Graded EMAT Performance Specification Validated in Blind Test

Abstract

The management of Stress Corrosion Cracking (SCC) represents one of the challenges for pipeline operating companies with pipelines potentially susceptible to SCC. In order to help better support the management of SCC, a graded performance specification for the high-resolution Electro-Magnetic Acoustical Transducer (EMAT) In-Line Inspection (ILI) technology is derived which provides higher levels of confidence for detection of crack-field anomalies with critical dimensions. This paper presents the process used to derive the graded performance values for the EMAT ILI technology with regard to SCC. The process covers the Probability of Detection (POD) and Probability of Identification (POI). A blind test was carried out to derive the graded performance specification. Test data set was compiled comprising EMAT data for several joints containing relevant anomalies and neighboring joints, some containing additional shallow SCC. These joints had been dug based on EMAT ILI data and all of the joints were evaluated with 360 degree Non-Destructive Examination (NDE) followed by destructive testing. For each target joint with relevant anomalies to be assessed, four additional joints were added in random order to generate a realistic density and distribution of anomalies. Furthermore, pipe joints with non-crack like anomalies, as well as pipe joints with mixed populations were included in the blind test data set to ensure a realistic feature population and to assess POI without side effects of a weighed feature population inside set of composed ILI data. The data set was then evaluated by multiple analysts and result from each analyst were evaluated and utilized to derive the POD and POI values for the graded specification. In addition, the full process of data analysis including team lead review was carried out for one of the analysts for comparison to the individual analyst results. Anomaly dimensions were compared against the true population to derive the POD and POI values. Furthermore, length and depth sizing performance was assessed.