Multi-Fidelity Analysis Approaches to Aircraft Icing

Various aircraft icing analysis approaches, ranging from simple 1D scaling analyses to 3D fully coupled computational fluid dynamics methods, have been validated against a swept-wing glaze icing tunnel experiment. The accuracy, utility, and computational expense of the varying approaches are compared to better inform analysis choices. The lower-fidelity methods (1D, 2D) can be extremely computationally efficient and are well suited for simple geometries and analyses where the ice thickness or growth potential is the primary goal. Higher-fidelity methods (3D) can be computationally expensive but are appropriate for complex geometries with secondary flow and applications where accurate ice shapes are critical (e.g., predicting aerodynamic impacts of ice accretion on aircraft surfaces). The tradeoff in computational cost for accuracy can be rather extreme. The simplest method (1D scaling analysis) yields limited outputs with order-of-magnitude accuracy (stagnation thickness within ±50%) but can rapidly reveal valuable scaling information. The highest-fidelity methods (fully coupled computational fluid dynamics) can provide a full picture of 3D ice accretion and aerodynamic impacts, but they can take days or weeks to solve even moderate-duration ice accumulations.