Sensitivity of system reliability of corroding pipelines to modeling of stochastic growth of corrosion defects

The time-dependent system reliability of pressurized pipeline segments containing multiple active corrosion defects is evaluated to investigate the sensitivity of the system reliability to various modeling options regarding the corrosion growth. The gamma and inverse Gaussian processes are employed to model the defect growth, whereas the dependence among the growths of different defects is characterized using the Gaussian copula and sum-of-stochastic-process approach. The analysis results indicate that all else being equal the system reliability is insensitive to using the Gaussian copula or sum-of-stochastic-process approach to model the dependence among the growths of different defects. Furthermore, using the inverse Gaussian process to model the defect growth leads to slightly higher failure probabilities than using the gamma process. Finally, the results suggest that the use of one year as the time increment to simulate Gaussian copula-based dependent defect growths in the reliability analysis is adequate for the relatively slow corrosion growth that is typical for buried pipelines. (C) 2017 Elsevier Ltd. All rights reserved.