Simplified Bayesian Network Model for Stress Corrosion Cracking

Paper ID: C2025-00126 ABSTRACT: Establishing trust in failure models requires a robust validation process. This is difficult because large amounts of field data are needed. Relying on a few field tests is insufficient, as a comprehensive validation demands a thorough comparison between corrosion model predictions and real-world observations. The present study by the California Energy Commission presents the validation of near neural and high pH stress corrosion cracking (SCC) model. The validation was carried out by comparing the external corrosion rate model results with field data. Good agreement was found between modelled corrosion rates and ILI run comparison data.
This study has two main benefits to pipeline operators. First, the validation helped identify areas where the SCC model may need to be adjusted to provide more accurate predictions for specific pipeline locations and operating conditions. Second, a simple set composed of 27 external SCC crack growth rates rate distributions (as a function of pipeline exposure, mitigation, and resistance) have been generated by the SCC model and validated with the ILI data. This publication provides 27 reliable and simple SCC rate estimates that can be utilized by pipeline operators without the need to use a complex SCC model or use specialized Bayesian network software.