Optimization of a space transportation system including design sensitivities

A new method for the performance rating of a space transportation system is presented. The method combines a commonly used quality criterion, the so-c alled growth factor, with a penalty function that penalizes high sensitivit ies with respect to inaccuracies in the employed physical computation model s. The method is applied to a space transportation system with horizontal t akeoff and landing, an air-breathing first stage, and a rocket-propelled se cond stage. Using a multimembered evolution strategy, an optimization of th e lower stage propulsion system is performed with and without considering d esign sensitivities. The results show that very critical situations can ari se if the sensitivity to inaccuracies in the physical models is neglected. Furthermore, remarkable reductions of the design sensitivity by more than a n order of magnitude are achievable with only moderate losses in the total system performance.