Risk comparison of crew launch vehicle concepts

NASA is embarking on a new era of human spaceflight, one in which commercial service providers will sell astronaut transportation services to NASA. Thus NASA will have limited insight into the design and manufacturing processes of space transportation vehicles. In this new paradigm, setting appropriate safety requirements and goals, for the service providers to meet, and a standard process for evaluating the safety of commercial rides, will be of paramount importance to ensuring the safety of the astronauts. Good systems engineering practice emphasizes the importance of having valid and verifiable requirements. In the case of safety, carrying an unrealistically high, and thus unverifiable requirement can actually reduce the safety of vehicles under development because it can lead to focus on quantification of known failure modes rather than on a search for the unknowns, focus on process rather than experience, lead to a false sense of security, and a tendency to game the analysis to meet the requirement. A probability of loss of crew (LOC) requirement cannot be strictly verified, as there will be too few flights for statistics and probability forecasts are only “opinions” of the forecaster, well-educated opinions hopefully, but opinions nonetheless. ("Not differently is the aim of logic on the other hand. This cannot tell me if my opinions are right or wrong, that is nonsense, but only if they are coherent or if there is among them an intrinsic inconsistency. And the calculus of probabilities is only the logic of practical convictions, that are subjected to a more or less large degree of doubt"[1]). But the reliability and safety that were actually achieved by previous vehicles, with reasonable expectations of growth, inform the range of LOC probabilities that can be credibly achieved by new systems. This paper shows how this process can inform the development of rational requirements with the example of launch vehicle safety.