Multivariable Signal Processing for Characterization of Failure Modes in Thin-Ply Hybrid Laminates Using Acoustic Emission Sensors.

The aim of this study was to find the correlation between failure modes and acoustic emission (AE) events in a comprehensive range of thin-ply pseudo-ductile hybrid composite laminates when loaded under uniaxial tension. The investigated hybrid laminates were Unidirectional (UD), Quasi-Isotropic (QI) and open-hole QI configurations composed of S-glass and several thin carbon prepregs. The laminates exhibited stress-strain responses that follow the elastic-yielding-hardening pattern commonly observed in ductile metals. The laminates experienced different sizes of gradual failure modes of carbon ply fragmentation and dispersed delamination. To analyze the correlation between these failure modes and AE signals, a multivariable clustering method was employed using Gaussian mixture model. The clustering results and visual observations were used to determine two AE clusters, corresponding to fragmentation and delamination modes, with high amplitude, energy, and duration signals linked to fragmentation. In contrast to the common belief, there was no correlation between the high frequency signals and the carbon fibre fragmentation. The multivariable AE analysis was able to identify fibre fracture and delamination and their sequence. However, the quantitative assessment of these failure modes was influenced by the nature of failure that depends on various factors, such as stacking sequence, material properties, energy release rate, and geometry.

Detection of Barely Visible Impact Damage in Polymeric Laminated Composites Using a Biomimetic Tactile Whisker.

This is a novel investigation on the possibility of detecting barely visible impact damage (BVID) in composite materials by whisking across the surface via tactile whisker sensors that resemble rats' whiskers. A series of drop tower low-velocity impact tests were performed on quasi-isotropic composite plates. The plates were made from unidirectional T800 carbon/MTM49-3 epoxy prepregs with the stacking sequence of [45/0/90/-45]4S. Investigating the specimens' surface by the naked eye does not reveal any significant damage, rather than a small dent on the surface, with no tangible difference in the different impact energy levels. Ultrasonic C-scan observations showed the existence of BVID in all the impact energy levels, with an increasing trend in the damage size by increasing the impact energy level. The collected data from whisker sensors were analyzed using the support vector machine classifier, based on their vibrational properties, to identify the impacted region and classify the impact severity. It was observed that after training for 13 whisker contacts, the BVID severity can be classified with an accuracy of 100%. This is offering a new BVID detection technique, with a high potential for automation and high reliability that can be used as an alternative or combined with available inspection systems.