Early Warning Method to Detect Mud Motor Stallings Incidents in Drilling Operations Using Machine Learning

In recent years and with the increase of complex operations such as directional drilling, there has been an increase use of mud motors to drive performance and deliver more complex well design. The challenge is that motor failures increased due to the complexity of the operation which tends to have a significant impact on the well delivery and cost. Like any equipment failure, it is impacted by the conditions the tool has been through and it often provides prior warning signs that if missed will have significant non-productive time. One of the major motor failure signs is the motor uncontrolled stalling. The objective of this paper is to present an early warning method to detect and alert of mud motor stalling to the rig crew to react before it leads to other undesired conditions. Motor stall occurs when the mud motor stops working which will cause stalls with a high-pressure reading. Using daily drilling reports (DDRs) and real-time data surface parameters, subsurface data, a dataset of incidents was created by SMEs and it is fed into a machine learning model to detect early warning signs of stalling and provide the rig crew with detailed reports in real-time. After applying the proposed data-driven method, the stalling events can be accurately detected with +80% accuracy and a low false positive rate. To ensure proper mud motor monitoring; DDR, BHA, and relevant data such as motor specifications are considered. The result of such work is impacted by the data quality of real-time data such as uncalibrated channels, improper units, and significant noise in the data. This work extends the existing literature by using surface and subsurface data to accurately detect early warning signs of motor stalling. Our work presents a complete framework for how stalling events can be detected and reported in real-time.