Multiscale Interface Modeling for Structural Health Monitoring and Failure Prediction in Adhesive Joints and Directionally Dependent Materials

Adhesive bonding enables the creation of lightweight structures. However, predictive modeling of single-lap joints under process-to-service loading remains a non-trivial task. This dissertation develops and validates a process-aware interface framework that unifies cohesive representations of the adhesive layer and thermomechanical curing effects consistent with ASTM-standard testing, and anisotropic damage formulation for directionally dependent materials. Within ABAQUS, four adhesive-layer idealizations listed as a single-row cohesive element, a middle-row cohesive element, a solid continuum layer, and a property-graded layer are benchmarked for mesh objectivity and mode-mixity evolution at overlap ends and failure-mode transitions. Later findings show that the solid cohesive layer provides a reliable local peel/shear stress field and damage onset, while a properly tuned property-graded or “middle-CZ” representation reproduces global load displacement with markedly lower cost; single-row continuum interfaces under-resolve peak peel near adherend terminations. Furthermore, a coupled thermomechanical analysis quantifies cure-induced residual fields using embedded optical fiber cable as in-situ sensors, with cohesive parameters calibrated by inverse identification against ASTM single-lap shear tests. The calibrated models reconstruct residual-stress gradients that bias early mode mixity, and they assess sensor intrusiveness through a diameter-to-adhesive-thickness scaling. The optical fiber’s stiffness perturbs local fields near the mid-thickness yet leaves far-field joint strength essentially unchanged when placement and diameter are selected within identified bounds. The simulations also rationalize the optical fiber strain under-reading in short edge zones via a finite strain-transfer length before the fiber tracks the adhesive strain. Crack propagation spanning adhesive and adherend regions is resolved through a formulation that preserves path continuity across dissimilar elastic and damage properties. To extend failure prediction to directionally dependent materials, an anisotropic damage model is formulated by combining a positive/negative projection, cast as a linear complementarity problem, with an auxiliary isotropic mapping. For transversely isotropic cases, a single auxiliary Poisson parameter identified within a narrow interval restores most lateral-strain recovery upon unloading, enforces unilateral damage, preserves the symmetry of the fourth-order consistent tangent, and yields objective energy dissipation with stable crack-path evolution. Collectively, the framework closes the process–structure–performance linkage for bonded materials and paves the way for the modeling of composites and sandwich structures under realistic manufacturing histories.