Seismic Fragility Analysis for a Gas-Plant System using Fault Tree-based Bayesian Network

A fault tree-based Bayesian network (FT-BN) framework was proposed for the seismic probabilistic safety assessment (sPSA) of gas plants.The framework models the components of a system as random variables, and constructs a network based on the correlations derived from plant processes.The constructed BN demonstrates strong potential for making risk-informed decisions when different types of information are available.However, performing sPSA for plant systems requires detailed fragility data for individual components, which remains limited in existing literature.To address this gap, a literature review was conducted, revealing that the Hazus Earthquake Model Technical Manual (2024) developed by the Federal Emergency Management Agency (FEMA) provides fragility curves for entire systems as well as for individual plant equipment.The objective of this research is to validate that the FT-BN can derive system-level fragility from component-level fragilities and to compare the resulting system fragility with that provided by FEMA.To achieve this, several hypothetical plot-plans were assumed, and corresponding FT-BNs were constructed.Numerical investigations confirmed that the system fragilities estimated using the FT-BN were generally more conservative than those presented by FEMA.These findings suggest that the FT-BN framework enables a more comprehensive assessment by incorporating the unique characteristics of each system.