Bayesian Network-Based Reliability Assessment of an Integrated Subsea Production Architecture

Growing operational challenges in deepwater production have intensified the demand for subsea systems that combine robustness, reliability, and efficiency in maintenance and risk control. This work advances the reliability assessment of subsea systems by extending the analysis beyond the Subsea Christmas Tree (XT) to encompass the entire subsea architecture. A subsea architecture is composed of multiple interconnected equipment, such as flowlines, manifolds, and the XT itself, whose interactions define the overall system performance. However, most studies in the literature focus on isolated components, which limits the understanding of global reliability behavior and its implications for design and maintenance strategies. Furthermore, traditional reliability tools face challenges in representing dynamic interdependencies and dealing with inherent uncertainties of complex subsea environments. To overcome these limitations, this study proposes an integrated framework that combines Event Sequence Diagrams (ESDs) and Fault Trees (FTs) within a unified Bayesian Network (BN) structure. This approach enables both predictive and diagnostic reasoning through bidirectional inference, allowing the identification of the most probable causes of system failure and the estimation of failure probabilities for individual components. Such capability provides theoretical support for decision-making during the design phase and the planning of maintenance routines under uncertain conditions. The proposed method is applied to a satellite-type subsea layout, where each well is directly connected to the Floating Production Storage and Offloading (FPSO) unit. This study explicitly includes the XT in the reliability modeling, recognizing its crucial role even in architectures where it is often assumed as a standard element and, therefore, neglected in comparative analyses. Preliminary results demonstrate that the proposed Bayesian framework can effectively predict the most failure-prone modules and critical elements within the architecture.

Keywords: Bayesian Networks, Reliability Analysis, Subsea Systems, Event Sequence Diagrams, Fault Trees.