Modelling multiple delamination and intralaminar cracks using a single-layer shell approach

Paper i proceeding, 2015

The introduction of fibre reinforced polymers in the automotive industry is strongly dependent on accurate and efficient CAE tools to predict the correct energy absorption in car crash analyses. From experimental observations during axial crushing, it is obvious that in order to obtain good predictability in the simulations, delamination needs to be accounted for. However, due to industrial restrictions on the total simulation time of full scale crash analyses, detailed modelling of each ply as represented by separate elements through the thickness – to promote delamination modelling by cohesive interface elements – is impossible. In order to provide an alternative approach, we have recently proposed an enriched XFEM shell element formulation with the benefit that multiple propagating delamination cracks can be kinematically represented independently of the finite element mesh. Hence, a structural model of a thinwalled laminate can thereby initially be built up by a single layer of shell elements through the thickness. During loading, the model is then enriched locally (and adaptively) in critical areas where delamination is predicted. However, as is well known, delamination is far from the only relevant failure mechanism which needs to be considered in the simulation of crash-like events. In combination with multiple delamination cracks, severe intralaminar failure transverse matrix failure, fibre kinking, matrix shear failure etc.) is observed in the highly strained areas of the structure. These mechanisms are often described in the literature via progressive damage models based on an embedded cohesive zone approach in which the cohesive degradation behaviour is ’smeared’ over the element experiencing damage evolution in a way such that pathological mesh dependence (which is always present for a local continuum damage model) should be avoided. In contrast to this, some authors have with success modelled intralaminar crack propagation explicitly by using the XFEM (or phantom node method), although for the case where each ply is discretised explicitly by separate shell or solid elements. In this contribution, we investigate the potential of further extending the previously developed enriched shell element capable of representing multiple delaminations to also handle intralaminar crack propagation without having to explicitly model each ply by separate (shell) elements through the thickness. Instead, the addditive enrichment for multiple delaminations is combined with additional enrichment functions confined only to individual plies, making it possible to represent individual intralaminar cracks with different orientations in different plies.