Analysis of Wingtip Devices Using CFD Simulations and A CVAE Model

The wingtip of an airfoil is considered a critical aerodynamic interface. While the size of a wingtip is small relative to the wing, the importance of its effect on fuel consumption, weight control and aerodynamic behavior during turbulence is significant. This study investigates the design of wingtip devices with CFD simulations and a generative ML model. We hypothesize that swept and grid wingtip designs offer superior aerodynamic performance for rotary-wing aircraft compared to alternative designs. To achieve this goal, we first analyzed aerodynamic efficiency of selected wingtips from data collected from the CFD simulations utilizing OpenFOAM platform. Specifically, we selected wingtips with grid, tip, sail, and swept types, analyzed lift and drag curves at multiple angles of attack and examined the unique impact of spanwise velocity on boundary layer stability. Our results show that the swept and grid wingtips performed the best. Additionally, we proposed a new rapid-iteration design framework using a Conditional Variational Autoencoder (CVAE) to address the challenge of obtaining data samples and identifying optimal aerodynamic relationships. Using the latent space, our CVAE model identified relationships between various designs based on combined aerodynamic properties and revealed the semantic differences of swept and grid wingtips compared to the other types. The advantage to this method is that it allows for much quicker iterations and the identification of promising designs. Our work suggests that it will be important to examine factors like the number of grid elements, their individual shapes, and spacings for how to create the ideal wingtip device.