Dear Rob, thank you for your reply. I did use a slip wall, as shown in the figure. While we can see the turbulent kinetic energy is zero at free surface. That means zero value boundary conditon is used for k and epsilon transportation equation. The turbulence viscosity is also zero at free surface. But free surface is not wall bounded, turbulence should happen there.

So I'm thinking could I modify boundary condition for k and epsilon transportation equation, like change zero value to zero flux for k and epsion at free surface?