Multi-CubeSat mission planning enabled through parallel computing

Small CubeSat-class satellites are opening up new avenues for science and technology development within the space industry. What was once a purely educational tool has quickly become the newest international exploration platform for low earth orbit missions. These sub-10 kg satellites ride into space as tertiary payloads, kicked out of their launch vehicles after all other satellites have reached their target orbit, and are left to survive in whatever ride-share provided orbit they are deposited into. Due to their small mass and volume, it has been infeasible until very recently to put any form of on-board propulsion on these spacecraft without a significant sacrifice of the science objectives. Current research at the University of Illinois and other institutions will soon lead to the flight of CubeSat-class low thrust, low-power, yet low-specific impulse propellant systems. This technology should enable CubeSats to acquire new orbits from their launch positions, to rendezvous and dock together, or to reconfigure their constellations within orbit. Because of the atypical combination of low-thrust with high propellant mass consumption, a new toolset is needed to assist with planning of both single and cooperative multi-satellite missions. This work contributes a new framework for the calculation of high-fidelity trajectories in low Earth orbit. A shooting method is reformulated as a non-linear programming problem, and wrapped by a novel mesh refinement algorithm, which updates the time discretization based on a cumulative thrust density function. The states are propagated using a higher order explicit Dormand-Prince integrator with an error-adaptive step size. The necessary derivatives and Jacobian are developed in real time using algorithmic differentiation, which allows for significantly higher accuracy over traditional finite difference methods. This framework is tested against analytical methods developed by Wiesel for in-plane, and Edelbaum and Kechichian for out of plane, and is shown to match or surpass their results. This work further contributes to the field by developing an extended framework that allows the simultaneous integration of multiple satellites using parallel processing on a super computer, and lays out the necessary constraints to define cooperative intercept, rendezvous and orbit reconfiguration problems. Finally, this dissertation develops a new approach to parallel algorithmic differentiation, allowing the concurrent calculation of multiple derivatives in a user-transparent manner, simultaneously while propagating multiple satellites. This is accomplished by using a Cartesian processor grid and a new parallel communication scheme to maintain the most data locality per processor, enabling orders of magnitude speedup by comparison to both serial, and the previously developed parallel processing approach. This new parallel algorithmic differentiation technique is demonstrated with a series of test cases, developing cooperative maneuvers for from two to four satellites experiencing non-linear orbit perturbations. Rendezvous from different altitudes, and from different phases of the same orbit are demonstrated, as well as constellation reconfigurations. Finally, a four satellite cooperative maneuver demonstrates the practical application of distributing satellites into a target constellation from the same launch vehicle.

Name thesis:degree_name

Ph.D.

Level thesis:degree_level

Dissertation

Discipline thesis:degree_discipline

Aerospace Engineering

Grantor

University of Illinois at Urbana-Champaign

Year dc:date

2013

Author dc:creator

• Ghosh, Alexander

Contributors dc:contributor

• Coverstone, Victoria L.
• Conway, Bruce A.
• Burton, Rodney L.
• Swenson, Gary R.

• satellite
• trajectory
• trajectory optimization
• optimization
• cubesat
• picosatellite
• earth orbit
• Orbit
• low thrust
• low-thrust
• automatic differentiation
• algorithmic differentiation
• parallel processing

Statement dc:rights

• Copyright 2013 Alexander Ghosh

Language dc:language

en

Handle dc:identifier

http://hdl.handle.net/2142/45554

OAI identifier oai:identifier

oai:www.ideals.illinois.edu:2142/45554