Efficient Simulation of Crack Propagation in Adhesive Bonds

Abstract

Accurate yet computationally affordable modelling of interface crack propagation is required for the development and certification of large bonded assemblies. Conventional cohesive zone modelling (CZM) resolves the fracture process zone (FPZ) with traction–separation laws that typically require element edges below 1 mm. This size limitation makes it impractical at industrial relevant scales. This thesis investigates an energy-release-rate cohesive (ERRC) approach for largeelement modelling of adhesive-interface crack propagation. The method combines a virtual crack closure technique (VCCT) propagation criterion with a local cohesive release law, so that crack growth is triggered by the energy release rate while the newly created crack surfaces dissipates the prescribed fracture energy progressively. The main contribution of the work is the reformulation of this approach as a userdefined solid element in LS-DYNA, intended for adhesive interfaces discretised with solid adherends and finite-thickness bondline representation. The implemented interface is represented by lower–upper nodal pairs, where each pair carries as discrete state: tied, cohesive, or open. In this thesis, two implementation variants were developed: a single-core reference (SCR) implementation and a multi-core capable (MCC) implementation. For the DCB benchmark, the implemented method reproduced the expected meshaligned Mode I crack-growth behaviour. The crack-length evolution followed the corrected beam theory (CBT) reference within the resolution of one discrete interfaceelement edge, and the interface-energy diagnostics showed that the released pairs followed the intended fracture-energy target. Representative Mode I propagation was obtained using an in-plane interface element length of 4mm, substantially larger than the sub-millimetre element sizes typically required to resolve a conventional FPZ. The ENF benchmark confirmed that the discrete nodal-pair formulation can produce a coherent Mode II-dominated propagation chain. However, the global force– displacement response showed pronounced dynamic oscillations, particularly for the MCC implementation. The dissipation diagnostics showed that the implemented damage variable limits the release state, but that the reconstructed path-work quantities can become unreliable during rapid, single-cycle Mode II release events. The ENF results therefore verify important parts of the formulation but also show that further stabilisation and quasi-static assessment are required before the method can be considered robust for Mode II-dominated loading. Overall, the work demonstrated the feasibility of extending the ERRC method to solid-element adhesive-interface formulation in LS-DYNA. The implementation provides a promising route for efficient large-element simulation of interfacial crack propagation in bonded structures. At the present stage, the method should be interpreted as a proof of concept for regular, mesh aligned benchmark problems rather than a fully general industrial fracture-modelling tool.