Topolgy optimization of wings with serrated trailing edges using Design-by-Morphing and Bayesian optimization

Aerodynamic efficiency and noise suppression are two seemingly conflicting objectives that aerodynamicists, particularly in the wind energy sector, need to strike a balance between. Serrations on the trailing edges of wings, perhaps inspired by natural predators such as owls, is an active area of research for aerodynamicists. However, the inhibitive cost of high-fidelity simulations to optimize the designs of such wings using conventional topology optimization methods, coupled with designer-biased topological spaces, forces aerodynamicists to work within narrow design spaces where radical improvements are unlikely. In our current work, we optimize such wings with serrated trailing edges using a two-pronged approach: (1) Design-by-Morphing (DbM), a data-driven design strategy for creating a continuous and constraint-free design search space by morphing homeomorphic shapes, that can produce radical out-of-sample designs, something which is unique from conventional design strategies; and (2) a Mixed-variable, Multi-Objective Bayesian Optimization (MixMOBO) algorithm, that can optimize expensive, black-box problems with a few functions calls. DbM-MixMOBO produces non-intuitive groundbreaking designs which exhibit considerable improvements.