A numerical approach for the prediction of roughness evolution during in-flight icing is presented. The roughness distribution in space and time is determined mathematically through the analysis of the behavior of the water layer flowing over the icing surface, with special consideration of the transition between stationary and moving beads. A Lagrangian model is used to identify the main statistical features of the stochastic process of bead nucleation, growth and coalescence under glaze ice conditions. The results provide information on average parameters that are then implemented in FENSAP-ICE. Computed maximum and average roughness values are successfully compared with experimental data on a NACA 0012 airfoil
