Risk assessment of disconnect operations of MNPP using similarity aggregation method based fuzzy dynamic Bayesian networks model

Abstract

Marine Nuclear Power Platforms (MNPP) have the potential to provide a safe and efficient energy supply for offshore oil extraction and remote islands. During the service, MNPP is subject to disconnect operations and return to the dock for repair and maintenance. However, the failure of disconnect operations of MNPP can have catastrophic consequences. It is essential to carry out an accurate risk assessment of the disconnect operation process. In this study, a comprehensive dynamic quantitative dynamic risk assessment method is proposed to study the whole process of MNPP disconnect operations from preparation to towing. Firstly, a similarity aggregation method (SAM) is used to solve the expert fuzzy opinion aggregation problem, and then the probability of failure of the event consistent with the group consensus opinion is obtained. Secondly, a dynamic Bayesian networks (DBNs) model is developed for three stages, including YOKE offloading operation, YOKE dropping operation, and equipment recovery operation, and then predicts the dynamic probability of failure during the disconnect operations. Finally, the prediction, sensitivity, and diagnostic analysis techniques in DBNs are used to simulate the evolution of the whole risk system. Meanwhile, the basic events probabilities were updated based on new evidence, and the most likely causes of accidents were found. Safety control measures are developed to effectively reduce the risks faced by the MNPP in the disconnect operations.