A Modified FMEA Framework Incorporating Confidence-Enhanced Probabilistic Linguistic Modeling and Weighted Hypergraph Propagation for LNG Storage Tank Risk Assessment

Abstract

LNG storage tanks are essential facilities for large-scale storage and transportation of cryogenic energy. Because of the flammable, explosive, and ultra-low-temperature characteristics of liquefied natural gas, failures in such systems may result in serious consequences for operational safety and the surrounding environment. Effective identification and prioritization of potential failure modes are therefore crucial for safe operation. Failure mode and effects analysis (FMEA) has been widely applied in risk assessment, yet conventional FMEA methods still show limited capability in describing uncertain linguistic evaluation information, reflecting the reliability of expert judgments, and representing high-order coupling relationships among failure modes. To address these issues, this study develops a modified FMEA framework that integrates confidence-enhanced probabilistic linguistic modeling with weighted hypergraph propagation for LNG storage tank risk assessment. In the proposed framework, confidence-enhanced probabilistic linguistic term sets are employed to represent the fuzziness, probabilistic preference, and reliability differences contained in expert assessments. A confidence-adaptive scoring function is further constructed to strengthen the discrimination of risk quantification by capturing structural differences in probability distributions without introducing externally specified parameters. Meanwhile, the importance of risk factors is determined through a combined subjective–objective weighting strategy, and a weighted hypergraph propagation mechanism is established to characterize high-order structural associations among failure modes and to revise baseline risk levels through a node–hyperedge–node transmission process. A case study of a large LNG storage tank system in Tangshan, China, is carried out to examine the applicability and effectiveness of the proposed framework. The results demonstrate that the proposed method can effectively integrate complex expert evaluation information with structural coupling effects, while sensitivity and comparative analyses further confirm its robustness and suitability for failure risk prioritization in LNG storage tanks.

Bibliographic Information