Gueugnaut, Berthier, Darut, Constantinescu
# 2004 Milan
It is well known today that rock impingement is one of the major cause of long term failure of PE pipes when in the ground. This phenomenon which results of an intermittent or permanent contact of a hard object – most often not directly injurious – with the PE pipe can lead to premature brittle failure within 10-20 years after laying. The localized increase in the tension in the pipe generates the progressive cracking from the inside toward the outside, in line with the hard point, over a few millimeters in the pipe inner surface. This cracking reveals in the form of a split – generally longitudinal – whose size and geometry depend on the external loading applied to the pipe and the intrinsic SCG resistance of the PE material. A lot of experimental - and a few numerical - studies have been carried out regarding rock impingement resulting in a lot of data sometimes contradictory or unrelevant for further lifetime predictions. On the experimental level the wide variety of both the geometries and the loading conditions applied generally lead to individual sets of data which are in turn very difficult to interpret and to generalize. It is commonly demonstrated that using a detergent accelerates the rock impingement phenomenon but in turn forbids any prediction in terms of long term performance of the pipe in air. Furthermore using a detergent can lead to non-realistic failures if the loading conditions are not representative. Moreover, - and there is really the problem – applying constant load, constant strain or initial constant deformation can lead to different and non consistent results. On the numerical level, the only few attempts available are definitely not capable of describing correctly the problem since they do not take into account the real elastic-viscous-plastic behavior of PE as a whole. Then today there is a real need for a correct description of the rock impingement process from the laboratory testing to the modellization both being intimately linked. As a matter of fact the input conditions of a representative experiment – temperature, load, strain - should be fixed by both a fine analysis of samples from the field and the output conditions of the model. The analysis of failed samples allows one to evaluate very accurately the “current” size and geometry of the zone on the pipe engaged in the impingement process. In parallel, these data are taken into account in the model in order to define more precisely the spatial extent of the perturbated zone under constant load or initial constant deformation. The link with SCG is made using the anelastic strain as a hidden parameter to describe the damage in the pipe midwall. Once these conditions have been fixed by the model, they are applied on a very limited number of pressure experiments in order to validate the numerical data with regards to both the failure type and the time to failure.