Probability of failure and risk assessment of propulsion structural components

SUMMARYBecause of the increasing need to account for the uncertalntles in mate-rial properties, 1oadlng conditions, geometry, etc., a methodology has beendeveloped to determine structural rellabllity and to assess the associatedrisk. The methodology consists of a probabilistlc structural analysls by aprobabilistic finite element computer code NESSUS (Numerlcal Evaluatlon of Sto-chastic Structures Under Stress) and a generic probablllstlc materlal propertymodel. The methodology Is versatile and is equally applicable to structuresoperating at high- and cryogenic-temperature environments. Results obtalneddemonstrate that the issues of structural rellability and risk can be formallyevaluated by using the methodology developed whlch is inclusive of the uncer-tainties in material properties, structural parameters, and loading conditions.The methodology is described in some detail with illustrative examples.INTRODUCTIONThe probabilistic structural analysis methods (PSAM) (ref. I) have beendeveloped to analyze the effects of fluctuatlng loads, varlable material prop-erties, uncertaintles in analytical models and geometry, etc., especially forhigh-performance structures such as space shuttle main engine (SSME) turbopumpblades. In the deterministic approach, uncertainties in the responses are notquantified, and the actual safety margin remains unknown. Risk is calculatedafter extensive servlce experience. However, probabilistic structural analy-sis provides a rational alternative method to quantify uncertainties in struc-tural performance and durabllity. NESSUS (Numerical Evaluation of StochasticStructures Under Stress) is a probabilistic structural analysis computer codedeveloped under the PSAM project which integrates finite element methods andreliability algorithms (refs. 2 and 3). This code is capable of predictingthe scatters of structural response variables (such as stress, dlsplacement,natural frequencies, and buckling loads) subjected to all kinds of uncertain-ties. These are subsequently compared with their probable failure modes toassess the risk of component fracture. Probable failure modes are defined fordifferent structures and thelr respective service environments. For example,failure events (such as stress greater than strength, displacement exceeds