The evaluation and development of the current corrosion defect assessment procedures for pipelines has been based on experimental burst tests of line pipe. In these tests, external corrosion has often been simulated with machined defects of simple geometry. As a result, assessment procedures which model the corrosion defect geometry with only a few parameters, such as ASME B31G, show reasonable agreement with the experiments. However, the degree of conservatism in these assessment methods is undefined when they are applied to complex corrosion defects. The authors have burst over 40 pipes removed from service due to corrosion defects. All corrosion defects on each pipe were measured in detail and the material properties were determined from tensile tests. The currently accepted assessment procedures for corroded line pipe (B31G and RSTRENG) have been applied to the database. The degree of conservatism in these procedures is quantified and a statistical model for the failure predictions is proposed.

This content is only available via PDF.