplet formation in T-shape devices, and within 1 mm, oil droplets seem to dissolve in water. I tried both phase-field and level set methods and while phase-field did not want to converge from the very beginning (the initialization of the interface seem to be problematic), the level set method first starts but there are some convergence issues once the dispersed phase enters the junction.

Could you give me an insight on how you manage to solve these issues in the past four months?
Best,

Axel
Hi Esther: Did you find out what was wrong in your model? I am currently stuck with the exact same problem, I am investigating droplet formation in T-shape devices, and within 1 mm, oil droplets seem to dissolve in water. I tried both phase-field and level set methods and while phase-field did not want to converge from the very beginning (the initialization of the interface seem to be problematic), the level set method first starts but there are some convergence issues once the dispersed phase enters the junction. Could you give me an insight on how you manage to solve these issues in the past four months? Best, Axel

Please login with a confirmed email address before reporting spam

Posted: 7 years ago 12.06.2017, 05:06 GMT-4
hi
i have same problem
did you find a solution for that?
thanks in advanced
hi i have same problem did you find a solution for that? thanks in advanced

Please login with a confirmed email address before reporting spam

Posted: 7 years ago 12.06.2017, 07:44 GMT-4
Hi ivar
I have a three phase flow problem. These three materials are immiscible (water, oil, air) but two of them getting mixed. I red another discussion: ''Problem: Oil Droplets Seem To Dissolve In Water (Two-Phase Flow, Level Set)''
In there, they overcome that problem with changing of non-conservative form to conservative form in advanced physic setting of two phase flow.
but in three phase flow, this option doesn't exist.
and I think the default equation is non-conservative.
is there any way to change it to conservative form?

best regards
Hi ivar I have a three phase flow problem. These three materials are immiscible (water, oil, air) but two of them getting mixed. I red another discussion: ''Problem: Oil Droplets Seem To Dissolve In Water (Two-Phase Flow, Level Set)'' In there, they overcome that problem with changing of non-conservative form to conservative form in advanced physic setting of two phase flow. but in three phase flow, this option doesn't exist. and I think the default equation is non-conservative. is there any way to change it to conservative form? best regards

Note that while COMSOL employees may participate in the discussion forum, COMSOL® software users who are on-subscription should submit their questions via the Support Center for a more comprehensive response from the Technical Support team.