Predicting Fracture Arrest Based on a Relationship Between Charpy Vee-Notch Toughness and Dynamic Crack-Propagation Resistance

摘要

Dynamic fracture in a gas-transmission pipeline has consequences that require pipelines be designed to virtually preclude its occurrence. Because the related phenomenology is complex, technology to help avoid such incident has been based on full-scale experiments, which have been done since the early 1970s, when the possibility of a dynamic ductile fracture was initially recognized. Empirical models were developed as a means to represent these experiments in a design or analysis setting. The Charpy vee-notch (CVN) fracture energy was found to adequately characterize dynamic fracture resistance for these early experiments. However, as interest shifted to larger diameter, higher pressure, higher BTU 'rich' gases, which require very high toughness steels to ensure fracture arrest, the predicted arrest toughness based on such models and CVN energy was found to be nonconservative. Recently proposed alternatives to the early CVN-based approaches, like that using crack-tip opening angle (CTOA) as a measure of fracture resistance, offer promise to avoid prediction errors when dealing with high-toughness steels. However, the experimental aspects are difficult to use on a production basis in steel and pipe mills, and the results of the analysis are case-specific in utility and complex in formulation, requiring numerical solution. For these reasons, a more fundamental explanation for why CVN-based methods have experienced problems was sought and implemented as part of the development of the fracture control plan for the Alliance Pipeline. This paper builds on that explanation, which postulates CVN-based models failed to correctly predict arrest-toughness in applications involving high-toughness steels because of significant differences in their fracture behavior as compared to that of the much lower toughness steels used in their empirical calibration. An overview of this explanation and support for it are introduced. The related recently proposed approach to correct CVN-based models is then briefly reviewed. Results from ongoing experiments done to augment the earlier experiments are presented in complement to prior results. The correction is then used in conjunction with Battelle's arrest-toughness model and indicated to accurately predict arrest toughness for the full-scale test database, including high pressure testing involving rich gases and very high toughness steels beyond that considered previously.