Prediction of CFRP longitudinal tensile strength distribution via probabilistic modeling of carbon fiber strength variability

The reliability of carbon fiber-reinforced polymer (CFRP) composite structures is often limited by strength variability introduced during fabrication, particularly at the fiber filament scale. This study combines finite element (FE) modeling with statistical analysis to investigate the influence of filament-level strength uncertainty on the tensile behavior of fiber bundles. The accuracy of the FE simulation is validated against experimental data. Subsequently, several probabilistic distribution models, including Weibull, normal, and log-normal, are employed to characterize the variability in longitudinal tensile strength. The Kolmogorov–Smirnov (K–S) test and the maximum likelihood criterion are used to identify the optimal probability distribution for axial tensile strength. The results show that the tensile strength most closely follows a log-normal distribution. These findings provide valuable insights into the stochastic nature of CFRP mechanical properties and establish a foundation for reliability-based design and structural optimization.