Input-output based analysis of convective velocity in turbulent channels*

This work employs an input-output based approach to computing convective velocities in turbulent channel flows based on stochastically forced Navier-Stokes equations linearized about a turbulent mean velocity profile. We use this model to examine both the mean convective velocity for fluctuating quantities as a function of wall-normal location, and the convective velocity associated with individual streamwise--spanwise wavelength pairs at each wall-normal location. We focus on the streamwise velocity and demonstrate the model's ability to reproduce the shape of the convective velocity profile, particularly its tendency toward a constant value in the near-wall region. The model predicted convective velocities also shows the Reynolds number invariance observed in computations using DNS data. We then isolate the contributions of each scale as well as each linear term in the momentum equation to examine their contribution to the convective velocity and their role in the known deviation from Taylor's hypothesis in the near-wall region.