Launch and Production Schedule Modeling for Sustained Earth Observation Constellations

Satellite missions are typically thought of singly or in the small blocks corresponding to a funded program. For example, in weather and environmental remote sensing the US geostationary program yet to launch comprises four satellites (GOES-R through GOES-U) and five polar LEO (SNPP launched and four JPSS yet to launch). However, in many mission areas the need is to provide a continuous and sustained on-orbit capability far beyond the borders of a single mission or even a single program. In weather and environmental remote sensing we want decades long data records and there is a strong disutility to any gap in service. This presents both a problem and opportunity when designing constellations. In designing and analyzing launch and production policies we should consider not a single satellite mission or a single funded program but a whole sequence of production and launch programs (most of which are in the unplanned future). We might discover in considering the far “out-years” that decisions made today may have complex consequences in the future, and policies that seem logical in the context of a single program may not be effective when considered over the lifetime of several programs.

This paper will discuss Aerospace Corporation efforts to build integrated satellite production and launch models, and associated analytical approaches, that apply to long-term sustainment of weather and environmental satellite constellations. In building the models and techniques we have had to reconcile disparate and complex issues.

1. Funding authorities expect to see deterministic plans (e.g.; flyout charts) but the actual lifetime behavior of satellites and launches is stochastic. Some of the most effective launch and production policies are event-driven and adaptive and thus impossible to precisely schedule long in advance.
2. Satellite lifetimes are random, and we know that our standard engineering estimates of lifetime probability distributions are usually wrong, although we have an empirical understanding of the plausible uncertainty space. We desire launch and production policies that are robust both against the uncertainty in satellite lifetime and against our uncertainty in the lifetime distribution.
3. Sponsors care both about avoiding any service gaps and efficiency of satellite utilization. The optimum point between these objectives is not well-defined, and in any case we desire policies that systematically improve both, or a demonstration that joint improvement is impossible.
4. We do not, and realistically cannot, produce satellites on-demand, we produce in blocks. We need feedback mechanisms (robust to the long lags inherent in satellite development programs) to signal when new blocks should begin what constraints exist on their length.

The paper will demonstrate graphical and simulation methods for generating and analyzing a variety of deterministic and event-driven launch and production policies.