Innovative Turbulence Analysis Methodology

Our three-phase methodology enhances turbulence understanding through data curation, hybrid training, and validation for real-time forecasting and integration with advanced simulation tools.

A wind turbine rises above a thick layer of fog, with its blades silhouetted against the sky. Below, a forested area is partially visible through the mist, adding a sense of tranquility and mystery to the scene.
A wind turbine rises above a thick layer of fog, with its blades silhouetted against the sky. Below, a forested area is partially visible through the mist, adding a sense of tranquility and mystery to the scene.

Turbulence Methodology

Utilizing advanced techniques for turbulence data extraction and analysis.

A commercial airplane is parked on a runway, with a focus on the side view through a wire fence. The airplane has the logo of Far Eastern Air Transport, and is set against the backdrop of a cloudy sky and distant mountains. A picket fence and grass are visible in the foreground.
A commercial airplane is parked on a runway, with a focus on the side view through a wire fence. The airplane has the logo of Far Eastern Air Transport, and is set against the backdrop of a cloudy sky and distant mountains. A picket fence and grass are visible in the foreground.
Hybrid Training

Our hybrid training combines pretraining on ERA5 with specialization for cascade dynamics, ensuring a robust understanding of turbulence phenomena through physics-aware loss regularization.

Waves with foam and bubbles rise in a turbulent ocean under an overcast sky. The texture of the water is detailed, displaying various shades of gray and blue.
Waves with foam and bubbles rise in a turbulent ocean under an overcast sky. The texture of the water is detailed, displaying various shades of gray and blue.
Real-time Validation

We validate our models through a priori testing and real-time forecasting, enhancing speed and accuracy in turbulence predictions while integrating seamlessly with OpenFOAM preprocessors.

Three-Phase Methodology

Our methodology integrates data curation, hybrid training, and validation for turbulence forecasting.

Data Curation
An airplane wing with a smooth, metallic surface is visible against a backdrop of layered, wavy clouds stretching to the horizon. A subtle gradient of colors from light blue to warm orange hints at either sunrise or sunset, creating a serene and calm atmosphere.
An airplane wing with a smooth, metallic surface is visible against a backdrop of layered, wavy clouds stretching to the horizon. A subtle gradient of colors from light blue to warm orange hints at either sunrise or sunset, creating a serene and calm atmosphere.

Extract and encode turbulence data with multiscale features for enhanced analysis.

A close-up view of an airplane's vertical stabilizer or tail fin against a backdrop of a blue sky transitioning to a lighter gradient near the horizon. The tail fin has a logo featuring a stylized bird.
A close-up view of an airplane's vertical stabilizer or tail fin against a backdrop of a blue sky transitioning to a lighter gradient near the horizon. The tail fin has a logo featuring a stylized bird.
A large passenger airplane is visible in mid-landing on a runway. The weather appears foggy or misty, with limited visibility. In the foreground, there is wire fencing and scattered snow-covered vegetation, suggesting a winter setting.
A large passenger airplane is visible in mid-landing on a runway. The weather appears foggy or misty, with limited visibility. In the foreground, there is wire fencing and scattered snow-covered vegetation, suggesting a winter setting.
Hybrid Training

Specialized training using physics-aware loss to improve cascade dynamics understanding and predictions.

Real-time validation through vortex visualization and deployment for accurate forecasting and speedup.

Validation Process
Methodology

Utilizing advanced techniques for turbulence analysis.

© 2025. All rights reserved.