Modelling of Crack Arrestors on Gas-Pipelines

Design of gas pipelines implies demanding tasks such as the definition of material requirements in terms of ductile fracture propagation control. This task is especially challenging when high strength steels (like X100 and beyond), whose use is imposing as a result of the consolidated worldwide trend towards the realization of long distance/high pressure gas pipelines, are considered. In fact as recent research projects clearly demonstrated (Mannucci [1], Demofonti [2] and Papka [3]) it is not possible to rely on the self-arrestability of such a class of high strength steel and as a consequence new solutions must be adopted, especially when severe operating conditions (rich gas, low temperature, etc.) are envisaged. In this context, the adoption of additional mechanical devices, such as Crack Arrestors (CA), represents a valid alternative since, if properly designed, they can externally provide a structural positive contribution in terms of resistance opposed to ductile fracture propagation. Optimal design of CA presupposes the knowledge of the efficiency of each possible CA type, but also the influence of its main geometric parameters (thickness, radial clearance, axial length…) on the constraint applied on fracturing pipe, in the specific operative condition. The use of a specific finite element code named PICPRO, which was developed by CSM jointly with the University of Tor Vergata and actually used in the frame of a BP project, allows to perform numerical simulations of running shear fracture along a pressurised steel pipeline also considering the constraint applied to the running fracture by the crack arrestor; PICPRO was used to perform a rapid comparison within all the potential CA solutions and it allowed the successful identification of the best one for the case under consideration (Mannucci [1]). In the present paper some new algorithms, expressly developed to simulate crack advance inside a crack arrestor, are presented. Exploration of main CA design parameters imposes the use of algorithms different from those based on the search and application of contact elements, in order to reduce calculation time and to give the possibility to perform an higher number of simulations. Hence, two new algorithms are on purpose developed and hereby presented: the first one for the evaluation of the interaction pipe-CA and the second one for managing the fracture propagation criteria even if crack flanks cannot freely open because of crack arrestor presence