An exploration of PRA methodology used in spacecraft design

Abstract

In order to achieve an optimal design of a complex space system that meets all constraints, the requirements placed upon the performance, mass, cost, and risk of the system must be considered, understood and traded against each other during the conceptual design of the system to avoid costly redesigns or project cancellation later in the development process [1]. A design process that follows this tenet of riskinformed design will need detailed insight into the relative risks facing the system, as well as quantitative estimates that can be produced through probabilistic risk assessment (PRA), in order to evaluate design decisions based upon the impact to all requirements on a co-equal basis [2].In this study, four types of methodologies used to produce risk estimates for spacecraft and satellites are examined. These include two traditional PRA methodologies [3,4], an innovative approach [5], and a top-down approach [6], all of which are explored by using the propulsion subsystem of the Lunar Reconnaissance Orbiter (LRO) as a comparative basis for the methodologies considered [7]. Similarities, differences, benefits, and drawbacks of various bottom-up, componentbased PRA approaches and the top-down approach are elucidated in terms of the process of modeling a system, the actionable information produced for the design team, and the overall quantitative risk evaluation of the system as compared to similar heritage space systems.Results of the various PRA methodologies are examined at the level of component failure rates, single-component failure probabilities, single-function failure probabilities where redundancy exists in the design, as well as the subsystem failure probability for the nominal LRO mission.Ultimately, all of the bottom-up, component-based PRA methods capture only the risk of a mature system and miss the risk contribution of design defects, which have been shown to be key drivers of reliability in single-use developmental systems [8,9]. Therefore, further steps must be taken to incorporate this contribution in future PRA methodologies.