Scheduling satellite launch missions: an MILP approach

The paper deals with an MILP model to schedule satellite launches with alternative launchers and different mission profiles, subject to resource constraints. The model is part of a simulation tool developed within a joint research project with the European space agency. The focus is on geostationary transfer orbit (GTO) launches of payloads, which are associated with a given time window for launch, a payload mass, and a potential revenue. A launch requires the payload, a launcher compatible with both payload mass and mission profile, a launch complex for that launcher, and a launch range (i.e., resources that are shared by the launch complexes, including a mission control station). We consider three launcher types, which differ in cost and performance, are produced at a limited rate, and cannot be stocked in large amounts. One of the launchers is also able to carry out dual launch missions, i.e., missions in which two payloads are launched together, provided that their joint mass does not exceed launcher's mass capacity and their time windows overlap. After each launch, the launch complexes and the launch range need some latency time to be reset. Two natural objectives are minimizing the number of lost payloads and maximizing profit. Here we report experiences with a discrete-time MILP model formulation, which is rather flexible and can be extended to cope with additional problem features. Natural concerns, such as computational effort and the effect of time discretization, are addressed in the paper.