Geometrical effects in the impact response of composite structures

• Main Author: Yang, Fanjing
• Published: University of Liverpool 2010
• Subjects: 620.1
• Online Access: http://ethos.bl.uk/OrderDetails.do?uin=uk.bl.ethos.548776

The aim of this research is to investigate geometrical effects in the low velocity impact response of a wide range of composite structures. Attention focuses on understanding the impact response of circular CFRP and GFRP laminates, as well as the behaviour of aluminium honeycomb sandwich beams and determining the influence of target geometry on impact response. The impact contact force histories predicted by the mathematical models are presented and compared to the experimental results. The resulting damage in the structures was also examined by evaluating cross-sections of damaged structures. A series of low velocity impact tests have been carried out on a (0°,90°) glass fibre reinforced epoxy resin in order to investigate the influence of varying key parameters on the damage initiation threshold. Initial tests have confirmed observations made by previous researchers, that is that the impact force required to initiate damage, Peril, varies linearly with h3/1, where t is the target thickness. This relationship has been shown to apply for test temperatures between 23 and 90°C. The experimental evidence suggests that the influence of test temperature on damage initiation is complex, with the initiation force increasing with temperature for thinner laminates and decreasing in thicker panels. It has also been shown that this threshold does not depend on target size, for the range of plate geometries investigated here. A final series of tests to investigate the influence of impactor geometry have shown that Pcrit increases with indentor diameter. The damage initiation threshold was predicted using two previously-published models, one based on the Mode II interlaminar fracture toughness of the composite, GIIe, and the other on the interlaminar shear strength, ILSS. Here, it was shown that the latter can successfully predict the variation of Peril with both target size and impactor geometry. The second part of the study focuses on investigating scaling effects in the low velocity impact response of sandwich structures and plain composite materials using an instrumented drop-weight impact tower. Tests on carbon fibre reinforced epoxy laminates indicated that at energies above the first failure threshold, damage does not obey a simple scaling law, instead, becoming more severe as the scale size is increased. An examination of the damaged samples indicated that fibre damage was greater in the larger samples for a given scaled impact energy. Fibre damage took the form of large cracks extending across the panel in the 0 and 90 degree directions, suggesting that the energy absorbed in this mode of failure should scale with the square of the scale factor, i.e. n2 In contrast, the initial impact energy was scaled according to n3 indicating that there will be an excess of energy in the larger panels for a given scaled impact energy. Impact tests on the sandwich structures resulted in a more localised mode of fracture, involving perforation of the upper skin and localised crushing of the core, when the impact energy was sufficiently high. The resulting load-displacement traces indicated that the larger scale sizes absorbed more energy during impact than their smaller scale sizes. Cross-sections of perforated sandwich structures showed that, for a given scaled impact energy, the type and severity of damage were similar in all panels. This research also presents an experimental study for the normalised low-velocity impact response of composite plates. It is demonstrated that a characterisation diagram that shows the relationship of three non-dimensional parameters with the normalised maximum impact force can be used to fully characterise the response. With the governing non-dimensional parameters obtained experimentally, it is shown that impact tests having the same non-dimensional parameters, have dynamic similarity and the same non-dimensional response. Furthermore, the experiments can be placed in appropriate dynamic regions in which simplified dynamic models can be used to predict the response.