Large-scale Deterministic Debris Simulation (LADDS)

The increasing number of satellite launches poses challenges to space debris tracking, analysis and prediction. Current methods to simulate the evolution of space debris rely on Monte Carlo approaches to tackle the computational challenges of tracking a large number of possible collisions of space debris. The goal of this study is to create a deterministic collision simulation using recent high-performance computing (HPC) advances to tackle the computational costs. This should allow smaller time steps and higher simulation fidelity.

This project is conducted in an ARIADNA study with the Chair of Scientific Computing in Computer Science at the Technical University Munich and ESA's Space Debris Office at ESOC.

Project overview

Recent years have seen a substantial increase in satellite launches. In 2020, for the first time, over a thousand objects were launched into low earth orbit [1] (see Fig. 1). Many of these launches immediately contribute to the space debris environment with, e.g., spent upper stages [2]. The debris concentration is further amplified by fragmentation events, such as explosions or collisions, which can create large numbers of small fragments [3] (see Fig. 2).

Current statistical estimates put the number of debris objects at 128 million for objects between 1 mm and 1 cm, 900 000 for 1 cm to 10 cm and 34 000 for objects larger than 10 cms [4]. Space debris has become a central topic for space agencies [5] and ESA in particular, which has been pioneering this field with its Clean Space initiative and recently commissioned the first active debris removal mission [6].

This study explores an alternative route and aims to incorporate state-of-the-art techniques from HPC to achieve the necessary computational efficiency and scaling to create a simulation that relies on deterministic conjunction tracking instead. This bears the promise to combat the inherent variability of Monte Carlo simulations [7]. A deterministic approach has become conceivable now, as hardware and software to run petascale numerical simulations are available [8,9]. The study relies on utilizing AutoPas [9], a software library for autotuned particle simulation, which has seen successful application in the context of molecular dynamics simulations [10-12].

References

1. Lemmens, Stijn, and Francesca Letizia. "ESA annual space environment report." ESA Space Debris Office, Darmstadt, Germany, Tech. Rep. GEN-DB-LOG-00288-OPS-SD (2020).
2. European Space Agency. “About space debris,” esa.int. [Online]. Available: http://www.esa.int/Safety_Security/Space_Debris/About_space_debris. [Accessed: 04-Feb-2021]
3. Frey, Stefan, Camilla Colombo, and Stijn Lemmens. "Application of density-based propagation to fragment clouds using the Starling suite." 1st International Orbital Debris Conference (IOC). 2019.
4. ESA Space Debris Office. “Space Environment Statistics.” Space Environment Statistics · Space Debris User Portal, 08-Jan-2021. [Online]. Available: https://sdup.esoc.esa.int/discosweb/statistics/. [Accessed: 04-Feb-2021]
5. Kessler, Donald J., et al. "The kessler syndrome: implications to future space operations." Advances in the Astronautical Sciences 137.8 (2010): 2010.
6. European Space Agency. “ESA commissions world's first space debris removal.” [Online]. Available: https://www.esa.int/Safety_Security/Clean_Space/ESA_commissions_world_s_first_space_debris_removal. [Accessed: 04-Feb-2021]
7. Lidtke, Aleksander A., Hugh G. Lewis, and Roberto Armellin. "Statistical analysis of the inherent variability in the results of evolutionary debris models." Advances in Space Research 59.7 (2017): 1698-1714.
8. Breuer, Alexander, et al. "Sustained petascale performance of seismic simulations with SeisSol on SuperMUC." International Supercomputing Conference. Springer, Cham, 2014.
9. Gratl, Fabio Alexander, et al. "Autopas: Auto-tuning for particle simulations." 2019 IEEE International Parallel and Distributed Processing Symposium Workshops (IPDPSW). IEEE, 2019.
10. Gratl, Fabio Alexander, et al. "Optimizing Molecular Dynamics Simulations with Dynamic Auto-Tuning." SIAM Conference on Parallel Processing for Scientific Computing 2020. 2020.
11. Seckler, Steffen, et al. "AutoPas & ls1 mardyn: Enabling Auto-Tuning in MPI+ X Load-Balanced Molecular Dynamics Simulations." Particles 2019. 2019.
12. Gratl, Fabio Alexander. "Leveraging Node-Level Performance for Molecular Dynamics through Auto-Tuning." SIAM CSE. 2019.

Outcome