[harry123]

I'm doing a water jet simulation using the VOF method. When I reduce the time step size of the simulation or make the grid finer, the results obtained becomes worse. I would have guessed by making grid finer, I should get better results. Am I wrong ?

I use RANS turbulence model, Eulerian Multiphase, VOF, implicit unsteady.

The simulation seem to be quite sensitive to change in time step size also, is it normal for VOF simulations or would I be doing something wrong ?

[Jonny6001]

Interesting question

[harry123]

Asking again since no reply was found !!

I have attached the plot of variation of water volume phase along axis and experimental data from journal is plotted alongside.

Can someone give a clue please ?

[marmot]

There is a link between mesh size, time step and number of inner iterations. 1st you need to make sure you have somewhat of a good convergence before going to the next time step. Next you need to make sure your VOF surface is not moving more than 1 or 2 cells in a time step, than you get smearing and have to use this sharpening factor. I found it better to have smaller time step, less inner iteration, ~5, rather than large time step and 20 inner iterations. So you need to look at how the VOF surface is moving for the mesh density you have.

VOF is fun but it does take time.

You should also read about CFL_l and CFL_u, if you make them larger in theory your vof surface could travel more than 1 or 2 cells but than sharpening factor should be turned on.

[JIM SU]

Dear OF user's,


Please anyone tell me if at OP 2.1 or 2.3 which the better schemes of discretization, I have to reconstruct the interface when are solving the gamma equation for two-phase flow. I would like to know if OP have a similar Geo-Reconstruct or CICSAM scheme discretization implemented on Fluent or maybe have a better scheme. I have seen in the literature that the interfaceCompression scheme includes more smear on interface , Is it true?

My kind regards

JIM SU