Grid independence is that results does not change with change in size of the grid although value s from different grids may lie close (within certain  %). usual way of going 2x, x, x/2 or any suitable proportions of each other

Interesting comment but u need to be more specific what are u actually looking at in two phase flow e.g. you said u r are validating dozen of data... there can be variation in flow itself may be L-L, L-T or T-T, which will further effect type of flow, VF, PD, etc.

VOF is more suitable for flow interface visualization while TFM for hydrodynamic characteristics etc.

u need to  give more details.

I agree with Shazia that VOF is suitable for flow simulation with interface of fluids, such as stratified, wavy and slug flow.

it looks to me that your question is holding two questions. one on the grid and one on the validation.

Two of my students have simulated slug water/air and slug oil/gas flows using the VOF. They have used   'Directed Mesh technique' in STAR-CCM+. We found that a 'structured hexahedral grid' is more suitable when solving a case of two phase flow in pipe. you may refer to our papers in research gate. This may be not the same as your case of nozzle and you have to consider the effect of changing the grid dimension in the longitudinal, i.e. axial, direction, where you have change in the cross sectional area. 

On the validation of your fluid flow simulation, It is preferable you validate against experimental results. If they are not available,  the well established correlations are also a tool for validation.

Thank Shazaia you for your response. 

What I understand from your answer is that beside the mesh, I should be very careful about choosing the proper simulation scheme. 

I would like to give more detail about the case I am simulating. 

It is a air-water Nozzle where air from one side and water from the other side are mixed in the mixing chamber. the main problem is that although I am able to achieve good agreements in cases where the throat diameter is constant, however, as soon as the throat diameter is expanded the previous mesh does not perform as good as before and looses quality in terms of predicting Air/Liquid Ratio. I should note that from a certain number of mesh and refinements the flow structures remains constant.

So my conclusion based on your comment would be the simulation scheme. do you agree? And I have to say I might be pro at meshing but I am very new to two phase modeling so I would be pleased if you clarify the flow regimes you mentioned. 

Dear Hussain,

I found your paper Titled as : Numerical Study on Fluid Structure Interaction of Slug Flow in a Horizontal Pipeline.  I am pretty sure it will come in handy and Thank you for that. 

I found a very interesting point in you comment about changing the mesh in various directions when the cross sectional area is changing. This is very true,

In fact whenever I simulate each case separately I can get quite good agreements. The problem occurs when I need to find a single mesh for all cases so that I can use the mesh for predicting cases where no experimental data is available. I have searched for mesh correlations but till now I failed to find any for my case. 

The best I could do, was to construct a mesh where it is able to qualitatively follow the experiment result on ALR (air/liquid ratio) but the quantitative comparison was not accurate. 

Thank you Hussain for agreeing and Foad for answering.

I have done two phase flows simulations with CFD Fluent and thus agree with Hussain that  'structured hexahedral grid' is more appropriate with grid intensity increasing towards throat.  

I would suggest that u keep the mesh intensity high after the throat (though more computational effort & increase converging difficulty) because flow realignment is expected after throat (in case of expansion) to some D lengths. Simulation of this is vital as how the two phase are behaving in this area and lot of your simulation results depends on this area.

As for details u still did not mentioned what are your cases or test matrix is i.e. what parameters are your varying in each of the simulated cases?? velocities?? ALR?? x or y in same geometry.

Thank you for the secondment, Shazia.

Again, I agree with Shazia that the simulation domain is to be extended out of the nozzle exit. You may study the back pressure effect on the flow in the nozzle domain.

Foad: By the way, what is the two phase type of the water/air flow in the nozzle? 

Prof. Hussain