Nuclear power plants (NPPs) are at risk from various internal and externalflood hazards that can lead to extensive damage. Events such as the tsunami atFukushima NPP or the river inundation at Fort Calhoun NPP underscore the ne-cessity of conducting flooding probabilistic safety assessments (PSA) for NPPs.The response of components to flooding conditions can have considerable impacton the stable operation of NPPs, however, flood PSA is not common and flood-ing related risk analysis lacks data. To help address this shortcoming, compo-nents were tested in the Portal Evaluation Tank (PET) to provide the necessarydata, initially preforming experiments on non-watertight doors . A Bayesian re-gression model, utilizing explanatory variables, was developed for assessing NPPcomponent fragility in flooding events. The concluding model uses the binomialdistribution with a logistic regression model for probability of failure and usesnon-informative priors so the experimental data takes preference in the analysis.The fragility model was validated using several metrics, including three types ofBayesian p-values. Finally, an application of the component fragility model wasdemonstrated to show how improvements might be made to 3D flooding simula-tions. |