Micromechanics Surrogate Model for Fatigue Life Prediction of Polymer and Ceramic Matrix Composites

Fatigue modeling of composites, particularly using a multiscale approach, can be very costly, due the number of iterations that must occur when evaluating the damage state of each constituent within each ply, thus making it difficult for engineering in early design stages to evaluate a large number of potential candidate material configurations. Previous work has shown that replacing physics-based simulation for an 8-ply composite with a surrogate model using recurrent neural networks for the entire multiscale simulation resulted in a speed-up of roughly 145 times but was limited in scope by restricting the total number of plies, only considering uniaxial loading, and only allowing polymer matrix material properties. In the current work, the methodology for training a surrogate model is extended to both polymer and ceramic matrix composites (PMC and CMC, respectively), which contain different microstructures and constituent properties, and is generalized to only replace the micromechanics at the ply level. During training, each ply is subject to various multiaxial loads, and is able to predict both the damage state due to the current loading and the number of cycles required for the next damage increment, following the physics-based fatigue model implemented in the Micromechanics Analysis Code with Generalized Method of Cells (MAC/GMC) tool. The developed machine learning model will allow engineers to quickly get a reasonably accurate estimate of the fatigue life a PMC or CMC with any arbitrary number of plies subject to any multiaxial load, thus addressing the shortcoming of the previously built model.