Assessing relevant roughness scales for accurate predictions of iced airfoils in both glaze and rime conditions

Accurately predicting aircraft performance degradation due to icing is crucial for certifying next-generation aircraft designs. Current ice accretion models often simulate ice growth as two-dimensional and lacking small scale distributed roughness observed in ice accretion experiments. To evaluate the impact of local roughness on aerodynamic loads of iced airfoils, we conducted RANS and wall modeled LES analyses on a hierarchy of filtered ice shapes under rime and glaze conditions. Baseline shapes of realistic ice were derived from laser scans of ice formed on a NACA23012 airfoil at the NASA Glenn IRT. We generated increasingly smoothed ice shapes using a volume-preserving Laplacian smoothing algorithm to the limit of two-dimensional shapes like those generated from ice accretion models. For glaze ice with horn-like structures, filtered roughness scales minimally affect lift, drag, and pressure profiles. However, for rime ice conditions, removing the smallest roughness scales significantly alters the lift curve to the extent of mispredicting the stall condition. This highlights the need for either artificially generated roughness in rime-iced environments or improved sub-grid roughness models.