Development of machine learning tools for aerospace design: wind tunnel investigations on a speed bump model

In recent years, Physics-Informed Neural Networks (PINNs) and Optimizing a DIscrete Loss (ODIL) have emerged as novel approaches to solving PDEs in many applications, including the Navier-Stokes equations in fluid mechanics. Through close collaboration with Boeing, we seek to develop a tool based on physics-guided machine learning that can bridge gaps in the experimental and computational efforts supporting the current aerospace design process. Flow field data on a Boeing Speed Bump model is collected in the subsonic Lucas Wind Tunnel facility at Caltech, to be used for the training and validation of advanced models. Static pressure tap measurements and multi-hole pressure probe measurements are presented. Potential directions for the development of machine learning tools and broader visions for the collaboration are discussed.