Non-linear Failure Analysis of FRP Laminates composed of UD laminae

- A comparison of the author’s predictions with test results within the worldwide Failure Exercise in the UK -

by Prof. Dr. Ralf Cuntze

The paper presents the author’s contribution to a worldwide exercise on ‘Failure theo-ries for fibre-reinforced plastic (FRP) laminates composed of unidirectional (UD) laminae’. An assessment is made of the correlation between predictions and experimental results provided for a couple of test cases involving biaxial initial and final failure envelopes and stress strain curves of various unidirectional and multidirectional laminates. The theory, implemented into a computer code, employs 3D failure criteria for UD laminae together with non-linear modelling of laminae and laminate. The criteria are based on the author’s Failure Mode Concept (FMC), which provides a set of failure criteria formulated on lamina level that allows for a prediction of the critical lamina failure mode. The generated computer code captures relatively large laminate strains to failure. Special emphasis has been put on the difference between an isolated and an embedded UD lamina. The correlation between the predictions and the experimental data is very satisfactory for multi-axially loaded UD laminae, and is generally satisfactory for the laminates, especially where Fibre Failure (FF) is the dominant mode, which is the case for well-designed (according to netting theory) laminates. Discrepancy between predictions and test data was largest for the ±55∞-angle ply glass/epoxy laminates that experienced very large deformations usually not permitted in structural practice. To tackle such an extreme test case (TC) would involve post failure progression analysis far beyond the occurrence of the various Inter Fibre Failure (IFF) modes. There was a lack of UD test data domains in some of the test cases. Furthermore, in some lamina and laminate test cases ‘jumping’ of experimental data is observed that cannot be explained by scatter. Where possible, a re-evaluation of experimental results was executed which reduced the discrepancy between prediction and experiment very much.