Stuck Pipe Prediction Using Automated Real-Time Modeling and Data Analysis

Abstract

Abstract A real-time method is presented to predict impending stuck pipe with sufficient warning to prevent it. The new method uses automated analysis of real-time modeling coupled with real-time data analysis. It can be applied to all well types for any well operation including drilling, casing running, completion activities, and re-entries. The method uses leading indicators of stuck pipe that were identified using historical data sets of 36 stuck pipe incidents in the Eagle Ford, Utica, Permian, and Gulf of Mexico. Four case histories show the utility of the new method in four shale horizontal wells for both drilling and off-bottom operations. The new method combines two types of analysis: (1) deviation of real-time data from real-time model predictions using hydraulics and torque and drag (T&D) software, and (2) trend analysis (i.e., rate of change) of real-time data. Parameters used are pump pressure, flow rate, torque, rotary speed, hookload and drag, and weight on bit along with static inputs such as BHA and drillstring configuration and directional surveys. Additional parameters such as downhole ECD are used when available and improve the results. But the method is designed to monitor all well types and provide a stuck-pipe-risk log even using only basic instrumentation. A novel algorithm predicts the probability of stuck pipe which is presented in a real-time log. Results demonstrate there is no single leading indicator present in all stuck pipe incidents. Consequently, relying on a single specific pattern (such as increasing pump pressure at constant flow rate) leads to inability to predict stuck pipe in some cases. Our early detection method relies on multiple indicators to strengthen the likelihood of detecting an increase in stuck pipe while avoiding false alerts. Deviations of parameters from model predictions have been used with success in the past by experts but the new method provides contextual awareness and provides thresholds that automate the process. A key element is the use of filtering for rig activity. Rate of change calculations provide valuable insight into rapidly deteriorating wellbore conditions when deviation from model predictions does not respond quickly enough over a short depth or time interval. The detection method was tested on four historical wells in which four stuck pipe and two near-miss events occurred. Testing showed that an alert was generated, on average, 38 minutes before the event. There were no false alerts recorded in these wells.