39 Space Shuttle External Tank Probabilistic Risk Assessment — Risk Logic Development and Quantification of the Welds (PSAM-0226)

Welds are one of the most critical components in aerospace manufacturing. From an operational and safety point of view the consistency and quality of the welds are critical when used to join propellant tanks and lines. The National Aeronautics and Space Administration (NASA) Space Shuttle External Tank (ET) is no exception. Without high quality welds, the tanks could collapse under the weight of the cryogenic propellant, internal cryogenic thermal extreme, and flight loads — leading to catastrophic results. This paper presents a unique application of probabilistic failure analysis to welds based on a stress versus strength model that incorporates non-destructive examination (NDE) data. The theoretical basis for conducting the weld analysis presented herein was established in the previously written paper, Weld Analysis Methods for Aerospace Systems [1]; however, no quantification was presented. This paper includes quantified examples and adds the case for Lognormal Load-Normal Strength probability distributions to the Normal Load-Normal Strength probability distributions considered in this paper and in Weld Analysis Methods for Aerospace Systems.

Weld failure can be initiated by: loads exceeding the designed strength, mission-initiated flaws, or pre-existing flaws. Mission-initiated flaws are those induced during the mission under normal or outlier loads and assumed to occur randomly. Pre-existing flaws are weld defects due to pre-existing material anomalies or process errors that are missed by inspection. The frequency of occurrence for the pre-existing flaws was estimated using a weld process event tree. The weld process event tree was developed based on the series of inspection and testing steps that are performed on all ET critical welds in the quality control process (e.g., dye penetrant, X-ray, proof-test, and post proof-test). The event tree defined the probability of occurrence of three potential fault outcomes for each weld — good, repaired, and flawed. The probability of failure for each of the three weld conditions was calculated using stress versus strength analysis, which was based on weld properties derived from material databases. The strength flight allowable property was adjusted using expert elicitation to account for the effect of process and mission-initiated flaws. The flight failure probability for each weld case was calculated based on the probability of occurrence and failure for each case. The three cases were then summed to derive the total probability of weld failure in flight.

The stress versus strength analysis uses established methods that have been adapted for weld analysis. The coupling of the weld fabrication process event tree with stress versus strength analysis and expert elicitation is believed to be unique and offers the most meaningful numerical estimate of the Space Shuttle ET weld risks to date. The approach can also be applied to parent metals in tank structures and other components, such as supports and propellant lines. The stress versus strength model was incorporated into the Space Shuttle Probabilistic Risk Assessment (SPRA), which includes the undesirable impacts of weld failure. The results may also be used for a comparative analysis of weld process improvement candidates.