An Artificial Neural Network Approach for the Prediction of Water-Based Drilling Fluid Rheological Behaviour

Abstract

It is well known that high temperatures, which change the rheological properties of the drillingfluid and can frequently cause problems in deep wells, is a major problem during drilling. Theimportance of the estimation and control of the rheological parameters of the drilling fluid and thehydraulics of the well increases as the depth of the well drilled is being increased to explore newoil, gas or geothermal reserves. Since it is difficult to measure these parameters with standardfield and laboratory viscometers, different conventional measurements and regression-analysistechniques are routinely used to approximate the true rheological parameters. In this study,water-based drilling fluid was initially prepared and rheological properties of the fluids weremeasured under elevated temperatures using high temperature rheometer (Fann Model 50 SL).Then, the shear stresses of drilling fluid are predicted using artificial neural network (ANN)method depending on the elevated temperature and shear rate. The results obtained from the hightemperature rheometer and artificial neural network were compared with each other and analyzed.Consequently, it is observed that the artificial neural network could be used with goodengineering accuracy to directly estimate the shear stress of drilling fluids without complexprocedures. The testing process shows that the average percentage error was found to beapproximately 2% for the prediction of shear stress values. Hence, rheological parameters of thedrilling fluid could be determined quickly and controllability was facilitated using artificial neuralnetwork structure developed.