Performance of Two Distinct Cohesive Layer Models for Tracking Composite Delamination

Two distinct cohesive layer models are developed for numerical simulation of delamination growth in composite layered specimens tested under static loading. One of these designated as the UMAT (user supplied material) model has a small, but finite thickness and the other designated as the UEL (user element) model has zero initial thickness. Crack growth in double cantilever beam specimens as well as in two test configurations of a composite plate carrying some discontinuities subjected to lateral load are studied using the models. It turns out that UEL model, though slightly more involved, is able to predict both crack initiation and large crack growth with sufficient accuracy. The UMAT model too consistently predicts crack initiation, but is unable to predict the crack growth accurately. It gives consistently higher loads for given crack extensions and predicts that the crack growth shuts off prematurely. Careful examination of the stresses in the cohesive layer of the UMAT model, in the upstream of the crack tip indicates that a ‘neck’ develops due to compressive stresses at some distance from the crack tip. Apparently it is the formation of this neck that ‘locks’ the crack from growing and is the cause of the inaccurate results given by the model.