A novel pipe-cap system for corrosion protection and security

摘要

Magnetic flux leakage (MFL) inspection tools have been used successfully for many years to estimate the dimension and severity of corrosion features in transmission and distribution pipelines.MFL is an indirect measurement system and a mathematical model or algorithm is used to translate MFL signals recorded by the tool into estimations of pipeline corrosion feature depth, length, and width, respectively.The derivation of these mathematical models can be anything from a simple calibration process of a representative sample of manufactured features in a pull rig, to an in-depth development process of highly complex algorithms.rnOne of the challenges to MFL technology is that the signal response,and hence sizing performance, is directly related to the shape of the corrosion. More importantly, since the MFL measurement technique is indirect and models are 'trained', in some cases when the geometries of real corrosion differ significantly from those used to develop sizing algorithms, higher than expected sizing error can occur.This situation is of course of most importance when corrosion is aggressive and can lead rapidly to deep, potentially injurious, defects.rnThis paper describes a process which uses a large volume of statistically significant high-resolution nondestructive examination (NDE) data as a primary means to train and optimize MFL sizing models when defect geometry is a major contributor to poor MFL inspection-sizing performance.The practical application of the method to a Canadian oil pipeline is discussed in detail along with the approach used to achieve greater than 90% sizing confidence on field verified automatic UT data.