[adenlan]

Hello everyone,

I have a question about free surface flow and convergence. I made a simple model where I defined the free surface and hydrostatic pressure with CCL expressions following the different free surface tutorials. The fluid is non moving, the domain is a 10 x 5 m rectangle, the mesh is hexa, and refined at the free surface. BCs are walls except for the top which is an opening. The same optios as the free surface tutorial are applied except for the turbulence modle(laminar because the fluid is not moving).

I thought this hydrostatic model would converge really quickly since conditions are , in my opinion, well defined. Yet the model in steady state does not converge at all (in terms of residual).

Do I miss out on something obvious ?
Is it normal that residuals are typically high with free surface flows ?


Thank you for your attention and future (I hope) answers !

Aden

[ghorrocks]

Getting convergence on a steady state simulation with a free surface is often difficult. It is because the free surface always has waves on it. Even numerical noise is enough to start them off. So I often run them as transient simulations and they converge fine.

[vishal.indomitable]

There are some issues for convergence in ANSYS-CFX for free surface flow.

[ghorrocks]

The issues are due to the surface waves as I said in my first post. Every other CFD code I have tried also suffers from it to. No matter what software I use, free surface simulations are always done transient.

There is also a problem in CFX in that the surface tension is included in an explicit manner and therefore not properly linearised in the implicit main equations. This means surface tension simulations are not as fast as they could be on CFX. But this only applies to surface tension simulations, general free surface simulations are fine.