Abstract

The Pipeline and Hazardous Materials Safety Administration (PHMSA) introduced a new final rule that became effective on October 5th, 2022. This rule encompasses various noteworthy provisions pertaining to leak detection, including its applicability to a wide spectrum of infrastructure, encompassing “Type A” gas gathering, gas transmission, and hazardous liquid pipelines, provided they have diameters of 6” (0.15 m) or more. In accordance with the newly established rule, there is a mandated requirement to promptly notify ruptures within a 15-minute timeframe for all “Type A” gas gathering, gas transmission, and hazardous liquid pipelines, provided they have diameters of 6” (0.15 m) or more. Furthermore, the rule stipulates that in the event of an identified rupture, the response protocol necessitates the closure of Remote Mainline Valves (RMVs) to isolate the affected pipeline segment within a 30-minute window. Compliance with the new rules for gathering pipelines, which connect terminals and processing facilities, is challenging. This is mainly because these pipelines often lack robust SCADA integration and adequate instrumentation. Implementing these improvements for short pipeline segments can also be cost-prohibitive.

In response to PHMSA's increasingly stringent regulations, this paper delves into an examination of the effectiveness of a leak detection system that exclusively relies on pressure measurements to ensure compliance. This system utilizes an unsupervised machine learning algorithm, leveraging high-frequency pressure data to promptly issue alerts in response to potential leaks. Throughout this exploration, the paper endeavors to provide readers with experimental data obtained from various fluid withdrawal tests performed on pipelines of differing operational complexities. Furthermore, it addresses the challenges associated with the use of pipeline pressure readings for leak detection purposes. Finally, the paper discusses the implementation of continuous machine learning to enhance the reliability of the leak detection system.

This content is only available via PDF.
You can access this article if you purchase or spend a download.