Evaluating Reservoir Pressure Gradient Trend for the Delaware Basin’s Potash Area Using Machine Learning & Geophysical Log Cross-Sections Approach

Abstract

The reservoir pressure trend prediction for the potash area of Delaware Basin would enhance its optimum producible depths selection. It is significant for safe drilling, effective, and efficient governmental drilling permits approval in the area. Avoiding kicks, blowouts, fluid loss, pipe differential sticking, and heaving shales prevention improved wellbore control. This also leads to dependable wellbore integrity and better reservoir or well fluids control which are some of the benefits of proper reservoir pressure trend prediction. This study used the reservoir pressures predicted by Multilinear Regression machine learning model to verify the reservoir pressures calculated using drilling data from the potash area. Then, pressure trends are built for the area with Petra using geophysical log cross-sections. The results from these pressure trends are presented in 2-Dimensional and 3-Dimensional forms for the area to connect permitting optimum safely producible depths with hydrocarbon production. The study utilized drilling and well logs data from about 229 wells. All the wells were drilled and completed within the Potash Area to at least the base of Wolfcamp formation. The geophysical log cross-sections were created in 2- and 3-Dimensional forms using Petra, Matlab, and R machine languages. For the Multilinear Regression model over 330,000 data points from model parameters such as Deep & Shallow Laterolog Resistivities, Gamma Ray log, Neutron & Density Porosity Limestone logs, Sonic logs, caliper log, depth, lithology, mud weight, Photoelectric Cross-section, average porosity, water saturation, corrected bulk density log, and bulk density log were used. The datasets were grouped into 70 percent training and 30 percent testing randomly. The Multilinear Regression model predicted the reservoir pressures with high accuracy where the coefficient of determination (R2) is greater than 0.990. The Root Mean Square Error (RMSE) ranges from 0.0086 to 0.034 psi/ft between the predicted and the measured reservoir pressure data. The validation of the Regression model was done using another dataset. The reservoir pressures were predicted by the model with high accuracy using the validation dataset. The coefficient of determination (R2) is 0.99. This study shows that the regression model is reliable and can predict the reservoir pressures for the area accurately using well logs, drilling data, and geophysical data. Furthermore, verified reservoir pressures is used to build reservoir pressure trends for the area. The reservoir pressure trend can then be used to select the optimum producible depths in the area in order to promote safe, cost efficient, and optimum hydrocarbon recovery in the area. This study will also promote concurrent operations in prospecting for, developing, and producing oil and gas and potash deposits owned by the United States within the Designated Potash Area (DPA) (BLM Secretary Order, 2012).