Burst Database Verification Study for Corroded Line-Pipe

A significant amount of effort over the past 20 years has been directed towards the experimental testing of corroded linepipe subjected to static internal pressure. This effort has led to the development of a number of empirical limit state models (e.g., ANSI/ASME B31G). Various studies have recently been completed to determine which limit state model is best suited for use in closed-form reliability evaluations using published burst databases compiled by the Pipeline Research Council International (PRCI) and Det Norske Veritas (DNV). Unfortunately, the results of these studies are inconsistent with little to no explanation why. This paper in-part accounts for the skewed results of previous studies by cleansing the PRCI and DNV burst databases of corroded line-pipe tests of flawed information and thus assembling the verifiable 2003 University of California, Berkeley (UCB) Burst Database. The errors found in these databases are statistically significant enough that extreme caution should be used when applying them in a reliability design, assessment, or management framework. Lastly, the relative performance of seven limit state models were analyzed with the 2003 UCB Burst Database and conclude that although a unbiased variation of ASME B31G is competitive, RAM PIPE I is the best overall limit state model of the seven evaluated.