Impact damage characterisation of composite laminates using a statistical approach

Detecting impact damage is an important factor in maintaining the structural integrity of aerospace composite structures. Since impacts can cause severe reductions in stiffness and strength of composite structures, there is a need to investigate the material’s stiffness and strength after an impact event. The work discussed in this paper is concerned with one overall goal, which is to determine if an impact has caused damage and then to determine the extent of that damage using only the structural responses that are acquired from low-profile surface-mounted transducers. The work presented in this paper demonstrates that responses recorded from these sensors can provide sufficient information to infer impact damage. Previous work by various authors has established that it is possible to distinguish between damaging and non-damaging impacts if the impact force time-history is available. Therefore, it is the requirement that only response data be used here that distinguishes this work from the previous studies. Following a systematic series of experiments, on the induction of impact damage in Carbon Fibre Reinforced Polymer (CFRP) laminates, a damage model was proposed from this work which can provide preliminary information on the type and extent of damages through observations made from Scanning Electron Microscopy (SEM) and X-ray radiography. The model can be used to provide a general understanding on the prediction of damage and failure progression in CFRP as a function of the number of layers and impact energies. To accommodate the results from SEM and X-ray, discussions on outlier analysis and visualisation, which emphasise the idea of discordancy from the discipline of statistics, are presented. Results from discordancy tests, for both univariate and multivariate data, can be used to clearly separate the data (non-damaging and damaging impacts as well as the separation of the type of failure modes) by a calculated threshold value. Both of these features (univariate and multivariate) can be used as damage indicators and proxies for determining the extent of damage.