Practical Applications of Diagnostic Data Science in Drilling and Completions

Abstract

The application of data science remains relatively new to the oil and gas industry but continues to gain traction on many projects due to its potential to assist in solving complex problems. The amount and quality of the right type of data can be as much of a limitation as the complex algorithms and programing required. The scope of any data science project should look for easy wins early on and not attempt an all-encompassing solution with the click of a button (although that would be amazing). This paper focuses on several specific applications of data applied to a sizable database to extract useful solutions and provide an approach for data science on future projects. The first step when applying data analytics is to build a suitable database. This might appear rudimentary at first glance, but historical data is seldom catalogued optimally for future projects. This is especially true if specific portions of the recorded data were not known to be of use in solving future problems. The approach to improving the quality of the database for this paper is to establish requirements for the data science objectives and apply this to past, present and future data. Once the data are in the right "format", the extensive process of quality control can begin. Although this part of the paper is not the most exciting, it might be the most important, as most programing yields the same "garbage in = garbage out" equation. After the data have found a home and are quality checked, the data science can be applied. Case studies are presented based on the application of diagnostic data from an extensive project/well database. To leverage historical data in new projects, metrics are created as a benchmarking tool. The case studies in this paper include metrics such as the Known Lateral Contribution (KLC), Heel-to-Toe Ratio (HTR), Communication Intensity (CI), Proppant Efficiency (PE) and stage level performance. These results are compared to additional stimulation and geological information. This paper includes case studies that apply data science to diagnostics on a large scale to deliver actionable results. The results discussed will allow for the utilization of this approach in future projects and provide a roadmap to better understand diagnostic results as they relate to drilling and completion activity.