Northeastern Section - 57th Annual Meeting - 2022

Paper No. 27-2
Presentation Time: 8:00 AM-12:00 PM

AN EVALUATION OF REAL-TIME, ONBOARD IMPLEMENTATION OF GIS AND GROUNDWATER FLOW MODELING TO CONDUCT AUTONOMOUS SEARCH FOR ABANDONED PETROLEUM WELLS USING UAS MAGNETOMETRY


BOWIE, Mariah1, HARTLOVE, Juliana1, HELMKE, Martin1, SPATAFORE, Joseph2, SCHULTZ, Gregory3 and MASTERS, Andrew3, (1)Department of Earth and Space Sciences, West Chester University of Pennsylvania, 750 S. Church St., West Chester, PA 19383, (2)Wood Environment and Infrastructure Solutions, Inc., 751 Arbor Way, Suite 180, Blue Bell, PA 19422, (3)White River Technologies, Inc., 115 Etna Road, Bldg. 3 Ste 1, Lebanon, NH 03766

UAS (Unoccupied Aerial Systems) equipped with magnetometers are now routinely employed to locate abandoned petroleum wells. These surveys generally rely on pre-assembled, autonomous grid missions that are time-consuming to construct. Such inefficiencies become particularly challenging when clearing large land areas that would be required to locate the estimated 3 million abandoned petroleum wells in the United States.

We propose an integration of technologies, including an autonomous UAS, magnetometer, onboard computer, geographic information system (GIS), and groundwater modeling software, to improve the efficiency of abandoned well detection. This study tested this approach using an autonomous vehicle equipped with a Pixhawk Cube flight controller and Raspberry Pi 4B companion computer in a hardware-in-the-loop (HIL) simulation. Magnetic anomaly data previously collected by a White River Technologies MagPi total field atomic magnetometer were used to represent the magnetic anomaly of an abandoned well casing. Terrain avoidance, viewshed analysis, and flight planning were performed on the vehicle in real time (during mission) using PyQGIS. Magnetic anomaly inversion and target location were performed on the UAS using MODFLOW, PEST, and FlowPy compiled for the Raspberry Pi. Adaptive waypoint commands were sent to the vehicle in-mission using the PyMavlink library.

This experiment demonstrates that integrating UAS and sensor technologies with onboard, real-time GIS flight planning and magnetic inversion with adaptive flight planning is possible, providing far more capability than previously achieved for this purpose. Stochastic simulations reveal that magnetic inversion used to direct flight paths is twice as efficient as traditional methods for locating abandoned wells. On-board, real-time flight planning further improves performance and offers a robust solution for practical deployment of UAS magnetometry.