PALMO: A High-Fidelity Machine Learning Airfoil Performance Database

​​​​

Task: OVERFLOW Machine Learning Airfoil Performance Database Predicting airfoil performance for different airfoils and flow conditions.

​

The OVERFLOW Machine Learning Airfoil Performance (PALMO) database is a publicly available, high-fidelity dataset designed to advance airfoil performance modeling for aerospace applications. Built using OVERFLOW simulations with high-order accuracy and Spalart-Allmaras turbulence modeling, PALMO provides precise aerodynamic data across a broad range of conditions.

​

At its core, PALMO features an airfoil base cube, parameterized over Mach number, Reynolds number, and angle of attack, covering NACA 4-series airfoils with varying thickness and camber (NACA 0006 to NACA 4424). With 52,480 simulations run on NASA’s HECC supercomputer, the database enables engineers and researchers to develop accurate airfoil performance look-up tables without additional high-performance computing.

​

PALMO is also a benchmark dataset for machine learning, supporting AI-driven surrogate models for aerodynamic predictions. Early validation shows that a PALMO-trained model predicted airfoil performance for an unseen NACA 3415 airfoil within 2.1% of CFD-generated results, demonstrating its reliability for aerospace design and optimization.

​

By offering high-resolution airfoil performance coefficients, PALMO provides a valuable resource for engineers, researchers, and AI practitioners working on aircraft design, aerodynamic analysis, and machine learning applications in aerospace.