Crack or crack-like anomaly is one of the major threats to the safety and structural integrity of oil and gas pipelines. Various assessment models have been developed and used within pipeline industry to predict the burst capacity for pipelines containing longitudinally-oriented surface cracks. These models have different level of conservatism, accuracy, and precision which significantly impacts pipeline operators’ integrity mitigation decisions such as pressure restriction, excavation, and repair, and also lead to different level of safety.
This paper compares the accuracy and precision of the most commonly used crack assessment models, i.e. Modified Ln-Sec, CorLAS, API 579 Level 2 and the recent-published PRCI MAT-8 model using in-service and hydrostatic testing failure data. A total number of 12 in-service and 63 hydrostatic test pipe ruptures due to stress corrosion cracking (SCC) with actual burst pressure, material property, and detailed crack size measurements are collected, and used to derive the probabilistic characteristics of the model errors associated with each model. Compared to the burst tests conducted in the laboratory and investigated in other previous studies, the results obtained from in-service and hydrostatic test ruptures are more representative of the real boundary conditions in pipeline operation. All the assumptions and empirical correlations associated with each model are discussed in details. The analysis result suggests that CorLAS is the most accurate model with the least uncertainty (or highest precision). Mitigation activities can be optimized without compromising safety by using the most accurate and precise model.