Toward real-time detection of pipelines and spills

Thousands of kilometers of oil industry gathering pipelines have been installed in North Dakota, USA, to move enormous volumes of fluids from the wellheads of thousands of unconventional oil wells to various processing facilities. This increase in pipeline infrastructure requires state-of-the-art technology for pipeline monitoring to achieve the goal of zero incidents. To be effective, technology developed for monitoring pipeline systems should handle large amounts of data efficiently and provide real-time information. This paper presents the results from the first phase of a three-phase project to develop technology to detect pipeline location and leaks in real time. In this phase, raw magnetic data and images from a magnetometer and camera system attached to an unmanned aircraft system (UAS) platform were used for detecting buried steel pipelines in controlled experiments. Machine learning and data analytics workflows were successfully developed to integrate magnetic and image data using cloud and edge computing for real-time pipeline detection.