Optimization of installed compact and robust nacelles using surrogate models

Abstract

The design and optimization of aero-engine nacelles in a configuration installed on the airframe may be an important consideration to realize the cycle benefits of new ultra-high bypass ratio aero-engines. However, this is typically a high-dimensional design problem and there is a need to reduce the associated computational costs. This work presents a method for aerodynamic nacelle optimization for an installed configuration and provides further knowledge about the characteristics of this design space. The methodology includes single fidelity surrogate models built with inviscid flow solutions. Gaussian process regression and artificial neural networks are tested as modelling techniques. Viscous computations are used to assess the optimized designs at cruise and off-design windmilling diversion condition. This approach yielded an optimal design with a reduction in fuel burn of about 0.56% relative to a design optimized in isolated configuration without considering the powerplant integration effects. The optimal design also met the robustness criteria in terms of limited flow separation at the windmilling diversion conditions.