Unveiling wing turbulence dynamics through explainable deep learning
Abstract
The three-dimensional organization of coherent structures in the turbulent flow over aircraft wings remains elusive, limiting our ability to reduce fuel consumption. Here, we use explainable artificial intelligence to characterize these structures by predicted flow evolution rather than classical predefined kinematic criteria. After training a deep neural network to predict the short-term evolution of the flow, we identify the highest-relevance flow regions using Shapley-value attribution methods. We have found that as the flow decelerates towards the trailing edge, predictive importance shifts from near-wall low-speed structures to three-dimensional pairs of high- and low-speed fluid regions. Matching no single classical structure family, these pairs progressively dominate the dynamically relevant flow. Most stand spanwise side-by-side, enclosing a near-vertical momentum interface with the strongest velocity jump. Their geometry remains invariant in viscous units while their volume expands when approaching separation. Our results reveal a predictive organization that traditional paradigms overlook, opening new ways to control turbulent flows.
Related
- Unveiling two-phase flow behavior and flooding precursors in a Rotating Packed Bed through deep learning-based image analysis
- Explainable Deep Learning for Price-Trade Dynamics: From Black-Box Forecasts to Effective Parametric Models
- Dynamics Creation through Neural Dynamical Transfer Learning
- Full Poincaré Beam Through Fresnel-Scaled FSO Turbulence: An Experimental Statistical Investigation and Performance Analysis
- Unveiling Three-Dimensional Skyrmion Transitions Through Vortices and Monopoles
BibTeX
@misc{molinacasino2026unveiling,
title = {{Unveiling wing turbulence dynamics through explainable deep learning}},
author = {Samuel Molina-Casino and Andrés Cremades and Sergio Hoyas and José I. Cardesa and François Chedevergne and Ricardo Vinuesa},
howpublished = {arXiv},
year = {2026},
month = sep,
doi = {10.48550/arxiv.2609.05015},
eprint = {2609.05015},
archivePrefix = {arXiv},
url = {https://arxiv.org/abs/2609.05015},
note = {CC BY-SA 4.0},
}
RIS
TY - UNPB
TI - Unveiling wing turbulence dynamics through explainable deep learning
AU - Samuel Molina-Casino
AU - Andrés Cremades
AU - Sergio Hoyas
AU - José I. Cardesa
AU - François Chedevergne
AU - Ricardo Vinuesa
PY - 2026
DA - 2026/09/04
PB - arXiv
DO - 10.48550/arxiv.2609.05015
UR - https://arxiv.org/abs/2609.05015
AB - The three-dimensional organization of coherent structures in the turbulent flow over aircraft wings remains elusive, limiting our ability to reduce fuel consumption. Here, we use explainable artificial intelligence to characterize these structures by predicted flow evolution rather than classical predefined kinematic criteria. After training a deep neural network to predict the short-term evolution of the flow, we identify the highest-relevance flow regions using Shapley-value attribution methods. We have found that as the flow decelerates towards the trailing edge, predictive importance shifts from near-wall low-speed structures to three-dimensional pairs of high- and low-speed fluid regions. Matching no single classical structure family, these pairs progressively dominate the dynamically relevant flow. Most stand spanwise side-by-side, enclosing a near-vertical momentum interface with the strongest velocity jump. Their geometry remains invariant in viscous units while their volume expands when approaching separation. Our results reveal a predictive organization that traditional paradigms overlook, opening new ways to control turbulent flows.
ER -