196 Mission Reliability Evaluation for a Space Propulsion System Phased-Mission Benchmark Problem (PSAM-0425) Available to Purchase

This paper summarizes results of the authors' participation in an exercise to apply various methods to evaluation of the mission reliability of a simplified propulsion system design proposed by the session organizers. The implicit point of the exercise is to promote discussion of strengths and weaknesses of the various methods that participants choose to employ for this particular problem type.

Key characteristics of the problem include the following:

• multiple phases, in which mission success requirements vary considerably;

• significant redundancy;

• long operating times;

• a mixture of failure modes, including passive failures, failures to run, failures to change state, and demand failures to start or shut down;

• low overall reliability, so that failure events are not “rare” in the sense of the rare-event approximation;

• consideration of common cause failure (CCF); and

• serial actuation of units, such that challenges to a given unit depend on the behavior of lower-numbered units.

Simulation techniques have been used for many years to address complex reliability-availability problems of this general type. TIGER is one of many programs using simulation to address such problems; it is used by the US Navy to evaluate reliability /availability / maintainability (RMA) metrics for complex entities such as ships, over complex design reference missions within which the operational needs and mission success requirements vary from one phase to another. Using programs like TIGER, it is possible to treat problems of this kind involving very large numbers of components having different time-dependent reliability characteristics.

The authors have analyzed this problem using simulation approaches, including TIGER. Some consideration was also given to a cut-set-based, sum-of-disjoint-products approach previously proposed for phased-mission analysis, but it turns out that this problem is not amenable to that approach. However, the problem is found to be quite amenable to simulation-based approaches.