Reliability assessment of buried pipelines based on different corrosion rate models

Different corrosion rate (CR) distributions have been derived from various corrosion growth models and used to perform reliability analyses of underground pipelines. The CR distributions considered in this work included a single-value distribution, based on the NACE-recommended CR for buried pipelines; a CR distribution derived from the Linear Growth corrosion model; Time-dependent and Time-independent CR distributions derived from a soil corrosion model; and a CR distribution derived from a Markov chain corrosion model. A Monte Carlo reliability framework capable of incorporating these CR distributions has been developed and applied to both synthetic and field-gathered corrosion data. The use of synthetic data assisted in evaluating the performance of each CR model with a consideration of corrosion defects of different sizes and ages. The application of the reliability framework to repeated in-line inspection data revealed the importance of careful selection of the CR distribution for an accurate assessment of the reliability of the inspected pipelines. It was shown that the best CR distribution is one that considers the ages and sizes of the corrosion defects as well as the observed dependence of the corrosion defect depth on time. Among the CRs considered in this study, the Markov chain-derived distribution was found to best satisfy these requirements.