Abstract

It is the objective of every fan aerodynamicist to design blades that perform favorably under manufacturing variations. This is a challenging task, and for the fan it is particularly important as any offsets in its performance have a significant impact on the overall engine specific fuel consumption. While the use of multi-objective optimizers and robust design methodologies are no doubt of importance, they often fail to offer pedigree rules that can be applied to retro-fit existing fan blades or aid in the design of new fan blades. This motivates the present work. Our approach in this paper is to leverage active subspaces: an emerging set of ideas for dimension reduction in parametric computational studies. A key component of active subspaces lies in determining which linear combination of select inputs drive the model output. It is from this reduced subspace that we seek to develop pedigree rules for design.