Reflections on roughness modelling in turbulent flow

Abstract

The general topic of practical modelling of roughness in turbulent flow is discussed. Double averaging is a useful framework, but most models cannot be construed as term-by-term closures to the double averaged equations. Double averaging justifies a drag representation. Drag models are effective for both Reynolds averaged and eddy resolving simulation. Boundary condition models are primarily viable for Reynolds averaged closure. Those models are calibrated with the Hama roughness function for the log-law. But a perplexing observation is that the VonKarman constant depends on roughness height.