This conference paper was submitted for presentation at the NAFEMS World Congress 2025, held in Salzburg, Austria from May 19–22, 2025.

Abstract

The performance of composite materials with thermoset resin is influenced by the processing conditions because of inherent residual stresses developed during cure. Thus, when analyzing composites, the effect of the cure process must be done in conjunction to the mechanical load analysis. The source of residual stresses are non-mechanical strains due to chemical shrinkage and thermal strains. Moreover, the non-uniform temperature distribution through the thickness also affects the stress development within the composite, especially when the composite is thick. In this work, we will present a finite element framework to perform cure analysis of fiber-reinforced composites and laminated structures. More generally, the cure-hardening instantaneous linear elasticity (CHILE) constitutive model is used to model the modulus evolution in the matrix during cure. Specifically, we will discuss a machine learning implementation using an artificial neural network (ANN) model, which serves as input to the CHILE model, to model the cure evolution of the matrix material. The ANN model is implemented inside our user subroutines. The present cure model takes into account the cure kinetics as well as the heat generation due to the chemical reaction. As an effective yet simple strategy, we will use micromechanics models to inform the effective thermal properties to capture the overall effects of cure process. In the micromechanics model itself, the fibers are modeled with a linear-elastic transverse isotropic material behavior, whereas the matrix behavior is captured using the CHILE model. The FE framework is used to capture the curvature of asymmetrical cross ply panels, with two different layups and sizes. The results are compared with experimentally measured panel curvatures. The integrated analysis framework enables a high-fidelity progressive failure analysis of the composite laminate following a virtual curing step. Thus, the impact of the cure process on the mechanical performance of the laminate, for a given loading scenario, can be assessed. The ongoing work is devoted to expanding the capabilities of our modeling framework to include thermoplastic resin systems and other architectures such as textile composites.