Optimizing Electrical Submersible Pump Operations with AI/ML-Driven Real-Time Event Detection Systems

In the Bakken, holes in tubing caused by rod-on-tubing wear are one of the most prevalent mechanisms of downhole failures in rod pumped applications, especially in deep, highly deviated wells. A common mitigation method involves using borided tubing in sections where tubing splits occur, typically near the pump where compressive and buckling forces are highest. Installing borided tubing along the entire length of this section would be favorable, however, this approach is cost-prohibitive and wasteful if wellhead Electromagnetic Interference (EMI) scanning determines that the tubing is unfit for reuse. The objective of this study is to explore economical ways to extend the borided section of tubing by focusing on the accuracy and precision of the data interpreted from EMI scans of the borided tubing. 
The methods in this study involved collaborating with Stress Engineering Services to utilize their Bore Erosion Measurement and Inspection System (BEMIS™) for high resolution mapping of surface wall loss in used borided joints of tubing. With more than 30% wall loss, previous EMI scanning during workovers suggested that these joints of tubing were deemed unusable (red/green grade). Pipe samples were scanned at the wellhead, then separated and transported to a designated location to benchmark their relative thickness readings against the BEMIS™ device measurements.
The results of the scanning study evolved through three phases with increasing scope. In the first phase, two red/green joints were cut into 5-6’ lengths and shipped to Stress Engineering in Houston. The results from this phase did not detect any defects. In the second phase, thirty-eight red/green joints were sent to Houston, resulting in a 97% pass rate. Of the thirty-eight joints scanned, 89% were still in yellow condition, three joints were in blue condition, and only one had a surface defect greater than 30%. The third phase involved scanning 170 joints of red/green tubing, which resulted in a 94% pass rate. Although the distribution of blue tubing increased in the third phase, the gap between the BEMIS™ system and EMI scanning was evident. A portion of the surface features found during the laser scanning were deeper than the boride coating penetration depths, but the 3D rendering showed these areas were isolated and few in quantity. Through three phases of tests, the consistent pass rate allowed ConocoPhillips to confirm that EMI scanning is incompatible with accurately reading true wall loss in borided pipe, often skewing high and leading to significant waste.
In conclusion, the data from this scanning project has given ConocoPhillips the confidence to re-run significant quantities of borided pipe. This approach allows for cost-effective reduction in the purchase of new borided pipe and extends the borided section to combat wear. However, there are still risks associated with re-using borided tubing and limitations inherent to the technology used in this study. Variances in the boride coating and potential wall loss missed by the device remain possibilities. Despite these risks, the accuracy and reliability of the results from this trial provide high confidence that significant cost savings and improved runtime on rod pump wells can be achieved.
This project could not have been done without support and assistance from Stress Engineering personnel Brandon McGinn and Jason Waligura and technical support from Craig Zimmerman with Bluewater Thermal Solutions.