Abstract
Building on the recent success of simulations that homogenize fiber plies ("ply-scale models"), experimenters and theorists are now seeking to predict the details of failure events within a single ply in a laminate, where the discreteness of fibers must enter the problem. This new frontier of research demands a multi-scale representation of failure, bridging the ply and infra ply scales. Such a multi-scale model promises to revolutionize the optimization of composite materials and the prediction of their safe life. However, the infra-ply scale is much more difficult than the ply-scale problem and will not be easily conquered. The key experimental challenges is acquiring 3D data that reveal the random microstructure and damage events in the interior of the composite with very high resolution (similar to 1 mu m). Key theoretical challenges include understanding how to represent the essential stochastic characteristics of the 3D microstructure, how to model the failure events that evolve within it, and resolving the fundamental question of what is predictable and what cannot be predicted because of the extreme sensitivity of the failure sequence in a random material to initial conditions.
