Homogenization Methods for Multiscale Process Modeling Through Full Factorial Design

Abstract

Crucial to developing the next generation of high-performance composites are multiscale process models which have the capacity to predict in-service performance affecting residual stresses and deformations generated during manufacturing. Length scale and geometrical complexity of composites increase in tandem; thus, process models must rely upon methods for homogenization of the curing process. This work proposes a novel, high-fidelity computational approach which leverages finite element (FE) analysis at the microscale to predict effective composite properties at arbitrary cure and temperature states. The approach is applied to a high-fidelity representative volume element (RVE) and low-fidelity repeating unit cell (RUC) that models a 75% fiber volume fraction AS4-carbon fiber/epoxy composite microstructure. Elastic properties of the composite are characterized across several temperature and cure states for each FE model. Two classical closed-form composite homogenization approaches are also used to predict effective properties at each cure state. Predictions obtained by each method are compared against those of the RVE on a state-by-state and property-to-property basis.