Results in Engineering (Jun 2023)
Horizontal well completions using data analytics
Abstract
There is no quick way to measure completion configurations using simple production data, and assessing them is a challenge for many operators. It has been demonstrated by the O&G industry that completion configurations of horizontal wells influence initial well production potential and long-term performance. The present paper introduces simple, but accurate models to predict completion variables for horizontal wells in the Wolfcamp shale. These include fracture stages, clusters and cluster spacing, and perforations. We believe that these models will help define the optimum completion variables for horizontal wells in unconventional resources. The model development is based on the analysis of hundreds of horizontal wells that include the production history and parameters affecting their production behavior, including but not limited to, well completion configurations, size of proppant, type of fluid, stages, and completed interval of the lateral. We find that six key parameters are essential to precisely predict and optimize the completion variables, namely county, reservoir type, proppant amount, fluid type, rectangular overlap between wells, and initial production. Relationships among these and other parameters and their effect on the production behavior of horizontal wells were evaluated using state-of-the-art data analytics (machine learning). Multivariate linear regression models were devised to predict the four completion variables. Publicly available Well Production Performance data were used as a separate criterion in cross-validating the model predictions out-of-sample. The results of this assessment demonstrate the precision of our models with an absolute relative error of the order of 13%. The true practical advantage of this work is not only in guiding future selective completion variables for horizontal wells in the shale play, but also in providing comparative metrics in assessing different completion styles of various basins using the production history of the offset wells.
