How does this model perform grid encryption?

  • Last Post 15 June 2018
  • Topic Is Solved
h395523899 posted this 01 June 2018

This model is mainly used in fluent.

Simulated liquid-liquid two-phase displacement flow.

Size 10m*20m*10mm thin plate channel. Each side of 10m*20m has a hemispherical cavity.

Hemispherical cavity diameter 3m

One liquid enters from the 10m*10mm face and exits the other.



1. Does the entrance and exit need to be refined? (No refinement at 20 meters)

2. 10 meters, 10 mm edges need to be refined. Should y + be considered? What is the distance from the first node to the 20m edge of the 10m edge?

3. Is mesh refinement necessary at the contact between the thin plate and the hemispherical cavity?

4. If you need refinement in 3, there will be four tangents at the round mouth of the hemispherical cavity. These four tangential expansions will also bring refined gridlines. Will this lead to errors?

I uploaded my model, thank you。

Attached Files

Order By: Standard | Newest | Votes
h395523899 posted this 02 June 2018

Can anyone help me see it

kkanade posted this 08 June 2018

Hello h395523899


Can you please elaborate on what do you mean by encryption at inlet & outlet or at edges?



h395523899 posted this 09 June 2018

1. I changed a grid method, such as graph. It should be better than before.

2. I want to know the extent of mesh refinement.

About grid thinning. The reason is that the simulation process appears at the boundary as shown in the diagram.

The simulated water replaces oil. But at the boundary, water is faster than oil. This is not in conformity with the facts.

I think it is related to the thinning of the wall mesh, so I want to refine the wall mesh. But I don't know about the extent to which it is refined. Maybe the thickness of 10mm is not suitable for y+ calculation..

Volume fraction cloud image, blue is water

3. mesh refinement is carried out in the new grid. But when the boundary is enlarged, it still appears as picture.

Water speed cloud

4. I mainly simulate the change of the volume fraction of the oil in the hemisphere. Is there any need to refine the part of the thin plate and hemispherical contact (including the inner hemisphere)?

peteroznewman posted this 09 June 2018

I think Encryption is a poor translation of what should have been Inflation.

kkanade posted this 11 June 2018

Hello h395523899

Which models you are using? Can you pleas let me know your solver settings. 




h395523899 posted this 12 June 2018

Do not consider gravity

Euler-Eurale two-phase flow

Standard k-e model

Material for the material library to give water and oil

Lf-inlet: The velocity of the water at the entrance is 5m/s, the speed of the oil is 0, and the volume factor is 0.

Lf-outlet is outflow.

Rd-inlet1 and 2 are interior, others are wall,

The initial state setting patch makes the oil volume coefficient in the model 1

Step size 0.1

kkanade posted this 13 June 2018


Please use k-w sst. 

Use pressure outlet

Use smaller time step. Usually deta t = ((Vcell_min) 1/3) / U

Vcell_min can be obtained by grid check.

You can also think of using adaptive time step



h395523899 posted this 13 June 2018

If the simulated water enters a container filled with diesel fuel, observe the change in the diesel volume fraction in the container. For example, the diesel volume fraction becomes zero over time.

Which of the three euler-euler, euler-euler+multi fluid vof, and vof models should be used?

h395523899 posted this 13 June 2018

If Volume statistics:

    Minimum volume (m3): 1.000000e-12

  ((Vcell_min) 1/3) / U

Does this mean ((1.000000e-12)1/3)/speed?

So my step will reach 3.333e-14, which is too small.

Kremella posted this 15 June 2018

Hello h395523899,


Here is a post by Keyur which clarifies which multi-phase model to use.

Applicability of euler-euler and euler-euler+multi fluid vof and vof models

Hope this helps answer you narrow down on the approach.

Best Regards,


h395523899 posted this 15 June 2018

When I use the vof model or the Multi fluid vof model it does not converge.

Using k-w sst. also does not converge.

Divergence does not occur only when using the Euler-Euler RNG model or the standard model.

Kremella posted this 15 June 2018


Firstly, you should select these models based on the physics you are trying to solve and not because you are able to get a converged solution. 

Secondly, your time step calculation for VOF should be: ((1.000000e-12)^(1/3))/speed

Perhaps your divergence issue is because you are not using the appropriate time-step size? 

Let us know what you get.

Best Regards,