Dear Gentleman,

The process is liquid emptying out from the container, then a ball (big, not a particle) will move with the liquid to the outlet.

I want to model this process. It seems the first big-stone in front of me is that the grid problem. How can I design the moving grid of the ball?

I'd like to hear your good feeling on modeling this process

Thx a lot

sheng

If you are choosing commercial code to do the work, a friend from Ricardo told me their code is good in moving mesh. based on predicted moving object positions, you set up several stages of mesh before computation. After compress/stretch mesh to a certain degree, map the solution of current stage onto new mesh of next stage in computation. The stretch and squeeze of mesh will be kept to minimal. You can contact www.ricardo.com for details.

If ball motion is unknown, you are coding the solver, and you don't want to remap solution or over stretch the mesh, I would try to use porosity in stationary mesh. The liquid fraction is 0 when an element is fully occupied by the ball, and 0~1 when partiall occupied. This approach is easier to program, but the accuracy for boundary layer around the ball is a problem.

You haven't specified the problem exactly, so the operating parameters are unknown (e.g., volumetric flow rate, relative size of the ball to the container and the exit port, the extent to which you wish to get a solution - do you want to see a solution up to the point that the ball reaches the exit port or including its motion through the port, etc.)

The reason for the above questions is simple: Sometimes we tend to over-complicate analytically simple problems by solving them using CFD.

Depending on the complexity of the problem, you can get quite accurate solutions using basic 1-D type flow analysis and your fluids knowledge. Try this route first before you spend alot of time worrying about meshing.

Adrin Gharakhani, Sc.D.

This is emptying process. Due to gravity force, the liquid in the container will flow through the bottom outlet. When the bath height decrease, the concentration of inclusion will increase at outlet.

Then, we want to put a big ball as a flow control device in order to prevent some inclusion particles flow out.

If without ball, I can simulate the process as a typical filling-out process (the container is 3D clinder). The predicted inclusion concentration at outlet also can be predicted. Now, considering A BIG BALL, I can put a 2D circle inside a 2D plane (by embedding-fine grid or multi-block), however, I am really confuse on how to put a 3D sphere into a clinder. By the way, I use commerical software based on FDM as grid generation tool.

Thanks a lot for your kind suggestion!

Sheng