A computational investigation of failure modes in hybrid titanium composite laminates

Abstract

High strength, low weight Fibre Metal Laminates (FMLs) are increasingly finding use in military and commercial aircraft applications by combining benefits of both fibre reinforced composite and metal components such as fracture toughness and fatigue resistance. Glass fibre reinforced composite laminates combined with aluminium (GLARE) was recently selected for the upper fuselage skin structure of Airbus A380 aircraft. Hybrid Titanium Composite Laminates (HTCLs), made from thin titanium sheets and Fibre Reinforced Polymer (FRP), are more suitable in high strength environments at elevated temperatures and may therefore be used in future aerospace applications. Currently, the combination of titanium and FRP is mainly applied to surgical implants or biomedical devices. In order to improve design and performance of FMLs, it is essential to understand their failure mechanisms under various loading conditions. Major failure modes in composite laminates are delamination, matrix cracking and fibre fracture. Delamination can be initiated by stress concentrations at free edges, interfacial voids or transverse matrix cracks. The latter is referred to as Matrix Cracking Induced Delamination (MCID). An additional critical failure mode in FMLs is debonding between composite and metal sheets. These failure modes cause stress redistribution and can lead to complete loss of load-carrying capability resulting in structural collapse. Virtual testing of FMLs can not only reduce physical testing, it also offers cost-effective design flexibility. FML structures can be virtually optimised to suit specific applications before time consuming and expensive manufacturing. Therefore, effective and accurate models are required to describe failure mechanisms and their interaction. The Phantom Node Method (PNM) is a numerical concept to simulate failure initiation and evolution in heterogeneous materials such as FMLs within the Finite Element Method (FEM). It includes the well-established Cohesive Zone Method (CZM) to approximate nonlinear fracture processes in interfaces or cracks. Elemental locality and the use of standard shape functions make PNM conceptually simpler and hence easier to implement and modify compared to similar enrichment methods such as X-FEM. This research project aims to develop, validate and use computational methods within FEM in order to investigate failure modes in composite laminates and HTCLs. Thereby, standard simulation tools as well as novel conceptual methods such as PNM are to be applied. Experimental investigations provide important modelling properties as well as data to validate numerical results. A novel Top Surface Analysis (TSA) technique using Digital Image Correlation in a combination with the J-integral characterises interface properties in Double Cantilever Beam specimens. It is demonstrated that TSA can estimate strain energy release rates in symmetric composite laminates and thin asymmetric HTCL samples. A numerical analysis using the Virtual Crack Closure Technique in the standard FE-software Abaqus confirms experimental results and assumptions made by TSA. The findings are used to numerically investigate failure progression in HTCLs under impact by Continuum Damage Mechanics and CZM within an explicit FEM analysis. Major failure modes are experimentally and numerically found to be debonding in the titanium-composite interface, matrix cracking and interlaminar delamination. The specific structure of HTCLs with energy absorbing titanium sheets allows for various simplifications to reduce model complexity and computational cost without loss of accuracy. In order to model complex interaction of matrix cracks and delamination in MCID, PNM is extended to incorporate interfacial failure at matrix crack tips. In addition to material nonlinearities governed by CZM, advanced PNM accounts for geometrically nonlinear effects by incorporating the total Lagrangian formulation for large displacements. The new concept is applied to glass fibre reinforced composite laminates under tension. It is able to accurately predict typical quantitative measures such as matrix crack density and stiffness degradation. A progressive analysis of MCID in carbon fibre reinforced composites under flexural loading predicts individual delamination growth at matrix cracks quantitatively and qualitatively. Advanced PNM proves to be a promising effective numerical tool which is applicable to standard FEM software packages.