Nacelle aerodynamic design and optimisation

Abstract

For a required thrust level, the reduction of the specific fuel consumption of an aero-engine can be achieved through an improvement of the propulsive efficiency. This is typically enabled by higher bypass ratios and lower fan pressure ratios. This may result in a larger fan diameter that increases the overall aircraft drag and weight as well as the interference effects between the engine and the airframe. Additionally, the next generation of Ultra-High Bypass Ratio (UHBPR) aero-engines are expected to be installed in a more close-coupled position with the airframe due to multi-disciplinary constraints such as structural loads or ground clearance. Consequently, the aerodynamic interference effects of the installation further increase and the potential benefits of the new UHBPR aero-engine cycles might be eroded. As a result, new design technologies that consider the engine coupled with the airframe have to be developed. The main focus of this project is on the aerodynamic shape design of compact nacelles in an airframe-installed configuration. This is a challenging design problem governed by computationally expensive numerical simulations, transonic non-linear flow physics and a high dimensional design space. Moreover, this design problem is typically subjected to computational constraints to meet industrial time scales. Therefore, surrogate modelling techniques to accelerate the optimisation process are a key aspect of the work. Single-fidelity and multi-fidelity surrogate models based on Artificial Neural Networks and Kriging interpolation were assessed. While a Reynolds-Averaged Navier-Stokes (RANS) Computational Fluid Dynamics (CFD) method was used as high-fidelity, an inviscid CFD model was used as a low-cost, low-fidelity method. The research quantified the effect of key aspects in the installed nacelle design problem such as the design variables considered, the surrogate modelling technique, the sampling size of the design space and the CFD fidelity. From the traditional Eulerbased method to assess the aerodynamic integration of the aero-engine in the airframe, a key novel contribution of this research is a systematic evaluation of different modelling approaches to quantify trades between aerodynamic performance and computational effort. The work showed that the size of the design space population for the surrogate models used, even with lower fidelity data, was the most important parameter to determine the design space gradients and identify the optimum design. At the same computational cost, the added complexity of multi-fidelity methods provided no benefits. Overall, for the first time this research established that for installed nacelle configurations, the industrial design problem is best addressed using single-fidelity Kriging models based on Euler CFD data with a posteriori RANS CFD evaluation of the optimal design. It was successfully shown that using this novel approach the installed nacelle design could be improved to provide a 0.6% reduction in cruise fuel burn, while also providing the design space maps to enable trade studies on key nacelle geometric parameters.