Deriving rock strength from MSE and drilling data
As the industry works to reduce costs and enhance completion techniques, engineered completions have emerged as a promising method to improve hydraulic fracturing efficiency. However, the method remains cost and labor intensive, limiting widespread adoption. A cost effective and easily implemented approach to engineered completions is needed. A data driven method utilizing Mechanical Specific Energy (MSE) has been proposed to denote relatively homogeneous sections of rock along the wellbore using only commonly available drilling data. This work investigates the MSE based engineered completions methods presented in the literature, and argues that the parameters which drive the MSE term may be more compelling indicators of rock heterogeneity. Additionally, automated pattern recognition methods to identify characteristic parameter response behaviors to either a rock strength or drilling efficiency change are explored. A Random Forest algorithm for defining characteristic parameter behaviors is presented and discussed, indicating promise for machine learning methods to define a library of parameter responses to energy changes that can be automatically detected while drilling the well, with positive implications for both completions design and drilling optimization.