Dynamic risk assessment of subsea pipelines leak using precursor data

Abstract

Accidental leak in subsea pipelines poses a severe threat to human life, environment, assets and corporate reputation. Quantitative risk assessment of such events is consistently a challenging task due to data scarcity. In this paper, a new methodology comprising of Hierarchical Bayesian analysis (HBA) implemented with the Bayesian network (BN) is proposed to assess the risk of subsea pipelines leak. This methodology capture both data and model uncertainty. HBA is utilized to handle the uncertainty among failure data from different sources; BN is used to capture the dependencies among primary events and safety barriers. Failure probabilities from HBA approach are used as prior beliefs, and the occurrence probabilities of different scenarios are derived from BN reasoning in uncertain conditions. Furthermore, the proposed methodology makes use of a concept of fuzzy loss ratio to evaluate the consequence of different scenarios. Integrating probability with consequence gives a reasonable risk assessment result. As new observation data becomes available, the assessed risk can be updated to generate a dynamic risk profile. An industrial case study demonstrates the application of the methodology. This methodology can serve as a helpful tool to risk management of subsea pipelines.