Reliable prediction of stagnation-point heat flux is essential in hypersonic vehicle design, but high-fidelity CFD is expensive. We encode nose-cone geometries with a variational autoencoder (VAE) into a compact latent space and learn a conditional MLP mapping (latent, flight conditions) → heat-flux. The surrogate is accurate on a small subset of the design space and generalizes beyond training conditions, enabling rapid design evaluation with much lower computational cost.
Yaşar, H. A., & Sevinc, O. K. (2025). Rapid Stagnation Heat Flux Prediction on Parametric Nose Cones in Hypersonic Flow via Variational Autoencoding. Proceedings of the 11th European Conference for Aerospace Sciences (EUCASS 2025), Rome, Italy. https://doi.org/10.13009/EUCASS2025-722
@inproceedings{YasarSevinc2025EUCASS,
author = {H{\"u}seyin Avni Ya{\c{s}}ar and Oguz Kaan Sevinc},
title = {Rapid Stagnation Heat Flux Prediction on Parametric Nose Cones in Hypersonic Flow via Variational Autoencoding},
booktitle = {Proceedings of the 11th European Conference for Aerospace Sciences (EUCASS 2025)},
address = {Rome, Italy},
month = {June--July},
year = {2025},
doi = {10.13009/EUCASS2025-722},
url = {https://doi.org/10.13009/EUCASS2025-722},
publisher = {EUCASS Association}
}