CFD Based Uncertainty Analysis of Airfoil Ice Accumulation

Abstract

It is the intent of this paper to perform an uncertainty quantification study of both leading edge and spanwise-ridge airfoil ice accumulations. The leading edge ice shapes are parameterized by the height of the horn and the angle with respect to the chord. The ridge ice shapes are parameterized by the chord-wise position and the radius. In both cases, 2-D RANS simulations are performed on modified ice shapes obtained by sampling from a bivariate Gaussian distribution of the two geometric variables. In the process of conducting this study, we created a preliminary database for the validation of future research into aerodynamic uncertainties due to wing icing. This data can also be used to establish a degree of con dence in iced-wing performance. In order to execute our study we needed to test thousands of profiles. The speed of the flow solver allows the computation of 500 ice profiles in 2 days. This computational power was achieved through the design, purchase, and assembly of a low-cost Linux node based desktop with a 16-core AMD processor. These rapid simulations provide the necessary function evaluations for implementing a Quasi-Monte Carlo method used to characterize the uncertainties. The relative importance of the geometric ice parameters are presented and discussed.