Masafumi Isaji
(Advisor: Prof. Koki Ho)
will defend a doctoral thesis entitled,
Space Logistics and Mission Planning Optimization via Nonconvex Mixed-Integer Nonlinear Programming
On
Thursday, July 17th at 12 p.m. 
CODA 1215 Midtown
Abstract
As we strive to establish a long-term or permanent presence in space, it is imperative to develop an optimization framework that enables space logistics and mission planning with high fidelity while maintaining tractable computational costs. Realistic space logistics problems often involve (i) integer variables to represent discrete mission decisions and (ii) nonlinear and nonconvex relationships to capture the complexity of space systems design or spacecraft dynamics. However, most existing methods and frameworks rely on mixed-integer linear programming (MILP), which cannot fully represent these nonlinear and nonconvex characteristics. To address both aspects simultaneously, this thesis formulates and solves space logistics and mission planning problems as nonconvex mixed-integer nonlinear programs (MINLPs).
The first application considers integrated space logistics, which concurrently optimizes transportation scheduling and spacecraft sizing, where the sizing component involves nonconvex and implicit functions. Since this problem is incompatible with standard MINLP solvers, a decomposition framework is developed and applied to efficiently identify high-quality feasible solutions. In the second application, the framework is then extended to enable global optimization of MINLPs involving concave functions, a common structure in space systems. This capability is demonstrated through a case study on simultaneous optimization of mission operations and systems design for crewed Mars missions. Another application considers integrated space logistics with spacecraft staging optimization. To accommodate the increased problem complexity, the decomposition method is modified to handle multiple classes of nonconvexity, enhancing modeling capability at the cost of higher computational effort. Overall, this thesis introduces nonconvex MINLP formulations and corresponding solution approaches to space logistics and mission planning, establishing a new direction for addressing problems with greater complexity and fidelity.
Committee
•    Prof. Koki Ho – School of Aerospace Engineering (advisor)
•    Prof. Brian Gunter – School of Aerospace Engineering
•    Prof. Glenn Lightsey – School of Mechanical Engineering
•    Prof. Nick Sahinidis – School of Industrial and Systems Engineering, School of Chemical and Biomolecular Engineering
•    Dr. Stephen Edwards – Advanced Concepts Office, NASA Marshall Space Flight Center