Development of probabilistic risk assessment methodology against strong wind hazards in floating nuclear power plants - Part I: hazard assessment

摘要: This study introduces a probabilistic risk assessment (PRA) framework against evaluating extreme wind hazards on floating nuclear power plants (FNPPs), extending conventional land-based nuclear plant methodologies through wind speed-based hazard curve development. We establish an integrated analytical approach incorporating 1) multi-source meteorological data integration, 2) coupled wind field modeling with extreme value theory, and 3) Bayesian uncertainty quantification. Hazard curves were systematically derived for three wind phenomena – straight-line winds, typhoons, and tornadoes – with consideration of their combination. The methodology advances inverse uncertainty quantification through Markov Chain Monte Carlo (MCMC) parameter calibration for typhoon and straight-line wind models, enabling probabilistic propagation of parameter uncertainties into hazard estimates. Tornado risk assessment integrates stochastic encounter modeling with classification error-adjusted conditional probabilities, enhanced through Bayesian updating techniques. Key findings reveal substantial epistemic uncertainties in low-probability hazard regions, primarily attributable to data sparsity in extreme wind records and empirical limitations of current wind field parameterizations. This work establishes a foundational methodology for marine nuclear infrastructure risk assessment while highlighting critical knowledge gaps in extreme wind characterization for floating systems.