Development of Partial Safety Factors for Fitness-For-Service Assessment of Pressure Vessels Using First Order Reliability Method

This paper addresses the reliability based acceptance criteria for evaluating fitness-for-service (FFS) of reactor pressure vessels of Nuclear Power Plant, for which the cracks are reported during in-service inspection (ISI). The FFS for cracked vessels is ensured by applying the principles of fracture mechanics with crack geometry, material fracture properties and applied loading being the decision making variables. The ISI of reactor pressure vessel is performed under challenging reactor environment and thus significant uncertainty is associated with the reported crack geometry. The fracture toughness of the vessel is strongly dependent on the irradiation levels and duration seen by the vessel material and is typically estimated using empirical models and also by using tests on surveillance coupons. This toughness value cannot be established with certainty under this scenario. The FFS is performed for loads including internal pressure and thermal loads. The current procedure in traditional codes like ASME Section XI, employs a deterministic criteria in which the safety is preserved against these uncertainties by employing a single safety factor which is applied on the fracture toughness value. The application of a single factor of safety when there are multiple uncertain variables, results in different reliability levels for different combination of variables. Use of partial safety factor based approach to arrive at consistent reliability levels in such situations is well established. This paper presents the derivation of the partial safety factors for the limit state defined in ASME Section XI, for the FFS of reactor pressure vessel. The deficiency of the existing method is first presented followed by detailed methodology to derive the partial safety factors as the acceptance criteria. The partial safety factors are derived by applying the principles of structural reliability using the first order reliability method.