Multiscale/cohesive zone model for composite laminate impact damage

Interface damage mechanics, or cohesive zone models, have been developed over the last decade as a method of modeling crack growth in a material or debonding between two different materials. These methods have alleviated many of the numerical problems inherent in crack modeling, including the large length scale difference between crack fronts and crack areas, strem singularities, and the adaptation of crack propagation criteria to non-linear materials. Cohesive zone models can also predict crack initiation at any number of predetermined possible crack locations. However, researchers have also found that numerical instabilities in the solutions emerge if the finite element mesh is too coarse relative to the crack process radius. Consequently, these have been practical only for very small structures, on the order of tens of millimeters, without the use of supercomputers. We will show that changing the order of numerical integration of the interface properties independently from their spatial discretization solves this convergence problem and in most cases decreases the total computation time, allowing for simulations of much larger structures. We will also show how these results are incorporated into our multilength scale model for predicting impact damage in laminated composite plates