Fatigue-life, fatigue-limits and delamination in oriented poly(vinyl chloried) pipes

The service life and fatigue performance expectations of poly(vinyl chloride) (PVC) pipes are increasing due to population growth and improvements in PVC performance. A fatigue-limit would simplify fatigue design. Fatigue-limits were proposed for unplasticised poly(vinyl chloride) (PVC-U) and modified poly(vinyl chloride) (PVC-M) [Brogden, 2000, Whittle and Teo, 2005], but the quantity of supporting data was small. Fatigue-life and fatigue-limit variability were not considered in the proposals. Debate within the Australian plastic pipes industry regarding fatigue-limits necessitated its verification. Furthermore, fatigue-limit variability and its effect on fatigue-life was not known. The Random Fatigue-Limit model [Pascual and Meeker, 1999] was applied to fatigue-life data sets for PVC-U and PVC-M [Brogden, 2000, Folkman et al., 2006, Joseph and Leevers, 1985], and to oriented poly(vinyl chloride) (PVC-O) fatigue-life data generated in this work. Fatigue-life variability was quantified for the first time and fatigue-limits were verified for all PVC pipe materials. Fatigue-limit variability was found to significantly affect the predicted fatigue-life variability. Fatigue design equations from the Random Fatigue-Limit models were proposed. The PVC-O pipe studied was a commercially sourced material. The PVC-O fatigue-life data generated in this work was acquired using the Ring and D-block testing geometry. This geometry enables a tension-tension load cycle to be applied, which simulates service loading conditions. Fatigue crack initiation and failure oc-curred on either internal surface of the ring, at the gap between the D-blocks, or on the external surface at angles between approximately ±45◦ . The stress state of the external surface of the ring was assessed and in some cases shown to exceed the stress of the internal surface failure. However, due to variation in radius of curvature and thickness of the commercial pipe, the stress on the external surface could not be accurately quantified. A pragmatic lower-bound fatigue stress was assumed to be the internal surface stress. Although in some tests, fatigue stress on the external surface may have been larger than lower-bound stress. The pragmatic lower-bound therefore predicts some premature fatigue failures that apply to this laboratory data only. The direction of fatigue crack propagation in PVC-O is not perpendicular to the tensile stress. This crack behaviour is referred to as delamination in this work. The angle of crack propagation is important for fracture mechanics-based fatigue design, because the angle affects the critical crack length. To improve the fatigue design for PVC-O, prediction of angles of crack propagation is essential. Unequal additive distribution in orthogonal directions through the pipe wall was suspected to cause delamination observed in PVC-O. However, x-ray photoelectron spectroscopy showed additive distribution was uniform. However, significant but small increases in additive concentration were identified on fatigue crack surfaces of PVC-O. Nevertheless, the measured increase was reasoned to be too small to create the delamination. A mixed-mode fracture mechanics model [Saouma et al., 1987] predicted angles of crack propagation in anisotropic solids. The model predictions correlated well with observed angles in PVC-O. However, the assumed fracture toughness of the PVC-O may be unrealistically large, and further refinement of the model for fatigue is required. It was concluded that the elastic and fracture toughness anisotropy of PVC-O caused the delamination.