Pattern Recognition Techniques Implementation on Data from In-Line Inspection (ILI)
Abstract
Onshore pipeline failure caused by corrosion represents about 16% of the overall number of incidents during the period from 2004 to 2011 according to databases such as CONCAWE and PHMSA. In-Line Inspection (ILI) is one of the available inspection techniques used to determine overall pipeline status, highlighted because it establishes a clear perspective of inner and outer condition of the pipe against the failure modes and wall thickness. Furthermore, it supports measures to prevent risk based on standards such as ASMEB31G or API579-1/ASME FFS-1. However, this approximation could represent a conservative assessment of the pipeline status, taking into account the uncertainty associated with ILI inspection tools such as MFL and UT. Several researches have been conducted to analyze available inspection techniques attempting to reduce noise generated by their inspection tools, and determine procedures in order to establish correct metal loss detection, excelling pattern recognition analysis and reliability concepts. Therefore this work seeks to transform a set of data obtained from two ILI runs, into useful information to support decision making in risk analysis based on pattern recognition techniques and reliability concepts, in order to obtain base failure frequencies for prior analysis from individual and grouped flaws. Moreover, growth corrosion and remaining life models supported on the standards mentioned above were evaluated using a pressure failure criteria. As a result it was obtained that the failure probability of the grouped flaws increases 10% in comparison with the corresponding flaws evaluated individually.
Description
PresentationSubject
Onshore pipelineCollections
Citation
Amaya-Gómez, Rafael; Sánchez-Silva, Mauricio; Sánchez-Akli, Andrés; Muñoz-Giraldo, Felipe (2015). Pattern Recognition Techniques Implementation on Data from In-Line Inspection (ILI). Mary Kay O'Connor Process Safety Center; Texas &M University. Libraries. Available electronically from https : / /hdl .handle .net /1969 .1 /193721.