Multiscale modeling of short fiber-reinforced thermoplastics under fatigue loading

The present work falls within a wider framework concerning the prediction of the mechanical response of short glass fiber-reinforced thermoplastics (SFRT), which are commonly employed in the automotive industry to reduce the overall weight of the components. More in detail, the main objective of the thesis is to develop a fatigue criterion for predicting the effect of different factors affecting the fatigue strength of such materials. In this interest, the influence of the composite complex morphology (local fiber orientation and fiber content) and of notches giving rise to stress concentrations is taken into account. In the present thesis, an experimental activity related to plain and notched specimens is firstly presented. In this context, data resulting from computed tomography (CT) analyses are shown. The latter serve to evaluate the specimens’ fiber orientation distributions, which are quantified by means of fiber orientation tensors (FOT). Furthermore, fatigue test data on the considered coupons, in the absence and in the presence of notches (with radii of 0.1 mm, 0.2 mm, 2 mm and 5 mm), are presented for different fiber orientations and weight fractions (15 wt%, 25 wt%, 35 wt% and 50 wt%). Secondly, being aware of the fact that the onset of a macroscopic crack is driven by the evolution of damage at the matrix level, a multiscale fatigue model relying on matrix stress distributions is presented. The calculation of the matrix stress cumulative distribution functions is achieved by formulating an analytical numerical pseudo-grain approach (i) permitting to avoid the generation, mesh and solution of complex microstructures, but only relying on the solution of simple unidirectional cells. The pseudo-grain method is subsequently included in the formulation of a fatigue criterion, for plain (ii), at first, and for notched specimens (iii), subsequently. The proposed fatigue criterion is eventually validated with a bulk of experimental data, partially presented in this work. Namely, fiber orientation tensors are used to properly assign the anisotropic elastic properties to the developed numerical models and the presented fatigue data are employed to assess the efficacy of the model in terms of fatigue strength prediction