In the aerodynamic performance of wind turbine blades and other airfoil-based systems. Accurate prediction of roughness effects is essential for reliable performance assessment and long-term power loss estimation in wind energy applications [1]. Roughness elements disrupt the viscous sublayer, enhance near-wall turbulence production, and alter the mean velocity profile through a downward shift of the logarithmic region, commonly characterized by the roughness function ΔU⁺. These mechanisms result in increased skin-friction coefficients and modified boundary layer development, directly affecting aerodynamic loads and overall performance [2]. Despite this established understanding, most experimental studies and model calibrations are based on canonical configurations, such as zero-pressure-gradient boundary layers, pipe flows, or channel flows. While these simplified cases provide valuable insight, they do not represent the complex flow physics encountered in practical applications, where significant favourable and adverse pressure gradients are common. This limitation is particularly evident in flows with adverse pressure gradients and incipient separation, which frequently occur on wind turbine blades operating at high angles of attack or under off-design conditions. In such regimes, surface roughness not only increases skin friction but also destabilizes the boundary layer, promoting earlier separation and amplifying performance losses [3]. This study aims to evaluate the predictive capability and limitations of RANS (Reynolds-Averaged Navier–Stokes) numerical simulations in capturing the effects of surface roughness on the aerodynamic characteristics of airfoils, by comparing numerical results with experimental data. An analysis is carried out to determine the equivalent roughness of a specific roughness configuration applied at the leading edge of airfoils, followed by numerical simulations to assess the resulting predictions. This roughness configuration corresponds to the standard roughness definition established by Abbott and Doenhoff in [4]. Turbulence models, including k–ε, k–ω SST, and Spalart–Allmaras, combined with various roughness corrections, are evaluated to quantify model accuracy and identify the conditions under which current roughness modeling approaches remain valid or require further development.