PROBABILISTIC FRACTURE-MECHANICS ANALYSIS OF A NUCLEAR PRESSURE-VESSEL FOR ALLOCATION OF IN-SERVICE INSPECTION

To find whether there are significant differences in fracture probability between various regions in a reactor pressure vessel a limited probabilistic fracture mechanics (PFM) study was carried out. The loading was assumed to be deterministic, whereas some of the other quantities involved were assumed to be of random character. A simple, mainly analytical probabilistic model was developed. The critical event was taken as unstable crack growth without any ensuing crack arrest. All random information on the J(R) vs DELTAa behaviour was assumed to be contained in a single scalar parameter. Various distributions such as the Weibull and the log-normal distributions were assumed for this parameter to judge the sensitivity of results to various assumptions. The initial size of pre-service defects was assumed to follow the OCTAVIA distribution. Possible time-dependent crack growth such as fatigue or stress corrosion cracking was treated in a simplified manner assuming deterministic material properties. Various regions (and crack geometries) were considered and the fracture and leakage probabilities were calculated for a number of load cases. The fracture probabilities were strongly dependent on the assumption made for the fracture toughness distribution, but the order of the fracture probabilities of the regions seemed to be relatively unaffected by this. The leakage probabilities are in most cases much lower than the fracture probabilities, indicating that consequence considerations are not very important for non-destructive test (NDT) allocation purposes. It is also concluded that probabilistic methods of the present kind may be a useful tool for judging which locations should be the primary targets for NDT.