That’s a nice first simulation, I have to say. Keep it up!
Now I don’t know much about architectural waterfalls, but it feels to me that it would be a single phase flow. In your second simulation, if there’s only water coming from the pipes, then there would be no air in the domain.
A couple of things about the setup. The boundary condition that you set to the outlet is causing the simulation to crash. You currently have a pressure inlet-outlet velocity BC for the outlet face. This BC will prevent any fluid from going out. You can change it to an ordinary pressure outlet.
If you really need this simulation to be multiphase, I believe you can use the local time step approach, which will help you achieve a “steady state” faster (saving you quite a lot of computational effort). If you decide to go for it, you can activate this option here:
Keep us informed