Extended fault modeling used in the space shuttle PRA

Abstract

A probabilistic risk assessment (PRA) has been completed for the space shuttle with NASA sponsorship and involvement. This current space shuttle PRA is an advancement over past PRAs conducted for the space shuttle in the technical approaches utilized and in the direct involvement of the NASA centers and prime contractors. One of the technical advancements is the extended fault modeling techniques used. A significant portion of the data collected by NASA for the space shuttle consists of faults, which are not yet failures but have the potential of becoming failures if not corrected. This fault data consists of leaks, cracks, material anomalies, and debonding faults. Detailed, quantitative fault models were developed for the space shuttle PRA which involved assessing the severity of the fault, detection effectiveness, recurrence control effectiveness, and mission-initiation potential. Each of these attributes was transformed into a quantitative weight to provide a systematic estimate of the probability of the fault becoming a failure in a mission. Using the methodology developed, mission failure probabilities were estimated from collected fault data. The methodology is an application of counter-factual theory and defect modeling which produces consistent estimates of failure rates from fault rates. Software was developed to analyze all the relevant fault data collected for given types of faults in given systems. The software allowed the PRA to be linked to NASA's fault databases. This also allows the PRA to be updated as new fault data is collected. This fault modeling and its implementation with FRAS was an important part of the space shuttle PRA.