RANS and LES simulation of airfoil ice accretion aerodynamics

RANS and LES modeling are applied to the geometrically complex problem of glaze ice accretion on fixed-wing and rotorcraft airfoils. The shortcomings of transport based RANS turbulence models for these systems is demonstrated using three experimental data sets. The roles of meshing topology, turbulence model choice, and 2D assumptions are quantified. Despite best practice implementation of RANS modeling, results show that these methods consistently under-predict stall onset and high-α lift. The reason for this poor performance is parametrically explored. It is shown that the smaller-scale (O[10² μm]) and larger scale (i.e., bluff "horns") geometric features of the ice shapes that give rise to a rich inertial 3D unsteady flow, at and aft of the leading edge, are primarily responsible, as these are largely inaccessible to RANS. Implicit LES results are presented that demonstrate improvement in predicted aerodynamic performance.