A new approach to determine the rheology parameters for water-based drilling fluid using artificial neural network

Rheological properties play key roles in hydraulic calculations and hole cleaning. Determination of the rheological properties in real time is very important to control the drilling operation and save time. Losing the rheological properties can lead to severe problems such as; pipe sticking, kick and blowout, increase in torque and drag, and hole cleaning issue. Rheological properties can be determine using the rheometer, which required time for measurement and also for cleaning the equipment after use. In the well site, rheological properties are measured twice a day. The objective of this paper is to develop a new technique to determine the rheological properties (plastic viscosity (PV), apparent viscosity (AV), yield point (YP), and flow behavior index (n)) of KCl water-based drilling fluid using the frequent measurements of the caliper parameters (drilling fluid density (D), Marsh funnel viscosity (MFV) and solid percent). D, MFV, and SV are measured frequently every 15-20 minutes in the well site. Artificial neural network (ANN) was used to build different models for PV, YP, AV, and n based on 3000 field data measurements. ANN was able to predict the rheological properties on a real time with a high accuracy. Four new empirical correlations were developed to predict PV, AV, YP, and n. the accuracy of these correlations are very high (correlation coefficient (CC) greater than 90% and the average absolute percentage error (AAPE) was less than 6%. The novelty of this technique is the prediction of the rheological properties in real time based on the caliper variables, which will help the drilling engineers to perform a safe drilling operation and eliminate the common drilling problems.