In 2010, I attended a training session on stuck-pipe prevention that left a lasting impression on me. One statement from the instructor resonated throughout the course: “The well always speaks to you; you just have to listen.”
The language of the well is conveyed through various parameters, including hook load, weight on bit, rotational speed, torque, pressure, and cuttings returns. The journey to understand this language has driven advancements in standardization, automation, technology, and science within planning and operations.
Over the past decade, the application of data science has significantly enhanced our understanding of these parameters and our ability to interpret them effectively. The technical papers discussed here present innovative methods and techniques for recognizing trends that may lead to stuck pipe.
Paper SPE 220991 hypothesizes a correlation between stuck-pipe incidents and increasing friction factors, implementing a graph attention network to evaluate predicted friction factors. The authors’ goal is to identify trends that may result in heightened friction because of factors such as cuttings buildup, excessive doglegs, or well caving. Their algorithm was able to provide an early warning 8 minutes before a stuck event, offering valuable time to take preventive action.
Similarly, paper SPE 221094 introduces a technique that uses time-series data from rigs to identify trends in measured parameters indicative of potential stuck conditions. Their model was trained on data from 19 historical wells and subsequently tested on 22 different wells, successfully detecting stuck-pipe incidents in all cases, with warning times as early as 2 hours.
Paper SPE 222762 presents a new method to assess the overall health of the wellbore by introducing a comprehensive index for the state of the hole. The overall health of the hole is quantified using a comprehensive stuck-pipe index (CSI), which indicates the runnability of a drillstring or a casing string. Several physics-based and data-based models and parameters, including curvature, tortuosity, mud properties, overpull limit, and offset data are used in evaluating CSI, which also can provide a warning for differential sticking. Such an approach can help in providing an elegant way to evaluate the health of the borehole and preemptively determine the stuck conditions.
Collectively, these papers contribute to a deeper understanding of the language of the well and its implications, furthering our ability to mitigate risks associated with stuck-pipe incidents.
Summarized Papers in This July 2025 Issue
SPE 220991 Approach Provides Early Warning of Stuck Pipe by Muchen Liu, China University of Petroleum, et al.
SPE 221094 Stuck-Pipe Prediction Method Uses Trigger Coefficients by Xiaoyan Shi, CNPC, et al.
SPE 222762 Time-Series Analysis Enables Early Stuck-Pipe Detection by Robello Samuel, Halliburton, et al.
Aman Srivastava, SPE, is a product owner with Halliburton Landmark. He holds a bachelor’s degree in mechanical engineering from the National Institute of Technology in Surat, India, and a master’s degree in petroleum engineering from the University of Oklahoma. Srivastava has more than 16 years of experience in oil and gas and holds a special interest in well-construction engineering and energy/sustainability. He received the Regional Management Award for the SPE Mid-Continent North America Region in 2023. Srivastava currently is an adviser for The Way Ahead magazine, a reviewer of peer-reviewed journals, and a committee member for SPE’s 2026 Annual Technical Conference and Exhibition.