Large-Scale MDO of a Small Satellite using a Novel Framework for the Solution of Coupled Systems and their Derivatives

Gradient-based multidisciplinary optimization is applied to a small satellite design problem. A novel MDO framework is developed by adopting a non-conventional definition of components which yields two key benefits for solving multidisciplinary analysis and optimization problems. The first is a unified solver which generalizes many existing methods for solving nonlinear and linear systems, and the second is an automated and efficient method for computing coupled derivatives. This framework is used to solve a small satellite optimization problem involving several disciplines including orbit dynamics, attitude dynamics, attitude control, temperature, solar power, battery, and communication. Multi-point optimizations involving over 35,000 design variables and 1.2 million unknowns require on the order of 5 hours to converge to significantly improved designs, and this efficient design optimization capability is used to compare potential launch options.