Rapid Design of Space System Architectures Taking Uncertainty Into Account

Daniel E. Hastings, Massachusetts Institute of Technology, Professor of Aeronautics & Astronautics

Space system architectures traditionally define "rocket science". However, while most often these architectures, once constructed, deliver excellent performance, it is extremely rare that they deliver the initially promised performance on the initially proposed cost and schedule. There are several reasons for this but a large amount of the blame lies in the fact that there is a little understanding of how to incorporate uncertainty into the design process in a way that allows decision makers to systematically take uncertainty into account.

The development of space systems is subject to not only cost, technical and market uncertainties, but also to uncertainties from the policy domain. This talk introduces an approach to quantify and compare space system architectures under uncertainty, with emphasis on policy uncertainty as well as technical uncertainty. Two key hypotheses are developed and explored. The first is that the cost of policy can be quantified through technical analysis of space system architectures under varying policies and the second is that uncertainties of space system architectures can be quantified and managed effectively by carrying portfolios of architectures, rather than any single architecture.

Enabling this approach is simulation modeling of space system architectures that are of interest in the design trade space. Using computer models to quantify the performance and cost characteristics, tradespaces of architectural characteristics can be developed and analyzed. These tradespaces are then explored to find the Pareto optimal fronts in the space based on some criteria (e.g. minimum cost, maximum function per cost etc.)The simulation also allows for the propagation of uncertainty in various architectural characteristics and an understanding of how those uncertainties propagate to system evaluation criteria.

The first step in the proposed approach is an analysis of the trade space of potential architectures. This analysis includes bounding the problem in terms of the architectural concepts that will be evaluated and the bounding of the policy uncertainties and scenarios that will be investigated in addition to the other uncertainties that have significant impact on architectural evaluation. The second step is to adjust the models to reflect the effects of these uncertainties of interest on the simulation. The third step is to quantify the impact of the uncertainties on the system evaluation criteria for each architecture of interest. Finally, we incorporate portfolio theory as an approach to manage uncertainty effectively.

To illustrate the approach in practice, we use several case studies. We examine the effect of various types of uncertainty on a military space based radar mission. We compare and contrast the choices of architectures on the basis of performance and on the basis of minimizing uncertainty. Finally we also consider a commercial case of a broadband space architecture and consider the choice of architectural portfolios when uncertainty is minimized versus other choices that may be made.