[gustavoliveira]

Hello,


I am trying to simulate a stepped channel using the multiphase/interFoam/RAS/weirOverflow example as base. For the stepped channel, I created a .stl file and start to tests some refinaments around the channel with the snappyHexMesh tool (and after I used the extrudeMesh tool to run in 2D).



If I use refinament levels (2 3), in the instant 2 seconds, the velocity magnitude is:





If I use refinament levels (2 4), in the instant 2 seconds, the velocity magnitude is:




The behavior in the two cases is very diferent. In the level (2 3) there are a air movement above the channel that I cant explain, while in the level (2 4) the water behavior is not what I expect.



Someone knows why?

[simrego]

Hi
On the left side as I can see you have giant cells. Try to resolve the interface better there. And also from (2, 3) -> (2, 4), is doubling the cells where the feature angle is high enough. I guess for you it is everywhere near the steps. So you halved the mesh size there. Of course you will get different results since in your 1st case it is clear from your figure that the resolution is not the best...
And what is your problem with the flow on the 2nd case? It is jumping over the steps. I think it can be possible at the begining of the simulation and probably the water will "fill" some steps as you go forward in time.


EDIT: also it seems like you are "loosing" water. maybe try to use less diffusive schemes if you can.

[Tobermory]

Agreed - you need to pay attention to the CFD basics. Improve your mesh, and make sure that you have enough refinement where you need it. Reduce your time step / Courant number. Pay attention to your numerical schemes. All of these will help reduce numerical diffusion, and improve your simulation.

Doing the above will probably increase your run time by orders of magnitude, but that's the price you have to pay for accuracy. Good luck my friend.