Microbiologically influenced corrosion (MIC) management using Bayesian inference

Microbiologically influenced corrosion (MIC) is a complex phenomenon that occurs when a microbial community is involved in the degradation of an asset (e.g. pipelines). It is widely recognized as a significant cause of hazardous hydrocarbon release and subsequently, fires, explosions, and economic and environmental impacts. This paper presents a new MIC management methodology. The proposed methodology assists in accurately monitoring MIC activity and accordingly develop strategies to manage it. The MIC monitoring and management activities are achieved using Continuous Bayesian Network (CBN) technique with Hierarchical Bayesian Analysis (HBA). The integration of HBA and CBN helps overcome the Bayesian network's discrete value limitations (BN) and source-to-source uncertainty for each node in the network. The methodology can provide the precise value of parameters, such as failure probability and MIC occurrence rate which are verified using observed data. The application of the methodology is demonstrated on a subsea pipeline. The study provides a better understanding of the influencing factors of MIC rate and failure probability. This assists in developing effective MIC management strategies.