Cost Variance Analysis in Constellation Architecture Studies

The NOAA Satellite Observing System Architecture (NSOSA) study has examined nearly 100 alternative weather satellite constellations, looking for the most cost effective arrangement of instruments, satellite orbits, and sustainment policies. Examining the cost-effectiveness of potential future satellite constellations requires quantification of each constellation’s long-term sustainment costs as well as short-term funding “spikes” required by production or launch. Cost estimates are well known to be uncertain, and the uncertainty in cost must be accommodated in a constellation architecture study. The paper discusses methods for cost uncertainty analysis developed as part of the NSOSA study. The methods start with probabilistic estimates at the satellite configuration level using the Formal Risk Assessment of System Cost Estimates (FRISK) model. The FRISK method component maturity and historical cost models to predict a cost probability distribution for each unique satellite design and production effort. These methods are then extended to combinations of configurations forming constellations. An important complexity is the strong correlation of cost uncertainty between constellations with partial configuration sharing. We develop a full model of correlated cost distributions that can be assessed across the nearly 100 NSOSA alternative configurations. The model allows both absolute and relative cost uncertainty comparisons. Absolute comparisons assess the likelihood of given alternatives breaking budgetary or lifecycle cost constraints. Relative comparisons assess the likelihood one alternative is more costly than another, with proper accounting for the likely correlation between alternatives.