Dynamic risk assessment of deep-water dual gradient drilling with SMD system using an uncertain DBN-based comprehensive method

The subsea mudlift drilling system (SMD) belongs to the system of Dual Gradient drilling. When the natural gas hydrate layer is encountered with the SMD system during the drilling process, significant risks and failures are resulted by the decomposition and secondary formation of the hydrate. To predict failures, the environmental factors, human factors, and equipment factors are analyzed in this study. Firstly, after the fault tree is established, it is transformed into Bayesian Network (BN) using the mapping algorithm. Secondly, the Leaky Noisy-OR node is added to BN and the uncertain influence of the logical relationship is considered. Then, the established BN is transformed into the uncertainty Dynamic Bayesian Networks (UDBNs) through the transition probability matrix, if the dynamic uncertainty of equipment factors and human factors are considered. In addition, the cognitive reliability and error analysis method (CREAM) are used to determine the prior probability of human factors in the DBN model. Moreover, we also use fuzzy theory and expert judgment to quantify the prior probability of equipment failure. At the end of the experiment, the ultimate result shows that the UDBNs model, derived by the existing data, can be used to predict the risk of lost circulation (LC), sticking, and blockage during the drilling process and the dynamic success probability at different stages of the shut-in process. The correctness of the established UDBNs model is verified by the Petri nets method.