Abstract
A Multi-point Approximation Method (MAM) coupled with adjoint is presented to increase the efficiency of a modern jet-engine fan blade. The study performed makes use of Rolls-Royce in-house suite of codes and its discrete adjoint capability. The adjoint gradient is used along with MAM to create a Design Of Experiment to enhance the optimization process. A generalized Free-Form Deformation (FFD) technique is used to parametrize the geometry, creating a design space of 180 parameters. The resulting optimum blade at design conditions is then evaluated at off-design conditions to produce the characteristic curve, which is compared with real test data. Finally, a preliminary Active Design Subspace (ADS) representing the fan efficiency is created to evaluate the robustness of the objective function in respect to the most significant design parameters. The ADS allows to collapse a large design space of the order of hundreds parameters to the few most important variables, measuring their contribution. This map is valuable in many respects to the fan designers and manufacture engineers to identify any ridges where the performance may deteriorate rapidly, hence a more robust part of the design space can easily be visualized and identified.
