Dynamic risk management of assets susceptible to pitting corrosion

Abstract

This paper presents a methodology to assess and dynamically update the risk of process components affected by pitting corrosion. The proposed framework considers the time-dependent growth of pits and uses the non-homogenous Markov process to model the maximum pit depth. The developed pit depth model is incorporated into a limit state function to estimate the failure probability of affected components. Economic consequences are estimated considering both business and accidental losses due to failure. The estimated risk is updated using Bayesian analysis as new inspection data become available. Different risk management strategies including prevention, control and mitigation measures are also studied.

An assessment of causes and failure likelihood of cross-country pipelines under uncertainty using bayesian networks

“…Research on corrosion prediction of the pipeline has been reported mostly using numerical methods such as Monte Carlo simulation. However, BN has also been employed to predict pipeline corrosion with promising results [31][32][33][34]. Abubakirov [31], for instance, used Dynamic Bayesian Network (DBN) to estimate both internal and external corrosion damage and assess the probability of failure to support inspection intervals optimisation.…”

Dynamic risk analysis of flammable liquid road tanker based on fuzzy Bayesian network

“…The core of dynamic risk analysis is to use existing data to update the probability of occurrence of specific event scenarios. The dynamic risk analysis of the processing system is a method to update the risk of the operating system according to the operation of the control system, safety barrier, human, and other factors 5 …”

“…The dynamic risk analysis of the processing system is a method to update the risk of the operating system according to the operation of the control system, safety barrier, human, and other factors. 5 For dynamic risk assessment, there are two widely used evaluation and analysis methods: one is based on the Bayesian network (BN) method, which uses the basic Bayesian principle to update the failure or accident probability in the form of likelihood function by inputting new observations or evidence in real time. There are many research results on dynamic risk analysis and condition assessment.…”