Obtaining RMS of turbulent velocity fluctuations
 

I need to collect the RMS (standard deviation) of the velocity fluctuations in my model due to turbulence (i.e. u', v', and w'). There is one thing that I'm confused about. Do I need to run this as an unsteady solution or can I run it as steady state. I don't see how the two differ in my case because my boundary conditions are steady and my solution has converged on a solution. So now can I just run it using a steady solution for another thousand iterations or so and collect the info on the fluctuations in the local velocities? Will the iterations of the steady state solution resolve the fluctuations due to turbulence? If not, why would time steps, which I just see as groups of iterations, resolve the fluctuations correctly.