Residual Strength Prediction of a Pipeline with Interacting Corrosion Defects Subjected to Combined Loading Using Artificial Neural Networks

Abstract

Predicting the failure pressure of corroded pipelines has long been a topic of great interest for researchers all around the world. There are several methods and guidelines available to estimate the failure pressures of pipelines with the most commonly used being DNV-RP-F101. However, despite being the most comprehensive method, neither DNV-RP-F101 nor any other widely used corrosion assessment methods consider interacting defects subjected to both internal pressure and longitudinal compressive stress, despite this scenario being extremely common in the real world. In this work, the relationship between interacting corrosion defects and applied loadings (internal pressure and longitudinal compressive stress) with the failure pressure of pipeline is investigated by predicting the failure pressure using an artificial neural network (ANN). Data regarding the failure pressure of pipelines with interacting defects and combined loading is collected from FEA and full-scale burst tests. These data are then fed to an artificial neural network, allowing it to provide appropriate results for new cases, i.e., for other defects, corrosion defect parameters and loadings. This research will provide a gateway to help predict the failure pressure of pipes with interacting corrosion defects subjected to both internal pressure and longitudinal compressive stress using ANN. The greatest advantage of utilizing an artificial neural network is its ability and ease to improve failure prediction accuracy. Over time, more failure data can be fed into the artificial neural network, allowing it to “learn”, thus being able to provide better failure pressure predictions.