Research on Friction Pressure Prediction of hydraulic fracturing Based on RBF Neural Network

Abstract

With the increasing of fracturing scale and displacement, the risk of engineering operation is increased because the high friction of string. Therefore, it is very important to predict the friction in fracturing operation. Used the local approximation characteristics of the RBF, the model between different factors and friction is established based on the data of laboratory test and field test. Thus, the corresponding relationship between multiple factors and friction is formed. This model predicts the friction of fracturing fluid in the future. The friction error predicted by this method is only within 9%. The prediction error of the same region is smaller than that of the classical mathematical model. This method effectively avoids the problems of complex friction mechanism and modeling difficulty. It solves the complex nonlinear problems and provides a new idea for the friction prediction method.