Numerical and Experimental Investigation of Hygrothermal Aging in Laminated Composites

Although being a crucial step in structural design of laminated composites, prediction of their long-term mechanical performance remains a challenging task for which no comprehensive and reliable solution is currently available. Nevertheless, structures such as wind turbine blades, of which laminated composites constitute the main load bearing parts, must be designed to withstand 20 years of service while being subjected to a combination of fatigue loads and interaction with often extreme environmental conditions. In the end, a compromise is reached by compensating the lack of knowledge on the complex material degradation and failure mechanisms spanning multiple spatial and time scales that determine mechanical performance by adopting higher safety factors. This in turn leads to heavier, less efficient and more expensive designs. A better understanding of these mechanisms through discerning experiments and the development of fast and accurate numerical prediction tools are therefore necessary.

This work focuses on the phenomenon of hygrothermal aging (a combination of high temperatures and moisture ingression) on unidirectional laminated composites. The complexities of the aging problem, a combination of physical and chemical degradation mechanisms that affect fibers, resin and interface differently, are investigated through a combination of experiments, microscopic observation techniques and state-of-the-art numerical modeling. The result is an efficient multiscale and multiphysics framework for the prediction of failure and hygrothermal degradation in composites.

First, an experimental campaign is conducted on unidirectional glass/epoxy composite samples and on pure epoxy specimens immersed in water at 50C and tested quasi-statically and in fatigue. By comparing results of unaged, partially saturated, saturated and redried samples, the contributions of reversible and irreversible hygrothermal aging mechanisms are measured. The results indicate a strong correlation of degradation with the water concentration field inside the specimens. Furthermore, significant differences in strength reduction between composites and pure resin specimens point to damage in the fiber-matrix interfaces.

In order to realistically model the diffusion process that drives degradation, an experimental/numerical study is conducted on the anisotropic diffusion behavior of laminated composites. Thin material slices extracted from a thick composite panel are immersed until saturation and the obtained anisotropic diffusivity parameters are numerically reproduced through a microscopic diffusion model with periodic concentration field. The existence of an interphase transition region around the fibers is confirmed through microscopic experiments and included in the model through a level set field.

Since both the diffusion process and the resultant material degradation are highly influenced by the microstructure of the material, a multiphysics and multiscale analysis approach becomes necessary. A numerical framework for modeling of the aging process is proposed combining a macroscopic Fickian diffusion analysis with a multiscale stress equilibrium analysis based on the FE2 method. Since the multiscale approach does not rely on any constitutive hypotheses at the macroscale, complex failure behavior combined with plasticization and differential swelling can be accurately captured.

In order to expand the framework to allow for modeling of cyclic loading and cyclic environmental exposure, a number of additional model ingredients are developed. Firstly, a new constitutive model for epoxy combining viscoelasticity, viscoplasticity and a damage formulation with rate-dependent fracture onset is presented. The model is calibrated through a series of quasi-static and fatigue experiments on pure resin specimens at multiple strain rates and both before and after hygrothermal aging. The calibrated model is able to accurately capture the observed strain rate dependency and stiffness and strength degradations after aging, as well as correctly capturing damage activation in low-cycle fatigue. Secondly, the significant computational cost associated with the use of a cyclic multiphysics/multiscale analysis with nested micromodels is alleviated through a number of acceleration techniques. Time homogenization is used to explicitly divide the loading into a nonlinear macrochronological part and a linear computationally inexpensive microchronological one. Furthermore, the size of the microscopic boundary value problem is reduced through a combination of Proper Orthogonal Decomposition (POD) and the Empirical Cubature Method (ECM), resulting in a hyper-reduced model. The resultant reduced and time homogenized micromodel allows for speed-ups higher than 1000, dramatically accelerating the solution of the problem.

The modified version of the framework is used to numerically reproduce the experimentally obtained interlaminar shear behavior of composite samples aged for different durations. Use of the multiphysics/multiscale approach allows for accurately describing the stress state in specimens with non-uniform water concentration fields. The viscoelsatic/viscoplastic resin model is capable of capturing differences in stress response between the very slow conditioning phase and the much faster mechanical test. The model is completed by a cohesive-zone model for fiber-matrix interface debonding including friction calibrated with a set of Single Fiber Fragmentation tests performed on dry and saturated samples.