f Shale Developments: Use of Modern Data Mining Methods to Interpret Similarities and Differences Between Gas and Oil Completion/Stimulation Strategies

The presentation and discussion will focus on results of statistical analysis of Barnett Shale gas and Bakken Shale oil production result drivers. Large data sets including key reservoir parameter proxies, well architecture information, completion variables, and stimulation data were compiled, quality controlled, and analyzed to identify key production influences. The analysis and interpretation took into account both controllable influences on well productivity, e.g., well length, azimuth, completion type, stimulation size and materials, along with uncontrollable influences, e.g., fracture bounding bed presence / absence, fracturing into unknown Geohazards. The most obvious production driver is well location, a fair proxy for reservoir quality in the “shale” plays. Well architecture, including azimuth, length, and drift angle may or may not be a major determinant of productivity. It is apparent from both the Barnett and Bakken data sets that longer well lengths do not produce proportionately more hydrocarbons. Specific completion and stimulation parameters, e.g., use of coarse-mesh proppants are also important productivity drivers in certain circumstances.