Risk assessment on deepwater drilling well control based on dynamic Bayesian network

Abstract

Deepwater drilling involves complex operations and equipment, so it is faced with various operational challenges including well control accidents. This paper proposes a dynamic risk assessment model for evaluating the safety of deepwater drilling operations. The dynamic risk assessment process includes three key steps: constructing fault tree models to analyze risk factors leading to a blowout accident, developing dynamic Bayesian network model based on the constructed fault trees, and performing dynamic risk analysis to evaluate the safety of well control operation. The proposed model includes risk factors about kick cause, kick detection, shut-in operation and kill operation, which covers the full process of a blowout. The proposed model could analyze the risk of blowout more comprehensively and the influencing degree of these four phases could also be clarified. Besides, the modular modelling method could update the structure and parameters of the developed model easily if new factors or data are added. The results show kick cause has the greatest impact on blowout accidents, followed by shut-in operation, kill operation and kick detection. Mutual information analysis and uncertainty analysis is performed to investigate the effects of risk factors on blowout. Finally, some corresponding preventive measures for blowouts are proposed.