10 A Probabilistic Method to Predict Nominal Wall Thickness

Abstract

Under the PHMSA 2019 Gas Transmission Rule, Operators must opportunistically verify material properties, such as nominal wall thickness (NWT), for pipeline features lacking traceable, verifiable, and complete (TVC) records. Non-destructive examination (NDE) such as ultrasonic testing (UT) might be performed on one or more locations to measure the wall thickness. The number of measurements collected can vary by several orders of magnitude depending on the technique used, ranging from manual 12-point UT measurements to automated UT measurements which can scan the entire pipe surface generating tens of thousands of measurements. Prior to the PHMSA 2019 Gas Transmission Rule, Operators had to assess NWT following §192.109, which by design resulted in conservative outcomes. Because NWT has a first order impact on design pressure, excessive conservatism risks reducing operating capacity. This can have serious financial and operational consequences including pressure reductions and disruptions to normal operations. Operators could benefit from developing their own procedures for verifying unknown NWT using NDE under §192.607; however, they must demonstrate that the new procedures meet the special requirements for nondestructive methods and provide an equivalent or better level of safety to §192.109.

The Pacific Gas and Electric Company (PG&E) has developed a methodology, known as the Confidence Interval Method (CIM), to assess unknown NWT intended to satisfy the requirements of §192.607 by demonstrating a conservative accounting for measurement inaccuracy and uncertainty. PG&E collaborated with Kiefner and Associates, Inc. to implement CIM in a software tool. CIM uses statistics and probability theory to combine in situ wall thickness measurements, industry standards for NWT dimensions, and manufacturing tolerances encapsulated in historical editions of American Petroleum Institute Specification 5L Line Pipe. A comparison between CIM and §192.109 was performed using a validation dataset consisting of in-situ wall thickness measurements taken on pipes with TVC NWT of record. It was observed that CIM more accurately predicted the TVC NWT of record compared with §192.109. CIM addresses the uncertainty in measurement data, and when necessary, makes conservative assumptions with 95% confidence levels. The paper will also discuss how the number of measurements influences the NWT assessment by comparing results where 12-point UT and automated UT were both performed and make recommendations on the minimum number of measurements that should be collected.

Read/Download PaperGradeAvailable after the event