Lever
# 2004 Milan
It has long been established that the primary long-term failure mode of PE fittings is Slow Crack Growth (SCG) from the area of highest stress concentration. This area of highest stress concentration is usually situated at the edge of the fusion joint, be it a conventional socket, sidewall or electrofusion joint. Pipes and fittings are tested for a given pressure rating by subjecting assemblies to pressure tests at 80°C. The hydrostatic approval tests are designed with reference to the Stress Rupture Curves for the given material class as determined by ISO 9080 or ASTM D2837. These curves are generated using data gathered from the testing of standard pipe specimens. The validity of these curves with respect to more complex stress distributions has been questioned. Indeed, taking stress concentrations in pipe/fitting assemblies into account it is apparent that the fittings in the test assemblies are subjected to stresses, in certain areas, which would lead to ductile failure in pipe specimens. How then, can we explain the SCG failures generated in the fittings? Furthermore, can we claim that a fitting that has met the standard test requirements will indeed last as long as pipe at the same pressure and temperature in service? The author has noted that when true stress in assemblies, as measured using Finite Element Methods, is plotted against time to failure via SCG at 80°C, the failure points lie on the SCG failure line extrapolated above the knee of the ISO 9080 stress rupture curve. This result implies that the current testing approach is indeed valid for fittings and that standard bi-directional shift factors or full knowledge of the ISO 9080 regression coefficients for a particular PE material can be used to predict assembly lifetime. Using this concept it is possible to propose a standardized procedure for specifying a PE material capable of withstanding the stress concentrations inherent in fittings and guarantee an acceptable system lifetime for a given application. Furthermore it should be possible to use SCG data generated from fittings to determine the knee in modern materials where no knee has been found using pipe test specimens.
