Abstract
The United States Pipeline and Hazardous Materials Safety Administration (PHMSA) recently revised the federal rules governing natural gas transport. PHMSA added a new section on the verification of pipeline material properties for pipeline assets with insufficient or incomplete records. This section permits the use of nondestructive examination (NDE) technologies to estimate material properties, which include yield strength (YS) and ultimate tensile strength (UTS), if several conditions are satisfied. These include that NDE measurement accuracy and uncertainty be conservatively accounted for, that the NDE technology be validated by experts, and that proper calibration procedures be implemented. One such NDE technology is Instrumented Indentation Testing (IIT), which can be used to estimate YS and UTS.
Precise quantification of any NDE technology’s precision and accuracy requires consistent identification of test errors: if an error occurs during a measurement such that the data should be excluded from subsequent analyses, analysts need to be alerted to the data characteristics prior to including these results. These testing errors are distinct from the inherent measurement uncertainty due to both random error and systematic error. Any NDE measurement will contain some degree of uncertainty; however, faulty measurements exhibiting clearly identifiable errors must be excluded from subsequent analyses to maintain the integrity of the data set.
Accordingly, this paper extends Pacific Gas and Electric’s (PG&E’s) previously reported efforts on IIT uncertainty quantification by presenting observations of a specific type of IIT error related to tool fixturing that has occurred during in-situ testing and describing the characteristics of how this error was exhibited in the test data. Once this test error was clearly identified, isolated, and was found repeatable; pre-processing algorithms were adapted to detect and alert NDE technicians to this error during testing, ultimately evolving NDE work procedures. This paper discusses this process from the initial recognition of a test error, to the adaptation of appropriate detection algorithms, and then finally to resulting revisions in operator procedures. Ultimately, these modifications have improved validation data quality and reduced the error rate of IIT measurements collected in the field.