Uncertainty and Leontief Systems: An Application to the Selection of Space Station System Designs

This paper deals with the problem of designing a critical component of a highly interactive technologically advanced system, one in which there is uncertainty as to technological and cost parameters. The particular example discussed in the paper is the choice of an energy module for Space Station, where two alternatives, PV and SD, were under consideration. If the uncertainty associated with technological and cost parameters were ignored, and a choice made solely on the basis of "best guesses" of NASA engineers as to parameter values, then SD, which has been implemented only in small scale laboratory applications, would be a clearcut "design to cost" winner. The situation is different when the "best guess" estimates are replaced by subjective probability distributions over technological and cost parameters. The paper outlines a "safety first" decision making model which suggests that the more reliable PV, which has been used on all previous space missions, becomes the preferred choice for highly risk averse decision makers.