In the context of damage tolerance for aeronautical structures, substantial research has focused on simulating skin-stiffener separation in stiffened composite panels. This separation is marked by unstable crack growth at the skin-stiffener interface, which can lead to structural collapse in the post-buckled regime. Recent post-buckling tests on thermoplastic butt-joint single stiffener panels indicate the development of delamination within the skin before crack propagation occurs at the skin-stiffener interface. This delamination is likely triggered by the crack extension process prior to the buckling tests, where the skin was subjected to out-of-plane loads away from the stiffener, promoting the extension of the artificial crack at the interface. It has been hypothesized that this delamination results from crack migration from the skin-stiffener interface into the ply interfaces within the skin. Crack migration, which involves complex interactions between delamination and matrix cracks, is crucial for improving numerical models. For accurate prediction, these models must capture both matrix crack and delamination interactions.

Cohesive zone models, in conjunction with the eXtended Finite Element Method (XFEM) and cohesive elements (CE), have been employed in literature to model the interaction between matrix cracks and delamination. While previous approaches often enrich cohesive elements using user subroutines, this thesis aims to leverage ABAQUS's built-in methods to model crack migration. A series of migration test simulations were conducted to evaluate the combined XFEM and CE approach. The LaRC05 failure criterion in ABAQUS was applied to initiate inclined matrix cracks within the plies, while delamination was modeled with standard 8-node linear cohesive elements. A three-dimensional mesoscale model of the test specimen was developed to simulate the migration test. The LaRC05 criterion successfully captured the orientation changes in matrix cracks due to changes in shear stress, consistent with experimental results. However, the predicted migration distance was 2-3 times greater than observed experimentally. A parametric study revealed that lower matrix strength and fracture energy facilitated migration, although increasing these parameters did not result in a consistent delay in migration, with discrepancies arising at higher values. Despite this, the methodology demonstrated the ability to predict crack migration tendencies and is considered suitable for structural-level applications.

Simulations of the butt-joint thermoplastic skin-stiffener panel under bending were also performed using a global-local modeling approach. The 19-ply skin was meshed with shell elements, while the local model explicitly represented the outer two plies (45/-45) with solid elements and the remaining 17 plies with shell elements. Three modeling approaches were explored: (i) damage only at the skin-stiffener interface, (ii) damage at both the skin-stiffener and ply interfaces (global-local model), and (iii) matrix cracks combined with delamination at both interfaces (global-local model). The first two approaches predicted mode I crack extension at the skin-stiffener interface, with no interlaminar damage in the second approach. However, the third approach using XFEM-CE predicted significant matrix cracking in the outer ply beneath the filler material, which further initiated delamination at the 45/-45 interface. This method successfully predicted delamination migration in the stiffened panel, demonstrating its capability to capture complex damage interactions at the structural level.