Multiscale Approaches Including ANN and M5P-Tree with SI and OBJ Assessment Tools to Predict the Shear Thinning of Bentonite Drilling Muds Modified with Clay Nanosize at Various Elevated Temperatures

Abstract

Improving the flowability behavior of the water-based drilling mud (WBM) using bentonite under high-temperature conditions is significant in the drilling borehole operations. In this study, the influence of bentonite and clay nanoparticle (CNP) contents on the non-Newtonian fluid (rheology) properties such as initial shear stress and fluid’s resistance to flow (viscosity) under high-temperature conditions were tested and quantified. The shear stress and shear strain rate relationships of WBM were simulated using two new rheological models of the vapor pressure and modified Hoerl models, and the modeling outcomes were validated with the Herschel–Bulkley (HB) model. This study also analyzed and formulated more than 268 data sets, including observational data and data obtained from previously published studies ranging with bentonite in the drilling muds from 2% to 8% (%wt. of water), 0% to 1% of CNP (%wt. of bentonite), and heating temperatures were ranging from 25°C to 100°C. The HB rheological model predicted initial shear stress better than the other two rheological models depending on several statistical assessments. The initial shear stress of drilling muds can be predicted well based on bentonite, CNP contents, and temperature using various simulation techniques. According to several statistical tests, the nonlinear regression and artificial neural network (ANN) performed stronger than other model techniques to predict the yield stress of WBM as a function of bentonite, CNP contents, and temperature.