Maintenance and integrity management of hydrocarbons pipelines face the challenges from uncertainties in the data available. This paper demonstrates a way for pipeline remaining service life prediction that integrates structural reliability analysis, accumulated corrosion knowledge, and inspection data on a sound mathematical foundation. Pipeline defects depth grows with time according to an empirical corrosion power law, and this is checked for leakage and rupture probability. The pipeline operating pressure is checked with the degraded failure pressure given by ASME B31G code for rupture likelihood. As corrosion process evolves with time, Dynamic Bayesian Network (DBN) is employed to model the stochastic corrosion deterioration process. From the results obtained, the proposed DBN model for pipeline reliability is advanced compared with other traditional structural reliability method whereby the updating ability brings in more accurate prediction results of structural reliability. The comparisons show that the DBN model can achieve a realistic result similar to the conventional method, Monte Carlo Simulation with very minor discrepancy.
Dynamic reliability analysis framework using fault tree and Dynamic Bayesian Network: A case study of NPP
