Reliability modeling of tensile properties in recycled carbon fibers: A comparative fit of statistical distributions pre- and post-textile processing

Due to their extensive use in the aerospace, automotive, wind energy, and sports industries, the recycling of carbon fiber composites has become increasingly important. Characterization of the tensile properties of these recycled fibers is essential, using single fiber tensile tests that provide data on tensile strength, elongation at break, and tensile modulus. This study compares four statistical distributions—2-parameter Weibull, 3-parameter Weibull, Normal, and Lognormal—to fit tensile data from two batches of recycled carbon fibers: one after recycling, and the same type of fiber after a further carding process. Maximum Likelihood Estimation (MLE) was used for the fitting process, while the Sum of Squared Errors (SSE) and Kolmogorov-Smirnov (KS) tests were employed to evaluate the fitting performance. The 2-parameter Weibull distribution provided the best fit overall for tensile strength, elongation at break, and tensile modulus. However, the Normal distribution performed slightly better for tensile strength and tensile modulus in the post-carding batch, indicating that different models might be more appropriate at various processing stages. This study illustrates the challenge of selecting a single model and balancing computational accuracy with mechanical consistency across different properties and conditions, providing insights into the complex behavior of discontinuous fiber composites and contributing to a better understanding of the mechanical challenges associated with recycled materials and their processing.