Learning Forum

[zakmt293]

Greetings

I am working on a simulation of single bubble growth in a super heated liquid domain.For phase change purpuses I am utilizing the lees model. I am changing the evaporation frequency to match with experimental time of bubble growth and mass evaporation. But interface deformation occurs with increase of frequency. Can any expert give me an idea that might help in modelling of my problem.

Thank You   

[Rob]

How well is the model converging, and how good is the mesh? Images posted into the comments will help. 

[zakmt293]

Greetings

I am using VOF model for performing the simulation. A UDF is used for the purpose of adding mass source of liquid only at  interface cells and is based upon the Lee Model. Only fluid domain is modeled with water as the working fluid. The domain is heated from the bottom to create a thermal boundary layer and then a vapor germs is placed in domain after some time. From the evaporation of liquid the bubble grow. The mass source is stores in UDM at interface cells and is then added through source Macros. As I increase the frequency of evaporation to match with experimental results deformation in interface occurs I would like to share the pics as well. I have changed the value of volumetric smoothing as well. I can share the UDF as well. If any other data is required I am willing to share it. I would be thankful for your guidance.

 

The contours of vapor volume fraction with respect to time are given as 

                                                        VOF of VAp at 0 sec

 

                                                           VOF of vap 1.7e-5sec

 

                                                         VOF of vap 4.7e-5 sec

 

                                                                Vof of vap at 2.61e-4 sec

                                                             From here the deformation starts

 

                                                  VOF of vap at 3.61e-4

 

 

                                                   VOF of vap at 6.61e-4 sec

 

                                                          VOF of vap at 1.061ms

 

 

 

[zakmt293]

And taking into account the Mesh it is made with point wise and is a 2d structural grid of cell more than 400000. The bottom is temp surface and the top is pressure outlet. The bottom is kept at temp of 395.1 while top is saturated temp. The left and right

side wall are no flux conditions.

The length is 10mm divided into separate connectors on both side of reference coordinate system and height is 5mm while the each element has dimension of 1.1 micro meter. With various recommendation on various forums  I refined it but not good results.

[Rob]

Replot the contours with node values off: this makes it easier to see if the result is "chequer-boarding" with isolated bits of liquid.  Convergence looks a little lumpy later on too: check the UDF maths and also that the model is well converged. 

[Rob]

Have a look at the interface with node values off & look at the UDF logic.  If I have 2-3 cells on the interface due to diffusion and run the UDF how many cells will be flagged and therefore altered? 

[zakmt293]

Greetings

Thank you for your answer Yes definitely. After the above comment two things came I believe could improve my results. 

1: One of them is to refine  the mesh more.

2: I should limit the cell and adopt the conditions like 

e.g if (C_VOF(c,tp)>0.9 && C_VOF (c,tp)<1)

And then check the results

 

Kindly if I am wrong I would be thankful if you can give me value able guidance.

 

[DrAmine]

What might help here is using the Anti-Diffusion feature, incrase resoltuion and reduce time step. You can use dynamic mesh adaption with some expressions: like whenver I am in the liquid core please adatp or based on the curvature.

 

I have a feeling that all effects seen there are due to spurious currents. Moreover do not use 2D planer here. Either 2D axisymmetric or better 3D solver should be used!

[zakmt293]

Greetings

Thank you for your comment Sir Amine. I would follow the above recommendations mentioned in 1st paragraph. 

But the dynamics mesh do I have to write a udf for it?As I don't have any knowledge of dynamics mesh.

The second question is in the 2nd paragraph. I would be comparing my results to an experiment in which bubble growth is modeled in a glass cell of inner cross sectional area of 10*10 square mm. Which means we cannot modeled it in terms of r and z for in case of axis symmetric could which could have been useful if circular geometry was present. I have gone through symmetric boundary condition but I am still confused in it whether it can be of any help.

In case of 3d geometry the computation resources would be much high and moreover wouldn't it disturb the interface reconstruction algorithms?

Thank You

Zeeshan Ahmad Khan

[Rob]

Amine means "dynamic mesh adaption". This means you automatically refine the mesh based on some criterion: it's not dynamic mesh which is the moving variety. 

 

[zakmt293]

Sorry sir 

Can you guide me  in other questions mentioned on just above your comments.

Thank You

Zeeshan Ahmad Khan

[DrAmine]

Better to use 3d or 3d with symmetrical boundaries for a 1/4 of the whole geometry