A global–local approach for progressive damage analysis of fiber-reinforced composite laminates

The present work applies the global–local technique to the progressive damage analysis of fiber-reinforced composite laminates. A one-way, loosely-coupled global–local approach is developed as a combination of a low-fidelity linear global analysis and a high-fidelity local nonlinear analysis of specific regions within the structure, where damage is expected to occur. The local model is based on higher-order structural theories derived using the Carrera Unified Formulation (CUF), and specifically, Lagrange polynomials are used to model each ply through its thickness, leading to a layer-wise model. Composite damage is described using the CODAM2 material model, which is based on continuum damage mechanics. Initial assessments compare the relative performance of 3D finite elements (FE), 1D-CUF, and the proposed global–local approach via the free-edge stress analysis of a stiffened composite plate. The proposed technique is then used to predict the tensile strength of an open-hole specimen. The last assessment simulates damage progression within an over-height compact tension specimen using the global–local approach. Verification and validation of results are carried out via refined models and experiments from literature. The results demonstrate the accurate evaluation of 3D stress fields and composite laminates’ mechanical response in the progressive damage regime. A multi-fold improvement in the computational cost is shown when compared to full-scale CUF analyses and indicates this technique’s strong potential towards the computationally-efficient high-fidelity analysis of complex and large-scale composite structures. •A global–local approach is used to evaluate damage onset and propagation in composite laminates.•The global model is 3D-FE from ABAQUS with coarse meshes.•The local model is 2D or 1D with layer-wise capabilities, and CODAM2 is used for the damage modeling.•Linear and nonlinear cases were considered.•The global–local approach has similar accuracy as a full global model but leads to significant reductions in the computational time.