Hi guys i am trying to solve a free surface problem and the geom is a 3D rectangular block with a circular inlet at one face and the rectangular outlet at the other face and i have been using an inlet velocity and opening boundary condition at the outlet.I am using laminar flow with the homogeneous model having standard free surface model.

But the residuals are not converging to set value and the solution is not stable also and i am using the time step of 0.1 sec. How to solve this problem, waiting for your responses.

Best regards, saqib mahmood.

Ensure you refine your mesh where you expect the interface between the fluid and gas to be. It's hard to say if 0.1 [s] is suitable without knowing the size of the geometry or the advection time. But reducing the timestep may help.

thanks johnny for your reply, but the dimensions of the geometry are:

Length = 10 cm width = 10 cm height = 10 cm

inlet diameter of 1 cm and velocity of 0.1m/sec and at outlet the diameter is about two times of the inlet and i want to make this a laminar solution using homogeneous model initially i assumed the box is full of air and then water start coming in the box and then the water level rises and i am using both pressure outlet and opening boundary conditions. Kindly please tell me about this advection time also. Because my convergence monitors are not converging to set values and it seems after 40 or 50 timesteps that the solution will never converge.

best regards, sam.

I had the same problem with free surface flow. Try the adaptive step size option, starting with an initial step size of 1e-6s or smaller. If you are using CFX11 you could select the volume fraction coupling from the advanced solver options. This may enhance convergence.

thanks alterego, but what about the maximum and minimum timestep if you are already giving initial time step and i want to ask one more question that i want to run this simulation as laminar and when the turbulence option is selected as laminar in CFX-Pre and i run the solution it give an error saying that Reynolds number is out of the range but in my case Reynolds number is 1600 and its laminar with a velocity of 0.15 m/sec and now the thing i am confused is that for internal flow i.e in my case i have a inlet of 1 cm diameter the characteristic dimension will be the diameter of the inlet or the length of the tank for Calculating Reynlods number.

regards, sam

CFX cannot guess what your characteristic dimension is, can it? So the characteristic Length used by CFX to <u>estimate</u> the Reybolds Number is the cubic root of the volume.

If your aren't obtaining convergence (what should be obtained within each timestep, and not after 40-50 timesteps (or did you mean 40-50 iterations?) ), decrease the timestep. I guess you will need a timestep some order of magnitudes lower than 0.1 s, specially at the biginning of your simulation. Follow the sugestion alterego

Regards

Thanks Rui and alterego,

By reducing the time step the convergence issue is resolved and the results are also very good and the solution ran pretty smooth.

regards, sam.