Implementation and Assessment of Roughness Modifications to 2-equation, 4-equation and Full Reynolds Stress Models

When it comes to modeling surface roughness for turbulent flows over rough surfaces, equiv-

alent sand grain roughness height (ks) based Reynolds-Averaged Navier-Stokes (RANS) models

are promising candidates due to their reasonable accuracy and low computational cost. We con-

sider three sublayer resolved turbulence models, Chien’s low Reynolds number k − ε, Durbin’s

k − ε − ̄v2 − f , and the SSG/LRR Full Reynolds Stress Model (FRSM), for rough wall appli-

cation. In RANS modeling of rough wall flows, surface roughness effects can be introduced in

turbulence models through a single parameter, ks. For example, a roughness-modified version of

Chien’s low Re k − ε turbulence model is available in the literature. In this work, we modify a

̄v2 − f and FRSM to accommodate surface-parameterized roughness. The roughness effects are

introduced in the ̄v2 − f model by modifying the wall boundary conditions as well as governing

equations, whereas FRSM is modified by altering wall boundary conditions alone. At first, we

demonstrate the smooth wall performance of the aforementioned turbulence models at both low

and high Reynolds numbers, and next, we compare the performance of the roughness-modified

versions of these models against results obtained from low Reynolds number Direct Numerical

Simulations (DNS) and high Reynolds number experiments.