Rare-event failure analysis of fine-grained isotropic graphite

Isotropic graphite produced through isostatic pressing has found applications across various industrial fields. Due to the inherent heterogeneity, measurements of tensile strength and fracture toughness of iostropic graphite exhibit considerable variability. While Weibull theory is commonly employed in the statistical analysis of graphite, it primarily fits a subset of data in the middle of the probability distribution, offering limited insight into rare events with low probability. However, rare-event fractures in isotropic graphite are critical for structures demanding extreme safety, such as high-temperature gas-cooled reactors. In this study, we conduct a statistical analysis of the rare-event fractures in isotropic fine-grained graphite, utilizing relatively large datasets of approximately 200 measurements. Tensile strength is measured through the Brazilian splitting tests, while fracture toughness is assessed using three-point bend tests. To analyze rare-event fractures, we employ the peak-over-threshold method combined with a generalized Pareto distribution, focusing on the tails of the probability distribution. Our analysis demonstrates that even small subsets of the dataset can reliably predict rare events with high confidence. The further application of the proposed method to nuclear graphite offers potential advancements in the design of graphite-moderated nuclear reactors with enhanced safety and reliability.