Fatigue crack growth prediction in polymer composites using XFEM cohesive segments

Introduction: Fatigue damage remains a major challenge in the design of composite structures, as cracks may initiate under low stress levels and evolve through multiple interacting failure mechanisms. To reduce reliance on extensive physical testing, finite element methods based on progressive damage analysis are increasingly used to predict fatigue-driven crack growth within structural substantiation workflows. However, most existing approaches require prior knowledge of the crack path, limiting their ability to represent complex, solution-dependent fracture trajectories. The eXtended Finite Element Method (XFEM) offers a promising alternative, as cracks can be modeled independently of the mesh, yet commercial implementations remain limited for fatigue loading. Materials and methods: This work introduces a fatigue-capable XFEM framework that combines the cohesive segments approach with a stress–life-based degradation law using only Abaqus built-in user subroutines and native elements. Crack initiation is governed by a fatigue endurance limit criterion, and damage evolution is driven by a user-defined internal variable that controls cohesive stiffness degradation along the crack interface. A dedicated bookkeeping strategy enables access and transfer of interfacial crack-opening quantities, ensuring consistent evaluation of the fatigue damage variable across the enriched interface. Results: The framework is verified against a classical Cohesive Zone Model (CZM) and applied to a two-dimensional single element model as well as Double Cantilever Beam (DCB) specimen under Mode I static and fatigue using the IM7/8552 carbon/epoxy material system. The results demonstrate excellent convergence between the proposed method and CZM approach for the single element model. Furthermore, the DCB results demonstrate accurate prediction of delamination growth and highlight the proposed formulation as a practical and more flexible alternative to the existing Virtual Crack Closure Technique (VCCT)-based XFEM fatigue capability in Abaqus, without requiring a pre-existing crack or inheriting the restrictions of linear elastic fracture mechanics. Conclusions: The proposed methodology effectively bridges cohesive zone modeling and XFEM, providing a tool for simulating progressive damage in composite structures without the limitations imposed by linear fracture mechanics. Future developments will extend the framework to mixed-mode loading and three-dimensional configurations, broadening its applicability to aerospace-grade laminates and structural components.