Técnicas de ensaios não destrutivos aliadas à inteligência artificial para avaliação da integridade estrutural de equipamentos offshore da indústria de petróleo e gás natural /

Abstract

Presence of discontinuities in pipelines can lead to the leakage of oil and/or its derivatives, damaging the transport of these substances, causing economic losses and causing pollution to the environment. With this concern, Non-Destructive Testing (NDT) is achieving a prominent position in the petrochemical industry to detect the reduction of pipeline thickness due to the corrosion phenomenon. In the present work, both the ultrasound technique and the magnetic test with Hall sensor were used to determine the integrity of the ducts. In order to do this, a medium carbon steel duct was used, in order to present internal holes of different depths, simulating different degrees of severity (corrosion levels). These NDTs were allied to Artificial Neural Networks (ANNs), with the objective of facilitating inspection, streamlining it and reducing the need for highly qualified professionals, allowing a greater accessibility of these techniques. The developed RNAs receive the information from the NDTs and provide as a response the thickness of the region under analysis, indicating whether or not there was a reduction in thickness, and how much it reduced, informing the existence or not of the phenomenon of corrosion. They are able to locate the discontinuities and determine how deep they are. Another interesting check was the fact of the importance of using more than one NDT, because one helps the other in its limitations, allowing to cover a greater field of applications.