Burn tests with acoustic emission monitoring of filament wound GFRP pipes

From both theory and experimental evidence, an optimum solution for filament wound tubes subjected to hydrostatic pressure is the 55° laminate, maximising strength and allowing for satisfactory production rates. The usual strength criteria predict an equally good behaviour if a (0°/90°2)s laminate is adopted. Although the latter solution would imply dedicated filament winding machines and longer fabrication times, it would also provide a sensibly higher flexural stiffness in the circumferential direction, required when buried pipes are concerned. At this time, no data are available, comparing the performances of 55° and (0°/90°2)s on a common basis. This was one of the objectives of the present work, where Glass Fibre Reinforced Plastic (GFRP) pipes, made of E-glass and Synolite NP244 polyester resin, were produced according to the lay-ups (90, 0, 90)n and (55)n, with n = 1-3, obtaining shell thicknesses in the range 1 to 3 mm. From the pipes, specially designed specimens were fabricated, suitable to yield valid failure modes. The samples were subjected to hydrostatic tests in a Hammel Sciteq P400 machine, by which an internal pressure was applied through water inlet. During the mechanical characterisation tests, the specimens were continuously monitored by two Acoustic Emission (AE) sensors linked to a Vallen AMSY4 system. The scope of AE analysis was twofold: a) to verify the usefulness of typical AE parameters [1-3], as the hits evolution, energy, and amplitude, in identifying the different failure modes occurring during loading; b) to attempt an effective localisation of the damage, rendered difficult by the acoustic anisotropy of the material [4]. After tests, some specimens were sectioned and microscopically examined, in order to verify their failure modes. It was found that, before the pipe burst, a dense network of resin microcracks, resulting in water leakage, is generated in the shell (Fig. 1a). This phenomenon impairs the possibility of actually achieving burst with evident fibre failure. It was found that the critical stress for leakage of the 0/90 is slightly lower than its 55 counterpart. Further, the AE response is strongly influenced by the laminate type, with the 55 laminate emitting much more than the 0/90 laminate. In all cases, the resin microcracks were characterised by low amplitude signals.