Rob
Ansys Employee

I'd also look at your turbulence boundary inputs, and (probably) increase the turbulent viscosity limiter value by a few orders of magnitude once you've confirmed the bc's are sensible. 

In terms of the convergence, look at the flow velocity and cell size. Divide cell (edge) size by  velocity and then that number by 10. This is a good starting point for a time step. You really need to converge each time step to get a physically sensible result. 

As an aside, are you using the University system? You may need to refine the mesh near the free surface and the student licence may struggle with the cell count.