As part of a safety case for a subsea 13Cr pipeline, the operator wished to demonstrate that if a circumferential through wall crack developed, the crack would remain stable as a leak rather than growing to a full bore rupture. An initial fracture mechanics analysis had suggested that the margins on crack length were too small to make such a “leak before break” argument. This paper reports an integrated programme of small scale testing, numerical modelling and full scale testing which showed that a leak before break case could be made.
13Cr martensitic steel generally shows excellent toughness at the service temperature, as does the super duplex weld metal that was used for the girth welds. However, as the pipeline had been installed by reeling, there was some concern that the toughness may have been reduced. Hence a programme of fracture toughness testing was designed to generate tearing resistance curves for both as-received and pre-strained parent material and weld metal. Deep and shallow through thickness notched specimen geometries were tested to explore the effect of constraint on the toughness. Finite element analysis was used to predict the stress intensity for a range of crack lengths, including the effects of misalignment. Non-linear analyses were used to estimate the limit load for the cracked pipe. The test results were used as input to tearing analyses to Level 3 of BS 7910. These showed that the tolerable length of a through wall crack exceeded the length of anticipated defects by a factor of at least two.
To confirm the fracture mechanics predictions, two full scale tests were carried out. These used pressure cycling to grow a through wall crack by fatigue. These cracks were stable under an internal pressure equal to the pipeline design pressure. The cracked specimens were then axially loaded to failure. Extensive tearing occurred before final failure at loads above those predicted by the fracture analysis, confirming the conservatism of the predictions.