Direct FE2 multiscale modeling of transverse matrix cracking in thermoplastic composites using stochastic representative volume elements

Abstract

Fiber-reinforced thermoplastic composites are critical in aerospace for their lightweight and durable properties. However, predicting transverse matrix cracking remains challenging due to microstructural heterogeneity and the stochastic nature of fiber arrangements. This study introduces a stochastic Direct FE 2 framework to model transverse failure in glass fiber/polypropylene composites. The approach integrates macroscale tensile simulations with randomly assigned representative volume elements at the microscale, moving beyond the conventional assumption of uniform microstructures. Automated Python scripting in Abaqus generates RVEs with stochastic fiber placements, modeling matrix plasticity via the Drucker-Prager criterion and ductile damage evolution. Simulations under transverse loading demonstrate that models with stochastic RVEs yield significantly more accurate elastic stiffness predictions compared to uniform RVE configurations, closely aligning with experimental data. The results capture matrix-dominated failure mechanisms, with cracks initiating at inter-fiber regions and propagating transversely.