What is it about?
In very low Earth orbit (VLEO), the atmosphere is so thin that individual air atoms travel freely and strike satellite surfaces one by one, instead of flowing around them like a continuous gas. Molecular dynamics (MD) simulations can capture these impacts with high accuracy by modeling each interaction with an aluminum oxide surface at the atomic scale. MD helps us see exactly how atoms rebound, but it’s limited to small surface patches and can’t handle the full geometry of an entire satellite. To bridge that gap, we take the detailed MD results and train an AI model to learn bouncing behavior for any incoming atom speed and angle. This AI-driven approach plugs into existing satellite-flow simulations, allowing us to calculate aerodynamic forces on a whole satellite more accurately than ever before.
Featured Image
Photo by Kevin Stadnyk on Unsplash
Why is it important?
As satellite launches increases and space becomes more accessible, low Earth orbit (LEO) is filling up fast - with thousands of active satellites and growing clouds of debris. This congestion increases the risk of collisions and cascading breakups, known as the Kessler syndrome, which could render parts of LEO unusable. Very low Earth orbit (VLEO) presents a promising alternative. It’s less crowded and naturally self-clearing thanks to atmospheric drag, which causes satellites to re-enter after a few months. But that same drag also limits mission lifetimes, making it essential to predict aerodynamic forces with high precision. To do that, we must understand how individual air atoms bounce off satellite surfaces. This enables a range of mission-critical improvements: • Satellite designs that minimize fuel consumption and extend orbital lifetimes • The ability to harness aerodynamic forces for altitude and orbit control, reducing dependence on propulsion systems • Enhanced drag modeling to support the development of atmospheric-breathing engines for ultra-efficient operations • More accurate force predictions that lead to smarter, more reliable mission planning
Read the Original
This page is a summary of: A machine learning framework for scattering kernel derivation using molecular dynamics data in very low Earth orbit, Physics of Fluids, September 2025, American Institute of Physics,
DOI: 10.1063/5.0287359.
You can read the full text:
Contributors
The following have contributed to this page







