Data-Driven Optimization of Drilling Parameters

Abstract A major challenge during drilling is to identify sub-formation change, enabling near-real-time adjustment of the drilling parameters to enhance performance. An algorithm that determines sub-formations and optimizes drilling parameters as formations change and then updates itself solves this problem. This paper presents case studies of wells in the New Mexico Delaware Basin with improved drilling performance, detailing the algorithm's design and ability to produce formation-specific drilling parameters. The workflow leverages historical drilling and geological information from offset wells, statistical analysis, physics-based models, and 1 Hz drilling data to identify sub-formations and transition zones in active wells. Rock drillability is calculated and used to identify the formation change via the algorithm. Simultaneously, optimal drilling parameters for the formation are generated, using an approach that relies on mechanical specific energy (MSE). The workflow was successfully applied to a significant population of wells in multiple fields. Analyzing the results, the main observations were as follows: reduced cumulative MSE and average drilling time, improved consistency, and expedited learning curve. These achievements to a large extent are due to improved downhole dynamics and the reduction of input variability due to the statistical analysis of the algorithm. In addition, self-learning capabilities are built into the algorithm, which can improve its own performance with the addition of offset data during field development. In general, three variables have major effects on drilling performance: surface controls, downhole equipment, and geology. Among these variables, geology is the most uncertain and yet crucial to success, as changes in rock type can significantly affect drilling performance. The power of the algorithm comes from the fact that it quantifies geological effects along with the effects of surface controls and BHA to significantly improve performance. This enables the algorithm to determine the formations that affect drilling operations and suggest the optimum drilling parameters for that formation zone. This work combines real-time and historical drilling and geological data to identify the sub-formation changes and proposes the optimized drilling parameters. As observed in the case studies, this can lead to a significant improvement in drilling performance, near-real-time optimization of the drilling parameters, and minimizing non-productive time due to sudden changes in formation type.