Multiscale experimental characterisation and modelling of transverse cracking in thin-ply composites

In this work, the damage dependency on ply thickness is investigated experimentally for three different thin-ply CFRP quasi-isotropic laminates, with constituents spanning a large range of properties. It is shown that free edge damage mechanisms, observed with in-situ microscopy, change with decreasing ply thickness. Whilst free-edge delamination dominates the damage behaviour when thicker plies are used, transverse cracking (TC) dominates the free edge behaviour of thinner plies with an increasing onset of damage with decreasing ply thickness. With some material constituents, this onset is increased so much that the translaminar fracture of the 0° plies precedes it and no transverse cracking is detected any more. Bulk damage, recorded by acoustic emission, is linked to major propagation of TC from the free edge through the width of the sample. It is shown to happen at higher strains than the free edge damage for all but the thickest plies tested, where free edge delamination prevails. Its onset is also shown to scale differently with respect to ply thickness than what is observed at the free edge. Assuming LEFM hypotheses, a critical ERR associated with transverse cracking is identified by inverting the in-situ strength model. A ply dependency of this apparent toughness is shown for all material tested with a decrease of its value with decreasing ply thickness.

A multiscale embedded-cell FE model is developed to gain more insight in the intricate damage mechanisms present at the micro-scale. All material properties used are obtained or derived from experimental tests, save for the fibre-matrix interface. A good agreement with the obtained experimental onset of TC in the bulk as well as at the free edge is achieved. A dominance of the interface properties on the TC development is demonstrated, and its appearance is shown to be triggered by debonding coalescence, leading to matrix micro-filament bridging. The crucial effect of thermal residual stresses is highlighted, as values leading to debonding or matrix failure after curing can be reached at the micro-scale for high temperature curing composites. An inversely proportional influence of the curing temperature on the onset of TC under the assumed modelling hypotheses is shown. Using this calibrated model to identify a TC-related critical ERR, the same trend of a decreasing apparent toughness with decreasing ply thickness than in the semi-experimental case is observed.

In light of these results, a mesoscale model is proposed with all the damage lumped into a single TC modelled using two superposed cohesive element zones, the first one representing debonding and the second one representing matrix micro bridging. An optimisation scheme is used to identify the two linear traction-separation laws implemented to match the experimentally measured AE onsets. The maximum cohesive stresses thus identified as well as critical ERR predictions are in good agreement with the values obtained from the micromechanical model. It was noticed that a simple linear cohesive law could predict the observed scaling of TC onset with respect to ply thickness with enough accuracy, which opens the way for new, improved in-situ strength models that could properly account for the finite size of the process zone without requiring a much larger experimental dataset.