Aircraft in flight are subject to numerous kinds of uncertain conditions such as turbulence, surface perturbations due to icing or contamination, as well as uncertainty due to manufacturing tolerances. Robust design is a nondeterministic optimization process that aims to take into account the impact of these uncertainties. In the present work, a multipoint, multilidelity design framework is developed in the Isight environment to explore new methodologies for the surrogate-based robust design of airfoils. To this end, class and shape function transformations are employed to model both the airfoil nominal geometry and its deviations, enabling an efficient exploration of the design space under uncertainty. Various surrogate modeling techniques are evaluated as part of this implementation, including advanced surrogate models based on the Kriging technique. In addition, various formulations of the design objective are considered to control the trade-off between design performance and design robustness, through the specification of different weights for the variance term in the objective. The design framework is then applied to both singlepoint and multipoint design optimizations, and the contribution to robustness provided by multipoint design is compared to the design robustness achieved by minimizing the statistical variance of the objective. A number of optimization cases are presented in which the trade-off between the mean and variance of the lift-to-drag ratio is evaluated and compared to deterministic designs, for both singlepoint and multipoint formulations.
展开▼
