CRITICALITY LEVEL ASSESSMENT FROM ILI DATA

摘要

In the last 10 years, technical and economical efforts have been made to improve pipeline integrity management. Those efforts focus on developing "searching tools", capable of identifying pipe mechanical damage due to slow landslides. We identified two main tools: geohazard mapping and inline inspection (OCP is using caliper with inertial navigation system INS). The INS system generates a substantial amount of information about pipe's geometry and deformation, reported as pitch, yaw and distance cover for each run. Since the caliper has been used for years, the pipeline's path of evolution over the years is already available. The INS data was merged with pipeline field inspections to develop an assessment tool based on Machine Learning Technology. This tool was applied to the complete path of the pipeline, analyzing each girth weld, thus obtaining a so called "criticality level" for each weld. Two models were evaluated, which differ on the size of the vicinity considered for each girth weld: 250m and 500m. The highest precision model was found with 250m, which already has allowed improvements in field inspections. This paper will describe this technique, capable of improving OCP's pipeline integrity management.