Graduate research · University of Dayton · AIAA SciTech 2024
Multi-fidelity surrogate models for uncertainty quantification
Built multi-fidelity sparse polynomial chaos surrogates for the UQ Challenge Problem: a NACA 2412 airfoil with a simple flap, analyzed in XFOIL. The surrogates reproduced 50,000-run Monte Carlo results for the aleatory, epistemic, discretization-error and uniform-input cases from as few as 11 high-fidelity runs, with compressed sensing performing best.
- Method
- Multi-fidelity sparse PCE with compressed sensing
- Test case
- NACA 2412 with flap, XFOIL
- Result
- Matched Monte Carlo with 11 to 151 high-fidelity runs
- Co-author
- Markus P. Rumpfkeil
