Speaker
Description
The recent surge in commercial space activity in Low-Earth Orbit (LEO) has led to increased interest in understanding the complex dynamics of space debris, estimating their long-term evolution, and defining an upper limit on sustainable satellite activity. The open-source MIT Orbital Capacity Assessment Tool (MOCAT) tackles this challenge through an integrated suite of models that offer varying levels of fidelity and computational efficiency.
MOCAT Monte Carlo (MOCAT-MC) represents the high-fidelity model that propagates individual space objects (SOs), models their interactions in terms of collisions and explosions, and is computationally efficient to predict millions of SOs over a 200-year period. MOCAT-MC has been validated against the Inter-Agency Space Debris Coordination Committee (IADC) study, which in 2013 had several space agencies use their MC tools to compare the performance for a strict future scenario.
MOCAT Source-Sink Evolutionary Model (MOCAT-SSEM) is the low- to medium-fidelity model that makes use of differential equations for the representation and evolution of the number of SOs, evolving distributions over orbital shells and species rather than propagating individual objects. The reduction of the fidelity of the model leads to large speed-ups with century scale simulations running in seconds. The original SSEM adopts two simplifying assumptions: all objects are on circular orbits, and fragments generated by collisions are retained within the altitude shell in which the collision occurs. Recent developments include debris fragment spreading, where fragment ejection velocities are used to redistribute collision fragments across neighboring altitude shells while retaining a circular orbit assumption, and an elliptical model that introduces explicit semi-major axis and eccentricity bins for the inclusion of elliptical orbits and extending fragment redistribution to operate across both altitude and eccentricity bins. Moreover, MOCAT-SSEM incorporates a feedback proportional-derivative controller and a nonlinear model predictive controller to optimize Active Debris Removal (ADR) subject to a defined cost objective, a constrained nonlinear programming optimization framework to compute the optimal orbital capacity of the low region of LEO, and several Integrated Assessment Models (IAMs) to support techno-economic analysis of debris-mitigation policies and ADR strategies. MOCAT-SSEM has recently undergone a cross-fidelity verification and benchmark-based validation study against MOCAT-MC by using a composite metric.
MOCAT Machine Learning (MOCAT-ML) is a surrogate model of MOCAT-MC with significantly reduced computational cost. By utilizing a Convolutional Gated Recurrent Unit (ConvGRU) architecture, the model captures complex spatiotemporal patterns to forecast the evolution of space object density maps. Recent advancements have extended this framework to provide accurate 100-year predictions in seconds, offering a scalable and efficient alternative to traditional physics-based simulators for long-term orbital capacity assessment.
MOCAT Quasi-Deterministic (MOCAT-QD) proposes an alternative to MOCAT-MC: a quasi-deterministic all-vs-all evolutionary model whose variance is greatly reduced, requiring much fewer simulations than traditional Monte Carlo evolutionary models. MOCAT-QD offers solutions that are almost unbiased with respect to the original corresponding all-vs-all simulator, and reduce the variance by a factor of up to 1,500, with computational cost only about 1.5 times larger. MOCAT-QD is augmented to be able to provide not only the mean, but also a user-defined standard deviations from the mean, with a single run.
Together, these open-source multi-fidelity tools establish MOCAT as a comprehensive and versatile framework for characterizing LEO capacity and supporting the implementation of data-driven space sustainability strategies.
| Which section would you like to submit your abstract to? | Session 2: “Challenges of space debris modelling” |
|---|