Design optimisation of separate-jet exhausts with CFD in-the-loop and dimensionality reduction techniques

For Ultra-High Bypass Ratio aero-engines, the exhaust system is likely to play a significant role on the aerodynamics and performance of the aircraft. For this reason, relatively rapid methods for the aerodynamic design and optimisation of exhaust systems are required to inform design decisions at early stages of the design process. Previous exhaust optimisation works encompassed Response Surface Model (RSM) based optimisations of nozzle configurations that were parametrised with a significant number of design variables. The RSM were constructed with a large database of designs that were assessed with fine computational meshes and well resolved boundary layers. However, the large number of design variables and the computational cost required to evaluate each exhaust design limited the optimisation capabilities. This work develops a relatively more rapid exhaust optimisation method based on CFD in-the-loop and dimensionality reduction. The methodology is based on coarse meshes and wall functions to guide the optimisation process and is coupled with methods for the identification of the dominant design variables. For an UHBR aero-engine exhaust design space of 16 design variables, it was found that the velocity coefficient could be characterised with only seven parameters. Based on these results, various optimisation methods were developed and applied. These targeted the maximisation of the velocity coefficient by optimising just the 7 dominant design variables. With these approaches, a similar benefit in exhaust performance relative to the baseline optimisation method was obtained approximately 4 times faster.