Various corrections were previously proposed to account for wall roughness with the k–ω and shear stress transport (SST) models. A simplified analysis, based upon the wall region analysis, is proposed to characterize the behavior of these roughness corrections. As this analysis points out some deficiencies for each correction, two new corrections are proposed for the SST model, to reproduce different behaviors, mainly in the transition regime. The correction development is based upon a previously developed strategy. A large set of boundary layer experiments is used to compare the different roughness corrections, confirm the failures of previous proposals, and validate the present ones. Moreover, it assesses the proposed simplified analysis. It also evidences the difficulty to determine the equivalent sand grain roughness for a given surface. The Colebrook based correction is recommended while the Nikuradse based one can add information about the envelope of possible behaviors in the transition regime.