NASA Common Research Model (CRM)​​​​

 

Task: Predict Predict Surface Pressure and Skin Friction Distributions on the aircraft surface.

​

​

​​

This benchmark dataset provides high-fidelity aerodynamic data for the NASA Common Research Model (CRM), a widely used open-source aircraft configuration in the AIAA community. Designed to address the challenges of scaling surrogate modeling techniques to large-scale 3D cases, this dataset supports research in aerodynamic load predictions and control surface effects across a broad flight envelope.

​

The dataset includes 149 high-fidelity CFD simulations using the DLR TAU solver with Reynolds-Averaged Navier-Stokes (RANS) equations and Spalart-Allmaras turbulence modeling. Each simulation varies six key parameters: Mach number, angle of attack, inboard and outboard aileron deflection, elevator deflection, and horizontal tailplane deflection, all at a fixed 37,000 ft altitude. The computational grid consists of 43 million points, with 454,404 surface points, ensuring high-resolution aerodynamic predictions.

​

As a benchmark for surrogate modeling, this dataset enables the development and validation of AI-driven aerodynamic models. The Halton sequence-based Design of Experiment (DoE) ensures optimal parameter space coverage, with 105 training samples and 44 test samples. Each file contains surface coordinates, normal vectors, pressure coefficients, and skin friction coefficients, making it a valuable resource for aerospace engineers, researchers, and AI practitioners working on large-scale aerodynamic modeling and digital twin development.