Robust identification of carbon fibers and polymer matrix properties in composites from simple tensile tests

The behavior of constituents in carbon fiber-reinforced plastic composite is often challenging to measure. Carbon fibers typically exhibit transversely isotropic elastic behavior, yet their parameters are challenging to measure directly due to their small diameter. Likewise, matrix suppliers often provide only elastic parameters, and neat matrix samples may be unavailable or may behave differently in isolation than within the composite. To access these properties, an alternative is to identify constituent properties from tests at the macroscopic level on the composite and inverse homogenization. This work presents a methodology for unidirectional composites that combines a simple experimental protocol with an inverse identification procedure to determine the anisotropic elastic parameters of carbon fibers and the viscoelastic behavior of the matrix. Particular emphasis is placed on designing straightforward tests and making methodological choices that enhance robustness, such as representative volume element selection guided by sensitivity analysis. Two complementary validation strategies are implemented: (i) uncertainty quantification of the identified fiber parameters using a Monte Carlo procedure with noisy inputs, and (ii) experimental validation of the viscoelastic matrix model by predicting the creep response of a laminate configuration not used in the identification and comparing it with experimental data. The approach is applied to a prototype PAEK/carbon HS 194 gsm composite. The uncertainties on the reconstructed fiber parameters are found to be satisfactory given the simplicity of the input measurements. The identified matrix model accurately reproduces the response of the validation laminate.