[blackbow20]

Dear CFD friends


I have to simulate the discharge of a canal into a lake. In such cases, the canal (pipe) lies quite often partially below the water surface. This looks like this, for example: Link

I'd like to simulate this with a two-phase simulation (water/air) and stationary. Now I have some troubles finding a appropriate boundary condition. I tried to solve it with a pressure BC with pascal's law at the outlet patch using a step function (OutletStepFunction) as follows:

Let's assume we have a pipe diameter of 1 m. The lake water level is at 0.7 m. My step function returns:
- 1 if the z value is below 0.7 m (water phase)
- 0 if the z value is above 0.7 m (air phase)

At the outlet i defined the relative static pressure as follows:

p_static = stepfunction*rho_water*g*h

implementet in CFX for the relative static pressure:

OutletStepFunction*997[kg m^-3]*g*z

This should lead to the right static pressure below water surface and to the reference pressure above water surface. When i check the inlet pressure distribution after initialisation it looks quite right. But sadly this is not stable. I also tried it with two different patches at the outlet (water & air), but got the same problem.

Is there an error in my reasoning? Or what would be a good BC to solve this problem?

Thanks and best regards
blackbow

[ghorrocks]

Conceptually your approach is correct but as you noticed, getting this sort of thing to be numerically stable is very hard. Specifying a BC with both phases and a complex pressure field is always going to be challenging.

Often the best way to treat these cases is to include the pipe exit in your model and put the boundary condition further up the pipe (rather than making the pipe exit the BC). You then put a BC up the pipe which is purely a single phase. So bend the pipe so you can make the boundary completely under water or completely above the water. Then the BC is far simpler and much easier to converge.