Risk analysis for pipeline assets-The use of models for failure prediction in plastics pipelines

Abstract

A key component in any risk-based asset management strategy is the ability to predict when assets will fail. For pipe materials such as cast iron (CI) and asbestos cement (AC), sufficient failure data are available to allow statistical failure models to be developed. However, for newer materials such as polyvinylchloride (PVC) and polyethylene (PE), only limited failure data are available. Lifetime prediction for brittle polymers can be based on Linear Elastic Fracture Mechanics (LEFM) theory. LEFM uses the concept of a single parameter known as the applied Stress Intensity Factor (SIF) that characterizes the stress field at the tip of a crack in a plastics material. The dependence of crack growth rate on the applied SIF can be determined using small laboratory scale tests on coupon samples, which can then be applied to the geometry of a flawed asset in service via a geometrical correction factor. While this approach can be applied to comparatively brittle materials such as unplasticized polyvinylchloride (PVC-U) and older PE materials, newer PE materials can form a large craze zone at the crack tip, which renders LEFM invalid. An extension to LEFM theory, Elastic Plastic Fracture Mechanics (EPFM), can be used to model post-yield fracture that occurs after plastic deformation in ductile materials. However, care must be taken to distinguish plastic fracture from crack tip blunting. In the case of crack tip blunting failure mechanism, EPFM failure predictions cannot be transferred from small test geometries to larger pipes in service and other approaches such as craze mechanics theory must be applied. When developing a failure model for new polymers with high, slow crack growth resistance, it is essential that the underlying theory (e.g., LEFM) is validated. Recent advances in failure mechanism research has led to a novel test method which isolates the craze formation and separation process in tougher pipeline materials such as the newer bimodal PE materials. By comparing measured craze strength from this new method with material yield strength, criteria can be applied which allow selection of the appropriate lifetime prediction methodology. © Springer Science+Business Media, LLC 2009.