Mud motor failure analysis using surface sensor data features and trends
Mud motor failure is a significant contributor to non-productive time in North American land drilling operations. Currently, there is limited work on mud motor failure analysis using surface drilling data. The objective of this work was to apply data analytics methodologies on historical drilling datasets to identify features and trends in surface data that contribute most to indicating impending mud motor failure. The methodology involved investigating a large dataset of mud motor runs for certain on-bottom and off-bottom drilling events that are generally known to be leading causes of motor failure. Spikes in differential pressure, pick up practices, drill-off time, etc., were all investigated. Additionally, the impact of temperature on mud motor failure was studied. A combination of the investigated features was analyzed using statistical measures and supervised machine learning algorithms to determine the leading contributors to motor failure. The dataset consisted of 32 motor runs drilled in the lateral section of wells from early-to-mid 2019. These motor runs represented a mix of both failure and non-failure cases. The motor stalls were categorized as either low, medium, or high impact depending on the severity of the differential pressure spike. It was observed that high impact motor stalls and the rate of stalling correlated strongly with eventual failure. Likewise, prolonged exposure to high bottom-hole temperature (exceeding the motor temperature limit) increased the risk of motor damage, most likely through thermal expansion of the stator elastomer. The drill-off time was also found to have some influence on motor failure for the investigated cases. Using these features, statistical significance tests and supervised machine learning models were trained on the dataset to determine the features that contribute most to motor failure. This work shows the feasibility of performing mud motor failure analysis using readily available surface data. It is expected to inform drilling engineers on data analysis methodologies for eﬀicient failure analysis and help them plan future wells more eﬀiciently.