Co-optimizing spacecraft component selection, design, and operation with MINLP

The spacecraft early conceptual design effort often involves picking components from a catalog or database, combined with discretized decisions concerning operations, like the choice of ground station, type of orbit (LEO or MEO, syn-synchronous or polar), and, in the case of constellations, the number of planes and satellites per plane. One common strategy is to exhaustively enumerate all combinations, filter out unfeasible ones, and plot all resulting designs in the objective space where a Pareto frontier might be found between competing performance metrics. However, as the number of combinations grows exponentially in the number of options, this strategy quickly becomes limited. Based on the optimization field of mixed-integer nonlinear programming, a new method has shown promise in overcoming the challenges resulting from a large number of options. This paper enhances the previous method describing how to incorporate operational aspects of design: choice of ground station, the timing of communication downlinks, and modeling eclipse time windows. We apply the method to a simple Earth observation case study and evaluate the new optimization problems' computational performance as a function of time discretization level.