Traditional remaining strength assessment methods such as B31G [1] and API RP-579 [2] are inherently conservative to account for uncertainties and thus assure safety. With the advent of reliability based assessment methods, built-in conservatisms at the model level are undesirable. Instead, the statistical model descriptions are used to account for the model uncertainties. To date, several research studies have compared the accuracy of these assessment methods as burst pressure predictors with newer limit-state models [3], In these comparisons, the assessment methods such as B31G do not perform well as burst predictors because their original formulation was used (i.e., safety factor was not removed), leading some to reject B31G for use in a reliability based assessment. In this paper the effectiveness of the burst capacity prediction of these assessment models are compared by reformulating these assessment methods into appropriate limit-state models such that they are compared on the same basis. The comparison shows that the B31G based limit state model is one of the most accurate burst capacity predictors, well suited for use in a reliability based framework.
展开▼
